2004-10-27 00:31:04 +00:00
|
|
|
/*
|
1999-08-16 21:50:52 +00:00
|
|
|
*******************************************************************************
|
2008-01-03 10:07:47 +00:00
|
|
|
* Copyright (C) 1996-2008, International Business Machines
|
1999-11-23 01:30:04 +00:00
|
|
|
* Corporation and others. All Rights Reserved.
|
1999-08-16 21:50:52 +00:00
|
|
|
*******************************************************************************
|
2001-03-17 00:46:46 +00:00
|
|
|
* file name: ucol.cpp
|
|
|
|
* encoding: US-ASCII
|
|
|
|
* tab size: 8 (not used)
|
|
|
|
* indentation:4
|
|
|
|
*
|
2001-02-20 00:26:50 +00:00
|
|
|
* Modification history
|
|
|
|
* Date Name Comments
|
2001-03-17 00:46:46 +00:00
|
|
|
* 1996-1999 various members of ICU team maintained C API for collation framework
|
2001-04-23 01:53:49 +00:00
|
|
|
* 02/16/2001 synwee Added internal method getPrevSpecialCE
|
2001-03-02 01:14:03 +00:00
|
|
|
* 03/01/2001 synwee Added maxexpansion functionality.
|
2001-03-17 00:46:46 +00:00
|
|
|
* 03/16/2001 weiv Collation framework is rewritten in C and made UCA compliant
|
1999-08-16 21:50:52 +00:00
|
|
|
*/
|
2001-03-08 17:40:42 +00:00
|
|
|
|
2002-09-20 01:54:48 +00:00
|
|
|
#include "unicode/utypes.h"
|
|
|
|
|
|
|
|
#if !UCONFIG_NO_COLLATION
|
|
|
|
|
2002-07-12 21:42:24 +00:00
|
|
|
#include "unicode/coleitr.h"
|
|
|
|
#include "unicode/unorm.h"
|
|
|
|
#include "unicode/udata.h"
|
2004-12-09 04:23:42 +00:00
|
|
|
#include "unicode/ustring.h"
|
2002-07-12 21:42:24 +00:00
|
|
|
|
2001-03-08 17:40:42 +00:00
|
|
|
#include "ucol_imp.h"
|
2001-05-18 19:49:04 +00:00
|
|
|
#include "bocsu.h"
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-11-30 00:06:13 +00:00
|
|
|
#include "unormimp.h"
|
2003-01-23 01:52:34 +00:00
|
|
|
#include "unorm_it.h"
|
2001-11-30 00:06:13 +00:00
|
|
|
#include "umutex.h"
|
2004-12-09 04:23:42 +00:00
|
|
|
#include "cmemory.h"
|
2003-12-03 00:45:12 +00:00
|
|
|
#include "ucln_in.h"
|
2002-07-12 21:42:24 +00:00
|
|
|
#include "cstring.h"
|
2003-10-31 02:19:42 +00:00
|
|
|
#include "utracimp.h"
|
2004-10-18 02:43:33 +00:00
|
|
|
#include "putilimp.h"
|
2008-01-05 01:27:56 +00:00
|
|
|
#include "uassert.h"
|
2000-12-06 00:53:48 +00:00
|
|
|
|
2001-06-01 22:08:06 +00:00
|
|
|
#ifdef UCOL_DEBUG
|
2000-11-30 23:20:14 +00:00
|
|
|
#include <stdio.h>
|
2001-06-01 22:08:06 +00:00
|
|
|
#endif
|
2000-11-30 23:20:14 +00:00
|
|
|
|
2001-10-08 23:26:58 +00:00
|
|
|
U_NAMESPACE_USE
|
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/* added by synwee for trie manipulation*/
|
|
|
|
#define STAGE_1_SHIFT_ 10
|
|
|
|
#define STAGE_2_SHIFT_ 4
|
|
|
|
#define STAGE_2_MASK_AFTER_SHIFT_ 0x3F
|
|
|
|
#define STAGE_3_MASK_ 0xF
|
|
|
|
#define LAST_BYTE_MASK_ 0xFF
|
|
|
|
#define SECOND_LAST_BYTE_SHIFT_ 8
|
2001-01-04 00:45:41 +00:00
|
|
|
|
2001-04-20 22:29:53 +00:00
|
|
|
#define ZERO_CC_LIMIT_ 0xC0
|
|
|
|
|
2004-01-26 21:16:25 +00:00
|
|
|
// this is static pointer to the normalizer fcdTrieIndex
|
|
|
|
// it is always the same between calls to u_cleanup
|
|
|
|
// and therefore writing to it is not synchronized.
|
|
|
|
// It is cleaned in ucol_cleanup
|
|
|
|
static const uint16_t *fcdTrieIndex=NULL;
|
2001-01-04 00:45:41 +00:00
|
|
|
|
2005-07-06 21:27:21 +00:00
|
|
|
// These are values from UCA required for
|
|
|
|
// implicit generation and supressing sort key compression
|
|
|
|
// they should regularly be in the UCA, but if one
|
|
|
|
// is running without UCA, it could be a problem
|
2008-01-03 10:07:47 +00:00
|
|
|
static const int32_t maxRegularPrimary = 0xA0;
|
|
|
|
static const int32_t minImplicitPrimary = 0xE0;
|
|
|
|
static const int32_t maxImplicitPrimary = 0xE4;
|
2005-07-06 21:27:21 +00:00
|
|
|
|
2001-07-02 22:56:52 +00:00
|
|
|
U_CDECL_BEGIN
|
2001-03-02 00:19:43 +00:00
|
|
|
static UBool U_CALLCONV
|
2008-03-12 06:24:36 +00:00
|
|
|
ucol_cleanup(void)
|
|
|
|
{
|
|
|
|
fcdTrieIndex = NULL;
|
|
|
|
return TRUE;
|
2001-01-04 00:45:41 +00:00
|
|
|
}
|
2002-07-02 22:32:14 +00:00
|
|
|
|
|
|
|
static int32_t U_CALLCONV
|
|
|
|
_getFoldingOffset(uint32_t data) {
|
|
|
|
return (int32_t)(data&0xFFFFFF);
|
|
|
|
}
|
|
|
|
|
2001-07-02 22:56:52 +00:00
|
|
|
U_CDECL_END
|
2001-04-06 23:37:48 +00:00
|
|
|
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2008-03-12 06:24:36 +00:00
|
|
|
inline void IInit_collIterate(const UCollator *collator, const UChar *sourceString,
|
|
|
|
int32_t sourceLen, collIterate *s)
|
|
|
|
{
|
2001-04-23 01:53:49 +00:00
|
|
|
(s)->string = (s)->pos = (UChar *)(sourceString);
|
2001-05-10 17:49:24 +00:00
|
|
|
(s)->origFlags = 0;
|
2001-04-06 23:37:48 +00:00
|
|
|
(s)->flags = 0;
|
|
|
|
if (sourceLen >= 0) {
|
|
|
|
s->flags |= UCOL_ITER_HASLEN;
|
2001-04-26 01:15:34 +00:00
|
|
|
(s)->endp = (UChar *)sourceString+sourceLen;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* change to enable easier checking for end of string for fcdpositon */
|
|
|
|
(s)->endp = NULL;
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
2007-12-15 00:30:35 +00:00
|
|
|
(s)->extendCEs = NULL;
|
|
|
|
(s)->extendCEsSize = 0;
|
2001-04-23 01:53:49 +00:00
|
|
|
(s)->CEpos = (s)->toReturn = (s)->CEs;
|
2008-05-23 04:22:28 +00:00
|
|
|
(s)->offsetBuffer = NULL;
|
|
|
|
(s)->offsetBufferSize = 0;
|
|
|
|
(s)->offsetReturn = (s)->offsetStore = NULL;
|
|
|
|
(s)->offsetRepeatCount = (s)->offsetRepeatValue = 0;
|
2001-04-23 01:53:49 +00:00
|
|
|
(s)->writableBuffer = (s)->stackWritableBuffer;
|
|
|
|
(s)->writableBufSize = UCOL_WRITABLE_BUFFER_SIZE;
|
|
|
|
(s)->coll = (collator);
|
|
|
|
(s)->fcdPosition = 0;
|
2001-04-12 19:59:28 +00:00
|
|
|
if(collator->normalizationMode == UCOL_ON) {
|
2003-12-03 00:45:12 +00:00
|
|
|
(s)->flags |= UCOL_ITER_NORM;
|
2001-10-05 02:07:51 +00:00
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
if(collator->hiraganaQ == UCOL_ON && collator->strength >= UCOL_QUATERNARY) {
|
2007-12-26 19:55:00 +00:00
|
|
|
(s)->flags |= UCOL_HIRAGANA_Q;
|
2001-10-05 02:07:51 +00:00
|
|
|
}
|
2003-01-20 07:43:32 +00:00
|
|
|
(s)->iterator = NULL;
|
|
|
|
//(s)->iteratorIndex = 0;
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
2002-07-16 01:46:42 +00:00
|
|
|
uprv_init_collIterate(const UCollator *collator, const UChar *sourceString,
|
2001-04-06 23:37:48 +00:00
|
|
|
int32_t sourceLen, collIterate *s){
|
2001-04-12 19:59:28 +00:00
|
|
|
/* Out-of-line version for use from other files. */
|
2001-04-06 23:37:48 +00:00
|
|
|
IInit_collIterate(collator, sourceString, sourceLen, s);
|
|
|
|
}
|
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
/**
|
2001-04-18 01:34:26 +00:00
|
|
|
* Backup the state of the collIterate struct data
|
|
|
|
* @param data collIterate to backup
|
2001-04-23 01:53:49 +00:00
|
|
|
* @param backup storage
|
2001-04-18 01:34:26 +00:00
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-04-23 01:53:49 +00:00
|
|
|
inline void backupState(const collIterate *data, collIterateState *backup)
|
2001-04-18 01:34:26 +00:00
|
|
|
{
|
2001-04-23 01:53:49 +00:00
|
|
|
backup->fcdPosition = data->fcdPosition;
|
|
|
|
backup->flags = data->flags;
|
|
|
|
backup->origFlags = data->origFlags;
|
|
|
|
backup->pos = data->pos;
|
2001-05-02 01:36:29 +00:00
|
|
|
backup->bufferaddress = data->writableBuffer;
|
2001-04-20 22:29:53 +00:00
|
|
|
backup->buffersize = data->writableBufSize;
|
2006-04-12 23:36:56 +00:00
|
|
|
backup->iteratorMove = 0;
|
|
|
|
backup->iteratorIndex = 0;
|
2003-01-20 07:43:32 +00:00
|
|
|
if(data->iterator != NULL) {
|
2006-04-12 23:36:56 +00:00
|
|
|
//backup->iteratorIndex = data->iterator->getIndex(data->iterator, UITER_CURRENT);
|
|
|
|
backup->iteratorIndex = data->iterator->getState(data->iterator);
|
|
|
|
// no we try to fixup if we're using a normalizing iterator and we get UITER_NO_STATE
|
|
|
|
if(backup->iteratorIndex == UITER_NO_STATE) {
|
|
|
|
while((backup->iteratorIndex = data->iterator->getState(data->iterator)) == UITER_NO_STATE) {
|
|
|
|
backup->iteratorMove++;
|
|
|
|
data->iterator->move(data->iterator, -1, UITER_CURRENT);
|
|
|
|
}
|
|
|
|
data->iterator->move(data->iterator, backup->iteratorMove, UITER_CURRENT);
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
2003-01-20 07:43:32 +00:00
|
|
|
}
|
2001-04-18 01:34:26 +00:00
|
|
|
}
|
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
/**
|
2001-04-18 01:34:26 +00:00
|
|
|
* Loads the state into the collIterate struct data
|
|
|
|
* @param data collIterate to backup
|
2001-04-23 01:53:49 +00:00
|
|
|
* @param backup storage
|
2001-05-11 01:13:08 +00:00
|
|
|
* @param forwards boolean to indicate if forwards iteration is used,
|
2001-05-02 01:36:29 +00:00
|
|
|
* false indicates backwards iteration
|
2001-04-18 01:34:26 +00:00
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-11 01:13:08 +00:00
|
|
|
inline void loadState(collIterate *data, const collIterateState *backup,
|
2001-05-02 01:36:29 +00:00
|
|
|
UBool forwards)
|
2001-04-18 01:34:26 +00:00
|
|
|
{
|
2006-04-12 23:36:56 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
2001-04-18 01:34:26 +00:00
|
|
|
data->flags = backup->flags;
|
|
|
|
data->origFlags = backup->origFlags;
|
2003-01-20 07:43:32 +00:00
|
|
|
if(data->iterator != NULL) {
|
2006-04-12 23:36:56 +00:00
|
|
|
//data->iterator->move(data->iterator, backup->iteratorIndex, UITER_ZERO);
|
|
|
|
data->iterator->setState(data->iterator, backup->iteratorIndex, &status);
|
|
|
|
if(backup->iteratorMove != 0) {
|
|
|
|
data->iterator->move(data->iterator, backup->iteratorMove, UITER_CURRENT);
|
|
|
|
}
|
2003-01-20 07:43:32 +00:00
|
|
|
}
|
2001-04-18 01:34:26 +00:00
|
|
|
data->pos = backup->pos;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
if ((data->flags & UCOL_ITER_INNORMBUF) &&
|
2001-05-02 01:36:29 +00:00
|
|
|
data->writableBuffer != backup->bufferaddress) {
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
|
|
|
this is when a new buffer has been reallocated and we'll have to
|
2001-04-20 22:29:53 +00:00
|
|
|
calculate the new position.
|
|
|
|
note the new buffer has to contain the contents of the old buffer.
|
|
|
|
*/
|
2001-05-02 01:36:29 +00:00
|
|
|
if (forwards) {
|
2001-05-11 01:13:08 +00:00
|
|
|
data->pos = data->writableBuffer +
|
2001-05-02 01:36:29 +00:00
|
|
|
(data->pos - backup->bufferaddress);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* backwards direction */
|
2001-05-11 01:13:08 +00:00
|
|
|
uint32_t temp = backup->buffersize -
|
2001-05-02 01:36:29 +00:00
|
|
|
(data->pos - backup->bufferaddress);
|
|
|
|
data->pos = data->writableBuffer + (data->writableBufSize - temp);
|
|
|
|
}
|
2001-04-25 23:28:55 +00:00
|
|
|
}
|
|
|
|
if ((data->flags & UCOL_ITER_INNORMBUF) == 0) {
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
2001-04-25 23:28:55 +00:00
|
|
|
this is alittle tricky.
|
2001-05-11 01:13:08 +00:00
|
|
|
if we are initially not in the normalization buffer, even if we
|
2001-04-25 23:28:55 +00:00
|
|
|
normalize in the later stage, the data in the buffer will be
|
|
|
|
ignored, since we skip back up to the data string.
|
|
|
|
however if we are already in the normalization buffer, any
|
2001-05-11 01:13:08 +00:00
|
|
|
further normalization will pull data into the normalization
|
2001-04-25 23:28:55 +00:00
|
|
|
buffer and modify the fcdPosition.
|
2001-05-11 01:13:08 +00:00
|
|
|
since we are keeping the data in the buffer for use, the
|
2001-04-25 23:28:55 +00:00
|
|
|
fcdPosition can not be reverted back.
|
|
|
|
arrgghh....
|
|
|
|
*/
|
|
|
|
data->fcdPosition = backup->fcdPosition;
|
|
|
|
}
|
2001-04-18 01:34:26 +00:00
|
|
|
}
|
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* collIter_eos()
|
|
|
|
* Checks for a collIterate being positioned at the end of
|
|
|
|
* its source string.
|
|
|
|
*
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-06-06 23:26:50 +00:00
|
|
|
inline UBool collIter_eos(collIterate *s) {
|
2003-01-20 07:43:32 +00:00
|
|
|
if(s->flags & UCOL_USE_ITERATOR) {
|
|
|
|
return !(s->iterator->hasNext(s->iterator));
|
|
|
|
}
|
2001-06-06 23:26:50 +00:00
|
|
|
if ((s->flags & UCOL_ITER_HASLEN) == 0 && *s->pos != 0) {
|
|
|
|
// Null terminated string, but not at null, so not at end.
|
|
|
|
// Whether in main or normalization buffer doesn't matter.
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
// String with length. Can't be in normalization buffer, which is always
|
|
|
|
// null termintated.
|
|
|
|
if (s->flags & UCOL_ITER_HASLEN) {
|
|
|
|
return (s->pos == s->endp);
|
|
|
|
}
|
|
|
|
|
|
|
|
// We are at a null termination, could be either normalization buffer or main string.
|
|
|
|
if ((s->flags & UCOL_ITER_INNORMBUF) == 0) {
|
|
|
|
// At null at end of main string.
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
// At null at end of normalization buffer. Need to check whether there there are
|
|
|
|
// any characters left in the main buffer.
|
2003-01-20 07:43:32 +00:00
|
|
|
if(s->origFlags & UCOL_USE_ITERATOR) {
|
|
|
|
return !(s->iterator->hasNext(s->iterator));
|
|
|
|
} else if ((s->origFlags & UCOL_ITER_HASLEN) == 0) {
|
2001-06-06 23:26:50 +00:00
|
|
|
// Null terminated main string. fcdPosition is the 'return' position into main buf.
|
|
|
|
return (*s->fcdPosition == 0);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Main string with an end pointer.
|
|
|
|
return s->fcdPosition == s->endp;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2002-07-02 22:32:14 +00:00
|
|
|
/*
|
|
|
|
* collIter_bos()
|
|
|
|
* Checks for a collIterate being positioned at the start of
|
|
|
|
* its source string.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
static
|
|
|
|
inline UBool collIter_bos(collIterate *source) {
|
2003-01-20 07:43:32 +00:00
|
|
|
// if we're going backwards, we need to know whether there is more in the
|
|
|
|
// iterator, even if we are in the side buffer
|
|
|
|
if(source->flags & UCOL_USE_ITERATOR || source->origFlags & UCOL_USE_ITERATOR) {
|
|
|
|
return !source->iterator->hasPrevious(source->iterator);
|
|
|
|
}
|
2002-07-02 22:32:14 +00:00
|
|
|
if (source->pos <= source->string ||
|
|
|
|
((source->flags & UCOL_ITER_INNORMBUF) &&
|
|
|
|
*(source->pos - 1) == 0 && source->fcdPosition == NULL)) {
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
2007-07-24 08:30:41 +00:00
|
|
|
/*static
|
2003-02-08 02:16:54 +00:00
|
|
|
inline UBool collIter_SimpleBos(collIterate *source) {
|
|
|
|
// if we're going backwards, we need to know whether there is more in the
|
|
|
|
// iterator, even if we are in the side buffer
|
|
|
|
if(source->flags & UCOL_USE_ITERATOR || source->origFlags & UCOL_USE_ITERATOR) {
|
|
|
|
return !source->iterator->hasPrevious(source->iterator);
|
|
|
|
}
|
|
|
|
if (source->pos == source->string) {
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
return FALSE;
|
2007-07-24 08:30:41 +00:00
|
|
|
}*/
|
2003-02-08 02:16:54 +00:00
|
|
|
//return (data->pos == data->string) ||
|
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
|
2001-05-10 17:49:24 +00:00
|
|
|
/**
|
|
|
|
* Checks and free writable buffer if it is not the original stack buffer
|
|
|
|
* in collIterate. This function does not reassign the writable buffer.
|
|
|
|
* @param data collIterate struct to determine and free the writable buffer
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-10 17:49:24 +00:00
|
|
|
inline void freeHeapWritableBuffer(collIterate *data)
|
|
|
|
{
|
|
|
|
if (data->writableBuffer != data->stackWritableBuffer) {
|
|
|
|
uprv_free(data->writableBuffer);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/****************************************************************************/
|
|
|
|
/* Following are the open/close functions */
|
|
|
|
/* */
|
|
|
|
/****************************************************************************/
|
2003-04-24 07:00:27 +00:00
|
|
|
|
2006-05-21 01:27:15 +00:00
|
|
|
static UCollator*
|
|
|
|
ucol_initFromBinary(const uint8_t *bin, int32_t length,
|
2004-12-09 04:23:42 +00:00
|
|
|
const UCollator *base,
|
2006-05-21 01:27:15 +00:00
|
|
|
UCollator *fillIn,
|
2004-12-09 04:23:42 +00:00
|
|
|
UErrorCode *status)
|
2003-05-08 22:21:05 +00:00
|
|
|
{
|
2006-05-21 01:27:15 +00:00
|
|
|
UCollator *result = fillIn;
|
2006-05-21 17:52:09 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
2004-12-09 04:23:42 +00:00
|
|
|
return NULL;
|
2003-11-08 08:29:31 +00:00
|
|
|
}
|
2005-07-06 21:27:21 +00:00
|
|
|
/*
|
2004-12-09 04:23:42 +00:00
|
|
|
if(base == NULL) {
|
|
|
|
// we don't support null base yet
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return NULL;
|
2002-02-28 07:20:52 +00:00
|
|
|
}
|
2005-07-06 21:27:21 +00:00
|
|
|
*/
|
|
|
|
// We need these and we could be running without UCA
|
2008-01-05 01:27:56 +00:00
|
|
|
uprv_uca_initImplicitConstants(status);
|
2004-12-09 04:23:42 +00:00
|
|
|
UCATableHeader *colData = (UCATableHeader *)bin;
|
|
|
|
// do we want version check here? We're trying to figure out whether collators are compatible
|
2005-07-07 22:48:45 +00:00
|
|
|
if((base && (uprv_memcmp(colData->UCAVersion, base->image->UCAVersion, sizeof(UVersionInfo)) != 0 ||
|
|
|
|
uprv_memcmp(colData->UCDVersion, base->image->UCDVersion, sizeof(UVersionInfo)) != 0)) ||
|
|
|
|
colData->version[0] != UCOL_BUILDER_VERSION)
|
2004-12-09 04:23:42 +00:00
|
|
|
{
|
|
|
|
*status = U_COLLATOR_VERSION_MISMATCH;
|
|
|
|
return NULL;
|
2006-05-21 17:52:09 +00:00
|
|
|
}
|
|
|
|
else {
|
2004-12-09 04:23:42 +00:00
|
|
|
if((uint32_t)length > (paddedsize(sizeof(UCATableHeader)) + paddedsize(sizeof(UColOptionSet)))) {
|
|
|
|
result = ucol_initCollator((const UCATableHeader *)bin, result, base, status);
|
|
|
|
if(U_FAILURE(*status)){
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
result->hasRealData = TRUE;
|
2006-05-21 17:52:09 +00:00
|
|
|
}
|
|
|
|
else {
|
2004-12-09 04:23:42 +00:00
|
|
|
if(base) {
|
|
|
|
result = ucol_initCollator(base->image, result, base, status);
|
|
|
|
ucol_setOptionsFromHeader(result, (UColOptionSet *)(bin+((const UCATableHeader *)bin)->options), status);
|
|
|
|
if(U_FAILURE(*status)){
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
result->hasRealData = FALSE;
|
2006-05-21 17:52:09 +00:00
|
|
|
}
|
|
|
|
else {
|
2004-12-09 04:23:42 +00:00
|
|
|
*status = U_USELESS_COLLATOR_ERROR;
|
|
|
|
return NULL;
|
|
|
|
}
|
2003-04-29 00:12:41 +00:00
|
|
|
}
|
2003-05-06 07:28:34 +00:00
|
|
|
result->freeImageOnClose = FALSE;
|
2001-02-26 10:28:56 +00:00
|
|
|
}
|
2007-12-18 01:19:16 +00:00
|
|
|
result->actualLocale = NULL;
|
2004-12-09 04:23:42 +00:00
|
|
|
result->validLocale = NULL;
|
|
|
|
result->requestedLocale = NULL;
|
|
|
|
result->rules = NULL;
|
|
|
|
result->rulesLength = 0;
|
|
|
|
result->freeRulesOnClose = FALSE;
|
2007-12-18 01:19:16 +00:00
|
|
|
result->ucaRules = NULL;
|
2004-12-09 04:23:42 +00:00
|
|
|
return result;
|
2003-05-01 16:02:29 +00:00
|
|
|
}
|
|
|
|
|
2006-05-21 01:27:15 +00:00
|
|
|
U_CAPI UCollator* U_EXPORT2
|
|
|
|
ucol_openBinary(const uint8_t *bin, int32_t length,
|
|
|
|
const UCollator *base,
|
|
|
|
UErrorCode *status)
|
|
|
|
{
|
|
|
|
return ucol_initFromBinary(bin, length, base, NULL, status);
|
|
|
|
}
|
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
ucol_cloneBinary(const UCollator *coll,
|
|
|
|
uint8_t *buffer, int32_t capacity,
|
|
|
|
UErrorCode *status)
|
|
|
|
{
|
|
|
|
int32_t length = 0;
|
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
return length;
|
|
|
|
}
|
|
|
|
if(capacity < 0) {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return length;
|
|
|
|
}
|
|
|
|
if(coll->hasRealData == TRUE) {
|
|
|
|
length = coll->image->size;
|
|
|
|
if(length <= capacity) {
|
|
|
|
uprv_memcpy(buffer, coll->image, length);
|
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
length = (int32_t)(paddedsize(sizeof(UCATableHeader))+paddedsize(sizeof(UColOptionSet)));
|
|
|
|
if(length <= capacity) {
|
|
|
|
/* build the UCATableHeader with minimal entries */
|
|
|
|
/* do not copy the header from the UCA file because its values are wrong! */
|
|
|
|
/* uprv_memcpy(result, UCA->image, sizeof(UCATableHeader)); */
|
|
|
|
|
|
|
|
/* reset everything */
|
|
|
|
uprv_memset(buffer, 0, length);
|
|
|
|
|
|
|
|
/* set the tailoring-specific values */
|
|
|
|
UCATableHeader *myData = (UCATableHeader *)buffer;
|
|
|
|
myData->size = length;
|
|
|
|
|
|
|
|
/* offset for the options, the only part of the data that is present after the header */
|
|
|
|
myData->options = sizeof(UCATableHeader);
|
|
|
|
|
|
|
|
/* need to always set the expansion value for an upper bound of the options */
|
|
|
|
myData->expansion = myData->options + sizeof(UColOptionSet);
|
|
|
|
|
|
|
|
myData->magic = UCOL_HEADER_MAGIC;
|
|
|
|
myData->isBigEndian = U_IS_BIG_ENDIAN;
|
|
|
|
myData->charSetFamily = U_CHARSET_FAMILY;
|
|
|
|
|
|
|
|
/* copy UCA's version; genrb will override all but the builder version with tailoring data */
|
|
|
|
uprv_memcpy(myData->version, coll->image->version, sizeof(UVersionInfo));
|
|
|
|
|
|
|
|
uprv_memcpy(myData->UCAVersion, coll->image->UCAVersion, sizeof(UVersionInfo));
|
|
|
|
uprv_memcpy(myData->UCDVersion, coll->image->UCDVersion, sizeof(UVersionInfo));
|
|
|
|
uprv_memcpy(myData->formatVersion, coll->image->formatVersion, sizeof(UVersionInfo));
|
|
|
|
myData->jamoSpecial = coll->image->jamoSpecial;
|
|
|
|
|
|
|
|
/* copy the collator options */
|
|
|
|
uprv_memcpy(buffer+paddedsize(sizeof(UCATableHeader)), coll->options, sizeof(UColOptionSet));
|
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return length;
|
|
|
|
}
|
|
|
|
|
2006-05-21 17:52:09 +00:00
|
|
|
U_CAPI UCollator* U_EXPORT2
|
|
|
|
ucol_safeClone(const UCollator *coll, void *stackBuffer, int32_t * pBufferSize, UErrorCode *status)
|
|
|
|
{
|
|
|
|
UCollator * localCollator;
|
|
|
|
int32_t bufferSizeNeeded = (int32_t)sizeof(UCollator);
|
|
|
|
char *stackBufferChars = (char *)stackBuffer;
|
|
|
|
int32_t imageSize = 0;
|
|
|
|
int32_t rulesSize = 0;
|
|
|
|
int32_t rulesPadding = 0;
|
|
|
|
uint8_t *image;
|
|
|
|
UChar *rules;
|
2006-05-21 22:22:24 +00:00
|
|
|
UBool colAllocated = FALSE;
|
|
|
|
UBool imageAllocated = FALSE;
|
2006-05-21 17:52:09 +00:00
|
|
|
|
|
|
|
if (status == NULL || U_FAILURE(*status)){
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if ((stackBuffer && !pBufferSize) || !coll){
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if (coll->rules && coll->freeRulesOnClose) {
|
|
|
|
rulesSize = (int32_t)(coll->rulesLength + 1)*sizeof(UChar);
|
|
|
|
rulesPadding = (int32_t)(bufferSizeNeeded % sizeof(UChar));
|
|
|
|
bufferSizeNeeded += rulesSize + rulesPadding;
|
|
|
|
}
|
2006-05-26 04:00:52 +00:00
|
|
|
|
2006-05-21 17:52:09 +00:00
|
|
|
if (stackBuffer && *pBufferSize <= 0){ /* 'preflighting' request - set needed size into *pBufferSize */
|
|
|
|
*pBufferSize = bufferSizeNeeded;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-05-26 04:00:52 +00:00
|
|
|
/* Pointers on 64-bit platforms need to be aligned
|
|
|
|
* on a 64-bit boundry in memory.
|
|
|
|
*/
|
|
|
|
if (U_ALIGNMENT_OFFSET(stackBuffer) != 0) {
|
|
|
|
int32_t offsetUp = (int32_t)U_ALIGNMENT_OFFSET_UP(stackBufferChars);
|
|
|
|
if (*pBufferSize > offsetUp) {
|
|
|
|
*pBufferSize -= offsetUp;
|
|
|
|
stackBufferChars += offsetUp;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* prevent using the stack buffer but keep the size > 0 so that we do not just preflight */
|
|
|
|
*pBufferSize = 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
stackBuffer = (void *)stackBufferChars;
|
|
|
|
|
|
|
|
if (stackBuffer == NULL || *pBufferSize < bufferSizeNeeded) {
|
2006-05-21 17:52:09 +00:00
|
|
|
/* allocate one here...*/
|
|
|
|
stackBufferChars = (char *)uprv_malloc(bufferSizeNeeded);
|
2008-01-16 20:12:48 +00:00
|
|
|
// Null pointer check.
|
|
|
|
if (stackBufferChars == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return NULL;
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2006-05-21 22:22:24 +00:00
|
|
|
colAllocated = TRUE;
|
2006-05-21 17:52:09 +00:00
|
|
|
if (U_SUCCESS(*status)) {
|
|
|
|
*status = U_SAFECLONE_ALLOCATED_WARNING;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
localCollator = (UCollator *)stackBufferChars;
|
|
|
|
rules = (UChar *)(stackBufferChars + sizeof(UCollator) + rulesPadding);
|
2006-06-06 06:42:34 +00:00
|
|
|
{
|
|
|
|
UErrorCode tempStatus = U_ZERO_ERROR;
|
|
|
|
imageSize = ucol_cloneBinary(coll, NULL, 0, &tempStatus);
|
|
|
|
}
|
|
|
|
if (coll->freeImageOnClose) {
|
2006-05-21 17:52:09 +00:00
|
|
|
image = (uint8_t *)uprv_malloc(imageSize);
|
2008-01-16 20:12:48 +00:00
|
|
|
// Null pointer check
|
|
|
|
if (image == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return NULL;
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2006-05-21 17:52:09 +00:00
|
|
|
ucol_cloneBinary(coll, image, imageSize, status);
|
2006-05-21 22:22:24 +00:00
|
|
|
imageAllocated = TRUE;
|
2006-05-21 17:52:09 +00:00
|
|
|
}
|
|
|
|
else {
|
|
|
|
image = (uint8_t *)coll->image;
|
|
|
|
}
|
2006-06-06 06:42:34 +00:00
|
|
|
localCollator = ucol_initFromBinary(image, imageSize, coll->UCA, localCollator, status);
|
|
|
|
if (U_FAILURE(*status)) {
|
|
|
|
return NULL;
|
|
|
|
}
|
2006-05-21 17:52:09 +00:00
|
|
|
|
|
|
|
if (coll->rules) {
|
|
|
|
if (coll->freeRulesOnClose) {
|
|
|
|
localCollator->rules = u_strcpy(rules, coll->rules);
|
|
|
|
//bufferEnd += rulesSize;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
localCollator->rules = coll->rules;
|
|
|
|
}
|
|
|
|
localCollator->freeRulesOnClose = FALSE;
|
|
|
|
localCollator->rulesLength = coll->rulesLength;
|
|
|
|
}
|
|
|
|
|
|
|
|
int32_t i;
|
|
|
|
for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
|
|
|
|
ucol_setAttribute(localCollator, (UColAttribute)i, ucol_getAttribute(coll, (UColAttribute)i, status), status);
|
|
|
|
}
|
2007-12-18 01:19:16 +00:00
|
|
|
// zero copies of pointers
|
|
|
|
localCollator->actualLocale = NULL;
|
2006-05-21 17:52:09 +00:00
|
|
|
localCollator->validLocale = NULL;
|
2007-12-18 01:19:16 +00:00
|
|
|
localCollator->requestedLocale = NULL;
|
|
|
|
localCollator->ucaRules = coll->ucaRules; // There should only be one copy here.
|
2006-05-21 22:22:24 +00:00
|
|
|
localCollator->freeOnClose = colAllocated;
|
|
|
|
localCollator->freeImageOnClose = imageAllocated;
|
2006-05-21 17:52:09 +00:00
|
|
|
return localCollator;
|
|
|
|
}
|
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_close(UCollator *coll)
|
|
|
|
{
|
2006-05-21 01:27:15 +00:00
|
|
|
UTRACE_ENTRY_OC(UTRACE_UCOL_CLOSE);
|
|
|
|
UTRACE_DATA1(UTRACE_INFO, "coll = %p", coll);
|
|
|
|
if(coll != NULL) {
|
|
|
|
// these are always owned by each UCollator struct,
|
|
|
|
// so we always free them
|
|
|
|
if(coll->validLocale != NULL) {
|
|
|
|
uprv_free(coll->validLocale);
|
|
|
|
}
|
2007-12-18 01:19:16 +00:00
|
|
|
if(coll->actualLocale != NULL) {
|
|
|
|
uprv_free(coll->actualLocale);
|
|
|
|
}
|
2006-05-21 01:27:15 +00:00
|
|
|
if(coll->requestedLocale != NULL) {
|
|
|
|
uprv_free(coll->requestedLocale);
|
|
|
|
}
|
2006-05-21 21:09:01 +00:00
|
|
|
if(coll->latinOneCEs != NULL) {
|
|
|
|
uprv_free(coll->latinOneCEs);
|
|
|
|
}
|
|
|
|
if(coll->options != NULL && coll->freeOptionsOnClose) {
|
|
|
|
uprv_free(coll->options);
|
|
|
|
}
|
|
|
|
if(coll->rules != NULL && coll->freeRulesOnClose) {
|
|
|
|
uprv_free((UChar *)coll->rules);
|
|
|
|
}
|
|
|
|
if(coll->image != NULL && coll->freeImageOnClose) {
|
|
|
|
uprv_free((UCATableHeader *)coll->image);
|
|
|
|
}
|
2006-05-21 01:27:15 +00:00
|
|
|
|
|
|
|
/* Here, it would be advisable to close: */
|
|
|
|
/* - UData for UCA (unless we stuff it in the root resb */
|
|
|
|
/* Again, do we need additional housekeeping... HMMM! */
|
|
|
|
UTRACE_DATA1(UTRACE_INFO, "coll->freeOnClose: %d", coll->freeOnClose);
|
|
|
|
if(coll->freeOnClose){
|
|
|
|
/* for safeClone, if freeOnClose is FALSE,
|
|
|
|
don't free the other instance data */
|
|
|
|
uprv_free(coll);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
UTRACE_EXIT();
|
2001-01-16 00:28:40 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* This one is currently used by genrb & tests. After constructing from rules (tailoring),*/
|
|
|
|
/* you should be able to get the binary chunk to write out... Doesn't look very full now */
|
2007-06-03 06:08:46 +00:00
|
|
|
U_CFUNC uint8_t* U_EXPORT2
|
2001-07-09 22:24:23 +00:00
|
|
|
ucol_cloneRuleData(const UCollator *coll, int32_t *length, UErrorCode *status)
|
2001-01-16 00:28:40 +00:00
|
|
|
{
|
2007-12-28 20:08:41 +00:00
|
|
|
uint8_t *result = NULL;
|
|
|
|
if(U_FAILURE(*status)) {
|
2002-06-29 09:32:36 +00:00
|
|
|
return NULL;
|
|
|
|
}
|
2007-12-28 20:08:41 +00:00
|
|
|
if(coll->hasRealData == TRUE) {
|
|
|
|
*length = coll->image->size;
|
|
|
|
result = (uint8_t *)uprv_malloc(*length);
|
|
|
|
/* test for NULL */
|
|
|
|
if (result == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
uprv_memcpy(result, coll->image, *length);
|
|
|
|
} else {
|
|
|
|
*length = (int32_t)(paddedsize(sizeof(UCATableHeader))+paddedsize(sizeof(UColOptionSet)));
|
|
|
|
result = (uint8_t *)uprv_malloc(*length);
|
|
|
|
/* test for NULL */
|
|
|
|
if (result == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return NULL;
|
|
|
|
}
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* build the UCATableHeader with minimal entries */
|
|
|
|
/* do not copy the header from the UCA file because its values are wrong! */
|
|
|
|
/* uprv_memcpy(result, UCA->image, sizeof(UCATableHeader)); */
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* reset everything */
|
|
|
|
uprv_memset(result, 0, *length);
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* set the tailoring-specific values */
|
|
|
|
UCATableHeader *myData = (UCATableHeader *)result;
|
|
|
|
myData->size = *length;
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* offset for the options, the only part of the data that is present after the header */
|
|
|
|
myData->options = sizeof(UCATableHeader);
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* need to always set the expansion value for an upper bound of the options */
|
|
|
|
myData->expansion = myData->options + sizeof(UColOptionSet);
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
myData->magic = UCOL_HEADER_MAGIC;
|
|
|
|
myData->isBigEndian = U_IS_BIG_ENDIAN;
|
|
|
|
myData->charSetFamily = U_CHARSET_FAMILY;
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* copy UCA's version; genrb will override all but the builder version with tailoring data */
|
|
|
|
uprv_memcpy(myData->version, coll->image->version, sizeof(UVersionInfo));
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
uprv_memcpy(myData->UCAVersion, coll->image->UCAVersion, sizeof(UVersionInfo));
|
|
|
|
uprv_memcpy(myData->UCDVersion, coll->image->UCDVersion, sizeof(UVersionInfo));
|
|
|
|
uprv_memcpy(myData->formatVersion, coll->image->formatVersion, sizeof(UVersionInfo));
|
|
|
|
myData->jamoSpecial = coll->image->jamoSpecial;
|
2003-09-18 23:18:23 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
/* copy the collator options */
|
|
|
|
uprv_memcpy(result+paddedsize(sizeof(UCATableHeader)), coll->options, sizeof(UColOptionSet));
|
|
|
|
}
|
|
|
|
return result;
|
2001-02-26 10:28:56 +00:00
|
|
|
}
|
|
|
|
|
2001-03-30 00:23:46 +00:00
|
|
|
void ucol_setOptionsFromHeader(UCollator* result, UColOptionSet * opts, UErrorCode *status) {
|
2007-12-28 20:08:41 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
return;
|
|
|
|
}
|
2002-09-17 04:53:35 +00:00
|
|
|
result->caseFirst = (UColAttributeValue)opts->caseFirst;
|
|
|
|
result->caseLevel = (UColAttributeValue)opts->caseLevel;
|
|
|
|
result->frenchCollation = (UColAttributeValue)opts->frenchCollation;
|
|
|
|
result->normalizationMode = (UColAttributeValue)opts->normalizationMode;
|
|
|
|
result->strength = (UColAttributeValue)opts->strength;
|
2002-10-08 16:10:49 +00:00
|
|
|
result->variableTopValue = opts->variableTopValue;
|
2002-09-17 04:53:35 +00:00
|
|
|
result->alternateHandling = (UColAttributeValue)opts->alternateHandling;
|
|
|
|
result->hiraganaQ = (UColAttributeValue)opts->hiraganaQ;
|
2003-07-25 23:57:51 +00:00
|
|
|
result->numericCollation = (UColAttributeValue)opts->numericCollation;
|
2001-02-26 10:28:56 +00:00
|
|
|
|
|
|
|
result->caseFirstisDefault = TRUE;
|
|
|
|
result->caseLevelisDefault = TRUE;
|
|
|
|
result->frenchCollationisDefault = TRUE;
|
|
|
|
result->normalizationModeisDefault = TRUE;
|
|
|
|
result->strengthisDefault = TRUE;
|
|
|
|
result->variableTopValueisDefault = TRUE;
|
2001-10-02 16:49:57 +00:00
|
|
|
result->hiraganaQisDefault = TRUE;
|
2003-07-25 23:57:51 +00:00
|
|
|
result->numericCollationisDefault = TRUE;
|
2001-03-02 00:19:43 +00:00
|
|
|
|
2002-09-04 06:02:13 +00:00
|
|
|
ucol_updateInternalState(result, status);
|
2001-03-30 00:23:46 +00:00
|
|
|
|
|
|
|
result->options = opts;
|
2001-02-26 10:28:56 +00:00
|
|
|
}
|
|
|
|
|
2001-04-10 22:06:25 +00:00
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
/**
|
2001-05-17 01:06:25 +00:00
|
|
|
* Approximate determination if a character is at a contraction end.
|
2001-06-06 23:26:50 +00:00
|
|
|
* Guaranteed to be TRUE if a character is at the end of a contraction,
|
2001-05-17 01:06:25 +00:00
|
|
|
* otherwise it is not deterministic.
|
|
|
|
* @param c character to be determined
|
|
|
|
* @param coll collator
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-10 17:49:24 +00:00
|
|
|
inline UBool ucol_contractionEndCP(UChar c, const UCollator *coll) {
|
2001-05-10 22:12:53 +00:00
|
|
|
if (c < coll->minContrEndCP) {
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
int32_t hash = c;
|
|
|
|
uint8_t htbyte;
|
|
|
|
if (hash >= UCOL_UNSAFECP_TABLE_SIZE*8) {
|
2007-12-28 21:16:45 +00:00
|
|
|
if (U16_IS_TRAIL(c)) {
|
|
|
|
return TRUE;
|
|
|
|
}
|
2001-05-10 22:12:53 +00:00
|
|
|
hash = (hash & UCOL_UNSAFECP_TABLE_MASK) + 256;
|
|
|
|
}
|
|
|
|
htbyte = coll->contrEndCP[hash>>3];
|
2001-05-11 01:13:08 +00:00
|
|
|
return (((htbyte >> (hash & 7)) & 1) == 1);
|
2001-05-10 17:49:24 +00:00
|
|
|
}
|
2001-04-10 22:06:25 +00:00
|
|
|
|
2001-03-03 03:35:17 +00:00
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
/*
|
|
|
|
* i_getCombiningClass()
|
|
|
|
* A fast, at least partly inline version of u_getCombiningClass()
|
|
|
|
* This is a candidate for further optimization. Used heavily
|
|
|
|
* in contraction processing.
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2005-05-19 06:43:35 +00:00
|
|
|
inline uint8_t i_getCombiningClass(UChar32 c, const UCollator *coll) {
|
2001-06-06 23:26:50 +00:00
|
|
|
uint8_t sCC = 0;
|
2005-05-19 06:43:35 +00:00
|
|
|
if ((c >= 0x300 && ucol_unsafeCP(c, coll)) || c > 0xFFFF) {
|
2001-06-06 23:26:50 +00:00
|
|
|
sCC = u_getCombiningClass(c);
|
|
|
|
}
|
|
|
|
return sCC;
|
|
|
|
}
|
|
|
|
|
2003-09-24 00:17:15 +00:00
|
|
|
UCollator* ucol_initCollator(const UCATableHeader *image, UCollator *fillIn, const UCollator *UCA, UErrorCode *status) {
|
2001-04-18 19:31:05 +00:00
|
|
|
UChar c;
|
2001-01-15 07:28:54 +00:00
|
|
|
UCollator *result = fillIn;
|
2001-01-29 22:09:24 +00:00
|
|
|
if(U_FAILURE(*status) || image == NULL) {
|
2001-01-05 00:47:25 +00:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(result == NULL) {
|
2001-01-15 19:02:30 +00:00
|
|
|
result = (UCollator *)uprv_malloc(sizeof(UCollator));
|
2001-01-05 00:47:25 +00:00
|
|
|
if(result == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
result->freeOnClose = TRUE;
|
|
|
|
} else {
|
|
|
|
result->freeOnClose = FALSE;
|
|
|
|
}
|
|
|
|
|
2008-03-12 06:24:36 +00:00
|
|
|
// init FCD data
|
|
|
|
if (fcdTrieIndex == NULL) {
|
|
|
|
// The result is constant, until the library is reloaded.
|
|
|
|
fcdTrieIndex = unorm_getFCDTrie(status);
|
|
|
|
ucln_i18n_registerCleanup(UCLN_I18N_UCOL, ucol_cleanup);
|
|
|
|
}
|
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
result->image = image;
|
2006-05-21 01:27:15 +00:00
|
|
|
result->mapping.getFoldingOffset = _getFoldingOffset;
|
2001-01-05 00:47:25 +00:00
|
|
|
const uint8_t *mapping = (uint8_t*)result->image+result->image->mappingPosition;
|
2006-05-21 01:27:15 +00:00
|
|
|
utrie_unserialize(&result->mapping, mapping, result->image->endExpansionCE - result->image->mappingPosition, status);
|
|
|
|
if(U_FAILURE(*status)) {
|
2001-01-05 00:47:25 +00:00
|
|
|
if(result->freeOnClose == TRUE) {
|
|
|
|
uprv_free(result);
|
|
|
|
result = NULL;
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2001-12-19 07:00:45 +00:00
|
|
|
/*result->latinOneMapping = (uint32_t*)((uint8_t*)result->image+result->image->latinOneMapping);*/
|
2006-05-21 01:27:15 +00:00
|
|
|
result->latinOneMapping = UTRIE_GET32_LATIN1(&result->mapping);
|
2001-01-05 00:47:25 +00:00
|
|
|
result->contractionCEs = (uint32_t*)((uint8_t*)result->image+result->image->contractionCEs);
|
|
|
|
result->contractionIndex = (UChar*)((uint8_t*)result->image+result->image->contractionIndex);
|
|
|
|
result->expansion = (uint32_t*)((uint8_t*)result->image+result->image->expansion);
|
2001-03-30 00:23:46 +00:00
|
|
|
|
|
|
|
result->options = (UColOptionSet*)((uint8_t*)result->image+result->image->options);
|
|
|
|
result->freeOptionsOnClose = FALSE;
|
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
/* set attributes */
|
2002-09-17 04:53:35 +00:00
|
|
|
result->caseFirst = (UColAttributeValue)result->options->caseFirst;
|
|
|
|
result->caseLevel = (UColAttributeValue)result->options->caseLevel;
|
|
|
|
result->frenchCollation = (UColAttributeValue)result->options->frenchCollation;
|
|
|
|
result->normalizationMode = (UColAttributeValue)result->options->normalizationMode;
|
|
|
|
result->strength = (UColAttributeValue)result->options->strength;
|
2002-10-08 16:10:49 +00:00
|
|
|
result->variableTopValue = result->options->variableTopValue;
|
2002-09-17 04:53:35 +00:00
|
|
|
result->alternateHandling = (UColAttributeValue)result->options->alternateHandling;
|
|
|
|
result->hiraganaQ = (UColAttributeValue)result->options->hiraganaQ;
|
2003-07-25 23:57:51 +00:00
|
|
|
result->numericCollation = (UColAttributeValue)result->options->numericCollation;
|
2001-01-09 00:52:18 +00:00
|
|
|
|
|
|
|
result->caseFirstisDefault = TRUE;
|
|
|
|
result->caseLevelisDefault = TRUE;
|
|
|
|
result->frenchCollationisDefault = TRUE;
|
|
|
|
result->normalizationModeisDefault = TRUE;
|
|
|
|
result->strengthisDefault = TRUE;
|
|
|
|
result->variableTopValueisDefault = TRUE;
|
2001-02-27 21:01:11 +00:00
|
|
|
result->alternateHandlingisDefault = TRUE;
|
2001-10-02 16:49:57 +00:00
|
|
|
result->hiraganaQisDefault = TRUE;
|
2003-07-25 23:57:51 +00:00
|
|
|
result->numericCollationisDefault = TRUE;
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2006-05-15 05:08:49 +00:00
|
|
|
/*result->scriptOrder = NULL;*/
|
2001-01-18 00:46:19 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
result->rules = NULL;
|
2001-09-27 01:01:30 +00:00
|
|
|
result->rulesLength = 0;
|
2007-12-18 01:19:16 +00:00
|
|
|
result->freeRulesOnClose = FALSE;
|
2001-09-27 01:01:30 +00:00
|
|
|
|
|
|
|
/* get the version info from UCATableHeader and populate the Collator struct*/
|
2006-05-15 05:08:49 +00:00
|
|
|
result->dataVersion[0] = result->image->version[0]; /* UCA Builder version*/
|
|
|
|
result->dataVersion[1] = result->image->version[1]; /* UCA Tailoring rules version*/
|
2006-05-21 01:27:15 +00:00
|
|
|
result->dataVersion[2] = 0;
|
|
|
|
result->dataVersion[3] = 0;
|
2001-03-02 01:14:03 +00:00
|
|
|
|
2001-03-03 03:35:17 +00:00
|
|
|
result->unsafeCP = (uint8_t *)result->image + result->image->unsafeCP;
|
2001-04-18 19:31:05 +00:00
|
|
|
result->minUnsafeCP = 0;
|
|
|
|
for (c=0; c<0x300; c++) { // Find the smallest unsafe char.
|
|
|
|
if (ucol_unsafeCP(c, result)) break;
|
|
|
|
}
|
|
|
|
result->minUnsafeCP = c;
|
2001-03-03 03:35:17 +00:00
|
|
|
|
2001-05-10 22:12:53 +00:00
|
|
|
result->contrEndCP = (uint8_t *)result->image + result->image->contrEndCP;
|
|
|
|
result->minContrEndCP = 0;
|
|
|
|
for (c=0; c<0x300; c++) { // Find the Contraction-ending char.
|
|
|
|
if (ucol_contractionEndCP(c, result)) break;
|
|
|
|
}
|
|
|
|
result->minContrEndCP = c;
|
|
|
|
|
2001-03-02 01:14:03 +00:00
|
|
|
/* max expansion tables */
|
2001-04-23 01:53:49 +00:00
|
|
|
result->endExpansionCE = (uint32_t*)((uint8_t*)result->image +
|
2001-03-02 01:14:03 +00:00
|
|
|
result->image->endExpansionCE);
|
2001-04-23 01:53:49 +00:00
|
|
|
result->lastEndExpansionCE = result->endExpansionCE +
|
2001-03-02 01:14:03 +00:00
|
|
|
result->image->endExpansionCECount - 1;
|
2001-04-23 01:53:49 +00:00
|
|
|
result->expansionCESize = (uint8_t*)result->image +
|
2001-03-02 01:14:03 +00:00
|
|
|
result->image->expansionCESize;
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-03-02 01:14:03 +00:00
|
|
|
|
2003-03-27 20:09:38 +00:00
|
|
|
//result->errorCode = *status;
|
2002-09-04 06:02:13 +00:00
|
|
|
|
|
|
|
result->latinOneCEs = NULL;
|
|
|
|
|
|
|
|
result->latinOneRegenTable = FALSE;
|
|
|
|
result->latinOneFailed = FALSE;
|
2003-09-24 00:17:15 +00:00
|
|
|
result->UCA = UCA;
|
2002-09-04 06:02:13 +00:00
|
|
|
|
|
|
|
ucol_updateInternalState(result, status);
|
|
|
|
|
2007-06-13 22:53:28 +00:00
|
|
|
/* Normally these will be set correctly later. This is the default if you use UCA or the default. */
|
2007-12-18 01:19:16 +00:00
|
|
|
result->ucaRules = NULL;
|
|
|
|
result->actualLocale = NULL;
|
2007-06-13 22:53:28 +00:00
|
|
|
result->validLocale = NULL;
|
|
|
|
result->requestedLocale = NULL;
|
|
|
|
result->hasRealData = FALSE; // real data lives in .dat file...
|
|
|
|
result->freeImageOnClose = FALSE;
|
2001-04-10 22:06:25 +00:00
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
/* new Mark's code */
|
|
|
|
|
|
|
|
/**
|
|
|
|
* For generation of Implicit CEs
|
|
|
|
* @author Davis
|
|
|
|
*
|
|
|
|
* Cleaned up so that changes can be made more easily.
|
|
|
|
* Old values:
|
|
|
|
# First Implicit: E26A792D
|
|
|
|
# Last Implicit: E3DC70C0
|
|
|
|
# First CJK: E0030300
|
|
|
|
# Last CJK: E0A9DD00
|
|
|
|
# First CJK_A: E0A9DF00
|
|
|
|
# Last CJK_A: E0DE3100
|
|
|
|
*/
|
2002-06-13 18:34:41 +00:00
|
|
|
/* Following is a port of Mark's code for new treatment of implicits.
|
2003-12-03 00:45:12 +00:00
|
|
|
* It is positioned here, since ucol_initUCA need to initialize the
|
2002-06-13 18:34:41 +00:00
|
|
|
* variables below according to the data in the fractional UCA.
|
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2002-06-13 18:34:41 +00:00
|
|
|
/**
|
2007-12-28 20:08:41 +00:00
|
|
|
* Function used to:
|
|
|
|
* a) collapse the 2 different Han ranges from UCA into one (in the right order), and
|
|
|
|
* b) bump any non-CJK characters by 10FFFF.
|
|
|
|
* The relevant blocks are:
|
|
|
|
* A: 4E00..9FFF; CJK Unified Ideographs
|
|
|
|
* F900..FAFF; CJK Compatibility Ideographs
|
|
|
|
* B: 3400..4DBF; CJK Unified Ideographs Extension A
|
|
|
|
* 20000..XX; CJK Unified Ideographs Extension B (and others later on)
|
|
|
|
* As long as
|
|
|
|
* no new B characters are allocated between 4E00 and FAFF, and
|
|
|
|
* no new A characters are outside of this range,
|
|
|
|
* (very high probability) this simple code will work.
|
|
|
|
* The reordered blocks are:
|
|
|
|
* Block1 is CJK
|
|
|
|
* Block2 is CJK_COMPAT_USED
|
|
|
|
* Block3 is CJK_A
|
|
|
|
* (all contiguous)
|
|
|
|
* Any other CJK gets its normal code point
|
|
|
|
* Any non-CJK gets +10FFFF
|
|
|
|
* When we reorder Block1, we make sure that it is at the very start,
|
|
|
|
* so that it will use a 3-byte form.
|
|
|
|
* Warning: the we only pick up the compatibility characters that are
|
|
|
|
* NOT decomposed, so that block is smaller!
|
|
|
|
*/
|
2002-06-13 18:34:41 +00:00
|
|
|
|
|
|
|
// CONSTANTS
|
2004-01-17 00:06:19 +00:00
|
|
|
static const UChar32
|
2002-06-13 18:34:41 +00:00
|
|
|
NON_CJK_OFFSET = 0x110000,
|
2004-01-16 16:40:30 +00:00
|
|
|
UCOL_MAX_INPUT = 0x220001; // 2 * Unicode range + 2
|
2004-01-16 07:09:12 +00:00
|
|
|
|
|
|
|
/**
|
2008-08-06 17:57:45 +00:00
|
|
|
* Precomputed by initImplicitConstants()
|
2004-01-16 07:09:12 +00:00
|
|
|
*/
|
|
|
|
static int32_t
|
|
|
|
final3Multiplier = 0,
|
|
|
|
final4Multiplier = 0,
|
|
|
|
final3Count = 0,
|
|
|
|
final4Count = 0,
|
|
|
|
medialCount = 0,
|
|
|
|
min3Primary = 0,
|
|
|
|
min4Primary = 0,
|
|
|
|
max4Primary = 0,
|
|
|
|
minTrail = 0,
|
|
|
|
maxTrail = 0,
|
|
|
|
max3Trail = 0,
|
|
|
|
max4Trail = 0,
|
|
|
|
min4Boundary = 0;
|
2002-06-13 18:34:41 +00:00
|
|
|
|
2002-07-25 17:38:34 +00:00
|
|
|
static const UChar32
|
2002-06-13 18:34:41 +00:00
|
|
|
CJK_BASE = 0x4E00,
|
|
|
|
CJK_LIMIT = 0x9FFF+1,
|
|
|
|
CJK_COMPAT_USED_BASE = 0xFA0E,
|
|
|
|
CJK_COMPAT_USED_LIMIT = 0xFA2F+1,
|
|
|
|
CJK_A_BASE = 0x3400,
|
|
|
|
CJK_A_LIMIT = 0x4DBF+1,
|
|
|
|
CJK_B_BASE = 0x20000,
|
|
|
|
CJK_B_LIMIT = 0x2A6DF+1;
|
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
static UChar32 swapCJK(UChar32 i) {
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
if (i >= CJK_BASE) {
|
|
|
|
if (i < CJK_LIMIT) return i - CJK_BASE;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
if (i < CJK_COMPAT_USED_BASE) return i + NON_CJK_OFFSET;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
if (i < CJK_COMPAT_USED_LIMIT) return i - CJK_COMPAT_USED_BASE
|
|
|
|
+ (CJK_LIMIT - CJK_BASE);
|
|
|
|
if (i < CJK_B_BASE) return i + NON_CJK_OFFSET;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
if (i < CJK_B_LIMIT) return i; // non-BMP-CJK
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
return i + NON_CJK_OFFSET; // non-CJK
|
|
|
|
}
|
|
|
|
if (i < CJK_A_BASE) return i + NON_CJK_OFFSET;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
if (i < CJK_A_LIMIT) return i - CJK_A_BASE
|
2004-11-11 23:34:58 +00:00
|
|
|
+ (CJK_LIMIT - CJK_BASE)
|
2004-01-16 07:09:12 +00:00
|
|
|
+ (CJK_COMPAT_USED_LIMIT - CJK_COMPAT_USED_BASE);
|
|
|
|
return i + NON_CJK_OFFSET; // non-CJK
|
2002-06-13 18:34:41 +00:00
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2004-01-16 23:44:58 +00:00
|
|
|
U_CAPI UChar32 U_EXPORT2
|
|
|
|
uprv_uca_getRawFromCodePoint(UChar32 i) {
|
|
|
|
return swapCJK(i)+1;
|
|
|
|
}
|
|
|
|
|
|
|
|
U_CAPI UChar32 U_EXPORT2
|
|
|
|
uprv_uca_getCodePointFromRaw(UChar32 i) {
|
|
|
|
i--;
|
|
|
|
UChar32 result = 0;
|
|
|
|
if(i >= NON_CJK_OFFSET) {
|
|
|
|
result = i - NON_CJK_OFFSET;
|
|
|
|
} else if(i >= CJK_B_BASE) {
|
|
|
|
result = i;
|
|
|
|
} else if(i < CJK_A_LIMIT + (CJK_LIMIT - CJK_BASE) + (CJK_COMPAT_USED_LIMIT - CJK_COMPAT_USED_BASE)) { // rest of CJKs, compacted
|
|
|
|
if(i < CJK_LIMIT - CJK_BASE) {
|
|
|
|
result = i + CJK_BASE;
|
|
|
|
} else if(i < (CJK_LIMIT - CJK_BASE) + (CJK_COMPAT_USED_LIMIT - CJK_COMPAT_USED_BASE)) {
|
|
|
|
result = i + CJK_COMPAT_USED_BASE - (CJK_LIMIT - CJK_BASE);
|
|
|
|
} else {
|
|
|
|
result = i + CJK_A_BASE - (CJK_LIMIT - CJK_BASE) - (CJK_COMPAT_USED_LIMIT - CJK_COMPAT_USED_BASE);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
result = -1;
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
2002-06-13 18:34:41 +00:00
|
|
|
|
|
|
|
// GET IMPLICIT PRIMARY WEIGHTS
|
|
|
|
// Return value is left justified primary key
|
2004-01-16 07:09:12 +00:00
|
|
|
U_CAPI uint32_t U_EXPORT2
|
|
|
|
uprv_uca_getImplicitFromRaw(UChar32 cp) {
|
|
|
|
/*
|
2004-01-16 16:40:30 +00:00
|
|
|
if (cp < 0 || cp > UCOL_MAX_INPUT) {
|
2004-01-16 07:09:12 +00:00
|
|
|
throw new IllegalArgumentException("Code point out of range " + Utility.hex(cp));
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
int32_t last0 = cp - min4Boundary;
|
2002-06-13 18:34:41 +00:00
|
|
|
if (last0 < 0) {
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t last1 = cp / final3Count;
|
|
|
|
last0 = cp % final3Count;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t last2 = last1 / medialCount;
|
|
|
|
last1 %= medialCount;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
last0 = minTrail + last0*final3Multiplier; // spread out, leaving gap at start
|
|
|
|
last1 = minTrail + last1; // offset
|
|
|
|
last2 = min3Primary + last2; // offset
|
2002-06-13 18:34:41 +00:00
|
|
|
/*
|
2004-01-16 07:09:12 +00:00
|
|
|
if (last2 >= min4Primary) {
|
|
|
|
throw new IllegalArgumentException("4-byte out of range: " + Utility.hex(cp) + ", " + Utility.hex(last2));
|
2004-11-11 23:34:58 +00:00
|
|
|
}
|
2002-06-13 18:34:41 +00:00
|
|
|
*/
|
2004-01-16 07:09:12 +00:00
|
|
|
return (last2 << 24) + (last1 << 16) + (last0 << 8);
|
2002-06-13 18:34:41 +00:00
|
|
|
} else {
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t last1 = last0 / final4Count;
|
|
|
|
last0 %= final4Count;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t last2 = last1 / medialCount;
|
|
|
|
last1 %= medialCount;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t last3 = last2 / medialCount;
|
|
|
|
last2 %= medialCount;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
|
|
|
last0 = minTrail + last0*final4Multiplier; // spread out, leaving gap at start
|
2004-01-16 07:09:12 +00:00
|
|
|
last1 = minTrail + last1; // offset
|
|
|
|
last2 = minTrail + last2; // offset
|
|
|
|
last3 = min4Primary + last3; // offset
|
|
|
|
/*
|
|
|
|
if (last3 > max4Primary) {
|
|
|
|
throw new IllegalArgumentException("4-byte out of range: " + Utility.hex(cp) + ", " + Utility.hex(last3));
|
2004-11-11 23:34:58 +00:00
|
|
|
}
|
2004-01-16 07:09:12 +00:00
|
|
|
*/
|
|
|
|
return (last3 << 24) + (last2 << 16) + (last1 << 8) + last0;
|
|
|
|
}
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-06-03 06:08:46 +00:00
|
|
|
static uint32_t U_EXPORT2
|
2004-01-16 07:09:12 +00:00
|
|
|
uprv_uca_getImplicitPrimary(UChar32 cp) {
|
|
|
|
//if (DEBUG) System.out.println("Incoming: " + Utility.hex(cp));
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
cp = swapCJK(cp);
|
|
|
|
cp++;
|
|
|
|
// we now have a range of numbers from 0 to 21FFFF.
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
//if (DEBUG) System.out.println("CJK swapped: " + Utility.hex(cp));
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
return uprv_uca_getImplicitFromRaw(cp);
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
/**
|
|
|
|
* Converts implicit CE into raw integer ("code point")
|
|
|
|
* @param implicit
|
|
|
|
* @return -1 if illegal format
|
|
|
|
*/
|
|
|
|
U_CAPI UChar32 U_EXPORT2
|
|
|
|
uprv_uca_getRawFromImplicit(uint32_t implicit) {
|
|
|
|
UChar32 result;
|
|
|
|
UChar32 b3 = implicit & 0xFF;
|
2007-12-28 20:08:41 +00:00
|
|
|
UChar32 b2 = (implicit >> 8) & 0xFF;
|
|
|
|
UChar32 b1 = (implicit >> 16) & 0xFF;
|
|
|
|
UChar32 b0 = (implicit >> 24) & 0xFF;
|
2004-01-16 07:09:12 +00:00
|
|
|
|
|
|
|
// simple parameter checks
|
|
|
|
if (b0 < min3Primary || b0 > max4Primary
|
2007-12-28 20:08:41 +00:00
|
|
|
|| b1 < minTrail || b1 > maxTrail)
|
|
|
|
return -1;
|
2004-01-16 07:09:12 +00:00
|
|
|
// normal offsets
|
|
|
|
b1 -= minTrail;
|
|
|
|
|
|
|
|
// take care of the final values, and compose
|
|
|
|
if (b0 < min4Primary) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (b2 < minTrail || b2 > max3Trail || b3 != 0)
|
|
|
|
return -1;
|
2004-01-16 07:09:12 +00:00
|
|
|
b2 -= minTrail;
|
|
|
|
UChar32 remainder = b2 % final3Multiplier;
|
2007-12-26 19:27:43 +00:00
|
|
|
if (remainder != 0)
|
|
|
|
return -1;
|
2004-01-16 07:09:12 +00:00
|
|
|
b0 -= min3Primary;
|
|
|
|
b2 /= final3Multiplier;
|
|
|
|
result = ((b0 * medialCount) + b1) * final3Count + b2;
|
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (b2 < minTrail || b2 > maxTrail
|
|
|
|
|| b3 < minTrail || b3 > max4Trail)
|
|
|
|
return -1;
|
2004-01-16 07:09:12 +00:00
|
|
|
b2 -= minTrail;
|
|
|
|
b3 -= minTrail;
|
|
|
|
UChar32 remainder = b3 % final4Multiplier;
|
2007-12-26 19:27:43 +00:00
|
|
|
if (remainder != 0)
|
|
|
|
return -1;
|
2004-01-16 07:09:12 +00:00
|
|
|
b3 /= final4Multiplier;
|
|
|
|
b0 -= min4Primary;
|
|
|
|
result = (((b0 * medialCount) + b1) * medialCount + b2) * final4Count + b3 + min4Boundary;
|
|
|
|
}
|
|
|
|
// final check
|
2007-12-26 19:27:43 +00:00
|
|
|
if (result < 0 || result > UCOL_MAX_INPUT)
|
|
|
|
return -1;
|
2004-01-16 07:09:12 +00:00
|
|
|
return result;
|
|
|
|
}
|
2002-06-13 18:34:41 +00:00
|
|
|
|
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
static inline int32_t divideAndRoundUp(int a, int b) {
|
|
|
|
return 1 + (a-1)/b;
|
2002-06-13 18:34:41 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* this function is either called from initUCA or from genUCA before
|
|
|
|
* doing canonical closure for the UCA.
|
|
|
|
*/
|
2004-01-16 07:09:12 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Set up to generate implicits.
|
2008-08-06 17:57:45 +00:00
|
|
|
* Maintenance Note: this function may end up being called more than once, due
|
|
|
|
* to threading races during initialization. Make sure that
|
|
|
|
* none of the Constants is ever transiently assigned an
|
|
|
|
* incorrect value.
|
2004-01-16 07:09:12 +00:00
|
|
|
* @param minPrimary
|
|
|
|
* @param maxPrimary
|
|
|
|
* @param minTrail final byte
|
|
|
|
* @param maxTrail final byte
|
|
|
|
* @param gap3 the gap we leave for tailoring for 3-byte forms
|
|
|
|
* @param gap4 the gap we leave for tailoring for 4-byte forms
|
|
|
|
*/
|
2004-11-11 23:34:58 +00:00
|
|
|
static void initImplicitConstants(int minPrimary, int maxPrimary,
|
|
|
|
int minTrailIn, int maxTrailIn,
|
2004-01-16 07:09:12 +00:00
|
|
|
int gap3, int primaries3count,
|
|
|
|
UErrorCode *status) {
|
|
|
|
// some simple parameter checks
|
2007-12-28 20:08:41 +00:00
|
|
|
if ((minPrimary < 0 || minPrimary >= maxPrimary || maxPrimary > 0xFF)
|
|
|
|
|| (minTrailIn < 0 || minTrailIn >= maxTrailIn || maxTrailIn > 0xFF)
|
|
|
|
|| (primaries3count < 1))
|
|
|
|
{
|
2004-01-16 07:09:12 +00:00
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return;
|
|
|
|
};
|
|
|
|
|
|
|
|
minTrail = minTrailIn;
|
|
|
|
maxTrail = maxTrailIn;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
min3Primary = minPrimary;
|
|
|
|
max4Primary = maxPrimary;
|
|
|
|
// compute constants for use later.
|
|
|
|
// number of values we can use in trailing bytes
|
|
|
|
// leave room for empty values between AND above, e.g. if gap = 2
|
|
|
|
// range 3..7 => +3 -4 -5 -6 -7: so 1 value
|
|
|
|
// range 3..8 => +3 -4 -5 +6 -7 -8: so 2 values
|
|
|
|
// range 3..9 => +3 -4 -5 +6 -7 -8 -9: so 2 values
|
|
|
|
final3Multiplier = gap3 + 1;
|
|
|
|
final3Count = (maxTrail - minTrail + 1) / final3Multiplier;
|
|
|
|
max3Trail = minTrail + (final3Count - 1) * final3Multiplier;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
// medials can use full range
|
|
|
|
medialCount = (maxTrail - minTrail + 1);
|
|
|
|
// find out how many values fit in each form
|
|
|
|
int32_t threeByteCount = medialCount * final3Count;
|
|
|
|
// now determine where the 3/4 boundary is.
|
|
|
|
// we use 3 bytes below the boundary, and 4 above
|
|
|
|
int32_t primariesAvailable = maxPrimary - minPrimary + 1;
|
|
|
|
int32_t primaries4count = primariesAvailable - primaries3count;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t min3ByteCoverage = primaries3count * threeByteCount;
|
|
|
|
min4Primary = minPrimary + primaries3count;
|
|
|
|
min4Boundary = min3ByteCoverage;
|
|
|
|
// Now expand out the multiplier for the 4 bytes, and redo.
|
|
|
|
|
2004-01-16 16:40:30 +00:00
|
|
|
int32_t totalNeeded = UCOL_MAX_INPUT - min4Boundary;
|
2004-01-16 07:09:12 +00:00
|
|
|
int32_t neededPerPrimaryByte = divideAndRoundUp(totalNeeded, primaries4count);
|
|
|
|
int32_t neededPerFinalByte = divideAndRoundUp(neededPerPrimaryByte, medialCount * medialCount);
|
|
|
|
int32_t gap4 = (maxTrail - minTrail - 1) / neededPerFinalByte;
|
|
|
|
if (gap4 < 1) {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
final4Multiplier = gap4 + 1;
|
|
|
|
final4Count = neededPerFinalByte;
|
|
|
|
max4Trail = minTrail + (final4Count - 1) * final4Multiplier;
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2004-01-16 07:09:12 +00:00
|
|
|
/**
|
|
|
|
* Supply parameters for generating implicit CEs
|
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
2008-01-05 01:27:56 +00:00
|
|
|
uprv_uca_initImplicitConstants(UErrorCode *status) {
|
2004-01-16 07:09:12 +00:00
|
|
|
// 13 is the largest 4-byte gap we can use without getting 2 four-byte forms.
|
2005-07-06 21:27:21 +00:00
|
|
|
//initImplicitConstants(minPrimary, maxPrimary, 0x04, 0xFE, 1, 1, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
initImplicitConstants(minImplicitPrimary, maxImplicitPrimary, 0x04, 0xFE, 1, 1, status);
|
2002-06-13 18:34:41 +00:00
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
|
2001-04-23 21:29:14 +00:00
|
|
|
/* collIterNormalize Incremental Normalization happens here. */
|
|
|
|
/* pick up the range of chars identifed by FCD, */
|
|
|
|
/* normalize it into the collIterate's writable buffer, */
|
|
|
|
/* switch the collIterate's state to use the writable buffer. */
|
|
|
|
/* */
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2001-04-23 21:29:14 +00:00
|
|
|
void collIterNormalize(collIterate *collationSource)
|
|
|
|
{
|
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
2003-01-20 07:43:32 +00:00
|
|
|
|
|
|
|
int32_t normLen;
|
2001-04-23 21:29:14 +00:00
|
|
|
UChar *srcP = collationSource->pos - 1; /* Start of chars to normalize */
|
|
|
|
UChar *endP = collationSource->fcdPosition; /* End of region to normalize+1 */
|
|
|
|
|
2001-10-19 17:36:02 +00:00
|
|
|
normLen = unorm_decompose(collationSource->writableBuffer, (int32_t)collationSource->writableBufSize,
|
2001-09-27 01:01:30 +00:00
|
|
|
srcP, (int32_t)(endP - srcP),
|
2003-02-15 02:02:13 +00:00
|
|
|
FALSE, 0,
|
2001-09-27 01:01:30 +00:00
|
|
|
&status);
|
2001-10-19 17:36:02 +00:00
|
|
|
if(status == U_BUFFER_OVERFLOW_ERROR || status == U_STRING_NOT_TERMINATED_WARNING) {
|
2001-09-27 01:01:30 +00:00
|
|
|
// reallocate and terminate
|
|
|
|
if(!u_growBufferFromStatic(collationSource->stackWritableBuffer,
|
|
|
|
&collationSource->writableBuffer,
|
|
|
|
(int32_t *)&collationSource->writableBufSize, normLen + 1,
|
2001-10-19 17:36:02 +00:00
|
|
|
0)
|
2001-09-27 01:01:30 +00:00
|
|
|
) {
|
2001-05-17 23:09:35 +00:00
|
|
|
#ifdef UCOL_DEBUG
|
2001-09-27 01:01:30 +00:00
|
|
|
fprintf(stderr, "collIterNormalize(), out of memory\n");
|
2001-05-17 23:09:35 +00:00
|
|
|
#endif
|
2001-04-23 21:29:14 +00:00
|
|
|
return;
|
|
|
|
}
|
2001-10-19 17:36:02 +00:00
|
|
|
status = U_ZERO_ERROR;
|
|
|
|
normLen = unorm_decompose(collationSource->writableBuffer, (int32_t)collationSource->writableBufSize,
|
|
|
|
srcP, (int32_t)(endP - srcP),
|
2003-02-15 02:02:13 +00:00
|
|
|
FALSE, 0,
|
2001-10-19 17:36:02 +00:00
|
|
|
&status);
|
|
|
|
}
|
|
|
|
if (U_FAILURE(status)) {
|
|
|
|
#ifdef UCOL_DEBUG
|
|
|
|
fprintf(stderr, "collIterNormalize(), unorm_decompose() failed, status = %s\n", u_errorName(status));
|
|
|
|
#endif
|
|
|
|
return;
|
2001-04-23 21:29:14 +00:00
|
|
|
}
|
2001-04-24 03:18:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(collationSource->writableBuffer != collationSource->stackWritableBuffer) {
|
|
|
|
collationSource->flags |= UCOL_ITER_ALLOCATED;
|
|
|
|
}
|
|
|
|
collationSource->pos = collationSource->writableBuffer;
|
|
|
|
collationSource->origFlags = collationSource->flags;
|
|
|
|
collationSource->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
collationSource->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN | UCOL_USE_ITERATOR);
|
2001-04-23 21:29:14 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
// This function takes the iterator and extracts normalized stuff up to the next boundary
|
|
|
|
// It is similar in the end results to the collIterNormalize, but for the cases when we
|
|
|
|
// use an iterator
|
2007-07-24 08:30:41 +00:00
|
|
|
/*static
|
2003-01-20 07:43:32 +00:00
|
|
|
inline void normalizeIterator(collIterate *collationSource) {
|
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
|
|
|
UBool wasNormalized = FALSE;
|
2003-01-23 01:52:34 +00:00
|
|
|
//int32_t iterIndex = collationSource->iterator->getIndex(collationSource->iterator, UITER_CURRENT);
|
|
|
|
uint32_t iterIndex = collationSource->iterator->getState(collationSource->iterator);
|
2003-12-03 00:45:12 +00:00
|
|
|
int32_t normLen = unorm_next(collationSource->iterator, collationSource->writableBuffer,
|
2003-01-20 07:43:32 +00:00
|
|
|
(int32_t)collationSource->writableBufSize, UNORM_FCD, 0, TRUE, &wasNormalized, &status);
|
|
|
|
if(status == U_BUFFER_OVERFLOW_ERROR || normLen == (int32_t)collationSource->writableBufSize) {
|
|
|
|
// reallocate and terminate
|
|
|
|
if(!u_growBufferFromStatic(collationSource->stackWritableBuffer,
|
|
|
|
&collationSource->writableBuffer,
|
|
|
|
(int32_t *)&collationSource->writableBufSize, normLen + 1,
|
|
|
|
0)
|
|
|
|
) {
|
|
|
|
#ifdef UCOL_DEBUG
|
|
|
|
fprintf(stderr, "normalizeIterator(), out of memory\n");
|
|
|
|
#endif
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
status = U_ZERO_ERROR;
|
2003-01-23 01:52:34 +00:00
|
|
|
//collationSource->iterator->move(collationSource->iterator, iterIndex, UITER_ZERO);
|
|
|
|
collationSource->iterator->setState(collationSource->iterator, iterIndex, &status);
|
2003-12-03 00:45:12 +00:00
|
|
|
normLen = unorm_next(collationSource->iterator, collationSource->writableBuffer,
|
2003-01-20 07:43:32 +00:00
|
|
|
(int32_t)collationSource->writableBufSize, UNORM_FCD, 0, TRUE, &wasNormalized, &status);
|
|
|
|
}
|
|
|
|
// Terminate the buffer - we already checked that it is big enough
|
2003-12-03 00:45:12 +00:00
|
|
|
collationSource->writableBuffer[normLen] = 0;
|
2003-01-20 07:43:32 +00:00
|
|
|
if(collationSource->writableBuffer != collationSource->stackWritableBuffer) {
|
|
|
|
collationSource->flags |= UCOL_ITER_ALLOCATED;
|
|
|
|
}
|
|
|
|
collationSource->pos = collationSource->writableBuffer;
|
|
|
|
collationSource->origFlags = collationSource->flags;
|
|
|
|
collationSource->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
collationSource->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN | UCOL_USE_ITERATOR);
|
2007-07-24 08:30:41 +00:00
|
|
|
}*/
|
2003-01-20 07:43:32 +00:00
|
|
|
|
2001-04-23 21:29:14 +00:00
|
|
|
|
|
|
|
/* Incremental FCD check and normalize */
|
|
|
|
/* Called from getNextCE when normalization state is suspect. */
|
|
|
|
/* When entering, the state is known to be this: */
|
|
|
|
/* o We are working in the main buffer of the collIterate, not the side */
|
|
|
|
/* writable buffer. When in the side buffer, normalization mode is always off, */
|
|
|
|
/* so we won't get here. */
|
|
|
|
/* o The leading combining class from the current character is 0 or */
|
|
|
|
/* the trailing combining class of the previous char was zero. */
|
|
|
|
/* True because the previous call to this function will have always exited */
|
|
|
|
/* that way, and we get called for every char where cc might be non-zero. */
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-04-24 03:18:54 +00:00
|
|
|
inline UBool collIterFCD(collIterate *collationSource) {
|
2001-08-17 00:21:18 +00:00
|
|
|
UChar c, c2;
|
|
|
|
const UChar *srcP, *endP;
|
2001-04-23 21:29:14 +00:00
|
|
|
uint8_t leadingCC;
|
|
|
|
uint8_t prevTrailingCC = 0;
|
|
|
|
uint16_t fcd;
|
|
|
|
UBool needNormalize = FALSE;
|
|
|
|
|
|
|
|
srcP = collationSource->pos-1;
|
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
if (collationSource->flags & UCOL_ITER_HASLEN) {
|
|
|
|
endP = collationSource->endp;
|
|
|
|
} else {
|
|
|
|
endP = NULL;
|
|
|
|
}
|
2001-04-23 21:29:14 +00:00
|
|
|
|
|
|
|
// Get the trailing combining class of the current character. If it's zero,
|
|
|
|
// we are OK.
|
2001-08-17 00:21:18 +00:00
|
|
|
c = *srcP++;
|
2001-04-23 21:29:14 +00:00
|
|
|
/* trie access */
|
2001-08-17 00:21:18 +00:00
|
|
|
fcd = unorm_getFCD16(fcdTrieIndex, c);
|
|
|
|
if (fcd != 0) {
|
2005-05-19 06:43:35 +00:00
|
|
|
if (U16_IS_LEAD(c)) {
|
|
|
|
if ((endP == NULL || srcP != endP) && U16_IS_TRAIL(c2=*srcP)) {
|
2001-08-17 00:21:18 +00:00
|
|
|
++srcP;
|
|
|
|
fcd = unorm_getFCD16FromSurrogatePair(fcdTrieIndex, fcd, c2);
|
|
|
|
} else {
|
|
|
|
fcd = 0;
|
2001-04-23 21:29:14 +00:00
|
|
|
}
|
2001-08-17 00:21:18 +00:00
|
|
|
}
|
2001-04-23 21:29:14 +00:00
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
prevTrailingCC = (uint8_t)(fcd & LAST_BYTE_MASK_);
|
|
|
|
|
|
|
|
if (prevTrailingCC != 0) {
|
|
|
|
// The current char has a non-zero trailing CC. Scan forward until we find
|
|
|
|
// a char with a leading cc of zero.
|
|
|
|
while (endP == NULL || srcP != endP)
|
|
|
|
{
|
|
|
|
const UChar *savedSrcP = srcP;
|
2001-04-23 21:29:14 +00:00
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
c = *srcP++;
|
|
|
|
/* trie access */
|
|
|
|
fcd = unorm_getFCD16(fcdTrieIndex, c);
|
2005-05-19 06:43:35 +00:00
|
|
|
if (fcd != 0 && U16_IS_LEAD(c)) {
|
|
|
|
if ((endP == NULL || srcP != endP) && U16_IS_TRAIL(c2=*srcP)) {
|
2001-08-17 00:21:18 +00:00
|
|
|
++srcP;
|
|
|
|
fcd = unorm_getFCD16FromSurrogatePair(fcdTrieIndex, fcd, c2);
|
|
|
|
} else {
|
|
|
|
fcd = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
leadingCC = (uint8_t)(fcd >> SECOND_LAST_BYTE_SHIFT_);
|
|
|
|
if (leadingCC == 0) {
|
|
|
|
srcP = savedSrcP; // Hit char that is not part of combining sequence.
|
|
|
|
// back up over it. (Could be surrogate pair!)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (leadingCC < prevTrailingCC) {
|
|
|
|
needNormalize = TRUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
prevTrailingCC = (uint8_t)(fcd & LAST_BYTE_MASK_);
|
|
|
|
}
|
2001-04-23 21:29:14 +00:00
|
|
|
}
|
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
collationSource->fcdPosition = (UChar *)srcP;
|
2001-04-23 21:29:14 +00:00
|
|
|
|
2001-04-24 03:18:54 +00:00
|
|
|
return needNormalize;
|
2001-04-23 21:29:14 +00:00
|
|
|
}
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/****************************************************************************/
|
|
|
|
/* Following are the CE retrieval functions */
|
|
|
|
/* */
|
|
|
|
/****************************************************************************/
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2005-07-06 21:27:21 +00:00
|
|
|
static uint32_t getImplicit(UChar32 cp, collIterate *collationSource);
|
|
|
|
static uint32_t getPrevImplicit(UChar32 cp, collIterate *collationSource);
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* there should be a macro version of this function in the header file */
|
|
|
|
/* This is the first function that tries to fetch a collation element */
|
|
|
|
/* If it's not succesfull or it encounters a more difficult situation */
|
|
|
|
/* some more sofisticated and slower functions are invoked */
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-04-18 19:31:05 +00:00
|
|
|
inline uint32_t ucol_IGetNextCE(const UCollator *coll, collIterate *collationSource, UErrorCode *status) {
|
2002-12-04 00:28:06 +00:00
|
|
|
uint32_t order = 0;
|
2001-02-06 00:36:48 +00:00
|
|
|
if (collationSource->CEpos > collationSource->toReturn) { /* Are there any CEs from previous expansions? */
|
2007-12-26 19:27:43 +00:00
|
|
|
order = *(collationSource->toReturn++); /* if so, return them */
|
|
|
|
if(collationSource->CEpos == collationSource->toReturn) {
|
|
|
|
collationSource->CEpos = collationSource->toReturn = collationSource->extendCEs ? collationSource->extendCEs : collationSource->CEs;
|
|
|
|
}
|
|
|
|
return order;
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
|
|
|
|
2002-12-04 00:28:06 +00:00
|
|
|
UChar ch = 0;
|
2008-05-23 04:22:28 +00:00
|
|
|
collationSource->offsetReturn = NULL;
|
2001-04-06 23:37:48 +00:00
|
|
|
|
2001-04-12 19:59:28 +00:00
|
|
|
for (;;) /* Loop handles case when incremental normalize switches */
|
|
|
|
{ /* to or from the side buffer / original string, and we */
|
2007-12-26 19:27:43 +00:00
|
|
|
/* need to start again to get the next character. */
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
if ((collationSource->flags & (UCOL_ITER_HASLEN | UCOL_ITER_INNORMBUF | UCOL_ITER_NORM | UCOL_HIRAGANA_Q | UCOL_USE_ITERATOR)) == 0)
|
2001-04-23 01:53:49 +00:00
|
|
|
{
|
2001-04-06 23:37:48 +00:00
|
|
|
// The source string is null terminated and we're not working from the side buffer,
|
|
|
|
// and we're not normalizing. This is the fast path.
|
2001-04-23 21:29:14 +00:00
|
|
|
// (We can be in the side buffer for Thai pre-vowel reordering even when not normalizing.)
|
2001-04-06 23:37:48 +00:00
|
|
|
ch = *collationSource->pos++;
|
2001-04-12 19:59:28 +00:00
|
|
|
if (ch != 0) {
|
2001-04-06 23:37:48 +00:00
|
|
|
break;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
|
|
|
else {
|
2001-04-06 23:37:48 +00:00
|
|
|
return UCOL_NO_MORE_CES;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (collationSource->flags & UCOL_ITER_HASLEN) {
|
|
|
|
// Normal path for strings when length is specified.
|
|
|
|
// (We can't be in side buffer because it is always null terminated.)
|
|
|
|
if (collationSource->pos >= collationSource->endp) {
|
|
|
|
// Ran off of the end of the main source string. We're done.
|
|
|
|
return UCOL_NO_MORE_CES;
|
|
|
|
}
|
|
|
|
ch = *collationSource->pos++;
|
|
|
|
}
|
2003-01-20 07:43:32 +00:00
|
|
|
else if(collationSource->flags & UCOL_USE_ITERATOR) {
|
2003-02-06 23:29:56 +00:00
|
|
|
UChar32 iterCh = collationSource->iterator->next(collationSource->iterator);
|
|
|
|
if(iterCh == U_SENTINEL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
return UCOL_NO_MORE_CES;
|
2003-01-20 07:43:32 +00:00
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
ch = (UChar)iterCh;
|
2003-01-20 07:43:32 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
else
|
|
|
|
{
|
2001-04-23 21:29:14 +00:00
|
|
|
// Null terminated string.
|
2001-04-06 23:37:48 +00:00
|
|
|
ch = *collationSource->pos++;
|
2001-04-23 21:29:14 +00:00
|
|
|
if (ch == 0) {
|
|
|
|
// Ran off end of buffer.
|
|
|
|
if ((collationSource->flags & UCOL_ITER_INNORMBUF) == 0) {
|
2001-06-20 18:14:51 +00:00
|
|
|
// Ran off end of main string. backing up one character.
|
|
|
|
collationSource->pos--;
|
2001-04-23 21:29:14 +00:00
|
|
|
return UCOL_NO_MORE_CES;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2001-05-25 19:30:01 +00:00
|
|
|
// Hit null in the normalize side buffer.
|
|
|
|
// Usually this means the end of the normalized data,
|
|
|
|
// except for one odd case: a null followed by combining chars,
|
|
|
|
// which is the case if we are at the start of the buffer.
|
|
|
|
if (collationSource->pos == collationSource->writableBuffer+1) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Null marked end of side buffer.
|
|
|
|
// Revert to the main string and
|
2001-04-23 21:29:14 +00:00
|
|
|
// loop back to top to try again to get a character.
|
|
|
|
collationSource->pos = collationSource->fcdPosition;
|
|
|
|
collationSource->flags = collationSource->origFlags;
|
|
|
|
continue;
|
|
|
|
}
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-10-05 02:07:51 +00:00
|
|
|
if(collationSource->flags&UCOL_HIRAGANA_Q) {
|
2008-04-17 05:19:19 +00:00
|
|
|
/* Codepoints \u3099-\u309C are both Hiragana and Katakana. Set the flag
|
|
|
|
* based on whether the previous codepoint was Hiragana or Katakana.
|
|
|
|
*/
|
|
|
|
if(((ch>=0x3040 && ch<=0x3096) || (ch >= 0x309d && ch <= 0x309f)) ||
|
|
|
|
((collationSource->flags & UCOL_WAS_HIRAGANA) && (ch >= 0x3099 && ch <= 0x309C))) {
|
2007-12-26 19:27:43 +00:00
|
|
|
collationSource->flags |= UCOL_WAS_HIRAGANA;
|
|
|
|
} else {
|
|
|
|
collationSource->flags &= ~UCOL_WAS_HIRAGANA;
|
|
|
|
}
|
2001-10-05 02:07:51 +00:00
|
|
|
}
|
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
// We've got a character. See if there's any fcd and/or normalization stuff to do.
|
2001-04-23 21:29:14 +00:00
|
|
|
// Note that UCOL_ITER_NORM flag is always zero when we are in the side buffer.
|
2001-04-12 19:59:28 +00:00
|
|
|
if ((collationSource->flags & UCOL_ITER_NORM) == 0) {
|
2001-04-06 23:37:48 +00:00
|
|
|
break;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
if (collationSource->fcdPosition >= collationSource->pos) {
|
|
|
|
// An earlier FCD check has already covered the current character.
|
|
|
|
// We can go ahead and process this char.
|
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-20 22:29:53 +00:00
|
|
|
if (ch < ZERO_CC_LIMIT_ ) {
|
2001-04-06 23:37:48 +00:00
|
|
|
// Fast fcd safe path. Trailing combining class == 0. This char is OK.
|
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
if (ch < NFC_ZERO_CC_BLOCK_LIMIT_) {
|
|
|
|
// We need to peek at the next character in order to tell if we are FCD
|
2001-04-12 19:59:28 +00:00
|
|
|
if ((collationSource->flags & UCOL_ITER_HASLEN) && collationSource->pos >= collationSource->endp) {
|
2001-04-06 23:37:48 +00:00
|
|
|
// We are at the last char of source string.
|
|
|
|
// It is always OK for FCD check.
|
|
|
|
break;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
// Not at last char of source string (or we'll check against terminating null). Do the FCD fast test
|
2001-04-12 19:59:28 +00:00
|
|
|
if (*collationSource->pos < NFC_ZERO_CC_BLOCK_LIMIT_) {
|
2001-04-06 23:37:48 +00:00
|
|
|
break;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
// Need a more complete FCD check and possible normalization.
|
2001-04-24 03:18:54 +00:00
|
|
|
if (collIterFCD(collationSource)) {
|
|
|
|
collIterNormalize(collationSource);
|
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
if ((collationSource->flags & UCOL_ITER_INNORMBUF) == 0) {
|
|
|
|
// No normalization was needed. Go ahead and process the char we already had.
|
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
// Some normalization happened. Next loop iteration will pick up a char
|
|
|
|
// from the normalization buffer.
|
2001-04-23 01:53:49 +00:00
|
|
|
|
|
|
|
} // end for (;;)
|
|
|
|
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (ch <= 0xFF) {
|
|
|
|
/* For latin-1 characters we never need to fall back to the UCA table */
|
|
|
|
/* because all of the UCA data is replicated in the latinOneMapping array */
|
|
|
|
order = coll->latinOneMapping[ch];
|
|
|
|
if (order > UCOL_NOT_FOUND) {
|
|
|
|
order = ucol_prv_getSpecialCE(coll, ch, order, collationSource, status);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
order = UTRIE_GET32_FROM_LEAD(&coll->mapping, ch);
|
|
|
|
if(order > UCOL_NOT_FOUND) { /* if a CE is special */
|
|
|
|
order = ucol_prv_getSpecialCE(coll, ch, order, collationSource, status); /* and try to get the special CE */
|
|
|
|
}
|
|
|
|
if(order == UCOL_NOT_FOUND && coll->UCA) { /* We couldn't find a good CE in the tailoring */
|
2001-09-20 20:16:39 +00:00
|
|
|
/* if we got here, the codepoint MUST be over 0xFF - so we look directly in the trie */
|
2006-05-21 01:27:15 +00:00
|
|
|
order = UTRIE_GET32_FROM_LEAD(&coll->UCA->mapping, ch);
|
2001-09-20 20:16:39 +00:00
|
|
|
|
|
|
|
if(order > UCOL_NOT_FOUND) { /* UCA also gives us a special CE */
|
2007-12-26 19:27:43 +00:00
|
|
|
order = ucol_prv_getSpecialCE(coll->UCA, ch, order, collationSource, status);
|
2001-09-20 20:16:39 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if(order == UCOL_NOT_FOUND) {
|
2005-07-06 21:27:21 +00:00
|
|
|
order = getImplicit(ch, collationSource);
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
return order; /* return the CE */
|
2001-02-06 00:36:48 +00:00
|
|
|
}
|
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
/* ucol_getNextCE, out-of-line version for use from other files. */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI uint32_t U_EXPORT2
|
|
|
|
ucol_getNextCE(const UCollator *coll, collIterate *collationSource, UErrorCode *status) {
|
2001-04-18 19:31:05 +00:00
|
|
|
return ucol_IGetNextCE(coll, collationSource, status);
|
2004-06-12 17:50:57 +00:00
|
|
|
}
|
2001-04-18 19:31:05 +00:00
|
|
|
|
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/**
|
2001-04-23 01:53:49 +00:00
|
|
|
* Incremental previous normalization happens here. Pick up the range of chars
|
|
|
|
* identifed by FCD, normalize it into the collIterate's writable buffer,
|
|
|
|
* switch the collIterate's state to use the writable buffer.
|
2001-04-12 00:08:26 +00:00
|
|
|
* @param data collation iterator data
|
|
|
|
*/
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2001-04-12 00:08:26 +00:00
|
|
|
void collPrevIterNormalize(collIterate *data)
|
|
|
|
{
|
2001-04-17 02:43:35 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
2001-04-24 03:18:54 +00:00
|
|
|
UChar *pEnd = data->pos; /* End normalize + 1 */
|
2001-04-12 18:25:07 +00:00
|
|
|
UChar *pStart;
|
2001-04-12 00:08:26 +00:00
|
|
|
uint32_t normLen;
|
2001-04-20 22:29:53 +00:00
|
|
|
UChar *pStartNorm;
|
2001-04-12 00:08:26 +00:00
|
|
|
|
2001-04-12 18:25:07 +00:00
|
|
|
/* Start normalize */
|
|
|
|
if (data->fcdPosition == NULL) {
|
|
|
|
pStart = data->string;
|
|
|
|
}
|
|
|
|
else {
|
2001-04-23 01:53:49 +00:00
|
|
|
pStart = data->fcdPosition + 1;
|
2001-04-12 18:25:07 +00:00
|
|
|
}
|
|
|
|
|
2001-04-24 03:18:54 +00:00
|
|
|
normLen = unorm_normalize(pStart, (pEnd - pStart) + 1, UNORM_NFD, 0,
|
2001-04-20 22:29:53 +00:00
|
|
|
data->writableBuffer, 0, &status);
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-04-20 22:29:53 +00:00
|
|
|
if (data->writableBufSize <= normLen) {
|
2001-05-10 17:49:24 +00:00
|
|
|
freeHeapWritableBuffer(data);
|
2001-04-23 01:53:49 +00:00
|
|
|
data->writableBuffer = (UChar *)uprv_malloc((normLen + 1) *
|
|
|
|
sizeof(UChar));
|
2002-07-20 06:00:04 +00:00
|
|
|
if(data->writableBuffer == NULL) { // something is wrong here, return
|
2008-02-13 09:35:50 +00:00
|
|
|
data->writableBufSize = 0; // Reset writableBufSize
|
|
|
|
return;
|
2002-07-20 06:00:04 +00:00
|
|
|
}
|
2001-09-27 01:01:30 +00:00
|
|
|
data->flags |= UCOL_ITER_ALLOCATED;
|
2001-04-23 01:53:49 +00:00
|
|
|
/* to handle the zero termination */
|
|
|
|
data->writableBufSize = normLen + 1;
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
status = U_ZERO_ERROR;
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
2001-04-20 22:29:53 +00:00
|
|
|
this puts the null termination infront of the normalized string instead
|
|
|
|
of the end
|
|
|
|
*/
|
|
|
|
pStartNorm = data->writableBuffer + (data->writableBufSize - normLen);
|
|
|
|
*(pStartNorm - 1) = 0;
|
2001-05-11 01:13:08 +00:00
|
|
|
unorm_normalize(pStart, (pEnd - pStart) + 1, UNORM_NFD, 0, pStartNorm,
|
2001-04-24 03:18:54 +00:00
|
|
|
normLen, &status);
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if (data->offsetBuffer == NULL) {
|
|
|
|
int32_t len = normLen >= UCOL_EXPAND_CE_BUFFER_SIZE ? normLen + 1 : UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
|
|
|
|
data->offsetBufferSize = len;
|
|
|
|
data->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * len);
|
|
|
|
data->offsetStore = data->offsetBuffer;
|
|
|
|
} else if(data->offsetBufferSize < (int32_t) normLen) {
|
|
|
|
int32_t storeIX = data->offsetStore - data->offsetBuffer;
|
|
|
|
int32_t *tob = (int32_t *) uprv_realloc(data->offsetBuffer, sizeof(int32_t) * (normLen + 1));
|
|
|
|
|
|
|
|
if (tob != NULL) {
|
|
|
|
data->offsetBuffer = tob;
|
|
|
|
data->offsetStore = &data->offsetBuffer[storeIX];
|
|
|
|
data->offsetBufferSize = normLen + 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The usual case at this point is that we've got a base
|
|
|
|
* character followed by marks that were normalized. If
|
|
|
|
* fcdPosition is NULL, that means that we backed up to
|
|
|
|
* the beginning of the string and there's no base character.
|
|
|
|
*
|
|
|
|
* Forward processing will usually normalize when it sees
|
|
|
|
* the first mark, so that mark will get it's natural offset
|
|
|
|
* and the rest will get the offset of the character following
|
|
|
|
* the marks. The base character will also get its natural offset.
|
|
|
|
*
|
|
|
|
* We write the offset of the base character, if there is one,
|
|
|
|
* followed by the offset of the first mark and then the offsets
|
|
|
|
* of the rest of the marks.
|
|
|
|
*/
|
|
|
|
int32_t firstMarkOffset = 0;
|
|
|
|
int32_t trailOffset = data->pos - data->string + 1;
|
|
|
|
int32_t trailCount = normLen - 1;
|
|
|
|
|
|
|
|
if (data->fcdPosition != NULL) {
|
|
|
|
int32_t baseOffset = data->fcdPosition - data->string;
|
|
|
|
UChar baseChar = *data->fcdPosition;
|
|
|
|
|
|
|
|
firstMarkOffset = baseOffset + 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the base character is the start of a contraction, forward processing
|
|
|
|
* will normalize the marks while checking for the contraction, which means
|
|
|
|
* that the offset of the first mark will the same as the other marks.
|
|
|
|
*
|
|
|
|
* **** THIS IS PROBABLY NOT A COMPLETE TEST ****
|
|
|
|
*/
|
|
|
|
if (baseChar >= 0x100) {
|
2008-05-26 19:18:33 +00:00
|
|
|
uint32_t baseOrder = UTRIE_GET32_FROM_LEAD(&data->coll->mapping, baseChar);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (baseOrder == UCOL_NOT_FOUND && data->coll->UCA) {
|
|
|
|
baseOrder = UTRIE_GET32_FROM_LEAD(&data->coll->UCA->mapping, baseChar);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (baseOrder > UCOL_NOT_FOUND && getCETag(baseOrder) == CONTRACTION_TAG) {
|
|
|
|
firstMarkOffset = trailOffset;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
*(data->offsetStore++) = baseOffset;
|
|
|
|
}
|
|
|
|
|
|
|
|
*(data->offsetStore++) = firstMarkOffset;
|
|
|
|
|
|
|
|
for (int32_t i = 0; i < trailCount; i += 1) {
|
|
|
|
*(data->offsetStore++) = trailOffset;
|
|
|
|
}
|
|
|
|
|
|
|
|
data->offsetRepeatValue = trailOffset;
|
|
|
|
|
|
|
|
data->offsetReturn = data->offsetStore - 1;
|
|
|
|
if (data->offsetReturn == data->offsetBuffer) {
|
|
|
|
data->offsetStore = data->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
2001-04-20 22:29:53 +00:00
|
|
|
data->pos = data->writableBuffer + data->writableBufSize;
|
2001-04-12 00:08:26 +00:00
|
|
|
data->origFlags = data->flags;
|
|
|
|
data->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
data->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN);
|
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
|
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/**
|
2001-04-23 01:53:49 +00:00
|
|
|
* Incremental FCD check for previous iteration and normalize. Called from
|
|
|
|
* getPrevCE when normalization state is suspect.
|
|
|
|
* When entering, the state is known to be this:
|
|
|
|
* o We are working in the main buffer of the collIterate, not the side
|
|
|
|
* writable buffer. When in the side buffer, normalization mode is always
|
|
|
|
* off, so we won't get here.
|
|
|
|
* o The leading combining class from the current character is 0 or the
|
|
|
|
* trailing combining class of the previous char was zero.
|
|
|
|
* True because the previous call to this function will have always exited
|
|
|
|
* that way, and we get called for every char where cc might be non-zero.
|
2001-04-12 00:08:26 +00:00
|
|
|
* @param data collation iterate struct
|
2001-05-11 01:13:08 +00:00
|
|
|
* @return normalization status, TRUE for normalization to be done, FALSE
|
2001-04-20 22:29:53 +00:00
|
|
|
* otherwise
|
2001-04-12 00:08:26 +00:00
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-11 01:13:08 +00:00
|
|
|
inline UBool collPrevIterFCD(collIterate *data)
|
2001-04-12 00:08:26 +00:00
|
|
|
{
|
2001-08-17 00:21:18 +00:00
|
|
|
const UChar *src, *start;
|
|
|
|
UChar c, c2;
|
2001-04-12 00:08:26 +00:00
|
|
|
uint8_t leadingCC;
|
|
|
|
uint8_t trailingCC = 0;
|
|
|
|
uint16_t fcd;
|
2001-04-20 22:29:53 +00:00
|
|
|
UBool result = FALSE;
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
start = data->string;
|
|
|
|
src = data->pos + 1;
|
2001-04-06 23:37:48 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/* Get the trailing combining class of the current character. */
|
2001-08-17 00:21:18 +00:00
|
|
|
c = *--src;
|
2005-05-19 06:43:35 +00:00
|
|
|
if (!U16_IS_SURROGATE(c)) {
|
2001-08-17 00:21:18 +00:00
|
|
|
fcd = unorm_getFCD16(fcdTrieIndex, c);
|
2005-05-19 06:43:35 +00:00
|
|
|
} else if (U16_IS_TRAIL(c) && start < src && U16_IS_LEAD(c2 = *(src - 1))) {
|
2001-08-17 00:21:18 +00:00
|
|
|
--src;
|
|
|
|
fcd = unorm_getFCD16(fcdTrieIndex, c2);
|
|
|
|
if (fcd != 0) {
|
|
|
|
fcd = unorm_getFCD16FromSurrogatePair(fcdTrieIndex, fcd, c);
|
|
|
|
}
|
|
|
|
} else /* unpaired surrogate */ {
|
|
|
|
fcd = 0;
|
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
leadingCC = (uint8_t)(fcd >> SECOND_LAST_BYTE_SHIFT_);
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
if (leadingCC != 0) {
|
|
|
|
/*
|
2001-04-23 01:53:49 +00:00
|
|
|
The current char has a non-zero leading combining class.
|
2001-04-12 00:08:26 +00:00
|
|
|
Scan backward until we find a char with a trailing cc of zero.
|
|
|
|
*/
|
2001-05-02 23:36:22 +00:00
|
|
|
for (;;)
|
2001-04-12 00:08:26 +00:00
|
|
|
{
|
2001-08-17 00:21:18 +00:00
|
|
|
if (start == src) {
|
|
|
|
data->fcdPosition = NULL;
|
|
|
|
return result;
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
c = *--src;
|
2005-05-19 06:43:35 +00:00
|
|
|
if (!U16_IS_SURROGATE(c)) {
|
2001-08-17 00:21:18 +00:00
|
|
|
fcd = unorm_getFCD16(fcdTrieIndex, c);
|
2005-05-19 06:43:35 +00:00
|
|
|
} else if (U16_IS_TRAIL(c) && start < src && U16_IS_LEAD(c2 = *(src - 1))) {
|
2001-08-17 00:21:18 +00:00
|
|
|
--src;
|
|
|
|
fcd = unorm_getFCD16(fcdTrieIndex, c2);
|
|
|
|
if (fcd != 0) {
|
|
|
|
fcd = unorm_getFCD16FromSurrogatePair(fcdTrieIndex, fcd, c);
|
|
|
|
}
|
|
|
|
} else /* unpaired surrogate */ {
|
|
|
|
fcd = 0;
|
|
|
|
}
|
2001-04-12 00:08:26 +00:00
|
|
|
|
|
|
|
trailingCC = (uint8_t)(fcd & LAST_BYTE_MASK_);
|
|
|
|
|
|
|
|
if (trailingCC == 0) {
|
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
if (leadingCC < trailingCC) {
|
2001-04-20 22:29:53 +00:00
|
|
|
result = TRUE;
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
|
|
|
leadingCC = (uint8_t)(fcd >> SECOND_LAST_BYTE_SHIFT_);
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-08-17 00:21:18 +00:00
|
|
|
data->fcdPosition = (UChar *)src;
|
2001-04-12 00:08:26 +00:00
|
|
|
|
2001-04-20 22:29:53 +00:00
|
|
|
return result;
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
|
|
|
|
2003-02-06 23:29:56 +00:00
|
|
|
/** gets a character from the string at a given offset
|
|
|
|
* Handles both normal and iterative cases.
|
|
|
|
* No error checking - caller beware!
|
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
inline static
|
2003-02-06 23:29:56 +00:00
|
|
|
UChar peekCharacter(collIterate *source, int32_t offset) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(source->pos != NULL) {
|
|
|
|
return *(source->pos + offset);
|
|
|
|
} else if(source->iterator != NULL) {
|
|
|
|
if(offset != 0) {
|
|
|
|
source->iterator->move(source->iterator, offset, UITER_CURRENT);
|
|
|
|
UChar toReturn = (UChar)source->iterator->next(source->iterator);
|
|
|
|
source->iterator->move(source->iterator, -offset-1, UITER_CURRENT);
|
|
|
|
return toReturn;
|
|
|
|
} else {
|
|
|
|
return (UChar)source->iterator->current(source->iterator);
|
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
return (UChar)U_SENTINEL;
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2001-05-17 01:06:25 +00:00
|
|
|
/**
|
|
|
|
* Determines if we are at the start of the data string in the backwards
|
|
|
|
* collation iterator
|
|
|
|
* @param data collation iterator
|
|
|
|
* @return TRUE if we are at the start
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-17 01:06:25 +00:00
|
|
|
inline UBool isAtStartPrevIterate(collIterate *data) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(data->pos == NULL && data->iterator != NULL) {
|
|
|
|
return !data->iterator->hasPrevious(data->iterator);
|
|
|
|
}
|
|
|
|
//return (collIter_bos(data)) ||
|
|
|
|
return (data->pos == data->string) ||
|
|
|
|
((data->flags & UCOL_ITER_INNORMBUF) &&
|
|
|
|
*(data->pos - 1) == 0 && data->fcdPosition == NULL);
|
2001-05-17 01:06:25 +00:00
|
|
|
}
|
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
static
|
2003-07-16 22:47:34 +00:00
|
|
|
inline void goBackOne(collIterate *data) {
|
|
|
|
# if 0
|
2007-12-26 19:27:43 +00:00
|
|
|
// somehow, it looks like we need to keep iterator synced up
|
|
|
|
// at all times, as above.
|
|
|
|
if(data->pos) {
|
|
|
|
data->pos--;
|
|
|
|
}
|
|
|
|
if(data->iterator) {
|
|
|
|
data->iterator->previous(data->iterator);
|
|
|
|
}
|
2003-07-16 22:47:34 +00:00
|
|
|
#endif
|
2007-12-26 19:27:43 +00:00
|
|
|
if(data->iterator && (data->flags & UCOL_USE_ITERATOR)) {
|
|
|
|
data->iterator->previous(data->iterator);
|
|
|
|
}
|
|
|
|
if(data->pos) {
|
|
|
|
data->pos --;
|
|
|
|
}
|
2003-07-16 22:47:34 +00:00
|
|
|
}
|
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/**
|
|
|
|
* Inline function that gets a simple CE.
|
2001-04-23 01:53:49 +00:00
|
|
|
* So what it does is that it will first check the expansion buffer. If the
|
|
|
|
* expansion buffer is not empty, ie the end pointer to the expansion buffer
|
|
|
|
* is different from the string pointer, we return the collation element at the
|
2001-04-12 00:08:26 +00:00
|
|
|
* return pointer and decrement it.
|
|
|
|
* For more complicated CEs it resorts to getComplicatedCE.
|
|
|
|
* @param coll collator data
|
|
|
|
* @param data collation iterator struct
|
|
|
|
* @param status error status
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-04-18 19:31:05 +00:00
|
|
|
inline uint32_t ucol_IGetPrevCE(const UCollator *coll, collIterate *data,
|
2001-04-23 01:53:49 +00:00
|
|
|
UErrorCode *status)
|
2001-04-12 00:08:26 +00:00
|
|
|
{
|
2004-06-12 17:50:57 +00:00
|
|
|
uint32_t result = (uint32_t)UCOL_NULLORDER;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (data->offsetReturn != NULL) {
|
|
|
|
if (data->offsetRepeatCount > 0) {
|
|
|
|
data->offsetRepeatCount -= 1;
|
|
|
|
} else {
|
|
|
|
if (data->offsetReturn == data->offsetBuffer) {
|
|
|
|
data->offsetReturn = NULL;
|
|
|
|
data->offsetStore = data->offsetBuffer;
|
|
|
|
} else {
|
|
|
|
data->offsetReturn -= 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if ((data->extendCEs && data->toReturn > data->extendCEs) ||
|
|
|
|
(!data->extendCEs && data->toReturn > data->CEs))
|
|
|
|
{
|
2008-05-23 04:22:28 +00:00
|
|
|
data->toReturn -= 1;
|
2001-04-23 01:53:49 +00:00
|
|
|
result = *(data->toReturn);
|
2007-12-15 00:30:35 +00:00
|
|
|
if (data->CEs == data->toReturn || data->extendCEs == data->toReturn) {
|
2001-09-07 21:56:18 +00:00
|
|
|
data->CEpos = data->toReturn;
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
2002-12-04 00:28:06 +00:00
|
|
|
UChar ch = 0;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
/*
|
|
|
|
Loop handles case when incremental normalize switches to or from the
|
|
|
|
side buffer / original string, and we need to start again to get the
|
2001-04-12 00:08:26 +00:00
|
|
|
next character.
|
|
|
|
*/
|
2001-05-02 23:36:22 +00:00
|
|
|
for (;;) {
|
2001-06-20 18:14:51 +00:00
|
|
|
if (data->flags & UCOL_ITER_HASLEN) {
|
2001-04-23 01:53:49 +00:00
|
|
|
/*
|
2001-04-12 00:08:26 +00:00
|
|
|
Normal path for strings when length is specified.
|
|
|
|
Not in side buffer because it is always null terminated.
|
|
|
|
*/
|
|
|
|
if (data->pos <= data->string) {
|
|
|
|
/* End of the main source string */
|
|
|
|
return UCOL_NO_MORE_CES;
|
|
|
|
}
|
2001-06-20 18:14:51 +00:00
|
|
|
data->pos --;
|
|
|
|
ch = *data->pos;
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
// we are using an iterator to go back. Pray for us!
|
|
|
|
else if (data->flags & UCOL_USE_ITERATOR) {
|
|
|
|
UChar32 iterCh = data->iterator->previous(data->iterator);
|
|
|
|
if(iterCh == U_SENTINEL) {
|
|
|
|
return UCOL_NO_MORE_CES;
|
|
|
|
} else {
|
|
|
|
ch = (UChar)iterCh;
|
|
|
|
}
|
|
|
|
}
|
2001-04-12 00:08:26 +00:00
|
|
|
else {
|
2001-06-20 18:14:51 +00:00
|
|
|
data->pos --;
|
|
|
|
ch = *data->pos;
|
2001-04-12 00:08:26 +00:00
|
|
|
/* we are in the side buffer. */
|
2001-06-20 18:14:51 +00:00
|
|
|
if (ch == 0) {
|
2001-04-23 01:53:49 +00:00
|
|
|
/*
|
|
|
|
At the start of the normalize side buffer.
|
2001-04-12 00:08:26 +00:00
|
|
|
Go back to string.
|
2001-04-12 18:25:07 +00:00
|
|
|
Because pointer points to the last accessed character,
|
|
|
|
hence we have to increment it by one here.
|
2001-04-12 00:08:26 +00:00
|
|
|
*/
|
2008-05-26 19:18:33 +00:00
|
|
|
data->flags = data->origFlags;
|
|
|
|
data->offsetRepeatValue = 0;
|
|
|
|
|
|
|
|
if (data->fcdPosition == NULL) {
|
2001-04-12 18:25:07 +00:00
|
|
|
data->pos = data->string;
|
|
|
|
return UCOL_NO_MORE_CES;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
data->pos = data->fcdPosition + 1;
|
|
|
|
}
|
2008-05-26 19:18:33 +00:00
|
|
|
|
|
|
|
continue;
|
2001-04-12 00:08:26 +00:00
|
|
|
}
|
|
|
|
}
|
2001-10-05 02:07:51 +00:00
|
|
|
|
|
|
|
if(data->flags&UCOL_HIRAGANA_Q) {
|
|
|
|
if(ch>=0x3040 && ch<=0x309f) {
|
|
|
|
data->flags |= UCOL_WAS_HIRAGANA;
|
|
|
|
} else {
|
|
|
|
data->flags &= ~UCOL_WAS_HIRAGANA;
|
|
|
|
}
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
/*
|
2003-12-03 00:45:12 +00:00
|
|
|
* got a character to determine if there's fcd and/or normalization
|
2001-06-20 18:14:51 +00:00
|
|
|
* stuff to do.
|
2001-04-12 00:08:26 +00:00
|
|
|
* if the current character is not fcd.
|
2001-04-25 23:28:55 +00:00
|
|
|
* if current character is at the start of the string
|
2001-04-12 00:08:26 +00:00
|
|
|
* Trailing combining class == 0.
|
|
|
|
* Note if pos is in the writablebuffer, norm is always 0
|
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
if (ch < ZERO_CC_LIMIT_ ||
|
2003-02-06 23:29:56 +00:00
|
|
|
// this should propel us out of the loop in the iterator case
|
2001-06-28 20:42:56 +00:00
|
|
|
(data->flags & UCOL_ITER_NORM) == 0 ||
|
2003-12-03 00:45:12 +00:00
|
|
|
(data->fcdPosition != NULL && data->fcdPosition <= data->pos)
|
2001-06-20 18:14:51 +00:00
|
|
|
|| data->string == data->pos) {
|
2001-04-12 00:08:26 +00:00
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
if (ch < NFC_ZERO_CC_BLOCK_LIMIT_) {
|
|
|
|
/* if next character is FCD */
|
|
|
|
if (data->pos == data->string) {
|
|
|
|
/* First char of string is always OK for FCD check */
|
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/* Not first char of string, do the FCD fast test */
|
|
|
|
if (*(data->pos - 1) < NFC_ZERO_CC_BLOCK_LIMIT_) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
/* Need a more complete FCD check and possible normalization. */
|
2001-04-20 22:29:53 +00:00
|
|
|
if (collPrevIterFCD(data)) {
|
|
|
|
collPrevIterNormalize(data);
|
|
|
|
}
|
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
if ((data->flags & UCOL_ITER_INNORMBUF) == 0) {
|
|
|
|
/* No normalization. Go ahead and process the char. */
|
|
|
|
break;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
Some normalization happened.
|
|
|
|
Next loop picks up a char from the normalization buffer.
|
2001-04-12 00:08:26 +00:00
|
|
|
*/
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
/* attempt to handle contractions, after removal of the backwards
|
2001-05-17 01:06:25 +00:00
|
|
|
contraction
|
|
|
|
*/
|
2001-06-28 20:42:56 +00:00
|
|
|
if (ucol_contractionEndCP(ch, coll) && !isAtStartPrevIterate(data)) {
|
2007-12-28 21:16:45 +00:00
|
|
|
result = ucol_prv_getSpecialPrevCE(coll, ch, UCOL_CONTRACTION, data, status);
|
2005-05-19 06:43:35 +00:00
|
|
|
} else {
|
2007-12-28 21:16:45 +00:00
|
|
|
if (ch <= 0xFF) {
|
|
|
|
result = coll->latinOneMapping[ch];
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
result = UTRIE_GET32_FROM_LEAD(&coll->mapping, ch);
|
2003-07-22 17:26:51 +00:00
|
|
|
}
|
2005-07-06 21:27:21 +00:00
|
|
|
if (result > UCOL_NOT_FOUND) {
|
2007-12-28 21:16:45 +00:00
|
|
|
result = ucol_prv_getSpecialPrevCE(coll, ch, result, data, status);
|
|
|
|
}
|
|
|
|
if (result == UCOL_NOT_FOUND) { // Not found in master list
|
|
|
|
if (!isAtStartPrevIterate(data) &&
|
|
|
|
ucol_contractionEndCP(ch, data->coll))
|
|
|
|
{
|
|
|
|
result = UCOL_CONTRACTION;
|
|
|
|
} else {
|
|
|
|
if(coll->UCA) {
|
|
|
|
result = UTRIE_GET32_FROM_LEAD(&coll->UCA->mapping, ch);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (result > UCOL_NOT_FOUND) {
|
|
|
|
if(coll->UCA) {
|
|
|
|
result = ucol_prv_getSpecialPrevCE(coll->UCA, ch, result, data, status);
|
|
|
|
}
|
|
|
|
}
|
2001-04-18 19:31:05 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2005-07-06 21:27:21 +00:00
|
|
|
if(result == UCOL_NOT_FOUND) {
|
2007-12-28 21:16:45 +00:00
|
|
|
result = getPrevImplicit(ch, data);
|
2005-07-06 21:27:21 +00:00
|
|
|
}
|
2005-05-19 06:43:35 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2001-04-12 00:08:26 +00:00
|
|
|
return result;
|
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
|
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
/* ucol_getPrevCE, out-of-line version for use from other files. */
|
2007-06-03 06:08:46 +00:00
|
|
|
U_CFUNC uint32_t U_EXPORT2
|
2001-11-21 01:08:55 +00:00
|
|
|
ucol_getPrevCE(const UCollator *coll, collIterate *data,
|
2001-04-18 19:31:05 +00:00
|
|
|
UErrorCode *status) {
|
|
|
|
return ucol_IGetPrevCE(coll, data, status);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2001-03-09 00:50:37 +00:00
|
|
|
/* this should be connected to special Jamo handling */
|
2007-06-03 06:08:46 +00:00
|
|
|
U_CFUNC uint32_t U_EXPORT2
|
2001-11-21 01:08:55 +00:00
|
|
|
ucol_getFirstCE(const UCollator *coll, UChar u, UErrorCode *status) {
|
2007-12-28 20:08:41 +00:00
|
|
|
collIterate colIt;
|
|
|
|
uint32_t order;
|
|
|
|
IInit_collIterate(coll, &u, 1, &colIt);
|
|
|
|
order = ucol_IGetNextCE(coll, &colIt, status);
|
|
|
|
/*UCOL_GETNEXTCE(order, coll, colIt, status);*/
|
|
|
|
return order;
|
2001-03-09 00:50:37 +00:00
|
|
|
}
|
|
|
|
|
2001-04-24 03:18:54 +00:00
|
|
|
/**
|
2001-05-11 01:13:08 +00:00
|
|
|
* Inserts the argument character into the end of the buffer pushing back the
|
2001-04-24 03:18:54 +00:00
|
|
|
* null terminator.
|
2001-04-28 01:22:25 +00:00
|
|
|
* @param data collIterate struct data
|
|
|
|
* @param pNull pointer to the null termination
|
2001-04-24 03:18:54 +00:00
|
|
|
* @param ch character to be appended
|
2001-04-25 23:28:55 +00:00
|
|
|
* @return the position of the new addition
|
2001-04-24 03:18:54 +00:00
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-11 01:13:08 +00:00
|
|
|
inline UChar * insertBufferEnd(collIterate *data, UChar *pNull, UChar ch)
|
2001-04-24 03:18:54 +00:00
|
|
|
{
|
2007-12-28 20:08:41 +00:00
|
|
|
uint32_t size = data->writableBufSize;
|
|
|
|
UChar *newbuffer;
|
|
|
|
static const uint32_t INCSIZE = 5;
|
2001-04-24 03:18:54 +00:00
|
|
|
|
2001-04-28 01:22:25 +00:00
|
|
|
if ((data->writableBuffer + size) > (pNull + 1)) {
|
|
|
|
*pNull = ch;
|
|
|
|
*(pNull + 1) = 0;
|
|
|
|
return pNull;
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
2001-04-24 03:18:54 +00:00
|
|
|
buffer will always be null terminated at the end.
|
|
|
|
giving extra space since it is likely that more characters will be added.
|
|
|
|
*/
|
2007-12-28 20:08:41 +00:00
|
|
|
size += INCSIZE;
|
2001-04-24 03:18:54 +00:00
|
|
|
newbuffer = (UChar *)uprv_malloc(sizeof(UChar) * size);
|
2002-07-20 06:00:04 +00:00
|
|
|
if(newbuffer != NULL) { // something wrong, but no status
|
2007-12-28 20:08:41 +00:00
|
|
|
uprv_memcpy(newbuffer, data->writableBuffer,
|
|
|
|
data->writableBufSize * sizeof(UChar));
|
2001-04-24 03:18:54 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
freeHeapWritableBuffer(data);
|
|
|
|
data->writableBufSize = size;
|
|
|
|
data->writableBuffer = newbuffer;
|
2001-04-25 23:28:55 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
newbuffer = newbuffer + data->writableBufSize;
|
|
|
|
*newbuffer = ch;
|
|
|
|
*(newbuffer + 1) = 0;
|
2002-07-20 06:00:04 +00:00
|
|
|
}
|
2001-04-25 23:28:55 +00:00
|
|
|
return newbuffer;
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
|
|
|
|
2001-05-10 17:49:24 +00:00
|
|
|
/**
|
2001-05-11 01:13:08 +00:00
|
|
|
* Inserts the argument string into the end of the buffer pushing back the
|
2001-05-10 17:49:24 +00:00
|
|
|
* null terminator.
|
|
|
|
* @param data collIterate struct data
|
|
|
|
* @param pNull pointer to the null termination
|
|
|
|
* @param string to be appended
|
|
|
|
* @param length of the string to be appended
|
|
|
|
* @return the position of the new addition
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-10 17:49:24 +00:00
|
|
|
inline UChar * insertBufferEnd(collIterate *data, UChar *pNull, UChar *str,
|
2001-06-03 23:40:41 +00:00
|
|
|
int32_t length)
|
2001-05-10 17:49:24 +00:00
|
|
|
{
|
|
|
|
uint32_t size = pNull - data->writableBuffer;
|
|
|
|
UChar *newbuffer;
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-05-10 17:49:24 +00:00
|
|
|
if (data->writableBuffer + data->writableBufSize > pNull + length + 1) {
|
|
|
|
uprv_memcpy(pNull, str, length * sizeof(UChar));
|
|
|
|
*(pNull + length) = 0;
|
|
|
|
return pNull;
|
|
|
|
}
|
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
2001-05-10 17:49:24 +00:00
|
|
|
buffer will always be null terminated at the end.
|
|
|
|
giving extra space since it is likely that more characters will be added.
|
|
|
|
*/
|
|
|
|
newbuffer = (UChar *)uprv_malloc(sizeof(UChar) * (size + length + 1));
|
2002-07-20 06:00:04 +00:00
|
|
|
if(newbuffer != NULL) {
|
|
|
|
uprv_memcpy(newbuffer, data->writableBuffer, size * sizeof(UChar));
|
|
|
|
uprv_memcpy(newbuffer + size, str, length * sizeof(UChar));
|
2001-05-10 17:49:24 +00:00
|
|
|
|
2002-07-20 06:00:04 +00:00
|
|
|
freeHeapWritableBuffer(data);
|
|
|
|
data->writableBufSize = size + length + 1;
|
|
|
|
data->writableBuffer = newbuffer;
|
|
|
|
}
|
2001-05-10 17:49:24 +00:00
|
|
|
|
|
|
|
return newbuffer;
|
|
|
|
}
|
|
|
|
|
2001-04-24 03:18:54 +00:00
|
|
|
/**
|
|
|
|
* Special normalization function for contraction in the forwards iterator.
|
|
|
|
* This normalization sequence will place the current character at source->pos
|
|
|
|
* and its following normalized sequence into the buffer.
|
2001-05-11 01:13:08 +00:00
|
|
|
* The fcd position, pos will be changed.
|
2001-04-24 03:18:54 +00:00
|
|
|
* pos will now point to positions in the buffer.
|
|
|
|
* Flags will be changed accordingly.
|
|
|
|
* @param data collation iterator data
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-04-24 03:18:54 +00:00
|
|
|
inline void normalizeNextContraction(collIterate *data)
|
2001-05-11 01:13:08 +00:00
|
|
|
{
|
2001-04-24 03:18:54 +00:00
|
|
|
UChar *buffer = data->writableBuffer;
|
|
|
|
uint32_t buffersize = data->writableBufSize;
|
|
|
|
uint32_t strsize;
|
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
2001-04-25 23:28:55 +00:00
|
|
|
/* because the pointer points to the next character */
|
2001-05-11 01:13:08 +00:00
|
|
|
UChar *pStart = data->pos - 1;
|
2001-04-24 03:18:54 +00:00
|
|
|
UChar *pEnd;
|
|
|
|
uint32_t normLen;
|
|
|
|
UChar *pStartNorm;
|
|
|
|
|
2001-04-25 23:28:55 +00:00
|
|
|
if ((data->flags & UCOL_ITER_INNORMBUF) == 0) {
|
2001-04-24 03:18:54 +00:00
|
|
|
*data->writableBuffer = *(pStart - 1);
|
|
|
|
strsize = 1;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
strsize = u_strlen(data->writableBuffer);
|
|
|
|
}
|
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
pEnd = data->fcdPosition;
|
|
|
|
|
|
|
|
normLen = unorm_normalize(pStart, pEnd - pStart, UNORM_NFD, 0, buffer, 0,
|
2001-04-24 03:18:54 +00:00
|
|
|
&status);
|
|
|
|
|
|
|
|
if (buffersize <= normLen + strsize) {
|
|
|
|
uint32_t size = strsize + normLen + 1;
|
|
|
|
UChar *temp = (UChar *)uprv_malloc(size * sizeof(UChar));
|
2002-07-20 06:00:04 +00:00
|
|
|
if(temp != NULL) {
|
|
|
|
uprv_memcpy(temp, buffer, sizeof(UChar) * strsize);
|
|
|
|
freeHeapWritableBuffer(data);
|
|
|
|
data->writableBuffer = temp;
|
|
|
|
data->writableBufSize = size;
|
|
|
|
data->flags |= UCOL_ITER_ALLOCATED;
|
2008-01-16 20:12:48 +00:00
|
|
|
} else {
|
2008-02-13 09:35:50 +00:00
|
|
|
return; // Avoid writing past bound of buffer->writableBuffer.
|
2002-07-20 06:00:04 +00:00
|
|
|
}
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
status = U_ZERO_ERROR;
|
|
|
|
pStartNorm = buffer + strsize;
|
|
|
|
/* null-termination will be added here */
|
2001-05-11 01:13:08 +00:00
|
|
|
unorm_normalize(pStart, pEnd - pStart, UNORM_NFD, 0, pStartNorm,
|
2001-04-25 23:28:55 +00:00
|
|
|
normLen + 1, &status);
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-04-24 03:18:54 +00:00
|
|
|
data->pos = data->writableBuffer + strsize;
|
|
|
|
data->origFlags = data->flags;
|
|
|
|
data->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
data->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Contraction character management function that returns the next character
|
|
|
|
* for the forwards iterator.
|
2001-05-11 01:13:08 +00:00
|
|
|
* Does nothing if the next character is in buffer and not the first character
|
2001-04-24 03:18:54 +00:00
|
|
|
* in it.
|
|
|
|
* Else it checks next character in data string to see if it is normalizable.
|
|
|
|
* If it is not, the character is simply copied into the buffer, else
|
2001-05-11 01:13:08 +00:00
|
|
|
* the whole normalized substring is copied into the buffer, including the
|
2001-04-24 03:18:54 +00:00
|
|
|
* current character.
|
|
|
|
* @param data collation element iterator data
|
|
|
|
* @return next character
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-11 01:13:08 +00:00
|
|
|
inline UChar getNextNormalizedChar(collIterate *data)
|
2001-04-24 03:18:54 +00:00
|
|
|
{
|
2001-04-25 23:28:55 +00:00
|
|
|
UChar nextch;
|
2001-04-24 03:18:54 +00:00
|
|
|
UChar ch;
|
2003-01-20 07:43:32 +00:00
|
|
|
// Here we need to add the iterator code. One problem is the way
|
|
|
|
// end of string is handled. If we just return next char, it could
|
|
|
|
// be the sentinel. Most of the cases already check for this, but we
|
|
|
|
// need to be sure.
|
2001-06-06 23:26:50 +00:00
|
|
|
if ((data->flags & (UCOL_ITER_NORM | UCOL_ITER_INNORMBUF)) == 0 ) {
|
|
|
|
/* if no normalization and not in buffer. */
|
2003-01-20 07:43:32 +00:00
|
|
|
if(data->flags & UCOL_USE_ITERATOR) {
|
|
|
|
return (UChar)data->iterator->next(data->iterator);
|
|
|
|
} else {
|
2001-06-06 23:26:50 +00:00
|
|
|
return *(data->pos ++);
|
2003-01-20 07:43:32 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2003-01-23 01:52:34 +00:00
|
|
|
//if (data->flags & UCOL_ITER_NORM && data->flags & UCOL_USE_ITERATOR) {
|
|
|
|
//normalizeIterator(data);
|
|
|
|
//}
|
2001-06-06 23:26:50 +00:00
|
|
|
|
2001-05-10 17:49:24 +00:00
|
|
|
UChar *pEndWritableBuffer = NULL;
|
2001-06-06 23:26:50 +00:00
|
|
|
UBool innormbuf = (UBool)(data->flags & UCOL_ITER_INNORMBUF);
|
|
|
|
if ((innormbuf && *data->pos != 0) ||
|
2001-05-11 01:13:08 +00:00
|
|
|
(data->fcdPosition != NULL && !innormbuf &&
|
2001-05-10 17:49:24 +00:00
|
|
|
data->pos < data->fcdPosition)) {
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
2001-04-24 03:18:54 +00:00
|
|
|
if next character is in normalized buffer, no further normalization
|
|
|
|
is required
|
|
|
|
*/
|
|
|
|
return *(data->pos ++);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (data->flags & UCOL_ITER_HASLEN) {
|
|
|
|
/* in data string */
|
|
|
|
if (data->pos + 1 == data->endp) {
|
2001-04-25 23:28:55 +00:00
|
|
|
return *(data->pos ++);
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
2001-04-26 01:15:34 +00:00
|
|
|
if (innormbuf) {
|
2003-12-03 00:45:12 +00:00
|
|
|
// inside the normalization buffer, but at the end
|
|
|
|
// (since we encountered zero). This means, in the
|
|
|
|
// case we're using char iterator, that we need to
|
|
|
|
// do another round of normalization.
|
2003-01-23 01:52:34 +00:00
|
|
|
//if(data->origFlags & UCOL_USE_ITERATOR) {
|
2003-01-20 07:43:32 +00:00
|
|
|
// we need to restore original flags,
|
|
|
|
// otherwise, we'll lose them
|
2003-01-23 01:52:34 +00:00
|
|
|
//data->flags = data->origFlags;
|
|
|
|
//normalizeIterator(data);
|
|
|
|
//return *(data->pos++);
|
|
|
|
//} else {
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
|
|
|
in writable buffer, at this point fcdPosition can not be
|
2001-04-24 03:18:54 +00:00
|
|
|
pointing to the end of the data string. see contracting tag.
|
|
|
|
*/
|
2003-02-20 08:18:37 +00:00
|
|
|
if(data->fcdPosition) {
|
2001-04-25 23:28:55 +00:00
|
|
|
if (*(data->fcdPosition + 1) == 0 ||
|
|
|
|
data->fcdPosition + 1 == data->endp) {
|
|
|
|
/* at the end of the string, dump it into the normalizer */
|
2001-05-11 01:13:08 +00:00
|
|
|
data->pos = insertBufferEnd(data, data->pos,
|
2001-04-28 01:22:25 +00:00
|
|
|
*(data->fcdPosition)) + 1;
|
2008-01-16 20:12:48 +00:00
|
|
|
// Check if data->pos received a null pointer
|
|
|
|
if (data->pos == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
return (UChar)-1; // Return to indicate error.
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2001-04-25 23:28:55 +00:00
|
|
|
return *(data->fcdPosition ++);
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
2001-05-10 17:49:24 +00:00
|
|
|
pEndWritableBuffer = data->pos;
|
2001-04-24 03:18:54 +00:00
|
|
|
data->pos = data->fcdPosition;
|
2003-02-20 08:18:37 +00:00
|
|
|
} else if(data->origFlags & UCOL_USE_ITERATOR) {
|
|
|
|
// if we are here, we're using a normalizing iterator.
|
|
|
|
// we should just continue further.
|
|
|
|
data->flags = data->origFlags;
|
|
|
|
data->pos = NULL;
|
|
|
|
return (UChar)data->iterator->next(data->iterator);
|
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
//}
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
|
|
|
else {
|
|
|
|
if (*(data->pos + 1) == 0) {
|
|
|
|
return *(data->pos ++);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2001-04-25 23:28:55 +00:00
|
|
|
ch = *data->pos ++;
|
|
|
|
nextch = *data->pos;
|
2001-05-11 01:13:08 +00:00
|
|
|
|
|
|
|
/*
|
2001-04-24 03:18:54 +00:00
|
|
|
* if the current character is not fcd.
|
|
|
|
* Trailing combining class == 0.
|
|
|
|
*/
|
2001-05-11 01:13:08 +00:00
|
|
|
if ((data->fcdPosition == NULL || data->fcdPosition < data->pos) &&
|
2001-04-24 03:18:54 +00:00
|
|
|
(nextch >= NFC_ZERO_CC_BLOCK_LIMIT_ ||
|
2001-04-25 23:28:55 +00:00
|
|
|
ch >= NFC_ZERO_CC_BLOCK_LIMIT_)) {
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
|
|
|
Need a more complete FCD check and possible normalization.
|
|
|
|
normalize substring will be appended to buffer
|
2001-04-24 03:18:54 +00:00
|
|
|
*/
|
2001-04-25 23:28:55 +00:00
|
|
|
if (collIterFCD(data)) {
|
2001-04-24 03:18:54 +00:00
|
|
|
normalizeNextContraction(data);
|
2001-04-25 23:28:55 +00:00
|
|
|
return *(data->pos ++);
|
|
|
|
}
|
2001-05-10 17:49:24 +00:00
|
|
|
else if (innormbuf) {
|
2001-05-11 01:13:08 +00:00
|
|
|
/* fcdposition shifted even when there's no normalization, if we
|
|
|
|
don't input the rest into this, we'll get the wrong position when
|
2001-05-10 17:49:24 +00:00
|
|
|
we reach the end of the writableBuffer */
|
2001-06-03 23:40:41 +00:00
|
|
|
int32_t length = data->fcdPosition - data->pos + 1;
|
2001-05-11 01:13:08 +00:00
|
|
|
data->pos = insertBufferEnd(data, pEndWritableBuffer,
|
2001-05-10 17:49:24 +00:00
|
|
|
data->pos - 1, length);
|
2008-01-16 20:12:48 +00:00
|
|
|
// Check if data->pos received a null pointer
|
|
|
|
if (data->pos == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
return (UChar)-1; // Return to indicate error.
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2001-05-10 17:49:24 +00:00
|
|
|
return *(data->pos ++);
|
|
|
|
}
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-04-26 01:15:34 +00:00
|
|
|
if (innormbuf) {
|
2001-05-11 01:13:08 +00:00
|
|
|
/*
|
|
|
|
no normalization is to be done hence only one character will be
|
2001-04-24 03:18:54 +00:00
|
|
|
appended to the buffer.
|
|
|
|
*/
|
2001-05-10 17:49:24 +00:00
|
|
|
data->pos = insertBufferEnd(data, pEndWritableBuffer, ch) + 1;
|
2008-01-16 20:12:48 +00:00
|
|
|
// Check if data->pos received a null pointer
|
|
|
|
if (data->pos == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
return (UChar)-1; // Return to indicate error.
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2001-04-24 03:18:54 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-04-24 03:18:54 +00:00
|
|
|
/* points back to the pos in string */
|
|
|
|
return ch;
|
|
|
|
}
|
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
|
|
|
|
|
2001-06-06 23:26:50 +00:00
|
|
|
/**
|
|
|
|
* Function to copy the buffer into writableBuffer and sets the fcd position to
|
2001-05-14 18:57:54 +00:00
|
|
|
* the correct position
|
|
|
|
* @param source data string source
|
|
|
|
* @param buffer character buffer
|
|
|
|
* @param tempdb current position in buffer that has been used up
|
|
|
|
*/
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-06-06 23:26:50 +00:00
|
|
|
inline void setDiscontiguosAttribute(collIterate *source, UChar *buffer,
|
2001-05-14 18:57:54 +00:00
|
|
|
UChar *tempdb)
|
|
|
|
{
|
2001-06-06 23:26:50 +00:00
|
|
|
/* okay confusing part here. to ensure that the skipped characters are
|
|
|
|
considered later, we need to place it in the appropriate position in the
|
|
|
|
normalization buffer and reassign the pos pointer. simple case if pos
|
|
|
|
reside in string, simply copy to normalization buffer and
|
|
|
|
fcdposition = pos, pos = start of normalization buffer. if pos in
|
|
|
|
normalization buffer, we'll insert the copy infront of pos and point pos
|
|
|
|
to the start of the normalization buffer. why am i doing these copies?
|
2001-10-20 01:09:31 +00:00
|
|
|
well, so that the whole chunk of codes in the getNextCE, ucol_prv_getSpecialCE does
|
2001-05-14 18:57:54 +00:00
|
|
|
not require any changes, which be really painful. */
|
|
|
|
uint32_t length = u_strlen(buffer);;
|
|
|
|
if (source->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
u_strcpy(tempdb, source->pos);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
source->fcdPosition = source->pos;
|
|
|
|
source->origFlags = source->flags;
|
|
|
|
source->flags |= UCOL_ITER_INNORMBUF;
|
2003-02-14 07:46:20 +00:00
|
|
|
source->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN | UCOL_USE_ITERATOR);
|
2001-05-14 18:57:54 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (length >= source->writableBufSize) {
|
|
|
|
freeHeapWritableBuffer(source);
|
|
|
|
source->writableBuffer =
|
|
|
|
(UChar *)uprv_malloc((length + 1) * sizeof(UChar));
|
2002-07-20 06:00:04 +00:00
|
|
|
if(source->writableBuffer == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
source->writableBufSize = 0; // Reset size
|
|
|
|
return;
|
2002-07-20 06:00:04 +00:00
|
|
|
}
|
2001-05-14 18:57:54 +00:00
|
|
|
source->writableBufSize = length;
|
|
|
|
}
|
|
|
|
|
|
|
|
u_strcpy(source->writableBuffer, buffer);
|
|
|
|
source->pos = source->writableBuffer;
|
|
|
|
}
|
|
|
|
|
2001-05-10 17:49:24 +00:00
|
|
|
/**
|
|
|
|
* Function to get the discontiguos collation element within the source.
|
|
|
|
* Note this function will set the position to the appropriate places.
|
|
|
|
* @param coll current collator used
|
|
|
|
* @param source data string source
|
|
|
|
* @param constart index to the start character in the contraction table
|
|
|
|
* @return discontiguos collation element offset
|
|
|
|
*/
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2001-06-06 23:26:50 +00:00
|
|
|
uint32_t getDiscontiguous(const UCollator *coll, collIterate *source,
|
2001-05-10 17:49:24 +00:00
|
|
|
const UChar *constart)
|
|
|
|
{
|
|
|
|
/* source->pos currently points to the second combining character after
|
|
|
|
the start character */
|
|
|
|
UChar *temppos = source->pos;
|
2002-07-27 05:16:44 +00:00
|
|
|
UChar buffer[4*UCOL_MAX_BUFFER];
|
2001-05-10 17:49:24 +00:00
|
|
|
UChar *tempdb = buffer;
|
|
|
|
const UChar *tempconstart = constart;
|
|
|
|
uint8_t tempflags = source->flags;
|
2001-05-14 18:57:54 +00:00
|
|
|
UBool multicontraction = FALSE;
|
|
|
|
UChar *tempbufferpos = 0;
|
2003-02-14 07:46:20 +00:00
|
|
|
collIterateState discState;
|
|
|
|
|
|
|
|
backupState(source, &discState);
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2003-02-06 23:29:56 +00:00
|
|
|
//*tempdb = *(source->pos - 1);
|
2006-12-19 06:06:30 +00:00
|
|
|
*tempdb = peekCharacter(source, -1);
|
|
|
|
tempdb++;
|
|
|
|
for (;;) {
|
2001-05-10 17:49:24 +00:00
|
|
|
UChar *UCharOffset;
|
|
|
|
UChar schar,
|
|
|
|
tchar;
|
|
|
|
uint32_t result;
|
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
if (((source->flags & UCOL_ITER_HASLEN) && source->pos >= source->endp)
|
2003-02-06 23:29:56 +00:00
|
|
|
|| (peekCharacter(source, 0) == 0 &&
|
|
|
|
//|| (*source->pos == 0 &&
|
2001-05-10 17:49:24 +00:00
|
|
|
((source->flags & UCOL_ITER_INNORMBUF) == 0 ||
|
2001-05-11 01:13:08 +00:00
|
|
|
source->fcdPosition == NULL ||
|
2001-05-10 17:49:24 +00:00
|
|
|
source->fcdPosition == source->endp ||
|
|
|
|
*(source->fcdPosition) == 0 ||
|
|
|
|
u_getCombiningClass(*(source->fcdPosition)) == 0)) ||
|
2001-05-11 01:13:08 +00:00
|
|
|
/* end of string in null terminated string or stopped by a
|
|
|
|
null character, note fcd does not always point to a base
|
2001-05-10 17:49:24 +00:00
|
|
|
character after the discontiguos change */
|
2003-02-06 23:29:56 +00:00
|
|
|
u_getCombiningClass(peekCharacter(source, 0)) == 0) {
|
|
|
|
//u_getCombiningClass(*(source->pos)) == 0) {
|
2001-05-12 00:12:55 +00:00
|
|
|
//constart = (UChar *)coll->image + getContractOffset(CE);
|
2001-05-14 18:57:54 +00:00
|
|
|
if (multicontraction) {
|
|
|
|
*tempbufferpos = 0;
|
|
|
|
source->pos = temppos - 1;
|
|
|
|
setDiscontiguosAttribute(source, buffer, tempdb);
|
2001-06-06 23:26:50 +00:00
|
|
|
return *(coll->contractionCEs +
|
2001-05-14 18:57:54 +00:00
|
|
|
(tempconstart - coll->contractionIndex));
|
|
|
|
}
|
2001-05-12 00:12:55 +00:00
|
|
|
constart = tempconstart;
|
2001-05-10 17:49:24 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
UCharOffset = (UChar *)(tempconstart + 1); /* skip the backward offset*/
|
|
|
|
schar = getNextNormalizedChar(source);
|
2001-05-11 01:13:08 +00:00
|
|
|
|
|
|
|
while (schar > (tchar = *UCharOffset)) {
|
2001-05-10 17:49:24 +00:00
|
|
|
UCharOffset++;
|
|
|
|
}
|
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
if (schar != tchar) {
|
|
|
|
/* not the correct codepoint. we stuff the current codepoint into
|
2001-05-10 17:49:24 +00:00
|
|
|
the discontiguos buffer and try the next character */
|
|
|
|
*tempdb = schar;
|
|
|
|
tempdb ++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
else {
|
2001-05-11 01:13:08 +00:00
|
|
|
if (u_getCombiningClass(schar) ==
|
2003-02-06 23:29:56 +00:00
|
|
|
u_getCombiningClass(peekCharacter(source, -2))) {
|
|
|
|
//u_getCombiningClass(*(source->pos - 2))) {
|
2001-05-10 17:49:24 +00:00
|
|
|
*tempdb = schar;
|
|
|
|
tempdb ++;
|
|
|
|
continue;
|
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
result = *(coll->contractionCEs +
|
2001-05-10 17:49:24 +00:00
|
|
|
(UCharOffset - coll->contractionIndex));
|
|
|
|
}
|
|
|
|
*tempdb = 0;
|
|
|
|
|
|
|
|
if (result == UCOL_NOT_FOUND) {
|
|
|
|
break;
|
2001-05-11 01:13:08 +00:00
|
|
|
} else if (isContraction(result)) {
|
2001-05-10 17:49:24 +00:00
|
|
|
/* this is a multi-contraction*/
|
|
|
|
tempconstart = (UChar *)coll->image + getContractOffset(result);
|
2001-05-14 18:57:54 +00:00
|
|
|
if (*(coll->contractionCEs + (constart - coll->contractionIndex))
|
|
|
|
!= UCOL_NOT_FOUND) {
|
|
|
|
multicontraction = TRUE;
|
|
|
|
temppos = source->pos + 1;
|
|
|
|
tempbufferpos = buffer + u_strlen(buffer);
|
2001-05-10 17:49:24 +00:00
|
|
|
}
|
2001-05-14 18:57:54 +00:00
|
|
|
} else {
|
|
|
|
setDiscontiguosAttribute(source, buffer, tempdb);
|
2001-05-10 17:49:24 +00:00
|
|
|
return result;
|
|
|
|
}
|
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
|
|
|
|
/* no problems simply reverting just like that,
|
|
|
|
if we are in string before getting into this function, points back to
|
2001-05-10 17:49:24 +00:00
|
|
|
string hence no problem.
|
2001-05-11 01:13:08 +00:00
|
|
|
if we are in normalization buffer before getting into this function,
|
|
|
|
since we'll never use another normalization within this function, we
|
|
|
|
know that fcdposition points to a base character. the normalization buffer
|
2001-05-10 17:49:24 +00:00
|
|
|
never change, hence this revert works. */
|
2003-02-14 07:46:20 +00:00
|
|
|
loadState(source, &discState, TRUE);
|
|
|
|
goBackOne(source);
|
|
|
|
|
|
|
|
//source->pos = temppos - 1;
|
2001-05-10 17:49:24 +00:00
|
|
|
source->flags = tempflags;
|
|
|
|
return *(coll->contractionCEs + (constart - coll->contractionIndex));
|
|
|
|
}
|
|
|
|
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2002-07-02 22:32:14 +00:00
|
|
|
inline UBool isNonChar(UChar32 cp) {
|
2007-12-28 20:08:41 +00:00
|
|
|
return (UBool)((cp & 0xFFFE) == 0xFFFE || (0xFDD0 <= cp && cp <= 0xFDEF) || (0xD800 <= cp && cp <= 0xDFFF));
|
2001-09-20 20:16:39 +00:00
|
|
|
}
|
2002-06-13 18:34:41 +00:00
|
|
|
|
|
|
|
/* now uses Mark's getImplicitPrimary code */
|
|
|
|
static
|
|
|
|
inline uint32_t getImplicit(UChar32 cp, collIterate *collationSource) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(isNonChar(cp)) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
uint32_t r = uprv_uca_getImplicitPrimary(cp);
|
|
|
|
*(collationSource->CEpos++) = ((r & 0x0000FFFF)<<16) | 0x000000C0;
|
2008-05-23 04:22:28 +00:00
|
|
|
collationSource->offsetRepeatCount += 1;
|
2007-12-26 19:27:43 +00:00
|
|
|
return (r & UCOL_PRIMARYMASK) | 0x00000505; // This was 'order'
|
2002-06-13 18:34:41 +00:00
|
|
|
}
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2003-05-07 05:33:50 +00:00
|
|
|
/**
|
|
|
|
* Inserts the argument character into the front of the buffer replacing the
|
|
|
|
* front null terminator.
|
|
|
|
* @param data collation element iterator data
|
|
|
|
* @param pNull pointer to the null terminator
|
|
|
|
* @param ch character to be appended
|
|
|
|
* @return positon of added character
|
|
|
|
*/
|
|
|
|
static
|
|
|
|
inline UChar * insertBufferFront(collIterate *data, UChar *pNull, UChar ch)
|
|
|
|
{
|
2007-12-28 20:08:41 +00:00
|
|
|
uint32_t size = data->writableBufSize;
|
|
|
|
UChar *end;
|
|
|
|
UChar *newbuffer;
|
|
|
|
static const uint32_t INCSIZE = 5;
|
2003-05-07 05:33:50 +00:00
|
|
|
|
|
|
|
if (pNull > data->writableBuffer + 1) {
|
|
|
|
*pNull = ch;
|
|
|
|
*(pNull - 1) = 0;
|
|
|
|
return pNull;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
buffer will always be null terminated infront.
|
|
|
|
giving extra space since it is likely that more characters will be added.
|
|
|
|
*/
|
2007-12-28 20:08:41 +00:00
|
|
|
size += INCSIZE;
|
2003-05-07 05:33:50 +00:00
|
|
|
newbuffer = (UChar *)uprv_malloc(sizeof(UChar) * size);
|
|
|
|
if(newbuffer == NULL) {
|
2007-12-28 20:08:41 +00:00
|
|
|
return NULL;
|
2003-05-07 05:33:50 +00:00
|
|
|
}
|
2007-12-28 20:08:41 +00:00
|
|
|
end = newbuffer + INCSIZE;
|
2003-05-07 05:33:50 +00:00
|
|
|
uprv_memcpy(end, data->writableBuffer,
|
|
|
|
data->writableBufSize * sizeof(UChar));
|
|
|
|
*end = ch;
|
|
|
|
*(end - 1) = 0;
|
|
|
|
|
|
|
|
freeHeapWritableBuffer(data);
|
|
|
|
|
|
|
|
data->writableBufSize = size;
|
|
|
|
data->writableBuffer = newbuffer;
|
|
|
|
return end;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Special normalization function for contraction in the previous iterator.
|
|
|
|
* This normalization sequence will place the current character at source->pos
|
|
|
|
* and its following normalized sequence into the buffer.
|
|
|
|
* The fcd position, pos will be changed.
|
|
|
|
* pos will now point to positions in the buffer.
|
|
|
|
* Flags will be changed accordingly.
|
|
|
|
* @param data collation iterator data
|
|
|
|
*/
|
|
|
|
static
|
2006-04-04 20:15:46 +00:00
|
|
|
inline void normalizePrevContraction(collIterate *data, UErrorCode *status)
|
2003-05-07 05:33:50 +00:00
|
|
|
{
|
|
|
|
uint32_t nulltermsize;
|
2006-04-04 20:15:46 +00:00
|
|
|
UErrorCode localstatus = U_ZERO_ERROR;
|
2003-05-07 05:33:50 +00:00
|
|
|
UChar *pEnd = data->pos + 1; /* End normalize + 1 */
|
|
|
|
UChar *pStart;
|
|
|
|
uint32_t normLen;
|
|
|
|
UChar *pStartNorm;
|
|
|
|
|
|
|
|
if (data->flags & UCOL_ITER_HASLEN) {
|
|
|
|
/*
|
|
|
|
normalization buffer not used yet, we'll pull down the next
|
|
|
|
character into the end of the buffer
|
|
|
|
*/
|
2008-02-13 09:35:50 +00:00
|
|
|
*(data->writableBuffer + (data->writableBufSize - 1)) = *(data->pos + 1);
|
|
|
|
nulltermsize = data->writableBufSize - 1;
|
2003-05-07 05:33:50 +00:00
|
|
|
}
|
|
|
|
else {
|
2008-02-13 09:35:50 +00:00
|
|
|
nulltermsize = data->writableBufSize;
|
|
|
|
UChar *temp = data->writableBuffer + (nulltermsize - 1);
|
2003-05-07 05:33:50 +00:00
|
|
|
while (*(temp --) != 0) {
|
|
|
|
nulltermsize --;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Start normalize */
|
|
|
|
if (data->fcdPosition == NULL) {
|
|
|
|
pStart = data->string;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
pStart = data->fcdPosition + 1;
|
|
|
|
}
|
|
|
|
|
2008-02-13 09:35:50 +00:00
|
|
|
normLen = unorm_normalize(pStart, pEnd - pStart, UNORM_NFD, 0, data->writableBuffer, 0,
|
2006-04-04 20:15:46 +00:00
|
|
|
&localstatus);
|
2003-05-07 05:33:50 +00:00
|
|
|
|
|
|
|
if (nulltermsize <= normLen) {
|
2008-02-13 09:35:50 +00:00
|
|
|
uint32_t size = data->writableBufSize - nulltermsize + normLen + 1;
|
2003-05-07 05:33:50 +00:00
|
|
|
UChar *temp = (UChar *)uprv_malloc(size * sizeof(UChar));
|
2006-04-04 20:15:46 +00:00
|
|
|
if (temp == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return;
|
2003-05-07 05:33:50 +00:00
|
|
|
}
|
2006-04-04 20:15:46 +00:00
|
|
|
nulltermsize = normLen + 1;
|
2008-02-13 09:35:50 +00:00
|
|
|
uprv_memcpy(temp + normLen, data->writableBuffer,
|
|
|
|
sizeof(UChar) * (data->writableBufSize - nulltermsize));
|
2006-04-04 20:15:46 +00:00
|
|
|
freeHeapWritableBuffer(data);
|
|
|
|
data->writableBuffer = temp;
|
|
|
|
data->writableBufSize = size;
|
2003-05-07 05:33:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
this puts the null termination infront of the normalized string instead
|
|
|
|
of the end
|
|
|
|
*/
|
2008-02-13 09:35:50 +00:00
|
|
|
pStartNorm = data->writableBuffer + (nulltermsize - normLen);
|
2003-05-07 05:33:50 +00:00
|
|
|
*(pStartNorm - 1) = 0;
|
|
|
|
unorm_normalize(pStart, pEnd - pStart, UNORM_NFD, 0, pStartNorm, normLen,
|
2006-04-04 20:15:46 +00:00
|
|
|
status);
|
2003-05-07 05:33:50 +00:00
|
|
|
|
|
|
|
data->pos = data->writableBuffer + nulltermsize;
|
|
|
|
data->origFlags = data->flags;
|
|
|
|
data->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
data->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Contraction character management function that returns the previous character
|
|
|
|
* for the backwards iterator.
|
|
|
|
* Does nothing if the previous character is in buffer and not the first
|
|
|
|
* character in it.
|
|
|
|
* Else it checks previous character in data string to see if it is
|
|
|
|
* normalizable.
|
|
|
|
* If it is not, the character is simply copied into the buffer, else
|
|
|
|
* the whole normalized substring is copied into the buffer, including the
|
|
|
|
* current character.
|
|
|
|
* @param data collation element iterator data
|
|
|
|
* @return previous character
|
|
|
|
*/
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2006-04-04 20:15:46 +00:00
|
|
|
inline UChar getPrevNormalizedChar(collIterate *data, UErrorCode *status)
|
2003-05-07 05:33:50 +00:00
|
|
|
{
|
|
|
|
UChar prevch;
|
|
|
|
UChar ch;
|
|
|
|
UChar *start;
|
|
|
|
UBool innormbuf = (UBool)(data->flags & UCOL_ITER_INNORMBUF);
|
|
|
|
UChar *pNull = NULL;
|
|
|
|
if ((data->flags & (UCOL_ITER_NORM | UCOL_ITER_INNORMBUF)) == 0 ||
|
|
|
|
(innormbuf && *(data->pos - 1) != 0)) {
|
|
|
|
/*
|
|
|
|
if no normalization.
|
|
|
|
if previous character is in normalized buffer, no further normalization
|
|
|
|
is required
|
|
|
|
*/
|
|
|
|
if(data->flags & UCOL_USE_ITERATOR) {
|
|
|
|
data->iterator->move(data->iterator, -1, UITER_CURRENT);
|
|
|
|
return (UChar)data->iterator->next(data->iterator);
|
|
|
|
} else {
|
|
|
|
return *(data->pos - 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
start = data->pos;
|
2008-04-04 22:47:43 +00:00
|
|
|
if ((data->fcdPosition==NULL)||(data->flags & UCOL_ITER_HASLEN)) {
|
2003-05-07 05:33:50 +00:00
|
|
|
/* in data string */
|
|
|
|
if ((start - 1) == data->string) {
|
|
|
|
return *(start - 1);
|
|
|
|
}
|
|
|
|
start --;
|
|
|
|
ch = *start;
|
|
|
|
prevch = *(start - 1);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/*
|
|
|
|
in writable buffer, at this point fcdPosition can not be NULL.
|
|
|
|
see contracting tag.
|
|
|
|
*/
|
|
|
|
if (data->fcdPosition == data->string) {
|
|
|
|
/* at the start of the string, just dump it into the normalizer */
|
|
|
|
insertBufferFront(data, data->pos - 1, *(data->fcdPosition));
|
|
|
|
data->fcdPosition = NULL;
|
|
|
|
return *(data->pos - 1);
|
|
|
|
}
|
|
|
|
pNull = data->pos - 1;
|
|
|
|
start = data->fcdPosition;
|
|
|
|
ch = *start;
|
|
|
|
prevch = *(start - 1);
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* if the current character is not fcd.
|
|
|
|
* Trailing combining class == 0.
|
|
|
|
*/
|
|
|
|
if (data->fcdPosition > start &&
|
|
|
|
(ch >= NFC_ZERO_CC_BLOCK_LIMIT_ || prevch >= NFC_ZERO_CC_BLOCK_LIMIT_))
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
Need a more complete FCD check and possible normalization.
|
|
|
|
normalize substring will be appended to buffer
|
|
|
|
*/
|
|
|
|
UChar *backuppos = data->pos;
|
|
|
|
data->pos = start;
|
|
|
|
if (collPrevIterFCD(data)) {
|
2006-04-04 20:15:46 +00:00
|
|
|
normalizePrevContraction(data, status);
|
2003-05-07 05:33:50 +00:00
|
|
|
return *(data->pos - 1);
|
|
|
|
}
|
|
|
|
data->pos = backuppos;
|
|
|
|
data->fcdPosition ++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (innormbuf) {
|
|
|
|
/*
|
|
|
|
no normalization is to be done hence only one character will be
|
|
|
|
appended to the buffer.
|
|
|
|
*/
|
|
|
|
insertBufferFront(data, pNull, ch);
|
|
|
|
data->fcdPosition --;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ch;
|
|
|
|
}
|
2001-10-15 00:09:23 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* This function handles the special CEs like contractions, expansions, surrogates, Thai */
|
2001-09-20 20:16:39 +00:00
|
|
|
/* It is called by getNextCE */
|
2001-10-20 01:09:31 +00:00
|
|
|
|
|
|
|
uint32_t ucol_prv_getSpecialCE(const UCollator *coll, UChar ch, uint32_t CE, collIterate *source, UErrorCode *status) {
|
2007-12-26 19:27:43 +00:00
|
|
|
collIterateState entryState;
|
|
|
|
backupState(source, &entryState);
|
|
|
|
UChar32 cp = ch;
|
2001-09-27 23:18:14 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
for (;;) {
|
|
|
|
// This loop will repeat only in the case of contractions, and only when a contraction
|
|
|
|
// is found and the first CE resulting from that contraction is itself a special
|
|
|
|
// (an expansion, for example.) All other special CE types are fully handled the
|
|
|
|
// first time through, and the loop exits.
|
|
|
|
|
|
|
|
const uint32_t *CEOffset = NULL;
|
|
|
|
switch(getCETag(CE)) {
|
|
|
|
case NOT_FOUND_TAG:
|
|
|
|
/* This one is not found, and we'll let somebody else bother about it... no more games */
|
|
|
|
return CE;
|
|
|
|
case SPEC_PROC_TAG:
|
|
|
|
{
|
|
|
|
// Special processing is getting a CE that is preceded by a certain prefix
|
|
|
|
// Currently this is only needed for optimizing Japanese length and iteration marks.
|
|
|
|
// When we encouter a special processing tag, we go backwards and try to see if
|
|
|
|
// we have a match.
|
|
|
|
// Contraction tables are used - so the whole process is not unlike contraction.
|
|
|
|
// prefix data is stored backwards in the table.
|
|
|
|
const UChar *UCharOffset;
|
|
|
|
UChar schar, tchar;
|
|
|
|
collIterateState prefixState;
|
|
|
|
backupState(source, &prefixState);
|
|
|
|
loadState(source, &entryState, TRUE);
|
|
|
|
goBackOne(source); // We want to look at the point where we entered - actually one
|
|
|
|
// before that...
|
|
|
|
|
|
|
|
for(;;) {
|
|
|
|
// This loop will run once per source string character, for as long as we
|
|
|
|
// are matching a potential contraction sequence
|
|
|
|
|
|
|
|
// First we position ourselves at the begining of contraction sequence
|
|
|
|
const UChar *ContractionStart = UCharOffset = (UChar *)coll->image+getContractOffset(CE);
|
|
|
|
if (collIter_bos(source)) {
|
|
|
|
CE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
schar = getPrevNormalizedChar(source, status);
|
|
|
|
goBackOne(source);
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(schar > (tchar = *UCharOffset)) { /* since the contraction codepoints should be ordered, we skip all that are smaller */
|
|
|
|
UCharOffset++;
|
|
|
|
}
|
2001-03-03 09:27:42 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (schar == tchar) {
|
|
|
|
// Found the source string char in the table.
|
|
|
|
// Pick up the corresponding CE from the table.
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(UCharOffset - coll->contractionIndex));
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
// Source string char was not in the table.
|
|
|
|
// We have not found the prefix.
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(ContractionStart - coll->contractionIndex));
|
|
|
|
}
|
2001-06-06 23:26:50 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(!isPrefix(CE)) {
|
|
|
|
// The source string char was in the contraction table, and the corresponding
|
|
|
|
// CE is not a prefix CE. We found the prefix, break
|
|
|
|
// out of loop, this CE will end up being returned. This is the normal
|
|
|
|
// way out of prefix handling when the source actually contained
|
|
|
|
// the prefix.
|
|
|
|
break;
|
2005-05-19 06:43:35 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
if(CE != UCOL_NOT_FOUND) { // we found something and we can merilly continue
|
|
|
|
loadState(source, &prefixState, TRUE);
|
|
|
|
if(source->origFlags & UCOL_USE_ITERATOR) {
|
|
|
|
source->flags = source->origFlags;
|
2005-05-19 06:43:35 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else { // prefix search was a failure, we have to backup all the way to the start
|
|
|
|
loadState(source, &entryState, TRUE);
|
2001-06-06 23:26:50 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
2001-05-16 18:23:26 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
case CONTRACTION_TAG:
|
|
|
|
{
|
|
|
|
/* This should handle contractions */
|
|
|
|
collIterateState state;
|
|
|
|
backupState(source, &state);
|
|
|
|
uint32_t firstCE = *(coll->contractionCEs + ((UChar *)coll->image+getContractOffset(CE) - coll->contractionIndex)); //UCOL_NOT_FOUND;
|
|
|
|
const UChar *UCharOffset;
|
|
|
|
UChar schar, tchar;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
/* This loop will run once per source string character, for as long as we */
|
|
|
|
/* are matching a potential contraction sequence */
|
|
|
|
|
|
|
|
/* First we position ourselves at the begining of contraction sequence */
|
|
|
|
const UChar *ContractionStart = UCharOffset = (UChar *)coll->image+getContractOffset(CE);
|
|
|
|
|
|
|
|
if (collIter_eos(source)) {
|
|
|
|
// Ran off the end of the source string.
|
|
|
|
CE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
|
|
|
|
// So we'll pick whatever we have at the point...
|
|
|
|
if (CE == UCOL_NOT_FOUND) {
|
|
|
|
// back up the source over all the chars we scanned going into this contraction.
|
|
|
|
CE = firstCE;
|
|
|
|
loadState(source, &state, TRUE);
|
|
|
|
if(source->origFlags & UCOL_USE_ITERATOR) {
|
|
|
|
source->flags = source->origFlags;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
2001-06-06 23:26:50 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint8_t maxCC = (uint8_t)(*(UCharOffset)&0xFF); /*get the discontiguos stuff */ /* skip the backward offset, see above */
|
|
|
|
uint8_t allSame = (uint8_t)(*(UCharOffset++)>>8);
|
2003-07-25 05:31:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
schar = getNextNormalizedChar(source);
|
|
|
|
while(schar > (tchar = *UCharOffset)) { /* since the contraction codepoints should be ordered, we skip all that are smaller */
|
|
|
|
UCharOffset++;
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (schar == tchar) {
|
|
|
|
// Found the source string char in the contraction table.
|
|
|
|
// Pick up the corresponding CE from the table.
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(UCharOffset - coll->contractionIndex));
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
// Source string char was not in contraction table.
|
|
|
|
// Unless we have a discontiguous contraction, we have finished
|
|
|
|
// with this contraction.
|
|
|
|
// in order to do the proper detection, we
|
|
|
|
// need to see if we're dealing with a supplementary
|
|
|
|
/* We test whether the next two char are surrogate pairs.
|
|
|
|
* This test is done if the iterator is not NULL.
|
|
|
|
* If there is no surrogate pair, the iterator
|
|
|
|
* goes back one if needed. */
|
|
|
|
UChar32 miss = schar;
|
|
|
|
if (source->iterator) {
|
|
|
|
UChar32 surrNextChar; /* the next char in the iteration to test */
|
|
|
|
int32_t prevPos; /* holds the previous position before move forward of the source iterator */
|
2007-12-31 20:25:48 +00:00
|
|
|
if(U16_IS_LEAD(schar) && source->iterator->hasNext(source->iterator)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
prevPos = source->iterator->index;
|
|
|
|
surrNextChar = getNextNormalizedChar(source);
|
|
|
|
if (U16_IS_TRAIL(surrNextChar)) {
|
|
|
|
miss = U16_GET_SUPPLEMENTARY(schar, surrNextChar);
|
|
|
|
} else if (prevPos < source->iterator->index){
|
|
|
|
goBackOne(source);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else if (U16_IS_LEAD(schar)) {
|
|
|
|
miss = U16_GET_SUPPLEMENTARY(schar, getNextNormalizedChar(source));
|
|
|
|
}
|
2003-07-25 05:31:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint8_t sCC;
|
|
|
|
if (miss < 0x300 ||
|
|
|
|
maxCC == 0 ||
|
|
|
|
(sCC = i_getCombiningClass(miss, coll)) == 0 ||
|
|
|
|
sCC>maxCC ||
|
|
|
|
(allSame != 0 && sCC == maxCC) ||
|
|
|
|
collIter_eos(source))
|
|
|
|
{
|
|
|
|
// Contraction can not be discontiguous.
|
|
|
|
goBackOne(source); // back up the source string by one,
|
|
|
|
// because the character we just looked at was
|
|
|
|
// not part of the contraction. */
|
|
|
|
if(U_IS_SUPPLEMENTARY(miss)) {
|
|
|
|
goBackOne(source);
|
|
|
|
}
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(ContractionStart - coll->contractionIndex));
|
|
|
|
} else {
|
|
|
|
//
|
|
|
|
// Contraction is possibly discontiguous.
|
|
|
|
// Scan more of source string looking for a match
|
|
|
|
//
|
|
|
|
UChar tempchar;
|
|
|
|
/* find the next character if schar is not a base character
|
|
|
|
and we are not yet at the end of the string */
|
|
|
|
tempchar = getNextNormalizedChar(source);
|
|
|
|
// probably need another supplementary thingie here
|
|
|
|
goBackOne(source);
|
|
|
|
if (i_getCombiningClass(tempchar, coll) == 0) {
|
|
|
|
goBackOne(source);
|
|
|
|
if(U_IS_SUPPLEMENTARY(miss)) {
|
|
|
|
goBackOne(source);
|
|
|
|
}
|
|
|
|
/* Spit out the last char of the string, wasn't tasty enough */
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(ContractionStart - coll->contractionIndex));
|
|
|
|
} else {
|
|
|
|
CE = getDiscontiguous(coll, source, ContractionStart);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} // else after if(schar == tchar)
|
|
|
|
|
|
|
|
if(CE == UCOL_NOT_FOUND) {
|
|
|
|
/* The Source string did not match the contraction that we were checking. */
|
|
|
|
/* Back up the source position to undo the effects of having partially */
|
|
|
|
/* scanned through what ultimately proved to not be a contraction. */
|
|
|
|
loadState(source, &state, TRUE);
|
|
|
|
CE = firstCE;
|
|
|
|
break;
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(!isContraction(CE)) {
|
|
|
|
// The source string char was in the contraction table, and the corresponding
|
|
|
|
// CE is not a contraction CE. We completed the contraction, break
|
|
|
|
// out of loop, this CE will end up being returned. This is the normal
|
|
|
|
// way out of contraction handling when the source actually contained
|
|
|
|
// the contraction.
|
|
|
|
break;
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2003-07-25 05:31:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// The source string char was in the contraction table, and the corresponding
|
|
|
|
// CE is IS a contraction CE. We will continue looping to check the source
|
|
|
|
// string for the remaining chars in the contraction.
|
|
|
|
uint32_t tempCE = *(coll->contractionCEs + (ContractionStart - coll->contractionIndex));
|
|
|
|
if(tempCE != UCOL_NOT_FOUND) {
|
|
|
|
// We have scanned a a section of source string for which there is a
|
|
|
|
// CE from the contraction table. Remember the CE and scan position, so
|
|
|
|
// that we can return to this point if further scanning fails to
|
|
|
|
// match a longer contraction sequence.
|
|
|
|
firstCE = tempCE;
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
goBackOne(source);
|
|
|
|
backupState(source, &state);
|
|
|
|
getNextNormalizedChar(source);
|
|
|
|
|
|
|
|
// Another way to do this is:
|
|
|
|
//collIterateState tempState;
|
|
|
|
//backupState(source, &tempState);
|
|
|
|
//goBackOne(source);
|
|
|
|
//backupState(source, &state);
|
|
|
|
//loadState(source, &tempState, TRUE);
|
|
|
|
|
|
|
|
// The problem is that for incomplete contractions we have to remember the previous
|
|
|
|
// position. Before, the only thing I needed to do was state.pos--;
|
|
|
|
// After iterator introduction and especially after introduction of normalizing
|
|
|
|
// iterators, it became much more difficult to decrease the saved state.
|
|
|
|
// I'm not yet sure which of the two methods above is faster.
|
|
|
|
}
|
|
|
|
} // for(;;)
|
|
|
|
break;
|
|
|
|
} // case CONTRACTION_TAG:
|
|
|
|
case LONG_PRIMARY_TAG:
|
|
|
|
{
|
|
|
|
*(source->CEpos++) = ((CE & 0xFF)<<24)|UCOL_CONTINUATION_MARKER;
|
|
|
|
CE = ((CE & 0xFFFF00) << 8) | (UCOL_BYTE_COMMON << 8) | UCOL_BYTE_COMMON;
|
2008-05-23 04:22:28 +00:00
|
|
|
source->offsetRepeatCount += 1;
|
2007-12-26 19:27:43 +00:00
|
|
|
return CE;
|
2003-07-25 05:31:54 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
case EXPANSION_TAG:
|
2006-07-28 21:47:47 +00:00
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
/* This should handle expansion. */
|
|
|
|
/* NOTE: we can encounter both continuations and expansions in an expansion! */
|
|
|
|
/* I have to decide where continuations are going to be dealt with */
|
|
|
|
uint32_t size;
|
|
|
|
uint32_t i; /* general counter */
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
CEOffset = (uint32_t *)coll->image+getExpansionOffset(CE); /* find the offset to expansion table */
|
|
|
|
size = getExpansionCount(CE);
|
|
|
|
CE = *CEOffset++;
|
2008-05-23 04:22:28 +00:00
|
|
|
//source->offsetRepeatCount = -1;
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(size != 0) { /* if there are less than 16 elements in expansion, we don't terminate */
|
|
|
|
for(i = 1; i<size; i++) {
|
|
|
|
*(source->CEpos++) = *CEOffset++;
|
2008-05-23 04:22:28 +00:00
|
|
|
source->offsetRepeatCount += 1;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
} else { /* else, we do */
|
|
|
|
while(*CEOffset != 0) {
|
|
|
|
*(source->CEpos++) = *CEOffset++;
|
2008-05-23 04:22:28 +00:00
|
|
|
source->offsetRepeatCount += 1;
|
2007-12-14 09:23:07 +00:00
|
|
|
}
|
2006-07-28 21:47:47 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return CE;
|
2006-07-28 21:47:47 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
case DIGIT_TAG:
|
|
|
|
{
|
2004-11-11 23:34:58 +00:00
|
|
|
/*
|
2007-12-26 19:27:43 +00:00
|
|
|
We do a check to see if we want to collate digits as numbers; if so we generate
|
|
|
|
a custom collation key. Otherwise we pull out the value stored in the expansion table.
|
|
|
|
*/
|
|
|
|
//uint32_t size;
|
|
|
|
uint32_t i; /* general counter */
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (source->coll->numericCollation == UCOL_ON){
|
|
|
|
collIterateState digitState = {0,0,0,0,0,0,0,0};
|
|
|
|
UChar32 char32 = 0;
|
|
|
|
|
|
|
|
uint32_t digIndx = 0;
|
|
|
|
uint32_t endIndex = 0;
|
|
|
|
uint32_t trailingZeroIndex = 0;
|
|
|
|
|
|
|
|
uint32_t primWeight = 0;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
int32_t digVal = 0;
|
|
|
|
uint8_t collateVal = 0;
|
|
|
|
|
|
|
|
UBool nonZeroValReached = FALSE;
|
|
|
|
|
|
|
|
uint8_t *numTempBuf;
|
|
|
|
uint8_t stackNumTempBuf[UCOL_MAX_BUFFER]; // I just need a temporary place to store my generated CEs.
|
|
|
|
uint32_t numTempBufSize = UCOL_MAX_BUFFER;
|
|
|
|
|
|
|
|
numTempBuf = stackNumTempBuf;
|
|
|
|
/*
|
|
|
|
We parse the source string until we hit a char that's NOT a digit.
|
|
|
|
Use this u_charDigitValue. This might be slow because we have to
|
|
|
|
handle surrogates...
|
|
|
|
*/
|
|
|
|
/*
|
|
|
|
if (U16_IS_LEAD(ch)){
|
|
|
|
if (!collIter_eos(source)) {
|
|
|
|
backupState(source, &digitState);
|
|
|
|
UChar trail = getNextNormalizedChar(source);
|
|
|
|
if(U16_IS_TRAIL(trail)) {
|
|
|
|
char32 = U16_GET_SUPPLEMENTARY(ch, trail);
|
|
|
|
} else {
|
|
|
|
loadState(source, &digitState, TRUE);
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
char32 = ch;
|
|
|
|
}
|
2003-07-25 05:31:54 +00:00
|
|
|
} else {
|
|
|
|
char32 = ch;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
digVal = u_charDigitValue(char32);
|
|
|
|
*/
|
|
|
|
digVal = u_charDigitValue(cp); // if we have arrived here, we have
|
|
|
|
// already processed possible supplementaries that trigered the digit tag -
|
|
|
|
// all supplementaries are marked in the UCA.
|
|
|
|
/*
|
|
|
|
We pad a zero in front of the first element anyways. This takes
|
|
|
|
care of the (probably) most common case where people are sorting things followed
|
|
|
|
by a single digit
|
|
|
|
*/
|
|
|
|
digIndx++;
|
|
|
|
for(;;){
|
|
|
|
// Make sure we have enough space.
|
|
|
|
if (digIndx >= ((numTempBufSize - 2) * 2) + 1)
|
|
|
|
{
|
|
|
|
numTempBufSize *= 2;
|
|
|
|
if (numTempBuf == stackNumTempBuf){
|
|
|
|
numTempBuf = (uint8_t *)uprv_malloc(sizeof(uint8_t) * numTempBufSize);
|
2008-01-16 20:12:48 +00:00
|
|
|
// Null pointer check
|
|
|
|
if (numTempBuf == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return 0;
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
uprv_memcpy(numTempBuf, stackNumTempBuf, UCOL_MAX_BUFFER);
|
|
|
|
} else {
|
|
|
|
uint8_t *temp = (uint8_t *)uprv_realloc(numTempBuf, numTempBufSize);
|
|
|
|
if (temp == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
/* The original contents weren't freed. */
|
|
|
|
uprv_free(temp);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
numTempBuf = temp;
|
|
|
|
}
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// Skipping over leading zeroes.
|
|
|
|
if (digVal != 0) {
|
|
|
|
nonZeroValReached = TRUE;
|
|
|
|
}
|
|
|
|
if (nonZeroValReached) {
|
|
|
|
/*
|
|
|
|
We parse the digit string into base 100 numbers (this fits into a byte).
|
|
|
|
We only add to the buffer in twos, thus if we are parsing an odd character,
|
|
|
|
that serves as the 'tens' digit while the if we are parsing an even one, that
|
|
|
|
is the 'ones' digit. We dumped the parsed base 100 value (collateVal) into
|
|
|
|
a buffer. We multiply each collateVal by 2 (to give us room) and add 5 (to avoid
|
|
|
|
overlapping magic CE byte values). The last byte we subtract 1 to ensure it is less
|
|
|
|
than all the other bytes.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (digIndx % 2 == 1){
|
|
|
|
collateVal += (uint8_t)digVal;
|
|
|
|
|
|
|
|
// We don't enter the low-order-digit case unless we've already seen
|
|
|
|
// the high order, or for the first digit, which is always non-zero.
|
|
|
|
if (collateVal != 0)
|
|
|
|
trailingZeroIndex = 0;
|
|
|
|
|
|
|
|
numTempBuf[(digIndx/2) + 2] = collateVal*2 + 6;
|
|
|
|
collateVal = 0;
|
|
|
|
}
|
|
|
|
else{
|
|
|
|
// We drop the collation value into the buffer so if we need to do
|
|
|
|
// a "front patch" we don't have to check to see if we're hitting the
|
|
|
|
// last element.
|
|
|
|
collateVal = (uint8_t)(digVal * 10);
|
|
|
|
|
|
|
|
// Check for trailing zeroes.
|
|
|
|
if (collateVal == 0)
|
|
|
|
{
|
|
|
|
if (!trailingZeroIndex)
|
|
|
|
trailingZeroIndex = (digIndx/2) + 2;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
trailingZeroIndex = 0;
|
|
|
|
|
|
|
|
numTempBuf[(digIndx/2) + 2] = collateVal*2 + 6;
|
|
|
|
}
|
|
|
|
digIndx++;
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// Get next character.
|
|
|
|
if (!collIter_eos(source)){
|
|
|
|
ch = getNextNormalizedChar(source);
|
|
|
|
if (U16_IS_LEAD(ch)){
|
|
|
|
if (!collIter_eos(source)) {
|
|
|
|
backupState(source, &digitState);
|
|
|
|
UChar trail = getNextNormalizedChar(source);
|
|
|
|
if(U16_IS_TRAIL(trail)) {
|
|
|
|
char32 = U16_GET_SUPPLEMENTARY(ch, trail);
|
|
|
|
} else {
|
|
|
|
loadState(source, &digitState, TRUE);
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((digVal = u_charDigitValue(char32)) == -1){
|
|
|
|
// Resetting position to point to the next unprocessed char. We
|
|
|
|
// overshot it when doing our test/set for numbers.
|
|
|
|
if (char32 > 0xFFFF) { // For surrogates.
|
|
|
|
loadState(source, &digitState, TRUE);
|
|
|
|
//goBackOne(source);
|
|
|
|
}
|
|
|
|
goBackOne(source);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (nonZeroValReached == FALSE){
|
|
|
|
digIndx = 2;
|
|
|
|
numTempBuf[2] = 6;
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
endIndex = trailingZeroIndex ? trailingZeroIndex : ((digIndx/2) + 2) ;
|
|
|
|
if (digIndx % 2 != 0){
|
|
|
|
/*
|
|
|
|
We missed a value. Since digIndx isn't even, stuck too many values into the buffer (this is what
|
|
|
|
we get for padding the first byte with a zero). "Front-patch" now by pushing all nybbles forward.
|
|
|
|
Doing it this way ensures that at least 50% of the time (statistically speaking) we'll only be doing a
|
|
|
|
single pass and optimizes for strings with single digits. I'm just assuming that's the more common case.
|
|
|
|
*/
|
|
|
|
|
|
|
|
for(i = 2; i < endIndex; i++){
|
|
|
|
numTempBuf[i] = (((((numTempBuf[i] - 6)/2) % 10) * 10) +
|
|
|
|
(((numTempBuf[i+1])-6)/2) / 10) * 2 + 6;
|
|
|
|
}
|
|
|
|
--digIndx;
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// Subtract one off of the last byte.
|
|
|
|
numTempBuf[endIndex-1] -= 1;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/*
|
|
|
|
We want to skip over the first two slots in the buffer. The first slot
|
|
|
|
is reserved for the header byte UCOL_CODAN_PLACEHOLDER. The second slot is for the
|
|
|
|
sign/exponent byte: 0x80 + (decimalPos/2) & 7f.
|
|
|
|
*/
|
|
|
|
numTempBuf[0] = UCOL_CODAN_PLACEHOLDER;
|
|
|
|
numTempBuf[1] = (uint8_t)(0x80 + ((digIndx/2) & 0x7F));
|
|
|
|
|
|
|
|
// Now transfer the collation key to our collIterate struct.
|
|
|
|
// The total size for our collation key is endIndx bumped up to the next largest even value divided by two.
|
|
|
|
//size = ((endIndex+1) & ~1)/2;
|
|
|
|
CE = (((numTempBuf[0] << 8) | numTempBuf[1]) << UCOL_PRIMARYORDERSHIFT) | //Primary weight
|
|
|
|
(UCOL_BYTE_COMMON << UCOL_SECONDARYORDERSHIFT) | // Secondary weight
|
|
|
|
UCOL_BYTE_COMMON; // Tertiary weight.
|
|
|
|
i = 2; // Reset the index into the buffer.
|
|
|
|
while(i < endIndex)
|
|
|
|
{
|
|
|
|
primWeight = numTempBuf[i++] << 8;
|
|
|
|
if ( i < endIndex)
|
|
|
|
primWeight |= numTempBuf[i++];
|
|
|
|
*(source->CEpos++) = (primWeight << UCOL_PRIMARYORDERSHIFT) | UCOL_CONTINUATION_MARKER;
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (numTempBuf != stackNumTempBuf)
|
|
|
|
uprv_free(numTempBuf);
|
|
|
|
} else {
|
|
|
|
// no numeric mode, we'll just switch to whatever we stashed and continue
|
|
|
|
CEOffset = (uint32_t *)coll->image+getExpansionOffset(CE); /* find the offset to expansion table */
|
|
|
|
CE = *CEOffset++;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return CE;
|
|
|
|
}
|
|
|
|
/* various implicits optimization */
|
|
|
|
case IMPLICIT_TAG: /* everything that is not defined otherwise */
|
|
|
|
/* UCA is filled with these. Tailorings are NOT_FOUND */
|
|
|
|
return getImplicit(cp, source);
|
2007-12-28 20:08:41 +00:00
|
|
|
case CJK_IMPLICIT_TAG: /* 0x3400-0x4DB5, 0x4E00-0x9FA5, 0xF900-0xFA2D*/
|
|
|
|
// TODO: remove CJK_IMPLICIT_TAG completely - handled by the getImplicit
|
|
|
|
return getImplicit(cp, source);
|
2007-12-26 19:27:43 +00:00
|
|
|
case HANGUL_SYLLABLE_TAG: /* AC00-D7AF*/
|
|
|
|
{
|
2007-12-28 20:08:41 +00:00
|
|
|
static const uint32_t
|
2007-12-26 19:27:43 +00:00
|
|
|
SBase = 0xAC00, LBase = 0x1100, VBase = 0x1161, TBase = 0x11A7;
|
|
|
|
//const uint32_t LCount = 19;
|
2007-12-28 20:08:41 +00:00
|
|
|
static const uint32_t VCount = 21;
|
|
|
|
static const uint32_t TCount = 28;
|
2007-12-26 19:27:43 +00:00
|
|
|
//const uint32_t NCount = VCount * TCount; // 588
|
|
|
|
//const uint32_t SCount = LCount * NCount; // 11172
|
|
|
|
uint32_t L = ch - SBase;
|
|
|
|
|
|
|
|
// divide into pieces
|
|
|
|
|
|
|
|
uint32_t T = L % TCount; // we do it in this order since some compilers can do % and / in one operation
|
|
|
|
L /= TCount;
|
|
|
|
uint32_t V = L % VCount;
|
|
|
|
L /= VCount;
|
|
|
|
|
|
|
|
// offset them
|
|
|
|
|
|
|
|
L += LBase;
|
|
|
|
V += VBase;
|
|
|
|
T += TBase;
|
|
|
|
|
|
|
|
// return the first CE, but first put the rest into the expansion buffer
|
|
|
|
if (!source->coll->image->jamoSpecial) { // FAST PATH
|
|
|
|
|
|
|
|
*(source->CEpos++) = UTRIE_GET32_FROM_LEAD(&coll->mapping, V);
|
|
|
|
if (T != TBase) {
|
|
|
|
*(source->CEpos++) = UTRIE_GET32_FROM_LEAD(&coll->mapping, T);
|
|
|
|
}
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return UTRIE_GET32_FROM_LEAD(&coll->mapping, L);
|
|
|
|
|
|
|
|
} else { // Jamo is Special
|
|
|
|
// Since Hanguls pass the FCD check, it is
|
|
|
|
// guaranteed that we won't be in
|
|
|
|
// the normalization buffer if something like this happens
|
|
|
|
// However, if we are using a uchar iterator and normalization
|
|
|
|
// is ON, the Hangul that lead us here is going to be in that
|
|
|
|
// normalization buffer. Here we want to restore the uchar
|
|
|
|
// iterator state and pull out of the normalization buffer
|
|
|
|
if(source->iterator != NULL && source->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
source->flags = source->origFlags; // restore the iterator
|
|
|
|
source->pos = NULL;
|
|
|
|
}
|
|
|
|
// Move Jamos into normalization buffer
|
|
|
|
source->writableBuffer[0] = (UChar)L;
|
|
|
|
source->writableBuffer[1] = (UChar)V;
|
|
|
|
if (T != TBase) {
|
|
|
|
source->writableBuffer[2] = (UChar)T;
|
|
|
|
source->writableBuffer[3] = 0;
|
|
|
|
} else {
|
|
|
|
source->writableBuffer[2] = 0;
|
|
|
|
}
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
source->fcdPosition = source->pos; // Indicate where to continue in main input string
|
|
|
|
// after exhausting the writableBuffer
|
|
|
|
source->pos = source->writableBuffer;
|
|
|
|
source->origFlags = source->flags;
|
|
|
|
source->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
source->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN);
|
2001-09-27 17:17:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return(UCOL_IGNORABLE);
|
|
|
|
}
|
|
|
|
}
|
2007-12-28 20:08:41 +00:00
|
|
|
case SURROGATE_TAG:
|
|
|
|
/* we encountered a leading surrogate. We shall get the CE by using the following code unit */
|
|
|
|
/* two things can happen here: next code point can be a trailing surrogate - we will use it */
|
|
|
|
/* to retrieve the CE, or it is not a trailing surrogate (or the string is done). In that case */
|
|
|
|
/* we return 0 (completely ignorable - per UCA specification */
|
|
|
|
{
|
|
|
|
UChar trail;
|
|
|
|
collIterateState state;
|
|
|
|
backupState(source, &state);
|
|
|
|
if (collIter_eos(source) || !(U16_IS_TRAIL((trail = getNextNormalizedChar(source))))) {
|
|
|
|
// we chould have stepped one char forward and it might have turned that it
|
|
|
|
// was not a trail surrogate. In that case, we have to backup.
|
|
|
|
loadState(source, &state, TRUE);
|
|
|
|
return 0;
|
|
|
|
} else {
|
|
|
|
/* TODO: CE contain the data from the previous CE + the mask. It should at least be unmasked */
|
|
|
|
CE = UTRIE_GET32_FROM_OFFSET_TRAIL(&coll->mapping, CE&0xFFFFFF, trail);
|
|
|
|
if(CE == UCOL_NOT_FOUND) { // there are tailored surrogates in this block, but not this one.
|
|
|
|
// We need to backup
|
|
|
|
loadState(source, &state, TRUE);
|
|
|
|
return CE;
|
|
|
|
}
|
|
|
|
// calculate the supplementary code point value, if surrogate was not tailored
|
|
|
|
cp = ((((uint32_t)ch)<<10UL)+(trail)-(((uint32_t)0xd800<<10UL)+0xdc00-0x10000));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case LEAD_SURROGATE_TAG: /* D800-DBFF*/
|
|
|
|
UChar nextChar;
|
|
|
|
if( source->flags & UCOL_USE_ITERATOR) {
|
|
|
|
if(U_IS_TRAIL(nextChar = (UChar)source->iterator->current(source->iterator))) {
|
|
|
|
cp = U16_GET_SUPPLEMENTARY(ch, nextChar);
|
|
|
|
source->iterator->next(source->iterator);
|
|
|
|
return getImplicit(cp, source);
|
|
|
|
} else {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
} else if((((source->flags & UCOL_ITER_HASLEN) == 0 ) || (source->pos<source->endp)) &&
|
|
|
|
U_IS_TRAIL((nextChar=*source->pos))) {
|
|
|
|
cp = U16_GET_SUPPLEMENTARY(ch, nextChar);
|
|
|
|
source->pos++;
|
|
|
|
return getImplicit(cp, source);
|
|
|
|
} else {
|
|
|
|
return 0; /* completely ignorable */
|
|
|
|
}
|
|
|
|
case TRAIL_SURROGATE_TAG: /* DC00-DFFF*/
|
|
|
|
return 0; /* broken surrogate sequence */
|
2007-12-26 19:27:43 +00:00
|
|
|
case CHARSET_TAG:
|
|
|
|
/* not yet implemented */
|
|
|
|
/* probably after 1.8 */
|
|
|
|
return UCOL_NOT_FOUND;
|
|
|
|
default:
|
|
|
|
*status = U_INTERNAL_PROGRAM_ERROR;
|
|
|
|
CE=0;
|
|
|
|
break;
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
|
|
|
if (CE <= UCOL_NOT_FOUND) break;
|
|
|
|
}
|
|
|
|
return CE;
|
1999-08-16 21:50:52 +00:00
|
|
|
}
|
|
|
|
|
2002-06-13 18:34:41 +00:00
|
|
|
|
|
|
|
/* now uses Mark's getImplicitPrimary code */
|
|
|
|
static
|
|
|
|
inline uint32_t getPrevImplicit(UChar32 cp, collIterate *collationSource) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(isNonChar(cp)) {
|
|
|
|
return 0;
|
|
|
|
}
|
2002-06-13 18:34:41 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t r = uprv_uca_getImplicitPrimary(cp);
|
2002-06-13 18:34:41 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
*(collationSource->CEpos++) = (r & UCOL_PRIMARYMASK) | 0x00000505;
|
|
|
|
collationSource->toReturn = collationSource->CEpos;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (collationSource->offsetBuffer == NULL) {
|
|
|
|
collationSource->offsetBufferSize = UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
collationSource->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * UCOL_EXPAND_CE_BUFFER_SIZE);
|
|
|
|
collationSource->offsetStore = collationSource->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
|
|
|
// **** doesn't work if using iterator ****
|
|
|
|
if (collationSource->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
collationSource->offsetRepeatCount = 1;
|
|
|
|
} else {
|
|
|
|
int32_t firstOffset = (int32_t)(collationSource->pos - collationSource->string);
|
|
|
|
|
|
|
|
*(collationSource->offsetStore++) = firstOffset;
|
|
|
|
*(collationSource->offsetStore++) = firstOffset + 1;
|
|
|
|
|
|
|
|
collationSource->offsetReturn = collationSource->offsetStore - 1;
|
|
|
|
*(collationSource->offsetBuffer) = firstOffset;
|
|
|
|
if (collationSource->offsetReturn == collationSource->offsetBuffer) {
|
|
|
|
collationSource->offsetStore = collationSource->offsetBuffer;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return ((r & 0x0000FFFF)<<16) | 0x000000C0;
|
2002-06-13 18:34:41 +00:00
|
|
|
}
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2001-05-11 01:13:08 +00:00
|
|
|
/**
|
2002-06-13 18:34:41 +00:00
|
|
|
* This function handles the special CEs like contractions, expansions,
|
|
|
|
* surrogates, Thai.
|
|
|
|
* It is called by both getPrevCE
|
|
|
|
*/
|
2001-10-20 01:09:31 +00:00
|
|
|
uint32_t ucol_prv_getSpecialPrevCE(const UCollator *coll, UChar ch, uint32_t CE,
|
2001-04-23 01:53:49 +00:00
|
|
|
collIterate *source,
|
|
|
|
UErrorCode *status)
|
2001-02-20 00:26:50 +00:00
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
const uint32_t *CEOffset = NULL;
|
|
|
|
UChar *UCharOffset = NULL;
|
|
|
|
UChar schar;
|
|
|
|
const UChar *constart = NULL;
|
|
|
|
uint32_t size;
|
|
|
|
UChar buffer[UCOL_MAX_BUFFER];
|
|
|
|
uint32_t *endCEBuffer;
|
|
|
|
UChar *strbuffer;
|
|
|
|
int32_t noChars = 0;
|
|
|
|
int32_t CECount = 0;
|
|
|
|
|
|
|
|
for(;;)
|
2001-02-20 00:26:50 +00:00
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
/* the only ces that loops are thai and contractions */
|
|
|
|
switch (getCETag(CE))
|
|
|
|
{
|
|
|
|
case NOT_FOUND_TAG: /* this tag always returns */
|
|
|
|
return CE;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case SPEC_PROC_TAG:
|
|
|
|
{
|
|
|
|
// Special processing is getting a CE that is preceded by a certain prefix
|
|
|
|
// Currently this is only needed for optimizing Japanese length and iteration marks.
|
|
|
|
// When we encouter a special processing tag, we go backwards and try to see if
|
|
|
|
// we have a match.
|
|
|
|
// Contraction tables are used - so the whole process is not unlike contraction.
|
|
|
|
// prefix data is stored backwards in the table.
|
|
|
|
const UChar *UCharOffset;
|
|
|
|
UChar schar, tchar;
|
|
|
|
collIterateState prefixState;
|
|
|
|
backupState(source, &prefixState);
|
|
|
|
for(;;) {
|
|
|
|
// This loop will run once per source string character, for as long as we
|
|
|
|
// are matching a potential contraction sequence
|
|
|
|
|
|
|
|
// First we position ourselves at the begining of contraction sequence
|
|
|
|
const UChar *ContractionStart = UCharOffset = (UChar *)coll->image+getContractOffset(CE);
|
|
|
|
|
|
|
|
if (collIter_bos(source)) {
|
|
|
|
CE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
schar = getPrevNormalizedChar(source, status);
|
|
|
|
goBackOne(source);
|
2001-10-02 01:25:25 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(schar > (tchar = *UCharOffset)) { /* since the contraction codepoints should be ordered, we skip all that are smaller */
|
|
|
|
UCharOffset++;
|
|
|
|
}
|
2001-10-02 01:25:25 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (schar == tchar) {
|
|
|
|
// Found the source string char in the table.
|
|
|
|
// Pick up the corresponding CE from the table.
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(UCharOffset - coll->contractionIndex));
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
// if there is a completely ignorable code point in the middle of
|
|
|
|
// a prefix, we need to act as if it's not there
|
|
|
|
// assumption: 'real' noncharacters (*fffe, *ffff, fdd0-fdef are set to zero)
|
|
|
|
// lone surrogates cannot be set to zero as it would break other processing
|
|
|
|
uint32_t isZeroCE = UTRIE_GET32_FROM_LEAD(&coll->mapping, schar);
|
|
|
|
// it's easy for BMP code points
|
|
|
|
if(isZeroCE == 0) {
|
|
|
|
continue;
|
|
|
|
} else if(U16_IS_TRAIL(schar) || U16_IS_LEAD(schar)) {
|
|
|
|
// for supplementary code points, we have to check the next one
|
|
|
|
// situations where we are going to ignore
|
|
|
|
// 1. beginning of the string: schar is a lone surrogate
|
|
|
|
// 2. schar is a lone surrogate
|
|
|
|
// 3. schar is a trail surrogate in a valid surrogate sequence
|
|
|
|
// that is explicitly set to zero.
|
|
|
|
if (!collIter_bos(source)) {
|
|
|
|
UChar lead;
|
|
|
|
if(U16_IS_LEAD(lead = getPrevNormalizedChar(source, status))) {
|
|
|
|
isZeroCE = UTRIE_GET32_FROM_LEAD(&coll->mapping, lead);
|
|
|
|
if(getCETag(isZeroCE) == SURROGATE_TAG) {
|
|
|
|
uint32_t finalCE = UTRIE_GET32_FROM_OFFSET_TRAIL(&coll->mapping, isZeroCE&0xFFFFFF, schar);
|
|
|
|
if(finalCE == 0) {
|
|
|
|
// this is a real, assigned completely ignorable code point
|
|
|
|
goBackOne(source);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// lone surrogate, completely ignorable
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// lone surrogate at the beggining, completely ignorable
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// Source string char was not in the table.
|
|
|
|
// We have not found the prefix.
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(ContractionStart - coll->contractionIndex));
|
2002-07-02 22:32:14 +00:00
|
|
|
}
|
2001-10-02 01:25:25 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(!isPrefix(CE)) {
|
|
|
|
// The source string char was in the contraction table, and the corresponding
|
|
|
|
// CE is not a prefix CE. We found the prefix, break
|
|
|
|
// out of loop, this CE will end up being returned. This is the normal
|
|
|
|
// way out of prefix handling when the source actually contained
|
|
|
|
// the prefix.
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
loadState(source, &prefixState, TRUE);
|
|
|
|
break;
|
|
|
|
}
|
2001-10-02 01:25:25 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case CONTRACTION_TAG:
|
|
|
|
/* to ensure that the backwards and forwards iteration matches, we
|
|
|
|
take the current region of most possible match and pass it through
|
|
|
|
the forward iteration. this will ensure that the obstinate problem of
|
|
|
|
overlapping contractions will not occur.
|
|
|
|
*/
|
|
|
|
schar = peekCharacter(source, 0);
|
|
|
|
constart = (UChar *)coll->image + getContractOffset(CE);
|
|
|
|
if (isAtStartPrevIterate(source)
|
|
|
|
/* commented away contraction end checks after adding the checks
|
|
|
|
in getPrevCE */) {
|
|
|
|
/* start of string or this is not the end of any contraction */
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(constart - coll->contractionIndex));
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
strbuffer = buffer;
|
|
|
|
UCharOffset = strbuffer + (UCOL_MAX_BUFFER - 1);
|
|
|
|
*(UCharOffset --) = 0;
|
|
|
|
noChars = 0;
|
|
|
|
// have to swap thai characters
|
|
|
|
while (ucol_unsafeCP(schar, coll)) {
|
|
|
|
*(UCharOffset) = schar;
|
|
|
|
noChars++;
|
|
|
|
UCharOffset --;
|
|
|
|
schar = getPrevNormalizedChar(source, status);
|
|
|
|
goBackOne(source);
|
|
|
|
// TODO: when we exhaust the contraction buffer,
|
|
|
|
// it needs to get reallocated. The problem is
|
|
|
|
// that the size depends on the string which is
|
|
|
|
// not iterated over. However, since we're travelling
|
|
|
|
// backwards, we already had to set the iterator at
|
|
|
|
// the end - so we might as well know where we are?
|
|
|
|
if (UCharOffset + 1 == buffer) {
|
|
|
|
/* we have exhausted the buffer */
|
|
|
|
int32_t newsize = 0;
|
|
|
|
if(source->pos) { // actually dealing with a position
|
|
|
|
newsize = source->pos - source->string + 1;
|
|
|
|
} else { // iterator
|
|
|
|
newsize = 4 * UCOL_MAX_BUFFER;
|
|
|
|
}
|
|
|
|
strbuffer = (UChar *)uprv_malloc(sizeof(UChar) *
|
|
|
|
(newsize + UCOL_MAX_BUFFER));
|
|
|
|
/* test for NULL */
|
|
|
|
if (strbuffer == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return UCOL_NO_MORE_CES;
|
|
|
|
}
|
|
|
|
UCharOffset = strbuffer + newsize;
|
|
|
|
uprv_memcpy(UCharOffset, buffer,
|
|
|
|
UCOL_MAX_BUFFER * sizeof(UChar));
|
|
|
|
UCharOffset --;
|
|
|
|
}
|
|
|
|
if ((source->pos && (source->pos == source->string ||
|
|
|
|
((source->flags & UCOL_ITER_INNORMBUF) &&
|
|
|
|
*(source->pos - 1) == 0 && source->fcdPosition == NULL)))
|
|
|
|
|| (source->iterator && !source->iterator->hasPrevious(source->iterator))) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/* adds the initial base character to the string */
|
2001-05-10 17:49:24 +00:00
|
|
|
*(UCharOffset) = schar;
|
2002-07-02 22:32:14 +00:00
|
|
|
noChars++;
|
2007-12-26 19:27:43 +00:00
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
int32_t offsetBias;
|
|
|
|
|
|
|
|
#if 0
|
|
|
|
if (source->offsetReturn != NULL) {
|
|
|
|
source->offsetStore = source->offsetReturn - noChars;
|
|
|
|
}
|
|
|
|
|
|
|
|
// **** doesn't work if using iterator ****
|
|
|
|
if (source->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
if (source->fcdPosition == NULL) {
|
|
|
|
offsetBias = 0;
|
|
|
|
} else {
|
|
|
|
offsetBias = (int32_t)(source->fcdPosition - source->string);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
offsetBias = (int32_t)(source->pos - source->string);
|
|
|
|
}
|
|
|
|
|
|
|
|
#else
|
|
|
|
// **** doesn't work if using iterator ****
|
|
|
|
if (source->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
#if 1
|
|
|
|
offsetBias = -1;
|
|
|
|
#else
|
|
|
|
if (source->fcdPosition == NULL) {
|
|
|
|
offsetBias = 0;
|
|
|
|
} else {
|
|
|
|
offsetBias = (int32_t)(source->fcdPosition - source->string);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
} else {
|
|
|
|
offsetBias = (int32_t)(source->pos - source->string);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* a new collIterate is used to simplify things, since using the current
|
|
|
|
collIterate will mean that the forward and backwards iteration will
|
|
|
|
share and change the same buffers. we don't want to get into that. */
|
|
|
|
collIterate temp;
|
2008-05-23 04:22:28 +00:00
|
|
|
int32_t rawOffset;
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
//IInit_collIterate(coll, UCharOffset, -1, &temp);
|
|
|
|
IInit_collIterate(coll, UCharOffset, noChars, &temp);
|
|
|
|
temp.flags &= ~UCOL_ITER_NORM;
|
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
rawOffset = temp.pos - temp.string; // should always be zero?
|
2007-12-26 19:27:43 +00:00
|
|
|
CE = ucol_IGetNextCE(coll, &temp, status);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (source->extendCEs) {
|
|
|
|
endCEBuffer = source->extendCEs + source->extendCEsSize;
|
|
|
|
CECount = (source->CEpos - source->extendCEs)/sizeof(uint32_t);
|
|
|
|
} else {
|
|
|
|
endCEBuffer = source->CEs + UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
CECount = (source->CEpos - source->CEs)/sizeof(uint32_t);
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (source->offsetBuffer == NULL) {
|
|
|
|
source->offsetBufferSize = UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
source->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * UCOL_EXPAND_CE_BUFFER_SIZE);
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while (CE != UCOL_NO_MORE_CES) {
|
|
|
|
*(source->CEpos ++) = CE;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (offsetBias >= 0) {
|
|
|
|
*(source->offsetStore ++) = rawOffset + offsetBias;
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
CECount++;
|
|
|
|
if (source->CEpos == endCEBuffer) {
|
|
|
|
/* ran out of CE space, reallocate to new buffer.
|
|
|
|
If reallocation fails, reset pointers and bail out,
|
|
|
|
there's no guarantee of the right character position after
|
|
|
|
this bail*/
|
|
|
|
if (source->extendCEs == NULL) {
|
|
|
|
source->extendCEs = (uint32_t *)uprv_malloc(sizeof(uint32_t) *
|
|
|
|
(source->extendCEsSize =UCOL_EXPAND_CE_BUFFER_SIZE + UCOL_EXPAND_CE_BUFFER_EXTEND_SIZE));
|
|
|
|
if (source->extendCEs == NULL) {
|
2007-12-26 19:55:00 +00:00
|
|
|
// Handle error later.
|
2007-12-26 19:27:43 +00:00
|
|
|
CECount = -1;
|
|
|
|
} else {
|
|
|
|
source->extendCEs = (uint32_t *)uprv_memcpy(source->extendCEs, source->CEs, UCOL_EXPAND_CE_BUFFER_SIZE * sizeof(uint32_t));
|
|
|
|
}
|
|
|
|
} else {
|
2007-12-26 19:55:00 +00:00
|
|
|
uint32_t *tempBufCE = (uint32_t *)uprv_realloc(source->extendCEs,
|
2007-12-26 19:27:43 +00:00
|
|
|
sizeof(uint32_t) * (source->extendCEsSize += UCOL_EXPAND_CE_BUFFER_EXTEND_SIZE));
|
2007-12-26 19:55:00 +00:00
|
|
|
if (tempBufCE == NULL) {
|
|
|
|
// Handle error later.
|
2007-12-26 19:27:43 +00:00
|
|
|
CECount = -1;
|
2007-12-26 19:55:00 +00:00
|
|
|
}
|
|
|
|
else {
|
|
|
|
source->extendCEs = tempBufCE;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (CECount == -1) {
|
2007-12-26 19:55:00 +00:00
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
source->extendCEsSize = 0;
|
2007-12-26 19:27:43 +00:00
|
|
|
source->CEpos = source->CEs;
|
|
|
|
freeHeapWritableBuffer(&temp);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (strbuffer != buffer) {
|
|
|
|
uprv_free(strbuffer);
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return (uint32_t)UCOL_NULLORDER;
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
source->CEpos = source->extendCEs + CECount;
|
|
|
|
endCEBuffer = source->extendCEs + source->extendCEsSize;
|
2004-11-11 23:34:58 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (offsetBias >= 0 && source->offsetStore >= &source->offsetBuffer[source->offsetBufferSize]) {
|
|
|
|
int32_t storeIX = source->offsetStore - source->offsetBuffer;
|
|
|
|
int32_t *tob = (int32_t *) uprv_realloc(source->offsetBuffer,
|
|
|
|
sizeof(int32_t) * (source->offsetBufferSize + UCOL_EXPAND_CE_BUFFER_EXTEND_SIZE));
|
|
|
|
|
|
|
|
if (tob != NULL) {
|
|
|
|
source->offsetBuffer = tob;
|
|
|
|
source->offsetStore = &source->offsetBuffer[storeIX];
|
|
|
|
source->offsetBufferSize += UCOL_EXPAND_CE_BUFFER_EXTEND_SIZE;
|
|
|
|
} else {
|
|
|
|
// memory error...
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
source->CEpos = source->CEs;
|
|
|
|
freeHeapWritableBuffer(&temp);
|
|
|
|
|
|
|
|
if (strbuffer != buffer) {
|
|
|
|
uprv_free(strbuffer);
|
|
|
|
}
|
|
|
|
|
|
|
|
return (uint32_t) UCOL_NULLORDER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
rawOffset = temp.pos - temp.string;
|
2007-12-26 19:27:43 +00:00
|
|
|
CE = ucol_IGetNextCE(coll, &temp, status);
|
2001-05-17 01:06:25 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (source->offsetRepeatValue != 0) {
|
|
|
|
if (CECount > noChars) {
|
|
|
|
source->offsetRepeatCount += temp.offsetRepeatCount;
|
|
|
|
} else {
|
|
|
|
// **** does this really skip the right offsets? ****
|
|
|
|
source->offsetReturn -= (noChars - CECount);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
freeHeapWritableBuffer(&temp);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (strbuffer != buffer) {
|
|
|
|
uprv_free(strbuffer);
|
2001-04-18 01:34:26 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (offsetBias >= 0) {
|
|
|
|
source->offsetReturn = source->offsetStore - 1;
|
|
|
|
if (source->offsetReturn == source->offsetBuffer) {
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
source->toReturn = source->CEpos - 1;
|
|
|
|
if (source->toReturn == source->CEs) {
|
|
|
|
source->CEpos = source->CEs;
|
2001-05-17 21:34:04 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return *(source->toReturn);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case LONG_PRIMARY_TAG:
|
|
|
|
{
|
|
|
|
*(source->CEpos++) = ((CE & 0xFFFF00) << 8) | (UCOL_BYTE_COMMON << 8) | UCOL_BYTE_COMMON;
|
|
|
|
*(source->CEpos++) = ((CE & 0xFF)<<24)|UCOL_CONTINUATION_MARKER;
|
|
|
|
source->toReturn = source->CEpos - 1;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (source->offsetBuffer == NULL) {
|
|
|
|
source->offsetBufferSize = UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
source->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * UCOL_EXPAND_CE_BUFFER_SIZE);
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (source->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
source->offsetRepeatCount = 1;
|
|
|
|
} else {
|
|
|
|
int32_t firstOffset = (int32_t)(source->pos - source->string);
|
|
|
|
|
|
|
|
*(source->offsetStore++) = firstOffset;
|
|
|
|
*(source->offsetStore++) = firstOffset + 1;
|
|
|
|
|
|
|
|
source->offsetReturn = source->offsetStore - 1;
|
|
|
|
*(source->offsetBuffer) = firstOffset;
|
|
|
|
if (source->offsetReturn == source->offsetBuffer) {
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return *(source->toReturn);
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case EXPANSION_TAG: /* this tag always returns */
|
2008-05-23 04:22:28 +00:00
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
/*
|
|
|
|
This should handle expansion.
|
|
|
|
NOTE: we can encounter both continuations and expansions in an expansion!
|
|
|
|
I have to decide where continuations are going to be dealt with
|
|
|
|
*/
|
2008-05-23 04:22:28 +00:00
|
|
|
int32_t firstOffset = (int32_t)(source->pos - source->string);
|
|
|
|
|
|
|
|
// **** doesn't work if using iterator ****
|
|
|
|
if (source->offsetReturn != NULL) {
|
2008-05-26 19:18:33 +00:00
|
|
|
if (! (source->flags & UCOL_ITER_INNORMBUF) && source->offsetReturn == source->offsetBuffer) {
|
2008-05-23 04:22:28 +00:00
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}else {
|
|
|
|
firstOffset = -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (source->offsetBuffer == NULL) {
|
|
|
|
source->offsetBufferSize = UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
source->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * UCOL_EXPAND_CE_BUFFER_SIZE);
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* find the offset to expansion table */
|
|
|
|
CEOffset = (uint32_t *)coll->image + getExpansionOffset(CE);
|
|
|
|
size = getExpansionCount(CE);
|
|
|
|
if (size != 0) {
|
|
|
|
/*
|
|
|
|
if there are less than 16 elements in expansion, we don't terminate
|
|
|
|
*/
|
|
|
|
uint32_t count;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
for (count = 0; count < size; count++) {
|
|
|
|
*(source->CEpos ++) = *CEOffset++;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (firstOffset >= 0) {
|
|
|
|
*(source->offsetStore ++) = firstOffset + 1;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
/* else, we do */
|
|
|
|
while (*CEOffset != 0) {
|
|
|
|
*(source->CEpos ++) = *CEOffset ++;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (firstOffset >= 0) {
|
|
|
|
*(source->offsetStore ++) = firstOffset + 1;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (firstOffset >= 0) {
|
|
|
|
source->offsetReturn = source->offsetStore - 1;
|
|
|
|
*(source->offsetBuffer) = firstOffset;
|
|
|
|
if (source->offsetReturn == source->offsetBuffer) {
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
source->offsetRepeatCount += size - 1;
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
source->toReturn = source->CEpos - 1;
|
|
|
|
// in case of one element expansion, we
|
|
|
|
// want to immediately return CEpos
|
|
|
|
if(source->toReturn == source->CEs) {
|
|
|
|
source->CEpos = source->CEs;
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return *(source->toReturn);
|
2008-05-23 04:22:28 +00:00
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case DIGIT_TAG:
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
We do a check to see if we want to collate digits as numbers; if so we generate
|
|
|
|
a custom collation key. Otherwise we pull out the value stored in the expansion table.
|
|
|
|
*/
|
|
|
|
//uint32_t size;
|
|
|
|
uint32_t i; /* general counter */
|
2003-07-25 05:31:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (source->coll->numericCollation == UCOL_ON){
|
|
|
|
collIterateState state = {0,0,0,0,0,0,0,0};
|
|
|
|
UChar32 char32 = 0;
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t digIndx = 0;
|
|
|
|
uint32_t endIndex = 0;
|
|
|
|
uint32_t leadingZeroIndex = 0;
|
|
|
|
uint32_t trailingZeroCount = 0;
|
2003-07-25 05:31:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t primWeight = 0;
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
int32_t digVal = 0;
|
|
|
|
uint8_t collateVal = 0;
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
UBool nonZeroValReached = FALSE;
|
2003-07-25 05:31:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint8_t *numTempBuf;
|
|
|
|
uint8_t stackNumTempBuf[UCOL_MAX_BUFFER]; // I just need a temporary place to store my generated CEs.
|
|
|
|
uint32_t numTempBufSize = UCOL_MAX_BUFFER;
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
numTempBuf = stackNumTempBuf;
|
|
|
|
/*
|
|
|
|
We parse the source string until we hit a char that's NOT a digit.
|
|
|
|
Use this u_charDigitValue. This might be slow because we have to
|
|
|
|
handle surrogates...
|
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if (U16_IS_TRAIL (ch)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (!collIter_bos(source)){
|
|
|
|
UChar lead = getPrevNormalizedChar(source, status);
|
|
|
|
if(U16_IS_LEAD(lead)) {
|
|
|
|
char32 = U16_GET_SUPPLEMENTARY(lead,ch);
|
|
|
|
goBackOne(source);
|
|
|
|
} else {
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
digVal = u_charDigitValue(char32);
|
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
for(;;) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// Make sure we have enough space.
|
2008-05-23 04:22:28 +00:00
|
|
|
if (digIndx >= ((numTempBufSize - 2) * 2) + 1) {
|
2007-12-26 19:27:43 +00:00
|
|
|
numTempBufSize *= 2;
|
2008-05-23 04:22:28 +00:00
|
|
|
if (numTempBuf == stackNumTempBuf) {
|
2007-12-26 19:27:43 +00:00
|
|
|
numTempBuf = (uint8_t *)uprv_malloc(sizeof(uint8_t) * numTempBufSize);
|
2008-01-16 20:12:48 +00:00
|
|
|
// Null pointer check
|
|
|
|
if (numTempBuf == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return 0;
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
uprv_memcpy(numTempBuf, stackNumTempBuf, UCOL_MAX_BUFFER);
|
2008-05-23 04:22:28 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
uint8_t *temp = (uint8_t *)uprv_realloc(numTempBuf, numTempBufSize);
|
|
|
|
if (temp == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
/* The original contents weren't freed. */
|
|
|
|
uprv_free(temp);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
numTempBuf = temp;
|
|
|
|
}
|
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// Skip over trailing zeroes, and keep a count of them.
|
|
|
|
if (digVal != 0)
|
|
|
|
nonZeroValReached = TRUE;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
if (nonZeroValReached) {
|
2007-12-26 19:27:43 +00:00
|
|
|
/*
|
|
|
|
We parse the digit string into base 100 numbers (this fits into a byte).
|
|
|
|
We only add to the buffer in twos, thus if we are parsing an odd character,
|
|
|
|
that serves as the 'tens' digit while the if we are parsing an even one, that
|
|
|
|
is the 'ones' digit. We dumped the parsed base 100 value (collateVal) into
|
|
|
|
a buffer. We multiply each collateVal by 2 (to give us room) and add 5 (to avoid
|
|
|
|
overlapping magic CE byte values). The last byte we subtract 1 to ensure it is less
|
|
|
|
than all the other bytes.
|
|
|
|
|
|
|
|
Since we're doing in this reverse we want to put the first digit encountered into the
|
|
|
|
ones place and the second digit encountered into the tens place.
|
|
|
|
*/
|
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if ((digIndx + trailingZeroCount) % 2 == 1) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// High-order digit case (tens place)
|
|
|
|
collateVal += (uint8_t)(digVal * 10);
|
|
|
|
|
|
|
|
// We cannot set leadingZeroIndex unless it has been set for the
|
|
|
|
// low-order digit. Therefore, all we can do for the high-order
|
|
|
|
// digit is turn it off, never on.
|
|
|
|
// The only time we will have a high digit without a low is for
|
|
|
|
// the very first non-zero digit, so no zero check is necessary.
|
|
|
|
if (collateVal != 0)
|
|
|
|
leadingZeroIndex = 0;
|
|
|
|
|
|
|
|
numTempBuf[(digIndx/2) + 2] = collateVal*2 + 6;
|
|
|
|
collateVal = 0;
|
2008-05-23 04:22:28 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
// Low-order digit case (ones place)
|
|
|
|
collateVal = (uint8_t)digVal;
|
|
|
|
|
|
|
|
// Check for leading zeroes.
|
2008-05-23 04:22:28 +00:00
|
|
|
if (collateVal == 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (!leadingZeroIndex)
|
|
|
|
leadingZeroIndex = (digIndx/2) + 2;
|
2008-05-23 04:22:28 +00:00
|
|
|
} else
|
2007-12-26 19:27:43 +00:00
|
|
|
leadingZeroIndex = 0;
|
|
|
|
|
|
|
|
// No need to write to buffer; the case of a last odd digit
|
|
|
|
// is handled below.
|
|
|
|
}
|
|
|
|
++digIndx;
|
2008-05-23 04:22:28 +00:00
|
|
|
} else
|
2007-12-26 19:27:43 +00:00
|
|
|
++trailingZeroCount;
|
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if (!collIter_bos(source)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
ch = getPrevNormalizedChar(source, status);
|
|
|
|
//goBackOne(source);
|
2008-05-23 04:22:28 +00:00
|
|
|
if (U16_IS_TRAIL(ch)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
backupState(source, &state);
|
2008-05-23 04:22:28 +00:00
|
|
|
if (!collIter_bos(source)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
goBackOne(source);
|
|
|
|
UChar lead = getPrevNormalizedChar(source, status);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(U16_IS_LEAD(lead)) {
|
|
|
|
char32 = U16_GET_SUPPLEMENTARY(lead,ch);
|
|
|
|
} else {
|
|
|
|
loadState(source, &state, FALSE);
|
|
|
|
char32 = ch;
|
|
|
|
}
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
} else
|
2007-12-26 19:27:43 +00:00
|
|
|
char32 = ch;
|
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if ((digVal = u_charDigitValue(char32)) == -1) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (char32 > 0xFFFF) {// For surrogates.
|
|
|
|
loadState(source, &state, FALSE);
|
|
|
|
}
|
|
|
|
// Don't need to "reverse" the goBackOne call,
|
|
|
|
// as this points to the next position to process..
|
|
|
|
//if (char32 > 0xFFFF) // For surrogates.
|
|
|
|
//getNextNormalizedChar(source);
|
|
|
|
break;
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
goBackOne(source);
|
|
|
|
}else
|
|
|
|
break;
|
2007-12-14 09:23:07 +00:00
|
|
|
}
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if (! nonZeroValReached) {
|
2007-12-26 19:27:43 +00:00
|
|
|
digIndx = 2;
|
|
|
|
trailingZeroCount = 0;
|
|
|
|
numTempBuf[2] = 6;
|
2004-11-11 23:34:58 +00:00
|
|
|
}
|
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if ((digIndx + trailingZeroCount) % 2 != 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
numTempBuf[((digIndx)/2) + 2] = collateVal*2 + 6;
|
|
|
|
digIndx += 1; // The implicit leading zero
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
if (trailingZeroCount % 2 != 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// We had to consume one trailing zero for the low digit
|
|
|
|
// of the least significant byte
|
|
|
|
digIndx += 1; // The trailing zero not in the exponent
|
|
|
|
trailingZeroCount -= 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
endIndex = leadingZeroIndex ? leadingZeroIndex : ((digIndx/2) + 2) ;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// Subtract one off of the last byte. Really the first byte here, but it's reversed...
|
|
|
|
numTempBuf[2] -= 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
We want to skip over the first two slots in the buffer. The first slot
|
|
|
|
is reserved for the header byte UCOL_CODAN_PLACEHOLDER. The second slot is for the
|
|
|
|
sign/exponent byte: 0x80 + (decimalPos/2) & 7f.
|
|
|
|
The exponent must be adjusted by the number of leading zeroes, and the number of
|
|
|
|
trailing zeroes.
|
|
|
|
*/
|
|
|
|
numTempBuf[0] = UCOL_CODAN_PLACEHOLDER;
|
|
|
|
uint32_t exponent = (digIndx+trailingZeroCount)/2;
|
|
|
|
if (leadingZeroIndex)
|
|
|
|
exponent -= ((digIndx/2) + 2 - leadingZeroIndex);
|
|
|
|
numTempBuf[1] = (uint8_t)(0x80 + (exponent & 0x7F));
|
|
|
|
|
|
|
|
// Now transfer the collation key to our collIterate struct.
|
|
|
|
// The total size for our collation key is endIndx bumped up to the next largest even value divided by two.
|
|
|
|
//size = ((endIndex+1) & ~1)/2;
|
|
|
|
*(source->CEpos++) = (((numTempBuf[0] << 8) | numTempBuf[1]) << UCOL_PRIMARYORDERSHIFT) | //Primary weight
|
|
|
|
(UCOL_BYTE_COMMON << UCOL_SECONDARYORDERSHIFT) | // Secondary weight
|
|
|
|
UCOL_BYTE_COMMON; // Tertiary weight.
|
|
|
|
i = endIndex - 1; // Reset the index into the buffer.
|
2008-05-23 04:22:28 +00:00
|
|
|
while(i >= 2) {
|
2007-12-26 19:27:43 +00:00
|
|
|
primWeight = numTempBuf[i--] << 8;
|
|
|
|
if ( i >= 2)
|
|
|
|
primWeight |= numTempBuf[i--];
|
|
|
|
*(source->CEpos++) = (primWeight << UCOL_PRIMARYORDERSHIFT) | UCOL_CONTINUATION_MARKER;
|
2004-11-11 23:34:58 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if (numTempBuf != stackNumTempBuf)
|
|
|
|
uprv_free(numTempBuf);
|
|
|
|
|
|
|
|
source->toReturn = source->CEpos -1;
|
|
|
|
return *(source->toReturn);
|
2008-05-23 04:22:28 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
CEOffset = (uint32_t *)coll->image + getExpansionOffset(CE);
|
|
|
|
CE = *(CEOffset++);
|
2004-11-11 23:34:58 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case HANGUL_SYLLABLE_TAG: /* AC00-D7AF*/
|
|
|
|
{
|
2007-12-28 20:08:41 +00:00
|
|
|
static const uint32_t
|
2007-12-26 19:27:43 +00:00
|
|
|
SBase = 0xAC00, LBase = 0x1100, VBase = 0x1161, TBase = 0x11A7;
|
|
|
|
//const uint32_t LCount = 19;
|
2007-12-28 20:08:41 +00:00
|
|
|
static const uint32_t VCount = 21;
|
|
|
|
static const uint32_t TCount = 28;
|
2007-12-26 19:27:43 +00:00
|
|
|
//const uint32_t NCount = VCount * TCount; /* 588 */
|
|
|
|
//const uint32_t SCount = LCount * NCount; /* 11172 */
|
|
|
|
|
|
|
|
uint32_t L = ch - SBase;
|
|
|
|
/*
|
|
|
|
divide into pieces.
|
|
|
|
we do it in this order since some compilers can do % and / in one
|
|
|
|
operation
|
|
|
|
*/
|
|
|
|
uint32_t T = L % TCount;
|
|
|
|
L /= TCount;
|
|
|
|
uint32_t V = L % VCount;
|
|
|
|
L /= VCount;
|
2004-11-11 23:34:58 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* offset them */
|
|
|
|
L += LBase;
|
|
|
|
V += VBase;
|
|
|
|
T += TBase;
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2008-05-23 04:22:28 +00:00
|
|
|
if (source->offsetBuffer == NULL) {
|
|
|
|
source->offsetBufferSize = UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
source->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * UCOL_EXPAND_CE_BUFFER_SIZE);
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
|
|
|
int32_t firstOffset = (int32_t)(source->pos - source->string);
|
|
|
|
|
|
|
|
*(source->offsetStore++) = firstOffset;
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/*
|
2008-05-23 04:22:28 +00:00
|
|
|
* return the first CE, but first put the rest into the expansion buffer
|
|
|
|
*/
|
|
|
|
if (!source->coll->image->jamoSpecial) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*(source->CEpos++) = UTRIE_GET32_FROM_LEAD(&coll->mapping, L);
|
|
|
|
*(source->CEpos++) = UTRIE_GET32_FROM_LEAD(&coll->mapping, V);
|
2008-05-23 04:22:28 +00:00
|
|
|
*(source->offsetStore++) = firstOffset + 1;
|
|
|
|
|
|
|
|
if (T != TBase) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*(source->CEpos++) = UTRIE_GET32_FROM_LEAD(&coll->mapping, T);
|
2008-05-23 04:22:28 +00:00
|
|
|
*(source->offsetStore++) = firstOffset + 1;
|
|
|
|
}
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
source->toReturn = source->CEpos - 1;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
|
|
|
source->offsetReturn = source->offsetStore - 1;
|
|
|
|
if (source->offsetReturn == source->offsetBuffer) {
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
|
|
|
return *(source->toReturn);
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
// Since Hanguls pass the FCD check, it is
|
|
|
|
// guaranteed that we won't be in
|
|
|
|
// the normalization buffer if something like this happens
|
|
|
|
// Move Jamos into normalization buffer
|
|
|
|
/*
|
|
|
|
Move the Jamos into the
|
|
|
|
normalization buffer
|
|
|
|
*/
|
|
|
|
UChar *tempbuffer = source->writableBuffer +
|
|
|
|
(source->writableBufSize - 1);
|
|
|
|
*(tempbuffer) = 0;
|
|
|
|
if (T != TBase) {
|
|
|
|
*(tempbuffer - 1) = (UChar)T;
|
|
|
|
*(tempbuffer - 2) = (UChar)V;
|
|
|
|
*(tempbuffer - 3) = (UChar)L;
|
|
|
|
*(tempbuffer - 4) = 0;
|
|
|
|
} else {
|
|
|
|
*(tempbuffer - 1) = (UChar)V;
|
|
|
|
*(tempbuffer - 2) = (UChar)L;
|
|
|
|
*(tempbuffer - 3) = 0;
|
|
|
|
}
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/*
|
|
|
|
Indicate where to continue in main input string after exhausting
|
|
|
|
the writableBuffer
|
|
|
|
*/
|
|
|
|
if (source->pos == source->string) {
|
|
|
|
source->fcdPosition = NULL;
|
|
|
|
} else {
|
|
|
|
source->fcdPosition = source->pos-1;
|
|
|
|
}
|
2001-09-24 02:24:09 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
source->pos = tempbuffer;
|
|
|
|
source->origFlags = source->flags;
|
|
|
|
source->flags |= UCOL_ITER_INNORMBUF;
|
|
|
|
source->flags &= ~(UCOL_ITER_NORM | UCOL_ITER_HASLEN);
|
2001-09-20 20:16:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return(UCOL_IGNORABLE);
|
|
|
|
}
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
case IMPLICIT_TAG: /* everything that is not defined otherwise */
|
2008-05-23 04:22:28 +00:00
|
|
|
#if 0
|
|
|
|
if (source->offsetBuffer == NULL) {
|
|
|
|
source->offsetBufferSize = UCOL_EXPAND_CE_BUFFER_SIZE;
|
|
|
|
source->offsetBuffer = (int32_t *) uprv_malloc(sizeof(int32_t) * UCOL_EXPAND_CE_BUFFER_SIZE);
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
|
|
|
|
// **** doesn't work if using iterator ****
|
|
|
|
if (source->flags & UCOL_ITER_INNORMBUF) {
|
|
|
|
source->offsetRepeatCount = 1;
|
|
|
|
} else {
|
|
|
|
int32_t firstOffset = (int32_t)(source->pos - source->string);
|
|
|
|
|
|
|
|
*(source->offsetStore++) = firstOffset;
|
|
|
|
*(source->offsetStore++) = firstOffset + 1;
|
|
|
|
|
|
|
|
source->offsetReturn = source->offsetStore - 1;
|
|
|
|
if (source->offsetReturn == source->offsetBuffer) {
|
|
|
|
source->offsetStore = source->offsetBuffer;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
return getPrevImplicit(ch, source);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
// TODO: Remove CJK implicits as they are handled by the getImplicitPrimary function
|
|
|
|
case CJK_IMPLICIT_TAG: /* 0x3400-0x4DB5, 0x4E00-0x9FA5, 0xF900-0xFA2D*/
|
|
|
|
return getPrevImplicit(ch, source);
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
case SURROGATE_TAG: /* This is a surrogate pair */
|
|
|
|
/* essentialy an engaged lead surrogate. */
|
|
|
|
/* if you have encountered it here, it means that a */
|
|
|
|
/* broken sequence was encountered and this is an error */
|
|
|
|
return 0;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case LEAD_SURROGATE_TAG: /* D800-DBFF*/
|
|
|
|
return 0; /* broken surrogate sequence */
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
case TRAIL_SURROGATE_TAG: /* DC00-DFFF*/
|
|
|
|
{
|
|
|
|
UChar32 cp = 0;
|
|
|
|
UChar prevChar;
|
|
|
|
UChar *prev;
|
|
|
|
if (isAtStartPrevIterate(source)) {
|
|
|
|
/* we are at the start of the string, wrong place to be at */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if (source->pos != source->writableBuffer) {
|
|
|
|
prev = source->pos - 1;
|
|
|
|
} else {
|
|
|
|
prev = source->fcdPosition;
|
|
|
|
}
|
|
|
|
prevChar = *prev;
|
2001-09-27 17:17:54 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* Handles Han and Supplementary characters here.*/
|
|
|
|
if (U16_IS_LEAD(prevChar)) {
|
|
|
|
cp = ((((uint32_t)prevChar)<<10UL)+(ch)-(((uint32_t)0xd800<<10UL)+0xdc00-0x10000));
|
|
|
|
source->pos = prev;
|
|
|
|
} else {
|
|
|
|
return 0; /* completely ignorable */
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return getPrevImplicit(cp, source);
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* UCA is filled with these. Tailorings are NOT_FOUND */
|
|
|
|
/* not yet implemented */
|
|
|
|
case CHARSET_TAG: /* this tag always returns */
|
|
|
|
/* probably after 1.8 */
|
|
|
|
return UCOL_NOT_FOUND;
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
default: /* this tag always returns */
|
|
|
|
*status = U_INTERNAL_PROGRAM_ERROR;
|
|
|
|
CE=0;
|
|
|
|
break;
|
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (CE <= UCOL_NOT_FOUND) {
|
|
|
|
break;
|
2001-09-20 20:16:39 +00:00
|
|
|
}
|
2001-03-07 21:01:53 +00:00
|
|
|
}
|
2008-05-23 04:22:28 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return CE;
|
2001-02-20 00:26:50 +00:00
|
|
|
}
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* This should really be a macro */
|
|
|
|
/* However, it is used only when stack buffers are not sufficiently big, and then we're messed up performance wise */
|
|
|
|
/* anyway */
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2001-03-14 07:49:03 +00:00
|
|
|
uint8_t *reallocateBuffer(uint8_t **secondaries, uint8_t *secStart, uint8_t *second, uint32_t *secSize, uint32_t newSize, UErrorCode *status) {
|
2001-05-17 23:09:35 +00:00
|
|
|
#ifdef UCOL_DEBUG
|
2007-12-26 19:27:43 +00:00
|
|
|
fprintf(stderr, ".");
|
2001-05-17 18:19:42 +00:00
|
|
|
#endif
|
2007-12-26 19:27:43 +00:00
|
|
|
uint8_t *newStart = NULL;
|
|
|
|
uint32_t offset = *secondaries-secStart;
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(secStart==second) {
|
|
|
|
newStart=(uint8_t*)uprv_malloc(newSize);
|
|
|
|
if(newStart==NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
uprv_memcpy(newStart, secStart, *secondaries-secStart);
|
|
|
|
} else {
|
|
|
|
newStart=(uint8_t*)uprv_realloc(secStart, newSize);
|
|
|
|
if(newStart==NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
2008-02-16 22:58:32 +00:00
|
|
|
/* Since we're reallocating, return original reference so we don't loose it. */
|
|
|
|
return secStart;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2001-01-09 00:52:18 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
*secondaries=newStart+offset;
|
|
|
|
*secSize=newSize;
|
|
|
|
return newStart;
|
2001-01-09 00:52:18 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* This should really be a macro */
|
|
|
|
/* This function is used to reverse parts of a buffer. We need this operation when doing continuation */
|
|
|
|
/* secondaries in French */
|
|
|
|
/*
|
2001-01-09 00:52:18 +00:00
|
|
|
void uprv_ucol_reverse_buffer(uint8_t *start, uint8_t *end) {
|
|
|
|
uint8_t temp;
|
|
|
|
while(start<end) {
|
|
|
|
temp = *start;
|
|
|
|
*start++ = *end;
|
|
|
|
*end-- = temp;
|
|
|
|
}
|
|
|
|
}
|
2001-01-16 00:28:40 +00:00
|
|
|
*/
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
#define uprv_ucol_reverse_buffer(TYPE, start, end) { \
|
|
|
|
TYPE tempA; \
|
|
|
|
while((start)<(end)) { \
|
|
|
|
tempA = *(start); \
|
|
|
|
*(start)++ = *(end); \
|
|
|
|
*(end)-- = tempA; \
|
|
|
|
} \
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
2001-01-15 07:28:54 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/****************************************************************************/
|
|
|
|
/* Following are the sortkey generation functions */
|
|
|
|
/* */
|
|
|
|
/****************************************************************************/
|
2001-01-15 07:28:54 +00:00
|
|
|
|
2001-09-18 18:37:57 +00:00
|
|
|
/**
|
|
|
|
* Merge two sort keys.
|
|
|
|
* This is useful, for example, to combine sort keys from first and last names
|
|
|
|
* to sort such pairs.
|
|
|
|
* Merged sort keys consider on each collation level the first part first entirely,
|
|
|
|
* then the second one.
|
|
|
|
* It is possible to merge multiple sort keys by consecutively merging
|
|
|
|
* another one with the intermediate result.
|
|
|
|
*
|
|
|
|
* The length of the merge result is the sum of the lengths of the input sort keys
|
|
|
|
* minus 1.
|
|
|
|
*
|
|
|
|
* @param src1 the first sort key
|
|
|
|
* @param src1Length the length of the first sort key, including the zero byte at the end;
|
|
|
|
* can be -1 if the function is to find the length
|
|
|
|
* @param src2 the second sort key
|
|
|
|
* @param src2Length the length of the second sort key, including the zero byte at the end;
|
|
|
|
* can be -1 if the function is to find the length
|
|
|
|
* @param dest the buffer where the merged sort key is written,
|
|
|
|
* can be NULL if destCapacity==0
|
|
|
|
* @param destCapacity the number of bytes in the dest buffer
|
|
|
|
* @return the length of the merged sort key, src1Length+src2Length-1;
|
|
|
|
* can be larger than destCapacity, or 0 if an error occurs (only for illegal arguments),
|
|
|
|
* in which cases the contents of dest is undefined
|
|
|
|
*
|
|
|
|
* @draft
|
|
|
|
*/
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
2001-09-18 18:37:57 +00:00
|
|
|
ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
|
|
|
|
const uint8_t *src2, int32_t src2Length,
|
|
|
|
uint8_t *dest, int32_t destCapacity) {
|
|
|
|
int32_t destLength;
|
|
|
|
uint8_t b;
|
|
|
|
|
|
|
|
/* check arguments */
|
|
|
|
if( src1==NULL || src1Length<-2 || src1Length==0 || (src1Length>0 && src1[src1Length-1]!=0) ||
|
|
|
|
src2==NULL || src2Length<-2 || src2Length==0 || (src2Length>0 && src2[src2Length-1]!=0) ||
|
|
|
|
destCapacity<0 || (destCapacity>0 && dest==NULL)
|
|
|
|
) {
|
|
|
|
/* error, attempt to write a zero byte and return 0 */
|
|
|
|
if(dest!=NULL && destCapacity>0) {
|
|
|
|
*dest=0;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* check lengths and capacity */
|
|
|
|
if(src1Length<0) {
|
|
|
|
src1Length=(int32_t)uprv_strlen((const char *)src1)+1;
|
|
|
|
}
|
|
|
|
if(src2Length<0) {
|
|
|
|
src2Length=(int32_t)uprv_strlen((const char *)src2)+1;
|
|
|
|
}
|
|
|
|
|
|
|
|
destLength=src1Length+src2Length-1;
|
|
|
|
if(destLength>destCapacity) {
|
|
|
|
/* the merged sort key does not fit into the destination */
|
|
|
|
return destLength;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* merge the sort keys with the same number of levels */
|
|
|
|
while(*src1!=0 && *src2!=0) { /* while both have another level */
|
|
|
|
/* copy level from src1 not including 00 or 01 */
|
|
|
|
while((b=*src1)>=2) {
|
|
|
|
++src1;
|
|
|
|
*dest++=b;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* add a 02 merge separator */
|
|
|
|
*dest++=2;
|
|
|
|
|
|
|
|
/* copy level from src2 not including 00 or 01 */
|
|
|
|
while((b=*src2)>=2) {
|
|
|
|
++src2;
|
|
|
|
*dest++=b;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* if both sort keys have another level, then add a 01 level separator and continue */
|
|
|
|
if(*src1==1 && *src2==1) {
|
|
|
|
++src1;
|
|
|
|
++src2;
|
|
|
|
*dest++=1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* here, at least one sort key is finished now, but the other one
|
|
|
|
* might have some contents left from containing more levels;
|
|
|
|
* that contents is just appended to the result
|
|
|
|
*/
|
|
|
|
if(*src1!=0) {
|
|
|
|
/* src1 is not finished, therefore *src2==0, and src1 is appended */
|
|
|
|
src2=src1;
|
|
|
|
}
|
|
|
|
/* append src2, "the other, unfinished sort key" */
|
|
|
|
uprv_strcpy((char *)dest, (const char *)src2);
|
|
|
|
|
|
|
|
/* trust that neither sort key contained illegally embedded zero bytes */
|
|
|
|
return destLength;
|
|
|
|
}
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* sortkey API */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_getSortKey(const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
uint8_t *result,
|
|
|
|
int32_t resultLength)
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
UTRACE_ENTRY(UTRACE_UCOL_GET_SORTKEY);
|
|
|
|
if (UTRACE_LEVEL(UTRACE_VERBOSE)) {
|
|
|
|
UTRACE_DATA3(UTRACE_VERBOSE, "coll=%p, source string = %vh ", coll, source,
|
|
|
|
((sourceLength==-1 && source!=NULL) ? u_strlen(source) : sourceLength));
|
|
|
|
}
|
2003-09-20 01:21:23 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
|
|
|
int32_t keySize = 0;
|
2003-05-09 20:10:01 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(source != NULL) {
|
|
|
|
// source == NULL is actually an error situation, but we would need to
|
|
|
|
// have an error code to return it. Until we introduce a new
|
|
|
|
// API, it stays like this
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* this uses the function pointer that is set in updateinternalstate */
|
|
|
|
/* currently, there are two funcs: */
|
|
|
|
/*ucol_calcSortKey(...);*/
|
|
|
|
/*ucol_calcSortKeySimpleTertiary(...);*/
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
keySize = coll->sortKeyGen(coll, source, sourceLength, &result, resultLength, FALSE, &status);
|
2008-02-18 00:46:38 +00:00
|
|
|
//if (U_FAILURE(status) && status != U_BUFFER_OVERFLOW_ERROR && result && resultLength > 0) {
|
2008-02-17 06:50:59 +00:00
|
|
|
// That's not good. Something unusual happened.
|
|
|
|
// We don't know how much we initialized before we failed.
|
|
|
|
// NULL terminate for safety.
|
|
|
|
// We have no way say that we have generated a partial sort key.
|
2008-02-18 00:46:38 +00:00
|
|
|
//result[0] = 0;
|
|
|
|
//keySize = 0;
|
|
|
|
//}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
UTRACE_DATA2(UTRACE_VERBOSE, "Sort Key = %vb", result, keySize);
|
|
|
|
UTRACE_EXIT_STATUS(status);
|
|
|
|
return keySize;
|
2001-01-16 00:28:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* this function is called by the C++ API for sortkey generation */
|
2001-09-27 01:01:30 +00:00
|
|
|
U_CFUNC int32_t
|
|
|
|
ucol_getSortKeyWithAllocation(const UCollator *coll,
|
|
|
|
const UChar *source, int32_t sourceLength,
|
|
|
|
uint8_t **pResult,
|
|
|
|
UErrorCode *pErrorCode) {
|
|
|
|
*pResult = 0;
|
|
|
|
return coll->sortKeyGen(coll, source, sourceLength, pResult, 0, TRUE, pErrorCode);
|
2001-01-15 07:28:54 +00:00
|
|
|
}
|
|
|
|
|
2001-09-24 00:20:59 +00:00
|
|
|
#define UCOL_FSEC_BUF_SIZE 256
|
2001-01-16 00:28:40 +00:00
|
|
|
|
|
|
|
/* This function tries to get the size of a sortkey. It will be invoked if the size of resulting buffer is 0 */
|
|
|
|
/* or if we run out of space while making a sortkey and want to return ASAP */
|
2002-08-02 06:55:51 +00:00
|
|
|
int32_t ucol_getSortKeySize(const UCollator *coll, collIterate *s, int32_t currentSize, UColAttributeValue strength, int32_t len) {
|
2001-01-05 00:47:25 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
2005-07-06 21:27:21 +00:00
|
|
|
//const UCAConstants *UCAconsts = (UCAConstants *)((uint8_t *)coll->UCA->image + coll->image->UCAConsts);
|
2001-02-28 19:30:41 +00:00
|
|
|
uint8_t compareSec = (uint8_t)((strength >= UCOL_SECONDARY)?0:0xFF);
|
|
|
|
uint8_t compareTer = (uint8_t)((strength >= UCOL_TERTIARY)?0:0xFF);
|
|
|
|
uint8_t compareQuad = (uint8_t)((strength >= UCOL_QUATERNARY)?0:0xFF);
|
2001-01-05 00:47:25 +00:00
|
|
|
UBool compareIdent = (strength == UCOL_IDENTICAL);
|
2001-01-09 00:52:18 +00:00
|
|
|
UBool doCase = (coll->caseLevel == UCOL_ON);
|
2001-03-15 02:49:35 +00:00
|
|
|
UBool shifted = (coll->alternateHandling == UCOL_SHIFTED);
|
2001-10-20 01:09:31 +00:00
|
|
|
//UBool qShifted = shifted && (compareQuad == 0);
|
2001-10-10 01:48:36 +00:00
|
|
|
UBool doHiragana = (coll->hiraganaQ == UCOL_ON) && (compareQuad == 0);
|
2001-01-26 00:12:23 +00:00
|
|
|
UBool isFrenchSec = (coll->frenchCollation == UCOL_ON) && (compareSec == 0);
|
2001-09-24 00:20:59 +00:00
|
|
|
uint8_t fSecsBuff[UCOL_FSEC_BUF_SIZE];
|
|
|
|
uint8_t *fSecs = fSecsBuff;
|
|
|
|
uint32_t fSecsLen = 0, fSecsMaxLen = UCOL_FSEC_BUF_SIZE;
|
|
|
|
uint8_t *frenchStartPtr = NULL, *frenchEndPtr = NULL;
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2001-06-26 22:24:10 +00:00
|
|
|
uint32_t variableTopValue = coll->variableTopValue;
|
2003-12-03 00:45:12 +00:00
|
|
|
uint8_t UCOL_COMMON_BOT4 = (uint8_t)((coll->variableTopValue>>8)+1);
|
2001-10-02 16:49:57 +00:00
|
|
|
if(doHiragana) {
|
2007-12-28 20:08:41 +00:00
|
|
|
UCOL_COMMON_BOT4++;
|
|
|
|
/* allocate one more space for hiragana */
|
2001-10-02 16:49:57 +00:00
|
|
|
}
|
2001-02-28 19:30:41 +00:00
|
|
|
uint8_t UCOL_BOT_COUNT4 = (uint8_t)(0xFF - UCOL_COMMON_BOT4);
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2001-03-22 21:16:20 +00:00
|
|
|
uint32_t order = UCOL_NO_MORE_CES;
|
2001-01-09 00:52:18 +00:00
|
|
|
uint8_t primary1 = 0;
|
|
|
|
uint8_t primary2 = 0;
|
2001-01-05 00:47:25 +00:00
|
|
|
uint8_t secondary = 0;
|
|
|
|
uint8_t tertiary = 0;
|
2001-01-09 00:52:18 +00:00
|
|
|
int32_t caseShift = 0;
|
2001-01-18 00:46:19 +00:00
|
|
|
uint32_t c2 = 0, c3 = 0, c4 = 0; /* variables for compression */
|
2001-03-02 00:19:43 +00:00
|
|
|
|
|
|
|
uint8_t caseSwitch = coll->caseSwitch;
|
|
|
|
uint8_t tertiaryMask = coll->tertiaryMask;
|
2001-05-10 22:33:50 +00:00
|
|
|
uint8_t tertiaryCommon = coll->tertiaryCommon;
|
2001-03-02 00:19:43 +00:00
|
|
|
|
2001-02-26 10:28:56 +00:00
|
|
|
UBool wasShifted = FALSE;
|
|
|
|
UBool notIsContinuation = FALSE;
|
2001-05-03 23:33:29 +00:00
|
|
|
uint8_t leadPrimary = 0;
|
2001-04-23 01:53:49 +00:00
|
|
|
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
for(;;) {
|
2007-12-26 19:27:43 +00:00
|
|
|
order = ucol_IGetNextCE(coll, s, &status);
|
|
|
|
if(order == UCOL_NO_MORE_CES) {
|
|
|
|
break;
|
|
|
|
}
|
2001-05-29 04:59:29 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(order == 0) {
|
2001-03-15 02:49:35 +00:00
|
|
|
continue;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
notIsContinuation = !isContinuation(order);
|
2001-01-09 00:52:18 +00:00
|
|
|
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(notIsContinuation) {
|
2001-05-10 22:33:50 +00:00
|
|
|
tertiary = (uint8_t)((order & UCOL_BYTE_SIZE_MASK));
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
2001-05-10 22:33:50 +00:00
|
|
|
tertiary = (uint8_t)((order & UCOL_REMOVE_CONTINUATION));
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
secondary = (uint8_t)((order >>= 8) & UCOL_BYTE_SIZE_MASK);
|
|
|
|
primary2 = (uint8_t)((order >>= 8) & UCOL_BYTE_SIZE_MASK);
|
|
|
|
primary1 = (uint8_t)(order >> 8);
|
2001-01-09 00:52:18 +00:00
|
|
|
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(shifted && ((notIsContinuation && order <= variableTopValue && primary1 > 0)
|
2003-12-03 00:45:12 +00:00
|
|
|
|| (!notIsContinuation && wasShifted))
|
2002-06-05 21:14:41 +00:00
|
|
|
|| (wasShifted && primary1 == 0)) { /* amendment to the UCA says that primary ignorables */
|
2007-12-26 19:27:43 +00:00
|
|
|
/* and other ignorables should be removed if following a shifted code point */
|
|
|
|
if(primary1 == 0) { /* if we were shifted and we got an ignorable code point */
|
|
|
|
/* we should just completely ignore it */
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if(compareQuad == 0) {
|
|
|
|
if(c4 > 0) {
|
|
|
|
currentSize += (c2/UCOL_BOT_COUNT4)+1;
|
|
|
|
c4 = 0;
|
|
|
|
}
|
|
|
|
currentSize++;
|
|
|
|
if(primary2 != 0) {
|
|
|
|
currentSize++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
wasShifted = TRUE;
|
|
|
|
} else {
|
2001-02-26 10:28:56 +00:00
|
|
|
wasShifted = FALSE;
|
2001-01-09 00:52:18 +00:00
|
|
|
/* Note: This code assumes that the table is well built i.e. not having 0 bytes where they are not supposed to be. */
|
|
|
|
/* Usually, we'll have non-zero primary1 & primary2, except in cases of LatinOne and friends, when primary2 will */
|
2001-04-25 18:00:41 +00:00
|
|
|
/* calculate sortkey size */
|
2001-04-23 23:41:46 +00:00
|
|
|
if(primary1 != UCOL_IGNORABLE) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(notIsContinuation) {
|
|
|
|
if(leadPrimary == primary1) {
|
|
|
|
currentSize++;
|
|
|
|
} else {
|
|
|
|
if(leadPrimary != 0) {
|
|
|
|
currentSize++;
|
|
|
|
}
|
|
|
|
if(primary2 == UCOL_IGNORABLE) {
|
|
|
|
/* one byter, not compressed */
|
|
|
|
currentSize++;
|
|
|
|
leadPrimary = 0;
|
|
|
|
}
|
|
|
|
else if(primary1<UCOL_BYTE_FIRST_NON_LATIN_PRIMARY ||
|
|
|
|
//(primary1 > (UCOL_RESET_TOP_VALUE>>24) && primary1 < (UCOL_NEXT_TOP_VALUE>>24))) {
|
|
|
|
//(primary1 > (*UCAconsts->UCA_LAST_NON_VARIABLE>>24) && primary1 < (*UCAconsts->UCA_FIRST_IMPLICIT>>24))) {
|
|
|
|
(primary1 > maxRegularPrimary && primary1 < minImplicitPrimary))
|
|
|
|
{
|
|
|
|
/* not compressible */
|
|
|
|
leadPrimary = 0;
|
|
|
|
currentSize+=2;
|
|
|
|
}
|
|
|
|
else { /* compress */
|
|
|
|
leadPrimary = primary1;
|
|
|
|
currentSize+=2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { /* we are in continuation, so we're gonna add primary to the key don't care about compression */
|
2001-04-23 23:41:46 +00:00
|
|
|
currentSize++;
|
2007-12-26 19:27:43 +00:00
|
|
|
if(primary2 != UCOL_IGNORABLE) {
|
|
|
|
currentSize++;
|
|
|
|
}
|
2001-04-24 23:32:40 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
}
|
2001-01-09 00:52:18 +00:00
|
|
|
|
|
|
|
if(secondary > compareSec) { /* I think that != 0 test should be != IGNORABLE */
|
2007-12-26 19:27:43 +00:00
|
|
|
if(!isFrenchSec){
|
|
|
|
if (secondary == UCOL_COMMON2 && notIsContinuation) {
|
|
|
|
c2++;
|
2001-01-26 00:12:23 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(c2 > 0) {
|
|
|
|
if (secondary > UCOL_COMMON2) { // not necessary for 4th level.
|
|
|
|
currentSize += (c2/(uint32_t)UCOL_TOP_COUNT2)+1;
|
|
|
|
} else {
|
|
|
|
currentSize += (c2/(uint32_t)UCOL_BOT_COUNT2)+1;
|
|
|
|
}
|
|
|
|
c2 = 0;
|
|
|
|
}
|
|
|
|
currentSize++;
|
2001-01-26 00:12:23 +00:00
|
|
|
}
|
2001-09-24 00:20:59 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
fSecs[fSecsLen++] = secondary;
|
|
|
|
if(fSecsLen == fSecsMaxLen) {
|
2008-02-13 09:35:50 +00:00
|
|
|
uint8_t *fSecsTemp;
|
2007-12-26 19:27:43 +00:00
|
|
|
if(fSecs == fSecsBuff) {
|
2008-02-13 09:35:50 +00:00
|
|
|
fSecsTemp = (uint8_t *)uprv_malloc(2*fSecsLen);
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
2008-02-13 09:35:50 +00:00
|
|
|
fSecsTemp = (uint8_t *)uprv_realloc(fSecs, 2*fSecsLen);
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2008-02-13 09:35:50 +00:00
|
|
|
if(fSecsTemp == NULL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
status = U_MEMORY_ALLOCATION_ERROR;
|
2008-02-13 09:35:50 +00:00
|
|
|
return 0;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2008-02-13 09:35:50 +00:00
|
|
|
fSecs = fSecsTemp;
|
2007-12-26 19:27:43 +00:00
|
|
|
fSecsMaxLen *= 2;
|
|
|
|
}
|
|
|
|
if(notIsContinuation) {
|
|
|
|
if (frenchStartPtr != NULL) {
|
|
|
|
/* reverse secondaries from frenchStartPtr up to frenchEndPtr */
|
|
|
|
uprv_ucol_reverse_buffer(uint8_t, frenchStartPtr, frenchEndPtr);
|
|
|
|
frenchStartPtr = NULL;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (frenchStartPtr == NULL) {
|
|
|
|
frenchStartPtr = fSecs+fSecsLen-2;
|
|
|
|
}
|
|
|
|
frenchEndPtr = fSecs+fSecsLen-1;
|
|
|
|
}
|
2001-09-24 00:20:59 +00:00
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2006-01-28 08:26:55 +00:00
|
|
|
if(doCase && (primary1 > 0 || strength >= UCOL_SECONDARY)) {
|
|
|
|
// do the case level if we need to do it. We don't want to calculate
|
|
|
|
// case level for primary ignorables if we have only primary strength and case level
|
2007-12-26 19:27:43 +00:00
|
|
|
// otherwise we would break well formedness of CEs
|
|
|
|
if (caseShift == 0) {
|
2001-05-10 22:33:50 +00:00
|
|
|
currentSize++;
|
|
|
|
caseShift = UCOL_CASE_SHIFT_START;
|
2001-05-11 01:13:08 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if((tertiary&0x3F) > 0 && notIsContinuation) {
|
|
|
|
caseShift--;
|
|
|
|
if((tertiary &0xC0) != 0) {
|
|
|
|
if (caseShift == 0) {
|
|
|
|
currentSize++;
|
|
|
|
caseShift = UCOL_CASE_SHIFT_START;
|
|
|
|
}
|
|
|
|
caseShift--;
|
|
|
|
}
|
|
|
|
}
|
2001-05-10 22:33:50 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(notIsContinuation) {
|
|
|
|
tertiary ^= caseSwitch;
|
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2001-05-10 22:33:50 +00:00
|
|
|
tertiary &= tertiaryMask;
|
2001-01-09 00:52:18 +00:00
|
|
|
if(tertiary > compareTer) { /* I think that != 0 test should be != IGNORABLE */
|
2007-12-26 19:27:43 +00:00
|
|
|
if (tertiary == tertiaryCommon && notIsContinuation) {
|
|
|
|
c3++;
|
|
|
|
} else {
|
|
|
|
if(c3 > 0) {
|
|
|
|
if((tertiary > tertiaryCommon && tertiaryCommon == UCOL_COMMON3_NORMAL)
|
|
|
|
|| (tertiary <= tertiaryCommon && tertiaryCommon == UCOL_COMMON3_UPPERFIRST)) {
|
|
|
|
currentSize += (c3/(uint32_t)coll->tertiaryTopCount)+1;
|
|
|
|
} else {
|
|
|
|
currentSize += (c3/(uint32_t)coll->tertiaryBottomCount)+1;
|
|
|
|
}
|
|
|
|
c3 = 0;
|
|
|
|
}
|
|
|
|
currentSize++;
|
2001-01-18 00:46:19 +00:00
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
2001-01-09 00:52:18 +00:00
|
|
|
|
2001-10-02 16:49:57 +00:00
|
|
|
if(/*qShifted*/(compareQuad==0) && notIsContinuation) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(s->flags & UCOL_WAS_HIRAGANA) { // This was Hiragana and we need to note it
|
|
|
|
if(c4>0) { // Close this part
|
|
|
|
currentSize += (c4/UCOL_BOT_COUNT4)+1;
|
|
|
|
c4 = 0;
|
|
|
|
}
|
|
|
|
currentSize++; // Add the Hiragana
|
|
|
|
} else { // This wasn't Hiragana, so we can continue adding stuff
|
|
|
|
c4++;
|
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
2000-11-20 19:17:17 +00:00
|
|
|
|
2001-09-24 00:20:59 +00:00
|
|
|
if(!isFrenchSec){
|
2007-12-26 19:27:43 +00:00
|
|
|
if(c2 > 0) {
|
|
|
|
currentSize += (c2/(uint32_t)UCOL_BOT_COUNT2)+((c2%(uint32_t)UCOL_BOT_COUNT2 != 0)?1:0);
|
|
|
|
}
|
2001-09-24 00:20:59 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t i = 0;
|
|
|
|
if(frenchStartPtr != NULL) {
|
|
|
|
uprv_ucol_reverse_buffer(uint8_t, frenchStartPtr, frenchEndPtr);
|
|
|
|
}
|
|
|
|
for(i = 0; i<fSecsLen; i++) {
|
|
|
|
secondary = *(fSecs+fSecsLen-i-1);
|
|
|
|
/* This is compression code. */
|
|
|
|
if (secondary == UCOL_COMMON2) {
|
|
|
|
++c2;
|
2001-09-24 00:20:59 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(c2 > 0) {
|
|
|
|
if (secondary > UCOL_COMMON2) { // not necessary for 4th level.
|
|
|
|
currentSize += (c2/(uint32_t)UCOL_TOP_COUNT2)+((c2%(uint32_t)UCOL_TOP_COUNT2 != 0)?1:0);
|
|
|
|
} else {
|
|
|
|
currentSize += (c2/(uint32_t)UCOL_BOT_COUNT2)+((c2%(uint32_t)UCOL_BOT_COUNT2 != 0)?1:0);
|
|
|
|
}
|
|
|
|
c2 = 0;
|
|
|
|
}
|
|
|
|
currentSize++;
|
2001-09-24 00:20:59 +00:00
|
|
|
}
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if(c2 > 0) {
|
|
|
|
currentSize += (c2/(uint32_t)UCOL_BOT_COUNT2)+((c2%(uint32_t)UCOL_BOT_COUNT2 != 0)?1:0);
|
|
|
|
}
|
|
|
|
if(fSecs != fSecsBuff) {
|
|
|
|
uprv_free(fSecs);
|
|
|
|
}
|
2001-01-23 06:58:22 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if(c3 > 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
currentSize += (c3/(uint32_t)coll->tertiaryBottomCount) + ((c3%(uint32_t)coll->tertiaryBottomCount != 0)?1:0);
|
2001-01-23 06:58:22 +00:00
|
|
|
}
|
|
|
|
|
2001-03-15 02:49:35 +00:00
|
|
|
if(c4 > 0 && compareQuad == 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
currentSize += (c4/(uint32_t)UCOL_BOT_COUNT4)+((c4%(uint32_t)UCOL_BOT_COUNT4 != 0)?1:0);
|
2001-01-23 06:58:22 +00:00
|
|
|
}
|
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
if(compareIdent) {
|
2007-12-26 19:27:43 +00:00
|
|
|
currentSize += u_lengthOfIdenticalLevelRun(s->string, len);
|
2001-04-19 19:02:01 +00:00
|
|
|
}
|
|
|
|
return currentSize;
|
|
|
|
}
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-05-10 22:33:50 +00:00
|
|
|
inline void doCaseShift(uint8_t **cases, uint32_t &caseShift) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (caseShift == 0) {
|
|
|
|
*(*cases)++ = UCOL_CASE_BYTE_START;
|
|
|
|
caseShift = UCOL_CASE_SHIFT_START;
|
|
|
|
}
|
2001-05-10 22:33:50 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2002-07-27 05:16:44 +00:00
|
|
|
// Adds a value to the buffer if it's safe to add. Increments the number of added values, so that we
|
|
|
|
// know how many values we wanted to add, even if we didn't add them all
|
|
|
|
static
|
|
|
|
inline void addWithIncrement(uint8_t *&primaries, uint8_t *limit, uint32_t &size, const uint8_t value) {
|
2007-12-26 19:27:43 +00:00
|
|
|
size++;
|
|
|
|
if(primaries < limit) {
|
|
|
|
*(primaries)++ = value;
|
|
|
|
}
|
2002-07-27 05:16:44 +00:00
|
|
|
}
|
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
// Packs the secondary buffer when processing French locale. Adds the terminator.
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2002-07-27 05:16:44 +00:00
|
|
|
inline uint8_t *packFrench(uint8_t *primaries, uint8_t *primEnd, uint8_t *secondaries, uint32_t *secsize, uint8_t *frenchStartPtr, uint8_t *frenchEndPtr) {
|
2007-12-26 19:27:43 +00:00
|
|
|
uint8_t secondary;
|
|
|
|
int32_t count2 = 0;
|
|
|
|
uint32_t i = 0, size = 0;
|
|
|
|
// we use i here since the key size already accounts for terminators, so we'll discard the increment
|
|
|
|
addWithIncrement(primaries, primEnd, i, UCOL_LEVELTERMINATOR);
|
|
|
|
/* If there are any unresolved continuation secondaries, reverse them here so that we can reverse the whole secondary thing */
|
|
|
|
if(frenchStartPtr != NULL) {
|
|
|
|
uprv_ucol_reverse_buffer(uint8_t, frenchStartPtr, frenchEndPtr);
|
|
|
|
}
|
|
|
|
for(i = 0; i<*secsize; i++) {
|
|
|
|
secondary = *(secondaries-i-1);
|
|
|
|
/* This is compression code. */
|
|
|
|
if (secondary == UCOL_COMMON2) {
|
|
|
|
++count2;
|
2001-09-24 00:20:59 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if (count2 > 0) {
|
|
|
|
if (secondary > UCOL_COMMON2) { // not necessary for 4th level.
|
|
|
|
while (count2 > UCOL_TOP_COUNT2) {
|
|
|
|
addWithIncrement(primaries, primEnd, size, (uint8_t)(UCOL_COMMON_TOP2 - UCOL_TOP_COUNT2));
|
|
|
|
count2 -= (uint32_t)UCOL_TOP_COUNT2;
|
|
|
|
}
|
|
|
|
addWithIncrement(primaries, primEnd, size, (uint8_t)(UCOL_COMMON_TOP2 - (count2-1)));
|
|
|
|
} else {
|
|
|
|
while (count2 > UCOL_BOT_COUNT2) {
|
|
|
|
addWithIncrement(primaries, primEnd, size, (uint8_t)(UCOL_COMMON_BOT2 + UCOL_BOT_COUNT2));
|
|
|
|
count2 -= (uint32_t)UCOL_BOT_COUNT2;
|
|
|
|
}
|
|
|
|
addWithIncrement(primaries, primEnd, size, (uint8_t)(UCOL_COMMON_BOT2 + (count2-1)));
|
|
|
|
}
|
|
|
|
count2 = 0;
|
|
|
|
}
|
|
|
|
addWithIncrement(primaries, primEnd, size, secondary);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (count2 > 0) {
|
|
|
|
while (count2 > UCOL_BOT_COUNT2) {
|
2002-07-27 05:16:44 +00:00
|
|
|
addWithIncrement(primaries, primEnd, size, (uint8_t)(UCOL_COMMON_BOT2 + UCOL_BOT_COUNT2));
|
2001-09-24 00:20:59 +00:00
|
|
|
count2 -= (uint32_t)UCOL_BOT_COUNT2;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
addWithIncrement(primaries, primEnd, size, (uint8_t)(UCOL_COMMON_BOT2 + (count2-1)));
|
2001-09-24 00:20:59 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
*secsize = size;
|
|
|
|
return primaries;
|
2001-09-24 00:20:59 +00:00
|
|
|
}
|
|
|
|
|
2008-02-23 07:21:00 +00:00
|
|
|
#define DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY 0
|
2008-02-19 20:10:55 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* This is the sortkey work horse function */
|
2002-07-11 17:06:51 +00:00
|
|
|
U_CFUNC int32_t U_CALLCONV
|
2001-01-15 07:28:54 +00:00
|
|
|
ucol_calcSortKey(const UCollator *coll,
|
2001-01-05 00:47:25 +00:00
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
uint8_t **result,
|
2001-02-28 19:01:23 +00:00
|
|
|
uint32_t resultLength,
|
2001-11-14 21:55:21 +00:00
|
|
|
UBool allocateSKBuffer,
|
2001-01-05 06:36:10 +00:00
|
|
|
UErrorCode *status)
|
2001-01-05 00:47:25 +00:00
|
|
|
{
|
2005-07-06 21:27:21 +00:00
|
|
|
//const UCAConstants *UCAconsts = (UCAConstants *)((uint8_t *)coll->UCA->image + coll->image->UCAConsts);
|
2003-09-24 00:17:15 +00:00
|
|
|
|
2001-01-05 06:36:10 +00:00
|
|
|
uint32_t i = 0; /* general purpose counter */
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-01-05 06:36:10 +00:00
|
|
|
/* Stack allocated buffers for buffers we use */
|
2001-03-14 07:49:03 +00:00
|
|
|
uint8_t prim[UCOL_PRIMARY_MAX_BUFFER], second[UCOL_SECONDARY_MAX_BUFFER], tert[UCOL_TERTIARY_MAX_BUFFER], caseB[UCOL_CASE_MAX_BUFFER], quad[UCOL_QUAD_MAX_BUFFER];
|
1999-10-22 00:43:39 +00:00
|
|
|
|
2001-01-05 06:36:10 +00:00
|
|
|
uint8_t *primaries = *result, *secondaries = second, *tertiaries = tert, *cases = caseB, *quads = quad;
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-01-05 06:36:10 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
2008-02-13 09:35:50 +00:00
|
|
|
return 0;
|
2001-01-05 06:36:10 +00:00
|
|
|
}
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-11-14 21:55:21 +00:00
|
|
|
if(primaries == NULL && allocateSKBuffer == TRUE) {
|
2001-03-14 07:49:03 +00:00
|
|
|
primaries = *result = prim;
|
|
|
|
resultLength = UCOL_PRIMARY_MAX_BUFFER;
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
uint32_t secSize = UCOL_SECONDARY_MAX_BUFFER, terSize = UCOL_TERTIARY_MAX_BUFFER,
|
2001-03-14 07:49:03 +00:00
|
|
|
caseSize = UCOL_CASE_MAX_BUFFER, quadSize = UCOL_QUAD_MAX_BUFFER;
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
uint32_t sortKeySize = 1; /* it is always \0 terminated */
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-03-14 07:49:03 +00:00
|
|
|
UChar normBuffer[UCOL_NORMALIZATION_MAX_BUFFER];
|
2001-01-05 00:47:25 +00:00
|
|
|
UChar *normSource = normBuffer;
|
2001-03-14 07:49:03 +00:00
|
|
|
int32_t normSourceLen = UCOL_NORMALIZATION_MAX_BUFFER;
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-03-03 03:35:17 +00:00
|
|
|
int32_t len = (sourceLength == -1 ? u_strlen(source) : sourceLength);
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-01-09 00:52:18 +00:00
|
|
|
UColAttributeValue strength = coll->strength;
|
1999-08-16 21:50:52 +00:00
|
|
|
|
2001-02-28 19:30:41 +00:00
|
|
|
uint8_t compareSec = (uint8_t)((strength >= UCOL_SECONDARY)?0:0xFF);
|
|
|
|
uint8_t compareTer = (uint8_t)((strength >= UCOL_TERTIARY)?0:0xFF);
|
|
|
|
uint8_t compareQuad = (uint8_t)((strength >= UCOL_QUATERNARY)?0:0xFF);
|
2001-01-05 00:47:25 +00:00
|
|
|
UBool compareIdent = (strength == UCOL_IDENTICAL);
|
2001-01-09 00:52:18 +00:00
|
|
|
UBool doCase = (coll->caseLevel == UCOL_ON);
|
2001-01-16 00:28:40 +00:00
|
|
|
UBool isFrenchSec = (coll->frenchCollation == UCOL_ON) && (compareSec == 0);
|
2001-03-15 02:49:35 +00:00
|
|
|
UBool shifted = (coll->alternateHandling == UCOL_SHIFTED);
|
2001-10-20 01:09:31 +00:00
|
|
|
//UBool qShifted = shifted && (compareQuad == 0);
|
2001-10-10 01:48:36 +00:00
|
|
|
UBool doHiragana = (coll->hiraganaQ == UCOL_ON) && (compareQuad == 0);
|
2006-05-15 05:08:49 +00:00
|
|
|
/*const uint8_t *scriptOrder = coll->scriptOrder;*/
|
2001-01-08 06:51:18 +00:00
|
|
|
|
2001-10-02 16:49:57 +00:00
|
|
|
uint32_t variableTopValue = coll->variableTopValue;
|
2001-10-05 02:07:51 +00:00
|
|
|
// TODO: UCOL_COMMON_BOT4 should be a function of qShifted. If we have no
|
|
|
|
// qShifted, we don't need to set UCOL_COMMON_BOT4 so high.
|
2001-10-02 16:49:57 +00:00
|
|
|
uint8_t UCOL_COMMON_BOT4 = (uint8_t)((coll->variableTopValue>>8)+1);
|
2001-10-05 02:07:51 +00:00
|
|
|
uint8_t UCOL_HIRAGANA_QUAD = 0;
|
2001-10-02 16:49:57 +00:00
|
|
|
if(doHiragana) {
|
2007-12-26 19:27:43 +00:00
|
|
|
UCOL_HIRAGANA_QUAD=UCOL_COMMON_BOT4++;
|
|
|
|
/* allocate one more space for hiragana, value for hiragana */
|
2001-10-02 16:49:57 +00:00
|
|
|
}
|
|
|
|
uint8_t UCOL_BOT_COUNT4 = (uint8_t)(0xFF - UCOL_COMMON_BOT4);
|
|
|
|
|
2001-01-08 06:51:18 +00:00
|
|
|
/* support for special features like caselevel and funky secondaries */
|
|
|
|
uint8_t *frenchStartPtr = NULL;
|
|
|
|
uint8_t *frenchEndPtr = NULL;
|
|
|
|
uint32_t caseShift = 0;
|
2000-12-07 07:22:55 +00:00
|
|
|
|
2001-10-02 16:49:57 +00:00
|
|
|
sortKeySize += ((compareSec?0:1) + (compareTer?0:1) + (doCase?1:0) + /*(qShifted?1:0)*/(compareQuad?0:1) + (compareIdent?1:0));
|
2000-12-07 07:22:55 +00:00
|
|
|
|
2001-09-27 01:01:30 +00:00
|
|
|
/* If we need to normalize, we'll do it all at once at the beginning! */
|
|
|
|
UNormalizationMode normMode;
|
2001-03-14 02:45:39 +00:00
|
|
|
if(compareIdent) {
|
2001-09-27 01:01:30 +00:00
|
|
|
normMode = UNORM_NFD;
|
|
|
|
} else if(coll->normalizationMode != UCOL_OFF) {
|
|
|
|
normMode = UNORM_FCD;
|
|
|
|
} else {
|
|
|
|
normMode = UNORM_NONE;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(normMode != UNORM_NONE && UNORM_YES != unorm_quickCheck(source, len, normMode, status)) {
|
2001-10-19 17:36:02 +00:00
|
|
|
len = unorm_internalNormalize(normSource, normSourceLen,
|
2001-09-27 01:01:30 +00:00
|
|
|
source, len,
|
|
|
|
normMode, FALSE,
|
|
|
|
status);
|
2001-10-19 17:36:02 +00:00
|
|
|
if(*status == U_BUFFER_OVERFLOW_ERROR) {
|
|
|
|
normSourceLen = len;
|
|
|
|
normSource = (UChar *)uprv_malloc(len*U_SIZEOF_UCHAR);
|
|
|
|
if(normSource == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
len = unorm_internalNormalize(normSource, normSourceLen,
|
|
|
|
source, len,
|
|
|
|
normMode, FALSE,
|
|
|
|
status);
|
|
|
|
}
|
|
|
|
|
2001-01-05 06:36:10 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
2001-09-27 01:01:30 +00:00
|
|
|
return 0;
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
2001-09-27 01:01:30 +00:00
|
|
|
source = normSource;
|
|
|
|
}
|
2000-12-07 07:22:55 +00:00
|
|
|
|
2001-09-27 01:01:30 +00:00
|
|
|
collIterate s;
|
|
|
|
IInit_collIterate(coll, (UChar *)source, len, &s);
|
|
|
|
if(source == normSource) {
|
|
|
|
s.flags &= ~UCOL_ITER_NORM;
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
2000-12-07 07:22:55 +00:00
|
|
|
|
2001-03-23 07:48:40 +00:00
|
|
|
if(resultLength == 0 || primaries == NULL) {
|
2008-02-13 09:35:50 +00:00
|
|
|
int32_t keyLen = ucol_getSortKeySize(coll, &s, sortKeySize, strength, len);
|
|
|
|
if(normSource != normBuffer) {
|
|
|
|
uprv_free(normSource);
|
|
|
|
}
|
|
|
|
return keyLen;
|
2000-12-07 07:22:55 +00:00
|
|
|
}
|
2006-07-06 06:30:06 +00:00
|
|
|
uint8_t *primarySafeEnd = primaries + resultLength - 1;
|
|
|
|
if(strength > UCOL_PRIMARY) {
|
|
|
|
primarySafeEnd--;
|
|
|
|
}
|
2000-12-07 07:22:55 +00:00
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
uint32_t minBufferSize = UCOL_MAX_BUFFER;
|
2000-11-07 00:00:17 +00:00
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
uint8_t *primStart = primaries;
|
|
|
|
uint8_t *secStart = secondaries;
|
|
|
|
uint8_t *terStart = tertiaries;
|
2001-01-05 06:36:10 +00:00
|
|
|
uint8_t *caseStart = cases;
|
|
|
|
uint8_t *quadStart = quads;
|
2000-11-22 23:52:43 +00:00
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
uint32_t order = 0;
|
2000-11-22 00:04:03 +00:00
|
|
|
|
2001-01-08 06:51:18 +00:00
|
|
|
uint8_t primary1 = 0;
|
|
|
|
uint8_t primary2 = 0;
|
2001-01-05 00:47:25 +00:00
|
|
|
uint8_t secondary = 0;
|
|
|
|
uint8_t tertiary = 0;
|
2001-03-02 00:19:43 +00:00
|
|
|
uint8_t caseSwitch = coll->caseSwitch;
|
|
|
|
uint8_t tertiaryMask = coll->tertiaryMask;
|
2007-12-18 01:19:16 +00:00
|
|
|
int8_t tertiaryAddition = coll->tertiaryAddition;
|
2001-05-10 22:33:50 +00:00
|
|
|
uint8_t tertiaryTop = coll->tertiaryTop;
|
2001-05-17 23:14:44 +00:00
|
|
|
uint8_t tertiaryBottom = coll->tertiaryBottom;
|
2001-05-10 22:33:50 +00:00
|
|
|
uint8_t tertiaryCommon = coll->tertiaryCommon;
|
|
|
|
uint8_t caseBits = 0;
|
2000-11-22 00:04:03 +00:00
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
UBool finished = FALSE;
|
2001-01-24 16:18:48 +00:00
|
|
|
UBool wasShifted = FALSE;
|
2001-02-06 00:36:48 +00:00
|
|
|
UBool notIsContinuation = FALSE;
|
2000-11-22 23:52:43 +00:00
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
uint32_t prevBuffSize = 0;
|
2001-01-15 07:28:54 +00:00
|
|
|
|
2001-01-23 06:58:22 +00:00
|
|
|
uint32_t count2 = 0, count3 = 0, count4 = 0;
|
2001-05-03 23:33:29 +00:00
|
|
|
uint8_t leadPrimary = 0;
|
2001-01-18 00:46:19 +00:00
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
for(;;) {
|
|
|
|
for(i=prevBuffSize; i<minBufferSize; ++i) {
|
2000-11-07 00:00:17 +00:00
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
order = ucol_IGetNextCE(coll, &s, status);
|
2001-01-24 16:18:48 +00:00
|
|
|
if(order == UCOL_NO_MORE_CES) {
|
2001-01-05 00:47:25 +00:00
|
|
|
finished = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
2000-11-30 23:20:14 +00:00
|
|
|
|
2001-05-29 04:59:29 +00:00
|
|
|
if(order == 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
2001-03-15 02:49:35 +00:00
|
|
|
}
|
|
|
|
|
2001-05-29 04:59:29 +00:00
|
|
|
notIsContinuation = !isContinuation(order);
|
2001-01-08 06:51:18 +00:00
|
|
|
|
2001-05-10 22:33:50 +00:00
|
|
|
if(notIsContinuation) {
|
2007-12-26 19:27:43 +00:00
|
|
|
tertiary = (uint8_t)(order & UCOL_BYTE_SIZE_MASK);
|
2001-05-10 22:33:50 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
tertiary = (uint8_t)((order & UCOL_REMOVE_CONTINUATION));
|
2001-03-16 19:06:07 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-03-14 07:49:03 +00:00
|
|
|
secondary = (uint8_t)((order >>= 8) & UCOL_BYTE_SIZE_MASK);
|
|
|
|
primary2 = (uint8_t)((order >>= 8) & UCOL_BYTE_SIZE_MASK);
|
2001-05-29 04:59:29 +00:00
|
|
|
primary1 = (uint8_t)(order >> 8);
|
2001-01-12 08:18:12 +00:00
|
|
|
|
2006-05-15 05:08:49 +00:00
|
|
|
/*if(notIsContinuation && scriptOrder != NULL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
primary1 = scriptOrder[primary1];
|
2006-05-15 05:08:49 +00:00
|
|
|
}*/
|
2001-01-08 06:51:18 +00:00
|
|
|
|
2001-06-26 22:24:10 +00:00
|
|
|
if(shifted && ((notIsContinuation && order <= variableTopValue && primary1 > 0)
|
2007-12-26 19:27:43 +00:00
|
|
|
|| (!notIsContinuation && wasShifted))
|
2007-12-26 21:45:18 +00:00
|
|
|
|| (wasShifted && primary1 == 0)) /* amendment to the UCA says that primary ignorables */
|
|
|
|
{
|
|
|
|
/* and other ignorables should be removed if following a shifted code point */
|
|
|
|
if(primary1 == 0) { /* if we were shifted and we got an ignorable code point */
|
|
|
|
/* we should just completely ignore it */
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if(compareQuad == 0) {
|
|
|
|
if(count4 > 0) {
|
|
|
|
while (count4 > UCOL_BOT_COUNT4) {
|
|
|
|
*quads++ = (uint8_t)(UCOL_COMMON_BOT4 + UCOL_BOT_COUNT4);
|
|
|
|
count4 -= UCOL_BOT_COUNT4;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2007-12-26 21:45:18 +00:00
|
|
|
*quads++ = (uint8_t)(UCOL_COMMON_BOT4 + (count4-1));
|
|
|
|
count4 = 0;
|
|
|
|
}
|
|
|
|
/* We are dealing with a variable and we're treating them as shifted */
|
|
|
|
/* This is a shifted ignorable */
|
|
|
|
if(primary1 != 0) { /* we need to check this since we could be in continuation */
|
|
|
|
*quads++ = primary1;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2007-12-26 21:45:18 +00:00
|
|
|
if(primary2 != 0) {
|
|
|
|
*quads++ = primary2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
wasShifted = TRUE;
|
2001-01-08 06:51:18 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
wasShifted = FALSE;
|
|
|
|
/* Note: This code assumes that the table is well built i.e. not having 0 bytes where they are not supposed to be. */
|
|
|
|
/* Usually, we'll have non-zero primary1 & primary2, except in cases of LatinOne and friends, when primary2 will */
|
|
|
|
/* regular and simple sortkey calc */
|
|
|
|
if(primary1 != UCOL_IGNORABLE) {
|
|
|
|
if(notIsContinuation) {
|
|
|
|
if(leadPrimary == primary1) {
|
|
|
|
*primaries++ = primary2;
|
|
|
|
} else {
|
|
|
|
if(leadPrimary != 0) {
|
|
|
|
*primaries++ = (uint8_t)((primary1 > leadPrimary) ? UCOL_BYTE_UNSHIFTED_MAX : UCOL_BYTE_UNSHIFTED_MIN);
|
|
|
|
}
|
|
|
|
if(primary2 == UCOL_IGNORABLE) {
|
|
|
|
/* one byter, not compressed */
|
|
|
|
*primaries++ = primary1;
|
|
|
|
leadPrimary = 0;
|
|
|
|
} else if(primary1<UCOL_BYTE_FIRST_NON_LATIN_PRIMARY ||
|
|
|
|
//(primary1 > (*UCAconsts->UCA_LAST_NON_VARIABLE>>24) && primary1 < (*UCAconsts->UCA_FIRST_IMPLICIT>>24))) {
|
|
|
|
(primary1 > maxRegularPrimary && primary1 < minImplicitPrimary)) {
|
|
|
|
/* not compressible */
|
|
|
|
leadPrimary = 0;
|
|
|
|
*primaries++ = primary1;
|
2008-06-11 21:37:03 +00:00
|
|
|
if(primaries <= primarySafeEnd) {
|
|
|
|
*primaries++ = primary2;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else { /* compress */
|
|
|
|
*primaries++ = leadPrimary = primary1;
|
2008-06-11 21:37:03 +00:00
|
|
|
if(primaries <= primarySafeEnd) {
|
|
|
|
*primaries++ = primary2;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { /* we are in continuation, so we're gonna add primary to the key don't care about compression */
|
2001-04-25 18:00:41 +00:00
|
|
|
*primaries++ = primary1;
|
2008-06-11 21:37:03 +00:00
|
|
|
if((primary2 != UCOL_IGNORABLE) && (primaries <= primarySafeEnd)) {
|
|
|
|
*primaries++ = primary2; /* second part */
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
}
|
2001-04-23 23:41:46 +00:00
|
|
|
}
|
2001-01-08 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(secondary > compareSec) {
|
|
|
|
if(!isFrenchSec) {
|
|
|
|
/* This is compression code. */
|
|
|
|
if (secondary == UCOL_COMMON2 && notIsContinuation) {
|
|
|
|
++count2;
|
|
|
|
} else {
|
|
|
|
if (count2 > 0) {
|
|
|
|
if (secondary > UCOL_COMMON2) { // not necessary for 4th level.
|
|
|
|
while (count2 > UCOL_TOP_COUNT2) {
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_TOP2 - UCOL_TOP_COUNT2);
|
|
|
|
count2 -= (uint32_t)UCOL_TOP_COUNT2;
|
|
|
|
}
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_TOP2 - (count2-1));
|
|
|
|
} else {
|
|
|
|
while (count2 > UCOL_BOT_COUNT2) {
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + UCOL_BOT_COUNT2);
|
|
|
|
count2 -= (uint32_t)UCOL_BOT_COUNT2;
|
|
|
|
}
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + (count2-1));
|
|
|
|
}
|
|
|
|
count2 = 0;
|
|
|
|
}
|
|
|
|
*secondaries++ = secondary;
|
|
|
|
}
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
*secondaries++ = secondary;
|
|
|
|
/* Do the special handling for French secondaries */
|
|
|
|
/* We need to get continuation elements and do intermediate restore */
|
|
|
|
/* abc1c2c3de with french secondaries need to be edc1c2c3ba NOT edc3c2c1ba */
|
|
|
|
if(notIsContinuation) {
|
|
|
|
if (frenchStartPtr != NULL) {
|
|
|
|
/* reverse secondaries from frenchStartPtr up to frenchEndPtr */
|
|
|
|
uprv_ucol_reverse_buffer(uint8_t, frenchStartPtr, frenchEndPtr);
|
|
|
|
frenchStartPtr = NULL;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (frenchStartPtr == NULL) {
|
|
|
|
frenchStartPtr = secondaries - 2;
|
|
|
|
}
|
|
|
|
frenchEndPtr = secondaries-1;
|
|
|
|
}
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(doCase && (primary1 > 0 || strength >= UCOL_SECONDARY)) {
|
|
|
|
// do the case level if we need to do it. We don't want to calculate
|
|
|
|
// case level for primary ignorables if we have only primary strength and case level
|
|
|
|
// otherwise we would break well formedness of CEs
|
|
|
|
doCaseShift(&cases, caseShift);
|
|
|
|
if(notIsContinuation) {
|
|
|
|
caseBits = (uint8_t)(tertiary & 0xC0);
|
|
|
|
|
|
|
|
if(tertiary != 0) {
|
|
|
|
if(coll->caseFirst == UCOL_UPPER_FIRST) {
|
|
|
|
if((caseBits & 0xC0) == 0) {
|
|
|
|
*(cases-1) |= 1 << (--caseShift);
|
|
|
|
} else {
|
|
|
|
*(cases-1) |= 0 << (--caseShift);
|
|
|
|
/* second bit */
|
|
|
|
doCaseShift(&cases, caseShift);
|
|
|
|
*(cases-1) |= ((caseBits>>6)&1) << (--caseShift);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if((caseBits & 0xC0) == 0) {
|
|
|
|
*(cases-1) |= 0 << (--caseShift);
|
|
|
|
} else {
|
|
|
|
*(cases-1) |= 1 << (--caseShift);
|
|
|
|
/* second bit */
|
|
|
|
doCaseShift(&cases, caseShift);
|
|
|
|
*(cases-1) |= ((caseBits>>7)&1) << (--caseShift);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2001-05-10 22:33:50 +00:00
|
|
|
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
if(notIsContinuation) {
|
|
|
|
tertiary ^= caseSwitch;
|
|
|
|
}
|
2001-01-08 06:51:18 +00:00
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
tertiary &= tertiaryMask;
|
|
|
|
if(tertiary > compareTer) {
|
|
|
|
/* This is compression code. */
|
|
|
|
/* sequence size check is included in the if clause */
|
|
|
|
if (tertiary == tertiaryCommon && notIsContinuation) {
|
|
|
|
++count3;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(tertiary > tertiaryCommon && tertiaryCommon == UCOL_COMMON3_NORMAL) {
|
|
|
|
tertiary += tertiaryAddition;
|
|
|
|
} else if(tertiary <= tertiaryCommon && tertiaryCommon == UCOL_COMMON3_UPPERFIRST) {
|
|
|
|
tertiary -= tertiaryAddition;
|
|
|
|
}
|
|
|
|
if (count3 > 0) {
|
|
|
|
if ((tertiary > tertiaryCommon)) {
|
|
|
|
while (count3 > coll->tertiaryTopCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - coll->tertiaryTopCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryTopCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - (count3-1));
|
|
|
|
} else {
|
|
|
|
while (count3 > coll->tertiaryBottomCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + coll->tertiaryBottomCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryBottomCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + (count3-1));
|
|
|
|
}
|
|
|
|
count3 = 0;
|
|
|
|
}
|
|
|
|
*tertiaries++ = tertiary;
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2001-01-12 08:18:12 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(/*qShifted*/(compareQuad==0) && notIsContinuation) {
|
|
|
|
if(s.flags & UCOL_WAS_HIRAGANA) { // This was Hiragana and we need to note it
|
|
|
|
if(count4>0) { // Close this part
|
|
|
|
while (count4 > UCOL_BOT_COUNT4) {
|
|
|
|
*quads++ = (uint8_t)(UCOL_COMMON_BOT4 + UCOL_BOT_COUNT4);
|
|
|
|
count4 -= UCOL_BOT_COUNT4;
|
|
|
|
}
|
|
|
|
*quads++ = (uint8_t)(UCOL_COMMON_BOT4 + (count4-1));
|
|
|
|
count4 = 0;
|
|
|
|
}
|
|
|
|
*quads++ = UCOL_HIRAGANA_QUAD; // Add the Hiragana
|
|
|
|
} else { // This wasn't Hiragana, so we can continue adding stuff
|
|
|
|
count4++;
|
2001-10-05 02:07:51 +00:00
|
|
|
}
|
|
|
|
}
|
2001-01-08 06:51:18 +00:00
|
|
|
}
|
|
|
|
|
2001-02-06 00:36:48 +00:00
|
|
|
if(primaries > primarySafeEnd) { /* We have stepped over the primary buffer */
|
2007-12-26 19:27:43 +00:00
|
|
|
if(allocateSKBuffer == FALSE) { /* need to save our butts if we cannot reallocate */
|
|
|
|
IInit_collIterate(coll, (UChar *)source, len, &s);
|
|
|
|
if(source == normSource) {
|
|
|
|
s.flags &= ~UCOL_ITER_NORM;
|
|
|
|
}
|
|
|
|
sortKeySize = ucol_getSortKeySize(coll, &s, sortKeySize, strength, len);
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
finished = TRUE;
|
|
|
|
break;
|
|
|
|
} else { /* It's much nicer if we can actually reallocate */
|
|
|
|
int32_t sks = sortKeySize+(primaries - primStart)+(secondaries - secStart)+(tertiaries - terStart)+(cases-caseStart)+(quads-quadStart);
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sks, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
primarySafeEnd = primStart + resultLength - 1;
|
|
|
|
if(strength > UCOL_PRIMARY) {
|
|
|
|
primarySafeEnd--;
|
|
|
|
}
|
|
|
|
} else {
|
2008-02-19 20:10:55 +00:00
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
2007-12-26 19:27:43 +00:00
|
|
|
finished = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
2002-07-27 05:16:44 +00:00
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if(finished) {
|
|
|
|
break;
|
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
prevBuffSize = minBufferSize;
|
2007-02-16 19:06:24 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t frenchStartOffset = 0, frenchEndOffset = 0;
|
|
|
|
if (frenchStartPtr != NULL) {
|
|
|
|
frenchStartOffset = frenchStartPtr - secStart;
|
|
|
|
frenchEndOffset = frenchEndPtr - secStart;
|
|
|
|
}
|
|
|
|
secStart = reallocateBuffer(&secondaries, secStart, second, &secSize, 2*secSize, status);
|
|
|
|
terStart = reallocateBuffer(&tertiaries, terStart, tert, &terSize, 2*terSize, status);
|
|
|
|
caseStart = reallocateBuffer(&cases, caseStart, caseB, &caseSize, 2*caseSize, status);
|
|
|
|
quadStart = reallocateBuffer(&quads, quadStart, quad, &quadSize, 2*quadSize, status);
|
2008-02-19 20:10:55 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
2002-08-02 06:55:51 +00:00
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
if (frenchStartPtr != NULL) {
|
|
|
|
frenchStartPtr = secStart + frenchStartOffset;
|
|
|
|
frenchEndPtr = secStart + frenchEndOffset;
|
|
|
|
}
|
|
|
|
minBufferSize *= 2;
|
2000-11-30 23:20:14 +00:00
|
|
|
}
|
|
|
|
}
|
2000-12-14 01:11:11 +00:00
|
|
|
|
2001-03-14 07:49:03 +00:00
|
|
|
/* Here, we are generally done with processing */
|
|
|
|
/* bailing out would not be too productive */
|
|
|
|
|
2001-01-05 06:36:10 +00:00
|
|
|
if(U_SUCCESS(*status)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
sortKeySize += (primaries - primStart);
|
|
|
|
/* we have done all the CE's, now let's put them together to form a key */
|
|
|
|
if(compareSec == 0) {
|
|
|
|
if (count2 > 0) {
|
|
|
|
while (count2 > UCOL_BOT_COUNT2) {
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + UCOL_BOT_COUNT2);
|
|
|
|
count2 -= (uint32_t)UCOL_BOT_COUNT2;
|
|
|
|
}
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + (count2-1));
|
|
|
|
}
|
|
|
|
uint32_t secsize = secondaries-secStart;
|
|
|
|
if(!isFrenchSec) { // Regular situation, we know the length of secondaries
|
|
|
|
sortKeySize += secsize;
|
|
|
|
if(sortKeySize <= resultLength) {
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, secStart, secsize);
|
|
|
|
primaries += secsize;
|
|
|
|
} else {
|
|
|
|
if(allocateSKBuffer == TRUE) { /* need to save our butts if we cannot reallocate */
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, secStart, secsize);
|
|
|
|
primaries += secsize;
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { // French secondary is on. We will need to pack French. packFrench will add the level terminator
|
|
|
|
uint8_t *newPrim = packFrench(primaries, primStart+resultLength, secondaries, &secsize, frenchStartPtr, frenchEndPtr);
|
|
|
|
sortKeySize += secsize;
|
|
|
|
if(sortKeySize <= resultLength) { // if we managed to pack fine
|
|
|
|
primaries = newPrim; // update the primary pointer
|
|
|
|
} else { // overflow, need to reallocate and redo
|
|
|
|
if(allocateSKBuffer == TRUE) { /* need to save our butts if we cannot reallocate */
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
primaries = packFrench(primaries, primStart+resultLength, secondaries, &secsize, frenchStartPtr, frenchEndPtr);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
|
|
|
|
if(doCase) {
|
|
|
|
uint32_t casesize = cases - caseStart;
|
|
|
|
sortKeySize += casesize;
|
|
|
|
if(sortKeySize <= resultLength) {
|
2002-07-27 05:16:44 +00:00
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
2007-12-26 19:27:43 +00:00
|
|
|
uprv_memcpy(primaries, caseStart, casesize);
|
|
|
|
primaries += casesize;
|
2001-04-23 01:53:49 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(allocateSKBuffer == TRUE) {
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, caseStart, casesize);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
2001-03-14 07:49:03 +00:00
|
|
|
}
|
2002-07-27 05:16:44 +00:00
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(compareTer == 0) {
|
|
|
|
if (count3 > 0) {
|
|
|
|
if (coll->tertiaryCommon != UCOL_COMMON_BOT3) {
|
|
|
|
while (count3 >= coll->tertiaryTopCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - coll->tertiaryTopCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryTopCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - count3);
|
|
|
|
} else {
|
|
|
|
while (count3 > coll->tertiaryBottomCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + coll->tertiaryBottomCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryBottomCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + (count3-1));
|
|
|
|
}
|
2002-07-27 05:16:44 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t tersize = tertiaries - terStart;
|
|
|
|
sortKeySize += tersize;
|
2002-07-27 05:16:44 +00:00
|
|
|
if(sortKeySize <= resultLength) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, terStart, tersize);
|
|
|
|
primaries += tersize;
|
2002-07-27 05:16:44 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(allocateSKBuffer == TRUE) {
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, terStart, tersize);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
2001-03-14 07:49:03 +00:00
|
|
|
}
|
2002-07-27 05:16:44 +00:00
|
|
|
}
|
2001-01-05 06:36:10 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(compareQuad == 0/*qShifted == TRUE*/) {
|
|
|
|
if(count4 > 0) {
|
|
|
|
while (count4 > UCOL_BOT_COUNT4) {
|
|
|
|
*quads++ = (uint8_t)(UCOL_COMMON_BOT4 + UCOL_BOT_COUNT4);
|
|
|
|
count4 -= UCOL_BOT_COUNT4;
|
|
|
|
}
|
|
|
|
*quads++ = (uint8_t)(UCOL_COMMON_BOT4 + (count4-1));
|
|
|
|
}
|
|
|
|
uint32_t quadsize = quads - quadStart;
|
|
|
|
sortKeySize += quadsize;
|
|
|
|
if(sortKeySize <= resultLength) {
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, quadStart, quadsize);
|
|
|
|
primaries += quadsize;
|
|
|
|
} else {
|
|
|
|
if(allocateSKBuffer == TRUE) {
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, quadStart, quadsize);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
|
|
|
}
|
2001-05-29 04:59:29 +00:00
|
|
|
}
|
2001-01-05 00:47:25 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(compareIdent) {
|
|
|
|
sortKeySize += u_lengthOfIdenticalLevelRun(s.string, len);
|
|
|
|
if(sortKeySize <= resultLength) {
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
primaries += u_writeIdenticalLevelRun(s.string, len, primaries);
|
|
|
|
} else {
|
|
|
|
if(allocateSKBuffer == TRUE) {
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
u_writeIdenticalLevelRun(s.string, len, primaries);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*(primaries++) = '\0';
|
|
|
|
}
|
|
|
|
|
2001-11-14 21:55:21 +00:00
|
|
|
if(allocateSKBuffer == TRUE) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*result = (uint8_t*)uprv_malloc(sortKeySize);
|
|
|
|
/* test for NULL */
|
|
|
|
if (*result == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
2008-02-13 09:35:50 +00:00
|
|
|
goto cleanup;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
uprv_memcpy(*result, primStart, sortKeySize);
|
|
|
|
if(primStart != prim) {
|
|
|
|
uprv_free(primStart);
|
|
|
|
}
|
2001-03-14 07:49:03 +00:00
|
|
|
}
|
|
|
|
|
2008-02-13 09:35:50 +00:00
|
|
|
cleanup:
|
2008-02-19 20:10:55 +00:00
|
|
|
if (allocateSKBuffer == FALSE && resultLength > 0 && U_FAILURE(*status) && *status != U_BUFFER_OVERFLOW_ERROR) {
|
|
|
|
/* NULL terminate for safety */
|
|
|
|
**result = 0;
|
|
|
|
}
|
2008-02-13 09:35:50 +00:00
|
|
|
if(terStart != tert) {
|
|
|
|
uprv_free(terStart);
|
|
|
|
uprv_free(secStart);
|
|
|
|
uprv_free(caseStart);
|
|
|
|
uprv_free(quadStart);
|
|
|
|
}
|
2008-06-09 21:18:46 +00:00
|
|
|
|
|
|
|
/* To avoid memory leak, free the offset buffer if necessary. */
|
|
|
|
freeOffsetBuffer(&s);
|
2008-02-13 09:35:50 +00:00
|
|
|
|
|
|
|
if(normSource != normBuffer) {
|
|
|
|
uprv_free(normSource);
|
|
|
|
}
|
|
|
|
|
2001-01-05 00:47:25 +00:00
|
|
|
return sortKeySize;
|
|
|
|
}
|
|
|
|
|
2001-05-10 22:12:53 +00:00
|
|
|
|
2002-07-11 17:06:51 +00:00
|
|
|
U_CFUNC int32_t U_CALLCONV
|
2001-02-07 00:57:39 +00:00
|
|
|
ucol_calcSortKeySimpleTertiary(const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
uint8_t **result,
|
2001-03-14 02:45:39 +00:00
|
|
|
uint32_t resultLength,
|
2001-11-14 21:55:21 +00:00
|
|
|
UBool allocateSKBuffer,
|
2001-02-07 00:57:39 +00:00
|
|
|
UErrorCode *status)
|
|
|
|
{
|
2001-05-10 22:12:53 +00:00
|
|
|
U_ALIGN_CODE(16);
|
2003-09-24 00:17:15 +00:00
|
|
|
|
2005-07-06 21:27:21 +00:00
|
|
|
//const UCAConstants *UCAconsts = (UCAConstants *)((uint8_t *)coll->UCA->image + coll->image->UCAConsts);
|
2001-02-07 00:57:39 +00:00
|
|
|
uint32_t i = 0; /* general purpose counter */
|
|
|
|
|
|
|
|
/* Stack allocated buffers for buffers we use */
|
2001-03-14 07:49:03 +00:00
|
|
|
uint8_t prim[UCOL_PRIMARY_MAX_BUFFER], second[UCOL_SECONDARY_MAX_BUFFER], tert[UCOL_TERTIARY_MAX_BUFFER];
|
2001-02-07 00:57:39 +00:00
|
|
|
|
|
|
|
uint8_t *primaries = *result, *secondaries = second, *tertiaries = tert;
|
|
|
|
|
|
|
|
if(U_FAILURE(*status)) {
|
2007-09-24 16:05:53 +00:00
|
|
|
return 0;
|
2001-02-07 00:57:39 +00:00
|
|
|
}
|
|
|
|
|
2001-11-14 21:55:21 +00:00
|
|
|
if(primaries == NULL && allocateSKBuffer == TRUE) {
|
2001-03-14 07:49:03 +00:00
|
|
|
primaries = *result = prim;
|
|
|
|
resultLength = UCOL_PRIMARY_MAX_BUFFER;
|
2001-02-07 00:57:39 +00:00
|
|
|
}
|
|
|
|
|
2001-03-14 07:49:03 +00:00
|
|
|
uint32_t secSize = UCOL_SECONDARY_MAX_BUFFER, terSize = UCOL_TERTIARY_MAX_BUFFER;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-03-22 21:16:20 +00:00
|
|
|
uint32_t sortKeySize = 3; /* it is always \0 terminated plus separators for secondary and tertiary */
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-03-14 07:49:03 +00:00
|
|
|
UChar normBuffer[UCOL_NORMALIZATION_MAX_BUFFER];
|
2001-02-07 00:57:39 +00:00
|
|
|
UChar *normSource = normBuffer;
|
2001-03-14 07:49:03 +00:00
|
|
|
int32_t normSourceLen = UCOL_NORMALIZATION_MAX_BUFFER;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-05-10 22:12:53 +00:00
|
|
|
int32_t len = sourceLength;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-09-27 01:01:30 +00:00
|
|
|
/* If we need to normalize, we'll do it all at once at the beginning! */
|
|
|
|
if(coll->normalizationMode != UCOL_OFF && UNORM_YES != unorm_quickCheck(source, len, UNORM_FCD, status)) {
|
2001-10-19 17:36:02 +00:00
|
|
|
len = unorm_internalNormalize(normSource, normSourceLen,
|
2001-09-27 01:01:30 +00:00
|
|
|
source, len,
|
|
|
|
UNORM_FCD, FALSE,
|
|
|
|
status);
|
2001-10-19 17:36:02 +00:00
|
|
|
if(*status == U_BUFFER_OVERFLOW_ERROR) {
|
|
|
|
normSourceLen = len;
|
|
|
|
normSource = (UChar *)uprv_malloc(len*U_SIZEOF_UCHAR);
|
|
|
|
if(normSource == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
len = unorm_internalNormalize(normSource, normSourceLen,
|
|
|
|
source, len,
|
|
|
|
UNORM_FCD, FALSE,
|
|
|
|
status);
|
2007-09-24 16:05:53 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
/* Should never happen. */
|
|
|
|
uprv_free(normSource);
|
|
|
|
normSource = normBuffer;
|
|
|
|
}
|
2001-10-19 17:36:02 +00:00
|
|
|
}
|
|
|
|
|
2001-09-27 01:01:30 +00:00
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
source = normSource;
|
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
|
|
|
|
collIterate s;
|
2001-04-06 23:37:48 +00:00
|
|
|
IInit_collIterate(coll, (UChar *)source, len, &s);
|
2001-09-27 01:01:30 +00:00
|
|
|
if(source == normSource) {
|
|
|
|
s.flags &= ~UCOL_ITER_NORM;
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
|
2001-03-23 07:48:40 +00:00
|
|
|
if(resultLength == 0 || primaries == NULL) {
|
2002-08-02 06:55:51 +00:00
|
|
|
int32_t t = ucol_getSortKeySize(coll, &s, sortKeySize, coll->strength, len);
|
2001-05-17 23:09:35 +00:00
|
|
|
if(normSource != normBuffer) {
|
|
|
|
uprv_free(normSource);
|
2001-11-07 00:01:29 +00:00
|
|
|
}
|
2001-05-17 23:09:35 +00:00
|
|
|
return t;
|
|
|
|
}
|
|
|
|
|
2001-03-23 07:48:40 +00:00
|
|
|
uint8_t *primarySafeEnd = primaries + resultLength - 2;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
uint32_t minBufferSize = UCOL_MAX_BUFFER;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
|
|
|
uint8_t *primStart = primaries;
|
|
|
|
uint8_t *secStart = secondaries;
|
|
|
|
uint8_t *terStart = tertiaries;
|
|
|
|
|
|
|
|
uint32_t order = 0;
|
|
|
|
|
|
|
|
uint8_t primary1 = 0;
|
|
|
|
uint8_t primary2 = 0;
|
|
|
|
uint8_t secondary = 0;
|
|
|
|
uint8_t tertiary = 0;
|
2001-03-02 00:19:43 +00:00
|
|
|
uint8_t caseSwitch = coll->caseSwitch;
|
|
|
|
uint8_t tertiaryMask = coll->tertiaryMask;
|
2007-12-18 01:19:16 +00:00
|
|
|
int8_t tertiaryAddition = coll->tertiaryAddition;
|
2001-05-10 22:33:50 +00:00
|
|
|
uint8_t tertiaryTop = coll->tertiaryTop;
|
2001-05-17 23:14:44 +00:00
|
|
|
uint8_t tertiaryBottom = coll->tertiaryBottom;
|
2001-05-10 22:33:50 +00:00
|
|
|
uint8_t tertiaryCommon = coll->tertiaryCommon;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
uint32_t prevBuffSize = 0;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
|
|
|
UBool finished = FALSE;
|
|
|
|
UBool notIsContinuation = FALSE;
|
|
|
|
|
|
|
|
uint32_t count2 = 0, count3 = 0;
|
2001-05-03 23:33:29 +00:00
|
|
|
uint8_t leadPrimary = 0;
|
2001-02-07 00:57:39 +00:00
|
|
|
|
|
|
|
for(;;) {
|
|
|
|
for(i=prevBuffSize; i<minBufferSize; ++i) {
|
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
order = ucol_IGetNextCE(coll, &s, status);
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-05-29 04:59:29 +00:00
|
|
|
if(order == 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
2001-02-07 00:57:39 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if(order == UCOL_NO_MORE_CES) {
|
|
|
|
finished = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2001-05-29 04:59:29 +00:00
|
|
|
notIsContinuation = !isContinuation(order);
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-03-16 19:06:07 +00:00
|
|
|
if(notIsContinuation) {
|
2007-12-26 19:27:43 +00:00
|
|
|
tertiary = (uint8_t)((order & tertiaryMask));
|
2001-03-16 19:06:07 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
tertiary = (uint8_t)((order & UCOL_REMOVE_CONTINUATION));
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
2001-03-14 07:49:03 +00:00
|
|
|
secondary = (uint8_t)((order >>= 8) & UCOL_BYTE_SIZE_MASK);
|
|
|
|
primary2 = (uint8_t)((order >>= 8) & UCOL_BYTE_SIZE_MASK);
|
2001-05-29 04:59:29 +00:00
|
|
|
primary1 = (uint8_t)(order >> 8);
|
2001-02-07 00:57:39 +00:00
|
|
|
|
|
|
|
/* Note: This code assumes that the table is well built i.e. not having 0 bytes where they are not supposed to be. */
|
|
|
|
/* Usually, we'll have non-zero primary1 & primary2, except in cases of LatinOne and friends, when primary2 will */
|
|
|
|
/* be zero with non zero primary1. primary3 is different than 0 only for long primaries - see above. */
|
2001-05-29 04:59:29 +00:00
|
|
|
/* regular and simple sortkey calc */
|
2001-04-19 23:49:03 +00:00
|
|
|
if(primary1 != UCOL_IGNORABLE) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(notIsContinuation) {
|
|
|
|
if(leadPrimary == primary1) {
|
|
|
|
*primaries++ = primary2;
|
|
|
|
} else {
|
|
|
|
if(leadPrimary != 0) {
|
|
|
|
*primaries++ = (uint8_t)((primary1 > leadPrimary) ? UCOL_BYTE_UNSHIFTED_MAX : UCOL_BYTE_UNSHIFTED_MIN);
|
|
|
|
}
|
|
|
|
if(primary2 == UCOL_IGNORABLE) {
|
|
|
|
/* one byter, not compressed */
|
|
|
|
*primaries++ = primary1;
|
|
|
|
leadPrimary = 0;
|
|
|
|
} else if(primary1<UCOL_BYTE_FIRST_NON_LATIN_PRIMARY ||
|
|
|
|
//(primary1 > (UCOL_RESET_TOP_VALUE>>24) && primary1 < (UCOL_NEXT_TOP_VALUE>>24)))
|
|
|
|
//(primary1 > (*UCAconsts->UCA_LAST_NON_VARIABLE>>24) && primary1 < (*UCAconsts->UCA_FIRST_IMPLICIT>>24))) {
|
|
|
|
(primary1 > maxRegularPrimary && primary1 < minImplicitPrimary)) {
|
|
|
|
/* not compressible */
|
|
|
|
leadPrimary = 0;
|
|
|
|
*primaries++ = primary1;
|
|
|
|
*primaries++ = primary2;
|
|
|
|
} else { /* compress */
|
|
|
|
*primaries++ = leadPrimary = primary1;
|
|
|
|
*primaries++ = primary2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { /* we are in continuation, so we're gonna add primary to the key don't care about compression */
|
|
|
|
*primaries++ = primary1;
|
|
|
|
if(primary2 != UCOL_IGNORABLE) {
|
|
|
|
*primaries++ = primary2; /* second part */
|
|
|
|
}
|
2001-04-23 23:41:46 +00:00
|
|
|
}
|
2001-05-11 01:13:08 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2001-05-29 04:59:29 +00:00
|
|
|
if(secondary > 0) { /* I think that != 0 test should be != IGNORABLE */
|
2007-12-26 19:27:43 +00:00
|
|
|
/* This is compression code. */
|
|
|
|
if (secondary == UCOL_COMMON2 && notIsContinuation) {
|
|
|
|
++count2;
|
|
|
|
} else {
|
|
|
|
if (count2 > 0) {
|
|
|
|
if (secondary > UCOL_COMMON2) { // not necessary for 4th level.
|
|
|
|
while (count2 > UCOL_TOP_COUNT2) {
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_TOP2 - UCOL_TOP_COUNT2);
|
|
|
|
count2 -= (uint32_t)UCOL_TOP_COUNT2;
|
|
|
|
}
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_TOP2 - (count2-1));
|
|
|
|
} else {
|
|
|
|
while (count2 > UCOL_BOT_COUNT2) {
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + UCOL_BOT_COUNT2);
|
|
|
|
count2 -= (uint32_t)UCOL_BOT_COUNT2;
|
|
|
|
}
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + (count2-1));
|
|
|
|
}
|
|
|
|
count2 = 0;
|
2001-03-14 02:45:39 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
*secondaries++ = secondary;
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
}
|
|
|
|
|
2001-05-10 22:33:50 +00:00
|
|
|
if(notIsContinuation) {
|
2007-12-26 19:27:43 +00:00
|
|
|
tertiary ^= caseSwitch;
|
2001-05-10 22:33:50 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(tertiary > 0) {
|
|
|
|
/* This is compression code. */
|
|
|
|
/* sequence size check is included in the if clause */
|
|
|
|
if (tertiary == tertiaryCommon && notIsContinuation) {
|
|
|
|
++count3;
|
|
|
|
} else {
|
|
|
|
if(tertiary > tertiaryCommon && tertiaryCommon == UCOL_COMMON3_NORMAL) {
|
|
|
|
tertiary += tertiaryAddition;
|
|
|
|
} else if (tertiary <= tertiaryCommon && tertiaryCommon == UCOL_COMMON3_UPPERFIRST) {
|
|
|
|
tertiary -= tertiaryAddition;
|
2001-03-14 02:45:39 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if (count3 > 0) {
|
|
|
|
if ((tertiary > tertiaryCommon)) {
|
|
|
|
while (count3 > coll->tertiaryTopCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - coll->tertiaryTopCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryTopCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - (count3-1));
|
|
|
|
} else {
|
|
|
|
while (count3 > coll->tertiaryBottomCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + coll->tertiaryBottomCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryBottomCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + (count3-1));
|
|
|
|
}
|
|
|
|
count3 = 0;
|
|
|
|
}
|
|
|
|
*tertiaries++ = tertiary;
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if(primaries > primarySafeEnd) { /* We have stepped over the primary buffer */
|
2007-12-26 19:27:43 +00:00
|
|
|
if(allocateSKBuffer == FALSE) { /* need to save our butts if we cannot reallocate */
|
|
|
|
IInit_collIterate(coll, (UChar *)source, len, &s);
|
|
|
|
if(source == normSource) {
|
|
|
|
s.flags &= ~UCOL_ITER_NORM;
|
|
|
|
}
|
|
|
|
sortKeySize = ucol_getSortKeySize(coll, &s, sortKeySize, coll->strength, len);
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
finished = TRUE;
|
|
|
|
break;
|
|
|
|
} else { /* It's much nicer if we can actually reallocate */
|
|
|
|
int32_t sks = sortKeySize+(primaries - primStart)+(secondaries - secStart)+(tertiaries - terStart);
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sks, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
primarySafeEnd = primStart + resultLength - 2;
|
|
|
|
} else {
|
2008-02-19 20:10:55 +00:00
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
2007-12-26 19:27:43 +00:00
|
|
|
finished = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(finished) {
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
prevBuffSize = minBufferSize;
|
|
|
|
secStart = reallocateBuffer(&secondaries, secStart, second, &secSize, 2*secSize, status);
|
|
|
|
terStart = reallocateBuffer(&tertiaries, terStart, tert, &terSize, 2*terSize, status);
|
|
|
|
minBufferSize *= 2;
|
|
|
|
if(U_FAILURE(*status)) { // if we cannot reallocate buffers, we can at least give the sortkey size
|
2008-02-19 20:10:55 +00:00
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
2001-02-07 00:57:39 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if(U_SUCCESS(*status)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
sortKeySize += (primaries - primStart);
|
|
|
|
/* we have done all the CE's, now let's put them together to form a key */
|
|
|
|
if (count2 > 0) {
|
|
|
|
while (count2 > UCOL_BOT_COUNT2) {
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + UCOL_BOT_COUNT2);
|
|
|
|
count2 -= (uint32_t)UCOL_BOT_COUNT2;
|
|
|
|
}
|
|
|
|
*secondaries++ = (uint8_t)(UCOL_COMMON_BOT2 + (count2-1));
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t secsize = secondaries-secStart;
|
|
|
|
sortKeySize += secsize;
|
|
|
|
if(sortKeySize <= resultLength) {
|
2002-07-27 05:16:44 +00:00
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, secStart, secsize);
|
2007-12-26 19:27:43 +00:00
|
|
|
primaries += secsize;
|
2001-03-14 07:49:03 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(allocateSKBuffer == TRUE) {
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
*result = primStart;
|
|
|
|
uprv_memcpy(primaries, secStart, secsize);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_BUFFER_OVERFLOW_ERROR;
|
|
|
|
}
|
2001-03-14 07:49:03 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (count3 > 0) {
|
|
|
|
if (coll->tertiaryCommon != UCOL_COMMON3_NORMAL) {
|
|
|
|
while (count3 >= coll->tertiaryTopCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - coll->tertiaryTopCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryTopCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryTop - count3);
|
|
|
|
} else {
|
|
|
|
while (count3 > coll->tertiaryBottomCount) {
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + coll->tertiaryBottomCount);
|
|
|
|
count3 -= (uint32_t)coll->tertiaryBottomCount;
|
|
|
|
}
|
|
|
|
*tertiaries++ = (uint8_t)(tertiaryBottom + (count3-1));
|
|
|
|
}
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t tersize = tertiaries - terStart;
|
|
|
|
sortKeySize += tersize;
|
|
|
|
if(sortKeySize <= resultLength) {
|
2002-07-27 05:16:44 +00:00
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, terStart, tersize);
|
2007-12-26 19:27:43 +00:00
|
|
|
primaries += tersize;
|
2001-03-14 07:49:03 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(allocateSKBuffer == TRUE) {
|
|
|
|
primStart = reallocateBuffer(&primaries, *result, prim, &resultLength, 2*sortKeySize, status);
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
*result = primStart;
|
|
|
|
*(primaries++) = UCOL_LEVELTERMINATOR;
|
|
|
|
uprv_memcpy(primaries, terStart, tersize);
|
|
|
|
}
|
2008-02-19 20:10:55 +00:00
|
|
|
else {
|
|
|
|
/* We ran out of memory!? We can't recover. */
|
|
|
|
sortKeySize = DEFAULT_ERROR_SIZE_FOR_CALCSORTKEY;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
}
|
2001-03-14 07:49:03 +00:00
|
|
|
}
|
2001-02-07 00:57:39 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
*(primaries++) = '\0';
|
2001-02-07 00:57:39 +00:00
|
|
|
}
|
|
|
|
|
2001-11-14 21:55:21 +00:00
|
|
|
if(allocateSKBuffer == TRUE) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*result = (uint8_t*)uprv_malloc(sortKeySize);
|
|
|
|
/* test for NULL */
|
|
|
|
if (*result == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
2008-02-19 20:10:55 +00:00
|
|
|
goto cleanup;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
uprv_memcpy(*result, primStart, sortKeySize);
|
|
|
|
if(primStart != prim) {
|
|
|
|
uprv_free(primStart);
|
|
|
|
}
|
2001-03-14 07:49:03 +00:00
|
|
|
}
|
|
|
|
|
2008-02-19 20:10:55 +00:00
|
|
|
cleanup:
|
|
|
|
if (allocateSKBuffer == FALSE && resultLength > 0 && U_FAILURE(*status) && *status != U_BUFFER_OVERFLOW_ERROR) {
|
|
|
|
/* NULL terminate for safety */
|
|
|
|
**result = 0;
|
|
|
|
}
|
|
|
|
if(terStart != tert) {
|
|
|
|
uprv_free(terStart);
|
|
|
|
uprv_free(secStart);
|
|
|
|
}
|
2008-06-09 21:18:46 +00:00
|
|
|
|
|
|
|
/* To avoid memory leak, free the offset buffer if necessary. */
|
|
|
|
freeOffsetBuffer(&s);
|
|
|
|
|
2008-02-19 20:10:55 +00:00
|
|
|
if(normSource != normBuffer) {
|
|
|
|
uprv_free(normSource);
|
|
|
|
}
|
|
|
|
|
2001-02-07 00:57:39 +00:00
|
|
|
return sortKeySize;
|
|
|
|
}
|
|
|
|
|
2003-02-06 23:29:56 +00:00
|
|
|
static inline
|
|
|
|
UBool isShiftedCE(uint32_t CE, uint32_t LVT, UBool *wasShifted) {
|
2007-12-26 19:27:43 +00:00
|
|
|
UBool notIsContinuation = !isContinuation(CE);
|
|
|
|
uint8_t primary1 = (uint8_t)((CE >> 24) & 0xFF);
|
|
|
|
if(LVT && ((notIsContinuation && (CE & 0xFFFF0000)<= LVT && primary1 > 0)
|
|
|
|
|| (!notIsContinuation && *wasShifted))
|
|
|
|
|| (*wasShifted && primary1 == 0)) /* amendment to the UCA says that primary ignorables */
|
|
|
|
{
|
|
|
|
// The stuff below should probably be in the sortkey code... maybe not...
|
|
|
|
if(primary1 != 0) { /* if we were shifted and we got an ignorable code point */
|
|
|
|
/* we should just completely ignore it */
|
|
|
|
*wasShifted = TRUE;
|
|
|
|
//continue;
|
|
|
|
}
|
|
|
|
//*wasShifted = TRUE;
|
|
|
|
return TRUE;
|
|
|
|
} else {
|
|
|
|
*wasShifted = FALSE;
|
|
|
|
return FALSE;
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
|
|
|
}
|
2003-03-25 18:25:09 +00:00
|
|
|
static inline
|
|
|
|
void terminatePSKLevel(int32_t level, int32_t maxLevel, int32_t &i, uint8_t *dest) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(level < maxLevel) {
|
|
|
|
dest[i++] = UCOL_LEVELTERMINATOR;
|
|
|
|
} else {
|
|
|
|
dest[i++] = 0;
|
|
|
|
}
|
2003-03-25 18:25:09 +00:00
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
|
2003-03-07 07:00:53 +00:00
|
|
|
/** enumeration of level identifiers for partial sort key generation */
|
|
|
|
enum {
|
|
|
|
UCOL_PSK_PRIMARY = 0,
|
|
|
|
UCOL_PSK_SECONDARY = 1,
|
|
|
|
UCOL_PSK_CASE = 2,
|
|
|
|
UCOL_PSK_TERTIARY = 3,
|
|
|
|
UCOL_PSK_QUATERNARY = 4,
|
|
|
|
UCOL_PSK_QUIN = 5, /** This is an extra level, not used - but we have three bits to blow */
|
|
|
|
UCOL_PSK_IDENTICAL = 6,
|
|
|
|
UCOL_PSK_NULL = 7, /** level for the end of sort key. Will just produce zeros */
|
|
|
|
UCOL_PSK_LIMIT
|
|
|
|
};
|
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
/** collation state enum. *_SHIFT value is how much to shift right
|
|
|
|
* to get the state piece to the right. *_MASK value should be
|
2003-03-07 07:00:53 +00:00
|
|
|
* ANDed with the shifted state. This data is stored in state[1]
|
|
|
|
* field.
|
|
|
|
*/
|
|
|
|
enum {
|
|
|
|
UCOL_PSK_LEVEL_SHIFT = 0, /** level identificator. stores an enum value from above */
|
|
|
|
UCOL_PSK_LEVEL_MASK = 7, /** three bits */
|
2003-03-13 17:15:53 +00:00
|
|
|
UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_SHIFT = 3, /** number of bytes of primary or quaternary already written */
|
2003-12-03 00:45:12 +00:00
|
|
|
UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_MASK = 1,
|
2003-03-13 17:15:53 +00:00
|
|
|
/** can be only 0 or 1, since we get up to two bytes from primary or quaternary
|
|
|
|
* This field is also used to denote that the French secondary level is finished
|
|
|
|
*/
|
2003-03-07 07:00:53 +00:00
|
|
|
UCOL_PSK_WAS_SHIFTED_SHIFT = 4,/** was the last value shifted */
|
|
|
|
UCOL_PSK_WAS_SHIFTED_MASK = 1, /** can be 0 or 1 (Boolean) */
|
|
|
|
UCOL_PSK_USED_FRENCH_SHIFT = 5,/** how many French bytes have we already written */
|
|
|
|
UCOL_PSK_USED_FRENCH_MASK = 3, /** up to 4 bytes. See comment just below */
|
|
|
|
/** When we do French we need to reverse secondary values. However, continuations
|
2003-12-03 00:45:12 +00:00
|
|
|
* need to stay the same. So if you had abc1c2c3de, you need to have edc1c2c3ba
|
2003-03-07 07:00:53 +00:00
|
|
|
*/
|
2006-06-16 06:00:13 +00:00
|
|
|
UCOL_PSK_BOCSU_BYTES_SHIFT = 7,
|
|
|
|
UCOL_PSK_BOCSU_BYTES_MASK = 3,
|
|
|
|
UCOL_PSK_CONSUMED_CES_SHIFT = 9,
|
|
|
|
UCOL_PSK_CONSUMED_CES_MASK = 0x7FFFF
|
2003-03-07 07:00:53 +00:00
|
|
|
};
|
|
|
|
|
2006-06-16 06:00:13 +00:00
|
|
|
// macro calculating the number of expansion CEs available
|
|
|
|
#define uprv_numAvailableExpCEs(s) (s).CEpos - (s).toReturn
|
|
|
|
|
2003-03-07 07:00:53 +00:00
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
/** main sortkey part procedure. On the first call,
|
|
|
|
* you should pass in a collator, an iterator, empty state
|
2003-03-13 17:15:53 +00:00
|
|
|
* state[0] == state[1] == 0, a buffer to hold results
|
|
|
|
* number of bytes you need and an error code pointer.
|
|
|
|
* Make sure your buffer is big enough to hold the wanted
|
2003-12-03 00:45:12 +00:00
|
|
|
* number of sortkey bytes. I don't check.
|
|
|
|
* The only meaningful status you can get back is
|
|
|
|
* U_BUFFER_OVERFLOW_ERROR, which basically means that you
|
2003-03-13 17:15:53 +00:00
|
|
|
* have been dealt a raw deal and that you probably won't
|
|
|
|
* be able to use partial sortkey generation for this
|
|
|
|
* particular combination of string and collator. This
|
|
|
|
* is highly unlikely, but you should still check the error code.
|
2003-12-03 00:45:12 +00:00
|
|
|
* Any other status means that you're not in a sane situation
|
|
|
|
* anymore. After the first call, preserve state values and
|
2003-03-13 17:15:53 +00:00
|
|
|
* use them on subsequent calls to obtain more bytes of a sortkey.
|
|
|
|
* Use until the number of bytes written is smaller than the requested
|
|
|
|
* number of bytes. Generated sortkey is not compatible with the
|
|
|
|
* one generated by ucol_getSortKey, as we don't do any compression.
|
|
|
|
* However, levels are still terminated by a 1 (one) and the sortkey
|
|
|
|
* is terminated by a 0 (zero). Identical level is the same as in the
|
2003-12-03 00:45:12 +00:00
|
|
|
* regular sortkey - internal bocu-1 implementation is used.
|
|
|
|
* For curious, although you cannot do much about this, here is
|
2003-03-13 17:15:53 +00:00
|
|
|
* the structure of state words.
|
|
|
|
* state[0] - iterator state. Depends on the iterator implementation,
|
|
|
|
* but allows the iterator to continue where it stopped in
|
|
|
|
* the last iteration.
|
|
|
|
* state[1] - collation processing state. Here is the distribution
|
|
|
|
* of the bits:
|
|
|
|
* 0, 1, 2 - level of the sortkey - primary, secondary, case, tertiary
|
|
|
|
* quaternary, quin (we don't use this one), identical and
|
|
|
|
* null (producing only zeroes - first one to terminate the
|
|
|
|
* sortkey and subsequent to fill the buffer).
|
|
|
|
* 3 - byte count. Number of bytes written on the primary level.
|
|
|
|
* 4 - was shifted. Whether the previous iteration finished in the
|
|
|
|
* shifted state.
|
|
|
|
* 5, 6 - French continuation bytes written. See the comment in the enum
|
2006-06-16 06:00:13 +00:00
|
|
|
* 7,8 - Bocsu bytes used. Number of bytes from a bocu sequence on
|
2003-03-13 17:15:53 +00:00
|
|
|
* the identical level.
|
2007-12-26 19:27:43 +00:00
|
|
|
* 9..31 - CEs consumed. Number of getCE or next32 operations performed
|
2006-06-16 06:00:13 +00:00
|
|
|
* since thes last successful update of the iterator state.
|
2003-03-13 17:15:53 +00:00
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
2003-03-27 20:09:38 +00:00
|
|
|
ucol_nextSortKeyPart(const UCollator *coll,
|
2003-01-23 01:52:34 +00:00
|
|
|
UCharIterator *iter,
|
|
|
|
uint32_t state[2],
|
|
|
|
uint8_t *dest, int32_t count,
|
2007-12-26 19:27:43 +00:00
|
|
|
UErrorCode *status)
|
|
|
|
{
|
2003-01-23 01:52:34 +00:00
|
|
|
/* error checking */
|
|
|
|
if(status==NULL || U_FAILURE(*status)) {
|
|
|
|
return 0;
|
|
|
|
}
|
2003-09-24 00:58:35 +00:00
|
|
|
UTRACE_ENTRY(UTRACE_UCOL_NEXTSORTKEYPART);
|
2003-01-23 01:52:34 +00:00
|
|
|
if( coll==NULL || iter==NULL ||
|
|
|
|
state==NULL ||
|
|
|
|
count<0 || (count>0 && dest==NULL)
|
|
|
|
) {
|
|
|
|
*status=U_ILLEGAL_ARGUMENT_ERROR;
|
2007-08-29 02:57:42 +00:00
|
|
|
UTRACE_EXIT_STATUS(status);
|
|
|
|
return 0;
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
|
|
|
|
2003-09-24 00:58:35 +00:00
|
|
|
UTRACE_DATA6(UTRACE_VERBOSE, "coll=%p, iter=%p, state=%d %d, dest=%p, count=%d",
|
|
|
|
coll, iter, state[0], state[1], dest, count);
|
2003-01-23 01:52:34 +00:00
|
|
|
|
|
|
|
if(count==0) {
|
|
|
|
/* nothing to do */
|
2003-11-01 01:18:42 +00:00
|
|
|
UTRACE_EXIT_VALUE(0);
|
2003-01-23 01:52:34 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
/** Setting up situation according to the state we got from the previous iteration */
|
|
|
|
// The state of the iterator from the previous invocation
|
|
|
|
uint32_t iterState = state[0];
|
|
|
|
// Has the last iteration ended in the shifted state
|
|
|
|
UBool wasShifted = ((state[1] >> UCOL_PSK_WAS_SHIFTED_SHIFT) & UCOL_PSK_WAS_SHIFTED_MASK)?TRUE:FALSE;
|
|
|
|
// What is the current level of the sortkey?
|
|
|
|
int32_t level= (state[1] >> UCOL_PSK_LEVEL_SHIFT) & UCOL_PSK_LEVEL_MASK;
|
|
|
|
// Have we written only one byte from a two byte primary in the previous iteration?
|
|
|
|
// Also on secondary level - have we finished with the French secondary?
|
2003-12-03 00:45:12 +00:00
|
|
|
int32_t byteCountOrFrenchDone = (state[1] >> UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_SHIFT) & UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_MASK;
|
2003-03-13 17:15:53 +00:00
|
|
|
// number of bytes in the continuation buffer for French
|
|
|
|
int32_t usedFrench = (state[1] >> UCOL_PSK_USED_FRENCH_SHIFT) & UCOL_PSK_USED_FRENCH_MASK;
|
2006-06-16 06:00:13 +00:00
|
|
|
// Number of bytes already written from a bocsu sequence. Since
|
2007-12-26 19:27:43 +00:00
|
|
|
// the longes bocsu sequence is 4 long, this can be up to 3.
|
2006-06-16 06:00:13 +00:00
|
|
|
int32_t bocsuBytesUsed = (state[1] >> UCOL_PSK_BOCSU_BYTES_SHIFT) & UCOL_PSK_BOCSU_BYTES_MASK;
|
|
|
|
// Number of elements that need to be consumed in this iteration because
|
2007-12-26 19:27:43 +00:00
|
|
|
// the iterator returned UITER_NO_STATE at the end of the last iteration,
|
2006-06-16 06:00:13 +00:00
|
|
|
// so we had to save the last valid state.
|
|
|
|
int32_t cces = (state[1] >> UCOL_PSK_CONSUMED_CES_SHIFT) & UCOL_PSK_CONSUMED_CES_MASK;
|
2003-03-13 17:15:53 +00:00
|
|
|
|
|
|
|
/** values that depend on the collator attributes */
|
2003-12-03 00:45:12 +00:00
|
|
|
// strength of the collator.
|
2003-01-23 01:52:34 +00:00
|
|
|
int32_t strength = ucol_getAttribute(coll, UCOL_STRENGTH, status);
|
2003-03-25 18:25:09 +00:00
|
|
|
// maximal level of the partial sortkey. Need to take whether case level is done
|
|
|
|
int32_t maxLevel = 0;
|
|
|
|
if(strength < UCOL_TERTIARY) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(ucol_getAttribute(coll, UCOL_CASE_LEVEL, status) == UCOL_ON) {
|
|
|
|
maxLevel = UCOL_PSK_CASE;
|
|
|
|
} else {
|
|
|
|
maxLevel = strength;
|
|
|
|
}
|
2003-03-25 18:25:09 +00:00
|
|
|
} else {
|
|
|
|
if(strength == UCOL_TERTIARY) {
|
2007-12-26 19:27:43 +00:00
|
|
|
maxLevel = UCOL_PSK_TERTIARY;
|
2003-03-25 18:25:09 +00:00
|
|
|
} else if(strength == UCOL_QUATERNARY) {
|
2007-12-26 19:27:43 +00:00
|
|
|
maxLevel = UCOL_PSK_QUATERNARY;
|
2003-03-25 18:25:09 +00:00
|
|
|
} else { // identical
|
2007-12-26 19:27:43 +00:00
|
|
|
maxLevel = UCOL_IDENTICAL;
|
2003-03-25 18:25:09 +00:00
|
|
|
}
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
// value for the quaternary level if Hiragana is encountered. Used for JIS X 4061 collation
|
2003-12-03 00:45:12 +00:00
|
|
|
uint8_t UCOL_HIRAGANA_QUAD =
|
2003-01-23 01:52:34 +00:00
|
|
|
(ucol_getAttribute(coll, UCOL_HIRAGANA_QUATERNARY_MODE, status) == UCOL_ON)?0xFE:0xFF;
|
2003-02-06 23:29:56 +00:00
|
|
|
// Boundary value that decides whether a CE is shifted or not
|
|
|
|
uint32_t LVT = (coll->alternateHandling == UCOL_SHIFTED)?(coll->variableTopValue<<16):0;
|
2003-03-13 17:15:53 +00:00
|
|
|
// Are we doing French collation?
|
|
|
|
UBool doingFrench = (ucol_getAttribute(coll, UCOL_FRENCH_COLLATION, status) == UCOL_ON);
|
2003-02-06 23:29:56 +00:00
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
/** initializing the collation state */
|
|
|
|
UBool notIsContinuation = FALSE;
|
|
|
|
uint32_t CE = UCOL_NO_MORE_CES;
|
2003-01-23 01:52:34 +00:00
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
collIterate s;
|
2003-01-23 01:52:34 +00:00
|
|
|
IInit_collIterate(coll, NULL, -1, &s);
|
|
|
|
s.iterator = iter;
|
|
|
|
s.flags |= UCOL_USE_ITERATOR;
|
2003-03-13 17:15:53 +00:00
|
|
|
// This variable tells us whether we have produced some other levels in this iteration
|
2003-12-03 00:45:12 +00:00
|
|
|
// before we moved to the identical level. In that case, we need to switch the
|
2003-03-13 17:15:53 +00:00
|
|
|
// type of the iterator.
|
2003-02-14 07:46:20 +00:00
|
|
|
UBool doingIdenticalFromStart = FALSE;
|
2003-03-13 17:15:53 +00:00
|
|
|
// Normalizing iterator
|
2003-05-06 18:53:42 +00:00
|
|
|
// The division for the array length may truncate the array size to
|
|
|
|
// a little less than UNORM_ITER_SIZE, but that size is dimensioned too high
|
|
|
|
// for all platforms anyway.
|
2003-03-17 21:20:36 +00:00
|
|
|
UAlignedMemory stackNormIter[UNORM_ITER_SIZE/sizeof(UAlignedMemory)];
|
2003-01-23 01:52:34 +00:00
|
|
|
UNormIterator *normIter = NULL;
|
2003-03-13 17:15:53 +00:00
|
|
|
// If the normalization is turned on for the collator and we are below identical level
|
|
|
|
// we will use a FCD normalizing iterator
|
2003-02-06 23:29:56 +00:00
|
|
|
if(ucol_getAttribute(coll, UCOL_NORMALIZATION_MODE, status) == UCOL_ON && level < UCOL_PSK_IDENTICAL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
normIter = unorm_openIter(stackNormIter, sizeof(stackNormIter), status);
|
|
|
|
s.iterator = unorm_setIter(normIter, iter, UNORM_FCD, status);
|
|
|
|
s.flags &= ~UCOL_ITER_NORM;
|
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
UTRACE_EXIT_STATUS(*status);
|
|
|
|
return 0;
|
|
|
|
}
|
2003-02-14 07:46:20 +00:00
|
|
|
} else if(level == UCOL_PSK_IDENTICAL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// for identical level, we need a NFD iterator. We need to instantiate it here, since we
|
|
|
|
// will be updating the state - and this cannot be done on an ordinary iterator.
|
|
|
|
normIter = unorm_openIter(stackNormIter, sizeof(stackNormIter), status);
|
|
|
|
s.iterator = unorm_setIter(normIter, iter, UNORM_NFD, status);
|
|
|
|
s.flags &= ~UCOL_ITER_NORM;
|
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
UTRACE_EXIT_STATUS(*status);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
doingIdenticalFromStart = TRUE;
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
// This is the tentative new state of the iterator. The problem
|
|
|
|
// is that the iterator might return an undefined state, in
|
|
|
|
// which case we should save the last valid state and increase
|
|
|
|
// the iterator skip value.
|
2003-01-23 01:52:34 +00:00
|
|
|
uint32_t newState = 0;
|
2003-03-13 17:15:53 +00:00
|
|
|
|
|
|
|
// First, we set the iterator to the last valid position
|
|
|
|
// from the last iteration. This was saved in state[0].
|
2003-01-23 01:52:34 +00:00
|
|
|
if(iterState == 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
/* initial state */
|
|
|
|
if(level == UCOL_PSK_SECONDARY && doingFrench && !byteCountOrFrenchDone) {
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_LIMIT);
|
|
|
|
} else {
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
} else {
|
|
|
|
/* reset to previous state */
|
2007-12-26 19:27:43 +00:00
|
|
|
s.iterator->setState(s.iterator, iterState, status);
|
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
UTRACE_EXIT_STATUS(*status);
|
|
|
|
return 0;
|
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-16 06:00:13 +00:00
|
|
|
// This variable tells us whether we can attempt to update the state
|
|
|
|
// of iterator. Situations where we don't want to update iterator state
|
|
|
|
// are the existence of expansion CEs that are not yet processed, and
|
|
|
|
// finishing the case level without enough space in the buffer to insert
|
|
|
|
// a level terminator.
|
|
|
|
UBool canUpdateState = TRUE;
|
|
|
|
|
|
|
|
// Consume all the CEs that were consumed at the end of the previous
|
|
|
|
// iteration without updating the iterator state. On identical level,
|
2007-12-26 19:27:43 +00:00
|
|
|
// consume the code points.
|
2006-06-16 06:00:13 +00:00
|
|
|
int32_t counter = cces;
|
|
|
|
if(level < UCOL_PSK_IDENTICAL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
while(counter-->0) {
|
|
|
|
// If we're doing French and we are on the secondary level,
|
|
|
|
// we go backwards.
|
|
|
|
if(level == UCOL_PSK_SECONDARY && doingFrench) {
|
|
|
|
CE = ucol_IGetPrevCE(coll, &s, status);
|
|
|
|
} else {
|
|
|
|
CE = ucol_IGetNextCE(coll, &s, status);
|
|
|
|
}
|
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
/* should not happen */
|
|
|
|
*status=U_INTERNAL_PROGRAM_ERROR;
|
|
|
|
UTRACE_EXIT_STATUS(*status);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
}
|
2006-06-16 06:00:13 +00:00
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
while(counter-->0) {
|
|
|
|
uiter_next32(s.iterator);
|
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
// French secondary needs to know whether the iterator state of zero came from previous level OR
|
|
|
|
// from a new invocation...
|
|
|
|
UBool wasDoingPrimary = FALSE;
|
|
|
|
// destination buffer byte counter. When this guy
|
|
|
|
// gets to count, we're done with the iteration
|
2003-12-03 00:45:12 +00:00
|
|
|
int32_t i = 0;
|
|
|
|
// used to count the zero bytes written after we
|
2003-03-13 17:15:53 +00:00
|
|
|
// have finished with the sort key
|
2003-01-23 01:52:34 +00:00
|
|
|
int32_t j = 0;
|
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
|
|
|
|
// Hm.... I think we're ready to plunge in. Basic story is as following:
|
|
|
|
// we have a fall through case based on level. This is used for initial
|
|
|
|
// positioning on iteration start. Every level processor contains a
|
|
|
|
// for(;;) which will be broken when we exhaust all the CEs. Other
|
|
|
|
// way to exit is a goto saveState, which happens when we have filled
|
|
|
|
// out our buffer.
|
|
|
|
switch(level) {
|
2003-12-03 00:45:12 +00:00
|
|
|
case UCOL_PSK_PRIMARY:
|
2007-12-26 19:27:43 +00:00
|
|
|
wasDoingPrimary = TRUE;
|
|
|
|
for(;;) {
|
|
|
|
if(i==count) {
|
|
|
|
goto saveState;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
// We should save the state only if we
|
|
|
|
// are sure that we are done with the
|
|
|
|
// previous iterator state
|
2007-12-26 19:27:43 +00:00
|
|
|
if(canUpdateState && byteCountOrFrenchDone == 0) {
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
CE = ucol_IGetNextCE(coll, &s, status);
|
2006-06-16 06:00:13 +00:00
|
|
|
cces++;
|
2003-03-13 17:15:53 +00:00
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
// Add the level separator
|
2003-03-25 18:25:09 +00:00
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
2007-12-26 19:27:43 +00:00
|
|
|
byteCountOrFrenchDone=0;
|
2003-03-13 17:15:53 +00:00
|
|
|
// Restart the iteration an move to the
|
|
|
|
// second level
|
2003-12-03 00:45:12 +00:00
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
2006-06-16 06:00:13 +00:00
|
|
|
cces = 0;
|
2007-12-26 19:27:43 +00:00
|
|
|
level = UCOL_PSK_SECONDARY;
|
2003-01-23 01:52:34 +00:00
|
|
|
break;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
if(!isShiftedCE(CE, LVT, &wasShifted)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
CE >>= UCOL_PRIMARYORDERSHIFT; /* get primary */
|
|
|
|
if(CE != 0) {
|
|
|
|
if(byteCountOrFrenchDone == 0) {
|
|
|
|
// get the second byte of primary
|
|
|
|
dest[i++]=(uint8_t)(CE >> 8);
|
|
|
|
} else {
|
|
|
|
byteCountOrFrenchDone = 0;
|
|
|
|
}
|
|
|
|
if((CE &=0xff)!=0) {
|
|
|
|
if(i==count) {
|
|
|
|
/* overflow */
|
|
|
|
byteCountOrFrenchDone = 1;
|
|
|
|
cces--;
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
dest[i++]=(uint8_t)CE;
|
|
|
|
}
|
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2006-06-16 06:00:13 +00:00
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
2007-12-26 19:27:43 +00:00
|
|
|
canUpdateState = FALSE;
|
2003-03-13 17:15:53 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
canUpdateState = TRUE;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_SECONDARY:
|
|
|
|
if(strength >= UCOL_SECONDARY) {
|
|
|
|
if(!doingFrench) {
|
|
|
|
for(;;) {
|
|
|
|
if(i == count) {
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
// We should save the state only if we
|
|
|
|
// are sure that we are done with the
|
|
|
|
// previous iterator state
|
|
|
|
if(canUpdateState) {
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
CE = ucol_IGetNextCE(coll, &s, status);
|
|
|
|
cces++;
|
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
// Add the level separator
|
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
|
|
|
byteCountOrFrenchDone = 0;
|
|
|
|
// Restart the iteration an move to the
|
|
|
|
// second level
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
|
|
|
cces = 0;
|
|
|
|
level = UCOL_PSK_CASE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if(!isShiftedCE(CE, LVT, &wasShifted)) {
|
|
|
|
CE >>= 8; /* get secondary */
|
|
|
|
if(CE != 0) {
|
|
|
|
dest[i++]=(uint8_t)CE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
} else {
|
|
|
|
canUpdateState = TRUE;
|
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else { // French secondary processing
|
|
|
|
uint8_t frenchBuff[UCOL_MAX_BUFFER];
|
|
|
|
int32_t frenchIndex = 0;
|
|
|
|
// Here we are going backwards.
|
|
|
|
// If the iterator is at the beggining, it should be
|
|
|
|
// moved to end.
|
|
|
|
if(wasDoingPrimary) {
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_LIMIT);
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
for(;;) {
|
|
|
|
if(i == count) {
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
if(canUpdateState) {
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
CE = ucol_IGetPrevCE(coll, &s, status);
|
|
|
|
cces++;
|
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
// Add the level separator
|
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
|
|
|
byteCountOrFrenchDone = 0;
|
|
|
|
// Restart the iteration an move to the next level
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
|
|
|
level = UCOL_PSK_CASE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if(isContinuation(CE)) { // if it's a continuation, we want to save it and
|
|
|
|
// reverse when we get a first non-continuation CE.
|
|
|
|
CE >>= 8;
|
|
|
|
frenchBuff[frenchIndex++] = (uint8_t)CE;
|
|
|
|
} else if(!isShiftedCE(CE, LVT, &wasShifted)) {
|
|
|
|
CE >>= 8; /* get secondary */
|
|
|
|
if(!frenchIndex) {
|
|
|
|
if(CE != 0) {
|
|
|
|
dest[i++]=(uint8_t)CE;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
frenchBuff[frenchIndex++] = (uint8_t)CE;
|
|
|
|
frenchIndex -= usedFrench;
|
|
|
|
usedFrench = 0;
|
|
|
|
while(i < count && frenchIndex) {
|
|
|
|
dest[i++] = frenchBuff[--frenchIndex];
|
|
|
|
usedFrench++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
} else {
|
|
|
|
canUpdateState = TRUE;
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
level = UCOL_PSK_CASE;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_CASE:
|
2007-12-26 19:27:43 +00:00
|
|
|
if(ucol_getAttribute(coll, UCOL_CASE_LEVEL, status) == UCOL_ON) {
|
|
|
|
uint32_t caseShift = UCOL_CASE_SHIFT_START;
|
|
|
|
uint8_t caseByte = UCOL_CASE_BYTE_START;
|
|
|
|
uint8_t caseBits = 0;
|
|
|
|
|
|
|
|
for(;;) {
|
|
|
|
if(i == count) {
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
// We should save the state only if we
|
|
|
|
// are sure that we are done with the
|
|
|
|
// previous iterator state
|
|
|
|
if(canUpdateState) {
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
CE = ucol_IGetNextCE(coll, &s, status);
|
|
|
|
cces++;
|
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
// On the case level we might have an unfinished
|
|
|
|
// case byte. Add one if it's started.
|
|
|
|
if(caseShift != UCOL_CASE_SHIFT_START) {
|
|
|
|
dest[i++] = caseByte;
|
|
|
|
}
|
|
|
|
cces = 0;
|
|
|
|
// We have finished processing CEs on this level.
|
|
|
|
// However, we don't know if we have enough space
|
|
|
|
// to add a case level terminator.
|
|
|
|
if(i < count) {
|
|
|
|
// Add the level separator
|
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
|
|
|
// Restart the iteration and move to the
|
|
|
|
// next level
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
|
|
|
level = UCOL_PSK_TERTIARY;
|
|
|
|
} else {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
}
|
|
|
|
break;
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(!isShiftedCE(CE, LVT, &wasShifted)) {
|
|
|
|
if(!isContinuation(CE) && ((CE & UCOL_PRIMARYMASK) != 0 || strength > UCOL_PRIMARY)) {
|
|
|
|
// do the case level if we need to do it. We don't want to calculate
|
|
|
|
// case level for primary ignorables if we have only primary strength and case level
|
|
|
|
// otherwise we would break well formedness of CEs
|
|
|
|
CE = (uint8_t)(CE & UCOL_BYTE_SIZE_MASK);
|
|
|
|
caseBits = (uint8_t)(CE & 0xC0);
|
|
|
|
// this copies the case level logic from the
|
|
|
|
// sort key generation code
|
|
|
|
if(CE != 0) {
|
|
|
|
if(coll->caseFirst == UCOL_UPPER_FIRST) {
|
|
|
|
if((caseBits & 0xC0) == 0) {
|
|
|
|
caseByte |= 1 << (--caseShift);
|
|
|
|
} else {
|
|
|
|
caseByte |= 0 << (--caseShift);
|
|
|
|
/* second bit */
|
|
|
|
if(caseShift == 0) {
|
|
|
|
dest[i++] = caseByte;
|
|
|
|
caseShift = UCOL_CASE_SHIFT_START;
|
|
|
|
caseByte = UCOL_CASE_BYTE_START;
|
|
|
|
}
|
|
|
|
caseByte |= ((caseBits>>6)&1) << (--caseShift);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if((caseBits & 0xC0) == 0) {
|
|
|
|
caseByte |= 0 << (--caseShift);
|
|
|
|
} else {
|
|
|
|
caseByte |= 1 << (--caseShift);
|
|
|
|
/* second bit */
|
|
|
|
if(caseShift == 0) {
|
|
|
|
dest[i++] = caseByte;
|
|
|
|
caseShift = UCOL_CASE_SHIFT_START;
|
|
|
|
caseByte = UCOL_CASE_BYTE_START;
|
|
|
|
}
|
|
|
|
caseByte |= ((caseBits>>7)&1) << (--caseShift);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// Not sure this is correct for the case level - revisit
|
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
} else {
|
|
|
|
canUpdateState = TRUE;
|
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
level = UCOL_PSK_TERTIARY;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_TERTIARY:
|
2007-12-26 19:27:43 +00:00
|
|
|
if(strength >= UCOL_TERTIARY) {
|
|
|
|
for(;;) {
|
|
|
|
if(i == count) {
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
// We should save the state only if we
|
|
|
|
// are sure that we are done with the
|
|
|
|
// previous iterator state
|
|
|
|
if(canUpdateState) {
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
CE = ucol_IGetNextCE(coll, &s, status);
|
|
|
|
cces++;
|
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
// Add the level separator
|
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
|
|
|
byteCountOrFrenchDone = 0;
|
|
|
|
// Restart the iteration an move to the
|
|
|
|
// second level
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
|
|
|
cces = 0;
|
|
|
|
level = UCOL_PSK_QUATERNARY;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if(!isShiftedCE(CE, LVT, &wasShifted)) {
|
|
|
|
notIsContinuation = !isContinuation(CE);
|
2003-03-13 17:15:53 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(notIsContinuation) {
|
|
|
|
CE = (uint8_t)(CE & UCOL_BYTE_SIZE_MASK);
|
|
|
|
CE ^= coll->caseSwitch;
|
|
|
|
CE &= coll->tertiaryMask;
|
|
|
|
} else {
|
|
|
|
CE = (uint8_t)((CE & UCOL_REMOVE_CONTINUATION));
|
|
|
|
}
|
2003-02-06 23:29:56 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(CE != 0) {
|
|
|
|
dest[i++]=(uint8_t)CE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
} else {
|
|
|
|
canUpdateState = TRUE;
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
// if we're not doing tertiary
|
|
|
|
// skip to the end
|
|
|
|
level = UCOL_PSK_NULL;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_QUATERNARY:
|
2007-12-26 19:27:43 +00:00
|
|
|
if(strength >= UCOL_QUATERNARY) {
|
|
|
|
for(;;) {
|
|
|
|
if(i == count) {
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
// We should save the state only if we
|
|
|
|
// are sure that we are done with the
|
|
|
|
// previous iterator state
|
|
|
|
if(canUpdateState) {
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
CE = ucol_IGetNextCE(coll, &s, status);
|
|
|
|
cces++;
|
|
|
|
if(CE==UCOL_NO_MORE_CES) {
|
|
|
|
// Add the level separator
|
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
|
|
|
//dest[i++] = UCOL_LEVELTERMINATOR;
|
|
|
|
byteCountOrFrenchDone = 0;
|
|
|
|
// Restart the iteration an move to the
|
|
|
|
// second level
|
|
|
|
s.iterator->move(s.iterator, 0, UITER_START);
|
|
|
|
cces = 0;
|
|
|
|
level = UCOL_PSK_QUIN;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if(CE==0)
|
|
|
|
continue;
|
|
|
|
if(isShiftedCE(CE, LVT, &wasShifted)) {
|
|
|
|
CE >>= 16; /* get primary */
|
|
|
|
if(CE != 0) {
|
|
|
|
if(byteCountOrFrenchDone == 0) {
|
|
|
|
dest[i++]=(uint8_t)(CE >> 8);
|
|
|
|
} else {
|
|
|
|
byteCountOrFrenchDone = 0;
|
|
|
|
}
|
|
|
|
if((CE &=0xff)!=0) {
|
|
|
|
if(i==count) {
|
|
|
|
/* overflow */
|
|
|
|
byteCountOrFrenchDone = 1;
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
dest[i++]=(uint8_t)CE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
notIsContinuation = !isContinuation(CE);
|
|
|
|
if(notIsContinuation) {
|
|
|
|
if(s.flags & UCOL_WAS_HIRAGANA) { // This was Hiragana and we need to note it
|
|
|
|
dest[i++] = UCOL_HIRAGANA_QUAD;
|
|
|
|
} else {
|
|
|
|
dest[i++] = 0xFF;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(uprv_numAvailableExpCEs(s)) {
|
|
|
|
canUpdateState = FALSE;
|
|
|
|
} else {
|
|
|
|
canUpdateState = TRUE;
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
// if we're not doing quaternary
|
|
|
|
// skip to the end
|
|
|
|
level = UCOL_PSK_NULL;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_QUIN:
|
2007-12-26 19:27:43 +00:00
|
|
|
level = UCOL_PSK_IDENTICAL;
|
2003-03-13 17:15:53 +00:00
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_IDENTICAL:
|
2007-12-26 19:27:43 +00:00
|
|
|
if(strength >= UCOL_IDENTICAL) {
|
|
|
|
UChar32 first, second;
|
|
|
|
int32_t bocsuBytesWritten = 0;
|
|
|
|
// We always need to do identical on
|
|
|
|
// the NFD form of the string.
|
|
|
|
if(normIter == NULL) {
|
|
|
|
// we arrived from the level below and
|
|
|
|
// normalization was not turned on.
|
|
|
|
// therefore, we need to make a fresh NFD iterator
|
|
|
|
normIter = unorm_openIter(stackNormIter, sizeof(stackNormIter), status);
|
|
|
|
s.iterator = unorm_setIter(normIter, iter, UNORM_NFD, status);
|
|
|
|
} else if(!doingIdenticalFromStart) {
|
|
|
|
// there is an iterator, but we did some other levels.
|
|
|
|
// therefore, we have a FCD iterator - need to make
|
|
|
|
// a NFD one.
|
|
|
|
// normIter being at the beginning does not guarantee
|
|
|
|
// that the underlying iterator is at the beginning
|
|
|
|
iter->move(iter, 0, UITER_START);
|
|
|
|
s.iterator = unorm_setIter(normIter, iter, UNORM_NFD, status);
|
|
|
|
}
|
|
|
|
// At this point we have a NFD iterator that is positioned
|
|
|
|
// in the right place
|
|
|
|
if(U_FAILURE(*status)) {
|
|
|
|
UTRACE_EXIT_STATUS(*status);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
first = uiter_previous32(s.iterator);
|
|
|
|
// maybe we're at the start of the string
|
|
|
|
if(first == U_SENTINEL) {
|
|
|
|
first = 0;
|
|
|
|
} else {
|
|
|
|
uiter_next32(s.iterator);
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
j = 0;
|
|
|
|
for(;;) {
|
|
|
|
if(i == count) {
|
|
|
|
if(j+1 < bocsuBytesWritten) {
|
|
|
|
bocsuBytesUsed = j+1;
|
|
|
|
}
|
|
|
|
goto saveState;
|
|
|
|
}
|
|
|
|
|
|
|
|
// On identical level, we will always save
|
|
|
|
// the state if we reach this point, since
|
|
|
|
// we don't depend on getNextCE for content
|
|
|
|
// all the content is in our buffer and we
|
|
|
|
// already either stored the full buffer OR
|
|
|
|
// otherwise we won't arrive here.
|
|
|
|
newState = s.iterator->getState(s.iterator);
|
|
|
|
if(newState != UITER_NO_STATE) {
|
|
|
|
iterState = newState;
|
|
|
|
cces = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
uint8_t buff[4];
|
|
|
|
second = uiter_next32(s.iterator);
|
|
|
|
cces++;
|
|
|
|
|
|
|
|
// end condition for identical level
|
|
|
|
if(second == U_SENTINEL) {
|
|
|
|
terminatePSKLevel(level, maxLevel, i, dest);
|
|
|
|
level = UCOL_PSK_NULL;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
bocsuBytesWritten = u_writeIdenticalLevelRunTwoChars(first, second, buff);
|
|
|
|
first = second;
|
2003-03-13 17:15:53 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
j = 0;
|
|
|
|
if(bocsuBytesUsed != 0) {
|
|
|
|
while(bocsuBytesUsed-->0) {
|
|
|
|
j++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
while(i < count && j < bocsuBytesWritten) {
|
|
|
|
dest[i++] = buff[j++];
|
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
level = UCOL_PSK_NULL;
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
/* fall through to next level */
|
|
|
|
case UCOL_PSK_NULL:
|
2007-12-26 19:27:43 +00:00
|
|
|
j = i;
|
|
|
|
while(j<count) {
|
|
|
|
dest[j++]=0;
|
|
|
|
}
|
|
|
|
break;
|
2003-03-13 17:15:53 +00:00
|
|
|
default:
|
2007-12-26 19:27:43 +00:00
|
|
|
*status = U_INTERNAL_PROGRAM_ERROR;
|
|
|
|
UTRACE_EXIT_STATUS(*status);
|
|
|
|
return 0;
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
2003-01-23 01:52:34 +00:00
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
saveState:
|
|
|
|
// Now we need to return stuff. First we want to see whether we have
|
2003-01-23 01:52:34 +00:00
|
|
|
// done everything for the current state of iterator.
|
2006-06-16 06:00:13 +00:00
|
|
|
if(byteCountOrFrenchDone
|
2007-12-26 19:27:43 +00:00
|
|
|
|| canUpdateState == FALSE
|
|
|
|
|| (newState = s.iterator->getState(s.iterator)) == UITER_NO_STATE)
|
|
|
|
{
|
|
|
|
// Any of above mean that the previous transaction
|
|
|
|
// wasn't finished and that we should store the
|
|
|
|
// previous iterator state.
|
|
|
|
state[0] = iterState;
|
2003-01-23 01:52:34 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
// The transaction is complete. We will continue in the next iteration.
|
2003-01-23 01:52:34 +00:00
|
|
|
state[0] = s.iterator->getState(s.iterator);
|
2006-06-16 06:00:13 +00:00
|
|
|
cces = 0;
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2006-06-16 06:00:13 +00:00
|
|
|
// Store the number of bocsu bytes written.
|
|
|
|
if((bocsuBytesUsed & UCOL_PSK_BOCSU_BYTES_MASK) != bocsuBytesUsed) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*status = U_INDEX_OUTOFBOUNDS_ERROR;
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
2006-06-16 06:00:13 +00:00
|
|
|
state[1] = (bocsuBytesUsed & UCOL_PSK_BOCSU_BYTES_MASK) << UCOL_PSK_BOCSU_BYTES_SHIFT;
|
2003-03-13 17:15:53 +00:00
|
|
|
|
|
|
|
// Next we put in the level of comparison
|
2003-12-03 00:45:12 +00:00
|
|
|
state[1] |= ((level & UCOL_PSK_LEVEL_MASK) << UCOL_PSK_LEVEL_SHIFT);
|
2003-03-13 17:15:53 +00:00
|
|
|
|
|
|
|
// If we are doing French, we need to store whether we have just finished the French level
|
|
|
|
if(level == UCOL_PSK_SECONDARY && doingFrench) {
|
2007-12-26 19:27:43 +00:00
|
|
|
state[1] |= (((state[0] == 0) & UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_MASK) << UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_SHIFT);
|
2003-03-13 17:15:53 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
state[1] |= ((byteCountOrFrenchDone & UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_MASK) << UCOL_PSK_BYTE_COUNT_OR_FRENCH_DONE_SHIFT);
|
2003-03-13 17:15:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// Was the latest CE shifted
|
2003-02-06 23:29:56 +00:00
|
|
|
if(wasShifted) {
|
2007-12-26 19:27:43 +00:00
|
|
|
state[1] |= 1 << UCOL_PSK_WAS_SHIFTED_SHIFT;
|
2003-02-06 23:29:56 +00:00
|
|
|
}
|
2006-06-16 06:00:13 +00:00
|
|
|
// Check for cces overflow
|
|
|
|
if((cces & UCOL_PSK_CONSUMED_CES_MASK) != cces) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*status = U_INDEX_OUTOFBOUNDS_ERROR;
|
2003-03-04 06:31:20 +00:00
|
|
|
}
|
2006-06-16 06:00:13 +00:00
|
|
|
// Store cces
|
|
|
|
state[1] |= ((cces & UCOL_PSK_CONSUMED_CES_MASK) << UCOL_PSK_CONSUMED_CES_SHIFT);
|
2003-02-06 23:29:56 +00:00
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
// Check for French overflow
|
2003-03-04 06:31:20 +00:00
|
|
|
if((usedFrench & UCOL_PSK_USED_FRENCH_MASK) != usedFrench) {
|
2007-12-26 19:27:43 +00:00
|
|
|
*status = U_INDEX_OUTOFBOUNDS_ERROR;
|
2003-03-04 06:31:20 +00:00
|
|
|
}
|
2003-03-13 17:15:53 +00:00
|
|
|
// Store number of bytes written in the French secondary continuation sequence
|
2003-02-06 23:29:56 +00:00
|
|
|
state[1] |= ((usedFrench & UCOL_PSK_USED_FRENCH_MASK) << UCOL_PSK_USED_FRENCH_SHIFT);
|
|
|
|
|
2003-01-23 01:52:34 +00:00
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
// If we have used normalizing iterator, get rid of it
|
2003-01-23 01:52:34 +00:00
|
|
|
if(normIter != NULL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
unorm_closeIter(normIter);
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
|
|
|
|
2008-06-09 21:18:46 +00:00
|
|
|
/* To avoid memory leak, free the offset buffer if necessary. */
|
|
|
|
freeOffsetBuffer(&s);
|
|
|
|
|
2003-03-13 17:15:53 +00:00
|
|
|
// Return number of meaningful sortkey bytes.
|
2003-09-24 00:58:35 +00:00
|
|
|
UTRACE_DATA4(UTRACE_VERBOSE, "dest = %vb, state=%d %d",
|
|
|
|
dest,i, state[0], state[1]);
|
2003-11-01 01:18:42 +00:00
|
|
|
UTRACE_EXIT_VALUE(i);
|
2003-01-23 01:52:34 +00:00
|
|
|
return i;
|
|
|
|
}
|
|
|
|
|
2002-01-21 23:54:34 +00:00
|
|
|
/**
|
2002-01-24 23:02:02 +00:00
|
|
|
* Produce a bound for a given sortkey and a number of levels.
|
2002-01-21 23:54:34 +00:00
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
2002-01-24 23:02:02 +00:00
|
|
|
ucol_getBound(const uint8_t *source,
|
2002-01-21 23:54:34 +00:00
|
|
|
int32_t sourceLength,
|
2002-01-24 23:02:02 +00:00
|
|
|
UColBoundMode boundType,
|
|
|
|
uint32_t noOfLevels,
|
2002-01-21 23:54:34 +00:00
|
|
|
uint8_t *result,
|
|
|
|
int32_t resultLength,
|
2007-12-26 19:27:43 +00:00
|
|
|
UErrorCode *status)
|
|
|
|
{
|
|
|
|
// consistency checks
|
|
|
|
if(status == NULL || U_FAILURE(*status)) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if(source == NULL) {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return 0;
|
2002-01-21 23:54:34 +00:00
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
int32_t sourceIndex = 0;
|
|
|
|
// Scan the string until we skip enough of the key OR reach the end of the key
|
|
|
|
do {
|
|
|
|
sourceIndex++;
|
|
|
|
if(source[sourceIndex] == UCOL_LEVELTERMINATOR) {
|
|
|
|
noOfLevels--;
|
|
|
|
}
|
|
|
|
} while (noOfLevels > 0
|
|
|
|
&& (source[sourceIndex] != 0 || sourceIndex < sourceLength));
|
|
|
|
|
|
|
|
if((source[sourceIndex] == 0 || sourceIndex == sourceLength)
|
|
|
|
&& noOfLevels > 0) {
|
|
|
|
*status = U_SORT_KEY_TOO_SHORT_WARNING;
|
|
|
|
}
|
2002-01-21 23:54:34 +00:00
|
|
|
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// READ ME: this code assumes that the values for boundType
|
|
|
|
// enum will not changes. They are set so that the enum value
|
|
|
|
// corresponds to the number of extra bytes each bound type
|
|
|
|
// needs.
|
|
|
|
if(result != NULL && resultLength >= sourceIndex+boundType) {
|
|
|
|
uprv_memcpy(result, source, sourceIndex);
|
|
|
|
switch(boundType) {
|
|
|
|
// Lower bound just gets terminated. No extra bytes
|
2007-12-28 20:08:41 +00:00
|
|
|
case UCOL_BOUND_LOWER: // = 0
|
|
|
|
break;
|
|
|
|
// Upper bound needs one extra byte
|
|
|
|
case UCOL_BOUND_UPPER: // = 1
|
|
|
|
result[sourceIndex++] = 2;
|
|
|
|
break;
|
|
|
|
// Upper long bound needs two extra bytes
|
|
|
|
case UCOL_BOUND_UPPER_LONG: // = 2
|
|
|
|
result[sourceIndex++] = 0xFF;
|
|
|
|
result[sourceIndex++] = 0xFF;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return 0;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
|
|
|
result[sourceIndex++] = 0;
|
2002-01-21 23:54:34 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return sourceIndex;
|
|
|
|
} else {
|
|
|
|
return sourceIndex+boundType+1;
|
|
|
|
}
|
2002-01-21 23:54:34 +00:00
|
|
|
}
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/****************************************************************************/
|
|
|
|
/* Following are the functions that deal with the properties of a collator */
|
|
|
|
/* there are new APIs and some compatibility APIs */
|
|
|
|
/****************************************************************************/
|
2002-09-04 06:02:13 +00:00
|
|
|
|
|
|
|
static inline void
|
|
|
|
ucol_addLatinOneEntry(UCollator *coll, UChar ch, uint32_t CE,
|
2007-12-26 19:27:43 +00:00
|
|
|
int32_t *primShift, int32_t *secShift, int32_t *terShift)
|
|
|
|
{
|
|
|
|
uint8_t primary1 = 0, primary2 = 0, secondary = 0, tertiary = 0;
|
|
|
|
UBool reverseSecondary = FALSE;
|
|
|
|
if(!isContinuation(CE)) {
|
|
|
|
tertiary = (uint8_t)((CE & coll->tertiaryMask));
|
|
|
|
tertiary ^= coll->caseSwitch;
|
|
|
|
reverseSecondary = TRUE;
|
|
|
|
} else {
|
|
|
|
tertiary = (uint8_t)((CE & UCOL_REMOVE_CONTINUATION));
|
|
|
|
tertiary &= UCOL_REMOVE_CASE;
|
|
|
|
reverseSecondary = FALSE;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
secondary = (uint8_t)((CE >>= 8) & UCOL_BYTE_SIZE_MASK);
|
|
|
|
primary2 = (uint8_t)((CE >>= 8) & UCOL_BYTE_SIZE_MASK);
|
|
|
|
primary1 = (uint8_t)(CE >> 8);
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(primary1 != 0) {
|
|
|
|
coll->latinOneCEs[ch] |= (primary1 << *primShift);
|
|
|
|
*primShift -= 8;
|
|
|
|
}
|
|
|
|
if(primary2 != 0) {
|
|
|
|
if(*primShift < 0) {
|
|
|
|
coll->latinOneCEs[ch] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+ch] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+ch] = UCOL_BAIL_OUT_CE;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
coll->latinOneCEs[ch] |= (primary2 << *primShift);
|
|
|
|
*primShift -= 8;
|
|
|
|
}
|
|
|
|
if(secondary != 0) {
|
|
|
|
if(reverseSecondary && coll->frenchCollation == UCOL_ON) { // reverse secondary
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+ch] >>= 8; // make space for secondary
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+ch] |= (secondary << 24);
|
|
|
|
} else { // normal case
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+ch] |= (secondary << *secShift);
|
|
|
|
}
|
|
|
|
*secShift -= 8;
|
|
|
|
}
|
|
|
|
if(tertiary != 0) {
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+ch] |= (tertiary << *terShift);
|
|
|
|
*terShift -= 8;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline UBool
|
|
|
|
ucol_resizeLatinOneTable(UCollator *coll, int32_t size, UErrorCode *status) {
|
|
|
|
uint32_t *newTable = (uint32_t *)uprv_malloc(size*sizeof(uint32_t)*3);
|
|
|
|
if(newTable == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
coll->latinOneFailed = TRUE;
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
int32_t sizeToCopy = ((size<coll->latinOneTableLen)?size:coll->latinOneTableLen)*sizeof(uint32_t);
|
|
|
|
uprv_memset(newTable, 0, size*sizeof(uint32_t)*3);
|
|
|
|
uprv_memcpy(newTable, coll->latinOneCEs, sizeToCopy);
|
|
|
|
uprv_memcpy(newTable+size, coll->latinOneCEs+coll->latinOneTableLen, sizeToCopy);
|
|
|
|
uprv_memcpy(newTable+2*size, coll->latinOneCEs+2*coll->latinOneTableLen, sizeToCopy);
|
|
|
|
coll->latinOneTableLen = size;
|
|
|
|
uprv_free(coll->latinOneCEs);
|
|
|
|
coll->latinOneCEs = newTable;
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
|
2003-05-16 22:53:54 +00:00
|
|
|
static UBool
|
|
|
|
ucol_setUpLatinOne(UCollator *coll, UErrorCode *status) {
|
2007-12-26 19:27:43 +00:00
|
|
|
UBool result = TRUE;
|
2002-09-04 06:02:13 +00:00
|
|
|
if(coll->latinOneCEs == NULL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->latinOneCEs = (uint32_t *)uprv_malloc(sizeof(uint32_t)*UCOL_LATINONETABLELEN*3);
|
|
|
|
if(coll->latinOneCEs == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
coll->latinOneTableLen = UCOL_LATINONETABLELEN;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
UChar ch = 0;
|
|
|
|
UCollationElements *it = ucol_openElements(coll, &ch, 1, status);
|
2008-01-16 20:12:48 +00:00
|
|
|
// Check for null pointer
|
2008-02-19 20:10:55 +00:00
|
|
|
if (U_FAILURE(*status)) {
|
2008-02-13 09:35:50 +00:00
|
|
|
return FALSE;
|
2008-01-16 20:12:48 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
uprv_memset(coll->latinOneCEs, 0, sizeof(uint32_t)*coll->latinOneTableLen*3);
|
|
|
|
|
|
|
|
int32_t primShift = 24, secShift = 24, terShift = 24;
|
|
|
|
uint32_t CE = 0;
|
|
|
|
int32_t contractionOffset = UCOL_ENDOFLATINONERANGE+1;
|
|
|
|
|
|
|
|
// TODO: make safe if you get more than you wanted...
|
|
|
|
for(ch = 0; ch <= UCOL_ENDOFLATINONERANGE; ch++) {
|
|
|
|
primShift = 24; secShift = 24; terShift = 24;
|
|
|
|
if(ch < 0x100) {
|
|
|
|
CE = coll->latinOneMapping[ch];
|
|
|
|
} else {
|
|
|
|
CE = UTRIE_GET32_FROM_LEAD(&coll->mapping, ch);
|
|
|
|
if(CE == UCOL_NOT_FOUND && coll->UCA) {
|
|
|
|
CE = UTRIE_GET32_FROM_LEAD(&coll->UCA->mapping, ch);
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if(CE < UCOL_NOT_FOUND) {
|
|
|
|
ucol_addLatinOneEntry(coll, ch, CE, &primShift, &secShift, &terShift);
|
|
|
|
} else {
|
|
|
|
switch (getCETag(CE)) {
|
|
|
|
case EXPANSION_TAG:
|
|
|
|
case DIGIT_TAG:
|
|
|
|
ucol_setText(it, &ch, 1, status);
|
|
|
|
while((int32_t)(CE = ucol_next(it, status)) != UCOL_NULLORDER) {
|
2002-09-04 06:02:13 +00:00
|
|
|
if(primShift < 0 || secShift < 0 || terShift < 0) {
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->latinOneCEs[ch] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+ch] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+ch] = UCOL_BAIL_OUT_CE;
|
|
|
|
break;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
ucol_addLatinOneEntry(coll, ch, CE, &primShift, &secShift, &terShift);
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
|
|
|
case CONTRACTION_TAG:
|
|
|
|
// here is the trick
|
|
|
|
// F2 is contraction. We do something very similar to contractions
|
|
|
|
// but have two indices, one in the real contraction table and the
|
|
|
|
// other to where we stuffed things. This hopes that we don't have
|
|
|
|
// many contractions (this should work for latin-1 tables).
|
|
|
|
{
|
|
|
|
if((CE & 0x00FFF000) != 0) {
|
|
|
|
*status = U_UNSUPPORTED_ERROR;
|
|
|
|
goto cleanup_after_failure;
|
|
|
|
}
|
|
|
|
|
|
|
|
const UChar *UCharOffset = (UChar *)coll->image+getContractOffset(CE);
|
|
|
|
|
|
|
|
CE |= (contractionOffset & 0xFFF) << 12; // insert the offset in latin-1 table
|
|
|
|
|
|
|
|
coll->latinOneCEs[ch] = CE;
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+ch] = CE;
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+ch] = CE;
|
|
|
|
|
|
|
|
// We're going to jump into contraction table, pick the elements
|
|
|
|
// and use them
|
|
|
|
do {
|
|
|
|
CE = *(coll->contractionCEs +
|
|
|
|
(UCharOffset - coll->contractionIndex));
|
|
|
|
if(CE > UCOL_NOT_FOUND && getCETag(CE) == EXPANSION_TAG) {
|
|
|
|
uint32_t size;
|
|
|
|
uint32_t i; /* general counter */
|
|
|
|
uint32_t *CEOffset = (uint32_t *)coll->image+getExpansionOffset(CE); /* find the offset to expansion table */
|
|
|
|
size = getExpansionCount(CE);
|
|
|
|
//CE = *CEOffset++;
|
|
|
|
if(size != 0) { /* if there are less than 16 elements in expansion, we don't terminate */
|
|
|
|
for(i = 0; i<size; i++) {
|
|
|
|
if(primShift < 0 || secShift < 0 || terShift < 0) {
|
|
|
|
coll->latinOneCEs[(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
ucol_addLatinOneEntry(coll, (UChar)contractionOffset, *CEOffset++, &primShift, &secShift, &terShift);
|
|
|
|
}
|
|
|
|
} else { /* else, we do */
|
|
|
|
while(*CEOffset != 0) {
|
|
|
|
if(primShift < 0 || secShift < 0 || terShift < 0) {
|
|
|
|
coll->latinOneCEs[(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
ucol_addLatinOneEntry(coll, (UChar)contractionOffset, *CEOffset++, &primShift, &secShift, &terShift);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
contractionOffset++;
|
|
|
|
} else if(CE < UCOL_NOT_FOUND) {
|
|
|
|
ucol_addLatinOneEntry(coll, (UChar)contractionOffset++, CE, &primShift, &secShift, &terShift);
|
|
|
|
} else {
|
|
|
|
coll->latinOneCEs[(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[coll->latinOneTableLen+(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
coll->latinOneCEs[2*coll->latinOneTableLen+(UChar)contractionOffset] = UCOL_BAIL_OUT_CE;
|
|
|
|
contractionOffset++;
|
|
|
|
}
|
|
|
|
UCharOffset++;
|
|
|
|
primShift = 24; secShift = 24; terShift = 24;
|
|
|
|
if(contractionOffset == coll->latinOneTableLen) { // we need to reallocate
|
|
|
|
if(!ucol_resizeLatinOneTable(coll, 2*coll->latinOneTableLen, status)) {
|
|
|
|
goto cleanup_after_failure;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} while(*UCharOffset != 0xFFFF);
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2008-05-01 01:18:54 +00:00
|
|
|
break;;
|
|
|
|
case SPEC_PROC_TAG:
|
|
|
|
{
|
|
|
|
// 0xB7 is a precontext character defined in UCA5.1, a special
|
|
|
|
// handle is implemeted in order to save LatinOne table for
|
|
|
|
// most locales.
|
|
|
|
if (ch==0xb7) {
|
|
|
|
ucol_addLatinOneEntry(coll, ch, CE, &primShift, &secShift, &terShift);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
goto cleanup_after_failure;
|
|
|
|
}
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
goto cleanup_after_failure;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
// compact table
|
|
|
|
if(contractionOffset < coll->latinOneTableLen) {
|
|
|
|
if(!ucol_resizeLatinOneTable(coll, contractionOffset, status)) {
|
|
|
|
goto cleanup_after_failure;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
ucol_closeElements(it);
|
|
|
|
return result;
|
2006-03-30 05:29:25 +00:00
|
|
|
|
|
|
|
cleanup_after_failure:
|
2007-12-26 19:27:43 +00:00
|
|
|
// status should already be set before arriving here.
|
|
|
|
coll->latinOneFailed = TRUE;
|
|
|
|
ucol_closeElements(it);
|
|
|
|
return FALSE;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void ucol_updateInternalState(UCollator *coll, UErrorCode *status) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
if(coll->caseFirst == UCOL_UPPER_FIRST) {
|
|
|
|
coll->caseSwitch = UCOL_CASE_SWITCH;
|
|
|
|
} else {
|
|
|
|
coll->caseSwitch = UCOL_NO_CASE_SWITCH;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(coll->caseLevel == UCOL_ON || coll->caseFirst == UCOL_OFF) {
|
|
|
|
coll->tertiaryMask = UCOL_REMOVE_CASE;
|
|
|
|
coll->tertiaryCommon = UCOL_COMMON3_NORMAL;
|
|
|
|
coll->tertiaryAddition = (int8_t)UCOL_FLAG_BIT_MASK_CASE_SW_OFF; /* Should be 0x80 */
|
|
|
|
coll->tertiaryTop = UCOL_COMMON_TOP3_CASE_SW_OFF;
|
|
|
|
coll->tertiaryBottom = UCOL_COMMON_BOT3;
|
|
|
|
} else {
|
|
|
|
coll->tertiaryMask = UCOL_KEEP_CASE;
|
|
|
|
coll->tertiaryAddition = UCOL_FLAG_BIT_MASK_CASE_SW_ON;
|
|
|
|
if(coll->caseFirst == UCOL_UPPER_FIRST) {
|
|
|
|
coll->tertiaryCommon = UCOL_COMMON3_UPPERFIRST;
|
|
|
|
coll->tertiaryTop = UCOL_COMMON_TOP3_CASE_SW_UPPER;
|
|
|
|
coll->tertiaryBottom = UCOL_COMMON_BOTTOM3_CASE_SW_UPPER;
|
|
|
|
} else {
|
|
|
|
coll->tertiaryCommon = UCOL_COMMON3_NORMAL;
|
|
|
|
coll->tertiaryTop = UCOL_COMMON_TOP3_CASE_SW_LOWER;
|
|
|
|
coll->tertiaryBottom = UCOL_COMMON_BOTTOM3_CASE_SW_LOWER;
|
|
|
|
}
|
|
|
|
}
|
2001-05-15 17:38:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* Set the compression values */
|
|
|
|
uint8_t tertiaryTotal = (uint8_t)(coll->tertiaryTop - UCOL_COMMON_BOT3-1);
|
|
|
|
coll->tertiaryTopCount = (uint8_t)(UCOL_PROPORTION3*tertiaryTotal); /* we multilply double with int, but need only int */
|
|
|
|
coll->tertiaryBottomCount = (uint8_t)(tertiaryTotal - coll->tertiaryTopCount);
|
2001-05-15 17:38:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(coll->caseLevel == UCOL_OFF && coll->strength == UCOL_TERTIARY
|
2008-02-17 06:50:59 +00:00
|
|
|
&& coll->frenchCollation == UCOL_OFF && coll->alternateHandling == UCOL_NON_IGNORABLE)
|
|
|
|
{
|
|
|
|
coll->sortKeyGen = ucol_calcSortKeySimpleTertiary;
|
2002-09-04 06:02:13 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->sortKeyGen = ucol_calcSortKey;
|
|
|
|
}
|
|
|
|
if(coll->caseLevel == UCOL_OFF && coll->strength <= UCOL_TERTIARY && coll->numericCollation == UCOL_OFF
|
2008-02-17 06:50:59 +00:00
|
|
|
&& coll->alternateHandling == UCOL_NON_IGNORABLE && !coll->latinOneFailed)
|
|
|
|
{
|
|
|
|
if(coll->latinOneCEs == NULL || coll->latinOneRegenTable) {
|
|
|
|
if(ucol_setUpLatinOne(coll, status)) { // if we succeed in building latin1 table, we'll use it
|
|
|
|
//fprintf(stderr, "F");
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->latinOneUse = TRUE;
|
2008-02-17 06:50:59 +00:00
|
|
|
} else {
|
|
|
|
coll->latinOneUse = FALSE;
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2008-02-17 06:50:59 +00:00
|
|
|
if(*status == U_UNSUPPORTED_ERROR) {
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
}
|
|
|
|
} else { // latin1Table exists and it doesn't need to be regenerated, just use it
|
|
|
|
coll->latinOneUse = TRUE;
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else {
|
|
|
|
coll->latinOneUse = FALSE;
|
2006-03-30 05:29:25 +00:00
|
|
|
}
|
|
|
|
}
|
2001-03-02 00:19:43 +00:00
|
|
|
}
|
2001-01-16 00:28:40 +00:00
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI uint32_t U_EXPORT2
|
|
|
|
ucol_setVariableTop(UCollator *coll, const UChar *varTop, int32_t len, UErrorCode *status) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(U_FAILURE(*status) || coll == NULL) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if(len == -1) {
|
|
|
|
len = u_strlen(varTop);
|
|
|
|
}
|
|
|
|
if(len == 0) {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
return 0;
|
|
|
|
}
|
2001-06-25 04:01:49 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
collIterate s;
|
|
|
|
IInit_collIterate(coll, varTop, len, &s);
|
2001-06-25 04:01:49 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t CE = ucol_IGetNextCE(coll, &s, status);
|
2001-06-26 22:24:10 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
/* here we check if we have consumed all characters */
|
|
|
|
/* you can put in either one character or a contraction */
|
|
|
|
/* you shouldn't put more... */
|
|
|
|
if(s.pos != s.endp || CE == UCOL_NO_MORE_CES) {
|
|
|
|
*status = U_CE_NOT_FOUND_ERROR;
|
|
|
|
return 0;
|
|
|
|
}
|
2001-06-26 22:24:10 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
uint32_t nextCE = ucol_IGetNextCE(coll, &s, status);
|
2001-06-26 22:24:10 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(isContinuation(nextCE) && (nextCE & UCOL_PRIMARYMASK) != 0) {
|
|
|
|
*status = U_PRIMARY_TOO_LONG_ERROR;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if(coll->variableTopValue != (CE & UCOL_PRIMARYMASK)>>16) {
|
|
|
|
coll->variableTopValueisDefault = FALSE;
|
|
|
|
coll->variableTopValue = (CE & UCOL_PRIMARYMASK)>>16;
|
|
|
|
}
|
2008-06-09 21:18:46 +00:00
|
|
|
|
|
|
|
/* To avoid memory leak, free the offset buffer if necessary. */
|
|
|
|
freeOffsetBuffer(&s);
|
2001-06-25 04:01:49 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
return CE & UCOL_PRIMARYMASK;
|
2001-06-25 04:01:49 +00:00
|
|
|
}
|
|
|
|
|
2002-06-29 09:32:36 +00:00
|
|
|
U_CAPI uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(U_FAILURE(*status) || coll == NULL) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
return coll->variableTopValue<<16;
|
2001-06-26 22:24:10 +00:00
|
|
|
}
|
2001-06-25 04:01:49 +00:00
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
|
|
|
ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(U_FAILURE(*status) || coll == NULL) {
|
|
|
|
return;
|
|
|
|
}
|
2004-05-20 21:03:31 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(coll->variableTopValue != (varTop & UCOL_PRIMARYMASK)>>16) {
|
|
|
|
coll->variableTopValueisDefault = FALSE;
|
|
|
|
coll->variableTopValue = (varTop & UCOL_PRIMARYMASK)>>16;
|
|
|
|
}
|
2001-06-26 22:24:10 +00:00
|
|
|
}
|
2001-01-16 00:28:40 +00:00
|
|
|
/* Attribute setter API */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
|
|
|
ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status) {
|
2001-06-26 22:24:10 +00:00
|
|
|
if(U_FAILURE(*status) || coll == NULL) {
|
|
|
|
return;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
UColAttributeValue oldFrench = coll->frenchCollation;
|
|
|
|
UColAttributeValue oldCaseFirst = coll->caseFirst;
|
2001-03-03 03:35:17 +00:00
|
|
|
switch(attr) {
|
2003-07-25 05:31:54 +00:00
|
|
|
case UCOL_NUMERIC_COLLATION: /* sort substrings of digits as numbers */
|
2007-12-26 19:27:43 +00:00
|
|
|
if(value == UCOL_ON) {
|
|
|
|
coll->numericCollation = UCOL_ON;
|
|
|
|
coll->numericCollationisDefault = FALSE;
|
|
|
|
} else if (value == UCOL_OFF) {
|
|
|
|
coll->numericCollation = UCOL_OFF;
|
|
|
|
coll->numericCollationisDefault = FALSE;
|
|
|
|
} else if (value == UCOL_DEFAULT) {
|
|
|
|
coll->numericCollationisDefault = TRUE;
|
|
|
|
coll->numericCollation = (UColAttributeValue)coll->options->numericCollation;
|
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
}
|
|
|
|
break;
|
2001-10-02 16:49:57 +00:00
|
|
|
case UCOL_HIRAGANA_QUATERNARY_MODE: /* special quaternary values for Hiragana */
|
2007-12-26 19:27:43 +00:00
|
|
|
if(value == UCOL_ON) {
|
|
|
|
coll->hiraganaQ = UCOL_ON;
|
|
|
|
coll->hiraganaQisDefault = FALSE;
|
|
|
|
} else if (value == UCOL_OFF) {
|
|
|
|
coll->hiraganaQ = UCOL_OFF;
|
|
|
|
coll->hiraganaQisDefault = FALSE;
|
|
|
|
} else if (value == UCOL_DEFAULT) {
|
|
|
|
coll->hiraganaQisDefault = TRUE;
|
|
|
|
coll->hiraganaQ = (UColAttributeValue)coll->options->hiraganaQ;
|
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
}
|
|
|
|
break;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_FRENCH_COLLATION: /* attribute for direction of secondary weights*/
|
|
|
|
if(value == UCOL_ON) {
|
|
|
|
coll->frenchCollation = UCOL_ON;
|
2001-01-16 00:28:40 +00:00
|
|
|
coll->frenchCollationisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_OFF) {
|
|
|
|
coll->frenchCollation = UCOL_OFF;
|
2001-01-16 00:28:40 +00:00
|
|
|
coll->frenchCollationisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_DEFAULT) {
|
2001-01-16 00:28:40 +00:00
|
|
|
coll->frenchCollationisDefault = TRUE;
|
2002-09-17 04:53:35 +00:00
|
|
|
coll->frenchCollation = (UColAttributeValue)coll->options->frenchCollation;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR ;
|
|
|
|
}
|
|
|
|
break;
|
2001-01-05 00:47:25 +00:00
|
|
|
case UCOL_ALTERNATE_HANDLING: /* attribute for handling variable elements*/
|
2001-03-03 03:35:17 +00:00
|
|
|
if(value == UCOL_SHIFTED) {
|
|
|
|
coll->alternateHandling = UCOL_SHIFTED;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->alternateHandlingisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_NON_IGNORABLE) {
|
|
|
|
coll->alternateHandling = UCOL_NON_IGNORABLE;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->alternateHandlingisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_DEFAULT) {
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->alternateHandlingisDefault = TRUE;
|
2002-09-17 04:53:35 +00:00
|
|
|
coll->alternateHandling = (UColAttributeValue)coll->options->alternateHandling ;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR ;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case UCOL_CASE_FIRST: /* who goes first, lower case or uppercase */
|
|
|
|
if(value == UCOL_LOWER_FIRST) {
|
|
|
|
coll->caseFirst = UCOL_LOWER_FIRST;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->caseFirstisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_UPPER_FIRST) {
|
|
|
|
coll->caseFirst = UCOL_UPPER_FIRST;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->caseFirstisDefault = FALSE;
|
2001-03-02 00:19:43 +00:00
|
|
|
} else if (value == UCOL_OFF) {
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->caseFirst = UCOL_OFF;
|
|
|
|
coll->caseFirstisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_DEFAULT) {
|
2002-09-17 04:53:35 +00:00
|
|
|
coll->caseFirst = (UColAttributeValue)coll->options->caseFirst;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->caseFirstisDefault = TRUE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR ;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case UCOL_CASE_LEVEL: /* do we have an extra case level */
|
|
|
|
if(value == UCOL_ON) {
|
|
|
|
coll->caseLevel = UCOL_ON;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->caseLevelisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_OFF) {
|
|
|
|
coll->caseLevel = UCOL_OFF;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->caseLevelisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_DEFAULT) {
|
2002-09-17 04:53:35 +00:00
|
|
|
coll->caseLevel = (UColAttributeValue)coll->options->caseLevel;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->caseLevelisDefault = TRUE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR ;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case UCOL_NORMALIZATION_MODE: /* attribute for normalization */
|
|
|
|
if(value == UCOL_ON) {
|
2001-01-05 00:47:25 +00:00
|
|
|
coll->normalizationMode = UCOL_ON;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->normalizationModeisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_OFF) {
|
2001-01-05 00:47:25 +00:00
|
|
|
coll->normalizationMode = UCOL_OFF;
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->normalizationModeisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value == UCOL_DEFAULT) {
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->normalizationModeisDefault = TRUE;
|
2002-09-17 04:53:35 +00:00
|
|
|
coll->normalizationMode = (UColAttributeValue)coll->options->normalizationMode;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR ;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case UCOL_STRENGTH: /* attribute for strength */
|
2001-01-05 00:47:25 +00:00
|
|
|
if (value == UCOL_DEFAULT) {
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->strengthisDefault = TRUE;
|
2002-09-17 04:53:35 +00:00
|
|
|
coll->strength = (UColAttributeValue)coll->options->strength;
|
2001-03-03 03:35:17 +00:00
|
|
|
} else if (value <= UCOL_IDENTICAL) {
|
2001-01-09 00:52:18 +00:00
|
|
|
coll->strengthisDefault = FALSE;
|
2001-03-03 03:35:17 +00:00
|
|
|
coll->strength = value;
|
|
|
|
} else {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR ;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case UCOL_ATTRIBUTE_COUNT:
|
|
|
|
default:
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
break;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
if(oldFrench != coll->frenchCollation || oldCaseFirst != coll->caseFirst) {
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->latinOneRegenTable = TRUE;
|
2003-12-03 00:45:12 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
coll->latinOneRegenTable = FALSE;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
ucol_updateInternalState(coll, status);
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI UColAttributeValue U_EXPORT2
|
|
|
|
ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status) {
|
2001-06-26 22:24:10 +00:00
|
|
|
if(U_FAILURE(*status) || coll == NULL) {
|
|
|
|
return UCOL_DEFAULT;
|
|
|
|
}
|
2001-03-03 03:35:17 +00:00
|
|
|
switch(attr) {
|
2003-07-25 05:31:54 +00:00
|
|
|
case UCOL_NUMERIC_COLLATION:
|
2003-12-03 00:45:12 +00:00
|
|
|
return coll->numericCollation;
|
2001-10-02 16:49:57 +00:00
|
|
|
case UCOL_HIRAGANA_QUATERNARY_MODE:
|
|
|
|
return coll->hiraganaQ;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_FRENCH_COLLATION: /* attribute for direction of secondary weights*/
|
2001-03-23 23:48:26 +00:00
|
|
|
return coll->frenchCollation;
|
2001-01-05 00:47:25 +00:00
|
|
|
case UCOL_ALTERNATE_HANDLING: /* attribute for handling variable elements*/
|
2001-03-23 23:48:26 +00:00
|
|
|
return coll->alternateHandling;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_CASE_FIRST: /* who goes first, lower case or uppercase */
|
2001-03-23 23:48:26 +00:00
|
|
|
return coll->caseFirst;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_CASE_LEVEL: /* do we have an extra case level */
|
2001-03-23 23:48:26 +00:00
|
|
|
return coll->caseLevel;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_NORMALIZATION_MODE: /* attribute for normalization */
|
2001-03-23 23:48:26 +00:00
|
|
|
return coll->normalizationMode;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_STRENGTH: /* attribute for strength */
|
2001-03-23 23:48:26 +00:00
|
|
|
return coll->strength;
|
2001-03-03 03:35:17 +00:00
|
|
|
case UCOL_ATTRIBUTE_COUNT:
|
|
|
|
default:
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return UCOL_DEFAULT;
|
2001-01-05 00:47:25 +00:00
|
|
|
}
|
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_setStrength( UCollator *coll,
|
|
|
|
UCollationStrength strength)
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
|
|
|
ucol_setAttribute(coll, UCOL_STRENGTH, strength, &status);
|
2001-01-16 00:28:40 +00:00
|
|
|
}
|
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI UCollationStrength U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_getStrength(const UCollator *coll)
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
|
|
|
return ucol_getAttribute(coll, UCOL_STRENGTH, &status);
|
2001-01-16 00:28:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/****************************************************************************/
|
|
|
|
/* Following are misc functions */
|
|
|
|
/* there are new APIs and some compatibility APIs */
|
|
|
|
/****************************************************************************/
|
|
|
|
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
2001-04-23 01:53:49 +00:00
|
|
|
ucol_getVersion(const UCollator* coll,
|
|
|
|
UVersionInfo versionInfo)
|
2001-01-16 00:28:40 +00:00
|
|
|
{
|
2001-02-10 02:42:54 +00:00
|
|
|
/* RunTime version */
|
|
|
|
uint8_t rtVersion = UCOL_RUNTIME_VERSION;
|
2001-02-26 23:52:44 +00:00
|
|
|
/* Builder version*/
|
2003-04-24 07:00:27 +00:00
|
|
|
uint8_t bdVersion = coll->image->version[0];
|
2001-02-26 23:52:44 +00:00
|
|
|
|
2001-02-10 02:42:54 +00:00
|
|
|
/* Charset Version. Need to get the version from cnv files
|
2001-04-23 01:53:49 +00:00
|
|
|
* makeconv should populate cnv files with version and
|
2001-02-10 02:42:54 +00:00
|
|
|
* an api has to be provided in ucnv.h to obtain this version
|
|
|
|
*/
|
|
|
|
uint8_t csVersion = 0;
|
|
|
|
|
|
|
|
/* combine the version info */
|
2001-02-28 19:30:41 +00:00
|
|
|
uint16_t cmbVersion = (uint16_t)((rtVersion<<11) | (bdVersion<<6) | (csVersion));
|
2001-02-10 02:42:54 +00:00
|
|
|
|
2001-02-26 23:52:44 +00:00
|
|
|
/* Tailoring rules */
|
2001-02-28 19:30:41 +00:00
|
|
|
versionInfo[0] = (uint8_t)(cmbVersion>>8);
|
2001-02-10 02:42:54 +00:00
|
|
|
versionInfo[1] = (uint8_t)cmbVersion;
|
2003-04-24 07:00:27 +00:00
|
|
|
versionInfo[2] = coll->image->version[1];
|
2004-10-27 00:31:04 +00:00
|
|
|
if(coll->UCA) {
|
2008-08-08 15:04:18 +00:00
|
|
|
/* Include the minor number when getting the UCA version. (major & 1f) << 3 | (minor & 7) */
|
|
|
|
versionInfo[3] = (coll->UCA->image->UCAVersion[0] & 0x1f) << 3 | (coll->UCA->image->UCAVersion[1] & 0x07);
|
2004-10-27 00:31:04 +00:00
|
|
|
} else {
|
|
|
|
versionInfo[3] = 0;
|
|
|
|
}
|
2001-03-03 03:35:17 +00:00
|
|
|
}
|
2001-02-26 23:52:44 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
|
2001-03-22 18:45:31 +00:00
|
|
|
/* This internal API checks whether a character is tailored or not */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI UBool U_EXPORT2
|
2002-07-16 01:46:42 +00:00
|
|
|
ucol_isTailored(const UCollator *coll, const UChar u, UErrorCode *status) {
|
2007-06-13 19:03:09 +00:00
|
|
|
if(U_FAILURE(*status) || coll == NULL || coll == coll->UCA) {
|
2001-03-22 18:45:31 +00:00
|
|
|
return FALSE;
|
2007-06-13 19:03:09 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
uint32_t CE = UCOL_NOT_FOUND;
|
|
|
|
const UChar *ContractionStart = NULL;
|
|
|
|
if(u < 0x100) { /* latin-1 */
|
|
|
|
CE = coll->latinOneMapping[u];
|
|
|
|
if(coll->UCA && CE == coll->UCA->latinOneMapping[u]) {
|
|
|
|
return FALSE;
|
|
|
|
}
|
2001-03-22 18:45:31 +00:00
|
|
|
} else { /* regular */
|
2007-06-13 19:03:09 +00:00
|
|
|
CE = UTRIE_GET32_FROM_LEAD(&coll->mapping, u);
|
2001-03-22 18:45:31 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if(isContraction(CE)) {
|
2007-06-13 19:03:09 +00:00
|
|
|
ContractionStart = (UChar *)coll->image+getContractOffset(CE);
|
|
|
|
CE = *(coll->contractionCEs + (ContractionStart- coll->contractionIndex));
|
2001-03-22 18:45:31 +00:00
|
|
|
}
|
|
|
|
|
2007-06-13 19:03:09 +00:00
|
|
|
return (UBool)(CE != UCOL_NOT_FOUND);
|
2001-03-22 18:45:31 +00:00
|
|
|
}
|
|
|
|
|
2001-03-03 03:35:17 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/****************************************************************************/
|
|
|
|
/* Following are the string compare functions */
|
|
|
|
/* */
|
|
|
|
/****************************************************************************/
|
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
|
|
|
|
/* ucol_checkIdent internal function. Does byte level string compare. */
|
|
|
|
/* Used by strcoll if strength == identical and strings */
|
|
|
|
/* are otherwise equal. Moved out-of-line because this */
|
|
|
|
/* is a rare case. */
|
|
|
|
/* */
|
|
|
|
/* Comparison must be done on NFD normalized strings. */
|
|
|
|
/* FCD is not good enough. */
|
|
|
|
/* */
|
|
|
|
/* TODO: make an incremental NFD Comparison function, which could */
|
|
|
|
/* be of general use */
|
|
|
|
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2003-01-20 07:43:32 +00:00
|
|
|
UCollationResult ucol_checkIdent(collIterate *sColl, collIterate *tColl, UBool normalize, UErrorCode *status)
|
2001-04-06 23:37:48 +00:00
|
|
|
{
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
// TODO: When we have an UChar iterator, we need to access the whole string. One
|
2003-01-20 07:43:32 +00:00
|
|
|
// useful modification would be a UChar iterator extract API, since reset next next...
|
|
|
|
// is not optimal.
|
|
|
|
// TODO: Handle long strings. Do the same in compareUsingSortKeys.
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
// When we arrive here, we can have normal strings or UCharIterators. Currently they are both
|
2003-12-03 00:45:12 +00:00
|
|
|
// of same type, but that doesn't really mean that it will stay that way.
|
2003-03-17 21:20:36 +00:00
|
|
|
|
2003-05-06 18:53:42 +00:00
|
|
|
// The division for the array length may truncate the array size to
|
|
|
|
// a little less than UNORM_ITER_SIZE, but that size is dimensioned too high
|
|
|
|
// for all platforms anyway.
|
2003-03-17 21:20:36 +00:00
|
|
|
UAlignedMemory stackNormIter1[UNORM_ITER_SIZE/sizeof(UAlignedMemory)];
|
|
|
|
UAlignedMemory stackNormIter2[UNORM_ITER_SIZE/sizeof(UAlignedMemory)];
|
2003-02-06 23:29:56 +00:00
|
|
|
//UChar sStackBuf[256], tStackBuf[256];
|
|
|
|
//int32_t sBufSize = 256, tBufSize = 256;
|
2003-01-20 07:43:32 +00:00
|
|
|
int32_t comparison;
|
|
|
|
int32_t sLen = 0;
|
|
|
|
UChar *sBuf = NULL;
|
|
|
|
int32_t tLen = 0;
|
|
|
|
UChar *tBuf = NULL;
|
|
|
|
UBool freeSBuf = FALSE, freeTBuf = FALSE;
|
|
|
|
|
|
|
|
if (sColl->flags & UCOL_USE_ITERATOR) {
|
2007-12-26 19:27:43 +00:00
|
|
|
UNormIterator *sNIt = NULL, *tNIt = NULL;
|
|
|
|
sNIt = unorm_openIter(stackNormIter1, sizeof(stackNormIter1), status);
|
|
|
|
tNIt = unorm_openIter(stackNormIter2, sizeof(stackNormIter2), status);
|
|
|
|
sColl->iterator->move(sColl->iterator, 0, UITER_START);
|
|
|
|
tColl->iterator->move(tColl->iterator, 0, UITER_START);
|
|
|
|
UCharIterator *sIt = unorm_setIter(sNIt, sColl->iterator, UNORM_NFD, status);
|
|
|
|
UCharIterator *tIt = unorm_setIter(tNIt, tColl->iterator, UNORM_NFD, status);
|
|
|
|
comparison = u_strCompareIter(sIt, tIt, TRUE);
|
|
|
|
unorm_closeIter(sNIt);
|
|
|
|
unorm_closeIter(tNIt);
|
2003-01-20 07:43:32 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
sLen = (sColl->flags & UCOL_ITER_HASLEN) ? sColl->endp - sColl->string : -1;
|
|
|
|
sBuf = sColl->string;
|
|
|
|
tLen = (tColl->flags & UCOL_ITER_HASLEN) ? tColl->endp - tColl->string : -1;
|
|
|
|
tBuf = tColl->string;
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (normalize) {
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
if (unorm_quickCheck(sBuf, sLen, UNORM_NFD, status) != UNORM_YES) {
|
|
|
|
sLen = unorm_decompose(sColl->writableBuffer, (int32_t)sColl->writableBufSize,
|
|
|
|
sBuf, sLen,
|
|
|
|
FALSE, 0,
|
|
|
|
status);
|
|
|
|
if(*status == U_BUFFER_OVERFLOW_ERROR) {
|
|
|
|
if(!u_growBufferFromStatic(sColl->stackWritableBuffer,
|
|
|
|
&sColl->writableBuffer,
|
|
|
|
(int32_t *)&sColl->writableBufSize, sLen,
|
|
|
|
0)
|
|
|
|
)
|
|
|
|
{
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return UCOL_LESS; /* TODO set *status = U_MEMORY_ALLOCATION_ERROR; */
|
|
|
|
}
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
sLen = unorm_decompose(sColl->writableBuffer, (int32_t)sColl->writableBufSize,
|
|
|
|
sBuf, sLen,
|
|
|
|
FALSE, 0,
|
|
|
|
status);
|
|
|
|
}
|
|
|
|
if(freeSBuf) {
|
|
|
|
uprv_free(sBuf);
|
|
|
|
freeSBuf = FALSE;
|
|
|
|
}
|
|
|
|
sBuf = sColl->writableBuffer;
|
|
|
|
if (sBuf != sColl->stackWritableBuffer) {
|
|
|
|
sColl->flags |= UCOL_ITER_ALLOCATED;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
if (unorm_quickCheck(tBuf, tLen, UNORM_NFD, status) != UNORM_YES) {
|
|
|
|
tLen = unorm_decompose(tColl->writableBuffer, (int32_t)tColl->writableBufSize,
|
|
|
|
tBuf, tLen,
|
|
|
|
FALSE, 0,
|
|
|
|
status);
|
|
|
|
if(*status == U_BUFFER_OVERFLOW_ERROR) {
|
|
|
|
if(!u_growBufferFromStatic(tColl->stackWritableBuffer,
|
|
|
|
&tColl->writableBuffer,
|
|
|
|
(int32_t *)&tColl->writableBufSize, tLen,
|
|
|
|
0)
|
|
|
|
)
|
|
|
|
{
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
return UCOL_LESS; /* TODO set *status = U_MEMORY_ALLOCATION_ERROR; */
|
|
|
|
}
|
|
|
|
*status = U_ZERO_ERROR;
|
|
|
|
tLen = unorm_decompose(tColl->writableBuffer, (int32_t)tColl->writableBufSize,
|
|
|
|
tBuf, tLen,
|
|
|
|
FALSE, 0,
|
|
|
|
status);
|
|
|
|
}
|
|
|
|
if(freeTBuf) {
|
|
|
|
uprv_free(tBuf);
|
|
|
|
freeTBuf = FALSE;
|
|
|
|
}
|
|
|
|
tBuf = tColl->writableBuffer;
|
|
|
|
if (tBuf != tColl->stackWritableBuffer) {
|
|
|
|
tColl->flags |= UCOL_ITER_ALLOCATED;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (sLen == -1 && tLen == -1) {
|
|
|
|
comparison = u_strcmpCodePointOrder(sBuf, tBuf);
|
|
|
|
} else {
|
|
|
|
if (sLen == -1) {
|
|
|
|
sLen = u_strlen(sBuf);
|
|
|
|
}
|
|
|
|
if (tLen == -1) {
|
|
|
|
tLen = u_strlen(tBuf);
|
|
|
|
}
|
|
|
|
comparison = u_memcmpCodePointOrder(sBuf, tBuf, uprv_min(sLen, tLen));
|
|
|
|
if (comparison == 0) {
|
|
|
|
comparison = sLen - tLen;
|
|
|
|
}
|
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2001-09-27 01:01:30 +00:00
|
|
|
if (comparison < 0) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
} else if (comparison == 0) {
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
} else /* comparison > 0 */ {
|
|
|
|
return UCOL_GREATER;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
/* CEBuf - A struct and some inline functions to handle the saving */
|
|
|
|
/* of CEs in a buffer within ucol_strcoll */
|
|
|
|
|
|
|
|
#define UCOL_CEBUF_SIZE 512
|
|
|
|
typedef struct ucol_CEBuf {
|
|
|
|
uint32_t *buf;
|
|
|
|
uint32_t *endp;
|
|
|
|
uint32_t *pos;
|
|
|
|
uint32_t localArray[UCOL_CEBUF_SIZE];
|
|
|
|
} ucol_CEBuf;
|
2001-04-06 23:37:48 +00:00
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2001-04-18 19:31:05 +00:00
|
|
|
inline void UCOL_INIT_CEBUF(ucol_CEBuf *b) {
|
|
|
|
(b)->buf = (b)->pos = (b)->localArray;
|
|
|
|
(b)->endp = (b)->buf + UCOL_CEBUF_SIZE;
|
2004-12-30 07:25:51 +00:00
|
|
|
}
|
2001-04-18 19:31:05 +00:00
|
|
|
|
2001-10-20 01:09:31 +00:00
|
|
|
static
|
2008-02-10 20:17:14 +00:00
|
|
|
void ucol_CEBuf_Expand(ucol_CEBuf *b, collIterate *ci, UErrorCode *status) {
|
2001-04-06 23:37:48 +00:00
|
|
|
uint32_t oldSize;
|
|
|
|
uint32_t newSize;
|
|
|
|
uint32_t *newBuf;
|
|
|
|
|
2001-04-18 19:31:05 +00:00
|
|
|
ci->flags |= UCOL_ITER_ALLOCATED;
|
2001-04-06 23:37:48 +00:00
|
|
|
oldSize = b->pos - b->buf;
|
|
|
|
newSize = oldSize * 2;
|
|
|
|
newBuf = (uint32_t *)uprv_malloc(newSize * sizeof(uint32_t));
|
2008-02-10 20:17:14 +00:00
|
|
|
if(newBuf == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
}
|
|
|
|
else {
|
2007-12-26 19:27:43 +00:00
|
|
|
uprv_memcpy(newBuf, b->buf, oldSize * sizeof(uint32_t));
|
|
|
|
if (b->buf != b->localArray) {
|
|
|
|
uprv_free(b->buf);
|
|
|
|
}
|
|
|
|
b->buf = newBuf;
|
|
|
|
b->endp = b->buf + newSize;
|
|
|
|
b->pos = b->buf + oldSize;
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
|
|
|
|
2001-10-22 05:30:22 +00:00
|
|
|
static
|
2008-02-10 20:17:14 +00:00
|
|
|
inline void UCOL_CEBUF_PUT(ucol_CEBuf *b, uint32_t ce, collIterate *ci, UErrorCode *status) {
|
2001-05-17 23:09:35 +00:00
|
|
|
if (b->pos == b->endp) {
|
2008-02-10 20:17:14 +00:00
|
|
|
ucol_CEBuf_Expand(b, ci, status);
|
|
|
|
}
|
|
|
|
if (U_SUCCESS(*status)) {
|
|
|
|
*(b)->pos++ = ce;
|
2007-12-26 21:44:39 +00:00
|
|
|
}
|
2004-12-30 07:25:51 +00:00
|
|
|
}
|
2001-04-18 19:31:05 +00:00
|
|
|
|
2001-10-13 16:20:01 +00:00
|
|
|
/* This is a trick string compare function that goes in and uses sortkeys to compare */
|
|
|
|
/* It is used when compare gets in trouble and needs to bail out */
|
2003-01-20 07:43:32 +00:00
|
|
|
static UCollationResult ucol_compareUsingSortKeys(collIterate *sColl,
|
2006-03-30 16:54:13 +00:00
|
|
|
collIterate *tColl,
|
|
|
|
UErrorCode *status)
|
2001-10-13 16:20:01 +00:00
|
|
|
{
|
|
|
|
uint8_t sourceKey[UCOL_MAX_BUFFER], targetKey[UCOL_MAX_BUFFER];
|
|
|
|
uint8_t *sourceKeyP = sourceKey;
|
|
|
|
uint8_t *targetKeyP = targetKey;
|
|
|
|
int32_t sourceKeyLen = UCOL_MAX_BUFFER, targetKeyLen = UCOL_MAX_BUFFER;
|
2003-01-20 07:43:32 +00:00
|
|
|
const UCollator *coll = sColl->coll;
|
|
|
|
UChar *source = NULL;
|
|
|
|
UChar *target = NULL;
|
2006-03-30 16:54:13 +00:00
|
|
|
int32_t result = UCOL_EQUAL;
|
2003-01-20 07:43:32 +00:00
|
|
|
UChar sStackBuf[256], tStackBuf[256];
|
|
|
|
int32_t sourceLength = (sColl->flags&UCOL_ITER_HASLEN)?(sColl->endp-sColl->string):-1;
|
|
|
|
int32_t targetLength = (tColl->flags&UCOL_ITER_HASLEN)?(tColl->endp-tColl->string):-1;
|
|
|
|
|
|
|
|
// TODO: Handle long strings. Do the same in ucol_checkIdent.
|
|
|
|
if(sColl->flags & UCOL_USE_ITERATOR) {
|
2006-03-30 16:54:13 +00:00
|
|
|
sColl->iterator->move(sColl->iterator, 0, UITER_START);
|
|
|
|
tColl->iterator->move(tColl->iterator, 0, UITER_START);
|
|
|
|
source = sStackBuf;
|
|
|
|
UChar *sBufp = source;
|
|
|
|
target = tStackBuf;
|
|
|
|
UChar *tBufp = target;
|
|
|
|
while(sColl->iterator->hasNext(sColl->iterator)) {
|
|
|
|
*sBufp++ = (UChar)sColl->iterator->next(sColl->iterator);
|
|
|
|
}
|
|
|
|
while(tColl->iterator->hasNext(tColl->iterator)) {
|
|
|
|
*tBufp++ = (UChar)tColl->iterator->next(tColl->iterator);
|
|
|
|
}
|
|
|
|
sourceLength = sBufp - source;
|
|
|
|
targetLength = tBufp - target;
|
2003-01-20 07:43:32 +00:00
|
|
|
} else { // no iterators
|
2006-03-30 16:54:13 +00:00
|
|
|
sourceLength = (sColl->flags&UCOL_ITER_HASLEN)?(sColl->endp-sColl->string):-1;
|
|
|
|
targetLength = (tColl->flags&UCOL_ITER_HASLEN)?(tColl->endp-tColl->string):-1;
|
|
|
|
source = sColl->string;
|
|
|
|
target = tColl->string;
|
2003-01-20 07:43:32 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2001-10-13 16:20:01 +00:00
|
|
|
|
|
|
|
sourceKeyLen = ucol_getSortKey(coll, source, sourceLength, sourceKeyP, sourceKeyLen);
|
|
|
|
if(sourceKeyLen > UCOL_MAX_BUFFER) {
|
|
|
|
sourceKeyP = (uint8_t*)uprv_malloc(sourceKeyLen*sizeof(uint8_t));
|
2006-03-30 16:54:13 +00:00
|
|
|
if(sourceKeyP == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
goto cleanup_and_do_compare;
|
2002-07-20 06:00:04 +00:00
|
|
|
}
|
2006-03-30 16:54:13 +00:00
|
|
|
sourceKeyLen = ucol_getSortKey(coll, source, sourceLength, sourceKeyP, sourceKeyLen);
|
2001-10-13 16:20:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
targetKeyLen = ucol_getSortKey(coll, target, targetLength, targetKeyP, targetKeyLen);
|
|
|
|
if(targetKeyLen > UCOL_MAX_BUFFER) {
|
|
|
|
targetKeyP = (uint8_t*)uprv_malloc(targetKeyLen*sizeof(uint8_t));
|
2006-03-30 16:54:13 +00:00
|
|
|
if(targetKeyP == NULL) {
|
|
|
|
*status = U_MEMORY_ALLOCATION_ERROR;
|
|
|
|
goto cleanup_and_do_compare;
|
2002-07-20 06:00:04 +00:00
|
|
|
}
|
2006-03-30 16:54:13 +00:00
|
|
|
targetKeyLen = ucol_getSortKey(coll, target, targetLength, targetKeyP, targetKeyLen);
|
2001-10-13 16:20:01 +00:00
|
|
|
}
|
|
|
|
|
2006-03-30 16:54:13 +00:00
|
|
|
result = uprv_strcmp((const char*)sourceKeyP, (const char*)targetKeyP);
|
2001-10-13 16:20:01 +00:00
|
|
|
|
2006-03-30 16:54:13 +00:00
|
|
|
cleanup_and_do_compare:
|
|
|
|
if(sourceKeyP != NULL && sourceKeyP != sourceKey) {
|
2001-10-13 16:20:01 +00:00
|
|
|
uprv_free(sourceKeyP);
|
|
|
|
}
|
|
|
|
|
2006-03-30 16:54:13 +00:00
|
|
|
if(targetKeyP != NULL && targetKeyP != targetKey) {
|
2001-10-13 16:20:01 +00:00
|
|
|
uprv_free(targetKeyP);
|
|
|
|
}
|
|
|
|
|
|
|
|
if(result<0) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
} else if(result>0) {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
} else {
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
|
|
|
}
|
2001-04-06 23:37:48 +00:00
|
|
|
|
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
static inline UCollationResult
|
2003-01-20 07:43:32 +00:00
|
|
|
ucol_strcollRegular( collIterate *sColl, collIterate *tColl,
|
|
|
|
// const UCollator *coll,
|
|
|
|
// const UChar *source,
|
|
|
|
// int32_t sourceLength,
|
|
|
|
// const UChar *target,
|
|
|
|
// int32_t targetLength,
|
2002-09-04 06:02:13 +00:00
|
|
|
UErrorCode *status)
|
2001-01-05 00:47:25 +00:00
|
|
|
{
|
2001-05-10 22:12:53 +00:00
|
|
|
U_ALIGN_CODE(16);
|
2001-01-05 00:47:25 +00:00
|
|
|
|
2003-01-20 07:43:32 +00:00
|
|
|
const UCollator *coll = sColl->coll;
|
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-05-22 22:26:58 +00:00
|
|
|
// setting up the collator parameters
|
2001-03-14 02:45:39 +00:00
|
|
|
UColAttributeValue strength = coll->strength;
|
|
|
|
UBool initialCheckSecTer = (strength >= UCOL_SECONDARY);
|
|
|
|
|
|
|
|
UBool checkSecTer = initialCheckSecTer;
|
|
|
|
UBool checkTertiary = (strength >= UCOL_TERTIARY);
|
|
|
|
UBool checkQuad = (strength >= UCOL_QUATERNARY);
|
|
|
|
UBool checkIdent = (strength == UCOL_IDENTICAL);
|
|
|
|
UBool checkCase = (coll->caseLevel == UCOL_ON);
|
|
|
|
UBool isFrenchSec = (coll->frenchCollation == UCOL_ON) && checkSecTer;
|
2001-03-15 02:49:35 +00:00
|
|
|
UBool shifted = (coll->alternateHandling == UCOL_SHIFTED);
|
|
|
|
UBool qShifted = shifted && checkQuad;
|
2001-10-10 01:48:36 +00:00
|
|
|
UBool doHiragana = (coll->hiraganaQ == UCOL_ON) && checkQuad;
|
2001-03-14 02:45:39 +00:00
|
|
|
|
2008-04-17 05:19:19 +00:00
|
|
|
if(doHiragana && shifted) {
|
2007-12-26 19:27:43 +00:00
|
|
|
return (ucol_compareUsingSortKeys(sColl, tColl, status));
|
2001-10-13 16:20:01 +00:00
|
|
|
}
|
2001-05-22 22:26:58 +00:00
|
|
|
uint8_t caseSwitch = coll->caseSwitch;
|
|
|
|
uint8_t tertiaryMask = coll->tertiaryMask;
|
|
|
|
|
|
|
|
// This is the lowest primary value that will not be ignored if shifted
|
2001-06-26 22:24:10 +00:00
|
|
|
uint32_t LVT = (shifted)?(coll->variableTopValue<<16):0;
|
2001-05-22 22:26:58 +00:00
|
|
|
|
2001-01-15 07:28:54 +00:00
|
|
|
UCollationResult result = UCOL_EQUAL;
|
2001-10-05 02:07:51 +00:00
|
|
|
UCollationResult hirResult = UCOL_EQUAL;
|
2001-01-05 00:47:25 +00:00
|
|
|
|
2001-05-22 22:26:58 +00:00
|
|
|
// Preparing the CE buffers. They will be filled during the primary phase
|
2001-04-06 23:37:48 +00:00
|
|
|
ucol_CEBuf sCEs;
|
|
|
|
ucol_CEBuf tCEs;
|
|
|
|
UCOL_INIT_CEBUF(&sCEs);
|
|
|
|
UCOL_INIT_CEBUF(&tCEs);
|
|
|
|
|
2001-01-24 16:18:48 +00:00
|
|
|
uint32_t secS = 0, secT = 0;
|
2001-01-24 00:35:19 +00:00
|
|
|
uint32_t sOrder=0, tOrder=0;
|
2001-05-22 22:26:58 +00:00
|
|
|
|
|
|
|
// Non shifted primary processing is quite simple
|
2001-01-25 06:51:18 +00:00
|
|
|
if(!shifted) {
|
2001-01-26 00:12:23 +00:00
|
|
|
for(;;) {
|
2007-12-26 19:27:43 +00:00
|
|
|
|
|
|
|
// We fetch CEs until we hit a non ignorable primary or end.
|
|
|
|
do {
|
|
|
|
// We get the next CE
|
|
|
|
sOrder = ucol_IGetNextCE(coll, sColl, status);
|
|
|
|
// Stuff it in the buffer
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
// And keep just the primary part.
|
2001-05-21 22:30:49 +00:00
|
|
|
sOrder &= UCOL_PRIMARYMASK;
|
2007-12-26 19:27:43 +00:00
|
|
|
} while(sOrder == 0);
|
2001-01-22 23:48:48 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
// see the comments on the above block
|
|
|
|
do {
|
|
|
|
tOrder = ucol_IGetNextCE(coll, tColl, status);
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2001-05-21 22:30:49 +00:00
|
|
|
tOrder &= UCOL_PRIMARYMASK;
|
2007-12-26 19:27:43 +00:00
|
|
|
} while(tOrder == 0);
|
|
|
|
|
|
|
|
// if both primaries are the same
|
|
|
|
if(sOrder == tOrder) {
|
|
|
|
// and there are no more CEs, we advance to the next level
|
|
|
|
if(sOrder == UCOL_NO_MORE_CES_PRIMARY) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if(doHiragana && hirResult == UCOL_EQUAL) {
|
|
|
|
if((sColl->flags & UCOL_WAS_HIRAGANA) != (tColl->flags & UCOL_WAS_HIRAGANA)) {
|
|
|
|
hirResult = ((sColl->flags & UCOL_WAS_HIRAGANA) > (tColl->flags & UCOL_WAS_HIRAGANA))
|
|
|
|
? UCOL_LESS:UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// if two primaries are different, we are done
|
|
|
|
result = (sOrder < tOrder) ? UCOL_LESS: UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
2000-11-20 19:17:17 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} // no primary difference... do the rest from the buffers
|
|
|
|
} else { // shifted - do a slightly more complicated processing :)
|
|
|
|
for(;;) {
|
|
|
|
UBool sInShifted = FALSE;
|
|
|
|
UBool tInShifted = FALSE;
|
|
|
|
// This version of code can be refactored. However, it seems easier to understand this way.
|
|
|
|
// Source loop. Sam as the target loop.
|
|
|
|
for(;;) {
|
|
|
|
sOrder = ucol_IGetNextCE(coll, sColl, status);
|
|
|
|
if(sOrder == UCOL_NO_MORE_CES) {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
2007-12-26 21:37:40 +00:00
|
|
|
} else if(sOrder == 0 || (sInShifted && (sOrder & UCOL_PRIMARYMASK) == 0)) {
|
|
|
|
/* UCA amendment - ignore ignorables that follow shifted code points */
|
|
|
|
continue;
|
2007-12-26 19:27:43 +00:00
|
|
|
} else if(isContinuation(sOrder)) {
|
|
|
|
if((sOrder & UCOL_PRIMARYMASK) > 0) { /* There is primary value */
|
|
|
|
if(sInShifted) {
|
|
|
|
sOrder = (sOrder & UCOL_PRIMARYMASK) | 0xC0; /* preserve interesting continuation */
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
} else {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else { /* Just lower level values */
|
|
|
|
if(sInShifted) {
|
|
|
|
continue;
|
|
|
|
} else {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { /* regular */
|
|
|
|
if((sOrder & UCOL_PRIMARYMASK) > LVT) {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
if((sOrder & UCOL_PRIMARYMASK) > 0) {
|
|
|
|
sInShifted = TRUE;
|
|
|
|
sOrder &= UCOL_PRIMARYMASK;
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
} else {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&sCEs, sOrder, sColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
sInShifted = FALSE;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sOrder &= UCOL_PRIMARYMASK;
|
|
|
|
sInShifted = FALSE;
|
2001-01-16 00:28:40 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
for(;;) {
|
|
|
|
tOrder = ucol_IGetNextCE(coll, tColl, status);
|
|
|
|
if(tOrder == UCOL_NO_MORE_CES) {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
2007-12-26 21:37:40 +00:00
|
|
|
} else if(tOrder == 0 || (tInShifted && (tOrder & UCOL_PRIMARYMASK) == 0)) {
|
|
|
|
/* UCA amendment - ignore ignorables that follow shifted code points */
|
|
|
|
continue;
|
2007-12-26 19:27:43 +00:00
|
|
|
} else if(isContinuation(tOrder)) {
|
|
|
|
if((tOrder & UCOL_PRIMARYMASK) > 0) { /* There is primary value */
|
|
|
|
if(tInShifted) {
|
|
|
|
tOrder = (tOrder & UCOL_PRIMARYMASK) | 0xC0; /* preserve interesting continuation */
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
} else {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else { /* Just lower level values */
|
|
|
|
if(tInShifted) {
|
|
|
|
continue;
|
|
|
|
} else {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { /* regular */
|
|
|
|
if((tOrder & UCOL_PRIMARYMASK) > LVT) {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
if((tOrder & UCOL_PRIMARYMASK) > 0) {
|
|
|
|
tInShifted = TRUE;
|
|
|
|
tOrder &= UCOL_PRIMARYMASK;
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
} else {
|
2008-02-10 20:17:14 +00:00
|
|
|
UCOL_CEBUF_PUT(&tCEs, tOrder, tColl, status);
|
2007-12-26 19:27:43 +00:00
|
|
|
tInShifted = FALSE;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2001-10-05 02:07:51 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
tOrder &= UCOL_PRIMARYMASK;
|
|
|
|
tInShifted = FALSE;
|
|
|
|
|
|
|
|
if(sOrder == tOrder) {
|
|
|
|
/*
|
|
|
|
if(doHiragana && hirResult == UCOL_EQUAL) {
|
|
|
|
if((sColl.flags & UCOL_WAS_HIRAGANA) != (tColl.flags & UCOL_WAS_HIRAGANA)) {
|
|
|
|
hirResult = ((sColl.flags & UCOL_WAS_HIRAGANA) > (tColl.flags & UCOL_WAS_HIRAGANA))
|
|
|
|
? UCOL_LESS:UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
if(sOrder == UCOL_NO_MORE_CES_PRIMARY) {
|
|
|
|
break;
|
|
|
|
} else {
|
2007-12-26 21:44:39 +00:00
|
|
|
sOrder = 0;
|
|
|
|
tOrder = 0;
|
2007-12-26 19:27:43 +00:00
|
|
|
continue;
|
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
result = (sOrder < tOrder) ? UCOL_LESS : UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
2000-11-29 00:16:15 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} /* no primary difference... do the rest from the buffers */
|
2001-01-25 06:51:18 +00:00
|
|
|
}
|
2000-11-29 00:16:15 +00:00
|
|
|
|
2001-01-25 06:51:18 +00:00
|
|
|
/* now, we're gonna reexamine collected CEs */
|
2001-04-06 23:37:48 +00:00
|
|
|
uint32_t *sCE;
|
|
|
|
uint32_t *tCE;
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
/* This is the secondary level of comparison */
|
2001-01-25 06:51:18 +00:00
|
|
|
if(checkSecTer) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(!isFrenchSec) { /* normal */
|
|
|
|
sCE = sCEs.buf;
|
|
|
|
tCE = tCEs.buf;
|
|
|
|
for(;;) {
|
|
|
|
while (secS == 0) {
|
|
|
|
secS = *(sCE++) & UCOL_SECONDARYMASK;
|
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(secT == 0) {
|
|
|
|
secT = *(tCE++) & UCOL_SECONDARYMASK;
|
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(secS == secT) {
|
|
|
|
if(secS == UCOL_NO_MORE_CES_SECONDARY) {
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
secS = 0; secT = 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
result = (secS < secT) ? UCOL_LESS : UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else { /* do the French */
|
|
|
|
uint32_t *sCESave = NULL;
|
|
|
|
uint32_t *tCESave = NULL;
|
|
|
|
sCE = sCEs.pos-2; /* this could also be sCEs-- if needs to be optimized */
|
|
|
|
tCE = tCEs.pos-2;
|
|
|
|
for(;;) {
|
|
|
|
while (secS == 0 && sCE >= sCEs.buf) {
|
|
|
|
if(sCESave == 0) {
|
|
|
|
secS = *(sCE--);
|
|
|
|
if(isContinuation(secS)) {
|
2007-12-26 21:37:40 +00:00
|
|
|
while(isContinuation(secS = *(sCE--)))
|
|
|
|
;
|
2007-12-26 19:27:43 +00:00
|
|
|
/* after this, secS has the start of continuation, and sCEs points before that */
|
|
|
|
sCESave = sCE; /* we save it, so that we know where to come back AND that we need to go forward */
|
|
|
|
sCE+=2; /* need to point to the first continuation CP */
|
|
|
|
/* However, now you can just continue doing stuff */
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
secS = *(sCE++);
|
|
|
|
if(!isContinuation(secS)) { /* This means we have finished with this cont */
|
|
|
|
sCE = sCESave; /* reset the pointer to before continuation */
|
|
|
|
sCESave = 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
secS &= UCOL_SECONDARYMASK; /* remove the continuation bit */
|
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(secT == 0 && tCE >= tCEs.buf) {
|
|
|
|
if(tCESave == 0) {
|
|
|
|
secT = *(tCE--);
|
|
|
|
if(isContinuation(secT)) {
|
2007-12-26 21:44:39 +00:00
|
|
|
while(isContinuation(secT = *(tCE--)))
|
|
|
|
;
|
2007-12-26 19:27:43 +00:00
|
|
|
/* after this, secS has the start of continuation, and sCEs points before that */
|
|
|
|
tCESave = tCE; /* we save it, so that we know where to come back AND that we need to go forward */
|
|
|
|
tCE+=2; /* need to point to the first continuation CP */
|
|
|
|
/* However, now you can just continue doing stuff */
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
secT = *(tCE++);
|
|
|
|
if(!isContinuation(secT)) { /* This means we have finished with this cont */
|
|
|
|
tCE = tCESave; /* reset the pointer to before continuation */
|
|
|
|
tCESave = 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
secT &= UCOL_SECONDARYMASK; /* remove the continuation bit */
|
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(secS == secT) {
|
|
|
|
if(secS == UCOL_NO_MORE_CES_SECONDARY || (sCE < sCEs.buf && tCE < tCEs.buf)) {
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
secS = 0; secT = 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
result = (secS < secT) ? UCOL_LESS : UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
}
|
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
}
|
|
|
|
|
2001-01-26 00:12:23 +00:00
|
|
|
/* doing the case bit */
|
2001-01-25 06:51:18 +00:00
|
|
|
if(checkCase) {
|
2007-12-26 19:27:43 +00:00
|
|
|
sCE = sCEs.buf;
|
|
|
|
tCE = tCEs.buf;
|
|
|
|
for(;;) {
|
|
|
|
while((secS & UCOL_REMOVE_CASE) == 0) {
|
|
|
|
if(!isContinuation(*sCE++)) {
|
|
|
|
secS =*(sCE-1);
|
|
|
|
if(((secS & UCOL_PRIMARYMASK) != 0) || strength > UCOL_PRIMARY) {
|
|
|
|
// primary ignorables should not be considered on the case level when the strength is primary
|
|
|
|
// otherwise, the CEs stop being well-formed
|
|
|
|
secS &= UCOL_TERT_CASE_MASK;
|
|
|
|
secS ^= caseSwitch;
|
|
|
|
} else {
|
|
|
|
secS = 0;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
secS = 0;
|
|
|
|
}
|
2006-01-28 08:26:55 +00:00
|
|
|
}
|
2001-01-26 00:12:23 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while((secT & UCOL_REMOVE_CASE) == 0) {
|
|
|
|
if(!isContinuation(*tCE++)) {
|
|
|
|
secT = *(tCE-1);
|
|
|
|
if(((secT & UCOL_PRIMARYMASK) != 0) || strength > UCOL_PRIMARY) {
|
|
|
|
// primary ignorables should not be considered on the case level when the strength is primary
|
|
|
|
// otherwise, the CEs stop being well-formed
|
|
|
|
secT &= UCOL_TERT_CASE_MASK;
|
|
|
|
secT ^= caseSwitch;
|
|
|
|
} else {
|
|
|
|
secT = 0;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
secT = 0;
|
|
|
|
}
|
2006-01-28 08:26:55 +00:00
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if((secS & UCOL_CASE_BIT_MASK) < (secT & UCOL_CASE_BIT_MASK)) {
|
|
|
|
result = UCOL_LESS;
|
|
|
|
goto commonReturn;
|
|
|
|
} else if((secS & UCOL_CASE_BIT_MASK) > (secT & UCOL_CASE_BIT_MASK)) {
|
|
|
|
result = UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
|
|
|
}
|
2001-03-02 00:19:43 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if((secS & UCOL_REMOVE_CASE) == UCOL_NO_MORE_CES_TERTIARY || (secT & UCOL_REMOVE_CASE) == UCOL_NO_MORE_CES_TERTIARY ) {
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
secS = 0;
|
|
|
|
secT = 0;
|
|
|
|
}
|
2001-03-02 00:19:43 +00:00
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
}
|
|
|
|
|
2001-02-28 19:01:23 +00:00
|
|
|
/* Tertiary level */
|
2001-01-25 06:51:18 +00:00
|
|
|
if(checkTertiary) {
|
2007-12-26 19:27:43 +00:00
|
|
|
secS = 0;
|
|
|
|
secT = 0;
|
|
|
|
sCE = sCEs.buf;
|
|
|
|
tCE = tCEs.buf;
|
|
|
|
for(;;) {
|
|
|
|
while((secS & UCOL_REMOVE_CASE) == 0) {
|
|
|
|
secS = *(sCE++) & tertiaryMask;
|
|
|
|
if(!isContinuation(secS)) {
|
|
|
|
secS ^= caseSwitch;
|
|
|
|
} else {
|
|
|
|
secS &= UCOL_REMOVE_CASE;
|
|
|
|
}
|
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while((secT & UCOL_REMOVE_CASE) == 0) {
|
|
|
|
secT = *(tCE++) & tertiaryMask;
|
|
|
|
if(!isContinuation(secT)) {
|
|
|
|
secT ^= caseSwitch;
|
|
|
|
} else {
|
|
|
|
secT &= UCOL_REMOVE_CASE;
|
|
|
|
}
|
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(secS == secT) {
|
|
|
|
if((secS & UCOL_REMOVE_CASE) == 1) {
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
secS = 0; secT = 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
result = (secS < secT) ? UCOL_LESS : UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
2001-01-25 06:51:18 +00:00
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2001-10-05 02:07:51 +00:00
|
|
|
if(qShifted /*checkQuad*/) {
|
2007-12-26 19:27:43 +00:00
|
|
|
UBool sInShifted = TRUE;
|
|
|
|
UBool tInShifted = TRUE;
|
|
|
|
secS = 0;
|
|
|
|
secT = 0;
|
|
|
|
sCE = sCEs.buf;
|
|
|
|
tCE = tCEs.buf;
|
|
|
|
for(;;) {
|
|
|
|
while(secS == 0 && secS != UCOL_NO_MORE_CES || (isContinuation(secS) && !sInShifted)) {
|
|
|
|
secS = *(sCE++);
|
|
|
|
if(isContinuation(secS)) {
|
|
|
|
if(!sInShifted) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else if(secS > LVT || (secS & UCOL_PRIMARYMASK) == 0) { /* non continuation */
|
|
|
|
secS = UCOL_PRIMARYMASK;
|
|
|
|
sInShifted = FALSE;
|
|
|
|
} else {
|
|
|
|
sInShifted = TRUE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
secS &= UCOL_PRIMARYMASK;
|
2000-11-29 00:16:15 +00:00
|
|
|
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(secT == 0 && secT != UCOL_NO_MORE_CES || (isContinuation(secT) && !tInShifted)) {
|
|
|
|
secT = *(tCE++);
|
|
|
|
if(isContinuation(secT)) {
|
|
|
|
if(!tInShifted) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else if(secT > LVT || (secT & UCOL_PRIMARYMASK) == 0) {
|
|
|
|
secT = UCOL_PRIMARYMASK;
|
|
|
|
tInShifted = FALSE;
|
|
|
|
} else {
|
|
|
|
tInShifted = TRUE;
|
|
|
|
}
|
2001-06-02 01:01:18 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
secT &= UCOL_PRIMARYMASK;
|
2001-01-25 06:51:18 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(secS == secT) {
|
|
|
|
if(secS == UCOL_NO_MORE_CES_PRIMARY) {
|
|
|
|
break;
|
|
|
|
} else {
|
|
|
|
secS = 0; secT = 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
result = (secS < secT) ? UCOL_LESS : UCOL_GREATER;
|
|
|
|
goto commonReturn;
|
|
|
|
}
|
2001-04-23 01:53:49 +00:00
|
|
|
}
|
2001-10-05 02:07:51 +00:00
|
|
|
} else if(doHiragana && hirResult != UCOL_EQUAL) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// If we're fine on quaternaries, we might be different
|
|
|
|
// on Hiragana. This, however, might fail us in shifted.
|
|
|
|
result = hirResult;
|
|
|
|
goto commonReturn;
|
2001-01-25 06:51:18 +00:00
|
|
|
}
|
2000-11-20 19:17:17 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* For IDENTICAL comparisons, we use a bitwise character comparison */
|
2001-04-06 23:37:48 +00:00
|
|
|
/* as a tiebreaker if all else is equal. */
|
|
|
|
/* Getting here should be quite rare - strings are not identical - */
|
|
|
|
/* that is checked first, but compared == through all other checks. */
|
2001-02-28 19:01:23 +00:00
|
|
|
if(checkIdent)
|
2000-11-20 19:17:17 +00:00
|
|
|
{
|
2001-11-01 00:00:15 +00:00
|
|
|
//result = ucol_checkIdent(&sColl, &tColl, coll->normalizationMode == UCOL_ON);
|
2003-01-20 07:43:32 +00:00
|
|
|
result = ucol_checkIdent(sColl, tColl, TRUE, status);
|
2001-04-06 23:37:48 +00:00
|
|
|
}
|
2001-03-14 02:45:39 +00:00
|
|
|
|
2001-04-06 23:37:48 +00:00
|
|
|
commonReturn:
|
2003-01-20 07:43:32 +00:00
|
|
|
if ((sColl->flags | tColl->flags) & UCOL_ITER_ALLOCATED) {
|
|
|
|
freeHeapWritableBuffer(sColl);
|
|
|
|
freeHeapWritableBuffer(tColl);
|
2000-11-20 19:17:17 +00:00
|
|
|
|
2001-04-23 01:53:49 +00:00
|
|
|
if (sCEs.buf != sCEs.localArray ) {
|
|
|
|
uprv_free(sCEs.buf);
|
|
|
|
}
|
|
|
|
if (tCEs.buf != tCEs.localArray ) {
|
|
|
|
uprv_free(tCEs.buf);
|
2001-04-18 19:31:05 +00:00
|
|
|
}
|
2001-04-12 19:59:28 +00:00
|
|
|
}
|
2000-11-20 19:17:17 +00:00
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
static inline uint32_t
|
|
|
|
ucol_getLatinOneContraction(const UCollator *coll, int32_t strength,
|
|
|
|
uint32_t CE, const UChar *s, int32_t *index, int32_t len)
|
|
|
|
{
|
|
|
|
const UChar *UCharOffset = (UChar *)coll->image+getContractOffset(CE&0xFFF);
|
|
|
|
int32_t latinOneOffset = (CE & 0x00FFF000) >> 12;
|
|
|
|
int32_t offset = 1;
|
|
|
|
UChar schar = 0, tchar = 0;
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
for(;;) {
|
|
|
|
if(len == -1) {
|
|
|
|
if(s[*index] == 0) { // end of string
|
|
|
|
return(coll->latinOneCEs[strength*coll->latinOneTableLen+latinOneOffset]);
|
|
|
|
} else {
|
|
|
|
schar = s[*index];
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if(*index == len) {
|
|
|
|
return(coll->latinOneCEs[strength*coll->latinOneTableLen+latinOneOffset]);
|
|
|
|
} else {
|
|
|
|
schar = s[*index];
|
|
|
|
}
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(schar > (tchar = *(UCharOffset+offset))) { /* since the contraction codepoints should be ordered, we skip all that are smaller */
|
|
|
|
offset++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (schar == tchar) {
|
|
|
|
(*index)++;
|
|
|
|
return(coll->latinOneCEs[strength*coll->latinOneTableLen+latinOneOffset+offset]);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
if(schar & 0xFF00 /*> UCOL_ENDOFLATIN1RANGE*/) {
|
|
|
|
return UCOL_BAIL_OUT_CE;
|
|
|
|
}
|
|
|
|
// skip completely ignorables
|
|
|
|
uint32_t isZeroCE = UTRIE_GET32_FROM_LEAD(&coll->mapping, schar);
|
|
|
|
if(isZeroCE == 0) { // we have to ignore completely ignorables
|
|
|
|
(*index)++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
return(coll->latinOneCEs[strength*coll->latinOneTableLen+latinOneOffset]);
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2003-12-03 00:45:12 +00:00
|
|
|
/**
|
|
|
|
* This is a fast strcoll, geared towards text in Latin-1.
|
2002-09-04 06:02:13 +00:00
|
|
|
* It supports contractions of size two, French secondaries
|
|
|
|
* and case switching. You can use it with strengths primary
|
|
|
|
* to tertiary. It does not support shifted and case level.
|
|
|
|
* It relies on the table build by setupLatin1Table. If it
|
|
|
|
* doesn't understand something, it will go to the regular
|
2003-12-03 00:45:12 +00:00
|
|
|
* strcoll.
|
2002-09-04 06:02:13 +00:00
|
|
|
*/
|
2003-12-03 00:45:12 +00:00
|
|
|
static inline UCollationResult
|
2002-09-04 06:02:13 +00:00
|
|
|
ucol_strcollUseLatin1( const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sLen,
|
|
|
|
const UChar *target,
|
|
|
|
int32_t tLen,
|
2003-12-03 00:45:12 +00:00
|
|
|
UErrorCode *status)
|
2002-09-04 06:02:13 +00:00
|
|
|
{
|
|
|
|
U_ALIGN_CODE(16);
|
|
|
|
int32_t strength = coll->strength;
|
|
|
|
|
|
|
|
int32_t sIndex = 0, tIndex = 0;
|
|
|
|
UChar sChar = 0, tChar = 0;
|
|
|
|
uint32_t sOrder=0, tOrder=0;
|
|
|
|
|
2006-02-09 08:59:20 +00:00
|
|
|
UBool endOfSource = FALSE;
|
2002-09-04 06:02:13 +00:00
|
|
|
|
|
|
|
uint32_t *elements = coll->latinOneCEs;
|
|
|
|
|
|
|
|
UBool haveContractions = FALSE; // if we have contractions in our string
|
|
|
|
// we cannot do French secondary
|
|
|
|
|
|
|
|
// Do the primary level
|
|
|
|
for(;;) {
|
2007-12-26 19:27:43 +00:00
|
|
|
while(sOrder==0) { // this loop skips primary ignorables
|
|
|
|
// sOrder=getNextlatinOneCE(source);
|
|
|
|
if(sLen==-1) { // handling zero terminated strings
|
|
|
|
sChar=source[sIndex++];
|
|
|
|
if(sChar==0) {
|
|
|
|
endOfSource = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else { // handling strings with known length
|
|
|
|
if(sIndex==sLen) {
|
|
|
|
endOfSource = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
sChar=source[sIndex++];
|
|
|
|
}
|
|
|
|
if(sChar&0xFF00) { // if we encounter non-latin-1, we bail out (sChar > 0xFF, but this is faster on win32)
|
|
|
|
//fprintf(stderr, "R");
|
|
|
|
goto returnRegular;
|
|
|
|
//return ucol_strcollRegular(coll, source, sLen, target, tLen, status);
|
|
|
|
}
|
|
|
|
sOrder = elements[sChar];
|
|
|
|
if(sOrder >= UCOL_NOT_FOUND) { // if we got a special
|
|
|
|
// specials can basically be either contractions or bail-out signs. If we get anything
|
|
|
|
// else, we'll bail out anywasy
|
|
|
|
if(getCETag(sOrder) == CONTRACTION_TAG) {
|
|
|
|
sOrder = ucol_getLatinOneContraction(coll, UCOL_PRIMARY, sOrder, source, &sIndex, sLen);
|
|
|
|
haveContractions = TRUE; // if there are contractions, we cannot do French secondary
|
|
|
|
// However, if there are contractions in the table, but we always use just one char,
|
|
|
|
// we might be able to do French. This should be checked out.
|
|
|
|
}
|
|
|
|
if(sOrder >= UCOL_NOT_FOUND /*== UCOL_BAIL_OUT_CE*/) {
|
|
|
|
//fprintf(stderr, "S");
|
|
|
|
goto returnRegular;
|
|
|
|
//return ucol_strcollRegular(coll, source, sLen, target, tLen, status);
|
|
|
|
}
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(tOrder==0) { // this loop skips primary ignorables
|
|
|
|
// tOrder=getNextlatinOneCE(target);
|
|
|
|
if(tLen==-1) { // handling zero terminated strings
|
|
|
|
tChar=target[tIndex++];
|
|
|
|
if(tChar==0) {
|
|
|
|
if(endOfSource) { // this is different than source loop,
|
|
|
|
// as we already know that source loop is done here,
|
|
|
|
// so we can either finish the primary loop if both
|
|
|
|
// strings are done or anounce the result if only
|
|
|
|
// target is done. Same below.
|
|
|
|
goto endOfPrimLoop;
|
|
|
|
} else {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else { // handling strings with known length
|
|
|
|
if(tIndex==tLen) {
|
|
|
|
if(endOfSource) {
|
|
|
|
goto endOfPrimLoop;
|
|
|
|
} else {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
tChar=target[tIndex++];
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if(tChar&0xFF00) { // if we encounter non-latin-1, we bail out (sChar > 0xFF, but this is faster on win32)
|
|
|
|
//fprintf(stderr, "R");
|
|
|
|
goto returnRegular;
|
|
|
|
//return ucol_strcollRegular(coll, source, sLen, target, tLen, status);
|
|
|
|
}
|
|
|
|
tOrder = elements[tChar];
|
|
|
|
if(tOrder >= UCOL_NOT_FOUND) {
|
|
|
|
// Handling specials, see the comments for source
|
|
|
|
if(getCETag(tOrder) == CONTRACTION_TAG) {
|
|
|
|
tOrder = ucol_getLatinOneContraction(coll, UCOL_PRIMARY, tOrder, target, &tIndex, tLen);
|
|
|
|
haveContractions = TRUE;
|
|
|
|
}
|
|
|
|
if(tOrder >= UCOL_NOT_FOUND /*== UCOL_BAIL_OUT_CE*/) {
|
|
|
|
//fprintf(stderr, "S");
|
|
|
|
goto returnRegular;
|
|
|
|
//return ucol_strcollRegular(coll, source, sLen, target, tLen, status);
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if(endOfSource) { // source is finished, but target is not, say the result.
|
2002-09-04 06:02:13 +00:00
|
|
|
return UCOL_LESS;
|
|
|
|
}
|
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(sOrder == tOrder) { // if we have same CEs, we continue the loop
|
|
|
|
sOrder = 0; tOrder = 0;
|
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
// compare current top bytes
|
|
|
|
if(((sOrder^tOrder)&0xFF000000)!=0) {
|
|
|
|
// top bytes differ, return difference
|
|
|
|
if(sOrder < tOrder) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
} else if(sOrder > tOrder) {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
// instead of return (int32_t)(sOrder>>24)-(int32_t)(tOrder>>24);
|
|
|
|
// since we must return enum value
|
|
|
|
}
|
|
|
|
|
|
|
|
// top bytes match, continue with following bytes
|
|
|
|
sOrder<<=8;
|
|
|
|
tOrder<<=8;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
endOfPrimLoop:
|
2003-12-03 00:45:12 +00:00
|
|
|
// after primary loop, we definitely know the sizes of strings,
|
2002-09-04 06:02:13 +00:00
|
|
|
// so we set it and use simpler loop for secondaries and tertiaries
|
|
|
|
sLen = sIndex; tLen = tIndex;
|
|
|
|
if(strength >= UCOL_SECONDARY) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// adjust the table beggining
|
|
|
|
elements += coll->latinOneTableLen;
|
|
|
|
endOfSource = FALSE;
|
|
|
|
|
|
|
|
if(coll->frenchCollation == UCOL_OFF) { // non French
|
|
|
|
// This loop is a simplified copy of primary loop
|
|
|
|
// at this point we know that whole strings are latin-1, so we don't
|
|
|
|
// check for that. We also know that we only have contractions as
|
|
|
|
// specials.
|
|
|
|
sIndex = 0; tIndex = 0;
|
|
|
|
for(;;) {
|
|
|
|
while(sOrder==0) {
|
|
|
|
if(sIndex==sLen) {
|
|
|
|
endOfSource = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
sChar=source[sIndex++];
|
|
|
|
sOrder = elements[sChar];
|
|
|
|
if(sOrder > UCOL_NOT_FOUND) {
|
|
|
|
sOrder = ucol_getLatinOneContraction(coll, UCOL_SECONDARY, sOrder, source, &sIndex, sLen);
|
|
|
|
}
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(tOrder==0) {
|
|
|
|
if(tIndex==tLen) {
|
|
|
|
if(endOfSource) {
|
|
|
|
goto endOfSecLoop;
|
|
|
|
} else {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
tChar=target[tIndex++];
|
|
|
|
tOrder = elements[tChar];
|
|
|
|
if(tOrder > UCOL_NOT_FOUND) {
|
|
|
|
tOrder = ucol_getLatinOneContraction(coll, UCOL_SECONDARY, tOrder, target, &tIndex, tLen);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if(endOfSource) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(sOrder == tOrder) {
|
|
|
|
sOrder = 0; tOrder = 0;
|
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
// see primary loop for comments on this
|
|
|
|
if(((sOrder^tOrder)&0xFF000000)!=0) {
|
|
|
|
if(sOrder < tOrder) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
} else if(sOrder > tOrder) {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sOrder<<=8;
|
|
|
|
tOrder<<=8;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
} else { // French
|
|
|
|
if(haveContractions) { // if we have contractions, we have to bail out
|
|
|
|
// since we don't really know how to handle them here
|
|
|
|
goto returnRegular;
|
|
|
|
//return ucol_strcollRegular(coll, source, sLen, target, tLen, status);
|
|
|
|
}
|
|
|
|
// For French, we go backwards
|
|
|
|
sIndex = sLen; tIndex = tLen;
|
|
|
|
for(;;) {
|
|
|
|
while(sOrder==0) {
|
|
|
|
if(sIndex==0) {
|
|
|
|
endOfSource = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
sChar=source[--sIndex];
|
|
|
|
sOrder = elements[sChar];
|
|
|
|
// don't even look for contractions
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while(tOrder==0) {
|
|
|
|
if(tIndex==0) {
|
|
|
|
if(endOfSource) {
|
|
|
|
goto endOfSecLoop;
|
|
|
|
} else {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
tChar=target[--tIndex];
|
|
|
|
tOrder = elements[tChar];
|
|
|
|
// don't even look for contractions
|
|
|
|
}
|
|
|
|
if(endOfSource) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(sOrder == tOrder) {
|
|
|
|
sOrder = 0; tOrder = 0;
|
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
// see the primary loop for comments
|
|
|
|
if(((sOrder^tOrder)&0xFF000000)!=0) {
|
|
|
|
if(sOrder < tOrder) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
} else if(sOrder > tOrder) {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sOrder<<=8;
|
|
|
|
tOrder<<=8;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
|
|
|
endOfSecLoop:
|
|
|
|
if(strength >= UCOL_TERTIARY) {
|
2007-12-26 19:27:43 +00:00
|
|
|
// tertiary loop is the same as secondary (except no French)
|
|
|
|
elements += coll->latinOneTableLen;
|
|
|
|
sIndex = 0; tIndex = 0;
|
|
|
|
endOfSource = FALSE;
|
|
|
|
for(;;) {
|
|
|
|
while(sOrder==0) {
|
|
|
|
if(sIndex==sLen) {
|
|
|
|
endOfSource = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
sChar=source[sIndex++];
|
|
|
|
sOrder = elements[sChar];
|
|
|
|
if(sOrder > UCOL_NOT_FOUND) {
|
|
|
|
sOrder = ucol_getLatinOneContraction(coll, UCOL_TERTIARY, sOrder, source, &sIndex, sLen);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
while(tOrder==0) {
|
|
|
|
if(tIndex==tLen) {
|
|
|
|
if(endOfSource) {
|
|
|
|
return UCOL_EQUAL; // if both strings are at the end, they are equal
|
|
|
|
} else {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
tChar=target[tIndex++];
|
|
|
|
tOrder = elements[tChar];
|
|
|
|
if(tOrder > UCOL_NOT_FOUND) {
|
|
|
|
tOrder = ucol_getLatinOneContraction(coll, UCOL_TERTIARY, tOrder, target, &tIndex, tLen);
|
|
|
|
}
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
if(endOfSource) {
|
2007-12-26 19:27:43 +00:00
|
|
|
return UCOL_LESS;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if(sOrder == tOrder) {
|
|
|
|
sOrder = 0; tOrder = 0;
|
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
if(((sOrder^tOrder)&0xff000000)!=0) {
|
|
|
|
if(sOrder < tOrder) {
|
|
|
|
return UCOL_LESS;
|
|
|
|
} else if(sOrder > tOrder) {
|
|
|
|
return UCOL_GREATER;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sOrder<<=8;
|
|
|
|
tOrder<<=8;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2003-12-03 00:45:12 +00:00
|
|
|
}
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
return UCOL_EQUAL;
|
2003-01-20 07:43:32 +00:00
|
|
|
|
|
|
|
returnRegular:
|
|
|
|
// Preparing the context objects for iterating over strings
|
|
|
|
collIterate sColl, tColl;
|
|
|
|
|
|
|
|
IInit_collIterate(coll, source, sLen, &sColl);
|
|
|
|
IInit_collIterate(coll, target, tLen, &tColl);
|
2003-12-03 00:45:12 +00:00
|
|
|
return ucol_strcollRegular(&sColl, &tColl, status);
|
|
|
|
}
|
2003-01-20 07:43:32 +00:00
|
|
|
|
|
|
|
|
|
|
|
U_CAPI UCollationResult U_EXPORT2
|
|
|
|
ucol_strcollIter( const UCollator *coll,
|
|
|
|
UCharIterator *sIter,
|
|
|
|
UCharIterator *tIter,
|
2007-12-26 19:27:43 +00:00
|
|
|
UErrorCode *status)
|
|
|
|
{
|
|
|
|
if(!status || U_FAILURE(*status)) {
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
2003-09-24 00:58:35 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
UTRACE_ENTRY(UTRACE_UCOL_STRCOLLITER);
|
|
|
|
UTRACE_DATA3(UTRACE_VERBOSE, "coll=%p, sIter=%p, tIter=%p", coll, sIter, tIter);
|
2003-09-24 00:58:35 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (sIter == tIter) {
|
|
|
|
UTRACE_EXIT_VALUE_STATUS(UCOL_EQUAL, *status)
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
|
|
|
if(sIter == NULL || tIter == NULL || coll == NULL) {
|
|
|
|
*status = U_ILLEGAL_ARGUMENT_ERROR;
|
|
|
|
UTRACE_EXIT_VALUE_STATUS(UCOL_EQUAL, *status)
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
UCollationResult result = UCOL_EQUAL;
|
|
|
|
|
|
|
|
// Preparing the context objects for iterating over strings
|
|
|
|
collIterate sColl, tColl;
|
|
|
|
// The division for the array length may truncate the array size to
|
|
|
|
// a little less than UNORM_ITER_SIZE, but that size is dimensioned too high
|
|
|
|
// for all platforms anyway.
|
|
|
|
UAlignedMemory stackNormIter1[UNORM_ITER_SIZE/sizeof(UAlignedMemory)];
|
|
|
|
UAlignedMemory stackNormIter2[UNORM_ITER_SIZE/sizeof(UAlignedMemory)];
|
|
|
|
UNormIterator *sNormIter = NULL, *tNormIter = NULL;
|
2003-01-23 01:52:34 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
IInit_collIterate(coll, NULL, -1, &sColl);
|
|
|
|
sColl.iterator = sIter;
|
|
|
|
sColl.flags |= UCOL_USE_ITERATOR;
|
|
|
|
IInit_collIterate(coll, NULL, -1, &tColl);
|
|
|
|
tColl.flags |= UCOL_USE_ITERATOR;
|
|
|
|
tColl.iterator = tIter;
|
2003-12-03 00:45:12 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(ucol_getAttribute(coll, UCOL_NORMALIZATION_MODE, status) == UCOL_ON) {
|
|
|
|
sNormIter = unorm_openIter(stackNormIter1, sizeof(stackNormIter1), status);
|
|
|
|
sColl.iterator = unorm_setIter(sNormIter, sIter, UNORM_FCD, status);
|
|
|
|
sColl.flags &= ~UCOL_ITER_NORM;
|
|
|
|
|
|
|
|
tNormIter = unorm_openIter(stackNormIter2, sizeof(stackNormIter2), status);
|
|
|
|
tColl.iterator = unorm_setIter(tNormIter, tIter, UNORM_FCD, status);
|
|
|
|
tColl.flags &= ~UCOL_ITER_NORM;
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
UChar32 sChar = U_SENTINEL, tChar = U_SENTINEL;
|
2003-01-23 01:52:34 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
while((sChar = sColl.iterator->next(sColl.iterator)) ==
|
|
|
|
(tChar = tColl.iterator->next(tColl.iterator))) {
|
|
|
|
if(sChar == U_SENTINEL) {
|
|
|
|
result = UCOL_EQUAL;
|
|
|
|
goto end_compare;
|
|
|
|
}
|
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(sChar == U_SENTINEL) {
|
|
|
|
tChar = tColl.iterator->previous(tColl.iterator);
|
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if(tChar == U_SENTINEL) {
|
2003-02-20 01:06:06 +00:00
|
|
|
sChar = sColl.iterator->previous(sColl.iterator);
|
2007-12-26 19:27:43 +00:00
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
sChar = sColl.iterator->previous(sColl.iterator);
|
|
|
|
tChar = tColl.iterator->previous(tColl.iterator);
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
if (ucol_unsafeCP((UChar)sChar, coll) || ucol_unsafeCP((UChar)tChar, coll))
|
|
|
|
{
|
|
|
|
// We are stopped in the middle of a contraction.
|
|
|
|
// Scan backwards through the == part of the string looking for the start of the contraction.
|
|
|
|
// It doesn't matter which string we scan, since they are the same in this region.
|
|
|
|
do
|
|
|
|
{
|
|
|
|
sChar = sColl.iterator->previous(sColl.iterator);
|
|
|
|
tChar = tColl.iterator->previous(tColl.iterator);
|
|
|
|
}
|
|
|
|
while (sChar != U_SENTINEL && ucol_unsafeCP((UChar)sChar, coll));
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
if(U_SUCCESS(*status)) {
|
|
|
|
result = ucol_strcollRegular(&sColl, &tColl, status);
|
|
|
|
}
|
2003-02-20 01:06:06 +00:00
|
|
|
|
2003-02-25 21:32:33 +00:00
|
|
|
end_compare:
|
2007-12-26 19:27:43 +00:00
|
|
|
if(sNormIter || tNormIter) {
|
|
|
|
unorm_closeIter(sNormIter);
|
|
|
|
unorm_closeIter(tNormIter);
|
|
|
|
}
|
2003-01-20 07:43:32 +00:00
|
|
|
|
2007-12-26 19:27:43 +00:00
|
|
|
UTRACE_EXIT_VALUE_STATUS(result, *status)
|
|
|
|
return result;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/* */
|
|
|
|
/* ucol_strcoll Main public API string comparison function */
|
|
|
|
/* */
|
|
|
|
U_CAPI UCollationResult U_EXPORT2
|
|
|
|
ucol_strcoll( const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
const UChar *target,
|
2007-12-28 20:08:41 +00:00
|
|
|
int32_t targetLength)
|
|
|
|
{
|
2002-09-04 06:02:13 +00:00
|
|
|
U_ALIGN_CODE(16);
|
2003-09-24 00:58:35 +00:00
|
|
|
|
2003-09-20 01:21:23 +00:00
|
|
|
UTRACE_ENTRY(UTRACE_UCOL_STRCOLL);
|
2003-09-24 00:58:35 +00:00
|
|
|
if (UTRACE_LEVEL(UTRACE_VERBOSE)) {
|
2007-12-28 20:08:41 +00:00
|
|
|
UTRACE_DATA3(UTRACE_VERBOSE, "coll=%p, source=%p, target=%p", coll, source, target);
|
|
|
|
UTRACE_DATA2(UTRACE_VERBOSE, "source string = %vh ", source, sourceLength);
|
|
|
|
UTRACE_DATA2(UTRACE_VERBOSE, "target string = %vh ", target, targetLength);
|
2003-09-24 00:58:35 +00:00
|
|
|
}
|
|
|
|
|
2003-05-09 20:10:01 +00:00
|
|
|
if(source == NULL || target == NULL) {
|
2007-12-28 20:08:41 +00:00
|
|
|
// do not crash, but return. Should have
|
|
|
|
// status argument to return error.
|
|
|
|
UTRACE_EXIT_VALUE(UCOL_EQUAL);
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Quick check if source and target are same strings. */
|
|
|
|
/* They should either both be NULL terminated or the explicit length should be set on both. */
|
|
|
|
if (source==target && sourceLength==targetLength) {
|
|
|
|
UTRACE_EXIT_VALUE(UCOL_EQUAL);
|
|
|
|
return UCOL_EQUAL;
|
2003-05-09 20:10:01 +00:00
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
|
|
|
|
/* Scan the strings. Find: */
|
|
|
|
/* The length of any leading portion that is equal */
|
|
|
|
/* Whether they are exactly equal. (in which case we just return) */
|
|
|
|
const UChar *pSrc = source;
|
|
|
|
const UChar *pTarg = target;
|
|
|
|
int32_t equalLength;
|
|
|
|
|
|
|
|
if (sourceLength == -1 && targetLength == -1) {
|
|
|
|
// Both strings are null terminated.
|
2007-12-28 20:08:41 +00:00
|
|
|
// Scan through any leading equal portion.
|
|
|
|
while (*pSrc == *pTarg && *pSrc != 0) {
|
2002-09-04 06:02:13 +00:00
|
|
|
pSrc++;
|
|
|
|
pTarg++;
|
|
|
|
}
|
|
|
|
if (*pSrc == 0 && *pTarg == 0) {
|
2003-11-01 01:18:42 +00:00
|
|
|
UTRACE_EXIT_VALUE(UCOL_EQUAL);
|
2002-09-04 06:02:13 +00:00
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
|
|
|
equalLength = pSrc - source;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
// One or both strings has an explicit length.
|
|
|
|
const UChar *pSrcEnd = source + sourceLength;
|
|
|
|
const UChar *pTargEnd = target + targetLength;
|
|
|
|
|
|
|
|
// Scan while the strings are bitwise ==, or until one is exhausted.
|
2007-12-26 19:27:43 +00:00
|
|
|
for (;;) {
|
|
|
|
if (pSrc == pSrcEnd || pTarg == pTargEnd) {
|
|
|
|
break;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
if ((*pSrc == 0 && sourceLength == -1) || (*pTarg == 0 && targetLength == -1)) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (*pSrc != *pTarg) {
|
|
|
|
break;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
2007-12-26 19:27:43 +00:00
|
|
|
pSrc++;
|
|
|
|
pTarg++;
|
|
|
|
}
|
|
|
|
equalLength = pSrc - source;
|
|
|
|
|
|
|
|
// If we made it all the way through both strings, we are done. They are ==
|
|
|
|
if ((pSrc ==pSrcEnd || (pSrcEnd <pSrc && *pSrc==0)) && /* At end of src string, however it was specified. */
|
2007-12-28 20:08:41 +00:00
|
|
|
(pTarg==pTargEnd || (pTargEnd<pTarg && *pTarg==0))) /* and also at end of dest string */
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
UTRACE_EXIT_VALUE(UCOL_EQUAL);
|
|
|
|
return UCOL_EQUAL;
|
|
|
|
}
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
if (equalLength > 0) {
|
|
|
|
/* There is an identical portion at the beginning of the two strings. */
|
|
|
|
/* If the identical portion ends within a contraction or a comibining */
|
|
|
|
/* character sequence, back up to the start of that sequence. */
|
2007-12-28 20:08:41 +00:00
|
|
|
|
|
|
|
// These values should already be set by the code above.
|
|
|
|
//pSrc = source + equalLength; /* point to the first differing chars */
|
|
|
|
//pTarg = target + equalLength;
|
2002-09-04 06:02:13 +00:00
|
|
|
if (pSrc != source+sourceLength && ucol_unsafeCP(*pSrc, coll) ||
|
|
|
|
pTarg != target+targetLength && ucol_unsafeCP(*pTarg, coll))
|
|
|
|
{
|
|
|
|
// We are stopped in the middle of a contraction.
|
|
|
|
// Scan backwards through the == part of the string looking for the start of the contraction.
|
|
|
|
// It doesn't matter which string we scan, since they are the same in this region.
|
|
|
|
do
|
|
|
|
{
|
|
|
|
equalLength--;
|
|
|
|
pSrc--;
|
|
|
|
}
|
|
|
|
while (equalLength>0 && ucol_unsafeCP(*pSrc, coll));
|
|
|
|
}
|
|
|
|
|
|
|
|
source += equalLength;
|
|
|
|
target += equalLength;
|
|
|
|
if (sourceLength > 0) {
|
|
|
|
sourceLength -= equalLength;
|
|
|
|
}
|
|
|
|
if (targetLength > 0) {
|
|
|
|
targetLength -= equalLength;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-12-28 20:08:41 +00:00
|
|
|
UErrorCode status = U_ZERO_ERROR;
|
|
|
|
UCollationResult returnVal;
|
2003-01-23 01:52:34 +00:00
|
|
|
if(!coll->latinOneUse || (sourceLength > 0 && *source&0xff00) || (targetLength > 0 && *target&0xff00)) {
|
2007-12-28 20:08:41 +00:00
|
|
|
collIterate sColl, tColl;
|
2007-12-26 19:27:43 +00:00
|
|
|
// Preparing the context objects for iterating over strings
|
|
|
|
IInit_collIterate(coll, source, sourceLength, &sColl);
|
|
|
|
IInit_collIterate(coll, target, targetLength, &tColl);
|
|
|
|
returnVal = ucol_strcollRegular(&sColl, &tColl, &status);
|
2003-01-23 01:52:34 +00:00
|
|
|
} else {
|
2007-12-26 19:27:43 +00:00
|
|
|
returnVal = ucol_strcollUseLatin1(coll, source, sourceLength, target, targetLength, &status);
|
2003-01-23 01:52:34 +00:00
|
|
|
}
|
2003-11-01 01:18:42 +00:00
|
|
|
UTRACE_EXIT_VALUE(returnVal);
|
2003-09-20 01:21:23 +00:00
|
|
|
return returnVal;
|
2002-09-04 06:02:13 +00:00
|
|
|
}
|
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* convenience function for comparing strings */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI UBool U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_greater( const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
const UChar *target,
|
|
|
|
int32_t targetLength)
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
return (ucol_strcoll(coll, source, sourceLength, target, targetLength)
|
|
|
|
== UCOL_GREATER);
|
2001-01-16 00:28:40 +00:00
|
|
|
}
|
2001-01-04 00:45:41 +00:00
|
|
|
|
2001-01-16 00:28:40 +00:00
|
|
|
/* convenience function for comparing strings */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI UBool U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_greaterOrEqual( const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
const UChar *target,
|
|
|
|
int32_t targetLength)
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
return (ucol_strcoll(coll, source, sourceLength, target, targetLength)
|
|
|
|
!= UCOL_LESS);
|
2001-01-16 00:28:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* convenience function for comparing strings */
|
2001-11-21 01:08:55 +00:00
|
|
|
U_CAPI UBool U_EXPORT2
|
2001-01-16 00:28:40 +00:00
|
|
|
ucol_equal( const UCollator *coll,
|
|
|
|
const UChar *source,
|
|
|
|
int32_t sourceLength,
|
|
|
|
const UChar *target,
|
|
|
|
int32_t targetLength)
|
|
|
|
{
|
2007-12-26 19:27:43 +00:00
|
|
|
return (ucol_strcoll(coll, source, sourceLength, target, targetLength)
|
|
|
|
== UCOL_EQUAL);
|
2001-01-15 07:28:54 +00:00
|
|
|
}
|
2001-08-16 00:55:16 +00:00
|
|
|
|
2003-11-12 04:17:35 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
|
|
|
ucol_getUCAVersion(const UCollator* coll, UVersionInfo info) {
|
2007-12-26 19:27:43 +00:00
|
|
|
if(coll && coll->UCA) {
|
|
|
|
uprv_memcpy(info, coll->UCA->image->UCAVersion, sizeof(UVersionInfo));
|
|
|
|
}
|
2003-11-12 04:17:35 +00:00
|
|
|
}
|
|
|
|
|
2002-09-20 01:54:48 +00:00
|
|
|
#endif /* #if !UCONFIG_NO_COLLATION */
|