AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
7917df1e80
scuffed-code/icu4c/source/common/umutablecptrie.cpp
Markus Scherer 98f9170004 ICU-20250 make UnicodeSet(intprop=value) faster 
- fastpath for UnicodeSet.add(new last range)
- fewer UnicodeSet memory allocations:
  initial internal list array, exponential array growth,
  allocate strings list/set only when first one is added
- faster CodePointTrie.getRange(): fewer calls to filter function
- revert UnicodeSet(intprop=value) from trie ranges to range starts + lookup
- cache per-int-prop range starts: fewer lookups
2018-11-12 16:05:57 -08:00

1853 lines
64 KiB
C++
Raw Blame History

// © 2017 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
// umutablecptrie.cpp (inspired by utrie2_builder.cpp)
// created: 2017dec29 Markus W. Scherer
// #define UCPTRIE_DEBUG
#ifdef UCPTRIE_DEBUG
# include <stdio.h>
#endif
#include "unicode/utypes.h"
#include "unicode/ucptrie.h"
#include "unicode/umutablecptrie.h"
#include "unicode/uobject.h"
#include "unicode/utf16.h"
#include "cmemory.h"
#include "uassert.h"
#include "ucptrie_impl.h"
// ICU-20235 In case Microsoft math.h has defined this, undefine it.
#ifdef OVERFLOW
#undef OVERFLOW
#endif
U_NAMESPACE_BEGIN
namespace {
constexpr int32_t MAX_UNICODE = 0x10ffff;
constexpr int32_t UNICODE_LIMIT = 0x110000;
constexpr int32_t BMP_LIMIT = 0x10000;
constexpr int32_t ASCII_LIMIT = 0x80;
constexpr int32_t I_LIMIT = UNICODE_LIMIT >> UCPTRIE_SHIFT_3;
constexpr int32_t BMP_I_LIMIT = BMP_LIMIT >> UCPTRIE_SHIFT_3;
constexpr int32_t ASCII_I_LIMIT = ASCII_LIMIT >> UCPTRIE_SHIFT_3;
constexpr int32_t SMALL_DATA_BLOCKS_PER_BMP_BLOCK = (1 << (UCPTRIE_FAST_SHIFT - UCPTRIE_SHIFT_3));
// Flag values for data blocks.
constexpr uint8_t ALL_SAME = 0;
constexpr uint8_t MIXED = 1;
constexpr uint8_t SAME_AS = 2;
/** Start with allocation of 16k data entries. */
constexpr int32_t INITIAL_DATA_LENGTH = ((int32_t)1 << 14);
/** Grow about 8x each time. */
constexpr int32_t MEDIUM_DATA_LENGTH = ((int32_t)1 << 17);
/**
* Maximum length of the build-time data array.
* One entry per 0x110000 code points.
*/
constexpr int32_t MAX_DATA_LENGTH = UNICODE_LIMIT;
// Flag values for index-3 blocks while compacting/building.
constexpr uint8_t I3_NULL = 0;
constexpr uint8_t I3_BMP = 1;
constexpr uint8_t I3_16 = 2;
constexpr uint8_t I3_18 = 3;
constexpr int32_t INDEX_3_18BIT_BLOCK_LENGTH = UCPTRIE_INDEX_3_BLOCK_LENGTH + UCPTRIE_INDEX_3_BLOCK_LENGTH / 8;
class AllSameBlocks;
class MixedBlocks;
class MutableCodePointTrie : public UMemory {
public:
MutableCodePointTrie(uint32_t initialValue, uint32_t errorValue, UErrorCode &errorCode);
MutableCodePointTrie(const MutableCodePointTrie &other, UErrorCode &errorCode);
MutableCodePointTrie(const MutableCodePointTrie &other) = delete;
~MutableCodePointTrie();
MutableCodePointTrie &operator=(const MutableCodePointTrie &other) = delete;
static MutableCodePointTrie *fromUCPMap(const UCPMap *map, UErrorCode &errorCode);
static MutableCodePointTrie *fromUCPTrie(const UCPTrie *trie, UErrorCode &errorCode);
uint32_t get(UChar32 c) const;
int32_t getRange(UChar32 start, UCPMapValueFilter *filter, const void *context,
uint32_t *pValue) const;
void set(UChar32 c, uint32_t value, UErrorCode &errorCode);
void setRange(UChar32 start, UChar32 end, uint32_t value, UErrorCode &errorCode);
UCPTrie *build(UCPTrieType type, UCPTrieValueWidth valueWidth, UErrorCode &errorCode);
private:
void clear();
bool ensureHighStart(UChar32 c);
int32_t allocDataBlock(int32_t blockLength);
int32_t getDataBlock(int32_t i);
void maskValues(uint32_t mask);
UChar32 findHighStart() const;
int32_t compactWholeDataBlocks(int32_t fastILimit, AllSameBlocks &allSameBlocks);
int32_t compactData(
int32_t fastILimit, uint32_t *newData, int32_t newDataCapacity,
int32_t dataNullIndex, MixedBlocks &mixedBlocks, UErrorCode &errorCode);
int32_t compactIndex(int32_t fastILimit, MixedBlocks &mixedBlocks, UErrorCode &errorCode);
int32_t compactTrie(int32_t fastILimit, UErrorCode &errorCode);
uint32_t *index = nullptr;
int32_t indexCapacity = 0;
int32_t index3NullOffset = -1;
uint32_t *data = nullptr;
int32_t dataCapacity = 0;
int32_t dataLength = 0;
int32_t dataNullOffset = -1;
uint32_t origInitialValue;
uint32_t initialValue;
uint32_t errorValue;
UChar32 highStart;
uint32_t highValue;
#ifdef UCPTRIE_DEBUG
public:
const char *name;
#endif
private:
/** Temporary array while building the final data. */
uint16_t *index16 = nullptr;
uint8_t flags[UNICODE_LIMIT >> UCPTRIE_SHIFT_3];
};
MutableCodePointTrie::MutableCodePointTrie(uint32_t iniValue, uint32_t errValue, UErrorCode &errorCode) :
origInitialValue(iniValue), initialValue(iniValue), errorValue(errValue),
highStart(0), highValue(initialValue)
#ifdef UCPTRIE_DEBUG
, name("open")
#endif
{
if (U_FAILURE(errorCode)) { return; }
index = (uint32_t *)uprv_malloc(BMP_I_LIMIT * 4);
data = (uint32_t *)uprv_malloc(INITIAL_DATA_LENGTH * 4);
if (index == nullptr || data == nullptr) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return;
}
indexCapacity = BMP_I_LIMIT;
dataCapacity = INITIAL_DATA_LENGTH;
}
MutableCodePointTrie::MutableCodePointTrie(const MutableCodePointTrie &other, UErrorCode &errorCode) :
index3NullOffset(other.index3NullOffset),
dataNullOffset(other.dataNullOffset),
origInitialValue(other.origInitialValue), initialValue(other.initialValue),
errorValue(other.errorValue),
highStart(other.highStart), highValue(other.highValue)
#ifdef UCPTRIE_DEBUG
, name("mutable clone")
#endif
{
if (U_FAILURE(errorCode)) { return; }
int32_t iCapacity = highStart <= BMP_LIMIT ? BMP_I_LIMIT : I_LIMIT;
index = (uint32_t *)uprv_malloc(iCapacity * 4);
data = (uint32_t *)uprv_malloc(other.dataCapacity * 4);
if (index == nullptr || data == nullptr) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return;
}
indexCapacity = iCapacity;
dataCapacity = other.dataCapacity;
int32_t iLimit = highStart >> UCPTRIE_SHIFT_3;
uprv_memcpy(flags, other.flags, iLimit);
uprv_memcpy(index, other.index, iLimit * 4);
uprv_memcpy(data, other.data, (size_t)other.dataLength * 4);
dataLength = other.dataLength;
U_ASSERT(other.index16 == nullptr);
}
MutableCodePointTrie::~MutableCodePointTrie() {
uprv_free(index);
uprv_free(data);
uprv_free(index16);
}
MutableCodePointTrie *MutableCodePointTrie::fromUCPMap(const UCPMap *map, UErrorCode &errorCode) {
// Use the highValue as the initialValue to reduce the highStart.
uint32_t errorValue = ucpmap_get(map, -1);
uint32_t initialValue = ucpmap_get(map, 0x10ffff);
LocalPointer<MutableCodePointTrie> mutableTrie(
new MutableCodePointTrie(initialValue, errorValue, errorCode),
errorCode);
if (U_FAILURE(errorCode)) {
return nullptr;
}
UChar32 start = 0, end;
uint32_t value;
while ((end = ucpmap_getRange(map, start, UCPMAP_RANGE_NORMAL, 0,
nullptr, nullptr, &value)) >= 0) {
if (value != initialValue) {
if (start == end) {
mutableTrie->set(start, value, errorCode);
} else {
mutableTrie->setRange(start, end, value, errorCode);
}
}
start = end + 1;
}
if (U_SUCCESS(errorCode)) {
return mutableTrie.orphan();
} else {
return nullptr;
}
}
MutableCodePointTrie *MutableCodePointTrie::fromUCPTrie(const UCPTrie *trie, UErrorCode &errorCode) {
// Use the highValue as the initialValue to reduce the highStart.
uint32_t errorValue;
uint32_t initialValue;
switch (trie->valueWidth) {
case UCPTRIE_VALUE_BITS_16:
errorValue = trie->data.ptr16[trie->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET];
initialValue = trie->data.ptr16[trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET];
break;
case UCPTRIE_VALUE_BITS_32:
errorValue = trie->data.ptr32[trie->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET];
initialValue = trie->data.ptr32[trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET];
break;
case UCPTRIE_VALUE_BITS_8:
errorValue = trie->data.ptr8[trie->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET];
initialValue = trie->data.ptr8[trie->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET];
break;
default:
// Unreachable if the trie is properly initialized.
errorCode = U_ILLEGAL_ARGUMENT_ERROR;
return nullptr;
}
LocalPointer<MutableCodePointTrie> mutableTrie(
new MutableCodePointTrie(initialValue, errorValue, errorCode),
errorCode);
if (U_FAILURE(errorCode)) {
return nullptr;
}
UChar32 start = 0, end;
uint32_t value;
while ((end = ucptrie_getRange(trie, start, UCPMAP_RANGE_NORMAL, 0,
nullptr, nullptr, &value)) >= 0) {
if (value != initialValue) {
if (start == end) {
mutableTrie->set(start, value, errorCode);
} else {
mutableTrie->setRange(start, end, value, errorCode);
}
}
start = end + 1;
}
if (U_SUCCESS(errorCode)) {
return mutableTrie.orphan();
} else {
return nullptr;
}
}
void MutableCodePointTrie::clear() {
index3NullOffset = dataNullOffset = -1;
dataLength = 0;
highValue = initialValue = origInitialValue;
highStart = 0;
uprv_free(index16);
index16 = nullptr;
}
uint32_t MutableCodePointTrie::get(UChar32 c) const {
if ((uint32_t)c > MAX_UNICODE) {
return errorValue;
}
if (c >= highStart) {
return highValue;
}
int32_t i = c >> UCPTRIE_SHIFT_3;
if (flags[i] == ALL_SAME) {
return index[i];
} else {
return data[index[i] + (c & UCPTRIE_SMALL_DATA_MASK)];
}
}
inline uint32_t maybeFilterValue(uint32_t value, uint32_t initialValue, uint32_t nullValue,
UCPMapValueFilter *filter, const void *context) {
if (value == initialValue) {
value = nullValue;
} else if (filter != nullptr) {
value = filter(context, value);
}
return value;
}
UChar32 MutableCodePointTrie::getRange(
UChar32 start, UCPMapValueFilter *filter, const void *context,
uint32_t *pValue) const {
if ((uint32_t)start > MAX_UNICODE) {
return U_SENTINEL;
}
if (start >= highStart) {
if (pValue != nullptr) {
uint32_t value = highValue;
if (filter != nullptr) { value = filter(context, value); }
*pValue = value;
}
return MAX_UNICODE;
}
uint32_t nullValue = initialValue;
if (filter != nullptr) { nullValue = filter(context, nullValue); }
UChar32 c = start;
uint32_t trieValue, value;
bool haveValue = false;
int32_t i = c >> UCPTRIE_SHIFT_3;
do {
if (flags[i] == ALL_SAME) {
uint32_t trieValue2 = index[i];
if (haveValue) {
if (trieValue2 != trieValue) {
if (filter == nullptr ||
maybeFilterValue(trieValue2, initialValue, nullValue,
filter, context) != value) {
return c - 1;
}
trieValue = trieValue2; // may or may not help
}
} else {
trieValue = trieValue2;
value = maybeFilterValue(trieValue2, initialValue, nullValue, filter, context);
if (pValue != nullptr) { *pValue = value; }
haveValue = true;
}
c = (c + UCPTRIE_SMALL_DATA_BLOCK_LENGTH) & ~UCPTRIE_SMALL_DATA_MASK;
} else /* MIXED */ {
int32_t di = index[i] + (c & UCPTRIE_SMALL_DATA_MASK);
uint32_t trieValue2 = data[di];
if (haveValue) {
if (trieValue2 != trieValue) {
if (filter == nullptr ||
maybeFilterValue(trieValue2, initialValue, nullValue,
filter, context) != value) {
return c - 1;
}
trieValue = trieValue2; // may or may not help
}
} else {
trieValue = trieValue2;
value = maybeFilterValue(trieValue2, initialValue, nullValue, filter, context);
if (pValue != nullptr) { *pValue = value; }
haveValue = true;
}
while ((++c & UCPTRIE_SMALL_DATA_MASK) != 0) {
trieValue2 = data[++di];
if (trieValue2 != trieValue) {
if (filter == nullptr ||
maybeFilterValue(trieValue2, initialValue, nullValue,
filter, context) != value) {
return c - 1;
}
}
trieValue = trieValue2; // may or may not help
}
}
++i;
} while (c < highStart);
U_ASSERT(haveValue);
if (maybeFilterValue(highValue, initialValue, nullValue,
filter, context) != value) {
return c - 1;
} else {
return MAX_UNICODE;
}
}
void
writeBlock(uint32_t *block, uint32_t value) {
uint32_t *limit = block + UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
while (block < limit) {
*block++ = value;
}
}
bool MutableCodePointTrie::ensureHighStart(UChar32 c) {
if (c >= highStart) {
// Round up to a UCPTRIE_CP_PER_INDEX_2_ENTRY boundary to simplify compaction.
c = (c + UCPTRIE_CP_PER_INDEX_2_ENTRY) & ~(UCPTRIE_CP_PER_INDEX_2_ENTRY - 1);
int32_t i = highStart >> UCPTRIE_SHIFT_3;
int32_t iLimit = c >> UCPTRIE_SHIFT_3;
if (iLimit > indexCapacity) {
uint32_t *newIndex = (uint32_t *)uprv_malloc(I_LIMIT * 4);
if (newIndex == nullptr) { return false; }
uprv_memcpy(newIndex, index, i * 4);
uprv_free(index);
index = newIndex;
indexCapacity = I_LIMIT;
}
do {
flags[i] = ALL_SAME;
index[i] = initialValue;
} while(++i < iLimit);
highStart = c;
}
return true;
}
int32_t MutableCodePointTrie::allocDataBlock(int32_t blockLength) {
int32_t newBlock = dataLength;
int32_t newTop = newBlock + blockLength;
if (newTop > dataCapacity) {
int32_t capacity;
if (dataCapacity < MEDIUM_DATA_LENGTH) {
capacity = MEDIUM_DATA_LENGTH;
} else if (dataCapacity < MAX_DATA_LENGTH) {
capacity = MAX_DATA_LENGTH;
} else {
// Should never occur.
// Either MAX_DATA_LENGTH is incorrect,
// or the code writes more values than should be possible.
return -1;
}
uint32_t *newData = (uint32_t *)uprv_malloc(capacity * 4);
if (newData == nullptr) {
return -1;
}
uprv_memcpy(newData, data, (size_t)dataLength * 4);
uprv_free(data);
data = newData;
dataCapacity = capacity;
}
dataLength = newTop;
return newBlock;
}
/**
* No error checking for illegal arguments.
*
* @return -1 if no new data block available (out of memory in data array)
* @internal
*/
int32_t MutableCodePointTrie::getDataBlock(int32_t i) {
if (flags[i] == MIXED) {
return index[i];
}
if (i < BMP_I_LIMIT) {
int32_t newBlock = allocDataBlock(UCPTRIE_FAST_DATA_BLOCK_LENGTH);
if (newBlock < 0) { return newBlock; }
int32_t iStart = i & ~(SMALL_DATA_BLOCKS_PER_BMP_BLOCK -1);
int32_t iLimit = iStart + SMALL_DATA_BLOCKS_PER_BMP_BLOCK;
do {
U_ASSERT(flags[iStart] == ALL_SAME);
writeBlock(data + newBlock, index[iStart]);
flags[iStart] = MIXED;
index[iStart++] = newBlock;
newBlock += UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
} while (iStart < iLimit);
return index[i];
} else {
int32_t newBlock = allocDataBlock(UCPTRIE_SMALL_DATA_BLOCK_LENGTH);
if (newBlock < 0) { return newBlock; }
writeBlock(data + newBlock, index[i]);
flags[i] = MIXED;
index[i] = newBlock;
return newBlock;
}
}
void MutableCodePointTrie::set(UChar32 c, uint32_t value, UErrorCode &errorCode) {
if (U_FAILURE(errorCode)) {
return;
}
if ((uint32_t)c > MAX_UNICODE) {
errorCode = U_ILLEGAL_ARGUMENT_ERROR;
return;
}
int32_t block;
if (!ensureHighStart(c) || (block = getDataBlock(c >> UCPTRIE_SHIFT_3)) < 0) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return;
}
data[block + (c & UCPTRIE_SMALL_DATA_MASK)] = value;
}
void
fillBlock(uint32_t *block, UChar32 start, UChar32 limit, uint32_t value) {
uint32_t *pLimit = block + limit;
block += start;
while (block < pLimit) {
*block++ = value;
}
}
void MutableCodePointTrie::setRange(UChar32 start, UChar32 end, uint32_t value, UErrorCode &errorCode) {
if (U_FAILURE(errorCode)) {
return;
}
if ((uint32_t)start > MAX_UNICODE || (uint32_t)end > MAX_UNICODE || start > end) {
errorCode = U_ILLEGAL_ARGUMENT_ERROR;
return;
}
if (!ensureHighStart(end)) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return;
}
UChar32 limit = end + 1;
if (start & UCPTRIE_SMALL_DATA_MASK) {
// Set partial block at [start..following block boundary[.
int32_t block = getDataBlock(start >> UCPTRIE_SHIFT_3);
if (block < 0) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return;
}
UChar32 nextStart = (start + UCPTRIE_SMALL_DATA_MASK) & ~UCPTRIE_SMALL_DATA_MASK;
if (nextStart <= limit) {
fillBlock(data + block, start & UCPTRIE_SMALL_DATA_MASK, UCPTRIE_SMALL_DATA_BLOCK_LENGTH,
value);
start = nextStart;
} else {
fillBlock(data + block, start & UCPTRIE_SMALL_DATA_MASK, limit & UCPTRIE_SMALL_DATA_MASK,
value);
return;
}
}
// Number of positions in the last, partial block.
int32_t rest = limit & UCPTRIE_SMALL_DATA_MASK;
// Round down limit to a block boundary.
limit &= ~UCPTRIE_SMALL_DATA_MASK;
// Iterate over all-value blocks.
while (start < limit) {
int32_t i = start >> UCPTRIE_SHIFT_3;
if (flags[i] == ALL_SAME) {
index[i] = value;
} else /* MIXED */ {
fillBlock(data + index[i], 0, UCPTRIE_SMALL_DATA_BLOCK_LENGTH, value);
}
start += UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
}
if (rest > 0) {
// Set partial block at [last block boundary..limit[.
int32_t block = getDataBlock(start >> UCPTRIE_SHIFT_3);
if (block < 0) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return;
}
fillBlock(data + block, 0, rest, value);
}
}
/* compaction --------------------------------------------------------------- */
void MutableCodePointTrie::maskValues(uint32_t mask) {
initialValue &= mask;
errorValue &= mask;
highValue &= mask;
int32_t iLimit = highStart >> UCPTRIE_SHIFT_3;
for (int32_t i = 0; i < iLimit; ++i) {
if (flags[i] == ALL_SAME) {
index[i] &= mask;
}
}
for (int32_t i = 0; i < dataLength; ++i) {
data[i] &= mask;
}
}
template<typename UIntA, typename UIntB>
bool equalBlocks(const UIntA *s, const UIntB *t, int32_t length) {
while (length > 0 && *s == *t) {
++s;
++t;
--length;
}
return length == 0;
}
bool allValuesSameAs(const uint32_t *p, int32_t length, uint32_t value) {
const uint32_t *pLimit = p + length;
while (p < pLimit && *p == value) { ++p; }
return p == pLimit;
}
/** Search for an identical block. */
int32_t findSameBlock(const uint16_t *p, int32_t pStart, int32_t length,
const uint16_t *q, int32_t qStart, int32_t blockLength) {
// Ensure that we do not even partially get past length.
length -= blockLength;
q += qStart;
while (pStart <= length) {
if (equalBlocks(p + pStart, q, blockLength)) {
return pStart;
}
++pStart;
}
return -1;
}
int32_t findAllSameBlock(const uint32_t *p, int32_t start, int32_t limit,
uint32_t value, int32_t blockLength) {
// Ensure that we do not even partially get past limit.
limit -= blockLength;
for (int32_t block = start; block <= limit; ++block) {
if (p[block] == value) {
for (int32_t i = 1;; ++i) {
if (i == blockLength) {
return block;
}
if (p[block + i] != value) {
block += i;
break;
}
}
}
}
return -1;
}
/**
* Look for maximum overlap of the beginning of the other block
* with the previous, adjacent block.
*/
template<typename UIntA, typename UIntB>
int32_t getOverlap(const UIntA *p, int32_t length,
const UIntB *q, int32_t qStart, int32_t blockLength) {
int32_t overlap = blockLength - 1;
U_ASSERT(overlap <= length);
q += qStart;
while (overlap > 0 && !equalBlocks(p + (length - overlap), q, overlap)) {
--overlap;
}
return overlap;
}
int32_t getAllSameOverlap(const uint32_t *p, int32_t length, uint32_t value,
int32_t blockLength) {
int32_t min = length - (blockLength - 1);
int32_t i = length;
while (min < i && p[i - 1] == value) { --i; }
return length - i;
}
bool isStartOfSomeFastBlock(uint32_t dataOffset, const uint32_t index[], int32_t fastILimit) {
for (int32_t i = 0; i < fastILimit; i += SMALL_DATA_BLOCKS_PER_BMP_BLOCK) {
if (index[i] == dataOffset) {
return true;
}
}
return false;
}
/**
* Finds the start of the last range in the trie by enumerating backward.
* Indexes for code points higher than this will be omitted.
*/
UChar32 MutableCodePointTrie::findHighStart() const {
int32_t i = highStart >> UCPTRIE_SHIFT_3;
while (i > 0) {
bool match;
if (flags[--i] == ALL_SAME) {
match = index[i] == highValue;
} else /* MIXED */ {
const uint32_t *p = data + index[i];
for (int32_t j = 0;; ++j) {
if (j == UCPTRIE_SMALL_DATA_BLOCK_LENGTH) {
match = true;
break;
}
if (p[j] != highValue) {
match = false;
break;
}
}
}
if (!match) {
return (i + 1) << UCPTRIE_SHIFT_3;
}
}
return 0;
}
class AllSameBlocks {
public:
static constexpr int32_t NEW_UNIQUE = -1;
static constexpr int32_t OVERFLOW = -2;
AllSameBlocks() : length(0), mostRecent(-1) {}
int32_t findOrAdd(int32_t index, int32_t count, uint32_t value) {
if (mostRecent >= 0 && values[mostRecent] == value) {
refCounts[mostRecent] += count;
return indexes[mostRecent];
}
for (int32_t i = 0; i < length; ++i) {
if (values[i] == value) {
mostRecent = i;
refCounts[i] += count;
return indexes[i];
}
}
if (length == CAPACITY) {
return OVERFLOW;
}
mostRecent = length;
indexes[length] = index;
values[length] = value;
refCounts[length++] = count;
return NEW_UNIQUE;
}
/** Replaces the block which has the lowest reference count. */
void add(int32_t index, int32_t count, uint32_t value) {
U_ASSERT(length == CAPACITY);
int32_t least = -1;
int32_t leastCount = I_LIMIT;
for (int32_t i = 0; i < length; ++i) {
U_ASSERT(values[i] != value);
if (refCounts[i] < leastCount) {
least = i;
leastCount = refCounts[i];
}
}
U_ASSERT(least >= 0);
mostRecent = least;
indexes[least] = index;
values[least] = value;
refCounts[least] = count;
}
int32_t findMostUsed() const {
if (length == 0) { return -1; }
int32_t max = -1;
int32_t maxCount = 0;
for (int32_t i = 0; i < length; ++i) {
if (refCounts[i] > maxCount) {
max = i;
maxCount = refCounts[i];
}
}
return indexes[max];
}
private:
static constexpr int32_t CAPACITY = 32;
int32_t length;
int32_t mostRecent;
int32_t indexes[CAPACITY];
uint32_t values[CAPACITY];
int32_t refCounts[CAPACITY];
};
// Custom hash table for mixed-value blocks to be found anywhere in the
// compacted data or index so far.
class MixedBlocks {
public:
MixedBlocks() {}
~MixedBlocks() {
uprv_free(table);
}
bool init(int32_t maxLength, int32_t newBlockLength) {
// We store actual data indexes + 1 to reserve 0 for empty entries.
int32_t maxDataIndex = maxLength - newBlockLength + 1;
int32_t newLength;
if (maxDataIndex <= 0xfff) { // 4k
newLength = 6007;
shift = 12;
mask = 0xfff;
} else if (maxDataIndex <= 0x7fff) { // 32k
newLength = 50021;
shift = 15;
mask = 0x7fff;
} else if (maxDataIndex <= 0x1ffff) { // 128k
newLength = 200003;
shift = 17;
mask = 0x1ffff;
} else {
// maxDataIndex up to around MAX_DATA_LENGTH, ca. 1.1M
newLength = 1500007;
shift = 21;
mask = 0x1fffff;
}
if (newLength > capacity) {
uprv_free(table);
table = (uint32_t *)uprv_malloc(newLength * 4);
if (table == nullptr) {
return false;
}
capacity = newLength;
}
length = newLength;
uprv_memset(table, 0, length * 4);
blockLength = newBlockLength;
return true;
}
template<typename UInt>
void extend(const UInt *data, int32_t minStart, int32_t prevDataLength, int32_t newDataLength) {
int32_t start = prevDataLength - blockLength;
if (start >= minStart) {
++start; // Skip the last block that we added last time.
} else {
start = minStart; // Begin with the first full block.
}
for (int32_t end = newDataLength - blockLength; start <= end; ++start) {
uint32_t hashCode = makeHashCode(data, start);
addEntry(data, start, hashCode, start);
}
}
template<typename UIntA, typename UIntB>
int32_t findBlock(const UIntA *data, const UIntB *blockData, int32_t blockStart) const {
uint32_t hashCode = makeHashCode(blockData, blockStart);
int32_t entryIndex = findEntry(data, blockData, blockStart, hashCode);
if (entryIndex >= 0) {
return (table[entryIndex] & mask) - 1;
} else {
return -1;
}
}
int32_t findAllSameBlock(const uint32_t *data, uint32_t blockValue) const {
uint32_t hashCode = makeHashCode(blockValue);
int32_t entryIndex = findEntry(data, blockValue, hashCode);
if (entryIndex >= 0) {
return (table[entryIndex] & mask) - 1;
} else {
return -1;
}
}
private:
template<typename UInt>
uint32_t makeHashCode(const UInt *blockData, int32_t blockStart) const {
int32_t blockLimit = blockStart + blockLength;
uint32_t hashCode = blockData[blockStart++];
do {
hashCode = 37 * hashCode + blockData[blockStart++];
} while (blockStart < blockLimit);
return hashCode;
}
uint32_t makeHashCode(uint32_t blockValue) const {
uint32_t hashCode = blockValue;
for (int32_t i = 1; i < blockLength; ++i) {
hashCode = 37 * hashCode + blockValue;
}
return hashCode;
}
template<typename UInt>
void addEntry(const UInt *data, int32_t blockStart, uint32_t hashCode, int32_t dataIndex) {
U_ASSERT(0 <= dataIndex && dataIndex < (int32_t)mask);
int32_t entryIndex = findEntry(data, data, blockStart, hashCode);
if (entryIndex < 0) {
table[~entryIndex] = (hashCode << shift) | (dataIndex + 1);
}
}
template<typename UIntA, typename UIntB>
int32_t findEntry(const UIntA *data, const UIntB *blockData, int32_t blockStart,
uint32_t hashCode) const {
uint32_t shiftedHashCode = hashCode << shift;
int32_t initialEntryIndex = (hashCode % (length - 1)) + 1; // 1..length-1
for (int32_t entryIndex = initialEntryIndex;;) {
uint32_t entry = table[entryIndex];
if (entry == 0) {
return ~entryIndex;
}
if ((entry & ~mask) == shiftedHashCode) {
int32_t dataIndex = (entry & mask) - 1;
if (equalBlocks(data + dataIndex, blockData + blockStart, blockLength)) {
return entryIndex;
}
}
entryIndex = nextIndex(initialEntryIndex, entryIndex);
}
}
int32_t findEntry(const uint32_t *data, uint32_t blockValue, uint32_t hashCode) const {
uint32_t shiftedHashCode = hashCode << shift;
int32_t initialEntryIndex = (hashCode % (length - 1)) + 1; // 1..length-1
for (int32_t entryIndex = initialEntryIndex;;) {
uint32_t entry = table[entryIndex];
if (entry == 0) {
return ~entryIndex;
}
if ((entry & ~mask) == shiftedHashCode) {
int32_t dataIndex = (entry & mask) - 1;
if (allValuesSameAs(data + dataIndex, blockLength, blockValue)) {
return entryIndex;
}
}
entryIndex = nextIndex(initialEntryIndex, entryIndex);
}
}
inline int32_t nextIndex(int32_t initialEntryIndex, int32_t entryIndex) const {
// U_ASSERT(0 < initialEntryIndex && initialEntryIndex < length);
return (entryIndex + initialEntryIndex) % length;
}
// Hash table.
// The length is a prime number, larger than the maximum data length.
// The "shift" lower bits store a data index + 1.
// The remaining upper bits store a partial hashCode of the block data values.
uint32_t *table = nullptr;
int32_t capacity = 0;
int32_t length = 0;
int32_t shift = 0;
uint32_t mask = 0;
int32_t blockLength = 0;
};
int32_t MutableCodePointTrie::compactWholeDataBlocks(int32_t fastILimit, AllSameBlocks &allSameBlocks) {
#ifdef UCPTRIE_DEBUG
bool overflow = false;
#endif
// ASCII data will be stored as a linear table, even if the following code
// does not yet count it that way.
int32_t newDataCapacity = ASCII_LIMIT;
// Add room for a small data null block in case it would match the start of
// a fast data block where dataNullOffset must not be set in that case.
newDataCapacity += UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
// Add room for special values (errorValue, highValue) and padding.
newDataCapacity += 4;
int32_t iLimit = highStart >> UCPTRIE_SHIFT_3;
int32_t blockLength = UCPTRIE_FAST_DATA_BLOCK_LENGTH;
int32_t inc = SMALL_DATA_BLOCKS_PER_BMP_BLOCK;
for (int32_t i = 0; i < iLimit; i += inc) {
if (i == fastILimit) {
blockLength = UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
inc = 1;
}
uint32_t value = index[i];
if (flags[i] == MIXED) {
// Really mixed?
const uint32_t *p = data + value;
value = *p;
if (allValuesSameAs(p + 1, blockLength - 1, value)) {
flags[i] = ALL_SAME;
index[i] = value;
// Fall through to ALL_SAME handling.
} else {
newDataCapacity += blockLength;
continue;
}
} else {
U_ASSERT(flags[i] == ALL_SAME);
if (inc > 1) {
// Do all of the fast-range data block's ALL_SAME parts have the same value?
bool allSame = true;
int32_t next_i = i + inc;
for (int32_t j = i + 1; j < next_i; ++j) {
U_ASSERT(flags[j] == ALL_SAME);
if (index[j] != value) {
allSame = false;
break;
}
}
if (!allSame) {
// Turn it into a MIXED block.
if (getDataBlock(i) < 0) {
return -1;
}
newDataCapacity += blockLength;
continue;
}
}
}
// Is there another ALL_SAME block with the same value?
int32_t other = allSameBlocks.findOrAdd(i, inc, value);
if (other == AllSameBlocks::OVERFLOW) {
// The fixed-size array overflowed. Slow check for a duplicate block.
#ifdef UCPTRIE_DEBUG
if (!overflow) {
puts("UCPTrie AllSameBlocks overflow");
overflow = true;
}
#endif
int32_t jInc = SMALL_DATA_BLOCKS_PER_BMP_BLOCK;
for (int32_t j = 0;; j += jInc) {
if (j == i) {
allSameBlocks.add(i, inc, value);
break;
}
if (j == fastILimit) {
jInc = 1;
}
if (flags[j] == ALL_SAME && index[j] == value) {
allSameBlocks.add(j, jInc + inc, value);
other = j;
break;
// We could keep counting blocks with the same value
// before we add the first one, which may improve compaction in rare cases,
// but it would make it slower.
}
}
}
if (other >= 0) {
flags[i] = SAME_AS;
index[i] = other;
} else {
// New unique same-value block.
newDataCapacity += blockLength;
}
}
return newDataCapacity;
}
#ifdef UCPTRIE_DEBUG
# define DEBUG_DO(expr) expr
#else
# define DEBUG_DO(expr)
#endif
#ifdef UCPTRIE_DEBUG
// Braille symbols: U+28xx = UTF-8 E2 A0 80..E2 A3 BF
int32_t appendValue(char s[], int32_t length, uint32_t value) {
value ^= value >> 16;
value ^= value >> 8;
s[length] = 0xE2;
s[length + 1] = (char)(0xA0 + ((value >> 6) & 3));
s[length + 2] = (char)(0x80 + (value & 0x3F));
return length + 3;
}
void printBlock(const uint32_t *block, int32_t blockLength, uint32_t value,
UChar32 start, int32_t overlap, uint32_t initialValue) {
char s[UCPTRIE_FAST_DATA_BLOCK_LENGTH * 3 + 3];
int32_t length = 0;
int32_t i;
for (i = 0; i < overlap; ++i) {
length = appendValue(s, length, 0); // Braille blank
}
s[length++] = '|';
for (; i < blockLength; ++i) {
if (block != nullptr) {
value = block[i];
}
if (value == initialValue) {
value = 0x40; // Braille lower left dot
}
length = appendValue(s, length, value);
}
s[length] = 0;
start += overlap;
if (start <= 0xffff) {
printf(" %04lX %s|\n", (long)start, s);
} else if (start <= 0xfffff) {
printf(" %5lX %s|\n", (long)start, s);
} else {
printf(" %6lX %s|\n", (long)start, s);
}
}
#endif
/**
* Compacts a build-time trie.
*
* The compaction
* - removes blocks that are identical with earlier ones
* - overlaps each new non-duplicate block as much as possible with the previously-written one
* - works with fast-range data blocks whose length is a multiple of that of
* higher-code-point data blocks
*
* It does not try to find an optimal order of writing, deduplicating, and overlapping blocks.
*/
int32_t MutableCodePointTrie::compactData(
int32_t fastILimit, uint32_t *newData, int32_t newDataCapacity,
int32_t dataNullIndex, MixedBlocks &mixedBlocks, UErrorCode &errorCode) {
#ifdef UCPTRIE_DEBUG
int32_t countSame=0, sumOverlaps=0;
bool printData = dataLength == 29088 /* line.brk */ ||
// dataLength == 30048 /* CanonIterData */ ||
dataLength == 50400 /* zh.txt~stroke */;
#endif
// The linear ASCII data has been copied into newData already.
int32_t newDataLength = 0;
for (int32_t i = 0; newDataLength < ASCII_LIMIT;
newDataLength += UCPTRIE_FAST_DATA_BLOCK_LENGTH, i += SMALL_DATA_BLOCKS_PER_BMP_BLOCK) {
index[i] = newDataLength;
#ifdef UCPTRIE_DEBUG
if (printData) {
printBlock(newData + newDataLength, UCPTRIE_FAST_DATA_BLOCK_LENGTH, 0, newDataLength, 0, initialValue);
}
#endif
}
int32_t blockLength = UCPTRIE_FAST_DATA_BLOCK_LENGTH;
if (!mixedBlocks.init(newDataCapacity, blockLength)) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
mixedBlocks.extend(newData, 0, 0, newDataLength);
int32_t iLimit = highStart >> UCPTRIE_SHIFT_3;
int32_t inc = SMALL_DATA_BLOCKS_PER_BMP_BLOCK;
int32_t fastLength = 0;
for (int32_t i = ASCII_I_LIMIT; i < iLimit; i += inc) {
if (i == fastILimit) {
blockLength = UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
inc = 1;
fastLength = newDataLength;
if (!mixedBlocks.init(newDataCapacity, blockLength)) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
mixedBlocks.extend(newData, 0, 0, newDataLength);
}
if (flags[i] == ALL_SAME) {
uint32_t value = index[i];
// Find an earlier part of the data array of length blockLength
// that is filled with this value.
int32_t n = mixedBlocks.findAllSameBlock(newData, value);
// If we find a match, and the current block is the data null block,
// and it is not a fast block but matches the start of a fast block,
// then we need to continue looking.
// This is because this small block is shorter than the fast block,
// and not all of the rest of the fast block is filled with this value.
// Otherwise trie.getRange() would detect that the fast block starts at
// dataNullOffset and assume incorrectly that it is filled with the null value.
while (n >= 0 && i == dataNullIndex && i >= fastILimit && n < fastLength &&
isStartOfSomeFastBlock(n, index, fastILimit)) {
n = findAllSameBlock(newData, n + 1, newDataLength, value, blockLength);
}
if (n >= 0) {
DEBUG_DO(++countSame);
index[i] = n;
} else {
n = getAllSameOverlap(newData, newDataLength, value, blockLength);
DEBUG_DO(sumOverlaps += n);
#ifdef UCPTRIE_DEBUG
if (printData) {
printBlock(nullptr, blockLength, value, i << UCPTRIE_SHIFT_3, n, initialValue);
}
#endif
index[i] = newDataLength - n;
int32_t prevDataLength = newDataLength;
while (n < blockLength) {
newData[newDataLength++] = value;
++n;
}
mixedBlocks.extend(newData, 0, prevDataLength, newDataLength);
}
} else if (flags[i] == MIXED) {
const uint32_t *block = data + index[i];
int32_t n = mixedBlocks.findBlock(newData, block, 0);
if (n >= 0) {
DEBUG_DO(++countSame);
index[i] = n;
} else {
n = getOverlap(newData, newDataLength, block, 0, blockLength);
DEBUG_DO(sumOverlaps += n);
#ifdef UCPTRIE_DEBUG
if (printData) {
printBlock(block, blockLength, 0, i << UCPTRIE_SHIFT_3, n, initialValue);
}
#endif
index[i] = newDataLength - n;
int32_t prevDataLength = newDataLength;
while (n < blockLength) {
newData[newDataLength++] = block[n++];
}
mixedBlocks.extend(newData, 0, prevDataLength, newDataLength);
}
} else /* SAME_AS */ {
uint32_t j = index[i];
index[i] = index[j];
}
}
#ifdef UCPTRIE_DEBUG
/* we saved some space */
printf("compacting UCPTrie: count of 32-bit data words %lu->%lu countSame=%ld sumOverlaps=%ld\n",
(long)dataLength, (long)newDataLength, (long)countSame, (long)sumOverlaps);
#endif
return newDataLength;
}
int32_t MutableCodePointTrie::compactIndex(int32_t fastILimit, MixedBlocks &mixedBlocks,
UErrorCode &errorCode) {
int32_t fastIndexLength = fastILimit >> (UCPTRIE_FAST_SHIFT - UCPTRIE_SHIFT_3);
if ((highStart >> UCPTRIE_FAST_SHIFT) <= fastIndexLength) {
// Only the linear fast index, no multi-stage index tables.
index3NullOffset = UCPTRIE_NO_INDEX3_NULL_OFFSET;
return fastIndexLength;
}
// Condense the fast index table.
// Also, does it contain an index-3 block with all dataNullOffset?
uint16_t fastIndex[UCPTRIE_BMP_INDEX_LENGTH]; // fastIndexLength
int32_t i3FirstNull = -1;
for (int32_t i = 0, j = 0; i < fastILimit; ++j) {
uint32_t i3 = index[i];
fastIndex[j] = (uint16_t)i3;
if (i3 == (uint32_t)dataNullOffset) {
if (i3FirstNull < 0) {
i3FirstNull = j;
} else if (index3NullOffset < 0 &&
(j - i3FirstNull + 1) == UCPTRIE_INDEX_3_BLOCK_LENGTH) {
index3NullOffset = i3FirstNull;
}
} else {
i3FirstNull = -1;
}
// Set the index entries that compactData() skipped.
// Needed when the multi-stage index covers the fast index range as well.
int32_t iNext = i + SMALL_DATA_BLOCKS_PER_BMP_BLOCK;
while (++i < iNext) {
i3 += UCPTRIE_SMALL_DATA_BLOCK_LENGTH;
index[i] = i3;
}
}
if (!mixedBlocks.init(fastIndexLength, UCPTRIE_INDEX_3_BLOCK_LENGTH)) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
mixedBlocks.extend(fastIndex, 0, 0, fastIndexLength);
// Examine index-3 blocks. For each determine one of:
// - same as the index-3 null block
// - same as a fast-index block
// - 16-bit indexes
// - 18-bit indexes
// We store this in the first flags entry for the index-3 block.
//
// Also determine an upper limit for the index-3 table length.
int32_t index3Capacity = 0;
i3FirstNull = index3NullOffset;
bool hasLongI3Blocks = false;
// If the fast index covers the whole BMP, then
// the multi-stage index is only for supplementary code points.
// Otherwise, the multi-stage index covers all of Unicode.
int32_t iStart = fastILimit < BMP_I_LIMIT ? 0 : BMP_I_LIMIT;
int32_t iLimit = highStart >> UCPTRIE_SHIFT_3;
for (int32_t i = iStart; i < iLimit;) {
int32_t j = i;
int32_t jLimit = i + UCPTRIE_INDEX_3_BLOCK_LENGTH;
uint32_t oredI3 = 0;
bool isNull = true;
do {
uint32_t i3 = index[j];
oredI3 |= i3;
if (i3 != (uint32_t)dataNullOffset) {
isNull = false;
}
} while (++j < jLimit);
if (isNull) {
flags[i] = I3_NULL;
if (i3FirstNull < 0) {
if (oredI3 <= 0xffff) {
index3Capacity += UCPTRIE_INDEX_3_BLOCK_LENGTH;
} else {
index3Capacity += INDEX_3_18BIT_BLOCK_LENGTH;
hasLongI3Blocks = true;
}
i3FirstNull = 0;
}
} else {
if (oredI3 <= 0xffff) {
int32_t n = mixedBlocks.findBlock(fastIndex, index, i);
if (n >= 0) {
flags[i] = I3_BMP;
index[i] = n;
} else {
flags[i] = I3_16;
index3Capacity += UCPTRIE_INDEX_3_BLOCK_LENGTH;
}
} else {
flags[i] = I3_18;
index3Capacity += INDEX_3_18BIT_BLOCK_LENGTH;
hasLongI3Blocks = true;
}
}
i = j;
}
int32_t index2Capacity = (iLimit - iStart) >> UCPTRIE_SHIFT_2_3;
// Length of the index-1 table, rounded up.
int32_t index1Length = (index2Capacity + UCPTRIE_INDEX_2_MASK) >> UCPTRIE_SHIFT_1_2;
// Index table: Fast index, index-1, index-3, index-2.
// +1 for possible index table padding.
int32_t index16Capacity = fastIndexLength + index1Length + index3Capacity + index2Capacity + 1;
index16 = (uint16_t *)uprv_malloc(index16Capacity * 2);
if (index16 == nullptr) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
uprv_memcpy(index16, fastIndex, fastIndexLength * 2);
if (!mixedBlocks.init(index16Capacity, UCPTRIE_INDEX_3_BLOCK_LENGTH)) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
MixedBlocks longI3Blocks;
if (hasLongI3Blocks) {
if (!longI3Blocks.init(index16Capacity, INDEX_3_18BIT_BLOCK_LENGTH)) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
}
// Compact the index-3 table and write an uncompacted version of the index-2 table.
uint16_t index2[UNICODE_LIMIT >> UCPTRIE_SHIFT_2]; // index2Capacity
int32_t i2Length = 0;
i3FirstNull = index3NullOffset;
int32_t index3Start = fastIndexLength + index1Length;
int32_t indexLength = index3Start;
for (int32_t i = iStart; i < iLimit; i += UCPTRIE_INDEX_3_BLOCK_LENGTH) {
int32_t i3;
uint8_t f = flags[i];
if (f == I3_NULL && i3FirstNull < 0) {
// First index-3 null block. Write & overlap it like a normal block, then remember it.
f = dataNullOffset <= 0xffff ? I3_16 : I3_18;
i3FirstNull = 0;
}
if (f == I3_NULL) {
i3 = index3NullOffset;
} else if (f == I3_BMP) {
i3 = index[i];
} else if (f == I3_16) {
int32_t n = mixedBlocks.findBlock(index16, index, i);
if (n >= 0) {
i3 = n;
} else {
if (indexLength == index3Start) {
// No overlap at the boundary between the index-1 and index-3 tables.
n = 0;
} else {
n = getOverlap(index16, indexLength,
index, i, UCPTRIE_INDEX_3_BLOCK_LENGTH);
}
i3 = indexLength - n;
int32_t prevIndexLength = indexLength;
while (n < UCPTRIE_INDEX_3_BLOCK_LENGTH) {
index16[indexLength++] = index[i + n++];
}
mixedBlocks.extend(index16, index3Start, prevIndexLength, indexLength);
if (hasLongI3Blocks) {
longI3Blocks.extend(index16, index3Start, prevIndexLength, indexLength);
}
}
} else {
U_ASSERT(f == I3_18);
U_ASSERT(hasLongI3Blocks);
// Encode an index-3 block that contains one or more data indexes exceeding 16 bits.
int32_t j = i;
int32_t jLimit = i + UCPTRIE_INDEX_3_BLOCK_LENGTH;
int32_t k = indexLength;
do {
++k;
uint32_t v = index[j++];
uint32_t upperBits = (v & 0x30000) >> 2;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 4;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 6;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 8;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 10;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 12;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 14;
index16[k++] = v;
v = index[j++];
upperBits |= (v & 0x30000) >> 16;
index16[k++] = v;
index16[k - 9] = upperBits;
} while (j < jLimit);
int32_t n = longI3Blocks.findBlock(index16, index16, indexLength);
if (n >= 0) {
i3 = n | 0x8000;
} else {
if (indexLength == index3Start) {
// No overlap at the boundary between the index-1 and index-3 tables.
n = 0;
} else {
n = getOverlap(index16, indexLength,
index16, indexLength, INDEX_3_18BIT_BLOCK_LENGTH);
}
i3 = (indexLength - n) | 0x8000;
int32_t prevIndexLength = indexLength;
if (n > 0) {
int32_t start = indexLength;
while (n < INDEX_3_18BIT_BLOCK_LENGTH) {
index16[indexLength++] = index16[start + n++];
}
} else {
indexLength += INDEX_3_18BIT_BLOCK_LENGTH;
}
mixedBlocks.extend(index16, index3Start, prevIndexLength, indexLength);
if (hasLongI3Blocks) {
longI3Blocks.extend(index16, index3Start, prevIndexLength, indexLength);
}
}
}
if (index3NullOffset < 0 && i3FirstNull >= 0) {
index3NullOffset = i3;
}
// Set the index-2 table entry.
index2[i2Length++] = i3;
}
U_ASSERT(i2Length == index2Capacity);
U_ASSERT(indexLength <= index3Start + index3Capacity);
if (index3NullOffset < 0) {
index3NullOffset = UCPTRIE_NO_INDEX3_NULL_OFFSET;
}
if (indexLength >= (UCPTRIE_NO_INDEX3_NULL_OFFSET + UCPTRIE_INDEX_3_BLOCK_LENGTH)) {
// The index-3 offsets exceed 15 bits, or
// the last one cannot be distinguished from the no-null-block value.
errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
return 0;
}
// Compact the index-2 table and write the index-1 table.
static_assert(UCPTRIE_INDEX_2_BLOCK_LENGTH == UCPTRIE_INDEX_3_BLOCK_LENGTH,
"must re-init mixedBlocks");
int32_t blockLength = UCPTRIE_INDEX_2_BLOCK_LENGTH;
int32_t i1 = fastIndexLength;
for (int32_t i = 0; i < i2Length; i += blockLength) {
int32_t n;
if ((i2Length - i) >= blockLength) {
// normal block
U_ASSERT(blockLength == UCPTRIE_INDEX_2_BLOCK_LENGTH);
n = mixedBlocks.findBlock(index16, index2, i);
} else {
// highStart is inside the last index-2 block. Shorten it.
blockLength = i2Length - i;
n = findSameBlock(index16, index3Start, indexLength,
index2, i, blockLength);
}
int32_t i2;
if (n >= 0) {
i2 = n;
} else {
if (indexLength == index3Start) {
// No overlap at the boundary between the index-1 and index-3/2 tables.
n = 0;
} else {
n = getOverlap(index16, indexLength, index2, i, blockLength);
}
i2 = indexLength - n;
int32_t prevIndexLength = indexLength;
while (n < blockLength) {
index16[indexLength++] = index2[i + n++];
}
mixedBlocks.extend(index16, index3Start, prevIndexLength, indexLength);
}
// Set the index-1 table entry.
index16[i1++] = i2;
}
U_ASSERT(i1 == index3Start);
U_ASSERT(indexLength <= index16Capacity);
#ifdef UCPTRIE_DEBUG
/* we saved some space */
printf("compacting UCPTrie: count of 16-bit index words %lu->%lu\n",
(long)iLimit, (long)indexLength);
#endif
return indexLength;
}
int32_t MutableCodePointTrie::compactTrie(int32_t fastILimit, UErrorCode &errorCode) {
// Find the real highStart and round it up.
U_ASSERT((highStart & (UCPTRIE_CP_PER_INDEX_2_ENTRY - 1)) == 0);
highValue = get(MAX_UNICODE);
int32_t realHighStart = findHighStart();
realHighStart = (realHighStart + (UCPTRIE_CP_PER_INDEX_2_ENTRY - 1)) &
~(UCPTRIE_CP_PER_INDEX_2_ENTRY - 1);
if (realHighStart == UNICODE_LIMIT) {
highValue = initialValue;
}
#ifdef UCPTRIE_DEBUG
printf("UCPTrie: highStart U+%06lx highValue 0x%lx initialValue 0x%lx\n",
(long)realHighStart, (long)highValue, (long)initialValue);
#endif
// We always store indexes and data values for the fast range.
// Pin highStart to the top of that range while building.
UChar32 fastLimit = fastILimit << UCPTRIE_SHIFT_3;
if (realHighStart < fastLimit) {
for (int32_t i = (realHighStart >> UCPTRIE_SHIFT_3); i < fastILimit; ++i) {
flags[i] = ALL_SAME;
index[i] = highValue;
}
highStart = fastLimit;
} else {
highStart = realHighStart;
}
uint32_t asciiData[ASCII_LIMIT];
for (int32_t i = 0; i < ASCII_LIMIT; ++i) {
asciiData[i] = get(i);
}
// First we look for which data blocks have the same value repeated over the whole block,
// deduplicate such blocks, find a good null data block (for faster enumeration),
// and get an upper bound for the necessary data array length.
AllSameBlocks allSameBlocks;
int32_t newDataCapacity = compactWholeDataBlocks(fastILimit, allSameBlocks);
if (newDataCapacity < 0) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
uint32_t *newData = (uint32_t *)uprv_malloc(newDataCapacity * 4);
if (newData == nullptr) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
uprv_memcpy(newData, asciiData, sizeof(asciiData));
int32_t dataNullIndex = allSameBlocks.findMostUsed();
MixedBlocks mixedBlocks;
int32_t newDataLength = compactData(fastILimit, newData, newDataCapacity,
dataNullIndex, mixedBlocks, errorCode);
if (U_FAILURE(errorCode)) { return 0; }
U_ASSERT(newDataLength <= newDataCapacity);
uprv_free(data);
data = newData;
dataCapacity = newDataCapacity;
dataLength = newDataLength;
if (dataLength > (0x3ffff + UCPTRIE_SMALL_DATA_BLOCK_LENGTH)) {
// The offset of the last data block is too high to be stored in the index table.
errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
return 0;
}
if (dataNullIndex >= 0) {
dataNullOffset = index[dataNullIndex];
#ifdef UCPTRIE_DEBUG
if (data[dataNullOffset] != initialValue) {
printf("UCPTrie initialValue %lx -> more common nullValue %lx\n",
(long)initialValue, (long)data[dataNullOffset]);
}
#endif
initialValue = data[dataNullOffset];
} else {
dataNullOffset = UCPTRIE_NO_DATA_NULL_OFFSET;
}
int32_t indexLength = compactIndex(fastILimit, mixedBlocks, errorCode);
highStart = realHighStart;
return indexLength;
}
UCPTrie *MutableCodePointTrie::build(UCPTrieType type, UCPTrieValueWidth valueWidth, UErrorCode &errorCode) {
if (U_FAILURE(errorCode)) {
return nullptr;
}
if (type < UCPTRIE_TYPE_FAST || UCPTRIE_TYPE_SMALL < type ||
valueWidth < UCPTRIE_VALUE_BITS_16 || UCPTRIE_VALUE_BITS_8 < valueWidth) {
errorCode = U_ILLEGAL_ARGUMENT_ERROR;
return nullptr;
}
// The mutable trie always stores 32-bit values.
// When we build a UCPTrie for a smaller value width, we first mask off unused bits
// before compacting the data.
switch (valueWidth) {
case UCPTRIE_VALUE_BITS_32:
break;
case UCPTRIE_VALUE_BITS_16:
maskValues(0xffff);
break;
case UCPTRIE_VALUE_BITS_8:
maskValues(0xff);
break;
default:
break;
}
UChar32 fastLimit = type == UCPTRIE_TYPE_FAST ? BMP_LIMIT : UCPTRIE_SMALL_LIMIT;
int32_t indexLength = compactTrie(fastLimit >> UCPTRIE_SHIFT_3, errorCode);
if (U_FAILURE(errorCode)) {
clear();
return nullptr;
}
// Ensure data table alignment: The index length must be even for uint32_t data.
if (valueWidth == UCPTRIE_VALUE_BITS_32 && (indexLength & 1) != 0) {
index16[indexLength++] = 0xffee; // arbitrary value
}
// Make the total trie structure length a multiple of 4 bytes by padding the data table,
// and store special values as the last two data values.
int32_t length = indexLength * 2;
if (valueWidth == UCPTRIE_VALUE_BITS_16) {
if (((indexLength ^ dataLength) & 1) != 0) {
// padding
data[dataLength++] = errorValue;
}
if (data[dataLength - 1] != errorValue || data[dataLength - 2] != highValue) {
data[dataLength++] = highValue;
data[dataLength++] = errorValue;
}
length += dataLength * 2;
} else if (valueWidth == UCPTRIE_VALUE_BITS_32) {
// 32-bit data words never need padding to a multiple of 4 bytes.
if (data[dataLength - 1] != errorValue || data[dataLength - 2] != highValue) {
if (data[dataLength - 1] != highValue) {
data[dataLength++] = highValue;
}
data[dataLength++] = errorValue;
}
length += dataLength * 4;
} else {
int32_t and3 = (length + dataLength) & 3;
if (and3 == 0 && data[dataLength - 1] == errorValue && data[dataLength - 2] == highValue) {
// all set
} else if(and3 == 3 && data[dataLength - 1] == highValue) {
data[dataLength++] = errorValue;
} else {
while (and3 != 2) {
data[dataLength++] = highValue;
and3 = (and3 + 1) & 3;
}
data[dataLength++] = highValue;
data[dataLength++] = errorValue;
}
length += dataLength;
}
// Calculate the total length of the UCPTrie as a single memory block.
length += sizeof(UCPTrie);
U_ASSERT((length & 3) == 0);
uint8_t *bytes = (uint8_t *)uprv_malloc(length);
if (bytes == nullptr) {
errorCode = U_MEMORY_ALLOCATION_ERROR;
clear();
return nullptr;
}
UCPTrie *trie = reinterpret_cast<UCPTrie *>(bytes);
uprv_memset(trie, 0, sizeof(UCPTrie));
trie->indexLength = indexLength;
trie->dataLength = dataLength;
trie->highStart = highStart;
// Round up shifted12HighStart to a multiple of 0x1000 for easy testing from UTF-8 lead bytes.
// Runtime code needs to then test for the real highStart as well.
trie->shifted12HighStart = (highStart + 0xfff) >> 12;
trie->type = type;
trie->valueWidth = valueWidth;
trie->index3NullOffset = index3NullOffset;
trie->dataNullOffset = dataNullOffset;
trie->nullValue = initialValue;
bytes += sizeof(UCPTrie);
// Fill the index and data arrays.
uint16_t *dest16 = (uint16_t *)bytes;
trie->index = dest16;
if (highStart <= fastLimit) {
// Condense only the fast index from the mutable-trie index.
for (int32_t i = 0, j = 0; j < indexLength; i += SMALL_DATA_BLOCKS_PER_BMP_BLOCK, ++j) {
*dest16++ = (uint16_t)index[i]; // dest16[j]
}
} else {
uprv_memcpy(dest16, index16, indexLength * 2);
dest16 += indexLength;
}
bytes += indexLength * 2;
// Write the data array.
const uint32_t *p = data;
switch (valueWidth) {
case UCPTRIE_VALUE_BITS_16:
// Write 16-bit data values.
trie->data.ptr16 = dest16;
for (int32_t i = dataLength; i > 0; --i) {
*dest16++ = (uint16_t)*p++;
}
break;
case UCPTRIE_VALUE_BITS_32:
// Write 32-bit data values.
trie->data.ptr32 = (uint32_t *)bytes;
uprv_memcpy(bytes, p, (size_t)dataLength * 4);
break;
case UCPTRIE_VALUE_BITS_8:
// Write 8-bit data values.
trie->data.ptr8 = bytes;
for (int32_t i = dataLength; i > 0; --i) {
*bytes++ = (uint8_t)*p++;
}
break;
default:
// Will not occur, valueWidth checked at the beginning.
break;
}
#ifdef UCPTRIE_DEBUG
trie->name = name;
ucptrie_printLengths(trie, "");
#endif
clear();
return trie;
}
} // namespace
U_NAMESPACE_END
U_NAMESPACE_USE
U_CAPI UMutableCPTrie * U_EXPORT2
umutablecptrie_open(uint32_t initialValue, uint32_t errorValue, UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
LocalPointer<MutableCodePointTrie> trie(
new MutableCodePointTrie(initialValue, errorValue, *pErrorCode), *pErrorCode);
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
return reinterpret_cast<UMutableCPTrie *>(trie.orphan());
}
U_CAPI UMutableCPTrie * U_EXPORT2
umutablecptrie_clone(const UMutableCPTrie *other, UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
if (other == nullptr) {
return nullptr;
}
LocalPointer<MutableCodePointTrie> clone(
new MutableCodePointTrie(*reinterpret_cast<const MutableCodePointTrie *>(other), *pErrorCode), *pErrorCode);
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
return reinterpret_cast<UMutableCPTrie *>(clone.orphan());
}
U_CAPI void U_EXPORT2
umutablecptrie_close(UMutableCPTrie *trie) {
delete reinterpret_cast<MutableCodePointTrie *>(trie);
}
U_CAPI UMutableCPTrie * U_EXPORT2
umutablecptrie_fromUCPMap(const UCPMap *map, UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
if (map == nullptr) {
*pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
return nullptr;
}
return reinterpret_cast<UMutableCPTrie *>(MutableCodePointTrie::fromUCPMap(map, *pErrorCode));
}
U_CAPI UMutableCPTrie * U_EXPORT2
umutablecptrie_fromUCPTrie(const UCPTrie *trie, UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
if (trie == nullptr) {
*pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
return nullptr;
}
return reinterpret_cast<UMutableCPTrie *>(MutableCodePointTrie::fromUCPTrie(trie, *pErrorCode));
}
U_CAPI uint32_t U_EXPORT2
umutablecptrie_get(const UMutableCPTrie *trie, UChar32 c) {
return reinterpret_cast<const MutableCodePointTrie *>(trie)->get(c);
}
namespace {
UChar32 getRange(const void *trie, UChar32 start,
UCPMapValueFilter *filter, const void *context, uint32_t *pValue) {
return reinterpret_cast<const MutableCodePointTrie *>(trie)->
getRange(start, filter, context, pValue);
}
} // namespace
U_CAPI UChar32 U_EXPORT2
umutablecptrie_getRange(const UMutableCPTrie *trie, UChar32 start,
UCPMapRangeOption option, uint32_t surrogateValue,
UCPMapValueFilter *filter, const void *context, uint32_t *pValue) {
return ucptrie_internalGetRange(getRange, trie, start,
option, surrogateValue,
filter, context, pValue);
}
U_CAPI void U_EXPORT2
umutablecptrie_set(UMutableCPTrie *trie, UChar32 c, uint32_t value, UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return;
}
reinterpret_cast<MutableCodePointTrie *>(trie)->set(c, value, *pErrorCode);
}
U_CAPI void U_EXPORT2
umutablecptrie_setRange(UMutableCPTrie *trie, UChar32 start, UChar32 end,
uint32_t value, UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return;
}
reinterpret_cast<MutableCodePointTrie *>(trie)->setRange(start, end, value, *pErrorCode);
}
/* Compact and internally serialize the trie. */
U_CAPI UCPTrie * U_EXPORT2
umutablecptrie_buildImmutable(UMutableCPTrie *trie, UCPTrieType type, UCPTrieValueWidth valueWidth,
UErrorCode *pErrorCode) {
if (U_FAILURE(*pErrorCode)) {
return nullptr;
}
return reinterpret_cast<MutableCodePointTrie *>(trie)->build(type, valueWidth, *pErrorCode);
}
#ifdef UCPTRIE_DEBUG
U_CFUNC void umutablecptrie_setName(UMutableCPTrie *trie, const char *name) {
reinterpret_cast<MutableCodePointTrie *>(trie)->name = name;
}
#endif
Reference in New Issue View Git Blame Copy Permalink