AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ICU-1428 ScriptRun - support matching for more paired characters.

Browse Source 
X-SVN-Rev: 7052
This commit is contained in:
Eric Mader 2001-11-21 19:01:01 +00:00
parent cec5db5134
commit b51d954ade
2 changed files with 147 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

147
icu4c/source/samples/layout/scrptrun.cpp
View File

@ -16,6 +16,92 @@
#include "scrptrun.h"
#define ARRAY_SIZE(array) (sizeof array / sizeof array[0])
UChar32 ScriptRun::pairedChars[] = {
0x0028, 0x0029, // ascii paired punctuation
0x003c, 0x003e,
0x005b, 0x005d,
0x007b, 0x007d,
0x00ab, 0x00bb, // guillemets
0x2018, 0x2019, // general punctuation
0x201c, 0x201d,
0x2039, 0x203a,
0x3008, 0x3009, // chinese paired punctuation
0x300a, 0x300b,
0x300c, 0x300d,
0x300e, 0x300f,
0x3010, 0x3011,
0x3014, 0x3015,
0x3016, 0x3017,
0x3018, 0x3019,
0x301a, 0x301b
};
const int32_t ScriptRun::pairedCharCount = ARRAY_SIZE(pairedChars);
const int32_t ScriptRun::pairedCharPower = 1 << highBit(pairedCharCount);
const int32_t ScriptRun::pairedCharExtra = pairedCharCount - pairedCharPower;
int8_t ScriptRun::highBit(int32_t value)
{
if (value <= 0) {
return -32;
}
int8_t bit = 0;
if (value >= 1 << 16) {
value >>= 16;
bit += 16;
}
if (value >= 1 << 8) {
value >>= 8;
bit += 8;
}
if (value >= 1 << 4) {
value >>= 4;
bit += 4;
}
if (value >= 1 << 2) {
value >>= 2;
bit += 2;
}
if (value >= 1 << 1) {
value >>= 1;
bit += 1;
}
return bit;
}
int32_t ScriptRun::getPairIndex(UChar32 ch)
{
int32_t probe = pairedCharPower;
int32_t index = 0;
if (ch >= pairedChars[pairedCharExtra]) {
index = pairedCharExtra;
}
while (probe > (1 << 0)) {
probe >>= 1;
if (ch >= pairedChars[index + probe]) {
index += probe;
}
}
if (pairedChars[index] != ch) {
index = -1;
}
return index;
}
UBool ScriptRun::sameScript(int32_t scriptOne, int32_t scriptTwo)
{
return scriptOne <= USCRIPT_INHERITED || scriptTwo <= USCRIPT_INHERITED || scriptOne == scriptTwo;
@ -23,8 +109,10 @@ UBool ScriptRun::sameScript(int32_t scriptOne, int32_t scriptTwo)
UBool ScriptRun::next()
{
int32_t startSP = parenSP; // used to find the first new open character
UErrorCode error = U_ZERO_ERROR;
// if we've fallen off the end of the text, we're done
if (scriptEnd >= charLimit) {
return false;
}
@ -35,10 +123,14 @@ UBool ScriptRun::next()
UChar high = charArray[scriptEnd];
UChar32 ch = high;
if (scriptEnd < charLimit - 1 && high >= 0xD800 && high <= 0xDBFF)
// if the character is a high surrogate and it's not the last one
// in the text, see if it's followed by a low surrogate
if (high >= 0xD800 && high <= 0xDBFF && scriptEnd < charLimit - 1)
{
UChar low = charArray[scriptEnd + 1];
// if it is followed by a low surrogate,
// consume it and form the full character
if (low >= 0xDC00 && low <= 0xDFFF) {
ch = (high - 0xD800) * 0x0400 + low - 0xDC00 + 0x10000;
scriptEnd += 1;
@ -46,34 +138,59 @@ UBool ScriptRun::next()
}
UScriptCode sc = uscript_getScript(ch, &error);
int32_t pairIndex = getPairIndex(ch);
if (ch == ')' && parenSP >= 0) {
sc = parenStack[parenSP];
// Paired character handling:
//
// if it's an open character, push it onto the stack.
// if it's a close character, find the matching open on the
// stack, and use that script code. Any non-matching open
// characters above it on the stack will be poped.
if (pairIndex >= 0) {
if ((pairIndex & 1) == 0) {
parenStack[++parenSP].pairIndex = pairIndex;
parenStack[parenSP].scriptCode = scriptCode;
} else if (parenSP >= 0) {
int32_t pi = pairIndex & ~1;
while (parenSP >= 0 && parenStack[parenSP].pairIndex != pi) {
parenSP -= 1;
}
if (parenSP < startSP) {
startSP = parenSP;
}
if (parenSP >= 0) {
sc = parenStack[parenSP].scriptCode;
}
}
}
if (sameScript(scriptCode, sc)) {
if (scriptCode <= USCRIPT_INHERITED && sc > USCRIPT_INHERITED) {
scriptCode = sc;
// now that we have a final script code, fix any open
// characters we pushed before we knew the script code.
while (startSP < parenSP) {
parenStack[++startSP].scriptCode = scriptCode;
}
}
if (ch == ')' && parenSP >= 0) {
parenSP -= 1;
}
// if this character is a close paired character,
// pop it from the stack
if (pairIndex >= 0 && (pairIndex & 1) != 0 && parenSP >= 0) {
parenSP -= 1;
startSP -= 1;
}
} else {
int32_t i;
// if the run broke on a surrogate pair,
// end it before the high surrogate
if (ch >= 0x10000) {
scriptEnd -= 1;
}
for (i = scriptStart; i < scriptEnd; i += 1) {
if (charArray[i] == '(') {
parenStack[++parenSP] = scriptCode;
}
}
break;
}
}

16
icu4c/source/samples/layout/scrptrun.h
View File

@ -24,6 +24,12 @@ struct ScriptRecord
UScriptCode scriptCode;
};
struct ParenStackEntry
{
int32_t pairIndex;
UScriptCode scriptCode;
};
class ScriptRun
{
public:
@ -59,8 +65,16 @@ private:
int32_t scriptEnd;
UScriptCode scriptCode;
UScriptCode parenStack[128];
ParenStackEntry parenStack[128];
int32_t parenSP;
static int8_t highBit(int32_t value);
static int32_t getPairIndex(UChar32 ch);
static UChar32 pairedChars[];
static const int32_t pairedCharCount;
static const int32_t pairedCharPower;
static const int32_t pairedCharExtra;
};
inline ScriptRun::ScriptRun()