1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/usc_impl.c Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,361 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (C) 1999-2009, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +**********************************************************************
1.9 +*
1.10 +* File USC_IMPL.C
1.11 +*
1.12 +* Modification History:
1.13 +*
1.14 +* Date Name Description
1.15 +* 07/08/2002 Eric Mader Creation.
1.16 +******************************************************************************
1.17 +*/
1.18 +
1.19 +#include "unicode/uscript.h"
1.20 +#include "usc_impl.h"
1.21 +#include "cmemory.h"
1.22 +
1.23 +#define ARRAY_SIZE(array) (sizeof array / sizeof array[0])
1.24 +
1.25 +#define PAREN_STACK_DEPTH 32
1.26 +
1.27 +#define MOD(sp) ((sp) % PAREN_STACK_DEPTH)
1.28 +#define LIMIT_INC(sp) (((sp) < PAREN_STACK_DEPTH)? (sp) + 1 : PAREN_STACK_DEPTH)
1.29 +#define INC(sp,count) (MOD((sp) + (count)))
1.30 +#define INC1(sp) (INC(sp, 1))
1.31 +#define DEC(sp,count) (MOD((sp) + PAREN_STACK_DEPTH - (count)))
1.32 +#define DEC1(sp) (DEC(sp, 1))
1.33 +#define STACK_IS_EMPTY(scriptRun) ((scriptRun)->pushCount <= 0)
1.34 +#define STACK_IS_NOT_EMPTY(scriptRun) (! STACK_IS_EMPTY(scriptRun))
1.35 +#define TOP(scriptRun) ((scriptRun)->parenStack[(scriptRun)->parenSP])
1.36 +#define SYNC_FIXUP(scriptRun) ((scriptRun)->fixupCount = 0)
1.37 +
1.38 +struct ParenStackEntry
1.39 +{
1.40 + int32_t pairIndex;
1.41 + UScriptCode scriptCode;
1.42 +};
1.43 +
1.44 +struct UScriptRun
1.45 +{
1.46 + int32_t textLength;
1.47 + const UChar *textArray;
1.48 +
1.49 + int32_t scriptStart;
1.50 + int32_t scriptLimit;
1.51 + UScriptCode scriptCode;
1.52 +
1.53 + struct ParenStackEntry parenStack[PAREN_STACK_DEPTH];
1.54 + int32_t parenSP;
1.55 + int32_t pushCount;
1.56 + int32_t fixupCount;
1.57 +};
1.58 +
1.59 +static int8_t highBit(int32_t value);
1.60 +
1.61 +static const UChar32 pairedChars[] = {
1.62 + 0x0028, 0x0029, /* ascii paired punctuation */
1.63 + 0x003c, 0x003e,
1.64 + 0x005b, 0x005d,
1.65 + 0x007b, 0x007d,
1.66 + 0x00ab, 0x00bb, /* guillemets */
1.67 + 0x2018, 0x2019, /* general punctuation */
1.68 + 0x201c, 0x201d,
1.69 + 0x2039, 0x203a,
1.70 + 0x3008, 0x3009, /* chinese paired punctuation */
1.71 + 0x300a, 0x300b,
1.72 + 0x300c, 0x300d,
1.73 + 0x300e, 0x300f,
1.74 + 0x3010, 0x3011,
1.75 + 0x3014, 0x3015,
1.76 + 0x3016, 0x3017,
1.77 + 0x3018, 0x3019,
1.78 + 0x301a, 0x301b
1.79 +};
1.80 +
1.81 +static void push(UScriptRun *scriptRun, int32_t pairIndex, UScriptCode scriptCode)
1.82 +{
1.83 + scriptRun->pushCount = LIMIT_INC(scriptRun->pushCount);
1.84 + scriptRun->fixupCount = LIMIT_INC(scriptRun->fixupCount);
1.85 +
1.86 + scriptRun->parenSP = INC1(scriptRun->parenSP);
1.87 + scriptRun->parenStack[scriptRun->parenSP].pairIndex = pairIndex;
1.88 + scriptRun->parenStack[scriptRun->parenSP].scriptCode = scriptCode;
1.89 +}
1.90 +
1.91 +static void pop(UScriptRun *scriptRun)
1.92 +{
1.93 + if (STACK_IS_EMPTY(scriptRun)) {
1.94 + return;
1.95 + }
1.96 +
1.97 + if (scriptRun->fixupCount > 0) {
1.98 + scriptRun->fixupCount -= 1;
1.99 + }
1.100 +
1.101 + scriptRun->pushCount -= 1;
1.102 + scriptRun->parenSP = DEC1(scriptRun->parenSP);
1.103 +
1.104 + /* If the stack is now empty, reset the stack
1.105 + pointers to their initial values.
1.106 + */
1.107 + if (STACK_IS_EMPTY(scriptRun)) {
1.108 + scriptRun->parenSP = -1;
1.109 + }
1.110 +}
1.111 +
1.112 +static void fixup(UScriptRun *scriptRun, UScriptCode scriptCode)
1.113 +{
1.114 + int32_t fixupSP = DEC(scriptRun->parenSP, scriptRun->fixupCount);
1.115 +
1.116 + while (scriptRun->fixupCount-- > 0) {
1.117 + fixupSP = INC1(fixupSP);
1.118 + scriptRun->parenStack[fixupSP].scriptCode = scriptCode;
1.119 + }
1.120 +}
1.121 +
1.122 +static int8_t
1.123 +highBit(int32_t value)
1.124 +{
1.125 + int8_t bit = 0;
1.126 +
1.127 + if (value <= 0) {
1.128 + return -32;
1.129 + }
1.130 +
1.131 + if (value >= 1 << 16) {
1.132 + value >>= 16;
1.133 + bit += 16;
1.134 + }
1.135 +
1.136 + if (value >= 1 << 8) {
1.137 + value >>= 8;
1.138 + bit += 8;
1.139 + }
1.140 +
1.141 + if (value >= 1 << 4) {
1.142 + value >>= 4;
1.143 + bit += 4;
1.144 + }
1.145 +
1.146 + if (value >= 1 << 2) {
1.147 + value >>= 2;
1.148 + bit += 2;
1.149 + }
1.150 +
1.151 + if (value >= 1 << 1) {
1.152 + value >>= 1;
1.153 + bit += 1;
1.154 + }
1.155 +
1.156 + return bit;
1.157 +}
1.158 +
1.159 +static int32_t
1.160 +getPairIndex(UChar32 ch)
1.161 +{
1.162 + int32_t pairedCharCount = ARRAY_SIZE(pairedChars);
1.163 + int32_t pairedCharPower = 1 << highBit(pairedCharCount);
1.164 + int32_t pairedCharExtra = pairedCharCount - pairedCharPower;
1.165 +
1.166 + int32_t probe = pairedCharPower;
1.167 + int32_t pairIndex = 0;
1.168 +
1.169 + if (ch >= pairedChars[pairedCharExtra]) {
1.170 + pairIndex = pairedCharExtra;
1.171 + }
1.172 +
1.173 + while (probe > (1 << 0)) {
1.174 + probe >>= 1;
1.175 +
1.176 + if (ch >= pairedChars[pairIndex + probe]) {
1.177 + pairIndex += probe;
1.178 + }
1.179 + }
1.180 +
1.181 + if (pairedChars[pairIndex] != ch) {
1.182 + pairIndex = -1;
1.183 + }
1.184 +
1.185 + return pairIndex;
1.186 +}
1.187 +
1.188 +static UBool
1.189 +sameScript(UScriptCode scriptOne, UScriptCode scriptTwo)
1.190 +{
1.191 + return scriptOne <= USCRIPT_INHERITED || scriptTwo <= USCRIPT_INHERITED || scriptOne == scriptTwo;
1.192 +}
1.193 +
1.194 +U_CAPI UScriptRun * U_EXPORT2
1.195 +uscript_openRun(const UChar *src, int32_t length, UErrorCode *pErrorCode)
1.196 +{
1.197 + UScriptRun *result = NULL;
1.198 +
1.199 + if (pErrorCode == NULL || U_FAILURE(*pErrorCode)) {
1.200 + return NULL;
1.201 + }
1.202 +
1.203 + result = uprv_malloc(sizeof (UScriptRun));
1.204 +
1.205 + if (result == NULL) {
1.206 + *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1.207 + return NULL;
1.208 + }
1.209 +
1.210 + uscript_setRunText(result, src, length, pErrorCode);
1.211 +
1.212 + /* Release the UScriptRun if uscript_setRunText() returns an error */
1.213 + if (U_FAILURE(*pErrorCode)) {
1.214 + uprv_free(result);
1.215 + result = NULL;
1.216 + }
1.217 +
1.218 + return result;
1.219 +}
1.220 +
1.221 +U_CAPI void U_EXPORT2
1.222 +uscript_closeRun(UScriptRun *scriptRun)
1.223 +{
1.224 + if (scriptRun != NULL) {
1.225 + uprv_free(scriptRun);
1.226 + }
1.227 +}
1.228 +
1.229 +U_CAPI void U_EXPORT2
1.230 +uscript_resetRun(UScriptRun *scriptRun)
1.231 +{
1.232 + if (scriptRun != NULL) {
1.233 + scriptRun->scriptStart = 0;
1.234 + scriptRun->scriptLimit = 0;
1.235 + scriptRun->scriptCode = USCRIPT_INVALID_CODE;
1.236 + scriptRun->parenSP = -1;
1.237 + scriptRun->pushCount = 0;
1.238 + scriptRun->fixupCount = 0;
1.239 + }
1.240 +}
1.241 +
1.242 +U_CAPI void U_EXPORT2
1.243 +uscript_setRunText(UScriptRun *scriptRun, const UChar *src, int32_t length, UErrorCode *pErrorCode)
1.244 +{
1.245 + if (pErrorCode == NULL || U_FAILURE(*pErrorCode)) {
1.246 + return;
1.247 + }
1.248 +
1.249 + if (scriptRun == NULL || length < 0 || ((src == NULL) != (length == 0))) {
1.250 + *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
1.251 + return;
1.252 + }
1.253 +
1.254 + scriptRun->textArray = src;
1.255 + scriptRun->textLength = length;
1.256 +
1.257 + uscript_resetRun(scriptRun);
1.258 +}
1.259 +
1.260 +U_CAPI UBool U_EXPORT2
1.261 +uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript)
1.262 +{
1.263 + UErrorCode error = U_ZERO_ERROR;
1.264 +
1.265 + /* if we've fallen off the end of the text, we're done */
1.266 + if (scriptRun == NULL || scriptRun->scriptLimit >= scriptRun->textLength) {
1.267 + return FALSE;
1.268 + }
1.269 +
1.270 + SYNC_FIXUP(scriptRun);
1.271 + scriptRun->scriptCode = USCRIPT_COMMON;
1.272 +
1.273 + for (scriptRun->scriptStart = scriptRun->scriptLimit; scriptRun->scriptLimit < scriptRun->textLength; scriptRun->scriptLimit += 1) {
1.274 + UChar high = scriptRun->textArray[scriptRun->scriptLimit];
1.275 + UChar32 ch = high;
1.276 + UScriptCode sc;
1.277 + int32_t pairIndex;
1.278 +
1.279 + /*
1.280 + * if the character is a high surrogate and it's not the last one
1.281 + * in the text, see if it's followed by a low surrogate
1.282 + */
1.283 + if (high >= 0xD800 && high <= 0xDBFF && scriptRun->scriptLimit < scriptRun->textLength - 1) {
1.284 + UChar low = scriptRun->textArray[scriptRun->scriptLimit + 1];
1.285 +
1.286 + /*
1.287 + * if it is followed by a low surrogate,
1.288 + * consume it and form the full character
1.289 + */
1.290 + if (low >= 0xDC00 && low <= 0xDFFF) {
1.291 + ch = (high - 0xD800) * 0x0400 + low - 0xDC00 + 0x10000;
1.292 + scriptRun->scriptLimit += 1;
1.293 + }
1.294 + }
1.295 +
1.296 + sc = uscript_getScript(ch, &error);
1.297 + pairIndex = getPairIndex(ch);
1.298 +
1.299 + /*
1.300 + * Paired character handling:
1.301 + *
1.302 + * if it's an open character, push it onto the stack.
1.303 + * if it's a close character, find the matching open on the
1.304 + * stack, and use that script code. Any non-matching open
1.305 + * characters above it on the stack will be poped.
1.306 + */
1.307 + if (pairIndex >= 0) {
1.308 + if ((pairIndex & 1) == 0) {
1.309 + push(scriptRun, pairIndex, scriptRun->scriptCode);
1.310 + } else {
1.311 + int32_t pi = pairIndex & ~1;
1.312 +
1.313 + while (STACK_IS_NOT_EMPTY(scriptRun) && TOP(scriptRun).pairIndex != pi) {
1.314 + pop(scriptRun);
1.315 + }
1.316 +
1.317 + if (STACK_IS_NOT_EMPTY(scriptRun)) {
1.318 + sc = TOP(scriptRun).scriptCode;
1.319 + }
1.320 + }
1.321 + }
1.322 +
1.323 + if (sameScript(scriptRun->scriptCode, sc)) {
1.324 + if (scriptRun->scriptCode <= USCRIPT_INHERITED && sc > USCRIPT_INHERITED) {
1.325 + scriptRun->scriptCode = sc;
1.326 +
1.327 + fixup(scriptRun, scriptRun->scriptCode);
1.328 + }
1.329 +
1.330 + /*
1.331 + * if this character is a close paired character,
1.332 + * pop the matching open character from the stack
1.333 + */
1.334 + if (pairIndex >= 0 && (pairIndex & 1) != 0) {
1.335 + pop(scriptRun);
1.336 + }
1.337 + } else {
1.338 + /*
1.339 + * if the run broke on a surrogate pair,
1.340 + * end it before the high surrogate
1.341 + */
1.342 + if (ch >= 0x10000) {
1.343 + scriptRun->scriptLimit -= 1;
1.344 + }
1.345 +
1.346 + break;
1.347 + }
1.348 + }
1.349 +
1.350 +
1.351 + if (pRunStart != NULL) {
1.352 + *pRunStart = scriptRun->scriptStart;
1.353 + }
1.354 +
1.355 + if (pRunLimit != NULL) {
1.356 + *pRunLimit = scriptRun->scriptLimit;
1.357 + }
1.358 +
1.359 + if (pRunScript != NULL) {
1.360 + *pRunScript = scriptRun->scriptCode;
1.361 + }
1.362 +
1.363 + return TRUE;
1.364 +}
