michael@0: /*
michael@0: **********************************************************************
michael@0: *   Copyright (C) 1999-2009, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *
michael@0: * File USC_IMPL.C
michael@0: *
michael@0: * Modification History:
michael@0: *
michael@0: *   Date        Name        Description
michael@0: *   07/08/2002  Eric Mader  Creation.
michael@0: ******************************************************************************
michael@0: */
michael@0: 
michael@0: #include "unicode/uscript.h"
michael@0: #include "usc_impl.h"
michael@0: #include "cmemory.h"
michael@0: 
michael@0: #define ARRAY_SIZE(array) (sizeof array  / sizeof array[0])
michael@0: 
michael@0: #define PAREN_STACK_DEPTH 32
michael@0: 
michael@0: #define MOD(sp) ((sp) % PAREN_STACK_DEPTH)
michael@0: #define LIMIT_INC(sp) (((sp) < PAREN_STACK_DEPTH)? (sp) + 1 : PAREN_STACK_DEPTH)
michael@0: #define INC(sp,count) (MOD((sp) + (count)))
michael@0: #define INC1(sp) (INC(sp, 1))
michael@0: #define DEC(sp,count) (MOD((sp) + PAREN_STACK_DEPTH - (count)))
michael@0: #define DEC1(sp) (DEC(sp, 1))
michael@0: #define STACK_IS_EMPTY(scriptRun) ((scriptRun)->pushCount <= 0)
michael@0: #define STACK_IS_NOT_EMPTY(scriptRun) (! STACK_IS_EMPTY(scriptRun))
michael@0: #define TOP(scriptRun) ((scriptRun)->parenStack[(scriptRun)->parenSP])
michael@0: #define SYNC_FIXUP(scriptRun) ((scriptRun)->fixupCount = 0)
michael@0: 
michael@0: struct ParenStackEntry
michael@0: {
michael@0:     int32_t pairIndex;
michael@0:     UScriptCode scriptCode;
michael@0: };
michael@0: 
michael@0: struct UScriptRun
michael@0: {
michael@0:     int32_t textLength;
michael@0:     const UChar *textArray;
michael@0: 
michael@0:     int32_t scriptStart;
michael@0:     int32_t scriptLimit;
michael@0:     UScriptCode scriptCode;
michael@0: 
michael@0:     struct ParenStackEntry parenStack[PAREN_STACK_DEPTH];
michael@0:     int32_t parenSP;
michael@0:     int32_t pushCount;
michael@0:     int32_t fixupCount;
michael@0: };
michael@0: 
michael@0: static int8_t highBit(int32_t value);
michael@0: 
michael@0: static const UChar32 pairedChars[] = {
michael@0:     0x0028, 0x0029, /* ascii paired punctuation */
michael@0:     0x003c, 0x003e,
michael@0:     0x005b, 0x005d,
michael@0:     0x007b, 0x007d,
michael@0:     0x00ab, 0x00bb, /* guillemets */
michael@0:     0x2018, 0x2019, /* general punctuation */
michael@0:     0x201c, 0x201d,
michael@0:     0x2039, 0x203a,
michael@0:     0x3008, 0x3009, /* chinese paired punctuation */
michael@0:     0x300a, 0x300b,
michael@0:     0x300c, 0x300d,
michael@0:     0x300e, 0x300f,
michael@0:     0x3010, 0x3011,
michael@0:     0x3014, 0x3015,
michael@0:     0x3016, 0x3017,
michael@0:     0x3018, 0x3019,
michael@0:     0x301a, 0x301b
michael@0: };
michael@0: 
michael@0: static void push(UScriptRun *scriptRun, int32_t pairIndex, UScriptCode scriptCode)
michael@0: {
michael@0:     scriptRun->pushCount  = LIMIT_INC(scriptRun->pushCount);
michael@0:     scriptRun->fixupCount = LIMIT_INC(scriptRun->fixupCount);
michael@0:     
michael@0:     scriptRun->parenSP = INC1(scriptRun->parenSP);
michael@0:     scriptRun->parenStack[scriptRun->parenSP].pairIndex  = pairIndex;
michael@0:     scriptRun->parenStack[scriptRun->parenSP].scriptCode = scriptCode;
michael@0: }
michael@0: 
michael@0: static void pop(UScriptRun *scriptRun)
michael@0: {
michael@0:     if (STACK_IS_EMPTY(scriptRun)) {
michael@0:         return;
michael@0:     }
michael@0:     
michael@0:     if (scriptRun->fixupCount > 0) {
michael@0:         scriptRun->fixupCount -= 1;
michael@0:     }
michael@0:     
michael@0:     scriptRun->pushCount -= 1;
michael@0:     scriptRun->parenSP = DEC1(scriptRun->parenSP);
michael@0:     
michael@0:     /* If the stack is now empty, reset the stack
michael@0:        pointers to their initial values.
michael@0:      */
michael@0:     if (STACK_IS_EMPTY(scriptRun)) {
michael@0:         scriptRun->parenSP = -1;
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void fixup(UScriptRun *scriptRun, UScriptCode scriptCode)
michael@0: {
michael@0:     int32_t fixupSP = DEC(scriptRun->parenSP, scriptRun->fixupCount);
michael@0:     
michael@0:     while (scriptRun->fixupCount-- > 0) {
michael@0:         fixupSP = INC1(fixupSP);
michael@0:         scriptRun->parenStack[fixupSP].scriptCode = scriptCode;
michael@0:     }
michael@0: }
michael@0: 
michael@0: static int8_t
michael@0: highBit(int32_t value)
michael@0: {
michael@0:     int8_t bit = 0;
michael@0: 
michael@0:     if (value <= 0) {
michael@0:         return -32;
michael@0:     }
michael@0: 
michael@0:     if (value >= 1 << 16) {
michael@0:         value >>= 16;
michael@0:         bit += 16;
michael@0:     }
michael@0: 
michael@0:     if (value >= 1 << 8) {
michael@0:         value >>= 8;
michael@0:         bit += 8;
michael@0:     }
michael@0: 
michael@0:     if (value >= 1 << 4) {
michael@0:         value >>= 4;
michael@0:         bit += 4;
michael@0:     }
michael@0: 
michael@0:     if (value >= 1 << 2) {
michael@0:         value >>= 2;
michael@0:         bit += 2;
michael@0:     }
michael@0: 
michael@0:     if (value >= 1 << 1) {
michael@0:         value >>= 1;
michael@0:         bit += 1;
michael@0:     }
michael@0: 
michael@0:     return bit;
michael@0: }
michael@0: 
michael@0: static int32_t
michael@0: getPairIndex(UChar32 ch)
michael@0: {
michael@0:     int32_t pairedCharCount = ARRAY_SIZE(pairedChars);
michael@0:     int32_t pairedCharPower = 1 << highBit(pairedCharCount);
michael@0:     int32_t pairedCharExtra = pairedCharCount - pairedCharPower;
michael@0: 
michael@0:     int32_t probe = pairedCharPower;
michael@0:     int32_t pairIndex = 0;
michael@0: 
michael@0:     if (ch >= pairedChars[pairedCharExtra]) {
michael@0:         pairIndex = pairedCharExtra;
michael@0:     }
michael@0: 
michael@0:     while (probe > (1 << 0)) {
michael@0:         probe >>= 1;
michael@0: 
michael@0:         if (ch >= pairedChars[pairIndex + probe]) {
michael@0:             pairIndex += probe;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     if (pairedChars[pairIndex] != ch) {
michael@0:         pairIndex = -1;
michael@0:     }
michael@0: 
michael@0:     return pairIndex;
michael@0: }
michael@0: 
michael@0: static UBool
michael@0: sameScript(UScriptCode scriptOne, UScriptCode scriptTwo)
michael@0: {
michael@0:     return scriptOne <= USCRIPT_INHERITED || scriptTwo <= USCRIPT_INHERITED || scriptOne == scriptTwo;
michael@0: }
michael@0: 
michael@0: U_CAPI UScriptRun * U_EXPORT2
michael@0: uscript_openRun(const UChar *src, int32_t length, UErrorCode *pErrorCode)
michael@0: {
michael@0:     UScriptRun *result = NULL;
michael@0: 
michael@0:     if (pErrorCode == NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     result = uprv_malloc(sizeof (UScriptRun));
michael@0: 
michael@0:     if (result == NULL) {
michael@0:         *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     uscript_setRunText(result, src, length, pErrorCode);
michael@0: 
michael@0:     /* Release the UScriptRun if uscript_setRunText() returns an error */
michael@0:     if (U_FAILURE(*pErrorCode)) {
michael@0:         uprv_free(result);
michael@0:         result = NULL;
michael@0:     }
michael@0: 
michael@0:     return result;
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: uscript_closeRun(UScriptRun *scriptRun)
michael@0: {
michael@0:     if (scriptRun != NULL) {
michael@0:         uprv_free(scriptRun);
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: uscript_resetRun(UScriptRun *scriptRun)
michael@0: {
michael@0:     if (scriptRun != NULL) {
michael@0:         scriptRun->scriptStart = 0;
michael@0:         scriptRun->scriptLimit = 0;
michael@0:         scriptRun->scriptCode  = USCRIPT_INVALID_CODE;
michael@0:         scriptRun->parenSP     = -1;
michael@0:         scriptRun->pushCount   =  0;
michael@0:         scriptRun->fixupCount  =  0;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: uscript_setRunText(UScriptRun *scriptRun, const UChar *src, int32_t length, UErrorCode *pErrorCode)
michael@0: {
michael@0:     if (pErrorCode == NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     if (scriptRun == NULL || length < 0 || ((src == NULL) != (length == 0))) {
michael@0:         *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     scriptRun->textArray  = src;
michael@0:     scriptRun->textLength = length;
michael@0: 
michael@0:     uscript_resetRun(scriptRun);
michael@0: }
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript)
michael@0: {
michael@0:     UErrorCode error = U_ZERO_ERROR;
michael@0: 
michael@0:     /* if we've fallen off the end of the text, we're done */
michael@0:     if (scriptRun == NULL || scriptRun->scriptLimit >= scriptRun->textLength) {
michael@0:         return FALSE;
michael@0:     }
michael@0:     
michael@0:     SYNC_FIXUP(scriptRun);
michael@0:     scriptRun->scriptCode = USCRIPT_COMMON;
michael@0: 
michael@0:     for (scriptRun->scriptStart = scriptRun->scriptLimit; scriptRun->scriptLimit < scriptRun->textLength; scriptRun->scriptLimit += 1) {
michael@0:         UChar   high = scriptRun->textArray[scriptRun->scriptLimit];
michael@0:         UChar32 ch   = high;
michael@0:         UScriptCode sc;
michael@0:         int32_t pairIndex;
michael@0: 
michael@0:         /*
michael@0:          * if the character is a high surrogate and it's not the last one
michael@0:          * in the text, see if it's followed by a low surrogate
michael@0:          */
michael@0:         if (high >= 0xD800 && high <= 0xDBFF && scriptRun->scriptLimit < scriptRun->textLength - 1) {
michael@0:             UChar low = scriptRun->textArray[scriptRun->scriptLimit + 1];
michael@0: 
michael@0:             /*
michael@0:              * if it is followed by a low surrogate,
michael@0:              * consume it and form the full character
michael@0:              */
michael@0:             if (low >= 0xDC00 && low <= 0xDFFF) {
michael@0:                 ch = (high - 0xD800) * 0x0400 + low - 0xDC00 + 0x10000;
michael@0:                 scriptRun->scriptLimit += 1;
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         sc = uscript_getScript(ch, &error);
michael@0:         pairIndex = getPairIndex(ch);
michael@0: 
michael@0:         /*
michael@0:          * Paired character handling:
michael@0:          *
michael@0:          * if it's an open character, push it onto the stack.
michael@0:          * if it's a close character, find the matching open on the
michael@0:          * stack, and use that script code. Any non-matching open
michael@0:          * characters above it on the stack will be poped.
michael@0:          */
michael@0:         if (pairIndex >= 0) {
michael@0:             if ((pairIndex & 1) == 0) {
michael@0:                 push(scriptRun, pairIndex, scriptRun->scriptCode);
michael@0:             } else {
michael@0:                 int32_t pi = pairIndex & ~1;
michael@0: 
michael@0:                 while (STACK_IS_NOT_EMPTY(scriptRun) && TOP(scriptRun).pairIndex != pi) {
michael@0:                     pop(scriptRun);
michael@0:                 }
michael@0: 
michael@0:                 if (STACK_IS_NOT_EMPTY(scriptRun)) {
michael@0:                     sc = TOP(scriptRun).scriptCode;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         if (sameScript(scriptRun->scriptCode, sc)) {
michael@0:             if (scriptRun->scriptCode <= USCRIPT_INHERITED && sc > USCRIPT_INHERITED) {
michael@0:                 scriptRun->scriptCode = sc;
michael@0: 
michael@0:                 fixup(scriptRun, scriptRun->scriptCode);
michael@0:             }
michael@0: 
michael@0:             /*
michael@0:              * if this character is a close paired character,
michael@0:              * pop the matching open character from the stack
michael@0:              */
michael@0:             if (pairIndex >= 0 && (pairIndex & 1) != 0) {
michael@0:                 pop(scriptRun);
michael@0:             }
michael@0:         } else {
michael@0:             /*
michael@0:              * if the run broke on a surrogate pair,
michael@0:              * end it before the high surrogate
michael@0:              */
michael@0:             if (ch >= 0x10000) {
michael@0:                 scriptRun->scriptLimit -= 1;
michael@0:             }
michael@0: 
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0: 
michael@0: 
michael@0:     if (pRunStart != NULL) {
michael@0:         *pRunStart = scriptRun->scriptStart;
michael@0:     }
michael@0: 
michael@0:     if (pRunLimit != NULL) {
michael@0:         *pRunLimit = scriptRun->scriptLimit;
michael@0:     }
michael@0: 
michael@0:     if (pRunScript != NULL) {
michael@0:         *pRunScript = scriptRun->scriptCode;
michael@0:     }
michael@0: 
michael@0:     return TRUE;
michael@0: }