michael@0: /*
michael@0:  *******************************************************************************
michael@0:  *
michael@0:  *   Copyright (C) 1999-2001, International Business Machines
michael@0:  *   Corporation and others.  All Rights Reserved.
michael@0:  *
michael@0:  *******************************************************************************
michael@0:  *   file name:  scrptrun.cpp
michael@0:  *
michael@0:  *   created on: 10/17/2001
michael@0:  *   created by: Eric R. Mader
michael@0:  */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uscript.h"
michael@0: 
michael@0: #include "scrptrun.h"
michael@0: 
michael@0: #define ARRAY_SIZE(array) (sizeof array  / sizeof array[0])
michael@0: 
michael@0: const char ScriptRun::fgClassID=0;
michael@0: 
michael@0: UChar32 ScriptRun::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: const int32_t ScriptRun::pairedCharCount = ARRAY_SIZE(pairedChars);
michael@0: const int32_t ScriptRun::pairedCharPower = 1 << highBit(pairedCharCount);
michael@0: const int32_t ScriptRun::pairedCharExtra = pairedCharCount - pairedCharPower;
michael@0: 
michael@0: int8_t ScriptRun::highBit(int32_t value)
michael@0: {
michael@0:     if (value <= 0) {
michael@0:         return -32;
michael@0:     }
michael@0: 
michael@0:     int8_t bit = 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: int32_t ScriptRun::getPairIndex(UChar32 ch)
michael@0: {
michael@0:     int32_t probe = pairedCharPower;
michael@0:     int32_t index = 0;
michael@0: 
michael@0:     if (ch >= pairedChars[pairedCharExtra]) {
michael@0:         index = pairedCharExtra;
michael@0:     }
michael@0: 
michael@0:     while (probe > (1 << 0)) {
michael@0:         probe >>= 1;
michael@0: 
michael@0:         if (ch >= pairedChars[index + probe]) {
michael@0:             index += probe;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     if (pairedChars[index] != ch) {
michael@0:         index = -1;
michael@0:     }
michael@0: 
michael@0:     return index;
michael@0: }
michael@0: 
michael@0: UBool ScriptRun::sameScript(int32_t scriptOne, int32_t scriptTwo)
michael@0: {
michael@0:     return scriptOne <= USCRIPT_INHERITED || scriptTwo <= USCRIPT_INHERITED || scriptOne == scriptTwo;
michael@0: }
michael@0: 
michael@0: UBool ScriptRun::next()
michael@0: {
michael@0:     int32_t startSP  = parenSP;  // used to find the first new open character
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 (scriptEnd >= charLimit) {
michael@0:         return false;
michael@0:     }
michael@0:     
michael@0:     scriptCode = USCRIPT_COMMON;
michael@0: 
michael@0:     for (scriptStart = scriptEnd; scriptEnd < charLimit; scriptEnd += 1) {
michael@0:         UChar   high = charArray[scriptEnd];
michael@0:         UChar32 ch   = high;
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:         if (high >= 0xD800 && high <= 0xDBFF && scriptEnd < charLimit - 1)
michael@0:         {
michael@0:             UChar low = charArray[scriptEnd + 1];
michael@0: 
michael@0:             // if it is followed by a low surrogate,
michael@0:             // consume it and form the full character
michael@0:             if (low >= 0xDC00 && low <= 0xDFFF) {
michael@0:                 ch = (high - 0xD800) * 0x0400 + low - 0xDC00 + 0x10000;
michael@0:                 scriptEnd += 1;
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         UScriptCode sc = uscript_getScript(ch, &error);
michael@0:         int32_t pairIndex = getPairIndex(ch);
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:         if (pairIndex >= 0) {
michael@0:             if ((pairIndex & 1) == 0) {
michael@0:                 parenStack[++parenSP].pairIndex = pairIndex;
michael@0:                 parenStack[parenSP].scriptCode  = scriptCode;
michael@0:             } else if (parenSP >= 0) {
michael@0:                 int32_t pi = pairIndex & ~1;
michael@0: 
michael@0:                 while (parenSP >= 0 && parenStack[parenSP].pairIndex != pi) {
michael@0:                     parenSP -= 1;
michael@0:                 }
michael@0: 
michael@0:                 if (parenSP < startSP) {
michael@0:                     startSP = parenSP;
michael@0:                 }
michael@0: 
michael@0:                 if (parenSP >= 0) {
michael@0:                     sc = parenStack[parenSP].scriptCode;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         if (sameScript(scriptCode, sc)) {
michael@0:             if (scriptCode <= USCRIPT_INHERITED && sc > USCRIPT_INHERITED) {
michael@0:                 scriptCode = sc;
michael@0: 
michael@0:                 // now that we have a final script code, fix any open
michael@0:                 // characters we pushed before we knew the script code.
michael@0:                 while (startSP < parenSP) {
michael@0:                     parenStack[++startSP].scriptCode = scriptCode;
michael@0:                 }
michael@0:             }
michael@0: 
michael@0:             // if this character is a close paired character,
michael@0:             // pop it from the stack
michael@0:             if (pairIndex >= 0 && (pairIndex & 1) != 0 && parenSP >= 0) {
michael@0:                 parenSP -= 1;
michael@0:                 startSP -= 1;
michael@0:             }
michael@0:         } else {
michael@0:             // if the run broke on a surrogate pair,
michael@0:             // end it before the high surrogate
michael@0:             if (ch >= 0x10000) {
michael@0:                 scriptEnd -= 1;
michael@0:             }
michael@0: 
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     return true;
michael@0: }
michael@0: