bool
gfxScriptItemizer::Next(PRUint32& aRunStart, PRUint32& aRunLimit,
PRInt32& aRunScript)
{
/* if we've fallen off the end of the text, we're done */
if (scriptLimit >= textLength) {
return PR_FALSE;
}
SYNC_FIXUP();
scriptCode = HB_SCRIPT_COMMON;
for (scriptStart = scriptLimit; scriptLimit < textLength; scriptLimit += 1) {
PRUint32 ch;
PRInt32 sc;
PRInt32 pairIndex;
PRUint32 startOfChar = scriptLimit;
ch = textPtr[scriptLimit];
/*
* MODIFICATION for Gecko - clear the paired-character stack
* when we see a space character, because we cannot trust
* context outside the current "word" when doing textrun
* construction
*/
if (ch == 0x20) {
while (STACK_IS_NOT_EMPTY()) {
pop();
}
sc = HB_SCRIPT_COMMON;
pairIndex = -1;
} else {
/* decode UTF-16 (may be surrogate pair) */
if (NS_IS_HIGH_SURROGATE(ch) && scriptLimit < textLength - 1) {
PRUint32 low = textPtr[scriptLimit + 1];
if (NS_IS_LOW_SURROGATE(low)) {
ch = SURROGATE_TO_UCS4(ch, low);
scriptLimit += 1;
}
}
sc = gfxUnicodeProperties::GetScriptCode(ch);
pairIndex = getPairIndex(ch);
/*
* 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) {
push(pairIndex, scriptCode);
} else {
PRInt32 pi = pairIndex & ~1;
while (STACK_IS_NOT_EMPTY() && TOP().pairIndex != pi) {
pop();
}
if (STACK_IS_NOT_EMPTY()) {
sc = TOP().scriptCode;
}
}
}
}
if (sameScript(scriptCode, sc)) {
if (scriptCode <= HB_SCRIPT_INHERITED && sc > HB_SCRIPT_INHERITED) {
scriptCode = sc;
fixup(scriptCode);
}
/*
* if this character is a close paired character,
* pop the matching open character from the stack
*/
if (pairIndex >= 0 && (pairIndex & 1) != 0) {
pop();
}
} else {
/*
* reset scriptLimit in case it was advanced during reading a
* multiple-code-unit character
*/
scriptLimit = startOfChar;
break;
}
}
aRunStart = scriptStart;
aRunLimit = scriptLimit;
aRunScript = scriptCode;
return PR_TRUE;
}
bool
gfxScriptItemizer::Next(uint32_t& aRunStart, uint32_t& aRunLimit,
int32_t& aRunScript)
{
/* if we've fallen off the end of the text, we're done */
if (scriptLimit >= textLength) {
return false;
}
SYNC_FIXUP();
scriptCode = MOZ_SCRIPT_COMMON;
for (scriptStart = scriptLimit; scriptLimit < textLength; scriptLimit += 1) {
uint32_t ch;
int32_t sc;
uint32_t startOfChar = scriptLimit;
ch = textPtr[scriptLimit];
/* decode UTF-16 (may be surrogate pair) */
if (NS_IS_HIGH_SURROGATE(ch) && scriptLimit < textLength - 1) {
uint32_t low = textPtr[scriptLimit + 1];
if (NS_IS_LOW_SURROGATE(low)) {
ch = SURROGATE_TO_UCS4(ch, low);
scriptLimit += 1;
}
}
// Get the nsCharProps2 record for the current character,
// so we can read the script and (if needed) the gen category
// without needing to do two multi-level lookups.
// NOTE that this means we're relying on an implementation detail
// of the nsUnicodeProperties tables, and might have to revise this
// if the nsCharProps records used there are modified in future.
const nsCharProps2& charProps = GetCharProps2(ch);
// Initialize gc to UNASSIGNED; we'll only set it to the true GC
// if the character has script=COMMON, otherwise we don't care.
uint8_t gc = HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED;
sc = charProps.mScriptCode;
if (sc == MOZ_SCRIPT_COMMON) {
/*
* 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 popped.
*
* We only do this if the script is COMMON; for chars with
* specific script assignments, we just use them as-is.
*/
gc = charProps.mCategory;
if (gc == HB_UNICODE_GENERAL_CATEGORY_OPEN_PUNCTUATION) {
uint32_t endPairChar = mozilla::unicode::GetMirroredChar(ch);
if (endPairChar != ch) {
push(endPairChar, scriptCode);
}
} else if (gc == HB_UNICODE_GENERAL_CATEGORY_CLOSE_PUNCTUATION &&
HasMirroredChar(ch))
{
while (STACK_IS_NOT_EMPTY() && TOP().endPairChar != ch) {
pop();
}
if (STACK_IS_NOT_EMPTY()) {
sc = TOP().scriptCode;
}
}
}
if (SameScript(scriptCode, sc)) {
if (scriptCode <= MOZ_SCRIPT_INHERITED &&
sc > MOZ_SCRIPT_INHERITED)
{
scriptCode = sc;
fixup(scriptCode);
}
/*
* if this character is a close paired character,
* pop the matching open character from the stack
*/
if (gc == HB_UNICODE_GENERAL_CATEGORY_CLOSE_PUNCTUATION &&
HasMirroredChar(ch)) {
pop();
}
} else {
/*
* reset scriptLimit in case it was advanced during reading a
* multiple-code-unit character
*/
scriptLimit = startOfChar;
break;
}
}
aRunStart = scriptStart;
aRunLimit = scriptLimit;
aRunScript = scriptCode;
return true;
}
