wine/gecko
0
0
Fork 0
mirror of https://gitlab.winehq.org/wine/wine-gecko.git synced 2024-09-13 09:24:08 -07:00
Code Issues Packages Projects Releases Wiki Activity

Not Really Part Of The Build. Update nsLineBreaker interface to allow aSink to be null, in which case the computed linebreaks are just discarded. We use this to optimize cases where no breaking is enabled and the results of linebreaking are not required. Also exposes IsSpace and IsComplexChar methods that the linebreaker specification relies on. Renamed CJK stuff to ComplexChar because it's going to include Thai and probably other languages.

Browse Source
This commit is contained in:
roc+@cs.cmu.edu 2007-06-06 21:17:55 -07:00
parent 52b25218e7
commit c71327187f
2 changed files with 76 additions and 52 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
content/base/public/nsLineBreaker.h
View File

@ -80,10 +80,25 @@ class nsLineBreaker {
public:
nsLineBreaker();
~nsLineBreaker();
static inline PRBool IsSpace(PRUnichar u)
{
return u == 0x0020 || u == 0x200b/*ZWSP*/ || u == '\n' || u == '\t';
}
static inline PRBool IsComplexChar(PRUnichar u)
{
return (0x1100 <= u && u <= 0x11ff) ||
(0x2e80 <= u && u <= 0xd7ff) ||
(0xf900 <= u && u <= 0xfaff) ||
(0xff00 <= u && u <= 0xffef);
}
// Normally, break opportunities exist at the end of each run of whitespace
// (Unicode ZWSP (U+200B) and ASCII space (U+0020)). Break opportunities can
// also exist inside runs of non-whitespace, as determined by nsILineBreaker.
// (see IsSpace above). Break opportunities can also exist inside runs of
// non-whitespace, as determined by nsILineBreaker. We pass a whitespace-
// delimited word to nsILineBreaker if it contains at least one character
// matching IsComplexChar.
// We provide flags to control on a per-chunk basis where breaks are allowed.
// At any character boundary, exactly one text chunk governs whether a
// break is allowed at that boundary.
@ -111,11 +126,15 @@ public:
/**
* Feed Unicode text into the linebreaker for analysis. aLength must be
* nonzero.
* @param aSink can be null if the breaks are not actually needed (we may
* still be setting up state for later breaks)
*/
nsresult AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32 aLength,
PRUint32 aFlags, nsILineBreakSink* aSink);
/**
* Feed 8-bit text into the linebreaker for analysis. aLength must be nonzero.
* @param aSink can be null if the breaks are not actually needed (we may
* still be setting up state for later breaks)
*/
nsresult AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aLength,
PRUint32 aFlags, nsILineBreakSink* aSink);
@ -155,7 +174,7 @@ private:
nsAutoTArray<PRUnichar,100> mCurrentWord;
// All the items that contribute to mCurrentWord
nsAutoTArray<TextItem,2> mTextItems;
PRPackedBool mCurrentWordContainsCJK;
PRPackedBool mCurrentWordContainsComplexChar;
// True if the previous character was whitespace
PRPackedBool mAfterSpace;

103
content/base/src/nsLineBreaker.cpp
View File

@ -40,31 +40,8 @@
#include "nsContentUtils.h"
#include "nsILineBreaker.h"
#define UNICODE_ZWSP 0x200b
static inline int
IS_SPACE(PRUnichar u)
{
return u == 0x0020 || u == UNICODE_ZWSP;
}
static inline int
IS_SPACE(PRUint8 u)
{
return u == 0x0020;
}
static inline int
IS_CJK_CHAR(PRUnichar u)
{
return (0x1100 <= u && u <= 0x11ff) ||
(0x2e80 <= u && u <= 0xd7ff) ||
(0xf900 <= u && u <= 0xfaff) ||
(0xff00 <= u && u <= 0xffef);
}
nsLineBreaker::nsLineBreaker()
: mCurrentWordContainsCJK(PR_FALSE),
: mCurrentWordContainsComplexChar(PR_FALSE),
mAfterSpace(PR_FALSE)
{
}
@ -81,7 +58,7 @@ nsLineBreaker::FlushCurrentWord()
if (!breakState.AppendElements(mCurrentWord.Length()))
return NS_ERROR_OUT_OF_MEMORY;
if (!mCurrentWordContainsCJK) {
if (!mCurrentWordContainsComplexChar) {
// Just set everything internal to "no break"!
memset(breakState.Elements(), PR_FALSE, mCurrentWord.Length());
} else {
@ -107,14 +84,16 @@ nsLineBreaker::FlushCurrentWord()
// it was already set correctly earlier and we don't know what the true
// value should be.
PRUint32 skipSet = i == 0 ? 1 : 0;
ti->mSink->SetBreaks(ti->mSinkOffset + skipSet, ti->mLength - skipSet,
breakState.Elements() + offset + skipSet);
if (ti->mSink) {
ti->mSink->SetBreaks(ti->mSinkOffset + skipSet, ti->mLength - skipSet,
breakState.Elements() + offset + skipSet);
}
offset += ti->mLength;
}
mCurrentWord.Clear();
mTextItems.Clear();
mCurrentWordContainsCJK = PR_FALSE;
mCurrentWordContainsComplexChar = PR_FALSE;
return NS_OK;
}
@ -130,10 +109,10 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
if (mCurrentWord.Length() > 0) {
NS_ASSERTION(!mAfterSpace, "These should not be set");
while (offset < aLength && !IS_SPACE(aText[offset])) {
while (offset < aLength && !IsSpace(aText[offset])) {
mCurrentWord.AppendElement(aText[offset]);
if (!mCurrentWordContainsCJK && IS_CJK_CHAR(aText[offset])) {
mCurrentWordContainsCJK = PR_TRUE;
if (!mCurrentWordContainsComplexChar && IsComplexChar(aText[offset])) {
mCurrentWordContainsComplexChar = PR_TRUE;
}
++offset;
}
@ -156,19 +135,28 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
return NS_ERROR_OUT_OF_MEMORY;
PRUint32 start = offset;
if (!aSink && !aFlags) {
// Skip to the space before the last word, since we don't need the breaks
offset = aLength;
while (offset > start) {
--offset;
if (IsSpace(aText[offset]))
break;
}
}
PRUint32 wordStart = offset;
PRBool wordHasCJK = PR_FALSE;
PRBool wordHasComplexChar = PR_FALSE;
for (;;) {
PRUnichar ch = aText[offset];
PRBool isSpace = IS_SPACE(ch);
PRBool isSpace = IsSpace(ch);
breakState[offset] = mAfterSpace && !isSpace &&
(aFlags & (offset == 0 ? BREAK_ALLOW_INITIAL : BREAK_ALLOW_INSIDE));
mAfterSpace = isSpace;
if (isSpace) {
if (offset > wordStart && wordHasCJK) {
if (offset > wordStart && wordHasComplexChar) {
if (aFlags & BREAK_ALLOW_INSIDE) {
// Save current start-of-word state because GetJISx4051Breaks will
// set it to false
@ -178,7 +166,7 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
breakState.Elements() + wordStart);
breakState[wordStart] = currentStart;
}
wordHasCJK = PR_FALSE;
wordHasComplexChar = PR_FALSE;
}
++offset;
@ -186,13 +174,13 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
break;
wordStart = offset;
} else {
if (!wordHasCJK && IS_CJK_CHAR(ch)) {
wordHasCJK = PR_TRUE;
if (!wordHasComplexChar && IsComplexChar(ch)) {
wordHasComplexChar = PR_TRUE;
}
++offset;
if (offset >= aLength) {
// Save this word
mCurrentWordContainsCJK = wordHasCJK;
mCurrentWordContainsComplexChar = wordHasComplexChar;
PRUint32 len = offset - wordStart;
PRUnichar* elems = mCurrentWord.AppendElements(len);
if (!elems)
@ -206,7 +194,9 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
}
}
aSink->SetBreaks(start, offset - start, breakState.Elements() + start);
if (aSink) {
aSink->SetBreaks(start, offset - start, breakState.Elements() + start);
}
return NS_OK;
}
@ -222,7 +212,7 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aL
if (mCurrentWord.Length() > 0) {
NS_ASSERTION(!mAfterSpace, "These should not be set");
while (offset < aLength && !IS_SPACE(aText[offset])) {
while (offset < aLength && !IsSpace(aText[offset])) {
mCurrentWord.AppendElement(aText[offset]);
++offset;
}
@ -243,22 +233,35 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aL
}
nsAutoTArray<PRPackedBool,4000> breakState;
if (!breakState.AppendElements(aLength))
return NS_ERROR_OUT_OF_MEMORY;
if (aSink) {
if (!breakState.AppendElements(aLength))
return NS_ERROR_OUT_OF_MEMORY;
}
PRUint32 start = offset;
if (!aSink && !aFlags) {
// Skip to the space before the last word, since we don't need the breaks
offset = aLength;
while (offset > start) {
--offset;
if (IsSpace(aText[offset]))
break;
}
}
PRUint32 wordStart = offset;
for (;;) {
PRUint8 ch = aText[offset];
PRBool isSpace = IS_SPACE(ch);
PRBool isSpace = IsSpace(ch);
breakState[offset] = mAfterSpace && !isSpace &&
(aFlags & (offset == 0 ? BREAK_ALLOW_INITIAL : BREAK_ALLOW_INSIDE));
if (aSink) {
breakState[offset] = mAfterSpace && !isSpace &&
(aFlags & (offset == 0 ? BREAK_ALLOW_INITIAL : BREAK_ALLOW_INSIDE));
}
mAfterSpace = isSpace;
if (isSpace) {
// The current word can't have any special (CJK/Thai) characters inside it
// The current word can't have any complex characters inside it
// because this is 8-bit text, so just ignore it
++offset;
if (offset >= aLength)
@ -268,7 +271,7 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aL
++offset;
if (offset >= aLength) {
// Save this word
mCurrentWordContainsCJK = PR_FALSE;
mCurrentWordContainsComplexChar = PR_FALSE;
PRUint32 len = offset - wordStart;
PRUnichar* elems = mCurrentWord.AppendElements(len);
if (!elems)
@ -282,12 +285,14 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aL
offset = wordStart + 1;
break;
}
// We can't break inside words in 8-bit text (no CJK characters), so
// We can't break inside words in 8-bit text (no complex characters), so
// there is no need to do anything else to handle words
}
}
aSink->SetBreaks(start, offset - start, breakState.Elements() + start);
if (aSink) {
aSink->SetBreaks(start, offset - start, breakState.Elements() + start);
}
return NS_OK;
}