aboutsummaryrefslogtreecommitdiffhomepage
path: root/lexlib
diff options
context:
space:
mode:
Diffstat (limited to 'lexlib')
-rw-r--r--lexlib/LexAccessor.h15
-rw-r--r--lexlib/StyleContext.h115
2 files changed, 74 insertions, 56 deletions
diff --git a/lexlib/LexAccessor.h b/lexlib/LexAccessor.h
index 4223f302d..92e719360 100644
--- a/lexlib/LexAccessor.h
+++ b/lexlib/LexAccessor.h
@@ -126,6 +126,21 @@ public:
return startNext - 1;
}
}
+ int GetRelativePosition(int start, int characterOffset, int *character, int *width) {
+ if (documentVersion >= dvLineEnd) {
+ return (static_cast<IDocumentWithLineEnd *>(pAccess))->GetRelativePosition(
+ start, characterOffset, character, width);
+ } else {
+ // Old version -> byte-oriented only
+ // Handle doc range overflow
+ int posNew = start + characterOffset;
+ if ((posNew < 0) || (posNew > Length()))
+ return -1;
+ *character = SafeGetCharAt(posNew, 0);
+ *width = 1;
+ return start + characterOffset;
+ }
+ }
int LevelAt(int line) const {
return pAccess->GetLevel(line);
}
diff --git a/lexlib/StyleContext.h b/lexlib/StyleContext.h
index 2c010645b..0b5dee379 100644
--- a/lexlib/StyleContext.h
+++ b/lexlib/StyleContext.h
@@ -51,35 +51,27 @@ class StyleContext {
LexAccessor &styler;
unsigned int endPos;
unsigned int lengthDocument;
+
+ // Used for optimizing GetRelativeCharacter
+ unsigned int posRelative;
+ unsigned int currentPosLastRelative;
+ int offsetRelative;
+
StyleContext &operator=(const StyleContext &);
- void GetNextChar(unsigned int pos) {
- chNext = static_cast<unsigned char>(styler.SafeGetCharAt(pos+1, 0));
- if (styler.Encoding() == encUnicode) {
- if (chNext >= 0x80) {
- unsigned char bytes[4] = { static_cast<unsigned char>(chNext), 0, 0, 0 };
- for (int trail=1; trail<3; trail++) {
- bytes[trail] = static_cast<unsigned char>(styler.SafeGetCharAt(pos+1+trail, 0));
- if (!((bytes[trail] >= 0x80) && (bytes[trail] < 0xc0))) {
- bytes[trail] = 0;
- break;
- }
- }
- chNext = UnicodeCodePoint(bytes);
- }
- } else if (styler.Encoding() == encDBCS) {
- if (styler.IsLeadByte(static_cast<char>(chNext))) {
- chNext = chNext << 8;
- chNext |= static_cast<unsigned char>(styler.SafeGetCharAt(pos+2, 0));
- }
+ void GetNextChar() {
+ if (styler.Encoding() == enc8bit) {
+ chNext = static_cast<unsigned char>(styler.SafeGetCharAt(currentPos+width, 0));
+ widthNext = 1;
+ } else {
+ styler.GetRelativePosition(currentPos+width, 0, &chNext, &widthNext);
}
- // End of line?
- // Trigger on CR only (Mac style) or either on LF from CR+LF (Dos/Win)
- // or on LF alone (Unix). Avoid triggering two times on Dos/Win.
+ // End of line determined from line end position, allowing CR, LF,
+ // CRLF and Unicode line ends as set by document.
if (currentLine < lineDocEnd)
- atLineEnd = static_cast<int>(pos) >= (lineStartNext-1);
+ atLineEnd = static_cast<int>(currentPos) >= (lineStartNext-1);
else // Last line
- atLineEnd = static_cast<int>(pos) >= lineStartNext;
+ atLineEnd = static_cast<int>(currentPos) >= lineStartNext;
}
public:
@@ -92,12 +84,17 @@ public:
int state;
int chPrev;
int ch;
+ int width;
int chNext;
+ int widthNext;
StyleContext(unsigned int startPos, unsigned int length,
int initStyle, LexAccessor &styler_, char chMask=31) :
styler(styler_),
endPos(startPos + length),
+ posRelative(0),
+ currentPosLastRelative(0x7FFFFFFF),
+ offsetRelative(0),
currentPos(startPos),
currentLine(-1),
lineStartNext(-1),
@@ -105,7 +102,9 @@ public:
state(initStyle & chMask), // Mask off all bits which aren't in the chMask.
chPrev(0),
ch(0),
- chNext(0) {
+ width(0),
+ chNext(0),
+ widthNext(1) {
styler.StartAt(startPos, chMask);
styler.StartSegment(startPos);
currentLine = styler.GetLine(startPos);
@@ -115,21 +114,14 @@ public:
endPos++;
lineDocEnd = styler.GetLine(lengthDocument);
atLineStart = static_cast<unsigned int>(styler.LineStart(currentLine)) == startPos;
- unsigned int pos = currentPos;
- ch = static_cast<unsigned char>(styler.SafeGetCharAt(pos, 0));
- if (styler.Encoding() == encUnicode) {
- // Get the current char
- GetNextChar(pos-1);
- ch = chNext;
- pos += BytesInUnicodeCodePoint(ch) - 1;
- } else if (styler.Encoding() == encDBCS) {
- if (styler.IsLeadByte(static_cast<char>(ch))) {
- pos++;
- ch = ch << 8;
- ch |= static_cast<unsigned char>(styler.SafeGetCharAt(pos, 0));
- }
- }
- GetNextChar(pos);
+
+ // Variable width is now 0 so GetNextChar gets the char at currentPos into chNext/widthNext
+ width = 0;
+ GetNextChar();
+ ch = chNext;
+ width = widthNext;
+
+ GetNextChar();
}
void Complete() {
styler.ColourTo(currentPos - ((currentPos > lengthDocument) ? 2 : 1), state);
@@ -146,23 +138,10 @@ public:
lineStartNext = styler.LineStart(currentLine+1);
}
chPrev = ch;
- if (styler.Encoding() == encUnicode) {
- currentPos += BytesInUnicodeCodePoint(ch);
- } else if (styler.Encoding() == encDBCS) {
- currentPos++;
- if (ch >= 0x100)
- currentPos++;
- } else {
- currentPos++;
- }
+ currentPos += width;
ch = chNext;
- if (styler.Encoding() == encUnicode) {
- GetNextChar(currentPos + BytesInUnicodeCodePoint(ch)-1);
- } else if (styler.Encoding() == encDBCS) {
- GetNextChar(currentPos + ((ch >= 0x100) ? 1 : 0));
- } else {
- GetNextChar(currentPos);
- }
+ width = widthNext;
+ GetNextChar();
} else {
atLineStart = false;
chPrev = ' ';
@@ -200,6 +179,30 @@ public:
int GetRelative(int n) {
return static_cast<unsigned char>(styler.SafeGetCharAt(currentPos+n, 0));
}
+ int GetRelativeCharacter(int n) {
+ if (n == 0)
+ return ch;
+ if (styler.Encoding() == enc8bit) {
+ // fast version for single byte encodings
+ return static_cast<unsigned char>(styler.SafeGetCharAt(currentPos + n, 0));
+ } else {
+ int ch = 0;
+ int width = 0;
+ //styler.GetRelativePosition(currentPos, n, &ch, &width);
+ if ((currentPosLastRelative != currentPos) ||
+ ((n > 0) && ((offsetRelative < 0) || (n < offsetRelative))) ||
+ ((n < 0) && ((offsetRelative > 0) || (n > offsetRelative)))) {
+ posRelative = currentPos;
+ offsetRelative = 0;
+ }
+ int diffRelative = n - offsetRelative;
+ int posNew = styler.GetRelativePosition(posRelative, diffRelative, &ch, &width);
+ posRelative = posNew;
+ currentPosLastRelative = currentPos;
+ offsetRelative = n;
+ return ch;
+ }
+ }
bool Match(char ch0) const {
return ch == static_cast<unsigned char>(ch0);
}
generated by cgit v1.2.3 (git 2.25.1) at 2025-10-12 10:08:38 +0000