Patch from Kein-Hong Man improves regular expression detection

when disambiguating character on previous line or separated from
'/' by a comment.
Includes trailing lowercase regex flags in the regex style.

2 files changed, 56 insertions, 12 deletions

diff --git a/src/LexCPP.cxx b/src/LexCPP.cxx
index cd2755243..40f3a60bd 100644
--- a/src/LexCPP.cxx
+++ b/src/LexCPP.cxx
@@ -36,10 +36,17 @@ static inline bool IsAWordStart(int ch) {
 }
 
 static inline bool IsADoxygenChar(int ch) {
-	return (islower(ch) || ch == '$' || ch == '@' ||
-	        ch == '\\' || ch == '&' || ch == '<' ||
-	        ch == '>' || ch == '#' || ch == '{' ||
-	        ch == '}' || ch == '[' || ch == ']');
+	return (ch < 0x80 && islower(ch)) || ch == '$' || ch == '@' ||
+		ch == '\\' || ch == '&' || ch == '<' ||
+		ch == '>' || ch == '#' || ch == '{' ||
+		ch == '}' || ch == '[' || ch == ']';
+}
+
+static bool IsSpaceEquiv(int state) {
+	return (state <= SCE_C_COMMENTDOC) ||
+		// including SCE_C_DEFAULT, SCE_C_COMMENT, SCE_C_COMMENTLINE
+		(state == SCE_C_COMMENTLINEDOC) || (state == SCE_C_COMMENTDOCKEYWORD) ||
+		(state == SCE_C_COMMENTDOCKEYWORDERROR);
 }
 
 static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
@@ -74,6 +81,16 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 		}
 	}
 
+	// look back to set chPrevNonWhite properly for better regex colouring
+	if (startPos > 0) {
+		int back = startPos;
+		while (--back && IsSpaceEquiv(styler.StyleAt(back)))
+			;
+		if (styler.StyleAt(back) == SCE_C_OPERATOR) {
+			chPrevNonWhite = styler.SafeGetCharAt(back);
+		}
+	}
+
 	StyleContext sc(startPos, length, initStyle, styler);
 
 	for (; sc.More(); sc.Forward()) {
@@ -86,7 +103,6 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 			}
 			// Reset states to begining of colourise so no surprises
 			// if different sets of lines lexed.
-			chPrevNonWhite = ' ';
 			visibleChars = 0;
 			lastWordWasUUID = false;
 		}
@@ -224,7 +240,10 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 				if (sc.atLineStart) {
 					sc.SetState(SCE_C_DEFAULT);
 				} else if (sc.ch == '/') {
-					sc.ForwardSetState(SCE_C_DEFAULT);
+					sc.Forward();
+					while ((sc.ch < 0x80) && islower(sc.ch))
+						sc.Forward();    // gobble regex flags
+					sc.SetState(SCE_C_DEFAULT);
 				} else if (sc.ch == '\\') {
 					// Gobble up the quoted character
 					if (sc.chNext == '\\' || sc.chNext == '/') {
@@ -305,7 +324,7 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 			}
 		}
 
-		if (!IsASpace(sc.ch)) {
+		if (!IsASpace(sc.ch) && !IsSpaceEquiv(sc.state)) {
 			chPrevNonWhite = sc.ch;
 			visibleChars++;
 		}
@@ -316,9 +335,9 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 
 static bool IsStreamCommentStyle(int style) {
 	return style == SCE_C_COMMENT ||
-	       style == SCE_C_COMMENTDOC ||
-	       style == SCE_C_COMMENTDOCKEYWORD ||
-	       style == SCE_C_COMMENTDOCKEYWORDERROR;
+		style == SCE_C_COMMENTDOC ||
+		style == SCE_C_COMMENTDOCKEYWORD ||
+		style == SCE_C_COMMENTDOCKEYWORDERROR;
 }
 
 // Store both the current line's fold level and the next lines in the
diff --git a/src/LexHTML.cxx b/src/LexHTML.cxx
index b822852a5..a268c2b0d 100644
--- a/src/LexHTML.cxx
+++ b/src/LexHTML.cxx
@@ -2,7 +2,7 @@
 /** @file LexHTML.cxx
  ** Lexer for HTML.
  **/
-// Copyright 1998-2003 by Neil Hodgson <neilh@scintilla.org>
+// Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>
 // The License.txt file describes the conditions under which this software may be distributed.
 
 #include <stdlib.h>
@@ -503,12 +503,28 @@ static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initSty
 	char chPrev = ' ';
 	char ch = ' ';
 	char chPrevNonWhite = ' ';
+	// look back to set chPrevNonWhite properly for better regex colouring
+	if (scriptLanguage == eScriptJS && startPos > 0) {
+		int back = startPos;
+		int style = 0;
+		while (--back) {
+			style = styler.StyleAt(back);
+			if (style < SCE_HJ_DEFAULT || style > SCE_HJ_COMMENTDOC)
+				// includes SCE_HJ_COMMENT & SCE_HJ_COMMENTLINE
+				break;
+		}
+		if (style == SCE_HJ_SYMBOLS) {
+			chPrevNonWhite = styler.SafeGetCharAt(back);
+		}
+	}
+
 	styler.StartSegment(startPos);
 	const int lengthDoc = startPos + length;
 	for (int i = startPos; i < lengthDoc; i++) {
 		const char chPrev2 = chPrev;
 		chPrev = ch;
-		if (ch != ' ' && ch != '\t')
+		if (!isspacechar(ch) && state != SCE_HJ_COMMENT &&
+			state != SCE_HJ_COMMENTLINE && state != SCE_HJ_COMMENTDOC)
 			chPrevNonWhite = ch;
 		ch = styler[i];
 		char chNext = styler.SafeGetCharAt(i + 1);
@@ -1236,12 +1252,14 @@ static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initSty
 			if (ch == '/' && chPrev == '*') {
 				styler.ColourTo(i, StateToPrint);
 				state = SCE_HJ_DEFAULT;
+				ch = ' ';
 			}
 			break;
 		case SCE_HJ_COMMENTLINE:
 			if (ch == '\r' || ch == '\n') {
 				styler.ColourTo(i - 1, statePrintForState(SCE_HJ_COMMENTLINE, inScriptType));
 				state = SCE_HJ_DEFAULT;
+				ch = ' ';
 			}
 			break;
 		case SCE_HJ_DOUBLESTRING:
@@ -1289,6 +1307,13 @@ static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initSty
 			break;
 		case SCE_HJ_REGEX:
 			if (ch == '\r' || ch == '\n' || ch == '/') {
+				if (ch == '/') {
+					while (chNext < 0x80 && islower(chNext)) {   // gobble regex flags
+						i++;
+						ch = chNext;
+						chNext = styler.SafeGetCharAt(i + 1);
+					}
+				}
 				styler.ColourTo(i, StateToPrint);
 				state = SCE_HJ_DEFAULT;
 			} else if (ch == '\\') {