Changed operator bool in WordList to return false if WordList is empty.

Many lexer changes from Philippe Lhoste.
VB handles preprocessor and hex constants.
C++ optionally leaves preprocessor state after the preprocessor command.
HTML terminates incomplete entities earlier and marks them as bad attributes.

3 files changed, 18 insertions, 51 deletions

diff --git a/src/LexCPP.cxx b/src/LexCPP.cxx
index d62375e9c..cad24a773 100644
--- a/src/LexCPP.cxx
+++ b/src/LexCPP.cxx
@@ -46,6 +46,7 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 	styler.StartAt(startPos);
 	
 	bool fold = styler.GetPropertyInt("fold");
+	bool stylingWithinPreprocessor = styler.GetPropertyInt("styling.within.preprocessor");
 	int lineCurrent = styler.GetLine(startPos);
 	int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 	int levelCurrent = levelPrev;
@@ -118,15 +119,12 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 			} else if (ch == '#') {
 				styler.ColourTo(i-1, state);
 				state = SCE_C_PREPROCESSOR;
-#ifdef COLORIZE_ALL_PREPROC /* PL 2000/05/18 */
-#else  /* OLD PL 2000/05/18 */
 				// Skip whitespace between # and preprocessor word
 				do {
 					i++;
 					ch = chNext;
 					chNext = styler.SafeGetCharAt(i + 1);
 				} while (isspace(ch));
-#endif /* OLD PL 2000/05/18 */
 			} else if (isoperator(ch)) {
 				styler.ColourTo(i-1, state);
 				styler.ColourTo(i, SCE_C_OPERATOR);
@@ -149,10 +147,6 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 					state = SCE_C_STRING;
 				} else if (ch == '\'') {
 					state = SCE_C_CHARACTER;
-#ifdef OLD /* PL 2000/05/18 -- A preprocessor symbol shouldn't start in the middle of a word! */
-				} else if (ch == '#') {
-					state = SCE_C_PREPROCESSOR;
-#endif /* OLD PL 2000/05/18 */
 				} else if (isoperator(ch)) {
 					styler.ColourTo(i, SCE_C_OPERATOR);
 					if ((ch == '{') || (ch == '}')) {
@@ -162,17 +156,17 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 			}
 		} else {
 			if (state == SCE_C_PREPROCESSOR) {
-#ifdef COLORIZE_ALL_PREPROC /* PL 2000/05/18 -- Stop after the preprocessor word */
-				if ((ch == '\r' || ch == '\n') && !(chPrev == '\\' || chPrev == '\r')) {
-					styler.ColourTo(i-1, state);
-					state = SCE_C_DEFAULT;
-				}
-#else  /* OLD PL 2000/05/18 */
-				if (isspace(ch)) {
-					styler.ColourTo(i-1, state);
-					state = SCE_C_DEFAULT;
+				if (stylingWithinPreprocessor) {
+					if (isspace(ch)) {
+						styler.ColourTo(i-1, state);
+						state = SCE_C_DEFAULT;
+					}
+				} else {
+					if ((ch == '\r' || ch == '\n') && !(chPrev == '\\' || chPrev == '\r')) {
+						styler.ColourTo(i-1, state);
+						state = SCE_C_DEFAULT;
+					}
 				}
-#endif /* OLD PL 2000/05/18 */
 			} else if (state == SCE_C_COMMENT) {
 				if (ch == '/' && chPrev == '*') {
 					if (((i > styler.GetStartSegment() + 2) || (
@@ -248,10 +242,6 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 					state = SCE_C_STRING;
 				} else if (ch == '\'') {
 					state = SCE_C_CHARACTER;
-#ifdef OLD /* PL 2000/05/18 -- A preprocessor symbol SHOULD be the first non-white char. of the line! */
-				} else if (ch == '#') {
-					state = SCE_C_PREPROCESSOR;
-#endif /* OLD PL 2000/05/18 */
 				} else if (iswordstart(ch)) {
 					state = SCE_C_WORD;
 				} else if (isoperator(ch)) {
diff --git a/src/LexHTML.cxx b/src/LexHTML.cxx
index 0b31cba39..9a9377292 100644
--- a/src/LexHTML.cxx
+++ b/src/LexHTML.cxx
@@ -54,6 +54,9 @@ static void classifyAttribHTML(unsigned int start, unsigned int end, WordList &k
 		if (keywords.InList(s))
 			chAttr = SCE_H_ATTRIBUTE;
 	}
+	if ((chAttr == SCE_H_ATTRIBUTEUNKNOWN) && !keywords)
+		// No keywords -> all are known
+		chAttr = SCE_H_ATTRIBUTE;
 	styler.ColourTo(end, chAttr);
 }
 
@@ -81,6 +84,9 @@ static int classifyTagHTML(unsigned int start, unsigned int end,
 				chAttr = SCE_H_SCRIPT;
 		}
 	}
+	if ((chAttr == SCE_H_TAGUNKNOWN) && !keywords)
+		// No keywords -> all are known
+		chAttr = SCE_H_TAG;
 	styler.ColourTo(end, chAttr);
 	return chAttr;
 }
@@ -353,13 +359,10 @@ static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initSty
 				styler.ColourTo(i, state);
 				state = SCE_H_DEFAULT;
 			}
-#ifdef OLD /* PL 2000/05/18 -- An bad entity may stop on a non-alphabetic character */
-#else  /* OLD PL 2000/05/18 */
 			if (ch != '#' && !isalnum(ch)) {	// Should check that '#' follows '&', but it is unlikely anyway...
 				styler.ColourTo(i, SCE_H_TAGUNKNOWN);
 				state = SCE_H_DEFAULT;
 			}
-#endif /* OLD PL 2000/05/18 */
 		} else if (state == SCE_H_TAGUNKNOWN) {
 			if (!ishtmlwordchar(ch) && ch != '/' && ch != '-') {
 				int eClass = classifyTagHTML(styler.GetStartSegment(), i - 1, keywords, styler);
@@ -368,7 +371,7 @@ static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initSty
 					scriptLanguage = eScriptJS;
 					eClass = SCE_H_TAG;
 				}
-				if (ch == '>') {
+				if ((ch == '>') && (eClass != SCE_H_COMMENT)) {
 					styler.ColourTo(i, SCE_H_TAG);
 					if (lastTagWasScript) {
 						if (scriptLanguage == eScriptVBS)
diff --git a/src/LexVB.cxx b/src/LexVB.cxx
index b5b3bf9ab..feb9fe24a 100644
--- a/src/LexVB.cxx
+++ b/src/LexVB.cxx
@@ -20,38 +20,21 @@
 static int classifyWordVB(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler) {
 
 	char s[100];
-#ifdef OLD /* PL 2000/05/18 -- Include hexadecimal number */
-	bool wordIsNumber = isdigit(styler[start]) || (styler[start] == '.');
-#else  /* OLD PL 2000/05/18 */
 	bool wordIsNumber = isdigit(styler[start]) || (styler[start] == '.') ||
 		(styler[start] == '&' && tolower(styler[start+1]) == 'h');
-#endif /* OLD PL 2000/05/18 */
 	unsigned int i;
 	for (i = 0; i < end - start + 1 && i < 30; i++) {
 		s[i] = static_cast<char>(tolower(styler[start + i]));
-#ifdef OLD /* PL 2000/05/18 -- Little optimization */
-		s[i + 1] = '\0';
-	}
-#else  /* OLD PL 2000/05/18 */
 	}
 	s[i] = '\0';
-#endif /* OLD PL 2000/05/18 */
 	char chAttr = SCE_C_DEFAULT;
 	if (wordIsNumber)
 		chAttr = SCE_C_NUMBER;
 	else {
-#ifdef OLD /* PL 2000/05/18 */
-		if (keywords.InList(s)) {
-			chAttr = SCE_C_WORD;
-			if (strcmp(s, "rem") == 0)
-				chAttr = SCE_C_COMMENTLINE;
-		}
-#else  /* OLD PL 2000/05/18 */
 		if (strcmp(s, "rem") == 0)
 			chAttr = SCE_C_COMMENTLINE;
 		else if (keywords.InList(s))
 			chAttr = SCE_C_WORD;
-#endif /* OLD PL 2000/05/18 */
 	}
 	styler.ColourTo(end, chAttr);
 	if (chAttr == SCE_C_COMMENTLINE)
@@ -91,8 +74,6 @@ static void ColouriseVBDoc(unsigned int startPos, int length, int initStyle,
 			} else if (ch == '\"') {
 				styler.ColourTo(i - 1, state);
 				state = SCE_C_STRING;
-#ifdef OLD /* PL 2000/05/18 */
-#else  /* OLD PL 2000/05/18 */
 			} else if (ch == '#') {
 				styler.ColourTo(i - 1, state);
 				state = SCE_C_PREPROCESSOR;
@@ -102,7 +83,6 @@ static void ColouriseVBDoc(unsigned int startPos, int length, int initStyle,
 			} else if (isoperator(ch)) {
 				styler.ColourTo(i - 1, state);
 				styler.ColourTo(i, SCE_C_OPERATOR);
-#endif /* OLD PL 2000/05/18 */
 			}
 		} else if (state == SCE_C_WORD) {
 			if (!iswordchar(ch)) {
@@ -112,12 +92,9 @@ static void ColouriseVBDoc(unsigned int startPos, int length, int initStyle,
 						state = SCE_C_COMMENTLINE;
 					} else if (ch == '\"') {
 						state = SCE_C_STRING;
-#ifdef OLD /* PL 2000/05/18 */
-#else  /* OLD PL 2000/05/18 */
 					} else if (isoperator(ch)) {
 						styler.ColourTo(i - 1, state);
 						styler.ColourTo(i, SCE_C_OPERATOR);
-#endif /* OLD PL 2000/05/18 */
 					}
 				}
 			}
@@ -136,14 +113,11 @@ static void ColouriseVBDoc(unsigned int startPos, int length, int initStyle,
 					ch = chNext;
 					chNext = styler.SafeGetCharAt(i + 1);
 				}
-#ifdef OLD /* PL 2000/05/18 */
-#else  /* OLD PL 2000/05/18 */
 			} else if (state == SCE_C_PREPROCESSOR) {
 				if (ch == '\r' || ch == '\n') {
 					styler.ColourTo(i - 1, state);
 					state = SCE_C_DEFAULT;
 				}
-#endif /* OLD PL 2000/05/18 */
 			}
 			if (state == SCE_C_DEFAULT) {    // One of the above succeeded
 				if (ch == '\'') {