Patch from Jakub to optionally implement more POSIX compatible regular

expressions. \(..\) changes to (..)
Fixes problem where find previous would not find earlier matches on same
line.

7 files changed, 84 insertions, 38 deletions

diff --git a/include/Scintilla.h b/include/Scintilla.h
index ecdda8de1..a04dc6c50 100644
--- a/include/Scintilla.h
+++ b/include/Scintilla.h
@@ -285,6 +285,7 @@ typedef sptr_t (*SciFnDirect)(sptr_t ptr, unsigned int iMessage, uptr_t wParam,
 #define SCFIND_MATCHCASE 4
 #define SCFIND_WORDSTART 0x00100000
 #define SCFIND_REGEXP 0x00200000
+#define SCFIND_POSIX 0x00400000
 #define SCI_FINDTEXT 2150
 #define SCI_FORMATRANGE 2151
 #define SCI_GETFIRSTVISIBLELINE 2152
diff --git a/include/Scintilla.iface b/include/Scintilla.iface
index 984b4a3d8..a0c73053e 100644
--- a/include/Scintilla.iface
+++ b/include/Scintilla.iface
@@ -701,6 +701,7 @@ val SCFIND_WHOLEWORD=2
 val SCFIND_MATCHCASE=4
 val SCFIND_WORDSTART=0x00100000
 val SCFIND_REGEXP=0x00200000
+val SCFIND_POSIX=0x00400000
 
 # Find some text in the document.
 fun position FindText=2150(int flags, findtext ft)
diff --git a/src/Document.cxx b/src/Document.cxx
index 20a53b132..a1d5beff6 100644
--- a/src/Document.cxx
+++ b/src/Document.cxx
@@ -847,7 +847,7 @@ public:
  * Has not been tested with backwards DBCS searches yet.
  */
 long Document::FindText(int minPos, int maxPos, const char *s,
-                        bool caseSensitive, bool word, bool wordStart, bool regExp,
+                        bool caseSensitive, bool word, bool wordStart, bool regExp, bool posix,
                         int *length) {
 	if (regExp) {
 		if (!pre)
@@ -864,7 +864,7 @@ long Document::FindText(int minPos, int maxPos, const char *s,
 		startPos = MovePositionOutsideChar(startPos, 1, false);
 		endPos = MovePositionOutsideChar(endPos, 1, false);
 
-		const char *errmsg = pre->Compile(s, *length, caseSensitive);
+		const char *errmsg = pre->Compile(s, *length, caseSensitive, posix);
 		if (errmsg) {
 			return -1;
 		}
@@ -888,16 +888,30 @@ long Document::FindText(int minPos, int maxPos, const char *s,
 		for (int line = lineRangeStart; line != lineRangeBreak; line += increment) {
 			int startOfLine = LineStart(line);
 			int endOfLine = LineEnd(line);
-			if ((increment == 1) && (line == lineRangeStart)) {
-				if ((startPos != startOfLine) && (s[0] == '^'))
-					continue;	// Can't match start of line if start position after start of line
-				startOfLine = startPos;
-			}
-			if ((increment == 1) && (line == lineRangeEnd)) {
-				if ((endPos != endOfLine) && (searchEnd == '$'))
-					continue;	// Can't match end of line if end position before end of line
-				endOfLine = endPos;
+			if (increment == 1) {
+				if (line == lineRangeStart) {
+					if ((startPos != startOfLine) && (s[0] == '^'))
+						continue;	// Can't match start of line if start position after start of line
+					startOfLine = startPos;
+				}
+				if (line == lineRangeEnd) {
+					if ((endPos != endOfLine) && (searchEnd == '$'))
+						continue;	// Can't match end of line if end position before end of line
+					endOfLine = endPos;
+				}
+			} else {
+				if (line == lineRangeEnd) {
+					if ((endPos != startOfLine) && (s[0] == '^'))
+						continue;	// Can't match start of line if end position after start of line
+					startOfLine = endPos;
+				}
+				if (line == lineRangeStart) {
+					if ((startPos != endOfLine) && (searchEnd == '$'))
+						continue;	// Can't match end of line if start position before end of line
+					endOfLine = startPos;
+				}
 			}
+			
 			DocumentIndexer di(this, endOfLine);
 			int success = pre->Execute(di, startOfLine, endOfLine);
 			if (success) {
diff --git a/src/Document.h b/src/Document.h
index 30fdc625e..597e33171 100644
--- a/src/Document.h
+++ b/src/Document.h
@@ -192,7 +192,7 @@ public:
 	int NextWordStart(int pos, int delta);
 	int Length() { return cb.Length(); }
 	long FindText(int minPos, int maxPos, const char *s,
-		bool caseSensitive, bool word, bool wordStart, bool regExp, int *length);
+		bool caseSensitive, bool word, bool wordStart, bool regExp, bool posix, int *length);
 	long FindText(int iMessage, unsigned long wParam, long lParam);
 	const char *SubstituteByPosition(const char *text, int *length);
 	int LinesTotal();
diff --git a/src/Editor.cxx b/src/Editor.cxx
index b3509510d..7569d5a60 100644
--- a/src/Editor.cxx
+++ b/src/Editor.cxx
@@ -3988,7 +3988,7 @@ void Editor::Indent(bool forwards) {
  */
 long Editor::FindText(
     uptr_t wParam,     	///< Search modes : @c SCFIND_MATCHCASE, @c SCFIND_WHOLEWORD,
-    ///< @c SCFIND_WORDSTART or @c SCFIND_REGEXP.
+    ///< @c SCFIND_WORDSTART, @c SCFIND_REGEXP or @c SCFIND_POSIX.
     sptr_t lParam) {			///< @c TextToFind structure: The text to search for in the given range.
 
 	TextToFind *ft = reinterpret_cast<TextToFind *>(lParam);
@@ -3998,6 +3998,7 @@ long Editor::FindText(
 	                         (wParam & SCFIND_WHOLEWORD) != 0,
 	                         (wParam & SCFIND_WORDSTART) != 0,
 	                         (wParam & SCFIND_REGEXP) != 0,
+	                         (wParam & SCFIND_POSIX) != 0,
 	                         &lengthFound);
 	if (pos != -1) {
 		ft->chrgText.cpMin = pos;
@@ -4041,6 +4042,7 @@ long Editor::SearchText(
 		                     (wParam & SCFIND_WHOLEWORD) != 0,
 		                     (wParam & SCFIND_WORDSTART) != 0,
 		                     (wParam & SCFIND_REGEXP) != 0,
+		                     (wParam & SCFIND_POSIX) != 0,
 		                     &lengthFound);
 	} else {
 		pos = pdoc->FindText(searchAnchor, 0, txt,
@@ -4048,6 +4050,7 @@ long Editor::SearchText(
 		                     (wParam & SCFIND_WHOLEWORD) != 0,
 		                     (wParam & SCFIND_WORDSTART) != 0,
 		                     (wParam & SCFIND_REGEXP) != 0,
+		                     (wParam & SCFIND_POSIX) != 0,
 		                     &lengthFound);
 	}
 
@@ -4069,6 +4072,7 @@ long Editor::SearchInTarget(const char *text, int length) {
 	                         (searchFlags & SCFIND_WHOLEWORD) != 0,
 	                         (searchFlags & SCFIND_WORDSTART) != 0,
 	                         (searchFlags & SCFIND_REGEXP) != 0,
+	                         (searchFlags & SCFIND_POSIX) != 0,
 	                         &lengthFound);
 	if (pos != -1) {
 		targetStart = pos;
diff --git a/src/RESearch.cxx b/src/RESearch.cxx
index 77d897782..430a65e1f 100644
--- a/src/RESearch.cxx
+++ b/src/RESearch.cxx
@@ -30,6 +30,15 @@
  * Modification history:
  *
  * $Log$
+ * Revision 1.8  2003/03/04 10:53:59  nyamatongwe
+ * Patch from Jakub to optionally implement more POSIX compatible regular
+ * expressions. \(..\) changes to (..)
+ * Fixes problem where find previous would not find earlier matches on same
+ * line.
+ *
+ * Revision 1.8  2003/03/03 20:12:56  vrana
+ * Added posix syntax.
+ *
  * Revision 1.7  2002/09/28 00:33:28  nyamatongwe
  * Fixed problem with character ranges caused by expansion to 8 bits.
  *
@@ -332,7 +341,7 @@ const char escapeValue(char ch) {
 	return 0;
 }
 
-const char *RESearch::Compile(const char *pat, int length, bool caseSensitive) {
+const char *RESearch::Compile(const char *pat, int length, bool caseSensitive, bool posix) {
 	char *mp=nfa;          /* nfa pointer       */
 	char *lp;              /* saved pointer..   */
 	char *sp=nfa;          /* another one..     */
@@ -467,25 +476,6 @@ const char *RESearch::Compile(const char *pat, int length, bool caseSensitive) {
 			i++;
 			switch(*++p) {
 
-			case '(':
-				if (tagc < MAXTAG) {
-					tagstk[++tagi] = tagc;
-					*mp++ = BOT;
-					*mp++ = static_cast<char>(tagc++);
-				}
-				else
-					return badpat("Too many \\(\\) pairs");
-				break;
-			case ')':
-				if (*sp == BOT)
-					return badpat("Null pattern inside \\(\\)");
-				if (tagi > 0) {
-					*mp++ = static_cast<char>(EOT);
-					*mp++ = static_cast<char>(tagstk[tagi--]);
-				}
-				else
-					return badpat("Unmatched \\)");
-				break;
 			case '<':
 				*mp++ = BOW;
 				break;
@@ -524,13 +514,49 @@ const char *RESearch::Compile(const char *pat, int length, bool caseSensitive) {
 				*mp++ = escapeValue(*p);
 				break;
 			default:
-				*mp++ = CHR;
-				*mp++ = *p;
+				if (!posix && *p == '(') {
+					if (tagc < MAXTAG) {
+						tagstk[++tagi] = tagc;
+						*mp++ = BOT;
+						*mp++ = static_cast<char>(tagc++);
+					}
+					else
+						return badpat("Too many \\(\\) pairs");
+				} else if (!posix && *p == ')') {
+					if (*sp == BOT)
+						return badpat("Null pattern inside \\(\\)");
+					if (tagi > 0) {
+						*mp++ = static_cast<char>(EOT);
+						*mp++ = static_cast<char>(tagstk[tagi--]);
+					}
+					else
+						return badpat("Unmatched \\)");
+				} else {
+					*mp++ = CHR;
+					*mp++ = *p;
+				}
 			}
 			break;
 
 		default :               /* an ordinary char  */
-			if (caseSensitive) {
+			if (posix && *p == '(') {
+				if (tagc < MAXTAG) {
+					tagstk[++tagi] = tagc;
+					*mp++ = BOT;
+					*mp++ = static_cast<char>(tagc++);
+				}
+				else
+					return badpat("Too many () pairs");
+			} else if (posix && *p == ')') {
+				if (*sp == BOT)
+					return badpat("Null pattern inside ()");
+				if (tagi > 0) {
+					*mp++ = static_cast<char>(EOT);
+					*mp++ = static_cast<char>(tagstk[tagi--]);
+				}
+				else
+					return badpat("Unmatched )");
+			} else if (caseSensitive) {
 				*mp++ = CHR;
 				*mp++ = *p;
 			} else {
@@ -545,7 +571,7 @@ const char *RESearch::Compile(const char *pat, int length, bool caseSensitive) {
 		sp = lp;
 	}
 	if (tagi > 0)
-		return badpat("Unmatched \\(");
+		return badpat((posix ? "Unmatched (" : "Unmatched \\("));
 	*mp = END;
 	sta = OKP;
 	return 0;
diff --git a/src/RESearch.h b/src/RESearch.h
index 9d0f20389..8ca9cd0b8 100644
--- a/src/RESearch.h
+++ b/src/RESearch.h
@@ -32,7 +32,7 @@ public:
 	bool GrabMatches(CharacterIndexer &ci);
 	void ChSet(char c);
 	void ChSetWithCase(char c, bool caseSensitive);
-	const char *Compile(const char *pat, int length, bool caseSensitive);
+	const char *Compile(const char *pat, int length, bool caseSensitive, bool posix);
 	int Execute(CharacterIndexer &ci, int lp, int endp);
 	void ModifyWord(char *s);
 	int Substitute(CharacterIndexer &ci, char *src, char *dst);