Regular expression find support.

8 files changed, 151 insertions, 77 deletions

diff --git a/gtk/makefile b/gtk/makefile
index 77d7225e3..7ee7a70db 100644
--- a/gtk/makefile
+++ b/gtk/makefile
@@ -35,7 +35,7 @@ all: $(COMPLIB) $(LEXOBJS)
 $(COMPLIB): DocumentAccessor.o WindowAccessor.o KeyWords.o Document.o CallTip.o \
 	ScintillaBase.o ContractionState.o Editor.o PropSet.o PlatGTK.o \
 	KeyMap.o LineMarker.o ScintillaGTK.o CellBuffer.o ViewStyle.o \
-	Style.o Indicator.o AutoComplete.o
+	PosRegExp.o Style.o Indicator.o AutoComplete.o
 	$(AR) rc $@ $^
 
 AutoComplete.o: AutoComplete.cxx Platform.h AutoComplete.h
@@ -43,8 +43,8 @@ CallTip.o: CallTip.cxx Platform.h CallTip.h
 CellBuffer.o: CellBuffer.cxx Platform.h Scintilla.h WinDefs.h \
  CellBuffer.h
 ContractionState.o: ContractionState.cxx Platform.h ContractionState.h
-Document.o: Document.cxx Platform.h Scintilla.h WinDefs.h CellBuffer.h \
- Document.h
+Document.o: Document.cxx Platform.h Scintilla.h WinDefs.h PosRegExp.h \

+ CellBuffer.h Document.h
 DocumentAccessor.o: DocumentAccessor.cxx Platform.h PropSet.h SString.h \
  Accessor.h DocumentAccessor.h Scintilla.h WinDefs.h
 Editor.o: Editor.cxx Platform.h Scintilla.h WinDefs.h \
@@ -80,6 +80,7 @@ LexVB.o: LexVB.cxx Platform.h PropSet.h SString.h \
 LineMarker.o: LineMarker.cxx Platform.h Scintilla.h WinDefs.h \
  LineMarker.h
 PlatGTK.o: PlatGTK.cxx Platform.h Scintilla.h ScintillaWidget.h WinDefs.h
+PosRegExp.o: PosRegExp.cxx PosRegExp.h
 PropSet.o: PropSet.cxx Platform.h PropSet.h SString.h
 SciTEBase.o: SciTEBase.cxx Platform.h WinDefs.h SciTE.h PropSet.h \
  SString.h Accessor.h KeyWords.h Scintilla.h SciLexer.h SciTEBase.h
diff --git a/gtk/scintilla.mak b/gtk/scintilla.mak
index 6398b4122..0d813f63a 100644
--- a/gtk/scintilla.mak
+++ b/gtk/scintilla.mak
@@ -107,6 +107,7 @@ SOBJS=\
 	$(DIR_O)\KeyMap.obj \
 	$(DIR_O)\LineMarker.obj \
 	$(DIR_O)\PlatGTK.obj \
+	$(DIR_O)\PosRegExp.obj \
 	$(DIR_O)\ScintillaBase.obj \
 	$(DIR_O)\ScintillaGTK.obj \
 	$(DIR_O)\Style.obj \
@@ -138,6 +139,7 @@ LOBJS=\
 	$(DIR_O)\KeyWords.obj \
 	$(DIR_O)\LineMarker.obj \
 	$(DIR_O)\PlatGTK.obj \
+	$(DIR_O)\PosRegExp.obj \
 	$(DIR_O)\PropSet.obj \
 	$(DIR_O)\ScintillaBaseL.obj \
 	$(DIR_O)\ScintillaGTKL.obj \
@@ -203,8 +205,8 @@ $(DIR_O)\CellBuffer.obj: ..\src\CellBuffer.cxx ..\include\Platform.h ..\include\
 
 $(DIR_O)\ContractionState.obj: ..\src\ContractionState.cxx ..\include\Platform.h ..\src\ContractionState.h
 
-$(DIR_O)\Document.obj: ..\src\Document.cxx ..\include\Platform.h ..\include\Scintilla.h ..\src\CellBuffer.h \
- ..\src\Document.h
+$(DIR_O)\Document.obj: ..\src\Document.cxx ..\include\Platform.h ..\include\Scintilla.h ..\include\PosRegExp.h \
+ ..\src\CellBuffer.h ..\src\Document.h
 
 $(DIR_O)\DocumentAccessor.obj: ..\src\DocumentAccessor.cxx ..\include\Platform.h ..\include\PropSet.h ..\include\Accessor.h ..\src\DocumentAccessor.h ..\include\Scintilla.h
 
@@ -253,6 +255,8 @@ $(DIR_O)\LineMarker.obj: ..\src\LineMarker.cxx ..\include\Platform.h ..\include\
 
 $(DIR_O)\PlatWin.obj: PlatWin.cxx ..\include\Platform.h PlatformRes.h ..\src\UniConversion.h
 
+$(DIR_O)\PosRegExp.obj: ..\src\PosRegExp.cxx ..\include\PosRegExp.h
+
 $(DIR_O)\PropSet.obj: ..\src\PropSet.cxx ..\include\Platform.h ..\include\PropSet.h
 
 $(DIR_O)\ScintillaBase.obj: ..\src\ScintillaBase.cxx ..\include\Platform.h ..\include\Scintilla.h \
diff --git a/src/Document.cxx b/src/Document.cxx
index 6be928cf7..babe25bcf 100644
--- a/src/Document.cxx
+++ b/src/Document.cxx
@@ -16,6 +16,7 @@
 #include "SVector.h"
 #include "CellBuffer.h"
 #include "Document.h"
+#include "PosRegExp.h"
 
 // This is ASCII specific but is safe with chars >= 0x80
 inline bool isspacechar(unsigned char ch) {
@@ -742,70 +743,118 @@ bool Document::IsWordAt(int start, int end) {
 	return IsWordStartAt(start) && IsWordEndAt(end);
 }
 
+char Document::DocCharAt(int pos, void *param) {
+	return reinterpret_cast<Document*>(param)->CharAt(pos);
+}
+
 // Find text in document, supporting both forward and backward
 // searches (just pass minPos > maxPos to do a backward search)
 // 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 forward = minPos <= maxPos;
-	int increment = forward ? 1 : -1;
-
-	// Range endpoints should not be inside DBCS characters, but just in case, move them.
-	int startPos = MovePositionOutsideChar(minPos, increment, false);
-	int endPos = MovePositionOutsideChar(maxPos, increment, false);
-
-	// Compute actual search ranges needed
-	int lengthFind = strlen(s);
-	int endSearch = endPos;
-	if (startPos <= endPos) {
-		endSearch = endPos - lengthFind + 1;
-	}
-	//Platform::DebugPrintf("Find %d %d %s %d\n", startPos, endPos, ft->lpstrText, lengthFind);
-	char firstChar = s[0];
-	if (!caseSensitive)
-		firstChar = static_cast<char>(toupper(firstChar));
-	int pos = startPos;
-	while (forward ? (pos < endSearch) : (pos >= endSearch)) {
-		char ch = CharAt(pos);
-		if (caseSensitive) {
-			if (ch == firstChar) {
-				bool found = true;
-				for (int posMatch = 1; posMatch < lengthFind && found; posMatch++) {
-					ch = CharAt(pos + posMatch);
-					if (ch != s[posMatch])
-						found = false;
-				}
-				if (found) {
-					if ((!word && !wordStart) ||
-					        word && IsWordAt(pos, pos + lengthFind) ||
-					        wordStart && IsWordStartAt(pos))
-						return pos;
-				}
-			}
+                        bool caseSensitive, bool word, bool wordStart, bool regExp,
+			int *length) {
+	if (regExp) {
+		char *pat = new char[strlen(s) + 4];
+		if (!pat)
+			return -1;
+		
+		strcpy(pat, "/");
+		int startPos;
+		int endPos;
+
+		if (minPos <= maxPos) {
+			startPos = minPos;
+			endPos = maxPos;
 		} else {
-			if (toupper(ch) == firstChar) {
-				bool found = true;
-				for (int posMatch = 1; posMatch < lengthFind && found; posMatch++) {
-					ch = CharAt(pos + posMatch);
-					if (toupper(ch) != toupper(s[posMatch]))
-						found = false;
+			startPos = maxPos;
+			endPos = minPos;
+		}
+
+		// Range endpoints should not be inside DBCS characters, but just in case, move them.
+		startPos = MovePositionOutsideChar(startPos, 1, false);
+		endPos = MovePositionOutsideChar(endPos, 1, false);
+
+		strcat(pat, s);
+		strcat(pat, "/");
+		PosRegExp re;
+		if (!caseSensitive)
+			strcat(pat, "i");
+		if (!re.SetExpr(pat)) {
+			delete []pat;
+			return -1;
+		}
+		re.param = this;
+		re.CharAt = DocCharAt;
+		SMatches matches;
+		if (!re.Parse(startPos, 0, endPos, &matches)) {
+			delete []pat;
+			return -1;
+		}
+		*length = matches.e[0] - matches.s[0];
+		delete []pat;
+		return matches.s[0];
+	} else {
+				
+		bool forward = minPos <= maxPos;
+		int increment = forward ? 1 : -1;
+	
+		// Range endpoints should not be inside DBCS characters, but just in case, move them.
+		int startPos = MovePositionOutsideChar(minPos, increment, false);
+		int endPos = MovePositionOutsideChar(maxPos, increment, false);
+	
+		// Compute actual search ranges needed
+		int lengthFind = strlen(s);
+		int endSearch = endPos;
+		if (startPos <= endPos) {
+			endSearch = endPos - lengthFind + 1;
+		}
+		//Platform::DebugPrintf("Find %d %d %s %d\n", startPos, endPos, ft->lpstrText, lengthFind);
+		char firstChar = s[0];
+		if (!caseSensitive)
+			firstChar = static_cast<char>(toupper(firstChar));
+		int pos = startPos;
+		while (forward ? (pos < endSearch) : (pos >= endSearch)) {
+			char ch = CharAt(pos);
+			if (caseSensitive) {
+				if (ch == firstChar) {
+					bool found = true;
+					for (int posMatch = 1; posMatch < lengthFind && found; posMatch++) {
+						ch = CharAt(pos + posMatch);
+						if (ch != s[posMatch])
+							found = false;
+					}
+					if (found) {
+						if ((!word && !wordStart) ||
+							word && IsWordAt(pos, pos + lengthFind) ||
+							wordStart && IsWordStartAt(pos))
+							return pos;
+					}
 				}
-				if (found) {
-					if ((!word && !wordStart) ||
-					        word && IsWordAt(pos, pos + lengthFind) ||
-					        wordStart && IsWordStartAt(pos))
-						return pos;
+			} else {
+				if (toupper(ch) == firstChar) {
+					bool found = true;
+					for (int posMatch = 1; posMatch < lengthFind && found; posMatch++) {
+						ch = CharAt(pos + posMatch);
+						if (toupper(ch) != toupper(s[posMatch]))
+							found = false;
+					}
+					if (found) {
+						if ((!word && !wordStart) ||
+							word && IsWordAt(pos, pos + lengthFind) ||
+							wordStart && IsWordStartAt(pos))
+							return pos;
+					}
 				}
 			}
-		}
-		pos += increment;
-		if (dbcsCodePage) {
-			// Ensure trying to match from start of character
-			pos = MovePositionOutsideChar(pos, increment, false);
+			pos += increment;
+			if (dbcsCodePage) {
+				// Ensure trying to match from start of character
+				pos = MovePositionOutsideChar(pos, increment, false);
+			}
 		}
 	}
 	//Platform::DebugPrintf("Not found\n");
-	return - 1;
+	return -1;
 }
 
 int Document::LinesTotal() {
diff --git a/src/Document.h b/src/Document.h
index 4b0b9b1db..cfb131608 100644
--- a/src/Document.h
+++ b/src/Document.h
@@ -173,7 +173,7 @@ 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 caseSensitive, bool word, bool wordStart, bool regExp, int *length);
 	long FindText(int iMessage, unsigned long wParam, long lParam);
 	int LinesTotal();
 	
@@ -196,6 +196,8 @@ public:
 	const WatcherWithUserData *GetWatchers() const { return watchers; }
 	int GetLenWatchers() const { return lenWatchers; }
 	
+	static char DocCharAt(int pos, void *param);
+	
 	bool IsWordPartSeparator(char ch);
 	int WordPartLeft(int pos);
 	int WordPartRight(int pos);
diff --git a/src/Editor.cxx b/src/Editor.cxx
index afc127c07..479157e39 100644
--- a/src/Editor.cxx
+++ b/src/Editor.cxx
@@ -2458,13 +2458,17 @@ void Editor::Indent(bool forwards) {
 
 long Editor::FindText(unsigned int iMessage, unsigned long wParam, long lParam) {
 	TextToFind *ft = reinterpret_cast<TextToFind *>(lParam);
+	int lengthFound = strlen(ft->lpstrText);
 	int pos = pdoc->FindText(ft->chrg.cpMin, ft->chrg.cpMax, ft->lpstrText,
-	                         wParam & SCFIND_MATCHCASE, wParam & SCFIND_WHOLEWORD,
-	                         wParam & SCFIND_WORDSTART);
+		wParam & SCFIND_MATCHCASE, 
+		wParam & SCFIND_WHOLEWORD,
+		wParam & SCFIND_WORDSTART,
+		wParam & SCFIND_REGEXP,
+		&lengthFound);
 	if (pos != -1) {
 		if (iMessage != EM_FINDTEXT) {
 			ft->chrgText.cpMin = pos;
-			ft->chrgText.cpMax = pos + strlen(ft->lpstrText);
+			ft->chrgText.cpMax = pos + lengthFound;
 		}
 	}
 	return pos;
@@ -2489,21 +2493,25 @@ void Editor::SearchAnchor() {
 long Editor::SearchText(unsigned int iMessage, unsigned long wParam, long lParam) {
 	const char *txt = reinterpret_cast<char *>(lParam);
 	int pos;
-
+	int lengthFound = strlen(txt);
 	if (iMessage == SCI_SEARCHNEXT) {
 		pos = pdoc->FindText(searchAnchor, pdoc->Length(), txt,
-		                     wParam & SCFIND_MATCHCASE,
-		                     wParam & SCFIND_WHOLEWORD,
-		                     wParam & SCFIND_WORDSTART);
+					wParam & SCFIND_MATCHCASE,
+					wParam & SCFIND_WHOLEWORD,
+					wParam & SCFIND_WORDSTART,
+					wParam & SCFIND_REGEXP,
+					&lengthFound);
 	} else {
 		pos = pdoc->FindText(searchAnchor, 0, txt,
-		                     wParam & SCFIND_MATCHCASE,
-		                     wParam & SCFIND_WHOLEWORD,
-		                     wParam & SCFIND_WORDSTART);
+					wParam & SCFIND_MATCHCASE,
+					wParam & SCFIND_WHOLEWORD,
+					wParam & SCFIND_WORDSTART,
+					wParam & SCFIND_REGEXP,
+					&lengthFound);
 	}
 
 	if (pos != -1) {
-		SetSelection(pos, pos + strlen(txt));
+		SetSelection(pos, pos + lengthFound);
 	}
 
 	return pos;
diff --git a/vcbuild/SciLexer.dsp b/vcbuild/SciLexer.dsp
index 6ff3245c7..1631607f0 100644
--- a/vcbuild/SciLexer.dsp
+++ b/vcbuild/SciLexer.dsp
@@ -186,6 +186,10 @@ SOURCE=..\win32\PlatWin.cxx
 # End Source File
 # Begin Source File
 
+SOURCE=..\src\PosRegExp.cxx
+# End Source File
+# Begin Source File
+
 SOURCE=..\src\PropSet.cxx
 # End Source File
 # Begin Source File
diff --git a/win32/makefile b/win32/makefile
index 8874825cd..b0e5d0609 100644
--- a/win32/makefile
+++ b/win32/makefile
@@ -39,15 +39,16 @@ LEXOBJS	=  LexAVE.o LexConf.o LexCPP.o LexHTML.o LexLua.o LexOthers.o \
 
 SOBJS	= ScintillaWin.o ScintillaBase.o Editor.o Document.o \
 	ContractionState.o CellBuffer.o CallTip.o \
-	ScintRes.o PlatWin.o KeyMap.o Indicator.o LineMarker.o Style.o \
-	ViewStyle.o AutoComplete.o UniConversion.o
+	ScintRes.o PlatWin.o KeyMap.o Indicator.o LineMarker.o PosRegExp.o \
+	Style.o ViewStyle.o AutoComplete.o UniConversion.o
 $(COMPONENT): $(SOBJS)
 	$(DLLWRAP) --target i386-mingw32 -o $@ $(SOBJS) $(LDFLAGS) -s --relocatable
 
 LOBJS	= ScintillaWinL.o ScintillaBaseL.o Editor.o Document.o \
 	ContractionState.o CellBuffer.o CallTip.o \
-	ScintRes.o PlatWin.o KeyMap.o Indicator.o LineMarker.o Style.o ViewStyle.o \
-	AutoComplete.o UniConversion.o KeyWords.o DocumentAccessor.o PropSet.o $(LEXOBJS)
+	ScintRes.o PlatWin.o KeyMap.o Indicator.o LineMarker.o PosRegExp.o \
+	Style.o ViewStyle.o AutoComplete.o UniConversion.o KeyWords.o \
+	DocumentAccessor.o PropSet.o $(LEXOBJS)
 $(LEXCOMPONENT): $(LOBJS)
 	$(DLLWRAP) --target i386-mingw32 -o $@ $(LOBJS) $(LDFLAGS) -s --relocatable
 
@@ -55,8 +56,9 @@ AutoComplete.o: AutoComplete.cxx Platform.h AutoComplete.h
 CallTip.o: CallTip.cxx Platform.h CallTip.h
 CellBuffer.o: CellBuffer.cxx Platform.h Scintilla.h CellBuffer.h
 ContractionState.o: ContractionState.cxx Platform.h ContractionState.h
-Document.o: Document.cxx Platform.h Scintilla.h CellBuffer.h \
+Document.o: Document.cxx Platform.h Scintilla.h PosRegExp.h CellBuffer.h \
  Document.h
+PosRegExp.o: PosRegExp.cxx PosRegExp.h
 DocumentAccessor.o: DocumentAccessor.cxx Platform.h PropSet.h \
  SString.h Accessor.h DocumentAccessor.h Scintilla.h
 Editor.o: Editor.cxx Platform.h Scintilla.h ContractionState.h \
diff --git a/win32/scintilla.mak b/win32/scintilla.mak
index fdf3c3865..3e3dbcf1a 100644
--- a/win32/scintilla.mak
+++ b/win32/scintilla.mak
@@ -95,6 +95,7 @@ SOBJS=\
 	$(DIR_O)\KeyMap.obj \
 	$(DIR_O)\LineMarker.obj \
 	$(DIR_O)\PlatWin.obj \
+	$(DIR_O)\PosRegExp.obj \
 	$(DIR_O)\ScintillaBase.obj \
 	$(DIR_O)\ScintillaWin.obj \
 	$(DIR_O)\Style.obj \
@@ -127,6 +128,7 @@ LOBJS=\
 	$(DIR_O)\KeyWords.obj \
 	$(DIR_O)\LineMarker.obj \
 	$(DIR_O)\PlatWin.obj \
+	$(DIR_O)\PosRegExp.obj \
 	$(DIR_O)\PropSet.obj \
 	$(DIR_O)\ScintillaBaseL.obj \
 	$(DIR_O)\ScintillaWinL.obj \
@@ -186,8 +188,8 @@ $(DIR_O)\CellBuffer.obj: ..\src\CellBuffer.cxx ..\include\Platform.h ..\include\
 
 $(DIR_O)\ContractionState.obj: ..\src\ContractionState.cxx ..\include\Platform.h ..\src\ContractionState.h
 
-$(DIR_O)\Document.obj: ..\src\Document.cxx ..\include\Platform.h ..\include\Scintilla.h ..\src\CellBuffer.h \
- ..\src\Document.h
+$(DIR_O)\Document.obj: ..\src\Document.cxx ..\include\Platform.h ..\include\Scintilla.h \
+ ..\include\PosRegExp.h ..\src\CellBuffer.h ..\src\Document.h
 
 $(DIR_O)\DocumentAccessor.obj: ..\src\DocumentAccessor.cxx ..\include\Platform.h ..\include\PropSet.h \
  ..\include\SString.h ..\include\Accessor.h ..\src\DocumentAccessor.h ..\include\Scintilla.h
@@ -252,6 +254,8 @@ $(DIR_O)\LineMarker.obj: ..\src\LineMarker.cxx ..\include\Platform.h ..\include\
 
 $(DIR_O)\PlatWin.obj: PlatWin.cxx ..\include\Platform.h PlatformRes.h ..\src\UniConversion.h
 
+$(DIR_O)\PosRegExp.obj: ..\src\PosRegExp.cxx ..\include\PosRegExp.h
+
 $(DIR_O)\PropSet.obj: ..\src\PropSet.cxx ..\include\Platform.h ..\include\PropSet.h \
  ..\include\SString.h