Start of new lexer infrastructure.

Lexers can have a fold function as well as a lexer function.
They can be identified by string name as well as an integer ID and may
ask to be automatically assigned that ID.

8 files changed, 111 insertions, 21 deletions

diff --git a/include/KeyWords.h b/include/KeyWords.h
index d589d1228..43de26fe6 100644
--- a/include/KeyWords.h
+++ b/include/KeyWords.h
@@ -9,17 +9,30 @@ typedef void (*LexerFunction)(unsigned int startPos, int lengthDoc, int initStyl
                   WordList *keywordlists[], Accessor &styler);
                   
 /**
+ * A LexerModule is responsible for lexing and folding a particular language.
+ * The class maintains a list of LexerModules which can be searched to find a
+ * module appropriate to a particular language.
  */
 class LexerModule {
-	static LexerModule *base;
 	LexerModule *next;
 	int language;
-	LexerFunction fn;
+	const char *languageName;
+	LexerFunction fnLexer;
+	LexerFunction fnFolder;
+	
+	static LexerModule *base;
+	static int nextLanguage;
 
 public:
-	LexerModule(int language_, LexerFunction fn_);
-	static void Colourise(unsigned int startPos, int lengthDoc, int initStyle,
-                  int language, WordList *keywordlists[], Accessor &styler);
+	LexerModule(int language_, LexerFunction fnLexer_, 
+		const char *languageName_=0, LexerFunction fnFolder_=0);
+	int GetLanguage() { return language; }
+	void Lex(unsigned int startPos, int lengthDoc, int initStyle,
+                  WordList *keywordlists[], Accessor &styler);
+	void Fold(unsigned int startPos, int lengthDoc, int initStyle,
+                  WordList *keywordlists[], Accessor &styler);
+	static LexerModule *Find(int language);
+	static LexerModule *Find(const char *languageName);
 };
 
 /**
diff --git a/include/SciLexer.h b/include/SciLexer.h
index 88e8b954b..9274b31de 100644
--- a/include/SciLexer.h
+++ b/include/SciLexer.h
@@ -35,6 +35,7 @@
 #define SCLEX_PASCAL 18
 #define SCLEX_AVE 19
 #define SCLEX_ADA 20
+#define SCLEX_AUTOMATIC 1000
 #define SCE_P_DEFAULT 0
 #define SCE_P_COMMENTLINE 1
 #define SCE_P_NUMBER 2
diff --git a/include/Scintilla.h b/include/Scintilla.h
index fa1003ab4..54c02de47 100644
--- a/include/Scintilla.h
+++ b/include/Scintilla.h
@@ -411,6 +411,7 @@ typedef sptr_t (*SciFnDirect)(sptr_t ptr, unsigned int iMessage, uptr_t wParam,
 #define SCI_COLOURISE 4003
 #define SCI_SETPROPERTY 4004
 #define SCI_SETKEYWORDS 4005
+#define SCI_SETLEXERLANGUAGE 4006
 #define SC_MOD_INSERTTEXT 0x1
 #define SC_MOD_DELETETEXT 0x2
 #define SC_MOD_CHANGESTYLE 0x4
diff --git a/include/Scintilla.iface b/include/Scintilla.iface
index 97f183ffc..be9468184 100644
--- a/include/Scintilla.iface
+++ b/include/Scintilla.iface
@@ -1102,6 +1102,9 @@ set void SetProperty=4004(string key, string value)
 # Set up the key words used by the lexer.
 set void SetKeyWords=4005(int keywordSet, string keyWords)
 
+# Set the lexing language of the document based on string name.
+set void SetLexerLanguage=4006(, string language)
+
 # Notifications
 # Type of modification and the action which caused the modification
 # These are defined as a bit mask to make it easy to specify which notifications are wanted.
@@ -1174,6 +1177,9 @@ val SCLEX_CONF=17
 val SCLEX_PASCAL=18
 val SCLEX_AVE=19
 val SCLEX_ADA=20
+# When a lexer specifies its language as SCLEX_AUTOMATIC it receives a 
+# value assigned in sequence from SCLEX_AUTOMATIC+1.
+val SCLEX_AUTOMATIC=1000
 # Lexical states for SCLEX_PYTHON
 val SCE_P_DEFAULT=0
 val SCE_P_COMMENTLINE=1
diff --git a/src/KeyWords.cxx b/src/KeyWords.cxx
index a275c1963..f28fcecfb 100644
--- a/src/KeyWords.cxx
+++ b/src/KeyWords.cxx
@@ -20,32 +20,70 @@
 #include "SciLexer.h"
 
 LexerModule *LexerModule::base = 0;
+int LexerModule::nextLanguage = SCLEX_AUTOMATIC+1;
 
-LexerModule::LexerModule(int language_, LexerFunction fn_) :
-	language(language_), fn(fn_) {
+LexerModule::LexerModule(int language_, LexerFunction fnLexer_,
+	const char *languageName_, LexerFunction fnFolder_) :
+	language(language_), 
+	languageName(languageName_), 
+	fnLexer(fnLexer_), 
+	fnFolder(fnFolder_) {
 	next = base;
 	base = this;
+	if (language == SCLEX_AUTOMATIC) {
+		language = nextLanguage;
+		nextLanguage++;
+	}
 }
 
-void LexerModule::Colourise(unsigned int startPos, int lengthDoc, int initStyle,
-		int language, WordList *keywordlists[], Accessor &styler) {
+LexerModule *LexerModule::Find(int language) {
 	LexerModule *lm = base;
 	while (lm) {
 		if (lm->language == language) {
-			lm->fn(startPos, lengthDoc, initStyle, keywordlists, styler);
-			return;
+			return lm;
 		}
 		lm = lm->next;
 	}
-	// Unknown language
+	return 0;
+}
+
+LexerModule *LexerModule::Find(const char *languageName) {
+	if (languageName) {
+		LexerModule *lm = base;
+		while (lm) {
+			if (lm->languageName && 0 == strcmp(lm->languageName, languageName)) {
+				return lm;
+			}
+			lm = lm->next;
+		}
+	}
+	return 0;
+}
+
+void LexerModule::Lex(unsigned int startPos, int lengthDoc, int initStyle,
+	  WordList *keywordlists[], Accessor &styler) {
+	if (fnLexer)
+		fnLexer(startPos, lengthDoc, initStyle, keywordlists, styler);
+}
+
+void LexerModule::Fold(unsigned int startPos, int lengthDoc, int initStyle,
+	  WordList *keywordlists[], Accessor &styler) {
+	if (fnFolder)
+		fnFolder(startPos, lengthDoc, initStyle, keywordlists, styler);
+}
+
+static void ColouriseNullDoc(unsigned int startPos, int length, int, WordList *[],
+                            Accessor &styler) {
 	// Null language means all style bytes are 0 so just mark the end - no need to fill in.
-	if (lengthDoc > 0) {
-		styler.StartAt(startPos + lengthDoc - 1);
-		styler.StartSegment(startPos + lengthDoc - 1);
-		styler.ColourTo(startPos + lengthDoc - 1, 0);
+	if (length > 0) {
+		styler.StartAt(startPos + length - 1);
+		styler.StartSegment(startPos + length - 1);
+		styler.ColourTo(startPos + length - 1, 0);
 	}
 }
 
+LexerModule lmNull(SCLEX_NULL, ColouriseNullDoc, "null");
+
 #ifdef __vms
 
 // The following code forces a reference to all of the Scintilla lexers.
@@ -55,6 +93,7 @@ void LexerModule::Colourise(unsigned int startPos, int lengthDoc, int initStyle,
 // Taken from wxWindow's stc.cpp. Walter.
 
 int wxForceScintillaLexers(void) {
+  extern LexerModule lmAda;
   extern LexerModule lmCPP;
   extern LexerModule lmHTML;
   extern LexerModule lmXML;
@@ -68,7 +107,8 @@ int wxForceScintillaLexers(void) {
   extern LexerModule lmVB;
 
   if (
-      &lmCPP
+      &lmAda
+      && &lmCPP
       && &lmHTML
       && &lmXML
       && &lmProps
diff --git a/src/LexCPP.cxx b/src/LexCPP.cxx
index 966601ed3..826ac259f 100644
--- a/src/LexCPP.cxx
+++ b/src/LexCPP.cxx
@@ -328,7 +328,6 @@ static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, Wo
 	}
 	styler.ColourTo(lengthDoc - 1, state);
 	styler.Flush();
-	FoldCppDoc(startPos, length, initStyle, keywordlists, styler);
 }
 
-LexerModule lmCPP(SCLEX_CPP, ColouriseCppDoc);
+LexerModule lmCPP(SCLEX_CPP, ColouriseCppDoc, "cpp", FoldCppDoc);
diff --git a/src/ScintillaBase.cxx b/src/ScintillaBase.cxx
index fd2bef184..4eb0cf7a0 100644
--- a/src/ScintillaBase.cxx
+++ b/src/ScintillaBase.cxx
@@ -38,6 +38,7 @@ ScintillaBase::ScintillaBase() {
 	listType = 0;
 #ifdef SCI_LEXER
 	lexLanguage = SCLEX_CONTAINER;
+	lexCurrent = 0;
 	for (int wl = 0;wl < numWordLists;wl++)
 		keyWordLists[wl] = new WordList;
 #endif
@@ -338,6 +339,22 @@ void ScintillaBase::ButtonDown(Point pt, unsigned int curTime, bool shift, bool
 }
 
 #ifdef SCI_LEXER
+void ScintillaBase::SetLexer(uptr_t wParam) {
+	lexLanguage = wParam;
+	lexCurrent = LexerModule::Find(lexLanguage);
+	if (!lexCurrent)
+		lexCurrent = LexerModule::Find(SCLEX_NULL);
+}
+
+void ScintillaBase::SetLexerLanguage(const char *languageName) {
+	lexLanguage = SCLEX_CONTAINER;
+	lexCurrent = LexerModule::Find(languageName);
+	if (!lexCurrent)
+		lexCurrent = LexerModule::Find(SCLEX_NULL);
+	if (lexCurrent)
+		lexLanguage = lexCurrent->GetLanguage();
+}
+
 void ScintillaBase::Colourise(int start, int end) {
 	int lengthDoc = pdoc->Length();
 	if (end == -1)
@@ -352,8 +369,12 @@ void ScintillaBase::Colourise(int start, int end) {
 		styleStart = styler.StyleAt(start - 1);
 	styler.SetCodePage(pdoc->dbcsCodePage);
 
-	LexerModule::Colourise(start, len, styleStart, lexLanguage, keyWordLists, styler);
-	styler.Flush();
+	if (lexCurrent) {	// Should always succeed as null lexer should always be available
+		lexCurrent->Lex(start, len, styleStart, keyWordLists, styler);
+		styler.Flush();
+		lexCurrent->Fold(start, len, styleStart, keyWordLists, styler);
+		styler.Flush();
+	}
 }
 #endif
 
@@ -482,6 +503,7 @@ sptr_t ScintillaBase::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lPara
 
 #ifdef SCI_LEXER
 	case SCI_SETLEXER:
+		SetLexer(wParam);
 		lexLanguage = wParam;
 		break;
 
@@ -504,6 +526,11 @@ sptr_t ScintillaBase::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lPara
 			keyWordLists[wParam]->Set(reinterpret_cast<const char *>(lParam));
 		}
 		break;
+
+	case SCI_SETLEXERLANGUAGE:
+		SetLexerLanguage(reinterpret_cast<const char *>(lParam));
+		break;
+
 #endif
 
 	default:
diff --git a/src/ScintillaBase.h b/src/ScintillaBase.h
index 758f75a4c..13172031a 100644
--- a/src/ScintillaBase.h
+++ b/src/ScintillaBase.h
@@ -40,9 +40,12 @@ protected:
 	
 #ifdef SCI_LEXER
 	int lexLanguage;
+	LexerModule *lexCurrent;
 	PropSet props;
 	enum {numWordLists=5};
 	WordList *keyWordLists[numWordLists];
+	void SetLexer(uptr_t wParam);
+	void SetLexerLanguage(const char *languageName);
 	void Colourise(int start, int end);
 #endif