New lexer for Fortran.

10 files changed, 327 insertions, 4 deletions

diff --git a/doc/ScintillaHistory.html b/doc/ScintillaHistory.html
index 8a3d8f481..cb058cce4 100644
--- a/doc/ScintillaHistory.html
+++ b/doc/ScintillaHistory.html
@@ -108,6 +108,7 @@
 	<li>Josh Wingstrom</li>
 	<li>Bruce Dodson</li>
 	<li>Sergey Koshcheyev</li>
+	<li>Chuan-jian Shen</li>
     </ul>
     <p>
        Images used in GTK+ version
diff --git a/gtk/makefile b/gtk/makefile
index 14b3819ad..cc262eb71 100644
--- a/gtk/makefile
+++ b/gtk/makefile
@@ -43,8 +43,8 @@ endif
 #**LEXOBJS=\\\n\(\*.o \)
 LEXOBJS=\
 LexAda.o LexAsm.o LexAVE.o LexBaan.o LexBullant.o LexConf.o LexCPP.o \
-LexCrontab.o LexEiffel.o LexHTML.o LexLisp.o LexLua.o LexMatlab.o LexOthers.o \
-LexPascal.o LexPerl.o LexPython.o LexRuby.o LexSQL.o LexVB.o 
+LexCrontab.o LexEiffel.o LexFortran.o LexHTML.o LexLisp.o LexLua.o LexMatlab.o \
+LexOthers.o LexPascal.o LexPerl.o LexPython.o LexRuby.o LexSQL.o LexVB.o 
 #--Autogenerated -- end of automatically generated section
 
 # The LEXOBJS have to be treated specially as the functions in them are not called from external code
diff --git a/gtk/scintilla.mak b/gtk/scintilla.mak
index 96447587f..ed6ae7434 100644
--- a/gtk/scintilla.mak
+++ b/gtk/scintilla.mak
@@ -127,6 +127,7 @@ LEXOBJS=\
 	$(DIR_O)\LexCPP.obj \
 	$(DIR_O)\LexCrontab.obj \
 	$(DIR_O)\LexEiffel.obj \
+	$(DIR_O)\LexFortran.obj \
 	$(DIR_O)\LexHTML.obj \
 	$(DIR_O)\LexLisp.obj \
 	$(DIR_O)\LexLua.obj \
@@ -264,6 +265,8 @@ $(DIR_O)\LexCrontab.obj: ..\src\LexCrontab.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexEiffel.obj: ..\src\LexEiffel.cxx $(LEX_HEADERS)
 
+$(DIR_O)\LexFortran.obj: ..\src\LexFortran.cxx $(LEX_HEADERS)
+
 $(DIR_O)\LexHTML.obj: ..\src\LexHTML.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexLisp.obj: ..\src\LexLisp.cxx $(LEX_HEADERS)
diff --git a/include/SciLexer.h b/include/SciLexer.h
index 083809ce3..8137db8f3 100644
--- a/include/SciLexer.h
+++ b/include/SciLexer.h
@@ -50,6 +50,8 @@
 #define SCLEX_SCRIPTOL 33
 #define SCLEX_ASM 34
 #define SCLEX_CPPNOCASE 35
+#define SCLEX_FORTRAN 36
+#define SCLEX_F77 37
 #define SCLEX_AUTOMATIC 1000
 #define SCE_P_DEFAULT 0
 #define SCE_P_COMMENTLINE 1
@@ -408,6 +410,21 @@
 #define SCE_ASM_REGISTER 8
 #define SCE_ASM_DIRECTIVE 9
 #define SCE_ASM_DIRECTIVEOPERAND 10
+#define SCE_F_DEFAULT 0
+#define SCE_F_COMMENT 1
+#define SCE_F_NUMBER 2
+#define SCE_F_STRING1 3
+#define SCE_F_STRING2 4
+#define SCE_F_STRINGEOL 5
+#define SCE_F_OPERATOR 6
+#define SCE_F_IDENTIFIER 7
+#define SCE_F_WORD 8
+#define SCE_F_WORD2 9
+#define SCE_F_WORD3 10
+#define SCE_F_PREPROCESSOR 11
+#define SCE_F_OPERATOR2 12
+#define SCE_F_LABEL 13
+#define SCE_F_CONTINUATION 14
 //--Autogenerated -- end of section automatically generated from Scintilla.iface
 
 #endif
diff --git a/include/Scintilla.iface b/include/Scintilla.iface
index 31344aa6f..fbe531780 100644
--- a/include/Scintilla.iface
+++ b/include/Scintilla.iface
@@ -1462,6 +1462,8 @@ val SCLEX_MATLAB=32
 val SCLEX_SCRIPTOL=33
 val SCLEX_ASM=34
 val SCLEX_CPPNOCASE=35
+val SCLEX_FORTRAN=36
+val SCLEX_F77=37
 
 # When a lexer specifies its language as SCLEX_AUTOMATIC it receives a
 # value assigned in sequence from SCLEX_AUTOMATIC+1.
@@ -1888,6 +1890,24 @@ val SCE_ASM_MATHINSTRUCTION=7
 val SCE_ASM_REGISTER=8
 val SCE_ASM_DIRECTIVE=9
 val SCE_ASM_DIRECTIVEOPERAND=10
+# Lexical states for SCLEX_FORTRAN
+lex Fortran=SCLEX_FORTRAN SCE_F_
+lex F77=SCLEX_F77 SCE_F_
+val SCE_F_DEFAULT=0
+val SCE_F_COMMENT=1
+val SCE_F_NUMBER=2
+val SCE_F_STRING1=3
+val SCE_F_STRING2=4
+val SCE_F_STRINGEOL=5
+val SCE_F_OPERATOR=6
+val SCE_F_IDENTIFIER=7
+val SCE_F_WORD=8
+val SCE_F_WORD2=9
+val SCE_F_WORD3=10
+val SCE_F_PREPROCESSOR=11
+val SCE_F_OPERATOR2=12
+val SCE_F_LABEL=13
+val SCE_F_CONTINUATION=14
 
 # Events
 
diff --git a/src/KeyWords.cxx b/src/KeyWords.cxx
index 7f77f2fad..d9682b606 100644
--- a/src/KeyWords.cxx
+++ b/src/KeyWords.cxx
@@ -153,6 +153,8 @@ int Scintilla_LinkLexers() {
 	LINK_LEXER(lmNncrontab);
 	LINK_LEXER(lmEiffel);
 	LINK_LEXER(lmEiffelkw);
+	LINK_LEXER(lmFortran);
+	LINK_LEXER(lmF77);
 	LINK_LEXER(lmHTML);
 	LINK_LEXER(lmXML);
 	LINK_LEXER(lmASP);
diff --git a/src/LexFortran.cxx b/src/LexFortran.cxx
new file mode 100644
index 000000000..85e84c58a
--- /dev/null
+++ b/src/LexFortran.cxx
@@ -0,0 +1,274 @@
+// Scintilla source code edit control
+/** @file LexFortran.cxx
+ ** Lexer for Fortran.
+ ** Writen by Chuan-jian Shen, Last changed Oct. 2002
+ **/
+// Copyright 1998-2001 by Neil Hodgson <neilh@scintilla.org>
+// The License.txt file describes the conditions under which this software may be distributed.
+
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+#include <stdio.h>
+#include <stdarg.h>
+
+#include "Platform.h"
+
+#include "PropSet.h"
+#include "Accessor.h"
+#include "StyleContext.h"
+#include "KeyWords.h"
+#include "Scintilla.h"
+#include "SciLexer.h"
+
+static inline bool IsAWordChar(const int ch) {
+	return (ch < 0x80) && (isalnum(ch) || ch == '_' || ch == '%');
+}
+
+static inline bool IsAWordStart(const int ch) {
+	return (ch < 0x80) && (isalnum(ch));
+}
+
+static void ColouriseFortranDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
+                            Accessor &styler, bool isFixFormat) {
+
+	WordList &keywords = *keywordlists[0];
+	WordList &keywords2 = *keywordlists[1];
+	WordList &keywords3 = *keywordlists[2];
+
+	int posLineStart = 0;
+
+	StyleContext sc(startPos, length, initStyle, styler);
+
+	for (; sc.More(); sc.Forward()) {
+
+		if (sc.atLineStart) {
+			posLineStart = sc.currentPos;
+			if (sc.state == SCE_F_STRING1) {
+			    // Prevent SCE_F_STRINGEOL from leaking back to previous line
+			    sc.SetState(SCE_F_STRING1);
+			}
+			else if(sc.state == SCE_F_STRING2) {
+				sc.SetState(SCE_F_STRING2);
+			}
+		}
+
+		// Handle line continuation generically.
+		if (sc.ch == '&') {
+			if (sc.chNext == '\n' || sc.chNext == '\r') {
+				sc.Forward();
+				if (sc.ch == '\r' && sc.chNext == '\n') {
+					sc.Forward();
+				}
+				continue;
+			}
+		}
+
+		// Determine if the current state should terminate.
+		if (sc.state == SCE_F_OPERATOR) {
+			sc.SetState(SCE_F_DEFAULT);
+		} else if (sc.state == SCE_F_NUMBER) {
+			if (!IsAWordChar(sc.ch)) {
+				sc.SetState(SCE_F_DEFAULT);
+			}
+		} else if (sc.state == SCE_F_IDENTIFIER) {
+			if (!IsAWordChar(sc.ch) || (sc.ch == '%')) {
+				char s[100];
+				sc.GetCurrentLowered(s, sizeof(s));
+				if (keywords.InList(s)) {
+					sc.ChangeState(SCE_F_WORD);
+				} else if (keywords2.InList(s)) {
+					sc.ChangeState(SCE_F_WORD2);
+				} else if (keywords3.InList(s)) {
+					sc.ChangeState(SCE_F_WORD3);
+				}
+				sc.SetState(SCE_F_DEFAULT);
+			}
+		} else if (sc.state == SCE_F_COMMENT) {
+			if (sc.ch == '\r' || sc.ch == '\n') {
+				sc.SetState(SCE_F_DEFAULT);
+			}
+		} else if (sc.state == SCE_F_STRING1) {
+			if (sc.ch == '&') {
+				sc.Forward();
+			} else if (sc.ch == '\'') {
+				sc.Forward();
+				if (sc.chNext == '\'') {
+					sc.Forward();
+				} else {
+					sc.SetState(SCE_F_DEFAULT);
+				}
+			} else if (sc.atLineEnd) {
+				sc.ChangeState(SCE_F_STRINGEOL);
+				sc.ForwardSetState(SCE_F_DEFAULT);
+			}
+		} else if (sc.state == SCE_F_STRING2) {
+			if (sc.atLineEnd) {
+				sc.ChangeState(SCE_F_STRINGEOL);
+				sc.ForwardSetState(SCE_F_DEFAULT);
+			} else if (sc.ch == '&') {
+				sc.Forward();
+			} else if (sc.ch == '\"') {
+				sc.Forward();
+				if (sc.chNext == '\"') {
+					sc.Forward();
+				} else {
+					sc.SetState(SCE_F_DEFAULT);
+				}
+			}
+		} else if (sc.state == SCE_F_OPERATOR2) {
+			if (sc.ch == '.') {
+				sc.ForwardSetState(SCE_F_DEFAULT);
+			}
+		} else if (sc.state == SCE_F_CONTINUATION) {
+			sc.SetState(SCE_F_DEFAULT);
+		} else if (sc.state == SCE_F_LABEL) {
+			if (sc.currentPos >= posLineStart+5) {
+				sc.SetState(SCE_F_DEFAULT);
+			}
+		}
+
+		// Determine if a new state should be entered.
+		if (sc.state == SCE_F_DEFAULT) {
+			int toLineStart = sc.currentPos - posLineStart;
+			if (isFixFormat && (toLineStart < 6 || toLineStart > 72)) {
+				if (sc.atLineStart && (tolower(sc.ch) == 'c' || sc.ch == '*')) {
+					sc.SetState(SCE_F_COMMENT);
+				} else if (toLineStart > 72) {
+					sc.SetState(SCE_F_COMMENT);
+				} else if (toLineStart < 5 && !IsASpace(sc.ch)) {
+					sc.SetState(SCE_F_LABEL);
+				} else if (toLineStart == 5 && (sc.ch != ' ' || sc.ch != '0')) {
+					sc.SetState(SCE_F_CONTINUATION);
+				}
+			} else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
+				sc.SetState(SCE_F_NUMBER);
+			} else if (sc.ch == '.' && isalpha(sc.chNext)) {
+				sc.SetState(SCE_F_OPERATOR2);
+			} else if (IsAWordStart(sc.ch)) {
+				sc.SetState(SCE_F_IDENTIFIER);
+			} else if (sc.ch == '!') {
+				sc.SetState(SCE_F_COMMENT);
+			} else if (sc.ch == '\"') {
+				sc.SetState(SCE_F_STRING2);
+			} else if (sc.ch == '\'') {
+				sc.SetState(SCE_F_STRING1);
+			} else if (isoperator(static_cast<char>(sc.ch))) {
+				sc.SetState(SCE_F_OPERATOR);
+			}
+		}
+	}
+	sc.Complete();
+}
+
+static int classifyFoldPointFortran(const char* s, const char* prevWord) {
+	int lev = 0;
+	if (strcmp(prevWord, "end") == 0) return lev;
+	if (strcmp(s, "associate") == 0 || strcmp(s, "block") == 0
+	    || strcmp(s, "blockdata") == 0 || strcmp(s, "case") == 0
+	    || strcmp(s, "do") == 0 || strcmp(s, "enum") ==0
+	    || strcmp(s, "forall") == 0 || strcmp(s, "function") == 0
+	    || strcmp(s, "interface") == 0 || strcmp(s, "module") == 0
+	    || strcmp(s, "program") == 0 || strcmp(s, "subroutine") == 0
+	    || strcmp(s, "then") == 0 || strcmp(s, "type") == 0
+	    || strcmp(s, "where") == 0) {
+		lev = 1;
+	} else if (strcmp(s, "end") == 0
+	    || strcmp(s, "endassociate") == 0 || strcmp(s, "endblock") == 0
+	    || strcmp(s, "endblockdata") == 0 || strcmp(s, "endselect") == 0
+	    || strcmp(s, "enddo") == 0 || strcmp(s, "endenum") ==0
+	    || strcmp(s, "endif") == 0
+	    || strcmp(s, "endforall") == 0 || strcmp(s, "endfunction") == 0
+	    || strcmp(s, "endinterface") == 0 || strcmp(s, "endmodule") == 0
+	    || strcmp(s, "endprogram") == 0 || strcmp(s, "endsubroutine") == 0
+	    || strcmp(s, "endtype") == 0 || strcmp(s, "endwhere") == 0) {
+		lev = -1;
+	}
+	return lev;
+}
+static void FoldFortranDoc(unsigned int startPos, int length, int initStyle, WordList *[], Accessor &styler) {
+	//~ bool foldComment = styler.GetPropertyInt("fold.comment") != 0;
+	// Do not know how to fold the comment at the moment.
+	bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
+	unsigned int endPos = startPos + length;
+	int visibleChars = 0;
+	int lineCurrent = styler.GetLine(startPos);
+	int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
+	int levelCurrent = levelPrev;
+	char chNext = styler[startPos];
+	int styleNext = styler.StyleAt(startPos);
+	int style = initStyle;
+
+	int lastStart = 0;
+	char prevWord[32] = "";
+
+	for (unsigned int i = startPos; i < endPos; i++) {
+		char ch = chNext;
+		chNext = styler.SafeGetCharAt(i + 1);
+		int stylePrev = style;
+		style = styleNext;
+		styleNext = styler.StyleAt(i + 1);
+		bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
+
+		if (stylePrev == SCE_F_DEFAULT && style == SCE_F_WORD)
+		{
+			// Store last word start point.
+			lastStart = i;
+		}
+
+		if (style == SCE_F_WORD) {
+			if(iswordchar(ch) && !iswordchar(chNext)) {
+				char s[32];
+				unsigned int j;
+				for(j = 0; ( j < 31 ) && ( j < i-lastStart+1 ); j++) {
+					s[j] = static_cast<char>(tolower(styler[lastStart + j]));
+				}
+				s[j] = '\0';
+				levelCurrent += classifyFoldPointFortran(s, prevWord);
+				strcpy(prevWord, s);
+			}
+		}
+		if (atEOL) {
+			int lev = levelPrev;
+			if (visibleChars == 0 && foldCompact)
+				lev |= SC_FOLDLEVELWHITEFLAG;
+			if ((levelCurrent > levelPrev) && (visibleChars > 0))
+				lev |= SC_FOLDLEVELHEADERFLAG;
+			if (lev != styler.LevelAt(lineCurrent)) {
+				styler.SetLevel(lineCurrent, lev);
+			}
+			lineCurrent++;
+			levelPrev = levelCurrent;
+			visibleChars = 0;
+			strcpy(prevWord, "");
+		}
+
+		if (!isspacechar(ch))
+			visibleChars++;
+	}
+
+	// Fill in the real level of the next line, keeping the current flags as they will be filled in later
+	int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
+	styler.SetLevel(lineCurrent, levelPrev | flagsNext);
+}
+
+static const char * const FortranWordLists[] = {
+	"Primary keywords and identifiers",
+	"Intrinsic functions",
+	"Extended and user defined functions",
+	0,
+};
+
+static void ColouriseFortranDocFreeFormat(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
+                            Accessor &styler) {
+	ColouriseFortranDoc(startPos, length, initStyle, keywordlists, styler, false);
+}
+
+static void ColouriseFortranDocFixFormat(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
+                            Accessor &styler) {
+	ColouriseFortranDoc(startPos, length, initStyle, keywordlists, styler, true);
+}
+
+
+LexerModule lmFortran(SCLEX_FORTRAN, ColouriseFortranDocFreeFormat, "fortran", FoldFortranDoc, FortranWordLists);
+LexerModule lmF77(SCLEX_F77, ColouriseFortranDocFixFormat, "f77", FoldFortranDoc, FortranWordLists);
diff --git a/win32/makefile b/win32/makefile
index c9097c4f2..cc066f919 100644
--- a/win32/makefile
+++ b/win32/makefile
@@ -41,8 +41,8 @@ deps:
 #**LEXOBJS=\\\n\(\*.o \)
 LEXOBJS=\
 LexAda.o LexAsm.o LexAVE.o LexBaan.o LexBullant.o LexConf.o LexCPP.o \
-LexCrontab.o LexEiffel.o LexHTML.o LexLisp.o LexLua.o LexMatlab.o LexOthers.o \
-LexPascal.o LexPerl.o LexPython.o LexRuby.o LexSQL.o LexVB.o 
+LexCrontab.o LexEiffel.o LexFortran.o LexHTML.o LexLisp.o LexLua.o LexMatlab.o \
+LexOthers.o LexPascal.o LexPerl.o LexPython.o LexRuby.o LexSQL.o LexVB.o 
 #--Autogenerated -- end of automatically generated section
 
 SOBJS	= ScintillaWin.o ScintillaBase.o Editor.o Document.o \
diff --git a/win32/scintilla.mak b/win32/scintilla.mak
index ce7c64707..9274c51f3 100644
--- a/win32/scintilla.mak
+++ b/win32/scintilla.mak
@@ -117,6 +117,7 @@ LEXOBJS=\
 	$(DIR_O)\LexCPP.obj \
 	$(DIR_O)\LexCrontab.obj \
 	$(DIR_O)\LexEiffel.obj \
+	$(DIR_O)\LexFortran.obj \
 	$(DIR_O)\LexHTML.obj \
 	$(DIR_O)\LexLisp.obj \
 	$(DIR_O)\LexLua.obj \
@@ -254,6 +255,8 @@ $(DIR_O)\LexCrontab.obj: ..\src\LexCrontab.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexEiffel.obj: ..\src\LexEiffel.cxx $(LEX_HEADERS)
 
+$(DIR_O)\LexFortran.obj: ..\src\LexFortran.cxx $(LEX_HEADERS)
+
 $(DIR_O)\LexHTML.obj: ..\src\LexHTML.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexLisp.obj: ..\src\LexLisp.cxx $(LEX_HEADERS)
diff --git a/win32/scintilla_vc6.mak b/win32/scintilla_vc6.mak
index 5b9a70d93..bb8e11b7a 100644
--- a/win32/scintilla_vc6.mak
+++ b/win32/scintilla_vc6.mak
@@ -119,6 +119,7 @@ LEXOBJS=\
 	$(DIR_O)\LexCPP.obj \
 	$(DIR_O)\LexCrontab.obj \
 	$(DIR_O)\LexEiffel.obj \
+	$(DIR_O)\LexFortran.obj \
 	$(DIR_O)\LexHTML.obj \
 	$(DIR_O)\LexLisp.obj \
 	$(DIR_O)\LexLua.obj \
@@ -256,6 +257,8 @@ $(DIR_O)\LexCrontab.obj: ..\src\LexCrontab.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexEiffel.obj: ..\src\LexEiffel.cxx $(LEX_HEADERS)
 
+$(DIR_O)\LexFortran.obj: ..\src\LexFortran.cxx $(LEX_HEADERS)
+
 $(DIR_O)\LexHTML.obj: ..\src\LexHTML.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexLisp.obj: ..\src\LexLisp.cxx $(LEX_HEADERS)