Lexer for SML. Feature Request #2710950.

12 files changed, 277 insertions, 3 deletions

diff --git a/doc/ScintillaHistory.html b/doc/ScintillaHistory.html
index 7d57921ec..9c3f53f9b 100644
--- a/doc/ScintillaHistory.html
+++ b/doc/ScintillaHistory.html
@@ -329,6 +329,9 @@
 	<td>Sergey Kishchenko</td>
 	<td>Kai Liu</td>
 	<td>Andreas Rumpf</td>
+	<td>James Moffatt</td>
+      </tr><tr>
+	<td>Yuzhou Xin</td>
     </tr>
     </table>
     <p>
diff --git a/gtk/makefile b/gtk/makefile
index a22250976..d7c05a6f7 100644
--- a/gtk/makefile
+++ b/gtk/makefile
@@ -72,7 +72,7 @@ LexLout.o LexLua.o LexMagik.o LexMatlab.o LexMetapost.o LexMMIXAL.o LexMPT.o \
 LexMSSQL.o LexMySQL.o LexNimrod.o LexNsis.o LexOpal.o LexOthers.o LexPascal.o \
 LexPB.o LexPerl.o LexPLM.o LexPOV.o LexPowerPro.o LexPowerShell.o \
 LexProgress.o LexPS.o LexPython.o LexR.o LexRebol.o LexRuby.o LexScriptol.o \
-LexSmalltalk.o LexSorcus.o LexSpecman.o LexSpice.o LexSQL.o LexTACL.o \
+LexSmalltalk.o LexSML.o LexSorcus.o LexSpecman.o LexSpice.o LexSQL.o LexTACL.o \
 LexTADS3.o LexTAL.o LexTCL.o LexTeX.o LexVB.o LexVerilog.o LexVHDL.o LexYAML.o
 #--Autogenerated -- end of automatically generated section
 
diff --git a/gtk/scintilla.mak b/gtk/scintilla.mak
index 88d795502..c4924ed6a 100644
--- a/gtk/scintilla.mak
+++ b/gtk/scintilla.mak
@@ -206,6 +206,7 @@ LEXOBJS=\
 	$(DIR_O)\LexRuby.obj \
 	$(DIR_O)\LexScriptol.obj \
 	$(DIR_O)\LexSmalltalk.obj \
+	$(DIR_O)\LexSML.obj \
 	$(DIR_O)\LexSorcus.obj \
 	$(DIR_O)\LexSpecman.obj \
 	$(DIR_O)\LexSpice.obj \
@@ -487,6 +488,8 @@ $(DIR_O)\LexScriptol.obj: ..\src\LexScriptol.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexSmalltalk.obj: ..\src\LexSmalltalk.cxx $(LEX_HEADERS)
 
+$(DIR_O)\LexSML.obj: ..\src\LexSML.cxx $(LEX_HEADERS)
+
 $(DIR_O)\LexSorcus.obj: ..\src\LexSorcus.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexSpecman.obj: ..\src\LexSpecman.cxx $(LEX_HEADERS)
diff --git a/include/SciLexer.h b/include/SciLexer.h
index 3a06b0d5a..56c9cbdc5 100644
--- a/include/SciLexer.h
+++ b/include/SciLexer.h
@@ -109,6 +109,7 @@
 #define SCLEX_SORCUS 94
 #define SCLEX_POWERPRO 95
 #define SCLEX_NIMROD 96
+#define SCLEX_SML 97
 #define SCLEX_AUTOMATIC 1000
 #define SCE_P_DEFAULT 0
 #define SCE_P_COMMENTLINE 1
@@ -1320,6 +1321,21 @@
 #define SCE_POWERPRO_VERBATIM 14
 #define SCE_POWERPRO_ALTQUOTE 15
 #define SCE_POWERPRO_FUNCTION 16
+#define SCE_SML_DEFAULT 0
+#define SCE_SML_IDENTIFIER 1
+#define SCE_SML_TAGNAME 2
+#define SCE_SML_KEYWORD 3
+#define SCE_SML_KEYWORD2 4
+#define SCE_SML_KEYWORD3 5
+#define SCE_SML_LINENUM 6
+#define SCE_SML_OPERATOR 7
+#define SCE_SML_NUMBER 8
+#define SCE_SML_CHAR 9
+#define SCE_SML_STRING 11
+#define SCE_SML_COMMENT 12
+#define SCE_SML_COMMENT1 13
+#define SCE_SML_COMMENT2 14
+#define SCE_SML_COMMENT3 15
 #define SCLEX_ASP 29
 #define SCLEX_PHP 30
 /* --Autogenerated -- end of section automatically generated from Scintilla.iface */
diff --git a/include/Scintilla.iface b/include/Scintilla.iface
index 638cb7700..f13c144a9 100644
--- a/include/Scintilla.iface
+++ b/include/Scintilla.iface
@@ -2021,6 +2021,7 @@ val SCLEX_TACL=93
 val SCLEX_SORCUS=94
 val SCLEX_POWERPRO=95
 val SCLEX_NIMROD=96
+val SCLEX_SML=97
 
 # When a lexer specifies its language as SCLEX_AUTOMATIC it receives a
 # value assigned in sequence from SCLEX_AUTOMATIC+1.
@@ -3389,6 +3390,7 @@ val SCE_SORCUS_OPERATOR=7
 val SCE_SORCUS_NUMBER=8
 val SCE_SORCUS_CONSTANT=9
 # Lexical state for SCLEX_POWERPRO
+lex PowerPro=SCLEX_POWERPRO SCE_POWERPRO_
 val SCE_POWERPRO_DEFAULT=0
 val SCE_POWERPRO_COMMENTBLOCK=1
 val SCE_POWERPRO_COMMENTLINE=2
@@ -3406,6 +3408,23 @@ val SCE_POWERPRO_STRINGEOL=13
 val SCE_POWERPRO_VERBATIM=14
 val SCE_POWERPRO_ALTQUOTE=15
 val SCE_POWERPRO_FUNCTION=16
+# Lexical states for SCLEX_SML
+lex SML=SCLEX_SML SCE_SML_
+val SCE_SML_DEFAULT=0
+val SCE_SML_IDENTIFIER=1
+val SCE_SML_TAGNAME=2
+val SCE_SML_KEYWORD=3
+val SCE_SML_KEYWORD2=4
+val SCE_SML_KEYWORD3=5
+val SCE_SML_LINENUM=6
+val SCE_SML_OPERATOR=7
+val SCE_SML_NUMBER=8
+val SCE_SML_CHAR=9
+val SCE_SML_STRING=11
+val SCE_SML_COMMENT=12
+val SCE_SML_COMMENT1=13
+val SCE_SML_COMMENT2=14
+val SCE_SML_COMMENT3=15
 
 # Events
 
diff --git a/macosx/makefile b/macosx/makefile
index 2c9d13282..462326965 100644
--- a/macosx/makefile
+++ b/macosx/makefile
@@ -72,7 +72,7 @@ LexLout.o LexLua.o LexMagik.o LexMatlab.o LexMetapost.o LexMMIXAL.o LexMPT.o \
 LexMSSQL.o LexMySQL.o LexNimrod.o LexNsis.o LexOpal.o LexOthers.o LexPascal.o \
 LexPB.o LexPerl.o LexPLM.o LexPOV.o LexPowerPro.o LexPowerShell.o \
 LexProgress.o LexPS.o LexPython.o LexR.o LexRebol.o LexRuby.o LexScriptol.o \
-LexSmalltalk.o LexSorcus.o LexSpecman.o LexSpice.o LexSQL.o LexTACL.o \
+LexSmalltalk.o LexSML.o LexSorcus.o LexSpecman.o LexSpice.o LexSQL.o LexTACL.o \
 LexTADS3.o LexTAL.o LexTCL.o LexTeX.o LexVB.o LexVerilog.o LexVHDL.o LexYAML.o
 #--Autogenerated -- end of automatically generated section
 
diff --git a/src/KeyWords.cxx b/src/KeyWords.cxx
index 5c8b1625f..f02511e5e 100644
--- a/src/KeyWords.cxx
+++ b/src/KeyWords.cxx
@@ -221,6 +221,7 @@ int Scintilla_LinkLexers() {
 	LINK_LEXER(lmRuby);
 	LINK_LEXER(lmScriptol);
 	LINK_LEXER(lmSmalltalk);
+	LINK_LEXER(lmSML);
 	LINK_LEXER(lmSorc);
 	LINK_LEXER(lmSpecman);
 	LINK_LEXER(lmSpice);
diff --git a/src/LexSML.cxx b/src/LexSML.cxx
new file mode 100644
index 000000000..dadfc0fb3
--- /dev/null
+++ b/src/LexSML.cxx
@@ -0,0 +1,222 @@
+// Scintilla source code edit control
+/** @file LexSML.cxx
+ ** Lexer for SML.
+ **/
+// Copyright 2009 by James Moffatt and Yuzhou Xin
+// 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"
+
+inline int  issml(int c) {return isalnum(c) || c == '_';}
+inline int issmlf(int c) {return isalpha(c) || c == '_';}
+inline int issmld(int c) {return isdigit(c) || c == '_';}
+
+
+#ifdef SCI_NAMESPACE
+using namespace Scintilla;
+#endif
+
+void ColouriseSMLDoc(
+	unsigned int startPos, int length,
+	int initStyle,
+	WordList *keywordlists[],
+	Accessor &styler)
+{
+	StyleContext sc(startPos, length, initStyle, styler);
+	int nesting = 0;
+	if (sc.state < SCE_SML_STRING)
+		sc.state = SCE_SML_DEFAULT;
+	if (sc.state >= SCE_SML_COMMENT)
+		nesting = (sc.state & 0x0f) - SCE_SML_COMMENT;
+
+	int chBase = 0, chToken = 0, chLit = 0;
+	WordList& keywords  = *keywordlists[0];
+	WordList& keywords2 = *keywordlists[1];
+	WordList& keywords3 = *keywordlists[2];
+	const int useMagic = styler.GetPropertyInt("lexer.caml.magic", 0);
+
+	while (sc.More()) {
+		int state2 = -1;
+		int chColor = sc.currentPos - 1;
+		bool advance = true;
+
+		switch (sc.state & 0x0f) {
+		case SCE_SML_DEFAULT:
+			chToken = sc.currentPos;
+			if (issmlf(sc.ch))
+				state2 = SCE_SML_IDENTIFIER;
+			else if (sc.Match('`') && issmlf(sc.chNext))
+				state2 = SCE_SML_TAGNAME;
+			else if (sc.Match('#')&&isdigit(sc.chNext))
+					state2 = SCE_SML_LINENUM;
+			else if (sc.Match('#','\"')){
+					state2 = SCE_SML_CHAR,chLit = 0;
+					sc.Forward();
+					
+				}
+			else if (isdigit(sc.ch)) {
+				state2 = SCE_SML_NUMBER, chBase = 10;
+				if (sc.Match('0') && strchr("xX", sc.chNext))
+					chBase = 16, sc.Forward();}
+			else if (sc.Match('\"')&&sc.chPrev!='#')
+				state2 = SCE_SML_STRING;
+			else if (sc.Match('(', '*')){
+				state2 = SCE_SML_COMMENT,
+					sc.ch = ' ',
+					sc.Forward();}
+			else if (strchr("!~"
+					"=<>@^+-*/"
+					"()[];,:.#", sc.ch))
+				state2 = SCE_SML_OPERATOR;
+			break;
+
+		case SCE_SML_IDENTIFIER:
+			if (!(issml(sc.ch) || sc.Match('\''))) {
+				const int n = sc.currentPos - chToken;
+				if (n < 24) {
+					char t[24];
+					for (int i = -n; i < 0; i++)
+						t[n + i] = static_cast<char>(sc.GetRelative(i));
+					t[n] = '\0';
+					if ((n == 1 && sc.chPrev == '_') || keywords.InList(t))
+						sc.ChangeState(SCE_SML_KEYWORD);
+					else if (keywords2.InList(t))
+						sc.ChangeState(SCE_SML_KEYWORD2);
+					else if (keywords3.InList(t))
+						sc.ChangeState(SCE_SML_KEYWORD3);
+				}
+				state2 = SCE_SML_DEFAULT, advance = false;
+			}
+			break;
+
+		case SCE_SML_TAGNAME:
+			if (!(issml(sc.ch) || sc.Match('\'')))
+				state2 = SCE_SML_DEFAULT, advance = false;
+			break;
+
+		case SCE_SML_LINENUM:
+			if (!isdigit(sc.ch))
+				state2 = SCE_SML_DEFAULT, advance = false;
+			break;
+
+		case SCE_SML_OPERATOR: {
+			const char* o = 0;
+			if (issml(sc.ch) || isspace(sc.ch)
+				|| (o = strchr(")]};,\'\"`#", sc.ch),o)
+				|| !strchr("!$%&*+-./:<=>?@^|~", sc.ch)) {
+				if (o && strchr(")]};,", sc.ch)) {
+					if ((sc.Match(')') && sc.chPrev == '(')
+						|| (sc.Match(']') && sc.chPrev == '['))
+						sc.ChangeState(SCE_SML_KEYWORD);
+					chColor++;
+				} else
+					advance = false;
+				state2 = SCE_SML_DEFAULT;
+			}
+			break;
+		}
+
+		case SCE_SML_NUMBER:
+			if (issmld(sc.ch) || IsADigit(sc.ch, chBase))
+				break;
+			if ((sc.Match('l') || sc.Match('L') || sc.Match('n'))
+				&& (issmld(sc.chPrev) || IsADigit(sc.chPrev, chBase)))
+				break;
+			if (chBase == 10) {
+				if (sc.Match('.') && issmld(sc.chPrev))
+					break;
+				if ((sc.Match('e') || sc.Match('E'))
+					&& (issmld(sc.chPrev) || sc.chPrev == '.'))
+					break;
+				if ((sc.Match('+') || sc.Match('-'))
+					&& (sc.chPrev == 'e' || sc.chPrev == 'E'))
+					break;
+			}
+			state2 = SCE_SML_DEFAULT, advance = false;
+			break;
+
+		case SCE_SML_CHAR:
+			if (sc.Match('\\')) {
+				chLit = 1;
+				if (sc.chPrev == '\\')
+					sc.ch = ' ';
+			} else if ((sc.Match('\"') && sc.chPrev != '\\') || sc.atLineEnd) {
+				state2 = SCE_SML_DEFAULT;
+				chLit = 1;
+				if (sc.Match('\"'))
+					chColor++;
+				else
+					sc.ChangeState(SCE_SML_IDENTIFIER);
+			} else if (chLit < 1 && sc.currentPos - chToken >= 3)
+				sc.ChangeState(SCE_SML_IDENTIFIER), advance = false;
+			break;
+
+		case SCE_SML_STRING:
+			if (sc.Match('\\') && sc.chPrev == '\\')
+				sc.ch = ' ';
+			else if (sc.Match('\"') && sc.chPrev != '\\')
+				state2 = SCE_SML_DEFAULT, chColor++;
+			break;
+
+		case SCE_SML_COMMENT:
+		case SCE_SML_COMMENT1:
+		case SCE_SML_COMMENT2:
+		case SCE_SML_COMMENT3:
+			if (sc.Match('(', '*'))
+				state2 = sc.state + 1, chToken = sc.currentPos,
+					sc.ch = ' ',
+					sc.Forward(), nesting++;
+			else if (sc.Match(')') && sc.chPrev == '*') {
+				if (nesting)
+					state2 = (sc.state & 0x0f) - 1, chToken = 0, nesting--;
+				else
+					state2 = SCE_SML_DEFAULT;
+				chColor++;
+			} else if (useMagic && sc.currentPos - chToken == 4
+				&& sc.Match('c') && sc.chPrev == 'r' && sc.GetRelative(-2) == '@')
+				sc.state |= 0x10;
+			break;
+		}
+
+		if (state2 >= 0)
+			styler.ColourTo(chColor, sc.state), sc.ChangeState(state2);
+		if (advance)
+			sc.Forward();
+	}
+
+	sc.Complete();
+}
+
+void FoldSMLDoc(
+	unsigned int startPos, int length,
+	int initStyle,
+	WordList *keywordlists[],
+	Accessor &styler)
+{
+	//supress "not used" warnings
+	startPos || length || initStyle || keywordlists[0] || styler.Length();
+}
+
+static const char * const SMLWordListDesc[] = {
+	"Keywords",
+	"Keywords2",
+	"Keywords3",
+	0
+};
+
+LexerModule lmSML(SCLEX_SML, ColouriseSMLDoc, "SML", FoldSMLDoc, SMLWordListDesc);
+ 	  	 
diff --git a/vcbuild/SciLexer.dsp b/vcbuild/SciLexer.dsp
index 7e0de8a66..3c12f40de 100644
--- a/vcbuild/SciLexer.dsp
+++ b/vcbuild/SciLexer.dsp
@@ -398,6 +398,10 @@ SOURCE=..\src\LexSmalltalk.cxx
 # End Source File
 # Begin Source File
 
+SOURCE=..\src\LexSML.cxx
+# End Source File
+# Begin Source File
+
 SOURCE=..\src\LexSorcus.cxx
 # End Source File
 # Begin Source File
diff --git a/win32/makefile b/win32/makefile
index 697b3cdc5..986191e57 100644
--- a/win32/makefile
+++ b/win32/makefile
@@ -60,7 +60,7 @@ LexLout.o LexLua.o LexMagik.o LexMatlab.o LexMetapost.o LexMMIXAL.o LexMPT.o \
 LexMSSQL.o LexMySQL.o LexNimrod.o LexNsis.o LexOpal.o LexOthers.o LexPascal.o \
 LexPB.o LexPerl.o LexPLM.o LexPOV.o LexPowerPro.o LexPowerShell.o \
 LexProgress.o LexPS.o LexPython.o LexR.o LexRebol.o LexRuby.o LexScriptol.o \
-LexSmalltalk.o LexSorcus.o LexSpecman.o LexSpice.o LexSQL.o LexTACL.o \
+LexSmalltalk.o LexSML.o LexSorcus.o LexSpecman.o LexSpice.o LexSQL.o LexTACL.o \
 LexTADS3.o LexTAL.o LexTCL.o LexTeX.o LexVB.o LexVerilog.o LexVHDL.o LexYAML.o
 #--Autogenerated -- end of automatically generated section
 
diff --git a/win32/scintilla.mak b/win32/scintilla.mak
index 7f4831452..d7ff22dd8 100644
--- a/win32/scintilla.mak
+++ b/win32/scintilla.mak
@@ -176,6 +176,7 @@ LEXOBJS=\
 	$(DIR_O)\LexRuby.obj \
 	$(DIR_O)\LexScriptol.obj \
 	$(DIR_O)\LexSmalltalk.obj \
+	$(DIR_O)\LexSML.obj \
 	$(DIR_O)\LexSorcus.obj \
 	$(DIR_O)\LexSpecman.obj \
 	$(DIR_O)\LexSpice.obj \
@@ -438,6 +439,8 @@ $(DIR_O)\LexScriptol.obj: ..\src\LexScriptol.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexSmalltalk.obj: ..\src\LexSmalltalk.cxx $(LEX_HEADERS)
 
+$(DIR_O)\LexSML.obj: ..\src\LexSML.cxx $(LEX_HEADERS)
+
 $(DIR_O)\LexSorcus.obj: ..\src\LexSorcus.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexSpecman.obj: ..\src\LexSpecman.cxx $(LEX_HEADERS)
diff --git a/win32/scintilla_vc6.mak b/win32/scintilla_vc6.mak
index 29aba7347..1f24df3c8 100644
--- a/win32/scintilla_vc6.mak
+++ b/win32/scintilla_vc6.mak
@@ -178,6 +178,7 @@ LEXOBJS=\
 	$(DIR_O)\LexRuby.obj \
 	$(DIR_O)\LexScriptol.obj \
 	$(DIR_O)\LexSmalltalk.obj \
+	$(DIR_O)\LexSML.obj \
 	$(DIR_O)\LexSorcus.obj \
 	$(DIR_O)\LexSpecman.obj \
 	$(DIR_O)\LexSpice.obj \
@@ -440,6 +441,8 @@ $(DIR_O)\LexScriptol.obj: ..\src\LexScriptol.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexSmalltalk.obj: ..\src\LexSmalltalk.cxx $(LEX_HEADERS)
 
+$(DIR_O)\LexSML.obj: ..\src\LexSML.cxx $(LEX_HEADERS)
+
 $(DIR_O)\LexSorcus.obj: ..\src\LexSorcus.cxx $(LEX_HEADERS)
 
 $(DIR_O)\LexSpecman.obj: ..\src\LexSpecman.cxx $(LEX_HEADERS)