[wxWidgets.git] / src / stc / scintilla / lexers / LexCaml.cxx

// Scintilla source code edit control
/** @file LexCaml.cxx
 ** Lexer for Objective Caml.
 **/
// Copyright 2005-2009 by Robert Roessler <robertr@rftp.com>
// The License.txt file describes the conditions under which this software may be distributed.
/*	Release History
	20050204 Initial release.
	20050205 Quick compiler standards/"cleanliness" adjustment.
	20050206 Added cast for IsLeadByte().
	20050209 Changes to "external" build support.
	20050306 Fix for 1st-char-in-doc "corner" case.
	20050502 Fix for [harmless] one-past-the-end coloring.
	20050515 Refined numeric token recognition logic.
	20051125 Added 2nd "optional" keywords class.
	20051129 Support "magic" (read-only) comments for RCaml.
	20051204 Swtich to using StyleContext infrastructure.
	20090629 Add full Standard ML '97 support.
*/

#include <stdlib.h>
#include <string.h>
#include <stdio.h>
#include <stdarg.h>
#include <assert.h>
#include <ctype.h>

#include "ILexer.h"
#include "Scintilla.h"
#include "SciLexer.h"

#include "PropSetSimple.h"
#include "WordList.h"
#include "LexAccessor.h"
#include "Accessor.h"
#include "StyleContext.h"
#include "CharacterSet.h"
#include "LexerModule.h"

//	Since the Microsoft __iscsym[f] funcs are not ANSI...
inline int  iscaml(int c) {return isalnum(c) || c == '_';}
inline int iscamlf(int c) {return isalpha(c) || c == '_';}

static const int baseT[24] = {
	0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,	/* A - L */
	0, 0, 8, 0, 0, 0, 0, 0, 0, 0, 0,16	/* M - X */
};

#ifdef SCI_NAMESPACE
using namespace Scintilla;
#endif

#ifdef BUILD_AS_EXTERNAL_LEXER
/*
	(actually seems to work!)
*/
#include <string>
#include "WindowAccessor.h"
#include "ExternalLexer.h"

#undef EXT_LEXER_DECL
#define EXT_LEXER_DECL __declspec( dllexport ) __stdcall

#if PLAT_WIN
#include <windows.h>
#endif

static void ColouriseCamlDoc(
	unsigned int startPos, int length,
	int initStyle,
	WordList *keywordlists[],
	Accessor &styler);

static void FoldCamlDoc(
	unsigned int startPos, int length,
	int initStyle,
	WordList *keywordlists[],
	Accessor &styler);

static void InternalLexOrFold(int lexOrFold, unsigned int startPos, int length,
	int initStyle, char *words[], WindowID window, char *props);

static const char* LexerName = "caml";

#ifdef TRACE
void Platform::DebugPrintf(const char *format, ...) {
	char buffer[2000];
	va_list pArguments;
	va_start(pArguments, format);
	vsprintf(buffer,format,pArguments);
	va_end(pArguments);
	Platform::DebugDisplay(buffer);
}
#else
void Platform::DebugPrintf(const char *, ...) {
}
#endif

bool Platform::IsDBCSLeadByte(int codePage, char ch) {
	return ::IsDBCSLeadByteEx(codePage, ch) != 0;
}

long Platform::SendScintilla(WindowID w, unsigned int msg, unsigned long wParam, long lParam) {
	return ::SendMessage(reinterpret_cast<HWND>(w), msg, wParam, lParam);
}

long Platform::SendScintillaPointer(WindowID w, unsigned int msg, unsigned long wParam, void *lParam) {
	return ::SendMessage(reinterpret_cast<HWND>(w), msg, wParam,
		reinterpret_cast<LPARAM>(lParam));
}

void EXT_LEXER_DECL Fold(unsigned int lexer, unsigned int startPos, int length,
	int initStyle, char *words[], WindowID window, char *props)
{
	// below useless evaluation(s) to supress "not used" warnings
	lexer;
	// build expected data structures and do the Fold
	InternalLexOrFold(1, startPos, length, initStyle, words, window, props);

}

int EXT_LEXER_DECL GetLexerCount()
{
	return 1;	// just us [Objective] Caml lexers here!
}

void EXT_LEXER_DECL GetLexerName(unsigned int Index, char *name, int buflength)
{
	// below useless evaluation(s) to supress "not used" warnings
	Index;
	// return as much of our lexer name as will fit (what's up with Index?)
	if (buflength > 0) {
		buflength--;
		int n = strlen(LexerName);
		if (n > buflength)
			n = buflength;
		memcpy(name, LexerName, n), name[n] = '\0';
	}
}

void EXT_LEXER_DECL Lex(unsigned int lexer, unsigned int startPos, int length,
	int initStyle, char *words[], WindowID window, char *props)
{
	// below useless evaluation(s) to supress "not used" warnings
	lexer;
	// build expected data structures and do the Lex
	InternalLexOrFold(0, startPos, length, initStyle, words, window, props);
}

static void InternalLexOrFold(int foldOrLex, unsigned int startPos, int length,
	int initStyle, char *words[], WindowID window, char *props)
{
	// create and initialize a WindowAccessor (including contained PropSet)
	PropSetSimple ps;
	ps.SetMultiple(props);
	WindowAccessor wa(window, ps);
	// create and initialize WordList(s)
	int nWL = 0;
	for (; words[nWL]; nWL++) ;	// count # of WordList PTRs needed
	WordList** wl = new WordList* [nWL + 1];// alloc WordList PTRs
	int i = 0;
	for (; i < nWL; i++) {
		wl[i] = new WordList();	// (works or THROWS bad_alloc EXCEPTION)
		wl[i]->Set(words[i]);
	}
	wl[i] = 0;
	// call our "internal" folder/lexer (... then do Flush!)
	if (foldOrLex)
		FoldCamlDoc(startPos, length, initStyle, wl, wa);
	else
		ColouriseCamlDoc(startPos, length, initStyle, wl, wa);
	wa.Flush();
	// clean up before leaving
	for (i = nWL - 1; i >= 0; i--)
		delete wl[i];
	delete [] wl;
}

static
#endif	/* BUILD_AS_EXTERNAL_LEXER */

void ColouriseCamlDoc(
	unsigned int startPos, int length,
	int initStyle,
	WordList *keywordlists[],
	Accessor &styler)
{
	// initialize styler
	StyleContext sc(startPos, length, initStyle, styler);

	int chBase = 0, chToken = 0, chLit = 0;
	WordList& keywords  = *keywordlists[0];
	WordList& keywords2 = *keywordlists[1];
	WordList& keywords3 = *keywordlists[2];
	const bool isSML = keywords.InList("andalso");
	const int useMagic = styler.GetPropertyInt("lexer.caml.magic", 0);

	// set up [initial] state info (terminating states that shouldn't "bleed")
	const int state_ = sc.state & 0x0f;
	if (state_ <= SCE_CAML_CHAR
		|| (isSML && state_ == SCE_CAML_STRING))
		sc.state = SCE_CAML_DEFAULT;
	int nesting = (state_ >= SCE_CAML_COMMENT)? (state_ - SCE_CAML_COMMENT): 0;

	// foreach char in range...
	while (sc.More()) {
		// set up [per-char] state info
		int state2 = -1;				// (ASSUME no state change)
		int chColor = sc.currentPos - 1;// (ASSUME standard coloring range)
		bool advance = true;			// (ASSUME scanner "eats" 1 char)

		// step state machine
		switch (sc.state & 0x0f) {
		case SCE_CAML_DEFAULT:
			chToken = sc.currentPos;	// save [possible] token start (JIC)
			// it's wide open; what do we have?
			if (iscamlf(sc.ch))
				state2 = SCE_CAML_IDENTIFIER;
			else if (!isSML && sc.Match('`') && iscamlf(sc.chNext))
				state2 = SCE_CAML_TAGNAME;
			else if (!isSML && sc.Match('#') && isdigit(sc.chNext))
				state2 = SCE_CAML_LINENUM;
			else if (isdigit(sc.ch)) {
				// it's a number, assume base 10
				state2 = SCE_CAML_NUMBER, chBase = 10;
				if (sc.Match('0')) {
					// there MAY be a base specified...
					const char* baseC = "bBoOxX";
					if (isSML) {
						if (sc.chNext == 'w')
							sc.Forward();	// (consume SML "word" indicator)
						baseC = "x";
					}
					// ... change to specified base AS REQUIRED
					if (strchr(baseC, sc.chNext))
						chBase = baseT[tolower(sc.chNext) - 'a'], sc.Forward();
				}
			} else if (!isSML && sc.Match('\''))	// (Caml char literal?)
				state2 = SCE_CAML_CHAR, chLit = 0;
			else if (isSML && sc.Match('#', '"'))	// (SML char literal?)
				state2 = SCE_CAML_CHAR, sc.Forward();
			else if (sc.Match('"'))
				state2 = SCE_CAML_STRING;
			else if (sc.Match('(', '*'))
				state2 = SCE_CAML_COMMENT, sc.Forward(), sc.ch = ' '; // (*)...
			else if (strchr("!?~"			/* Caml "prefix-symbol" */
					"=<>@^|&+-*/$%"			/* Caml "infix-symbol" */
					"()[]{};,:.#", sc.ch)	// Caml "bracket" or ;,:.#
											// SML "extra" ident chars
				|| (isSML && (sc.Match('\\') || sc.Match('`'))))
				state2 = SCE_CAML_OPERATOR;
			break;

		case SCE_CAML_IDENTIFIER:
			// [try to] interpret as [additional] identifier char
			if (!(iscaml(sc.ch) || sc.Match('\''))) {
				const int n = sc.currentPos - chToken;
				if (n < 24) {
					// length is believable as keyword, [re-]construct token
					char t[24];
					for (int i = -n; i < 0; i++)
						t[n + i] = static_cast<char>(sc.GetRelative(i));
					t[n] = '\0';
					// special-case "_" token as KEYWORD
					if ((n == 1 && sc.chPrev == '_') || keywords.InList(t))
						sc.ChangeState(SCE_CAML_KEYWORD);
					else if (keywords2.InList(t))
						sc.ChangeState(SCE_CAML_KEYWORD2);
					else if (keywords3.InList(t))
						sc.ChangeState(SCE_CAML_KEYWORD3);
				}
				state2 = SCE_CAML_DEFAULT, advance = false;
			}
			break;

		case SCE_CAML_TAGNAME:
			// [try to] interpret as [additional] tagname char
			if (!(iscaml(sc.ch) || sc.Match('\'')))
				state2 = SCE_CAML_DEFAULT, advance = false;
			break;

		/*case SCE_CAML_KEYWORD:
		case SCE_CAML_KEYWORD2:
		case SCE_CAML_KEYWORD3:
			// [try to] interpret as [additional] keyword char
			if (!iscaml(ch))
				state2 = SCE_CAML_DEFAULT, advance = false;
			break;*/

		case SCE_CAML_LINENUM:
			// [try to] interpret as [additional] linenum directive char
			if (!isdigit(sc.ch))
				state2 = SCE_CAML_DEFAULT, advance = false;
			break;

		case SCE_CAML_OPERATOR: {
			// [try to] interpret as [additional] operator char
			const char* o = 0;
			if (iscaml(sc.ch) || isspace(sc.ch)			// ident or whitespace
				|| (o = strchr(")]};,\'\"#", sc.ch),o)	// "termination" chars
				|| (!isSML && sc.Match('`'))			// Caml extra term char
				|| (!strchr("!$%&*+-./:<=>?@^|~", sc.ch)// "operator" chars
														// SML extra ident chars
					&& !(isSML && (sc.Match('\\') || sc.Match('`'))))) {
				// check for INCLUSIVE termination
				if (o && strchr(")]};,", sc.ch)) {
					if ((sc.Match(')') && sc.chPrev == '(')
						|| (sc.Match(']') && sc.chPrev == '['))
						// special-case "()" and "[]" tokens as KEYWORDS
						sc.ChangeState(SCE_CAML_KEYWORD);
					chColor++;
				} else
					advance = false;
				state2 = SCE_CAML_DEFAULT;
			}
			break;
		}

		case SCE_CAML_NUMBER:
			// [try to] interpret as [additional] numeric literal char
			if ((!isSML && sc.Match('_')) || IsADigit(sc.ch, chBase))
				break;
			// how about an integer suffix?
			if (!isSML && (sc.Match('l') || sc.Match('L') || sc.Match('n'))
				&& (sc.chPrev == '_' || IsADigit(sc.chPrev, chBase)))
				break;
			// or a floating-point literal?
			if (chBase == 10) {
				// with a decimal point?
				if (sc.Match('.')
					&& ((!isSML && sc.chPrev == '_')
						|| IsADigit(sc.chPrev, chBase)))
					break;
				// with an exponent? (I)
				if ((sc.Match('e') || sc.Match('E'))
					&& ((!isSML && (sc.chPrev == '.' || sc.chPrev == '_'))
						|| IsADigit(sc.chPrev, chBase)))
					break;
				// with an exponent? (II)
				if (((!isSML && (sc.Match('+') || sc.Match('-')))
						|| (isSML && sc.Match('~')))
					&& (sc.chPrev == 'e' || sc.chPrev == 'E'))
					break;
			}
			// it looks like we have run out of number
			state2 = SCE_CAML_DEFAULT, advance = false;
			break;

		case SCE_CAML_CHAR:
			if (!isSML) {
				// [try to] interpret as [additional] char literal char
				if (sc.Match('\\')) {
					chLit = 1;	// (definitely IS a char literal)
					if (sc.chPrev == '\\')
						sc.ch = ' ';	// (...\\')
				// should we be terminating - one way or another?
				} else if ((sc.Match('\'') && sc.chPrev != '\\')
					|| sc.atLineEnd) {
					state2 = SCE_CAML_DEFAULT;
					if (sc.Match('\''))
						chColor++;
					else
						sc.ChangeState(SCE_CAML_IDENTIFIER);
				// ... maybe a char literal, maybe not
				} else if (chLit < 1 && sc.currentPos - chToken >= 2)
					sc.ChangeState(SCE_CAML_IDENTIFIER), advance = false;
				break;
			}/* else
				// fall through for SML char literal (handle like string) */

		case SCE_CAML_STRING:
			// [try to] interpret as [additional] [SML char/] string literal char
			if (isSML && sc.Match('\\') && sc.chPrev != '\\' && isspace(sc.chNext))
				state2 = SCE_CAML_WHITE;
			else if (sc.Match('\\') && sc.chPrev == '\\')
				sc.ch = ' ';	// (...\\")
			// should we be terminating - one way or another?
			else if ((sc.Match('"') && sc.chPrev != '\\')
				|| (isSML && sc.atLineEnd)) {
				state2 = SCE_CAML_DEFAULT;
				if (sc.Match('"'))
					chColor++;
			}
			break;

		case SCE_CAML_WHITE:
			// [try to] interpret as [additional] SML embedded whitespace char
			if (sc.Match('\\')) {
				// style this puppy NOW...
				state2 = SCE_CAML_STRING, sc.ch = ' ' /* (...\") */, chColor++,
					styler.ColourTo(chColor, SCE_CAML_WHITE), styler.Flush();
				// ... then backtrack to determine original SML literal type
				int p = chColor - 2;
				for (; p >= 0 && styler.StyleAt(p) == SCE_CAML_WHITE; p--) ;
				if (p >= 0)
					state2 = static_cast<int>(styler.StyleAt(p));
				// take care of state change NOW
				sc.ChangeState(state2), state2 = -1;
			}
			break;

		case SCE_CAML_COMMENT:
		case SCE_CAML_COMMENT1:
		case SCE_CAML_COMMENT2:
		case SCE_CAML_COMMENT3:
			// we're IN a comment - does this start a NESTED comment?
			if (sc.Match('(', '*'))
				state2 = sc.state + 1, chToken = sc.currentPos,
					sc.Forward(), sc.ch = ' ' /* (*)... */, nesting++;
			// [try to] interpret as [additional] comment char
			else if (sc.Match(')') && sc.chPrev == '*') {
				if (nesting)
					state2 = (sc.state & 0x0f) - 1, chToken = 0, nesting--;
				else
					state2 = SCE_CAML_DEFAULT;
				chColor++;
			// enable "magic" (read-only) comment AS REQUIRED
			} else if (useMagic && sc.currentPos - chToken == 4
				&& sc.Match('c') && sc.chPrev == 'r' && sc.GetRelative(-2) == '@')
				sc.state |= 0x10;	// (switch to read-only comment style)
			break;
		}

		// handle state change and char coloring AS REQUIRED
		if (state2 >= 0)
			styler.ColourTo(chColor, sc.state), sc.ChangeState(state2);
		// move to next char UNLESS re-scanning current char
		if (advance)
			sc.Forward();
	}

	// do any required terminal char coloring (JIC)
	sc.Complete();
}

#ifdef BUILD_AS_EXTERNAL_LEXER
static
#endif	/* BUILD_AS_EXTERNAL_LEXER */
void FoldCamlDoc(
	unsigned int, int,
	int,
	WordList *[],
	Accessor &)
{
}

static const char * const camlWordListDesc[] = {
	"Keywords",		// primary Objective Caml keywords
	"Keywords2",	// "optional" keywords (typically from Pervasives)
	"Keywords3",	// "optional" keywords (typically typenames)
	0
};

#ifndef BUILD_AS_EXTERNAL_LEXER
LexerModule lmCaml(SCLEX_CAML, ColouriseCamlDoc, "caml", FoldCamlDoc, camlWordListDesc);
#endif	/* BUILD_AS_EXTERNAL_LEXER */
Commit	Line	Data
1e9bafca RD	1	// Scintilla source code edit control
	2	/** @file LexCaml.cxx
	3	** Lexer for Objective Caml.
	4	**/
9e96e16f	5	// Copyright 2005-2009 by Robert Roessler <robertr@rftp.com>
1e9bafca RD	6	// The License.txt file describes the conditions under which this software may be distributed.
	7	/* Release History
	8	20050204 Initial release.
	9	20050205 Quick compiler standards/"cleanliness" adjustment.
	10	20050206 Added cast for IsLeadByte().
	11	20050209 Changes to "external" build support.
	12	20050306 Fix for 1st-char-in-doc "corner" case.
	13	20050502 Fix for [harmless] one-past-the-end coloring.
	14	20050515 Refined numeric token recognition logic.
	15	20051125 Added 2nd "optional" keywords class.
	16	20051129 Support "magic" (read-only) comments for RCaml.
	17	20051204 Swtich to using StyleContext infrastructure.
9e96e16f	18	20090629 Add full Standard ML '97 support.
1e9bafca RD	19	*/
	20
	21	#include <stdlib.h>
	22	#include <string.h>
1e9bafca RD	23	#include <stdio.h>
1e9bafca RD	24	#include <stdarg.h>
1dcf666d RD	25	#include <assert.h>
1dcf666d RD	26	#include <ctype.h>
1e9bafca	27
1dcf666d RD	28	#include "ILexer.h"
	29	#include "Scintilla.h"
	30	#include "SciLexer.h"
1e9bafca	31
9e96e16f	32	#include "PropSetSimple.h"
1dcf666d RD	33	#include "WordList.h"
1dcf666d RD	34	#include "LexAccessor.h"
1e9bafca RD	35	#include "Accessor.h"
1e9bafca RD	36	#include "StyleContext.h"
1dcf666d RD	37	#include "CharacterSet.h"
1dcf666d RD	38	#include "LexerModule.h"
1e9bafca RD	39
	40	// Since the Microsoft __iscsym[f] funcs are not ANSI...
	41	inline int iscaml(int c) {return isalnum(c) \|\| c == '_';}
	42	inline int iscamlf(int c) {return isalpha(c) \|\| c == '_';}
1e9bafca RD	43
	44	static const int baseT[24] = {
	45	0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A - L */
	46	0, 0, 8, 0, 0, 0, 0, 0, 0, 0, 0,16 /* M - X */
	47	};
	48
7e0c58e9 RD	49	#ifdef SCI_NAMESPACE
	50	using namespace Scintilla;
	51	#endif
	52
1e9bafca RD	53	#ifdef BUILD_AS_EXTERNAL_LEXER
	54	/*
	55	(actually seems to work!)
	56	*/
1dcf666d	57	#include <string>
1e9bafca RD	58	#include "WindowAccessor.h"
	59	#include "ExternalLexer.h"
	60
1dcf666d RD	61	#undef EXT_LEXER_DECL
	62	#define EXT_LEXER_DECL __declspec( dllexport ) __stdcall
	63
1e9bafca RD	64	#if PLAT_WIN
	65	#include <windows.h>
	66	#endif
	67
	68	static void ColouriseCamlDoc(
	69	unsigned int startPos, int length,
	70	int initStyle,
	71	WordList *keywordlists[],
	72	Accessor &styler);
	73
	74	static void FoldCamlDoc(
	75	unsigned int startPos, int length,
	76	int initStyle,
	77	WordList *keywordlists[],
	78	Accessor &styler);
	79
	80	static void InternalLexOrFold(int lexOrFold, unsigned int startPos, int length,
	81	int initStyle, char words[], WindowID window, char props);
	82
	83	static const char* LexerName = "caml";
	84
	85	#ifdef TRACE
	86	void Platform::DebugPrintf(const char *format, ...) {
	87	char buffer[2000];
	88	va_list pArguments;
	89	va_start(pArguments, format);
	90	vsprintf(buffer,format,pArguments);
	91	va_end(pArguments);
	92	Platform::DebugDisplay(buffer);
	93	}
	94	#else
	95	void Platform::DebugPrintf(const char *, ...) {
	96	}
	97	#endif
	98
	99	bool Platform::IsDBCSLeadByte(int codePage, char ch) {
	100	return ::IsDBCSLeadByteEx(codePage, ch) != 0;
	101	}
	102
	103	long Platform::SendScintilla(WindowID w, unsigned int msg, unsigned long wParam, long lParam) {
	104	return ::SendMessage(reinterpret_cast<HWND>(w), msg, wParam, lParam);
	105	}
	106
	107	long Platform::SendScintillaPointer(WindowID w, unsigned int msg, unsigned long wParam, void *lParam) {
	108	return ::SendMessage(reinterpret_cast<HWND>(w), msg, wParam,
	109	reinterpret_cast<LPARAM>(lParam));
	110	}
	111
	112	void EXT_LEXER_DECL Fold(unsigned int lexer, unsigned int startPos, int length,
	113	int initStyle, char words[], WindowID window, char props)
	114	{
	115	// below useless evaluation(s) to supress "not used" warnings
	116	lexer;
	117	// build expected data structures and do the Fold
	118	InternalLexOrFold(1, startPos, length, initStyle, words, window, props);
	119
	120	}
	121
	122	int EXT_LEXER_DECL GetLexerCount()
	123	{
	124	return 1; // just us [Objective] Caml lexers here!
	125	}
	126
	127	void EXT_LEXER_DECL GetLexerName(unsigned int Index, char *name, int buflength)
128	{
129	// below useless evaluation(s) to supress "not used" warnings
130	Index;
131	// return as much of our lexer name as will fit (what's up with Index?)
132	if (buflength > 0) {
133	buflength--;
134	int n = strlen(LexerName);
135	if (n > buflength)
136	n = buflength;
137	memcpy(name, LexerName, n), name[n] = '\0';
138	}
139	}
140
141	void EXT_LEXER_DECL Lex(unsigned int lexer, unsigned int startPos, int length,
142	int initStyle, char words[], WindowID window, char props)
143	{
144	// below useless evaluation(s) to supress "not used" warnings
145	lexer;
146	// build expected data structures and do the Lex
147	InternalLexOrFold(0, startPos, length, initStyle, words, window, props);
148	}
149
150	static void InternalLexOrFold(int foldOrLex, unsigned int startPos, int length,
151	int initStyle, char words[], WindowID window, char props)
152	{
153	// create and initialize a WindowAccessor (including contained PropSet)
9e96e16f	154	PropSetSimple ps;
1e9bafca RD	155	ps.SetMultiple(props);
	156	WindowAccessor wa(window, ps);
	157	// create and initialize WordList(s)
	158	int nWL = 0;
	159	for (; words[nWL]; nWL++) ; // count # of WordList PTRs needed
	160	WordList** wl = new WordList* [nWL + 1];// alloc WordList PTRs
	161	int i = 0;
	162	for (; i < nWL; i++) {
	163	wl[i] = new WordList(); // (works or THROWS bad_alloc EXCEPTION)
	164	wl[i]->Set(words[i]);
	165	}
	166	wl[i] = 0;
	167	// call our "internal" folder/lexer (... then do Flush!)
	168	if (foldOrLex)
	169	FoldCamlDoc(startPos, length, initStyle, wl, wa);
	170	else
	171	ColouriseCamlDoc(startPos, length, initStyle, wl, wa);
	172	wa.Flush();
	173	// clean up before leaving
	174	for (i = nWL - 1; i >= 0; i--)
	175	delete wl[i];
	176	delete [] wl;
	177	}
	178
	179	static
	180	#endif /* BUILD_AS_EXTERNAL_LEXER */
	181
	182	void ColouriseCamlDoc(
	183	unsigned int startPos, int length,
	184	int initStyle,
	185	WordList *keywordlists[],
	186	Accessor &styler)
	187	{
	188	// initialize styler
	189	StyleContext sc(startPos, length, initStyle, styler);
1e9bafca RD	190
	191	int chBase = 0, chToken = 0, chLit = 0;
	192	WordList& keywords = *keywordlists[0];
	193	WordList& keywords2 = *keywordlists[1];
	194	WordList& keywords3 = *keywordlists[2];
9e96e16f	195	const bool isSML = keywords.InList("andalso");
1e9bafca RD	196	const int useMagic = styler.GetPropertyInt("lexer.caml.magic", 0);
1e9bafca RD	197
9e96e16f RD	198	// set up [initial] state info (terminating states that shouldn't "bleed")
	199	const int state_ = sc.state & 0x0f;
	200	if (state_ <= SCE_CAML_CHAR
	201	\|\| (isSML && state_ == SCE_CAML_STRING))
	202	sc.state = SCE_CAML_DEFAULT;
	203	int nesting = (state_ >= SCE_CAML_COMMENT)? (state_ - SCE_CAML_COMMENT): 0;
	204
1e9bafca RD	205	// foreach char in range...
	206	while (sc.More()) {
	207	// set up [per-char] state info
9e96e16f	208	int state2 = -1; // (ASSUME no state change)
1e9bafca	209	int chColor = sc.currentPos - 1;// (ASSUME standard coloring range)
9e96e16f	210	bool advance = true; // (ASSUME scanner "eats" 1 char)
1e9bafca RD	211
	212	// step state machine
	213	switch (sc.state & 0x0f) {
	214	case SCE_CAML_DEFAULT:
	215	chToken = sc.currentPos; // save [possible] token start (JIC)
	216	// it's wide open; what do we have?
	217	if (iscamlf(sc.ch))
	218	state2 = SCE_CAML_IDENTIFIER;
9e96e16f	219	else if (!isSML && sc.Match('`') && iscamlf(sc.chNext))
1e9bafca	220	state2 = SCE_CAML_TAGNAME;
9e96e16f	221	else if (!isSML && sc.Match('#') && isdigit(sc.chNext))
1e9bafca RD	222	state2 = SCE_CAML_LINENUM;
1e9bafca RD	223	else if (isdigit(sc.ch)) {
9e96e16f	224	// it's a number, assume base 10
1e9bafca	225	state2 = SCE_CAML_NUMBER, chBase = 10;
9e96e16f RD	226	if (sc.Match('0')) {
	227	// there MAY be a base specified...
	228	const char* baseC = "bBoOxX";
	229	if (isSML) {
	230	if (sc.chNext == 'w')
	231	sc.Forward(); // (consume SML "word" indicator)
	232	baseC = "x";
	233	}
	234	// ... change to specified base AS REQUIRED
	235	if (strchr(baseC, sc.chNext))
	236	chBase = baseT[tolower(sc.chNext) - 'a'], sc.Forward();
	237	}
	238	} else if (!isSML && sc.Match('\'')) // (Caml char literal?)
1e9bafca	239	state2 = SCE_CAML_CHAR, chLit = 0;
9e96e16f RD	240	else if (isSML && sc.Match('#', '"')) // (SML char literal?)
	241	state2 = SCE_CAML_CHAR, sc.Forward();
	242	else if (sc.Match('"'))
1e9bafca RD	243	state2 = SCE_CAML_STRING;
1e9bafca RD	244	else if (sc.Match('(', '*'))
9e96e16f RD	245	state2 = SCE_CAML_COMMENT, sc.Forward(), sc.ch = ' '; // (*)...
	246	else if (strchr("!?~" /* Caml "prefix-symbol" */
	247	"=<>@^\|&+-/$%" / Caml "infix-symbol" */
	248	"()[]{};,:.#", sc.ch) // Caml "bracket" or ;,:.#
	249	// SML "extra" ident chars
	250	\|\| (isSML && (sc.Match('\\') \|\| sc.Match('`'))))
1e9bafca RD	251	state2 = SCE_CAML_OPERATOR;
	252	break;
	253
	254	case SCE_CAML_IDENTIFIER:
	255	// [try to] interpret as [additional] identifier char
	256	if (!(iscaml(sc.ch) \|\| sc.Match('\''))) {
	257	const int n = sc.currentPos - chToken;
	258	if (n < 24) {
	259	// length is believable as keyword, [re-]construct token
	260	char t[24];
	261	for (int i = -n; i < 0; i++)
	262	t[n + i] = static_cast<char>(sc.GetRelative(i));
	263	t[n] = '\0';
	264	// special-case "_" token as KEYWORD
	265	if ((n == 1 && sc.chPrev == '_') \|\| keywords.InList(t))
	266	sc.ChangeState(SCE_CAML_KEYWORD);
	267	else if (keywords2.InList(t))
	268	sc.ChangeState(SCE_CAML_KEYWORD2);
	269	else if (keywords3.InList(t))
	270	sc.ChangeState(SCE_CAML_KEYWORD3);
	271	}
	272	state2 = SCE_CAML_DEFAULT, advance = false;
	273	}
	274	break;
	275
	276	case SCE_CAML_TAGNAME:
	277	// [try to] interpret as [additional] tagname char
	278	if (!(iscaml(sc.ch) \|\| sc.Match('\'')))
	279	state2 = SCE_CAML_DEFAULT, advance = false;
	280	break;
	281
	282	/*case SCE_CAML_KEYWORD:
	283	case SCE_CAML_KEYWORD2:
	284	case SCE_CAML_KEYWORD3:
	285	// [try to] interpret as [additional] keyword char
	286	if (!iscaml(ch))
	287	state2 = SCE_CAML_DEFAULT, advance = false;
	288	break;*/
	289
	290	case SCE_CAML_LINENUM:
	291	// [try to] interpret as [additional] linenum directive char
	292	if (!isdigit(sc.ch))
	293	state2 = SCE_CAML_DEFAULT, advance = false;
	294	break;
	295
	296	case SCE_CAML_OPERATOR: {
	297	// [try to] interpret as [additional] operator char
	298	const char* o = 0;
9e96e16f RD	299	if (iscaml(sc.ch) \|\| isspace(sc.ch) // ident or whitespace
	300	\|\| (o = strchr(")]};,\'\"#", sc.ch),o) // "termination" chars
	301	\|\| (!isSML && sc.Match('`')) // Caml extra term char
	302	\|\| (!strchr("!$%&*+-./:<=>?@^\|~", sc.ch)// "operator" chars
	303	// SML extra ident chars
	304	&& !(isSML && (sc.Match('\\') \|\| sc.Match('`'))))) {
1e9bafca RD	305	// check for INCLUSIVE termination
	306	if (o && strchr(")]};,", sc.ch)) {
	307	if ((sc.Match(')') && sc.chPrev == '(')
	308	\|\| (sc.Match(']') && sc.chPrev == '['))
	309	// special-case "()" and "[]" tokens as KEYWORDS
	310	sc.ChangeState(SCE_CAML_KEYWORD);
	311	chColor++;
	312	} else
	313	advance = false;
	314	state2 = SCE_CAML_DEFAULT;
	315	}
	316	break;
	317	}
	318
	319	case SCE_CAML_NUMBER:
	320	// [try to] interpret as [additional] numeric literal char
9e96e16f	321	if ((!isSML && sc.Match('_')) \|\| IsADigit(sc.ch, chBase))
1e9bafca RD	322	break;
1e9bafca RD	323	// how about an integer suffix?
9e96e16f RD	324	if (!isSML && (sc.Match('l') \|\| sc.Match('L') \|\| sc.Match('n'))
9e96e16f RD	325	&& (sc.chPrev == '_' \|\| IsADigit(sc.chPrev, chBase)))
1e9bafca RD	326	break;
	327	// or a floating-point literal?
	328	if (chBase == 10) {
	329	// with a decimal point?
9e96e16f RD	330	if (sc.Match('.')
	331	&& ((!isSML && sc.chPrev == '_')
	332	\|\| IsADigit(sc.chPrev, chBase)))
1e9bafca RD	333	break;
	334	// with an exponent? (I)
	335	if ((sc.Match('e') \|\| sc.Match('E'))
9e96e16f RD	336	&& ((!isSML && (sc.chPrev == '.' \|\| sc.chPrev == '_'))
9e96e16f RD	337	\|\| IsADigit(sc.chPrev, chBase)))
1e9bafca RD	338	break;
1e9bafca RD	339	// with an exponent? (II)
9e96e16f RD	340	if (((!isSML && (sc.Match('+') \|\| sc.Match('-')))
9e96e16f RD	341	\|\| (isSML && sc.Match('~')))
1e9bafca RD	342	&& (sc.chPrev == 'e' \|\| sc.chPrev == 'E'))
	343	break;
	344	}
	345	// it looks like we have run out of number
	346	state2 = SCE_CAML_DEFAULT, advance = false;
	347	break;
	348
	349	case SCE_CAML_CHAR:
9e96e16f RD	350	if (!isSML) {
	351	// [try to] interpret as [additional] char literal char
	352	if (sc.Match('\\')) {
	353	chLit = 1; // (definitely IS a char literal)
	354	if (sc.chPrev == '\\')
	355	sc.ch = ' '; // (...\\')
	356	// should we be terminating - one way or another?
	357	} else if ((sc.Match('\'') && sc.chPrev != '\\')
	358	\|\| sc.atLineEnd) {
	359	state2 = SCE_CAML_DEFAULT;
	360	if (sc.Match('\''))
	361	chColor++;
	362	else
	363	sc.ChangeState(SCE_CAML_IDENTIFIER);
	364	// ... maybe a char literal, maybe not
	365	} else if (chLit < 1 && sc.currentPos - chToken >= 2)
	366	sc.ChangeState(SCE_CAML_IDENTIFIER), advance = false;
	367	break;
	368	}/* else
	369	// fall through for SML char literal (handle like string) */
	370
	371	case SCE_CAML_STRING:
	372	// [try to] interpret as [additional] [SML char/] string literal char
	373	if (isSML && sc.Match('\\') && sc.chPrev != '\\' && isspace(sc.chNext))
	374	state2 = SCE_CAML_WHITE;
	375	else if (sc.Match('\\') && sc.chPrev == '\\')
	376	sc.ch = ' '; // (...\\")
1e9bafca	377	// should we be terminating - one way or another?
9e96e16f RD	378	else if ((sc.Match('"') && sc.chPrev != '\\')
9e96e16f RD	379	\|\| (isSML && sc.atLineEnd)) {
1e9bafca	380	state2 = SCE_CAML_DEFAULT;
9e96e16f	381	if (sc.Match('"'))
1e9bafca	382	chColor++;
9e96e16f	383	}
1e9bafca RD	384	break;
1e9bafca RD	385
9e96e16f RD	386	case SCE_CAML_WHITE:
	387	// [try to] interpret as [additional] SML embedded whitespace char
	388	if (sc.Match('\\')) {
	389	// style this puppy NOW...
	390	state2 = SCE_CAML_STRING, sc.ch = ' ' /* (...\") */, chColor++,
	391	styler.ColourTo(chColor, SCE_CAML_WHITE), styler.Flush();
	392	// ... then backtrack to determine original SML literal type
	393	int p = chColor - 2;
	394	for (; p >= 0 && styler.StyleAt(p) == SCE_CAML_WHITE; p--) ;
	395	if (p >= 0)
	396	state2 = static_cast<int>(styler.StyleAt(p));
	397	// take care of state change NOW
	398	sc.ChangeState(state2), state2 = -1;
	399	}
1e9bafca RD	400	break;
	401
	402	case SCE_CAML_COMMENT:
	403	case SCE_CAML_COMMENT1:
	404	case SCE_CAML_COMMENT2:
	405	case SCE_CAML_COMMENT3:
	406	// we're IN a comment - does this start a NESTED comment?
	407	if (sc.Match('(', '*'))
	408	state2 = sc.state + 1, chToken = sc.currentPos,
9e96e16f	409	sc.Forward(), sc.ch = ' ' /* ()... /, nesting++;
1e9bafca RD	410	// [try to] interpret as [additional] comment char
	411	else if (sc.Match(')') && sc.chPrev == '*') {
	412	if (nesting)
	413	state2 = (sc.state & 0x0f) - 1, chToken = 0, nesting--;
	414	else
	415	state2 = SCE_CAML_DEFAULT;
	416	chColor++;
	417	// enable "magic" (read-only) comment AS REQUIRED
	418	} else if (useMagic && sc.currentPos - chToken == 4
	419	&& sc.Match('c') && sc.chPrev == 'r' && sc.GetRelative(-2) == '@')
	420	sc.state \|= 0x10; // (switch to read-only comment style)
	421	break;
	422	}
	423
9e96e16f	424	// handle state change and char coloring AS REQUIRED
1e9bafca RD	425	if (state2 >= 0)
	426	styler.ColourTo(chColor, sc.state), sc.ChangeState(state2);
	427	// move to next char UNLESS re-scanning current char
	428	if (advance)
	429	sc.Forward();
	430	}
	431
	432	// do any required terminal char coloring (JIC)
	433	sc.Complete();
	434	}
	435
	436	#ifdef BUILD_AS_EXTERNAL_LEXER
	437	static
	438	#endif /* BUILD_AS_EXTERNAL_LEXER */
	439	void FoldCamlDoc(
1dcf666d RD	440	unsigned int, int,
	441	int,
	442	WordList *[],
	443	Accessor &)
1e9bafca	444	{
1e9bafca RD	445	}
	446
	447	static const char * const camlWordListDesc[] = {
	448	"Keywords", // primary Objective Caml keywords
	449	"Keywords2", // "optional" keywords (typically from Pervasives)
	450	"Keywords3", // "optional" keywords (typically typenames)
	451	0
	452	};
	453
	454	#ifndef BUILD_AS_EXTERNAL_LEXER
	455	LexerModule lmCaml(SCLEX_CAML, ColouriseCamlDoc, "caml", FoldCamlDoc, camlWordListDesc);
	456	#endif /* BUILD_AS_EXTERNAL_LEXER */