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

// Scintilla source code edit control
/** @file LexSML.cxx
 ** Lexer for SML.
 **/
// Copyright 2009 by James Moffatt and Yuzhou Xin
// Modified from LexCaml.cxx by Robert Roessler <robertr@rftp.com> Copyright 2005
// The License.txt file describes the conditions under which this software may be distributed.

#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 "WordList.h"
#include "LexAccessor.h"
#include "Accessor.h"
#include "StyleContext.h"
#include "CharacterSet.h"
#include "LexerModule.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, int,
	int,
	WordList *[],
	Accessor &)
{
}

static const char * const SMLWordListDesc[] = {
	"Keywords",
	"Keywords2",
	"Keywords3",
	0
};

LexerModule lmSML(SCLEX_SML, ColouriseSMLDoc, "SML", FoldSMLDoc, SMLWordListDesc);
Commit	Line	Data
9e96e16f RD	1	// Scintilla source code edit control
	2	/** @file LexSML.cxx
	3	** Lexer for SML.
	4	**/
	5	// Copyright 2009 by James Moffatt and Yuzhou Xin
	6	// Modified from LexCaml.cxx by Robert Roessler <robertr@rftp.com> Copyright 2005
	7	// The License.txt file describes the conditions under which this software may be distributed.
	8
9e96e16f RD	9	#include <stdlib.h>
9e96e16f RD	10	#include <string.h>
9e96e16f RD	11	#include <stdio.h>
9e96e16f RD	12	#include <stdarg.h>
1dcf666d RD	13	#include <assert.h>
1dcf666d RD	14	#include <ctype.h>
9e96e16f	15
1dcf666d RD	16	#include "ILexer.h"
	17	#include "Scintilla.h"
	18	#include "SciLexer.h"
9e96e16f	19
1dcf666d RD	20	#include "WordList.h"
1dcf666d RD	21	#include "LexAccessor.h"
9e96e16f RD	22	#include "Accessor.h"
9e96e16f RD	23	#include "StyleContext.h"
1dcf666d RD	24	#include "CharacterSet.h"
1dcf666d RD	25	#include "LexerModule.h"
9e96e16f RD	26
	27	inline int issml(int c) {return isalnum(c) \|\| c == '_';}
	28	inline int issmlf(int c) {return isalpha(c) \|\| c == '_';}
	29	inline int issmld(int c) {return isdigit(c) \|\| c == '_';}
	30
	31
	32	#ifdef SCI_NAMESPACE
	33	using namespace Scintilla;
	34	#endif
	35
	36	void ColouriseSMLDoc(
	37	unsigned int startPos, int length,
	38	int initStyle,
	39	WordList *keywordlists[],
	40	Accessor &styler)
	41	{
	42	StyleContext sc(startPos, length, initStyle, styler);
	43	int nesting = 0;
	44	if (sc.state < SCE_SML_STRING)
	45	sc.state = SCE_SML_DEFAULT;
	46	if (sc.state >= SCE_SML_COMMENT)
	47	nesting = (sc.state & 0x0f) - SCE_SML_COMMENT;
	48
	49	int chBase = 0, chToken = 0, chLit = 0;
	50	WordList& keywords = *keywordlists[0];
	51	WordList& keywords2 = *keywordlists[1];
	52	WordList& keywords3 = *keywordlists[2];
	53	const int useMagic = styler.GetPropertyInt("lexer.caml.magic", 0);
	54
	55	while (sc.More()) {
	56	int state2 = -1;
	57	int chColor = sc.currentPos - 1;
	58	bool advance = true;
	59
	60	switch (sc.state & 0x0f) {
	61	case SCE_SML_DEFAULT:
	62	chToken = sc.currentPos;
	63	if (issmlf(sc.ch))
	64	state2 = SCE_SML_IDENTIFIER;
	65	else if (sc.Match('`') && issmlf(sc.chNext))
	66	state2 = SCE_SML_TAGNAME;
	67	else if (sc.Match('#')&&isdigit(sc.chNext))
	68	state2 = SCE_SML_LINENUM;
	69	else if (sc.Match('#','\"')){
	70	state2 = SCE_SML_CHAR,chLit = 0;
	71	sc.Forward();
1dcf666d	72
9e96e16f RD	73	}
	74	else if (isdigit(sc.ch)) {
	75	state2 = SCE_SML_NUMBER, chBase = 10;
	76	if (sc.Match('0') && strchr("xX", sc.chNext))
	77	chBase = 16, sc.Forward();}
	78	else if (sc.Match('\"')&&sc.chPrev!='#')
	79	state2 = SCE_SML_STRING;
	80	else if (sc.Match('(', '*')){
	81	state2 = SCE_SML_COMMENT,
	82	sc.ch = ' ',
	83	sc.Forward();}
	84	else if (strchr("!~"
	85	"=<>@^+-*/"
	86	"()[];,:.#", sc.ch))
	87	state2 = SCE_SML_OPERATOR;
	88	break;
	89
	90	case SCE_SML_IDENTIFIER:
	91	if (!(issml(sc.ch) \|\| sc.Match('\''))) {
	92	const int n = sc.currentPos - chToken;
	93	if (n < 24) {
	94	char t[24];
	95	for (int i = -n; i < 0; i++)
	96	t[n + i] = static_cast<char>(sc.GetRelative(i));
	97	t[n] = '\0';
	98	if ((n == 1 && sc.chPrev == '_') \|\| keywords.InList(t))
	99	sc.ChangeState(SCE_SML_KEYWORD);
	100	else if (keywords2.InList(t))
	101	sc.ChangeState(SCE_SML_KEYWORD2);
	102	else if (keywords3.InList(t))
	103	sc.ChangeState(SCE_SML_KEYWORD3);
	104	}
	105	state2 = SCE_SML_DEFAULT, advance = false;
	106	}
	107	break;
	108
	109	case SCE_SML_TAGNAME:
	110	if (!(issml(sc.ch) \|\| sc.Match('\'')))
	111	state2 = SCE_SML_DEFAULT, advance = false;
	112	break;
	113
	114	case SCE_SML_LINENUM:
	115	if (!isdigit(sc.ch))
	116	state2 = SCE_SML_DEFAULT, advance = false;
	117	break;
	118
	119	case SCE_SML_OPERATOR: {
	120	const char* o = 0;
	121	if (issml(sc.ch) \|\| isspace(sc.ch)
	122	\|\| (o = strchr(")]};,\'\"`#", sc.ch),o)
	123	\|\| !strchr("!$%&*+-./:<=>?@^\|~", sc.ch)) {
	124	if (o && strchr(")]};,", sc.ch)) {
	125	if ((sc.Match(')') && sc.chPrev == '(')
	126	\|\| (sc.Match(']') && sc.chPrev == '['))
	127	sc.ChangeState(SCE_SML_KEYWORD);
	128	chColor++;
	129	} else
	130	advance = false;
	131	state2 = SCE_SML_DEFAULT;
	132	}
	133	break;
	134	}
	135
	136	case SCE_SML_NUMBER:
137	if (issmld(sc.ch) \|\| IsADigit(sc.ch, chBase))
138	break;
139	if ((sc.Match('l') \|\| sc.Match('L') \|\| sc.Match('n'))
140	&& (issmld(sc.chPrev) \|\| IsADigit(sc.chPrev, chBase)))
141	break;
142	if (chBase == 10) {
143	if (sc.Match('.') && issmld(sc.chPrev))
144	break;
145	if ((sc.Match('e') \|\| sc.Match('E'))
146	&& (issmld(sc.chPrev) \|\| sc.chPrev == '.'))
147	break;
148	if ((sc.Match('+') \|\| sc.Match('-'))
149	&& (sc.chPrev == 'e' \|\| sc.chPrev == 'E'))
150	break;
151	}
152	state2 = SCE_SML_DEFAULT, advance = false;
153	break;
154
155	case SCE_SML_CHAR:
156	if (sc.Match('\\')) {
157	chLit = 1;
158	if (sc.chPrev == '\\')
159	sc.ch = ' ';
160	} else if ((sc.Match('\"') && sc.chPrev != '\\') \|\| sc.atLineEnd) {
161	state2 = SCE_SML_DEFAULT;
162	chLit = 1;
163	if (sc.Match('\"'))
164	chColor++;
165	else
166	sc.ChangeState(SCE_SML_IDENTIFIER);
167	} else if (chLit < 1 && sc.currentPos - chToken >= 3)
168	sc.ChangeState(SCE_SML_IDENTIFIER), advance = false;
169	break;
170
171	case SCE_SML_STRING:
172	if (sc.Match('\\') && sc.chPrev == '\\')
173	sc.ch = ' ';
174	else if (sc.Match('\"') && sc.chPrev != '\\')
175	state2 = SCE_SML_DEFAULT, chColor++;
176	break;
177
178	case SCE_SML_COMMENT:
179	case SCE_SML_COMMENT1:
180	case SCE_SML_COMMENT2:
181	case SCE_SML_COMMENT3:
182	if (sc.Match('(', '*'))
183	state2 = sc.state + 1, chToken = sc.currentPos,
184	sc.ch = ' ',
185	sc.Forward(), nesting++;
186	else if (sc.Match(')') && sc.chPrev == '*') {
187	if (nesting)
188	state2 = (sc.state & 0x0f) - 1, chToken = 0, nesting--;
189	else
190	state2 = SCE_SML_DEFAULT;
191	chColor++;
192	} else if (useMagic && sc.currentPos - chToken == 4
193	&& sc.Match('c') && sc.chPrev == 'r' && sc.GetRelative(-2) == '@')
194	sc.state \|= 0x10;
195	break;
196	}
197
198	if (state2 >= 0)
199	styler.ColourTo(chColor, sc.state), sc.ChangeState(state2);
200	if (advance)
201	sc.Forward();
202	}
203
204	sc.Complete();
205	}
206
207	void FoldSMLDoc(
1dcf666d RD	208	unsigned int, int,
	209	int,
	210	WordList *[],
	211	Accessor &)
9e96e16f	212	{
9e96e16f RD	213	}
	214
	215	static const char * const SMLWordListDesc[] = {
	216	"Keywords",
	217	"Keywords2",
	218	"Keywords3",
	219	0
	220	};
	221
	222	LexerModule lmSML(SCLEX_SML, ColouriseSMLDoc, "SML", FoldSMLDoc, SMLWordListDesc);
1dcf666d	223