1 // Scintilla source code edit control
2 /** @file LexRebol.cxx
3 ** Lexer for REBOL.
4 ** Written by Pascal Hurni, inspired from LexLua by Paul Winwood & Marcos E. Wurzius & Philippe Lhoste
5 **
6 ** History:
7 ** 2005-04-07 First release.
8 ** 2005-04-10 Closing parens and brackets go now in default style
9 ** String and comment nesting should be more safe
10 **/
11 // Copyright 2005 by Pascal Hurni <pascal_hurni@fastmail.fm>
12 // The License.txt file describes the conditions under which this software may be distributed.
13
14 #include <stdlib.h>
15 #include <string.h>
16 #include <stdio.h>
17 #include <stdarg.h>
18 #include <assert.h>
19 #include <ctype.h>
20
21 #include "ILexer.h"
22 #include "Scintilla.h"
23 #include "SciLexer.h"
24
25 #include "WordList.h"
26 #include "LexAccessor.h"
27 #include "Accessor.h"
28 #include "StyleContext.h"
29 #include "CharacterSet.h"
30 #include "LexerModule.h"
31
32 using namespace Scintilla;
33
IsAWordChar(const int ch)34 static inline bool IsAWordChar(const int ch) {
35 return (isalnum(ch) || ch == '?' || ch == '!' || ch == '.' || ch == '\'' || ch == '+' || ch == '-' || ch == '*' || ch == '&' || ch == '|' || ch == '=' || ch == '_' || ch == '~');
36 }
37
IsAWordStart(const int ch,const int ch2)38 static inline bool IsAWordStart(const int ch, const int ch2) {
39 return ((ch == '+' || ch == '-' || ch == '.') && !isdigit(ch2)) ||
40 (isalpha(ch) || ch == '?' || ch == '!' || ch == '\'' || ch == '*' || ch == '&' || ch == '|' || ch == '=' || ch == '_' || ch == '~');
41 }
42
IsAnOperator(const int ch,const int ch2,const int ch3)43 static inline bool IsAnOperator(const int ch, const int ch2, const int ch3) {
44 // One char operators
45 if (IsASpaceOrTab(ch2)) {
46 return ch == '+' || ch == '-' || ch == '*' || ch == '/' || ch == '<' || ch == '>' || ch == '=' || ch == '?';
47 }
48
49 // Two char operators
50 if (IsASpaceOrTab(ch3)) {
51 return (ch == '*' && ch2 == '*') ||
52 (ch == '/' && ch2 == '/') ||
53 (ch == '<' && (ch2 == '=' || ch2 == '>')) ||
54 (ch == '>' && ch2 == '=') ||
55 (ch == '=' && (ch2 == '=' || ch2 == '?')) ||
56 (ch == '?' && ch2 == '?');
57 }
58
59 return false;
60 }
61
IsBinaryStart(const int ch,const int ch2,const int ch3,const int ch4)62 static inline bool IsBinaryStart(const int ch, const int ch2, const int ch3, const int ch4) {
63 return (ch == '#' && ch2 == '{') ||
64 (IsADigit(ch) && ch2 == '#' && ch3 == '{' ) ||
65 (IsADigit(ch) && IsADigit(ch2) && ch3 == '#' && ch4 == '{' );
66 }
67
68
ColouriseRebolDoc(Sci_PositionU startPos,Sci_Position length,int initStyle,WordList * keywordlists[],Accessor & styler)69 static void ColouriseRebolDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *keywordlists[], Accessor &styler) {
70
71 WordList &keywords = *keywordlists[0];
72 WordList &keywords2 = *keywordlists[1];
73 WordList &keywords3 = *keywordlists[2];
74 WordList &keywords4 = *keywordlists[3];
75 WordList &keywords5 = *keywordlists[4];
76 WordList &keywords6 = *keywordlists[5];
77 WordList &keywords7 = *keywordlists[6];
78 WordList &keywords8 = *keywordlists[7];
79
80 Sci_Position currentLine = styler.GetLine(startPos);
81 // Initialize the braced string {.. { ... } ..} nesting level, if we are inside such a string.
82 int stringLevel = 0;
83 if (initStyle == SCE_REBOL_BRACEDSTRING || initStyle == SCE_REBOL_COMMENTBLOCK) {
84 stringLevel = styler.GetLineState(currentLine - 1);
85 }
86
87 bool blockComment = initStyle == SCE_REBOL_COMMENTBLOCK;
88 int dotCount = 0;
89
90 // Do not leak onto next line
91 if (initStyle == SCE_REBOL_COMMENTLINE) {
92 initStyle = SCE_REBOL_DEFAULT;
93 }
94
95 StyleContext sc(startPos, length, initStyle, styler);
96 if (startPos == 0) {
97 sc.SetState(SCE_REBOL_PREFACE);
98 }
99 for (; sc.More(); sc.Forward()) {
100
101 //--- What to do at line end ?
102 if (sc.atLineEnd) {
103 // Can be either inside a {} string or simply at eol
104 if (sc.state != SCE_REBOL_BRACEDSTRING && sc.state != SCE_REBOL_COMMENTBLOCK &&
105 sc.state != SCE_REBOL_BINARY && sc.state != SCE_REBOL_PREFACE)
106 sc.SetState(SCE_REBOL_DEFAULT);
107
108 // Update the line state, so it can be seen by next line
109 currentLine = styler.GetLine(sc.currentPos);
110 switch (sc.state) {
111 case SCE_REBOL_BRACEDSTRING:
112 case SCE_REBOL_COMMENTBLOCK:
113 // Inside a braced string, we set the line state
114 styler.SetLineState(currentLine, stringLevel);
115 break;
116 default:
117 // Reset the line state
118 styler.SetLineState(currentLine, 0);
119 break;
120 }
121
122 // continue with next char
123 continue;
124 }
125
126 //--- What to do on white-space ?
127 if (IsASpaceOrTab(sc.ch))
128 {
129 // Return to default if any of these states
130 if (sc.state == SCE_REBOL_OPERATOR || sc.state == SCE_REBOL_CHARACTER ||
131 sc.state == SCE_REBOL_NUMBER || sc.state == SCE_REBOL_PAIR ||
132 sc.state == SCE_REBOL_TUPLE || sc.state == SCE_REBOL_FILE ||
133 sc.state == SCE_REBOL_DATE || sc.state == SCE_REBOL_TIME ||
134 sc.state == SCE_REBOL_MONEY || sc.state == SCE_REBOL_ISSUE ||
135 sc.state == SCE_REBOL_URL || sc.state == SCE_REBOL_EMAIL) {
136 sc.SetState(SCE_REBOL_DEFAULT);
137 }
138 }
139
140 //--- Specialize state ?
141 // URL, Email look like identifier
142 if (sc.state == SCE_REBOL_IDENTIFIER)
143 {
144 if (sc.ch == ':' && !IsASpace(sc.chNext)) {
145 sc.ChangeState(SCE_REBOL_URL);
146 } else if (sc.ch == '@') {
147 sc.ChangeState(SCE_REBOL_EMAIL);
148 } else if (sc.ch == '$') {
149 sc.ChangeState(SCE_REBOL_MONEY);
150 }
151 }
152 // Words look like identifiers
153 if (sc.state == SCE_REBOL_IDENTIFIER || (sc.state >= SCE_REBOL_WORD && sc.state <= SCE_REBOL_WORD8)) {
154 // Keywords ?
155 if (!IsAWordChar(sc.ch) || sc.Match('/')) {
156 char s[100];
157 sc.GetCurrentLowered(s, sizeof(s));
158 blockComment = strcmp(s, "comment") == 0;
159 if (keywords8.InList(s)) {
160 sc.ChangeState(SCE_REBOL_WORD8);
161 } else if (keywords7.InList(s)) {
162 sc.ChangeState(SCE_REBOL_WORD7);
163 } else if (keywords6.InList(s)) {
164 sc.ChangeState(SCE_REBOL_WORD6);
165 } else if (keywords5.InList(s)) {
166 sc.ChangeState(SCE_REBOL_WORD5);
167 } else if (keywords4.InList(s)) {
168 sc.ChangeState(SCE_REBOL_WORD4);
169 } else if (keywords3.InList(s)) {
170 sc.ChangeState(SCE_REBOL_WORD3);
171 } else if (keywords2.InList(s)) {
172 sc.ChangeState(SCE_REBOL_WORD2);
173 } else if (keywords.InList(s)) {
174 sc.ChangeState(SCE_REBOL_WORD);
175 }
176 // Keep same style if there are refinements
177 if (!sc.Match('/')) {
178 sc.SetState(SCE_REBOL_DEFAULT);
179 }
180 }
181 // special numbers
182 } else if (sc.state == SCE_REBOL_NUMBER) {
183 switch (sc.ch) {
184 case 'x': sc.ChangeState(SCE_REBOL_PAIR);
185 break;
186 case ':': sc.ChangeState(SCE_REBOL_TIME);
187 break;
188 case '-':
189 case '/': sc.ChangeState(SCE_REBOL_DATE);
190 break;
191 case '.': if (++dotCount >= 2) sc.ChangeState(SCE_REBOL_TUPLE);
192 break;
193 }
194 }
195
196 //--- Determine if the current state should terminate
197 if (sc.state == SCE_REBOL_QUOTEDSTRING || sc.state == SCE_REBOL_CHARACTER) {
198 if (sc.ch == '^' && sc.chNext == '\"') {
199 sc.Forward();
200 } else if (sc.ch == '\"') {
201 sc.ForwardSetState(SCE_REBOL_DEFAULT);
202 }
203 } else if (sc.state == SCE_REBOL_BRACEDSTRING || sc.state == SCE_REBOL_COMMENTBLOCK) {
204 if (sc.ch == '}') {
205 if (--stringLevel == 0) {
206 sc.ForwardSetState(SCE_REBOL_DEFAULT);
207 }
208 } else if (sc.ch == '{') {
209 stringLevel++;
210 }
211 } else if (sc.state == SCE_REBOL_BINARY) {
212 if (sc.ch == '}') {
213 sc.ForwardSetState(SCE_REBOL_DEFAULT);
214 }
215 } else if (sc.state == SCE_REBOL_TAG) {
216 if (sc.ch == '>') {
217 sc.ForwardSetState(SCE_REBOL_DEFAULT);
218 }
219 } else if (sc.state == SCE_REBOL_PREFACE) {
220 if (sc.MatchIgnoreCase("rebol"))
221 {
222 int i;
223 for (i=5; IsASpaceOrTab(styler.SafeGetCharAt(sc.currentPos+i, 0)); i++);
224 if (sc.GetRelative(i) == '[')
225 sc.SetState(SCE_REBOL_DEFAULT);
226 }
227 }
228
229 //--- Parens and bracket changes to default style when the current is a number
230 if (sc.state == SCE_REBOL_NUMBER || sc.state == SCE_REBOL_PAIR || sc.state == SCE_REBOL_TUPLE ||
231 sc.state == SCE_REBOL_MONEY || sc.state == SCE_REBOL_ISSUE || sc.state == SCE_REBOL_EMAIL ||
232 sc.state == SCE_REBOL_URL || sc.state == SCE_REBOL_DATE || sc.state == SCE_REBOL_TIME) {
233 if (sc.ch == '(' || sc.ch == '[' || sc.ch == ')' || sc.ch == ']') {
234 sc.SetState(SCE_REBOL_DEFAULT);
235 }
236 }
237
238 //--- Determine if a new state should be entered.
239 if (sc.state == SCE_REBOL_DEFAULT) {
240 if (IsAnOperator(sc.ch, sc.chNext, sc.GetRelative(2))) {
241 sc.SetState(SCE_REBOL_OPERATOR);
242 } else if (IsBinaryStart(sc.ch, sc.chNext, sc.GetRelative(2), sc.GetRelative(3))) {
243 sc.SetState(SCE_REBOL_BINARY);
244 } else if (IsAWordStart(sc.ch, sc.chNext)) {
245 sc.SetState(SCE_REBOL_IDENTIFIER);
246 } else if (IsADigit(sc.ch) || sc.ch == '+' || sc.ch == '-' || /*Decimal*/ sc.ch == '.' || sc.ch == ',') {
247 dotCount = 0;
248 sc.SetState(SCE_REBOL_NUMBER);
249 } else if (sc.ch == '\"') {
250 sc.SetState(SCE_REBOL_QUOTEDSTRING);
251 } else if (sc.ch == '{') {
252 sc.SetState(blockComment ? SCE_REBOL_COMMENTBLOCK : SCE_REBOL_BRACEDSTRING);
253 ++stringLevel;
254 } else if (sc.ch == ';') {
255 sc.SetState(SCE_REBOL_COMMENTLINE);
256 } else if (sc.ch == '$') {
257 sc.SetState(SCE_REBOL_MONEY);
258 } else if (sc.ch == '%') {
259 sc.SetState(SCE_REBOL_FILE);
260 } else if (sc.ch == '<') {
261 sc.SetState(SCE_REBOL_TAG);
262 } else if (sc.ch == '#' && sc.chNext == '"') {
263 sc.SetState(SCE_REBOL_CHARACTER);
264 sc.Forward();
265 } else if (sc.ch == '#' && sc.chNext != '"' && sc.chNext != '{' ) {
266 sc.SetState(SCE_REBOL_ISSUE);
267 }
268 }
269 }
270 sc.Complete();
271 }
272
273
FoldRebolDoc(Sci_PositionU startPos,Sci_Position length,int,WordList * [],Accessor & styler)274 static void FoldRebolDoc(Sci_PositionU startPos, Sci_Position length, int /* initStyle */, WordList *[],
275 Accessor &styler) {
276 Sci_PositionU lengthDoc = startPos + length;
277 int visibleChars = 0;
278 Sci_Position lineCurrent = styler.GetLine(startPos);
279 int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
280 int levelCurrent = levelPrev;
281 char chNext = styler[startPos];
282 int styleNext = styler.StyleAt(startPos);
283 for (Sci_PositionU i = startPos; i < lengthDoc; i++) {
284 char ch = chNext;
285 chNext = styler.SafeGetCharAt(i + 1);
286 int style = styleNext;
287 styleNext = styler.StyleAt(i + 1);
288 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
289 if (style == SCE_REBOL_DEFAULT) {
290 if (ch == '[') {
291 levelCurrent++;
292 } else if (ch == ']') {
293 levelCurrent--;
294 }
295 }
296 if (atEOL) {
297 int lev = levelPrev;
298 if (visibleChars == 0)
299 lev |= SC_FOLDLEVELWHITEFLAG;
300 if ((levelCurrent > levelPrev) && (visibleChars > 0))
301 lev |= SC_FOLDLEVELHEADERFLAG;
302 if (lev != styler.LevelAt(lineCurrent)) {
303 styler.SetLevel(lineCurrent, lev);
304 }
305 lineCurrent++;
306 levelPrev = levelCurrent;
307 visibleChars = 0;
308 }
309 if (!isspacechar(ch))
310 visibleChars++;
311 }
312 // Fill in the real level of the next line, keeping the current flags as they will be filled in later
313 int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
314 styler.SetLevel(lineCurrent, levelPrev | flagsNext);
315 }
316
317 static const char * const rebolWordListDesc[] = {
318 "Keywords",
319 0
320 };
321
322 LexerModule lmREBOL(SCLEX_REBOL, ColouriseRebolDoc, "rebol", FoldRebolDoc, rebolWordListDesc);
323
324