]>
Commit | Line | Data |
---|---|---|
1e9bafca RD |
1 | // Scintilla source code edit control |
2 | /** @file LexTADS3.cxx | |
3 | ** Lexer for TADS3. | |
4 | **/ | |
7e0c58e9 RD |
5 | // Copyright 1998-2006 by Neil Hodgson <neilh@scintilla.org> |
6 | // The License.txt file describes the conditions under which this software may be distributed. | |
1e9bafca RD |
7 | |
8 | /* | |
9 | * TADS3 is a language designed by Michael J. Roberts for the writing of text | |
10 | * based games. TADS comes from Text Adventure Development System. It has good | |
11 | * support for the processing and outputting of formatted text and much of a | |
12 | * TADS program listing consists of strings. | |
13 | * | |
14 | * TADS has two types of strings, those enclosed in single quotes (') and those | |
15 | * enclosed in double quotes ("). These strings have different symantics and | |
16 | * can be given different highlighting if desired. | |
17 | * | |
18 | * There can be embedded within both types of strings html tags | |
19 | * ( <tag key=value> ), library directives ( <.directive> ), and message | |
20 | * parameters ( {The doctor's/his} ). | |
21 | * | |
22 | * Double quoted strings can also contain interpolated expressions | |
23 | * ( << rug.moved ? ' and a hole in the floor. ' : nil >> ). These expressions | |
24 | * may themselves contain single or double quoted strings, although the double | |
25 | * quoted strings may not contain interpolated expressions. | |
26 | * | |
27 | * These embedded constructs influence the output and formatting and are an | |
28 | * important part of a program and require highlighting. | |
29 | * | |
30 | * LINKS | |
31 | * http://www.tads.org/ | |
32 | */ | |
33 | ||
34 | #include <stdlib.h> | |
35 | #include <string.h> | |
1e9bafca RD |
36 | #include <stdio.h> |
37 | #include <stdarg.h> | |
1dcf666d RD |
38 | #include <assert.h> |
39 | #include <ctype.h> | |
1e9bafca | 40 | |
1dcf666d RD |
41 | #include "ILexer.h" |
42 | #include "Scintilla.h" | |
43 | #include "SciLexer.h" | |
1e9bafca | 44 | |
1dcf666d RD |
45 | #include "WordList.h" |
46 | #include "LexAccessor.h" | |
1e9bafca RD |
47 | #include "Accessor.h" |
48 | #include "StyleContext.h" | |
1dcf666d RD |
49 | #include "CharacterSet.h" |
50 | #include "LexerModule.h" | |
1e9bafca | 51 | |
7e0c58e9 RD |
52 | #ifdef SCI_NAMESPACE |
53 | using namespace Scintilla; | |
54 | #endif | |
55 | ||
1e9bafca RD |
56 | static const int T3_SINGLE_QUOTE = 1; |
57 | static const int T3_INT_EXPRESSION = 2; | |
7e0c58e9 RD |
58 | static const int T3_INT_EXPRESSION_IN_TAG = 4; |
59 | static const int T3_HTML_SQUOTE = 8; | |
1e9bafca RD |
60 | |
61 | static inline bool IsEOL(const int ch, const int chNext) { | |
7e0c58e9 RD |
62 | return (ch == '\r' && chNext != '\n') || (ch == '\n'); |
63 | } | |
64 | ||
65 | /* | |
66 | * Test the current character to see if it's the START of an EOL sequence; | |
67 | * if so, skip ahead to the last character of the sequence and return true, | |
68 | * and if not just return false. There are a few places where we want to | |
69 | * check to see if a newline sequence occurs at a particular point, but | |
70 | * where a caller expects a subroutine to stop only upon reaching the END | |
71 | * of a newline sequence (in particular, CR-LF on Windows). That's why | |
72 | * IsEOL() above only returns true on CR if the CR isn't followed by an LF | |
73 | * - it doesn't want to admit that there's a newline until reaching the END | |
74 | * of the sequence. We meet both needs by saying that there's a newline | |
75 | * when we see the CR in a CR-LF, but skipping the CR before returning so | |
1dcf666d | 76 | * that the caller's caller will see that we've stopped at the LF. |
7e0c58e9 RD |
77 | */ |
78 | static inline bool IsEOLSkip(StyleContext &sc) | |
79 | { | |
80 | /* test for CR-LF */ | |
81 | if (sc.ch == '\r' && sc.chNext == '\n') | |
82 | { | |
83 | /* got CR-LF - skip the CR and indicate that we're at a newline */ | |
84 | sc.Forward(); | |
85 | return true; | |
86 | } | |
87 | ||
1dcf666d | 88 | /* |
7e0c58e9 | 89 | * in other cases, we have at most a 1-character newline, so do the |
1dcf666d | 90 | * normal IsEOL test |
7e0c58e9 RD |
91 | */ |
92 | return IsEOL(sc.ch, sc.chNext); | |
1e9bafca RD |
93 | } |
94 | ||
1e9bafca | 95 | static inline bool IsATADS3Operator(const int ch) { |
7e0c58e9 RD |
96 | return ch == '=' || ch == '{' || ch == '}' || ch == '(' || ch == ')' |
97 | || ch == '[' || ch == ']' || ch == ',' || ch == ':' || ch == ';' | |
98 | || ch == '+' || ch == '-' || ch == '*' || ch == '/' || ch == '%' | |
99 | || ch == '?' || ch == '!' || ch == '<' || ch == '>' || ch == '|' | |
100 | || ch == '@' || ch == '&' || ch == '~'; | |
1e9bafca RD |
101 | } |
102 | ||
103 | static inline bool IsAWordChar(const int ch) { | |
7e0c58e9 | 104 | return isalnum(ch) || ch == '_'; |
1e9bafca RD |
105 | } |
106 | ||
107 | static inline bool IsAWordStart(const int ch) { | |
7e0c58e9 | 108 | return isalpha(ch) || ch == '_'; |
1e9bafca RD |
109 | } |
110 | ||
111 | static inline bool IsAHexDigit(const int ch) { | |
7e0c58e9 RD |
112 | int lch = tolower(ch); |
113 | return isdigit(lch) || lch == 'a' || lch == 'b' || lch == 'c' | |
114 | || lch == 'd' || lch == 'e' || lch == 'f'; | |
1e9bafca RD |
115 | } |
116 | ||
117 | static inline bool IsAnHTMLChar(int ch) { | |
7e0c58e9 | 118 | return isalnum(ch) || ch == '-' || ch == '_' || ch == '.'; |
1e9bafca RD |
119 | } |
120 | ||
121 | static inline bool IsADirectiveChar(int ch) { | |
7e0c58e9 | 122 | return isalnum(ch) || isspace(ch) || ch == '-' || ch == '/'; |
1e9bafca RD |
123 | } |
124 | ||
125 | static inline bool IsANumberStart(StyleContext &sc) { | |
7e0c58e9 RD |
126 | return isdigit(sc.ch) |
127 | || (!isdigit(sc.chPrev) && sc.ch == '.' && isdigit(sc.chNext)); | |
1e9bafca RD |
128 | } |
129 | ||
130 | inline static void ColouriseTADS3Operator(StyleContext &sc) { | |
7e0c58e9 RD |
131 | int initState = sc.state; |
132 | int c = sc.ch; | |
133 | sc.SetState(c == '{' || c == '}' ? SCE_T3_BRACE : SCE_T3_OPERATOR); | |
134 | sc.ForwardSetState(initState); | |
1e9bafca RD |
135 | } |
136 | ||
137 | static void ColouriseTADSHTMLString(StyleContext &sc, int &lineState) { | |
7e0c58e9 RD |
138 | int endState = sc.state; |
139 | int chQuote = sc.ch; | |
140 | int chString = (lineState & T3_SINGLE_QUOTE) ? '\'' : '"'; | |
141 | if (endState == SCE_T3_HTML_STRING) { | |
142 | if (lineState&T3_SINGLE_QUOTE) { | |
143 | endState = SCE_T3_S_STRING; | |
144 | chString = '\''; | |
145 | } else if (lineState&T3_INT_EXPRESSION) { | |
146 | endState = SCE_T3_X_STRING; | |
147 | chString = '"'; | |
148 | } else { | |
149 | endState = SCE_T3_HTML_DEFAULT; | |
150 | chString = '"'; | |
151 | } | |
152 | chQuote = (lineState & T3_HTML_SQUOTE) ? '\'' : '"'; | |
153 | } else { | |
154 | sc.SetState(SCE_T3_HTML_STRING); | |
155 | sc.Forward(); | |
156 | } | |
157 | if (chQuote == '"') | |
1dcf666d | 158 | lineState &= ~T3_HTML_SQUOTE; |
7e0c58e9 RD |
159 | else |
160 | lineState |= T3_HTML_SQUOTE; | |
161 | ||
162 | while (sc.More()) { | |
163 | if (IsEOL(sc.ch, sc.chNext)) { | |
164 | return; | |
165 | } | |
166 | if (sc.ch == chQuote) { | |
167 | sc.ForwardSetState(endState); | |
168 | return; | |
169 | } | |
170 | if (sc.Match('\\', static_cast<char>(chQuote))) { | |
171 | sc.Forward(2); | |
172 | sc.SetState(endState); | |
173 | return; | |
174 | } | |
175 | if (sc.ch == chString) { | |
176 | sc.SetState(SCE_T3_DEFAULT); | |
177 | return; | |
178 | } | |
179 | ||
180 | if (sc.Match('<', '<')) { | |
181 | lineState |= T3_INT_EXPRESSION | T3_INT_EXPRESSION_IN_TAG; | |
182 | sc.SetState(SCE_T3_X_DEFAULT); | |
183 | sc.Forward(2); | |
184 | return; | |
185 | } | |
186 | ||
187 | if (sc.Match('\\', static_cast<char>(chQuote)) | |
188 | || sc.Match('\\', static_cast<char>(chString)) | |
189 | || sc.Match('\\', '\\')) { | |
190 | sc.Forward(2); | |
191 | } else { | |
192 | sc.Forward(); | |
193 | } | |
194 | } | |
1e9bafca RD |
195 | } |
196 | ||
197 | static void ColouriseTADS3HTMLTagStart(StyleContext &sc) { | |
7e0c58e9 RD |
198 | sc.SetState(SCE_T3_HTML_TAG); |
199 | sc.Forward(); | |
200 | if (sc.ch == '/') { | |
201 | sc.Forward(); | |
202 | } | |
203 | while (IsAnHTMLChar(sc.ch)) { | |
204 | sc.Forward(); | |
205 | } | |
1e9bafca RD |
206 | } |
207 | ||
208 | static void ColouriseTADS3HTMLTag(StyleContext &sc, int &lineState) { | |
7e0c58e9 RD |
209 | int endState = sc.state; |
210 | int chQuote = '"'; | |
211 | int chString = '\''; | |
212 | switch (endState) { | |
213 | case SCE_T3_S_STRING: | |
214 | ColouriseTADS3HTMLTagStart(sc); | |
215 | sc.SetState(SCE_T3_HTML_DEFAULT); | |
216 | chQuote = '\''; | |
217 | chString = '"'; | |
218 | break; | |
219 | case SCE_T3_D_STRING: | |
220 | case SCE_T3_X_STRING: | |
221 | ColouriseTADS3HTMLTagStart(sc); | |
222 | sc.SetState(SCE_T3_HTML_DEFAULT); | |
223 | break; | |
224 | case SCE_T3_HTML_DEFAULT: | |
225 | if (lineState&T3_SINGLE_QUOTE) { | |
226 | endState = SCE_T3_S_STRING; | |
227 | chQuote = '\''; | |
228 | chString = '"'; | |
229 | } else if (lineState&T3_INT_EXPRESSION) { | |
230 | endState = SCE_T3_X_STRING; | |
231 | } else { | |
232 | endState = SCE_T3_D_STRING; | |
233 | } | |
234 | break; | |
235 | } | |
236 | ||
237 | while (sc.More()) { | |
238 | if (IsEOL(sc.ch, sc.chNext)) { | |
239 | return; | |
240 | } | |
241 | if (sc.Match('/', '>')) { | |
242 | sc.SetState(SCE_T3_HTML_TAG); | |
243 | sc.Forward(2); | |
244 | sc.SetState(endState); | |
245 | return; | |
246 | } | |
247 | if (sc.ch == '>') { | |
248 | sc.SetState(SCE_T3_HTML_TAG); | |
249 | sc.ForwardSetState(endState); | |
250 | return; | |
251 | } | |
252 | if (sc.ch == chQuote) { | |
253 | sc.SetState(endState); | |
254 | return; | |
255 | } | |
256 | if (sc.Match('\\', static_cast<char>(chQuote))) { | |
257 | sc.Forward(); | |
258 | ColouriseTADSHTMLString(sc, lineState); | |
259 | if (sc.state == SCE_T3_X_DEFAULT) | |
260 | break; | |
261 | } else if (sc.ch == chString) { | |
262 | ColouriseTADSHTMLString(sc, lineState); | |
263 | } else if (sc.ch == '=') { | |
264 | ColouriseTADS3Operator(sc); | |
265 | } else { | |
266 | sc.Forward(); | |
267 | } | |
268 | } | |
1e9bafca RD |
269 | } |
270 | ||
271 | static void ColouriseTADS3Keyword(StyleContext &sc, | |
7e0c58e9 RD |
272 | WordList *keywordlists[], unsigned int endPos) { |
273 | char s[250]; | |
274 | WordList &keywords = *keywordlists[0]; | |
275 | WordList &userwords1 = *keywordlists[1]; | |
276 | WordList &userwords2 = *keywordlists[2]; | |
277 | WordList &userwords3 = *keywordlists[3]; | |
278 | int initState = sc.state; | |
279 | sc.SetState(SCE_T3_IDENTIFIER); | |
280 | while (sc.More() && (IsAWordChar(sc.ch))) { | |
281 | sc.Forward(); | |
282 | } | |
283 | sc.GetCurrent(s, sizeof(s)); | |
284 | if ( strcmp(s, "is") == 0 || strcmp(s, "not") == 0) { | |
285 | // have to find if "in" is next | |
286 | int n = 1; | |
287 | while (n + sc.currentPos < endPos && IsASpaceOrTab(sc.GetRelative(n))) | |
288 | n++; | |
289 | if (sc.GetRelative(n) == 'i' && sc.GetRelative(n+1) == 'n') { | |
290 | sc.Forward(n+2); | |
291 | sc.ChangeState(SCE_T3_KEYWORD); | |
292 | } | |
293 | } else if (keywords.InList(s)) { | |
294 | sc.ChangeState(SCE_T3_KEYWORD); | |
295 | } else if (userwords3.InList(s)) { | |
296 | sc.ChangeState(SCE_T3_USER3); | |
297 | } else if (userwords2.InList(s)) { | |
298 | sc.ChangeState(SCE_T3_USER2); | |
299 | } else if (userwords1.InList(s)) { | |
300 | sc.ChangeState(SCE_T3_USER1); | |
301 | } | |
302 | sc.SetState(initState); | |
1e9bafca RD |
303 | } |
304 | ||
305 | static void ColouriseTADS3MsgParam(StyleContext &sc, int &lineState) { | |
7e0c58e9 RD |
306 | int endState = sc.state; |
307 | int chQuote = '"'; | |
308 | switch (endState) { | |
309 | case SCE_T3_S_STRING: | |
310 | sc.SetState(SCE_T3_MSG_PARAM); | |
311 | sc.Forward(); | |
312 | chQuote = '\''; | |
313 | break; | |
314 | case SCE_T3_D_STRING: | |
315 | case SCE_T3_X_STRING: | |
316 | sc.SetState(SCE_T3_MSG_PARAM); | |
317 | sc.Forward(); | |
318 | break; | |
319 | case SCE_T3_MSG_PARAM: | |
320 | if (lineState&T3_SINGLE_QUOTE) { | |
321 | endState = SCE_T3_S_STRING; | |
322 | chQuote = '\''; | |
323 | } else if (lineState&T3_INT_EXPRESSION) { | |
324 | endState = SCE_T3_X_STRING; | |
325 | } else { | |
326 | endState = SCE_T3_D_STRING; | |
327 | } | |
328 | break; | |
329 | } | |
330 | while (sc.More() && sc.ch != '}' && sc.ch != chQuote) { | |
331 | if (IsEOL(sc.ch, sc.chNext)) { | |
332 | return; | |
333 | } | |
334 | if (sc.ch == '\\') { | |
335 | sc.Forward(); | |
336 | } | |
337 | sc.Forward(); | |
338 | } | |
339 | if (sc.ch == chQuote) { | |
340 | sc.SetState(endState); | |
341 | } else { | |
342 | sc.ForwardSetState(endState); | |
343 | } | |
1e9bafca RD |
344 | } |
345 | ||
346 | static void ColouriseTADS3LibDirective(StyleContext &sc, int &lineState) { | |
7e0c58e9 RD |
347 | int initState = sc.state; |
348 | int chQuote = '"'; | |
349 | switch (initState) { | |
350 | case SCE_T3_S_STRING: | |
351 | sc.SetState(SCE_T3_LIB_DIRECTIVE); | |
352 | sc.Forward(2); | |
353 | chQuote = '\''; | |
354 | break; | |
355 | case SCE_T3_D_STRING: | |
356 | sc.SetState(SCE_T3_LIB_DIRECTIVE); | |
357 | sc.Forward(2); | |
358 | break; | |
359 | case SCE_T3_LIB_DIRECTIVE: | |
360 | if (lineState&T3_SINGLE_QUOTE) { | |
361 | initState = SCE_T3_S_STRING; | |
362 | chQuote = '\''; | |
363 | } else { | |
364 | initState = SCE_T3_D_STRING; | |
365 | } | |
366 | break; | |
367 | } | |
368 | while (sc.More() && IsADirectiveChar(sc.ch)) { | |
369 | if (IsEOL(sc.ch, sc.chNext)) { | |
370 | return; | |
371 | } | |
372 | sc.Forward(); | |
373 | }; | |
374 | if (sc.ch == '>' || !sc.More()) { | |
375 | sc.ForwardSetState(initState); | |
376 | } else if (sc.ch == chQuote) { | |
377 | sc.SetState(initState); | |
378 | } else { | |
379 | sc.ChangeState(initState); | |
380 | sc.Forward(); | |
381 | } | |
1e9bafca RD |
382 | } |
383 | ||
384 | static void ColouriseTADS3String(StyleContext &sc, int &lineState) { | |
7e0c58e9 RD |
385 | int chQuote = sc.ch; |
386 | int endState = sc.state; | |
387 | switch (sc.state) { | |
388 | case SCE_T3_DEFAULT: | |
389 | case SCE_T3_X_DEFAULT: | |
390 | if (chQuote == '"') { | |
391 | if (sc.state == SCE_T3_DEFAULT) { | |
392 | sc.SetState(SCE_T3_D_STRING); | |
393 | } else { | |
394 | sc.SetState(SCE_T3_X_STRING); | |
395 | } | |
396 | lineState &= ~T3_SINGLE_QUOTE; | |
397 | } else { | |
398 | sc.SetState(SCE_T3_S_STRING); | |
399 | lineState |= T3_SINGLE_QUOTE; | |
400 | } | |
401 | sc.Forward(); | |
402 | break; | |
403 | case SCE_T3_S_STRING: | |
404 | chQuote = '\''; | |
405 | endState = lineState&T3_INT_EXPRESSION ? | |
406 | SCE_T3_X_DEFAULT : SCE_T3_DEFAULT; | |
407 | break; | |
408 | case SCE_T3_D_STRING: | |
409 | chQuote = '"'; | |
410 | endState = SCE_T3_DEFAULT; | |
411 | break; | |
412 | case SCE_T3_X_STRING: | |
413 | chQuote = '"'; | |
414 | endState = SCE_T3_X_DEFAULT; | |
415 | break; | |
416 | } | |
417 | while (sc.More()) { | |
418 | if (IsEOL(sc.ch, sc.chNext)) { | |
419 | return; | |
420 | } | |
421 | if (sc.ch == chQuote) { | |
422 | sc.ForwardSetState(endState); | |
423 | return; | |
424 | } | |
425 | if (sc.state == SCE_T3_D_STRING && sc.Match('<', '<')) { | |
426 | lineState |= T3_INT_EXPRESSION; | |
427 | sc.SetState(SCE_T3_X_DEFAULT); | |
428 | sc.Forward(2); | |
429 | return; | |
430 | } | |
431 | if (sc.Match('\\', static_cast<char>(chQuote)) | |
432 | || sc.Match('\\', '\\')) { | |
433 | sc.Forward(2); | |
434 | } else if (sc.ch == '{') { | |
435 | ColouriseTADS3MsgParam(sc, lineState); | |
436 | } else if (sc.Match('<', '.')) { | |
437 | ColouriseTADS3LibDirective(sc, lineState); | |
438 | } else if (sc.ch == '<') { | |
439 | ColouriseTADS3HTMLTag(sc, lineState); | |
440 | if (sc.state == SCE_T3_X_DEFAULT) | |
441 | return; | |
442 | } else { | |
443 | sc.Forward(); | |
444 | } | |
445 | } | |
1e9bafca RD |
446 | } |
447 | ||
448 | static void ColouriseTADS3Comment(StyleContext &sc, int endState) { | |
7e0c58e9 RD |
449 | sc.SetState(SCE_T3_BLOCK_COMMENT); |
450 | while (sc.More()) { | |
451 | if (IsEOL(sc.ch, sc.chNext)) { | |
452 | return; | |
453 | } | |
454 | if (sc.Match('*', '/')) { | |
455 | sc.Forward(2); | |
456 | sc.SetState(endState); | |
457 | return; | |
458 | } | |
459 | sc.Forward(); | |
460 | } | |
1e9bafca RD |
461 | } |
462 | ||
463 | static void ColouriseToEndOfLine(StyleContext &sc, int initState, int endState) { | |
7e0c58e9 RD |
464 | sc.SetState(initState); |
465 | while (sc.More()) { | |
466 | if (sc.ch == '\\') { | |
467 | sc.Forward(); | |
468 | if (IsEOLSkip(sc)) { | |
469 | return; | |
470 | } | |
471 | } | |
472 | if (IsEOL(sc.ch, sc.chNext)) { | |
473 | sc.SetState(endState); | |
474 | return; | |
475 | } | |
476 | sc.Forward(); | |
477 | } | |
1e9bafca RD |
478 | } |
479 | ||
480 | static void ColouriseTADS3Number(StyleContext &sc) { | |
7e0c58e9 RD |
481 | int endState = sc.state; |
482 | bool inHexNumber = false; | |
483 | bool seenE = false; | |
484 | bool seenDot = sc.ch == '.'; | |
485 | sc.SetState(SCE_T3_NUMBER); | |
486 | if (sc.More()) { | |
487 | sc.Forward(); | |
488 | } | |
489 | if (sc.chPrev == '0' && tolower(sc.ch) == 'x') { | |
490 | inHexNumber = true; | |
491 | sc.Forward(); | |
492 | } | |
493 | while (sc.More()) { | |
494 | if (inHexNumber) { | |
495 | if (!IsAHexDigit(sc.ch)) { | |
496 | break; | |
497 | } | |
498 | } else if (!isdigit(sc.ch)) { | |
499 | if (!seenE && tolower(sc.ch) == 'e') { | |
500 | seenE = true; | |
501 | seenDot = true; | |
502 | if (sc.chNext == '+' || sc.chNext == '-') { | |
503 | sc.Forward(); | |
504 | } | |
505 | } else if (!seenDot && sc.ch == '.') { | |
506 | seenDot = true; | |
507 | } else { | |
508 | break; | |
509 | } | |
510 | } | |
511 | sc.Forward(); | |
512 | } | |
513 | sc.SetState(endState); | |
1e9bafca RD |
514 | } |
515 | ||
516 | static void ColouriseTADS3Doc(unsigned int startPos, int length, int initStyle, | |
7e0c58e9 RD |
517 | WordList *keywordlists[], Accessor &styler) { |
518 | int visibleChars = 0; | |
519 | int bracketLevel = 0; | |
520 | int lineState = 0; | |
521 | unsigned int endPos = startPos + length; | |
522 | int lineCurrent = styler.GetLine(startPos); | |
523 | if (lineCurrent > 0) { | |
524 | lineState = styler.GetLineState(lineCurrent-1); | |
525 | } | |
526 | StyleContext sc(startPos, length, initStyle, styler); | |
527 | ||
528 | while (sc.More()) { | |
529 | ||
530 | if (IsEOL(sc.ch, sc.chNext)) { | |
531 | styler.SetLineState(lineCurrent, lineState); | |
532 | lineCurrent++; | |
533 | visibleChars = 0; | |
534 | sc.Forward(); | |
535 | if (sc.ch == '\n') { | |
536 | sc.Forward(); | |
537 | } | |
538 | } | |
539 | ||
540 | switch(sc.state) { | |
541 | case SCE_T3_PREPROCESSOR: | |
542 | case SCE_T3_LINE_COMMENT: | |
543 | ColouriseToEndOfLine(sc, sc.state, lineState&T3_INT_EXPRESSION ? | |
544 | SCE_T3_X_DEFAULT : SCE_T3_DEFAULT); | |
545 | break; | |
546 | case SCE_T3_S_STRING: | |
547 | case SCE_T3_D_STRING: | |
548 | case SCE_T3_X_STRING: | |
549 | ColouriseTADS3String(sc, lineState); | |
550 | visibleChars++; | |
551 | break; | |
552 | case SCE_T3_MSG_PARAM: | |
553 | ColouriseTADS3MsgParam(sc, lineState); | |
554 | break; | |
555 | case SCE_T3_LIB_DIRECTIVE: | |
556 | ColouriseTADS3LibDirective(sc, lineState); | |
557 | break; | |
558 | case SCE_T3_HTML_DEFAULT: | |
559 | ColouriseTADS3HTMLTag(sc, lineState); | |
560 | break; | |
561 | case SCE_T3_HTML_STRING: | |
562 | ColouriseTADSHTMLString(sc, lineState); | |
563 | break; | |
564 | case SCE_T3_BLOCK_COMMENT: | |
565 | ColouriseTADS3Comment(sc, lineState&T3_INT_EXPRESSION ? | |
566 | SCE_T3_X_DEFAULT : SCE_T3_DEFAULT); | |
567 | break; | |
568 | case SCE_T3_DEFAULT: | |
569 | case SCE_T3_X_DEFAULT: | |
570 | if (IsASpaceOrTab(sc.ch)) { | |
571 | sc.Forward(); | |
572 | } else if (sc.ch == '#' && visibleChars == 0) { | |
573 | ColouriseToEndOfLine(sc, SCE_T3_PREPROCESSOR, sc.state); | |
574 | } else if (sc.Match('/', '*')) { | |
575 | ColouriseTADS3Comment(sc, sc.state); | |
576 | visibleChars++; | |
577 | } else if (sc.Match('/', '/')) { | |
578 | ColouriseToEndOfLine(sc, SCE_T3_LINE_COMMENT, sc.state); | |
579 | } else if (sc.ch == '"') { | |
580 | bracketLevel = 0; | |
581 | ColouriseTADS3String(sc, lineState); | |
582 | visibleChars++; | |
583 | } else if (sc.ch == '\'') { | |
584 | ColouriseTADS3String(sc, lineState); | |
585 | visibleChars++; | |
586 | } else if (sc.state == SCE_T3_X_DEFAULT && bracketLevel == 0 | |
587 | && sc.Match('>', '>')) { | |
588 | sc.Forward(2); | |
589 | sc.SetState(SCE_T3_D_STRING); | |
590 | if (lineState & T3_INT_EXPRESSION_IN_TAG) | |
591 | sc.SetState(SCE_T3_HTML_STRING); | |
592 | lineState &= ~(T3_SINGLE_QUOTE|T3_INT_EXPRESSION | |
593 | |T3_INT_EXPRESSION_IN_TAG); | |
594 | } else if (IsATADS3Operator(sc.ch)) { | |
595 | if (sc.state == SCE_T3_X_DEFAULT) { | |
596 | if (sc.ch == '(') { | |
597 | bracketLevel++; | |
598 | } else if (sc.ch == ')' && bracketLevel > 0) { | |
599 | bracketLevel--; | |
600 | } | |
601 | } | |
602 | ColouriseTADS3Operator(sc); | |
603 | visibleChars++; | |
604 | } else if (IsANumberStart(sc)) { | |
605 | ColouriseTADS3Number(sc); | |
606 | visibleChars++; | |
607 | } else if (IsAWordStart(sc.ch)) { | |
608 | ColouriseTADS3Keyword(sc, keywordlists, endPos); | |
609 | visibleChars++; | |
610 | } else if (sc.Match("...")) { | |
611 | sc.SetState(SCE_T3_IDENTIFIER); | |
612 | sc.Forward(3); | |
613 | sc.SetState(SCE_T3_DEFAULT); | |
614 | } else { | |
615 | sc.Forward(); | |
616 | visibleChars++; | |
617 | } | |
618 | break; | |
619 | default: | |
620 | sc.SetState(SCE_T3_DEFAULT); | |
621 | sc.Forward(); | |
622 | } | |
623 | } | |
624 | sc.Complete(); | |
1e9bafca RD |
625 | } |
626 | ||
627 | /* | |
628 | TADS3 has two styles of top level block (TLB). Eg | |
629 | ||
630 | // default style | |
631 | silverKey : Key 'small silver key' 'small silver key' | |
7e0c58e9 | 632 | "A small key glints in the sunlight. " |
1e9bafca RD |
633 | ; |
634 | ||
635 | and | |
636 | ||
637 | silverKey : Key { | |
7e0c58e9 RD |
638 | 'small silver key' |
639 | 'small silver key' | |
640 | "A small key glints in the sunlight. " | |
1e9bafca RD |
641 | } |
642 | ||
643 | Some constructs mandate one or the other, but usually the author has may choose | |
644 | either. | |
645 | ||
646 | T3_SEENSTART is used to indicate that a braceless TLB has been (potentially) | |
647 | seen and is also used to match the closing ';' of the default style. | |
648 | ||
649 | T3_EXPECTINGIDENTIFIER and T3_EXPECTINGPUNCTUATION are used to keep track of | |
650 | what characters may be seen without incrementing the block level. The general | |
651 | pattern is identifier <punc> identifier, acceptable punctuation characters | |
652 | are ':', ',', '(' and ')'. No attempt is made to ensure that punctuation | |
653 | characters are syntactically correct, eg parentheses match. A ')' always | |
654 | signifies the start of a block. We just need to check if it is followed by a | |
655 | '{', in which case we let the brace handling code handle the folding level. | |
656 | ||
657 | expectingIdentifier == false && expectingIdentifier == false | |
658 | Before the start of a TLB. | |
659 | ||
660 | expectingIdentifier == true && expectingIdentifier == true | |
661 | Currently in an identifier. Will accept identifier or punctuation. | |
662 | ||
663 | expectingIdentifier == true && expectingIdentifier == false | |
664 | Just seen a punctuation character & now waiting for an identifier to start. | |
665 | ||
666 | expectingIdentifier == false && expectingIdentifier == truee | |
667 | We were in an identifier and have seen space. Now waiting to see a punctuation | |
668 | character | |
669 | ||
670 | Space, comments & preprocessor directives are always acceptable and are | |
671 | equivalent. | |
672 | */ | |
673 | ||
674 | static const int T3_SEENSTART = 1 << 12; | |
675 | static const int T3_EXPECTINGIDENTIFIER = 1 << 13; | |
676 | static const int T3_EXPECTINGPUNCTUATION = 1 << 14; | |
677 | ||
678 | static inline bool IsStringTransition(int s1, int s2) { | |
7e0c58e9 RD |
679 | return s1 != s2 |
680 | && (s1 == SCE_T3_S_STRING || s1 == SCE_T3_X_STRING | |
9e96e16f | 681 | || (s1 == SCE_T3_D_STRING && s2 != SCE_T3_X_DEFAULT)) |
7e0c58e9 RD |
682 | && s2 != SCE_T3_LIB_DIRECTIVE |
683 | && s2 != SCE_T3_MSG_PARAM | |
684 | && s2 != SCE_T3_HTML_TAG | |
685 | && s2 != SCE_T3_HTML_STRING; | |
1e9bafca RD |
686 | } |
687 | ||
688 | static inline bool IsATADS3Punctuation(const int ch) { | |
7e0c58e9 | 689 | return ch == ':' || ch == ',' || ch == '(' || ch == ')'; |
1e9bafca RD |
690 | } |
691 | ||
692 | static inline bool IsAnIdentifier(const int style) { | |
7e0c58e9 RD |
693 | return style == SCE_T3_IDENTIFIER |
694 | || style == SCE_T3_USER1 | |
695 | || style == SCE_T3_USER2 | |
696 | || style == SCE_T3_USER3; | |
697 | } | |
698 | ||
699 | static inline bool IsAnOperator(const int style) { | |
1dcf666d | 700 | return style == SCE_T3_OPERATOR || style == SCE_T3_BRACE; |
1e9bafca RD |
701 | } |
702 | ||
703 | static inline bool IsSpaceEquivalent(const int ch, const int style) { | |
7e0c58e9 RD |
704 | return isspace(ch) |
705 | || style == SCE_T3_BLOCK_COMMENT | |
706 | || style == SCE_T3_LINE_COMMENT | |
707 | || style == SCE_T3_PREPROCESSOR; | |
1e9bafca RD |
708 | } |
709 | ||
710 | static char peekAhead(unsigned int startPos, unsigned int endPos, | |
7e0c58e9 RD |
711 | Accessor &styler) { |
712 | for (unsigned int i = startPos; i < endPos; i++) { | |
713 | int style = styler.StyleAt(i); | |
714 | char ch = styler[i]; | |
715 | if (!IsSpaceEquivalent(ch, style)) { | |
716 | if (IsAnIdentifier(style)) { | |
717 | return 'a'; | |
718 | } | |
719 | if (IsATADS3Punctuation(ch)) { | |
720 | return ':'; | |
721 | } | |
722 | if (ch == '{') { | |
723 | return '{'; | |
724 | } | |
725 | return '*'; | |
726 | } | |
727 | } | |
728 | return ' '; | |
1e9bafca RD |
729 | } |
730 | ||
731 | static void FoldTADS3Doc(unsigned int startPos, int length, int initStyle, | |
732 | WordList *[], Accessor &styler) { | |
7e0c58e9 RD |
733 | unsigned int endPos = startPos + length; |
734 | int lineCurrent = styler.GetLine(startPos); | |
735 | int levelCurrent = SC_FOLDLEVELBASE; | |
736 | if (lineCurrent > 0) | |
737 | levelCurrent = styler.LevelAt(lineCurrent-1) >> 16; | |
738 | int seenStart = levelCurrent & T3_SEENSTART; | |
739 | int expectingIdentifier = levelCurrent & T3_EXPECTINGIDENTIFIER; | |
740 | int expectingPunctuation = levelCurrent & T3_EXPECTINGPUNCTUATION; | |
741 | levelCurrent &= SC_FOLDLEVELNUMBERMASK; | |
742 | int levelMinCurrent = levelCurrent; | |
743 | int levelNext = levelCurrent; | |
744 | char chNext = styler[startPos]; | |
745 | int styleNext = styler.StyleAt(startPos); | |
746 | int style = initStyle; | |
747 | char ch = chNext; | |
748 | int stylePrev = style; | |
749 | bool redo = false; | |
750 | for (unsigned int i = startPos; i < endPos; i++) { | |
751 | if (redo) { | |
752 | redo = false; | |
753 | i--; | |
754 | } else { | |
755 | ch = chNext; | |
756 | chNext = styler.SafeGetCharAt(i + 1); | |
757 | stylePrev = style; | |
758 | style = styleNext; | |
759 | styleNext = styler.StyleAt(i + 1); | |
760 | } | |
761 | bool atEOL = IsEOL(ch, chNext); | |
762 | ||
763 | if (levelNext == SC_FOLDLEVELBASE) { | |
764 | if (IsSpaceEquivalent(ch, style)) { | |
765 | if (expectingPunctuation) { | |
766 | expectingIdentifier = 0; | |
767 | } | |
768 | if (style == SCE_T3_BLOCK_COMMENT) { | |
769 | levelNext++; | |
770 | } | |
771 | } else if (ch == '{') { | |
772 | levelNext++; | |
773 | seenStart = 0; | |
774 | } else if (ch == '\'' || ch == '"' || ch == '[') { | |
775 | levelNext++; | |
776 | if (seenStart) { | |
777 | redo = true; | |
778 | } | |
779 | } else if (ch == ';') { | |
780 | seenStart = 0; | |
781 | expectingIdentifier = 0; | |
782 | expectingPunctuation = 0; | |
783 | } else if (expectingIdentifier && expectingPunctuation) { | |
784 | if (IsATADS3Punctuation(ch)) { | |
785 | if (ch == ')' && peekAhead(i+1, endPos, styler) != '{') { | |
786 | levelNext++; | |
787 | } else { | |
788 | expectingPunctuation = 0; | |
789 | } | |
790 | } else if (!IsAnIdentifier(style)) { | |
791 | levelNext++; | |
792 | } | |
793 | } else if (expectingIdentifier && !expectingPunctuation) { | |
794 | if (!IsAnIdentifier(style)) { | |
795 | levelNext++; | |
796 | } else { | |
797 | expectingPunctuation = T3_EXPECTINGPUNCTUATION; | |
798 | } | |
799 | } else if (!expectingIdentifier && expectingPunctuation) { | |
800 | if (!IsATADS3Punctuation(ch)) { | |
801 | levelNext++; | |
802 | } else { | |
803 | if (ch == ')' && peekAhead(i+1, endPos, styler) != '{') { | |
804 | levelNext++; | |
805 | } else { | |
806 | expectingIdentifier = T3_EXPECTINGIDENTIFIER; | |
807 | expectingPunctuation = 0; | |
808 | } | |
809 | } | |
810 | } else if (!expectingIdentifier && !expectingPunctuation) { | |
811 | if (IsAnIdentifier(style)) { | |
812 | seenStart = T3_SEENSTART; | |
813 | expectingIdentifier = T3_EXPECTINGIDENTIFIER; | |
814 | expectingPunctuation = T3_EXPECTINGPUNCTUATION; | |
815 | } | |
816 | } | |
817 | ||
818 | if (levelNext != SC_FOLDLEVELBASE && style != SCE_T3_BLOCK_COMMENT) { | |
819 | expectingIdentifier = 0; | |
820 | expectingPunctuation = 0; | |
821 | } | |
822 | ||
823 | } else if (levelNext == SC_FOLDLEVELBASE+1 && seenStart | |
824 | && ch == ';' && IsAnOperator(style)) { | |
825 | levelNext--; | |
826 | seenStart = 0; | |
827 | } else if (style == SCE_T3_BLOCK_COMMENT) { | |
828 | if (stylePrev != SCE_T3_BLOCK_COMMENT) { | |
829 | levelNext++; | |
830 | } else if (styleNext != SCE_T3_BLOCK_COMMENT && !atEOL) { | |
831 | // Comments don't end at end of line and the next character may be unstyled. | |
832 | levelNext--; | |
833 | } | |
834 | } else if (ch == '\'' || ch == '"') { | |
835 | if (IsStringTransition(style, stylePrev)) { | |
836 | if (levelMinCurrent > levelNext) { | |
837 | levelMinCurrent = levelNext; | |
838 | } | |
839 | levelNext++; | |
840 | } else if (IsStringTransition(style, styleNext)) { | |
841 | levelNext--; | |
842 | } | |
843 | } else if (IsAnOperator(style)) { | |
844 | if (ch == '{' || ch == '[') { | |
845 | // Measure the minimum before a '{' to allow | |
846 | // folding on "} else {" | |
847 | if (levelMinCurrent > levelNext) { | |
848 | levelMinCurrent = levelNext; | |
849 | } | |
850 | levelNext++; | |
851 | } else if (ch == '}' || ch == ']') { | |
852 | levelNext--; | |
853 | } | |
854 | } | |
855 | ||
856 | if (atEOL) { | |
857 | if (seenStart && levelNext == SC_FOLDLEVELBASE) { | |
858 | switch (peekAhead(i+1, endPos, styler)) { | |
859 | case ' ': | |
860 | case '{': | |
861 | break; | |
862 | case '*': | |
863 | levelNext++; | |
864 | break; | |
865 | case 'a': | |
866 | if (expectingPunctuation) { | |
867 | levelNext++; | |
868 | } | |
869 | break; | |
870 | case ':': | |
871 | if (expectingIdentifier) { | |
872 | levelNext++; | |
873 | } | |
874 | break; | |
875 | } | |
876 | if (levelNext != SC_FOLDLEVELBASE) { | |
877 | expectingIdentifier = 0; | |
878 | expectingPunctuation = 0; | |
879 | } | |
880 | } | |
881 | int lev = levelMinCurrent | (levelNext | expectingIdentifier | |
882 | | expectingPunctuation | seenStart) << 16; | |
883 | if (levelMinCurrent < levelNext) | |
884 | lev |= SC_FOLDLEVELHEADERFLAG; | |
885 | if (lev != styler.LevelAt(lineCurrent)) { | |
886 | styler.SetLevel(lineCurrent, lev); | |
887 | } | |
888 | lineCurrent++; | |
889 | levelCurrent = levelNext; | |
890 | levelMinCurrent = levelCurrent; | |
891 | } | |
892 | } | |
1e9bafca RD |
893 | } |
894 | ||
895 | static const char * const tads3WordList[] = { | |
7e0c58e9 RD |
896 | "TADS3 Keywords", |
897 | "User defined 1", | |
898 | "User defined 2", | |
899 | "User defined 3", | |
900 | 0 | |
1e9bafca RD |
901 | }; |
902 | ||
903 | LexerModule lmTADS3(SCLEX_TADS3, ColouriseTADS3Doc, "tads3", FoldTADS3Doc, tads3WordList); |