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