1 | // Scintilla source code edit control |
2 | /** @file LexTAL.cxx |
3 | ** Lexer for TAL |
4 | ** Based on LexPascal.cxx |
5 | ** Written by Laurent le Tynevez |
6 | ** Updated by Simon Steele <s.steele@pnotepad.org> September 2002 |
7 | ** Updated by Mathias Rauen <scite@madshi.net> May 2003 (Delphi adjustments) |
8 | ** Updated by Rod Falck, Aug 2006 Converted to TAL |
9 | **/ |
10 | |
11 | #include <stdlib.h> |
12 | #include <string.h> |
13 | #include <stdio.h> |
14 | #include <stdarg.h> |
15 | #include <assert.h> |
16 | #include <ctype.h> |
17 | |
18 | #include <string> |
19 | #include <string_view> |
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 Lexilla; |
33 | |
34 | inline bool isTALoperator(char ch) |
35 | { |
36 | return ch == '\'' || ch == '@' || ch == '#' || isoperator(ch); |
37 | } |
38 | |
39 | inline bool isTALwordchar(char ch) |
40 | { |
41 | return ch == '$' || ch == '^' || iswordchar(ch); |
42 | } |
43 | |
44 | inline bool isTALwordstart(char ch) |
45 | { |
46 | return ch == '$' || ch == '^' || iswordstart(ch); |
47 | } |
48 | |
49 | static void getRange(Sci_PositionU start, |
50 | Sci_PositionU end, |
51 | Accessor &styler, |
52 | char *s, |
53 | Sci_PositionU len) { |
54 | Sci_PositionU i = 0; |
55 | while ((i < end - start + 1) && (i < len-1)) { |
56 | s[i] = static_cast<char>(tolower(styler[start + i])); |
57 | i++; |
58 | } |
59 | s[i] = '\0'; |
60 | } |
61 | |
62 | static bool (int style) { |
63 | return style == SCE_C_COMMENT || |
64 | style == SCE_C_COMMENTDOC || |
65 | style == SCE_C_COMMENTDOCKEYWORD || |
66 | style == SCE_C_COMMENTDOCKEYWORDERROR; |
67 | } |
68 | |
69 | static void ColourTo(Accessor &styler, Sci_PositionU end, unsigned int attr, bool bInAsm) { |
70 | if ((bInAsm) && (attr == SCE_C_OPERATOR || attr == SCE_C_NUMBER || attr == SCE_C_DEFAULT || attr == SCE_C_WORD || attr == SCE_C_IDENTIFIER)) { |
71 | styler.ColourTo(end, SCE_C_REGEX); |
72 | } else |
73 | styler.ColourTo(end, attr); |
74 | } |
75 | |
76 | // returns 1 if the item starts a class definition, and -1 if the word is "end", and 2 if the word is "asm" |
77 | static int classifyWordTAL(Sci_PositionU start, Sci_PositionU end, /*WordList &keywords*/WordList *keywordlists[], Accessor &styler, bool bInAsm) { |
78 | int ret = 0; |
79 | |
80 | WordList& keywords = *keywordlists[0]; |
81 | WordList& builtins = *keywordlists[1]; |
82 | WordList& nonreserved_keywords = *keywordlists[2]; |
83 | |
84 | char s[100]; |
85 | getRange(start, end, styler, s, sizeof(s)); |
86 | |
87 | char chAttr = SCE_C_IDENTIFIER; |
88 | if (isdigit(s[0]) || (s[0] == '.')) { |
89 | chAttr = SCE_C_NUMBER; |
90 | } |
91 | else { |
92 | if (keywords.InList(s)) { |
93 | chAttr = SCE_C_WORD; |
94 | |
95 | if (strcmp(s, "asm" ) == 0) { |
96 | ret = 2; |
97 | } |
98 | else if (strcmp(s, "end" ) == 0) { |
99 | ret = -1; |
100 | } |
101 | } |
102 | else if (s[0] == '$' || builtins.InList(s)) { |
103 | chAttr = SCE_C_WORD2; |
104 | } |
105 | else if (nonreserved_keywords.InList(s)) { |
106 | chAttr = SCE_C_UUID; |
107 | } |
108 | } |
109 | ColourTo(styler, end, chAttr, (bInAsm && ret != -1)); |
110 | return ret; |
111 | } |
112 | |
113 | static int classifyFoldPointTAL(const char* s) { |
114 | int lev = 0; |
115 | if (!(isdigit(s[0]) || (s[0] == '.'))) { |
116 | if (strcmp(s, "begin" ) == 0 || |
117 | strcmp(s, "block" ) == 0) { |
118 | lev=1; |
119 | } else if (strcmp(s, "end" ) == 0) { |
120 | lev=-1; |
121 | } |
122 | } |
123 | return lev; |
124 | } |
125 | |
126 | static void ColouriseTALDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *keywordlists[], |
127 | Accessor &styler) { |
128 | |
129 | styler.StartAt(startPos); |
130 | |
131 | int state = initStyle; |
132 | if (state == SCE_C_CHARACTER) // Does not leak onto next line |
133 | state = SCE_C_DEFAULT; |
134 | char chPrev = ' '; |
135 | char chNext = styler[startPos]; |
136 | Sci_PositionU lengthDoc = startPos + length; |
137 | |
138 | bool bInClassDefinition; |
139 | |
140 | Sci_Position currentLine = styler.GetLine(startPos); |
141 | if (currentLine > 0) { |
142 | styler.SetLineState(currentLine, styler.GetLineState(currentLine-1)); |
143 | bInClassDefinition = (styler.GetLineState(currentLine) == 1); |
144 | } else { |
145 | styler.SetLineState(currentLine, 0); |
146 | bInClassDefinition = false; |
147 | } |
148 | |
149 | bool bInAsm = (state == SCE_C_REGEX); |
150 | if (bInAsm) |
151 | state = SCE_C_DEFAULT; |
152 | |
153 | styler.StartSegment(startPos); |
154 | int visibleChars = 0; |
155 | for (Sci_PositionU i = startPos; i < lengthDoc; i++) { |
156 | char ch = chNext; |
157 | |
158 | chNext = styler.SafeGetCharAt(i + 1); |
159 | |
160 | if ((ch == '\r' && chNext != '\n') || (ch == '\n')) { |
161 | // Trigger on CR only (Mac style) or either on LF from CR+LF (Dos/Win) or on LF alone (Unix) |
162 | // Avoid triggering two times on Dos/Win |
163 | // End of line |
164 | if (state == SCE_C_CHARACTER) { |
165 | ColourTo(styler, i, state, bInAsm); |
166 | state = SCE_C_DEFAULT; |
167 | } |
168 | visibleChars = 0; |
169 | currentLine++; |
170 | styler.SetLineState(currentLine, (bInClassDefinition ? 1 : 0)); |
171 | } |
172 | |
173 | if (styler.IsLeadByte(ch)) { |
174 | chNext = styler.SafeGetCharAt(i + 2); |
175 | chPrev = ' '; |
176 | i += 1; |
177 | continue; |
178 | } |
179 | |
180 | if (state == SCE_C_DEFAULT) { |
181 | if (isTALwordstart(ch)) { |
182 | ColourTo(styler, i-1, state, bInAsm); |
183 | state = SCE_C_IDENTIFIER; |
184 | } else if (ch == '!' && chNext != '*') { |
185 | ColourTo(styler, i-1, state, bInAsm); |
186 | state = SCE_C_COMMENT; |
187 | } else if (ch == '!' && chNext == '*') { |
188 | ColourTo(styler, i-1, state, bInAsm); |
189 | state = SCE_C_COMMENTDOC; |
190 | } else if (ch == '-' && chNext == '-') { |
191 | ColourTo(styler, i-1, state, bInAsm); |
192 | state = SCE_C_COMMENTLINE; |
193 | } else if (ch == '"') { |
194 | ColourTo(styler, i-1, state, bInAsm); |
195 | state = SCE_C_STRING; |
196 | } else if (ch == '?' && visibleChars == 0) { |
197 | ColourTo(styler, i-1, state, bInAsm); |
198 | state = SCE_C_PREPROCESSOR; |
199 | } else if (isTALoperator(ch)) { |
200 | ColourTo(styler, i-1, state, bInAsm); |
201 | ColourTo(styler, i, SCE_C_OPERATOR, bInAsm); |
202 | } |
203 | } else if (state == SCE_C_IDENTIFIER) { |
204 | if (!isTALwordchar(ch)) { |
205 | int lStateChange = classifyWordTAL(styler.GetStartSegment(), i - 1, keywordlists, styler, bInAsm); |
206 | |
207 | if(lStateChange == 1) { |
208 | styler.SetLineState(currentLine, 1); |
209 | bInClassDefinition = true; |
210 | } else if(lStateChange == 2) { |
211 | bInAsm = true; |
212 | } else if(lStateChange == -1) { |
213 | styler.SetLineState(currentLine, 0); |
214 | bInClassDefinition = false; |
215 | bInAsm = false; |
216 | } |
217 | |
218 | state = SCE_C_DEFAULT; |
219 | chNext = styler.SafeGetCharAt(i + 1); |
220 | if (ch == '!' && chNext != '*') { |
221 | state = SCE_C_COMMENT; |
222 | } else if (ch == '!' && chNext == '*') { |
223 | ColourTo(styler, i-1, state, bInAsm); |
224 | state = SCE_C_COMMENTDOC; |
225 | } else if (ch == '-' && chNext == '-') { |
226 | state = SCE_C_COMMENTLINE; |
227 | } else if (ch == '"') { |
228 | state = SCE_C_STRING; |
229 | } else if (isTALoperator(ch)) { |
230 | ColourTo(styler, i, SCE_C_OPERATOR, bInAsm); |
231 | } |
232 | } |
233 | } else { |
234 | if (state == SCE_C_PREPROCESSOR) { |
235 | if ((ch == '\r' || ch == '\n') && !(chPrev == '\\' || chPrev == '\r')) { |
236 | ColourTo(styler, i-1, state, bInAsm); |
237 | state = SCE_C_DEFAULT; |
238 | } |
239 | } else if (state == SCE_C_COMMENT) { |
240 | if (ch == '!' || (ch == '\r' || ch == '\n') ) { |
241 | ColourTo(styler, i, state, bInAsm); |
242 | state = SCE_C_DEFAULT; |
243 | } |
244 | } else if (state == SCE_C_COMMENTDOC) { |
245 | if (ch == '!' || (ch == '\r' || ch == '\n')) { |
246 | if (((i > styler.GetStartSegment() + 2) || ( |
247 | (initStyle == SCE_C_COMMENTDOC) && |
248 | (styler.GetStartSegment() == static_cast<Sci_PositionU>(startPos))))) { |
249 | ColourTo(styler, i, state, bInAsm); |
250 | state = SCE_C_DEFAULT; |
251 | } |
252 | } |
253 | } else if (state == SCE_C_COMMENTLINE) { |
254 | if (ch == '\r' || ch == '\n') { |
255 | ColourTo(styler, i-1, state, bInAsm); |
256 | state = SCE_C_DEFAULT; |
257 | } |
258 | } else if (state == SCE_C_STRING) { |
259 | if (ch == '"') { |
260 | ColourTo(styler, i, state, bInAsm); |
261 | state = SCE_C_DEFAULT; |
262 | } |
263 | } |
264 | } |
265 | if (!isspacechar(ch)) |
266 | visibleChars++; |
267 | chPrev = ch; |
268 | } |
269 | ColourTo(styler, lengthDoc - 1, state, bInAsm); |
270 | } |
271 | |
272 | static void FoldTALDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *[], |
273 | Accessor &styler) { |
274 | bool = styler.GetPropertyInt("fold.comment" ) != 0; |
275 | bool foldPreprocessor = styler.GetPropertyInt("fold.preprocessor" ) != 0; |
276 | bool foldCompact = styler.GetPropertyInt("fold.compact" , 1) != 0; |
277 | Sci_PositionU endPos = startPos + length; |
278 | int visibleChars = 0; |
279 | Sci_Position lineCurrent = styler.GetLine(startPos); |
280 | int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK; |
281 | int levelCurrent = levelPrev; |
282 | char chNext = styler[startPos]; |
283 | int styleNext = styler.StyleAt(startPos); |
284 | int style = initStyle; |
285 | bool was_end = false; |
286 | bool section = false; |
287 | |
288 | Sci_Position lastStart = 0; |
289 | |
290 | for (Sci_PositionU i = startPos; i < endPos; i++) { |
291 | char ch = chNext; |
292 | chNext = styler.SafeGetCharAt(i + 1); |
293 | int stylePrev = style; |
294 | style = styleNext; |
295 | styleNext = styler.StyleAt(i + 1); |
296 | bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n'); |
297 | |
298 | if (stylePrev == SCE_C_DEFAULT && (style == SCE_C_WORD || style == SCE_C_UUID || style == SCE_C_PREPROCESSOR)) |
299 | { |
300 | // Store last word start point. |
301 | lastStart = i; |
302 | } |
303 | |
304 | if (stylePrev == SCE_C_WORD || style == SCE_C_UUID || stylePrev == SCE_C_PREPROCESSOR) { |
305 | if(isTALwordchar(ch) && !isTALwordchar(chNext)) { |
306 | char s[100]; |
307 | getRange(lastStart, i, styler, s, sizeof(s)); |
308 | if (stylePrev == SCE_C_PREPROCESSOR && strcmp(s, "?section" ) == 0) |
309 | { |
310 | section = true; |
311 | levelCurrent = 1; |
312 | levelPrev = 0; |
313 | } |
314 | else if (stylePrev == SCE_C_WORD || stylePrev == SCE_C_UUID) |
315 | { |
316 | if (strcmp(s, "block" ) == 0) |
317 | { |
318 | // block keyword is ignored immediately after end keyword |
319 | if (!was_end) |
320 | levelCurrent++; |
321 | } |
322 | else |
323 | levelCurrent += classifyFoldPointTAL(s); |
324 | if (strcmp(s, "end" ) == 0) |
325 | { |
326 | was_end = true; |
327 | } |
328 | else |
329 | { |
330 | was_end = false; |
331 | } |
332 | } |
333 | } |
334 | } |
335 | |
336 | if (foldComment && (style == SCE_C_COMMENTLINE)) { |
337 | if ((ch == '/') && (chNext == '/')) { |
338 | char chNext2 = styler.SafeGetCharAt(i + 2); |
339 | if (chNext2 == '{') { |
340 | levelCurrent++; |
341 | } else if (chNext2 == '}') { |
342 | levelCurrent--; |
343 | } |
344 | } |
345 | } |
346 | |
347 | if (foldPreprocessor && (style == SCE_C_PREPROCESSOR)) { |
348 | if (ch == '{' && chNext == '$') { |
349 | Sci_PositionU j=i+2; // skip {$ |
350 | while ((j<endPos) && IsASpaceOrTab(styler.SafeGetCharAt(j))) { |
351 | j++; |
352 | } |
353 | if (styler.Match(j, "region" ) || styler.Match(j, "if" )) { |
354 | levelCurrent++; |
355 | } else if (styler.Match(j, "end" )) { |
356 | levelCurrent--; |
357 | } |
358 | } |
359 | } |
360 | |
361 | if (foldComment && IsStreamCommentStyle(style)) { |
362 | if (!IsStreamCommentStyle(stylePrev)) { |
363 | levelCurrent++; |
364 | } else if (!IsStreamCommentStyle(styleNext) && !atEOL) { |
365 | // Comments don't end at end of line and the next character may be unstyled. |
366 | levelCurrent--; |
367 | } |
368 | } |
369 | |
370 | if (atEOL) { |
371 | int lev = levelPrev | SC_FOLDLEVELBASE; |
372 | if (visibleChars == 0 && foldCompact) |
373 | lev |= SC_FOLDLEVELWHITEFLAG; |
374 | if ((levelCurrent > levelPrev || section) && (visibleChars > 0)) |
375 | lev |= SC_FOLDLEVELHEADERFLAG; |
376 | if (lev != styler.LevelAt(lineCurrent)) { |
377 | styler.SetLevel(lineCurrent, lev); |
378 | } |
379 | lineCurrent++; |
380 | levelPrev = levelCurrent; |
381 | visibleChars = 0; |
382 | section = false; |
383 | } |
384 | |
385 | if (!isspacechar(ch)) |
386 | visibleChars++; |
387 | } |
388 | |
389 | // Fill in the real level of the next line, keeping the current flags as they will be filled in later |
390 | int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK; |
391 | styler.SetLevel(lineCurrent, levelPrev | flagsNext); |
392 | } |
393 | |
394 | static const char * const TALWordListDesc[] = { |
395 | "Keywords" , |
396 | "Builtins" , |
397 | 0 |
398 | }; |
399 | |
400 | LexerModule lmTAL(SCLEX_TAL, ColouriseTALDoc, "TAL" , FoldTALDoc, TALWordListDesc); |
401 | |