1 | // Scintilla source code edit control |
2 | /** @file LexTCL.cxx |
3 | ** Lexer for TCL language. |
4 | **/ |
5 | // Copyright 1998-2001 by Andre Arpin <arpin@kingston.net> |
6 | // The License.txt file describes the conditions under which this software may be distributed. |
7 | |
8 | #include <stdlib.h> |
9 | #include <string.h> |
10 | #include <stdio.h> |
11 | #include <stdarg.h> |
12 | #include <assert.h> |
13 | #include <ctype.h> |
14 | |
15 | #include <string> |
16 | #include <string_view> |
17 | |
18 | #include "ILexer.h" |
19 | #include "Scintilla.h" |
20 | #include "SciLexer.h" |
21 | |
22 | #include "WordList.h" |
23 | #include "LexAccessor.h" |
24 | #include "Accessor.h" |
25 | #include "StyleContext.h" |
26 | #include "CharacterSet.h" |
27 | #include "LexerModule.h" |
28 | |
29 | using namespace Lexilla; |
30 | |
31 | // Extended to accept accented characters |
32 | static inline bool IsAWordChar(int ch) { |
33 | return ch >= 0x80 || |
34 | (isalnum(ch) || ch == '_' || ch ==':' || ch=='.'); // : name space separator |
35 | } |
36 | |
37 | static inline bool IsAWordStart(int ch) { |
38 | return ch >= 0x80 || (ch ==':' || isalpha(ch) || ch == '_'); |
39 | } |
40 | |
41 | static inline bool IsANumberChar(int ch) { |
42 | // Not exactly following number definition (several dots are seen as OK, etc.) |
43 | // but probably enough in most cases. |
44 | return (ch < 0x80) && |
45 | (IsADigit(ch, 0x10) || toupper(ch) == 'E' || |
46 | ch == '.' || ch == '-' || ch == '+'); |
47 | } |
48 | |
49 | static void ColouriseTCLDoc(Sci_PositionU startPos, Sci_Position length, int , WordList *keywordlists[], Accessor &styler) { |
50 | #define (s) (s==SCE_TCL_COMMENT || s==SCE_TCL_COMMENTLINE || s==SCE_TCL_COMMENT_BOX || s==SCE_TCL_BLOCK_COMMENT) |
51 | const bool = styler.GetPropertyInt("fold.comment" ) != 0; |
52 | const bool foldCompact = styler.GetPropertyInt("fold.compact" , 1) != 0; |
53 | bool = false; |
54 | bool subBrace = false; // substitution begin with a brace ${.....} |
55 | enum tLineState {LS_DEFAULT, , LS_OPEN_DOUBLE_QUOTE, , LS_MASK_STATE = 0xf, |
56 | LS_COMMAND_EXPECTED = 16, LS_BRACE_ONLY = 32 |
57 | } lineState = LS_DEFAULT; |
58 | bool prevSlash = false; |
59 | int currentLevel = 0; |
60 | bool expected = 0; |
61 | bool subParen = 0; |
62 | |
63 | Sci_Position currentLine = styler.GetLine(startPos); |
64 | if (currentLine > 0) |
65 | currentLine--; |
66 | length += startPos - styler.LineStart(currentLine); |
67 | // make sure lines overlap |
68 | startPos = styler.LineStart(currentLine); |
69 | |
70 | WordList &keywords = *keywordlists[0]; |
71 | WordList &keywords2 = *keywordlists[1]; |
72 | WordList &keywords3 = *keywordlists[2]; |
73 | WordList &keywords4 = *keywordlists[3]; |
74 | WordList &keywords5 = *keywordlists[4]; |
75 | WordList &keywords6 = *keywordlists[5]; |
76 | WordList &keywords7 = *keywordlists[6]; |
77 | WordList &keywords8 = *keywordlists[7]; |
78 | WordList &keywords9 = *keywordlists[8]; |
79 | |
80 | if (currentLine > 0) { |
81 | int ls = styler.GetLineState(currentLine - 1); |
82 | lineState = tLineState(ls & LS_MASK_STATE); |
83 | expected = LS_COMMAND_EXPECTED == tLineState(ls & LS_COMMAND_EXPECTED); |
84 | subBrace = LS_BRACE_ONLY == tLineState(ls & LS_BRACE_ONLY); |
85 | currentLevel = styler.LevelAt(currentLine - 1) >> 17; |
86 | commentLevel = (styler.LevelAt(currentLine - 1) >> 16) & 1; |
87 | } else |
88 | styler.SetLevel(0, SC_FOLDLEVELBASE | SC_FOLDLEVELHEADERFLAG); |
89 | bool visibleChars = false; |
90 | |
91 | int previousLevel = currentLevel; |
92 | StyleContext sc(startPos, length, SCE_TCL_DEFAULT, styler); |
93 | for (; ; sc.Forward()) { |
94 | next: |
95 | if (sc.ch=='\r' && sc.chNext == '\n') // only ignore \r on PC process on the mac |
96 | continue; |
97 | bool atEnd = !sc.More(); // make sure we coloured the last word |
98 | if (lineState != LS_DEFAULT) { |
99 | sc.SetState(SCE_TCL_DEFAULT); |
100 | if (lineState == LS_OPEN_COMMENT) |
101 | sc.SetState(SCE_TCL_COMMENTLINE); |
102 | else if (lineState == LS_OPEN_DOUBLE_QUOTE) |
103 | sc.SetState(SCE_TCL_IN_QUOTE); |
104 | else if (lineState == LS_COMMENT_BOX && (sc.ch == '#' || (sc.ch == ' ' && sc.chNext=='#'))) |
105 | sc.SetState(SCE_TCL_COMMENT_BOX); |
106 | lineState = LS_DEFAULT; |
107 | } |
108 | if (subBrace) { // ${ overrides every thing even \ except } |
109 | if (sc.ch == '}') { |
110 | subBrace = false; |
111 | sc.SetState(SCE_TCL_OPERATOR); |
112 | sc.ForwardSetState(SCE_TCL_DEFAULT); |
113 | goto next; |
114 | } else |
115 | sc.SetState(SCE_TCL_SUB_BRACE); |
116 | if (!sc.atLineEnd) |
117 | continue; |
118 | } else if (sc.state == SCE_TCL_DEFAULT || sc.state ==SCE_TCL_OPERATOR) { |
119 | expected &= isspacechar(static_cast<unsigned char>(sc.ch)) || IsAWordStart(sc.ch) || sc.ch =='#'; |
120 | } else if (sc.state == SCE_TCL_SUBSTITUTION) { |
121 | switch (sc.ch) { |
122 | case '(': |
123 | subParen=true; |
124 | sc.SetState(SCE_TCL_OPERATOR); |
125 | sc.ForwardSetState(SCE_TCL_SUBSTITUTION); |
126 | continue; |
127 | case ')': |
128 | sc.SetState(SCE_TCL_OPERATOR); |
129 | subParen=false; |
130 | continue; |
131 | case '$': |
132 | continue; |
133 | case ',': |
134 | sc.SetState(SCE_TCL_OPERATOR); |
135 | if (subParen) { |
136 | sc.ForwardSetState(SCE_TCL_SUBSTITUTION); |
137 | goto next; // Already forwarded so avoid loop's Forward() |
138 | } |
139 | continue; |
140 | default : |
141 | // maybe spaces should be allowed ??? |
142 | if (!IsAWordChar(sc.ch)) { // probably the code is wrong |
143 | sc.SetState(SCE_TCL_DEFAULT); |
144 | subParen = 0; |
145 | } |
146 | break; |
147 | } |
148 | } else if (isComment(sc.state)) { |
149 | } else if (!IsAWordChar(sc.ch)) { |
150 | if ((sc.state == SCE_TCL_IDENTIFIER && expected) || sc.state == SCE_TCL_MODIFIER) { |
151 | char w[100]; |
152 | sc.GetCurrent(w, sizeof(w)); |
153 | char *s=w; |
154 | if (w[strlen(w)-1]=='\r') |
155 | w[strlen(w)-1]=0; |
156 | while (*s == ':') // ignore leading : like in ::set a 10 |
157 | ++s; |
158 | bool quote = sc.state == SCE_TCL_IN_QUOTE; |
159 | if (commentLevel || expected) { |
160 | if (keywords.InList(s)) { |
161 | sc.ChangeState(quote ? SCE_TCL_WORD_IN_QUOTE : SCE_TCL_WORD); |
162 | } else if (keywords2.InList(s)) { |
163 | sc.ChangeState(quote ? SCE_TCL_WORD_IN_QUOTE : SCE_TCL_WORD2); |
164 | } else if (keywords3.InList(s)) { |
165 | sc.ChangeState(quote ? SCE_TCL_WORD_IN_QUOTE : SCE_TCL_WORD3); |
166 | } else if (keywords4.InList(s)) { |
167 | sc.ChangeState(quote ? SCE_TCL_WORD_IN_QUOTE : SCE_TCL_WORD4); |
168 | } else if (sc.GetRelative(-static_cast<Sci_Position>(strlen(s))-1) == '{' && |
169 | keywords5.InList(s) && sc.ch == '}') { // {keyword} exactly no spaces |
170 | sc.ChangeState(SCE_TCL_EXPAND); |
171 | } |
172 | if (keywords6.InList(s)) { |
173 | sc.ChangeState(SCE_TCL_WORD5); |
174 | } else if (keywords7.InList(s)) { |
175 | sc.ChangeState(SCE_TCL_WORD6); |
176 | } else if (keywords8.InList(s)) { |
177 | sc.ChangeState(SCE_TCL_WORD7); |
178 | } else if (keywords9.InList(s)) { |
179 | sc.ChangeState(SCE_TCL_WORD8); |
180 | } |
181 | } |
182 | expected = false; |
183 | sc.SetState(quote ? SCE_TCL_IN_QUOTE : SCE_TCL_DEFAULT); |
184 | } else if (sc.state == SCE_TCL_MODIFIER || sc.state == SCE_TCL_IDENTIFIER) { |
185 | sc.SetState(SCE_TCL_DEFAULT); |
186 | } |
187 | } |
188 | if (atEnd) |
189 | break; |
190 | if (sc.atLineEnd) { |
191 | lineState = LS_DEFAULT; |
192 | currentLine = styler.GetLine(sc.currentPos); |
193 | if (foldComment && sc.state!=SCE_TCL_COMMENT && isComment(sc.state)) { |
194 | if (currentLevel == 0) { |
195 | ++currentLevel; |
196 | commentLevel = true; |
197 | } |
198 | } else { |
199 | if (visibleChars && commentLevel) { |
200 | --currentLevel; |
201 | --previousLevel; |
202 | commentLevel = false; |
203 | } |
204 | } |
205 | int flag = 0; |
206 | if (!visibleChars && foldCompact) |
207 | flag = SC_FOLDLEVELWHITEFLAG; |
208 | if (currentLevel > previousLevel) |
209 | flag = SC_FOLDLEVELHEADERFLAG; |
210 | styler.SetLevel(currentLine, flag + previousLevel + SC_FOLDLEVELBASE + (currentLevel << 17) + (commentLevel << 16)); |
211 | |
212 | // Update the line state, so it can be seen by next line |
213 | if (sc.state == SCE_TCL_IN_QUOTE) { |
214 | lineState = LS_OPEN_DOUBLE_QUOTE; |
215 | } else { |
216 | if (prevSlash) { |
217 | if (isComment(sc.state)) |
218 | lineState = LS_OPEN_COMMENT; |
219 | } else if (sc.state == SCE_TCL_COMMENT_BOX) |
220 | lineState = LS_COMMENT_BOX; |
221 | } |
222 | styler.SetLineState(currentLine, |
223 | (subBrace ? LS_BRACE_ONLY : 0) | |
224 | (expected ? LS_COMMAND_EXPECTED : 0) | lineState); |
225 | if (lineState == LS_COMMENT_BOX) |
226 | sc.ForwardSetState(SCE_TCL_COMMENT_BOX); |
227 | else if (lineState == LS_OPEN_DOUBLE_QUOTE) |
228 | sc.ForwardSetState(SCE_TCL_IN_QUOTE); |
229 | else |
230 | sc.ForwardSetState(SCE_TCL_DEFAULT); |
231 | prevSlash = false; |
232 | previousLevel = currentLevel; |
233 | visibleChars = false; |
234 | goto next; |
235 | } |
236 | |
237 | if (prevSlash) { |
238 | prevSlash = false; |
239 | if (sc.ch == '#' && IsANumberChar(sc.chNext)) |
240 | sc.ForwardSetState(SCE_TCL_NUMBER); |
241 | continue; |
242 | } |
243 | prevSlash = sc.ch == '\\'; |
244 | if (isComment(sc.state)) |
245 | continue; |
246 | if (sc.atLineStart) { |
247 | visibleChars = false; |
248 | if (sc.state!=SCE_TCL_IN_QUOTE && !isComment(sc.state)) |
249 | { |
250 | sc.SetState(SCE_TCL_DEFAULT); |
251 | expected = IsAWordStart(sc.ch)|| isspacechar(static_cast<unsigned char>(sc.ch)); |
252 | } |
253 | } |
254 | |
255 | switch (sc.state) { |
256 | case SCE_TCL_NUMBER: |
257 | if (!IsANumberChar(sc.ch)) |
258 | sc.SetState(SCE_TCL_DEFAULT); |
259 | break; |
260 | case SCE_TCL_IN_QUOTE: |
261 | if (sc.ch == '"') { |
262 | sc.ForwardSetState(SCE_TCL_DEFAULT); |
263 | visibleChars = true; // necessary if a " is the first and only character on a line |
264 | goto next; |
265 | } else if (sc.ch == '[' || sc.ch == ']' || sc.ch == '$') { |
266 | sc.SetState(SCE_TCL_OPERATOR); |
267 | expected = sc.ch == '['; |
268 | sc.ForwardSetState(SCE_TCL_IN_QUOTE); |
269 | goto next; |
270 | } |
271 | continue; |
272 | case SCE_TCL_OPERATOR: |
273 | sc.SetState(SCE_TCL_DEFAULT); |
274 | break; |
275 | } |
276 | |
277 | if (sc.ch == '#') { |
278 | if (visibleChars) { |
279 | if (sc.state != SCE_TCL_IN_QUOTE && expected) |
280 | sc.SetState(SCE_TCL_COMMENT); |
281 | } else { |
282 | sc.SetState(SCE_TCL_COMMENTLINE); |
283 | if (sc.chNext == '~') |
284 | sc.SetState(SCE_TCL_BLOCK_COMMENT); |
285 | if (sc.atLineStart && (sc.chNext == '#' || sc.chNext == '-')) |
286 | sc.SetState(SCE_TCL_COMMENT_BOX); |
287 | } |
288 | } |
289 | |
290 | if (!isspacechar(static_cast<unsigned char>(sc.ch))) { |
291 | visibleChars = true; |
292 | } |
293 | |
294 | if (sc.ch == '\\') { |
295 | prevSlash = true; |
296 | continue; |
297 | } |
298 | |
299 | // Determine if a new state should be entered. |
300 | if (sc.state == SCE_TCL_DEFAULT) { |
301 | if (IsAWordStart(sc.ch)) { |
302 | sc.SetState(SCE_TCL_IDENTIFIER); |
303 | } else if (IsADigit(sc.ch) && !IsAWordChar(sc.chPrev)) { |
304 | sc.SetState(SCE_TCL_NUMBER); |
305 | } else { |
306 | switch (sc.ch) { |
307 | case '\"': |
308 | sc.SetState(SCE_TCL_IN_QUOTE); |
309 | break; |
310 | case '{': |
311 | sc.SetState(SCE_TCL_OPERATOR); |
312 | expected = true; |
313 | ++currentLevel; |
314 | break; |
315 | case '}': |
316 | sc.SetState(SCE_TCL_OPERATOR); |
317 | expected = true; |
318 | --currentLevel; |
319 | break; |
320 | case '[': |
321 | expected = true; |
322 | [[fallthrough]]; |
323 | case ']': |
324 | case '(': |
325 | case ')': |
326 | sc.SetState(SCE_TCL_OPERATOR); |
327 | break; |
328 | case ';': |
329 | expected = true; |
330 | break; |
331 | case '$': |
332 | subParen = 0; |
333 | if (sc.chNext != '{') { |
334 | sc.SetState(SCE_TCL_SUBSTITUTION); |
335 | } else { |
336 | sc.SetState(SCE_TCL_OPERATOR); // $ |
337 | sc.Forward(); // { |
338 | sc.ForwardSetState(SCE_TCL_SUB_BRACE); |
339 | subBrace = true; |
340 | } |
341 | break; |
342 | case '#': |
343 | if ((isspacechar(static_cast<unsigned char>(sc.chPrev))|| |
344 | isoperator(static_cast<char>(sc.chPrev))) && IsADigit(sc.chNext,0x10)) |
345 | sc.SetState(SCE_TCL_NUMBER); |
346 | break; |
347 | case '-': |
348 | sc.SetState(IsADigit(sc.chNext)? SCE_TCL_NUMBER: SCE_TCL_MODIFIER); |
349 | break; |
350 | default: |
351 | if (isoperator(static_cast<char>(sc.ch))) { |
352 | sc.SetState(SCE_TCL_OPERATOR); |
353 | } |
354 | } |
355 | } |
356 | } |
357 | } |
358 | sc.Complete(); |
359 | } |
360 | |
361 | static const char *const tclWordListDesc[] = { |
362 | "TCL Keywords" , |
363 | "TK Keywords" , |
364 | "iTCL Keywords" , |
365 | "tkCommands" , |
366 | "expand" , |
367 | "user1" , |
368 | "user2" , |
369 | "user3" , |
370 | "user4" , |
371 | 0 |
372 | }; |
373 | |
374 | // this code supports folding in the colourizer |
375 | LexerModule lmTCL(SCLEX_TCL, ColouriseTCLDoc, "tcl" , 0, tclWordListDesc); |
376 | |