]>
Commit | Line | Data |
---|---|---|
591d01be RD |
1 | // Scintilla source code edit control |
2 | /** @file LexBash.cxx | |
3 | ** Lexer for Bash. | |
4 | **/ | |
1dcf666d | 5 | // Copyright 2004-2010 by Neil Hodgson <neilh@scintilla.org> |
9e96e16f | 6 | // Adapted from LexPerl by Kein-Hong Man 2004 |
591d01be RD |
7 | // The License.txt file describes the conditions under which this software may be distributed. |
8 | ||
9 | #include <stdlib.h> | |
10 | #include <string.h> | |
591d01be RD |
11 | #include <stdio.h> |
12 | #include <stdarg.h> | |
1dcf666d | 13 | #include <assert.h> |
591d01be | 14 | |
1dcf666d RD |
15 | #include "ILexer.h" |
16 | #include "Scintilla.h" | |
17 | #include "SciLexer.h" | |
591d01be | 18 | |
1dcf666d RD |
19 | #include "WordList.h" |
20 | #include "LexAccessor.h" | |
591d01be | 21 | #include "Accessor.h" |
9e96e16f | 22 | #include "StyleContext.h" |
9e96e16f | 23 | #include "CharacterSet.h" |
1dcf666d | 24 | #include "LexerModule.h" |
9e96e16f RD |
25 | |
26 | #ifdef SCI_NAMESPACE | |
27 | using namespace Scintilla; | |
28 | #endif | |
29 | ||
1dcf666d | 30 | #define HERE_DELIM_MAX 256 |
591d01be | 31 | |
7e0c58e9 RD |
32 | // define this if you want 'invalid octals' to be marked as errors |
33 | // usually, this is not a good idea, permissive lexing is better | |
34 | #undef PEDANTIC_OCTAL | |
35 | ||
1dcf666d RD |
36 | #define BASH_BASE_ERROR 65 |
37 | #define BASH_BASE_DECIMAL 66 | |
38 | #define BASH_BASE_HEX 67 | |
7e0c58e9 | 39 | #ifdef PEDANTIC_OCTAL |
1dcf666d RD |
40 | #define BASH_BASE_OCTAL 68 |
41 | #define BASH_BASE_OCTAL_ERROR 69 | |
7e0c58e9 | 42 | #endif |
591d01be | 43 | |
1dcf666d RD |
44 | // state constants for parts of a bash command segment |
45 | #define BASH_CMD_BODY 0 | |
46 | #define BASH_CMD_START 1 | |
47 | #define BASH_CMD_WORD 2 | |
48 | #define BASH_CMD_TEST 3 | |
49 | #define BASH_CMD_ARITH 4 | |
50 | #define BASH_CMD_DELIM 5 | |
51 | ||
9e96e16f | 52 | static inline int translateBashDigit(int ch) { |
591d01be RD |
53 | if (ch >= '0' && ch <= '9') { |
54 | return ch - '0'; | |
55 | } else if (ch >= 'a' && ch <= 'z') { | |
56 | return ch - 'a' + 10; | |
57 | } else if (ch >= 'A' && ch <= 'Z') { | |
58 | return ch - 'A' + 36; | |
59 | } else if (ch == '@') { | |
60 | return 62; | |
61 | } else if (ch == '_') { | |
62 | return 63; | |
63 | } | |
64 | return BASH_BASE_ERROR; | |
65 | } | |
66 | ||
9e96e16f RD |
67 | static inline int getBashNumberBase(char *s) { |
68 | int i = 0; | |
591d01be | 69 | int base = 0; |
9e96e16f RD |
70 | while (*s) { |
71 | base = base * 10 + (*s++ - '0'); | |
72 | i++; | |
591d01be | 73 | } |
9e96e16f | 74 | if (base > 64 || i > 2) { |
591d01be RD |
75 | return BASH_BASE_ERROR; |
76 | } | |
77 | return base; | |
78 | } | |
79 | ||
9e96e16f RD |
80 | static int opposite(int ch) { |
81 | if (ch == '(') return ')'; | |
82 | if (ch == '[') return ']'; | |
83 | if (ch == '{') return '}'; | |
84 | if (ch == '<') return '>'; | |
591d01be RD |
85 | return ch; |
86 | } | |
87 | ||
88 | static void ColouriseBashDoc(unsigned int startPos, int length, int initStyle, | |
9e96e16f | 89 | WordList *keywordlists[], Accessor &styler) { |
591d01be RD |
90 | |
91 | WordList &keywords = *keywordlists[0]; | |
1dcf666d RD |
92 | WordList cmdDelimiter, bashStruct, bashStruct_in; |
93 | cmdDelimiter.Set("| || |& & && ; ;; ( ) { }"); | |
94 | bashStruct.Set("if elif fi while until else then do done esac eval"); | |
95 | bashStruct_in.Set("for case select"); | |
591d01be | 96 | |
9e96e16f RD |
97 | CharacterSet setWordStart(CharacterSet::setAlpha, "_"); |
98 | // note that [+-] are often parts of identifiers in shell scripts | |
99 | CharacterSet setWord(CharacterSet::setAlphaNum, "._+-"); | |
1dcf666d | 100 | CharacterSet setBashOperator(CharacterSet::setNone, "^&%()-+=|{}[]:;>,*/<?!.~@"); |
9e96e16f RD |
101 | CharacterSet setSingleCharOp(CharacterSet::setNone, "rwxoRWXOezsfdlpSbctugkTBMACahGLNn"); |
102 | CharacterSet setParam(CharacterSet::setAlphaNum, "$_"); | |
103 | CharacterSet setHereDoc(CharacterSet::setAlpha, "_\\-+!"); | |
104 | CharacterSet setHereDoc2(CharacterSet::setAlphaNum, "_-+!"); | |
105 | CharacterSet setLeftShift(CharacterSet::setDigits, "=$"); | |
106 | ||
107 | class HereDocCls { // Class to manage HERE document elements | |
591d01be RD |
108 | public: |
109 | int State; // 0: '<<' encountered | |
110 | // 1: collect the delimiter | |
111 | // 2: here doc text (lines after the delimiter) | |
9e96e16f | 112 | int Quote; // the char after '<<' |
591d01be RD |
113 | bool Quoted; // true if Quote in ('\'','"','`') |
114 | bool Indent; // indented delimiter (for <<-) | |
115 | int DelimiterLength; // strlen(Delimiter) | |
116 | char *Delimiter; // the Delimiter, 256: sizeof PL_tokenbuf | |
117 | HereDocCls() { | |
118 | State = 0; | |
9e96e16f RD |
119 | Quote = 0; |
120 | Quoted = false; | |
121 | Indent = 0; | |
591d01be RD |
122 | DelimiterLength = 0; |
123 | Delimiter = new char[HERE_DELIM_MAX]; | |
124 | Delimiter[0] = '\0'; | |
125 | } | |
9e96e16f RD |
126 | void Append(int ch) { |
127 | Delimiter[DelimiterLength++] = static_cast<char>(ch); | |
128 | Delimiter[DelimiterLength] = '\0'; | |
129 | } | |
591d01be RD |
130 | ~HereDocCls() { |
131 | delete []Delimiter; | |
132 | } | |
133 | }; | |
134 | HereDocCls HereDoc; | |
135 | ||
9e96e16f | 136 | class QuoteCls { // Class to manage quote pairs (simplified vs LexPerl) |
591d01be | 137 | public: |
9e96e16f RD |
138 | int Count; |
139 | int Up, Down; | |
591d01be | 140 | QuoteCls() { |
591d01be RD |
141 | Count = 0; |
142 | Up = '\0'; | |
143 | Down = '\0'; | |
144 | } | |
9e96e16f | 145 | void Open(int u) { |
591d01be RD |
146 | Count++; |
147 | Up = u; | |
148 | Down = opposite(Up); | |
149 | } | |
9e96e16f RD |
150 | void Start(int u) { |
151 | Count = 0; | |
152 | Open(u); | |
153 | } | |
591d01be RD |
154 | }; |
155 | QuoteCls Quote; | |
156 | ||
591d01be | 157 | int numBase = 0; |
9e96e16f RD |
158 | int digit; |
159 | unsigned int endPos = startPos + length; | |
1dcf666d RD |
160 | int cmdState = BASH_CMD_START; |
161 | int testExprType = 0; | |
591d01be | 162 | |
1dcf666d RD |
163 | // Always backtracks to the start of a line that is not a continuation |
164 | // of the previous line (i.e. start of a bash command segment) | |
165 | int ln = styler.GetLine(startPos); | |
166 | for (;;) { | |
167 | startPos = styler.LineStart(ln); | |
168 | if (ln == 0 || styler.GetLineState(ln) == BASH_CMD_START) | |
169 | break; | |
170 | ln--; | |
591d01be | 171 | } |
1dcf666d | 172 | initStyle = SCE_SH_DEFAULT; |
591d01be | 173 | |
9e96e16f RD |
174 | StyleContext sc(startPos, endPos - startPos, initStyle, styler); |
175 | ||
176 | for (; sc.More(); sc.Forward()) { | |
177 | ||
1dcf666d RD |
178 | // handle line continuation, updates per-line stored state |
179 | if (sc.atLineStart) { | |
180 | ln = styler.GetLine(sc.currentPos); | |
181 | if (sc.state == SCE_SH_STRING | |
182 | || sc.state == SCE_SH_BACKTICKS | |
183 | || sc.state == SCE_SH_CHARACTER | |
184 | || sc.state == SCE_SH_HERE_Q | |
185 | || sc.state == SCE_SH_COMMENTLINE | |
186 | || sc.state == SCE_SH_PARAM) { | |
187 | // force backtrack while retaining cmdState | |
188 | styler.SetLineState(ln, BASH_CMD_BODY); | |
189 | } else { | |
190 | if (ln > 0) { | |
191 | if ((sc.GetRelative(-3) == '\\' && sc.GetRelative(-2) == '\r' && sc.chPrev == '\n') | |
192 | || sc.GetRelative(-2) == '\\') { // handle '\' line continuation | |
193 | // retain last line's state | |
194 | } else | |
195 | cmdState = BASH_CMD_START; | |
196 | } | |
197 | styler.SetLineState(ln, cmdState); | |
198 | } | |
199 | } | |
200 | ||
201 | // controls change of cmdState at the end of a non-whitespace element | |
202 | // states BODY|TEST|ARITH persist until the end of a command segment | |
203 | // state WORD persist, but ends with 'in' or 'do' construct keywords | |
204 | int cmdStateNew = BASH_CMD_BODY; | |
205 | if (cmdState == BASH_CMD_TEST || cmdState == BASH_CMD_ARITH || cmdState == BASH_CMD_WORD) | |
206 | cmdStateNew = cmdState; | |
207 | int stylePrev = sc.state; | |
208 | ||
9e96e16f RD |
209 | // Determine if the current state should terminate. |
210 | switch (sc.state) { | |
211 | case SCE_SH_OPERATOR: | |
212 | sc.SetState(SCE_SH_DEFAULT); | |
1dcf666d RD |
213 | if (cmdState == BASH_CMD_DELIM) // if command delimiter, start new command |
214 | cmdStateNew = BASH_CMD_START; | |
215 | else if (sc.chPrev == '\\') // propagate command state if line continued | |
216 | cmdStateNew = cmdState; | |
9e96e16f RD |
217 | break; |
218 | case SCE_SH_WORD: | |
219 | // "." never used in Bash variable names but used in file names | |
220 | if (!setWord.Contains(sc.ch)) { | |
1dcf666d RD |
221 | char s[500]; |
222 | char s2[10]; | |
9e96e16f | 223 | sc.GetCurrent(s, sizeof(s)); |
1dcf666d RD |
224 | // allow keywords ending in a whitespace or command delimiter |
225 | s2[0] = static_cast<char>(sc.ch); | |
226 | s2[1] = '\0'; | |
227 | bool keywordEnds = IsASpace(sc.ch) || cmdDelimiter.InList(s2); | |
228 | // 'in' or 'do' may be construct keywords | |
229 | if (cmdState == BASH_CMD_WORD) { | |
230 | if (strcmp(s, "in") == 0 && keywordEnds) | |
231 | cmdStateNew = BASH_CMD_BODY; | |
232 | else if (strcmp(s, "do") == 0 && keywordEnds) | |
233 | cmdStateNew = BASH_CMD_START; | |
234 | else | |
235 | sc.ChangeState(SCE_SH_IDENTIFIER); | |
236 | sc.SetState(SCE_SH_DEFAULT); | |
237 | break; | |
238 | } | |
239 | // a 'test' keyword starts a test expression | |
240 | if (strcmp(s, "test") == 0) { | |
241 | if (cmdState == BASH_CMD_START && keywordEnds) { | |
242 | cmdStateNew = BASH_CMD_TEST; | |
243 | testExprType = 0; | |
244 | } else | |
245 | sc.ChangeState(SCE_SH_IDENTIFIER); | |
246 | } | |
247 | // detect bash construct keywords | |
248 | else if (bashStruct.InList(s)) { | |
249 | if (cmdState == BASH_CMD_START && keywordEnds) | |
250 | cmdStateNew = BASH_CMD_START; | |
251 | else | |
252 | sc.ChangeState(SCE_SH_IDENTIFIER); | |
253 | } | |
254 | // 'for'|'case'|'select' needs 'in'|'do' to be highlighted later | |
255 | else if (bashStruct_in.InList(s)) { | |
256 | if (cmdState == BASH_CMD_START && keywordEnds) | |
257 | cmdStateNew = BASH_CMD_WORD; | |
258 | else | |
259 | sc.ChangeState(SCE_SH_IDENTIFIER); | |
260 | } | |
261 | // disambiguate option items and file test operators | |
262 | else if (s[0] == '-') { | |
263 | if (cmdState != BASH_CMD_TEST) | |
264 | sc.ChangeState(SCE_SH_IDENTIFIER); | |
265 | } | |
266 | // disambiguate keywords and identifiers | |
267 | else if (cmdState != BASH_CMD_START | |
268 | || !(keywords.InList(s) && keywordEnds)) { | |
9e96e16f | 269 | sc.ChangeState(SCE_SH_IDENTIFIER); |
591d01be | 270 | } |
9e96e16f | 271 | sc.SetState(SCE_SH_DEFAULT); |
591d01be | 272 | } |
9e96e16f RD |
273 | break; |
274 | case SCE_SH_IDENTIFIER: | |
275 | if (sc.chPrev == '\\') { // for escaped chars | |
276 | sc.ForwardSetState(SCE_SH_DEFAULT); | |
277 | } else if (!setWord.Contains(sc.ch)) { | |
278 | sc.SetState(SCE_SH_DEFAULT); | |
591d01be | 279 | } |
9e96e16f RD |
280 | break; |
281 | case SCE_SH_NUMBER: | |
282 | digit = translateBashDigit(sc.ch); | |
283 | if (numBase == BASH_BASE_DECIMAL) { | |
284 | if (sc.ch == '#') { | |
285 | char s[10]; | |
286 | sc.GetCurrent(s, sizeof(s)); | |
287 | numBase = getBashNumberBase(s); | |
288 | if (numBase != BASH_BASE_ERROR) | |
289 | break; | |
290 | } else if (IsADigit(sc.ch)) | |
291 | break; | |
292 | } else if (numBase == BASH_BASE_HEX) { | |
293 | if (IsADigit(sc.ch, 16)) | |
294 | break; | |
7e0c58e9 | 295 | #ifdef PEDANTIC_OCTAL |
9e96e16f RD |
296 | } else if (numBase == BASH_BASE_OCTAL || |
297 | numBase == BASH_BASE_OCTAL_ERROR) { | |
298 | if (digit <= 7) | |
299 | break; | |
591d01be | 300 | if (digit <= 9) { |
9e96e16f RD |
301 | numBase = BASH_BASE_OCTAL_ERROR; |
302 | break; | |
591d01be | 303 | } |
9e96e16f RD |
304 | #endif |
305 | } else if (numBase == BASH_BASE_ERROR) { | |
306 | if (digit <= 9) | |
307 | break; | |
308 | } else { // DD#DDDD number style handling | |
309 | if (digit != BASH_BASE_ERROR) { | |
310 | if (numBase <= 36) { | |
311 | // case-insensitive if base<=36 | |
312 | if (digit >= 36) digit -= 26; | |
313 | } | |
314 | if (digit < numBase) | |
315 | break; | |
591d01be RD |
316 | if (digit <= 9) { |
317 | numBase = BASH_BASE_ERROR; | |
9e96e16f RD |
318 | break; |
319 | } | |
591d01be | 320 | } |
9e96e16f RD |
321 | } |
322 | // fallthrough when number is at an end or error | |
323 | if (numBase == BASH_BASE_ERROR | |
7e0c58e9 | 324 | #ifdef PEDANTIC_OCTAL |
9e96e16f | 325 | || numBase == BASH_BASE_OCTAL_ERROR |
7e0c58e9 | 326 | #endif |
9e96e16f RD |
327 | ) { |
328 | sc.ChangeState(SCE_SH_ERROR); | |
591d01be | 329 | } |
9e96e16f RD |
330 | sc.SetState(SCE_SH_DEFAULT); |
331 | break; | |
332 | case SCE_SH_COMMENTLINE: | |
1dcf666d RD |
333 | if (sc.atLineEnd && sc.chPrev != '\\') { |
334 | sc.SetState(SCE_SH_DEFAULT); | |
591d01be | 335 | } |
9e96e16f RD |
336 | break; |
337 | case SCE_SH_HERE_DELIM: | |
591d01be RD |
338 | // From Bash info: |
339 | // --------------- | |
340 | // Specifier format is: <<[-]WORD | |
341 | // Optional '-' is for removal of leading tabs from here-doc. | |
342 | // Whitespace acceptable after <<[-] operator | |
343 | // | |
344 | if (HereDoc.State == 0) { // '<<' encountered | |
9e96e16f | 345 | HereDoc.Quote = sc.chNext; |
591d01be RD |
346 | HereDoc.Quoted = false; |
347 | HereDoc.DelimiterLength = 0; | |
348 | HereDoc.Delimiter[HereDoc.DelimiterLength] = '\0'; | |
9e96e16f RD |
349 | if (sc.chNext == '\'' || sc.chNext == '\"') { // a quoted here-doc delimiter (' or ") |
350 | sc.Forward(); | |
591d01be | 351 | HereDoc.Quoted = true; |
9e96e16f RD |
352 | HereDoc.State = 1; |
353 | } else if (!HereDoc.Indent && sc.chNext == '-') { // <<- indent case | |
591d01be | 354 | HereDoc.Indent = true; |
9e96e16f | 355 | } else if (setHereDoc.Contains(sc.chNext)) { |
591d01be | 356 | // an unquoted here-doc delimiter, no special handling |
9e96e16f RD |
357 | // TODO check what exactly bash considers part of the delim |
358 | HereDoc.State = 1; | |
359 | } else if (sc.chNext == '<') { // HERE string <<< | |
360 | sc.Forward(); | |
361 | sc.ForwardSetState(SCE_SH_DEFAULT); | |
362 | } else if (IsASpace(sc.chNext)) { | |
591d01be | 363 | // eat whitespace |
9e96e16f | 364 | } else if (setLeftShift.Contains(sc.chNext)) { |
591d01be | 365 | // left shift << or <<= operator cases |
9e96e16f RD |
366 | sc.ChangeState(SCE_SH_OPERATOR); |
367 | sc.ForwardSetState(SCE_SH_DEFAULT); | |
591d01be RD |
368 | } else { |
369 | // symbols terminates; deprecated zero-length delimiter | |
9e96e16f | 370 | HereDoc.State = 1; |
591d01be RD |
371 | } |
372 | } else if (HereDoc.State == 1) { // collect the delimiter | |
1dcf666d RD |
373 | if (setHereDoc2.Contains(sc.ch) || sc.chPrev == '\\') { |
374 | HereDoc.Append(sc.ch); | |
375 | } else if (HereDoc.Quoted && sc.ch == HereDoc.Quote) { // closing quote => end of delimiter | |
376 | sc.ForwardSetState(SCE_SH_DEFAULT); | |
377 | } else if (sc.ch == '\\') { | |
378 | // skip escape prefix | |
379 | } else { | |
380 | sc.SetState(SCE_SH_DEFAULT); | |
591d01be | 381 | } |
9e96e16f RD |
382 | if (HereDoc.DelimiterLength >= HERE_DELIM_MAX - 1) { // force blowup |
383 | sc.SetState(SCE_SH_ERROR); | |
384 | HereDoc.State = 0; | |
591d01be RD |
385 | } |
386 | } | |
9e96e16f RD |
387 | break; |
388 | case SCE_SH_HERE_Q: | |
389 | // HereDoc.State == 2 | |
390 | if (sc.atLineStart) { | |
391 | sc.SetState(SCE_SH_HERE_Q); | |
392 | int prefixws = 0; | |
393 | while (IsASpace(sc.ch) && !sc.atLineEnd) { // whitespace prefix | |
394 | sc.Forward(); | |
395 | prefixws++; | |
396 | } | |
397 | if (prefixws > 0) | |
398 | sc.SetState(SCE_SH_HERE_Q); | |
399 | while (!sc.atLineEnd) { | |
400 | sc.Forward(); | |
401 | } | |
402 | char s[HERE_DELIM_MAX]; | |
403 | sc.GetCurrent(s, sizeof(s)); | |
404 | if (sc.LengthCurrent() == 0) | |
405 | break; | |
406 | if (s[strlen(s) - 1] == '\r') | |
407 | s[strlen(s) - 1] = '\0'; | |
408 | if (strcmp(HereDoc.Delimiter, s) == 0) { | |
1dcf666d RD |
409 | if ((prefixws == 0) || // indentation rule |
410 | (prefixws > 0 && HereDoc.Indent)) { | |
9e96e16f RD |
411 | sc.SetState(SCE_SH_DEFAULT); |
412 | break; | |
591d01be RD |
413 | } |
414 | } | |
415 | } | |
9e96e16f RD |
416 | break; |
417 | case SCE_SH_SCALAR: // variable names | |
418 | if (!setParam.Contains(sc.ch)) { | |
419 | if (sc.LengthCurrent() == 1) { | |
591d01be | 420 | // Special variable: $(, $_ etc. |
9e96e16f | 421 | sc.ForwardSetState(SCE_SH_DEFAULT); |
591d01be | 422 | } else { |
9e96e16f | 423 | sc.SetState(SCE_SH_DEFAULT); |
591d01be RD |
424 | } |
425 | } | |
9e96e16f RD |
426 | break; |
427 | case SCE_SH_STRING: // delimited styles | |
9e96e16f RD |
428 | case SCE_SH_BACKTICKS: |
429 | case SCE_SH_PARAM: | |
430 | if (sc.ch == '\\' && Quote.Up != '\\') { | |
431 | sc.Forward(); | |
432 | } else if (sc.ch == Quote.Down) { | |
591d01be RD |
433 | Quote.Count--; |
434 | if (Quote.Count == 0) { | |
9e96e16f | 435 | sc.ForwardSetState(SCE_SH_DEFAULT); |
591d01be | 436 | } |
9e96e16f | 437 | } else if (sc.ch == Quote.Up) { |
591d01be RD |
438 | Quote.Count++; |
439 | } | |
9e96e16f | 440 | break; |
1dcf666d RD |
441 | case SCE_SH_CHARACTER: // singly-quoted strings |
442 | if (sc.ch == Quote.Down) { | |
443 | Quote.Count--; | |
444 | if (Quote.Count == 0) { | |
445 | sc.ForwardSetState(SCE_SH_DEFAULT); | |
446 | } | |
447 | } | |
448 | break; | |
9e96e16f RD |
449 | } |
450 | ||
451 | // Must check end of HereDoc state 1 before default state is handled | |
452 | if (HereDoc.State == 1 && sc.atLineEnd) { | |
453 | // Begin of here-doc (the line after the here-doc delimiter): | |
454 | // Lexically, the here-doc starts from the next line after the >>, but the | |
455 | // first line of here-doc seem to follow the style of the last EOL sequence | |
456 | HereDoc.State = 2; | |
457 | if (HereDoc.Quoted) { | |
458 | if (sc.state == SCE_SH_HERE_DELIM) { | |
459 | // Missing quote at end of string! We are stricter than bash. | |
460 | // Colour here-doc anyway while marking this bit as an error. | |
461 | sc.ChangeState(SCE_SH_ERROR); | |
462 | } | |
463 | // HereDoc.Quote always == '\'' | |
591d01be | 464 | } |
9e96e16f | 465 | sc.SetState(SCE_SH_HERE_Q); |
591d01be | 466 | } |
9e96e16f | 467 | |
1dcf666d RD |
468 | // update cmdState about the current command segment |
469 | if (stylePrev != SCE_SH_DEFAULT && sc.state == SCE_SH_DEFAULT) { | |
470 | cmdState = cmdStateNew; | |
471 | } | |
9e96e16f RD |
472 | // Determine if a new state should be entered. |
473 | if (sc.state == SCE_SH_DEFAULT) { | |
1dcf666d RD |
474 | if (sc.ch == '\\') { |
475 | // Bash can escape any non-newline as a literal | |
9e96e16f | 476 | sc.SetState(SCE_SH_IDENTIFIER); |
1dcf666d RD |
477 | if (sc.chNext == '\r' || sc.chNext == '\n') |
478 | sc.SetState(SCE_SH_OPERATOR); | |
9e96e16f RD |
479 | } else if (IsADigit(sc.ch)) { |
480 | sc.SetState(SCE_SH_NUMBER); | |
481 | numBase = BASH_BASE_DECIMAL; | |
482 | if (sc.ch == '0') { // hex,octal | |
483 | if (sc.chNext == 'x' || sc.chNext == 'X') { | |
484 | numBase = BASH_BASE_HEX; | |
485 | sc.Forward(); | |
486 | } else if (IsADigit(sc.chNext)) { | |
487 | #ifdef PEDANTIC_OCTAL | |
488 | numBase = BASH_BASE_OCTAL; | |
489 | #else | |
490 | numBase = BASH_BASE_HEX; | |
491 | #endif | |
492 | } | |
493 | } | |
494 | } else if (setWordStart.Contains(sc.ch)) { | |
495 | sc.SetState(SCE_SH_WORD); | |
496 | } else if (sc.ch == '#') { | |
497 | sc.SetState(SCE_SH_COMMENTLINE); | |
498 | } else if (sc.ch == '\"') { | |
499 | sc.SetState(SCE_SH_STRING); | |
500 | Quote.Start(sc.ch); | |
501 | } else if (sc.ch == '\'') { | |
502 | sc.SetState(SCE_SH_CHARACTER); | |
503 | Quote.Start(sc.ch); | |
504 | } else if (sc.ch == '`') { | |
505 | sc.SetState(SCE_SH_BACKTICKS); | |
506 | Quote.Start(sc.ch); | |
507 | } else if (sc.ch == '$') { | |
1dcf666d RD |
508 | if (sc.Match("$((")) { |
509 | sc.SetState(SCE_SH_OPERATOR); // handle '((' later | |
510 | continue; | |
511 | } | |
9e96e16f RD |
512 | sc.SetState(SCE_SH_SCALAR); |
513 | sc.Forward(); | |
514 | if (sc.ch == '{') { | |
515 | sc.ChangeState(SCE_SH_PARAM); | |
516 | } else if (sc.ch == '\'') { | |
1dcf666d | 517 | sc.ChangeState(SCE_SH_STRING); |
9e96e16f RD |
518 | } else if (sc.ch == '"') { |
519 | sc.ChangeState(SCE_SH_STRING); | |
520 | } else if (sc.ch == '(' || sc.ch == '`') { | |
521 | sc.ChangeState(SCE_SH_BACKTICKS); | |
9e96e16f RD |
522 | } else { |
523 | continue; // scalar has no delimiter pair | |
524 | } | |
525 | // fallthrough, open delim for $[{'"(`] | |
526 | Quote.Start(sc.ch); | |
527 | } else if (sc.Match('<', '<')) { | |
528 | sc.SetState(SCE_SH_HERE_DELIM); | |
529 | HereDoc.State = 0; | |
530 | HereDoc.Indent = false; | |
531 | } else if (sc.ch == '-' && // one-char file test operators | |
532 | setSingleCharOp.Contains(sc.chNext) && | |
533 | !setWord.Contains(sc.GetRelative(2)) && | |
534 | IsASpace(sc.chPrev)) { | |
535 | sc.SetState(SCE_SH_WORD); | |
536 | sc.Forward(); | |
537 | } else if (setBashOperator.Contains(sc.ch)) { | |
1dcf666d RD |
538 | char s[10]; |
539 | bool isCmdDelim = false; | |
9e96e16f | 540 | sc.SetState(SCE_SH_OPERATOR); |
1dcf666d RD |
541 | // handle opening delimiters for test/arithmetic expressions - ((,[[,[ |
542 | if (cmdState == BASH_CMD_START | |
543 | || cmdState == BASH_CMD_BODY) { | |
544 | if (sc.Match('(', '(')) { | |
545 | cmdState = BASH_CMD_ARITH; | |
546 | sc.Forward(); | |
547 | } else if (sc.Match('[', '[') && IsASpace(sc.GetRelative(2))) { | |
548 | cmdState = BASH_CMD_TEST; | |
549 | testExprType = 1; | |
550 | sc.Forward(); | |
551 | } else if (sc.ch == '[' && IsASpace(sc.chNext)) { | |
552 | cmdState = BASH_CMD_TEST; | |
553 | testExprType = 2; | |
554 | } | |
555 | } | |
556 | // special state -- for ((x;y;z)) in ... looping | |
557 | if (cmdState == BASH_CMD_WORD && sc.Match('(', '(')) { | |
558 | cmdState = BASH_CMD_ARITH; | |
559 | sc.Forward(); | |
560 | continue; | |
561 | } | |
562 | // handle command delimiters in command START|BODY|WORD state, also TEST if 'test' | |
563 | if (cmdState == BASH_CMD_START | |
564 | || cmdState == BASH_CMD_BODY | |
565 | || cmdState == BASH_CMD_WORD | |
566 | || (cmdState == BASH_CMD_TEST && testExprType == 0)) { | |
567 | s[0] = static_cast<char>(sc.ch); | |
568 | if (setBashOperator.Contains(sc.chNext)) { | |
569 | s[1] = static_cast<char>(sc.chNext); | |
570 | s[2] = '\0'; | |
571 | isCmdDelim = cmdDelimiter.InList(s); | |
572 | if (isCmdDelim) | |
573 | sc.Forward(); | |
574 | } | |
575 | if (!isCmdDelim) { | |
576 | s[1] = '\0'; | |
577 | isCmdDelim = cmdDelimiter.InList(s); | |
578 | } | |
579 | if (isCmdDelim) { | |
580 | cmdState = BASH_CMD_DELIM; | |
581 | continue; | |
582 | } | |
583 | } | |
584 | // handle closing delimiters for test/arithmetic expressions - )),]],] | |
585 | if (cmdState == BASH_CMD_ARITH && sc.Match(')', ')')) { | |
586 | cmdState = BASH_CMD_BODY; | |
587 | sc.Forward(); | |
588 | } else if (cmdState == BASH_CMD_TEST && IsASpace(sc.chPrev)) { | |
589 | if (sc.Match(']', ']') && testExprType == 1) { | |
590 | sc.Forward(); | |
591 | cmdState = BASH_CMD_BODY; | |
592 | } else if (sc.ch == ']' && testExprType == 2) { | |
593 | cmdState = BASH_CMD_BODY; | |
594 | } | |
595 | } | |
9e96e16f | 596 | } |
1dcf666d | 597 | }// sc.state |
591d01be | 598 | } |
9e96e16f | 599 | sc.Complete(); |
591d01be RD |
600 | } |
601 | ||
1e9bafca RD |
602 | static bool IsCommentLine(int line, Accessor &styler) { |
603 | int pos = styler.LineStart(line); | |
604 | int eol_pos = styler.LineStart(line + 1) - 1; | |
605 | for (int i = pos; i < eol_pos; i++) { | |
606 | char ch = styler[i]; | |
607 | if (ch == '#') | |
608 | return true; | |
609 | else if (ch != ' ' && ch != '\t') | |
610 | return false; | |
611 | } | |
612 | return false; | |
613 | } | |
614 | ||
591d01be | 615 | static void FoldBashDoc(unsigned int startPos, int length, int, WordList *[], |
9e96e16f | 616 | Accessor &styler) { |
591d01be RD |
617 | bool foldComment = styler.GetPropertyInt("fold.comment") != 0; |
618 | bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0; | |
619 | unsigned int endPos = startPos + length; | |
620 | int visibleChars = 0; | |
621 | int lineCurrent = styler.GetLine(startPos); | |
622 | int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK; | |
623 | int levelCurrent = levelPrev; | |
624 | char chNext = styler[startPos]; | |
625 | int styleNext = styler.StyleAt(startPos); | |
626 | for (unsigned int i = startPos; i < endPos; i++) { | |
627 | char ch = chNext; | |
628 | chNext = styler.SafeGetCharAt(i + 1); | |
629 | int style = styleNext; | |
630 | styleNext = styler.StyleAt(i + 1); | |
631 | bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n'); | |
9e96e16f | 632 | // Comment folding |
1e9bafca | 633 | if (foldComment && atEOL && IsCommentLine(lineCurrent, styler)) |
9e96e16f RD |
634 | { |
635 | if (!IsCommentLine(lineCurrent - 1, styler) | |
636 | && IsCommentLine(lineCurrent + 1, styler)) | |
637 | levelCurrent++; | |
638 | else if (IsCommentLine(lineCurrent - 1, styler) | |
639 | && !IsCommentLine(lineCurrent + 1, styler)) | |
640 | levelCurrent--; | |
641 | } | |
b8193d80 | 642 | if (style == SCE_SH_OPERATOR) { |
591d01be RD |
643 | if (ch == '{') { |
644 | levelCurrent++; | |
645 | } else if (ch == '}') { | |
646 | levelCurrent--; | |
647 | } | |
648 | } | |
1dcf666d RD |
649 | // Here Document folding |
650 | if (style == SCE_SH_HERE_DELIM) { | |
651 | if (ch == '<' && chNext == '<') { | |
652 | levelCurrent++; | |
653 | } | |
654 | } else if (style == SCE_SH_HERE_Q && styler.StyleAt(i+1) == SCE_PL_DEFAULT) { | |
655 | levelCurrent--; | |
656 | } | |
591d01be RD |
657 | if (atEOL) { |
658 | int lev = levelPrev; | |
659 | if (visibleChars == 0 && foldCompact) | |
660 | lev |= SC_FOLDLEVELWHITEFLAG; | |
661 | if ((levelCurrent > levelPrev) && (visibleChars > 0)) | |
662 | lev |= SC_FOLDLEVELHEADERFLAG; | |
663 | if (lev != styler.LevelAt(lineCurrent)) { | |
664 | styler.SetLevel(lineCurrent, lev); | |
665 | } | |
666 | lineCurrent++; | |
667 | levelPrev = levelCurrent; | |
668 | visibleChars = 0; | |
669 | } | |
670 | if (!isspacechar(ch)) | |
671 | visibleChars++; | |
672 | } | |
673 | // Fill in the real level of the next line, keeping the current flags as they will be filled in later | |
674 | int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK; | |
675 | styler.SetLevel(lineCurrent, levelPrev | flagsNext); | |
676 | } | |
677 | ||
678 | static const char * const bashWordListDesc[] = { | |
679 | "Keywords", | |
680 | 0 | |
681 | }; | |
682 | ||
683 | LexerModule lmBash(SCLEX_BASH, ColouriseBashDoc, "bash", FoldBashDoc, bashWordListDesc); |