1 // Scintilla source code edit control
2 /** @file LexLua.cxx
3 ** Lexer for Lua language.
4 **
5 ** Written by Paul Winwood.
6 ** Folder by Alexey Yutkin.
7 ** Modified by Marcos E. Wurzius & Philippe Lhoste
8 **/
9
10 #include <stdlib.h>
11 #include <string.h>
12 #include <stdio.h>
13 #include <stdarg.h>
14 #include <assert.h>
15 #include <ctype.h>
16
17 #include "ILexer.h"
18 #include "Scintilla.h"
19 #include "SciLexer.h"
20
21 #include "WordList.h"
22 #include "LexAccessor.h"
23 #include "Accessor.h"
24 #include "StyleContext.h"
25 #include "CharacterSet.h"
26 #include "LexerModule.h"
27
28 #ifdef SCI_NAMESPACE
29 using namespace Scintilla;
30 #endif
31
32 // Test for [=[ ... ]=] delimiters, returns 0 if it's only a [ or ],
33 // return 1 for [[ or ]], returns >=2 for [=[ or ]=] and so on.
34 // The maximum number of '=' characters allowed is 254.
LongDelimCheck(StyleContext & sc)35 static int LongDelimCheck(StyleContext &sc) {
36 int sep = 1;
37 while (sc.GetRelative(sep) == '=' && sep < 0xFF)
38 sep++;
39 if (sc.GetRelative(sep) == sc.ch)
40 return sep;
41 return 0;
42 }
43
ColouriseLuaDoc(unsigned int startPos,int length,int initStyle,WordList * keywordlists[],Accessor & styler)44 static void ColouriseLuaDoc(
45 unsigned int startPos,
46 int length,
47 int initStyle,
48 WordList *keywordlists[],
49 Accessor &styler) {
50
51 WordList &keywords = *keywordlists[0];
52 WordList &keywords2 = *keywordlists[1];
53 WordList &keywords3 = *keywordlists[2];
54 WordList &keywords4 = *keywordlists[3];
55 WordList &keywords5 = *keywordlists[4];
56 WordList &keywords6 = *keywordlists[5];
57 WordList &keywords7 = *keywordlists[6];
58 WordList &keywords8 = *keywordlists[7];
59
60 // Accepts accented characters
61 CharacterSet setWordStart(CharacterSet::setAlpha, "_", 0x80, true);
62 CharacterSet setWord(CharacterSet::setAlphaNum, "_", 0x80, true);
63 // Not exactly following number definition (several dots are seen as OK, etc.)
64 // but probably enough in most cases. [pP] is for hex floats.
65 CharacterSet setNumber(CharacterSet::setDigits, ".-+abcdefpABCDEFP");
66 CharacterSet setExponent(CharacterSet::setNone, "eEpP");
67 CharacterSet setLuaOperator(CharacterSet::setNone, "*/-+()={}~[];<>,.^%:#");
68 CharacterSet setEscapeSkip(CharacterSet::setNone, "\"'\\");
69
70 int currentLine = styler.GetLine(startPos);
71 // Initialize long string [[ ... ]] or block comment --[[ ... ]] nesting level,
72 // if we are inside such a string. Block comment was introduced in Lua 5.0,
73 // blocks with separators [=[ ... ]=] in Lua 5.1.
74 // Continuation of a string (\z whitespace escaping) is controlled by stringWs.
75 int nestLevel = 0;
76 int sepCount = 0;
77 int stringWs = 0;
78 if (initStyle == SCE_LUA_LITERALSTRING || initStyle == SCE_LUA_COMMENT ||
79 initStyle == SCE_LUA_STRING || initStyle == SCE_LUA_CHARACTER) {
80 int lineState = styler.GetLineState(currentLine - 1);
81 nestLevel = lineState >> 9;
82 sepCount = lineState & 0xFF;
83 stringWs = lineState & 0x100;
84 }
85
86 // Do not leak onto next line
87 if (initStyle == SCE_LUA_STRINGEOL || initStyle == SCE_LUA_COMMENTLINE || initStyle == SCE_LUA_PREPROCESSOR) {
88 initStyle = SCE_LUA_DEFAULT;
89 }
90
91 StyleContext sc(startPos, length, initStyle, styler);
92 if (startPos == 0 && sc.ch == '#') {
93 // shbang line: # is a comment only if first char of the script
94 sc.SetState(SCE_LUA_COMMENTLINE);
95 }
96 for (; sc.More(); sc.Forward()) {
97 if (sc.atLineEnd) {
98 // Update the line state, so it can be seen by next line
99 currentLine = styler.GetLine(sc.currentPos);
100 switch (sc.state) {
101 case SCE_LUA_LITERALSTRING:
102 case SCE_LUA_COMMENT:
103 case SCE_LUA_STRING:
104 case SCE_LUA_CHARACTER:
105 // Inside a literal string, block comment or string, we set the line state
106 styler.SetLineState(currentLine, (nestLevel << 9) | stringWs | sepCount);
107 break;
108 default:
109 // Reset the line state
110 styler.SetLineState(currentLine, 0);
111 break;
112 }
113 }
114 if (sc.atLineStart && (sc.state == SCE_LUA_STRING)) {
115 // Prevent SCE_LUA_STRINGEOL from leaking back to previous line
116 sc.SetState(SCE_LUA_STRING);
117 }
118
119 // Handle string line continuation
120 if ((sc.state == SCE_LUA_STRING || sc.state == SCE_LUA_CHARACTER) &&
121 sc.ch == '\\') {
122 if (sc.chNext == '\n' || sc.chNext == '\r') {
123 sc.Forward();
124 if (sc.ch == '\r' && sc.chNext == '\n') {
125 sc.Forward();
126 }
127 continue;
128 }
129 }
130
131 // Determine if the current state should terminate.
132 if (sc.state == SCE_LUA_OPERATOR) {
133 if (sc.ch == ':' && sc.chPrev == ':') { // :: <label> :: forward scan
134 sc.Forward();
135 int ln = 0, maxln = startPos + length - sc.currentPos;
136 int c;
137 while (ln < maxln) { // determine line extent
138 c = sc.GetRelative(ln);
139 if (c == '\r' || c == '\n')
140 break;
141 ln++;
142 }
143 maxln = ln; ln = 0;
144 while (ln < maxln) { // skip over spaces/tabs
145 if (!IsASpaceOrTab(sc.GetRelative(ln)))
146 break;
147 ln++;
148 }
149 int ws1 = ln;
150 if (setWordStart.Contains(sc.GetRelative(ln))) {
151 int i = 0;
152 char s[100];
153 while (ln < maxln) { // get potential label
154 c = sc.GetRelative(ln);
155 if (!setWord.Contains(c))
156 break;
157 if (i < 90)
158 s[i++] = c;
159 ln++;
160 }
161 s[i] = '\0'; int lbl = ln;
162 if (!keywords.InList(s)) {
163 while (ln < maxln) { // skip over spaces/tabs
164 if (!IsASpaceOrTab(sc.GetRelative(ln)))
165 break;
166 ln++;
167 }
168 int ws2 = ln - lbl;
169 if (sc.GetRelative(ln) == ':' && sc.GetRelative(ln + 1) == ':') {
170 // final :: found, complete valid label construct
171 sc.ChangeState(SCE_LUA_LABEL);
172 if (ws1) {
173 sc.SetState(SCE_LUA_DEFAULT);
174 sc.Forward(ws1);
175 }
176 sc.SetState(SCE_LUA_LABEL);
177 sc.Forward(lbl - ws1);
178 if (ws2) {
179 sc.SetState(SCE_LUA_DEFAULT);
180 sc.Forward(ws2);
181 }
182 sc.SetState(SCE_LUA_LABEL);
183 sc.Forward(2);
184 }
185 }
186 }
187 }
188 sc.SetState(SCE_LUA_DEFAULT);
189 } else if (sc.state == SCE_LUA_NUMBER) {
190 // We stop the number definition on non-numerical non-dot non-eEpP non-sign non-hexdigit char
191 if (!setNumber.Contains(sc.ch)) {
192 sc.SetState(SCE_LUA_DEFAULT);
193 } else if (sc.ch == '-' || sc.ch == '+') {
194 if (!setExponent.Contains(sc.chPrev))
195 sc.SetState(SCE_LUA_DEFAULT);
196 }
197 } else if (sc.state == SCE_LUA_IDENTIFIER) {
198 if (!(setWord.Contains(sc.ch) || sc.ch == '.') || sc.Match('.', '.')) {
199 char s[100];
200 sc.GetCurrent(s, sizeof(s));
201 if (keywords.InList(s)) {
202 sc.ChangeState(SCE_LUA_WORD);
203 if (strcmp(s, "goto") == 0) { // goto <label> forward scan
204 sc.SetState(SCE_LUA_DEFAULT);
205 while (IsASpaceOrTab(sc.ch) && !sc.atLineEnd)
206 sc.Forward();
207 if (setWordStart.Contains(sc.ch)) {
208 sc.SetState(SCE_LUA_LABEL);
209 sc.Forward();
210 while (setWord.Contains(sc.ch))
211 sc.Forward();
212 sc.GetCurrent(s, sizeof(s));
213 if (keywords.InList(s))
214 sc.ChangeState(SCE_LUA_WORD);
215 }
216 sc.SetState(SCE_LUA_DEFAULT);
217 }
218 } else if (keywords2.InList(s)) {
219 sc.ChangeState(SCE_LUA_WORD2);
220 } else if (keywords3.InList(s)) {
221 sc.ChangeState(SCE_LUA_WORD3);
222 } else if (keywords4.InList(s)) {
223 sc.ChangeState(SCE_LUA_WORD4);
224 } else if (keywords5.InList(s)) {
225 sc.ChangeState(SCE_LUA_WORD5);
226 } else if (keywords6.InList(s)) {
227 sc.ChangeState(SCE_LUA_WORD6);
228 } else if (keywords7.InList(s)) {
229 sc.ChangeState(SCE_LUA_WORD7);
230 } else if (keywords8.InList(s)) {
231 sc.ChangeState(SCE_LUA_WORD8);
232 }
233 sc.SetState(SCE_LUA_DEFAULT);
234 }
235 } else if (sc.state == SCE_LUA_COMMENTLINE || sc.state == SCE_LUA_PREPROCESSOR) {
236 if (sc.atLineEnd) {
237 sc.ForwardSetState(SCE_LUA_DEFAULT);
238 }
239 } else if (sc.state == SCE_LUA_STRING) {
240 if (stringWs) {
241 if (!IsASpace(sc.ch))
242 stringWs = 0;
243 }
244 if (sc.ch == '\\') {
245 if (setEscapeSkip.Contains(sc.chNext)) {
246 sc.Forward();
247 } else if (sc.chNext == 'z') {
248 sc.Forward();
249 stringWs = 0x100;
250 }
251 } else if (sc.ch == '\"') {
252 sc.ForwardSetState(SCE_LUA_DEFAULT);
253 } else if (stringWs == 0 && sc.atLineEnd) {
254 sc.ChangeState(SCE_LUA_STRINGEOL);
255 sc.ForwardSetState(SCE_LUA_DEFAULT);
256 }
257 } else if (sc.state == SCE_LUA_CHARACTER) {
258 if (stringWs) {
259 if (!IsASpace(sc.ch))
260 stringWs = 0;
261 }
262 if (sc.ch == '\\') {
263 if (setEscapeSkip.Contains(sc.chNext)) {
264 sc.Forward();
265 } else if (sc.chNext == 'z') {
266 sc.Forward();
267 stringWs = 0x100;
268 }
269 } else if (sc.ch == '\'') {
270 sc.ForwardSetState(SCE_LUA_DEFAULT);
271 } else if (stringWs == 0 && sc.atLineEnd) {
272 sc.ChangeState(SCE_LUA_STRINGEOL);
273 sc.ForwardSetState(SCE_LUA_DEFAULT);
274 }
275 } else if (sc.state == SCE_LUA_LITERALSTRING || sc.state == SCE_LUA_COMMENT) {
276 if (sc.ch == '[') {
277 int sep = LongDelimCheck(sc);
278 if (sep == 1 && sepCount == 1) { // [[-only allowed to nest
279 nestLevel++;
280 sc.Forward();
281 }
282 } else if (sc.ch == ']') {
283 int sep = LongDelimCheck(sc);
284 if (sep == 1 && sepCount == 1) { // un-nest with ]]-only
285 nestLevel--;
286 sc.Forward();
287 if (nestLevel == 0) {
288 sc.ForwardSetState(SCE_LUA_DEFAULT);
289 }
290 } else if (sep > 1 && sep == sepCount) { // ]=]-style delim
291 sc.Forward(sep);
292 sc.ForwardSetState(SCE_LUA_DEFAULT);
293 }
294 }
295 }
296
297 // Determine if a new state should be entered.
298 if (sc.state == SCE_LUA_DEFAULT) {
299 if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
300 sc.SetState(SCE_LUA_NUMBER);
301 if (sc.ch == '0' && toupper(sc.chNext) == 'X') {
302 sc.Forward();
303 }
304 } else if (setWordStart.Contains(sc.ch)) {
305 sc.SetState(SCE_LUA_IDENTIFIER);
306 } else if (sc.ch == '\"') {
307 sc.SetState(SCE_LUA_STRING);
308 stringWs = 0;
309 } else if (sc.ch == '\'') {
310 sc.SetState(SCE_LUA_CHARACTER);
311 stringWs = 0;
312 } else if (sc.ch == '[') {
313 sepCount = LongDelimCheck(sc);
314 if (sepCount == 0) {
315 sc.SetState(SCE_LUA_OPERATOR);
316 } else {
317 nestLevel = 1;
318 sc.SetState(SCE_LUA_LITERALSTRING);
319 sc.Forward(sepCount);
320 }
321 } else if (sc.Match('-', '-')) {
322 sc.SetState(SCE_LUA_COMMENTLINE);
323 if (sc.Match("--[")) {
324 sc.Forward(2);
325 sepCount = LongDelimCheck(sc);
326 if (sepCount > 0) {
327 nestLevel = 1;
328 sc.ChangeState(SCE_LUA_COMMENT);
329 sc.Forward(sepCount);
330 }
331 } else {
332 sc.Forward();
333 }
334 } else if (sc.atLineStart && sc.Match('$')) {
335 sc.SetState(SCE_LUA_PREPROCESSOR); // Obsolete since Lua 4.0, but still in old code
336 } else if (setLuaOperator.Contains(sc.ch)) {
337 sc.SetState(SCE_LUA_OPERATOR);
338 }
339 }
340 }
341
342 if (setWord.Contains(sc.chPrev) || sc.chPrev == '.') {
343 char s[100];
344 sc.GetCurrent(s, sizeof(s));
345 if (keywords.InList(s)) {
346 sc.ChangeState(SCE_LUA_WORD);
347 } else if (keywords2.InList(s)) {
348 sc.ChangeState(SCE_LUA_WORD2);
349 } else if (keywords3.InList(s)) {
350 sc.ChangeState(SCE_LUA_WORD3);
351 } else if (keywords4.InList(s)) {
352 sc.ChangeState(SCE_LUA_WORD4);
353 } else if (keywords5.InList(s)) {
354 sc.ChangeState(SCE_LUA_WORD5);
355 } else if (keywords6.InList(s)) {
356 sc.ChangeState(SCE_LUA_WORD6);
357 } else if (keywords7.InList(s)) {
358 sc.ChangeState(SCE_LUA_WORD7);
359 } else if (keywords8.InList(s)) {
360 sc.ChangeState(SCE_LUA_WORD8);
361 }
362 }
363
364 sc.Complete();
365 }
366
FoldLuaDoc(unsigned int startPos,int length,int,WordList * [],Accessor & styler)367 static void FoldLuaDoc(unsigned int startPos, int length, int /* initStyle */, WordList *[],
368 Accessor &styler) {
369 unsigned int lengthDoc = startPos + length;
370 int visibleChars = 0;
371 int lineCurrent = styler.GetLine(startPos);
372 int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
373 int levelCurrent = levelPrev;
374 char chNext = styler[startPos];
375 bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
376 int styleNext = styler.StyleAt(startPos);
377 char s[10];
378
379 for (unsigned int i = startPos; i < lengthDoc; i++) {
380 char ch = chNext;
381 chNext = styler.SafeGetCharAt(i + 1);
382 int style = styleNext;
383 styleNext = styler.StyleAt(i + 1);
384 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
385 if (style == SCE_LUA_WORD) {
386 if (ch == 'i' || ch == 'd' || ch == 'f' || ch == 'e' || ch == 'r' || ch == 'u') {
387 for (unsigned int j = 0; j < 8; j++) {
388 if (!iswordchar(styler[i + j])) {
389 break;
390 }
391 s[j] = styler[i + j];
392 s[j + 1] = '\0';
393 }
394
395 if ((strcmp(s, "if") == 0) || (strcmp(s, "do") == 0) || (strcmp(s, "function") == 0) || (strcmp(s, "repeat") == 0)) {
396 levelCurrent++;
397 }
398 if ((strcmp(s, "end") == 0) || (strcmp(s, "elseif") == 0) || (strcmp(s, "until") == 0)) {
399 levelCurrent--;
400 }
401 }
402 } else if (style == SCE_LUA_OPERATOR) {
403 if (ch == '{' || ch == '(') {
404 levelCurrent++;
405 } else if (ch == '}' || ch == ')') {
406 levelCurrent--;
407 }
408 } else if (style == SCE_LUA_LITERALSTRING || style == SCE_LUA_COMMENT) {
409 if (ch == '[') {
410 levelCurrent++;
411 } else if (ch == ']') {
412 levelCurrent--;
413 }
414 }
415
416 if (atEOL) {
417 int lev = levelPrev;
418 if (visibleChars == 0 && foldCompact) {
419 lev |= SC_FOLDLEVELWHITEFLAG;
420 }
421 if ((levelCurrent > levelPrev) && (visibleChars > 0)) {
422 lev |= SC_FOLDLEVELHEADERFLAG;
423 }
424 if (lev != styler.LevelAt(lineCurrent)) {
425 styler.SetLevel(lineCurrent, lev);
426 }
427 lineCurrent++;
428 levelPrev = levelCurrent;
429 visibleChars = 0;
430 }
431 if (!isspacechar(ch)) {
432 visibleChars++;
433 }
434 }
435 // Fill in the real level of the next line, keeping the current flags as they will be filled in later
436
437 int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
438 styler.SetLevel(lineCurrent, levelPrev | flagsNext);
439 }
440
441 static const char * const luaWordListDesc[] = {
442 "Keywords",
443 "Basic functions",
444 "String, (table) & math functions",
445 "(coroutines), I/O & system facilities",
446 "user1",
447 "user2",
448 "user3",
449 "user4",
450 0
451 };
452
453 LexerModule lmLua(SCLEX_LUA, ColouriseLuaDoc, "lua", FoldLuaDoc, luaWordListDesc);
454