]>
Commit | Line | Data |
---|---|---|
1 | /* | |
2 | ** $Id: llex.c,v 2.63 2013/03/16 21:10:18 roberto Exp $ | |
3 | ** Lexical Analyzer | |
4 | ** See Copyright Notice in lua.h | |
5 | */ | |
6 | ||
7 | ||
8 | #include <locale.h> | |
9 | #include <string.h> | |
10 | ||
11 | #define llex_c | |
12 | #define LUA_CORE | |
13 | ||
14 | #include "lua.h" | |
15 | ||
16 | #include "lctype.h" | |
17 | #include "ldo.h" | |
18 | #include "llex.h" | |
19 | #include "lobject.h" | |
20 | #include "lparser.h" | |
21 | #include "lstate.h" | |
22 | #include "lstring.h" | |
23 | #include "ltable.h" | |
24 | #include "lzio.h" | |
25 | ||
26 | ||
27 | ||
28 | #define next(ls) (ls->current = zgetc(ls->z)) | |
29 | ||
30 | ||
31 | ||
32 | #define currIsNewline(ls) (ls->current == '\n' || ls->current == '\r') | |
33 | ||
34 | ||
35 | /* ORDER RESERVED */ | |
36 | static const char *const luaX_tokens [] = { | |
37 | "and", "break", "do", "else", "elseif", | |
38 | "end", "false", "for", "function", "goto", "if", | |
39 | "in", "local", "nil", "not", "or", "repeat", | |
40 | "return", "then", "true", "until", "while", | |
41 | "..", "...", "==", ">=", "<=", "~=", "::", "<eof>", | |
42 | "<number>", "<name>", "<string>" | |
43 | }; | |
44 | ||
45 | ||
46 | #define save_and_next(ls) (save(ls, ls->current), next(ls)) | |
47 | ||
48 | ||
49 | static l_noret lexerror (LexState *ls, const char *msg, int token); | |
50 | ||
51 | ||
52 | static void save (LexState *ls, int c) { | |
53 | Mbuffer *b = ls->buff; | |
54 | if (luaZ_bufflen(b) + 1 > luaZ_sizebuffer(b)) { | |
55 | size_t newsize; | |
56 | if (luaZ_sizebuffer(b) >= MAX_SIZET/2) | |
57 | lexerror(ls, "lexical element too long", 0); | |
58 | newsize = luaZ_sizebuffer(b) * 2; | |
59 | luaZ_resizebuffer(ls->L, b, newsize); | |
60 | } | |
61 | b->buffer[luaZ_bufflen(b)++] = cast(char, c); | |
62 | } | |
63 | ||
64 | ||
65 | void luaX_init (lua_State *L) { | |
66 | int i; | |
67 | for (i=0; i<NUM_RESERVED; i++) { | |
68 | TString *ts = luaS_new(L, luaX_tokens[i]); | |
69 | luaS_fix(ts); /* reserved words are never collected */ | |
70 | ts->tsv.extra = cast_byte(i+1); /* reserved word */ | |
71 | } | |
72 | } | |
73 | ||
74 | ||
75 | const char *luaX_token2str (LexState *ls, int token) { | |
76 | if (token < FIRST_RESERVED) { /* single-byte symbols? */ | |
77 | lua_assert(token == cast(unsigned char, token)); | |
78 | return (lisprint(token)) ? luaO_pushfstring(ls->L, LUA_QL("%c"), token) : | |
79 | luaO_pushfstring(ls->L, "char(%d)", token); | |
80 | } | |
81 | else { | |
82 | const char *s = luaX_tokens[token - FIRST_RESERVED]; | |
83 | if (token < TK_EOS) /* fixed format (symbols and reserved words)? */ | |
84 | return luaO_pushfstring(ls->L, LUA_QS, s); | |
85 | else /* names, strings, and numerals */ | |
86 | return s; | |
87 | } | |
88 | } | |
89 | ||
90 | ||
91 | static const char *txtToken (LexState *ls, int token) { | |
92 | switch (token) { | |
93 | case TK_NAME: | |
94 | case TK_STRING: | |
95 | case TK_NUMBER: | |
96 | save(ls, '\0'); | |
97 | return luaO_pushfstring(ls->L, LUA_QS, luaZ_buffer(ls->buff)); | |
98 | default: | |
99 | return luaX_token2str(ls, token); | |
100 | } | |
101 | } | |
102 | ||
103 | ||
104 | static l_noret lexerror (LexState *ls, const char *msg, int token) { | |
105 | char buff[LUA_IDSIZE]; | |
106 | luaO_chunkid(buff, getstr(ls->source), LUA_IDSIZE); | |
107 | msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg); | |
108 | if (token) | |
109 | luaO_pushfstring(ls->L, "%s near %s", msg, txtToken(ls, token)); | |
110 | luaD_throw(ls->L, LUA_ERRSYNTAX); | |
111 | } | |
112 | ||
113 | ||
114 | l_noret luaX_syntaxerror (LexState *ls, const char *msg) { | |
115 | lexerror(ls, msg, ls->t.token); | |
116 | } | |
117 | ||
118 | ||
119 | /* | |
120 | ** creates a new string and anchors it in function's table so that | |
121 | ** it will not be collected until the end of the function's compilation | |
122 | ** (by that time it should be anchored in function's prototype) | |
123 | */ | |
124 | TString *luaX_newstring (LexState *ls, const char *str, size_t l) { | |
125 | lua_State *L = ls->L; | |
126 | TValue *o; /* entry for `str' */ | |
127 | TString *ts = luaS_newlstr(L, str, l); /* create new string */ | |
128 | setsvalue2s(L, L->top++, ts); /* temporarily anchor it in stack */ | |
129 | o = luaH_set(L, ls->fs->h, L->top - 1); | |
130 | if (ttisnil(o)) { /* not in use yet? (see 'addK') */ | |
131 | /* boolean value does not need GC barrier; | |
132 | table has no metatable, so it does not need to invalidate cache */ | |
133 | setbvalue(o, 1); /* t[string] = true */ | |
134 | luaC_checkGC(L); | |
135 | } | |
136 | L->top--; /* remove string from stack */ | |
137 | return ts; | |
138 | } | |
139 | ||
140 | ||
141 | /* | |
142 | ** increment line number and skips newline sequence (any of | |
143 | ** \n, \r, \n\r, or \r\n) | |
144 | */ | |
145 | static void inclinenumber (LexState *ls) { | |
146 | int old = ls->current; | |
147 | lua_assert(currIsNewline(ls)); | |
148 | next(ls); /* skip `\n' or `\r' */ | |
149 | if (currIsNewline(ls) && ls->current != old) | |
150 | next(ls); /* skip `\n\r' or `\r\n' */ | |
151 | if (++ls->linenumber >= MAX_INT) | |
152 | luaX_syntaxerror(ls, "chunk has too many lines"); | |
153 | } | |
154 | ||
155 | ||
156 | void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source, | |
157 | int firstchar) { | |
158 | ls->decpoint = '.'; | |
159 | ls->L = L; | |
160 | ls->current = firstchar; | |
161 | ls->lookahead.token = TK_EOS; /* no look-ahead token */ | |
162 | ls->z = z; | |
163 | ls->fs = NULL; | |
164 | ls->linenumber = 1; | |
165 | ls->lastline = 1; | |
166 | ls->source = source; | |
167 | ls->envn = luaS_new(L, LUA_ENV); /* create env name */ | |
168 | luaS_fix(ls->envn); /* never collect this name */ | |
169 | luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER); /* initialize buffer */ | |
170 | } | |
171 | ||
172 | ||
173 | ||
174 | /* | |
175 | ** ======================================================= | |
176 | ** LEXICAL ANALYZER | |
177 | ** ======================================================= | |
178 | */ | |
179 | ||
180 | ||
181 | ||
182 | static int check_next (LexState *ls, const char *set) { | |
183 | if (ls->current == '\0' || !strchr(set, ls->current)) | |
184 | return 0; | |
185 | save_and_next(ls); | |
186 | return 1; | |
187 | } | |
188 | ||
189 | ||
190 | /* | |
191 | ** change all characters 'from' in buffer to 'to' | |
192 | */ | |
193 | static void buffreplace (LexState *ls, char from, char to) { | |
194 | size_t n = luaZ_bufflen(ls->buff); | |
195 | char *p = luaZ_buffer(ls->buff); | |
196 | while (n--) | |
197 | if (p[n] == from) p[n] = to; | |
198 | } | |
199 | ||
200 | ||
201 | #if ANDROID | |
202 | #define getlocaldecpoint() '.' | |
203 | #elif !defined(getlocaledecpoint) | |
204 | #define getlocaledecpoint() (localeconv()->decimal_point[0]) | |
205 | #endif | |
206 | ||
207 | ||
208 | #define buff2d(b,e) luaO_str2d(luaZ_buffer(b), luaZ_bufflen(b) - 1, e) | |
209 | ||
210 | /* | |
211 | ** in case of format error, try to change decimal point separator to | |
212 | ** the one defined in the current locale and check again | |
213 | */ | |
214 | static void trydecpoint (LexState *ls, SemInfo *seminfo) { | |
215 | char old = ls->decpoint; | |
216 | ls->decpoint = getlocaledecpoint(); | |
217 | buffreplace(ls, old, ls->decpoint); /* try new decimal separator */ | |
218 | if (!buff2d(ls->buff, &seminfo->r)) { | |
219 | /* format error with correct decimal point: no more options */ | |
220 | buffreplace(ls, ls->decpoint, '.'); /* undo change (for error message) */ | |
221 | lexerror(ls, "malformed number", TK_NUMBER); | |
222 | } | |
223 | } | |
224 | ||
225 | ||
226 | /* LUA_NUMBER */ | |
227 | /* | |
228 | ** this function is quite liberal in what it accepts, as 'luaO_str2d' | |
229 | ** will reject ill-formed numerals. | |
230 | */ | |
231 | static void read_numeral (LexState *ls, SemInfo *seminfo) { | |
232 | const char *expo = "Ee"; | |
233 | int first = ls->current; | |
234 | lua_assert(lisdigit(ls->current)); | |
235 | save_and_next(ls); | |
236 | if (first == '0' && check_next(ls, "Xx")) /* hexadecimal? */ | |
237 | expo = "Pp"; | |
238 | for (;;) { | |
239 | if (check_next(ls, expo)) /* exponent part? */ | |
240 | check_next(ls, "+-"); /* optional exponent sign */ | |
241 | if (lisxdigit(ls->current) || ls->current == '.') | |
242 | save_and_next(ls); | |
243 | else break; | |
244 | } | |
245 | save(ls, '\0'); | |
246 | buffreplace(ls, '.', ls->decpoint); /* follow locale for decimal point */ | |
247 | if (!buff2d(ls->buff, &seminfo->r)) /* format error? */ | |
248 | trydecpoint(ls, seminfo); /* try to update decimal point separator */ | |
249 | } | |
250 | ||
251 | ||
252 | /* | |
253 | ** skip a sequence '[=*[' or ']=*]' and return its number of '='s or | |
254 | ** -1 if sequence is malformed | |
255 | */ | |
256 | static int skip_sep (LexState *ls) { | |
257 | int count = 0; | |
258 | int s = ls->current; | |
259 | lua_assert(s == '[' || s == ']'); | |
260 | save_and_next(ls); | |
261 | while (ls->current == '=') { | |
262 | save_and_next(ls); | |
263 | count++; | |
264 | } | |
265 | return (ls->current == s) ? count : (-count) - 1; | |
266 | } | |
267 | ||
268 | ||
269 | static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) { | |
270 | save_and_next(ls); /* skip 2nd `[' */ | |
271 | if (currIsNewline(ls)) /* string starts with a newline? */ | |
272 | inclinenumber(ls); /* skip it */ | |
273 | for (;;) { | |
274 | switch (ls->current) { | |
275 | case EOZ: | |
276 | lexerror(ls, (seminfo) ? "unfinished long string" : | |
277 | "unfinished long comment", TK_EOS); | |
278 | break; /* to avoid warnings */ | |
279 | case ']': { | |
280 | if (skip_sep(ls) == sep) { | |
281 | save_and_next(ls); /* skip 2nd `]' */ | |
282 | goto endloop; | |
283 | } | |
284 | break; | |
285 | } | |
286 | case '\n': case '\r': { | |
287 | save(ls, '\n'); | |
288 | inclinenumber(ls); | |
289 | if (!seminfo) luaZ_resetbuffer(ls->buff); /* avoid wasting space */ | |
290 | break; | |
291 | } | |
292 | default: { | |
293 | if (seminfo) save_and_next(ls); | |
294 | else next(ls); | |
295 | } | |
296 | } | |
297 | } endloop: | |
298 | if (seminfo) | |
299 | seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep), | |
300 | luaZ_bufflen(ls->buff) - 2*(2 + sep)); | |
301 | } | |
302 | ||
303 | ||
304 | static void escerror (LexState *ls, int *c, int n, const char *msg) { | |
305 | int i; | |
306 | luaZ_resetbuffer(ls->buff); /* prepare error message */ | |
307 | save(ls, '\\'); | |
308 | for (i = 0; i < n && c[i] != EOZ; i++) | |
309 | save(ls, c[i]); | |
310 | lexerror(ls, msg, TK_STRING); | |
311 | } | |
312 | ||
313 | ||
314 | static int readhexaesc (LexState *ls) { | |
315 | int c[3], i; /* keep input for error message */ | |
316 | int r = 0; /* result accumulator */ | |
317 | c[0] = 'x'; /* for error message */ | |
318 | for (i = 1; i < 3; i++) { /* read two hexadecimal digits */ | |
319 | c[i] = next(ls); | |
320 | if (!lisxdigit(c[i])) | |
321 | escerror(ls, c, i + 1, "hexadecimal digit expected"); | |
322 | r = (r << 4) + luaO_hexavalue(c[i]); | |
323 | } | |
324 | return r; | |
325 | } | |
326 | ||
327 | ||
328 | static int readdecesc (LexState *ls) { | |
329 | int c[3], i; | |
330 | int r = 0; /* result accumulator */ | |
331 | for (i = 0; i < 3 && lisdigit(ls->current); i++) { /* read up to 3 digits */ | |
332 | c[i] = ls->current; | |
333 | r = 10*r + c[i] - '0'; | |
334 | next(ls); | |
335 | } | |
336 | if (r > UCHAR_MAX) | |
337 | escerror(ls, c, i, "decimal escape too large"); | |
338 | return r; | |
339 | } | |
340 | ||
341 | ||
342 | static void read_string (LexState *ls, int del, SemInfo *seminfo) { | |
343 | save_and_next(ls); /* keep delimiter (for error messages) */ | |
344 | while (ls->current != del) { | |
345 | switch (ls->current) { | |
346 | case EOZ: | |
347 | lexerror(ls, "unfinished string", TK_EOS); | |
348 | break; /* to avoid warnings */ | |
349 | case '\n': | |
350 | case '\r': | |
351 | lexerror(ls, "unfinished string", TK_STRING); | |
352 | break; /* to avoid warnings */ | |
353 | case '\\': { /* escape sequences */ | |
354 | int c; /* final character to be saved */ | |
355 | next(ls); /* do not save the `\' */ | |
356 | switch (ls->current) { | |
357 | case 'a': c = '\a'; goto read_save; | |
358 | case 'b': c = '\b'; goto read_save; | |
359 | case 'f': c = '\f'; goto read_save; | |
360 | case 'n': c = '\n'; goto read_save; | |
361 | case 'r': c = '\r'; goto read_save; | |
362 | case 't': c = '\t'; goto read_save; | |
363 | case 'v': c = '\v'; goto read_save; | |
364 | case 'x': c = readhexaesc(ls); goto read_save; | |
365 | case '\n': case '\r': | |
366 | inclinenumber(ls); c = '\n'; goto only_save; | |
367 | case '\\': case '\"': case '\'': | |
368 | c = ls->current; goto read_save; | |
369 | case EOZ: goto no_save; /* will raise an error next loop */ | |
370 | case 'z': { /* zap following span of spaces */ | |
371 | next(ls); /* skip the 'z' */ | |
372 | while (lisspace(ls->current)) { | |
373 | if (currIsNewline(ls)) inclinenumber(ls); | |
374 | else next(ls); | |
375 | } | |
376 | goto no_save; | |
377 | } | |
378 | default: { | |
379 | if (!lisdigit(ls->current)) | |
380 | escerror(ls, &ls->current, 1, "invalid escape sequence"); | |
381 | /* digital escape \ddd */ | |
382 | c = readdecesc(ls); | |
383 | goto only_save; | |
384 | } | |
385 | } | |
386 | read_save: next(ls); /* read next character */ | |
387 | only_save: save(ls, c); /* save 'c' */ | |
388 | no_save: break; | |
389 | } | |
390 | default: | |
391 | save_and_next(ls); | |
392 | } | |
393 | } | |
394 | save_and_next(ls); /* skip delimiter */ | |
395 | seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1, | |
396 | luaZ_bufflen(ls->buff) - 2); | |
397 | } | |
398 | ||
399 | ||
400 | static int llex (LexState *ls, SemInfo *seminfo) { | |
401 | luaZ_resetbuffer(ls->buff); | |
402 | for (;;) { | |
403 | switch (ls->current) { | |
404 | case '\n': case '\r': { /* line breaks */ | |
405 | inclinenumber(ls); | |
406 | break; | |
407 | } | |
408 | case ' ': case '\f': case '\t': case '\v': { /* spaces */ | |
409 | next(ls); | |
410 | break; | |
411 | } | |
412 | case '-': { /* '-' or '--' (comment) */ | |
413 | next(ls); | |
414 | if (ls->current != '-') return '-'; | |
415 | /* else is a comment */ | |
416 | next(ls); | |
417 | if (ls->current == '[') { /* long comment? */ | |
418 | int sep = skip_sep(ls); | |
419 | luaZ_resetbuffer(ls->buff); /* `skip_sep' may dirty the buffer */ | |
420 | if (sep >= 0) { | |
421 | read_long_string(ls, NULL, sep); /* skip long comment */ | |
422 | luaZ_resetbuffer(ls->buff); /* previous call may dirty the buff. */ | |
423 | break; | |
424 | } | |
425 | } | |
426 | /* else short comment */ | |
427 | while (!currIsNewline(ls) && ls->current != EOZ) | |
428 | next(ls); /* skip until end of line (or end of file) */ | |
429 | break; | |
430 | } | |
431 | case '[': { /* long string or simply '[' */ | |
432 | int sep = skip_sep(ls); | |
433 | if (sep >= 0) { | |
434 | read_long_string(ls, seminfo, sep); | |
435 | return TK_STRING; | |
436 | } | |
437 | else if (sep == -1) return '['; | |
438 | else lexerror(ls, "invalid long string delimiter", TK_STRING); | |
439 | } | |
440 | case '=': { | |
441 | next(ls); | |
442 | if (ls->current != '=') return '='; | |
443 | else { next(ls); return TK_EQ; } | |
444 | } | |
445 | case '<': { | |
446 | next(ls); | |
447 | if (ls->current != '=') return '<'; | |
448 | else { next(ls); return TK_LE; } | |
449 | } | |
450 | case '>': { | |
451 | next(ls); | |
452 | if (ls->current != '=') return '>'; | |
453 | else { next(ls); return TK_GE; } | |
454 | } | |
455 | case '~': { | |
456 | next(ls); | |
457 | if (ls->current != '=') return '~'; | |
458 | else { next(ls); return TK_NE; } | |
459 | } | |
460 | case ':': { | |
461 | next(ls); | |
462 | if (ls->current != ':') return ':'; | |
463 | else { next(ls); return TK_DBCOLON; } | |
464 | } | |
465 | case '"': case '\'': { /* short literal strings */ | |
466 | read_string(ls, ls->current, seminfo); | |
467 | return TK_STRING; | |
468 | } | |
469 | case '.': { /* '.', '..', '...', or number */ | |
470 | save_and_next(ls); | |
471 | if (check_next(ls, ".")) { | |
472 | if (check_next(ls, ".")) | |
473 | return TK_DOTS; /* '...' */ | |
474 | else return TK_CONCAT; /* '..' */ | |
475 | } | |
476 | else if (!lisdigit(ls->current)) return '.'; | |
477 | /* else go through */ | |
478 | } | |
479 | case '0': case '1': case '2': case '3': case '4': | |
480 | case '5': case '6': case '7': case '8': case '9': { | |
481 | read_numeral(ls, seminfo); | |
482 | return TK_NUMBER; | |
483 | } | |
484 | case EOZ: { | |
485 | return TK_EOS; | |
486 | } | |
487 | default: { | |
488 | if (lislalpha(ls->current)) { /* identifier or reserved word? */ | |
489 | TString *ts; | |
490 | do { | |
491 | save_and_next(ls); | |
492 | } while (lislalnum(ls->current)); | |
493 | ts = luaX_newstring(ls, luaZ_buffer(ls->buff), | |
494 | luaZ_bufflen(ls->buff)); | |
495 | seminfo->ts = ts; | |
496 | if (isreserved(ts)) /* reserved word? */ | |
497 | return ts->tsv.extra - 1 + FIRST_RESERVED; | |
498 | else { | |
499 | return TK_NAME; | |
500 | } | |
501 | } | |
502 | else { /* single-char tokens (+ - / ...) */ | |
503 | int c = ls->current; | |
504 | next(ls); | |
505 | return c; | |
506 | } | |
507 | } | |
508 | } | |
509 | } | |
510 | } | |
511 | ||
512 | ||
513 | void luaX_next (LexState *ls) { | |
514 | ls->lastline = ls->linenumber; | |
515 | if (ls->lookahead.token != TK_EOS) { /* is there a look-ahead token? */ | |
516 | ls->t = ls->lookahead; /* use this one */ | |
517 | ls->lookahead.token = TK_EOS; /* and discharge it */ | |
518 | } | |
519 | else | |
520 | ls->t.token = llex(ls, &ls->t.seminfo); /* read next token */ | |
521 | } | |
522 | ||
523 | ||
524 | int luaX_lookahead (LexState *ls) { | |
525 | lua_assert(ls->lookahead.token == TK_EOS); | |
526 | ls->lookahead.token = llex(ls, &ls->lookahead.seminfo); | |
527 | return ls->lookahead.token; | |
528 | } | |
529 |