diff options
author | Roberto Ierusalimschy <roberto@inf.puc-rio.br> | 1997-09-16 16:25:59 -0300 |
---|---|---|
committer | Roberto Ierusalimschy <roberto@inf.puc-rio.br> | 1997-09-16 16:25:59 -0300 |
commit | 60cc473bcfce079d1525fcffcfdfbeb66e35afa2 (patch) | |
tree | 368fec478cec8637af6346e69ec66722ab38a9cf /lex.c | |
parent | 43a2ee6ea1b7825c1892de614cb38a3fe487a19f (diff) | |
download | lua-60cc473bcfce079d1525fcffcfdfbeb66e35afa2.tar.gz lua-60cc473bcfce079d1525fcffcfdfbeb66e35afa2.tar.bz2 lua-60cc473bcfce079d1525fcffcfdfbeb66e35afa2.zip |
Lexical Analizer
Diffstat (limited to 'lex.c')
-rw-r--r-- | lex.c | 466 |
1 files changed, 0 insertions, 466 deletions
@@ -1,466 +0,0 @@ | |||
1 | char *rcs_lex = "$Id: lex.c,v 3.8 1997/07/30 22:00:50 roberto Exp roberto $"; | ||
2 | |||
3 | |||
4 | #include <ctype.h> | ||
5 | #include <string.h> | ||
6 | |||
7 | #include "auxlib.h" | ||
8 | #include "luamem.h" | ||
9 | #include "tree.h" | ||
10 | #include "table.h" | ||
11 | #include "lex.h" | ||
12 | #include "inout.h" | ||
13 | #include "luadebug.h" | ||
14 | #include "parser.h" | ||
15 | |||
16 | #define MINBUFF 250 | ||
17 | |||
18 | static int current; /* look ahead character */ | ||
19 | static ZIO *lex_z; | ||
20 | |||
21 | |||
22 | #define next() (current = zgetc(lex_z)) | ||
23 | #define save(x) (yytext[tokensize++] = (x)) | ||
24 | #define save_and_next() (save(current), next()) | ||
25 | |||
26 | |||
27 | #define MAX_IFS 5 | ||
28 | |||
29 | /* "ifstate" keeps the state of each nested $if the lexical is dealing with. */ | ||
30 | |||
31 | static struct { | ||
32 | int elsepart; /* true if its in the $else part */ | ||
33 | int condition; /* true if $if condition is true */ | ||
34 | int skip; /* true if part must be skiped */ | ||
35 | } ifstate[MAX_IFS]; | ||
36 | |||
37 | static int iflevel; /* level of nested $if's */ | ||
38 | |||
39 | |||
40 | |||
41 | static void firstline (void) | ||
42 | { | ||
43 | int c = zgetc(lex_z); | ||
44 | if (c == '#') | ||
45 | while((c=zgetc(lex_z)) != '\n' && c != EOZ) /* skip first line */; | ||
46 | zungetc(lex_z); | ||
47 | } | ||
48 | |||
49 | |||
50 | void lua_setinput (ZIO *z) | ||
51 | { | ||
52 | current = '\n'; | ||
53 | lua_linenumber = 0; | ||
54 | iflevel = 0; | ||
55 | ifstate[0].skip = 0; | ||
56 | ifstate[0].elsepart = 1; /* to avoid a free $else */ | ||
57 | lex_z = z; | ||
58 | firstline(); | ||
59 | } | ||
60 | |||
61 | |||
62 | static void luaI_auxsyntaxerror (char *s) | ||
63 | { | ||
64 | luaL_verror("%s;\n> at line %d in file %s", | ||
65 | s, lua_linenumber, lua_parsedfile->str); | ||
66 | } | ||
67 | |||
68 | static void luaI_auxsynterrbf (char *s, char *token) | ||
69 | { | ||
70 | if (token[0] == 0) | ||
71 | token = "<eof>"; | ||
72 | luaL_verror("%s;\n> last token read: \"%s\" at line %d in file %s", | ||
73 | s, token, lua_linenumber, lua_parsedfile->str); | ||
74 | } | ||
75 | |||
76 | void luaI_syntaxerror (char *s) | ||
77 | { | ||
78 | luaI_auxsynterrbf(s, luaI_buffer(1)); | ||
79 | } | ||
80 | |||
81 | |||
82 | static struct | ||
83 | { | ||
84 | char *name; | ||
85 | int token; | ||
86 | } reserved [] = { | ||
87 | {"and", AND}, | ||
88 | {"do", DO}, | ||
89 | {"else", ELSE}, | ||
90 | {"elseif", ELSEIF}, | ||
91 | {"end", END}, | ||
92 | {"function", FUNCTION}, | ||
93 | {"if", IF}, | ||
94 | {"local", LOCAL}, | ||
95 | {"nil", NIL}, | ||
96 | {"not", NOT}, | ||
97 | {"or", OR}, | ||
98 | {"repeat", REPEAT}, | ||
99 | {"return", RETURN}, | ||
100 | {"then", THEN}, | ||
101 | {"until", UNTIL}, | ||
102 | {"while", WHILE} }; | ||
103 | |||
104 | |||
105 | #define RESERVEDSIZE (sizeof(reserved)/sizeof(reserved[0])) | ||
106 | |||
107 | |||
108 | void luaI_addReserved (void) | ||
109 | { | ||
110 | int i; | ||
111 | for (i=0; i<RESERVEDSIZE; i++) | ||
112 | { | ||
113 | TaggedString *ts = luaI_createstring(reserved[i].name); | ||
114 | ts->marked = reserved[i].token; /* reserved word (always > 255) */ | ||
115 | } | ||
116 | } | ||
117 | |||
118 | |||
119 | /* | ||
120 | ** Pragma handling | ||
121 | */ | ||
122 | |||
123 | #define PRAGMASIZE 20 | ||
124 | |||
125 | static void skipspace (void) | ||
126 | { | ||
127 | while (current == ' ' || current == '\t') next(); | ||
128 | } | ||
129 | |||
130 | |||
131 | static int checkcond (char *buff) | ||
132 | { | ||
133 | static char *opts[] = {"nil", "1"}; | ||
134 | int i = luaI_findstring(buff, opts); | ||
135 | if (i >= 0) return i; | ||
136 | else if (isalpha((unsigned char)buff[0]) || buff[0] == '_') | ||
137 | return luaI_globaldefined(buff); | ||
138 | else { | ||
139 | luaI_auxsynterrbf("invalid $if condition", buff); | ||
140 | return 0; /* to avoid warnings */ | ||
141 | } | ||
142 | } | ||
143 | |||
144 | |||
145 | static void readname (char *buff) | ||
146 | { | ||
147 | int i = 0; | ||
148 | skipspace(); | ||
149 | while (isalnum((unsigned char)current) || current == '_') { | ||
150 | if (i >= PRAGMASIZE) { | ||
151 | buff[PRAGMASIZE] = 0; | ||
152 | luaI_auxsynterrbf("pragma too long", buff); | ||
153 | } | ||
154 | buff[i++] = current; | ||
155 | next(); | ||
156 | } | ||
157 | buff[i] = 0; | ||
158 | } | ||
159 | |||
160 | |||
161 | static void inclinenumber (void); | ||
162 | |||
163 | |||
164 | static void ifskip (void) | ||
165 | { | ||
166 | while (ifstate[iflevel].skip) { | ||
167 | if (current == '\n') | ||
168 | inclinenumber(); | ||
169 | else if (current == EOZ) | ||
170 | luaI_auxsyntaxerror("input ends inside a $if"); | ||
171 | else next(); | ||
172 | } | ||
173 | } | ||
174 | |||
175 | |||
176 | static void inclinenumber (void) | ||
177 | { | ||
178 | static char *pragmas [] = | ||
179 | {"debug", "nodebug", "endinput", "end", "ifnot", "if", "else", NULL}; | ||
180 | next(); /* skip '\n' */ | ||
181 | ++lua_linenumber; | ||
182 | if (current == '$') { /* is a pragma? */ | ||
183 | char buff[PRAGMASIZE+1]; | ||
184 | int ifnot = 0; | ||
185 | int skip = ifstate[iflevel].skip; | ||
186 | next(); /* skip $ */ | ||
187 | readname(buff); | ||
188 | switch (luaI_findstring(buff, pragmas)) { | ||
189 | case 0: /* debug */ | ||
190 | if (!skip) lua_debug = 1; | ||
191 | break; | ||
192 | case 1: /* nodebug */ | ||
193 | if (!skip) lua_debug = 0; | ||
194 | break; | ||
195 | case 2: /* endinput */ | ||
196 | if (!skip) { | ||
197 | current = EOZ; | ||
198 | iflevel = 0; /* to allow $endinput inside a $if */ | ||
199 | } | ||
200 | break; | ||
201 | case 3: /* end */ | ||
202 | if (iflevel-- == 0) | ||
203 | luaI_auxsyntaxerror("unmatched $endif"); | ||
204 | break; | ||
205 | case 4: /* ifnot */ | ||
206 | ifnot = 1; | ||
207 | /* go through */ | ||
208 | case 5: /* if */ | ||
209 | if (iflevel == MAX_IFS-1) | ||
210 | luaI_auxsyntaxerror("too many nested `$ifs'"); | ||
211 | readname(buff); | ||
212 | iflevel++; | ||
213 | ifstate[iflevel].elsepart = 0; | ||
214 | ifstate[iflevel].condition = checkcond(buff) ? !ifnot : ifnot; | ||
215 | ifstate[iflevel].skip = skip || !ifstate[iflevel].condition; | ||
216 | break; | ||
217 | case 6: /* else */ | ||
218 | if (ifstate[iflevel].elsepart) | ||
219 | luaI_auxsyntaxerror("unmatched $else"); | ||
220 | ifstate[iflevel].elsepart = 1; | ||
221 | ifstate[iflevel].skip = | ||
222 | ifstate[iflevel-1].skip || ifstate[iflevel].condition; | ||
223 | break; | ||
224 | default: | ||
225 | luaI_auxsynterrbf("invalid pragma", buff); | ||
226 | } | ||
227 | skipspace(); | ||
228 | if (current == '\n') /* pragma must end with a '\n' ... */ | ||
229 | inclinenumber(); | ||
230 | else if (current != EOZ) /* or eof */ | ||
231 | luaI_auxsyntaxerror("invalid pragma format"); | ||
232 | ifskip(); | ||
233 | } | ||
234 | } | ||
235 | |||
236 | static int read_long_string (char *yytext, int buffsize) | ||
237 | { | ||
238 | int cont = 0; | ||
239 | int tokensize = 2; /* '[[' already stored */ | ||
240 | while (1) | ||
241 | { | ||
242 | if (buffsize-tokensize <= 2) /* may read more than 1 char in one cicle */ | ||
243 | yytext = luaI_buffer(buffsize *= 2); | ||
244 | switch (current) | ||
245 | { | ||
246 | case EOZ: | ||
247 | save(0); | ||
248 | return WRONGTOKEN; | ||
249 | case '[': | ||
250 | save_and_next(); | ||
251 | if (current == '[') | ||
252 | { | ||
253 | cont++; | ||
254 | save_and_next(); | ||
255 | } | ||
256 | continue; | ||
257 | case ']': | ||
258 | save_and_next(); | ||
259 | if (current == ']') | ||
260 | { | ||
261 | if (cont == 0) goto endloop; | ||
262 | cont--; | ||
263 | save_and_next(); | ||
264 | } | ||
265 | continue; | ||
266 | case '\n': | ||
267 | save('\n'); | ||
268 | inclinenumber(); | ||
269 | continue; | ||
270 | default: | ||
271 | save_and_next(); | ||
272 | } | ||
273 | } endloop: | ||
274 | save_and_next(); /* pass the second ']' */ | ||
275 | yytext[tokensize-2] = 0; /* erases ']]' */ | ||
276 | luaY_lval.pTStr = luaI_createtempstring(yytext+2); | ||
277 | yytext[tokensize-2] = ']'; /* restores ']]' */ | ||
278 | save(0); | ||
279 | return STRING; | ||
280 | } | ||
281 | |||
282 | int luaY_lex (void) | ||
283 | { | ||
284 | static int linelasttoken = 0; | ||
285 | double a; | ||
286 | int buffsize = MINBUFF; | ||
287 | char *yytext = luaI_buffer(buffsize); | ||
288 | yytext[1] = yytext[2] = yytext[3] = 0; | ||
289 | if (lua_debug) | ||
290 | luaI_codedebugline(linelasttoken); | ||
291 | linelasttoken = lua_linenumber; | ||
292 | while (1) { | ||
293 | int tokensize = 0; | ||
294 | switch (current) { | ||
295 | case '\n': | ||
296 | inclinenumber(); | ||
297 | linelasttoken = lua_linenumber; | ||
298 | continue; | ||
299 | |||
300 | case ' ': case '\t': case '\r': /* CR: to avoid problems with DOS */ | ||
301 | next(); | ||
302 | continue; | ||
303 | |||
304 | case '-': | ||
305 | save_and_next(); | ||
306 | if (current != '-') return '-'; | ||
307 | do { next(); } while (current != '\n' && current != EOZ); | ||
308 | continue; | ||
309 | |||
310 | case '[': | ||
311 | save_and_next(); | ||
312 | if (current != '[') return '['; | ||
313 | else | ||
314 | { | ||
315 | save_and_next(); /* pass the second '[' */ | ||
316 | return read_long_string(yytext, buffsize); | ||
317 | } | ||
318 | |||
319 | case '=': | ||
320 | save_and_next(); | ||
321 | if (current != '=') return '='; | ||
322 | else { save_and_next(); return EQ; } | ||
323 | |||
324 | case '<': | ||
325 | save_and_next(); | ||
326 | if (current != '=') return '<'; | ||
327 | else { save_and_next(); return LE; } | ||
328 | |||
329 | case '>': | ||
330 | save_and_next(); | ||
331 | if (current != '=') return '>'; | ||
332 | else { save_and_next(); return GE; } | ||
333 | |||
334 | case '~': | ||
335 | save_and_next(); | ||
336 | if (current != '=') return '~'; | ||
337 | else { save_and_next(); return NE; } | ||
338 | |||
339 | case '"': | ||
340 | case '\'': | ||
341 | { | ||
342 | int del = current; | ||
343 | save_and_next(); | ||
344 | while (current != del) | ||
345 | { | ||
346 | if (buffsize-tokensize <= 2) /* may read more than 1 char in one cicle */ | ||
347 | yytext = luaI_buffer(buffsize *= 2); | ||
348 | switch (current) | ||
349 | { | ||
350 | case EOZ: | ||
351 | case '\n': | ||
352 | save(0); | ||
353 | return WRONGTOKEN; | ||
354 | case '\\': | ||
355 | next(); /* do not save the '\' */ | ||
356 | switch (current) | ||
357 | { | ||
358 | case 'n': save('\n'); next(); break; | ||
359 | case 't': save('\t'); next(); break; | ||
360 | case 'r': save('\r'); next(); break; | ||
361 | case '\n': save('\n'); inclinenumber(); break; | ||
362 | default : save_and_next(); break; | ||
363 | } | ||
364 | break; | ||
365 | default: | ||
366 | save_and_next(); | ||
367 | } | ||
368 | } | ||
369 | next(); /* skip delimiter */ | ||
370 | save(0); | ||
371 | luaY_lval.pTStr = luaI_createtempstring(yytext+1); | ||
372 | tokensize--; | ||
373 | save(del); save(0); /* restore delimiter */ | ||
374 | return STRING; | ||
375 | } | ||
376 | |||
377 | case '.': | ||
378 | save_and_next(); | ||
379 | if (current == '.') | ||
380 | { | ||
381 | save_and_next(); | ||
382 | if (current == '.') | ||
383 | { | ||
384 | save_and_next(); | ||
385 | return DOTS; /* ... */ | ||
386 | } | ||
387 | else return CONC; /* .. */ | ||
388 | } | ||
389 | else if (!isdigit((unsigned char)current)) return '.'; | ||
390 | /* current is a digit: goes through to number */ | ||
391 | a=0.0; | ||
392 | goto fraction; | ||
393 | |||
394 | case '0': case '1': case '2': case '3': case '4': | ||
395 | case '5': case '6': case '7': case '8': case '9': | ||
396 | a=0.0; | ||
397 | do { | ||
398 | a=10.0*a+(current-'0'); | ||
399 | save_and_next(); | ||
400 | } while (isdigit((unsigned char)current)); | ||
401 | if (current == '.') { | ||
402 | save_and_next(); | ||
403 | if (current == '.') | ||
404 | luaI_syntaxerror( | ||
405 | "ambiguous syntax (decimal point x string concatenation)"); | ||
406 | } | ||
407 | fraction: | ||
408 | { double da=0.1; | ||
409 | while (isdigit((unsigned char)current)) | ||
410 | { | ||
411 | a+=(current-'0')*da; | ||
412 | da/=10.0; | ||
413 | save_and_next(); | ||
414 | } | ||
415 | if (current == 'e' || current == 'E') | ||
416 | { | ||
417 | int e=0; | ||
418 | int neg; | ||
419 | double ea; | ||
420 | save_and_next(); | ||
421 | neg=(current=='-'); | ||
422 | if (current == '+' || current == '-') save_and_next(); | ||
423 | if (!isdigit((unsigned char)current)) { | ||
424 | save(0); return WRONGTOKEN; } | ||
425 | do { | ||
426 | e=10.0*e+(current-'0'); | ||
427 | save_and_next(); | ||
428 | } while (isdigit((unsigned char)current)); | ||
429 | for (ea=neg?0.1:10.0; e>0; e>>=1) | ||
430 | { | ||
431 | if (e & 1) a*=ea; | ||
432 | ea*=ea; | ||
433 | } | ||
434 | } | ||
435 | luaY_lval.vReal = a; | ||
436 | save(0); | ||
437 | return NUMBER; | ||
438 | } | ||
439 | |||
440 | case EOZ: | ||
441 | save(0); | ||
442 | if (iflevel > 0) | ||
443 | luaI_syntaxerror("missing $endif"); | ||
444 | return 0; | ||
445 | |||
446 | default: | ||
447 | if (current != '_' && !isalpha((unsigned char)current)) { | ||
448 | save_and_next(); | ||
449 | return yytext[0]; | ||
450 | } | ||
451 | else { /* identifier or reserved word */ | ||
452 | TaggedString *ts; | ||
453 | do { | ||
454 | save_and_next(); | ||
455 | } while (isalnum((unsigned char)current) || current == '_'); | ||
456 | save(0); | ||
457 | ts = luaI_createtempstring(yytext); | ||
458 | if (ts->marked > 255) | ||
459 | return ts->marked; /* reserved word */ | ||
460 | luaY_lval.pTStr = ts; | ||
461 | return NAME; | ||
462 | } | ||
463 | } | ||
464 | } | ||
465 | } | ||
466 | |||