isl_stream_next_token: treat "-0" as two tokens '-' and '0'
[isl.git] / isl_stream.c
blob364389324a299ef310c43deadc6d8ea034d467d5
1 /*
2 * Copyright 2008-2009 Katholieke Universiteit Leuven
4 * Use of this software is governed by the GNU LGPLv2.1 license
6 * Written by Sven Verdoolaege, K.U.Leuven, Departement
7 * Computerwetenschappen, Celestijnenlaan 200A, B-3001 Leuven, Belgium
8 */
10 #include <ctype.h>
11 #include <string.h>
12 #include <strings.h>
13 #include <isl/ctx.h>
14 #include <isl_stream_private.h>
15 #include <isl/map.h>
16 #include <isl/aff.h>
18 struct isl_keyword {
19 char *name;
20 enum isl_token_type type;
23 static int same_name(const void *entry, const void *val)
25 const struct isl_keyword *keyword = (const struct isl_keyword *)entry;
27 return !strcmp(keyword->name, val);
30 enum isl_token_type isl_stream_register_keyword(struct isl_stream *s,
31 const char *name)
33 struct isl_hash_table_entry *entry;
34 struct isl_keyword *keyword;
35 uint32_t name_hash;
37 if (!s->keywords) {
38 s->keywords = isl_hash_table_alloc(s->ctx, 10);
39 if (!s->keywords)
40 return ISL_TOKEN_ERROR;
41 s->next_type = ISL_TOKEN_LAST;
44 name_hash = isl_hash_string(isl_hash_init(), name);
46 entry = isl_hash_table_find(s->ctx, s->keywords, name_hash,
47 same_name, name, 1);
48 if (!entry)
49 return ISL_TOKEN_ERROR;
50 if (entry->data) {
51 keyword = entry->data;
52 return keyword->type;
55 keyword = isl_calloc_type(s->ctx, struct isl_keyword);
56 if (!keyword)
57 return ISL_TOKEN_ERROR;
58 keyword->type = s->next_type++;
59 keyword->name = strdup(name);
60 if (!keyword->name) {
61 free(keyword);
62 return ISL_TOKEN_ERROR;
64 entry->data = keyword;
66 return keyword->type;
69 struct isl_token *isl_token_new(isl_ctx *ctx,
70 int line, int col, unsigned on_new_line)
72 struct isl_token *tok = isl_alloc_type(ctx, struct isl_token);
73 if (!tok)
74 return NULL;
75 tok->line = line;
76 tok->col = col;
77 tok->on_new_line = on_new_line;
78 tok->is_keyword = 0;
79 tok->u.s = NULL;
80 return tok;
83 void isl_token_free(struct isl_token *tok)
85 if (!tok)
86 return;
87 if (tok->type == ISL_TOKEN_VALUE)
88 isl_int_clear(tok->u.v);
89 else if (tok->type == ISL_TOKEN_MAP)
90 isl_map_free(tok->u.map);
91 else if (tok->type == ISL_TOKEN_AFF)
92 isl_pw_aff_free(tok->u.pwaff);
93 else
94 free(tok->u.s);
95 free(tok);
98 void isl_stream_error(struct isl_stream *s, struct isl_token *tok, char *msg)
100 int line = tok ? tok->line : s->line;
101 int col = tok ? tok->col : s->col;
102 fprintf(stderr, "syntax error (%d, %d): %s\n", line, col, msg);
103 if (tok) {
104 if (tok->type < 256)
105 fprintf(stderr, "got '%c'\n", tok->type);
106 else if (tok->type == ISL_TOKEN_IDENT)
107 fprintf(stderr, "got ident '%s'\n", tok->u.s);
108 else if (tok->is_keyword)
109 fprintf(stderr, "got keyword '%s'\n", tok->u.s);
110 else if (tok->type == ISL_TOKEN_VALUE) {
111 fprintf(stderr, "got value '");
112 isl_int_print(stderr, tok->u.v, 0);
113 fprintf(stderr, "'\n");
114 } else if (tok->type == ISL_TOKEN_MAP) {
115 isl_printer *p;
116 fprintf(stderr, "got map '");
117 p = isl_printer_to_file(s->ctx, stderr);
118 p = isl_printer_print_map(p, tok->u.map);
119 isl_printer_free(p);
120 fprintf(stderr, "'\n");
121 } else if (tok->type == ISL_TOKEN_AFF) {
122 isl_printer *p;
123 fprintf(stderr, "got affine expression '");
124 p = isl_printer_to_file(s->ctx, stderr);
125 p = isl_printer_print_pw_aff(p, tok->u.pwaff);
126 isl_printer_free(p);
127 fprintf(stderr, "'\n");
128 } else if (tok->u.s)
129 fprintf(stderr, "got token '%s'\n", tok->u.s);
130 else
131 fprintf(stderr, "got token type %d\n", tok->type);
135 static struct isl_stream* isl_stream_new(struct isl_ctx *ctx)
137 int i;
138 struct isl_stream *s = isl_alloc_type(ctx, struct isl_stream);
139 if (!s)
140 return NULL;
141 s->ctx = ctx;
142 isl_ctx_ref(s->ctx);
143 s->file = NULL;
144 s->str = NULL;
145 s->len = 0;
146 s->line = 1;
147 s->col = 0;
148 s->eof = 0;
149 s->c = -1;
150 s->n_un = 0;
151 for (i = 0; i < 5; ++i)
152 s->tokens[i] = NULL;
153 s->n_token = 0;
154 s->keywords = NULL;
155 s->size = 256;
156 s->buffer = isl_alloc_array(ctx, char, s->size);
157 if (!s->buffer)
158 goto error;
159 return s;
160 error:
161 isl_stream_free(s);
162 return NULL;
165 struct isl_stream* isl_stream_new_file(struct isl_ctx *ctx, FILE *file)
167 struct isl_stream *s = isl_stream_new(ctx);
168 if (!s)
169 return NULL;
170 s->file = file;
171 return s;
174 struct isl_stream* isl_stream_new_str(struct isl_ctx *ctx, const char *str)
176 struct isl_stream *s = isl_stream_new(ctx);
177 if (!s)
178 return NULL;
179 s->str = str;
180 return s;
183 static int stream_getc(struct isl_stream *s)
185 int c;
186 if (s->eof)
187 return -1;
188 if (s->n_un)
189 return s->c = s->un[--s->n_un];
190 if (s->file)
191 c = fgetc(s->file);
192 else {
193 c = *s->str++;
194 if (c == '\0')
195 c = -1;
197 if (c == -1)
198 s->eof = 1;
199 if (!s->eof) {
200 if (s->c == '\n') {
201 s->line++;
202 s->col = 0;
203 } else
204 s->col++;
206 s->c = c;
207 return c;
210 static void isl_stream_ungetc(struct isl_stream *s, int c)
212 isl_assert(s->ctx, s->n_un < 5, return);
213 s->un[s->n_un++] = c;
214 s->c = -1;
217 static int isl_stream_getc(struct isl_stream *s)
219 int c;
221 do {
222 c = stream_getc(s);
223 if (c != '\\')
224 return c;
225 c = stream_getc(s);
226 } while (c == '\n');
228 isl_stream_ungetc(s, c);
230 return '\\';
233 static int isl_stream_push_char(struct isl_stream *s, int c)
235 if (s->len >= s->size) {
236 char *buffer;
237 s->size = (3*s->size)/2;
238 buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
239 if (!buffer)
240 return -1;
241 s->buffer = buffer;
243 s->buffer[s->len++] = c;
244 return 0;
247 void isl_stream_push_token(struct isl_stream *s, struct isl_token *tok)
249 isl_assert(s->ctx, s->n_token < 5, return);
250 s->tokens[s->n_token++] = tok;
253 static enum isl_token_type check_keywords(struct isl_stream *s)
255 struct isl_hash_table_entry *entry;
256 struct isl_keyword *keyword;
257 uint32_t name_hash;
259 if (!strcasecmp(s->buffer, "exists"))
260 return ISL_TOKEN_EXISTS;
261 if (!strcasecmp(s->buffer, "and"))
262 return ISL_TOKEN_AND;
263 if (!strcasecmp(s->buffer, "or"))
264 return ISL_TOKEN_OR;
265 if (!strcasecmp(s->buffer, "not"))
266 return ISL_TOKEN_NOT;
267 if (!strcasecmp(s->buffer, "infty"))
268 return ISL_TOKEN_INFTY;
269 if (!strcasecmp(s->buffer, "infinity"))
270 return ISL_TOKEN_INFTY;
271 if (!strcasecmp(s->buffer, "NaN"))
272 return ISL_TOKEN_NAN;
273 if (!strcasecmp(s->buffer, "min"))
274 return ISL_TOKEN_MIN;
275 if (!strcasecmp(s->buffer, "max"))
276 return ISL_TOKEN_MAX;
277 if (!strcasecmp(s->buffer, "rat"))
278 return ISL_TOKEN_RAT;
279 if (!strcasecmp(s->buffer, "true"))
280 return ISL_TOKEN_TRUE;
281 if (!strcasecmp(s->buffer, "false"))
282 return ISL_TOKEN_FALSE;
283 if (!strcasecmp(s->buffer, "ceild"))
284 return ISL_TOKEN_CEILD;
285 if (!strcasecmp(s->buffer, "floord"))
286 return ISL_TOKEN_FLOORD;
287 if (!strcasecmp(s->buffer, "mod"))
288 return ISL_TOKEN_MOD;
290 if (!s->keywords)
291 return ISL_TOKEN_IDENT;
293 name_hash = isl_hash_string(isl_hash_init(), s->buffer);
294 entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
295 s->buffer, 0);
296 if (entry) {
297 keyword = entry->data;
298 return keyword->type;
301 return ISL_TOKEN_IDENT;
304 int isl_stream_skip_line(struct isl_stream *s)
306 int c;
308 while ((c = isl_stream_getc(s)) != -1 && c != '\n')
309 /* nothing */
312 return c == -1 ? -1 : 0;
315 static struct isl_token *next_token(struct isl_stream *s, int same_line)
317 int c;
318 struct isl_token *tok = NULL;
319 int line, col;
320 int old_line = s->line;
322 if (s->n_token) {
323 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
324 return NULL;
325 return s->tokens[--s->n_token];
328 if (same_line && s->c == '\n')
329 return NULL;
331 s->len = 0;
333 /* skip spaces and comment lines */
334 while ((c = isl_stream_getc(s)) != -1) {
335 if (c == '#') {
336 if (isl_stream_skip_line(s) < 0)
337 break;
338 c = '\n';
339 if (same_line)
340 break;
341 } else if (!isspace(c) || (same_line && c == '\n'))
342 break;
345 line = s->line;
346 col = s->col;
348 if (c == -1 || (same_line && c == '\n'))
349 return NULL;
350 if (c == '(' ||
351 c == ')' ||
352 c == '+' ||
353 c == '*' ||
354 c == '%' ||
355 c == '?' ||
356 c == '^' ||
357 c == '=' ||
358 c == '@' ||
359 c == '$' ||
360 c == ',' ||
361 c == '.' ||
362 c == ';' ||
363 c == '[' ||
364 c == ']' ||
365 c == '{' ||
366 c == '}') {
367 tok = isl_token_new(s->ctx, line, col, old_line != line);
368 if (!tok)
369 return NULL;
370 tok->type = (enum isl_token_type)c;
371 return tok;
373 if (c == '-') {
374 int c;
375 if ((c = isl_stream_getc(s)) == '>') {
376 tok = isl_token_new(s->ctx, line, col, old_line != line);
377 if (!tok)
378 return NULL;
379 tok->u.s = strdup("->");
380 tok->type = ISL_TOKEN_TO;
381 return tok;
383 if (c != -1)
384 isl_stream_ungetc(s, c);
385 if (!isdigit(c)) {
386 tok = isl_token_new(s->ctx, line, col, old_line != line);
387 if (!tok)
388 return NULL;
389 tok->type = (enum isl_token_type) '-';
390 return tok;
393 if (c == '-' || isdigit(c)) {
394 int minus = c == '-';
395 tok = isl_token_new(s->ctx, line, col, old_line != line);
396 if (!tok)
397 return NULL;
398 tok->type = ISL_TOKEN_VALUE;
399 isl_int_init(tok->u.v);
400 if (isl_stream_push_char(s, c))
401 goto error;
402 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
403 if (isl_stream_push_char(s, c))
404 goto error;
405 if (c != -1)
406 isl_stream_ungetc(s, c);
407 isl_stream_push_char(s, '\0');
408 isl_int_read(tok->u.v, s->buffer);
409 if (minus && isl_int_is_zero(tok->u.v)) {
410 tok->col++;
411 tok->on_new_line = 0;
412 isl_stream_push_token(s, tok);
413 tok = isl_token_new(s->ctx, line, col, old_line != line);
414 if (!tok)
415 return NULL;
416 tok->type = (enum isl_token_type) '-';
418 return tok;
420 if (isalpha(c) || c == '_') {
421 tok = isl_token_new(s->ctx, line, col, old_line != line);
422 if (!tok)
423 return NULL;
424 isl_stream_push_char(s, c);
425 while ((c = isl_stream_getc(s)) != -1 &&
426 (isalnum(c) || c == '_'))
427 isl_stream_push_char(s, c);
428 if (c != -1)
429 isl_stream_ungetc(s, c);
430 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
431 isl_stream_push_char(s, c);
432 if (c != -1)
433 isl_stream_ungetc(s, c);
434 isl_stream_push_char(s, '\0');
435 tok->type = check_keywords(s);
436 if (tok->type != ISL_TOKEN_IDENT)
437 tok->is_keyword = 1;
438 tok->u.s = strdup(s->buffer);
439 if (!tok->u.s)
440 goto error;
441 return tok;
443 if (c == '"') {
444 tok = isl_token_new(s->ctx, line, col, old_line != line);
445 if (!tok)
446 return NULL;
447 tok->type = ISL_TOKEN_STRING;
448 tok->u.s = NULL;
449 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
450 isl_stream_push_char(s, c);
451 if (c != '"') {
452 isl_stream_error(s, NULL, "unterminated string");
453 goto error;
455 isl_stream_push_char(s, '\0');
456 tok->u.s = strdup(s->buffer);
457 return tok;
459 if (c == ':') {
460 int c;
461 tok = isl_token_new(s->ctx, line, col, old_line != line);
462 if (!tok)
463 return NULL;
464 if ((c = isl_stream_getc(s)) == '=') {
465 tok->u.s = strdup(":=");
466 tok->type = ISL_TOKEN_DEF;
467 return tok;
469 if (c != -1)
470 isl_stream_ungetc(s, c);
471 tok->type = (enum isl_token_type) ':';
472 return tok;
474 if (c == '>') {
475 int c;
476 tok = isl_token_new(s->ctx, line, col, old_line != line);
477 if (!tok)
478 return NULL;
479 if ((c = isl_stream_getc(s)) == '=') {
480 tok->u.s = strdup(">=");
481 tok->type = ISL_TOKEN_GE;
482 return tok;
483 } else if (c == '>') {
484 if ((c = isl_stream_getc(s)) == '=') {
485 tok->u.s = strdup(">>=");
486 tok->type = ISL_TOKEN_LEX_GE;
487 return tok;
489 tok->u.s = strdup(">>");
490 tok->type = ISL_TOKEN_LEX_GT;
491 } else {
492 tok->u.s = strdup(">");
493 tok->type = ISL_TOKEN_GT;
495 if (c != -1)
496 isl_stream_ungetc(s, c);
497 return tok;
499 if (c == '<') {
500 int c;
501 tok = isl_token_new(s->ctx, line, col, old_line != line);
502 if (!tok)
503 return NULL;
504 if ((c = isl_stream_getc(s)) == '=') {
505 tok->u.s = strdup("<=");
506 tok->type = ISL_TOKEN_LE;
507 return tok;
508 } else if (c == '<') {
509 if ((c = isl_stream_getc(s)) == '=') {
510 tok->u.s = strdup("<<=");
511 tok->type = ISL_TOKEN_LEX_LE;
512 return tok;
514 tok->u.s = strdup("<<");
515 tok->type = ISL_TOKEN_LEX_LT;
516 } else {
517 tok->u.s = strdup("<");
518 tok->type = ISL_TOKEN_LT;
520 if (c != -1)
521 isl_stream_ungetc(s, c);
522 return tok;
524 if (c == '&') {
525 tok = isl_token_new(s->ctx, line, col, old_line != line);
526 if (!tok)
527 return NULL;
528 tok->type = ISL_TOKEN_AND;
529 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
530 tok->u.s = strdup("&");
531 isl_stream_ungetc(s, c);
532 } else
533 tok->u.s = strdup("&&");
534 return tok;
536 if (c == '|') {
537 tok = isl_token_new(s->ctx, line, col, old_line != line);
538 if (!tok)
539 return NULL;
540 tok->type = ISL_TOKEN_OR;
541 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
542 tok->u.s = strdup("|");
543 isl_stream_ungetc(s, c);
544 } else
545 tok->u.s = strdup("||");
546 return tok;
548 if (c == '/') {
549 tok = isl_token_new(s->ctx, line, col, old_line != line);
550 if (!tok)
551 return NULL;
552 if ((c = isl_stream_getc(s)) != '\\' && c != -1) {
553 tok->type = (enum isl_token_type) '/';
554 isl_stream_ungetc(s, c);
555 } else {
556 tok->u.s = strdup("/\\");
557 tok->type = ISL_TOKEN_AND;
559 return tok;
561 if (c == '\\') {
562 tok = isl_token_new(s->ctx, line, col, old_line != line);
563 if (!tok)
564 return NULL;
565 if ((c = isl_stream_getc(s)) != '/' && c != -1) {
566 tok->type = (enum isl_token_type) '\\';
567 isl_stream_ungetc(s, c);
568 } else {
569 tok->u.s = strdup("\\/");
570 tok->type = ISL_TOKEN_OR;
572 return tok;
574 if (c == '!') {
575 tok = isl_token_new(s->ctx, line, col, old_line != line);
576 if (!tok)
577 return NULL;
578 if ((c = isl_stream_getc(s)) == '=') {
579 tok->u.s = strdup("!=");
580 tok->type = ISL_TOKEN_NE;
581 return tok;
582 } else {
583 tok->type = ISL_TOKEN_NOT;
584 tok->u.s = strdup("!");
586 if (c != -1)
587 isl_stream_ungetc(s, c);
588 return tok;
591 tok = isl_token_new(s->ctx, line, col, old_line != line);
592 if (!tok)
593 return NULL;
594 tok->type = ISL_TOKEN_UNKNOWN;
595 return tok;
596 error:
597 isl_token_free(tok);
598 return NULL;
601 struct isl_token *isl_stream_next_token(struct isl_stream *s)
603 return next_token(s, 0);
606 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
608 return next_token(s, 1);
611 int isl_stream_eat_if_available(struct isl_stream *s, int type)
613 struct isl_token *tok;
615 tok = isl_stream_next_token(s);
616 if (!tok)
617 return 0;
618 if (tok->type == type) {
619 isl_token_free(tok);
620 return 1;
622 isl_stream_push_token(s, tok);
623 return 0;
626 int isl_stream_next_token_is(struct isl_stream *s, int type)
628 struct isl_token *tok;
629 int r;
631 tok = isl_stream_next_token(s);
632 if (!tok)
633 return 0;
634 r = tok->type == type;
635 isl_stream_push_token(s, tok);
636 return r;
639 char *isl_stream_read_ident_if_available(struct isl_stream *s)
641 struct isl_token *tok;
643 tok = isl_stream_next_token(s);
644 if (!tok)
645 return NULL;
646 if (tok->type == ISL_TOKEN_IDENT) {
647 char *ident = strdup(tok->u.s);
648 isl_token_free(tok);
649 return ident;
651 isl_stream_push_token(s, tok);
652 return NULL;
655 int isl_stream_eat(struct isl_stream *s, int type)
657 struct isl_token *tok;
659 tok = isl_stream_next_token(s);
660 if (!tok)
661 return -1;
662 if (tok->type == type) {
663 isl_token_free(tok);
664 return 0;
666 isl_stream_error(s, tok, "expecting other token");
667 isl_stream_push_token(s, tok);
668 return -1;
671 int isl_stream_is_empty(struct isl_stream *s)
673 struct isl_token *tok;
675 tok = isl_stream_next_token(s);
677 if (!tok)
678 return 1;
680 isl_stream_push_token(s, tok);
681 return 0;
684 static int free_keyword(void **p, void *user)
686 struct isl_keyword *keyword = *p;
688 free(keyword->name);
689 free(keyword);
691 return 0;
694 void isl_stream_flush_tokens(struct isl_stream *s)
696 int i;
698 if (!s)
699 return;
700 for (i = 0; i < s->n_token; ++i)
701 isl_token_free(s->tokens[i]);
702 s->n_token = 0;
705 void isl_stream_free(struct isl_stream *s)
707 if (!s)
708 return;
709 free(s->buffer);
710 if (s->n_token != 0) {
711 struct isl_token *tok = isl_stream_next_token(s);
712 isl_stream_error(s, tok, "unexpected token");
713 isl_token_free(tok);
715 if (s->keywords) {
716 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
717 isl_hash_table_free(s->ctx, s->keywords);
719 isl_ctx_deref(s->ctx);
720 free(s);