1 /*
2  * Copyright 2008-2009 Katholieke Universiteit Leuven
3  *
4  * Use of this software is governed by the MIT license
5  *
6  * Written by Sven Verdoolaege, K.U.Leuven, Departement
7  * Computerwetenschappen, Celestijnenlaan 200A, B-3001 Leuven, Belgium
8  */
9 
10 #include <ctype.h>
11 #include <string.h>
12 #include <isl_ctx_private.h>
13 #include <isl_stream_private.h>
14 #include <isl/map.h>
15 #include <isl/aff.h>
16 #include <isl_val_private.h>
17 #include <isl_options_private.h>
18 
19 struct isl_keyword {
20 	char			*name;
21 	enum isl_token_type	type;
22 };
23 
same_name(const void * entry,const void * val)24 static isl_bool same_name(const void *entry, const void *val)
25 {
26 	const struct isl_keyword *keyword = (const struct isl_keyword *)entry;
27 
28 	return isl_bool_ok(!strcmp(keyword->name, val));
29 }
30 
isl_stream_register_keyword(__isl_keep isl_stream * s,const char * name)31 enum isl_token_type isl_stream_register_keyword(__isl_keep isl_stream *s,
32 	const char *name)
33 {
34 	struct isl_hash_table_entry *entry;
35 	struct isl_keyword *keyword;
36 	uint32_t name_hash;
37 
38 	if (!s->keywords) {
39 		s->keywords = isl_hash_table_alloc(s->ctx, 10);
40 		if (!s->keywords)
41 			return ISL_TOKEN_ERROR;
42 		s->next_type = ISL_TOKEN_LAST;
43 	}
44 
45 	name_hash = isl_hash_string(isl_hash_init(), name);
46 
47 	entry = isl_hash_table_find(s->ctx, s->keywords, name_hash,
48 					same_name, name, 1);
49 	if (!entry)
50 		return ISL_TOKEN_ERROR;
51 	if (entry->data) {
52 		keyword = entry->data;
53 		return keyword->type;
54 	}
55 
56 	keyword = isl_calloc_type(s->ctx, struct isl_keyword);
57 	if (!keyword)
58 		return ISL_TOKEN_ERROR;
59 	keyword->type = s->next_type++;
60 	keyword->name = strdup(name);
61 	if (!keyword->name) {
62 		free(keyword);
63 		return ISL_TOKEN_ERROR;
64 	}
65 	entry->data = keyword;
66 
67 	return keyword->type;
68 }
69 
isl_token_new(isl_ctx * ctx,int line,int col,unsigned on_new_line)70 struct isl_token *isl_token_new(isl_ctx *ctx,
71 	int line, int col, unsigned on_new_line)
72 {
73 	struct isl_token *tok = isl_alloc_type(ctx, struct isl_token);
74 	if (!tok)
75 		return NULL;
76 	tok->line = line;
77 	tok->col = col;
78 	tok->on_new_line = on_new_line;
79 	tok->is_keyword = 0;
80 	tok->u.s = NULL;
81 	return tok;
82 }
83 
84 /* Return the type of "tok".
85  */
isl_token_get_type(struct isl_token * tok)86 int isl_token_get_type(struct isl_token *tok)
87 {
88 	return tok ? tok->type : ISL_TOKEN_ERROR;
89 }
90 
91 /* Given a token of type ISL_TOKEN_VALUE, return the value it represents.
92  */
isl_token_get_val(isl_ctx * ctx,struct isl_token * tok)93 __isl_give isl_val *isl_token_get_val(isl_ctx *ctx, struct isl_token *tok)
94 {
95 	if (!tok)
96 		return NULL;
97 	if (tok->type != ISL_TOKEN_VALUE)
98 		isl_die(ctx, isl_error_invalid, "not a value token",
99 			return NULL);
100 
101 	return isl_val_int_from_isl_int(ctx, tok->u.v);
102 }
103 
104 /* Given a token with a string representation, return a copy of this string.
105  */
isl_token_get_str(isl_ctx * ctx,struct isl_token * tok)106 __isl_give char *isl_token_get_str(isl_ctx *ctx, struct isl_token *tok)
107 {
108 	if (!tok)
109 		return NULL;
110 	if (!tok->u.s)
111 		isl_die(ctx, isl_error_invalid,
112 			"token does not have a string representation",
113 			return NULL);
114 
115 	return strdup(tok->u.s);
116 }
117 
isl_token_free(struct isl_token * tok)118 void isl_token_free(struct isl_token *tok)
119 {
120 	if (!tok)
121 		return;
122 	if (tok->type == ISL_TOKEN_VALUE)
123 		isl_int_clear(tok->u.v);
124 	else if (tok->type == ISL_TOKEN_MAP)
125 		isl_map_free(tok->u.map);
126 	else if (tok->type == ISL_TOKEN_AFF)
127 		isl_pw_aff_free(tok->u.pwaff);
128 	else
129 		free(tok->u.s);
130 	free(tok);
131 }
132 
isl_stream_error(__isl_keep isl_stream * s,struct isl_token * tok,char * msg)133 void isl_stream_error(__isl_keep isl_stream *s, struct isl_token *tok,
134 	char *msg)
135 {
136 	int line = tok ? tok->line : s->line;
137 	int col = tok ? tok->col : s->col;
138 
139 	isl_ctx_set_full_error(s->ctx, isl_error_invalid, "syntax error",
140 				__FILE__, __LINE__);
141 
142 	if (s->ctx->opt->on_error == ISL_ON_ERROR_CONTINUE)
143 		return;
144 	fprintf(stderr, "syntax error (%d, %d): %s\n", line, col, msg);
145 	if (tok) {
146 		if (tok->type < 256)
147 			fprintf(stderr, "got '%c'\n", tok->type);
148 		else if (tok->type == ISL_TOKEN_IDENT)
149 			fprintf(stderr, "got ident '%s'\n", tok->u.s);
150 		else if (tok->is_keyword)
151 			fprintf(stderr, "got keyword '%s'\n", tok->u.s);
152 		else if (tok->type == ISL_TOKEN_VALUE) {
153 			fprintf(stderr, "got value '");
154 			isl_int_print(stderr, tok->u.v, 0);
155 			fprintf(stderr, "'\n");
156 		} else if (tok->type == ISL_TOKEN_MAP) {
157 			isl_printer *p;
158 			fprintf(stderr, "got map '");
159 			p = isl_printer_to_file(s->ctx, stderr);
160 			p = isl_printer_print_map(p, tok->u.map);
161 			isl_printer_free(p);
162 			fprintf(stderr, "'\n");
163 		} else if (tok->type == ISL_TOKEN_AFF) {
164 			isl_printer *p;
165 			fprintf(stderr, "got affine expression '");
166 			p = isl_printer_to_file(s->ctx, stderr);
167 			p = isl_printer_print_pw_aff(p, tok->u.pwaff);
168 			isl_printer_free(p);
169 			fprintf(stderr, "'\n");
170 		} else if (tok->u.s)
171 			fprintf(stderr, "got token '%s'\n", tok->u.s);
172 		else
173 			fprintf(stderr, "got token type %d\n", tok->type);
174 	}
175 	if (s->ctx->opt->on_error == ISL_ON_ERROR_ABORT)
176 		abort();
177 }
178 
isl_stream_new(struct isl_ctx * ctx)179 static __isl_give isl_stream* isl_stream_new(struct isl_ctx *ctx)
180 {
181 	int i;
182 	isl_stream *s = isl_calloc_type(ctx, struct isl_stream);
183 	if (!s)
184 		return NULL;
185 	s->ctx = ctx;
186 	isl_ctx_ref(s->ctx);
187 	s->file = NULL;
188 	s->str = NULL;
189 	s->len = 0;
190 	s->line = 1;
191 	s->col = 1;
192 	s->eof = 0;
193 	s->last_line = 0;
194 	s->c = -1;
195 	s->n_un = 0;
196 	for (i = 0; i < 5; ++i)
197 		s->tokens[i] = NULL;
198 	s->n_token = 0;
199 	s->keywords = NULL;
200 	s->size = 256;
201 	s->buffer = isl_alloc_array(ctx, char, s->size);
202 	if (!s->buffer)
203 		goto error;
204 	return s;
205 error:
206 	isl_stream_free(s);
207 	return NULL;
208 }
209 
isl_stream_new_file(struct isl_ctx * ctx,FILE * file)210 __isl_give isl_stream* isl_stream_new_file(struct isl_ctx *ctx, FILE *file)
211 {
212 	isl_stream *s = isl_stream_new(ctx);
213 	if (!s)
214 		return NULL;
215 	s->file = file;
216 	return s;
217 }
218 
isl_stream_new_str(struct isl_ctx * ctx,const char * str)219 __isl_give isl_stream* isl_stream_new_str(struct isl_ctx *ctx, const char *str)
220 {
221 	isl_stream *s;
222 	if (!str)
223 		return NULL;
224 	s = isl_stream_new(ctx);
225 	if (!s)
226 		return NULL;
227 	s->str = str;
228 	return s;
229 }
230 
231 /* Read a character from the stream and advance s->line and s->col
232  * to point to the next character.
233  */
stream_getc(__isl_keep isl_stream * s)234 static int stream_getc(__isl_keep isl_stream *s)
235 {
236 	int c;
237 	if (s->eof)
238 		return -1;
239 	if (s->n_un)
240 		return s->c = s->un[--s->n_un];
241 	if (s->file)
242 		c = fgetc(s->file);
243 	else {
244 		c = *s->str++;
245 		if (c == '\0')
246 			c = -1;
247 	}
248 	if (c == -1)
249 		s->eof = 1;
250 	else if (c == '\n') {
251 		s->line++;
252 		s->col = 1;
253 	} else
254 		s->col++;
255 	s->c = c;
256 	return c;
257 }
258 
isl_stream_ungetc(__isl_keep isl_stream * s,int c)259 static void isl_stream_ungetc(__isl_keep isl_stream *s, int c)
260 {
261 	isl_assert(s->ctx, s->n_un < 5, return);
262 	s->un[s->n_un++] = c;
263 	s->c = -1;
264 }
265 
266 /* Read a character from the stream, skipping pairs of '\\' and '\n'.
267  * Set s->start_line and s->start_col to the line and column
268  * of the returned character.
269  */
isl_stream_getc(__isl_keep isl_stream * s)270 static int isl_stream_getc(__isl_keep isl_stream *s)
271 {
272 	int c;
273 
274 	do {
275 		s->start_line = s->line;
276 		s->start_col = s->col;
277 		c = stream_getc(s);
278 		if (c != '\\')
279 			return c;
280 		c = stream_getc(s);
281 	} while (c == '\n');
282 
283 	isl_stream_ungetc(s, c);
284 
285 	return '\\';
286 }
287 
isl_stream_push_char(__isl_keep isl_stream * s,int c)288 static int isl_stream_push_char(__isl_keep isl_stream *s, int c)
289 {
290 	if (s->len >= s->size) {
291 		char *buffer;
292 		s->size = (3*s->size)/2;
293 		buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
294 		if (!buffer)
295 			return -1;
296 		s->buffer = buffer;
297 	}
298 	s->buffer[s->len++] = c;
299 	return 0;
300 }
301 
isl_stream_push_token(__isl_keep isl_stream * s,struct isl_token * tok)302 void isl_stream_push_token(__isl_keep isl_stream *s, struct isl_token *tok)
303 {
304 	isl_assert(s->ctx, s->n_token < 5, return);
305 	s->tokens[s->n_token++] = tok;
306 }
307 
check_keywords(__isl_keep isl_stream * s)308 static enum isl_token_type check_keywords(__isl_keep isl_stream *s)
309 {
310 	struct isl_hash_table_entry *entry;
311 	struct isl_keyword *keyword;
312 	uint32_t name_hash;
313 
314 	if (!strcasecmp(s->buffer, "exists"))
315 		return ISL_TOKEN_EXISTS;
316 	if (!strcasecmp(s->buffer, "and"))
317 		return ISL_TOKEN_AND;
318 	if (!strcasecmp(s->buffer, "or"))
319 		return ISL_TOKEN_OR;
320 	if (!strcasecmp(s->buffer, "implies"))
321 		return ISL_TOKEN_IMPLIES;
322 	if (!strcasecmp(s->buffer, "not"))
323 		return ISL_TOKEN_NOT;
324 	if (!strcasecmp(s->buffer, "infty"))
325 		return ISL_TOKEN_INFTY;
326 	if (!strcasecmp(s->buffer, "infinity"))
327 		return ISL_TOKEN_INFTY;
328 	if (!strcasecmp(s->buffer, "NaN"))
329 		return ISL_TOKEN_NAN;
330 	if (!strcasecmp(s->buffer, "min"))
331 		return ISL_TOKEN_MIN;
332 	if (!strcasecmp(s->buffer, "max"))
333 		return ISL_TOKEN_MAX;
334 	if (!strcasecmp(s->buffer, "rat"))
335 		return ISL_TOKEN_RAT;
336 	if (!strcasecmp(s->buffer, "true"))
337 		return ISL_TOKEN_TRUE;
338 	if (!strcasecmp(s->buffer, "false"))
339 		return ISL_TOKEN_FALSE;
340 	if (!strcasecmp(s->buffer, "ceild"))
341 		return ISL_TOKEN_CEILD;
342 	if (!strcasecmp(s->buffer, "floord"))
343 		return ISL_TOKEN_FLOORD;
344 	if (!strcasecmp(s->buffer, "mod"))
345 		return ISL_TOKEN_MOD;
346 	if (!strcasecmp(s->buffer, "ceil"))
347 		return ISL_TOKEN_CEIL;
348 	if (!strcasecmp(s->buffer, "floor"))
349 		return ISL_TOKEN_FLOOR;
350 
351 	if (!s->keywords)
352 		return ISL_TOKEN_IDENT;
353 
354 	name_hash = isl_hash_string(isl_hash_init(), s->buffer);
355 	entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
356 					s->buffer, 0);
357 	if (!entry)
358 		return ISL_TOKEN_ERROR;
359 	if (entry != isl_hash_table_entry_none) {
360 		keyword = entry->data;
361 		return keyword->type;
362 	}
363 
364 	return ISL_TOKEN_IDENT;
365 }
366 
isl_stream_skip_line(__isl_keep isl_stream * s)367 int isl_stream_skip_line(__isl_keep isl_stream *s)
368 {
369 	int c;
370 
371 	while ((c = isl_stream_getc(s)) != -1 && c != '\n')
372 		/* nothing */
373 		;
374 
375 	return c == -1 ? -1 : 0;
376 }
377 
next_token(__isl_keep isl_stream * s,int same_line)378 static struct isl_token *next_token(__isl_keep isl_stream *s, int same_line)
379 {
380 	int c;
381 	struct isl_token *tok = NULL;
382 	int line, col;
383 	int old_line = s->last_line;
384 
385 	if (s->n_token) {
386 		if (same_line && s->tokens[s->n_token - 1]->on_new_line)
387 			return NULL;
388 		return s->tokens[--s->n_token];
389 	}
390 
391 	if (same_line && s->c == '\n')
392 		return NULL;
393 
394 	s->len = 0;
395 
396 	/* skip spaces and comment lines */
397 	while ((c = isl_stream_getc(s)) != -1) {
398 		if (c == '#') {
399 			if (isl_stream_skip_line(s) < 0)
400 				break;
401 			c = '\n';
402 			if (same_line)
403 				break;
404 		} else if (!isspace(c) || (same_line && c == '\n'))
405 			break;
406 	}
407 
408 	line = s->start_line;
409 	col = s->start_col;
410 
411 	if (c == -1 || (same_line && c == '\n'))
412 		return NULL;
413 	s->last_line = line;
414 
415 	if (c == '(' ||
416 	    c == ')' ||
417 	    c == '+' ||
418 	    c == '*' ||
419 	    c == '%' ||
420 	    c == '?' ||
421 	    c == '^' ||
422 	    c == '@' ||
423 	    c == '$' ||
424 	    c == ',' ||
425 	    c == '.' ||
426 	    c == ';' ||
427 	    c == '[' ||
428 	    c == ']' ||
429 	    c == '{' ||
430 	    c == '}') {
431 		tok = isl_token_new(s->ctx, line, col, old_line != line);
432 		if (!tok)
433 			return NULL;
434 		tok->type = (enum isl_token_type)c;
435 		return tok;
436 	}
437 	if (c == '-') {
438 		int c;
439 		if ((c = isl_stream_getc(s)) == '>') {
440 			tok = isl_token_new(s->ctx, line, col, old_line != line);
441 			if (!tok)
442 				return NULL;
443 			tok->u.s = strdup("->");
444 			tok->type = ISL_TOKEN_TO;
445 			return tok;
446 		}
447 		if (c != -1)
448 			isl_stream_ungetc(s, c);
449 		if (!isdigit(c)) {
450 			tok = isl_token_new(s->ctx, line, col, old_line != line);
451 			if (!tok)
452 				return NULL;
453 			tok->type = (enum isl_token_type) '-';
454 			return tok;
455 		}
456 	}
457 	if (c == '-' || isdigit(c)) {
458 		int minus = c == '-';
459 		tok = isl_token_new(s->ctx, line, col, old_line != line);
460 		if (!tok)
461 			return NULL;
462 		tok->type = ISL_TOKEN_VALUE;
463 		isl_int_init(tok->u.v);
464 		if (isl_stream_push_char(s, c))
465 			goto error;
466 		while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
467 			if (isl_stream_push_char(s, c))
468 				goto error;
469 		if (c != -1)
470 			isl_stream_ungetc(s, c);
471 		isl_stream_push_char(s, '\0');
472 		isl_int_read(tok->u.v, s->buffer);
473 		if (minus && isl_int_is_zero(tok->u.v)) {
474 			tok->col++;
475 			tok->on_new_line = 0;
476 			isl_stream_push_token(s, tok);
477 			tok = isl_token_new(s->ctx, line, col, old_line != line);
478 			if (!tok)
479 				return NULL;
480 			tok->type = (enum isl_token_type) '-';
481 		}
482 		return tok;
483 	}
484 	if (isalpha(c) || c == '_') {
485 		tok = isl_token_new(s->ctx, line, col, old_line != line);
486 		if (!tok)
487 			return NULL;
488 		isl_stream_push_char(s, c);
489 		while ((c = isl_stream_getc(s)) != -1 &&
490 				(isalnum(c) || c == '_'))
491 			isl_stream_push_char(s, c);
492 		if (c != -1)
493 			isl_stream_ungetc(s, c);
494 		while ((c = isl_stream_getc(s)) != -1 && c == '\'')
495 			isl_stream_push_char(s, c);
496 		if (c != -1)
497 			isl_stream_ungetc(s, c);
498 		isl_stream_push_char(s, '\0');
499 		tok->type = check_keywords(s);
500 		if (tok->type != ISL_TOKEN_IDENT)
501 			tok->is_keyword = 1;
502 		tok->u.s = strdup(s->buffer);
503 		if (!tok->u.s)
504 			goto error;
505 		return tok;
506 	}
507 	if (c == '"') {
508 		tok = isl_token_new(s->ctx, line, col, old_line != line);
509 		if (!tok)
510 			return NULL;
511 		tok->type = ISL_TOKEN_STRING;
512 		tok->u.s = NULL;
513 		while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
514 			isl_stream_push_char(s, c);
515 		if (c != '"') {
516 			isl_stream_error(s, NULL, "unterminated string");
517 			goto error;
518 		}
519 		isl_stream_push_char(s, '\0');
520 		tok->u.s = strdup(s->buffer);
521 		return tok;
522 	}
523 	if (c == '=') {
524 		int c;
525 		tok = isl_token_new(s->ctx, line, col, old_line != line);
526 		if (!tok)
527 			return NULL;
528 		if ((c = isl_stream_getc(s)) == '=') {
529 			tok->u.s = strdup("==");
530 			tok->type = ISL_TOKEN_EQ_EQ;
531 			return tok;
532 		}
533 		if (c != -1)
534 			isl_stream_ungetc(s, c);
535 		tok->type = (enum isl_token_type) '=';
536 		return tok;
537 	}
538 	if (c == ':') {
539 		int c;
540 		tok = isl_token_new(s->ctx, line, col, old_line != line);
541 		if (!tok)
542 			return NULL;
543 		if ((c = isl_stream_getc(s)) == '=') {
544 			tok->u.s = strdup(":=");
545 			tok->type = ISL_TOKEN_DEF;
546 			return tok;
547 		}
548 		if (c != -1)
549 			isl_stream_ungetc(s, c);
550 		tok->type = (enum isl_token_type) ':';
551 		return tok;
552 	}
553 	if (c == '>') {
554 		int c;
555 		tok = isl_token_new(s->ctx, line, col, old_line != line);
556 		if (!tok)
557 			return NULL;
558 		if ((c = isl_stream_getc(s)) == '=') {
559 			tok->u.s = strdup(">=");
560 			tok->type = ISL_TOKEN_GE;
561 			return tok;
562 		} else if (c == '>') {
563 			if ((c = isl_stream_getc(s)) == '=') {
564 				tok->u.s = strdup(">>=");
565 				tok->type = ISL_TOKEN_LEX_GE;
566 				return tok;
567 			}
568 			tok->u.s = strdup(">>");
569 			tok->type = ISL_TOKEN_LEX_GT;
570 		} else {
571 			tok->u.s = strdup(">");
572 			tok->type = ISL_TOKEN_GT;
573 		}
574 		if (c != -1)
575 			isl_stream_ungetc(s, c);
576 		return tok;
577 	}
578 	if (c == '<') {
579 		int c;
580 		tok = isl_token_new(s->ctx, line, col, old_line != line);
581 		if (!tok)
582 			return NULL;
583 		if ((c = isl_stream_getc(s)) == '=') {
584 			tok->u.s = strdup("<=");
585 			tok->type = ISL_TOKEN_LE;
586 			return tok;
587 		} else if (c == '<') {
588 			if ((c = isl_stream_getc(s)) == '=') {
589 				tok->u.s = strdup("<<=");
590 				tok->type = ISL_TOKEN_LEX_LE;
591 				return tok;
592 			}
593 			tok->u.s = strdup("<<");
594 			tok->type = ISL_TOKEN_LEX_LT;
595 		} else {
596 			tok->u.s = strdup("<");
597 			tok->type = ISL_TOKEN_LT;
598 		}
599 		if (c != -1)
600 			isl_stream_ungetc(s, c);
601 		return tok;
602 	}
603 	if (c == '&') {
604 		tok = isl_token_new(s->ctx, line, col, old_line != line);
605 		if (!tok)
606 			return NULL;
607 		tok->type = ISL_TOKEN_AND;
608 		if ((c = isl_stream_getc(s)) != '&' && c != -1) {
609 			tok->u.s = strdup("&");
610 			isl_stream_ungetc(s, c);
611 		} else
612 			tok->u.s = strdup("&&");
613 		return tok;
614 	}
615 	if (c == '|') {
616 		tok = isl_token_new(s->ctx, line, col, old_line != line);
617 		if (!tok)
618 			return NULL;
619 		tok->type = ISL_TOKEN_OR;
620 		if ((c = isl_stream_getc(s)) != '|' && c != -1) {
621 			tok->u.s = strdup("|");
622 			isl_stream_ungetc(s, c);
623 		} else
624 			tok->u.s = strdup("||");
625 		return tok;
626 	}
627 	if (c == '/') {
628 		tok = isl_token_new(s->ctx, line, col, old_line != line);
629 		if (!tok)
630 			return NULL;
631 		if ((c = isl_stream_getc(s)) == '\\') {
632 			tok->u.s = strdup("/\\");
633 			tok->type = ISL_TOKEN_AND;
634 			return tok;
635 		} else if (c == '/') {
636 			tok->u.s = strdup("//");
637 			tok->type = ISL_TOKEN_INT_DIV;
638 			return tok;
639 		} else {
640 			tok->type = (enum isl_token_type) '/';
641 		}
642 		if (c != -1)
643 			isl_stream_ungetc(s, c);
644 		return tok;
645 	}
646 	if (c == '\\') {
647 		tok = isl_token_new(s->ctx, line, col, old_line != line);
648 		if (!tok)
649 			return NULL;
650 		if ((c = isl_stream_getc(s)) != '/' && c != -1) {
651 			tok->type = (enum isl_token_type) '\\';
652 			isl_stream_ungetc(s, c);
653 		} else {
654 			tok->u.s = strdup("\\/");
655 			tok->type = ISL_TOKEN_OR;
656 		}
657 		return tok;
658 	}
659 	if (c == '!') {
660 		tok = isl_token_new(s->ctx, line, col, old_line != line);
661 		if (!tok)
662 			return NULL;
663 		if ((c = isl_stream_getc(s)) == '=') {
664 			tok->u.s = strdup("!=");
665 			tok->type = ISL_TOKEN_NE;
666 			return tok;
667 		} else {
668 			tok->type = ISL_TOKEN_NOT;
669 			tok->u.s = strdup("!");
670 		}
671 		if (c != -1)
672 			isl_stream_ungetc(s, c);
673 		return tok;
674 	}
675 
676 	tok = isl_token_new(s->ctx, line, col, old_line != line);
677 	if (!tok)
678 		return NULL;
679 	tok->type = ISL_TOKEN_UNKNOWN;
680 	return tok;
681 error:
682 	isl_token_free(tok);
683 	return NULL;
684 }
685 
isl_stream_next_token(__isl_keep isl_stream * s)686 struct isl_token *isl_stream_next_token(__isl_keep isl_stream *s)
687 {
688 	return next_token(s, 0);
689 }
690 
isl_stream_next_token_on_same_line(__isl_keep isl_stream * s)691 struct isl_token *isl_stream_next_token_on_same_line(__isl_keep isl_stream *s)
692 {
693 	return next_token(s, 1);
694 }
695 
isl_stream_eat_if_available(__isl_keep isl_stream * s,int type)696 int isl_stream_eat_if_available(__isl_keep isl_stream *s, int type)
697 {
698 	struct isl_token *tok;
699 
700 	tok = isl_stream_next_token(s);
701 	if (!tok)
702 		return 0;
703 	if (tok->type == type) {
704 		isl_token_free(tok);
705 		return 1;
706 	}
707 	isl_stream_push_token(s, tok);
708 	return 0;
709 }
710 
isl_stream_next_token_is(__isl_keep isl_stream * s,int type)711 int isl_stream_next_token_is(__isl_keep isl_stream *s, int type)
712 {
713 	struct isl_token *tok;
714 	int r;
715 
716 	tok = isl_stream_next_token(s);
717 	if (!tok)
718 		return 0;
719 	r = tok->type == type;
720 	isl_stream_push_token(s, tok);
721 	return r;
722 }
723 
isl_stream_read_ident_if_available(__isl_keep isl_stream * s)724 char *isl_stream_read_ident_if_available(__isl_keep isl_stream *s)
725 {
726 	struct isl_token *tok;
727 
728 	tok = isl_stream_next_token(s);
729 	if (!tok)
730 		return NULL;
731 	if (tok->type == ISL_TOKEN_IDENT) {
732 		char *ident = strdup(tok->u.s);
733 		isl_token_free(tok);
734 		return ident;
735 	}
736 	isl_stream_push_token(s, tok);
737 	return NULL;
738 }
739 
isl_stream_eat(__isl_keep isl_stream * s,int type)740 int isl_stream_eat(__isl_keep isl_stream *s, int type)
741 {
742 	struct isl_token *tok;
743 
744 	tok = isl_stream_next_token(s);
745 	if (!tok) {
746 		if (s->eof)
747 			isl_stream_error(s, NULL, "unexpected EOF");
748 		return -1;
749 	}
750 	if (tok->type == type) {
751 		isl_token_free(tok);
752 		return 0;
753 	}
754 	isl_stream_error(s, tok, "expecting other token");
755 	isl_stream_push_token(s, tok);
756 	return -1;
757 }
758 
isl_stream_is_empty(__isl_keep isl_stream * s)759 int isl_stream_is_empty(__isl_keep isl_stream *s)
760 {
761 	struct isl_token *tok;
762 
763 	tok = isl_stream_next_token(s);
764 
765 	if (!tok)
766 		return 1;
767 
768 	isl_stream_push_token(s, tok);
769 	return 0;
770 }
771 
free_keyword(void ** p,void * user)772 static isl_stat free_keyword(void **p, void *user)
773 {
774 	struct isl_keyword *keyword = *p;
775 
776 	free(keyword->name);
777 	free(keyword);
778 
779 	return isl_stat_ok;
780 }
781 
isl_stream_flush_tokens(__isl_keep isl_stream * s)782 void isl_stream_flush_tokens(__isl_keep isl_stream *s)
783 {
784 	int i;
785 
786 	if (!s)
787 		return;
788 	for (i = 0; i < s->n_token; ++i)
789 		isl_token_free(s->tokens[i]);
790 	s->n_token = 0;
791 }
792 
isl_stream_get_ctx(__isl_keep isl_stream * s)793 isl_ctx *isl_stream_get_ctx(__isl_keep isl_stream *s)
794 {
795 	return s ? s->ctx : NULL;
796 }
797 
isl_stream_free(__isl_take isl_stream * s)798 void isl_stream_free(__isl_take isl_stream *s)
799 {
800 	if (!s)
801 		return;
802 	free(s->buffer);
803 	if (s->n_token != 0) {
804 		struct isl_token *tok = isl_stream_next_token(s);
805 		isl_stream_error(s, tok, "unexpected token");
806 		isl_token_free(tok);
807 	}
808 	if (s->keywords) {
809 		isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
810 		isl_hash_table_free(s->ctx, s->keywords);
811 	}
812 	free(s->yaml_state);
813 	free(s->yaml_indent);
814 	isl_ctx_deref(s->ctx);
815 	free(s);
816 }
817 
818 /* Push "state" onto the stack of currently active YAML elements.
819  * The caller is responsible for setting the corresponding indentation.
820  * Return 0 on success and -1 on failure.
821  */
push_state(__isl_keep isl_stream * s,enum isl_yaml_state state)822 static int push_state(__isl_keep isl_stream *s, enum isl_yaml_state state)
823 {
824 	if (s->yaml_size < s->yaml_depth + 1) {
825 		int *indent;
826 		enum isl_yaml_state *state;
827 
828 		state = isl_realloc_array(s->ctx, s->yaml_state,
829 					enum isl_yaml_state, s->yaml_depth + 1);
830 		if (!state)
831 			return -1;
832 		s->yaml_state = state;
833 
834 		indent = isl_realloc_array(s->ctx, s->yaml_indent,
835 					int, s->yaml_depth + 1);
836 		if (!indent)
837 			return -1;
838 		s->yaml_indent = indent;
839 
840 		s->yaml_size = s->yaml_depth + 1;
841 	}
842 
843 	s->yaml_state[s->yaml_depth] = state;
844 	s->yaml_depth++;
845 
846 	return 0;
847 }
848 
849 /* Remove the innermost active YAML element from the stack.
850  * Return 0 on success and -1 on failure.
851  */
pop_state(__isl_keep isl_stream * s)852 static int pop_state(__isl_keep isl_stream *s)
853 {
854 	if (!s)
855 		return -1;
856 	if (s->yaml_depth < 1)
857 		isl_die(isl_stream_get_ctx(s), isl_error_invalid,
858 			"not in YAML construct", return -1);
859 
860 	s->yaml_depth--;
861 
862 	return 0;
863 }
864 
865 /* Set the state of the innermost active YAML element to "state".
866  * Return 0 on success and -1 on failure.
867  */
update_state(__isl_keep isl_stream * s,enum isl_yaml_state state)868 static int update_state(__isl_keep isl_stream *s, enum isl_yaml_state state)
869 {
870 	if (!s)
871 		return -1;
872 	if (s->yaml_depth < 1)
873 		isl_die(isl_stream_get_ctx(s), isl_error_invalid,
874 			"not in YAML construct", return -1);
875 
876 	s->yaml_state[s->yaml_depth - 1] = state;
877 
878 	return 0;
879 }
880 
881 /* Return the state of the innermost active YAML element.
882  * Return isl_yaml_none if we are not inside any YAML element.
883  */
current_state(__isl_keep isl_stream * s)884 static enum isl_yaml_state current_state(__isl_keep isl_stream *s)
885 {
886 	if (!s)
887 		return isl_yaml_none;
888 	if (s->yaml_depth < 1)
889 		return isl_yaml_none;
890 	return s->yaml_state[s->yaml_depth - 1];
891 }
892 
893 /* Set the indentation of the innermost active YAML element to "indent".
894  * If "indent" is equal to ISL_YAML_INDENT_FLOW, then this means
895  * that the current elemient is in flow format.
896  */
set_yaml_indent(__isl_keep isl_stream * s,int indent)897 static int set_yaml_indent(__isl_keep isl_stream *s, int indent)
898 {
899 	if (s->yaml_depth < 1)
900 		isl_die(s->ctx, isl_error_internal,
901 			"not in YAML element", return -1);
902 
903 	s->yaml_indent[s->yaml_depth - 1] = indent;
904 
905 	return 0;
906 }
907 
908 /* Return the indentation of the innermost active YAML element
909  * of -1 on error.
910  */
get_yaml_indent(__isl_keep isl_stream * s)911 static int get_yaml_indent(__isl_keep isl_stream *s)
912 {
913 	if (s->yaml_depth < 1)
914 		isl_die(s->ctx, isl_error_internal,
915 			"not in YAML element", return -1);
916 
917 	return s->yaml_indent[s->yaml_depth - 1];
918 }
919 
920 /* Move to the next state at the innermost level.
921  * Return 1 if successful.
922  * Return 0 if we are at the end of the innermost level.
923  * Return -1 on error.
924  *
925  * If we are in state isl_yaml_mapping_key_start, then we have just
926  * started a mapping and we are expecting a key.  If the mapping started
927  * with a '{', then we check if the next token is a '}'.  If so,
928  * then the mapping is empty and there is no next state at this level.
929  * Otherwise, we assume that there is at least one key (the one from
930  * which we derived the indentation in isl_stream_yaml_read_start_mapping.
931  *
932  * If we are in state isl_yaml_mapping_key, then the we expect a colon
933  * followed by a value, so there is always a next state unless
934  * some error occurs.
935  *
936  * If we are in state isl_yaml_mapping_val, then there may or may
937  * not be a subsequent key in the same mapping.
938  * In flow format, the next key is preceded by a comma.
939  * In block format, the next key has the same indentation as the first key.
940  * If the first token has a smaller indentation, then we have reached
941  * the end of the current mapping.
942  *
943  * If we are in state isl_yaml_sequence_start, then we have just
944  * started a sequence.  If the sequence started with a '[',
945  * then we check if the next token is a ']'.  If so, then the sequence
946  * is empty and there is no next state at this level.
947  * Otherwise, we assume that there is at least one element in the sequence
948  * (the one from which we derived the indentation in
949  * isl_stream_yaml_read_start_sequence.
950  *
951  * If we are in state isl_yaml_sequence, then there may or may
952  * not be a subsequent element in the same sequence.
953  * In flow format, the next element is preceded by a comma.
954  * In block format, the next element is introduced by a dash with
955  * the same indentation as that of the first element.
956  * If the first token is not a dash or if it has a smaller indentation,
957  * then we have reached the end of the current sequence.
958  */
isl_stream_yaml_next(__isl_keep isl_stream * s)959 int isl_stream_yaml_next(__isl_keep isl_stream *s)
960 {
961 	struct isl_token *tok;
962 	enum isl_yaml_state state;
963 	int indent;
964 
965 	state = current_state(s);
966 	if (state == isl_yaml_none)
967 		isl_die(s->ctx, isl_error_invalid,
968 			"not in YAML element", return -1);
969 	switch (state) {
970 	case isl_yaml_mapping_key_start:
971 		if (get_yaml_indent(s) == ISL_YAML_INDENT_FLOW &&
972 		    isl_stream_next_token_is(s, '}'))
973 			return 0;
974 		if (update_state(s, isl_yaml_mapping_key) < 0)
975 			return -1;
976 		return 1;
977 	case isl_yaml_mapping_key:
978 		tok = isl_stream_next_token(s);
979 		if (!tok) {
980 			if (s->eof)
981 				isl_stream_error(s, NULL, "unexpected EOF");
982 			return -1;
983 		}
984 		if (tok->type == ':') {
985 			isl_token_free(tok);
986 			if (update_state(s, isl_yaml_mapping_val) < 0)
987 				return -1;
988 			return 1;
989 		}
990 		isl_stream_error(s, tok, "expecting ':'");
991 		isl_stream_push_token(s, tok);
992 		return -1;
993 	case isl_yaml_mapping_val:
994 		if (get_yaml_indent(s) == ISL_YAML_INDENT_FLOW) {
995 			if (!isl_stream_eat_if_available(s, ','))
996 				return 0;
997 			if (update_state(s, isl_yaml_mapping_key) < 0)
998 				return -1;
999 			return 1;
1000 		}
1001 		tok = isl_stream_next_token(s);
1002 		if (!tok)
1003 			return 0;
1004 		indent = tok->col - 1;
1005 		isl_stream_push_token(s, tok);
1006 		if (indent < get_yaml_indent(s))
1007 			return 0;
1008 		if (update_state(s, isl_yaml_mapping_key) < 0)
1009 			return -1;
1010 		return 1;
1011 	case isl_yaml_sequence_start:
1012 		if (get_yaml_indent(s) == ISL_YAML_INDENT_FLOW) {
1013 			if (isl_stream_next_token_is(s, ']'))
1014 				return 0;
1015 			if (update_state(s, isl_yaml_sequence) < 0)
1016 				return -1;
1017 			return 1;
1018 		}
1019 		tok = isl_stream_next_token(s);
1020 		if (!tok) {
1021 			if (s->eof)
1022 				isl_stream_error(s, NULL, "unexpected EOF");
1023 			return -1;
1024 		}
1025 		if (tok->type == '-') {
1026 			isl_token_free(tok);
1027 			if (update_state(s, isl_yaml_sequence) < 0)
1028 				return -1;
1029 			return 1;
1030 		}
1031 		isl_stream_error(s, tok, "expecting '-'");
1032 		isl_stream_push_token(s, tok);
1033 		return 0;
1034 	case isl_yaml_sequence:
1035 		if (get_yaml_indent(s) == ISL_YAML_INDENT_FLOW)
1036 			return isl_stream_eat_if_available(s, ',');
1037 		tok = isl_stream_next_token(s);
1038 		if (!tok)
1039 			return 0;
1040 		indent = tok->col - 1;
1041 		if (indent < get_yaml_indent(s) || tok->type != '-') {
1042 			isl_stream_push_token(s, tok);
1043 			return 0;
1044 		}
1045 		isl_token_free(tok);
1046 		return 1;
1047 	default:
1048 		isl_die(s->ctx, isl_error_internal,
1049 			"unexpected state", return 0);
1050 	}
1051 }
1052 
1053 /* Start reading a YAML mapping.
1054  * Return 0 on success and -1 on error.
1055  *
1056  * If the first token on the stream is a '{' then we remove this token
1057  * from the stream and keep track of the fact that the mapping
1058  * is given in flow format.
1059  * Otherwise, we assume the first token is the first key of the mapping and
1060  * keep track of its indentation, but keep the token on the stream.
1061  * In both cases, the next token we expect is the first key of the mapping.
1062  */
isl_stream_yaml_read_start_mapping(__isl_keep isl_stream * s)1063 int isl_stream_yaml_read_start_mapping(__isl_keep isl_stream *s)
1064 {
1065 	struct isl_token *tok;
1066 	int indent;
1067 
1068 	if (push_state(s, isl_yaml_mapping_key_start) < 0)
1069 		return -1;
1070 
1071 	tok = isl_stream_next_token(s);
1072 	if (!tok) {
1073 		if (s->eof)
1074 			isl_stream_error(s, NULL, "unexpected EOF");
1075 		return -1;
1076 	}
1077 	if (isl_token_get_type(tok) == '{') {
1078 		isl_token_free(tok);
1079 		return set_yaml_indent(s, ISL_YAML_INDENT_FLOW);
1080 	}
1081 	indent = tok->col - 1;
1082 	isl_stream_push_token(s, tok);
1083 
1084 	return set_yaml_indent(s, indent);
1085 }
1086 
1087 /* Finish reading a YAML mapping.
1088  * Return 0 on success and -1 on error.
1089  *
1090  * If the mapping started with a '{', then we expect a '}' to close
1091  * the mapping.
1092  * Otherwise, we double-check that the next token (if any)
1093  * has a smaller indentation than that of the current mapping.
1094  */
isl_stream_yaml_read_end_mapping(__isl_keep isl_stream * s)1095 int isl_stream_yaml_read_end_mapping(__isl_keep isl_stream *s)
1096 {
1097 	struct isl_token *tok;
1098 	int indent;
1099 
1100 	if (get_yaml_indent(s) == ISL_YAML_INDENT_FLOW) {
1101 		if (isl_stream_eat(s, '}') < 0)
1102 			return -1;
1103 		return pop_state(s);
1104 	}
1105 
1106 	tok = isl_stream_next_token(s);
1107 	if (!tok)
1108 		return pop_state(s);
1109 
1110 	indent = tok->col - 1;
1111 	isl_stream_push_token(s, tok);
1112 
1113 	if (indent >= get_yaml_indent(s))
1114 		isl_die(isl_stream_get_ctx(s), isl_error_invalid,
1115 			"mapping not finished", return -1);
1116 
1117 	return pop_state(s);
1118 }
1119 
1120 /* Start reading a YAML sequence.
1121  * Return 0 on success and -1 on error.
1122  *
1123  * If the first token on the stream is a '[' then we remove this token
1124  * from the stream and keep track of the fact that the sequence
1125  * is given in flow format.
1126  * Otherwise, we assume the first token is the dash that introduces
1127  * the first element of the sequence and keep track of its indentation,
1128  * but keep the token on the stream.
1129  * In both cases, the next token we expect is the first element
1130  * of the sequence.
1131  */
isl_stream_yaml_read_start_sequence(__isl_keep isl_stream * s)1132 int isl_stream_yaml_read_start_sequence(__isl_keep isl_stream *s)
1133 {
1134 	struct isl_token *tok;
1135 	int indent;
1136 
1137 	if (push_state(s, isl_yaml_sequence_start) < 0)
1138 		return -1;
1139 
1140 	tok = isl_stream_next_token(s);
1141 	if (!tok) {
1142 		if (s->eof)
1143 			isl_stream_error(s, NULL, "unexpected EOF");
1144 		return -1;
1145 	}
1146 	if (isl_token_get_type(tok) == '[') {
1147 		isl_token_free(tok);
1148 		return set_yaml_indent(s, ISL_YAML_INDENT_FLOW);
1149 	}
1150 	indent = tok->col - 1;
1151 	isl_stream_push_token(s, tok);
1152 
1153 	return set_yaml_indent(s, indent);
1154 }
1155 
1156 /* Finish reading a YAML sequence.
1157  * Return 0 on success and -1 on error.
1158  *
1159  * If the sequence started with a '[', then we expect a ']' to close
1160  * the sequence.
1161  * Otherwise, we double-check that the next token (if any)
1162  * is not a dash or that it has a smaller indentation than
1163  * that of the current sequence.
1164  */
isl_stream_yaml_read_end_sequence(__isl_keep isl_stream * s)1165 int isl_stream_yaml_read_end_sequence(__isl_keep isl_stream *s)
1166 {
1167 	struct isl_token *tok;
1168 	int indent;
1169 	int dash;
1170 
1171 	if (get_yaml_indent(s) == ISL_YAML_INDENT_FLOW) {
1172 		if (isl_stream_eat(s, ']') < 0)
1173 			return -1;
1174 		return pop_state(s);
1175 	}
1176 
1177 	tok = isl_stream_next_token(s);
1178 	if (!tok)
1179 		return pop_state(s);
1180 
1181 	indent = tok->col - 1;
1182 	dash = tok->type == '-';
1183 	isl_stream_push_token(s, tok);
1184 
1185 	if (indent >= get_yaml_indent(s) && dash)
1186 		isl_die(isl_stream_get_ctx(s), isl_error_invalid,
1187 			"sequence not finished", return -1);
1188 
1189 	return pop_state(s);
1190 }
1191