isl_stream.c

   1 /*
   2  * Copyright 2008-2009 Katholieke Universiteit Leuven
   3  *
   4  * Use of this software is governed by the GNU LGPLv2.1 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 <strings.h>
  13 #include <isl/ctx.h>
  14 #include <isl/stream.h>
  15
  16 struct isl_keyword {
  17         char                    *name;
  18         enum isl_token_type     type;
  19 };
  20
  21 static int same_name(const void *entry, const void *val)
  22 {
  23         const struct isl_keyword *keyword = (const struct isl_keyword *)entry;
  24
  25         return !strcmp(keyword->name, val);
  26 }
  27
  28 enum isl_token_type isl_stream_register_keyword(struct isl_stream *s,
  29         const char *name)
  30 {
  31         struct isl_hash_table_entry *entry;
  32         struct isl_keyword *keyword;
  33         uint32_t name_hash;
  34
  35         if (!s->keywords) {
  36                 s->keywords = isl_hash_table_alloc(s->ctx, 10);
  37                 if (!s->keywords)
  38                         return ISL_TOKEN_ERROR;
  39                 s->next_type = ISL_TOKEN_LAST;
  40         }
  41
  42         name_hash = isl_hash_string(isl_hash_init(), name);
  43
  44         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash,
  45                                         same_name, name, 1);
  46         if (!entry)
  47                 return ISL_TOKEN_ERROR;
  48         if (entry->data) {
  49                 keyword = entry->data;
  50                 return keyword->type;
  51         }
  52
  53         keyword = isl_calloc_type(s->ctx, struct isl_keyword);
  54         if (!keyword)
  55                 return ISL_TOKEN_ERROR;
  56         keyword->type = s->next_type++;
  57         keyword->name = strdup(name);
  58         if (!keyword->name) {
  59                 free(keyword);
  60                 return ISL_TOKEN_ERROR;
  61         }
  62         entry->data = keyword;
  63
  64         return keyword->type;
  65 }
  66
  67 static struct isl_token *isl_token_new(struct isl_ctx *ctx,
  68         int line, int col, unsigned on_new_line)
  69 {
  70         struct isl_token *tok = isl_alloc_type(ctx, struct isl_token);
  71         if (!tok)
  72                 return NULL;
  73         tok->line = line;
  74         tok->col = col;
  75         tok->on_new_line = on_new_line;
  76         tok->is_keyword = 0;
  77         tok->u.s = NULL;
  78         return tok;
  79 }
  80
  81 void isl_token_free(struct isl_token *tok)
  82 {
  83         if (!tok)
  84                 return;
  85         if (tok->type == ISL_TOKEN_VALUE)
  86                 isl_int_clear(tok->u.v);
  87         else
  88                 free(tok->u.s);
  89         free(tok);
  90 }
  91
  92 void isl_stream_error(struct isl_stream *s, struct isl_token *tok, char *msg)
  93 {
  94         int line = tok ? tok->line : s->line;
  95         int col = tok ? tok->col : s->col;
  96         fprintf(stderr, "syntax error (%d, %d): %s\n", line, col, msg);
  97         if (tok) {
  98                 if (tok->type < 256)
  99                         fprintf(stderr, "got '%c'\n", tok->type);
 100                 else if (tok->type == ISL_TOKEN_IDENT)
 101                         fprintf(stderr, "got ident '%s'\n", tok->u.s);
 102                 else if (tok->is_keyword)
 103                         fprintf(stderr, "got keyword '%s'\n", tok->u.s);
 104                 else if (tok->type == ISL_TOKEN_VALUE) {
 105                         fprintf(stderr, "got value '");
 106                         isl_int_print(stderr, tok->u.v, 0);
 107                         fprintf(stderr, "'\n");
 108                 } else if (tok->u.s)
 109                         fprintf(stderr, "got token '%s'\n", tok->u.s);
 110                 else
 111                         fprintf(stderr, "got token type %d\n", tok->type);
 112         }
 113 }
 114
 115 static struct isl_stream* isl_stream_new(struct isl_ctx *ctx)
 116 {
 117         int i;
 118         struct isl_stream *s = isl_alloc_type(ctx, struct isl_stream);
 119         if (!s)
 120                 return NULL;
 121         s->ctx = ctx;
 122         isl_ctx_ref(s->ctx);
 123         s->file = NULL;
 124         s->str = NULL;
 125         s->len = 0;
 126         s->line = 1;
 127         s->col = 0;
 128         s->eof = 0;
 129         s->c = -1;
 130         for (i = 0; i < 5; ++i)
 131                 s->tokens[i] = NULL;
 132         s->n_token = 0;
 133         s->keywords = NULL;
 134         s->size = 256;
 135         s->buffer = isl_alloc_array(ctx, char, s->size);
 136         if (!s->buffer)
 137                 goto error;
 138         return s;
 139 error:
 140         isl_stream_free(s);
 141         return NULL;
 142 }
 143
 144 struct isl_stream* isl_stream_new_file(struct isl_ctx *ctx, FILE *file)
 145 {
 146         struct isl_stream *s = isl_stream_new(ctx);
 147         if (!s)
 148                 return NULL;
 149         s->file = file;
 150         return s;
 151 }
 152
 153 struct isl_stream* isl_stream_new_str(struct isl_ctx *ctx, const char *str)
 154 {
 155         struct isl_stream *s = isl_stream_new(ctx);
 156         if (!s)
 157                 return NULL;
 158         s->str = str;
 159         return s;
 160 }
 161
 162 static int isl_stream_getc(struct isl_stream *s)
 163 {
 164         int c;
 165         if (s->eof)
 166                 return -1;
 167         if (s->file)
 168                 c = fgetc(s->file);
 169         else {
 170                 c = *s->str++;
 171                 if (c == '\0')
 172                         c = -1;
 173         }
 174         if (c == -1)
 175                 s->eof = 1;
 176         if (!s->eof) {
 177                 if (s->c == '\n') {
 178                         s->line++;
 179                         s->col = 0;
 180                 } else
 181                         s->col++;
 182         }
 183         s->c = c;
 184         return c;
 185 }
 186
 187 static void isl_stream_ungetc(struct isl_stream *s, int c)
 188 {
 189         if (s->file)
 190                 ungetc(c, s->file);
 191         else
 192                 --s->str;
 193         s->c = -1;
 194 }
 195
 196 static int isl_stream_push_char(struct isl_stream *s, int c)
 197 {
 198         if (s->len >= s->size) {
 199                 s->size = (3*s->size)/2;
 200                 s->buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
 201                 if (!s->buffer)
 202                         return -1;
 203         }
 204         s->buffer[s->len++] = c;
 205         return 0;
 206 }
 207
 208 void isl_stream_push_token(struct isl_stream *s, struct isl_token *tok)
 209 {
 210         isl_assert(s->ctx, s->n_token < 5, return);
 211         s->tokens[s->n_token++] = tok;
 212 }
 213
 214 static enum isl_token_type check_keywords(struct isl_stream *s)
 215 {
 216         struct isl_hash_table_entry *entry;
 217         struct isl_keyword *keyword;
 218         uint32_t name_hash;
 219
 220         if (!strcasecmp(s->buffer, "exists"))
 221                 return ISL_TOKEN_EXISTS;
 222         if (!strcasecmp(s->buffer, "and"))
 223                 return ISL_TOKEN_AND;
 224         if (!strcasecmp(s->buffer, "or"))
 225                 return ISL_TOKEN_OR;
 226         if (!strcasecmp(s->buffer, "infty"))
 227                 return ISL_TOKEN_INFTY;
 228         if (!strcasecmp(s->buffer, "infinity"))
 229                 return ISL_TOKEN_INFTY;
 230         if (!strcasecmp(s->buffer, "NaN"))
 231                 return ISL_TOKEN_NAN;
 232         if (!strcasecmp(s->buffer, "max"))
 233                 return ISL_TOKEN_MAX;
 234
 235         if (!s->keywords)
 236                 return ISL_TOKEN_IDENT;
 237
 238         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 239         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 240                                         s->buffer, 0);
 241         if (entry) {
 242                 keyword = entry->data;
 243                 return keyword->type;
 244         }
 245
 246         return ISL_TOKEN_IDENT;
 247 }
 248
 249 int isl_stream_skip_line(struct isl_stream *s)
 250 {
 251         int c;
 252
 253         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 254                 /* nothing */
 255                 ;
 256
 257         return c == -1 ? -1 : 0;
 258 }
 259
 260 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 261 {
 262         int c;
 263         struct isl_token *tok = NULL;
 264         int line, col;
 265         int old_line = s->line;
 266
 267         if (s->n_token) {
 268                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 269                         return NULL;
 270                 return s->tokens[--s->n_token];
 271         }
 272
 273         if (same_line && s->c == '\n')
 274                 return NULL;
 275
 276         s->len = 0;
 277
 278         /* skip spaces and comment lines */
 279         while ((c = isl_stream_getc(s)) != -1) {
 280                 if (c == '#') {
 281                         if (isl_stream_skip_line(s) < 0)
 282                                 break;
 283                         c = '\n';
 284                         if (same_line)
 285                                 break;
 286                 } else if (!isspace(c) || (same_line && c == '\n'))
 287                         break;
 288         }
 289
 290         line = s->line;
 291         col = s->col;
 292
 293         if (c == -1 || (same_line && c == '\n'))
 294                 return NULL;
 295         if (c == '(' ||
 296             c == ')' ||
 297             c == '+' ||
 298             c == '/' ||
 299             c == '*' ||
 300             c == '%' ||
 301             c == '^' ||
 302             c == '=' ||
 303             c == '@' ||
 304             c == ',' ||
 305             c == '.' ||
 306             c == ';' ||
 307             c == '[' ||
 308             c == ']' ||
 309             c == '{' ||
 310             c == '}') {
 311                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 312                 if (!tok)
 313                         return NULL;
 314                 tok->type = (enum isl_token_type)c;
 315                 return tok;
 316         }
 317         if (c == '-') {
 318                 int c;
 319                 if ((c = isl_stream_getc(s)) == '>') {
 320                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 321                         if (!tok)
 322                                 return NULL;
 323                         tok->u.s = strdup("->");
 324                         tok->type = ISL_TOKEN_TO;
 325                         return tok;
 326                 }
 327                 if (c != -1)
 328                         isl_stream_ungetc(s, c);
 329                 if (!isdigit(c)) {
 330                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 331                         if (!tok)
 332                                 return NULL;
 333                         tok->type = (enum isl_token_type) '-';
 334                         return tok;
 335                 }
 336         }
 337         if (c == '-' || isdigit(c)) {
 338                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 339                 if (!tok)
 340                         return NULL;
 341                 tok->type = ISL_TOKEN_VALUE;
 342                 isl_int_init(tok->u.v);
 343                 if (isl_stream_push_char(s, c))
 344                         goto error;
 345                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 346                         if (isl_stream_push_char(s, c))
 347                                 goto error;
 348                 if (c != -1)
 349                         isl_stream_ungetc(s, c);
 350                 isl_stream_push_char(s, '\0');
 351                 isl_int_read(tok->u.v, s->buffer);
 352                 return tok;
 353         }
 354         if (isalpha(c) || c == '_') {
 355                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 356                 if (!tok)
 357                         return NULL;
 358                 isl_stream_push_char(s, c);
 359                 while ((c = isl_stream_getc(s)) != -1 &&
 360                                 (isalnum(c) || c == '_'))
 361                         isl_stream_push_char(s, c);
 362                 if (c != -1)
 363                         isl_stream_ungetc(s, c);
 364                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 365                         isl_stream_push_char(s, c);
 366                 if (c != -1)
 367                         isl_stream_ungetc(s, c);
 368                 isl_stream_push_char(s, '\0');
 369                 tok->type = check_keywords(s);
 370                 if (tok->type != ISL_TOKEN_IDENT)
 371                         tok->is_keyword = 1;
 372                 tok->u.s = strdup(s->buffer);
 373                 if (!tok->u.s)
 374                         goto error;
 375                 return tok;
 376         }
 377         if (c == '"') {
 378                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 379                 if (!tok)
 380                         return NULL;
 381                 tok->type = ISL_TOKEN_STRING;
 382                 tok->u.s = NULL;
 383                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 384                         isl_stream_push_char(s, c);
 385                 if (c != '"') {
 386                         isl_stream_error(s, NULL, "unterminated string");
 387                         goto error;
 388                 }
 389                 isl_stream_push_char(s, '\0');
 390                 tok->u.s = strdup(s->buffer);
 391                 return tok;
 392         }
 393         if (c == ':') {
 394                 int c;
 395                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 396                 if (!tok)
 397                         return NULL;
 398                 if ((c = isl_stream_getc(s)) == '=') {
 399                         tok->u.s = strdup(":=");
 400                         tok->type = ISL_TOKEN_DEF;
 401                         return tok;
 402                 }
 403                 if (c != -1)
 404                         isl_stream_ungetc(s, c);
 405                 tok->type = (enum isl_token_type) ':';
 406                 return tok;
 407         }
 408         if (c == '>') {
 409                 int c;
 410                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 411                 if (!tok)
 412                         return NULL;
 413                 if ((c = isl_stream_getc(s)) == '=') {
 414                         tok->u.s = strdup(">=");
 415                         tok->type = ISL_TOKEN_GE;
 416                         return tok;
 417                 } else if (c == '>') {
 418                         if ((c = isl_stream_getc(s)) == '=') {
 419                                 tok->u.s = strdup(">>=");
 420                                 tok->type = ISL_TOKEN_LEX_GE;
 421                                 return tok;
 422                         }
 423                         tok->u.s = strdup(">>");
 424                         tok->type = ISL_TOKEN_LEX_GT;
 425                 } else {
 426                         tok->u.s = strdup(">");
 427                         tok->type = ISL_TOKEN_GT;
 428                 }
 429                 if (c != -1)
 430                         isl_stream_ungetc(s, c);
 431                 return tok;
 432         }
 433         if (c == '<') {
 434                 int c;
 435                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 436                 if (!tok)
 437                         return NULL;
 438                 if ((c = isl_stream_getc(s)) == '=') {
 439                         tok->u.s = strdup("<=");
 440                         tok->type = ISL_TOKEN_LE;
 441                         return tok;
 442                 } else if (c == '<') {
 443                         if ((c = isl_stream_getc(s)) == '=') {
 444                                 tok->u.s = strdup("<<=");
 445                                 tok->type = ISL_TOKEN_LEX_LE;
 446                                 return tok;
 447                         }
 448                         tok->u.s = strdup("<<");
 449                         tok->type = ISL_TOKEN_LEX_LT;
 450                 } else {
 451                         tok->u.s = strdup("<");
 452                         tok->type = ISL_TOKEN_LT;
 453                 }
 454                 if (c != -1)
 455                         isl_stream_ungetc(s, c);
 456                 return tok;
 457         }
 458         if (c == '&') {
 459                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 460                 if (!tok)
 461                         return NULL;
 462                 tok->type = ISL_TOKEN_AND;
 463                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 464                         tok->u.s = strdup("&");
 465                         isl_stream_ungetc(s, c);
 466                 } else
 467                         tok->u.s = strdup("&&");
 468                 return tok;
 469         }
 470         if (c == '|') {
 471                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 472                 if (!tok)
 473                         return NULL;
 474                 tok->type = ISL_TOKEN_OR;
 475                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 476                         tok->u.s = strdup("|");
 477                         isl_stream_ungetc(s, c);
 478                 } else
 479                         tok->u.s = strdup("||");
 480                 return tok;
 481         }
 482
 483         tok = isl_token_new(s->ctx, line, col, old_line != line);
 484         if (!tok)
 485                 return NULL;
 486         tok->type = ISL_TOKEN_UNKNOWN;
 487         return tok;
 488 error:
 489         isl_token_free(tok);
 490         return NULL;
 491 }
 492
 493 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 494 {
 495         return next_token(s, 0);
 496 }
 497
 498 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 499 {
 500         return next_token(s, 1);
 501 }
 502
 503 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 504 {
 505         struct isl_token *tok;
 506
 507         tok = isl_stream_next_token(s);
 508         if (!tok)
 509                 return 0;
 510         if (tok->type == type) {
 511                 isl_token_free(tok);
 512                 return 1;
 513         }
 514         isl_stream_push_token(s, tok);
 515         return 0;
 516 }
 517
 518 int isl_stream_next_token_is(struct isl_stream *s, int type)
 519 {
 520         struct isl_token *tok;
 521         int r;
 522
 523         tok = isl_stream_next_token(s);
 524         if (!tok)
 525                 return 0;
 526         r = tok->type == type;
 527         isl_stream_push_token(s, tok);
 528         return r;
 529 }
 530
 531 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 532 {
 533         struct isl_token *tok;
 534
 535         tok = isl_stream_next_token(s);
 536         if (!tok)
 537                 return NULL;
 538         if (tok->type == ISL_TOKEN_IDENT) {
 539                 char *ident = strdup(tok->u.s);
 540                 isl_token_free(tok);
 541                 return ident;
 542         }
 543         isl_stream_push_token(s, tok);
 544         return NULL;
 545 }
 546
 547 int isl_stream_eat(struct isl_stream *s, int type)
 548 {
 549         struct isl_token *tok;
 550
 551         tok = isl_stream_next_token(s);
 552         if (!tok)
 553                 return -1;
 554         if (tok->type == type) {
 555                 isl_token_free(tok);
 556                 return 0;
 557         }
 558         isl_stream_error(s, tok, "expecting other token");
 559         isl_stream_push_token(s, tok);
 560         return -1;
 561 }
 562
 563 int isl_stream_is_empty(struct isl_stream *s)
 564 {
 565         struct isl_token *tok;
 566
 567         tok = isl_stream_next_token(s);
 568
 569         if (!tok)
 570                 return 1;
 571
 572         isl_stream_push_token(s, tok);
 573         return 0;
 574 }
 575
 576 static int free_keyword(void **p, void *user)
 577 {
 578         struct isl_keyword *keyword = *p;
 579
 580         free(keyword->name);
 581         free(keyword);
 582
 583         return 0;
 584 }
 585
 586 void isl_stream_flush_tokens(struct isl_stream *s)
 587 {
 588         int i;
 589
 590         if (!s)
 591                 return;
 592         for (i = 0; i < s->n_token; ++i)
 593                 isl_token_free(s->tokens[i]);
 594         s->n_token = 0;
 595 }
 596
 597 void isl_stream_free(struct isl_stream *s)
 598 {
 599         if (!s)
 600                 return;
 601         free(s->buffer);
 602         if (s->n_token != 0) {
 603                 struct isl_token *tok = isl_stream_next_token(s);
 604                 isl_stream_error(s, tok, "unexpected token");
 605                 isl_token_free(tok);
 606         }
 607         if (s->keywords) {
 608                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 609                 isl_hash_table_free(s->ctx, s->keywords);
 610         }
 611         isl_ctx_deref(s->ctx);
 612         free(s);
 613 }