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         s->n_un = 0;
 131         for (i = 0; i < 5; ++i)
 132                 s->tokens[i] = NULL;
 133         s->n_token = 0;
 134         s->keywords = NULL;
 135         s->size = 256;
 136         s->buffer = isl_alloc_array(ctx, char, s->size);
 137         if (!s->buffer)
 138                 goto error;
 139         return s;
 140 error:
 141         isl_stream_free(s);
 142         return NULL;
 143 }
 144
 145 struct isl_stream* isl_stream_new_file(struct isl_ctx *ctx, FILE *file)
 146 {
 147         struct isl_stream *s = isl_stream_new(ctx);
 148         if (!s)
 149                 return NULL;
 150         s->file = file;
 151         return s;
 152 }
 153
 154 struct isl_stream* isl_stream_new_str(struct isl_ctx *ctx, const char *str)
 155 {
 156         struct isl_stream *s = isl_stream_new(ctx);
 157         if (!s)
 158                 return NULL;
 159         s->str = str;
 160         return s;
 161 }
 162
 163 static int stream_getc(struct isl_stream *s)
 164 {
 165         int c;
 166         if (s->eof)
 167                 return -1;
 168         if (s->n_un)
 169                 return s->c = s->un[--s->n_un];
 170         if (s->file)
 171                 c = fgetc(s->file);
 172         else {
 173                 c = *s->str++;
 174                 if (c == '\0')
 175                         c = -1;
 176         }
 177         if (c == -1)
 178                 s->eof = 1;
 179         if (!s->eof) {
 180                 if (s->c == '\n') {
 181                         s->line++;
 182                         s->col = 0;
 183                 } else
 184                         s->col++;
 185         }
 186         s->c = c;
 187         return c;
 188 }
 189
 190 static void isl_stream_ungetc(struct isl_stream *s, int c)
 191 {
 192         isl_assert(s->ctx, s->n_un < 5, return);
 193         s->un[s->n_un++] = c;
 194         s->c = -1;
 195 }
 196
 197 static int isl_stream_getc(struct isl_stream *s)
 198 {
 199         int c;
 200
 201         do {
 202                 c = stream_getc(s);
 203                 if (c != '\\')
 204                         return c;
 205                 c = stream_getc(s);
 206         } while (c == '\n');
 207
 208         isl_stream_ungetc(s, c);
 209
 210         return '\\';
 211 }
 212
 213 static int isl_stream_push_char(struct isl_stream *s, int c)
 214 {
 215         if (s->len >= s->size) {
 216                 s->size = (3*s->size)/2;
 217                 s->buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
 218                 if (!s->buffer)
 219                         return -1;
 220         }
 221         s->buffer[s->len++] = c;
 222         return 0;
 223 }
 224
 225 void isl_stream_push_token(struct isl_stream *s, struct isl_token *tok)
 226 {
 227         isl_assert(s->ctx, s->n_token < 5, return);
 228         s->tokens[s->n_token++] = tok;
 229 }
 230
 231 static enum isl_token_type check_keywords(struct isl_stream *s)
 232 {
 233         struct isl_hash_table_entry *entry;
 234         struct isl_keyword *keyword;
 235         uint32_t name_hash;
 236
 237         if (!strcasecmp(s->buffer, "exists"))
 238                 return ISL_TOKEN_EXISTS;
 239         if (!strcasecmp(s->buffer, "and"))
 240                 return ISL_TOKEN_AND;
 241         if (!strcasecmp(s->buffer, "or"))
 242                 return ISL_TOKEN_OR;
 243         if (!strcasecmp(s->buffer, "not"))
 244                 return ISL_TOKEN_NOT;
 245         if (!strcasecmp(s->buffer, "infty"))
 246                 return ISL_TOKEN_INFTY;
 247         if (!strcasecmp(s->buffer, "infinity"))
 248                 return ISL_TOKEN_INFTY;
 249         if (!strcasecmp(s->buffer, "NaN"))
 250                 return ISL_TOKEN_NAN;
 251         if (!strcasecmp(s->buffer, "max"))
 252                 return ISL_TOKEN_MAX;
 253         if (!strcasecmp(s->buffer, "rat"))
 254                 return ISL_TOKEN_RAT;
 255         if (!strcasecmp(s->buffer, "true"))
 256                 return ISL_TOKEN_TRUE;
 257         if (!strcasecmp(s->buffer, "false"))
 258                 return ISL_TOKEN_FALSE;
 259
 260         if (!s->keywords)
 261                 return ISL_TOKEN_IDENT;
 262
 263         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 264         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 265                                         s->buffer, 0);
 266         if (entry) {
 267                 keyword = entry->data;
 268                 return keyword->type;
 269         }
 270
 271         return ISL_TOKEN_IDENT;
 272 }
 273
 274 int isl_stream_skip_line(struct isl_stream *s)
 275 {
 276         int c;
 277
 278         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 279                 /* nothing */
 280                 ;
 281
 282         return c == -1 ? -1 : 0;
 283 }
 284
 285 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 286 {
 287         int c;
 288         struct isl_token *tok = NULL;
 289         int line, col;
 290         int old_line = s->line;
 291
 292         if (s->n_token) {
 293                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 294                         return NULL;
 295                 return s->tokens[--s->n_token];
 296         }
 297
 298         if (same_line && s->c == '\n')
 299                 return NULL;
 300
 301         s->len = 0;
 302
 303         /* skip spaces and comment lines */
 304         while ((c = isl_stream_getc(s)) != -1) {
 305                 if (c == '#') {
 306                         if (isl_stream_skip_line(s) < 0)
 307                                 break;
 308                         c = '\n';
 309                         if (same_line)
 310                                 break;
 311                 } else if (!isspace(c) || (same_line && c == '\n'))
 312                         break;
 313         }
 314
 315         line = s->line;
 316         col = s->col;
 317
 318         if (c == -1 || (same_line && c == '\n'))
 319                 return NULL;
 320         if (c == '(' ||
 321             c == ')' ||
 322             c == '+' ||
 323             c == '*' ||
 324             c == '%' ||
 325             c == '^' ||
 326             c == '=' ||
 327             c == '@' ||
 328             c == '$' ||
 329             c == ',' ||
 330             c == '.' ||
 331             c == ';' ||
 332             c == '[' ||
 333             c == ']' ||
 334             c == '{' ||
 335             c == '}') {
 336                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 337                 if (!tok)
 338                         return NULL;
 339                 tok->type = (enum isl_token_type)c;
 340                 return tok;
 341         }
 342         if (c == '-') {
 343                 int c;
 344                 if ((c = isl_stream_getc(s)) == '>') {
 345                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 346                         if (!tok)
 347                                 return NULL;
 348                         tok->u.s = strdup("->");
 349                         tok->type = ISL_TOKEN_TO;
 350                         return tok;
 351                 }
 352                 if (c != -1)
 353                         isl_stream_ungetc(s, c);
 354                 if (!isdigit(c)) {
 355                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 356                         if (!tok)
 357                                 return NULL;
 358                         tok->type = (enum isl_token_type) '-';
 359                         return tok;
 360                 }
 361         }
 362         if (c == '-' || isdigit(c)) {
 363                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 364                 if (!tok)
 365                         return NULL;
 366                 tok->type = ISL_TOKEN_VALUE;
 367                 isl_int_init(tok->u.v);
 368                 if (isl_stream_push_char(s, c))
 369                         goto error;
 370                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 371                         if (isl_stream_push_char(s, c))
 372                                 goto error;
 373                 if (c != -1)
 374                         isl_stream_ungetc(s, c);
 375                 isl_stream_push_char(s, '\0');
 376                 isl_int_read(tok->u.v, s->buffer);
 377                 return tok;
 378         }
 379         if (isalpha(c) || c == '_') {
 380                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 381                 if (!tok)
 382                         return NULL;
 383                 isl_stream_push_char(s, c);
 384                 while ((c = isl_stream_getc(s)) != -1 &&
 385                                 (isalnum(c) || c == '_'))
 386                         isl_stream_push_char(s, c);
 387                 if (c != -1)
 388                         isl_stream_ungetc(s, c);
 389                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 390                         isl_stream_push_char(s, c);
 391                 if (c != -1)
 392                         isl_stream_ungetc(s, c);
 393                 isl_stream_push_char(s, '\0');
 394                 tok->type = check_keywords(s);
 395                 if (tok->type != ISL_TOKEN_IDENT)
 396                         tok->is_keyword = 1;
 397                 tok->u.s = strdup(s->buffer);
 398                 if (!tok->u.s)
 399                         goto error;
 400                 return tok;
 401         }
 402         if (c == '"') {
 403                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 404                 if (!tok)
 405                         return NULL;
 406                 tok->type = ISL_TOKEN_STRING;
 407                 tok->u.s = NULL;
 408                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 409                         isl_stream_push_char(s, c);
 410                 if (c != '"') {
 411                         isl_stream_error(s, NULL, "unterminated string");
 412                         goto error;
 413                 }
 414                 isl_stream_push_char(s, '\0');
 415                 tok->u.s = strdup(s->buffer);
 416                 return tok;
 417         }
 418         if (c == ':') {
 419                 int c;
 420                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 421                 if (!tok)
 422                         return NULL;
 423                 if ((c = isl_stream_getc(s)) == '=') {
 424                         tok->u.s = strdup(":=");
 425                         tok->type = ISL_TOKEN_DEF;
 426                         return tok;
 427                 }
 428                 if (c != -1)
 429                         isl_stream_ungetc(s, c);
 430                 tok->type = (enum isl_token_type) ':';
 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_GE;
 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_GE;
 446                                 return tok;
 447                         }
 448                         tok->u.s = strdup(">>");
 449                         tok->type = ISL_TOKEN_LEX_GT;
 450                 } else {
 451                         tok->u.s = strdup(">");
 452                         tok->type = ISL_TOKEN_GT;
 453                 }
 454                 if (c != -1)
 455                         isl_stream_ungetc(s, c);
 456                 return tok;
 457         }
 458         if (c == '<') {
 459                 int c;
 460                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 461                 if (!tok)
 462                         return NULL;
 463                 if ((c = isl_stream_getc(s)) == '=') {
 464                         tok->u.s = strdup("<=");
 465                         tok->type = ISL_TOKEN_LE;
 466                         return tok;
 467                 } else if (c == '<') {
 468                         if ((c = isl_stream_getc(s)) == '=') {
 469                                 tok->u.s = strdup("<<=");
 470                                 tok->type = ISL_TOKEN_LEX_LE;
 471                                 return tok;
 472                         }
 473                         tok->u.s = strdup("<<");
 474                         tok->type = ISL_TOKEN_LEX_LT;
 475                 } else {
 476                         tok->u.s = strdup("<");
 477                         tok->type = ISL_TOKEN_LT;
 478                 }
 479                 if (c != -1)
 480                         isl_stream_ungetc(s, c);
 481                 return tok;
 482         }
 483         if (c == '&') {
 484                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 485                 if (!tok)
 486                         return NULL;
 487                 tok->type = ISL_TOKEN_AND;
 488                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 489                         tok->u.s = strdup("&");
 490                         isl_stream_ungetc(s, c);
 491                 } else
 492                         tok->u.s = strdup("&&");
 493                 return tok;
 494         }
 495         if (c == '|') {
 496                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 497                 if (!tok)
 498                         return NULL;
 499                 tok->type = ISL_TOKEN_OR;
 500                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 501                         tok->u.s = strdup("|");
 502                         isl_stream_ungetc(s, c);
 503                 } else
 504                         tok->u.s = strdup("||");
 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                 if ((c = isl_stream_getc(s)) != '\\' && c != -1) {
 512                         tok->type = (enum isl_token_type) '/';
 513                         isl_stream_ungetc(s, c);
 514                 } else {
 515                         tok->u.s = strdup("/\\");
 516                         tok->type = ISL_TOKEN_AND;
 517                 }
 518                 return tok;
 519         }
 520         if (c == '\\') {
 521                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 522                 if (!tok)
 523                         return NULL;
 524                 if ((c = isl_stream_getc(s)) != '/' && c != -1) {
 525                         tok->type = (enum isl_token_type) '\\';
 526                         isl_stream_ungetc(s, c);
 527                 } else {
 528                         tok->u.s = strdup("\\/");
 529                         tok->type = ISL_TOKEN_OR;
 530                 }
 531                 return tok;
 532         }
 533         if (c == '!') {
 534                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 535                 if (!tok)
 536                         return NULL;
 537                 tok->type = ISL_TOKEN_NOT;
 538                 tok->u.s = strdup("!");
 539                 return tok;
 540         }
 541
 542         tok = isl_token_new(s->ctx, line, col, old_line != line);
 543         if (!tok)
 544                 return NULL;
 545         tok->type = ISL_TOKEN_UNKNOWN;
 546         return tok;
 547 error:
 548         isl_token_free(tok);
 549         return NULL;
 550 }
 551
 552 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 553 {
 554         return next_token(s, 0);
 555 }
 556
 557 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 558 {
 559         return next_token(s, 1);
 560 }
 561
 562 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 563 {
 564         struct isl_token *tok;
 565
 566         tok = isl_stream_next_token(s);
 567         if (!tok)
 568                 return 0;
 569         if (tok->type == type) {
 570                 isl_token_free(tok);
 571                 return 1;
 572         }
 573         isl_stream_push_token(s, tok);
 574         return 0;
 575 }
 576
 577 int isl_stream_next_token_is(struct isl_stream *s, int type)
 578 {
 579         struct isl_token *tok;
 580         int r;
 581
 582         tok = isl_stream_next_token(s);
 583         if (!tok)
 584                 return 0;
 585         r = tok->type == type;
 586         isl_stream_push_token(s, tok);
 587         return r;
 588 }
 589
 590 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 591 {
 592         struct isl_token *tok;
 593
 594         tok = isl_stream_next_token(s);
 595         if (!tok)
 596                 return NULL;
 597         if (tok->type == ISL_TOKEN_IDENT) {
 598                 char *ident = strdup(tok->u.s);
 599                 isl_token_free(tok);
 600                 return ident;
 601         }
 602         isl_stream_push_token(s, tok);
 603         return NULL;
 604 }
 605
 606 int isl_stream_eat(struct isl_stream *s, int type)
 607 {
 608         struct isl_token *tok;
 609
 610         tok = isl_stream_next_token(s);
 611         if (!tok)
 612                 return -1;
 613         if (tok->type == type) {
 614                 isl_token_free(tok);
 615                 return 0;
 616         }
 617         isl_stream_error(s, tok, "expecting other token");
 618         isl_stream_push_token(s, tok);
 619         return -1;
 620 }
 621
 622 int isl_stream_is_empty(struct isl_stream *s)
 623 {
 624         struct isl_token *tok;
 625
 626         tok = isl_stream_next_token(s);
 627
 628         if (!tok)
 629                 return 1;
 630
 631         isl_stream_push_token(s, tok);
 632         return 0;
 633 }
 634
 635 static int free_keyword(void **p, void *user)
 636 {
 637         struct isl_keyword *keyword = *p;
 638
 639         free(keyword->name);
 640         free(keyword);
 641
 642         return 0;
 643 }
 644
 645 void isl_stream_flush_tokens(struct isl_stream *s)
 646 {
 647         int i;
 648
 649         if (!s)
 650                 return;
 651         for (i = 0; i < s->n_token; ++i)
 652                 isl_token_free(s->tokens[i]);
 653         s->n_token = 0;
 654 }
 655
 656 void isl_stream_free(struct isl_stream *s)
 657 {
 658         if (!s)
 659                 return;
 660         free(s->buffer);
 661         if (s->n_token != 0) {
 662                 struct isl_token *tok = isl_stream_next_token(s);
 663                 isl_stream_error(s, tok, "unexpected token");
 664                 isl_token_free(tok);
 665         }
 666         if (s->keywords) {
 667                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 668                 isl_hash_table_free(s->ctx, s->keywords);
 669         }
 670         isl_ctx_deref(s->ctx);
 671         free(s);
 672 }