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                 char *buffer;
 217                 s->size = (3*s->size)/2;
 218                 buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
 219                 if (!buffer)
 220                         return -1;
 221                 s->buffer = buffer;
 222         }
 223         s->buffer[s->len++] = c;
 224         return 0;
 225 }
 226
 227 void isl_stream_push_token(struct isl_stream *s, struct isl_token *tok)
 228 {
 229         isl_assert(s->ctx, s->n_token < 5, return);
 230         s->tokens[s->n_token++] = tok;
 231 }
 232
 233 static enum isl_token_type check_keywords(struct isl_stream *s)
 234 {
 235         struct isl_hash_table_entry *entry;
 236         struct isl_keyword *keyword;
 237         uint32_t name_hash;
 238
 239         if (!strcasecmp(s->buffer, "exists"))
 240                 return ISL_TOKEN_EXISTS;
 241         if (!strcasecmp(s->buffer, "and"))
 242                 return ISL_TOKEN_AND;
 243         if (!strcasecmp(s->buffer, "or"))
 244                 return ISL_TOKEN_OR;
 245         if (!strcasecmp(s->buffer, "not"))
 246                 return ISL_TOKEN_NOT;
 247         if (!strcasecmp(s->buffer, "infty"))
 248                 return ISL_TOKEN_INFTY;
 249         if (!strcasecmp(s->buffer, "infinity"))
 250                 return ISL_TOKEN_INFTY;
 251         if (!strcasecmp(s->buffer, "NaN"))
 252                 return ISL_TOKEN_NAN;
 253         if (!strcasecmp(s->buffer, "max"))
 254                 return ISL_TOKEN_MAX;
 255         if (!strcasecmp(s->buffer, "rat"))
 256                 return ISL_TOKEN_RAT;
 257         if (!strcasecmp(s->buffer, "true"))
 258                 return ISL_TOKEN_TRUE;
 259         if (!strcasecmp(s->buffer, "false"))
 260                 return ISL_TOKEN_FALSE;
 261
 262         if (!s->keywords)
 263                 return ISL_TOKEN_IDENT;
 264
 265         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 266         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 267                                         s->buffer, 0);
 268         if (entry) {
 269                 keyword = entry->data;
 270                 return keyword->type;
 271         }
 272
 273         return ISL_TOKEN_IDENT;
 274 }
 275
 276 int isl_stream_skip_line(struct isl_stream *s)
 277 {
 278         int c;
 279
 280         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 281                 /* nothing */
 282                 ;
 283
 284         return c == -1 ? -1 : 0;
 285 }
 286
 287 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 288 {
 289         int c;
 290         struct isl_token *tok = NULL;
 291         int line, col;
 292         int old_line = s->line;
 293
 294         if (s->n_token) {
 295                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 296                         return NULL;
 297                 return s->tokens[--s->n_token];
 298         }
 299
 300         if (same_line && s->c == '\n')
 301                 return NULL;
 302
 303         s->len = 0;
 304
 305         /* skip spaces and comment lines */
 306         while ((c = isl_stream_getc(s)) != -1) {
 307                 if (c == '#') {
 308                         if (isl_stream_skip_line(s) < 0)
 309                                 break;
 310                         c = '\n';
 311                         if (same_line)
 312                                 break;
 313                 } else if (!isspace(c) || (same_line && c == '\n'))
 314                         break;
 315         }
 316
 317         line = s->line;
 318         col = s->col;
 319
 320         if (c == -1 || (same_line && c == '\n'))
 321                 return NULL;
 322         if (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             c == '{' ||
 337             c == '}') {
 338                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 339                 if (!tok)
 340                         return NULL;
 341                 tok->type = (enum isl_token_type)c;
 342                 return tok;
 343         }
 344         if (c == '-') {
 345                 int c;
 346                 if ((c = isl_stream_getc(s)) == '>') {
 347                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 348                         if (!tok)
 349                                 return NULL;
 350                         tok->u.s = strdup("->");
 351                         tok->type = ISL_TOKEN_TO;
 352                         return tok;
 353                 }
 354                 if (c != -1)
 355                         isl_stream_ungetc(s, c);
 356                 if (!isdigit(c)) {
 357                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 358                         if (!tok)
 359                                 return NULL;
 360                         tok->type = (enum isl_token_type) '-';
 361                         return tok;
 362                 }
 363         }
 364         if (c == '-' || isdigit(c)) {
 365                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 366                 if (!tok)
 367                         return NULL;
 368                 tok->type = ISL_TOKEN_VALUE;
 369                 isl_int_init(tok->u.v);
 370                 if (isl_stream_push_char(s, c))
 371                         goto error;
 372                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 373                         if (isl_stream_push_char(s, c))
 374                                 goto error;
 375                 if (c != -1)
 376                         isl_stream_ungetc(s, c);
 377                 isl_stream_push_char(s, '\0');
 378                 isl_int_read(tok->u.v, s->buffer);
 379                 return tok;
 380         }
 381         if (isalpha(c) || c == '_') {
 382                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 383                 if (!tok)
 384                         return NULL;
 385                 isl_stream_push_char(s, c);
 386                 while ((c = isl_stream_getc(s)) != -1 &&
 387                                 (isalnum(c) || c == '_'))
 388                         isl_stream_push_char(s, c);
 389                 if (c != -1)
 390                         isl_stream_ungetc(s, c);
 391                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 392                         isl_stream_push_char(s, c);
 393                 if (c != -1)
 394                         isl_stream_ungetc(s, c);
 395                 isl_stream_push_char(s, '\0');
 396                 tok->type = check_keywords(s);
 397                 if (tok->type != ISL_TOKEN_IDENT)
 398                         tok->is_keyword = 1;
 399                 tok->u.s = strdup(s->buffer);
 400                 if (!tok->u.s)
 401                         goto error;
 402                 return tok;
 403         }
 404         if (c == '"') {
 405                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 406                 if (!tok)
 407                         return NULL;
 408                 tok->type = ISL_TOKEN_STRING;
 409                 tok->u.s = NULL;
 410                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 411                         isl_stream_push_char(s, c);
 412                 if (c != '"') {
 413                         isl_stream_error(s, NULL, "unterminated string");
 414                         goto error;
 415                 }
 416                 isl_stream_push_char(s, '\0');
 417                 tok->u.s = strdup(s->buffer);
 418                 return tok;
 419         }
 420         if (c == ':') {
 421                 int c;
 422                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 423                 if (!tok)
 424                         return NULL;
 425                 if ((c = isl_stream_getc(s)) == '=') {
 426                         tok->u.s = strdup(":=");
 427                         tok->type = ISL_TOKEN_DEF;
 428                         return tok;
 429                 }
 430                 if (c != -1)
 431                         isl_stream_ungetc(s, c);
 432                 tok->type = (enum isl_token_type) ':';
 433                 return tok;
 434         }
 435         if (c == '>') {
 436                 int c;
 437                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 438                 if (!tok)
 439                         return NULL;
 440                 if ((c = isl_stream_getc(s)) == '=') {
 441                         tok->u.s = strdup(">=");
 442                         tok->type = ISL_TOKEN_GE;
 443                         return tok;
 444                 } else if (c == '>') {
 445                         if ((c = isl_stream_getc(s)) == '=') {
 446                                 tok->u.s = strdup(">>=");
 447                                 tok->type = ISL_TOKEN_LEX_GE;
 448                                 return tok;
 449                         }
 450                         tok->u.s = strdup(">>");
 451                         tok->type = ISL_TOKEN_LEX_GT;
 452                 } else {
 453                         tok->u.s = strdup(">");
 454                         tok->type = ISL_TOKEN_GT;
 455                 }
 456                 if (c != -1)
 457                         isl_stream_ungetc(s, c);
 458                 return tok;
 459         }
 460         if (c == '<') {
 461                 int c;
 462                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 463                 if (!tok)
 464                         return NULL;
 465                 if ((c = isl_stream_getc(s)) == '=') {
 466                         tok->u.s = strdup("<=");
 467                         tok->type = ISL_TOKEN_LE;
 468                         return tok;
 469                 } else if (c == '<') {
 470                         if ((c = isl_stream_getc(s)) == '=') {
 471                                 tok->u.s = strdup("<<=");
 472                                 tok->type = ISL_TOKEN_LEX_LE;
 473                                 return tok;
 474                         }
 475                         tok->u.s = strdup("<<");
 476                         tok->type = ISL_TOKEN_LEX_LT;
 477                 } else {
 478                         tok->u.s = strdup("<");
 479                         tok->type = ISL_TOKEN_LT;
 480                 }
 481                 if (c != -1)
 482                         isl_stream_ungetc(s, c);
 483                 return tok;
 484         }
 485         if (c == '&') {
 486                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 487                 if (!tok)
 488                         return NULL;
 489                 tok->type = ISL_TOKEN_AND;
 490                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 491                         tok->u.s = strdup("&");
 492                         isl_stream_ungetc(s, c);
 493                 } else
 494                         tok->u.s = strdup("&&");
 495                 return tok;
 496         }
 497         if (c == '|') {
 498                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 499                 if (!tok)
 500                         return NULL;
 501                 tok->type = ISL_TOKEN_OR;
 502                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 503                         tok->u.s = strdup("|");
 504                         isl_stream_ungetc(s, c);
 505                 } else
 506                         tok->u.s = strdup("||");
 507                 return tok;
 508         }
 509         if (c == '/') {
 510                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 511                 if (!tok)
 512                         return NULL;
 513                 if ((c = isl_stream_getc(s)) != '\\' && c != -1) {
 514                         tok->type = (enum isl_token_type) '/';
 515                         isl_stream_ungetc(s, c);
 516                 } else {
 517                         tok->u.s = strdup("/\\");
 518                         tok->type = ISL_TOKEN_AND;
 519                 }
 520                 return tok;
 521         }
 522         if (c == '\\') {
 523                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 524                 if (!tok)
 525                         return NULL;
 526                 if ((c = isl_stream_getc(s)) != '/' && c != -1) {
 527                         tok->type = (enum isl_token_type) '\\';
 528                         isl_stream_ungetc(s, c);
 529                 } else {
 530                         tok->u.s = strdup("\\/");
 531                         tok->type = ISL_TOKEN_OR;
 532                 }
 533                 return tok;
 534         }
 535         if (c == '!') {
 536                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 537                 if (!tok)
 538                         return NULL;
 539                 tok->type = ISL_TOKEN_NOT;
 540                 tok->u.s = strdup("!");
 541                 return tok;
 542         }
 543
 544         tok = isl_token_new(s->ctx, line, col, old_line != line);
 545         if (!tok)
 546                 return NULL;
 547         tok->type = ISL_TOKEN_UNKNOWN;
 548         return tok;
 549 error:
 550         isl_token_free(tok);
 551         return NULL;
 552 }
 553
 554 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 555 {
 556         return next_token(s, 0);
 557 }
 558
 559 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 560 {
 561         return next_token(s, 1);
 562 }
 563
 564 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 565 {
 566         struct isl_token *tok;
 567
 568         tok = isl_stream_next_token(s);
 569         if (!tok)
 570                 return 0;
 571         if (tok->type == type) {
 572                 isl_token_free(tok);
 573                 return 1;
 574         }
 575         isl_stream_push_token(s, tok);
 576         return 0;
 577 }
 578
 579 int isl_stream_next_token_is(struct isl_stream *s, int type)
 580 {
 581         struct isl_token *tok;
 582         int r;
 583
 584         tok = isl_stream_next_token(s);
 585         if (!tok)
 586                 return 0;
 587         r = tok->type == type;
 588         isl_stream_push_token(s, tok);
 589         return r;
 590 }
 591
 592 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 593 {
 594         struct isl_token *tok;
 595
 596         tok = isl_stream_next_token(s);
 597         if (!tok)
 598                 return NULL;
 599         if (tok->type == ISL_TOKEN_IDENT) {
 600                 char *ident = strdup(tok->u.s);
 601                 isl_token_free(tok);
 602                 return ident;
 603         }
 604         isl_stream_push_token(s, tok);
 605         return NULL;
 606 }
 607
 608 int isl_stream_eat(struct isl_stream *s, int type)
 609 {
 610         struct isl_token *tok;
 611
 612         tok = isl_stream_next_token(s);
 613         if (!tok)
 614                 return -1;
 615         if (tok->type == type) {
 616                 isl_token_free(tok);
 617                 return 0;
 618         }
 619         isl_stream_error(s, tok, "expecting other token");
 620         isl_stream_push_token(s, tok);
 621         return -1;
 622 }
 623
 624 int isl_stream_is_empty(struct isl_stream *s)
 625 {
 626         struct isl_token *tok;
 627
 628         tok = isl_stream_next_token(s);
 629
 630         if (!tok)
 631                 return 1;
 632
 633         isl_stream_push_token(s, tok);
 634         return 0;
 635 }
 636
 637 static int free_keyword(void **p, void *user)
 638 {
 639         struct isl_keyword *keyword = *p;
 640
 641         free(keyword->name);
 642         free(keyword);
 643
 644         return 0;
 645 }
 646
 647 void isl_stream_flush_tokens(struct isl_stream *s)
 648 {
 649         int i;
 650
 651         if (!s)
 652                 return;
 653         for (i = 0; i < s->n_token; ++i)
 654                 isl_token_free(s->tokens[i]);
 655         s->n_token = 0;
 656 }
 657
 658 void isl_stream_free(struct isl_stream *s)
 659 {
 660         if (!s)
 661                 return;
 662         free(s->buffer);
 663         if (s->n_token != 0) {
 664                 struct isl_token *tok = isl_stream_next_token(s);
 665                 isl_stream_error(s, tok, "unexpected token");
 666                 isl_token_free(tok);
 667         }
 668         if (s->keywords) {
 669                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 670                 isl_hash_table_free(s->ctx, s->keywords);
 671         }
 672         isl_ctx_deref(s->ctx);
 673         free(s);
 674 }