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                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 336                 if (!tok)
 337                         return NULL;
 338                 tok->type = (enum isl_token_type)c;
 339                 return tok;
 340         }
 341         if (c == '-') {
 342                 int c;
 343                 if ((c = isl_stream_getc(s)) == '>') {
 344                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 345                         if (!tok)
 346                                 return NULL;
 347                         tok->u.s = strdup("->");
 348                         tok->type = ISL_TOKEN_TO;
 349                         return tok;
 350                 }
 351                 if (c != -1)
 352                         isl_stream_ungetc(s, c);
 353                 if (!isdigit(c)) {
 354                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 355                         if (!tok)
 356                                 return NULL;
 357                         tok->type = (enum isl_token_type) '-';
 358                         return tok;
 359                 }
 360         }
 361         if (c == '-' || isdigit(c)) {
 362                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 363                 if (!tok)
 364                         return NULL;
 365                 tok->type = ISL_TOKEN_VALUE;
 366                 isl_int_init(tok->u.v);
 367                 if (isl_stream_push_char(s, c))
 368                         goto error;
 369                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 370                         if (isl_stream_push_char(s, c))
 371                                 goto error;
 372                 if (c != -1)
 373                         isl_stream_ungetc(s, c);
 374                 isl_stream_push_char(s, '\0');
 375                 isl_int_read(tok->u.v, s->buffer);
 376                 return tok;
 377         }
 378         if (isalpha(c) || c == '_') {
 379                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 380                 if (!tok)
 381                         return NULL;
 382                 isl_stream_push_char(s, c);
 383                 while ((c = isl_stream_getc(s)) != -1 &&
 384                                 (isalnum(c) || c == '_'))
 385                         isl_stream_push_char(s, c);
 386                 if (c != -1)
 387                         isl_stream_ungetc(s, c);
 388                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 389                         isl_stream_push_char(s, c);
 390                 if (c != -1)
 391                         isl_stream_ungetc(s, c);
 392                 isl_stream_push_char(s, '\0');
 393                 tok->type = check_keywords(s);
 394                 if (tok->type != ISL_TOKEN_IDENT)
 395                         tok->is_keyword = 1;
 396                 tok->u.s = strdup(s->buffer);
 397                 if (!tok->u.s)
 398                         goto error;
 399                 return tok;
 400         }
 401         if (c == '"') {
 402                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 403                 if (!tok)
 404                         return NULL;
 405                 tok->type = ISL_TOKEN_STRING;
 406                 tok->u.s = NULL;
 407                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 408                         isl_stream_push_char(s, c);
 409                 if (c != '"') {
 410                         isl_stream_error(s, NULL, "unterminated string");
 411                         goto error;
 412                 }
 413                 isl_stream_push_char(s, '\0');
 414                 tok->u.s = strdup(s->buffer);
 415                 return tok;
 416         }
 417         if (c == ':') {
 418                 int c;
 419                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 420                 if (!tok)
 421                         return NULL;
 422                 if ((c = isl_stream_getc(s)) == '=') {
 423                         tok->u.s = strdup(":=");
 424                         tok->type = ISL_TOKEN_DEF;
 425                         return tok;
 426                 }
 427                 if (c != -1)
 428                         isl_stream_ungetc(s, c);
 429                 tok->type = (enum isl_token_type) ':';
 430                 return tok;
 431         }
 432         if (c == '>') {
 433                 int c;
 434                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 435                 if (!tok)
 436                         return NULL;
 437                 if ((c = isl_stream_getc(s)) == '=') {
 438                         tok->u.s = strdup(">=");
 439                         tok->type = ISL_TOKEN_GE;
 440                         return tok;
 441                 } else if (c == '>') {
 442                         if ((c = isl_stream_getc(s)) == '=') {
 443                                 tok->u.s = strdup(">>=");
 444                                 tok->type = ISL_TOKEN_LEX_GE;
 445                                 return tok;
 446                         }
 447                         tok->u.s = strdup(">>");
 448                         tok->type = ISL_TOKEN_LEX_GT;
 449                 } else {
 450                         tok->u.s = strdup(">");
 451                         tok->type = ISL_TOKEN_GT;
 452                 }
 453                 if (c != -1)
 454                         isl_stream_ungetc(s, c);
 455                 return tok;
 456         }
 457         if (c == '<') {
 458                 int c;
 459                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 460                 if (!tok)
 461                         return NULL;
 462                 if ((c = isl_stream_getc(s)) == '=') {
 463                         tok->u.s = strdup("<=");
 464                         tok->type = ISL_TOKEN_LE;
 465                         return tok;
 466                 } else if (c == '<') {
 467                         if ((c = isl_stream_getc(s)) == '=') {
 468                                 tok->u.s = strdup("<<=");
 469                                 tok->type = ISL_TOKEN_LEX_LE;
 470                                 return tok;
 471                         }
 472                         tok->u.s = strdup("<<");
 473                         tok->type = ISL_TOKEN_LEX_LT;
 474                 } else {
 475                         tok->u.s = strdup("<");
 476                         tok->type = ISL_TOKEN_LT;
 477                 }
 478                 if (c != -1)
 479                         isl_stream_ungetc(s, c);
 480                 return tok;
 481         }
 482         if (c == '&') {
 483                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 484                 if (!tok)
 485                         return NULL;
 486                 tok->type = ISL_TOKEN_AND;
 487                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 488                         tok->u.s = strdup("&");
 489                         isl_stream_ungetc(s, c);
 490                 } else
 491                         tok->u.s = strdup("&&");
 492                 return tok;
 493         }
 494         if (c == '|') {
 495                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 496                 if (!tok)
 497                         return NULL;
 498                 tok->type = ISL_TOKEN_OR;
 499                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 500                         tok->u.s = strdup("|");
 501                         isl_stream_ungetc(s, c);
 502                 } else
 503                         tok->u.s = strdup("||");
 504                 return tok;
 505         }
 506         if (c == '/') {
 507                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 508                 if (!tok)
 509                         return NULL;
 510                 if ((c = isl_stream_getc(s)) != '\\' && c != -1) {
 511                         tok->type = (enum isl_token_type) '/';
 512                         isl_stream_ungetc(s, c);
 513                 } else {
 514                         tok->u.s = strdup("/\\");
 515                         tok->type = ISL_TOKEN_AND;
 516                 }
 517                 return tok;
 518         }
 519         if (c == '\\') {
 520                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 521                 if (!tok)
 522                         return NULL;
 523                 if ((c = isl_stream_getc(s)) != '/' && c != -1) {
 524                         tok->type = (enum isl_token_type) '\\';
 525                         isl_stream_ungetc(s, c);
 526                 } else {
 527                         tok->u.s = strdup("\\/");
 528                         tok->type = ISL_TOKEN_OR;
 529                 }
 530                 return tok;
 531         }
 532         if (c == '!') {
 533                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 534                 if (!tok)
 535                         return NULL;
 536                 tok->type = ISL_TOKEN_NOT;
 537                 tok->u.s = strdup("!");
 538                 return tok;
 539         }
 540
 541         tok = isl_token_new(s->ctx, line, col, old_line != line);
 542         if (!tok)
 543                 return NULL;
 544         tok->type = ISL_TOKEN_UNKNOWN;
 545         return tok;
 546 error:
 547         isl_token_free(tok);
 548         return NULL;
 549 }
 550
 551 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 552 {
 553         return next_token(s, 0);
 554 }
 555
 556 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 557 {
 558         return next_token(s, 1);
 559 }
 560
 561 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 562 {
 563         struct isl_token *tok;
 564
 565         tok = isl_stream_next_token(s);
 566         if (!tok)
 567                 return 0;
 568         if (tok->type == type) {
 569                 isl_token_free(tok);
 570                 return 1;
 571         }
 572         isl_stream_push_token(s, tok);
 573         return 0;
 574 }
 575
 576 int isl_stream_next_token_is(struct isl_stream *s, int type)
 577 {
 578         struct isl_token *tok;
 579         int r;
 580
 581         tok = isl_stream_next_token(s);
 582         if (!tok)
 583                 return 0;
 584         r = tok->type == type;
 585         isl_stream_push_token(s, tok);
 586         return r;
 587 }
 588
 589 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 590 {
 591         struct isl_token *tok;
 592
 593         tok = isl_stream_next_token(s);
 594         if (!tok)
 595                 return NULL;
 596         if (tok->type == ISL_TOKEN_IDENT) {
 597                 char *ident = strdup(tok->u.s);
 598                 isl_token_free(tok);
 599                 return ident;
 600         }
 601         isl_stream_push_token(s, tok);
 602         return NULL;
 603 }
 604
 605 int isl_stream_eat(struct isl_stream *s, int type)
 606 {
 607         struct isl_token *tok;
 608
 609         tok = isl_stream_next_token(s);
 610         if (!tok)
 611                 return -1;
 612         if (tok->type == type) {
 613                 isl_token_free(tok);
 614                 return 0;
 615         }
 616         isl_stream_error(s, tok, "expecting other token");
 617         isl_stream_push_token(s, tok);
 618         return -1;
 619 }
 620
 621 int isl_stream_is_empty(struct isl_stream *s)
 622 {
 623         struct isl_token *tok;
 624
 625         tok = isl_stream_next_token(s);
 626
 627         if (!tok)
 628                 return 1;
 629
 630         isl_stream_push_token(s, tok);
 631         return 0;
 632 }
 633
 634 static int free_keyword(void **p, void *user)
 635 {
 636         struct isl_keyword *keyword = *p;
 637
 638         free(keyword->name);
 639         free(keyword);
 640
 641         return 0;
 642 }
 643
 644 void isl_stream_flush_tokens(struct isl_stream *s)
 645 {
 646         int i;
 647
 648         if (!s)
 649                 return;
 650         for (i = 0; i < s->n_token; ++i)
 651                 isl_token_free(s->tokens[i]);
 652         s->n_token = 0;
 653 }
 654
 655 void isl_stream_free(struct isl_stream *s)
 656 {
 657         if (!s)
 658                 return;
 659         free(s->buffer);
 660         if (s->n_token != 0) {
 661                 struct isl_token *tok = isl_stream_next_token(s);
 662                 isl_stream_error(s, tok, "unexpected token");
 663                 isl_token_free(tok);
 664         }
 665         if (s->keywords) {
 666                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 667                 isl_hash_table_free(s->ctx, s->keywords);
 668         }
 669         isl_ctx_deref(s->ctx);
 670         free(s);
 671 }