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, "min"))
 254                 return ISL_TOKEN_MIN;
 255         if (!strcasecmp(s->buffer, "max"))
 256                 return ISL_TOKEN_MAX;
 257         if (!strcasecmp(s->buffer, "rat"))
 258                 return ISL_TOKEN_RAT;
 259         if (!strcasecmp(s->buffer, "true"))
 260                 return ISL_TOKEN_TRUE;
 261         if (!strcasecmp(s->buffer, "false"))
 262                 return ISL_TOKEN_FALSE;
 263         if (!strcasecmp(s->buffer, "ceild"))
 264                 return ISL_TOKEN_CEILD;
 265         if (!strcasecmp(s->buffer, "floord"))
 266                 return ISL_TOKEN_FLOORD;
 267
 268         if (!s->keywords)
 269                 return ISL_TOKEN_IDENT;
 270
 271         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 272         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 273                                         s->buffer, 0);
 274         if (entry) {
 275                 keyword = entry->data;
 276                 return keyword->type;
 277         }
 278
 279         return ISL_TOKEN_IDENT;
 280 }
 281
 282 int isl_stream_skip_line(struct isl_stream *s)
 283 {
 284         int c;
 285
 286         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 287                 /* nothing */
 288                 ;
 289
 290         return c == -1 ? -1 : 0;
 291 }
 292
 293 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 294 {
 295         int c;
 296         struct isl_token *tok = NULL;
 297         int line, col;
 298         int old_line = s->line;
 299
 300         if (s->n_token) {
 301                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 302                         return NULL;
 303                 return s->tokens[--s->n_token];
 304         }
 305
 306         if (same_line && s->c == '\n')
 307                 return NULL;
 308
 309         s->len = 0;
 310
 311         /* skip spaces and comment lines */
 312         while ((c = isl_stream_getc(s)) != -1) {
 313                 if (c == '#') {
 314                         if (isl_stream_skip_line(s) < 0)
 315                                 break;
 316                         c = '\n';
 317                         if (same_line)
 318                                 break;
 319                 } else if (!isspace(c) || (same_line && c == '\n'))
 320                         break;
 321         }
 322
 323         line = s->line;
 324         col = s->col;
 325
 326         if (c == -1 || (same_line && c == '\n'))
 327                 return NULL;
 328         if (c == '(' ||
 329             c == ')' ||
 330             c == '+' ||
 331             c == '*' ||
 332             c == '%' ||
 333             c == '^' ||
 334             c == '=' ||
 335             c == '@' ||
 336             c == '$' ||
 337             c == ',' ||
 338             c == '.' ||
 339             c == ';' ||
 340             c == '[' ||
 341             c == ']' ||
 342             c == '{' ||
 343             c == '}') {
 344                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 345                 if (!tok)
 346                         return NULL;
 347                 tok->type = (enum isl_token_type)c;
 348                 return tok;
 349         }
 350         if (c == '-') {
 351                 int c;
 352                 if ((c = isl_stream_getc(s)) == '>') {
 353                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 354                         if (!tok)
 355                                 return NULL;
 356                         tok->u.s = strdup("->");
 357                         tok->type = ISL_TOKEN_TO;
 358                         return tok;
 359                 }
 360                 if (c != -1)
 361                         isl_stream_ungetc(s, c);
 362                 if (!isdigit(c)) {
 363                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 364                         if (!tok)
 365                                 return NULL;
 366                         tok->type = (enum isl_token_type) '-';
 367                         return tok;
 368                 }
 369         }
 370         if (c == '-' || isdigit(c)) {
 371                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 372                 if (!tok)
 373                         return NULL;
 374                 tok->type = ISL_TOKEN_VALUE;
 375                 isl_int_init(tok->u.v);
 376                 if (isl_stream_push_char(s, c))
 377                         goto error;
 378                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 379                         if (isl_stream_push_char(s, c))
 380                                 goto error;
 381                 if (c != -1)
 382                         isl_stream_ungetc(s, c);
 383                 isl_stream_push_char(s, '\0');
 384                 isl_int_read(tok->u.v, s->buffer);
 385                 return tok;
 386         }
 387         if (isalpha(c) || c == '_') {
 388                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 389                 if (!tok)
 390                         return NULL;
 391                 isl_stream_push_char(s, c);
 392                 while ((c = isl_stream_getc(s)) != -1 &&
 393                                 (isalnum(c) || c == '_'))
 394                         isl_stream_push_char(s, c);
 395                 if (c != -1)
 396                         isl_stream_ungetc(s, c);
 397                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 398                         isl_stream_push_char(s, c);
 399                 if (c != -1)
 400                         isl_stream_ungetc(s, c);
 401                 isl_stream_push_char(s, '\0');
 402                 tok->type = check_keywords(s);
 403                 if (tok->type != ISL_TOKEN_IDENT)
 404                         tok->is_keyword = 1;
 405                 tok->u.s = strdup(s->buffer);
 406                 if (!tok->u.s)
 407                         goto error;
 408                 return tok;
 409         }
 410         if (c == '"') {
 411                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 412                 if (!tok)
 413                         return NULL;
 414                 tok->type = ISL_TOKEN_STRING;
 415                 tok->u.s = NULL;
 416                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 417                         isl_stream_push_char(s, c);
 418                 if (c != '"') {
 419                         isl_stream_error(s, NULL, "unterminated string");
 420                         goto error;
 421                 }
 422                 isl_stream_push_char(s, '\0');
 423                 tok->u.s = strdup(s->buffer);
 424                 return tok;
 425         }
 426         if (c == ':') {
 427                 int c;
 428                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 429                 if (!tok)
 430                         return NULL;
 431                 if ((c = isl_stream_getc(s)) == '=') {
 432                         tok->u.s = strdup(":=");
 433                         tok->type = ISL_TOKEN_DEF;
 434                         return tok;
 435                 }
 436                 if (c != -1)
 437                         isl_stream_ungetc(s, c);
 438                 tok->type = (enum isl_token_type) ':';
 439                 return tok;
 440         }
 441         if (c == '>') {
 442                 int c;
 443                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 444                 if (!tok)
 445                         return NULL;
 446                 if ((c = isl_stream_getc(s)) == '=') {
 447                         tok->u.s = strdup(">=");
 448                         tok->type = ISL_TOKEN_GE;
 449                         return tok;
 450                 } else if (c == '>') {
 451                         if ((c = isl_stream_getc(s)) == '=') {
 452                                 tok->u.s = strdup(">>=");
 453                                 tok->type = ISL_TOKEN_LEX_GE;
 454                                 return tok;
 455                         }
 456                         tok->u.s = strdup(">>");
 457                         tok->type = ISL_TOKEN_LEX_GT;
 458                 } else {
 459                         tok->u.s = strdup(">");
 460                         tok->type = ISL_TOKEN_GT;
 461                 }
 462                 if (c != -1)
 463                         isl_stream_ungetc(s, c);
 464                 return tok;
 465         }
 466         if (c == '<') {
 467                 int c;
 468                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 469                 if (!tok)
 470                         return NULL;
 471                 if ((c = isl_stream_getc(s)) == '=') {
 472                         tok->u.s = strdup("<=");
 473                         tok->type = ISL_TOKEN_LE;
 474                         return tok;
 475                 } else if (c == '<') {
 476                         if ((c = isl_stream_getc(s)) == '=') {
 477                                 tok->u.s = strdup("<<=");
 478                                 tok->type = ISL_TOKEN_LEX_LE;
 479                                 return tok;
 480                         }
 481                         tok->u.s = strdup("<<");
 482                         tok->type = ISL_TOKEN_LEX_LT;
 483                 } else {
 484                         tok->u.s = strdup("<");
 485                         tok->type = ISL_TOKEN_LT;
 486                 }
 487                 if (c != -1)
 488                         isl_stream_ungetc(s, c);
 489                 return tok;
 490         }
 491         if (c == '&') {
 492                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 493                 if (!tok)
 494                         return NULL;
 495                 tok->type = ISL_TOKEN_AND;
 496                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 497                         tok->u.s = strdup("&");
 498                         isl_stream_ungetc(s, c);
 499                 } else
 500                         tok->u.s = strdup("&&");
 501                 return tok;
 502         }
 503         if (c == '|') {
 504                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 505                 if (!tok)
 506                         return NULL;
 507                 tok->type = ISL_TOKEN_OR;
 508                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 509                         tok->u.s = strdup("|");
 510                         isl_stream_ungetc(s, c);
 511                 } else
 512                         tok->u.s = strdup("||");
 513                 return tok;
 514         }
 515         if (c == '/') {
 516                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 517                 if (!tok)
 518                         return NULL;
 519                 if ((c = isl_stream_getc(s)) != '\\' && c != -1) {
 520                         tok->type = (enum isl_token_type) '/';
 521                         isl_stream_ungetc(s, c);
 522                 } else {
 523                         tok->u.s = strdup("/\\");
 524                         tok->type = ISL_TOKEN_AND;
 525                 }
 526                 return tok;
 527         }
 528         if (c == '\\') {
 529                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 530                 if (!tok)
 531                         return NULL;
 532                 if ((c = isl_stream_getc(s)) != '/' && c != -1) {
 533                         tok->type = (enum isl_token_type) '\\';
 534                         isl_stream_ungetc(s, c);
 535                 } else {
 536                         tok->u.s = strdup("\\/");
 537                         tok->type = ISL_TOKEN_OR;
 538                 }
 539                 return tok;
 540         }
 541         if (c == '!') {
 542                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 543                 if (!tok)
 544                         return NULL;
 545                 tok->type = ISL_TOKEN_NOT;
 546                 tok->u.s = strdup("!");
 547                 return tok;
 548         }
 549
 550         tok = isl_token_new(s->ctx, line, col, old_line != line);
 551         if (!tok)
 552                 return NULL;
 553         tok->type = ISL_TOKEN_UNKNOWN;
 554         return tok;
 555 error:
 556         isl_token_free(tok);
 557         return NULL;
 558 }
 559
 560 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 561 {
 562         return next_token(s, 0);
 563 }
 564
 565 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 566 {
 567         return next_token(s, 1);
 568 }
 569
 570 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 571 {
 572         struct isl_token *tok;
 573
 574         tok = isl_stream_next_token(s);
 575         if (!tok)
 576                 return 0;
 577         if (tok->type == type) {
 578                 isl_token_free(tok);
 579                 return 1;
 580         }
 581         isl_stream_push_token(s, tok);
 582         return 0;
 583 }
 584
 585 int isl_stream_next_token_is(struct isl_stream *s, int type)
 586 {
 587         struct isl_token *tok;
 588         int r;
 589
 590         tok = isl_stream_next_token(s);
 591         if (!tok)
 592                 return 0;
 593         r = tok->type == type;
 594         isl_stream_push_token(s, tok);
 595         return r;
 596 }
 597
 598 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 599 {
 600         struct isl_token *tok;
 601
 602         tok = isl_stream_next_token(s);
 603         if (!tok)
 604                 return NULL;
 605         if (tok->type == ISL_TOKEN_IDENT) {
 606                 char *ident = strdup(tok->u.s);
 607                 isl_token_free(tok);
 608                 return ident;
 609         }
 610         isl_stream_push_token(s, tok);
 611         return NULL;
 612 }
 613
 614 int isl_stream_eat(struct isl_stream *s, int type)
 615 {
 616         struct isl_token *tok;
 617
 618         tok = isl_stream_next_token(s);
 619         if (!tok)
 620                 return -1;
 621         if (tok->type == type) {
 622                 isl_token_free(tok);
 623                 return 0;
 624         }
 625         isl_stream_error(s, tok, "expecting other token");
 626         isl_stream_push_token(s, tok);
 627         return -1;
 628 }
 629
 630 int isl_stream_is_empty(struct isl_stream *s)
 631 {
 632         struct isl_token *tok;
 633
 634         tok = isl_stream_next_token(s);
 635
 636         if (!tok)
 637                 return 1;
 638
 639         isl_stream_push_token(s, tok);
 640         return 0;
 641 }
 642
 643 static int free_keyword(void **p, void *user)
 644 {
 645         struct isl_keyword *keyword = *p;
 646
 647         free(keyword->name);
 648         free(keyword);
 649
 650         return 0;
 651 }
 652
 653 void isl_stream_flush_tokens(struct isl_stream *s)
 654 {
 655         int i;
 656
 657         if (!s)
 658                 return;
 659         for (i = 0; i < s->n_token; ++i)
 660                 isl_token_free(s->tokens[i]);
 661         s->n_token = 0;
 662 }
 663
 664 void isl_stream_free(struct isl_stream *s)
 665 {
 666         if (!s)
 667                 return;
 668         free(s->buffer);
 669         if (s->n_token != 0) {
 670                 struct isl_token *tok = isl_stream_next_token(s);
 671                 isl_stream_error(s, tok, "unexpected token");
 672                 isl_token_free(tok);
 673         }
 674         if (s->keywords) {
 675                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 676                 isl_hash_table_free(s->ctx, s->keywords);
 677         }
 678         isl_ctx_deref(s->ctx);
 679         free(s);
 680 }