isl_stream.c

   1 /*
   2  * Copyright 2008-2009 Katholieke Universiteit Leuven
   3  *
   4  * Use of this software is governed by the GNU LGPLv2.1 license
   5  *
   6  * Written by Sven Verdoolaege, K.U.Leuven, Departement
   7  * Computerwetenschappen, Celestijnenlaan 200A, B-3001 Leuven, Belgium
   8  */
   9
  10 #include <ctype.h>
  11 #include <string.h>
  12 #include <strings.h>
  13 #include <isl/ctx.h>
  14 #include <isl/stream.h>
  15
  16 struct isl_keyword {
  17         char                    *name;
  18         enum isl_token_type     type;
  19 };
  20
  21 static int same_name(const void *entry, const void *val)
  22 {
  23         const struct isl_keyword *keyword = (const struct isl_keyword *)entry;
  24
  25         return !strcmp(keyword->name, val);
  26 }
  27
  28 enum isl_token_type isl_stream_register_keyword(struct isl_stream *s,
  29         const char *name)
  30 {
  31         struct isl_hash_table_entry *entry;
  32         struct isl_keyword *keyword;
  33         uint32_t name_hash;
  34
  35         if (!s->keywords) {
  36                 s->keywords = isl_hash_table_alloc(s->ctx, 10);
  37                 if (!s->keywords)
  38                         return ISL_TOKEN_ERROR;
  39                 s->next_type = ISL_TOKEN_LAST;
  40         }
  41
  42         name_hash = isl_hash_string(isl_hash_init(), name);
  43
  44         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash,
  45                                         same_name, name, 1);
  46         if (!entry)
  47                 return ISL_TOKEN_ERROR;
  48         if (entry->data) {
  49                 keyword = entry->data;
  50                 return keyword->type;
  51         }
  52
  53         keyword = isl_calloc_type(s->ctx, struct isl_keyword);
  54         if (!keyword)
  55                 return ISL_TOKEN_ERROR;
  56         keyword->type = s->next_type++;
  57         keyword->name = strdup(name);
  58         if (!keyword->name) {
  59                 free(keyword);
  60                 return ISL_TOKEN_ERROR;
  61         }
  62         entry->data = keyword;
  63
  64         return keyword->type;
  65 }
  66
  67 static struct isl_token *isl_token_new(struct isl_ctx *ctx,
  68         int line, int col, unsigned on_new_line)
  69 {
  70         struct isl_token *tok = isl_alloc_type(ctx, struct isl_token);
  71         if (!tok)
  72                 return NULL;
  73         tok->line = line;
  74         tok->col = col;
  75         tok->on_new_line = on_new_line;
  76         tok->is_keyword = 0;
  77         tok->u.s = NULL;
  78         return tok;
  79 }
  80
  81 void isl_token_free(struct isl_token *tok)
  82 {
  83         if (!tok)
  84                 return;
  85         if (tok->type == ISL_TOKEN_VALUE)
  86                 isl_int_clear(tok->u.v);
  87         else
  88                 free(tok->u.s);
  89         free(tok);
  90 }
  91
  92 void isl_stream_error(struct isl_stream *s, struct isl_token *tok, char *msg)
  93 {
  94         int line = tok ? tok->line : s->line;
  95         int col = tok ? tok->col : s->col;
  96         fprintf(stderr, "syntax error (%d, %d): %s\n", line, col, msg);
  97         if (tok) {
  98                 if (tok->type < 256)
  99                         fprintf(stderr, "got '%c'\n", tok->type);
 100                 else if (tok->type == ISL_TOKEN_IDENT)
 101                         fprintf(stderr, "got ident '%s'\n", tok->u.s);
 102                 else if (tok->is_keyword)
 103                         fprintf(stderr, "got keyword '%s'\n", tok->u.s);
 104                 else if (tok->type == ISL_TOKEN_VALUE) {
 105                         fprintf(stderr, "got value '");
 106                         isl_int_print(stderr, tok->u.v, 0);
 107                         fprintf(stderr, "'\n");
 108                 } else if (tok->u.s)
 109                         fprintf(stderr, "got token '%s'\n", tok->u.s);
 110                 else
 111                         fprintf(stderr, "got token type %d\n", tok->type);
 112         }
 113 }
 114
 115 static struct isl_stream* isl_stream_new(struct isl_ctx *ctx)
 116 {
 117         int i;
 118         struct isl_stream *s = isl_alloc_type(ctx, struct isl_stream);
 119         if (!s)
 120                 return NULL;
 121         s->ctx = ctx;
 122         isl_ctx_ref(s->ctx);
 123         s->file = NULL;
 124         s->str = NULL;
 125         s->len = 0;
 126         s->line = 1;
 127         s->col = 0;
 128         s->eof = 0;
 129         s->c = -1;
 130         for (i = 0; i < 5; ++i)
 131                 s->tokens[i] = NULL;
 132         s->n_token = 0;
 133         s->keywords = NULL;
 134         s->size = 256;
 135         s->buffer = isl_alloc_array(ctx, char, s->size);
 136         if (!s->buffer)
 137                 goto error;
 138         return s;
 139 error:
 140         isl_stream_free(s);
 141         return NULL;
 142 }
 143
 144 struct isl_stream* isl_stream_new_file(struct isl_ctx *ctx, FILE *file)
 145 {
 146         struct isl_stream *s = isl_stream_new(ctx);
 147         if (!s)
 148                 return NULL;
 149         s->file = file;
 150         return s;
 151 }
 152
 153 struct isl_stream* isl_stream_new_str(struct isl_ctx *ctx, const char *str)
 154 {
 155         struct isl_stream *s = isl_stream_new(ctx);
 156         if (!s)
 157                 return NULL;
 158         s->str = str;
 159         return s;
 160 }
 161
 162 static int isl_stream_getc(struct isl_stream *s)
 163 {
 164         int c;
 165         if (s->eof)
 166                 return -1;
 167         if (s->file)
 168                 c = fgetc(s->file);
 169         else {
 170                 c = *s->str++;
 171                 if (c == '\0')
 172                         c = -1;
 173         }
 174         if (c == -1)
 175                 s->eof = 1;
 176         if (!s->eof) {
 177                 if (s->c == '\n') {
 178                         s->line++;
 179                         s->col = 0;
 180                 } else
 181                         s->col++;
 182         }
 183         s->c = c;
 184         return c;
 185 }
 186
 187 static void isl_stream_ungetc(struct isl_stream *s, int c)
 188 {
 189         if (s->file)
 190                 ungetc(c, s->file);
 191         else
 192                 --s->str;
 193         s->c = -1;
 194 }
 195
 196 static int isl_stream_push_char(struct isl_stream *s, int c)
 197 {
 198         if (s->len >= s->size) {
 199                 s->size = (3*s->size)/2;
 200                 s->buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
 201                 if (!s->buffer)
 202                         return -1;
 203         }
 204         s->buffer[s->len++] = c;
 205         return 0;
 206 }
 207
 208 void isl_stream_push_token(struct isl_stream *s, struct isl_token *tok)
 209 {
 210         isl_assert(s->ctx, s->n_token < 5, return);
 211         s->tokens[s->n_token++] = tok;
 212 }
 213
 214 static enum isl_token_type check_keywords(struct isl_stream *s)
 215 {
 216         struct isl_hash_table_entry *entry;
 217         struct isl_keyword *keyword;
 218         uint32_t name_hash;
 219
 220         if (!strcasecmp(s->buffer, "exists"))
 221                 return ISL_TOKEN_EXISTS;
 222         if (!strcasecmp(s->buffer, "and"))
 223                 return ISL_TOKEN_AND;
 224         if (!strcasecmp(s->buffer, "or"))
 225                 return ISL_TOKEN_OR;
 226         if (!strcasecmp(s->buffer, "not"))
 227                 return ISL_TOKEN_NOT;
 228         if (!strcasecmp(s->buffer, "infty"))
 229                 return ISL_TOKEN_INFTY;
 230         if (!strcasecmp(s->buffer, "infinity"))
 231                 return ISL_TOKEN_INFTY;
 232         if (!strcasecmp(s->buffer, "NaN"))
 233                 return ISL_TOKEN_NAN;
 234         if (!strcasecmp(s->buffer, "max"))
 235                 return ISL_TOKEN_MAX;
 236         if (!strcasecmp(s->buffer, "rat"))
 237                 return ISL_TOKEN_RAT;
 238         if (!strcasecmp(s->buffer, "true"))
 239                 return ISL_TOKEN_TRUE;
 240         if (!strcasecmp(s->buffer, "false"))
 241                 return ISL_TOKEN_FALSE;
 242
 243         if (!s->keywords)
 244                 return ISL_TOKEN_IDENT;
 245
 246         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 247         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 248                                         s->buffer, 0);
 249         if (entry) {
 250                 keyword = entry->data;
 251                 return keyword->type;
 252         }
 253
 254         return ISL_TOKEN_IDENT;
 255 }
 256
 257 int isl_stream_skip_line(struct isl_stream *s)
 258 {
 259         int c;
 260
 261         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 262                 /* nothing */
 263                 ;
 264
 265         return c == -1 ? -1 : 0;
 266 }
 267
 268 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 269 {
 270         int c;
 271         struct isl_token *tok = NULL;
 272         int line, col;
 273         int old_line = s->line;
 274
 275         if (s->n_token) {
 276                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 277                         return NULL;
 278                 return s->tokens[--s->n_token];
 279         }
 280
 281         if (same_line && s->c == '\n')
 282                 return NULL;
 283
 284         s->len = 0;
 285
 286         /* skip spaces and comment lines */
 287         while ((c = isl_stream_getc(s)) != -1) {
 288                 if (c == '#') {
 289                         if (isl_stream_skip_line(s) < 0)
 290                                 break;
 291                         c = '\n';
 292                         if (same_line)
 293                                 break;
 294                 } else if (!isspace(c) || (same_line && c == '\n'))
 295                         break;
 296         }
 297
 298         line = s->line;
 299         col = s->col;
 300
 301         if (c == -1 || (same_line && c == '\n'))
 302                 return NULL;
 303         if (c == '(' ||
 304             c == ')' ||
 305             c == '+' ||
 306             c == '/' ||
 307             c == '*' ||
 308             c == '%' ||
 309             c == '^' ||
 310             c == '=' ||
 311             c == '@' ||
 312             c == ',' ||
 313             c == '.' ||
 314             c == ';' ||
 315             c == '[' ||
 316             c == ']' ||
 317             c == '{' ||
 318             c == '}') {
 319                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 320                 if (!tok)
 321                         return NULL;
 322                 tok->type = (enum isl_token_type)c;
 323                 return tok;
 324         }
 325         if (c == '-') {
 326                 int c;
 327                 if ((c = isl_stream_getc(s)) == '>') {
 328                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 329                         if (!tok)
 330                                 return NULL;
 331                         tok->u.s = strdup("->");
 332                         tok->type = ISL_TOKEN_TO;
 333                         return tok;
 334                 }
 335                 if (c != -1)
 336                         isl_stream_ungetc(s, c);
 337                 if (!isdigit(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) '-';
 342                         return tok;
 343                 }
 344         }
 345         if (c == '-' || isdigit(c)) {
 346                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 347                 if (!tok)
 348                         return NULL;
 349                 tok->type = ISL_TOKEN_VALUE;
 350                 isl_int_init(tok->u.v);
 351                 if (isl_stream_push_char(s, c))
 352                         goto error;
 353                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 354                         if (isl_stream_push_char(s, c))
 355                                 goto error;
 356                 if (c != -1)
 357                         isl_stream_ungetc(s, c);
 358                 isl_stream_push_char(s, '\0');
 359                 isl_int_read(tok->u.v, s->buffer);
 360                 return tok;
 361         }
 362         if (isalpha(c) || c == '_') {
 363                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 364                 if (!tok)
 365                         return NULL;
 366                 isl_stream_push_char(s, c);
 367                 while ((c = isl_stream_getc(s)) != -1 &&
 368                                 (isalnum(c) || c == '_'))
 369                         isl_stream_push_char(s, c);
 370                 if (c != -1)
 371                         isl_stream_ungetc(s, c);
 372                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 373                         isl_stream_push_char(s, c);
 374                 if (c != -1)
 375                         isl_stream_ungetc(s, c);
 376                 isl_stream_push_char(s, '\0');
 377                 tok->type = check_keywords(s);
 378                 if (tok->type != ISL_TOKEN_IDENT)
 379                         tok->is_keyword = 1;
 380                 tok->u.s = strdup(s->buffer);
 381                 if (!tok->u.s)
 382                         goto error;
 383                 return tok;
 384         }
 385         if (c == '"') {
 386                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 387                 if (!tok)
 388                         return NULL;
 389                 tok->type = ISL_TOKEN_STRING;
 390                 tok->u.s = NULL;
 391                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 392                         isl_stream_push_char(s, c);
 393                 if (c != '"') {
 394                         isl_stream_error(s, NULL, "unterminated string");
 395                         goto error;
 396                 }
 397                 isl_stream_push_char(s, '\0');
 398                 tok->u.s = strdup(s->buffer);
 399                 return tok;
 400         }
 401         if (c == ':') {
 402                 int c;
 403                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 404                 if (!tok)
 405                         return NULL;
 406                 if ((c = isl_stream_getc(s)) == '=') {
 407                         tok->u.s = strdup(":=");
 408                         tok->type = ISL_TOKEN_DEF;
 409                         return tok;
 410                 }
 411                 if (c != -1)
 412                         isl_stream_ungetc(s, c);
 413                 tok->type = (enum isl_token_type) ':';
 414                 return tok;
 415         }
 416         if (c == '>') {
 417                 int c;
 418                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 419                 if (!tok)
 420                         return NULL;
 421                 if ((c = isl_stream_getc(s)) == '=') {
 422                         tok->u.s = strdup(">=");
 423                         tok->type = ISL_TOKEN_GE;
 424                         return tok;
 425                 } else if (c == '>') {
 426                         if ((c = isl_stream_getc(s)) == '=') {
 427                                 tok->u.s = strdup(">>=");
 428                                 tok->type = ISL_TOKEN_LEX_GE;
 429                                 return tok;
 430                         }
 431                         tok->u.s = strdup(">>");
 432                         tok->type = ISL_TOKEN_LEX_GT;
 433                 } else {
 434                         tok->u.s = strdup(">");
 435                         tok->type = ISL_TOKEN_GT;
 436                 }
 437                 if (c != -1)
 438                         isl_stream_ungetc(s, c);
 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_LE;
 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_LE;
 454                                 return tok;
 455                         }
 456                         tok->u.s = strdup("<<");
 457                         tok->type = ISL_TOKEN_LEX_LT;
 458                 } else {
 459                         tok->u.s = strdup("<");
 460                         tok->type = ISL_TOKEN_LT;
 461                 }
 462                 if (c != -1)
 463                         isl_stream_ungetc(s, c);
 464                 return tok;
 465         }
 466         if (c == '&') {
 467                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 468                 if (!tok)
 469                         return NULL;
 470                 tok->type = ISL_TOKEN_AND;
 471                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 472                         tok->u.s = strdup("&");
 473                         isl_stream_ungetc(s, c);
 474                 } else
 475                         tok->u.s = strdup("&&");
 476                 return tok;
 477         }
 478         if (c == '|') {
 479                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 480                 if (!tok)
 481                         return NULL;
 482                 tok->type = ISL_TOKEN_OR;
 483                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 484                         tok->u.s = strdup("|");
 485                         isl_stream_ungetc(s, c);
 486                 } else
 487                         tok->u.s = strdup("||");
 488                 return tok;
 489         }
 490         if (c == '!') {
 491                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 492                 if (!tok)
 493                         return NULL;
 494                 tok->type = ISL_TOKEN_NOT;
 495                 tok->u.s = strdup("!");
 496                 return tok;
 497         }
 498
 499         tok = isl_token_new(s->ctx, line, col, old_line != line);
 500         if (!tok)
 501                 return NULL;
 502         tok->type = ISL_TOKEN_UNKNOWN;
 503         return tok;
 504 error:
 505         isl_token_free(tok);
 506         return NULL;
 507 }
 508
 509 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 510 {
 511         return next_token(s, 0);
 512 }
 513
 514 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 515 {
 516         return next_token(s, 1);
 517 }
 518
 519 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 520 {
 521         struct isl_token *tok;
 522
 523         tok = isl_stream_next_token(s);
 524         if (!tok)
 525                 return 0;
 526         if (tok->type == type) {
 527                 isl_token_free(tok);
 528                 return 1;
 529         }
 530         isl_stream_push_token(s, tok);
 531         return 0;
 532 }
 533
 534 int isl_stream_next_token_is(struct isl_stream *s, int type)
 535 {
 536         struct isl_token *tok;
 537         int r;
 538
 539         tok = isl_stream_next_token(s);
 540         if (!tok)
 541                 return 0;
 542         r = tok->type == type;
 543         isl_stream_push_token(s, tok);
 544         return r;
 545 }
 546
 547 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 548 {
 549         struct isl_token *tok;
 550
 551         tok = isl_stream_next_token(s);
 552         if (!tok)
 553                 return NULL;
 554         if (tok->type == ISL_TOKEN_IDENT) {
 555                 char *ident = strdup(tok->u.s);
 556                 isl_token_free(tok);
 557                 return ident;
 558         }
 559         isl_stream_push_token(s, tok);
 560         return NULL;
 561 }
 562
 563 int isl_stream_eat(struct isl_stream *s, int type)
 564 {
 565         struct isl_token *tok;
 566
 567         tok = isl_stream_next_token(s);
 568         if (!tok)
 569                 return -1;
 570         if (tok->type == type) {
 571                 isl_token_free(tok);
 572                 return 0;
 573         }
 574         isl_stream_error(s, tok, "expecting other token");
 575         isl_stream_push_token(s, tok);
 576         return -1;
 577 }
 578
 579 int isl_stream_is_empty(struct isl_stream *s)
 580 {
 581         struct isl_token *tok;
 582
 583         tok = isl_stream_next_token(s);
 584
 585         if (!tok)
 586                 return 1;
 587
 588         isl_stream_push_token(s, tok);
 589         return 0;
 590 }
 591
 592 static int free_keyword(void **p, void *user)
 593 {
 594         struct isl_keyword *keyword = *p;
 595
 596         free(keyword->name);
 597         free(keyword);
 598
 599         return 0;
 600 }
 601
 602 void isl_stream_flush_tokens(struct isl_stream *s)
 603 {
 604         int i;
 605
 606         if (!s)
 607                 return;
 608         for (i = 0; i < s->n_token; ++i)
 609                 isl_token_free(s->tokens[i]);
 610         s->n_token = 0;
 611 }
 612
 613 void isl_stream_free(struct isl_stream *s)
 614 {
 615         if (!s)
 616                 return;
 617         free(s->buffer);
 618         if (s->n_token != 0) {
 619                 struct isl_token *tok = isl_stream_next_token(s);
 620                 isl_stream_error(s, tok, "unexpected token");
 621                 isl_token_free(tok);
 622         }
 623         if (s->keywords) {
 624                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 625                 isl_hash_table_free(s->ctx, s->keywords);
 626         }
 627         isl_ctx_deref(s->ctx);
 628         free(s);
 629 }