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, "infty"))
 227                 return ISL_TOKEN_INFTY;
 228         if (!strcasecmp(s->buffer, "infinity"))
 229                 return ISL_TOKEN_INFTY;
 230         if (!strcasecmp(s->buffer, "NaN"))
 231                 return ISL_TOKEN_NAN;
 232         if (!strcasecmp(s->buffer, "max"))
 233                 return ISL_TOKEN_MAX;
 234         if (!strcasecmp(s->buffer, "rat"))
 235                 return ISL_TOKEN_RAT;
 236
 237         if (!s->keywords)
 238                 return ISL_TOKEN_IDENT;
 239
 240         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 241         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 242                                         s->buffer, 0);
 243         if (entry) {
 244                 keyword = entry->data;
 245                 return keyword->type;
 246         }
 247
 248         return ISL_TOKEN_IDENT;
 249 }
 250
 251 int isl_stream_skip_line(struct isl_stream *s)
 252 {
 253         int c;
 254
 255         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 256                 /* nothing */
 257                 ;
 258
 259         return c == -1 ? -1 : 0;
 260 }
 261
 262 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 263 {
 264         int c;
 265         struct isl_token *tok = NULL;
 266         int line, col;
 267         int old_line = s->line;
 268
 269         if (s->n_token) {
 270                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 271                         return NULL;
 272                 return s->tokens[--s->n_token];
 273         }
 274
 275         if (same_line && s->c == '\n')
 276                 return NULL;
 277
 278         s->len = 0;
 279
 280         /* skip spaces and comment lines */
 281         while ((c = isl_stream_getc(s)) != -1) {
 282                 if (c == '#') {
 283                         if (isl_stream_skip_line(s) < 0)
 284                                 break;
 285                         c = '\n';
 286                         if (same_line)
 287                                 break;
 288                 } else if (!isspace(c) || (same_line && c == '\n'))
 289                         break;
 290         }
 291
 292         line = s->line;
 293         col = s->col;
 294
 295         if (c == -1 || (same_line && c == '\n'))
 296                 return NULL;
 297         if (c == '(' ||
 298             c == ')' ||
 299             c == '+' ||
 300             c == '/' ||
 301             c == '*' ||
 302             c == '%' ||
 303             c == '^' ||
 304             c == '=' ||
 305             c == '@' ||
 306             c == ',' ||
 307             c == '.' ||
 308             c == ';' ||
 309             c == '[' ||
 310             c == ']' ||
 311             c == '{' ||
 312             c == '}') {
 313                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 314                 if (!tok)
 315                         return NULL;
 316                 tok->type = (enum isl_token_type)c;
 317                 return tok;
 318         }
 319         if (c == '-') {
 320                 int c;
 321                 if ((c = isl_stream_getc(s)) == '>') {
 322                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 323                         if (!tok)
 324                                 return NULL;
 325                         tok->u.s = strdup("->");
 326                         tok->type = ISL_TOKEN_TO;
 327                         return tok;
 328                 }
 329                 if (c != -1)
 330                         isl_stream_ungetc(s, c);
 331                 if (!isdigit(c)) {
 332                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 333                         if (!tok)
 334                                 return NULL;
 335                         tok->type = (enum isl_token_type) '-';
 336                         return tok;
 337                 }
 338         }
 339         if (c == '-' || isdigit(c)) {
 340                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 341                 if (!tok)
 342                         return NULL;
 343                 tok->type = ISL_TOKEN_VALUE;
 344                 isl_int_init(tok->u.v);
 345                 if (isl_stream_push_char(s, c))
 346                         goto error;
 347                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 348                         if (isl_stream_push_char(s, c))
 349                                 goto error;
 350                 if (c != -1)
 351                         isl_stream_ungetc(s, c);
 352                 isl_stream_push_char(s, '\0');
 353                 isl_int_read(tok->u.v, s->buffer);
 354                 return tok;
 355         }
 356         if (isalpha(c) || c == '_') {
 357                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 358                 if (!tok)
 359                         return NULL;
 360                 isl_stream_push_char(s, c);
 361                 while ((c = isl_stream_getc(s)) != -1 &&
 362                                 (isalnum(c) || c == '_'))
 363                         isl_stream_push_char(s, c);
 364                 if (c != -1)
 365                         isl_stream_ungetc(s, c);
 366                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 367                         isl_stream_push_char(s, c);
 368                 if (c != -1)
 369                         isl_stream_ungetc(s, c);
 370                 isl_stream_push_char(s, '\0');
 371                 tok->type = check_keywords(s);
 372                 if (tok->type != ISL_TOKEN_IDENT)
 373                         tok->is_keyword = 1;
 374                 tok->u.s = strdup(s->buffer);
 375                 if (!tok->u.s)
 376                         goto error;
 377                 return tok;
 378         }
 379         if (c == '"') {
 380                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 381                 if (!tok)
 382                         return NULL;
 383                 tok->type = ISL_TOKEN_STRING;
 384                 tok->u.s = NULL;
 385                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 386                         isl_stream_push_char(s, c);
 387                 if (c != '"') {
 388                         isl_stream_error(s, NULL, "unterminated string");
 389                         goto error;
 390                 }
 391                 isl_stream_push_char(s, '\0');
 392                 tok->u.s = strdup(s->buffer);
 393                 return tok;
 394         }
 395         if (c == ':') {
 396                 int c;
 397                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 398                 if (!tok)
 399                         return NULL;
 400                 if ((c = isl_stream_getc(s)) == '=') {
 401                         tok->u.s = strdup(":=");
 402                         tok->type = ISL_TOKEN_DEF;
 403                         return tok;
 404                 }
 405                 if (c != -1)
 406                         isl_stream_ungetc(s, c);
 407                 tok->type = (enum isl_token_type) ':';
 408                 return tok;
 409         }
 410         if (c == '>') {
 411                 int c;
 412                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 413                 if (!tok)
 414                         return NULL;
 415                 if ((c = isl_stream_getc(s)) == '=') {
 416                         tok->u.s = strdup(">=");
 417                         tok->type = ISL_TOKEN_GE;
 418                         return tok;
 419                 } else if (c == '>') {
 420                         if ((c = isl_stream_getc(s)) == '=') {
 421                                 tok->u.s = strdup(">>=");
 422                                 tok->type = ISL_TOKEN_LEX_GE;
 423                                 return tok;
 424                         }
 425                         tok->u.s = strdup(">>");
 426                         tok->type = ISL_TOKEN_LEX_GT;
 427                 } else {
 428                         tok->u.s = strdup(">");
 429                         tok->type = ISL_TOKEN_GT;
 430                 }
 431                 if (c != -1)
 432                         isl_stream_ungetc(s, c);
 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_LE;
 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_LE;
 448                                 return tok;
 449                         }
 450                         tok->u.s = strdup("<<");
 451                         tok->type = ISL_TOKEN_LEX_LT;
 452                 } else {
 453                         tok->u.s = strdup("<");
 454                         tok->type = ISL_TOKEN_LT;
 455                 }
 456                 if (c != -1)
 457                         isl_stream_ungetc(s, c);
 458                 return tok;
 459         }
 460         if (c == '&') {
 461                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 462                 if (!tok)
 463                         return NULL;
 464                 tok->type = ISL_TOKEN_AND;
 465                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 466                         tok->u.s = strdup("&");
 467                         isl_stream_ungetc(s, c);
 468                 } else
 469                         tok->u.s = strdup("&&");
 470                 return tok;
 471         }
 472         if (c == '|') {
 473                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 474                 if (!tok)
 475                         return NULL;
 476                 tok->type = ISL_TOKEN_OR;
 477                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 478                         tok->u.s = strdup("|");
 479                         isl_stream_ungetc(s, c);
 480                 } else
 481                         tok->u.s = strdup("||");
 482                 return tok;
 483         }
 484
 485         tok = isl_token_new(s->ctx, line, col, old_line != line);
 486         if (!tok)
 487                 return NULL;
 488         tok->type = ISL_TOKEN_UNKNOWN;
 489         return tok;
 490 error:
 491         isl_token_free(tok);
 492         return NULL;
 493 }
 494
 495 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 496 {
 497         return next_token(s, 0);
 498 }
 499
 500 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 501 {
 502         return next_token(s, 1);
 503 }
 504
 505 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 506 {
 507         struct isl_token *tok;
 508
 509         tok = isl_stream_next_token(s);
 510         if (!tok)
 511                 return 0;
 512         if (tok->type == type) {
 513                 isl_token_free(tok);
 514                 return 1;
 515         }
 516         isl_stream_push_token(s, tok);
 517         return 0;
 518 }
 519
 520 int isl_stream_next_token_is(struct isl_stream *s, int type)
 521 {
 522         struct isl_token *tok;
 523         int r;
 524
 525         tok = isl_stream_next_token(s);
 526         if (!tok)
 527                 return 0;
 528         r = tok->type == type;
 529         isl_stream_push_token(s, tok);
 530         return r;
 531 }
 532
 533 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 534 {
 535         struct isl_token *tok;
 536
 537         tok = isl_stream_next_token(s);
 538         if (!tok)
 539                 return NULL;
 540         if (tok->type == ISL_TOKEN_IDENT) {
 541                 char *ident = strdup(tok->u.s);
 542                 isl_token_free(tok);
 543                 return ident;
 544         }
 545         isl_stream_push_token(s, tok);
 546         return NULL;
 547 }
 548
 549 int isl_stream_eat(struct isl_stream *s, int type)
 550 {
 551         struct isl_token *tok;
 552
 553         tok = isl_stream_next_token(s);
 554         if (!tok)
 555                 return -1;
 556         if (tok->type == type) {
 557                 isl_token_free(tok);
 558                 return 0;
 559         }
 560         isl_stream_error(s, tok, "expecting other token");
 561         isl_stream_push_token(s, tok);
 562         return -1;
 563 }
 564
 565 int isl_stream_is_empty(struct isl_stream *s)
 566 {
 567         struct isl_token *tok;
 568
 569         tok = isl_stream_next_token(s);
 570
 571         if (!tok)
 572                 return 1;
 573
 574         isl_stream_push_token(s, tok);
 575         return 0;
 576 }
 577
 578 static int free_keyword(void **p, void *user)
 579 {
 580         struct isl_keyword *keyword = *p;
 581
 582         free(keyword->name);
 583         free(keyword);
 584
 585         return 0;
 586 }
 587
 588 void isl_stream_flush_tokens(struct isl_stream *s)
 589 {
 590         int i;
 591
 592         if (!s)
 593                 return;
 594         for (i = 0; i < s->n_token; ++i)
 595                 isl_token_free(s->tokens[i]);
 596         s->n_token = 0;
 597 }
 598
 599 void isl_stream_free(struct isl_stream *s)
 600 {
 601         if (!s)
 602                 return;
 603         free(s->buffer);
 604         if (s->n_token != 0) {
 605                 struct isl_token *tok = isl_stream_next_token(s);
 606                 isl_stream_error(s, tok, "unexpected token");
 607                 isl_token_free(tok);
 608         }
 609         if (s->keywords) {
 610                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 611                 isl_hash_table_free(s->ctx, s->keywords);
 612         }
 613         isl_ctx_deref(s->ctx);
 614         free(s);
 615 }