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
 239         if (!s->keywords)
 240                 return ISL_TOKEN_IDENT;
 241
 242         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 243         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 244                                         s->buffer, 0);
 245         if (entry) {
 246                 keyword = entry->data;
 247                 return keyword->type;
 248         }
 249
 250         return ISL_TOKEN_IDENT;
 251 }
 252
 253 int isl_stream_skip_line(struct isl_stream *s)
 254 {
 255         int c;
 256
 257         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 258                 /* nothing */
 259                 ;
 260
 261         return c == -1 ? -1 : 0;
 262 }
 263
 264 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 265 {
 266         int c;
 267         struct isl_token *tok = NULL;
 268         int line, col;
 269         int old_line = s->line;
 270
 271         if (s->n_token) {
 272                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 273                         return NULL;
 274                 return s->tokens[--s->n_token];
 275         }
 276
 277         if (same_line && s->c == '\n')
 278                 return NULL;
 279
 280         s->len = 0;
 281
 282         /* skip spaces and comment lines */
 283         while ((c = isl_stream_getc(s)) != -1) {
 284                 if (c == '#') {
 285                         if (isl_stream_skip_line(s) < 0)
 286                                 break;
 287                         c = '\n';
 288                         if (same_line)
 289                                 break;
 290                 } else if (!isspace(c) || (same_line && c == '\n'))
 291                         break;
 292         }
 293
 294         line = s->line;
 295         col = s->col;
 296
 297         if (c == -1 || (same_line && c == '\n'))
 298                 return NULL;
 299         if (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             c == '{' ||
 314             c == '}') {
 315                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 316                 if (!tok)
 317                         return NULL;
 318                 tok->type = (enum isl_token_type)c;
 319                 return tok;
 320         }
 321         if (c == '-') {
 322                 int c;
 323                 if ((c = isl_stream_getc(s)) == '>') {
 324                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 325                         if (!tok)
 326                                 return NULL;
 327                         tok->u.s = strdup("->");
 328                         tok->type = ISL_TOKEN_TO;
 329                         return tok;
 330                 }
 331                 if (c != -1)
 332                         isl_stream_ungetc(s, c);
 333                 if (!isdigit(c)) {
 334                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 335                         if (!tok)
 336                                 return NULL;
 337                         tok->type = (enum isl_token_type) '-';
 338                         return tok;
 339                 }
 340         }
 341         if (c == '-' || isdigit(c)) {
 342                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 343                 if (!tok)
 344                         return NULL;
 345                 tok->type = ISL_TOKEN_VALUE;
 346                 isl_int_init(tok->u.v);
 347                 if (isl_stream_push_char(s, c))
 348                         goto error;
 349                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 350                         if (isl_stream_push_char(s, c))
 351                                 goto error;
 352                 if (c != -1)
 353                         isl_stream_ungetc(s, c);
 354                 isl_stream_push_char(s, '\0');
 355                 isl_int_read(tok->u.v, s->buffer);
 356                 return tok;
 357         }
 358         if (isalpha(c) || c == '_') {
 359                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 360                 if (!tok)
 361                         return NULL;
 362                 isl_stream_push_char(s, c);
 363                 while ((c = isl_stream_getc(s)) != -1 &&
 364                                 (isalnum(c) || c == '_'))
 365                         isl_stream_push_char(s, c);
 366                 if (c != -1)
 367                         isl_stream_ungetc(s, c);
 368                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 369                         isl_stream_push_char(s, c);
 370                 if (c != -1)
 371                         isl_stream_ungetc(s, c);
 372                 isl_stream_push_char(s, '\0');
 373                 tok->type = check_keywords(s);
 374                 if (tok->type != ISL_TOKEN_IDENT)
 375                         tok->is_keyword = 1;
 376                 tok->u.s = strdup(s->buffer);
 377                 if (!tok->u.s)
 378                         goto error;
 379                 return tok;
 380         }
 381         if (c == '"') {
 382                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 383                 if (!tok)
 384                         return NULL;
 385                 tok->type = ISL_TOKEN_STRING;
 386                 tok->u.s = NULL;
 387                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 388                         isl_stream_push_char(s, c);
 389                 if (c != '"') {
 390                         isl_stream_error(s, NULL, "unterminated string");
 391                         goto error;
 392                 }
 393                 isl_stream_push_char(s, '\0');
 394                 tok->u.s = strdup(s->buffer);
 395                 return tok;
 396         }
 397         if (c == ':') {
 398                 int c;
 399                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 400                 if (!tok)
 401                         return NULL;
 402                 if ((c = isl_stream_getc(s)) == '=') {
 403                         tok->u.s = strdup(":=");
 404                         tok->type = ISL_TOKEN_DEF;
 405                         return tok;
 406                 }
 407                 if (c != -1)
 408                         isl_stream_ungetc(s, c);
 409                 tok->type = (enum isl_token_type) ':';
 410                 return tok;
 411         }
 412         if (c == '>') {
 413                 int c;
 414                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 415                 if (!tok)
 416                         return NULL;
 417                 if ((c = isl_stream_getc(s)) == '=') {
 418                         tok->u.s = strdup(">=");
 419                         tok->type = ISL_TOKEN_GE;
 420                         return tok;
 421                 } else if (c == '>') {
 422                         if ((c = isl_stream_getc(s)) == '=') {
 423                                 tok->u.s = strdup(">>=");
 424                                 tok->type = ISL_TOKEN_LEX_GE;
 425                                 return tok;
 426                         }
 427                         tok->u.s = strdup(">>");
 428                         tok->type = ISL_TOKEN_LEX_GT;
 429                 } else {
 430                         tok->u.s = strdup(">");
 431                         tok->type = ISL_TOKEN_GT;
 432                 }
 433                 if (c != -1)
 434                         isl_stream_ungetc(s, c);
 435                 return tok;
 436         }
 437         if (c == '<') {
 438                 int c;
 439                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 440                 if (!tok)
 441                         return NULL;
 442                 if ((c = isl_stream_getc(s)) == '=') {
 443                         tok->u.s = strdup("<=");
 444                         tok->type = ISL_TOKEN_LE;
 445                         return tok;
 446                 } else if (c == '<') {
 447                         if ((c = isl_stream_getc(s)) == '=') {
 448                                 tok->u.s = strdup("<<=");
 449                                 tok->type = ISL_TOKEN_LEX_LE;
 450                                 return tok;
 451                         }
 452                         tok->u.s = strdup("<<");
 453                         tok->type = ISL_TOKEN_LEX_LT;
 454                 } else {
 455                         tok->u.s = strdup("<");
 456                         tok->type = ISL_TOKEN_LT;
 457                 }
 458                 if (c != -1)
 459                         isl_stream_ungetc(s, c);
 460                 return tok;
 461         }
 462         if (c == '&') {
 463                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 464                 if (!tok)
 465                         return NULL;
 466                 tok->type = ISL_TOKEN_AND;
 467                 if ((c = isl_stream_getc(s)) != '&' && c != -1) {
 468                         tok->u.s = strdup("&");
 469                         isl_stream_ungetc(s, c);
 470                 } else
 471                         tok->u.s = strdup("&&");
 472                 return tok;
 473         }
 474         if (c == '|') {
 475                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 476                 if (!tok)
 477                         return NULL;
 478                 tok->type = ISL_TOKEN_OR;
 479                 if ((c = isl_stream_getc(s)) != '|' && c != -1) {
 480                         tok->u.s = strdup("|");
 481                         isl_stream_ungetc(s, c);
 482                 } else
 483                         tok->u.s = strdup("||");
 484                 return tok;
 485         }
 486         if (c == '!') {
 487                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 488                 if (!tok)
 489                         return NULL;
 490                 tok->type = ISL_TOKEN_NOT;
 491                 tok->u.s = strdup("!");
 492                 return tok;
 493         }
 494
 495         tok = isl_token_new(s->ctx, line, col, old_line != line);
 496         if (!tok)
 497                 return NULL;
 498         tok->type = ISL_TOKEN_UNKNOWN;
 499         return tok;
 500 error:
 501         isl_token_free(tok);
 502         return NULL;
 503 }
 504
 505 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 506 {
 507         return next_token(s, 0);
 508 }
 509
 510 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 511 {
 512         return next_token(s, 1);
 513 }
 514
 515 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 516 {
 517         struct isl_token *tok;
 518
 519         tok = isl_stream_next_token(s);
 520         if (!tok)
 521                 return 0;
 522         if (tok->type == type) {
 523                 isl_token_free(tok);
 524                 return 1;
 525         }
 526         isl_stream_push_token(s, tok);
 527         return 0;
 528 }
 529
 530 int isl_stream_next_token_is(struct isl_stream *s, int type)
 531 {
 532         struct isl_token *tok;
 533         int r;
 534
 535         tok = isl_stream_next_token(s);
 536         if (!tok)
 537                 return 0;
 538         r = tok->type == type;
 539         isl_stream_push_token(s, tok);
 540         return r;
 541 }
 542
 543 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 544 {
 545         struct isl_token *tok;
 546
 547         tok = isl_stream_next_token(s);
 548         if (!tok)
 549                 return NULL;
 550         if (tok->type == ISL_TOKEN_IDENT) {
 551                 char *ident = strdup(tok->u.s);
 552                 isl_token_free(tok);
 553                 return ident;
 554         }
 555         isl_stream_push_token(s, tok);
 556         return NULL;
 557 }
 558
 559 int isl_stream_eat(struct isl_stream *s, int type)
 560 {
 561         struct isl_token *tok;
 562
 563         tok = isl_stream_next_token(s);
 564         if (!tok)
 565                 return -1;
 566         if (tok->type == type) {
 567                 isl_token_free(tok);
 568                 return 0;
 569         }
 570         isl_stream_error(s, tok, "expecting other token");
 571         isl_stream_push_token(s, tok);
 572         return -1;
 573 }
 574
 575 int isl_stream_is_empty(struct isl_stream *s)
 576 {
 577         struct isl_token *tok;
 578
 579         tok = isl_stream_next_token(s);
 580
 581         if (!tok)
 582                 return 1;
 583
 584         isl_stream_push_token(s, tok);
 585         return 0;
 586 }
 587
 588 static int free_keyword(void **p, void *user)
 589 {
 590         struct isl_keyword *keyword = *p;
 591
 592         free(keyword->name);
 593         free(keyword);
 594
 595         return 0;
 596 }
 597
 598 void isl_stream_flush_tokens(struct isl_stream *s)
 599 {
 600         int i;
 601
 602         if (!s)
 603                 return;
 604         for (i = 0; i < s->n_token; ++i)
 605                 isl_token_free(s->tokens[i]);
 606         s->n_token = 0;
 607 }
 608
 609 void isl_stream_free(struct isl_stream *s)
 610 {
 611         if (!s)
 612                 return;
 613         free(s->buffer);
 614         if (s->n_token != 0) {
 615                 struct isl_token *tok = isl_stream_next_token(s);
 616                 isl_stream_error(s, tok, "unexpected token");
 617                 isl_token_free(tok);
 618         }
 619         if (s->keywords) {
 620                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 621                 isl_hash_table_free(s->ctx, s->keywords);
 622         }
 623         isl_ctx_deref(s->ctx);
 624         free(s);
 625 }