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         return tok;
  77 }
  78
  79 void isl_token_free(struct isl_token *tok)
  80 {
  81         if (!tok)
  82                 return;
  83         if (tok->type == ISL_TOKEN_VALUE)
  84                 isl_int_clear(tok->u.v);
  85         else if (tok->type == ISL_TOKEN_IDENT || tok->type == ISL_TOKEN_STRING)
  86                 free(tok->u.s);
  87         free(tok);
  88 }
  89
  90 void isl_stream_error(struct isl_stream *s, struct isl_token *tok, char *msg)
  91 {
  92         int line = tok ? tok->line : s->line;
  93         int col = tok ? tok->col : s->col;
  94         fprintf(stderr, "syntax error (%d, %d): %s\n", line, col, msg);
  95         if (tok) {
  96                 if (tok->type < 256)
  97                         fprintf(stderr, "got '%c'\n", tok->type);
  98                 else if (tok->type == ISL_TOKEN_IDENT)
  99                         fprintf(stderr, "got ident '%s'\n", tok->u.s);
 100                 else
 101                         fprintf(stderr, "got token type %d\n", tok->type);
 102         }
 103 }
 104
 105 static struct isl_stream* isl_stream_new(struct isl_ctx *ctx)
 106 {
 107         int i;
 108         struct isl_stream *s = isl_alloc_type(ctx, struct isl_stream);
 109         if (!s)
 110                 return NULL;
 111         s->ctx = ctx;
 112         isl_ctx_ref(s->ctx);
 113         s->file = NULL;
 114         s->str = NULL;
 115         s->len = 0;
 116         s->line = 1;
 117         s->col = 0;
 118         s->eof = 0;
 119         s->c = -1;
 120         for (i = 0; i < 5; ++i)
 121                 s->tokens[i] = NULL;
 122         s->n_token = 0;
 123         s->keywords = NULL;
 124         s->size = 256;
 125         s->buffer = isl_alloc_array(ctx, char, s->size);
 126         if (!s->buffer)
 127                 goto error;
 128         return s;
 129 error:
 130         isl_stream_free(s);
 131         return NULL;
 132 }
 133
 134 struct isl_stream* isl_stream_new_file(struct isl_ctx *ctx, FILE *file)
 135 {
 136         struct isl_stream *s = isl_stream_new(ctx);
 137         if (!s)
 138                 return NULL;
 139         s->file = file;
 140         return s;
 141 }
 142
 143 struct isl_stream* isl_stream_new_str(struct isl_ctx *ctx, const char *str)
 144 {
 145         struct isl_stream *s = isl_stream_new(ctx);
 146         if (!s)
 147                 return NULL;
 148         s->str = str;
 149         return s;
 150 }
 151
 152 static int isl_stream_getc(struct isl_stream *s)
 153 {
 154         int c;
 155         if (s->eof)
 156                 return -1;
 157         if (s->file)
 158                 c = fgetc(s->file);
 159         else {
 160                 c = *s->str++;
 161                 if (c == '\0')
 162                         c = -1;
 163         }
 164         if (c == -1)
 165                 s->eof = 1;
 166         if (!s->eof) {
 167                 if (s->c == '\n') {
 168                         s->line++;
 169                         s->col = 0;
 170                 } else
 171                         s->col++;
 172         }
 173         s->c = c;
 174         return c;
 175 }
 176
 177 static void isl_stream_ungetc(struct isl_stream *s, int c)
 178 {
 179         if (s->file)
 180                 ungetc(c, s->file);
 181         else
 182                 --s->str;
 183         s->c = -1;
 184 }
 185
 186 static int isl_stream_push_char(struct isl_stream *s, int c)
 187 {
 188         if (s->len >= s->size) {
 189                 s->size = (3*s->size)/2;
 190                 s->buffer = isl_realloc_array(s->ctx, s->buffer, char, s->size);
 191                 if (!s->buffer)
 192                         return -1;
 193         }
 194         s->buffer[s->len++] = c;
 195         return 0;
 196 }
 197
 198 void isl_stream_push_token(struct isl_stream *s, struct isl_token *tok)
 199 {
 200         isl_assert(s->ctx, s->n_token < 5, return);
 201         s->tokens[s->n_token++] = tok;
 202 }
 203
 204 static enum isl_token_type check_keywords(struct isl_stream *s)
 205 {
 206         struct isl_hash_table_entry *entry;
 207         struct isl_keyword *keyword;
 208         uint32_t name_hash;
 209
 210         if (!strcasecmp(s->buffer, "exists"))
 211                 return ISL_TOKEN_EXISTS;
 212         if (!strcasecmp(s->buffer, "and"))
 213                 return ISL_TOKEN_AND;
 214         if (!strcasecmp(s->buffer, "or"))
 215                 return ISL_TOKEN_OR;
 216         if (!strcasecmp(s->buffer, "infty"))
 217                 return ISL_TOKEN_INFTY;
 218         if (!strcasecmp(s->buffer, "infinity"))
 219                 return ISL_TOKEN_INFTY;
 220         if (!strcasecmp(s->buffer, "NaN"))
 221                 return ISL_TOKEN_NAN;
 222
 223         if (!s->keywords)
 224                 return ISL_TOKEN_IDENT;
 225
 226         name_hash = isl_hash_string(isl_hash_init(), s->buffer);
 227         entry = isl_hash_table_find(s->ctx, s->keywords, name_hash, same_name,
 228                                         s->buffer, 0);
 229         if (entry) {
 230                 keyword = entry->data;
 231                 return keyword->type;
 232         }
 233
 234         return ISL_TOKEN_IDENT;
 235 }
 236
 237 int isl_stream_skip_line(struct isl_stream *s)
 238 {
 239         int c;
 240
 241         while ((c = isl_stream_getc(s)) != -1 && c != '\n')
 242                 /* nothing */
 243                 ;
 244
 245         return c == -1 ? -1 : 0;
 246 }
 247
 248 static struct isl_token *next_token(struct isl_stream *s, int same_line)
 249 {
 250         int c;
 251         struct isl_token *tok = NULL;
 252         int line, col;
 253         int old_line = s->line;
 254
 255         if (s->n_token) {
 256                 if (same_line && s->tokens[s->n_token - 1]->on_new_line)
 257                         return NULL;
 258                 return s->tokens[--s->n_token];
 259         }
 260
 261         if (same_line && s->c == '\n')
 262                 return NULL;
 263
 264         s->len = 0;
 265
 266         /* skip spaces and comment lines */
 267         while ((c = isl_stream_getc(s)) != -1) {
 268                 if (c == '#') {
 269                         if (isl_stream_skip_line(s) < 0)
 270                                 break;
 271                         c = '\n';
 272                         if (same_line)
 273                                 break;
 274                 } else if (!isspace(c) || (same_line && c == '\n'))
 275                         break;
 276         }
 277
 278         line = s->line;
 279         col = s->col;
 280
 281         if (c == -1 || (same_line && c == '\n'))
 282                 return NULL;
 283         if (c == '(' ||
 284             c == ')' ||
 285             c == '+' ||
 286             c == '/' ||
 287             c == '*' ||
 288             c == '%' ||
 289             c == '^' ||
 290             c == '=' ||
 291             c == '@' ||
 292             c == ',' ||
 293             c == '.' ||
 294             c == ';' ||
 295             c == '[' ||
 296             c == ']' ||
 297             c == '{' ||
 298             c == '}') {
 299                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 300                 if (!tok)
 301                         return NULL;
 302                 tok->type = (enum isl_token_type)c;
 303                 return tok;
 304         }
 305         if (c == '-') {
 306                 int c;
 307                 if ((c = isl_stream_getc(s)) == '>') {
 308                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 309                         if (!tok)
 310                                 return NULL;
 311                         tok->type = ISL_TOKEN_TO;
 312                         return tok;
 313                 }
 314                 if (c != -1)
 315                         isl_stream_ungetc(s, c);
 316                 if (!isdigit(c)) {
 317                         tok = isl_token_new(s->ctx, line, col, old_line != line);
 318                         if (!tok)
 319                                 return NULL;
 320                         tok->type = (enum isl_token_type) '-';
 321                         return tok;
 322                 }
 323         }
 324         if (c == '-' || isdigit(c)) {
 325                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 326                 if (!tok)
 327                         return NULL;
 328                 tok->type = ISL_TOKEN_VALUE;
 329                 isl_int_init(tok->u.v);
 330                 if (isl_stream_push_char(s, c))
 331                         goto error;
 332                 while ((c = isl_stream_getc(s)) != -1 && isdigit(c))
 333                         if (isl_stream_push_char(s, c))
 334                                 goto error;
 335                 if (c != -1)
 336                         isl_stream_ungetc(s, c);
 337                 isl_stream_push_char(s, '\0');
 338                 isl_int_read(tok->u.v, s->buffer);
 339                 return tok;
 340         }
 341         if (isalpha(c)) {
 342                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 343                 if (!tok)
 344                         return NULL;
 345                 isl_stream_push_char(s, c);
 346                 while ((c = isl_stream_getc(s)) != -1 &&
 347                                 (isalnum(c) || c == '_'))
 348                         isl_stream_push_char(s, c);
 349                 if (c != -1)
 350                         isl_stream_ungetc(s, c);
 351                 while ((c = isl_stream_getc(s)) != -1 && c == '\'')
 352                         isl_stream_push_char(s, c);
 353                 if (c != -1)
 354                         isl_stream_ungetc(s, c);
 355                 isl_stream_push_char(s, '\0');
 356                 tok->type = check_keywords(s);
 357                 if (tok->type == ISL_TOKEN_IDENT)
 358                         tok->u.s = strdup(s->buffer);
 359                 return tok;
 360         }
 361         if (c == '"') {
 362                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 363                 if (!tok)
 364                         return NULL;
 365                 tok->type = ISL_TOKEN_STRING;
 366                 tok->u.s = NULL;
 367                 while ((c = isl_stream_getc(s)) != -1 && c != '"' && c != '\n')
 368                         isl_stream_push_char(s, c);
 369                 if (c != '"') {
 370                         isl_stream_error(s, NULL, "unterminated string");
 371                         goto error;
 372                 }
 373                 isl_stream_push_char(s, '\0');
 374                 tok->u.s = strdup(s->buffer);
 375                 return tok;
 376         }
 377         if (c == ':') {
 378                 int c;
 379                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 380                 if (!tok)
 381                         return NULL;
 382                 if ((c = isl_stream_getc(s)) == '=') {
 383                         tok->type = ISL_TOKEN_DEF;
 384                         return tok;
 385                 }
 386                 if (c != -1)
 387                         isl_stream_ungetc(s, c);
 388                 tok->type = (enum isl_token_type) ':';
 389                 return tok;
 390         }
 391         if (c == '>') {
 392                 int c;
 393                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 394                 if (!tok)
 395                         return NULL;
 396                 if ((c = isl_stream_getc(s)) == '=') {
 397                         tok->type = ISL_TOKEN_GE;
 398                         return tok;
 399                 } else if (c == '>') {
 400                         if ((c = isl_stream_getc(s)) == '=') {
 401                                 tok->type = ISL_TOKEN_LEX_GE;
 402                                 return tok;
 403                         }
 404                         tok->type = ISL_TOKEN_LEX_GT;
 405                 } else
 406                         tok->type = ISL_TOKEN_GT;
 407                 if (c != -1)
 408                         isl_stream_ungetc(s, c);
 409                 return tok;
 410         }
 411         if (c == '<') {
 412                 int c;
 413                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 414                 if (!tok)
 415                         return NULL;
 416                 if ((c = isl_stream_getc(s)) == '=') {
 417                         tok->type = ISL_TOKEN_LE;
 418                         return tok;
 419                 } else if (c == '<') {
 420                         if ((c = isl_stream_getc(s)) == '=') {
 421                                 tok->type = ISL_TOKEN_LEX_LE;
 422                                 return tok;
 423                         }
 424                         tok->type = ISL_TOKEN_LEX_LT;
 425                 } else
 426                         tok->type = ISL_TOKEN_LT;
 427                 if (c != -1)
 428                         isl_stream_ungetc(s, c);
 429                 return tok;
 430         }
 431         if (c == '&') {
 432                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 433                 if (!tok)
 434                         return NULL;
 435                 tok->type = ISL_TOKEN_AND;
 436                 if ((c = isl_stream_getc(s)) != '&' && c != -1)
 437                         isl_stream_ungetc(s, c);
 438                 return tok;
 439         }
 440         if (c == '|') {
 441                 tok = isl_token_new(s->ctx, line, col, old_line != line);
 442                 if (!tok)
 443                         return NULL;
 444                 tok->type = ISL_TOKEN_OR;
 445                 if ((c = isl_stream_getc(s)) != '|' && c != -1)
 446                         isl_stream_ungetc(s, c);
 447                 return tok;
 448         }
 449
 450         tok = isl_token_new(s->ctx, line, col, old_line != line);
 451         if (!tok)
 452                 return NULL;
 453         tok->type = ISL_TOKEN_UNKNOWN;
 454         return tok;
 455 error:
 456         isl_token_free(tok);
 457         return NULL;
 458 }
 459
 460 struct isl_token *isl_stream_next_token(struct isl_stream *s)
 461 {
 462         return next_token(s, 0);
 463 }
 464
 465 struct isl_token *isl_stream_next_token_on_same_line(struct isl_stream *s)
 466 {
 467         return next_token(s, 1);
 468 }
 469
 470 int isl_stream_eat_if_available(struct isl_stream *s, int type)
 471 {
 472         struct isl_token *tok;
 473
 474         tok = isl_stream_next_token(s);
 475         if (!tok)
 476                 return 0;
 477         if (tok->type == type) {
 478                 isl_token_free(tok);
 479                 return 1;
 480         }
 481         isl_stream_push_token(s, tok);
 482         return 0;
 483 }
 484
 485 int isl_stream_next_token_is(struct isl_stream *s, int type)
 486 {
 487         struct isl_token *tok;
 488         int r;
 489
 490         tok = isl_stream_next_token(s);
 491         if (!tok)
 492                 return 0;
 493         r = tok->type == type;
 494         isl_stream_push_token(s, tok);
 495         return r;
 496 }
 497
 498 char *isl_stream_read_ident_if_available(struct isl_stream *s)
 499 {
 500         struct isl_token *tok;
 501
 502         tok = isl_stream_next_token(s);
 503         if (!tok)
 504                 return NULL;
 505         if (tok->type == ISL_TOKEN_IDENT) {
 506                 char *ident = strdup(tok->u.s);
 507                 isl_token_free(tok);
 508                 return ident;
 509         }
 510         isl_stream_push_token(s, tok);
 511         return NULL;
 512 }
 513
 514 int isl_stream_eat(struct isl_stream *s, int type)
 515 {
 516         struct isl_token *tok;
 517
 518         tok = isl_stream_next_token(s);
 519         if (!tok)
 520                 return -1;
 521         if (tok->type == type) {
 522                 isl_token_free(tok);
 523                 return 0;
 524         }
 525         isl_stream_error(s, tok, "expecting other token");
 526         isl_stream_push_token(s, tok);
 527         return -1;
 528 }
 529
 530 int isl_stream_is_empty(struct isl_stream *s)
 531 {
 532         struct isl_token *tok;
 533
 534         tok = isl_stream_next_token(s);
 535
 536         if (!tok)
 537                 return 1;
 538
 539         isl_stream_push_token(s, tok);
 540         return 0;
 541 }
 542
 543 static int free_keyword(void **p, void *user)
 544 {
 545         struct isl_keyword *keyword = *p;
 546
 547         free(keyword->name);
 548         free(keyword);
 549
 550         return 0;
 551 }
 552
 553 void isl_stream_flush_tokens(struct isl_stream *s)
 554 {
 555         int i;
 556
 557         if (!s)
 558                 return;
 559         for (i = 0; i < s->n_token; ++i)
 560                 isl_token_free(s->tokens[i]);
 561         s->n_token = 0;
 562 }
 563
 564 void isl_stream_free(struct isl_stream *s)
 565 {
 566         if (!s)
 567                 return;
 568         free(s->buffer);
 569         if (s->n_token != 0) {
 570                 struct isl_token *tok = isl_stream_next_token(s);
 571                 isl_stream_error(s, tok, "unexpected token");
 572                 isl_token_free(tok);
 573         }
 574         if (s->keywords) {
 575                 isl_hash_table_foreach(s->ctx, s->keywords, &free_keyword, NULL);
 576                 isl_hash_table_free(s->ctx, s->keywords);
 577         }
 578         isl_ctx_deref(s->ctx);
 579         free(s);
 580 }