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