src/print-xml.c

   1 /* Print an xml on generated parser, for Bison,
   2
   3    Copyright (C) 2007 Free Software Foundation, Inc.
   4
   5    This file is part of Bison, the GNU Compiler Compiler.
   6
   7    Bison is free software; you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation; either version 2, or (at your option)
  10    any later version.
  11
  12    Bison is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License
  18    along with Bison; see the file COPYING.  If not, write to
  19    the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  20    Boston, MA 02110-1301, USA.  */
  21
  22 #include <config.h>
  23 #include "system.h"
  24
  25 #include <stdarg.h>
  26
  27 #include <bitset.h>
  28 #include <quotearg.h>
  29
  30 #include "LR0.h"
  31 #include "closure.h"
  32 #include "conflicts.h"
  33 #include "files.h"
  34 #include "getargs.h"
  35 #include "gram.h"
  36 #include "lalr.h"
  37 #include "print.h"
  38 #include "print-xml.h"
  39 #include "reader.h"
  40 #include "reduce.h"
  41 #include "state.h"
  42 #include "symtab.h"
  43 #include "tables.h"
  44
  45 static bitset no_reduce_set;
  46 struct escape_buf
  47 {
  48   char *ptr;
  49   size_t size;
  50 };
  51 static struct escape_buf escape_bufs[2];
  52
  53
  54 /*----------------------------.
  55 | Print rules never reduced.  |
  56 `-----------------------------*/
  57
  58 static void
  59 print_rules_never_reduced (FILE *out, int level)
  60 {
  61   rule_number r;
  62   bool count = false;
  63
  64   for (r = 0; r < nrules + nuseless_productions; r++)
  65     {
  66       if (rule_never_reduced_p (&rules[r]))
  67         {
  68           count = true;
  69           break;
  70         }
  71     }
  72
  73   if (count) {
  74     xml_puts (out, level, "<rules-never-reduced>");
  75     grammar_rules_partial_print_xml (out, level - 1,
  76                                      false, rule_never_reduced_p);
  77     xml_puts (out, level, "</rules-never-reduced>");
  78   }
  79   else
  80     xml_puts (out, level, "<rules-never-reduced/>");
  81 }
  82
  83 /*--------------------------------.
  84 | Report information on a state.  |
  85 `--------------------------------*/
  86
  87 static void
  88 print_core (FILE *out, int level, state *s)
  89 {
  90   size_t i;
  91   item_number *sitems = s->items;
  92   size_t snritems = s->nitems;
  93
  94   /* Output all the items of a state, not only its kernel.  */
  95   if (report_flag & report_itemsets)
  96     {
  97       closure (sitems, snritems);
  98       sitems = itemset;
  99       snritems = nitemset;
 100     }
 101
 102   if (!snritems) {
 103     xml_puts (out, level, "<itemset/>");
 104     return;
 105   }
 106
 107   xml_puts (out, level, "<itemset>");
 108
 109   for (i = 0; i < snritems; i++)
 110     {
 111       bool printed = false;
 112       item_number *sp;
 113       item_number *sp1;
 114       rule_number r;
 115
 116       sp1 = sp = ritem + sitems[i];
 117
 118       while (*sp >= 0)
 119         sp++;
 120
 121       r = item_number_as_rule_number (*sp);
 122       sp = rules[r].rhs;
 123
 124       /* Display the lookahead tokens?  */
 125       if (report_flag & report_lookahead_tokens
 126           && item_number_is_rule_number (*sp1))
 127         {
 128           reductions *reds = s->reductions;
 129           int red = state_reduction_find (s, &rules[r]);
 130           /* Print item with lookaheads if there are. */
 131           if (reds->lookahead_tokens && red != -1)
 132             {
 133               xml_printf (out, level + 1,
 134                           "<item rule-number=\"%d\" point=\"%d\">",
 135                           rules[r].number, sp1 - sp);
 136               state_rule_lookahead_tokens_print_xml (s, &rules[r],
 137                                                      out, level + 2);
 138               xml_puts (out, level + 1, "</item>");
 139               printed = true;
 140             }
 141         }
 142
 143       if (!printed)
 144         {
 145           xml_printf (out, level + 1,
 146                       "<item rule-number=\"%d\" point=\"%d\"/>",
 147                       rules[r].number,
 148                       sp1 - sp);
 149         }
 150     }
 151   xml_puts (out, level, "</itemset>");
 152 }
 153
 154
 155 /*-----------------------------------------------------------.
 156 | Report the shifts if DISPLAY_SHIFTS_P or the gotos of S on |
 157 | OUT.                                                       |
 158 `-----------------------------------------------------------*/
 159
 160 static void
 161 print_transitions (state *s, FILE *out, int level)
 162 {
 163   transitions *trans = s->transitions;
 164   int n = 0;
 165   int i;
 166
 167   for (i = 0; i < trans->num; i++)
 168     if (!TRANSITION_IS_DISABLED (trans, i))
 169       {
 170         n++;
 171       }
 172
 173   /* Nothing to report. */
 174   if (!n) {
 175     xml_puts (out, level, "<transitions/>");
 176     return;
 177   }
 178
 179   /* Report lookahead tokens and shifts.  */
 180   xml_puts (out, level, "<transitions>");
 181
 182   for (i = 0; i < trans->num; i++)
 183     if (!TRANSITION_IS_DISABLED (trans, i)
 184         && TRANSITION_IS_SHIFT (trans, i))
 185       {
 186         symbol *sym = symbols[TRANSITION_SYMBOL (trans, i)];
 187         char const *tag = sym->tag;
 188         state *s1 = trans->states[i];
 189
 190         xml_printf (out, level + 1,
 191                     "<transition type=\"shift\" symbol=\"%s\" state=\"%d\"/>",
 192                     xml_escape (tag), s1->number);
 193       }
 194
 195   for (i = 0; i < trans->num; i++)
 196     if (!TRANSITION_IS_DISABLED (trans, i)
 197         && !TRANSITION_IS_SHIFT (trans, i))
 198       {
 199         symbol *sym = symbols[TRANSITION_SYMBOL (trans, i)];
 200         char const *tag = sym->tag;
 201         state *s1 = trans->states[i];
 202
 203         xml_printf (out, level + 1,
 204                     "<transition type=\"goto\" symbol=\"%s\" state=\"%d\"/>",
 205                     xml_escape (tag), s1->number);
 206       }
 207
 208   xml_puts (out, level, "</transitions>");
 209 }
 210
 211
 212 /*--------------------------------------------------------.
 213 | Report the explicit errors of S raised from %nonassoc.  |
 214 `--------------------------------------------------------*/
 215
 216 static void
 217 print_errs (FILE *out, int level, state *s)
 218 {
 219   errs *errp = s->errs;
 220   bool count = false;
 221   int i;
 222
 223   for (i = 0; i < errp->num; ++i)
 224     if (errp->symbols[i])
 225       count = true;
 226
 227   /* Nothing to report. */
 228   if (!count) {
 229     xml_puts (out, level, "<errors/>");
 230     return;
 231   }
 232
 233   /* Report lookahead tokens and errors.  */
 234   xml_puts (out, level, "<errors>");
 235   for (i = 0; i < errp->num; ++i)
 236     if (errp->symbols[i])
 237       {
 238         char const *tag = errp->symbols[i]->tag;
 239         xml_printf (out, level + 1,
 240                     "<error symbol=\"%s\">nonassociative</error>",
 241                     xml_escape (tag));
 242       }
 243   xml_puts (out, level, "</errors>");
 244 }
 245
 246
 247 /*-------------------------------------------------------------------------.
 248 | Report a reduction of RULE on LOOKAHEAD_TOKEN (which can be `default').  |
 249 | If not ENABLED, the rule is masked by a shift or a reduce (S/R and       |
 250 | R/R conflicts).                                                          |
 251 `-------------------------------------------------------------------------*/
 252
 253 static void
 254 print_reduction (FILE *out, int level, char const *lookahead_token,
 255                  rule *r, bool enabled)
 256 {
 257   if (r->number)
 258     xml_printf (out, level,
 259                 "<reduction symbol=\"%s\" rule=\"%d\" enabled=\"%s\"/>",
 260                 xml_escape (lookahead_token),
 261                 r->number,
 262                 enabled ? "true" : "false");
 263   else
 264     xml_printf (out, level,
 265                 "<reduction symbol=\"%s\" rule=\"accept\" enabled=\"%s\"/>",
 266                 xml_escape (lookahead_token),
 267                 enabled ? "true" : "false");
 268 }
 269
 270
 271 /*-------------------------------------------.
 272 | Report on OUT the reduction actions of S.  |
 273 `-------------------------------------------*/
 274
 275 static void
 276 print_reductions (FILE *out, int level, state *s)
 277 {
 278   transitions *trans = s->transitions;
 279   reductions *reds = s->reductions;
 280   rule *default_rule = NULL;
 281   int report = false;
 282   int i, j;
 283
 284   if (reds->num == 0) {
 285     xml_puts (out, level, "<reductions/>");
 286     return;
 287   }
 288
 289   if (yydefact[s->number] != 0)
 290     default_rule = &rules[yydefact[s->number] - 1];
 291
 292   bitset_zero (no_reduce_set);
 293   FOR_EACH_SHIFT (trans, i)
 294     bitset_set (no_reduce_set, TRANSITION_SYMBOL (trans, i));
 295   for (i = 0; i < s->errs->num; ++i)
 296     if (s->errs->symbols[i])
 297       bitset_set (no_reduce_set, s->errs->symbols[i]->number);
 298
 299   if (default_rule)
 300     report = true;
 301
 302   if (reds->lookahead_tokens)
 303     for (i = 0; i < ntokens; i++)
 304       {
 305         bool count = bitset_test (no_reduce_set, i);
 306
 307         for (j = 0; j < reds->num; ++j)
 308           if (bitset_test (reds->lookahead_tokens[j], i))
 309             {
 310               if (! count)
 311                 {
 312                   if (reds->rules[j] != default_rule)
 313                     report = true;
 314                   count = true;
 315                 }
 316               else
 317                 {
 318                   report = true;
 319                 }
 320             }
 321       }
 322
 323   /* Nothing to report. */
 324   if (!report) {
 325     xml_puts (out, level, "<reductions/>");
 326     return;
 327   }
 328
 329   xml_puts (out, level, "<reductions>");
 330
 331   /* Report lookahead tokens (or $default) and reductions.  */
 332   if (reds->lookahead_tokens)
 333     for (i = 0; i < ntokens; i++)
 334       {
 335         bool defaulted = false;
 336         bool count = bitset_test (no_reduce_set, i);
 337
 338         for (j = 0; j < reds->num; ++j)
 339           if (bitset_test (reds->lookahead_tokens[j], i))
 340             {
 341               if (! count)
 342                 {
 343                   if (reds->rules[j] != default_rule)
 344                     print_reduction (out, level + 1, symbols[i]->tag,
 345                                      reds->rules[j], true);
 346                   else
 347                     defaulted = true;
 348                   count = true;
 349                 }
 350               else
 351                 {
 352                   if (defaulted)
 353                     print_reduction (out, level + 1, symbols[i]->tag,
 354                                      default_rule, true);
 355                   defaulted = false;
 356                   print_reduction (out, level + 1, symbols[i]->tag,
 357                                    reds->rules[j], false);
 358                 }
 359             }
 360       }
 361
 362   if (default_rule)
 363     print_reduction (out, level + 1,
 364                      "$default", default_rule, true);
 365
 366   xml_puts (out, level, "</reductions>");
 367 }
 368
 369
 370 /*--------------------------------------------------------------.
 371 | Report on OUT all the actions (shifts, gotos, reductions, and |
 372 | explicit erros from %nonassoc) of S.                          |
 373 `--------------------------------------------------------------*/
 374
 375 static void
 376 print_actions (FILE *out, int level, state *s)
 377 {
 378   xml_puts (out, level, "<actions>");
 379   print_transitions (s, out, level + 1);
 380   print_errs (out, level + 1, s);
 381   print_reductions (out, level + 1, s);
 382   xml_puts (out, level, "</actions>");
 383 }
 384
 385
 386 /*----------------------------------.
 387 | Report all the data on S on OUT.  |
 388 `----------------------------------*/
 389
 390 static void
 391 print_state (FILE *out, int level, state *s)
 392 {
 393   fputc ('\n', out);
 394   xml_printf (out, level, "<state number=\"%d\">", s->number);
 395   print_core (out, level + 1, s);
 396   print_actions (out, level + 1, s);
 397   if ((report_flag & report_solved_conflicts) && s->solved_conflicts_xml)
 398     {
 399       xml_puts (out, level + 1, "<solved-conflicts>");
 400       fputs (s->solved_conflicts_xml, out);
 401       xml_puts (out, level + 1, "</solved-conflicts>");
 402     }
 403   else
 404     xml_puts (out, level + 1, "<solved-conflicts/>");
 405   xml_puts (out, level, "</state>");
 406 }
 407
 408
 409 /*-----------------------------------------.
 410 | Print information on the whole grammar.  |
 411 `-----------------------------------------*/
 412
 413 static void
 414 print_grammar (FILE *out, int level)
 415 {
 416   symbol_number i;
 417
 418   fputc ('\n', out);
 419   xml_puts (out, level, "<grammar>");
 420   grammar_rules_print_xml (out, level);
 421
 422   /* Terminals */
 423   xml_puts (out, level + 1, "<terminals>");
 424   for (i = 0; i < max_user_token_number + 1; i++)
 425     if (token_translations[i] != undeftoken->number)
 426       {
 427         char const *tag = symbols[token_translations[i]]->tag;
 428         rule_number r;
 429         item_number *rhsp;
 430
 431         xml_printf (out, level + 2,
 432                     "<terminal symbol-number=\"%d\" token-number=\"%d\""
 433                     " name=\"%s\">",
 434                     token_translations[i], i, xml_escape (tag));
 435
 436         for (r = 0; r < nrules; r++)
 437           for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
 438             if (item_number_as_symbol_number (*rhsp) == token_translations[i])
 439               {
 440                 xml_printf (out, level + 3, "<rule>%d</rule>", r);
 441                 break;
 442               }
 443         xml_puts (out, level + 2, "</terminal>");
 444       }
 445   xml_puts (out, level + 1, "</terminals>");
 446
 447   /* Nonterminals */
 448   xml_puts (out, level + 1, "<nonterminals>");
 449   for (i = ntokens; i < nsyms; i++)
 450     {
 451       int left_count = 0, right_count = 0;
 452       rule_number r;
 453       char const *tag = symbols[i]->tag;
 454
 455       for (r = 0; r < nrules; r++)
 456         {
 457           item_number *rhsp;
 458           if (rules[r].lhs->number == i)
 459             left_count++;
 460           for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
 461             if (item_number_as_symbol_number (*rhsp) == i)
 462               {
 463                 right_count++;
 464                 break;
 465               }
 466         }
 467
 468       xml_printf (out, level + 2,
 469                   "<nonterminal symbol-number=\"%d\" name=\"%s\">",
 470                   i, xml_escape (tag));
 471
 472       if (left_count > 0)
 473         {
 474           xml_puts (out, level + 3, "<left>");
 475           for (r = 0; r < nrules; r++)
 476             {
 477               if (rules[r].lhs->number == i)
 478                 xml_printf (out, level + 4, "<rule>%d</rule>", r);
 479             }
 480           xml_puts (out, level + 3, "</left>");
 481         }
 482
 483       if (right_count > 0)
 484         {
 485           xml_puts (out, level + 3, "<right>");
 486           for (r = 0; r < nrules; r++)
 487             {
 488               item_number *rhsp;
 489               for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
 490                 if (item_number_as_symbol_number (*rhsp) == i)
 491                   {
 492                     xml_printf (out, level + 4, "<rule>%d</rule>", r);
 493                     break;
 494                   }
 495             }
 496           xml_puts (out, level + 3, "</right>");
 497         }
 498
 499       xml_puts (out, level + 2, "</nonterminal>");
 500     }
 501   xml_puts (out, level + 1, "</nonterminals>");
 502   xml_puts (out, level, "</grammar>");
 503 }
 504
 505 void
 506 xml_puts (FILE *out, int level, char const *s)
 507 {
 508   int i;
 509   for (i = 0; i < level; i++)
 510     fputs ("  ", out);
 511   fputs (s, out);
 512   fputc ('\n', out);
 513 }
 514
 515 void
 516 xml_printf (FILE *out, int level, char const *fmt, ...)
 517 {
 518   int i;
 519   va_list arglist;
 520
 521   for (i = 0; i < level; i++)
 522     fputs ("  ", out);
 523
 524   va_start (arglist, fmt);
 525   vfprintf (out, fmt, arglist);
 526   va_end (arglist);
 527
 528   fputc ('\n', out);
 529 }
 530
 531 static char const *
 532 xml_escape_string (struct escape_buf *buf, char const *str)
 533 {
 534   size_t len = strlen (str);
 535   size_t max_expansion = sizeof "&quot;" - 1;
 536   char *p;
 537
 538   if (buf->size <= max_expansion * len)
 539     {
 540       buf->size = max_expansion * len + 1;
 541       buf->ptr = x2realloc (buf->ptr, &buf->size);
 542     }
 543   p = buf->ptr;
 544
 545   for (; *str; str++)
 546     switch (*str)
 547       {
 548       default: *p++ = *str; break;
 549       case '&': p = stpcpy (p, "&amp;" ); break;
 550       case '<': p = stpcpy (p, "&lt;"  ); break;
 551       case '>': p = stpcpy (p, "&gt;"  ); break;
 552       case '"': p = stpcpy (p, "&quot;"); break;
 553       }
 554
 555   *p = '\0';
 556   return buf->ptr;
 557 }
 558
 559 char const *
 560 xml_escape_n (int n, char const *str)
 561 {
 562   return xml_escape_string (escape_bufs + n, str);
 563 }
 564
 565 char const *
 566 xml_escape (char const *str)
 567 {
 568   return xml_escape_n (0, str);
 569 }
 570
 571 void
 572 print_xml (void)
 573 {
 574   state_number i;
 575   int level = 0;
 576
 577   FILE *out = xfopen (spec_xml_file, "w");
 578
 579   fputs ("<?xml version=\"1.0\"?>\n\n", out);
 580   xml_printf (out, level, "<bison-xml-report version=\"%s\">",
 581               xml_escape (VERSION));
 582
 583   fputc ('\n', out);
 584   xml_printf (out, level + 1, "<filename>%s</filename>",
 585               xml_escape (grammar_file));
 586
 587   /* print reductions */
 588   reduce_xml (out, level + 1);
 589
 590   /* print rules never reduced */
 591   print_rules_never_reduced (out, level + 1);
 592
 593   /* print grammar */
 594   print_grammar (out, level + 1);
 595
 596   if (report_flag & report_itemsets)
 597     new_closure (nritems);
 598   no_reduce_set =  bitset_create (ntokens, BITSET_FIXED);
 599
 600   /* print automaton */
 601   fputc ('\n', out);
 602   xml_puts (out, level + 1, "<automaton>");
 603   for (i = 0; i < nstates; i++)
 604     print_state (out, level + 2, states[i]);
 605   xml_puts (out, level + 1, "</automaton>");
 606
 607   bitset_free (no_reduce_set);
 608   if (report_flag & report_itemsets)
 609     free_closure ();
 610
 611   xml_puts (out, 0, "</bison-xml-report>");
 612
 613   free (escape_bufs[0].ptr);
 614   free (escape_bufs[1].ptr);
 615
 616   xfclose (out);
 617 }