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         {
 127           reductions *reds = s->reductions;
 128           int red = state_reduction_find (s, &rules[r]);
 129           /* Print item with lookaheads if there are. */
 130           if (reds->lookahead_tokens && red != -1)
 131             {
 132               xml_printf (out, level + 1,
 133                           "<item rule-number=\"%d\" point=\"%d\">",
 134                           rules[r].number, sp1 - sp);
 135               state_rule_lookahead_tokens_print_xml (s, &rules[r],
 136                                                      out, level + 2);
 137               xml_puts (out, level + 1, "</item>");
 138               printed = true;
 139             }
 140         }
 141
 142       if (!printed)
 143         {
 144           xml_printf (out, level + 1,
 145                       "<item rule-number=\"%d\" point=\"%d\"/>",
 146                       rules[r].number,
 147                       sp1 - sp);
 148         }
 149     }
 150   xml_puts (out, level, "</itemset>");
 151 }
 152
 153
 154 /*-----------------------------------------------------------.
 155 | Report the shifts if DISPLAY_SHIFTS_P or the gotos of S on |
 156 | OUT.                                                       |
 157 `-----------------------------------------------------------*/
 158
 159 static void
 160 print_transitions (state *s, FILE *out, int level)
 161 {
 162   transitions *trans = s->transitions;
 163   int n = 0;
 164   int i;
 165
 166   for (i = 0; i < trans->num; i++)
 167     if (!TRANSITION_IS_DISABLED (trans, i))
 168       {
 169         n++;
 170       }
 171
 172   /* Nothing to report. */
 173   if (!n) {
 174     xml_puts (out, level, "<transitions/>");
 175     return;
 176   }
 177
 178   /* Report lookahead tokens and shifts.  */
 179   xml_puts (out, level, "<transitions>");
 180
 181   for (i = 0; i < trans->num; i++)
 182     if (!TRANSITION_IS_DISABLED (trans, i)
 183         && TRANSITION_IS_SHIFT (trans, i))
 184       {
 185         symbol *sym = symbols[TRANSITION_SYMBOL (trans, i)];
 186         char const *tag = sym->tag;
 187         state *s1 = trans->states[i];
 188
 189         xml_printf (out, level + 1,
 190                     "<transition type=\"shift\" symbol=\"%s\" state=\"%d\"/>",
 191                     xml_escape (tag), s1->number);
 192       }
 193
 194   for (i = 0; i < trans->num; i++)
 195     if (!TRANSITION_IS_DISABLED (trans, i)
 196         && !TRANSITION_IS_SHIFT (trans, i))
 197       {
 198         symbol *sym = symbols[TRANSITION_SYMBOL (trans, i)];
 199         char const *tag = sym->tag;
 200         state *s1 = trans->states[i];
 201
 202         xml_printf (out, level + 1,
 203                     "<transition type=\"goto\" symbol=\"%s\" state=\"%d\"/>",
 204                     xml_escape (tag), s1->number);
 205       }
 206
 207   xml_puts (out, level, "</transitions>");
 208 }
 209
 210
 211 /*--------------------------------------------------------.
 212 | Report the explicit errors of S raised from %nonassoc.  |
 213 `--------------------------------------------------------*/
 214
 215 static void
 216 print_errs (FILE *out, int level, state *s)
 217 {
 218   errs *errp = s->errs;
 219   bool count = false;
 220   int i;
 221
 222   for (i = 0; i < errp->num; ++i)
 223     if (errp->symbols[i])
 224       count = true;
 225
 226   /* Nothing to report. */
 227   if (!count) {
 228     xml_puts (out, level, "<errors/>");
 229     return;
 230   }
 231
 232   /* Report lookahead tokens and errors.  */
 233   xml_puts (out, level, "<errors>");
 234   for (i = 0; i < errp->num; ++i)
 235     if (errp->symbols[i])
 236       {
 237         char const *tag = errp->symbols[i]->tag;
 238         xml_printf (out, level + 1,
 239                     "<error symbol=\"%s\">nonassociative</error>",
 240                     xml_escape (tag));
 241       }
 242   xml_puts (out, level, "</errors>");
 243 }
 244
 245
 246 /*-------------------------------------------------------------------------.
 247 | Report a reduction of RULE on LOOKAHEAD_TOKEN (which can be `default').  |
 248 | If not ENABLED, the rule is masked by a shift or a reduce (S/R and       |
 249 | R/R conflicts).                                                          |
 250 `-------------------------------------------------------------------------*/
 251
 252 static void
 253 print_reduction (FILE *out, int level, char const *lookahead_token,
 254                  rule *r, bool enabled)
 255 {
 256   if (r->number)
 257     xml_printf (out, level,
 258                 "<reduction symbol=\"%s\" rule=\"%d\" enabled=\"%s\"/>",
 259                 xml_escape (lookahead_token),
 260                 r->number,
 261                 enabled ? "true" : "false");
 262   else
 263     xml_printf (out, level,
 264                 "<reduction symbol=\"%s\" rule=\"accept\" enabled=\"%s\"/>",
 265                 xml_escape (lookahead_token),
 266                 enabled ? "true" : "false");
 267 }
 268
 269
 270 /*-------------------------------------------.
 271 | Report on OUT the reduction actions of S.  |
 272 `-------------------------------------------*/
 273
 274 static void
 275 print_reductions (FILE *out, int level, state *s)
 276 {
 277   transitions *trans = s->transitions;
 278   reductions *reds = s->reductions;
 279   rule *default_rule = NULL;
 280   int report = false;
 281   int i, j;
 282
 283   if (reds->num == 0) {
 284     xml_puts (out, level, "<reductions/>");
 285     return;
 286   }
 287
 288   if (yydefact[s->number] != 0)
 289     default_rule = &rules[yydefact[s->number] - 1];
 290
 291   bitset_zero (no_reduce_set);
 292   FOR_EACH_SHIFT (trans, i)
 293     bitset_set (no_reduce_set, TRANSITION_SYMBOL (trans, i));
 294   for (i = 0; i < s->errs->num; ++i)
 295     if (s->errs->symbols[i])
 296       bitset_set (no_reduce_set, s->errs->symbols[i]->number);
 297
 298   if (default_rule)
 299     report = true;
 300
 301   if (reds->lookahead_tokens)
 302     for (i = 0; i < ntokens; i++)
 303       {
 304         bool count = bitset_test (no_reduce_set, i);
 305
 306         for (j = 0; j < reds->num; ++j)
 307           if (bitset_test (reds->lookahead_tokens[j], i))
 308             {
 309               if (! count)
 310                 {
 311                   if (reds->rules[j] != default_rule)
 312                     report = true;
 313                   count = true;
 314                 }
 315               else
 316                 {
 317                   report = true;
 318                 }
 319             }
 320       }
 321
 322   /* Nothing to report. */
 323   if (!report) {
 324     xml_puts (out, level, "<reductions/>");
 325     return;
 326   }
 327
 328   xml_puts (out, level, "<reductions>");
 329
 330   /* Report lookahead tokens (or $default) and reductions.  */
 331   if (reds->lookahead_tokens)
 332     for (i = 0; i < ntokens; i++)
 333       {
 334         bool defaulted = false;
 335         bool count = bitset_test (no_reduce_set, i);
 336
 337         for (j = 0; j < reds->num; ++j)
 338           if (bitset_test (reds->lookahead_tokens[j], i))
 339             {
 340               if (! count)
 341                 {
 342                   if (reds->rules[j] != default_rule)
 343                     print_reduction (out, level + 1, symbols[i]->tag,
 344                                      reds->rules[j], true);
 345                   else
 346                     defaulted = true;
 347                   count = true;
 348                 }
 349               else
 350                 {
 351                   if (defaulted)
 352                     print_reduction (out, level + 1, symbols[i]->tag,
 353                                      default_rule, true);
 354                   defaulted = false;
 355                   print_reduction (out, level + 1, symbols[i]->tag,
 356                                    reds->rules[j], false);
 357                 }
 358             }
 359       }
 360
 361   if (default_rule)
 362     print_reduction (out, level + 1,
 363                      "$default", default_rule, true);
 364
 365   xml_puts (out, level, "</reductions>");
 366 }
 367
 368
 369 /*--------------------------------------------------------------.
 370 | Report on OUT all the actions (shifts, gotos, reductions, and |
 371 | explicit erros from %nonassoc) of S.                          |
 372 `--------------------------------------------------------------*/
 373
 374 static void
 375 print_actions (FILE *out, int level, state *s)
 376 {
 377   xml_puts (out, level, "<actions>");
 378   print_transitions (s, out, level + 1);
 379   print_errs (out, level + 1, s);
 380   print_reductions (out, level + 1, s);
 381   xml_puts (out, level, "</actions>");
 382 }
 383
 384
 385 /*----------------------------------.
 386 | Report all the data on S on OUT.  |
 387 `----------------------------------*/
 388
 389 static void
 390 print_state (FILE *out, int level, state *s)
 391 {
 392   fputc ('\n', out);
 393   xml_printf (out, level, "<state number=\"%d\">", s->number);
 394   print_core (out, level + 1, s);
 395   print_actions (out, level + 1, s);
 396   if ((report_flag & report_solved_conflicts) && s->solved_conflicts_xml)
 397     {
 398       xml_puts (out, level + 1, "<solved-conflicts>");
 399       fputs (s->solved_conflicts_xml, out);
 400       xml_puts (out, level + 1, "</solved-conflicts>");
 401     }
 402   else
 403     xml_puts (out, level + 1, "<solved-conflicts/>");
 404   xml_puts (out, level, "</state>");
 405 }
 406
 407
 408 /*-----------------------------------------.
 409 | Print information on the whole grammar.  |
 410 `-----------------------------------------*/
 411
 412 static void
 413 print_grammar (FILE *out, int level)
 414 {
 415   symbol_number i;
 416
 417   fputc ('\n', out);
 418   xml_puts (out, level, "<grammar>");
 419   grammar_rules_print_xml (out, level);
 420
 421   /* Terminals */
 422   xml_puts (out, level + 1, "<terminals>");
 423   for (i = 0; i < max_user_token_number + 1; i++)
 424     if (token_translations[i] != undeftoken->number)
 425       {
 426         char const *tag = symbols[token_translations[i]]->tag;
 427         rule_number r;
 428         item_number *rhsp;
 429
 430         xml_printf (out, level + 2,
 431                     "<terminal type=\"%d\" symbol=\"%s\">",
 432                     i, xml_escape (tag));
 433
 434         for (r = 0; r < nrules; r++)
 435           for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
 436             if (item_number_as_symbol_number (*rhsp) == token_translations[i])
 437               {
 438                 xml_printf (out, level + 3, "<rule>%d</rule>", r);
 439                 break;
 440               }
 441         xml_puts (out, level + 2, "</terminal>");
 442       }
 443   xml_puts (out, level + 1, "</terminals>");
 444
 445   /* Nonterminals */
 446   xml_puts (out, level + 1, "<nonterminals>");
 447   for (i = ntokens; i < nsyms; i++)
 448     {
 449       int left_count = 0, right_count = 0;
 450       rule_number r;
 451       char const *tag = symbols[i]->tag;
 452
 453       for (r = 0; r < nrules; r++)
 454         {
 455           item_number *rhsp;
 456           if (rules[r].lhs->number == i)
 457             left_count++;
 458           for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
 459             if (item_number_as_symbol_number (*rhsp) == i)
 460               {
 461                 right_count++;
 462                 break;
 463               }
 464         }
 465
 466       xml_printf (out, level + 2,
 467                   "<nonterminal type=\"%d\" symbol=\"%s\">",
 468                   i, xml_escape (tag));
 469
 470       if (left_count > 0)
 471         {
 472           xml_puts (out, level + 3, "<left>");
 473           for (r = 0; r < nrules; r++)
 474             {
 475               if (rules[r].lhs->number == i)
 476                 xml_printf (out, level + 4, "<rule>%d</rule>", r);
 477             }
 478           xml_puts (out, level + 3, "</left>");
 479         }
 480
 481       if (right_count > 0)
 482         {
 483           xml_puts (out, level + 3, "<right>");
 484           for (r = 0; r < nrules; r++)
 485             {
 486               item_number *rhsp;
 487               for (rhsp = rules[r].rhs; *rhsp >= 0; rhsp++)
 488                 if (item_number_as_symbol_number (*rhsp) == i)
 489                   {
 490                     xml_printf (out, level + 4, "<rule>%d</rule>", r);
 491                     break;
 492                   }
 493             }
 494           xml_puts (out, level + 3, "</right>");
 495         }
 496
 497       xml_puts (out, level + 2, "</nonterminal>");
 498     }
 499   xml_puts (out, level + 1, "</nonterminals>");
 500   xml_puts (out, level, "</grammar>");
 501 }
 502
 503 void
 504 xml_puts (FILE *out, int level, char const *s)
 505 {
 506   int i;
 507   for (i = 0; i < level; i++)
 508     fputs ("  ", out);
 509   fputs (s, out);
 510   fputc ('\n', out);
 511 }
 512
 513 void
 514 xml_printf (FILE *out, int level, char const *fmt, ...)
 515 {
 516   int i;
 517   va_list arglist;
 518
 519   for (i = 0; i < level; i++)
 520     fputs ("  ", out);
 521
 522   va_start (arglist, fmt);
 523   vfprintf (out, fmt, arglist);
 524   va_end (arglist);
 525
 526   fputc ('\n', out);
 527 }
 528
 529 static char const *
 530 xml_escape_string (struct escape_buf *buf, char const *str)
 531 {
 532   size_t len = strlen (str);
 533   size_t max_expansion = sizeof "&quot;" - 1;
 534   char *p;
 535
 536   if (buf->size <= max_expansion * len)
 537     {
 538       buf->size = max_expansion * len + 1;
 539       buf->ptr = x2realloc (buf->ptr, &buf->size);
 540     }
 541   p = buf->ptr;
 542
 543   for (; *str; str++)
 544     switch (*str)
 545       {
 546       default: *p++ = *str; break;
 547       case '&': p = stpcpy (p, "&amp;" ); break;
 548       case '<': p = stpcpy (p, "&lt;"  ); break;
 549       case '>': p = stpcpy (p, "&gt;"  ); break;
 550       case '"': p = stpcpy (p, "&quot;"); break;
 551       }
 552
 553   *p = '\0';
 554   return buf->ptr;
 555 }
 556
 557 char const *
 558 xml_escape_n (int n, char const *str)
 559 {
 560   return xml_escape_string (escape_bufs + n, str);
 561 }
 562
 563 char const *
 564 xml_escape (char const *str)
 565 {
 566   return xml_escape_n (0, str);
 567 }
 568
 569 void
 570 print_xml (void)
 571 {
 572   state_number i;
 573   int level = 0;
 574
 575   FILE *out = xfopen (spec_xml_file, "w");
 576
 577   fputs ("<?xml version=\"1.0\"?>\n\n", out);
 578   xml_printf (out, level, "<bison-xml-report version=\"%s\">",
 579               xml_escape (VERSION));
 580
 581   fputc ('\n', out);
 582   xml_printf (out, level + 1, "<filename>%s</filename>",
 583               xml_escape (grammar_file));
 584
 585   /* print reductions */
 586   reduce_xml (out, level + 1);
 587
 588   /* print rules never reduced */
 589   print_rules_never_reduced (out, level + 1);
 590
 591   /* print grammar */
 592   print_grammar (out, level + 1);
 593
 594   if (report_flag & report_itemsets)
 595     new_closure (nritems);
 596   no_reduce_set =  bitset_create (ntokens, BITSET_FIXED);
 597
 598   /* print automaton */
 599   fputc ('\n', out);
 600   xml_puts (out, level + 1, "<automaton>");
 601   for (i = 0; i < nstates; i++)
 602     print_state (out, level + 2, states[i]);
 603   xml_puts (out, level + 1, "</automaton>");
 604
 605   bitset_free (no_reduce_set);
 606   if (report_flag & report_itemsets)
 607     free_closure ();
 608
 609   xml_puts (out, 0, "</bison-xml-report>");
 610
 611   free (escape_bufs[0].ptr);
 612   free (escape_bufs[1].ptr);
 613
 614   xfclose (out);
 615 }