tests/regression.at

   1 # Bison Regressions.                               -*- Autotest -*-
   2
   3 # Copyright (C) 2001-2010 Free Software Foundation, Inc.
   4
   5 # This program is free software: you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation, either version 3 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
  17
  18 AT_BANNER([[Regression tests.]])
  19
  20
  21 ## ------------------ ##
  22 ## Trivial grammars.  ##
  23 ## ------------------ ##
  24
  25 AT_SETUP([Trivial grammars])
  26
  27 AT_DATA_GRAMMAR([input.y],
  28 [[%{
  29 void yyerror (char const *);
  30 int yylex (void);
  31 #define YYSTYPE int *
  32 %}
  33
  34 %error-verbose
  35
  36 %%
  37
  38 program: 'x';
  39 ]])
  40
  41 AT_BISON_CHECK([-o input.c input.y])
  42 AT_COMPILE([input.o], [-c input.c])
  43 AT_COMPILE([input.o], [-DYYDEBUG -c input.c])
  44
  45 AT_CLEANUP
  46
  47
  48
  49 ## ----------------- ##
  50 ## YYSTYPE typedef.  ##
  51 ## ----------------- ##
  52
  53 AT_SETUP([YYSTYPE typedef])
  54
  55 AT_DATA_GRAMMAR([input.y],
  56 [[%{
  57 void yyerror (char const *);
  58 int yylex (void);
  59 typedef union { char const *val; } YYSTYPE;
  60 %}
  61
  62 %type <val> program
  63
  64 %%
  65
  66 program: { $$ = ""; };
  67 ]])
  68
  69 AT_BISON_CHECK([-o input.c input.y])
  70 AT_COMPILE([input.o], [-c input.c])
  71
  72 AT_CLEANUP
  73
  74
  75
  76 ## ------------------------------------- ##
  77 ## Early token definitions with --yacc.  ##
  78 ## ------------------------------------- ##
  79
  80
  81 AT_SETUP([Early token definitions with --yacc])
  82
  83 # Found in GCJ: they expect the tokens to be defined before the user
  84 # prologue, so that they can use the token definitions in it.
  85
  86 AT_DATA_GRAMMAR([input.y],
  87 [[%{
  88 void yyerror (const char *s);
  89 int yylex (void);
  90 %}
  91
  92 %union
  93 {
  94   int val;
  95 };
  96 %{
  97 #ifndef MY_TOKEN
  98 # error "MY_TOKEN not defined."
  99 #endif
 100 %}
 101 %token MY_TOKEN
 102 %%
 103 exp: MY_TOKEN;
 104 %%
 105 ]])
 106
 107 AT_BISON_CHECK([-y -o input.c input.y])
 108 AT_COMPILE([input.o], [-c input.c])
 109
 110 AT_CLEANUP
 111
 112
 113
 114 ## ---------------------------------------- ##
 115 ## Early token definitions without --yacc.  ##
 116 ## ---------------------------------------- ##
 117
 118
 119 AT_SETUP([Early token definitions without --yacc])
 120
 121 # Found in GCJ: they expect the tokens to be defined before the user
 122 # prologue, so that they can use the token definitions in it.
 123
 124 AT_DATA_GRAMMAR([input.y],
 125 [[%{
 126 #include <stdio.h>
 127 void yyerror (const char *s);
 128 int yylex (void);
 129 void print_my_token (void);
 130 %}
 131
 132 %union
 133 {
 134   int val;
 135 };
 136 %{
 137 void
 138 print_my_token (void)
 139 {
 140   enum yytokentype my_token = MY_TOKEN;
 141   printf ("%d\n", my_token);
 142 }
 143 %}
 144 %token MY_TOKEN
 145 %%
 146 exp: MY_TOKEN;
 147 %%
 148 ]])
 149
 150 AT_BISON_CHECK([-o input.c input.y])
 151 AT_COMPILE([input.o], [-c input.c])
 152
 153 AT_CLEANUP
 154
 155
 156
 157 ## ---------------- ##
 158 ## Braces parsing.  ##
 159 ## ---------------- ##
 160
 161
 162 AT_SETUP([Braces parsing])
 163
 164 AT_DATA([input.y],
 165 [[/* Bison used to swallow the character after `}'. */
 166
 167 %%
 168 exp: { tests = {{{{{{{{{{}}}}}}}}}}; };
 169 %%
 170 ]])
 171
 172 AT_BISON_CHECK([-v -o input.c input.y])
 173
 174 AT_CHECK([grep 'tests = {{{{{{{{{{}}}}}}}}}};' input.c], 0, [ignore])
 175
 176 AT_CLEANUP
 177
 178
 179 ## ------------------ ##
 180 ## Duplicate string.  ##
 181 ## ------------------ ##
 182
 183
 184 AT_SETUP([Duplicate string])
 185
 186 AT_DATA([input.y],
 187 [[/* `Bison -v' used to dump core when two tokens are defined with the same
 188    string, as LE and GE below. */
 189
 190 %token NUM
 191 %token LE "<="
 192 %token GE "<="
 193
 194 %%
 195 exp: '(' exp ')' | NUM ;
 196 %%
 197 ]])
 198
 199 AT_BISON_CHECK([-v -o input.c input.y], 0, [],
 200 [[input.y:6.8-14: warning: symbol `"<="' used more than once as a literal string
 201 ]])
 202
 203 AT_CLEANUP
 204
 205
 206 ## ------------------- ##
 207 ## Rule Line Numbers.  ##
 208 ## ------------------- ##
 209
 210 AT_SETUP([Rule Line Numbers])
 211
 212 AT_KEYWORDS([report])
 213
 214 AT_DATA([input.y],
 215 [[%%
 216 expr:
 217 'a'
 218
 219 {
 220
 221 }
 222
 223 'b'
 224
 225 {
 226
 227 }
 228
 229 |
 230
 231
 232 {
 233
 234
 235 }
 236
 237 'c'
 238
 239 {
 240
 241 };
 242 ]])
 243
 244 AT_BISON_CHECK([-o input.c -v input.y])
 245
 246 # Check the contents of the report.
 247 AT_CHECK([cat input.output], [],
 248 [[Grammar
 249
 250     0 $accept: expr $end
 251
 252     1 $@1: /* empty */
 253
 254     2 expr: 'a' $@1 'b'
 255
 256     3 $@2: /* empty */
 257
 258     4 expr: $@2 'c'
 259
 260
 261 Terminals, with rules where they appear
 262
 263 $end (0) 0
 264 'a' (97) 2
 265 'b' (98) 2
 266 'c' (99) 4
 267 error (256)
 268
 269
 270 Nonterminals, with rules where they appear
 271
 272 $accept (6)
 273     on left: 0
 274 expr (7)
 275     on left: 2 4, on right: 0
 276 $@1 (8)
 277     on left: 1, on right: 2
 278 $@2 (9)
 279     on left: 3, on right: 4
 280
 281
 282 state 0
 283
 284     0 $accept: . expr $end
 285
 286     'a'  shift, and go to state 1
 287
 288     $default  reduce using rule 3 ($@2)
 289
 290     expr  go to state 2
 291     $@2   go to state 3
 292
 293
 294 state 1
 295
 296     2 expr: 'a' . $@1 'b'
 297
 298     $default  reduce using rule 1 ($@1)
 299
 300     $@1  go to state 4
 301
 302
 303 state 2
 304
 305     0 $accept: expr . $end
 306
 307     $end  shift, and go to state 5
 308
 309
 310 state 3
 311
 312     4 expr: $@2 . 'c'
 313
 314     'c'  shift, and go to state 6
 315
 316
 317 state 4
 318
 319     2 expr: 'a' $@1 . 'b'
 320
 321     'b'  shift, and go to state 7
 322
 323
 324 state 5
 325
 326     0 $accept: expr $end .
 327
 328     $default  accept
 329
 330
 331 state 6
 332
 333     4 expr: $@2 'c' .
 334
 335     $default  reduce using rule 4 (expr)
 336
 337
 338 state 7
 339
 340     2 expr: 'a' $@1 'b' .
 341
 342     $default  reduce using rule 2 (expr)
 343 ]])
 344
 345 AT_CLEANUP
 346
 347
 348
 349 ## ---------------------- ##
 350 ## Mixing %token styles.  ##
 351 ## ---------------------- ##
 352
 353
 354 AT_SETUP([Mixing %token styles])
 355
 356 # Taken from the documentation.
 357 AT_DATA([input.y],
 358 [[%token  <operator>  OR      "||"
 359 %token  <operator>  LE 134  "<="
 360 %left  OR  "<="
 361 %%
 362 exp: ;
 363 %%
 364 ]])
 365
 366 AT_BISON_CHECK([-v -o input.c input.y])
 367
 368 AT_CLEANUP
 369
 370
 371
 372 ## ---------------- ##
 373 ## Invalid inputs.  ##
 374 ## ---------------- ##
 375
 376
 377 AT_SETUP([Invalid inputs])
 378
 379 AT_DATA([input.y],
 380 [[%%
 381 ?
 382 default: 'a' }
 383 %&
 384 %a-does-not-exist
 385 %-
 386 %{
 387 ]])
 388
 389 AT_BISON_CHECK([input.y], [1], [],
 390 [[input.y:2.1: invalid character: `?'
 391 input.y:3.14: invalid character: `}'
 392 input.y:4.1: invalid character: `%'
 393 input.y:4.2: invalid character: `&'
 394 input.y:5.1-17: invalid directive: `%a-does-not-exist'
 395 input.y:6.1-2: invalid directive: `%-'
 396 input.y:7.1-8.0: missing `%}' at end of file
 397 input.y:7.1-8.0: syntax error, unexpected %{...%}
 398 ]])
 399
 400 AT_CLEANUP
 401
 402
 403 AT_SETUP([Invalid inputs with {}])
 404
 405 AT_DATA([input.y],
 406 [[
 407 %destructor
 408 %initial-action
 409 %lex-param
 410 %parse-param
 411 %printer
 412 %union
 413 ]])
 414
 415 AT_BISON_CHECK([input.y], [1], [],
 416 [[input.y:3.1-15: syntax error, unexpected %initial-action, expecting {...}
 417 ]])
 418
 419 AT_CLEANUP
 420
 421
 422
 423 ## ------------------- ##
 424 ## Token definitions.  ##
 425 ## ------------------- ##
 426
 427
 428 AT_SETUP([Token definitions])
 429
 430 # Bison managed, when fed with `%token 'f' "f"' to #define 'f'!
 431 AT_DATA_GRAMMAR([input.y],
 432 [%{
 433 #include <stdlib.h>
 434 #include <stdio.h>
 435 void yyerror (const char *s);
 436 int yylex (void);
 437 %}
 438 [%error-verbose
 439 %token MYEOF 0 "end of file"
 440 %token 'a' "a"
 441 %token B_TOKEN "b"
 442 %token C_TOKEN 'c'
 443 %token 'd' D_TOKEN
 444 %token SPECIAL "\\\'\?\"\a\b\f\n\r\t\v\001\201\x001\x000081??!"
 445 %token SPECIAL "\\\'\?\"\a\b\f\n\r\t\v\001\201\x001\x000081??!"
 446 %%
 447 exp: "a" "\\\'\?\"\a\b\f\n\r\t\v\001\201\x001\x000081??!";
 448 %%
 449 void
 450 yyerror (char const *s)
 451 {
 452   fprintf (stderr, "%s\n", s);
 453 }
 454
 455 int
 456 yylex (void)
 457 {
 458   static int called;
 459   if (called++)
 460     abort ();
 461   return SPECIAL;
 462 }
 463
 464 int
 465 main (void)
 466 {
 467   return yyparse ();
 468 }
 469 ]])
 470
 471 # Checking the warning message guarantees that the trigraph "??!" isn't
 472 # unnecessarily escaped here even though it would need to be if encoded in a
 473 # C-string literal.  Also notice that unnecessary escaping, such as "\?", from
 474 # the user specification is eliminated.
 475 AT_BISON_CHECK([-o input.c input.y], [[0]], [[]],
 476 [[input.y:22.8-14: warning: symbol SPECIAL redeclared
 477 input.y:22.8-63: warning: symbol `"\\'?\"\a\b\f\n\r\t\v\001\201\001\201??!"' used more than once as a literal string
 478 ]])
 479 AT_COMPILE([input])
 480
 481 # Checking the error message here guarantees that yytname, which does contain
 482 # C-string literals, does have the trigraph escaped correctly.  Thus, the
 483 # symbol name reported by the parser is exactly the same as that reported by
 484 # Bison itself.
 485 AT_DATA([experr],
 486 [[syntax error, unexpected "\\'?\"\a\b\f\n\r\t\v\001\201\001\201??!", expecting a
 487 ]])
 488 AT_PARSER_CHECK([./input], 1, [], [experr])
 489 AT_CLEANUP
 490
 491
 492
 493 ## -------------------- ##
 494 ## Characters Escapes.  ##
 495 ## -------------------- ##
 496
 497
 498 AT_SETUP([Characters Escapes])
 499
 500 AT_DATA_GRAMMAR([input.y],
 501 [%{
 502 void yyerror (const char *s);
 503 int yylex (void);
 504 %}
 505 [%%
 506 exp:
 507   '\'' "\'"
 508 | '\"' "\""
 509 | '"'  "'"
 510 ;
 511 ]])
 512 # Pacify font-lock-mode: "
 513
 514 AT_BISON_CHECK([-o input.c input.y])
 515 AT_COMPILE([input.o], [-c input.c])
 516 AT_CLEANUP
 517
 518
 519
 520 ## -------------- ##
 521 ## Web2c Report.  ##
 522 ## -------------- ##
 523
 524 # The generation of the reduction was once wrong in Bison, and made it
 525 # miss some reductions.  In the following test case, the reduction on
 526 # `undef_id_tok' in state 1 was missing.  This is stripped down from
 527 # the actual web2c.y.
 528
 529 AT_SETUP([Web2c Report])
 530
 531 AT_KEYWORDS([report])
 532
 533 AT_DATA([input.y],
 534 [[%token        undef_id_tok const_id_tok
 535
 536 %start CONST_DEC_PART
 537 \f
 538 %%
 539 CONST_DEC_PART:
 540          CONST_DEC_LIST
 541         ;
 542
 543 CONST_DEC_LIST:
 544           CONST_DEC
 545         | CONST_DEC_LIST CONST_DEC
 546         ;
 547
 548 CONST_DEC:
 549           { } undef_id_tok '=' const_id_tok ';'
 550         ;
 551 %%
 552 ]])
 553
 554 AT_BISON_CHECK([-v input.y])
 555 AT_CHECK([cat input.output], 0,
 556 [[Grammar
 557
 558     0 $accept: CONST_DEC_PART $end
 559
 560     1 CONST_DEC_PART: CONST_DEC_LIST
 561
 562     2 CONST_DEC_LIST: CONST_DEC
 563     3               | CONST_DEC_LIST CONST_DEC
 564
 565     4 $@1: /* empty */
 566
 567     5 CONST_DEC: $@1 undef_id_tok '=' const_id_tok ';'
 568
 569
 570 Terminals, with rules where they appear
 571
 572 $end (0) 0
 573 ';' (59) 5
 574 '=' (61) 5
 575 error (256)
 576 undef_id_tok (258) 5
 577 const_id_tok (259) 5
 578
 579
 580 Nonterminals, with rules where they appear
 581
 582 $accept (7)
 583     on left: 0
 584 CONST_DEC_PART (8)
 585     on left: 1, on right: 0
 586 CONST_DEC_LIST (9)
 587     on left: 2 3, on right: 1 3
 588 CONST_DEC (10)
 589     on left: 5, on right: 2 3
 590 $@1 (11)
 591     on left: 4, on right: 5
 592
 593
 594 state 0
 595
 596     0 $accept: . CONST_DEC_PART $end
 597
 598     $default  reduce using rule 4 ($@1)
 599
 600     CONST_DEC_PART  go to state 1
 601     CONST_DEC_LIST  go to state 2
 602     CONST_DEC       go to state 3
 603     $@1             go to state 4
 604
 605
 606 state 1
 607
 608     0 $accept: CONST_DEC_PART . $end
 609
 610     $end  shift, and go to state 5
 611
 612
 613 state 2
 614
 615     1 CONST_DEC_PART: CONST_DEC_LIST .
 616     3 CONST_DEC_LIST: CONST_DEC_LIST . CONST_DEC
 617
 618     undef_id_tok  reduce using rule 4 ($@1)
 619     $default      reduce using rule 1 (CONST_DEC_PART)
 620
 621     CONST_DEC  go to state 6
 622     $@1        go to state 4
 623
 624
 625 state 3
 626
 627     2 CONST_DEC_LIST: CONST_DEC .
 628
 629     $default  reduce using rule 2 (CONST_DEC_LIST)
 630
 631
 632 state 4
 633
 634     5 CONST_DEC: $@1 . undef_id_tok '=' const_id_tok ';'
 635
 636     undef_id_tok  shift, and go to state 7
 637
 638
 639 state 5
 640
 641     0 $accept: CONST_DEC_PART $end .
 642
 643     $default  accept
 644
 645
 646 state 6
 647
 648     3 CONST_DEC_LIST: CONST_DEC_LIST CONST_DEC .
 649
 650     $default  reduce using rule 3 (CONST_DEC_LIST)
 651
 652
 653 state 7
 654
 655     5 CONST_DEC: $@1 undef_id_tok . '=' const_id_tok ';'
 656
 657     '='  shift, and go to state 8
 658
 659
 660 state 8
 661
 662     5 CONST_DEC: $@1 undef_id_tok '=' . const_id_tok ';'
 663
 664     const_id_tok  shift, and go to state 9
 665
 666
 667 state 9
 668
 669     5 CONST_DEC: $@1 undef_id_tok '=' const_id_tok . ';'
 670
 671     ';'  shift, and go to state 10
 672
 673
 674 state 10
 675
 676     5 CONST_DEC: $@1 undef_id_tok '=' const_id_tok ';' .
 677
 678     $default  reduce using rule 5 (CONST_DEC)
 679 ]])
 680
 681 AT_CLEANUP
 682
 683
 684 ## --------------- ##
 685 ## Web2c Actions.  ##
 686 ## --------------- ##
 687
 688 # The generation of the mapping `state -> action' was once wrong in
 689 # extremely specific situations.  web2c.y exhibits this situation.
 690 # Below is a stripped version of the grammar.  It looks like one can
 691 # simplify it further, but just don't: it is tuned to exhibit a bug,
 692 # which disapears when applying sane grammar transformations.
 693 #
 694 # It used to be wrong on yydefact only:
 695 #
 696 # static const yytype_uint8 yydefact[] =
 697 #  {
 698 # -       2,     0,     1,     0,     0,     2,     3,     2,     5,     4,
 699 # +       2,     0,     1,     0,     0,     0,     3,     2,     5,     4,
 700 #         0,     0
 701 #  };
 702 #
 703 # but let's check all the tables.
 704
 705
 706 AT_SETUP([Web2c Actions])
 707
 708 AT_KEYWORDS([report])
 709
 710 AT_DATA([input.y],
 711 [[%%
 712 statement:  struct_stat;
 713 struct_stat:  /* empty. */ | if else;
 714 if: "if" "const" "then" statement;
 715 else: "else" statement;
 716 %%
 717 ]])
 718
 719 AT_BISON_CHECK([-v -o input.c input.y])
 720
 721 # Check only the tables.
 722 [sed -n 's/  *$//;/^static const.*\[\] =/,/^}/p' input.c >tables.c]
 723
 724 AT_CHECK([[cat tables.c]], 0,
 725 [[static const yytype_uint8 yytranslate[] =
 726 {
 727        0,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 728        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 729        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 730        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 731        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 732        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 733        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 734        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 735        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 736        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 737        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 738        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 739        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 740        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 741        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 742        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 743        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 744        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 745        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 746        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 747        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 748        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 749        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 750        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 751        2,     2,     2,     2,     2,     2,     2,     2,     2,     2,
 752        2,     2,     2,     2,     2,     2,     1,     2,     3,     4,
 753        5,     6
 754 };
 755 static const yytype_uint8 yyprhs[] =
 756 {
 757        0,     0,     3,     5,     6,     9,    14
 758 };
 759 static const yytype_int8 yyrhs[] =
 760 {
 761        8,     0,    -1,     9,    -1,    -1,    10,    11,    -1,     3,
 762        4,     5,     8,    -1,     6,     8,    -1
 763 };
 764 static const yytype_uint8 yyrline[] =
 765 {
 766        0,     2,     2,     3,     3,     4,     5
 767 };
 768 static const char *const yytname[] =
 769 {
 770   "$end", "error", "$undefined", "\"if\"", "\"const\"", "\"then\"",
 771   "\"else\"", "$accept", "statement", "struct_stat", "if", "else", 0
 772 };
 773 static const yytype_uint16 yytoknum[] =
 774 {
 775        0,   256,   257,   258,   259,   260,   261
 776 };
 777 static const yytype_uint8 yyr1[] =
 778 {
 779        0,     7,     8,     9,     9,    10,    11
 780 };
 781 static const yytype_uint8 yyr2[] =
 782 {
 783        0,     2,     1,     0,     2,     4,     2
 784 };
 785 static const yytype_uint8 yydefact[] =
 786 {
 787        3,     0,     0,     2,     0,     0,     1,     3,     4,     3,
 788        6,     5
 789 };
 790 static const yytype_int8 yydefgoto[] =
 791 {
 792       -1,     2,     3,     4,     8
 793 };
 794 static const yytype_int8 yypact[] =
 795 {
 796       -2,    -1,     4,    -8,     0,     2,    -8,    -2,    -8,    -2,
 797       -8,    -8
 798 };
 799 static const yytype_int8 yypgoto[] =
 800 {
 801       -8,    -7,    -8,    -8,    -8
 802 };
 803 static const yytype_uint8 yytable[] =
 804 {
 805       10,     1,    11,     5,     6,     0,     7,     9
 806 };
 807 static const yytype_int8 yycheck[] =
 808 {
 809        7,     3,     9,     4,     0,    -1,     6,     5
 810 };
 811 static const yytype_uint8 yystos[] =
 812 {
 813        0,     3,     8,     9,    10,     4,     0,     6,    11,     5,
 814        8,     8
 815 };
 816 ]])
 817
 818 AT_CLEANUP
 819
 820
 821 ## ------------------------- ##
 822 ## yycheck Bound Violation.  ##
 823 ## ------------------------- ##
 824
 825
 826 # _AT_DATA_DANCER_Y(BISON-OPTIONS)
 827 # --------------------------------
 828 # The following grammar, taken from Andrew Suffield's GPL'd implementation
 829 # of DGMTP, the Dancer Generic Message Transport Protocol, used to violate
 830 # yycheck's bounds where issuing a verbose error message.  Keep this test
 831 # so that possible bound checking compilers could check all the skeletons.
 832 m4_define([_AT_DATA_DANCER_Y],
 833 [AT_DATA_GRAMMAR([dancer.y],
 834 [%{
 835 static int yylex (AT_LALR1_CC_IF([int *], [void]));
 836 AT_LALR1_CC_IF([],
 837 [#include <stdlib.h>
 838 #include <stdio.h>
 839 static void yyerror (const char *);])
 840 %}
 841 $1
 842 %token ARROW INVALID NUMBER STRING DATA
 843 %defines
 844 %verbose
 845 %error-verbose
 846 /* Grammar follows */
 847 %%
 848 line: header body
 849    ;
 850
 851 header: '<' from ARROW to '>' type ':'
 852    | '<' ARROW to '>' type ':'
 853    | ARROW to type ':'
 854    | type ':'
 855    | '<' '>'
 856    ;
 857
 858 from: DATA
 859    | STRING
 860    | INVALID
 861    ;
 862
 863 to: DATA
 864    | STRING
 865    | INVALID
 866    ;
 867
 868 type: DATA
 869    | STRING
 870    | INVALID
 871    ;
 872
 873 body: /* empty */
 874    | body member
 875    ;
 876
 877 member: STRING
 878    | DATA
 879    | '+' NUMBER
 880    | '-' NUMBER
 881    | NUMBER
 882    | INVALID
 883    ;
 884 %%
 885 AT_LALR1_CC_IF(
 886 [/* A C++ error reporting function. */
 887 void
 888 yy::parser::error (const location&, const std::string& m)
 889 {
 890   std::cerr << m << std::endl;
 891 }
 892
 893 int
 894 yyparse ()
 895 {
 896   yy::parser parser;
 897 #if YYDEBUG
 898   parser.set_debug_level (YYDEBUG);
 899 #endif
 900   return parser.parse ();
 901 }
 902 ],
 903 [static void
 904 yyerror (const char *s)
 905 {
 906   fprintf (stderr, "%s\n", s);
 907 }])
 908
 909 static int
 910 yylex (AT_LALR1_CC_IF([int *lval], [void]))
 911 [{
 912   static int const tokens[] =
 913     {
 914       ':', -1
 915     };
 916   static size_t toknum;
 917   ]AT_LALR1_CC_IF([*lval = 0; /* Pacify GCC.  */])[
 918   if (! (toknum < sizeof tokens / sizeof *tokens))
 919     abort ();
 920   return tokens[toknum++];
 921 }]
 922
 923 int
 924 main (void)
 925 {
 926   return yyparse ();
 927 }
 928 ])
 929 ])# _AT_DATA_DANCER_Y
 930
 931
 932 # AT_CHECK_DANCER(BISON-OPTIONS)
 933 # ------------------------------
 934 # Generate the grammar, compile it, run it.
 935 m4_define([AT_CHECK_DANCER],
 936 [AT_SETUP([Dancer $1])
 937 AT_BISON_OPTION_PUSHDEFS([$1])
 938 _AT_DATA_DANCER_Y([$1])
 939 AT_BISON_CHECK([-o dancer.c dancer.y])
 940 AT_FULL_COMPILE([dancer])
 941 AT_PARSER_CHECK([./dancer], 1, [],
 942 [syntax error, unexpected ':'
 943 ])
 944 AT_BISON_OPTION_POPDEFS
 945 AT_CLEANUP
 946 ])
 947
 948 AT_CHECK_DANCER()
 949 AT_CHECK_DANCER([%glr-parser])
 950 AT_CHECK_DANCER([%skeleton "lalr1.cc"])
 951
 952
 953 ## ------------------------------------------ ##
 954 ## Diagnostic that expects two alternatives.  ##
 955 ## ------------------------------------------ ##
 956
 957
 958 # _AT_DATA_EXPECT2_Y(BISON-OPTIONS)
 959 # --------------------------------
 960 m4_define([_AT_DATA_EXPECT2_Y],
 961 [AT_DATA_GRAMMAR([expect2.y],
 962 [%{
 963 static int yylex (AT_LALR1_CC_IF([int *], [void]));
 964 AT_LALR1_CC_IF([],
 965 [#include <stdio.h>
 966 #include <stdlib.h>
 967 static void yyerror (const char *);])
 968 %}
 969 $1
 970 %defines
 971 %error-verbose
 972 %token A 1000
 973 %token B
 974
 975 %%
 976 program: /* empty */
 977  | program e ';'
 978  | program error ';';
 979
 980 e: e '+' t | t;
 981 t: A | B;
 982
 983 %%
 984 AT_LALR1_CC_IF(
 985 [/* A C++ error reporting function. */
 986 void
 987 yy::parser::error (const location&, const std::string& m)
 988 {
 989   std::cerr << m << std::endl;
 990 }
 991
 992 int
 993 yyparse ()
 994 {
 995   yy::parser parser;
 996   return parser.parse ();
 997 }
 998 ],
 999 [static void
1000 yyerror (const char *s)
1001 {
1002   fprintf (stderr, "%s\n", s);
1003 }])
1004
1005 static int
1006 yylex (AT_LALR1_CC_IF([int *lval], [void]))
1007 [{
1008   static int const tokens[] =
1009     {
1010       1000, '+', '+', -1
1011     };
1012   static size_t toknum;
1013   ]AT_LALR1_CC_IF([*lval = 0; /* Pacify GCC.  */])[
1014   if (! (toknum < sizeof tokens / sizeof *tokens))
1015     abort ();
1016   return tokens[toknum++];
1017 }]
1018
1019 int
1020 main (void)
1021 {
1022   return yyparse ();
1023 }
1024 ])
1025 ])# _AT_DATA_EXPECT2_Y
1026
1027
1028 # AT_CHECK_EXPECT2(BISON-OPTIONS)
1029 # ------------------------------
1030 # Generate the grammar, compile it, run it.
1031 m4_define([AT_CHECK_EXPECT2],
1032 [AT_SETUP([Expecting two tokens $1])
1033 AT_BISON_OPTION_PUSHDEFS([$1])
1034 _AT_DATA_EXPECT2_Y([$1])
1035 AT_BISON_CHECK([-o expect2.c expect2.y])
1036 AT_FULL_COMPILE([expect2])
1037 AT_PARSER_CHECK([./expect2], 1, [],
1038 [syntax error, unexpected '+', expecting A or B
1039 ])
1040 AT_BISON_OPTION_POPDEFS
1041 AT_CLEANUP
1042 ])
1043
1044 AT_CHECK_EXPECT2()
1045 AT_CHECK_EXPECT2([%glr-parser])
1046 AT_CHECK_EXPECT2([%skeleton "lalr1.cc"])
1047
1048
1049
1050 ## --------------------------------------------- ##
1051 ## Braced code in declaration in rules section.  ##
1052 ## --------------------------------------------- ##
1053
1054 AT_SETUP([Braced code in declaration in rules section])
1055
1056 # Bison once mistook braced code in a declaration in the rules section to be a
1057 # rule action.
1058
1059 AT_DATA_GRAMMAR([input.y],
1060 [[%{
1061 #include <stdio.h>
1062 static void yyerror (char const *msg);
1063 static int yylex (void);
1064 %}
1065
1066 %error-verbose
1067
1068 %%
1069
1070 start:
1071   {
1072     printf ("Bison would once convert this action to a midrule because of the"
1073             " subsequent braced code.\n");
1074   }
1075   ;
1076
1077 %destructor { fprintf (stderr, "DESTRUCTOR\n"); } 'a';
1078 %printer { fprintf (yyoutput, "PRINTER"); } 'a';
1079
1080 %%
1081
1082 static void
1083 yyerror (char const *msg)
1084 {
1085   fprintf (stderr, "%s\n", msg);
1086 }
1087
1088 static int
1089 yylex (void)
1090 {
1091   return 'a';
1092 }
1093
1094 int
1095 main (void)
1096 {
1097   yydebug = 1;
1098   return !yyparse ();
1099 }
1100 ]])
1101
1102 AT_BISON_CHECK([-t -o input.c input.y])
1103 AT_COMPILE([input])
1104 AT_PARSER_CHECK([./input], 0,
1105 [[Bison would once convert this action to a midrule because of the subsequent braced code.
1106 ]],
1107 [[Starting parse
1108 Entering state 0
1109 Reducing stack by rule 1 (line 20):
1110 -> $$ = nterm start ()
1111 Stack now 0
1112 Entering state 1
1113 Reading a token: Next token is token 'a' (PRINTER)
1114 syntax error, unexpected 'a', expecting $end
1115 Error: popping nterm start ()
1116 Stack now 0
1117 Cleanup: discarding lookahead token 'a' (PRINTER)
1118 DESTRUCTOR
1119 Stack now 0
1120 ]])
1121
1122 AT_CLEANUP
1123
1124
1125
1126 ## --------------------------------- ##
1127 ## String alias declared after use.  ##
1128 ## --------------------------------- ##
1129
1130 AT_SETUP([String alias declared after use])
1131
1132 # Bison once incorrectly asserted that the symbol number for either a token or
1133 # its alias was the highest symbol number so far at the point of the alias
1134 # declaration.  That was true unless the declaration appeared after their first
1135 # uses and other tokens appeared in between.
1136
1137 AT_DATA([input.y],
1138 [[%%
1139 start: 'a' "A" 'b';
1140 %token 'a' "A";
1141 ]])
1142
1143 AT_BISON_CHECK([-t -o input.c input.y])
1144
1145 AT_CLEANUP
1146
1147
1148
1149 ## -------------------------------- ##
1150 ## Extra lookahead sets in report.  ##
1151 ## -------------------------------- ##
1152
1153 AT_SETUP([[Extra lookahead sets in report]])
1154
1155 # Bison prints each reduction's lookahead set only next to the associated
1156 # state's one item that (1) is associated with the same rule as the reduction
1157 # and (2) has its dot at the end of its RHS.  Previously, Bison also
1158 # erroneously printed the lookahead set next to all of the state's other items
1159 # associated with the same rule.  This bug affected only the `.output' file and
1160 # not the generated parser source code.
1161
1162 AT_DATA([[input.y]],
1163 [[%%
1164 start: a | 'a' a 'a' ;
1165 a: 'a' ;
1166 ]])
1167
1168 AT_BISON_CHECK([[--report=all input.y]])
1169 AT_CHECK([[sed -n '/^state 1$/,/^state 2$/p' input.output]], [[0]],
1170 [[state 1
1171
1172     2 start: 'a' . a 'a'
1173     3 a: . 'a'
1174     3  | 'a' .  [$end]
1175
1176     'a'  shift, and go to state 4
1177
1178     $default  reduce using rule 3 (a)
1179
1180     a  go to state 5
1181
1182
1183 state 2
1184 ]])
1185
1186 AT_CLEANUP
1187
1188
1189
1190 ## ---------------------------------------- ##
1191 ## Token number in precedence declaration.  ##
1192 ## ---------------------------------------- ##
1193
1194 AT_SETUP([[Token number in precedence declaration]])
1195
1196 # POSIX says token numbers can be declared in %left, %right, and %nonassoc, but
1197 # we lost this in Bison 1.50.
1198
1199 AT_DATA_GRAMMAR([input.y],
1200 [[%{
1201   #include <stdio.h>
1202   void yyerror (char const *);
1203   int yylex (void);
1204 %}
1205
1206 %error-verbose
1207 %left TK1 1 TK2 2 "tok alias" 3
1208
1209 %%
1210
1211 start: TK1 sr_conflict "tok alias" ;
1212
1213 sr_conflict:
1214   TK2
1215   | TK2 "tok alias"
1216   ;
1217
1218 %%
1219
1220 void
1221 yyerror (char const *msg)
1222 {
1223   fprintf (stderr, "%s\n", msg);
1224 }
1225
1226 int
1227 yylex (void)
1228 {
1229   static int const input[] = { 1, 2, 3, 0 };
1230   static int const *inputp = input;
1231   return *inputp++;
1232 }
1233
1234 int
1235 main (void)
1236 {
1237   return yyparse ();
1238 }
1239 ]])
1240
1241 AT_BISON_CHECK([[-o input.c input.y]], [[0]],,
1242 [[input.y:24.5-19: warning: rule useless in parser due to conflicts: sr_conflict: TK2 "tok alias"
1243 ]])
1244 AT_COMPILE([[input]])
1245 AT_PARSER_CHECK([[./input]])
1246
1247 AT_CLEANUP
1248
1249
1250
1251 ## --------------------------- ##
1252 ## parse-gram.y: LALR = IELR.  ##
1253 ## --------------------------- ##
1254
1255 # If parse-gram.y's LALR and IELR parser tables ever begin to differ, we
1256 # need to fix parse-gram.y or start using IELR.
1257
1258 AT_SETUP([[parse-gram.y: LALR = IELR]])
1259
1260 # Avoid differences in synclines by telling bison that the output files
1261 # have the same name.
1262 [cp $abs_top_srcdir/src/parse-gram.y input.y]
1263 AT_BISON_CHECK([[-o input.c -Dlr.type=lalr input.y]])
1264 [mv input.c lalr.c]
1265 AT_BISON_CHECK([[-o input.c -Dlr.type=ielr input.y]])
1266 [mv input.c ielr.c]
1267 AT_CHECK([[diff -u lalr.c ielr.c]])
1268
1269 AT_CLEANUP
1270
1271
1272
1273 ## --------------------------------------- ##
1274 ## %error-verbose and YYSTACK_USE_ALLOCA.  ##
1275 ## --------------------------------------- ##
1276
1277 AT_SETUP([[%error-verbose and YYSTACK_USE_ALLOCA]])
1278
1279 AT_DATA_GRAMMAR([input.y],
1280 [[%code {
1281   #include <stdio.h>
1282   void yyerror (char const *);
1283   int yylex (void);
1284   #define YYSTACK_USE_ALLOCA 1
1285 }
1286
1287 %error-verbose
1288
1289 %%
1290
1291 start: check syntax_error syntax_error ;
1292
1293 check:
1294 {
1295   if (128 < sizeof yymsgbuf)
1296     {
1297       fprintf (stderr,
1298                "The initial size of yymsgbuf in yyparse has increased\n"
1299                "since this test group was last updated.  As a result,\n"
1300                "this test group may no longer manage to induce a\n"
1301                "reallocation of the syntax error message buffer.\n"
1302                "This test group must be adjusted to produce a longer\n"
1303                "error message.\n");
1304       YYABORT;
1305     }
1306 }
1307 ;
1308
1309 // Induce a syntax error message whose total length is more than
1310 // sizeof yymsgbuf in yyparse.  Each token here is 64 bytes.
1311 syntax_error:
1312   "123456789112345678921234567893123456789412345678951234567896123A"
1313 | "123456789112345678921234567893123456789412345678951234567896123B"
1314 | error 'a' 'b' 'c'
1315 ;
1316
1317 %%
1318
1319 void
1320 yyerror (char const *msg)
1321 {
1322   fprintf (stderr, "%s\n", msg);
1323 }
1324
1325 int
1326 yylex (void)
1327 {
1328   /* Induce two syntax error messages (which requires full error
1329      recovery by shifting 3 tokens) in order to detect any loss of the
1330      reallocated buffer.  */
1331   static char const *input = "abc";
1332   return *input++;
1333 }
1334
1335 int
1336 main (void)
1337 {
1338   return yyparse ();
1339 }
1340 ]])
1341
1342 AT_BISON_CHECK([[-o input.c input.y]])
1343 AT_COMPILE([[input]])
1344 AT_PARSER_CHECK([[./input]], [[1]], [],
1345 [[syntax error, unexpected 'a', expecting 123456789112345678921234567893123456789412345678951234567896123A or 123456789112345678921234567893123456789412345678951234567896123B
1346 syntax error, unexpected $end, expecting 123456789112345678921234567893123456789412345678951234567896123A or 123456789112345678921234567893123456789412345678951234567896123B
1347 ]])
1348
1349 AT_CLEANUP
1350
1351
1352
1353 ## ------------------------- ##
1354 ## %error-verbose overflow.  ##
1355 ## ------------------------- ##
1356
1357 # Imagine the case where YYSTACK_ALLOC_MAXIMUM = YYSIZE_MAXIMUM and an
1358 # invocation of yysyntax_error has caused yymsg_alloc to grow to exactly
1359 # YYSTACK_ALLOC_MAXIMUM (perhaps because the normal doubling of size had
1360 # to be clipped to YYSTACK_ALLOC_MAXIMUM).  In an old version of yacc.c,
1361 # a subsequent invocation of yysyntax_error that overflows during its
1362 # size calculation would return YYSIZE_MAXIMUM to yyparse.  Then,
1363 # yyparse would invoke yyerror using the old contents of yymsg.
1364
1365 AT_SETUP([[%error-verbose overflow]])
1366
1367 AT_DATA_GRAMMAR([input.y],
1368 [[%code {
1369   #include <stdio.h>
1370   void yyerror (char const *);
1371   int yylex (void);
1372
1373   /* This prevents this test case from having to induce error messages
1374      large enough to overflow size_t.  */
1375   #define YYSIZE_T unsigned char
1376
1377   /* Bring in malloc so yacc.c doesn't try to provide a malloc prototype
1378      using our YYSIZE_T.  */
1379   #include <stdlib.h>
1380
1381   /* Max depth is usually much smaller than YYSTACK_ALLOC_MAXIMUM, and
1382      we don't want gcc to warn everywhere this constant would be too big
1383      to make sense for our YYSIZE_T.  */
1384   #define YYMAXDEPTH 100
1385 }
1386
1387 %error-verbose
1388
1389 %%
1390
1391 start: syntax_error1 check syntax_error2 ;
1392
1393 // Induce a syntax error message whose total length causes yymsg in
1394 // yyparse to be reallocated to size YYSTACK_ALLOC_MAXIMUM, which
1395 // should be 255.  Each token here is 64 bytes.
1396 syntax_error1:
1397   "123456789112345678921234567893123456789412345678951234567896123A"
1398 | "123456789112345678921234567893123456789412345678951234567896123B"
1399 | "123456789112345678921234567893123456789412345678951234567896123C"
1400 | error 'a' 'b' 'c'
1401 ;
1402
1403 check:
1404 {
1405   if (yymsg_alloc != YYSTACK_ALLOC_MAXIMUM
1406       || YYSTACK_ALLOC_MAXIMUM != YYSIZE_MAXIMUM
1407       || YYSIZE_MAXIMUM != 255)
1408     {
1409       fprintf (stderr,
1410                "The assumptions of this test group are no longer\n"
1411                "valid, so it may no longer catch the error it was\n"
1412                "designed to catch.  Specifically, the following\n"
1413                "values should all be 255:\n\n");
1414       fprintf (stderr, "  yymsg_alloc = %d\n", yymsg_alloc);
1415       fprintf (stderr, "  YYSTACK_ALLOC_MAXIMUM = %d\n",
1416                YYSTACK_ALLOC_MAXIMUM);
1417       fprintf (stderr, "  YYSIZE_MAXIMUM = %d\n", YYSIZE_MAXIMUM);
1418       YYABORT;
1419     }
1420 }
1421 ;
1422
1423 // Now overflow.
1424 syntax_error2:
1425   "123456789112345678921234567893123456789412345678951234567896123A"
1426 | "123456789112345678921234567893123456789412345678951234567896123B"
1427 | "123456789112345678921234567893123456789412345678951234567896123C"
1428 | "123456789112345678921234567893123456789412345678951234567896123D"
1429 | "123456789112345678921234567893123456789412345678951234567896123E"
1430 ;
1431
1432 %%
1433
1434 void
1435 yyerror (char const *msg)
1436 {
1437   fprintf (stderr, "%s\n", msg);
1438 }
1439
1440 int
1441 yylex (void)
1442 {
1443   /* Induce two syntax error messages (which requires full error
1444      recovery by shifting 3 tokens).  */
1445   static char const *input = "abc";
1446   return *input++;
1447 }
1448
1449 int
1450 main (void)
1451 {
1452   /* Push parsers throw away the message buffer between tokens, so skip
1453      this test under maintainer-push-check.  */
1454   if (YYPUSH)
1455     return 77;
1456   return yyparse ();
1457 }
1458 ]])
1459
1460 AT_BISON_CHECK([[-o input.c input.y]])
1461
1462 # gcc warns about tautologies and fallacies involving comparisons for
1463 # unsigned char.  However, it doesn't produce these same warnings for
1464 # size_t and many other types when the warnings would seem to make just
1465 # as much sense.  We ignore the warnings.
1466 [CFLAGS="$NO_WERROR_CFLAGS"]
1467 AT_COMPILE([[input]])
1468
1469 AT_PARSER_CHECK([[./input]], [[2]], [],
1470 [[syntax error, unexpected 'a', expecting 123456789112345678921234567893123456789412345678951234567896123A or 123456789112345678921234567893123456789412345678951234567896123B or 123456789112345678921234567893123456789412345678951234567896123C
1471 syntax error
1472 memory exhausted
1473 ]])
1474
1475 AT_CLEANUP