| 1 | # | ||
| 2 | # Module Parse::Yapp::Lalr | ||
| 3 | # | ||
| 4 | # (c) Copyright 1998-2001 Francois Desarmenien, all rights reserved. | ||
| 5 | # (see the pod text in Parse::Yapp module for use and distribution rights) | ||
| 6 | # | ||
| 7 | package Parse::Yapp::Lalr; | ||
| 8 | @ISA=qw( Parse::Yapp::Grammar ); | ||
| 9 | |||
| 10 | require 5.004; | ||
| 11 | |||
| 12 | use Parse::Yapp::Grammar; | ||
| 13 | |||
| 14 | =for nobody | ||
| 15 | |||
| 16 | Parse::Yapp::Compile Object Structure: | ||
| 17 | -------------------------------------- | ||
| 18 | { | ||
| 19 | GRAMMAR => Parse::Yapp::Grammar, | ||
| 20 | STATES => [ { CORE => [ items... ], | ||
| 21 | ACTIONS => { term => action } | ||
| 22 | GOTOS => { nterm => stateno } | ||
| 23 | }... ] | ||
| 24 | CONFLICTS=>{ SOLVED => { stateno => [ ruleno, token, solved ] }, | ||
| 25 | FORCED => { TOTAL => [ nbsr, nbrr ], | ||
| 26 | DETAIL => { stateno => { TOTAL => [ nbsr, nbrr ] } | ||
| 27 | LIST => [ ruleno, token ] | ||
| 28 | } | ||
| 29 | } | ||
| 30 | } | ||
| 31 | } | ||
| 32 | |||
| 33 | 'items' are of form: [ ruleno, dotpos ] | ||
| 34 | 'term' in ACTIONS is '' means default action | ||
| 35 | 'action' may be: | ||
| 36 | undef: explicit error (nonassociativity) | ||
| 37 | 0 : accept | ||
| 38 | >0 : shift and go to state 'action' | ||
| 39 | <0 : reduce using rule -'action' | ||
| 40 | 'solved' may have values of: | ||
| 41 | 'shift' if solved as Shift | ||
| 42 | 'reduce' if solved as Reduce | ||
| 43 | 'error' if solved by discarding both Shift and Reduce (nonassoc) | ||
| 44 | |||
| 45 | SOLVED is a set of states containing Solved conflicts | ||
| 46 | FORCED are forced conflict resolutions | ||
| 47 | |||
| 48 | nbsr and nbrr are number of shift/reduce and reduce/reduce conflicts | ||
| 49 | |||
| 50 | TOTAL is the total number of SR/RR conflicts for the parser | ||
| 51 | |||
| 52 | DETAIL is the detail of conflicts for each state | ||
| 53 | TOTAL is the total number of SR/RR conflicts for a state | ||
| 54 | LIST is the list of discarded reductions (for display purpose only) | ||
| 55 | |||
| 56 | |||
| 57 | =cut | ||
| 58 | |||
| 59 | use strict; | ||
| 60 | |||
| 61 | use Carp; | ||
| 62 | |||
| 63 | ############### | ||
| 64 | # Constructor # | ||
| 65 | ############### | ||
| 66 | sub new { | ||
| 67 | my($class)=shift; | ||
| 68 | |||
| 69 | ref($class) | ||
| 70 | and $class=ref($class); | ||
| 71 | |||
| 72 | my($self)=$class->SUPER::new(@_); | ||
| 73 | $self->_Compile(); | ||
| 74 | bless($self,$class); | ||
| 75 | } | ||
| 76 | ########### | ||
| 77 | # Methods # | ||
| 78 | ########### | ||
| 79 | |||
| 80 | ########################### | ||
| 81 | # Method To View Warnings # | ||
| 82 | ########################### | ||
| 83 | sub Warnings { | ||
| 84 | my($self)=shift; | ||
| 85 | my($text); | ||
| 86 | my($nbsr,$nbrr)=@{$$self{CONFLICTS}{FORCED}{TOTAL}}; | ||
| 87 | |||
| 88 | $text=$self->SUPER::Warnings(); | ||
| 89 | |||
| 90 | $nbsr != $$self{GRAMMAR}{EXPECT} | ||
| 91 | and $text.="$nbsr shift/reduce conflict".($nbsr > 1 ? "s" : ""); | ||
| 92 | |||
| 93 | $nbrr | ||
| 94 | and do { | ||
| 95 | $nbsr | ||
| 96 | and $text.=" and "; | ||
| 97 | $text.="$nbrr reduce/reduce conflict".($nbrr > 1 ? "s" : ""); | ||
| 98 | }; | ||
| 99 | |||
| 100 | ( $nbsr != $$self{GRAMMAR}{EXPECT} | ||
| 101 | or $nbrr) | ||
| 102 | and $text.="\n"; | ||
| 103 | |||
| 104 | $text; | ||
| 105 | } | ||
| 106 | ############################# | ||
| 107 | # Method To View DFA States # | ||
| 108 | ############################# | ||
| 109 | sub ShowDfa { | ||
| 110 | my($self)=shift; | ||
| 111 | my($text); | ||
| 112 | my($grammar,$states)=($$self{GRAMMAR}, $$self{STATES}); | ||
| 113 | |||
| 114 | for my $stateno (0..$#$states) { | ||
| 115 | my(@shifts,@reduces,@errors,$default); | ||
| 116 | |||
| 117 | $text.="State $stateno:\n\n"; | ||
| 118 | |||
| 119 | #Dump Kernel Items | ||
| 120 | for (sort { $$a[0] <=> $$b[0] | ||
| 121 | or $$a[1] <=> $$b[1] } @{$$states[$stateno]{'CORE'}}) { | ||
| 122 | my($ruleno,$pos)=@$_; | ||
| 123 | my($lhs,$rhs)=@{$$grammar{RULES}[$ruleno]}[0,1]; | ||
| 124 | my(@rhscopy)=@$rhs; | ||
| 125 | |||
| 126 | $ruleno | ||
| 127 | or $rhscopy[-1] = '$end'; | ||
| 128 | |||
| 129 | splice(@rhscopy,$pos,0,'.'); | ||
| 130 | $text.= "\t$lhs -> ".join(' ',@rhscopy)."\t(Rule $ruleno)\n"; | ||
| 131 | } | ||
| 132 | |||
| 133 | #Prepare Actions | ||
| 134 | for (keys(%{$$states[$stateno]{ACTIONS}})) { | ||
| 135 | my($term,$action)=($_,$$states[$stateno]{ACTIONS}{$_}); | ||
| 136 | |||
| 137 | $term eq chr(0) | ||
| 138 | and $term = '$end'; | ||
| 139 | |||
| 140 | not defined($action) | ||
| 141 | and do { | ||
| 142 | push(@errors,$term); | ||
| 143 | next; | ||
| 144 | }; | ||
| 145 | |||
| 146 | $action > 0 | ||
| 147 | and do { | ||
| 148 | push(@shifts,[ $term, $action ]); | ||
| 149 | next; | ||
| 150 | }; | ||
| 151 | |||
| 152 | $action = -$action; | ||
| 153 | |||
| 154 | $term | ||
| 155 | or do { | ||
| 156 | $default= [ '$default', $action ]; | ||
| 157 | next; | ||
| 158 | }; | ||
| 159 | |||
| 160 | push(@reduces,[ $term, $action ]); | ||
| 161 | } | ||
| 162 | |||
| 163 | #Dump shifts | ||
| 164 | @shifts | ||
| 165 | and do { | ||
| 166 | $text.="\n"; | ||
| 167 | for (sort { $$a[0] cmp $$b[0] } @shifts) { | ||
| 168 | my($term,$shift)=@$_; | ||
| 169 | |||
| 170 | $text.="\t$term\tshift, and go to state $shift\n"; | ||
| 171 | } | ||
| 172 | }; | ||
| 173 | |||
| 174 | #Dump errors | ||
| 175 | @errors | ||
| 176 | and do { | ||
| 177 | $text.="\n"; | ||
| 178 | for my $term (sort { $a cmp $b } @errors) { | ||
| 179 | $text.="\t$term\terror (nonassociative)\n"; | ||
| 180 | } | ||
| 181 | }; | ||
| 182 | |||
| 183 | #Prepare reduces | ||
| 184 | exists($$self{CONFLICTS}{FORCED}{DETAIL}{$stateno}) | ||
| 185 | and push(@reduces,@{$$self{CONFLICTS}{FORCED}{DETAIL}{$stateno}{LIST}}); | ||
| 186 | |||
| 187 | @reduces=sort { $$a[0] cmp $$b[0] or $$a[1] <=> $$b[1] } @reduces; | ||
| 188 | |||
| 189 | defined($default) | ||
| 190 | and push(@reduces,$default); | ||
| 191 | |||
| 192 | #Dump reduces | ||
| 193 | @reduces | ||
| 194 | and do { | ||
| 195 | $text.="\n"; | ||
| 196 | for (@reduces) { | ||
| 197 | my($term,$ruleno)=@$_; | ||
| 198 | my($discard); | ||
| 199 | |||
| 200 | $ruleno < 0 | ||
| 201 | and do { | ||
| 202 | ++$discard; | ||
| 203 | $ruleno = -$ruleno; | ||
| 204 | }; | ||
| 205 | |||
| 206 | $text.= "\t$term\t".($discard ? "[" : ""); | ||
| 207 | if($ruleno) { | ||
| 208 | $text.= "reduce using rule $ruleno ". | ||
| 209 | "($$grammar{RULES}[$ruleno][0])"; | ||
| 210 | } | ||
| 211 | else { | ||
| 212 | $text.='accept'; | ||
| 213 | } | ||
| 214 | $text.=($discard ? "]" : "")."\n"; | ||
| 215 | } | ||
| 216 | }; | ||
| 217 | |||
| 218 | #Dump gotos | ||
| 219 | exists($$states[$stateno]{GOTOS}) | ||
| 220 | and do { | ||
| 221 | $text.= "\n"; | ||
| 222 | for (keys(%{$$states[$stateno]{GOTOS}})) { | ||
| 223 | $text.= "\t$_\tgo to state $$states[$stateno]{GOTOS}{$_}\n"; | ||
| 224 | } | ||
| 225 | }; | ||
| 226 | |||
| 227 | $text.="\n"; | ||
| 228 | } | ||
| 229 | $text; | ||
| 230 | } | ||
| 231 | |||
| 232 | ###################################### | ||
| 233 | # Method to get summary about parser # | ||
| 234 | ###################################### | ||
| 235 | sub Summary { | ||
| 236 | my($self)=shift; | ||
| 237 | my($text); | ||
| 238 | |||
| 239 | $text=$self->SUPER::Summary(); | ||
| 240 | $text.="Number of states : ". | ||
| 241 | scalar(@{$$self{STATES}})."\n"; | ||
| 242 | $text; | ||
| 243 | } | ||
| 244 | |||
| 245 | ####################################### | ||
| 246 | # Method To Get Infos about conflicts # | ||
| 247 | ####################################### | ||
| 248 | sub Conflicts { | ||
| 249 | my($self)=shift; | ||
| 250 | my($states)=$$self{STATES}; | ||
| 251 | my($conflicts)=$$self{CONFLICTS}; | ||
| 252 | my($text); | ||
| 253 | |||
| 254 | for my $stateno ( sort { $a <=> $b } keys(%{$$conflicts{SOLVED}})) { | ||
| 255 | |||
| 256 | for (@{$$conflicts{SOLVED}{$stateno}}) { | ||
| 257 | my($ruleno,$token,$how)=@$_; | ||
| 258 | |||
| 259 | $token eq chr(0) | ||
| 260 | and $token = '$end'; | ||
| 261 | |||
| 262 | $text.="Conflict in state $stateno between rule ". | ||
| 263 | "$ruleno and token $token resolved as $how.\n"; | ||
| 264 | } | ||
| 265 | }; | ||
| 266 | |||
| 267 | for my $stateno ( sort { $a <=> $b } keys(%{$$conflicts{FORCED}{DETAIL}})) { | ||
| 268 | my($nbsr,$nbrr)=@{$$conflicts{FORCED}{DETAIL}{$stateno}{TOTAL}}; | ||
| 269 | |||
| 270 | $text.="State $stateno contains "; | ||
| 271 | |||
| 272 | $nbsr | ||
| 273 | and $text.="$nbsr shift/reduce conflict". | ||
| 274 | ($nbsr > 1 ? "s" : ""); | ||
| 275 | |||
| 276 | $nbrr | ||
| 277 | and do { | ||
| 278 | $nbsr | ||
| 279 | and $text.=" and "; | ||
| 280 | |||
| 281 | $text.="$nbrr reduce/reduce conflict". | ||
| 282 | ($nbrr > 1 ? "s" : ""); | ||
| 283 | }; | ||
| 284 | $text.="\n"; | ||
| 285 | }; | ||
| 286 | |||
| 287 | $text; | ||
| 288 | } | ||
| 289 | |||
| 290 | ################################# | ||
| 291 | # Method to dump parsing tables # | ||
| 292 | ################################# | ||
| 293 | sub DfaTable { | ||
| 294 | my($self)=shift; | ||
| 295 | my($states)=$$self{STATES}; | ||
| 296 | my($stateno); | ||
| 297 | my($text); | ||
| 298 | |||
| 299 | $text="[\n\t{"; | ||
| 300 | |||
| 301 | $text.=join("\n\t},\n\t{", | ||
| 302 | map { | ||
| 303 | my($state)=$_; | ||
| 304 | my($text); | ||
| 305 | |||
| 306 | $text="#State ".$stateno++."\n\t\t"; | ||
| 307 | |||
| 308 | ( not exists($$state{ACTIONS}{''}) | ||
| 309 | or keys(%{$$state{ACTIONS}}) > 1) | ||
| 310 | and do { | ||
| 311 | |||
| 312 | $text.="ACTIONS => {\n\t\t\t"; | ||
| 313 | |||
| 314 | $text.=join(",\n\t\t\t", | ||
| 315 | map { | ||
| 316 | my($term,$action)=($_,$$state{ACTIONS}{$_}); | ||
| 317 | my($text); | ||
| 318 | |||
| 319 | if(substr($term,0,1) eq "'") { | ||
| 320 | $term=~s/([\@\$\"])/\\$1/g; | ||
| 321 | $term=~s/^'|'$/"/g; | ||
| 322 | } | ||
| 323 | else { | ||
| 324 | $term= $term eq chr(0) | ||
| 325 | ? "''" | ||
| 326 | : "'$term'"; | ||
| 327 | } | ||
| 328 | |||
| 329 | if(defined($action)) { | ||
| 330 | $action=int($action); | ||
| 331 | } | ||
| 332 | else { | ||
| 333 | $action='undef'; | ||
| 334 | } | ||
| 335 | |||
| 336 | "$term => $action"; | ||
| 337 | |||
| 338 | } grep { $_ } keys(%{$$state{ACTIONS}})); | ||
| 339 | |||
| 340 | $text.="\n\t\t}"; | ||
| 341 | }; | ||
| 342 | |||
| 343 | exists($$state{ACTIONS}{''}) | ||
| 344 | and do { | ||
| 345 | keys(%{$$state{ACTIONS}}) > 1 | ||
| 346 | and $text.=",\n\t\t"; | ||
| 347 | |||
| 348 | $text.="DEFAULT => $$state{ACTIONS}{''}"; | ||
| 349 | }; | ||
| 350 | |||
| 351 | exists($$state{GOTOS}) | ||
| 352 | and do { | ||
| 353 | $text.=",\n\t\tGOTOS => {\n\t\t\t"; | ||
| 354 | $text.=join(",\n\t\t\t", | ||
| 355 | map { | ||
| 356 | my($nterm,$stateno)=($_,$$state{GOTOS}{$_}); | ||
| 357 | my($text); | ||
| 358 | |||
| 359 | "'$nterm' => $stateno"; | ||
| 360 | |||
| 361 | } keys(%{$$state{GOTOS}})); | ||
| 362 | $text.="\n\t\t}"; | ||
| 363 | }; | ||
| 364 | |||
| 365 | $text; | ||
| 366 | |||
| 367 | }@$states); | ||
| 368 | |||
| 369 | $text.="\n\t}\n]"; | ||
| 370 | |||
| 371 | $text; | ||
| 372 | |||
| 373 | } | ||
| 374 | |||
| 375 | |||
| 376 | #################################### | ||
| 377 | # Method to build Dfa from Grammar # | ||
| 378 | #################################### | ||
| 379 | sub _Compile { | ||
| 380 | my($self)=shift; | ||
| 381 | my($grammar,$states); | ||
| 382 | |||
| 383 | $grammar=$self->{GRAMMAR}; | ||
| 384 | |||
| 385 | $states = _LR0($grammar); | ||
| 386 | |||
| 387 | $self->{CONFLICTS} = _LALR($grammar,$states); | ||
| 388 | |||
| 389 | $self->{STATES}=$states; | ||
| 390 | } | ||
| 391 | |||
| 392 | ######################### | ||
| 393 | # LR0 States Generation # | ||
| 394 | ######################### | ||
| 395 | # | ||
| 396 | ########################### | ||
| 397 | # General digraph routine # | ||
| 398 | ########################### | ||
| 399 | sub _Digraph { | ||
| 400 | my($rel,$F)=@_; | ||
| 401 | my(%N,@S); | ||
| 402 | my($infinity)=(~(1<<31)); | ||
| 403 | my($Traverse); | ||
| 404 | |||
| 405 | $Traverse = sub { | ||
| 406 | my($x,$d)=@_; | ||
| 407 | my($y); | ||
| 408 | |||
| 409 | push(@S,$x); | ||
| 410 | $N{$x}=$d; | ||
| 411 | |||
| 412 | exists($$rel{$x}) | ||
| 413 | and do { | ||
| 414 | for $y (keys(%{$$rel{$x}})) { | ||
| 415 | exists($N{$y}) | ||
| 416 | or &$Traverse($y,$d+1); | ||
| 417 | |||
| 418 | $N{$y} < $N{$x} | ||
| 419 | and $N{$x} = $N{$y}; | ||
| 420 | |||
| 421 | $$F{$x}|=$$F{$y}; | ||
| 422 | } | ||
| 423 | }; | ||
| 424 | |||
| 425 | $N{$x} == $d | ||
| 426 | and do { | ||
| 427 | for(;;) { | ||
| 428 | $y=pop(@S); | ||
| 429 | $N{$y}=$infinity; | ||
| 430 | $y eq $x | ||
| 431 | and last; | ||
| 432 | $$F{$y}=$$F{$x}; | ||
| 433 | } | ||
| 434 | }; | ||
| 435 | }; | ||
| 436 | |||
| 437 | for (keys(%$rel)) { | ||
| 438 | exists($N{$_}) | ||
| 439 | or &$Traverse($_,1); | ||
| 440 | } | ||
| 441 | } | ||
| 442 | ####################### | ||
| 443 | # Generate LR0 states # | ||
| 444 | ####################### | ||
| 445 | =for nobody | ||
| 446 | Formula used for closures: | ||
| 447 | |||
| 448 | CLOSE(A) = DCLOSE(A) u U (CLOSE(B) | A close B) | ||
| 449 | |||
| 450 | where: | ||
| 451 | |||
| 452 | DCLOSE(A) = { [ A -> alpha ] in P } | ||
| 453 | |||
| 454 | A close B iff [ A -> B gamma ] in P | ||
| 455 | |||
| 456 | =cut | ||
| 457 | sub _SetClosures { | ||
| 458 | my($grammar)=@_; | ||
| 459 | my($rel,$closures); | ||
| 460 | |||
| 461 | for my $symbol (keys(%{$$grammar{NTERM}})) { | ||
| 462 | $closures->{$symbol}=pack('b'.@{$$grammar{RULES}}); | ||
| 463 | |||
| 464 | for my $ruleno (@{$$grammar{NTERM}{$symbol}}) { | ||
| 465 | my($rhs)=$$grammar{RULES}[$ruleno][1]; | ||
| 466 | |||
| 467 | vec($closures->{$symbol},$ruleno,1)=1; | ||
| 468 | |||
| 469 | @$rhs > 0 | ||
| 470 | and exists($$grammar{NTERM}{$$rhs[0]}) | ||
| 471 | and ++$rel->{$symbol}{$$rhs[0]}; | ||
| 472 | } | ||
| 473 | } | ||
| 474 | _Digraph($rel,$closures); | ||
| 475 | |||
| 476 | $closures | ||
| 477 | } | ||
| 478 | |||
| 479 | sub _Closures { | ||
| 480 | my($grammar,$core,$closures)=@_; | ||
| 481 | my($ruleset)=pack('b'.@{$$grammar{RULES}}); | ||
| 482 | |||
| 483 | for (@$core) { | ||
| 484 | my($ruleno,$pos)=@$_; | ||
| 485 | my($rhs)=$$grammar{RULES}[$ruleno][1]; | ||
| 486 | |||
| 487 | $pos < @$rhs | ||
| 488 | and exists($closures->{$$rhs[$pos]}) | ||
| 489 | and $ruleset|=$closures->{$$rhs[$pos]}; | ||
| 490 | } | ||
| 491 | [ @$core, map { [ $_, 0 ] } | ||
| 492 | grep { vec($ruleset,$_,1) } | ||
| 493 | 0..$#{$$grammar{RULES}} ]; | ||
| 494 | } | ||
| 495 | |||
| 496 | sub _Transitions { | ||
| 497 | my($grammar,$cores,$closures,$states,$stateno)=@_; | ||
| 498 | my($core)=$$states[$stateno]{'CORE'}; | ||
| 499 | my(%transitions); | ||
| 500 | |||
| 501 | for (@{_Closures($grammar,$core,$closures)}) { | ||
| 502 | my($ruleno,$pos)=@$_; | ||
| 503 | my($rhs)=$$grammar{RULES}[$ruleno][1]; | ||
| 504 | |||
| 505 | $pos == @$rhs | ||
| 506 | and do { | ||
| 507 | push(@{$$states[$stateno]{ACTIONS}{''}},$ruleno); | ||
| 508 | next; | ||
| 509 | }; | ||
| 510 | push(@{$transitions{$$rhs[$pos]}},[ $ruleno, $pos+1 ]); | ||
| 511 | } | ||
| 512 | |||
| 513 | for (keys(%transitions)) { | ||
| 514 | my($symbol,$core)=($_,$transitions{$_}); | ||
| 515 | my($corekey)=join(',',map { join('.',@$_) } | ||
| 516 | sort { $$a[0] <=> $$b[0] | ||
| 517 | or $$a[1] <=> $$b[1] } | ||
| 518 | @$core); | ||
| 519 | my($tostateno); | ||
| 520 | |||
| 521 | exists($cores->{$corekey}) | ||
| 522 | or do { | ||
| 523 | push(@$states,{ 'CORE' => $core }); | ||
| 524 | $cores->{$corekey}=$#$states; | ||
| 525 | }; | ||
| 526 | |||
| 527 | $tostateno=$cores->{$corekey}; | ||
| 528 | push(@{$$states[$tostateno]{FROM}},$stateno); | ||
| 529 | |||
| 530 | exists($$grammar{TERM}{$_}) | ||
| 531 | and do { | ||
| 532 | $$states[$stateno]{ACTIONS}{$_} = [ $tostateno ]; | ||
| 533 | next; | ||
| 534 | }; | ||
| 535 | $$states[$stateno]{GOTOS}{$_} = $tostateno; | ||
| 536 | } | ||
| 537 | } | ||
| 538 | |||
| 539 | sub _LR0 { | ||
| 540 | my($grammar)=@_; | ||
| 541 | my($states) = []; | ||
| 542 | my($stateno); | ||
| 543 | my($closures); #$closures={ nterm => ruleset,... } | ||
| 544 | my($cores)={}; # { "itemlist" => stateno, ... } | ||
| 545 | # where "itemlist" has the form: | ||
| 546 | # "ruleno.pos,ruleno.pos" ordered by ruleno,pos | ||
| 547 | |||
| 548 | $closures = _SetClosures($grammar); | ||
| 549 | push(@$states,{ 'CORE' => [ [ 0, 0 ] ] }); | ||
| 550 | for($stateno=0;$stateno<@$states;++$stateno) { | ||
| 551 | _Transitions($grammar,$cores,$closures,$states,$stateno); | ||
| 552 | } | ||
| 553 | |||
| 554 | $states | ||
| 555 | } | ||
| 556 | |||
| 557 | ######################################################### | ||
| 558 | # Add Lookahead tokens where needed to make LALR states # | ||
| 559 | ######################################################### | ||
| 560 | =for nobody | ||
| 561 | Compute First sets for non-terminal using the following formula: | ||
| 562 | |||
| 563 | FIRST(A) = { a in T u { epsilon } | A l a } | ||
| 564 | u | ||
| 565 | U { FIRST(B) | B in V and A l B } | ||
| 566 | |||
| 567 | where: | ||
| 568 | |||
| 569 | A l x iff [ A -> X1 X2 .. Xn x alpha ] in P and Xi =>* epsilon, 1 <= i <= n | ||
| 570 | =cut | ||
| 571 | sub _SetFirst { | ||
| 572 | my($grammar,$termlst,$terminx)=@_; | ||
| 573 | my($rel,$first)=( {}, {} ); | ||
| 574 | |||
| 575 | for my $symbol (keys(%{$$grammar{NTERM}})) { | ||
| 576 | $first->{$symbol}=pack('b'.@$termlst); | ||
| 577 | |||
| 578 | RULE: | ||
| 579 | for my $ruleno (@{$$grammar{NTERM}{$symbol}}) { | ||
| 580 | my($rhs)=$$grammar{RULES}[$ruleno][1]; | ||
| 581 | |||
| 582 | for (@$rhs) { | ||
| 583 | exists($terminx->{$_}) | ||
| 584 | and do { | ||
| 585 | vec($first->{$symbol},$terminx->{$_},1)=1; | ||
| 586 | next RULE; | ||
| 587 | }; | ||
| 588 | ++$rel->{$symbol}{$_}; | ||
| 589 | exists($$grammar{NULLABLE}{$_}) | ||
| 590 | or next RULE; | ||
| 591 | } | ||
| 592 | vec($first->{$symbol},0,1)=1; | ||
| 593 | } | ||
| 594 | } | ||
| 595 | _Digraph($rel,$first); | ||
| 596 | |||
| 597 | $first | ||
| 598 | } | ||
| 599 | |||
| 600 | sub _Preds { | ||
| 601 | my($states,$stateno,$len)=@_; | ||
| 602 | my($queue, $preds); | ||
| 603 | |||
| 604 | $len | ||
| 605 | or return [ $stateno ]; | ||
| 606 | |||
| 607 | $queue=[ [ $stateno, $len ] ]; | ||
| 608 | while(@$queue) { | ||
| 609 | my($pred) = shift(@$queue); | ||
| 610 | my($stateno, $len) = @$pred; | ||
| 611 | |||
| 612 | $len == 1 | ||
| 613 | and do { | ||
| 614 | push(@$preds,@{$states->[$stateno]{FROM}}); | ||
| 615 | next; | ||
| 616 | }; | ||
| 617 | |||
| 618 | push(@$queue, map { [ $_, $len - 1 ] } | ||
| 619 | @{$states->[$stateno]{FROM}}); | ||
| 620 | } | ||
| 621 | |||
| 622 | # Pass @$preds through a hash to ensure unicity | ||
| 623 | [ keys( %{ +{ map { ($_,1) } @$preds } } ) ]; | ||
| 624 | } | ||
| 625 | |||
| 626 | sub _FirstSfx { | ||
| 627 | my($grammar,$firstset,$termlst,$terminx,$ruleno,$pos,$key)=@_; | ||
| 628 | my($first)=pack('b'.@$termlst); | ||
| 629 | my($rhs)=$$grammar{RULES}[$ruleno][1]; | ||
| 630 | |||
| 631 | for (;$pos < @$rhs;++$pos) { | ||
| 632 | exists($terminx->{$$rhs[$pos]}) | ||
| 633 | and do { | ||
| 634 | vec($first,$terminx->{$$rhs[$pos]},1)=1; | ||
| 635 | return($first); | ||
| 636 | }; | ||
| 637 | $first|=$firstset->{$$rhs[$pos]}; | ||
| 638 | |||
| 639 | vec($first,0,1) | ||
| 640 | and vec($first,0,1)=0; | ||
| 641 | |||
| 642 | exists($$grammar{NULLABLE}{$$rhs[$pos]}) | ||
| 643 | or return($first); | ||
| 644 | |||
| 645 | } | ||
| 646 | vec($first,0,1)=1; | ||
| 647 | $first; | ||
| 648 | } | ||
| 649 | |||
| 650 | =for noboby | ||
| 651 | Compute Follow sets using following formula: | ||
| 652 | |||
| 653 | FOLLOW(p,A) = READ(p,A) | ||
| 654 | u | ||
| 655 | U { FOLLOW(q,B) | (p,A) include (q,B) | ||
| 656 | |||
| 657 | where: | ||
| 658 | |||
| 659 | READ(p,A) = U { FIRST(beta) | [ A -> alpha A . beta ] in KERNEL(GOTO(p,A)) | ||
| 660 | } - { epsilon } | ||
| 661 | |||
| 662 | (p,a) include (q,B) iff [ B -> alpha A . beta ] in KERNEL(GOTO(p,A), | ||
| 663 | epsilon in FIRST(beta) and | ||
| 664 | q in PRED(p,alpha) | ||
| 665 | =cut | ||
| 666 | sub _ComputeFollows { | ||
| 667 | my($grammar,$states,$termlst)=@_; | ||
| 668 | my($firstset,$terminx); | ||
| 669 | my($inconsistent, $rel, $follows, $sfx)= ( {}, {}, {}, {} ); | ||
| 670 | |||
| 671 | %$terminx= map { ($termlst->[$_],$_) } 0..$#$termlst; | ||
| 672 | |||
| 673 | $firstset=_SetFirst($grammar,$termlst,$terminx); | ||
| 674 | |||
| 675 | for my $stateno (0..$#$states) { | ||
| 676 | my($state)=$$states[$stateno]; | ||
| 677 | |||
| 678 | exists($$state{ACTIONS}{''}) | ||
| 679 | and ( @{$$state{ACTIONS}{''}} > 1 | ||
| 680 | or keys(%{$$state{ACTIONS}}) > 1 ) | ||
| 681 | and do { | ||
| 682 | ++$inconsistent->{$stateno}; | ||
| 683 | |||
| 684 | for my $ruleno (@{$$state{ACTIONS}{''}}) { | ||
| 685 | my($lhs,$rhs)=@{$$grammar{RULES}[$ruleno]}[0,1]; | ||
| 686 | |||
| 687 | for my $predno (@{_Preds($states,$stateno,scalar(@$rhs))}) { | ||
| 688 | ++$rel->{"$stateno.$ruleno"}{"$predno.$lhs"}; | ||
| 689 | } | ||
| 690 | } | ||
| 691 | }; | ||
| 692 | |||
| 693 | exists($$state{GOTOS}) | ||
| 694 | or next; | ||
| 695 | |||
| 696 | for my $symbol (keys(%{$$state{GOTOS}})) { | ||
| 697 | my($tostate)=$$states[$$state{GOTOS}{$symbol}]; | ||
| 698 | my($goto)="$stateno.$symbol"; | ||
| 699 | |||
| 700 | $follows->{$goto}=pack('b'.@$termlst); | ||
| 701 | |||
| 702 | for my $item (@{$$tostate{'CORE'}}) { | ||
| 703 | my($ruleno,$pos)=@$item; | ||
| 704 | my($key)="$ruleno.$pos"; | ||
| 705 | |||
| 706 | exists($sfx->{$key}) | ||
| 707 | or $sfx->{$key} = _FirstSfx($grammar,$firstset, | ||
| 708 | $termlst,$terminx, | ||
| 709 | $ruleno,$pos,$key); | ||
| 710 | |||
| 711 | $follows->{$goto}|=$sfx->{$key}; | ||
| 712 | |||
| 713 | vec($follows->{$goto},0,1) | ||
| 714 | and do { | ||
| 715 | my($lhs)=$$grammar{RULES}[$ruleno][0]; | ||
| 716 | |||
| 717 | vec($follows->{$goto},0,1)=0; | ||
| 718 | |||
| 719 | for my $predno (@{_Preds($states,$stateno,$pos-1)}) { | ||
| 720 | ++$rel->{$goto}{"$predno.$lhs"}; | ||
| 721 | } | ||
| 722 | }; | ||
| 723 | } | ||
| 724 | } | ||
| 725 | } | ||
| 726 | _Digraph($rel,$follows); | ||
| 727 | |||
| 728 | ($follows,$inconsistent) | ||
| 729 | } | ||
| 730 | |||
| 731 | sub _ComputeLA { | ||
| 732 | my($grammar,$states)=@_; | ||
| 733 | my($termlst)= [ '',keys(%{$$grammar{TERM}}) ]; | ||
| 734 | |||
| 735 | my($follows,$inconsistent) = _ComputeFollows($grammar,$states,$termlst); | ||
| 736 | |||
| 737 | for my $stateno ( keys(%$inconsistent ) ) { | ||
| 738 | my($state)=$$states[$stateno]; | ||
| 739 | my($conflict); | ||
| 740 | |||
| 741 | #NB the sort is VERY important for conflicts resolution order | ||
| 742 | for my $ruleno (sort { $a <=> $b } | ||
| 743 | @{$$state{ACTIONS}{''}}) { | ||
| 744 | for my $term ( map { $termlst->[$_] } grep { | ||
| 745 | vec($follows->{"$stateno.$ruleno"},$_,1) } | ||
| 746 | 0..$#$termlst) { | ||
| 747 | exists($$state{ACTIONS}{$term}) | ||
| 748 | and ++$conflict; | ||
| 749 | push(@{$$state{ACTIONS}{$term}},-$ruleno); | ||
| 750 | } | ||
| 751 | } | ||
| 752 | delete($$state{ACTIONS}{''}); | ||
| 753 | $conflict | ||
| 754 | or delete($inconsistent->{$stateno}); | ||
| 755 | } | ||
| 756 | |||
| 757 | $inconsistent | ||
| 758 | } | ||
| 759 | |||
| 760 | ############################# | ||
| 761 | # Solve remaining conflicts # | ||
| 762 | ############################# | ||
| 763 | |||
| 764 | sub _SolveConflicts { | ||
| 765 | my($grammar,$states,$inconsistent)=@_; | ||
| 766 | my(%rulesprec,$RulePrec); | ||
| 767 | my($conflicts)={ SOLVED => {}, | ||
| 768 | FORCED => { TOTAL => [ 0, 0 ], | ||
| 769 | DETAIL => {} | ||
| 770 | } | ||
| 771 | }; | ||
| 772 | |||
| 773 | $RulePrec = sub { | ||
| 774 | my($ruleno)=@_; | ||
| 775 | my($rhs,$rprec)=@{$$grammar{RULES}[$ruleno]}[1,2]; | ||
| 776 | my($lastterm); | ||
| 777 | |||
| 778 | defined($rprec) | ||
| 779 | and return($rprec); | ||
| 780 | |||
| 781 | exists($rulesprec{$ruleno}) | ||
| 782 | and return($rulesprec{$ruleno}); | ||
| 783 | |||
| 784 | $lastterm=(grep { exists($$grammar{TERM}{$_}) } @$rhs)[-1]; | ||
| 785 | |||
| 786 | defined($lastterm) | ||
| 787 | and ref($$grammar{TERM}{$lastterm}) | ||
| 788 | and do { | ||
| 789 | $rulesprec{$ruleno}=$$grammar{TERM}{$lastterm}[1]; | ||
| 790 | return($rulesprec{$ruleno}); | ||
| 791 | }; | ||
| 792 | |||
| 793 | undef; | ||
| 794 | }; | ||
| 795 | |||
| 796 | for my $stateno (keys(%$inconsistent)) { | ||
| 797 | my($state)=$$states[$stateno]; | ||
| 798 | my($actions)=$$state{ACTIONS}; | ||
| 799 | my($nbsr,$nbrr); | ||
| 800 | |||
| 801 | for my $term ( keys(%$actions) ) { | ||
| 802 | my($act)=$$actions{$term}; | ||
| 803 | |||
| 804 | @$act > 1 | ||
| 805 | or next; | ||
| 806 | |||
| 807 | $$act[0] > 0 | ||
| 808 | and ref($$grammar{TERM}{$term}) | ||
| 809 | and do { | ||
| 810 | my($assoc,$tprec)=@{$$grammar{TERM}{$term}}; | ||
| 811 | my($k,$error); | ||
| 812 | |||
| 813 | for ($k=1;$k<@$act;++$k) { | ||
| 814 | my($ruleno)=-$$act[$k]; | ||
| 815 | my($rprec)=&$RulePrec($ruleno); | ||
| 816 | |||
| 817 | defined($rprec) | ||
| 818 | or next; | ||
| 819 | |||
| 820 | ( $tprec > $rprec | ||
| 821 | or ( $tprec == $rprec and $assoc eq 'RIGHT')) | ||
| 822 | and do { | ||
| 823 | push(@{$$conflicts{SOLVED}{$stateno}}, | ||
| 824 | [ $ruleno, $term, 'shift' ]); | ||
| 825 | splice(@$act,$k--,1); | ||
| 826 | next; | ||
| 827 | }; | ||
| 828 | ( $tprec < $rprec | ||
| 829 | or $assoc eq 'LEFT') | ||
| 830 | and do { | ||
| 831 | push(@{$$conflicts{SOLVED}{$stateno}}, | ||
| 832 | [ $ruleno, $term, 'reduce' ]); | ||
| 833 | $$act[0] > 0 | ||
| 834 | and do { | ||
| 835 | splice(@$act,0,1); | ||
| 836 | --$k; | ||
| 837 | }; | ||
| 838 | next; | ||
| 839 | }; | ||
| 840 | push(@{$$conflicts{SOLVED}{$stateno}}, | ||
| 841 | [ $ruleno, $term, 'error' ]); | ||
| 842 | splice(@$act,$k--,1); | ||
| 843 | $$act[0] > 0 | ||
| 844 | and do { | ||
| 845 | splice(@$act,0,1); | ||
| 846 | ++$error; | ||
| 847 | --$k; | ||
| 848 | }; | ||
| 849 | } | ||
| 850 | $error | ||
| 851 | and unshift(@$act,undef); | ||
| 852 | }; | ||
| 853 | |||
| 854 | @$act > 1 | ||
| 855 | and do { | ||
| 856 | $nbrr += @$act - 2; | ||
| 857 | ($$act[0] > 0 ? $nbsr : $nbrr) += 1; | ||
| 858 | push(@{$$conflicts{FORCED}{DETAIL}{$stateno}{LIST}}, | ||
| 859 | map { [ $term, $_ ] } splice(@$act,1)); | ||
| 860 | }; | ||
| 861 | } | ||
| 862 | |||
| 863 | $nbsr | ||
| 864 | and do { | ||
| 865 | $$conflicts{FORCED}{TOTAL}[0]+=$nbsr; | ||
| 866 | $$conflicts{FORCED}{DETAIL}{$stateno}{TOTAL}[0]+=$nbsr; | ||
| 867 | }; | ||
| 868 | |||
| 869 | $nbrr | ||
| 870 | and do { | ||
| 871 | $$conflicts{FORCED}{TOTAL}[1]+=$nbrr; | ||
| 872 | $$conflicts{FORCED}{DETAIL}{$stateno}{TOTAL}[1]+=$nbrr; | ||
| 873 | }; | ||
| 874 | |||
| 875 | } | ||
| 876 | |||
| 877 | $conflicts | ||
| 878 | } | ||
| 879 | |||
| 880 | ############################### | ||
| 881 | # Make default reduce actions # | ||
| 882 | ############################### | ||
| 883 | sub _SetDefaults { | ||
| 884 | my($states)=@_; | ||
| 885 | |||
| 886 | for my $state (@$states) { | ||
| 887 | my($actions)=$$state{ACTIONS}; | ||
| 888 | my(%reduces,$default,$nodefault); | ||
| 889 | |||
| 890 | exists($$actions{''}) | ||
| 891 | and do { | ||
| 892 | $$actions{''}[0] = -$$actions{''}[0]; | ||
| 893 | ++$nodefault; | ||
| 894 | }; | ||
| 895 | |||
| 896 | #shift error token => no default | ||
| 897 | exists($$actions{error}) | ||
| 898 | and $$actions{error}[0] > 0 | ||
| 899 | and ++$nodefault; | ||
| 900 | |||
| 901 | for my $term (keys(%$actions)) { | ||
| 902 | |||
| 903 | $$actions{$term}=$$actions{$term}[0]; | ||
| 904 | |||
| 905 | ( not defined($$actions{$term}) | ||
| 906 | or $$actions{$term} > 0 | ||
| 907 | or $nodefault) | ||
| 908 | and next; | ||
| 909 | |||
| 910 | push(@{$reduces{$$actions{$term}}},$term); | ||
| 911 | } | ||
| 912 | |||
| 913 | keys(%reduces) > 0 | ||
| 914 | or next; | ||
| 915 | |||
| 916 | $default=( map { $$_[0] } | ||
| 917 | sort { $$b[1] <=> $$a[1] or $$b[0] <=> $$a[0] } | ||
| 918 | map { [ $_, scalar(@{$reduces{$_}}) ] } | ||
| 919 | keys(%reduces))[0]; | ||
| 920 | |||
| 921 | delete(@$actions{ @{$reduces{$default}} }); | ||
| 922 | $$state{ACTIONS}{''}=$default; | ||
| 923 | } | ||
| 924 | } | ||
| 925 | |||
| 926 | sub _LALR { | ||
| 927 | my($grammar,$states) = @_; | ||
| 928 | my($conflicts,$inconsistent); | ||
| 929 | |||
| 930 | $inconsistent = _ComputeLA($grammar,$states); | ||
| 931 | |||
| 932 | $conflicts = _SolveConflicts($grammar,$states,$inconsistent); | ||
| 933 | _SetDefaults($states); | ||
| 934 | |||
| 935 | $conflicts | ||
| 936 | } | ||
| 937 | |||
| 938 | |||
| 939 | 1; |