ragel/xmlcodegen.cpp

   1 /*
   2  *  Copyright 2005-2007 Adrian Thurston <thurston@cs.queensu.ca>
   3  */
   4
   5 /*  This file is part of Ragel.
   6  *
   7  *  Ragel 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 of the License, or
  10  *  (at your option) any later version.
  11  *
  12  *  Ragel 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 Ragel; if not, write to the Free Software
  19  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  20  */
  21
  22
  23 #include "ragel.h"
  24 #include "xmlcodegen.h"
  25 #include "parsedata.h"
  26 #include "fsmgraph.h"
  27 #include <string.h>
  28
  29 using namespace std;
  30
  31 XMLCodeGen::XMLCodeGen( char *fsmName, ParseData *pd, FsmAp *fsm,
  32                 std::ostream &out )
  33 :
  34         fsmName(fsmName),
  35         pd(pd),
  36         fsm(fsm),
  37         out(out),
  38         nextActionTableId(0)
  39 {
  40 }
  41
  42
  43 void XMLCodeGen::writeActionList()
  44 {
  45         /* Determine which actions to write. */
  46         int nextActionId = 0;
  47         for ( ActionList::Iter act = pd->actionList; act.lte(); act++ ) {
  48                 if ( act->numRefs() > 0 || act->numCondRefs > 0 )
  49                         act->actionId = nextActionId++;
  50         }
  51
  52         /* Write the list. */
  53         out << "    <action_list length=\"" << nextActionId << "\">\n";
  54         for ( ActionList::Iter act = pd->actionList; act.lte(); act++ ) {
  55                 if ( act->actionId >= 0 )
  56                         writeAction( act );
  57         }
  58         out << "    </action_list>\n";
  59 }
  60
  61 void XMLCodeGen::writeActionTableList()
  62 {
  63         /* Must first order the action tables based on their id. */
  64         int numTables = nextActionTableId;
  65         RedActionTable **tables = new RedActionTable*[numTables];
  66         for ( ActionTableMap::Iter at = actionTableMap; at.lte(); at++ )
  67                 tables[at->id] = at;
  68
  69         out << "    <action_table_list length=\"" << numTables << "\">\n";
  70         for ( int t = 0; t < numTables; t++ ) {
  71                 out << "      <action_table id=\"" << t << "\" length=\"" <<
  72                                 tables[t]->key.length() << "\">";
  73                 for ( ActionTable::Iter atel = tables[t]->key; atel.lte(); atel++ ) {
  74                         out << atel->value->actionId;
  75                         if ( ! atel.last() )
  76                                 out << " ";
  77                 }
  78                 out << "</action_table>\n";
  79         }
  80         out << "    </action_table_list>\n";
  81
  82         delete[] tables;
  83 }
  84
  85 void XMLCodeGen::reduceActionTables()
  86 {
  87         /* Reduce the actions tables to a set. */
  88         for ( StateList::Iter st = fsm->stateList; st.lte(); st++ ) {
  89                 RedActionTable *actionTable = 0;
  90
  91                 /* Reduce To State Actions. */
  92                 if ( st->toStateActionTable.length() > 0 ) {
  93                         if ( actionTableMap.insert( st->toStateActionTable, &actionTable ) )
  94                                 actionTable->id = nextActionTableId++;
  95                 }
  96
  97                 /* Reduce From State Actions. */
  98                 if ( st->fromStateActionTable.length() > 0 ) {
  99                         if ( actionTableMap.insert( st->fromStateActionTable, &actionTable ) )
 100                                 actionTable->id = nextActionTableId++;
 101                 }
 102
 103                 /* Reduce EOF actions. */
 104                 if ( st->eofActionTable.length() > 0 ) {
 105                         if ( actionTableMap.insert( st->eofActionTable, &actionTable ) )
 106                                 actionTable->id = nextActionTableId++;
 107                 }
 108
 109                 /* Loop the transitions and reduce their actions. */
 110                 for ( TransList::Iter trans = st->outList; trans.lte(); trans++ ) {
 111                         if ( trans->actionTable.length() > 0 ) {
 112                                 if ( actionTableMap.insert( trans->actionTable, &actionTable ) )
 113                                         actionTable->id = nextActionTableId++;
 114                         }
 115                 }
 116         }
 117 }
 118
 119 void XMLCodeGen::appendTrans( TransListVect &outList, Key lowKey,
 120                 Key highKey, TransAp *trans )
 121 {
 122         if ( trans->toState != 0 || trans->actionTable.length() > 0 )
 123                 outList.append( TransEl( lowKey, highKey, trans ) );
 124 }
 125
 126 void XMLCodeGen::writeKey( Key key )
 127 {
 128         if ( keyOps->isSigned )
 129                 out << key.getVal();
 130         else
 131                 out << (unsigned long) key.getVal();
 132 }
 133
 134 void XMLCodeGen::writeTrans( Key lowKey, Key highKey, TransAp *trans )
 135 {
 136         /* First reduce the action. */
 137         RedActionTable *actionTable = 0;
 138         if ( trans->actionTable.length() > 0 )
 139                 actionTable = actionTableMap.find( trans->actionTable );
 140
 141         /* Write the transition. */
 142         out << "        <t>";
 143         writeKey( lowKey );
 144         out << " ";
 145         writeKey( highKey );
 146
 147         if ( trans->toState != 0 )
 148                 out << " " << trans->toState->alg.stateNum;
 149         else
 150                 out << " x";
 151
 152         if ( actionTable != 0 )
 153                 out << " " << actionTable->id;
 154         else
 155                 out << " x";
 156         out << "</t>\n";
 157 }
 158
 159 void XMLCodeGen::writeTransList( StateAp *state )
 160 {
 161         TransListVect outList;
 162
 163         /* If there is only are no ranges the task is simple. */
 164         if ( state->outList.length() > 0 ) {
 165                 /* Loop each source range. */
 166                 for ( TransList::Iter trans = state->outList; trans.lte(); trans++ ) {
 167                         /* Reduce the transition. If it reduced to anything then add it. */
 168                         appendTrans( outList, trans->lowKey, trans->highKey, trans );
 169                 }
 170         }
 171
 172         out << "      <trans_list length=\"" << outList.length() << "\">\n";
 173         for ( TransListVect::Iter tvi = outList; tvi.lte(); tvi++ )
 174                 writeTrans( tvi->lowKey, tvi->highKey, tvi->value );
 175         out << "      </trans_list>\n";
 176 }
 177
 178 void XMLCodeGen::writeText( InlineItem *item )
 179 {
 180         if ( item->prev == 0 || item->prev->type != InlineItem::Text )
 181                 out << "<text>";
 182         xmlEscapeHost( out, item->data, strlen(item->data) );
 183         if ( item->next == 0 || item->next->type != InlineItem::Text )
 184                 out << "</text>";
 185 }
 186
 187 void XMLCodeGen::writeGoto( InlineItem *item )
 188 {
 189         if ( pd->generatingSectionSubset )
 190                 out << "<goto>-1</goto>";
 191         else {
 192                 EntryMapEl *targ = fsm->entryPoints.find( item->nameTarg->id );
 193                 out << "<goto>" << targ->value->alg.stateNum << "</goto>";
 194         }
 195 }
 196
 197 void XMLCodeGen::writeCall( InlineItem *item )
 198 {
 199         if ( pd->generatingSectionSubset )
 200                 out << "<call>-1</call>";
 201         else {
 202                 EntryMapEl *targ = fsm->entryPoints.find( item->nameTarg->id );
 203                 out << "<call>" << targ->value->alg.stateNum << "</call>";
 204         }
 205 }
 206
 207 void XMLCodeGen::writeNext( InlineItem *item )
 208 {
 209         if ( pd->generatingSectionSubset )
 210                 out << "<next>-1</next>";
 211         else {
 212                 EntryMapEl *targ = fsm->entryPoints.find( item->nameTarg->id );
 213                 out << "<next>" << targ->value->alg.stateNum << "</next>";
 214         }
 215 }
 216
 217 void XMLCodeGen::writeGotoExpr( InlineItem *item )
 218 {
 219         out << "<goto_expr>";
 220         writeInlineList( item->children );
 221         out << "</goto_expr>";
 222 }
 223
 224 void XMLCodeGen::writeCallExpr( InlineItem *item )
 225 {
 226         out << "<call_expr>";
 227         writeInlineList( item->children );
 228         out << "</call_expr>";
 229 }
 230
 231 void XMLCodeGen::writeNextExpr( InlineItem *item )
 232 {
 233         out << "<next_expr>";
 234         writeInlineList( item->children );
 235         out << "</next_expr>";
 236 }
 237
 238 void XMLCodeGen::writeEntry( InlineItem *item )
 239 {
 240         if ( pd->generatingSectionSubset )
 241                 out << "<entry>-1</entry>";
 242         else {
 243                 EntryMapEl *targ = fsm->entryPoints.find( item->nameTarg->id );
 244                 out << "<entry>" << targ->value->alg.stateNum << "</entry>";
 245         }
 246 }
 247
 248 void XMLCodeGen::writeActionExec( InlineItem *item )
 249 {
 250         out << "<exec>";
 251         writeInlineList( item->children );
 252         out << "</exec>";
 253 }
 254
 255 void XMLCodeGen::writeLmOnLast( InlineItem *item )
 256 {
 257         out << "<set_tokend>1</set_tokend>";
 258
 259         if ( item->longestMatchPart->action != 0 ) {
 260                 out << "<sub_action>";
 261                 writeInlineList( item->longestMatchPart->action->inlineList );
 262                 out << "</sub_action>";
 263         }
 264 }
 265
 266 void XMLCodeGen::writeLmOnNext( InlineItem *item )
 267 {
 268         out << "<set_tokend>0</set_tokend>";
 269         out << "<hold></hold>";
 270
 271         if ( item->longestMatchPart->action != 0 ) {
 272                 out << "<sub_action>";
 273                 writeInlineList( item->longestMatchPart->action->inlineList );
 274                 out << "</sub_action>";
 275         }
 276 }
 277
 278 void XMLCodeGen::writeLmOnLagBehind( InlineItem *item )
 279 {
 280         out << "<exec><get_tokend></get_tokend></exec>";
 281
 282         if ( item->longestMatchPart->action != 0 ) {
 283                 out << "<sub_action>";
 284                 writeInlineList( item->longestMatchPart->action->inlineList );
 285                 out << "</sub_action>";
 286         }
 287 }
 288
 289 void XMLCodeGen::writeLmSwitch( InlineItem *item )
 290 {
 291
 292         LongestMatch *longestMatch = item->longestMatch;
 293         out << "<lm_switch>\n";
 294
 295         if ( longestMatch->lmSwitchHandlesError ) {
 296                 /* If the switch handles error then we should have also forced the
 297                  * error state. */
 298                 assert( fsm->errState != 0 );
 299
 300                 out << "      <sub_action id=\"0\">";
 301                 out << "<goto>" << fsm->errState->alg.stateNum << "</goto>";
 302                 out << "</sub_action>\n";
 303         }
 304
 305         for ( LmPartList::Iter lmi = *longestMatch->longestMatchList; lmi.lte(); lmi++ ) {
 306                 if ( lmi->inLmSelect && lmi->action != 0 ) {
 307                         /* Open the action. Write it with the context that sets up _p
 308                          * when doing control flow changes from inside the machine. */
 309                         out << "      <sub_action id=\"" << lmi->longestMatchId << "\">";
 310                         out << "<exec><get_tokend></get_tokend></exec>";
 311                         writeInlineList( lmi->action->inlineList );
 312                         out << "</sub_action>\n";
 313                 }
 314         }
 315
 316         out << "    </lm_switch>";
 317 }
 318
 319 void XMLCodeGen::writeInlineList( InlineList *inlineList )
 320 {
 321         for ( InlineList::Iter item = *inlineList; item.lte(); item++ ) {
 322                 switch ( item->type ) {
 323                 case InlineItem::Text:
 324                         writeText( item );
 325                         break;
 326                 case InlineItem::Goto:
 327                         writeGoto( item );
 328                         break;
 329                 case InlineItem::GotoExpr:
 330                         writeGotoExpr( item );
 331                         break;
 332                 case InlineItem::Call:
 333                         writeCall( item );
 334                         break;
 335                 case InlineItem::CallExpr:
 336                         writeCallExpr( item );
 337                         break;
 338                 case InlineItem::Next:
 339                         writeNext( item );
 340                         break;
 341                 case InlineItem::NextExpr:
 342                         writeNextExpr( item );
 343                         break;
 344                 case InlineItem::Break:
 345                         out << "<break></break>";
 346                         break;
 347                 case InlineItem::Ret:
 348                         out << "<ret></ret>";
 349                         break;
 350                 case InlineItem::PChar:
 351                         out << "<pchar></pchar>";
 352                         break;
 353                 case InlineItem::Char:
 354                         out << "<char></char>";
 355                         break;
 356                 case InlineItem::Curs:
 357                         out << "<curs></curs>";
 358                         break;
 359                 case InlineItem::Targs:
 360                         out << "<targs></targs>";
 361                         break;
 362                 case InlineItem::Entry:
 363                         writeEntry( item );
 364                         break;
 365
 366                 case InlineItem::Hold:
 367                         out << "<hold></hold>";
 368                         break;
 369                 case InlineItem::Exec:
 370                         writeActionExec( item );
 371                         break;
 372
 373                 case InlineItem::LmSetActId:
 374                         out << "<set_act>" <<
 375                                         item->longestMatchPart->longestMatchId <<
 376                                         "</set_act>";
 377                         break;
 378                 case InlineItem::LmSetTokEnd:
 379                         out << "<set_tokend>1</set_tokend>";
 380                         break;
 381
 382                 case InlineItem::LmOnLast:
 383                         writeLmOnLast( item );
 384                         break;
 385                 case InlineItem::LmOnNext:
 386                         writeLmOnNext( item );
 387                         break;
 388                 case InlineItem::LmOnLagBehind:
 389                         writeLmOnLagBehind( item );
 390                         break;
 391                 case InlineItem::LmSwitch:
 392                         writeLmSwitch( item );
 393                         break;
 394
 395                 case InlineItem::LmInitAct:
 396                         out << "<init_act></init_act>";
 397                         break;
 398                 case InlineItem::LmInitTokStart:
 399                         out << "<init_tokstart></init_tokstart>";
 400                         break;
 401                 case InlineItem::LmSetTokStart:
 402                         out << "<set_tokstart></set_tokstart>";
 403                         break;
 404                 }
 405         }
 406 }
 407
 408 void XMLCodeGen::writeAction( Action *action )
 409 {
 410         out << "      <action id=\"" << action->actionId << "\"";
 411         if ( action->name != 0 )
 412                 out << " name=\"" << action->name << "\"";
 413         out << " line=\"" << action->loc.line << "\" col=\"" << action->loc.col << "\">";
 414         writeInlineList( action->inlineList );
 415         out << "</action>\n";
 416 }
 417
 418 void xmlEscapeHost( std::ostream &out, char *data, long len )
 419 {
 420         char *end = data + len;
 421         while ( data != end ) {
 422                 switch ( *data ) {
 423                 case '<': out << "&lt;"; break;
 424                 case '>': out << "&gt;"; break;
 425                 case '&': out << "&amp;"; break;
 426                 default: out << *data; break;
 427                 }
 428                 data += 1;
 429         }
 430 }
 431
 432 void XMLCodeGen::writeStateActions( StateAp *state )
 433 {
 434         RedActionTable *toStateActions = 0;
 435         if ( state->toStateActionTable.length() > 0 )
 436                 toStateActions = actionTableMap.find( state->toStateActionTable );
 437
 438         RedActionTable *fromStateActions = 0;
 439         if ( state->fromStateActionTable.length() > 0 )
 440                 fromStateActions = actionTableMap.find( state->fromStateActionTable );
 441
 442         RedActionTable *eofActions = 0;
 443         if ( state->eofActionTable.length() > 0 )
 444                 eofActions = actionTableMap.find( state->eofActionTable );
 445
 446         if ( toStateActions != 0 || fromStateActions != 0 || eofActions != 0 ) {
 447                 out << "      <state_actions>";
 448                 if ( toStateActions != 0 )
 449                         out << toStateActions->id;
 450                 else
 451                         out << "x";
 452
 453                 if ( fromStateActions != 0 )
 454                         out << " " << fromStateActions->id;
 455                 else
 456                         out << " x";
 457
 458                 if ( eofActions != 0 )
 459                         out << " " << eofActions->id;
 460                 else
 461                         out << " x"; out << "</state_actions>\n";
 462         }
 463 }
 464
 465 void XMLCodeGen::writeStateConditions( StateAp *state )
 466 {
 467         if ( state->stateCondList.length() > 0 ) {
 468                 out << "      <cond_list length=\"" << state->stateCondList.length() << "\">\n";
 469                 for ( StateCondList::Iter scdi = state->stateCondList; scdi.lte(); scdi++ ) {
 470                         out << "        <c>";
 471                         writeKey( scdi->lowKey );
 472                         out << " ";
 473                         writeKey( scdi->highKey );
 474                         out << " ";
 475                         out << scdi->condSpace->condSpaceId;
 476                         out << "</c>\n";
 477                 }
 478                 out << "      </cond_list>\n";
 479         }
 480 }
 481
 482 void XMLCodeGen::writeStateList()
 483 {
 484         /* Write the list of states. */
 485         out << "    <state_list length=\"" << fsm->stateList.length() << "\">\n";
 486         for ( StateList::Iter st = fsm->stateList; st.lte(); st++ ) {
 487                 out << "      <state id=\"" << st->alg.stateNum << "\"";
 488                 if ( st->isFinState() )
 489                         out << " final=\"t\"";
 490                 out << ">\n";
 491
 492                 writeStateActions( st );
 493                 writeStateConditions( st );
 494                 writeTransList( st );
 495
 496                 out << "      </state>\n";
 497
 498                 if ( !st.last() )
 499                         out << "\n";
 500         }
 501         out << "    </state_list>\n";
 502 }
 503
 504 bool XMLCodeGen::writeNameInst( NameInst *nameInst )
 505 {
 506         bool written = false;
 507         if ( nameInst->parent != 0 )
 508                 written = writeNameInst( nameInst->parent );
 509
 510         if ( nameInst->name != 0 ) {
 511                 if ( written )
 512                         out << '_';
 513                 out << nameInst->name;
 514                 written = true;
 515         }
 516
 517         return written;
 518 }
 519
 520 void XMLCodeGen::writeEntryPoints()
 521 {
 522         /* List of entry points other than start state. */
 523         if ( fsm->entryPoints.length() > 0 || pd->lmRequiresErrorState ) {
 524                 out << "    <entry_points";
 525                 if ( pd->lmRequiresErrorState )
 526                         out << " error=\"t\"";
 527                 out << ">\n";
 528                 for ( EntryMap::Iter en = fsm->entryPoints; en.lte(); en++ ) {
 529                         /* Get the name instantiation from nameIndex. */
 530                         NameInst *nameInst = pd->nameIndex[en->key];
 531                         StateAp *state = en->value;
 532                         out << "      <entry name=\"";
 533                         writeNameInst( nameInst );
 534                         out << "\">" << state->alg.stateNum << "</entry>\n";
 535                 }
 536                 out << "    </entry_points>\n";
 537         }
 538 }
 539
 540 void XMLCodeGen::writeMachine()
 541 {
 542         /* Open the machine. */
 543         out << "  <machine>\n";
 544
 545         /* Action tables. */
 546         reduceActionTables();
 547
 548         writeActionList();
 549         writeActionTableList();
 550         writeConditions();
 551
 552         /* Start state. */
 553         out << "    <start_state>" << fsm->startState->alg.stateNum <<
 554                         "</start_state>\n";
 555
 556         /* Error state. */
 557         if ( fsm->errState != 0 ) {
 558                 out << "    <error_state>" << fsm->errState->alg.stateNum <<
 559                         "</error_state>\n";
 560         }
 561
 562         writeEntryPoints();
 563         writeStateList();
 564
 565         out << "  </machine>\n";
 566 }
 567
 568
 569 void XMLCodeGen::writeConditions()
 570 {
 571         if ( condData->condSpaceMap.length() > 0 ) {
 572                 long nextCondSpaceId = 0;
 573                 for ( CondSpaceMap::Iter cs = condData->condSpaceMap; cs.lte(); cs++ )
 574                         cs->condSpaceId = nextCondSpaceId++;
 575
 576                 out << "    <cond_space_list length=\"" << condData->condSpaceMap.length() << "\">\n";
 577                 for ( CondSpaceMap::Iter cs = condData->condSpaceMap; cs.lte(); cs++ ) {
 578                         out << "      <cond_space id=\"" << cs->condSpaceId <<
 579                                 "\" length=\"" << cs->condSet.length() << "\">";
 580                         writeKey( cs->baseKey );
 581                         for ( CondSet::Iter csi = cs->condSet; csi.lte(); csi++ )
 582                                 out << " " << (*csi)->actionId;
 583                         out << "</cond_space>\n";
 584                 }
 585                 out << "    </cond_space_list>\n";
 586         }
 587 }
 588
 589 void XMLCodeGen::writeExports()
 590 {
 591         if ( pd->exportList.length() > 0 ) {
 592                 out << "  <exports>\n";
 593                 for ( ExportList::Iter exp = pd->exportList; exp.lte(); exp++ ) {
 594                         out << "    <ex name=\"" << exp->name << "\">";
 595                         writeKey( exp->key );
 596                         out << "</ex>\n";
 597                 }
 598                 out << "  </exports>\n";
 599         }
 600 }
 601
 602 void XMLCodeGen::writeXML()
 603 {
 604         /* Open the definition. */
 605         out << "<ragel_def name=\"" << fsmName << "\">\n";
 606
 607         /* Alphabet type. */
 608         out << "  <alphtype>" << keyOps->alphType->internalName << "</alphtype>\n";
 609
 610         /* Getkey expression. */
 611         if ( pd->getKeyExpr != 0 ) {
 612                 out << "  <getkey>";
 613                 writeInlineList( pd->getKeyExpr );
 614                 out << "</getkey>\n";
 615         }
 616
 617         /* Access expression. */
 618         if ( pd->accessExpr != 0 ) {
 619                 out << "  <access>";
 620                 writeInlineList( pd->accessExpr );
 621                 out << "</access>\n";
 622         }
 623
 624         /*
 625          * Variable expressions.
 626          */
 627
 628         if ( pd->pExpr != 0 ) {
 629                 out << "  <p_expr>";
 630                 writeInlineList( pd->pExpr );
 631                 out << "</p_expr>\n";
 632         }
 633
 634         if ( pd->peExpr != 0 ) {
 635                 out << "  <pe_expr>";
 636                 writeInlineList( pd->peExpr );
 637                 out << "</pe_expr>\n";
 638         }
 639
 640         if ( pd->csExpr != 0 ) {
 641                 out << "  <cs_expr>";
 642                 writeInlineList( pd->csExpr );
 643                 out << "</cs_expr>\n";
 644         }
 645
 646         if ( pd->topExpr != 0 ) {
 647                 out << "  <top_expr>";
 648                 writeInlineList( pd->topExpr );
 649                 out << "</top_expr>\n";
 650         }
 651
 652         if ( pd->stackExpr != 0 ) {
 653                 out << "  <stack_expr>";
 654                 writeInlineList( pd->stackExpr );
 655                 out << "</stack_expr>\n";
 656         }
 657
 658         if ( pd->actExpr != 0 ) {
 659                 out << "  <act_expr>";
 660                 writeInlineList( pd->actExpr );
 661                 out << "</act_expr>\n";
 662         }
 663
 664         if ( pd->tokstartExpr != 0 ) {
 665                 out << "  <tokstart_expr>";
 666                 writeInlineList( pd->tokstartExpr );
 667                 out << "</tokstart_expr>\n";
 668         }
 669
 670         if ( pd->tokendExpr != 0 ) {
 671                 out << "  <tokend_expr>";
 672                 writeInlineList( pd->tokendExpr );
 673                 out << "</tokend_expr>\n";
 674         }
 675
 676         writeExports();
 677
 678         writeMachine();
 679
 680         out <<
 681                 "</ragel_def>\n";
 682 }
 683