2 * Copyright 2001-2008 Adrian Thurston <thurston@complang.org>
5 /* This file is part of Ragel.
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.
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.
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
30 #include "parsedata.h"
31 #include "parsetree.h"
32 #include "mergesort.h"
33 #include "xmlcodegen.h"
35 #include "inputdata.h"
39 char mainMachine[] = "main";
41 void Token::set( const char *str, int len )
44 data = new char[len+1];
45 memcpy( data, str, len );
49 void Token::append( const Token &other )
51 int newLength = length + other.length;
52 char *newString = new char[newLength+1];
53 memcpy( newString, data, length );
54 memcpy( newString + length, other.data, other.length );
55 newString[newLength] = 0;
60 /* Perform minimization after an operation according
61 * to the command line args. */
62 void afterOpMinimize( FsmAp *fsm, bool lastInSeq )
64 /* Switch on the prefered minimization algorithm. */
65 if ( minimizeOpt == MinimizeEveryOp || ( minimizeOpt == MinimizeMostOps && lastInSeq ) ) {
66 /* First clean up the graph. FsmAp operations may leave these
67 * lying around. There should be no dead end states. The subtract
68 * intersection operators are the only places where they may be
69 * created and those operators clean them up. */
70 fsm->removeUnreachableStates();
72 switch ( minimizeLevel ) {
74 fsm->minimizeApproximate();
76 case MinimizePartition1:
77 fsm->minimizePartition1();
79 case MinimizePartition2:
80 fsm->minimizePartition2();
83 fsm->minimizeStable();
89 /* Count the transitions in the fsm by walking the state list. */
90 int countTransitions( FsmAp *fsm )
93 StateAp *state = fsm->stateList.head;
94 while ( state != 0 ) {
95 numTrans += state->outList.length();
101 Key makeFsmKeyHex( char *str, const InputLoc &loc, ParseData *pd )
103 /* Reset errno so we can check for overflow or underflow. In the event of
104 * an error, sets the return val to the upper or lower bound being tested
107 unsigned int size = keyOps->alphType->size;
108 bool unusedBits = size < sizeof(unsigned long);
110 unsigned long ul = strtoul( str, 0, 16 );
112 if ( errno == ERANGE || ( unusedBits && ul >> (size * 8) ) ) {
113 error(loc) << "literal " << str << " overflows the alphabet type" << endl;
114 ul = 1 << (size * 8);
117 if ( unusedBits && keyOps->alphType->isSigned && ul >> (size * 8 - 1) )
118 ul |= ( -1L >> (size*8) ) << (size*8);
120 return Key( (long)ul );
123 Key makeFsmKeyDec( char *str, const InputLoc &loc, ParseData *pd )
125 /* Convert the number to a decimal. First reset errno so we can check
126 * for overflow or underflow. */
128 long long minVal = keyOps->alphType->minVal;
129 long long maxVal = keyOps->alphType->maxVal;
131 long long ll = strtoll( str, 0, 10 );
133 /* Check for underflow. */
134 if ( ( errno == ERANGE && ll < 0 ) || ll < minVal) {
135 error(loc) << "literal " << str << " underflows the alphabet type" << endl;
138 /* Check for overflow. */
139 else if ( ( errno == ERANGE && ll > 0 ) || ll > maxVal ) {
140 error(loc) << "literal " << str << " overflows the alphabet type" << endl;
144 if ( keyOps->alphType->isSigned )
145 return Key( (long)ll );
147 return Key( (unsigned long)ll );
150 /* Make an fsm key in int format (what the fsm graph uses) from an alphabet
151 * number returned by the parser. Validates that the number doesn't overflow
152 * the alphabet type. */
153 Key makeFsmKeyNum( char *str, const InputLoc &loc, ParseData *pd )
155 /* Switch on hex/decimal format. */
156 if ( str[0] == '0' && str[1] == 'x' )
157 return makeFsmKeyHex( str, loc, pd );
159 return makeFsmKeyDec( str, loc, pd );
162 /* Make an fsm int format (what the fsm graph uses) from a single character.
163 * Performs proper conversion depending on signed/unsigned property of the
165 Key makeFsmKeyChar( char c, ParseData *pd )
167 if ( keyOps->isSigned ) {
168 /* Copy from a char type. */
172 /* Copy from an unsigned byte type. */
173 return Key( (unsigned char)c );
177 /* Make an fsm key array in int format (what the fsm graph uses) from a string
178 * of characters. Performs proper conversion depending on signed/unsigned
179 * property of the alphabet. */
180 void makeFsmKeyArray( Key *result, char *data, int len, ParseData *pd )
182 if ( keyOps->isSigned ) {
183 /* Copy from a char star type. */
185 for ( int i = 0; i < len; i++ )
186 result[i] = Key(src[i]);
189 /* Copy from an unsigned byte ptr type. */
190 unsigned char *src = (unsigned char*) data;
191 for ( int i = 0; i < len; i++ )
192 result[i] = Key(src[i]);
196 /* Like makeFsmKeyArray except the result has only unique keys. They ordering
197 * will be changed. */
198 void makeFsmUniqueKeyArray( KeySet &result, char *data, int len,
199 bool caseInsensitive, ParseData *pd )
201 /* Use a transitions list for getting unique keys. */
202 if ( keyOps->isSigned ) {
203 /* Copy from a char star type. */
205 for ( int si = 0; si < len; si++ ) {
207 result.insert( key );
208 if ( caseInsensitive ) {
210 result.insert( key.toUpper() );
211 else if ( key.isUpper() )
212 result.insert( key.toLower() );
217 /* Copy from an unsigned byte ptr type. */
218 unsigned char *src = (unsigned char*) data;
219 for ( int si = 0; si < len; si++ ) {
221 result.insert( key );
222 if ( caseInsensitive ) {
224 result.insert( key.toUpper() );
225 else if ( key.isUpper() )
226 result.insert( key.toLower() );
232 FsmAp *dotFsm( ParseData *pd )
234 FsmAp *retFsm = new FsmAp();
235 retFsm->rangeFsm( keyOps->minKey, keyOps->maxKey );
239 FsmAp *dotStarFsm( ParseData *pd )
241 FsmAp *retFsm = new FsmAp();
242 retFsm->rangeStarFsm( keyOps->minKey, keyOps->maxKey );
246 /* Make a builtin type. Depends on the signed nature of the alphabet type. */
247 FsmAp *makeBuiltin( BuiltinMachine builtin, ParseData *pd )
249 /* FsmAp created to return. */
251 bool isSigned = keyOps->isSigned;
255 /* All characters. */
256 retFsm = dotFsm( pd );
260 /* Ascii characters 0 to 127. */
261 retFsm = new FsmAp();
262 retFsm->rangeFsm( 0, 127 );
266 /* Ascii extended characters. This is the full byte range. Dependent
267 * on signed, vs no signed. If the alphabet is one byte then just use
270 retFsm = new FsmAp();
271 retFsm->rangeFsm( -128, 127 );
274 retFsm = new FsmAp();
275 retFsm->rangeFsm( 0, 255 );
280 /* Alpha [A-Za-z]. */
281 FsmAp *upper = new FsmAp(), *lower = new FsmAp();
282 upper->rangeFsm( 'A', 'Z' );
283 lower->rangeFsm( 'a', 'z' );
284 upper->unionOp( lower );
285 upper->minimizePartition2();
291 retFsm = new FsmAp();
292 retFsm->rangeFsm( '0', '9' );
296 /* Alpha numerics [0-9A-Za-z]. */
297 FsmAp *digit = new FsmAp(), *lower = new FsmAp();
298 FsmAp *upper = new FsmAp();
299 digit->rangeFsm( '0', '9' );
300 upper->rangeFsm( 'A', 'Z' );
301 lower->rangeFsm( 'a', 'z' );
302 digit->unionOp( upper );
303 digit->unionOp( lower );
304 digit->minimizePartition2();
309 /* Lower case characters. */
310 retFsm = new FsmAp();
311 retFsm->rangeFsm( 'a', 'z' );
315 /* Upper case characters. */
316 retFsm = new FsmAp();
317 retFsm->rangeFsm( 'A', 'Z' );
321 /* Control characters. */
322 FsmAp *cntrl = new FsmAp();
323 FsmAp *highChar = new FsmAp();
324 cntrl->rangeFsm( 0, 31 );
325 highChar->concatFsm( 127 );
326 cntrl->unionOp( highChar );
327 cntrl->minimizePartition2();
332 /* Graphical ascii characters [!-~]. */
333 retFsm = new FsmAp();
334 retFsm->rangeFsm( '!', '~' );
338 /* Printable characters. Same as graph except includes space. */
339 retFsm = new FsmAp();
340 retFsm->rangeFsm( ' ', '~' );
345 FsmAp *range1 = new FsmAp();
346 FsmAp *range2 = new FsmAp();
347 FsmAp *range3 = new FsmAp();
348 FsmAp *range4 = new FsmAp();
349 range1->rangeFsm( '!', '/' );
350 range2->rangeFsm( ':', '@' );
351 range3->rangeFsm( '[', '`' );
352 range4->rangeFsm( '{', '~' );
353 range1->unionOp( range2 );
354 range1->unionOp( range3 );
355 range1->unionOp( range4 );
356 range1->minimizePartition2();
361 /* Whitespace: [\t\v\f\n\r ]. */
362 FsmAp *cntrl = new FsmAp();
363 FsmAp *space = new FsmAp();
364 cntrl->rangeFsm( '\t', '\r' );
365 space->concatFsm( ' ' );
366 cntrl->unionOp( space );
367 cntrl->minimizePartition2();
372 /* Hex digits [0-9A-Fa-f]. */
373 FsmAp *digit = new FsmAp();
374 FsmAp *upper = new FsmAp();
375 FsmAp *lower = new FsmAp();
376 digit->rangeFsm( '0', '9' );
377 upper->rangeFsm( 'A', 'F' );
378 lower->rangeFsm( 'a', 'f' );
379 digit->unionOp( upper );
380 digit->unionOp( lower );
381 digit->minimizePartition2();
386 retFsm = new FsmAp();
391 retFsm = new FsmAp();
399 /* Check if this name inst or any name inst below is referenced. */
400 bool NameInst::anyRefsRec()
405 /* Recurse on children until true. */
406 for ( NameVect::Iter ch = childVect; ch.lte(); ch++ ) {
407 if ( (*ch)->anyRefsRec() )
418 /* Initialize the structure that will collect info during the parse of a
420 ParseData::ParseData( const char *fileName, char *sectionName,
421 const InputLoc §ionLoc )
424 generatingSectionSubset(false),
426 /* 0 is reserved for global error actions. */
448 sectionName(sectionName),
449 sectionLoc(sectionLoc),
454 nextEpsilonResolvedLink(0),
455 nextLongestMatchId(1),
456 lmRequiresErrorState(false),
459 /* Initialize the dictionary of graphs. This is our symbol table. The
460 * initialization needs to be done on construction which happens at the
461 * beginning of a machine spec so any assignment operators can reference
466 /* Clean up the data collected during a parse. */
467 ParseData::~ParseData()
469 /* Delete all the nodes in the action list. Will cause all the
470 * string data that represents the actions to be deallocated. */
474 /* Make a name id in the current name instantiation scope if it is not
476 NameInst *ParseData::addNameInst( const InputLoc &loc, const char *data, bool isLabel )
478 /* Create the name instantitaion object and insert it. */
479 NameInst *newNameInst = new NameInst( loc, curNameInst, data, nextNameId++, isLabel );
480 curNameInst->childVect.append( newNameInst );
482 curNameInst->children.insertMulti( data, newNameInst );
486 void ParseData::initNameWalk()
488 curNameInst = rootName;
492 void ParseData::initExportsNameWalk()
494 curNameInst = exportsRootName;
498 /* Goes into the next child scope. The number of the child is already set up.
499 * We need this for the syncronous name tree and parse tree walk to work
500 * properly. It is reset on entry into a scope and advanced on poping of a
501 * scope. A call to enterNameScope should be accompanied by a corresponding
503 NameFrame ParseData::enterNameScope( bool isLocal, int numScopes )
505 /* Save off the current data. */
507 retFrame.prevNameInst = curNameInst;
508 retFrame.prevNameChild = curNameChild;
509 retFrame.prevLocalScope = localNameScope;
511 /* Enter into the new name scope. */
512 for ( int i = 0; i < numScopes; i++ ) {
513 curNameInst = curNameInst->childVect[curNameChild];
518 localNameScope = curNameInst;
523 /* Return from a child scope to a parent. The parent info must be specified as
524 * an argument and is obtained from the corresponding call to enterNameScope.
526 void ParseData::popNameScope( const NameFrame &frame )
528 /* Pop the name scope. */
529 curNameInst = frame.prevNameInst;
530 curNameChild = frame.prevNameChild+1;
531 localNameScope = frame.prevLocalScope;
534 void ParseData::resetNameScope( const NameFrame &frame )
536 /* Pop the name scope. */
537 curNameInst = frame.prevNameInst;
538 curNameChild = frame.prevNameChild;
539 localNameScope = frame.prevLocalScope;
543 void ParseData::unsetObsoleteEntries( FsmAp *graph )
545 /* Loop the reference names and increment the usage. Names that are no
546 * longer needed will be unset in graph. */
547 for ( NameVect::Iter ref = curNameInst->referencedNames; ref.lte(); ref++ ) {
549 NameInst *name = *ref;
552 /* If the name is no longer needed unset its corresponding entry. */
553 if ( name->numUses == name->numRefs ) {
554 assert( graph->entryPoints.find( name->id ) != 0 );
555 graph->unsetEntry( name->id );
556 assert( graph->entryPoints.find( name->id ) == 0 );
561 NameSet ParseData::resolvePart( NameInst *refFrom, const char *data, bool recLabelsOnly )
563 /* Queue needed for breadth-first search, load it with the start node. */
564 NameInstList nameQueue;
565 nameQueue.append( refFrom );
568 while ( nameQueue.length() > 0 ) {
569 /* Pull the next from location off the queue. */
570 NameInst *from = nameQueue.detachFirst();
572 /* Look for the name. */
573 NameMapEl *low, *high;
574 if ( from->children.findMulti( data, low, high ) ) {
575 /* Record all instances of the name. */
576 for ( ; low <= high; low++ )
577 result.insert( low->value );
580 /* Name not there, do breadth-first operation of appending all
581 * childrent to the processing queue. */
582 for ( NameVect::Iter name = from->childVect; name.lte(); name++ ) {
583 if ( !recLabelsOnly || (*name)->isLabel )
584 nameQueue.append( *name );
588 /* Queue exhausted and name never found. */
592 void ParseData::resolveFrom( NameSet &result, NameInst *refFrom,
593 const NameRef &nameRef, int namePos )
595 /* Look for the name in the owning scope of the factor with aug. */
596 NameSet partResult = resolvePart( refFrom, nameRef[namePos], false );
598 /* If there are more parts to the name then continue on. */
599 if ( ++namePos < nameRef.length() ) {
600 /* There are more components to the name, search using all the part
601 * results as the base. */
602 for ( NameSet::Iter name = partResult; name.lte(); name++ )
603 resolveFrom( result, *name, nameRef, namePos );
606 /* This is the last component, append the part results to the final
608 result.insert( partResult );
612 /* Write out a name reference. */
613 ostream &operator<<( ostream &out, const NameRef &nameRef )
616 if ( nameRef[pos] == 0 ) {
620 out << nameRef[pos++];
621 for ( ; pos < nameRef.length(); pos++ )
622 out << "::" << nameRef[pos];
626 ostream &operator<<( ostream &out, const NameInst &nameInst )
628 /* Count the number fully qualified name parts. */
630 NameInst *curParent = nameInst.parent;
631 while ( curParent != 0 ) {
633 curParent = curParent->parent;
636 /* Make an array and fill it in. */
637 curParent = nameInst.parent;
638 NameInst **parents = new NameInst*[numParents];
639 for ( int p = numParents-1; p >= 0; p-- ) {
640 parents[p] = curParent;
641 curParent = curParent->parent;
644 /* Write the parents out, skip the root. */
645 for ( int p = 1; p < numParents; p++ )
646 out << "::" << ( parents[p]->name != 0 ? parents[p]->name : "<ANON>" );
648 /* Write the name and cleanup. */
649 out << "::" << ( nameInst.name != 0 ? nameInst.name : "<ANON>" );
654 struct CmpNameInstLoc
656 static int compare( const NameInst *ni1, const NameInst *ni2 )
658 if ( ni1->loc.line < ni2->loc.line )
660 else if ( ni1->loc.line > ni2->loc.line )
662 else if ( ni1->loc.col < ni2->loc.col )
664 else if ( ni1->loc.col > ni2->loc.col )
670 void errorStateLabels( const NameSet &resolved )
672 MergeSort<NameInst*, CmpNameInstLoc> mergeSort;
673 mergeSort.sort( resolved.data, resolved.length() );
674 for ( NameSet::Iter res = resolved; res.lte(); res++ )
675 error((*res)->loc) << " -> " << **res << endl;
679 NameInst *ParseData::resolveStateRef( const NameRef &nameRef, InputLoc &loc, Action *action )
681 NameInst *nameInst = 0;
683 /* Do the local search if the name is not strictly a root level name
685 if ( nameRef[0] != 0 ) {
686 /* If the action is referenced, resolve all of them. */
687 if ( action != 0 && action->actionRefs.length() > 0 ) {
688 /* Look for the name in all referencing scopes. */
690 for ( ActionRefs::Iter actRef = action->actionRefs; actRef.lte(); actRef++ )
691 resolveFrom( resolved, *actRef, nameRef, 0 );
693 if ( resolved.length() > 0 ) {
694 /* Take the first one. */
695 nameInst = resolved[0];
696 if ( resolved.length() > 1 ) {
697 /* Complain about the multiple references. */
698 error(loc) << "state reference " << nameRef <<
699 " resolves to multiple entry points" << endl;
700 errorStateLabels( resolved );
706 /* If not found in the local scope, look in global. */
707 if ( nameInst == 0 ) {
709 int fromPos = nameRef[0] != 0 ? 0 : 1;
710 resolveFrom( resolved, rootName, nameRef, fromPos );
712 if ( resolved.length() > 0 ) {
713 /* Take the first. */
714 nameInst = resolved[0];
715 if ( resolved.length() > 1 ) {
716 /* Complain about the multiple references. */
717 error(loc) << "state reference " << nameRef <<
718 " resolves to multiple entry points" << endl;
719 errorStateLabels( resolved );
724 if ( nameInst == 0 ) {
725 /* If not found then complain. */
726 error(loc) << "could not resolve state reference " << nameRef << endl;
731 void ParseData::resolveNameRefs( InlineList *inlineList, Action *action )
733 for ( InlineList::Iter item = *inlineList; item.lte(); item++ ) {
734 switch ( item->type ) {
735 case InlineItem::Entry: case InlineItem::Goto:
736 case InlineItem::Call: case InlineItem::Next: {
737 /* Resolve, pass action for local search. */
738 NameInst *target = resolveStateRef( *item->nameRef, item->loc, action );
740 /* Name lookup error reporting is handled by resolveStateRef. */
742 /* Check if the target goes into a longest match. */
743 NameInst *search = target->parent;
744 while ( search != 0 ) {
745 if ( search->isLongestMatch ) {
746 error(item->loc) << "cannot enter inside a longest "
747 "match construction as an entry point" << endl;
750 search = search->parent;
753 /* Record the reference in the name. This will cause the
754 * entry point to survive to the end of the graph
755 * generating walk. */
756 target->numRefs += 1;
759 item->nameTarg = target;
766 /* Some of the item types may have children. */
767 if ( item->children != 0 )
768 resolveNameRefs( item->children, action );
772 /* Resolve references to labels in actions. */
773 void ParseData::resolveActionNameRefs()
775 for ( ActionList::Iter act = actionList; act.lte(); act++ ) {
776 /* Only care about the actions that are referenced. */
777 if ( act->actionRefs.length() > 0 )
778 resolveNameRefs( act->inlineList, act );
782 /* Walk a name tree starting at from and fill the name index. */
783 void ParseData::fillNameIndex( NameInst *from )
785 /* Fill the value for from in the name index. */
786 nameIndex[from->id] = from;
788 /* Recurse on the implicit final state and then all children. */
789 if ( from->final != 0 )
790 fillNameIndex( from->final );
791 for ( NameVect::Iter name = from->childVect; name.lte(); name++ )
792 fillNameIndex( *name );
795 void ParseData::makeRootNames()
797 /* Create the root name. */
798 rootName = new NameInst( InputLoc(), 0, 0, nextNameId++, false );
799 exportsRootName = new NameInst( InputLoc(), 0, 0, nextNameId++, false );
802 /* Build the name tree and supporting data structures. */
803 void ParseData::makeNameTree( GraphDictEl *dictEl )
805 /* Set up curNameInst for the walk. */
809 /* A start location has been specified. */
810 dictEl->value->makeNameTree( dictEl->loc, this );
813 /* First make the name tree. */
814 for ( GraphList::Iter glel = instanceList; glel.lte(); glel++ ) {
815 /* Recurse on the instance. */
816 glel->value->makeNameTree( glel->loc, this );
820 /* The number of nodes in the tree can now be given by nextNameId */
821 nameIndex = new NameInst*[nextNameId];
822 memset( nameIndex, 0, sizeof(NameInst*)*nextNameId );
823 fillNameIndex( rootName );
824 fillNameIndex( exportsRootName );
828 void ParseData::createBuiltin( const char *name, BuiltinMachine builtin )
830 Expression *expression = new Expression( builtin );
831 Join *join = new Join( expression );
832 MachineDef *machineDef = new MachineDef( join );
833 VarDef *varDef = new VarDef( name, machineDef );
834 GraphDictEl *graphDictEl = new GraphDictEl( name, varDef );
835 graphDict.insert( graphDictEl );
838 /* Initialize the graph dict with builtin types. */
839 void ParseData::initGraphDict( )
841 createBuiltin( "any", BT_Any );
842 createBuiltin( "ascii", BT_Ascii );
843 createBuiltin( "extend", BT_Extend );
844 createBuiltin( "alpha", BT_Alpha );
845 createBuiltin( "digit", BT_Digit );
846 createBuiltin( "alnum", BT_Alnum );
847 createBuiltin( "lower", BT_Lower );
848 createBuiltin( "upper", BT_Upper );
849 createBuiltin( "cntrl", BT_Cntrl );
850 createBuiltin( "graph", BT_Graph );
851 createBuiltin( "print", BT_Print );
852 createBuiltin( "punct", BT_Punct );
853 createBuiltin( "space", BT_Space );
854 createBuiltin( "xdigit", BT_Xdigit );
855 createBuiltin( "null", BT_Lambda );
856 createBuiltin( "zlen", BT_Lambda );
857 createBuiltin( "empty", BT_Empty );
860 /* Set the alphabet type. If the types are not valid returns false. */
861 bool ParseData::setAlphType( const InputLoc &loc, char *s1, char *s2 )
864 userAlphType = findAlphType( s1, s2 );
866 return userAlphType != 0;
869 /* Set the alphabet type. If the types are not valid returns false. */
870 bool ParseData::setAlphType( const InputLoc &loc, char *s1 )
873 userAlphType = findAlphType( s1 );
875 return userAlphType != 0;
878 bool ParseData::setVariable( char *var, InlineList *inlineList )
882 if ( strcmp( var, "p" ) == 0 )
884 else if ( strcmp( var, "pe" ) == 0 )
886 else if ( strcmp( var, "eof" ) == 0 )
887 eofExpr = inlineList;
888 else if ( strcmp( var, "cs" ) == 0 )
890 else if ( strcmp( var, "data" ) == 0 )
891 dataExpr = inlineList;
892 else if ( strcmp( var, "top" ) == 0 )
893 topExpr = inlineList;
894 else if ( strcmp( var, "stack" ) == 0 )
895 stackExpr = inlineList;
896 else if ( strcmp( var, "act" ) == 0 )
897 actExpr = inlineList;
898 else if ( strcmp( var, "ts" ) == 0 )
899 tokstartExpr = inlineList;
900 else if ( strcmp( var, "te" ) == 0 )
901 tokendExpr = inlineList;
908 /* Initialize the key operators object that will be referenced by all fsms
910 void ParseData::initKeyOps( )
912 /* Signedness and bounds. */
913 HostType *alphType = alphTypeSet ? userAlphType : hostLang->defaultAlphType;
914 thisKeyOps.setAlphType( alphType );
916 if ( lowerNum != 0 ) {
917 /* If ranges are given then interpret the alphabet type. */
918 thisKeyOps.minKey = makeFsmKeyNum( lowerNum, rangeLowLoc, this );
919 thisKeyOps.maxKey = makeFsmKeyNum( upperNum, rangeHighLoc, this );
922 thisCondData.lastCondKey = thisKeyOps.maxKey;
925 void ParseData::printNameInst( NameInst *nameInst, int level )
927 for ( int i = 0; i < level; i++ )
929 cerr << (nameInst->name != 0 ? nameInst->name : "<ANON>") <<
930 " id: " << nameInst->id <<
931 " refs: " << nameInst->numRefs <<
932 " uses: " << nameInst->numUses << endl;
933 for ( NameVect::Iter name = nameInst->childVect; name.lte(); name++ )
934 printNameInst( *name, level+1 );
937 /* Remove duplicates of unique actions from an action table. */
938 void ParseData::removeDups( ActionTable &table )
940 /* Scan through the table looking for unique actions to
941 * remove duplicates of. */
942 for ( int i = 0; i < table.length(); i++ ) {
943 /* Remove any duplicates ahead of i. */
944 for ( int r = i+1; r < table.length(); ) {
945 if ( table[r].value == table[i].value )
953 /* Remove duplicates from action lists. This operates only on transition and
954 * eof action lists and so should be called once all actions have been
955 * transfered to their final resting place. */
956 void ParseData::removeActionDups( FsmAp *graph )
958 /* Loop all states. */
959 for ( StateList::Iter state = graph->stateList; state.lte(); state++ ) {
960 /* Loop all transitions. */
961 for ( TransList::Iter trans = state->outList; trans.lte(); trans++ )
962 removeDups( trans->actionTable );
963 removeDups( state->toStateActionTable );
964 removeDups( state->fromStateActionTable );
965 removeDups( state->eofActionTable );
969 Action *ParseData::newAction( const char *name, InlineList *inlineList )
974 loc.fileName = "NONE";
976 Action *action = new Action( loc, name, inlineList, nextCondId++ );
977 action->actionRefs.append( rootName );
978 actionList.append( action );
982 void ParseData::initLongestMatchData()
984 if ( lmList.length() > 0 ) {
985 /* The initTokStart action resets the token start. */
986 InlineList *il1 = new InlineList;
987 il1->append( new InlineItem( InputLoc(), InlineItem::LmInitTokStart ) );
988 initTokStart = newAction( "initts", il1 );
989 initTokStart->isLmAction = true;
991 /* The initActId action gives act a default value. */
992 InlineList *il4 = new InlineList;
993 il4->append( new InlineItem( InputLoc(), InlineItem::LmInitAct ) );
994 initActId = newAction( "initact", il4 );
995 initActId->isLmAction = true;
997 /* The setTokStart action sets tokstart. */
998 InlineList *il5 = new InlineList;
999 il5->append( new InlineItem( InputLoc(), InlineItem::LmSetTokStart ) );
1000 setTokStart = newAction( "ts", il5 );
1001 setTokStart->isLmAction = true;
1003 /* The setTokEnd action sets tokend. */
1004 InlineList *il3 = new InlineList;
1005 il3->append( new InlineItem( InputLoc(), InlineItem::LmSetTokEnd ) );
1006 setTokEnd = newAction( "te", il3 );
1007 setTokEnd->isLmAction = true;
1009 /* The action will also need an ordering: ahead of all user action
1011 initTokStartOrd = curActionOrd++;
1012 initActIdOrd = curActionOrd++;
1013 setTokStartOrd = curActionOrd++;
1014 setTokEndOrd = curActionOrd++;
1018 /* After building the graph, do some extra processing to ensure the runtime
1019 * data of the longest mactch operators is consistent. */
1020 void ParseData::setLongestMatchData( FsmAp *graph )
1022 if ( lmList.length() > 0 ) {
1023 /* Make sure all entry points (targets of fgoto, fcall, fnext, fentry)
1024 * init the tokstart. */
1025 for ( EntryMap::Iter en = graph->entryPoints; en.lte(); en++ ) {
1026 /* This is run after duplicates are removed, we must guard against
1027 * inserting a duplicate. */
1028 ActionTable &actionTable = en->value->toStateActionTable;
1029 if ( ! actionTable.hasAction( initTokStart ) )
1030 actionTable.setAction( initTokStartOrd, initTokStart );
1033 /* Find the set of states that are the target of transitions with
1034 * actions that have calls. These states will be targeted by fret
1037 for ( StateList::Iter state = graph->stateList; state.lte(); state++ ) {
1038 for ( TransList::Iter trans = state->outList; trans.lte(); trans++ ) {
1039 for ( ActionTable::Iter ati = trans->actionTable; ati.lte(); ati++ ) {
1040 if ( ati->value->anyCall && trans->toState != 0 )
1041 states.insert( trans->toState );
1047 /* Init tokstart upon entering the above collected states. */
1048 for ( StateSet::Iter ps = states; ps.lte(); ps++ ) {
1049 /* This is run after duplicates are removed, we must guard against
1050 * inserting a duplicate. */
1051 ActionTable &actionTable = (*ps)->toStateActionTable;
1052 if ( ! actionTable.hasAction( initTokStart ) )
1053 actionTable.setAction( initTokStartOrd, initTokStart );
1058 /* Make the graph from a graph dict node. Does minimization and state sorting. */
1059 FsmAp *ParseData::makeInstance( GraphDictEl *gdNode )
1061 /* Build the graph from a walk of the parse tree. */
1062 FsmAp *graph = gdNode->value->walk( this );
1064 /* Resolve any labels that point to multiple states. Any labels that are
1065 * still around are referenced only by gotos and calls and they need to be
1066 * made into deterministic entry points. */
1067 graph->deterministicEntry();
1070 * All state construction is now complete.
1073 /* Transfer actions from the out action tables to eof action tables. */
1074 for ( StateSet::Iter state = graph->finStateSet; state.lte(); state++ )
1075 graph->transferOutActions( *state );
1077 /* Transfer global error actions. */
1078 for ( StateList::Iter state = graph->stateList; state.lte(); state++ )
1079 graph->transferErrorActions( state, 0 );
1081 if ( ::wantDupsRemoved )
1082 removeActionDups( graph );
1084 /* Remove unreachable states. There should be no dead end states. The
1085 * subtract and intersection operators are the only places where they may
1086 * be created and those operators clean them up. */
1087 graph->removeUnreachableStates();
1089 /* No more fsm operations are to be done. Action ordering numbers are
1090 * no longer of use and will just hinder minimization. Clear them. */
1091 graph->nullActionKeys();
1093 /* Transition priorities are no longer of use. We can clear them
1094 * because they will just hinder minimization as well. Clear them. */
1095 graph->clearAllPriorities();
1097 if ( minimizeOpt != MinimizeNone ) {
1098 /* Minimize here even if we minimized at every op. Now that function
1099 * keys have been cleared we may get a more minimal fsm. */
1100 switch ( minimizeLevel ) {
1101 case MinimizeApprox:
1102 graph->minimizeApproximate();
1104 case MinimizeStable:
1105 graph->minimizeStable();
1107 case MinimizePartition1:
1108 graph->minimizePartition1();
1110 case MinimizePartition2:
1111 graph->minimizePartition2();
1116 graph->compressTransitions();
1121 void ParseData::printNameTree()
1123 /* Print the name instance map. */
1124 for ( NameVect::Iter name = rootName->childVect; name.lte(); name++ )
1125 printNameInst( *name, 0 );
1127 cerr << "name index:" << endl;
1128 /* Show that the name index is correct. */
1129 for ( int ni = 0; ni < nextNameId; ni++ ) {
1131 const char *name = nameIndex[ni]->name;
1132 cerr << ( name != 0 ? name : "<ANON>" ) << endl;
1136 FsmAp *ParseData::makeSpecific( GraphDictEl *gdNode )
1138 /* Build the name tree and supporting data structures. */
1139 makeNameTree( gdNode );
1141 /* Resove name references from gdNode. */
1143 gdNode->value->resolveNameRefs( this );
1145 /* Do not resolve action references. Since we are not building the entire
1146 * graph there's a good chance that many name references will fail. This
1147 * is okay since generating part of the graph is usually only done when
1148 * inspecting the compiled machine. */
1150 /* Same story for extern entry point references. */
1152 /* Flag this case so that the XML code generator is aware that we haven't
1153 * looked up name references in actions. It can then avoid segfaulting. */
1154 generatingSectionSubset = true;
1156 /* Just building the specified graph. */
1158 FsmAp *mainGraph = makeInstance( gdNode );
1163 FsmAp *ParseData::makeAll()
1165 /* Build the name tree and supporting data structures. */
1168 /* Resove name references in the tree. */
1170 for ( GraphList::Iter glel = instanceList; glel.lte(); glel++ )
1171 glel->value->resolveNameRefs( this );
1173 /* Resolve action code name references. */
1174 resolveActionNameRefs();
1176 /* Force name references to the top level instantiations. */
1177 for ( NameVect::Iter inst = rootName->childVect; inst.lte(); inst++ )
1178 (*inst)->numRefs += 1;
1180 FsmAp *mainGraph = 0;
1181 FsmAp **graphs = new FsmAp*[instanceList.length()];
1184 /* Make all the instantiations, we know that main exists in this list. */
1186 for ( GraphList::Iter glel = instanceList; glel.lte(); glel++ ) {
1187 if ( strcmp( glel->key, mainMachine ) == 0 ) {
1188 /* Main graph is always instantiated. */
1189 mainGraph = makeInstance( glel );
1192 /* Instantiate and store in others array. */
1193 graphs[numOthers++] = makeInstance( glel );
1197 if ( mainGraph == 0 )
1198 mainGraph = graphs[--numOthers];
1200 if ( numOthers > 0 ) {
1201 /* Add all the other graphs into main. */
1202 mainGraph->globOp( graphs, numOthers );
1209 void ParseData::analyzeAction( Action *action, InlineList *inlineList )
1211 /* FIXME: Actions used as conditions should be very constrained. */
1212 for ( InlineList::Iter item = *inlineList; item.lte(); item++ ) {
1213 if ( item->type == InlineItem::Call || item->type == InlineItem::CallExpr )
1214 action->anyCall = true;
1216 /* Need to recurse into longest match items. */
1217 if ( item->type == InlineItem::LmSwitch ) {
1218 LongestMatch *lm = item->longestMatch;
1219 for ( LmPartList::Iter lmi = *lm->longestMatchList; lmi.lte(); lmi++ ) {
1220 if ( lmi->action != 0 )
1221 analyzeAction( action, lmi->action->inlineList );
1225 if ( item->type == InlineItem::LmOnLast ||
1226 item->type == InlineItem::LmOnNext ||
1227 item->type == InlineItem::LmOnLagBehind )
1229 LongestMatchPart *lmi = item->longestMatchPart;
1230 if ( lmi->action != 0 )
1231 analyzeAction( action, lmi->action->inlineList );
1234 if ( item->children != 0 )
1235 analyzeAction( action, item->children );
1240 /* Check actions for bad uses of fsm directives. We don't go inside longest
1241 * match items in actions created by ragel, since we just want the user
1243 void ParseData::checkInlineList( Action *act, InlineList *inlineList )
1245 for ( InlineList::Iter item = *inlineList; item.lte(); item++ ) {
1247 if ( act->numEofRefs > 0 ) {
1248 switch ( item->type ) {
1249 /* Currently no checks. */
1256 if ( item->children != 0 )
1257 checkInlineList( act, item->children );
1261 void ParseData::checkAction( Action *action )
1263 /* Check for actions with calls that are embedded within a longest match
1265 if ( !action->isLmAction && action->numRefs() > 0 && action->anyCall ) {
1266 for ( ActionRefs::Iter ar = action->actionRefs; ar.lte(); ar++ ) {
1267 NameInst *check = *ar;
1268 while ( check != 0 ) {
1269 if ( check->isLongestMatch ) {
1270 error(action->loc) << "within a scanner, fcall is permitted"
1271 " only in pattern actions" << endl;
1274 check = check->parent;
1279 checkInlineList( action, action->inlineList );
1283 void ParseData::analyzeGraph( FsmAp *graph )
1285 for ( ActionList::Iter act = actionList; act.lte(); act++ )
1286 analyzeAction( act, act->inlineList );
1288 for ( StateList::Iter st = graph->stateList; st.lte(); st++ ) {
1289 /* The transition list. */
1290 for ( TransList::Iter trans = st->outList; trans.lte(); trans++ ) {
1291 for ( ActionTable::Iter at = trans->actionTable; at.lte(); at++ )
1292 at->value->numTransRefs += 1;
1295 for ( ActionTable::Iter at = st->toStateActionTable; at.lte(); at++ )
1296 at->value->numToStateRefs += 1;
1298 for ( ActionTable::Iter at = st->fromStateActionTable; at.lte(); at++ )
1299 at->value->numFromStateRefs += 1;
1301 for ( ActionTable::Iter at = st->eofActionTable; at.lte(); at++ )
1302 at->value->numEofRefs += 1;
1304 for ( StateCondList::Iter sc = st->stateCondList; sc.lte(); sc++ ) {
1305 for ( CondSet::Iter sci = sc->condSpace->condSet; sci.lte(); sci++ )
1306 (*sci)->numCondRefs += 1;
1310 /* Checks for bad usage of directives in action code. */
1311 for ( ActionList::Iter act = actionList; act.lte(); act++ )
1315 void ParseData::makeExportsNameTree()
1317 /* Make a name tree for the exports. */
1318 initExportsNameWalk();
1320 /* First make the name tree. */
1321 for ( GraphDict::Iter gdel = graphDict; gdel.lte(); gdel++ ) {
1322 if ( gdel->value->isExport ) {
1323 /* Recurse on the instance. */
1324 gdel->value->makeNameTree( gdel->loc, this );
1329 void ParseData::makeExports()
1331 makeExportsNameTree();
1333 /* Resove name references in the tree. */
1334 initExportsNameWalk();
1335 for ( GraphDict::Iter gdel = graphDict; gdel.lte(); gdel++ ) {
1336 if ( gdel->value->isExport )
1337 gdel->value->resolveNameRefs( this );
1340 /* Make all the instantiations, we know that main exists in this list. */
1341 initExportsNameWalk();
1342 for ( GraphDict::Iter gdel = graphDict; gdel.lte(); gdel++ ) {
1343 /* Check if this var def is an export. */
1344 if ( gdel->value->isExport ) {
1345 /* Build the graph from a walk of the parse tree. */
1346 FsmAp *graph = gdel->value->walk( this );
1348 /* Build the graph from a walk of the parse tree. */
1349 if ( !graph->checkSingleCharMachine() ) {
1350 error(gdel->loc) << "bad export machine, must define "
1351 "a single character" << endl;
1354 /* Safe to extract the key and declare the export. */
1355 Key exportKey = graph->startState->outList.head->lowKey;
1356 exportList.append( new Export( gdel->value->name, exportKey ) );
1363 /* Construct the machine and catch failures which can occur during
1365 void ParseData::prepareMachineGen( GraphDictEl *graphDictEl )
1368 /* This machine construction can fail. */
1369 prepareMachineGenTBWrapped( graphDictEl );
1371 catch ( FsmConstructFail fail ) {
1372 switch ( fail.reason ) {
1373 case FsmConstructFail::CondNoKeySpace: {
1374 InputLoc &loc = alphTypeSet ? alphTypeLoc : sectionLoc;
1375 error(loc) << "sorry, no more characters are "
1376 "available in the alphabet space" << endl;
1377 error(loc) << " for conditions, please use a "
1378 "smaller alphtype or reduce" << endl;
1379 error(loc) << " the span of characters on which "
1380 "conditions are embedded" << endl;
1387 void ParseData::prepareMachineGenTBWrapped( GraphDictEl *graphDictEl )
1392 initLongestMatchData();
1394 /* Make the graph, do minimization. */
1395 if ( graphDictEl == 0 )
1396 sectionGraph = makeAll();
1398 sectionGraph = makeSpecific( graphDictEl );
1400 /* Compute exports from the export definitions. */
1403 /* If any errors have occured in the input file then don't write anything. */
1404 if ( gblErrorCount > 0 )
1407 analyzeGraph( sectionGraph );
1409 /* Depends on the graph analysis. */
1410 setLongestMatchData( sectionGraph );
1412 /* Decide if an error state is necessary.
1413 * 1. There is an error transition
1414 * 2. There is a gap in the transitions
1415 * 3. The longest match operator requires it. */
1416 if ( lmRequiresErrorState || sectionGraph->hasErrorTrans() )
1417 sectionGraph->errState = sectionGraph->addState();
1419 /* State numbers need to be assigned such that all final states have a
1420 * larger state id number than all non-final states. This enables the
1421 * first_final mechanism to function correctly. We also want states to be
1422 * ordered in a predictable fashion. So we first apply a depth-first
1423 * search, then do a stable sort by final state status, then assign
1426 sectionGraph->depthFirstOrdering();
1427 sectionGraph->sortStatesByFinal();
1428 sectionGraph->setStateNumbers( 0 );
1431 void ParseData::generateReduced( InputData &inputData )
1435 cgd = makeCodeGen( inputData.inputFileName, sectionName, *inputData.outStream );
1437 /* Make the generator. */
1438 BackendGen backendGen( sectionName, this, sectionGraph, cgd );
1440 /* Write out with it. */
1441 backendGen.makeBackend();
1443 if ( printStatistics ) {
1444 cerr << "fsm name : " << sectionName << endl;
1445 cerr << "num states: " << sectionGraph->stateList.length() << endl;
1450 void ParseData::generateXML( ostream &out )
1454 /* Make the generator. */
1455 XMLCodeGen codeGen( sectionName, this, sectionGraph, out );
1457 /* Write out with it. */
1460 if ( printStatistics ) {
1461 cerr << "fsm name : " << sectionName << endl;
1462 cerr << "num states: " << sectionGraph->stateList.length() << endl;