Started on the move back to a single executable. The backend programs are now a
[ragel.git] / rlgen-cd / fsmcodegen.cpp
blobab8f4ad3b3bbe1260e7acbfc6de3723f1141212e
1 /*
2 * Copyright 2001-2006 Adrian Thurston <thurston@cs.queensu.ca>
3 * 2004 Erich Ocean <eric.ocean@ampede.com>
4 * 2005 Alan West <alan@alanz.com>
5 */
7 /* This file is part of Ragel.
9 * Ragel is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * Ragel is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with Ragel; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
24 #include "rlgen-cd.h"
25 #include "fsmcodegen.h"
26 #include "redfsm.h"
27 #include "gendata.h"
28 #include <sstream>
29 #include <string>
30 #include <assert.h>
33 using std::ostream;
34 using std::ostringstream;
35 using std::string;
36 using std::cerr;
37 using std::endl;
39 void cdLineDirective( ostream &out, const char *fileName, int line )
41 if ( noLineDirectives )
42 out << "/* ";
44 /* Write the preprocessor line info for to the input file. */
45 out << "#line " << line << " \"";
46 for ( const char *pc = fileName; *pc != 0; pc++ ) {
47 if ( *pc == '\\' )
48 out << "\\\\";
49 else
50 out << *pc;
52 out << '"';
54 if ( noLineDirectives )
55 out << " */";
57 out << '\n';
60 void FsmCodeGen::genLineDirective( ostream &out )
62 std::streambuf *sbuf = out.rdbuf();
63 output_filter *filter = static_cast<output_filter*>(sbuf);
64 cdLineDirective( out, filter->fileName, filter->line + 1 );
68 /* Init code gen with in parameters. */
69 FsmCodeGen::FsmCodeGen( ostream &out )
71 CodeGenData(out)
75 unsigned int FsmCodeGen::arrayTypeSize( unsigned long maxVal )
77 long long maxValLL = (long long) maxVal;
78 HostType *arrayType = keyOps->typeSubsumes( maxValLL );
79 assert( arrayType != 0 );
80 return arrayType->size;
83 string FsmCodeGen::ARRAY_TYPE( unsigned long maxVal )
85 long long maxValLL = (long long) maxVal;
86 HostType *arrayType = keyOps->typeSubsumes( maxValLL );
87 assert( arrayType != 0 );
89 string ret = arrayType->data1;
90 if ( arrayType->data2 != 0 ) {
91 ret += " ";
92 ret += arrayType->data2;
94 return ret;
98 /* Write out the fsm name. */
99 string FsmCodeGen::FSM_NAME()
101 return fsmName;
104 /* Emit the offset of the start state as a decimal integer. */
105 string FsmCodeGen::START_STATE_ID()
107 ostringstream ret;
108 ret << redFsm->startState->id;
109 return ret.str();
112 /* Write out the array of actions. */
113 std::ostream &FsmCodeGen::ACTIONS_ARRAY()
115 out << "\t0, ";
116 int totalActions = 1;
117 for ( ActionTableMap::Iter act = redFsm->actionMap; act.lte(); act++ ) {
118 /* Write out the length, which will never be the last character. */
119 out << act->key.length() << ", ";
120 /* Put in a line break every 8 */
121 if ( totalActions++ % 8 == 7 )
122 out << "\n\t";
124 for ( ActionTable::Iter item = act->key; item.lte(); item++ ) {
125 out << item->value->actionId;
126 if ( ! (act.last() && item.last()) )
127 out << ", ";
129 /* Put in a line break every 8 */
130 if ( totalActions++ % 8 == 7 )
131 out << "\n\t";
134 out << "\n";
135 return out;
139 string FsmCodeGen::ACCESS()
141 ostringstream ret;
142 if ( accessExpr != 0 )
143 INLINE_LIST( ret, accessExpr, 0, false, false );
144 return ret.str();
148 string FsmCodeGen::P()
150 ostringstream ret;
151 if ( pExpr == 0 )
152 ret << "p";
153 else {
154 ret << "(";
155 INLINE_LIST( ret, pExpr, 0, false, false );
156 ret << ")";
158 return ret.str();
161 string FsmCodeGen::PE()
163 ostringstream ret;
164 if ( peExpr == 0 )
165 ret << "pe";
166 else {
167 ret << "(";
168 INLINE_LIST( ret, peExpr, 0, false, false );
169 ret << ")";
171 return ret.str();
174 string FsmCodeGen::EOFV()
176 ostringstream ret;
177 if ( eofExpr == 0 )
178 ret << "eof";
179 else {
180 ret << "(";
181 INLINE_LIST( ret, eofExpr, 0, false, false );
182 ret << ")";
184 return ret.str();
187 string FsmCodeGen::CS()
189 ostringstream ret;
190 if ( csExpr == 0 )
191 ret << ACCESS() << "cs";
192 else {
193 /* Emit the user supplied method of retrieving the key. */
194 ret << "(";
195 INLINE_LIST( ret, csExpr, 0, false, false );
196 ret << ")";
198 return ret.str();
201 string FsmCodeGen::TOP()
203 ostringstream ret;
204 if ( topExpr == 0 )
205 ret << ACCESS() + "top";
206 else {
207 ret << "(";
208 INLINE_LIST( ret, topExpr, 0, false, false );
209 ret << ")";
211 return ret.str();
214 string FsmCodeGen::STACK()
216 ostringstream ret;
217 if ( stackExpr == 0 )
218 ret << ACCESS() + "stack";
219 else {
220 ret << "(";
221 INLINE_LIST( ret, stackExpr, 0, false, false );
222 ret << ")";
224 return ret.str();
227 string FsmCodeGen::ACT()
229 ostringstream ret;
230 if ( actExpr == 0 )
231 ret << ACCESS() + "act";
232 else {
233 ret << "(";
234 INLINE_LIST( ret, actExpr, 0, false, false );
235 ret << ")";
237 return ret.str();
240 string FsmCodeGen::TOKSTART()
242 ostringstream ret;
243 if ( tokstartExpr == 0 )
244 ret << ACCESS() + "ts";
245 else {
246 ret << "(";
247 INLINE_LIST( ret, tokstartExpr, 0, false, false );
248 ret << ")";
250 return ret.str();
253 string FsmCodeGen::TOKEND()
255 ostringstream ret;
256 if ( tokendExpr == 0 )
257 ret << ACCESS() + "te";
258 else {
259 ret << "(";
260 INLINE_LIST( ret, tokendExpr, 0, false, false );
261 ret << ")";
263 return ret.str();
266 string FsmCodeGen::GET_WIDE_KEY()
268 if ( redFsm->anyConditions() )
269 return "_widec";
270 else
271 return GET_KEY();
274 string FsmCodeGen::GET_WIDE_KEY( RedStateAp *state )
276 if ( state->stateCondList.length() > 0 )
277 return "_widec";
278 else
279 return GET_KEY();
282 string FsmCodeGen::GET_KEY()
284 ostringstream ret;
285 if ( getKeyExpr != 0 ) {
286 /* Emit the user supplied method of retrieving the key. */
287 ret << "(";
288 INLINE_LIST( ret, getKeyExpr, 0, false, false );
289 ret << ")";
291 else {
292 /* Expression for retrieving the key, use simple dereference. */
293 ret << "(*" << P() << ")";
295 return ret.str();
298 /* Write out level number of tabs. Makes the nested binary search nice
299 * looking. */
300 string FsmCodeGen::TABS( int level )
302 string result;
303 while ( level-- > 0 )
304 result += "\t";
305 return result;
308 /* Write out a key from the fsm code gen. Depends on wether or not the key is
309 * signed. */
310 string FsmCodeGen::KEY( Key key )
312 ostringstream ret;
313 if ( keyOps->isSigned || !hostLang->explicitUnsigned )
314 ret << key.getVal();
315 else
316 ret << (unsigned long) key.getVal() << 'u';
317 return ret.str();
320 void FsmCodeGen::EXEC( ostream &ret, GenInlineItem *item, int targState, int inFinish )
322 /* The parser gives fexec two children. The double brackets are for D
323 * code. If the inline list is a single word it will get interpreted as a
324 * C-style cast by the D compiler. */
325 ret << "{" << P() << " = ((";
326 INLINE_LIST( ret, item->children, targState, inFinish, false );
327 ret << "))-1;}";
330 void FsmCodeGen::LM_SWITCH( ostream &ret, GenInlineItem *item,
331 int targState, int inFinish, bool csForced )
333 ret <<
334 " switch( " << ACT() << " ) {\n";
336 bool haveDefault = false;
337 for ( GenInlineList::Iter lma = *item->children; lma.lte(); lma++ ) {
338 /* Write the case label, the action and the case break. */
339 if ( lma->lmId < 0 ) {
340 ret << " default:\n";
341 haveDefault = true;
343 else
344 ret << " case " << lma->lmId << ":\n";
346 /* Write the block and close it off. */
347 ret << " {";
348 INLINE_LIST( ret, lma->children, targState, inFinish, csForced );
349 ret << "}\n";
351 ret << " break;\n";
354 if ( hostLang->lang == HostLang::D && !haveDefault )
355 ret << " default: break;";
357 ret <<
358 " }\n"
359 "\t";
362 void FsmCodeGen::SET_ACT( ostream &ret, GenInlineItem *item )
364 ret << ACT() << " = " << item->lmId << ";";
367 void FsmCodeGen::SET_TOKEND( ostream &ret, GenInlineItem *item )
369 /* The tokend action sets tokend. */
370 ret << TOKEND() << " = " << P();
371 if ( item->offset != 0 )
372 out << "+" << item->offset;
373 out << ";";
376 void FsmCodeGen::GET_TOKEND( ostream &ret, GenInlineItem *item )
378 ret << TOKEND();
381 void FsmCodeGen::INIT_TOKSTART( ostream &ret, GenInlineItem *item )
383 ret << TOKSTART() << " = " << NULL_ITEM() << ";";
386 void FsmCodeGen::INIT_ACT( ostream &ret, GenInlineItem *item )
388 ret << ACT() << " = 0;";
391 void FsmCodeGen::SET_TOKSTART( ostream &ret, GenInlineItem *item )
393 ret << TOKSTART() << " = " << P() << ";";
396 void FsmCodeGen::SUB_ACTION( ostream &ret, GenInlineItem *item,
397 int targState, bool inFinish, bool csForced )
399 if ( item->children->length() > 0 ) {
400 /* Write the block and close it off. */
401 ret << "{";
402 INLINE_LIST( ret, item->children, targState, inFinish, csForced );
403 ret << "}";
408 /* Write out an inline tree structure. Walks the list and possibly calls out
409 * to virtual functions than handle language specific items in the tree. */
410 void FsmCodeGen::INLINE_LIST( ostream &ret, GenInlineList *inlineList,
411 int targState, bool inFinish, bool csForced )
413 for ( GenInlineList::Iter item = *inlineList; item.lte(); item++ ) {
414 switch ( item->type ) {
415 case GenInlineItem::Text:
416 ret << item->data;
417 break;
418 case GenInlineItem::Goto:
419 GOTO( ret, item->targState->id, inFinish );
420 break;
421 case GenInlineItem::Call:
422 CALL( ret, item->targState->id, targState, inFinish );
423 break;
424 case GenInlineItem::Next:
425 NEXT( ret, item->targState->id, inFinish );
426 break;
427 case GenInlineItem::Ret:
428 RET( ret, inFinish );
429 break;
430 case GenInlineItem::PChar:
431 ret << P();
432 break;
433 case GenInlineItem::Char:
434 ret << GET_KEY();
435 break;
436 case GenInlineItem::Hold:
437 ret << P() << "--;";
438 break;
439 case GenInlineItem::Exec:
440 EXEC( ret, item, targState, inFinish );
441 break;
442 case GenInlineItem::Curs:
443 CURS( ret, inFinish );
444 break;
445 case GenInlineItem::Targs:
446 TARGS( ret, inFinish, targState );
447 break;
448 case GenInlineItem::Entry:
449 ret << item->targState->id;
450 break;
451 case GenInlineItem::GotoExpr:
452 GOTO_EXPR( ret, item, inFinish );
453 break;
454 case GenInlineItem::CallExpr:
455 CALL_EXPR( ret, item, targState, inFinish );
456 break;
457 case GenInlineItem::NextExpr:
458 NEXT_EXPR( ret, item, inFinish );
459 break;
460 case GenInlineItem::LmSwitch:
461 LM_SWITCH( ret, item, targState, inFinish, csForced );
462 break;
463 case GenInlineItem::LmSetActId:
464 SET_ACT( ret, item );
465 break;
466 case GenInlineItem::LmSetTokEnd:
467 SET_TOKEND( ret, item );
468 break;
469 case GenInlineItem::LmGetTokEnd:
470 GET_TOKEND( ret, item );
471 break;
472 case GenInlineItem::LmInitTokStart:
473 INIT_TOKSTART( ret, item );
474 break;
475 case GenInlineItem::LmInitAct:
476 INIT_ACT( ret, item );
477 break;
478 case GenInlineItem::LmSetTokStart:
479 SET_TOKSTART( ret, item );
480 break;
481 case GenInlineItem::SubAction:
482 SUB_ACTION( ret, item, targState, inFinish, csForced );
483 break;
484 case GenInlineItem::Break:
485 BREAK( ret, targState, csForced );
486 break;
490 /* Write out paths in line directives. Escapes any special characters. */
491 string FsmCodeGen::LDIR_PATH( char *path )
493 ostringstream ret;
494 for ( char *pc = path; *pc != 0; pc++ ) {
495 if ( *pc == '\\' )
496 ret << "\\\\";
497 else
498 ret << *pc;
500 return ret.str();
503 void FsmCodeGen::ACTION( ostream &ret, Action *action, int targState,
504 bool inFinish, bool csForced )
506 /* Write the preprocessor line info for going into the source file. */
507 cdLineDirective( ret, sourceFileName, action->loc.line );
509 /* Write the block and close it off. */
510 ret << "\t{";
511 INLINE_LIST( ret, action->inlineList, targState, inFinish, csForced );
512 ret << "}\n";
515 void FsmCodeGen::CONDITION( ostream &ret, Action *condition )
517 ret << "\n";
518 cdLineDirective( ret, sourceFileName, condition->loc.line );
519 INLINE_LIST( ret, condition->inlineList, 0, false, false );
522 string FsmCodeGen::ERROR_STATE()
524 ostringstream ret;
525 if ( redFsm->errState != 0 )
526 ret << redFsm->errState->id;
527 else
528 ret << "-1";
529 return ret.str();
532 string FsmCodeGen::FIRST_FINAL_STATE()
534 ostringstream ret;
535 if ( redFsm->firstFinState != 0 )
536 ret << redFsm->firstFinState->id;
537 else
538 ret << redFsm->nextStateId;
539 return ret.str();
542 void FsmCodeGen::writeInit()
544 out << " {\n";
546 if ( writeCS )
547 out << "\t" << CS() << " = " << START() << ";\n";
549 /* If there are any calls, then the stack top needs initialization. */
550 if ( redFsm->anyActionCalls() || redFsm->anyActionRets() )
551 out << "\t" << TOP() << " = 0;\n";
553 if ( hasLongestMatch ) {
554 out <<
555 " " << TOKSTART() << " = " << NULL_ITEM() << ";\n"
556 " " << TOKEND() << " = " << NULL_ITEM() << ";\n"
557 " " << ACT() << " = 0;\n";
559 out << " }\n";
562 string FsmCodeGen::DATA_PREFIX()
564 if ( dataPrefix )
565 return FSM_NAME() + "_";
566 return "";
569 /* Emit the alphabet data type. */
570 string FsmCodeGen::ALPH_TYPE()
572 string ret = keyOps->alphType->data1;
573 if ( keyOps->alphType->data2 != 0 ) {
574 ret += " ";
575 ret += + keyOps->alphType->data2;
577 return ret;
580 /* Emit the alphabet data type. */
581 string FsmCodeGen::WIDE_ALPH_TYPE()
583 string ret;
584 if ( redFsm->maxKey <= keyOps->maxKey )
585 ret = ALPH_TYPE();
586 else {
587 long long maxKeyVal = redFsm->maxKey.getLongLong();
588 HostType *wideType = keyOps->typeSubsumes( keyOps->isSigned, maxKeyVal );
589 assert( wideType != 0 );
591 ret = wideType->data1;
592 if ( wideType->data2 != 0 ) {
593 ret += " ";
594 ret += wideType->data2;
597 return ret;
600 void FsmCodeGen::STATE_IDS()
602 if ( redFsm->startState != 0 )
603 STATIC_VAR( "int", START() ) << " = " << START_STATE_ID() << ";\n";
605 if ( writeFirstFinal )
606 STATIC_VAR( "int" , FIRST_FINAL() ) << " = " << FIRST_FINAL_STATE() << ";\n";
608 if ( writeErr )
609 STATIC_VAR( "int", ERROR() ) << " = " << ERROR_STATE() << ";\n";
611 out << "\n";
613 if ( entryPointNames.length() > 0 ) {
614 for ( EntryNameVect::Iter en = entryPointNames; en.lte(); en++ ) {
615 STATIC_VAR( "int", DATA_PREFIX() + "en_" + *en ) <<
616 " = " << entryPointIds[en.pos()] << ";\n";
618 out << "\n";
624 * Language specific, but style independent code generators functions.
627 string CCodeGen::PTR_CONST()
629 return "const ";
632 std::ostream &CCodeGen::OPEN_ARRAY( string type, string name )
634 out << "static const " << type << " " << name << "[] = {\n";
635 return out;
638 std::ostream &CCodeGen::CLOSE_ARRAY()
640 return out << "};\n";
643 std::ostream &CCodeGen::STATIC_VAR( string type, string name )
645 out << "static const " << type << " " << name;
646 return out;
649 string CCodeGen::UINT( )
651 return "unsigned int";
654 string CCodeGen::ARR_OFF( string ptr, string offset )
656 return ptr + " + " + offset;
659 string CCodeGen::CAST( string type )
661 return "(" + type + ")";
664 string CCodeGen::NULL_ITEM()
666 return "0";
669 string CCodeGen::POINTER()
671 return " *";
674 std::ostream &CCodeGen::SWITCH_DEFAULT()
676 return out;
679 string CCodeGen::CTRL_FLOW()
681 return "";
684 void CCodeGen::writeExports()
686 if ( exportList.length() > 0 ) {
687 for ( ExportList::Iter ex = exportList; ex.lte(); ex++ ) {
688 out << "#define " << DATA_PREFIX() << "ex_" << ex->name << " " <<
689 KEY(ex->key) << "\n";
691 out << "\n";
696 * D Specific
699 string DCodeGen::NULL_ITEM()
701 return "null";
704 string DCodeGen::POINTER()
706 // multiple items seperated by commas can also be pointer types.
707 return "* ";
710 string DCodeGen::PTR_CONST()
712 return "";
715 std::ostream &DCodeGen::OPEN_ARRAY( string type, string name )
717 out << "static const " << type << "[] " << name << " = [\n";
718 return out;
721 std::ostream &DCodeGen::CLOSE_ARRAY()
723 return out << "];\n";
726 std::ostream &DCodeGen::STATIC_VAR( string type, string name )
728 out << "static const " << type << " " << name;
729 return out;
732 string DCodeGen::ARR_OFF( string ptr, string offset )
734 return "&" + ptr + "[" + offset + "]";
737 string DCodeGen::CAST( string type )
739 return "cast(" + type + ")";
742 string DCodeGen::UINT( )
744 return "uint";
747 std::ostream &DCodeGen::SWITCH_DEFAULT()
749 out << " default: break;\n";
750 return out;
753 string DCodeGen::CTRL_FLOW()
755 return "if (true) ";
758 void DCodeGen::writeExports()
760 if ( exportList.length() > 0 ) {
761 for ( ExportList::Iter ex = exportList; ex.lte(); ex++ ) {
762 out << "static const " << ALPH_TYPE() << " " << DATA_PREFIX() <<
763 "ex_" << ex->name << " = " << KEY(ex->key) << ";\n";
765 out << "\n";
770 * End D-specific code.
773 void FsmCodeGen::finishRagelDef()
775 if ( codeStyle == GenGoto || codeStyle == GenFGoto ||
776 codeStyle == GenIpGoto || codeStyle == GenSplit )
778 /* For directly executable machines there is no required state
779 * ordering. Choose a depth-first ordering to increase the
780 * potential for fall-throughs. */
781 redFsm->depthFirstOrdering();
783 else {
784 /* The frontend will do this for us, but it may be a good idea to
785 * force it if the intermediate file is edited. */
786 redFsm->sortByStateId();
789 /* Choose default transitions and the single transition. */
790 redFsm->chooseDefaultSpan();
792 /* Maybe do flat expand, otherwise choose single. */
793 if ( codeStyle == GenFlat || codeStyle == GenFFlat )
794 redFsm->makeFlat();
795 else
796 redFsm->chooseSingle();
798 /* If any errors have occured in the input file then don't write anything. */
799 if ( gblErrorCount > 0 )
800 return;
802 if ( codeStyle == GenSplit )
803 redFsm->partitionFsm( numSplitPartitions );
805 if ( codeStyle == GenIpGoto || codeStyle == GenSplit )
806 redFsm->setInTrans();
808 /* Anlayze Machine will find the final action reference counts, among
809 * other things. We will use these in reporting the usage
810 * of fsm directives in action code. */
811 analyzeMachine();
813 /* Determine if we should use indicies. */
814 calcIndexSize();
817 ostream &FsmCodeGen::source_warning( const GenInputLoc &loc )
819 cerr << sourceFileName << ":" << loc.line << ":" << loc.col << ": warning: ";
820 return cerr;
823 ostream &FsmCodeGen::source_error( const GenInputLoc &loc )
825 gblErrorCount += 1;
826 assert( sourceFileName != 0 );
827 cerr << sourceFileName << ":" << loc.line << ":" << loc.col << ": ";
828 return cerr;