Regex subexpression captures are now bound as local variables in token generation actions.

6 files changed, 64 insertions, 13 deletions

diff --git a/colm/bytecode.cpp b/colm/bytecode.cpp
index 08a1ac98..75417880 100644
--- a/colm/bytecode.cpp
+++ b/colm/bytecode.cpp
@@ -402,7 +402,7 @@ void Program::run()
 
 	if ( rtd->rootCodeLen > 0 ) {
 		CodeVect reverseCode;
-		Execution execution( this, reverseCode, 0, rtd->rootCode, 0, 0 );
+		Execution execution( this, reverseCode, 0, rtd->rootCode, 0, 0, 0 );
 		execution.execute( root );
 
 		/* Pull out the reverse code and free it. */
@@ -422,7 +422,7 @@ void Program::run()
 }
 
 Execution::Execution( Program *prg, CodeVect &reverseCode,
-		PdaRun *parser, Code *code, Tree *lhs, Head *matchText )
+		PdaRun *parser, Code *code, Tree *lhs, long genId, Head *matchText )
 : 
 	prg(prg), 
 	parser(parser), 
@@ -430,6 +430,7 @@ Execution::Execution( Program *prg, CodeVect &reverseCode,
 	frame(0), iframe(0),
 	lhs(lhs),
 	parsed(0),
+	genId(genId),
 	matchText(matchText),
 	reject(false), 
 	reverseCode(reverseCode),
@@ -945,6 +946,31 @@ again:
 			push( prg->global );
 			break;
 		}
+		case IN_INIT_CAPTURES: {
+			uchar ncaps;
+			read_byte( ncaps );
+
+			#ifdef COLM_LOG_BYTECODE
+			if ( colm_log_bytecode ) {
+				cerr << "IN_INIT_CAPTURES " << ncaps << endl;
+			}
+			#endif
+
+			/* If there are captures (this is a translate block) then copy them into
+			 * the local frame now. */
+			LangElInfo *lelInfo = prg->rtd->lelInfo;
+			char **mark = parser->fsmRun->mark;
+
+			for ( int i = 0; i < lelInfo[genId].numCaptureAttr; i++ ) {
+				CaptureAttr *ca = &prg->rtd->captureAttr[lelInfo[genId].captureAttr + i];
+				Head *data = string_alloc_new( prg, 
+						mark[ca->mark_enter], mark[ca->mark_leave] - mark[ca->mark_enter] );
+				Tree *string = construct_string( prg, data );
+				tree_upref( string );
+				set_local( frame, -1 - i, string );
+			}
+			break;
+		}
 		case IN_INIT_RHS_EL: {
 			Half position;
 			short field;
diff --git a/colm/bytecode.h b/colm/bytecode.h
index 1b3eb02a..d7dc0d00 100644
--- a/colm/bytecode.h
+++ b/colm/bytecode.h
@@ -110,6 +110,7 @@ typedef unsigned char uchar;
 #define IN_GET_TOKEN_POS_R       0x35
 
 #define IN_INIT_RHS_EL           0x3b
+#define IN_INIT_CAPTURES         0xaa
 
 #define IN_TRITER_FROM_REF       0x3c
 #define IN_TRITER_ADVANCE        0x3d
@@ -688,7 +689,8 @@ struct Program
 struct Execution
 {
 	Execution( Program *prg, CodeVect &reverseCode,
-			PdaRun *parser, Code *code, Tree *lhs, Head *matchText );
+			PdaRun *parser, Code *code, Tree *lhs,
+			long genId, Head *matchText );
 
 	Program *prg;
 	PdaTables *pdaTables;
@@ -702,6 +704,7 @@ struct Execution
 	Tree *lhs;
 	Tree *parsed;
 
+	long genId;
 	Head *matchText;
 	bool reject;
 
diff --git a/colm/compile.cpp b/colm/compile.cpp
index 5fce43ce..d4d2f503 100644
--- a/colm/compile.cpp
+++ b/colm/compile.cpp
@@ -2208,6 +2208,15 @@ void ParseData::compileTranslateBlock( KlangEl *langEl )
 	code.append( IN_INIT_LOCALS );
 	code.appendHalf( 0 );
 
+	if ( langEl->tokenDef->reCaptureVect.length() > 0 ) {
+		code.append( IN_INIT_CAPTURES );
+		code.append( langEl->tokenDef->reCaptureVect.length() );
+
+		ObjFieldList::Iter f = *curLocalFrame->objFieldList;
+		for ( int i = 0; i < langEl->tokenDef->reCaptureVect.length(); i++, f++ )
+			curLocalFrame->referenceField( this, f->value );
+	}
+
 	/* Set the local frame and compile the reduce block. */
 	block->compile( this, code );
 
diff --git a/colm/fsmrun.cpp b/colm/fsmrun.cpp
index fc9c45d2..6f184b35 100644
--- a/colm/fsmrun.cpp
+++ b/colm/fsmrun.cpp
@@ -265,7 +265,7 @@ void FsmRun::sendBackIgnore( Kid *ignore )
 		/* Check for reverse code. */
 		if ( ignore->tree->flags & AF_HAS_RCODE ) {
 			Execution execution( prg, parser->reverseCode, 
-					parser, 0, 0, 0 );
+					parser, 0, 0, 0, 0 );
 
 			/* Do the reverse exeuction. */
 			execution.rexecute( parser->root, parser->allReverseCode );
@@ -307,7 +307,7 @@ void FsmRun::sendBack( Kid *input )
 	/* Check for reverse code. */
 	if ( input->tree->flags & AF_HAS_RCODE ) {
 		Execution execution( prg, parser->reverseCode, 
-				parser, 0, 0, 0 );
+				parser, 0, 0, 0, 0 );
 
 		/* Do the reverse exeuction. */
 		execution.rexecute( parser->root, parser->allReverseCode );
@@ -441,10 +441,10 @@ void FsmRun::sendNamedLangEl()
 	send_handle_error( this, parser, input );
 }
 
-void execute_generation_action( Program *prg, PdaRun *parser, Code *code, Head *tokdata )
+void execute_generation_action( Program *prg, PdaRun *parser, Code *code, long id, Head *tokdata )
 {
 	/* Execute the translation. */
-	Execution execution( prg, parser->reverseCode, parser, code, 0, tokdata );
+	Execution execution( prg, parser->reverseCode, parser, code, 0, id, tokdata );
 	execution.execute( parser->root );
 
 	/* If there is revese code but nothing generated we need a noToken. */
@@ -481,7 +481,7 @@ void FsmRun::generationAction( int id, Head *tokdata, bool namedLangEl, int bind
 			parser->tables->rtd->lelInfo[id].frameId].codeWV;
 
 	/* Execute the action and process the queue. */
-	execute_generation_action( prg, parser, code, tokdata );
+	execute_generation_action( prg, parser, code, id, tokdata );
 
 	/* Finished with the match text. */
 	string_free( prg, tokdata );
@@ -518,8 +518,8 @@ Kid *FsmRun::makeToken( int id, Head *tokdata, bool namedLangEl, int bindId )
 			Head *data = string_alloc_new( prg, 
 					mark[ca->mark_enter], mark[ca->mark_leave] - mark[ca->mark_enter] );
 			Tree *string = construct_string( prg, data );
-			set_attr( input->tree, ca->offset, string );
 			tree_upref( string );
+			set_attr( input->tree, ca->offset, string );
 		}
 	}
 	
@@ -772,7 +772,7 @@ void FsmRun::sendEOF( )
 		Code *code = parser->tables->rtd->frameInfo[frameId].codeWV;
 
 		/* Execute the action and process the queue. */
-		execute_generation_action( prg, parser, code, 0 );
+		execute_generation_action( prg, parser, code, input->tree->id, 0 );
 
 		/* Send the generated tokens. */
 		send_queued_tokens( this, parser );
diff --git a/colm/lmparse.kl b/colm/lmparse.kl
index 6f974c21..d4618c00 100644
--- a/colm/lmparse.kl
+++ b/colm/lmparse.kl
@@ -91,6 +91,20 @@ block_open: '{'
 				"local", new ObjFieldMap(), new ObjFieldList, 
 				new ObjMethodMap(), pd->nextObjectId++ );
 		pd->curLocalFrame = $$->localFrame;
+
+		/* Add captures to the local frame. We Depend on these becoming the
+		 * first local variables so we can compute their location. */
+
+		/* Make local variables corresponding to the local capture vector. */
+		for ( ReCaptureVect::Iter c = reCaptureVect; c.lte(); c++ )
+		{
+			ObjField *objField = new ObjField( c->objField->loc,
+					c->objField->typeRef, c->objField->name );
+
+			/* Insert it into the field map. */
+			pd->curLocalFrame->objFieldMap->insert( objField->name, objField );
+			pd->curLocalFrame->objFieldList->append( objField );
+		}
 	};
 
 block_close: '}' 
@@ -1987,7 +2001,6 @@ opt_commit: KW_Commit final { $$->commit = true; };
 
 void Parser::init()
 {
-
 	/* Set up the root namespace. */
 	const char *rootNamespaceName = "___ROOT_NAMESPACE";
 	Namespace *rootNamespace = new Namespace( InputLoc(), 
diff --git a/colm/pdarun.cpp b/colm/pdarun.cpp
index 07bd8d09..b85a32c0 100644
--- a/colm/pdarun.cpp
+++ b/colm/pdarun.cpp
@@ -463,7 +463,7 @@ again:
 
 			/* Execution environment for the reduction code. */
 			Execution execution( prg, reverseCode, 
-					this, fi->codeWV, redLel->tree, 0 );
+					this, fi->codeWV, redLel->tree, 0, 0 );
 
 			/* Execute it. */
 			execution.execute( root );
@@ -629,7 +629,7 @@ parseError:
 
 			/* Check for an execution environment. */
 			if ( undoLel->tree->flags & AF_HAS_RCODE ) {
-				Execution execution( prg, reverseCode, this, 0, 0, 0 );
+				Execution execution( prg, reverseCode, this, 0, 0, 0, 0 );
 
 				/* Do the reverse exeuction. */
 				execution.rexecute( root, allReverseCode );