added trim control flag to print code, auto-trimming all colm print calls

The print implemenation now takes a trim flag. The colm print function now sets
this flag by default. This is a change to the colm language back to 0.5
semantics. The $ conversion uses this flag too (also 0.5 semantics), in the
previous commit it issued a tree trim operation. The % operation gives a string
conversion without triming.

17 files changed, 54 insertions, 43 deletions

diff --git a/colm/bytecode.c b/colm/bytecode.c
index b5606635..57d31c78 100644
--- a/colm/bytecode.c
+++ b/colm/bytecode.c
@@ -164,13 +164,13 @@ void parserSetContext( Program *prg, Tree **sp, Parser *parser, Tree *val )
 	parser->pdaRun->context = splitTree( prg, val );
 }
 
-Head *treeToStr( Program *prg, Tree **sp, Tree *tree )
+Head *treeToStr( Program *prg, Tree **sp, Tree *tree, int trim )
 {
 	/* Collect the tree data. */
 	StrCollect collect;
 	initStrCollect( &collect );
 
-	printTreeCollect( prg, sp, &collect, tree );
+	printTreeCollect( prg, sp, &collect, tree, trim );
 
 	/* Set up the input stream. */
 	Head *ret = stringAllocFull( prg, collect.data, collect.length );
@@ -189,7 +189,7 @@ Word streamAppend( Program *prg, Tree **sp, Tree *input, InputStream *inputStrea
 		/* Collect the tree data. */
 		StrCollect collect;
 		initStrCollect( &collect );
-		printTreeCollect( prg, sp, &collect, input );
+		printTreeCollect( prg, sp, &collect, input, true );
 
 		/* Load it into the input. */
 		appendData( inputStream, collect.data, collect.length );
@@ -349,7 +349,7 @@ long streamPush( Program *prg, Tree **sp, FsmRun *fsmRun, InputStream *in, Tree
 		/* Collect the tree data. */
 		StrCollect collect;
 		initStrCollect( &collect );
-		printTreeCollect( prg, sp, &collect, tree );
+		printTreeCollect( prg, sp, &collect, tree, true );
 
 		streamPushText( fsmRun, in, collect.data, collect.length );
 		long length = collect.length;
@@ -893,7 +893,7 @@ again:
 
 			while ( n-- > 0 ) {
 				Tree *tree = vm_pop();
-				printTreeFile( prg, sp, stdout, tree );
+				printTreeFile( prg, sp, stdout, tree, true );
 				treeDownref( prg, sp, tree );
 			}
 			break;
@@ -906,7 +906,7 @@ again:
 
 			while ( n-- > 0 ) {
 				Tree *tree = vm_pop();
-				printXmlStdout( prg, sp, tree, true );
+				printXmlStdout( prg, sp, tree, true, true );
 				treeDownref( prg, sp, tree );
 			}
 			break;
@@ -918,7 +918,7 @@ again:
 
 			while ( n-- > 0 ) {
 				Tree *tree = vm_pop();
-				printXmlStdout( prg, sp, tree, false );
+				printXmlStdout( prg, sp, tree, false, true );
 				treeDownref( prg, sp, tree );
 			}
 			break;
@@ -931,7 +931,7 @@ again:
 			Stream *stream = (Stream*)vm_pop();
 			while ( n-- > 0 ) {
 				Tree *tree = vm_pop();
-				printTreeFile( prg, sp, stream->file, tree );
+				printTreeFile( prg, sp, stream->file, tree, true );
 				treeDownref( prg, sp, tree );
 			}
 			treeDownref( prg, sp, (Tree*)stream );
@@ -1543,7 +1543,18 @@ again:
 			debug( REALM_BYTECODE, "IN_TREE_TO_STR\n" );
 
 			Tree *tree = vm_pop();
-			Head *res = treeToStr( prg, sp, tree );
+			Head *res = treeToStr( prg, sp, tree, true );
+			Tree *str = constructString( prg, res );
+			treeUpref( str );
+			vm_push( str );
+			treeDownref( prg, sp, tree );
+			break;
+		}
+		case IN_TREE_TO_STR_NOTRIM: {
+			debug( REALM_BYTECODE, "IN_TREE_TO_STR_NOTRIM\n" );
+
+			Tree *tree = vm_pop();
+			Head *res = treeToStr( prg, sp, tree, false );
 			Tree *str = constructString( prg, res );
 			treeUpref( str );
 			vm_push( str );
diff --git a/colm/bytecode.h b/colm/bytecode.h
index 1c8545d1..bb639e94 100644
--- a/colm/bytecode.h
+++ b/colm/bytecode.h
@@ -214,6 +214,7 @@ typedef unsigned char uchar;
 
 #define IN_INT_TO_STR            0x8b
 #define IN_TREE_TO_STR           0x8c
+#define IN_TREE_TO_STR_NOTRIM    0xfd
 
 #define IN_CREATE_TOKEN          0x8d
 #define IN_MAKE_TOKEN            0x8e
diff --git a/colm/codegen.cc b/colm/codegen.cc
index 3abd99c3..400e4b9a 100644
--- a/colm/codegen.cc
+++ b/colm/codegen.cc
@@ -77,7 +77,7 @@ void ParseData::generateExports()
 		"inline std::string printTreeStr( ColmProgram *prg, ColmTree *tree )\n"
 		"{\n"
 		"	std::string str;\n"
-		"	ColmPrintArgs printArgs = { &str, 1, 0, &appendString, \n"
+		"	ColmPrintArgs printArgs = { &str, 1, 0, 1, &appendString, \n"
 		"			&printNull, &printTermTree, &printNull };\n"
 		"	printTreeArgs( prg, vm_root(prg), &printArgs, tree );\n"
 		"	return str;\n"
diff --git a/colm/colm.h b/colm/colm.h
index 335259b4..4f169254 100644
--- a/colm/colm.h
+++ b/colm/colm.h
@@ -20,6 +20,7 @@ struct ColmPrintArgs
 	void *arg;
 	int comm;
 	int attr;
+	int trim;
 	void (*out)( struct ColmPrintArgs *args, const char *data, int length );
 	void (*openTree)( struct ColmProgram *prg, struct ColmTree **sp, 
 		struct ColmPrintArgs *args, struct ColmKid *parent, struct ColmKid *kid );
diff --git a/colm/compile.cc b/colm/compile.cc
index f948d33f..14084a5a 100644
--- a/colm/compile.cc
+++ b/colm/compile.cc
@@ -1720,22 +1720,19 @@ UniqueType *LangExpr::evaluate( ParseData *pd, CodeVect &code ) const
 					return pd->uniqueTypeBool;
 				}
 				case '$': {
-					UniqueType *rt = right->evaluate( pd, code );
-					if ( rt->typeId == TYPE_TREE && rt != pd->uniqueTypeInt && rt != pd->uniqueTypeStr && rt != pd->uniqueTypeBool ) 
-						code.append( IN_TREE_TRIM );
+					right->evaluate( pd, code );
 					code.append( IN_TREE_TO_STR );
 					return pd->uniqueTypeStr;
 					
 				}
 				case '%': {
 					right->evaluate( pd, code );
-					code.append( IN_TREE_TO_STR );
+					code.append( IN_TREE_TO_STR_NOTRIM );
 					return pd->uniqueTypeStr;
 				}
 				case '^': {
 					UniqueType *rt = right->evaluate( pd, code );
-					if ( rt->typeId == TYPE_TREE && rt != pd->uniqueTypeInt && rt != pd->uniqueTypeStr && rt != pd->uniqueTypeBool ) 
-						code.append( IN_TREE_TRIM );
+					code.append( IN_TREE_TRIM );
 					return rt;
 				}
 				case OP_Deref: {
diff --git a/colm/tree.c b/colm/tree.c
index 722a474c..66e5e025 100644
--- a/colm/tree.c
+++ b/colm/tree.c
@@ -2096,12 +2096,15 @@ enum VisitType
 	NonTerm,
 };
 
+#define TF_TERM_SEEN 0x1
+
 void printKid( Program *prg, Tree **sp, struct ColmPrintArgs *printArgs, Kid *kid )
 {
 	enum ReturnType rt;
 	Kid *parent = 0;
 	Kid *leadingIgnore = 0;
 	enum VisitType visitType;
+	int flags = 0;
 
 	/* Iterate the kids passed in. We are expecting a next, which will allow us
 	 * to print the trailing ignore list. */
@@ -2188,7 +2191,7 @@ rec_call:
 
 			/* Print the leading ignore list. Also implement the suppress right
 			 * in the process. */
-			if ( printArgs->comm ) {	
+			if ( printArgs->comm && (!printArgs->trim || (flags & TF_TERM_SEEN && kid->tree->id > 0)) ) {	
 				ignore = leadingIgnore;
 				while ( ignore != 0 ) {
 					if ( ignore->tree->flags & AF_SUPPRESS_RIGHT )
@@ -2232,6 +2235,9 @@ rec_call:
 		printArgs->openTree( prg, sp, printArgs, parent, kid );
 	}
 
+	if ( visitType == Term )
+		flags |= TF_TERM_SEEN;
+
 	if ( visitType == Term || visitType == IgnoreData ) {
 		/* Print contents. */
 		if ( kid->tree->id < prg->rtd->firstNonTermId ) {
@@ -2455,23 +2461,23 @@ void closeTreeXml( Program *prg, Tree **sp, struct ColmPrintArgs *args, Kid *par
 	args->out( args, ">", 1 );
 }
 
-void printTreeCollect( Program *prg, Tree **sp, StrCollect *collect, Tree *tree )
+void printTreeCollect( Program *prg, Tree **sp, StrCollect *collect, Tree *tree, int trim )
 {
-	struct ColmPrintArgs printArgs = { collect, 1, 0, &appendCollect, 
+	struct ColmPrintArgs printArgs = { collect, true, false, trim, &appendCollect, 
 			&printNull, &printTermTree, &printNull };
 	printTreeArgs( prg, sp, &printArgs, tree );
 }
 
-void printTreeFile( Program *prg, Tree **sp, FILE *out, Tree *tree )
+void printTreeFile( Program *prg, Tree **sp, FILE *out, Tree *tree, int trim )
 {
-	struct ColmPrintArgs printArgs = { out, 1, 0, &appendFile, 
+	struct ColmPrintArgs printArgs = { out, true, false, trim, &appendFile, 
 			&printNull, &printTermTree, &printNull };
 	printTreeArgs( prg, sp, &printArgs, tree );
 }
 
-void printXmlStdout( Program *prg, Tree **sp, Tree *tree, int commAttr )
+void printXmlStdout( Program *prg, Tree **sp, Tree *tree, int commAttr, int trim )
 {
-	struct ColmPrintArgs printArgs = { stdout, commAttr, commAttr, &appendFile, 
+	struct ColmPrintArgs printArgs = { stdout, commAttr, commAttr, trim, &appendFile, 
 			&openTreeXml, &printTermXml, &closeTreeXml };
 	printTreeArgs( prg, sp, &printArgs, tree );
 }
diff --git a/colm/tree.h b/colm/tree.h
index e897b07f..4425cfc5 100644
--- a/colm/tree.h
+++ b/colm/tree.h
@@ -328,9 +328,6 @@ Tree *treeRevIterPrevChild( struct ColmProgram *prg, Tree ***psp, RevTreeIter *i
 Tree *treeIterNextRepeat( struct ColmProgram *prg, Tree ***psp, TreeIter *iter );
 Tree *treeIterPrevRepeat( struct ColmProgram *prg, Tree ***psp, TreeIter *iter );
 
-void printXmlStdout( struct ColmProgram *prg, Tree **sp, Tree *tree, int commAttr );
-
-
 /* An automatically grown buffer for collecting tokens. Always reuses space;
  * never down resizes. */
 typedef struct _StrCollect
@@ -346,10 +343,9 @@ void strCollectAppend( StrCollect *collect, const char *data, long len );
 void strCollectClear( StrCollect *collect );
 Tree *treeTrim( struct ColmProgram *prg, Tree **sp, Tree *tree );
 
-void printTree( struct ColmProgram *prg, Tree **sp, StrCollect *collect, Tree *tree );
-void printTermTree( struct ColmProgram *prg, Tree **sp, struct ColmPrintArgs *printArgs, Kid *kid );
-void printTreeCollect( struct ColmProgram *prg, Tree **sp, StrCollect *collect, Tree *tree );
-void printTreeFile( struct ColmProgram *prg, Tree **sp, FILE *out, Tree *tree );
+void printTreeCollect( struct ColmProgram *prg, Tree **sp, StrCollect *collect, Tree *tree, int trim );
+void printTreeFile( struct ColmProgram *prg, Tree **sp, FILE *out, Tree *tree, int trim );
+void printXmlStdout( struct ColmProgram *prg, Tree **sp, Tree *tree, int commAttr, int trim );
 
 #if defined(__cplusplus)
 }
diff --git a/test/accum1.lm b/test/accum1.lm
index 716bb20a..186125a3 100644
--- a/test/accum1.lm
+++ b/test/accum1.lm
@@ -23,4 +23,4 @@ for Id: id in Input {
 
 S: start = Output()
 
-print( S )
+print( S '\n' )
diff --git a/test/accum2.lm b/test/accum2.lm
index b17b8557..e1eac4d3 100644
--- a/test/accum2.lm
+++ b/test/accum2.lm
@@ -30,4 +30,4 @@ cons SP: parser<ctx::start> []
 SP.ctx = cons ctx []
 send SP stdin
 Input: ctx::start = SP()
-print( Input )
+print( Input '\n' )
diff --git a/test/context1.lm b/test/context1.lm
index 6f585031..6f703956 100644
--- a/test/context1.lm
+++ b/test/context1.lm
@@ -30,4 +30,4 @@ context ctx
 
 CTX: ctx = cons ctx[]
 Input: ctx::start = parse ctx::start( CTX stdin )
-print( Input )
+print( Input '\n' )
diff --git a/test/func.lm b/test/func.lm
index 52f6f3c5..61665140 100644
--- a/test/func.lm
+++ b/test/func.lm
@@ -27,7 +27,7 @@ int main()
     InputFile: stream = open( 'func.in' "r" )
     P: program = parse program( InputFile )
 	func( P )
-	print( P )
+	print( P '\n' )
 }
 
 main()
diff --git a/test/ignore1.lm b/test/ignore1.lm
index ebf7413f..2922116d 100644
--- a/test/ignore1.lm
+++ b/test/ignore1.lm
@@ -45,9 +45,8 @@ def item_list
 
 parse Attrs: attr_list( stdin )
 
-print( Attrs '\n' )
+print( %Attrs '\n' )
 
-## Reconstruct the left hand side with the 
 construct IL: item_list 
 	["<wrapper .[Attrs]. ></wrapper>"]
 
diff --git a/test/ignore2.lm b/test/ignore2.lm
index 716bb20a..186125a3 100644
--- a/test/ignore2.lm
+++ b/test/ignore2.lm
@@ -23,4 +23,4 @@ for Id: id in Input {
 
 S: start = Output()
 
-print( S )
+print( S '\n' )
diff --git a/test/order2.lm b/test/order2.lm
index 33458524..8f45a17f 100644
--- a/test/order2.lm
+++ b/test/order2.lm
@@ -91,6 +91,6 @@ match P [C: c '%%' protocol*]
 Output << [%C]
 Output <<
 	"#include <assert.h>
-	"
+Output << '\n\n'
 
-print( Output.finish() )
+print( Output.finish() '\n\n' )
diff --git a/test/reparse.lm b/test/reparse.lm
index c6b6a95b..9b70b555 100644
--- a/test/reparse.lm
+++ b/test/reparse.lm
@@ -18,5 +18,5 @@ S: start = cons start[ Input ]
 
 Again: start = parse start( Input )
 
-print( Again )
+print( %Again )
 
diff --git a/test/til.lm b/test/til.lm
index 3dfe8e5b..dba96c1d 100644
--- a/test/til.lm
+++ b/test/til.lm
@@ -160,9 +160,9 @@ for S: statement* in P
                 "	[^First]
                 "	[^Rest]
                 "while [^Expr];
-                ^Following]
+                Following]
 		}
     }
 }
 
-print(P '\n')
+print( %P )
diff --git a/test/translate1.lm b/test/translate1.lm
index 472ef86a..b5c6ca9e 100644
--- a/test/translate1.lm
+++ b/test/translate1.lm
@@ -20,5 +20,5 @@ def start
 	[item*]
 
 Input: start = parse start( stdin )
-print( Input )
+print( %Input )