longest-match struct split between libfsm and here in ragel

Moved param check code to mainline as part of trimming down common file.
NameInst structs live in libfsm and can be removed from ragel.

7 files changed, 155 insertions, 203 deletions

diff --git a/src/common.cc b/src/common.cc
index 6e0f5c0c..548d094f 100644
--- a/src/common.cc
+++ b/src/common.cc
@@ -139,135 +139,6 @@ HostType *findAlphTypeInternal( const HostLang *hostLang, const char *s1 )
 	return 0;
 }
 
-/* Construct a new parameter checker with for paramSpec. */
-ParamCheck::ParamCheck( const char *paramSpec, int argc, const char **argv )
-:
-	state(noparam),
-	argOffset(0),
-	curArg(0),
-	iCurArg(1),
-	paramSpec(paramSpec), 
-	argc(argc), 
-	argv(argv)
-{
-}
-
-/* Check a single option. Returns the index of the next parameter.  Sets p to
- * the arg character if valid, 0 otherwise.  Sets parg to the parameter arg if
- * there is one, NULL otherwise. */
-bool ParamCheck::check()
-{
-	bool requiresParam;
-
-	if ( iCurArg >= argc ) {            /* Off the end of the arg list. */
-		state = noparam;
-		return false;
-	}
-
-	if ( argOffset != 0 && *argOffset == 0 ) {
-		/* We are at the end of an arg string. */
-		iCurArg += 1;
-		if ( iCurArg >= argc ) {
-			state = noparam;
-			return false;
-		}
-		argOffset = 0;
-	}
-
-	if ( argOffset == 0 ) {
-		/* Set the current arg. */
-		curArg = argv[iCurArg];
-
-		/* We are at the beginning of an arg string. */
-		if ( argv[iCurArg] == 0 ||        /* Argv[iCurArg] is null. */
-			 argv[iCurArg][0] != '-' ||   /* Not a param. */
-			 argv[iCurArg][1] == 0 ) {    /* Only a dash. */
-			parameter = 0;
-			paramArg = 0;
-
-			iCurArg += 1;
-			state = noparam;
-			return true;
-		}
-		argOffset = argv[iCurArg] + 1;
-	}
-
-	/* Get the arg char. */
-	char argChar = *argOffset;
-	
-	/* Loop over all the parms and look for a match. */
-	const char *pSpec = paramSpec;
-	while ( *pSpec != 0 ) {
-		char pSpecChar = *pSpec;
-
-		/* If there is a ':' following the char then
-		 * it requires a parm.  If a parm is required
-		 * then move ahead two in the parmspec. Otherwise
-		 * move ahead one in the parm spec. */
-		if ( pSpec[1] == ':' ) {
-			requiresParam = true;
-			pSpec += 2;
-		}
-		else {
-			requiresParam = false;
-			pSpec += 1;
-		}
-
-		/* Do we have a match. */
-		if ( argChar == pSpecChar ) {
-			if ( requiresParam ) {
-				if ( argOffset[1] == 0 ) {
-					/* The param must follow. */
-					if ( iCurArg + 1 == argc ) {
-						/* We are the last arg so there
-						 * cannot be a parameter to it. */
-						parameter = argChar;
-						paramArg = 0;
-						iCurArg += 1;
-						argOffset = 0;
-						state = invalid;
-						return true;
-					}
-					else {
-						/* the parameter to the arg is the next arg. */
-						parameter = pSpecChar;
-						paramArg = argv[iCurArg + 1];
-						iCurArg += 2;
-						argOffset = 0;
-						state = match;
-						return true;
-					}
-				}
-				else {
-					/* The param for the arg is built in. */
-					parameter = pSpecChar;
-					paramArg = argOffset + 1;
-					iCurArg += 1;
-					argOffset = 0;
-					state = match;
-					return true;
-				}
-			}
-			else {
-				/* Good, we matched the parm and no
-				 * arg is required. */
-				parameter = pSpecChar;
-				paramArg = 0;
-				argOffset += 1;
-				state = match;
-				return true;
-			}
-		}
-	}
-
-	/* We did not find a match. Bad Argument. */
-	parameter = argChar;
-	paramArg = 0;
-	argOffset += 1;
-	state = invalid;
-	return true;
-}
-
 std::streamsize output_filter::countAndWrite( const char *s, std::streamsize n )
 {
 	for ( int i = 0; i < n; i++ ) {
diff --git a/src/inputdata.cc b/src/inputdata.cc
index d7e075d8..33a0b5b8 100644
--- a/src/inputdata.cc
+++ b/src/inputdata.cc
@@ -29,6 +29,8 @@
 #include "reducer.h"
 #include "version.h"
 #include "pcheck.h"
+#include <libfsm/dot.h>
+
 #include <colm/colm.h>
 
 #include <stdlib.h>
@@ -265,6 +267,13 @@ void InputData::closeOutput()
 	}
 }
 
+void InputData::writeDot( ostream &out )
+{
+	ParseData *pd = dotGenPd;
+	GraphvizDotGen dotGen( this, pd->fsmCtx, pd->sectionGraph, pd->sectionName, pd->machineId, out );
+	dotGen.write();
+}
+
 void InputData::processDot()
 {
 	/* Compiles the DOT machines. */
@@ -678,6 +687,136 @@ void escapeLineDirectivePath( std::ostream &out, char *path )
 	}
 }
 
+/* Construct a new parameter checker with for paramSpec. */
+ParamCheck::ParamCheck( const char *paramSpec, int argc, const char **argv )
+:
+	state(noparam),
+	argOffset(0),
+	curArg(0),
+	iCurArg(1),
+	paramSpec(paramSpec), 
+	argc(argc), 
+	argv(argv)
+{
+}
+
+/* Check a single option. Returns the index of the next parameter.  Sets p to
+ * the arg character if valid, 0 otherwise.  Sets parg to the parameter arg if
+ * there is one, NULL otherwise. */
+bool ParamCheck::check()
+{
+	bool requiresParam;
+
+	if ( iCurArg >= argc ) {            /* Off the end of the arg list. */
+		state = noparam;
+		return false;
+	}
+
+	if ( argOffset != 0 && *argOffset == 0 ) {
+		/* We are at the end of an arg string. */
+		iCurArg += 1;
+		if ( iCurArg >= argc ) {
+			state = noparam;
+			return false;
+		}
+		argOffset = 0;
+	}
+
+	if ( argOffset == 0 ) {
+		/* Set the current arg. */
+		curArg = argv[iCurArg];
+
+		/* We are at the beginning of an arg string. */
+		if ( argv[iCurArg] == 0 ||        /* Argv[iCurArg] is null. */
+			 argv[iCurArg][0] != '-' ||   /* Not a param. */
+			 argv[iCurArg][1] == 0 ) {    /* Only a dash. */
+			parameter = 0;
+			paramArg = 0;
+
+			iCurArg += 1;
+			state = noparam;
+			return true;
+		}
+		argOffset = argv[iCurArg] + 1;
+	}
+
+	/* Get the arg char. */
+	char argChar = *argOffset;
+	
+	/* Loop over all the parms and look for a match. */
+	const char *pSpec = paramSpec;
+	while ( *pSpec != 0 ) {
+		char pSpecChar = *pSpec;
+
+		/* If there is a ':' following the char then
+		 * it requires a parm.  If a parm is required
+		 * then move ahead two in the parmspec. Otherwise
+		 * move ahead one in the parm spec. */
+		if ( pSpec[1] == ':' ) {
+			requiresParam = true;
+			pSpec += 2;
+		}
+		else {
+			requiresParam = false;
+			pSpec += 1;
+		}
+
+		/* Do we have a match. */
+		if ( argChar == pSpecChar ) {
+			if ( requiresParam ) {
+				if ( argOffset[1] == 0 ) {
+					/* The param must follow. */
+					if ( iCurArg + 1 == argc ) {
+						/* We are the last arg so there
+						 * cannot be a parameter to it. */
+						parameter = argChar;
+						paramArg = 0;
+						iCurArg += 1;
+						argOffset = 0;
+						state = invalid;
+						return true;
+					}
+					else {
+						/* the parameter to the arg is the next arg. */
+						parameter = pSpecChar;
+						paramArg = argv[iCurArg + 1];
+						iCurArg += 2;
+						argOffset = 0;
+						state = match;
+						return true;
+					}
+				}
+				else {
+					/* The param for the arg is built in. */
+					parameter = pSpecChar;
+					paramArg = argOffset + 1;
+					iCurArg += 1;
+					argOffset = 0;
+					state = match;
+					return true;
+				}
+			}
+			else {
+				/* Good, we matched the parm and no
+				 * arg is required. */
+				parameter = pSpecChar;
+				paramArg = 0;
+				argOffset += 1;
+				state = match;
+				return true;
+			}
+		}
+	}
+
+	/* We did not find a match. Bad Argument. */
+	parameter = argChar;
+	paramArg = 0;
+	argOffset += 1;
+	state = invalid;
+	return true;
+}
+
+
 void InputData::parseArgs( int argc, const char **argv )
 {
 	ParamCheck pc( "o:dnmleabjkS:M:I:vHh?-:sT:F:W:G:LpV", argc, argv );
diff --git a/src/parsedata.h b/src/parsedata.h
index 1df3aa2e..109f36e6 100644
--- a/src/parsedata.h
+++ b/src/parsedata.h
@@ -115,65 +115,12 @@ typedef AvlMap<std::string, int, CmpString> PriorDict;
 typedef AvlMapEl<std::string, int> LocalErrDictEl;
 typedef AvlMap<std::string, int, CmpString> LocalErrDict;
 
-struct NameMapVal
-{
-	Vector<NameInst*> vals;
-};
-
 /* Tree of instantiated names. */
 typedef AvlMapEl<std::string, NameMapVal*> NameMapEl;
 typedef AvlMap<std::string, NameMapVal*, CmpString> NameMap;
 typedef Vector<NameInst*> NameVect;
 typedef BstSet<NameInst*> NameSet;
 
-/* Node in the tree of instantiated names. */
-struct NameInst
-{
-	NameInst( const InputLoc &loc, NameInst *parent, std::string name, int id, bool isLabel ) : 
-		loc(loc), parent(parent), name(name), id(id), isLabel(isLabel),
-		isLongestMatch(false), numRefs(0), numUses(0), start(0), final(0) {}
-
-	~NameInst();
-
-	InputLoc loc;
-
-	/* Keep parent pointers in the name tree to retrieve 
-	 * fully qulified names. */
-	NameInst *parent;
-
-	std::string name;
-	int id;
-	bool isLabel;
-	bool isLongestMatch;
-
-	int numRefs;
-	int numUses;
-
-	/* Names underneath us, excludes anonymous names. */
-	NameMap children;
-
-	/* All names underneath us in order of appearance. */
-	NameVect childVect;
-
-	/* Join scopes need an implicit "final" target. */
-	NameInst *start, *final;
-
-	/* During a fsm generation walk, lists the names that are referenced by
-	 * epsilon operations in the current scope. After the link is made by the
-	 * epsilon reference and the join operation is complete, the label can
-	 * have its refcount decremented. Once there are no more references the
-	 * entry point can be removed from the fsm returned. */
-	NameVect referencedNames;
-
-	/* Pointers for the name search queue. */
-	NameInst *prev, *next;
-
-	/* Check if this name inst or any name inst below is referenced. */
-	bool anyRefsRec();
-};
-
-typedef DList<NameInst> NameInstList;
-
 /* Stack frame used in walking the name tree. */
 struct NameFrame 
 {
diff --git a/src/parsetree.cc b/src/parsetree.cc
index f51c35e8..f882bc2c 100644
--- a/src/parsetree.cc
+++ b/src/parsetree.cc
@@ -30,6 +30,7 @@
 
 /* Parsing. */
 #include <libfsm/ragel.h>
+#include <libfsm/action.h>
 #include "parsetree.h"
 #include "parsedata.h"
 
diff --git a/src/parsetree.h b/src/parsetree.h
index 37f2fda7..9c47c0c2 100644
--- a/src/parsetree.h
+++ b/src/parsetree.h
@@ -151,7 +151,6 @@ struct InlineItem;
 struct InlineList;
 
 /* Reference to a named state. */
-struct NameRef : public Vector<std::string> {};
 typedef Vector<NameRef*> NameRefList;
 typedef Vector<NameInst*> NameTargList;
 
@@ -284,30 +283,21 @@ struct VarDef
  * and consume the current character.
  */
 struct LongestMatchPart
+:
+	public FsmLongestMatchPart
 {
 	LongestMatchPart( Join *join, Action *action, 
 			const InputLoc &semiLoc, int longestMatchId )
 	: 
-		join(join), action(action), semiLoc(semiLoc), 
-		longestMatchId(longestMatchId), inLmSelect(false) { }
+		FsmLongestMatchPart( action, longestMatchId ),
+		join(join), semiLoc(semiLoc)
+	{ }
 
 	InputLoc getLoc();
 	
 	Join *join;
-	Action *action;
 	InputLoc semiLoc;
 
-	Action *setActId;
-	Action *actOnLast;
-	Action *actOnNext;
-	Action *actLagBehind;
-	Action *actNfaOnLast;
-	Action *actNfaOnNext;
-	Action *actNfaOnEof;
-	int longestMatchId;
-	bool inLmSelect;
-	LongestMatch *longestMatch;
-
 	LongestMatchPart *prev, *next;
 };
 
@@ -315,21 +305,25 @@ struct LongestMatchPart
 struct LmPartList : DList<LongestMatchPart> {};
 
 struct LongestMatch
+:
+	public FsmLongestMatch
 {
 	/* Construct with a list of joins */
 	LongestMatch( const InputLoc &loc, LmPartList *longestMatchList )
 	: 
+		FsmLongestMatch( new FsmLmPartList ),
 		loc(loc),
 		longestMatchList(longestMatchList),
-		lmSwitchHandlesError(false),
 		nfaConstruction(false)
-	{ }
+	{
+		for ( LongestMatchPart *lmPart = longestMatchList->head; lmPart != 0; lmPart = lmPart->next )
+			FsmLongestMatch::longestMatchList->append( lmPart );
+	}
 
 	InputLoc loc;
 	LmPartList *longestMatchList;
 	std::string name;
 	Action *lmActSelect;
-	bool lmSwitchHandlesError;
 	bool nfaConstruction;
 
 	LongestMatch *next, *prev;
diff --git a/test/Makefile.am b/test/Makefile.am
index abaa35f0..d70e20ae 100644
--- a/test/Makefile.am
+++ b/test/Makefile.am
@@ -1,8 +1,8 @@
 SUBDIRS = . ragel.d
 
-noinst_SCRIPTS = subject.mk subject.sh
+noinst_SCRIPTS = runtests subject.mk subject.sh
 
-EXTRA_DIST = subject.mk.in subject.sh.in runtests
+EXTRA_DIST = subject.mk.in subject.sh.in runtests.sh
 
 subject.mk: subject.mk.in Makefile
 	@$(top_srcdir)/sedsubst $< $@ -w,+x $(SED_SUBST)
diff --git a/test/ragel.d/Makefile.am b/test/ragel.d/Makefile.am
index a95381c5..f6845367 100644
--- a/test/ragel.d/Makefile.am
+++ b/test/ragel.d/Makefile.am
@@ -19,7 +19,7 @@
 #   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA 
 
 COLM_BIN = @COLM@
-RAGEL_LM = ../../ragel
+RAGEL_LM = ../../src
 
 COLM_xCPPFLAGS = # -I../../colm/include
 COLM_xLDFLAGS = # -L../../colm