Rewrite the lexer to be context-aware

It turns out Flex can do modal lexing.  This means that we can switch
mode for optional properties, and also for co-ordinates.  As a result,
the parser is much simpler and doesn't keel over all the time on valid
input.

2 files changed, 116 insertions, 127 deletions

diff --git a/tikzit/src/common/tikzlexer.lm b/tikzit/src/common/tikzlexer.lm
index a9a2e51..3fd7b53 100644
--- a/tikzit/src/common/tikzlexer.lm
+++ b/tikzit/src/common/tikzlexer.lm
@@ -1,5 +1,10 @@
 %option nounput
 %option yylineno
+%s props
+%s xcoord
+%s ycoord
+%s noderef
+FLOAT \-?[0-9]*(\.[0-9]+)?
 
 %{
 //
@@ -7,6 +12,9 @@
 //  TikZiT
 //  
 //  Copyright 2010 Chris Heunen. All rights reserved.
+//  Copyright 2012 Aleks Kissinger
+//  Copyright 2012 KJ
+//  Copyright 2013 Alex Merry
 //  
 //  
 //  This file is part of TikZiT.
@@ -45,48 +53,82 @@ extern int tokenpos;
             tokenpos = 0;
             yyless(1);
         }
-[ ]+                           { tokenpos += yyleng; } /* ignore whitespace */;
-[\t]+                          { tokenpos += 4*yyleng; } /* ignore whitespace */;
-\\begin                        { tokenpos += yyleng; return LATEXBEGIN; }
-\\end                          { tokenpos += yyleng; return LATEXEND; }
-\{tikzpicture\}                { tokenpos += yyleng; return TIKZPICTURE; }
-\{pgfonlayer\}                 { tokenpos += yyleng; return PGFONLAYER; }
-\(                             { tokenpos += yyleng; return LEFTPARENTHESIS; }
-\)                             { tokenpos += yyleng; return RIGHTPARENTHESIS; }	       
-\[                             { tokenpos += yyleng; return LEFTBRACKET; }
-\]                             { tokenpos += yyleng; return RIGHTBRACKET; }
-;                              { tokenpos += yyleng; return SEMICOLON; }
-,                              { tokenpos += yyleng; return COMMA; }
-\.                             { tokenpos += yyleng; return FULLSTOP; }
-=                              { tokenpos += yyleng; return EQUALS; }
-\\draw                         { tokenpos += yyleng; return DRAW; }
-to                             { tokenpos += yyleng; return TO; }
-\\node                         { tokenpos += yyleng; return NODE; }
-\\path                         { tokenpos += yyleng; return PATH; }
-node                           { tokenpos += yyleng; return ALTNODE; }
+<INITIAL,xcoord,ycoord,props,noderef>[ ]+  { tokenpos += yyleng; } /* ignore whitespace */;
+<INITIAL,xcoord,ycoord,props,noderef>[\t]+ { tokenpos += 8*yyleng; } /* ignore whitespace */;
+\\begin\{tikzpicture\}         { tokenpos += yyleng; return BEGIN_TIKZPICTURE_CMD; }
+\\end\{tikzpicture\}           { tokenpos += yyleng; return END_TIKZPICTURE_CMD; }
+\\begin\{pgfonlayer\}          { tokenpos += yyleng; return BEGIN_PGFONLAYER_CMD; }
+\\end\{pgfonlayer\}            { tokenpos += yyleng; return END_PGFONLAYER_CMD; }
+\\draw                         { tokenpos += yyleng; return DRAW_CMD; }
+\\node                         { tokenpos += yyleng; return NODE_CMD; }
+\\path                         { tokenpos += yyleng; return PATH_CMD; }
 rectangle                      { tokenpos += yyleng; return RECTANGLE; }
+node                           { tokenpos += yyleng; return NODE; }
 at                             { tokenpos += yyleng; return AT; }
+to                             { tokenpos += yyleng; return TO; }
+;                              { tokenpos += yyleng; return SEMICOLON; }
 
-[0-9]+ {
-    tokenpos += yyleng;
-	yylval.nsstr=[NSString stringWithUTF8String:yytext];
-	return NATURALNUMBER;
+\([ ]*{FLOAT}[ ]*,[ ]*{FLOAT}[ ]*\) {
+	tokenpos += 1;
+	yyless(1);
+	BEGIN(xcoord);
+}
+<xcoord>{FLOAT} {
+	tokenpos += yyleng;
+	yylval.pt.x=(float)strtod(yytext,NULL);
+	BEGIN(ycoord);
+}
+<ycoord>, { tokenpos += yyleng; }
+<ycoord>{FLOAT} {
+	tokenpos += yyleng;
+	yylval.pt.y=(float)strtod(yytext,NULL);
+}
+<ycoord>\) {
+	tokenpos += yyleng;
+	BEGIN(INITIAL);
+	return COORD;
 }
 
-(\-?[0-9]*\.[0-9]+)|(\-?[0-9]+) {
-    tokenpos += yyleng;
+ /* when we see "[", change parsing mode */
+\[ /*syntaxhlfix]*/ {
+	tokenpos += yyleng;
+	BEGIN(props);
+	return LEFTBRACKET;
+}
+<props>= { tokenpos += yyleng; return EQUALS; }
+<props>, { tokenpos += yyleng; return COMMA; }
+<props>[^=,\{\] \t]([^=,\{\]]*[^=,\{\] \t])? {
+	tokenpos += yyleng;
 	yylval.nsstr=[NSString stringWithUTF8String:yytext];
-	return REALNUMBER;
+	return PROPSTRING;
+}
+<props>\] {
+	tokenpos += yyleng;
+	BEGIN(INITIAL);
+	return RIGHTBRACKET;
 }
 
-\\?[a-zA-Z<>\-\'][a-zA-Z<>\-\'0-9]* { //'
-    tokenpos += yyleng;
+\( {
+	tokenpos += yyleng;
+	BEGIN(noderef);
+	return LEFTPARENTHESIS;
+}
+<noderef>\. {
+	tokenpos += yyleng;
+	return FULLSTOP;
+}
+<noderef>[^\.\{\)]+ {
+	tokenpos += yyleng;
 	yylval.nsstr=[NSString stringWithUTF8String:yytext];
-	return LWORD;
+	return REFSTRING;
+}
+<noderef>\) {
+	tokenpos += yyleng;
+	BEGIN(INITIAL);
+	return RIGHTPARENTHESIS;
 }
 
-
-\{ {
+<INITIAL,props>\{ {
 	NSMutableString *buf = [NSMutableString string];
 	unsigned int brace_depth = 1;
 	unsigned int escape = 0;
diff --git a/tikzit/src/common/tikzparser.ym b/tikzit/src/common/tikzparser.ym
index 2277e23..170106f 100644
--- a/tikzit/src/common/tikzparser.ym
+++ b/tikzit/src/common/tikzparser.ym
@@ -44,55 +44,46 @@ extern void yyerror(const char *str);
 
 %error-verbose
 
-%token LATEXBEGIN
-%token LATEXEND
-%token TIKZPICTURE
-%token PGFONLAYER
+%token BEGIN_TIKZPICTURE_CMD
+%token END_TIKZPICTURE_CMD
+%token BEGIN_PGFONLAYER_CMD
+%token END_PGFONLAYER_CMD
+%token DRAW_CMD
+%token NODE_CMD
+%token PATH_CMD
+%token RECTANGLE
+%token NODE
+%token AT
+%token TO
+%token SEMICOLON
+%token COMMA
+
 %token LEFTPARENTHESIS
 %token RIGHTPARENTHESIS
 %token LEFTBRACKET
 %token RIGHTBRACKET
-%token SEMICOLON
-%token COMMA
 %token FULLSTOP
 %token EQUALS
-%token DRAW
-%token TO
-%token NODE
-%token RECTANGLE
-%token PATH
-%token ALTNODE
-%token AT
-%token REALNUMBER
-%token NATURALNUMBER
-%token LWORD
+%token COORD
+%token PROPSTRING
+%token REFSTRING
 %token DELIMITEDSTRING
 
 %type<nsstr> nodename
-%type<nsstr> anchor
-%type<nsstr> anchorsym
 %type<nsstr> optanchor
-%type<nsstr> nodeid
-%type<pt> coords
-%type<nsstr> propsym
-%type<nsstr> propsyms
 %type<nsstr> val
-%type<nsstr> number
 
 %%
 
-tikzpicture: LATEXBEGIN TIKZPICTURE optproperties expressions LATEXEND TIKZPICTURE;
-expressions: expressions expression | ;
-expression: node | edge | boundingbox | ignore;
-
-ignore: LATEXBEGIN PGFONLAYER DELIMITEDSTRING | LATEXEND PGFONLAYER;
+tikzpicture: BEGIN_TIKZPICTURE_CMD optproperties tikzcmds END_TIKZPICTURE_CMD;
+tikzcmds: tikzcmds tikzcmd | ;
+tikzcmd: node | edge | boundingbox | ignore;
 
-number: REALNUMBER { $$ = $<nsstr>1; } | NATURALNUMBER { $$ = $<nsstr>1; };
+ignore: BEGIN_PGFONLAYER_CMD DELIMITEDSTRING | END_PGFONLAYER_CMD;
 
 optproperties: LEFTBRACKET properties RIGHTBRACKET | ;
 properties: property extraproperties;
-extraproperties: COMMA property extraproperties | property extraproperties | ;
-
+extraproperties: COMMA property extraproperties | ;
 property:
 	val EQUALS val
 	{
@@ -108,79 +99,33 @@ property:
 		[[a data] addObject:p];
 		[p release];
 	};
+val: PROPSTRING { $$ = $<nsstr>1; } | DELIMITEDSTRING { $$ = $<nsstr>1; };
 
-val: propsyms { $$ = $<nsstr>1; } | DELIMITEDSTRING { $$ = $<nsstr>1; };
-propsyms:
-	propsym { $$ = $<nsstr>1; }
-	| propsyms propsym
-	{
-		NSString *s = [$<nsstr>1 stringByAppendingFormat:@" %@", $<nsstr>2];
-		$$ = s;
-	};
-
-propsym:
-	LWORD { $$ = $<nsstr>1; }
-	| number { $$ = $<nsstr>1; }
-	| TO { $$ = @"to"; }
-	| ALTNODE { $$ = @"node"; }
-	| RECTANGLE { $$ = @"rectangle"; }
-	| AT { $$ = @"at"; }
-	| FULLSTOP { $$ = @"."; };
-
-
-nodecmd : NODE { [[TikzGraphAssembler currentAssembler] prepareNode]; };
-
-nodename: LEFTPARENTHESIS nodeid RIGHTPARENTHESIS { $$ = $<nsstr>2; };
-
-node:
-	nodecmd optproperties nodename AT coords nodelabel SEMICOLON
+nodecmd: NODE_CMD { [[TikzGraphAssembler currentAssembler] prepareNode]; };
+nodename: LEFTPARENTHESIS REFSTRING RIGHTPARENTHESIS { $$ = $<nsstr>2; };
+node: nodecmd optproperties nodename AT COORD DELIMITEDSTRING SEMICOLON
 	{
 		TikzGraphAssembler *a = [TikzGraphAssembler currentAssembler];
 		[[a currentNode] setName:$<nsstr>3];
 		[[a currentNode] setPoint:$<pt>5];
+		[[a currentNode] setLabel:$<nsstr>6];
 		[a finishNode];
 	};
 
-nodelabel:
-	DELIMITEDSTRING
+edgecmd : DRAW_CMD { [[TikzGraphAssembler currentAssembler] prepareEdge]; };
+edge: edgecmd optproperties source TO optedgenode target SEMICOLON
 	{
-		Node *n = [[TikzGraphAssembler currentAssembler] currentNode];
-		[n setLabel:$<nsstr>1];
-	}
-
-anchor: anchorsym { $$ = $<nsstr>1; }
-	| anchor anchorsym
-	{
-		NSString *s = [$<nsstr>1 stringByAppendingFormat:@" %@", $<nsstr>2];
-		$$ = s;
+		TikzGraphAssembler *a = [TikzGraphAssembler currentAssembler];
+		[a finishEdge];
 	};
-anchorsym: LWORD { $$ = $<nsstr>1; } | NATURALNUMBER { $$ = $<nsstr>1; };
-
-optanchor:  { $$ = @""; } | FULLSTOP anchor { $$ = $<nsstr>2; };
-
-source: LEFTPARENTHESIS nodeid optanchor RIGHTPARENTHESIS
+optanchor:  { $$ = @""; } | FULLSTOP REFSTRING { $$ = $<nsstr>2; };
+source: LEFTPARENTHESIS REFSTRING optanchor RIGHTPARENTHESIS
     {
         TikzGraphAssembler *a = [TikzGraphAssembler currentAssembler];
         [a setEdgeSource:$<nsstr>2
         anchor:$<nsstr>3];
     };
-
-nodeid: LWORD { $$ = $<nsstr>1; } | NATURALNUMBER { $$ = $<nsstr>1; };
-
-coords:
-	LEFTPARENTHESIS number COMMA number RIGHTPARENTHESIS
-	{
-		$$ = NSMakePoint([$<nsstr>2 floatValue], [$<nsstr>4 floatValue]);
-	};
-
-edgecmd : DRAW { [[TikzGraphAssembler currentAssembler] prepareEdge]; };
-edge:
-	edgecmd optproperties source TO optedgenode target SEMICOLON
-	{
-		TikzGraphAssembler *a = [TikzGraphAssembler currentAssembler];
-		[a finishEdge];
-	};
-target: LEFTPARENTHESIS nodeid optanchor RIGHTPARENTHESIS
+target: LEFTPARENTHESIS REFSTRING optanchor RIGHTPARENTHESIS
     {
         TikzGraphAssembler *a = [TikzGraphAssembler currentAssembler];
         [a setEdgeTarget:$<nsstr>2
@@ -194,18 +139,20 @@ target: LEFTPARENTHESIS nodeid optanchor RIGHTPARENTHESIS
     };
 selfloop: LEFTPARENTHESIS RIGHTPARENTHESIS;
 
-altnodecmd: ALTNODE { [[TikzGraphAssembler currentAssembler] prepareNode]; };
+edgenodecmd: NODE { [[TikzGraphAssembler currentAssembler] prepareNode]; };
 optedgenode:
-	| altnodecmd optproperties nodelabel
+	| edgenodecmd optproperties DELIMITEDSTRING
 	{
-		[[TikzGraphAssembler currentAssembler] finishNode];
+		TikzGraphAssembler *a = [TikzGraphAssembler currentAssembler];
+		[[a currentNode] setLabel:$<nsstr>3];
+		[a finishNode];
 	}
 
-bbox_ignoreprops:
-	| LEFTBRACKET LWORD LWORD LWORD LWORD RIGHTBRACKET;
-
+ignoreprop: val | val EQUALS val;
+ignoreprops: ignoreprop ignoreprops | ;
+optignoreprops: LEFTBRACKET ignoreprops RIGHTBRACKET;
 boundingbox:
-	PATH bbox_ignoreprops coords RECTANGLE coords SEMICOLON
+	PATH_CMD optignoreprops COORD RECTANGLE COORD SEMICOLON
 	{
 		Graph *g = [[TikzGraphAssembler currentAssembler] graph];
 		[g setBoundingBox:NSRectAroundPoints($<pt>3, $<pt>5)];