Guided convergence in Rascal: first steps

grammarware · Jun 17, 2012 · 0f5ef68 · 0f5ef68
1 parent 062817c
commit 0f5ef68
Show file tree

Hide file tree

Showing 18 changed files with 1,441 additions and 0 deletions.
diff --git a/shared/rascal/src/analyse/Metrics.rsc b/shared/rascal/src/analyse/Metrics.rsc
@@ -0,0 +1,20 @@
+@contributor{Vadim Zaytsev - vadim@grammarware.net - SWAT, CWI}
+module analyse::Metrics
+
+import syntax::BGF;
+
+public set[str] allNs(list[BGFProduction] ps) = definedNs(ps) + usedNs(ps);
+public set[str] allNs(BGFGrammar g) = allNs(g.prods);
+
+public set[str] topNs(list[BGFProduction] ps) = definedNs(ps) - usedNs(ps);
+public set[str] topNs(BGFGrammar g) = topNs(g.prods);
+
+public set[str] allTs(list[BGFProduction] ps) = {s | /terminal(str s) := ps};
+public set[str] allTs(BGFGrammar g) = allTs(g.prods);
+
+public set[str] usedNs(list[BGFProduction] ps) = {s | /nonterminal(str s) := ps};
+public set[str] usedNs(BGFGrammar g) = usedNs(g.prods);
+
+public set[str] definedNs(list[BGFProduction] ps) = {s | production(_,str s,_) <- ps};
+public set[str] definedNs(BGFGrammar g) = definedNs(g.prods);
+
diff --git a/shared/rascal/src/converge/Guided.rsc b/shared/rascal/src/converge/Guided.rsc
@@ -0,0 +1,33 @@
+@contributor{Vadim Zaytsev - vadim@grammarware.net - SWAT, CWI}
+module converge::Guided
+
+import IO;
+import Set;
+import Map;
+import List;
+import io::ReadBGF;
+import syntax::BGF;
+import analyse::Metrics;
+
+list[str] sources =
+	["antlr","dcg","ecore","emf","jaxb","om","python","rascal-a","rascal-c","sdf","txl","xsd"];
+
+public void main()
+{
+	map[str,BGFGrammar] bgfs = ();
+	println("Reading the grammars...");
+	for (src <- sources)
+		bgfs[src] = readBGF(|home:///projects/slps/topics/convergence/guided/bgf/<src>.bgf|);
+	println("Parsing the grammarbase with <size(bgfs)> grammars is done.");
+	println("Starting with the root.");
+	for (src <- sources)
+	{
+		println(" * Roots in <src>: <bgfs[src].roots>");
+		if (isEmpty(bgfs[src].roots))
+		{
+			bgfs[src].roots = toList(topNs(bgfs[src]));
+			println("   * Roots in <src> changed to tops: <bgfs[src].roots>");
+		}
+	}
+	println("Done.");
+}
diff --git a/shared/rascal/src/extract/RascalSyntax2BGF.rsc b/shared/rascal/src/extract/RascalSyntax2BGF.rsc
@@ -24,6 +24,8 @@ public void main()
 {
 	Module pt = parse(#Module,|project://fl/src/Concrete.rsc|);
 	Grammar g = modules2grammar("Concrete", {pt});
+	//Grammar g = module2grammar("Concrete", {pt});
+	iprintln(g);
 	iprintln(grammar2grammar(g));
 }
 
@@ -58,6 +60,7 @@ BGFExpression rhs2expr(list[Symbol] seq) = syntax::BGF::sequence([symbol2expr(s)
 BGFExpression symbol2expr(\sort(str x)) = syntax::BGF::nonterminal(x);
 BGFExpression symbol2expr(conditional(\sort(str x),{except(_)})) = syntax::BGF::nonterminal(x); // cannot represent better in BGF
 BGFExpression symbol2expr(\lex(str x)) = syntax::BGF::nonterminal(x);
+BGFExpression symbol2expr(ParseTree::\lit("\n")) = syntax::BGF::terminal("\\n"); //hack?
 BGFExpression symbol2expr(ParseTree::\lit(str x)) = syntax::BGF::terminal(x);
 BGFExpression symbol2expr(\iter-seps(Symbol item, list[Symbol] seps)) = iterseps2expr(item,[s | s <- seps, layouts(_) !:= s]);
 default BGFExpression symbol2expr(Symbol s) = epsilon();

diff --git a/shared/rascal/src/io/ReadBGF.rsc b/shared/rascal/src/io/ReadBGF.rsc
@@ -45,6 +45,7 @@ public BGFExpression mapexpr(Node n)
 		case element(none(),"value",[charData("string")]): return val(string());
 		case element(none(),"value",[charData("int")]): return val(integer());
 		case element(none(),"any",[]): return anything();
+		case element(none(),"terminal",[]): return terminal(" ");// lang::xml::DOM does not preserve whitespace, so this can mean something like a newline
 		case element(none(),"terminal",[charData(str s)]): return terminal(s);
 		case element(none(),"nonterminal",[charData(str s)]): return nonterminal(s);
 		case element(none(),"selectable",[element(none(),"selector",[charData(str s)]),expr]): return selectable(s,mapexpr(expr));

diff --git a/shared/rascal/src/transform/library/Util.rsc b/shared/rascal/src/transform/library/Util.rsc
@@ -97,6 +97,7 @@ public tuple[list[BGFProduction],list[BGFProduction],list[BGFProduction]] splitP
 	return <ps1,ps2,ps3>;
 }
 
+// TODO moved to analyse::Metrics
 public set[str] allNs(list[BGFProduction] ps) = definedNs(ps) + usedNs(ps);
 public set[str] allTs(list[BGFProduction] ps) = {s | /terminal(str s) := ps};
 public set[str] usedNs(list[BGFProduction] ps) = {s | /nonterminal(str s) := ps};

diff --git a/topics/convergence/guided/bgf/Makefile b/topics/convergence/guided/bgf/Makefile
@@ -0,0 +1,20 @@
+build:
+	../../../../shared/tools/antlr2bgf ../../../fl/java1/FL.g antlr.bgf
+	../../../../shared/tools/dcg2bgf ../../../fl/prolog1/Parser.pro dcg.bgf
+	../../../../shared/tools/sdf2bgf ../../../fl/asfsdf/Syntax.sdf sdf.bgf
+	../../../../shared/tools/xsd2bgf ../../../fl/xsd/fl.xsd xsd.bgf
+	../../../../shared/tools/java2bgf ../../../fl/java1 types om.bgf
+	../../../../shared/tools/java2bgf ../../../fl/java3 fl jaxb.bgf
+	../../../../shared/tools/ecore2bgf ../../../fl/emf1/fl.ecore ecore.bgf
+	../../../../shared/tools/ecore2bgf ../../../fl/emf2/model/fl.ecore emf.bgf
+	../../../../shared/tools/py2bgf ../../../fl/python/parser2.py python.bgf
+	../../../../shared/tools/rscs2bgf ../../../fl/rascal/src/Concrete.rsc rascal-c.bgf
+	../../../../shared/tools/rscd2bgf ../../../fl/rascal/src/Abstract.rsc rascal-a.bgf
+	../../../../shared/tools/txl2bgf ../../../fl/txl/FL.Txl txl.bgf
+
+test:
+	ls -1 *.bgf | xargs -n1 ../../../../shared/tools/validate bgf
+
+clean:
+	rm -rf *.bgf
+
diff --git a/topics/convergence/guided/bgf/antlr.bgf b/topics/convergence/guided/bgf/antlr.bgf
@@ -0,0 +1,247 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<bgf:grammar xmlns:bgf="http://planet-sl.org/bgf">
+<bgf:production>
+<nonterminal>program</nonterminal>
+<bgf:expression>
+<plus>
+<bgf:expression>
+<selectable>
+<selector>f</selector>
+<bgf:expression>
+<nonterminal>function</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+</plus>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>function</nonterminal>
+<bgf:expression>
+<sequence>
+<bgf:expression>
+<selectable>
+<selector>n</selector>
+<bgf:expression>
+<nonterminal>ID</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<plus>
+<bgf:expression>
+<selectable>
+<selector>a</selector>
+<bgf:expression>
+<nonterminal>ID</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+</plus>
+</bgf:expression>
+<bgf:expression>
+<terminal>=</terminal>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>e</selector>
+<bgf:expression>
+<nonterminal>expr</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<plus>
+<bgf:expression>
+<nonterminal>NEWLINE</nonterminal>
+</bgf:expression>
+</plus>
+</bgf:expression>
+</sequence>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>expr</nonterminal>
+<bgf:expression>
+<choice>
+<bgf:expression>
+<selectable>
+<selector>b</selector>
+<bgf:expression>
+<nonterminal>binary</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>a</selector>
+<bgf:expression>
+<nonterminal>apply</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>i</selector>
+<bgf:expression>
+<nonterminal>ifThenElse</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+</choice>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>binary</nonterminal>
+<bgf:expression>
+<sequence>
+<bgf:expression>
+<selectable>
+<selector>l</selector>
+<bgf:expression>
+<nonterminal>atom</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<star>
+<bgf:expression>
+<sequence>
+<bgf:expression>
+<selectable>
+<selector>o</selector>
+<bgf:expression>
+<nonterminal>ops</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>r</selector>
+<bgf:expression>
+<nonterminal>atom</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+</sequence>
+</bgf:expression>
+</star>
+</bgf:expression>
+</sequence>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>apply</nonterminal>
+<bgf:expression>
+<sequence>
+<bgf:expression>
+<selectable>
+<selector>i</selector>
+<bgf:expression>
+<nonterminal>ID</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<plus>
+<bgf:expression>
+<selectable>
+<selector>a</selector>
+<bgf:expression>
+<nonterminal>atom</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+</plus>
+</bgf:expression>
+</sequence>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>ifThenElse</nonterminal>
+<bgf:expression>
+<sequence>
+<bgf:expression>
+<terminal>if</terminal>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>c</selector>
+<bgf:expression>
+<nonterminal>expr</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<terminal>then</terminal>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>e1</selector>
+<bgf:expression>
+<nonterminal>expr</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<terminal>else</terminal>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>e2</selector>
+<bgf:expression>
+<nonterminal>expr</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+</sequence>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>atom</nonterminal>
+<bgf:expression>
+<choice>
+<bgf:expression>
+<nonterminal>ID</nonterminal>
+</bgf:expression>
+<bgf:expression>
+<nonterminal>INT</nonterminal>
+</bgf:expression>
+<bgf:expression>
+<sequence>
+<bgf:expression>
+<terminal>(</terminal>
+</bgf:expression>
+<bgf:expression>
+<selectable>
+<selector>e</selector>
+<bgf:expression>
+<nonterminal>expr</nonterminal>
+</bgf:expression>
+</selectable>
+</bgf:expression>
+<bgf:expression>
+<terminal>)</terminal>
+</bgf:expression>
+</sequence>
+</bgf:expression>
+</choice>
+</bgf:expression>
+</bgf:production>
+<bgf:production>
+<nonterminal>ops</nonterminal>
+<bgf:expression>
+<choice>
+<bgf:expression>
+<terminal>==</terminal>
+</bgf:expression>
+<bgf:expression>
+<terminal>+</terminal>
+</bgf:expression>
+<bgf:expression>
+<terminal>-</terminal>
+</bgf:expression>
+</choice>
+</bgf:expression>
+</bgf:production>
+</bgf:grammar>