X-Git-Url: http://git.megacz.com/?p=sbp.git;a=blobdiff_plain;f=src%2Fedu%2Fberkeley%2Fsbp%2Fmeta%2FGrammarAST.java;h=586d519d607defd2cefc7de8d1eefe6c7727c814;hp=997d0b6230b1f06fc05ec4fd5742c86aaa3aef10;hb=f95343b3b6e918434ff68924e79df22b9cb074e8;hpb=61a0c83fd40b98292b2dfe1eaba237eb804b2cb4 diff --git a/src/edu/berkeley/sbp/meta/GrammarAST.java b/src/edu/berkeley/sbp/meta/GrammarAST.java index 997d0b6..586d519 100644 --- a/src/edu/berkeley/sbp/meta/GrammarAST.java +++ b/src/edu/berkeley/sbp/meta/GrammarAST.java @@ -1,4 +1,4 @@ -// Copyright 2006 all rights reserved; see LICENSE file for BSD-style license +// Copyright 2006-2007 all rights reserved; see LICENSE file for BSD-style license package edu.berkeley.sbp.meta; import edu.berkeley.sbp.util.*; @@ -16,6 +16,18 @@ import java.io.*; */ public class GrammarAST { + public static interface ImportResolver { + public InputStream getImportStream(String importname); + } + + /** + * Returns a Union representing the metagrammar (meta.g); the Tree produced by + * parsing with this Union should be provided to buildFromAST + */ + public static Union getMetaGrammar() { + return buildFromAST(MetaGrammar.meta, "s", null); + } + /** * Create a grammar from a parse tree and binding resolver * @@ -23,36 +35,37 @@ public class GrammarAST { * @param s the name of the "start symbol" * @param gbr a GrammarBindingResolver that resolves grammatical reductions into tree-node-heads */ - public static Union buildFromAST(Tree grammarAST, String startingNonterminal, File[] includes) { - return new GrammarAST(includes, "").buildGrammar(grammarAST, startingNonterminal); + public static Union buildFromAST(Tree grammarAST, String startingNonterminal, ImportResolver resolver) { + return new GrammarAST(resolver, "").buildGrammar(grammarAST, startingNonterminal); } - public static Object illegalTag = ""; // this is the tag that should never appear in the non-dropped output FIXME + private static Object illegalTag = ""; // this is the tag that should never appear in the non-dropped output FIXME + + // Instance ////////////////////////////////////////////////////////////////////////////// private final String prefix; - private final File[] includes; + private final ImportResolver resolver; - public GrammarAST(File[] includes, String prefix) { + public GrammarAST(ImportResolver resolver, String prefix) { this.prefix = prefix; - this.includes = includes; + this.resolver = resolver; } // Methods ////////////////////////////////////////////////////////////////////////////// private Union buildGrammar(Tree t, String rootNonTerminal) { - return ((GrammarAST.GrammarNode)walk(t)).build(rootNonTerminal); + Object o = walk(t); + if (o instanceof Union) return (Union)o; + return ((GrammarAST.GrammarNode)o).build(rootNonTerminal); } - public Object[] walkChildren(Tree t) { + private Object[] walkChildren(Tree t) { Object[] ret = new Object[t.size()]; - for(int i=0; i 0) head = head.substring(head.indexOf('.')+1); @@ -77,36 +93,34 @@ public class GrammarAST { if (head.equals("Grammar")) return new GrammarNode(walkChildren(t)); if (head.equals("(")) return new UnionNode((Seq[][])walkChildren(t.child(0))); if (head.equals("Word")) return stringifyChildren(t); - if (head.equals("Elements")) return seq2((ElementNode[])Reflection.rebuild(walkChildren(t), ElementNode[].class)); + if (head.equals("Elements")) return new Seq((ElementNode[])Reflection.rebuild(walkChildren(t), ElementNode[].class)); if (head.equals("NonTerminalReference")) return new ReferenceNode(stringifyChildren(t.child(0))); - //if (head.equals(")")) return new ReferenceNode(stringifyChildren(t.child(0))); - if (head.equals("{")) return new XTree((Seq)walk(t.child(0))); - if (head.equals("::")) return tag((String)walk(t.child(0)), (Seq)walk(t.child(1))); - if (head.equals("++")) return plusmax((ElementNode)walk(t.child(0))); - if (head.equals("...")) return star(new TildeNode(new AtomNode())); - if (head.equals("+")) return plus((ElementNode)walk(t.child(0))); - if (head.equals("++/")) return plusmaxfollow((ElementNode)walk(t.child(0)), (ElementNode)walk(t.child(1))); - if (head.equals("+/")) return plusfollow((ElementNode)walk(t.child(0)), (ElementNode)walk(t.child(1))); - if (head.equals("**")) return starmax((ElementNode)walk(t.child(0))); - if (head.equals("*")) return star((ElementNode)walk(t.child(0))); - if (head.equals("**/")) return starmaxfollow((ElementNode)walk(t.child(0)), (ElementNode)walk(t.child(1))); - if (head.equals("*/")) return starfollow((ElementNode)walk(t.child(0)), (ElementNode)walk(t.child(1))); - if (head.equals("?")) return question((ElementNode)walk(t.child(0))); - if (head.equals("!")) return new DropNode((ElementNode)walk(t.child(0))); - if (head.equals("^")) return new LiteralNode((String)walk(t.child(0)), true); - if (head.equals("`")) { - ElementNode ret = (ElementNode)walk(t.child(0)); - ret.lifted = true; - return ret; - } + if (head.equals(")")) return new ReferenceNode(stringifyChildren(t.child(0)), true); + if (head.equals(":")) return new LabelNode(stringifyChildren(t.child(0)), walkElement(t.child(1))); + if (head.equals("::")) return walkSeq(t.child(1)).tag(walkString(t.child(0))); + if (head.equals("...")) return new DropNode(new RepeatNode(new TildeNode(new AtomNode()), null, true, true, false)); + + if (head.equals("++")) return new RepeatNode(walkElement(t.child(0)), null, false, true, true); + if (head.equals("+")) return new RepeatNode(walkElement(t.child(0)), null, false, true, false); + if (head.equals("++/")) return new RepeatNode(walkElement(t.child(0)), walkElement(t.child(1)), false, true, true); + if (head.equals("+/")) return new RepeatNode(walkElement(t.child(0)), walkElement(t.child(1)), false, true, false); + if (head.equals("**")) return new RepeatNode(walkElement(t.child(0)), null, true, true, true); + if (head.equals("*")) return new RepeatNode(walkElement(t.child(0)), null, true, true, false); + if (head.equals("**/")) return new RepeatNode(walkElement(t.child(0)), walkElement(t.child(1)), true, true, true); + if (head.equals("*/")) return new RepeatNode(walkElement(t.child(0)), walkElement(t.child(1)), true, true, false); + if (head.equals("?")) return new RepeatNode(walkElement(t.child(0)), null, true, false, false); + + if (head.equals("!")) return new DropNode(walkElement(t.child(0))); + if (head.equals("^")) return new LiteralNode(walkString(t.child(0)), true); + if (head.equals("`")) return new BacktickNode(walkElement(t.child(0))); if (head.equals("Quoted")) return stringifyChildren(t); if (head.equals("Literal")) return new LiteralNode(walkString(t.child(0))); if (head.equals("->")) return walkSeq(t.child(0)).follow(walkElement(t.child(1))); if (head.equals("DropNT")) return new NonTerminalNode(walkString(t.child(0)), (Seq[][])walkChildren(t.child(1)), false, null, true); if (head.equals("=")) return new NonTerminalNode(walkString(t.child(0)), (Seq[][])walk(t.child(2)), true, t.size()==2 ? null : walkString(t.child(1)), false); - if (head.equals("&")) return and2(walkSeq(t.child(0)), walkSeq(t.child(1))); - if (head.equals("&~")) return andnot2(walkSeq(t.child(0)), walkSeq(t.child(1))); + if (head.equals("&")) return walkSeq(t.child(0)).and(walkSeq(t.child(1))); + if (head.equals("&~")) return walkSeq(t.child(0)).andnot(walkSeq(t.child(1))); if (head.equals("/")) return (walkSeq(t.child(0))).separate(walkElement(t.child(1))); if (head.equals("()")) return new LiteralNode(""); if (head.equals("[")) return new AtomNode((char[][])Reflection.rebuild(walkChildren(t), char[][].class)); @@ -116,53 +130,48 @@ public class GrammarAST { if (head.equals("<<")) return new DropNode(new AtomNode(new char[] { CharAtom.right, CharAtom.right })); if (head.equals("~")) return new TildeNode(walkElement(t.child(0))); if (head.equals("~~")) return new Seq(new RepeatNode(new TildeNode(new AtomNode()), null, true, true, false)).andnot(walkSeq(t.child(0))); - if (head.equals("Range") && t.size()==2 && ">".equals(t.child(0).head())) - return new char[] { CharAtom.left, CharAtom.left }; - if (head.equals("Range") && t.size()==2 && "<".equals(t.child(0).head())) - return new char[] { CharAtom.right, CharAtom.right }; - if (head.equals("Range") && t.size()==1) return new char[] { unescape(t).charAt(0), unescape(t).charAt(0) }; - if (head.equals("Range")) return new char[] { unescape(t).charAt(0), unescape(t).charAt(1) }; + if (head.equals("Range")) { + if (t.size()==2 && ">".equals(t.child(0).head())) return new char[] { CharAtom.left, CharAtom.left }; + if (t.size()==2 && "<".equals(t.child(0).head())) return new char[] { CharAtom.right, CharAtom.right }; + if (t.size()==1) return new char[] { unescape(t).charAt(0), unescape(t).charAt(0) }; + return new char[] { unescape(t).charAt(0), unescape(t).charAt(1) }; + } if (head.equals("\"\"")) return ""; - if (head.equals("\n")) return "\n"; - if (head.equals("\r")) return "\r"; - if (head.equals("grammar.Grammar")) return walkChildren(t); - if (head.equals("SubGrammar")) return GrammarBuilder.buildFromAST(t.child(0), "s", includes); + if (head.equals("\n")) return "\n"; + if (head.equals("\t")) return "\t"; + if (head.equals("\r")) return "\r"; + if (head.equals("SubGrammar")) return GrammarAST.buildFromAST(t.child(0), "s", resolver); if (head.equals("NonTerminal")) - return new NonTerminalNode((String)walk(t.child(0)), + return new NonTerminalNode(walkString(t.child(0)), (Seq[][])walkChildren(t.child(1)), false, null, false); if (head.equals("Colons")) { - String tag = (String)walk(t.child(0)); + String tag = walkString(t.child(0)); Seq[][] seqs = (Seq[][])walk(t.child(1)); for(Seq[] seq : seqs) for(int i=0; i " + (head.equals("..."))); } + // Nodes ////////////////////////////////////////////////////////////////////////////// @@ -198,23 +207,47 @@ public class GrammarAST { public Union build(String rootNonterminal) { Context cx = new Context(this); Union u = null; - for(GrammarBuilder.NonTerminalNode nt : values()) + for(GrammarAST.NonTerminalNode nt : values()) if (nt.name.equals(rootNonterminal)) return (Union)cx.get(nt.name); return null; } } - public class UnionNode extends ElementNode { + /** a node in the AST which is resolved into an Element */ + private abstract class ElementNode { + public boolean isLifted() { return false; } + public boolean isDropped(Context cx) { return false; } + public Atom toAtom(Context cx) { throw new Error("can't convert a " + this.getClass().getName() + " to an atom: " + this); } + public abstract Element build(Context cx, NonTerminalNode cnt, boolean dropall); + } + + /** a union, produced by a ( .. | .. | .. ) construct */ + private class UnionNode extends ElementNode { + + /** each component of a union is a sequence */ public Seq[][] sequences; - public String sep = null; + + /** if the union is a NonTerminal specified as Foo*=..., this is true */ public boolean rep; + + /** if the union is a NonTerminal specified as Foo* /ws=..., then this is "ws" */ + public String sep = null; + + public UnionNode(Seq seq) { this(new Seq[][] { new Seq[] { seq } }); } public UnionNode(Seq[][] sequences) { this(sequences, false, null); } public UnionNode(Seq[][] sequences, boolean rep, String sep) { this.sequences = sequences; this.rep = rep; this.sep = sep; } + public boolean isDropped(Context cx) { + for(Seq[] seqs : sequences) + for(Seq seq : seqs) + if (!seq.isDropped(cx)) + return false; + return true; + } public Atom toAtom(Context cx) { Atom ret = null; for(Seq[] ss : sequences) @@ -228,7 +261,7 @@ public class GrammarAST { Union urep = null; if (rep) { urep = new Union(null, false); - urep.add(Sequence.create(new Element[0], cnt.name)); + urep.add(Sequence.create(cnt.name, new Element[0])); urep.add(sep==null ? Sequence.create(new Element[] { u }, 0) : Sequence.create(new Element[] { cx.get(sep), u }, 1)); @@ -247,7 +280,7 @@ public class GrammarAST { Sequence s = Sequence.create(cnt.name, new Element[] { u3, urep }, new boolean[] { false, false }, - true); + new boolean[] { false, true}); u2.add(s); } if (sequences.length==1) break; @@ -260,10 +293,11 @@ public class GrammarAST { } } - public class NonTerminalNode extends UnionNode { + /** a NonTerminal is always a union at the top level */ + private class NonTerminalNode extends UnionNode { public boolean alwaysDrop; public String name = null; - public boolean drop(Context cx) { return alwaysDrop; } + public boolean isDropped(Context cx) { return alwaysDrop; } public NonTerminalNode(String name, Seq[][] sequences, boolean rep, String sep, boolean alwaysDrop) { super(sequences, rep, sep==null?null:(prefix + sep)); this.name = prefix + name; @@ -272,20 +306,54 @@ public class GrammarAST { public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return cx.get(name); } } - public class Seq { - public boolean alwaysDrop = false; - public boolean drop(Context cx) { return alwaysDrop; } - HashSet and = new HashSet(); - HashSet not = new HashSet(); + /** a sequence */ + private class Seq { + /** elements of the sequence */ ElementNode[] elements; + /** follow-set, if explicit */ ElementNode follow; + /** tag to add when building the AST */ String tag = null; - public Seq(ElementNode e) { this(new ElementNode[] { e }); } - public Seq(ElementNode[] elements) { - this.elements = elements; + /** positive conjuncts */ + HashSet and = new HashSet(); + /** negative conjuncts */ + HashSet not = new HashSet(); + public boolean alwaysDrop = false; + public boolean isDropped(Context cx) { + if (alwaysDrop) return true; + if (tag!=null) return false; for(int i=0; i 0 && elements[0].lifted) - lift = true; + throw new RuntimeException("multiple non-dropped elements in sequence: " + Sequence.create("", els)); + boolean[] lifts = new boolean[elements.length]; + for(int i=0; i)_e.toAtom(cx).complement()); } public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return toAtom(cx); } } - public class DropNode extends ElementNodeWrapper { + private class DropNode extends ElementNodeWrapper { public DropNode(ElementNode e) { super(e); } - public boolean drop(Context cx) { return true; } + public boolean isDropped(Context cx) { return true; } } - public Seq and2(Seq s, Seq a) { a.alwaysDrop = true; return s.and(a); } - public Seq andnot2(Seq s, Seq a) { a.alwaysDrop = true; return s.andnot(a); } - public Seq arrow(Seq s, ElementNode e) { return s.follow(e); } - public Seq tag(String tagname, Seq s) { return s.tag(tagname); } - public Seq seq(ElementNode[] elements) { return new Seq(elements); } - public Seq seq2(ElementNode[] elements) { return new Seq(elements); } - public ElementNode plusmax(final ElementNode e) { return new RepeatNode(e, null, false, true, true); } - public ElementNode plus(final ElementNode e) { return new RepeatNode(e, null, false, true, false); } - public ElementNode plusmaxfollow(final ElementNode e, final ElementNode sep) { return new RepeatNode(e, sep, false, true, true); } - public ElementNode plusfollow(final ElementNode e, final ElementNode sep) { return new RepeatNode(e, sep, false, true, false); } - public ElementNode starmax(final ElementNode e) { return new RepeatNode(e, null, true, true, true); } - public ElementNode star(final ElementNode e) { return new RepeatNode(e, null, true, true, false); } - public ElementNode starmaxfollow(final ElementNode e, final ElementNode sep) { return new RepeatNode(e, sep, true, true, true); } - public ElementNode starfollow(final ElementNode e, final ElementNode sep) { return new RepeatNode(e, sep, true, true, false); } - public ElementNode question(final ElementNode e) { return new RepeatNode(e, null, true, false, false); } + /** provides a label on the fields of a Seq */ + private class LabelNode extends ElementNodeWrapper { + public final String label; + public LabelNode(String label, ElementNode e) { super(e); this.label = label; } + } ////////////////////////////////////////////////////////////////////////////// @@ -523,7 +580,7 @@ public class GrammarAST { } else { ret = new Union(name, false); map.put(name, ret); - nt.buildIntoPreallocatedUnion(this, nt, nt.drop(this), ret); + nt.buildIntoPreallocatedUnion(this, nt, nt.isDropped(this), ret); } return ret; }