*/
public class GrammarAST {
+ public static interface ImportResolver {
+ public InputStream getImportStream(String importname);
+ }
+
/**
* Returns a Union representing the metagrammar (<tt>meta.g</tt>); the Tree produced by
* parsing with this Union should be provided to <tt>buildFromAST</tt>
*/
public static Union getMetaGrammar() {
- return buildFromAST(MetaGrammar.meta, "s", new File[0]);
+ return buildFromAST(MetaGrammar.meta, "s", null);
}
/**
* @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);
}
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 //////////////////////////////////////////////////////////////////////////////
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)), 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 "";
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", includes);
+ if (head.equals("SubGrammar")) return GrammarAST.buildFromAST(t.child(0), "s", resolver);
if (head.equals("NonTerminal"))
return new NonTerminalNode(walkString(t.child(0)),
(Seq[][])walkChildren(t.child(1)), false, null, false);
return new NonTerminalNode(tag, seqs, false, null, false);
}
if (head.equals("#import")) {
- String fileName = (String)stringifyChildren(t.child(0));
- for(File f : includes) {
- File file = new File(f.getAbsolutePath()+File.separatorChar+fileName);
- if (!file.exists()) continue;
+ if (resolver != null) {
+ String fileName = (String)stringifyChildren(t.child(0));
try {
String newPrefix = t.size()<2 ? "" : (walkString(t.child(1))+".");
- FileInputStream fis = new FileInputStream(file);
+ InputStream fis = resolver.getImportStream(fileName);
+ if (fis==null)
+ throw new RuntimeException("unable to find #include file \""+fileName+"\"");
Tree tr = new CharParser(getMetaGrammar()).parse(fis).expand1();
- return (GrammarNode)new GrammarAST(includes, newPrefix).walk(tr);
+ return (GrammarNode)new GrammarAST(resolver, newPrefix).walk(tr);
} catch (Exception e) {
- throw new RuntimeException("while parsing " + file, e);
+ throw new RuntimeException("while parsing " + fileName, e);
}
+ } else {
+ throw new RuntimeException("no resolver given");
}
- throw new RuntimeException("unable to find #include file \""+fileName+"\"");
}
throw new RuntimeException("unknown head: \"" + head + "\" => " + (head.equals("...")));
}
return ret + " ]";
}
public Union build(String rootNonterminal) {
- Context cx = new Context(this);
+ BuildContext cx = new BuildContext(this);
Union u = null;
for(GrammarAST.NonTerminalNode nt : values())
if (nt.name.equals(rootNonterminal))
}
}
- /** a node in the AST which is resolved into an Element */
- private abstract class ElementNode {
- public boolean isLifted() { return false; }
- public boolean drop(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);
- }
-
- private class UnionNode extends ElementNode {
- public Seq[][] sequences;
- public String sep = null;
- public boolean rep;
- 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 drop(Context cx) {
- for(Seq[] seqs : sequences)
- for(Seq seq : seqs)
- if (!seq.drop(cx))
- return false;
- return true;
- }
- public Atom toAtom(Context cx) {
- Atom ret = null;
- for(Seq[] ss : sequences)
- for(Seq s : ss)
- ret = ret==null ? s.toAtom(cx) : (Atom)ret.union(s.toAtom(cx));
- return ret;
- }
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) {
- return buildIntoPreallocatedUnion(cx, cnt, dropall, new Union(null, false)); }
- public Element buildIntoPreallocatedUnion(Context cx, NonTerminalNode cnt, boolean dropall, Union u) {
- Union urep = null;
- if (rep) {
- urep = new Union(null, false);
- 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));
- }
- HashSet<Sequence> bad2 = new HashSet<Sequence>();
- for(int i=0; i<sequences.length; i++) {
- Seq[] group = sequences[i];
- Union u2 = new Union(null, false);
- if (sequences.length==1) u2 = u;
- for(int j=0; j<group.length; j++)
- if (!rep)
- group[j].build(cx, u2, cnt, dropall);
- else {
- Union u3 = new Union(null, false);
- group[j].build(cx, u3, cnt, dropall);
- Sequence s = Sequence.create(cnt.name,
- new Element[] { u3, urep },
- new boolean[] { false, false },
- new boolean[] { false, true});
- u2.add(s);
- }
- if (sequences.length==1) break;
- Sequence seq = Sequence.create(u2);
- for(Sequence s : bad2) seq = seq.andnot(s);
- u.add(seq);
- bad2.add(Sequence.create(u2));
- }
- return u;
- }
- }
-
- private class NonTerminalNode extends UnionNode {
- public boolean alwaysDrop;
- public String name = null;
- public boolean drop(Context cx) { return alwaysDrop; }
+ /** a NonTerminal is always a union at the top level */
+ private class NonTerminalNode {
+ public final boolean alwaysDrop;
+ public final String name;
+ public final ElementNode elementNode;
+ public final UnionNode unionNode;
public NonTerminalNode(String name, Seq[][] sequences, boolean rep, String sep, boolean alwaysDrop) {
- super(sequences, rep, sep==null?null:(prefix + sep));
this.name = prefix + name;
this.alwaysDrop = alwaysDrop;
+ this.unionNode = new UnionNode(sequences, rep, sep==null?null:(prefix + sep));
+ this.elementNode = alwaysDrop ? new DropNode(unionNode) : unionNode;
}
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return cx.get(name); }
+ public boolean isDropped(BuildContext cx) { return alwaysDrop; }
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) { return cx.get(name); }
+ public ElementNode getElementNode() { return elementNode; }
+ public UnionNode getUnionNode() { return unionNode; }
}
+ /** 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;
+ /** positive conjuncts */
+ HashSet<Seq> and = new HashSet<Seq>();
+ /** negative conjuncts */
+ HashSet<Seq> not = new HashSet<Seq>();
public boolean alwaysDrop = false;
- public boolean drop(Context cx) {
+ public boolean isDropped(BuildContext cx) {
if (alwaysDrop) return true;
if (tag!=null) return false;
for(int i=0; i<elements.length; i++)
- if (!elements[i].drop(cx))
+ if (!elements[i].isDropped(cx) || ((elements[i] instanceof LiteralNode) && ((LiteralNode)elements[i]).caret))
return false;
return true;
}
- HashSet<Seq> and = new HashSet<Seq>();
- HashSet<Seq> not = new HashSet<Seq>();
- ElementNode[] elements;
- ElementNode follow;
- String tag = null;
public Seq(ElementNode e) { this(new ElementNode[] { e }); }
public Seq(ElementNode[] elements) { this(elements, true); }
public Seq(ElementNode[] el, boolean check) {
}
}
}
- public Atom toAtom(Context cx) {
+ public Atom toAtom(BuildContext cx) {
if (elements.length != 1)
throw new Error("you attempted to use ->, **, ++, or a similar character-class"+
" operator on a [potentially] multicharacter production");
this.elements = elements;
return this;
}
- public Sequence build(Context cx, Union u, NonTerminalNode cnt, boolean dropall) {
+ public Sequence build(BuildContext cx, Union u, NonTerminalNode cnt, boolean dropall) {
Sequence ret = build0(cx, cnt, dropall);
for(Seq s : and) ret = ret.and(s.build(cx, null, cnt, true));
for(Seq s : not) ret = ret.andnot(s.build(cx, null, cnt, true));
if (u!=null) u.add(ret);
return ret;
}
- public Sequence build0(Context cx, NonTerminalNode cnt, boolean dropall) {
+ public Sequence build0(BuildContext cx, NonTerminalNode cnt, boolean dropall) {
boolean[] drops = new boolean[elements.length];
Element[] els = new Element[elements.length];
- dropall |= drop(cx);
+ dropall |= isDropped(cx);
for(int i=0; i<elements.length; i++) {
if (dropall) drops[i] = true;
- else drops[i] = elements[i].drop(cx);
+ else drops[i] = elements[i].isDropped(cx);
if (elements[i] instanceof LiteralNode && ((LiteralNode)elements[i]).caret) {
if (tag != null) throw new RuntimeException("cannot have multiple tags in a sequence: " + this);
tag = ((LiteralNode)elements[i]).getLiteralTag();
}
}
+ /** a node in the AST which is resolved into an Element */
+ private abstract class ElementNode {
+ /** the field name to be used when synthesizing AST classes; null if none suggested */
+ public String getFieldName() { return null; }
+ public boolean isLifted() { return false; }
+ public boolean isDropped(BuildContext cx) { return false; }
+ public Atom toAtom(BuildContext cx) { throw new Error("can't convert a " + this.getClass().getName() + " to an atom: " + this); }
+ public abstract Element build(BuildContext 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;
+
+ /** 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 String getFieldName() { return null; }
+ public boolean isLifted() { return false; }
+ public boolean isDropped(BuildContext cx) {
+ for(Seq[] seqs : sequences)
+ for(Seq seq : seqs)
+ if (!seq.isDropped(cx))
+ return false;
+ return true;
+ }
+ public Atom toAtom(BuildContext cx) {
+ Atom ret = null;
+ for(Seq[] ss : sequences)
+ for(Seq s : ss)
+ ret = ret==null ? s.toAtom(cx) : (Atom)ret.union(s.toAtom(cx));
+ return ret;
+ }
+
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) {
+ return buildIntoPreallocatedUnion(cx, cnt, dropall, new Union(null, false)); }
+ public Element buildIntoPreallocatedUnion(BuildContext cx, NonTerminalNode cnt, boolean dropall, Union u) {
+ Union urep = null;
+ if (rep) {
+ urep = new Union(null, false);
+ 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));
+ }
+ HashSet<Sequence> bad2 = new HashSet<Sequence>();
+ for(int i=0; i<sequences.length; i++) {
+ Seq[] group = sequences[i];
+ Union u2 = new Union(null, false);
+ if (sequences.length==1) u2 = u;
+ for(int j=0; j<group.length; j++)
+ if (!rep)
+ group[j].build(cx, u2, cnt, dropall);
+ else {
+ Union u3 = new Union(null, false);
+ group[j].build(cx, u3, cnt, dropall);
+ Sequence s = Sequence.create(cnt.name,
+ new Element[] { u3, urep },
+ new boolean[] { false, false },
+ new boolean[] { false, true});
+ u2.add(s);
+ }
+ if (sequences.length==1) break;
+ Sequence seq = Sequence.create(u2);
+ for(Sequence s : bad2) seq = seq.andnot(s);
+ u.add(seq);
+ bad2.add(Sequence.create(u2));
+ }
+ return u;
+ }
+ }
+
+ /** reference to a NonTerminal by name */
private class ReferenceNode extends ElementNode {
public String nonTerminal;
public boolean parenthesized;
this.nonTerminal = nonTerminal.indexOf('.')==-1 ? (prefix + nonTerminal) : nonTerminal;
this.parenthesized = parenthesized;
}
- public NonTerminalNode resolve(Context cx) {
+ public NonTerminalNode resolve(BuildContext cx) {
NonTerminalNode ret = cx.grammar.get(nonTerminal);
if (ret==null) throw new RuntimeException("undefined nonterminal: " + nonTerminal);
return ret;
}
- public Atom toAtom(Context cx) {
- ElementNode ret = cx.grammar.get(nonTerminal);
+ public Atom toAtom(BuildContext cx) {
+ ElementNode ret = cx.grammar.get(nonTerminal).getElementNode();
if (ret == null) throw new RuntimeException("unknown nonterminal \""+nonTerminal+"\"");
return ret.toAtom(cx);
}
- public boolean drop(Context cx) { return resolve(cx).drop(cx); }
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) {
+ public boolean isDropped(BuildContext cx) { return resolve(cx).isDropped(cx); }
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) {
Element ret = cx.get(nonTerminal);
if (ret == null) throw new RuntimeException("unknown nonterminal \""+nonTerminal+"\"");
return ret;
}
+ public String getFieldName() { return StringUtil.uncapitalize(nonTerminal); }
}
+ /** a literal string */
private class LiteralNode extends ElementNode {
private String string;
private final String thePrefix = prefix;
}
public String getLiteralTag() { return caret ? thePrefix+string : null; }
public String toString() { return "\""+string+"\""; }
- public boolean drop(Context cx) { return true; }
- public Atom toAtom(Context cx) {
+ public boolean isDropped(BuildContext cx) { return true; }
+ public Atom toAtom(BuildContext cx) {
if (string.length()!=1) return super.toAtom(cx);
Range.Set set = new Range.Set();
set.add(string.charAt(0), string.charAt(0));
return CharAtom.set(set);
}
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return CharAtom.string(string); }
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) { return CharAtom.string(string); }
}
+ /** an atom (usually a character class) */
private class AtomNode extends ElementNode {
char[][] ranges;
public AtomNode() { this(new char[0][]); }
public AtomNode(char[][] ranges) { this.ranges = ranges; }
public AtomNode(char[] range) { this.ranges = new char[][] { range }; }
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return toAtom(cx); }
- public Atom toAtom(Context cx) {
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) { return toAtom(cx); }
+ public Atom toAtom(BuildContext cx) {
Range.Set set = new Range.Set();
for(char[] r : ranges) set.add(r[0], r[1]);
return CharAtom.set(set);
}
}
+ /** a repetition */
private class RepeatNode extends ElementNode {
public ElementNode e, sep;
public final boolean zero, many, max;
public RepeatNode(ElementNode e, ElementNode sep, boolean zero, boolean many, boolean max) {
this.e = e; this.sep = sep; this.zero = zero; this.many = many; this.max = max;
}
- public Atom toAtom(Context cx) { return sep==null ? e.toAtom(cx) : super.toAtom(cx); }
- public boolean drop(Context cx) { return e.drop(cx); }
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) {
+ public Atom toAtom(BuildContext cx) { return sep==null ? e.toAtom(cx) : super.toAtom(cx); }
+ public boolean isDropped(BuildContext cx) { return e.isDropped(cx); }
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) {
Element ret = build(cx, cnt, dropall, illegalTag);
String must = "must be tagged unless they appear within a dropped expression or their contents are dropped: ";
- if (!dropall && !drop(cx) && !e.drop(cx))
+ if (!dropall && !isDropped(cx) && !e.isDropped(cx))
if (!many) throw new RuntimeException("options (?) " + must + ret);
else if (zero) throw new RuntimeException("zero-or-more repetitions (*) " + must + ret);
else throw new RuntimeException("one-or-more repetitions (+) " + must + ret);
return ret;
}
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall, Object repeatTag) {
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall, Object repeatTag) {
return (!max)
? Repeat.repeat(e.build(cx, null, dropall), zero, many, sep==null ? null : sep.build(cx, null, dropall), repeatTag)
: sep==null
}
}
+ /** helper class for syntactic constructs that wrap another construct */
private abstract class ElementNodeWrapper extends ElementNode {
protected ElementNode _e;
public ElementNodeWrapper(ElementNode e) { this._e = e; }
- public boolean drop(Context cx) { return _e.drop(cx); }
- public Atom toAtom(Context cx) { return _e.toAtom(cx); }
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return _e.build(cx, cnt, dropall); }
+ public boolean isDropped(BuildContext cx) { return _e.isDropped(cx); }
+ public Atom toAtom(BuildContext cx) { return _e.toAtom(cx); }
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) { return _e.build(cx, cnt, dropall); }
+ public String getFieldName() { return _e.getFieldName(); }
}
+ /** a backtick node indicating that, when building the AST, the node's children should be inserted here */
private class BacktickNode extends ElementNodeWrapper {
public BacktickNode(ElementNode e) { super(e); }
public boolean isLifted() { return true; }
}
+ /** negation */
private class TildeNode extends ElementNodeWrapper {
public TildeNode(ElementNode e) { super(e); }
- public Atom toAtom(Context cx) { return (Atom)((Topology<Character>)_e.toAtom(cx).complement()); }
- public Element build(Context cx, NonTerminalNode cnt, boolean dropall) { return toAtom(cx); }
+ public Atom toAtom(BuildContext cx) { return (Atom)((Topology<Character>)_e.toAtom(cx).complement()); }
+ public Element build(BuildContext cx, NonTerminalNode cnt, boolean dropall) { return toAtom(cx); }
}
private class DropNode extends ElementNodeWrapper {
public DropNode(ElementNode e) { super(e); }
- public boolean drop(Context cx) { return true; }
+ public boolean isDropped(BuildContext cx) { return true; }
+ }
+
+ /** 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; }
+ public String getFieldName() { return label; }
}
//////////////////////////////////////////////////////////////////////////////
- public class Context {
+ public class BuildContext {
public HashMap<String,Union> map = new HashMap<String,Union>();
public GrammarNode grammar;
- public Context(Tree t) { }
- public Context(GrammarNode g) { this.grammar = g; }
+ public BuildContext(Tree t) { }
+ public BuildContext(GrammarNode g) { this.grammar = g; }
public Union build() {
Union ret = null;
for(NonTerminalNode nt : grammar.values()) {
} else {
ret = new Union(name, false);
map.put(name, ret);
- nt.buildIntoPreallocatedUnion(this, nt, nt.drop(this), ret);
+ nt.getUnionNode().buildIntoPreallocatedUnion(this, nt, nt.isDropped(this), ret);
}
return ret;
}