-// $Id: CQLParser.java,v 1.23 2003-09-04 21:56:46 mike Exp $
+// $Id: CQLParser.java,v 1.31 2007-06-29 10:25:38 mike Exp $
package org.z3950.zing.cql;
import java.io.IOException;
/**
* Compiles CQL strings into parse trees of CQLNode subtypes.
*
- * @version $Id: CQLParser.java,v 1.23 2003-09-04 21:56:46 mike Exp $
+ * @version $Id: CQLParser.java,v 1.31 2007-06-29 10:25:38 mike Exp $
* @see <A href="http://zing.z3950.org/cql/index.html"
* >http://zing.z3950.org/cql/index.html</A>
*/
return root;
}
- private CQLNode parseQuery(String qualifier, CQLRelation relation)
+ private CQLNode parseQuery(String index, CQLRelation relation)
throws CQLParseException, IOException {
debug("in parseQuery()");
- CQLNode term = parseTerm(qualifier, relation);
- while (lexer.ttype != lexer.TT_EOF &&
- lexer.ttype != ')') {
- if (lexer.ttype == lexer.TT_AND) {
- match(lexer.TT_AND);
- CQLNode term2 = parseTerm(qualifier, relation);
- term = new CQLAndNode(term, term2);
- } else if (lexer.ttype == lexer.TT_OR) {
- match(lexer.TT_OR);
- CQLNode term2 = parseTerm(qualifier, relation);
- term = new CQLOrNode(term, term2);
- } else if (lexer.ttype == lexer.TT_NOT) {
- match(lexer.TT_NOT);
- CQLNode term2 = parseTerm(qualifier, relation);
- term = new CQLNotNode(term, term2);
+ CQLNode term = parseTerm(index, relation);
+ while (lexer.ttype != lexer.TT_EOF && lexer.ttype != ')') {
+ if (lexer.ttype == lexer.TT_AND ||
+ lexer.ttype == lexer.TT_OR ||
+ lexer.ttype == lexer.TT_NOT) {
+ int type = lexer.ttype;
+ String val = lexer.sval;
+ match(type);
+ ModifierSet ms = gatherModifiers(val);
+ CQLNode term2 = parseTerm(index, relation);
+ term = ((type == lexer.TT_AND) ? new CQLAndNode(term, term2, ms) :
+ (type == lexer.TT_OR) ? new CQLOrNode (term, term2, ms) :
+ new CQLNotNode(term, term2, ms));
} else if (lexer.ttype == lexer.TT_PROX) {
- match(lexer.TT_PROX);
+ match(lexer.ttype);
CQLProxNode proxnode = new CQLProxNode(term);
gatherProxParameters(proxnode);
- CQLNode term2 = parseTerm(qualifier, relation);
+ CQLNode term2 = parseTerm(index, relation);
proxnode.addSecondSubterm(term2);
term = (CQLNode) proxnode;
} else {
return term;
}
- private CQLNode parseTerm(String qualifier, CQLRelation relation)
+ private ModifierSet gatherModifiers(String base)
+ throws CQLParseException, IOException {
+ debug("in gatherModifiers()");
+
+ ModifierSet ms = new ModifierSet(base);
+ while (lexer.ttype == '/') {
+ match('/');
+ if (lexer.ttype != lexer.TT_WORD)
+ throw new CQLParseException("expected modifier, "
+ + "got " + lexer.render());
+ String type = lexer.sval.toLowerCase();
+ match(lexer.ttype);
+ if (!isRelation()) {
+ // It's a simple modifier consisting of type only
+ ms.addModifier(type);
+ } else {
+ // It's a complex modifier of the form type=value
+ String comparision = lexer.render(lexer.ttype, false);
+ match(lexer.ttype);
+
+ // Yuck
+ String value = lexer.ttype == lexer.TT_WORD ? lexer.sval :
+ (double) lexer.nval == (int) lexer.nval ?
+ new Integer((int) lexer.nval).toString() :
+ new Double((double) lexer.nval).toString();
+
+ matchSymbol("modifier value");
+ ms.addModifier(type, comparision, value);
+ }
+ }
+
+ return ms;
+ }
+
+ private CQLNode parseTerm(String index, CQLRelation relation)
throws CQLParseException, IOException {
debug("in parseTerm()");
if (lexer.ttype == '(') {
debug("parenthesised term");
match('(');
- CQLNode expr = parseQuery(qualifier, relation);
+ CQLNode expr = parseQuery(index, relation);
match(')');
return expr;
} else if (lexer.ttype == '>') {
match('>');
- return parsePrefix(qualifier, relation);
+ return parsePrefix(index, relation);
}
debug("non-parenthesised term");
- word = matchSymbol("qualifier or term");
- if (!isBaseRelation())
+ word = matchSymbol("index or term");
+ if (!isRelation() && lexer.ttype != lexer.TT_WORD)
break;
- qualifier = word;
- relation = new CQLRelation(lexer.ttype == lexer.TT_WORD ?
- lexer.sval :
- lexer.render(lexer.ttype, false));
+ index = word;
+ String relstr = (lexer.ttype == lexer.TT_WORD ?
+ lexer.sval : lexer.render(lexer.ttype, false));
+ relation = new CQLRelation(relstr);
match(lexer.ttype);
-
- while (lexer.ttype == '/') {
- match('/');
- if (lexer.ttype != lexer.TT_RELEVANT &&
- lexer.ttype != lexer.TT_FUZZY &&
- lexer.ttype != lexer.TT_STEM &&
- lexer.ttype != lexer.TT_PHONETIC &&
- lexer.ttype != lexer.TT_WORD)
- throw new CQLParseException("expected relation modifier, "
- + "got " + lexer.render());
- if (lexer.ttype == lexer.TT_WORD &&
- lexer.sval.indexOf('.') == -1)
- throw new CQLParseException("unknown first-class " +
- "relation modifier: " +
- lexer.sval);
-
- relation.addModifier(lexer.sval.toLowerCase());
- match(lexer.ttype);
- }
-
- debug("qualifier='" + qualifier + ", " +
+ ModifierSet ms = gatherModifiers(relstr);
+ relation.setModifiers(ms);
+ debug("index='" + index + ", " +
"relation='" + relation.toCQL() + "'");
}
- CQLTermNode node = new CQLTermNode(qualifier, relation, word);
+ CQLTermNode node = new CQLTermNode(index, relation, word);
debug("made term node " + node.toCQL());
return node;
}
- private CQLNode parsePrefix(String qualifier, CQLRelation relation)
+ private CQLNode parsePrefix(String index, CQLRelation relation)
throws CQLParseException, IOException {
debug("prefix mapping");
name = identifier;
identifier = matchSymbol("prefix-identifer");
}
- CQLNode term = parseQuery(qualifier, relation);
+ CQLNode term = parseQuery(index, relation);
return new CQLPrefixNode(name, identifier, term);
}
private void gatherProxRelation(CQLProxNode node)
throws CQLParseException, IOException {
- if (!isProxRelation())
+ if (!isRelation())
throw new CQLParseException("expected proximity relation, got " +
lexer.render());
- node.addModifier("relation", lexer.render(lexer.ttype, false));
+ node.addModifier("relation", null, lexer.render(lexer.ttype, false));
match(lexer.ttype);
debug("gPR matched " + lexer.render(lexer.ttype, false));
}
if (lexer.ttype != lexer.TT_NUMBER)
throw new CQLParseException("expected proximity distance, got " +
lexer.render());
- node.addModifier("distance", lexer.render(lexer.ttype, false));
+ node.addModifier("distance", null, lexer.render(lexer.ttype, false));
match(lexer.ttype);
debug("gPD matched " + lexer.render(lexer.ttype, false));
}
lexer.ttype != lexer.TT_ELEMENT)
throw new CQLParseException("expected proximity unit, got " +
lexer.render());
- node.addModifier("unit", lexer.render());
+ node.addModifier("unit", null, lexer.render());
match(lexer.ttype);
}
lexer.ttype != lexer.TT_UNORDERED)
throw new CQLParseException("expected proximity ordering, got " +
lexer.render());
- node.addModifier("ordering", lexer.render());
+ node.addModifier("ordering", null, lexer.render());
match(lexer.ttype);
}
- private boolean isBaseRelation()
- throws CQLParseException {
- debug("isBaseRelation: checking ttype=" + lexer.ttype +
- " (" + lexer.render() + ")");
-
- if (lexer.ttype == lexer.TT_WORD &&
- lexer.sval.indexOf('.') == -1)
- throw new CQLParseException("unknown first-class relation: " +
- lexer.sval);
-
- return (isProxRelation() ||
- lexer.ttype == lexer.TT_ANY ||
- lexer.ttype == lexer.TT_ALL ||
- lexer.ttype == lexer.TT_EXACT ||
- lexer.ttype == lexer.TT_SCR ||
- lexer.ttype == lexer.TT_WORD);
- }
-
// Checks for a relation that may be used inside a prox operator
- private boolean isProxRelation() {
- debug("isProxRelation: checking ttype=" + lexer.ttype +
+ private boolean isRelation() {
+ debug("isRelation: checking ttype=" + lexer.ttype +
" (" + lexer.render() + ")");
return (lexer.ttype == '<' ||
lexer.ttype == '>' ||
lexer.ttype == '"' ||
// The following is a complete list of keywords. Because
// they're listed here, they can be used unquoted as
- // qualifiers, terms, prefix names and prefix identifiers.
+ // indexes, terms, prefix names and prefix identifiers.
// ### Instead, we should ask the lexer whether what we
// have is a keyword, and let the knowledge reside there.
lexer.ttype == lexer.TT_AND ||
lexer.ttype == lexer.TT_OR ||
lexer.ttype == lexer.TT_NOT ||
lexer.ttype == lexer.TT_PROX ||
- lexer.ttype == lexer.TT_ANY ||
- lexer.ttype == lexer.TT_ALL ||
- lexer.ttype == lexer.TT_EXACT ||
lexer.ttype == lexer.TT_pWORD ||
lexer.ttype == lexer.TT_SENTENCE ||
lexer.ttype == lexer.TT_PARAGRAPH ||
lexer.ttype == lexer.TT_ELEMENT ||
lexer.ttype == lexer.TT_ORDERED ||
- lexer.ttype == lexer.TT_UNORDERED ||
- lexer.ttype == lexer.TT_RELEVANT ||
- lexer.ttype == lexer.TT_FUZZY ||
- lexer.ttype == lexer.TT_STEM ||
- lexer.ttype == lexer.TT_SCR ||
- lexer.ttype == lexer.TT_PHONETIC) {
+ lexer.ttype == lexer.TT_UNORDERED) {
String symbol = (lexer.ttype == lexer.TT_NUMBER) ?
lexer.render() : lexer.sval;
match(lexer.ttype);
char mode = 'x'; // x=XCQL, c=CQL, p=PQF
String pfile = null;
- Vector argv = new Vector();
+ Vector<String> argv = new Vector<String>();
for (int i = 0; i < args.length; i++) {
argv.add(args[i]);
}
+ if (argv.size() > 0 && argv.get(0).equals("-d")) {
+ DEBUG = true;
+ argv.remove(0);
+ }
+
if (argv.size() > 0 && argv.get(0).equals("-c")) {
mode = 'c';
argv.remove(0);
}
if (argv.size() > 1) {
- System.err.println(
- "Usage: CQLParser [-c] [-p <pqf-properties> [<CQL-query>]");
+ System.err.println("Usage: CQLParser [-d] [-c] [-p <pqf-properties> [<CQL-query>]");
System.err.println("If unspecified, query is read from stdin");
System.exit(1);
}
} catch (IOException ex) {
System.err.println("Can't render query: " + ex.getMessage());
System.exit(5);
- } catch (UnknownQualifierException ex) {
- System.err.println("Unknown qualifier: " + ex.getMessage());
+ } catch (UnknownIndexException ex) {
+ System.err.println("Unknown index: " + ex.getMessage());
System.exit(6);
} catch (UnknownRelationException ex) {
System.err.println("Unknown relation: " + ex.getMessage());
projects / cql-java-moved-to-github.git / blobdiff