-// $Id: CQLParser.java,v 1.28 2007-06-27 22:44:40 mike Exp $
+// $Id: CQLParser.java,v 1.36 2007-06-29 15:39:09 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 1.28 2007-06-27 22:44:40 mike Exp $
+ * @version $Id: CQLParser.java,v 1.36 2007-06-29 15:39:09 mike Exp $
* @see <A href="http://zing.z3950.org/cql/index.html"
* >http://zing.z3950.org/cql/index.html</A>
*/
public class CQLParser {
private CQLLexer lexer;
+ private int compat; // When false, implement CQL 1.2
+ public static int V1POINT1 = 12368;
+ public static int V1POINT2 = 12369;
+ public static int V1POINT1SORT = 12370;
+
static private boolean DEBUG = false;
static private boolean LEXDEBUG = false;
+ /**
+ * The new parser implements a dialect of CQL specified by the
+ * <tt>compat</tt> argument:
+ * <ul>
+ * <li>V1POINT1 - CQL version 1.1
+ * </li>
+ * <li>V1POINT2 - CQL version 1.2
+ * </li>
+ * <li>V1POINT1SORT - CQL version 1.1 but including
+ * <tt>sortby</tt> as specified for CQL 1.2.
+ * </li>
+ * </ul>
+ */
+ CQLParser(int compat) {
+ this.compat = compat;
+ }
+
+ /**
+ * The new parser implements CQL 1.2
+ */
+ CQLParser() {
+ this.compat = V1POINT2;
+ }
+
private static void debug(String str) {
if (DEBUG)
System.err.println("PARSEDEBUG: " + str);
lexer.nextToken();
debug("about to parseQuery()");
- CQLNode root = parseQuery("srw.serverChoice", new CQLRelation("scr"));
+ CQLNode root = parseQuery("cql.serverChoice",
+ new CQLRelation(compat == V1POINT2 ? "=" : "scr"));
if (lexer.ttype != lexer.TT_EOF)
throw new CQLParseException("junk after end: " + lexer.render());
debug("in parseQuery()");
CQLNode term = parseTerm(index, relation);
- while (lexer.ttype != lexer.TT_EOF &&
- lexer.ttype != ')') {
- if (lexer.ttype == lexer.TT_AND) {
- match(lexer.TT_AND);
- CQLNode term2 = parseTerm(index, relation);
- term = new CQLAndNode(term, term2);
- } else if (lexer.ttype == lexer.TT_OR) {
- match(lexer.TT_OR);
- CQLNode term2 = parseTerm(index, relation);
- term = new CQLOrNode(term, term2);
- } else if (lexer.ttype == lexer.TT_NOT) {
- match(lexer.TT_NOT);
+ while (lexer.ttype != lexer.TT_EOF && lexer.ttype != ')') {
+ if (lexer.ttype == lexer.TT_AND ||
+ lexer.ttype == lexer.TT_OR ||
+ lexer.ttype == lexer.TT_NOT ||
+ lexer.ttype == lexer.TT_PROX) {
+ int type = lexer.ttype;
+ String val = lexer.sval;
+ match(type);
+ ModifierSet ms = gatherModifiers(val);
CQLNode term2 = parseTerm(index, relation);
- term = new CQLNotNode(term, term2);
- } else if (lexer.ttype == lexer.TT_PROX) {
- match(lexer.TT_PROX);
- CQLProxNode proxnode = new CQLProxNode(term);
- gatherProxParameters(proxnode);
- CQLNode term2 = parseTerm(index, relation);
- proxnode.addSecondSubterm(term2);
- term = (CQLNode) proxnode;
+ term = ((type == lexer.TT_AND) ? new CQLAndNode(term, term2, ms) :
+ (type == lexer.TT_OR) ? new CQLOrNode (term, term2, ms) :
+ (type == lexer.TT_NOT) ? new CQLNotNode(term, term2, ms) :
+ new CQLProxNode(term, term2, ms));
} else {
throw new CQLParseException("expected boolean, got " +
lexer.render());
return term;
}
+ 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);
+ String value = matchSymbol("modifier value");
+ ms.addModifier(type, comparision, value);
+ }
+ }
+
+ return ms;
+ }
+
private CQLNode parseTerm(String index, CQLRelation relation)
throws CQLParseException, IOException {
debug("in parseTerm()");
debug("non-parenthesised term");
word = matchSymbol("index or term");
- if (!isBaseRelation())
+ if (!isRelation() && lexer.ttype != lexer.TT_WORD)
break;
index = word;
- relation = new CQLRelation(lexer.ttype == lexer.TT_WORD ?
- lexer.sval :
- lexer.render(lexer.ttype, false));
+ 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);
- }
-
+ ModifierSet ms = gatherModifiers(relstr);
+ relation.setModifiers(ms);
debug("index='" + index + ", " +
"relation='" + relation.toCQL() + "'");
}
return new CQLPrefixNode(name, identifier, term);
}
- private void gatherProxParameters(CQLProxNode node)
- throws CQLParseException, IOException {
- for (int i = 0; i < 4; i++) {
- if (lexer.ttype != '/')
- return; // end of proximity parameters
-
- match('/');
- if (lexer.ttype != '/') {
- // not an omitted default
- switch (i) {
- // Order should be: relation/distance/unit/ordering
- // For now, use MA's: unit/relation/distance/ordering
- case 0: gatherProxRelation(node); break;
- case 1: gatherProxDistance(node); break;
- case 2: gatherProxUnit(node); break;
- case 3: gatherProxOrdering(node); break;
- }
- }
- }
- }
-
- private void gatherProxRelation(CQLProxNode node)
- throws CQLParseException, IOException {
- if (!isProxRelation())
- throw new CQLParseException("expected proximity relation, got " +
- lexer.render());
- node.addModifier("relation", null, lexer.render(lexer.ttype, false));
- match(lexer.ttype);
- debug("gPR matched " + lexer.render(lexer.ttype, false));
- }
-
- private void gatherProxDistance(CQLProxNode node)
- throws CQLParseException, IOException {
- if (lexer.ttype != lexer.TT_NUMBER)
- throw new CQLParseException("expected proximity distance, got " +
- lexer.render());
- node.addModifier("distance", null, lexer.render(lexer.ttype, false));
- match(lexer.ttype);
- debug("gPD matched " + lexer.render(lexer.ttype, false));
- }
-
- private void gatherProxUnit(CQLProxNode node)
- throws CQLParseException, IOException {
- if (lexer.ttype != lexer.TT_pWORD &&
- lexer.ttype != lexer.TT_SENTENCE &&
- lexer.ttype != lexer.TT_PARAGRAPH &&
- lexer.ttype != lexer.TT_ELEMENT)
- throw new CQLParseException("expected proximity unit, got " +
- lexer.render());
- node.addModifier("unit", null, lexer.render());
- match(lexer.ttype);
- }
-
- private void gatherProxOrdering(CQLProxNode node)
- throws CQLParseException, IOException {
- if (lexer.ttype != lexer.TT_ORDERED &&
- lexer.ttype != lexer.TT_UNORDERED)
- throw new CQLParseException("expected proximity ordering, got " +
- 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 +
+ // Checks for a relation
+ private boolean isRelation() {
+ debug("isRelation: checking ttype=" + lexer.ttype +
" (" + lexer.render() + ")");
return (lexer.ttype == '<' ||
lexer.ttype == '>' ||
lexer.ttype == '=' ||
lexer.ttype == lexer.TT_LE ||
lexer.ttype == lexer.TT_GE ||
- lexer.ttype == lexer.TT_NE);
+ lexer.ttype == lexer.TT_NE ||
+ lexer.ttype == lexer.TT_EQEQ);
}
private void match(int token)
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_PROX) {
String symbol = (lexer.ttype == lexer.TT_NUMBER) ?
lexer.render() : lexer.sval;
match(lexer.ttype);
* </triple>
* </PRE>
* <P>
+ * @param -1
+ * CQL version 1.1 (default version 1.2)
+ * @param -d
+ * Debug mode: extra output written to stderr.
* @param -c
* Causes the output to be written in CQL rather than XCQL - that
* is, a query equivalent to that which was input, is output. In
argv.add(args[i]);
}
+ int compat = V1POINT2;
+ if (argv.size() > 0 && argv.get(0).equals("-1")) {
+ compat = V1POINT1;
+ argv.remove(0);
+ }
+
if (argv.size() > 0 && argv.get(0).equals("-d")) {
DEBUG = true;
argv.remove(0);
}
if (argv.size() > 1) {
- System.err.println("Usage: CQLParser [-d] [-c] [-p <pqf-properties> [<CQL-query>]");
+ System.err.println("Usage: CQLParser [-1] [-d] [-c] " +
+ "[-p <pqf-properties> [<CQL-query>]");
System.err.println("If unspecified, query is read from stdin");
System.exit(1);
}
cql = new String(bytes);
}
- CQLParser parser = new CQLParser();
+ CQLParser parser = new CQLParser(compat);
CQLNode root = null;
try {
root = parser.parse(cql);