-// ### not _quite_ finished yet ...
+// $Id: CQLParser.java,v 1.10 2002-10-30 09:19:26 mike Exp $
+
+package org.z3950.zing.cql;
+import java.io.IOException;
+
+
+/**
+ * Compiles a CQL string into a parse tree.
+ * ###
+ *
+ * @version $Id: CQLParser.java,v 1.10 2002-10-30 09:19:26 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;
+ static private boolean DEBUG = false;
+ static private boolean LEXDEBUG = false;
+
+ private static void debug(String str) {
+ if (DEBUG)
+ System.err.println("PARSEDEBUG: " + str);
+ }
+
+ public CQLNode parse(String cql)
+ throws CQLParseException, IOException {
+ lexer = new CQLLexer(cql, LEXDEBUG);
+
+ lexer.nextToken();
+ debug("about to parse_query()");
+ CQLNode root = parse_query("srw.serverChoice", new CQLRelation("="));
+ if (lexer.ttype != lexer.TT_EOF)
+ throw new CQLParseException("junk after end: " + lexer.render());
+
+ return root;
+ }
+
+ private CQLNode parse_query(String qualifier, CQLRelation relation)
+ throws CQLParseException, IOException {
+ debug("in parse_query()");
+
+ CQLNode term = parse_term(qualifier, relation);
+ while (lexer.ttype != lexer.TT_EOF &&
+ lexer.ttype != ')') {
+ if (lexer.ttype == lexer.TT_AND) {
+ match(lexer.TT_AND);
+ CQLNode term2 = parse_term(qualifier, relation);
+ term = new CQLAndNode(term, term2);
+ } else if (lexer.ttype == lexer.TT_OR) {
+ match(lexer.TT_OR);
+ CQLNode term2 = parse_term(qualifier, relation);
+ term = new CQLOrNode(term, term2);
+ } else if (lexer.ttype == lexer.TT_NOT) {
+ match(lexer.TT_NOT);
+ CQLNode term2 = parse_term(qualifier, relation);
+ term = new CQLNotNode(term, term2);
+ } else if (lexer.ttype == lexer.TT_PROX) {
+ // ### Handle "prox"
+ } else {
+ throw new CQLParseException("expected boolean, got " +
+ lexer.render());
+ }
+ }
+
+ debug("no more ops");
+ return term;
+ }
+
+ private CQLNode parse_term(String qualifier, CQLRelation relation)
+ throws CQLParseException, IOException {
+ debug("in parse_term()");
+
+ String word;
+ while (true) {
+ if (lexer.ttype == '(') {
+ debug("parenthesised term");
+ match('(');
+ CQLNode expr = parse_query(qualifier, relation);
+ match(')');
+ return expr;
+ } else if (lexer.ttype != lexer.TT_WORD && lexer.ttype != '"') {
+ throw new CQLParseException("expected qualifier or term, " +
+ "got " + lexer.render());
+ }
+
+ debug("non-parenthesised term");
+ word = lexer.sval;
+ match(lexer.ttype);
+ if (!isBaseRelation())
+ break;
+
+ qualifier = word;
+ relation = new CQLRelation(lexer.render(lexer.ttype, false));
+ match(lexer.ttype);
+
+ while (lexer.ttype == '/') {
+ match('/');
+ // ### could insist on known modifiers only
+ if (lexer.ttype != lexer.TT_WORD)
+ throw new CQLParseException("expected relation modifier, "
+ + "got " + lexer.render());
+ relation.addModifier(lexer.sval);
+ match(lexer.TT_WORD);
+ }
+
+ debug("qualifier='" + qualifier + ", " +
+ "relation='" + relation.toCQL() + "'");
+ }
+
+ CQLTermNode node = new CQLTermNode(qualifier, relation, word);
+ debug("made term node " + node.toCQL());
+ return node;
+ }
+
+ boolean isBaseRelation() {
+ debug("isBaseRelation: 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_ANY ||
+ lexer.ttype == lexer.TT_ALL ||
+ lexer.ttype == lexer.TT_EXACT);
+ }
+
+ private void match(int token)
+ throws CQLParseException, IOException {
+ debug("in match(" + lexer.render(token, true) + ")");
+ if (lexer.ttype != token)
+ throw new CQLParseException("expected " +
+ lexer.render(token, true) +
+ ", " + "got " + lexer.render());
+ int tmp = lexer.nextToken();
+ debug("match() got token=" + lexer.ttype + ", " +
+ "nval=" + lexer.nval + ", sval='" + lexer.sval + "'" +
+ " (tmp=" + tmp + ")");
+ }
+
+
+ // Test harness.
+ //
+ // e.g. echo '(au=Kerninghan or au=Ritchie) and ti=Unix' |
+ // java org.z3950.zing.cql.CQLParser
+ // yields:
+ // <triple>
+ // <boolean>and</boolean>
+ // <triple>
+ // <boolean>or</boolean>
+ // <searchClause>
+ // <index>au<index>
+ // <relation>=<relation>
+ // <term>Kerninghan<term>
+ // </searchClause>
+ // <searchClause>
+ // <index>au<index>
+ // <relation>=<relation>
+ // <term>Ritchie<term>
+ // </searchClause>
+ // </triple>
+ // <searchClause>
+ // <index>ti<index>
+ // <relation>=<relation>
+ // <term>Unix<term>
+ // </searchClause>
+ // </triple>
+ //
+ public static void main (String[] args) {
+ if (args.length != 0) {
+ System.err.println("Usage: " + args[0]);
+ System.exit(1);
+ }
+
+ byte[] bytes = new byte[10000];
+ try {
+ // Read in the whole of standard input in one go
+ int nbytes = System.in.read(bytes);
+ } catch (java.io.IOException ex) {
+ System.err.println("Can't read query: " + ex.getMessage());
+ System.exit(2);
+ }
+ String cql = new String(bytes);
+ CQLParser parser = new CQLParser();
+ CQLNode root;
+ try {
+ root = parser.parse(cql);
+ debug("root='" + root + "'");
+ System.out.println(root.toXCQL(0));
+ } catch (CQLParseException ex) {
+ System.err.println("Syntax error: " + ex.getMessage());
+ System.exit(3);
+ } catch (java.io.IOException ex) {
+ System.err.println("Can't compile query: " + ex.getMessage());
+ System.exit(4);
+ }
+ }
+}