-// $Header: /home/cvsroot/cql-java/src/org/z3950/zing/cql/CQLParser.java,v 1.2 2002-10-24 16:05:15 mike Exp $
+// $Id: CQLParser.java,v 1.9 2002-10-29 10:15:58 mike Exp $
package org.z3950.zing.cql;
import java.util.Properties;
import java.io.InputStream;
-import java.io.FileNotFoundException;
import java.io.IOException;
import java.io.StringReader;
import java.io.StreamTokenizer;
* Compiles a CQL string into a parse tree ...
* ###
*
- * @version $Id: CQLParser.java,v 1.2 2002-10-24 16:05:15 mike Exp $
+ * @version $Id: CQLParser.java,v 1.9 2002-10-29 10:15:58 mike Exp $
* @see <A href="http://zing.z3950.org/cql/index.html"
* >http://zing.z3950.org/cql/index.html</A>
*/
-class CQLCompiler {
- private String cql;
- private String qualset;
- private Properties qualsetProperties;
- private StreamTokenizer st;
+public class CQLParser {
+ private CQLLexer lexer;
+ static private boolean PARSEDEBUG = false;
+ static private boolean LEXDEBUG = true;
private class CQLParseException extends Exception {
CQLParseException(String s) { super(s); }
}
- public CQLCompiler(String cql, String qualset) {
- this.cql = cql;
- this.qualset = qualset;
+ static void debug(String str) {
+ if (PARSEDEBUG)
+ System.err.println("PARSEDEBUG: " + str);
}
- public String convertToPQN()
- throws FileNotFoundException, IOException {
-
- if (qualsetProperties == null) {
- // ### Could think about caching named qualifier sets
- // across compilations (i.e. shared, in a static
- // Hashtable, between multiple CQLCompiler
- // instances.) Probably not worth it.
- InputStream is = this.getClass().getResourceAsStream(qualset);
- if (is == null)
- throw new FileNotFoundException("getResourceAsStream(" +
- qualset + ")");
- qualsetProperties = new Properties();
- qualsetProperties.load(is);
- }
-
- st = new StreamTokenizer(new StringReader(cql));
- st.wordChars('/', '/');
- st.wordChars('0', '9'); // ### but 1 is still recognised as TT_NUM
- st.wordChars('.', '.');
- st.wordChars('-', '-');
- st.ordinaryChar('=');
- st.ordinaryChar(',');
- st.ordinaryChar('(');
- st.ordinaryChar(')');
-
-// int token;
-// while ((token = st.nextToken()) != st.TT_EOF) {
-// System.out.println("token=" + token + ", " +
-// "nval=" + st.nval + ", " +
-// "sval=" + st.sval);
-// }
-
- st.nextToken();
- String ret;
- try {
- ret = parse_expression();
- } catch (CQLParseException ex) {
- System.err.println("### Oops: " + ex);
- return null;
- }
+ public CQLNode parse(String cql)
+ throws CQLParseException, IOException {
+ lexer = new CQLLexer(cql, LEXDEBUG);
- if (st.ttype != st.TT_EOF) {
- System.err.println("### Extra bits: " + render(st));
- return null;
- }
+ lexer.nextToken();
+ debug("about to parse_query()");
+ CQLNode root = parse_query("srw.serverChoice", "=");
+ if (lexer.ttype != lexer.TT_EOF)
+ throw new CQLParseException("junk after end: " + lexer.render());
- // Interpret attributes as BIB-1 unless otherwise specified
- return "@attrset bib-1 " + ret;
+ return root;
}
- private String parse_expression()
+ private CQLNode parse_query(String qualifier, String relation)
throws CQLParseException, IOException {
- String term = parse_term();
+ debug("in parse_query()");
- while (st.ttype == st.TT_WORD) {
- String op = st.sval.toLowerCase();
- if (!st.sval.equals("and") &&
- !st.sval.equals("or") &&
- !st.sval.equals("not"))
- break;
- match(st.TT_WORD);
- String term2 = parse_term();
- term = "@" + op + " " + term + " " + term2;
+ CQLNode term = parse_term(qualifier, relation);
+ while (lexer.ttype == lexer.TT_WORD) {
+ String op = lexer.sval.toLowerCase();
+ debug("checking op '" + op + "'");
+ if (lexer.sval.equals("and")) {
+ match(lexer.TT_WORD);
+ CQLNode term2 = parse_term(qualifier, relation);
+ term = new CQLAndNode(term, term2);
+ } else if (lexer.sval.equals("or")) {
+ match(lexer.TT_WORD);
+ CQLNode term2 = parse_term(qualifier, relation);
+ term = new CQLOrNode(term, term2);
+ } else if (lexer.sval.equals("not")) {
+ match(lexer.TT_WORD);
+ CQLNode term2 = parse_term(qualifier, relation);
+ term = new CQLNotNode(term, term2);
+ } else if (lexer.sval.equals("prox")) {
+ // ### Handle "prox"
+ } else {
+ throw new CQLParseException("unrecognised boolean: '" +
+ lexer.sval + "'");
+ }
}
+ debug("no more ops");
return term;
}
- private String parse_term()
+ private CQLNode parse_term(String qualifier, String relation)
throws CQLParseException, IOException {
- if (st.ttype == '(') {
- match('(');
- String expr = parse_expression();
- match(')');
- return expr;
- }
+ debug("in parse_term()");
- String word = null;
- String attrs = "";
+ 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());
+ }
- // ### We treat ',' and '=' equivalently here, which isn't quite right.
- while (st.ttype == st.TT_WORD) {
- word = st.sval;
- match(st.TT_WORD);
- if (st.ttype != '=' && st.ttype != ',') {
- // end of qualifer list
+ debug("non-parenthesised term");
+ word = lexer.sval;
+ match(lexer.ttype);
+ if (!isRelation())
break;
- }
- String attr = qualsetProperties.getProperty(word);
- if (attr == null) {
- throw new CQLParseException("unrecognised qualifier: " + word);
- }
- attrs = attrs + attr + " ";
- match(st.ttype);
- word = null; // mark as not-yet-read
+ qualifier = word;
+ relation = lexer.render(false);
+ debug("got relation '" + relation + "'");
+ match(lexer.ttype);
+ debug("qualifier='" + qualifier + ", relation='" + relation + "'");
}
- if (word == null) {
- // got to the end of a "foo,bar=" sequence
- word = st.sval;
- if (st.ttype != '\'' || st.ttype != '"') {
- word = "\"" + word + "\"";
- match(st.ttype);
- } else {
- match(st.TT_WORD);
- }
- }
+ CQLTermNode node = new CQLTermNode(qualifier, relation, word);
+ debug("made term node " + node);
+ return node;
+ }
- return attrs + word;
+ boolean isRelation() {
+ // ### Handle any, all and exact
+ return (lexer.ttype == '<' ||
+ lexer.ttype == '>' ||
+ lexer.ttype == '=' ||
+ lexer.ttype == lexer.TT_LE ||
+ lexer.ttype == lexer.TT_GE ||
+ lexer.ttype == lexer.TT_NE);
}
private void match(int token)
throws CQLParseException, IOException {
- if (st.ttype != token)
- throw new CQLParseException("expected " + render(st, token, null) +
- ", " + "got " + render(st));
- st.nextToken();
+ debug("in match(" + lexer.render(token, null, true) + ")");
+ if (lexer.ttype != token)
+ throw new CQLParseException("expected " +
+ lexer.render(token, null, true) +
+ ", " + "got " + lexer.render());
+ lexer.nextToken();
}
- // ### This utility should surely be a method of the StreamTokenizer class
- private static String render(StreamTokenizer st) {
- return render(st, st.ttype, null);
- }
-
- private static String render(StreamTokenizer st, int token, String str) {
- String ret;
-
- switch (token) {
- case st.TT_EOF: return "EOF";
- case st.TT_EOL: return "EOL";
- case st.TT_NUMBER: return "number";
- case st.TT_WORD: ret = "word"; break;
- case '"': case '\'': ret = "string"; break;
- default: return "'" + String.valueOf((char) token) + "'";
- }
-
- if (str != null)
- ret += "(\"" + str + "\")";
- return ret;
- }
- // ### Not really the right place for this test harness.
+ // Test harness.
//
- // e.g. java uk.org.miketaylor.zoom.CQLCompiler
- // '(au=Kerninghan or au=Ritchie) and ti=Unix' qualset.properties
+ // e.g. echo '(au=Kerninghan or au=Ritchie) and ti=Unix' |
+ // java org.z3950.zing.cql.CQLParser
// yields:
- // @and
- // @or
- // @attr 1=1 @attr 4=1 Kerninghan
- // @attr 1=1 @attr 4=1 Ritchie
- // @attr 1=4 @attr 4=1 Unix
+ // <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 != 2) {
- System.err.println("Usage: CQLQuery <cql> <qualset>");
+ if (args.length != 0) {
+ System.err.println("Usage: " + args[0]);
System.exit(1);
}
- CQLCompiler cc = new CQLCompiler(args[0], args[1]);
+ byte[] bytes = new byte[10000];
try {
- String pqn = cc.convertToPQN();
- System.out.println(pqn);
- } catch (FileNotFoundException ex) {
- System.err.println("Can't find qualifier set: " + ex);
- System.exit(2);
- } catch (IOException ex) {
- System.err.println("Can't read qualifier set: " + ex);
+ // 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);
+ }
+ }
+}
+
+
+// This is a trivial subclass for java.io.StreamTokenizer which knows
+// about the multi-character tokens "<=", ">=" and "<>", and included
+// a render() method. Used only by CQLParser.
+//
+class CQLLexer extends StreamTokenizer {
+ private static boolean LEXDEBUG;
+ static int TT_LE = 1000; // The token "<="
+ static int TT_GE = 1001; // The token ">="
+ static int TT_NE = 1002; // The token "<>"
+
+ static void debug(String str) {
+ if (LEXDEBUG)
+ System.err.println("LEXDEBUG: " + str);
+ }
+
+ CQLLexer(String cql, boolean lexdebug) {
+ super(new StringReader(cql));
+ this.ordinaryChar('=');
+ this.ordinaryChar('<');
+ this.ordinaryChar('>');
+ this.ordinaryChar('/');
+ this.ordinaryChar('(');
+ this.ordinaryChar(')');
+ this.wordChars('\'', '\''); // prevent this from introducing strings
+ this.LEXDEBUG = lexdebug;
+ }
+
+ public int nextToken() throws java.io.IOException {
+ int token = super.nextToken();
+
+ if (token == '<') {
+ debug("token starts with '<' ...");
+ int t2 = super.nextToken();
+ if (t2 == '=') {
+ debug("token continues with '=' - it's '<='");
+ this.ttype = token = TT_LE;
+ } else if (t2 == '>') {
+ debug("token continues with '>' - it's '<>'");
+ this.ttype = token = TT_NE;
+ } else {
+ debug("next token is " + token + " (pushed back)");
+ //this.pushBack();
+ }
+ } else if (token == '>') {
+ debug("token starts with '>' ...");
+ int t2 = super.nextToken();
+ if (t2 == '=') {
+ debug("token continues with '=' - it's '>='");
+ this.ttype = token = TT_GE;
+ } else {
+ debug("next token is " + token + " (pushed back)");
+ //this.pushBack();
+ }
+ }
+
+ debug("token=" + token + ", " +
+ "nval=" + this.nval + ", " + "sval=" + this.sval);
+
+ return token;
+ }
+
+ String render() {
+ return this.render(this.ttype, null, true);
+ }
+
+ String render(boolean quoteChars) {
+ return this.render(this.ttype, null, quoteChars);
+ }
+
+ String render(int token, String str, boolean quoteChars) {
+ String ret;
+
+ if (token == this.TT_EOF) {
+ return "EOF";
+ } else if (token == this.TT_EOL) {
+ return "EOL";
+ } else if (token == this.TT_NUMBER) {
+ return "number: " + this.nval;
+ } else if (token == this.TT_WORD) {
+ return "word: " + this.sval;
+ } else if (token == '"') {
+ return "string: \"" + this.sval + "\"";
+ } else if (token == TT_LE) {
+ return "<=";
+ } else if (token == TT_GE) {
+ return ">=";
+ } else if (token == TT_NE) {
+ return "<>";
+ }
+
+ String res = String.valueOf((char) token);
+ if (quoteChars) res = "'" + res + "'";
+ return res;
}
}