First semi-working version.

Can parse: foo or (bar and "baz")
No qualifiers, no relations, no modifiers, no proximity :-)

diff --git a/src/org/z3950/zing/cql/CQLBooleanNode.java b/src/org/z3950/zing/cql/CQLBooleanNode.java
index c56aac4..9908395 100644 (file)
--- a/src/org/z3950/zing/cql/CQLBooleanNode.java
+++ b/src/org/z3950/zing/cql/CQLBooleanNode.java
@@ -1,4 +1,4 @@
-// $Id: CQLBooleanNode.java,v 1.2 2002-10-25 16:11:05 mike Exp $
+// $Id: CQLBooleanNode.java,v 1.3 2002-10-25 16:56:43 mike Exp $
 
 package org.z3950.zing.cql;
 
@@ -7,9 +7,23 @@ package org.z3950.zing.cql;
  * Represents a boolean node in a CQL parse-tree ...
  * ###
  *
- * @version    $Id: CQLBooleanNode.java,v 1.2 2002-10-25 16:11:05 mike Exp $
+ * @version    $Id: CQLBooleanNode.java,v 1.3 2002-10-25 16:56:43 mike Exp $
  */
 public abstract class CQLBooleanNode extends CQLNode {
     protected CQLNode left;
     protected CQLNode right;
+
+    abstract String op();
+
+    String toXCQL(int level) {
+       return(indent(level) + "<triple>\n" +
+              indent(level+1) + "<boolean>" + op() + "</boolean>\n" +
+              left.toXCQL(level+1) +
+              right.toXCQL(level+1) +
+              indent(level) + "</triple>\n");
+    }
+
+    String toCQL() {
+       return "(" + left.toCQL() + ") " + op() + " (" + right.toCQL() + ")";
+    }
 }

diff --git a/src/org/z3950/zing/cql/CQLParser.java b/src/org/z3950/zing/cql/CQLParser.java
index 7a3df72..c086cac 100644 (file)
--- a/src/org/z3950/zing/cql/CQLParser.java
+++ b/src/org/z3950/zing/cql/CQLParser.java
@@ -1,9 +1,8 @@
-// $Id: CQLParser.java,v 1.6 2002-10-25 16:11:05 mike Exp $
+// $Id: CQLParser.java,v 1.7 2002-10-25 16:56:43 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;
@@ -13,7 +12,7 @@ import java.io.StreamTokenizer;
  * Compiles a CQL string into a parse tree ...
  * ###
  *
- * @version    $Id: CQLParser.java,v 1.6 2002-10-25 16:11:05 mike Exp $
+ * @version    $Id: CQLParser.java,v 1.7 2002-10-25 16:56:43 mike Exp $
  * @see                <A href="http://zing.z3950.org/cql/index.html"
  *                     >http://zing.z3950.org/cql/index.html</A>
  */
@@ -30,49 +29,44 @@ public class CQLParser {
     }
 
     public CQLNode parse(String cql)
-       throws FileNotFoundException, IOException {
+       throws CQLParseException, IOException {
        this.cql = cql;
        st = new StreamTokenizer(new StringReader(cql));
-       // ### these settings are wrong
-       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('>');
+       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();
-       CQLNode root;
-       try {
-           root = parse_expression();
-       } catch (CQLParseException ex) {
-           System.err.println("### Oops: " + ex);
-           return null;
+       if (false) {
+           // Lexical debug
+           int token;
+           while ((token = st.nextToken()) != st.TT_EOF) {
+               System.out.println("token=" + token + ", " +
+                                  "nval=" + st.nval + ", " +
+                                  "sval=" + st.sval);
+           }
+           System.exit(0);
        }
 
-       if (st.ttype != st.TT_EOF) {
-           System.err.println("### Extra bits: " + render(st));
-           return null;
-       }
+       st.nextToken();
+       System.err.println("*about to parse_query()");
+       CQLNode root = parse_query();
+       if (st.ttype != st.TT_EOF)
+           throw new CQLParseException("junk after end: " + render(st));
 
        return root;
     }
 
-    private CQLNode parse_expression()
+    private CQLNode parse_query()
        throws CQLParseException, IOException {
-       CQLNode term = parse_term();
+       System.err.println("*in parse_query()");
 
+       CQLNode term = parse_term();
        while (st.ttype == st.TT_WORD) {
            String op = st.sval.toLowerCase();
+           System.err.println("*checking op '" + op + "'");
            if (st.sval.equals("and")) {
                match(st.TT_WORD);
                CQLNode term2 = parse_term();
@@ -86,26 +80,35 @@ public class CQLParser {
                CQLNode term2 = parse_term();
                term = new CQLNotNode(term, term2);
            }
+           // ### Need to handle "prox"
        }
 
+       System.err.println("*no more ops");
        return term;
     }
 
     private CQLNode parse_term()
        throws CQLParseException, IOException {
+       System.err.println("*in parse_term()");
        if (st.ttype == '(') {
            match('(');
-           CQLNode expr = parse_expression();
+           CQLNode expr = parse_query();
            match(')');
            return expr;
        }
 
+       System.err.println("*not a parenthesised term");
+       // ### Need to parse qualifier-relation pairs
        String word = st.sval;
-       return new CQLTermNode("x", "=", word);
+       match(st.ttype);
+       CQLTermNode node = new CQLTermNode("x", "=", word);
+       System.err.println("*made term node " + node);
+       return node;
     }
 
     private void match(int token)
        throws CQLParseException, IOException {
+       System.err.println("*in match(" + render(st, token, null) + ")");
        if (st.ttype != token)
            throw new CQLParseException("expected " + render(st, token, null) +
                                        ", " + "got " + render(st));
@@ -125,11 +128,11 @@ public class CQLParser {
        } else if (token == st.TT_EOL) {
            return "EOL";
        } else if (token == st.TT_NUMBER) {
-           return "number";
+           return "number: " + st.nval;
        } else if (token == st.TT_WORD) {
-           return "word";
-       } else if (token == '"' && token == '\'') {
-           return "string";
+           return "word: \"" + st.sval + "\"";
+       } else if (token == '"' || token == '\'') {
+           return "string: \"" + st.sval + "\"";
        }
 
         return "'" + String.valueOf((char) token) + "'";
@@ -149,8 +152,9 @@ public class CQLParser {
            System.exit(1);
        }
 
-       byte[] bytes = new byte[1000];
+       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);
@@ -161,10 +165,14 @@ public class CQLParser {
        CQLNode root;
        try {
            root = parser.parse(cql);
+           System.err.println("root='" + root + "'");
            System.out.println(root.toXCQL(0));
+       } catch (CQLParseException ex) {
+           System.err.println("Syntax error: " + ex);
+           System.exit(3);
        } catch (java.io.IOException ex) {
            System.err.println("Can't compile query: " + ex);
-           System.exit(3);
+           System.exit(4);
        }
     }
 }

diff --git a/src/org/z3950/zing/cql/CQLTermNode.java b/src/org/z3950/zing/cql/CQLTermNode.java
index 625aab0..64a0a84 100644 (file)
--- a/src/org/z3950/zing/cql/CQLTermNode.java
+++ b/src/org/z3950/zing/cql/CQLTermNode.java
@@ -1,4 +1,4 @@
-// $Id: CQLTermNode.java,v 1.2 2002-10-25 16:01:26 mike Exp $
+// $Id: CQLTermNode.java,v 1.3 2002-10-25 16:56:43 mike Exp $
 
 package org.z3950.zing.cql;
 
@@ -7,48 +7,49 @@ package org.z3950.zing.cql;
  * Represents a terminal node in a CQL parse-tree ...
  * ###
  *
- * @version    $Id: CQLTermNode.java,v 1.2 2002-10-25 16:01:26 mike Exp $
+ * @version    $Id: CQLTermNode.java,v 1.3 2002-10-25 16:56:43 mike Exp $
  */
 public class CQLTermNode extends CQLNode {
     private String qualifier;
     private String relation;
-    private String value;
+    private String term;
 
-    public CQLTermNode(String qualifier, String relation, String value) {
+    public CQLTermNode(String qualifier, String relation, String term) {
        this.qualifier = qualifier;
        this.relation = relation;
-       this.value = value;
+       this.term = term;
     }
 
     String toXCQL(int level) {
        return (indent(level) + "<searchClause>\n" +
                indent(level+1) + "<index>" + qualifier + "<index>\n" +
                indent(level+1) + "<relation>" + relation + "<relation>\n" +
-               indent(level+1) + "<term>" + value + "<term>\n" +
+               indent(level+1) + "<term>" + term + "<term>\n" +
                indent(level) + "</searchClause>\n");
     }
 
     String toCQL() {
-       String res = value;
+       String quotedTerm = term;
 
-       if (res.indexOf('"') != -1) {
+       if (quotedTerm.indexOf('"') != -1) {
            // ### precede each '"' with a '/'
        }
 
-       if (res.indexOf('"') != -1 ||
-           res.indexOf(' ') != -1 ||
-           res.indexOf('\t') != -1 ||
-           res.indexOf('=') != -1 ||
-           res.indexOf('<') != -1 ||
-           res.indexOf('>') != -1 ||
-           res.indexOf('/') != -1 ||
-           res.indexOf('(') != -1 ||
-           res.indexOf(')') != -1) {
-           res = '"' + res + '"';
+       // ### There must be a better way ...
+       if (quotedTerm.indexOf('"') != -1 ||
+           quotedTerm.indexOf(' ') != -1 ||
+           quotedTerm.indexOf('\t') != -1 ||
+           quotedTerm.indexOf('=') != -1 ||
+           quotedTerm.indexOf('<') != -1 ||
+           quotedTerm.indexOf('>') != -1 ||
+           quotedTerm.indexOf('/') != -1 ||
+           quotedTerm.indexOf('(') != -1 ||
+           quotedTerm.indexOf(')') != -1) {
+           quotedTerm = '"' + quotedTerm + '"';
        }
 
        // ### The qualifier may need quoting.
        // ### We don't always need spaces around `relation'.
-       return qualifier + " " + relation + " " + value;
+       return qualifier + " " + relation + " " + quotedTerm;
     }
 }