X-Git-Url: http://sru.miketaylor.org.uk/?a=blobdiff_plain;f=src%2Forg%2Fz3950%2Fzing%2Fcql%2FCQLParser.java;h=0aded47d1785d238da77fc5658171bd1fbd67594;hb=f8154c71944186a9b64ddb782082a2026c5a912f;hp=bc19d48a2c9a3f7f1d24032282d77f4ffd9de1f8;hpb=99c44fee61010958478dea758d7b1322cd7a9ae6;p=cql-java-moved-to-github.git

diff --git a/src/org/z3950/zing/cql/CQLParser.java b/src/org/z3950/zing/cql/CQLParser.java
index bc19d48..0aded47 100644
--- a/src/org/z3950/zing/cql/CQLParser.java
+++ b/src/org/z3950/zing/cql/CQLParser.java
@@ -1,207 +1,279 @@
-// $Id: CQLParser.java,v 1.3 2002-10-24 16:06:34 mike Exp $
+// $Id: CQLParser.java,v 1.11 2002-10-31 22:22:01 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 ...
- * ###
+ * Compiles a CQL string into a parse tree.
+ * ##
  *
- * @version	$Id: CQLParser.java,v 1.3 2002-10-24 16:06:34 mike Exp $
+ * @version	$Id: CQLParser.java,v 1.11 2002-10-31 22:22:01 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;
-
-    private class CQLParseException extends Exception {
-	CQLParseException(String s) { super(s); }
-    }
+public class CQLParser {
+    private CQLLexer lexer;
+    static private boolean DEBUG = false;
+    static private boolean LEXDEBUG = false;
 
-    public CQLCompiler(String cql, String qualset) {
-	this.cql = cql;
-	this.qualset = qualset;
+    private static void debug(String str) {
+	if (DEBUG)
+	    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", new CQLRelation("="));
+	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, CQLRelation 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_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) {
+		match(lexer.TT_PROX);
+		CQLProxNode proxnode = new CQLProxNode(term);
+		gatherProxParameters(proxnode);
+		CQLNode term2 = parse_term(qualifier, relation);
+		proxnode.addSecondSubterm(term2);
+		term = (CQLNode) proxnode;
+	    } else {
+		throw new CQLParseException("expected boolean, got " +
+					    lexer.render());
+	    }
 	}
 
+	debug("no more ops");
 	return term;
     }
 
-    private String parse_term()
+    private CQLNode parse_term(String qualifier, CQLRelation 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 (!isBaseRelation())
 		break;
-	    }
 
-	    String attr = qualsetProperties.getProperty(word);
-	    if (attr == null) {
-		throw new CQLParseException("unrecognised qualifier: " + word);
+	    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);
 	    }
-	    attrs = attrs + attr + " ";
-	    match(st.ttype);
-	    word = null;	// mark as not-yet-read
+
+	    debug("qualifier='" + qualifier + ", " +
+		  "relation='" + relation.toCQL() + "'");
 	}
 
-	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.toCQL());
+	return node;
+    }
+
+    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) {
+		    // Assumes order is: relation/distance/unit/ordering
+		case 0: gatherProxRelation(node); break;
+		case 1: gatherProxDistance(node); break;
+		case 2: gatherProxUnit(node); break;
+		case 3: gatherProxOrdering(node); break;
+		}
 	    }
 	}
+    }
 
-	return attrs + word;
+    private void gatherProxRelation(CQLProxNode node)
+	throws CQLParseException, IOException {
+	if (!isProxRelation())
+	    throw new CQLParseException("expected proximity relation, got " +
+					lexer.render());
+	node.addModifier("relation", lexer.render(lexer.ttype, false));
+	match(lexer.ttype);
+	debug("gPR matched " + lexer.render(lexer.ttype, false));
     }
 
-    private void match(int token)
+    private void gatherProxDistance(CQLProxNode node)
 	throws CQLParseException, IOException {
-	if (st.ttype != token)
-	    throw new CQLParseException("expected " + render(st, token, null) +
-					", " + "got " + render(st));
-	st.nextToken();
+	if (lexer.ttype != lexer.TT_NUMBER)
+	    throw new CQLParseException("expected proximity distance, got " +
+					lexer.render());
+	node.addModifier("distance", lexer.render(lexer.ttype, false));
+	match(lexer.ttype);
+	debug("gPD matched " + lexer.render(lexer.ttype, false));
     }
 
-    // ### This utility should surely be a method of the StreamTokenizer class
-    private static String render(StreamTokenizer st) {
-	return render(st, st.ttype, null);
+    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", lexer.render());
+	match(lexer.ttype);
     }
 
-    private static String render(StreamTokenizer st, int token, String str) {
-	String ret;
+    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", lexer.render());
+	match(lexer.ttype);
+    }
 
-	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) + "'";
-	}
+    boolean isBaseRelation() {
+	debug("isBaseRelation: checking ttype=" + lexer.ttype +
+	      " (" + lexer.render() + ")");
+	return (isProxRelation() ||
+		lexer.ttype == lexer.TT_ANY ||
+		lexer.ttype == lexer.TT_ALL ||
+		lexer.ttype == lexer.TT_EXACT);
+    }
 
-	if (str != null)
-	    ret += "(\"" + str + "\")";
-	return ret;
+    boolean isProxRelation() {
+	debug("isProxRelation: 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);
     }
 
-    // ### Not really the right place for this test harness.
+    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. 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 > 1) {
+	    System.err.println("Usage: CQLParser [<CQL-query>]");
+	    System.err.println("If unspecified, query is read from stdin");
 	    System.exit(1);
 	}
 
-	CQLCompiler cc = new CQLCompiler(args[0], args[1]);
+	String cql;
+	if (args.length == 1) {
+	    cql = args[0];
+	} else {
+	    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);
+	    }
+	    cql = new String(bytes);
+	}
+
+	CQLParser parser = new CQLParser();
+	CQLNode root;
 	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);
-	    System.exit(2);
+	    root = parser.parse(cql);
+	    debug("root='" + root + "'");
+	    System.out.println(root.toCQL());
+	} 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);
 	}
     }
 }