-// $Id: CQLParser.java,v 1.39 2007-08-06 15:54:48 mike Exp $
package org.z3950.zing.cql;
+
+import java.io.BufferedReader;
import java.io.IOException;
import java.util.Properties;
import java.io.InputStream;
import java.io.FileInputStream;
-import java.io.FileNotFoundException;
+import java.io.InputStreamReader;
import java.util.ArrayList;
+import java.util.HashSet;
import java.util.List;
+import java.util.Set;
/**
* Compiles CQL strings into parse trees of CQLNode subtypes.
*
- * @version $Id: CQLParser.java,v 1.39 2007-08-06 15:54:48 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
+ private CQLTokenizer lexer;
+ private final int compat; // When false, implement CQL 1.2
+ private final Set<String> customRelations = new HashSet<String>();
+
public static final int V1POINT1 = 12368;
public static final int V1POINT2 = 12369;
public static final int V1POINT1SORT = 12370;
+ public final boolean allowKeywordTerms;
static private boolean DEBUG = false;
static private boolean LEXDEBUG = false;
*/
public CQLParser(int compat) {
this.compat = compat;
+ this.allowKeywordTerms = true;
}
-
+
+ /**
+ * Official CQL grammar allows registered keywords like 'and/or/not/sortby/prox'
+ * to be used unquoted in terms. This constructor allows to create an instance
+ * of a parser that prohibits this behavior while sacrificing compatibility.
+ * @param compat CQL version compatibility
+ * @param allowKeywordTerms when false registered keywords are disallowed in unquoted terms
+ */
+ public CQLParser(int compat, boolean allowKeywordTerms) {
+ this.compat = compat;
+ this.allowKeywordTerms = allowKeywordTerms;
+ }
+
/**
* The new parser implements CQL 1.2
*/
public CQLParser() {
this.compat = V1POINT2;
+ this.allowKeywordTerms = true;
}
private static void debug(String str) {
if (DEBUG)
System.err.println("PARSEDEBUG: " + str);
}
+
+ /**
+ * Registers custom relation in this parser. Note that when a custom relation
+ * is registered the parser is no longer strictly compliant with the chosen spec.
+ * @param relation
+ * @return true if custom relation has not been registered already
+ */
+ public boolean registerCustomRelation(String relation) {
+ return customRelations.add(relation);
+ }
+
+ /**
+ * Unregisters previously registered custom relation in this instance of the parser.
+ * @param relation
+ * @return true is relation has been previously registered
+ */
+ public boolean unregisterCustomRelation(String relation) {
+ return customRelations.remove(relation);
+ }
/**
* Compiles a CQL query.
throws CQLParseException, IOException {
lexer = new CQLLexer(cql, LEXDEBUG);
- lexer.nextToken();
+ lexer.move();
debug("about to parseQuery()");
CQLNode root = parseTopLevelPrefixes("cql.serverChoice",
new CQLRelation(compat == V1POINT2 ? "=" : "scr"));
- if (lexer.ttype != CQLLexer.TT_EOF)
- throw new CQLParseException("junk after end: " + lexer.render());
+ if (lexer.what() != CQLTokenizer.TT_EOF)
+ throw new CQLParseException("junk after end: " + lexer.render(),
+ lexer.pos());
return root;
}
throws CQLParseException, IOException {
debug("top-level prefix mapping");
- if (lexer.ttype == '>') {
+ if (lexer.what() == '>') {
return parsePrefix(index, relation, true);
}
CQLNode node = parseQuery(index, relation);
if ((compat == V1POINT2 || compat == V1POINT1SORT) &&
- lexer.ttype == CQLLexer.TT_SORTBY) {
- match(lexer.ttype);
+ lexer.what() == CQLTokenizer.TT_SORTBY) {
+ match(lexer.what());
debug("sortspec");
CQLSortNode sortnode = new CQLSortNode(node);
- while (lexer.ttype != CQLLexer.TT_EOF) {
+ while (lexer.what() != CQLTokenizer.TT_EOF) {
String sortindex = matchSymbol("sort index");
ModifierSet ms = gatherModifiers(sortindex);
sortnode.addSortIndex(ms);
}
if (sortnode.keys.size() == 0) {
- throw new CQLParseException("no sort keys");
+ throw new CQLParseException("no sort keys", lexer.pos());
}
node = sortnode;
debug("in parseQuery()");
CQLNode term = parseTerm(index, relation);
- while (lexer.ttype != CQLLexer.TT_EOF &&
- lexer.ttype != ')' &&
- lexer.ttype != CQLLexer.TT_SORTBY) {
- if (lexer.ttype == CQLLexer.TT_AND ||
- lexer.ttype == CQLLexer.TT_OR ||
- lexer.ttype == CQLLexer.TT_NOT ||
- lexer.ttype == CQLLexer.TT_PROX) {
- int type = lexer.ttype;
- String val = lexer.sval;
+ while (lexer.what() != CQLTokenizer.TT_EOF &&
+ lexer.what() != ')' &&
+ lexer.what() != CQLTokenizer.TT_SORTBY) {
+ if (lexer.what() == CQLTokenizer.TT_AND ||
+ lexer.what() == CQLTokenizer.TT_OR ||
+ lexer.what() == CQLTokenizer.TT_NOT ||
+ lexer.what() == CQLTokenizer.TT_PROX) {
+ int type = lexer.what();
+ String val = lexer.value();
match(type);
ModifierSet ms = gatherModifiers(val);
CQLNode term2 = parseTerm(index, relation);
- term = ((type == CQLLexer.TT_AND) ? new CQLAndNode(term, term2, ms) :
- (type == CQLLexer.TT_OR) ? new CQLOrNode (term, term2, ms) :
- (type == CQLLexer.TT_NOT) ? new CQLNotNode(term, term2, ms) :
+ term = ((type == CQLTokenizer.TT_AND) ? new CQLAndNode(term, term2, ms) :
+ (type == CQLTokenizer.TT_OR) ? new CQLOrNode (term, term2, ms) :
+ (type == CQLTokenizer.TT_NOT) ? new CQLNotNode(term, term2, ms) :
new CQLProxNode(term, term2, ms));
} else {
throw new CQLParseException("expected boolean, got " +
- lexer.render());
+ lexer.render(), lexer.pos());
}
}
debug("in gatherModifiers()");
ModifierSet ms = new ModifierSet(base);
- while (lexer.ttype == '/') {
+ while (lexer.what() == '/') {
match('/');
- if (lexer.ttype != CQLLexer.TT_WORD)
+ if (lexer.what() != CQLTokenizer.TT_WORD)
throw new CQLParseException("expected modifier, "
- + "got " + lexer.render());
- String type = lexer.sval.toLowerCase();
- match(lexer.ttype);
+ + "got " + lexer.render(),
+ lexer.pos());
+ String type = lexer.value().toLowerCase();
+ match(lexer.what());
if (!isSymbolicRelation()) {
// 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 comparision = lexer.render(lexer.what(), false);
+ match(lexer.what());
String value = matchSymbol("modifier value");
ms.addModifier(type, comparision, value);
}
throws CQLParseException, IOException {
debug("in parseTerm()");
- String word;
+ String first;
+ StringBuilder all;
while (true) {
- if (lexer.ttype == '(') {
+ if (lexer.what() == '(') {
debug("parenthesised term");
match('(');
CQLNode expr = parseQuery(index, relation);
match(')');
return expr;
- } else if (lexer.ttype == '>') {
+ } else if (lexer.what() == '>') {
return parsePrefix(index, relation, false);
}
debug("non-parenthesised term");
- word = matchSymbol("index or term");
- while (lexer.ttype == CQLLexer.TT_WORD && !isRelation()) {
- word = word + " " + lexer.sval;
- match(CQLLexer.TT_WORD);
+ first = matchSymbol("index or term");
+ all = new StringBuilder(first);
+ //match relation only on second postion
+ while (isWordOrString() && (all.length() > first.length() || !isRelation())) {
+ all.append(" ").append(lexer.value());
+ match(lexer.what());
}
if (!isRelation())
- break;
-
- index = word;
- String relstr = (lexer.ttype == CQLLexer.TT_WORD ?
- lexer.sval : lexer.render(lexer.ttype, false));
+ break; //we're done if no relation
+
+ //we have relation, but it only makes sense if preceded by a single term
+ if (all.length() > first.length()) {
+ throw new CQLParseException("unexpected relation '"+lexer.value()+"'"
+ , lexer.pos());
+ }
+ index = first;
+ String relstr = (lexer.what() == CQLTokenizer.TT_WORD ?
+ lexer.value() : lexer.render(lexer.what(), false));
relation = new CQLRelation(relstr);
- match(lexer.ttype);
+ match(lexer.what());
ModifierSet ms = gatherModifiers(relstr);
relation.ms = ms;
debug("index='" + index + ", " +
"relation='" + relation.toCQL() + "'");
}
-
- CQLTermNode node = new CQLTermNode(index, relation, word);
+ CQLTermNode node = new CQLTermNode(index, relation, all.toString());
debug("made term node " + node.toCQL());
return node;
}
match('>');
String name = null;
String identifier = matchSymbol("prefix-name");
- if (lexer.ttype == '=') {
+ if (lexer.what() == '=') {
match('=');
name = identifier;
identifier = matchSymbol("prefix-identifer");
return new CQLPrefixNode(name, identifier, node);
}
+
+ private boolean isWordOrString() {
+ return CQLTokenizer.TT_WORD == lexer.what()
+ || CQLTokenizer.TT_STRING == lexer.what();
+ }
private boolean isRelation() {
- debug("isRelation: checking ttype=" + lexer.ttype +
+ debug("isRelation: checking what()=" + lexer.what() +
" (" + lexer.render() + ")");
- if (lexer.ttype == CQLLexer.TT_WORD &&
- (lexer.sval.indexOf('.') >= 0 ||
- lexer.sval.equals("any") ||
- lexer.sval.equals("all") ||
- lexer.sval.equals("within") ||
- lexer.sval.equals("encloses") ||
- lexer.sval.equals("exact") ||
- (lexer.sval.equals("scr") && compat != V1POINT2) ||
- (lexer.sval.equals("adj") && compat == V1POINT2)))
+ if (lexer.what() == CQLTokenizer.TT_WORD &&
+ (lexer.value().indexOf('.') >= 0 ||
+ lexer.value().equals("any") ||
+ lexer.value().equals("all") ||
+ lexer.value().equals("within") ||
+ lexer.value().equals("encloses") ||
+ (lexer.value().equals("exact") && compat != V1POINT2) ||
+ (lexer.value().equals("scr") && compat != V1POINT2) ||
+ (lexer.value().equals("adj") && compat == V1POINT2) ||
+ customRelations.contains(lexer.value())))
return true;
return isSymbolicRelation();
}
private boolean isSymbolicRelation() {
- debug("isSymbolicRelation: checking ttype=" + lexer.ttype +
+ debug("isSymbolicRelation: checking what()=" + lexer.what() +
" (" + lexer.render() + ")");
- return (lexer.ttype == '<' ||
- lexer.ttype == '>' ||
- lexer.ttype == '=' ||
- lexer.ttype == CQLLexer.TT_LE ||
- lexer.ttype == CQLLexer.TT_GE ||
- lexer.ttype == CQLLexer.TT_NE ||
- lexer.ttype == CQLLexer.TT_EQEQ);
+ return (lexer.what() == '<' ||
+ lexer.what() == '>' ||
+ lexer.what() == '=' ||
+ lexer.what() == CQLTokenizer.TT_LE ||
+ lexer.what() == CQLTokenizer.TT_GE ||
+ lexer.what() == CQLTokenizer.TT_NE ||
+ lexer.what() == CQLTokenizer.TT_EQEQ);
}
private void match(int token)
throws CQLParseException, IOException {
debug("in match(" + lexer.render(token, true) + ")");
- if (lexer.ttype != token)
+ if (lexer.what() != 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 + ")");
+ ", " + "got " + lexer.render(),
+ lexer.pos());
+ lexer.move();
+ debug("match() got token=" + lexer.what() + ", value()='" + lexer.value() + "'");
}
private String matchSymbol(String expected)
throws CQLParseException, IOException {
debug("in matchSymbol()");
- if (lexer.ttype == CQLLexer.TT_WORD ||
- lexer.ttype == CQLLexer.TT_NUMBER ||
- lexer.ttype == '"' ||
+ if (lexer.what() == CQLTokenizer.TT_WORD ||
+ lexer.what() == CQLTokenizer.TT_STRING ||
// The following is a complete list of keywords. Because
// they're listed here, they can be used unquoted as
// indexes, terms, prefix names and prefix identifiers.
- // ### Instead, we should ask the lexer whether what we
- // have is a keyword, and let the knowledge reside there.
- lexer.ttype == CQLLexer.TT_AND ||
- lexer.ttype == CQLLexer.TT_OR ||
- lexer.ttype == CQLLexer.TT_NOT ||
- lexer.ttype == CQLLexer.TT_PROX ||
- lexer.ttype == CQLLexer.TT_SORTBY) {
- String symbol = (lexer.ttype == CQLLexer.TT_NUMBER) ?
- lexer.render() : lexer.sval;
- match(lexer.ttype);
+ (allowKeywordTerms &&
+ lexer.what() == CQLTokenizer.TT_AND ||
+ lexer.what() == CQLTokenizer.TT_OR ||
+ lexer.what() == CQLTokenizer.TT_NOT ||
+ lexer.what() == CQLTokenizer.TT_PROX ||
+ lexer.what() == CQLTokenizer.TT_SORTBY)) {
+ String symbol = lexer.value();
+ match(lexer.what());
return symbol;
}
throw new CQLParseException("expected " + expected + ", " +
- "got " + lexer.render());
+ "got " + lexer.render(), lexer.pos());
}
if (argv.size() == 1) {
cql = (String) argv.get(0);
} else {
- byte[] bytes = new byte[10000];
+ BufferedReader buff = new BufferedReader(new InputStreamReader(System.in));
try {
- // Read in the whole of standard input in one go
- int nbytes = System.in.read(bytes);
+ // read a single line of input
+ cql = buff.readLine();
+ if (cql == null) {
+ System.err.println("Can't read query from stdin");
+ System.exit(2);
+ return;
+ }
} catch (IOException ex) {
System.err.println("Can't read query: " + ex.getMessage());
System.exit(2);
+ return;
}
- cql = new String(bytes);
}
CQLParser parser = new CQLParser(compat);
- CQLNode root = null;
+ CQLNode root;
try {
root = parser.parse(cql);
} catch (CQLParseException ex) {
System.err.println("Syntax error: " + ex.getMessage());
+ StringBuilder space = new StringBuilder(cql.length());
+ System.out.println(cql);
+ for (int i=0; i<ex.getPosition(); i++) space.append(" ");
+ space.append("^");
+ System.err.println(space.toString());
System.exit(3);
+ return; //compiler
} catch (IOException ex) {
System.err.println("Can't compile query: " + ex.getMessage());
System.exit(4);
+ return; //compiler
}
try {
if (mode == 'c') {
System.out.println(root.toCQL());
} else if (mode == 'p') {
+ try {
InputStream f = new FileInputStream(pfile);
- if (f == null)
- throw new FileNotFoundException(pfile);
-
Properties config = new Properties();
config.load(f);
f.close();
System.out.println(root.toPQF(config));
+ } catch (IOException ex) {
+ System.err.println("Can't load PQF properties:" +
+ ex.getMessage());
+ System.exit(5);
+ }
} else {
- System.out.print(root.toXCQL(0));
+ System.out.print(root.toXCQL());
}
- } catch (IOException ex) {
- System.err.println("Can't render query: " + ex.getMessage());
- System.exit(5);
} catch (UnknownIndexException ex) {
System.err.println("Unknown index: " + ex.getMessage());
System.exit(6);
System.err.println("Unknown position: " + ex.getMessage());
System.exit(9);
} catch (PQFTranslationException ex) {
- // We catch all of this class's subclasses, so --
- throw new Error("can't get a PQFTranslationException");
+ System.err.println("Cannot translate to PQF: " + ex.getMessage());
+ System.exit(10);
}
}
}