1 // $Id: CQLParser.java,v 1.39 2007-08-06 15:54:48 mike Exp $
3 package org.z3950.zing.cql;
4 import java.io.IOException;
5 import java.util.Vector;
6 import java.util.Properties;
7 import java.io.InputStream;
8 import java.io.FileInputStream;
9 import java.io.FileNotFoundException;
13 * Compiles CQL strings into parse trees of CQLNode subtypes.
15 * @version $Id: CQLParser.java,v 1.39 2007-08-06 15:54:48 mike Exp $
16 * @see <A href="http://zing.z3950.org/cql/index.html"
17 * >http://zing.z3950.org/cql/index.html</A>
19 public class CQLParser {
20 private CQLLexer lexer;
21 private int compat; // When false, implement CQL 1.2
22 public static int V1POINT1 = 12368;
23 public static int V1POINT2 = 12369;
24 public static int V1POINT1SORT = 12370;
26 static private boolean DEBUG = false;
27 static private boolean LEXDEBUG = false;
30 * The new parser implements a dialect of CQL specified by the
31 * <tt>compat</tt> argument:
33 * <li>V1POINT1 - CQL version 1.1
35 * <li>V1POINT2 - CQL version 1.2
37 * <li>V1POINT1SORT - CQL version 1.1 but including
38 * <tt>sortby</tt> as specified for CQL 1.2.
42 public CQLParser(int compat) {
47 * The new parser implements CQL 1.2
50 this.compat = V1POINT2;
53 private static void debug(String str) {
55 System.err.println("PARSEDEBUG: " + str);
59 * Compiles a CQL query.
61 * The resulting parse tree may be further processed by hand (see
62 * the individual node-types' documentation for details on the
63 * data structure) or, more often, simply rendered out in the
64 * desired form using one of the back-ends. <TT>toCQL()</TT>
65 * returns a decompiled CQL query equivalent to the one that was
66 * compiled in the first place; <TT>toXCQL()</TT> returns an
67 * XML snippet representing the query; and <TT>toPQF()</TT>
68 * returns the query rendered in Index Data's Prefix Query
71 * @param cql The query
72 * @return A CQLNode object which is the root of a parse
73 * tree representing the query. */
74 public CQLNode parse(String cql)
75 throws CQLParseException, IOException {
76 lexer = new CQLLexer(cql, LEXDEBUG);
79 debug("about to parseQuery()");
80 CQLNode root = parseTopLevelPrefixes("cql.serverChoice",
81 new CQLRelation(compat == V1POINT2 ? "=" : "scr"));
82 if (lexer.ttype != lexer.TT_EOF)
83 throw new CQLParseException("junk after end: " + lexer.render());
88 private CQLNode parseTopLevelPrefixes(String index, CQLRelation relation)
89 throws CQLParseException, IOException {
90 debug("top-level prefix mapping");
92 if (lexer.ttype == '>') {
93 return parsePrefix(index, relation, true);
96 CQLNode node = parseQuery(index, relation);
97 if ((compat == V1POINT2 || compat == V1POINT1SORT) &&
98 lexer.ttype == lexer.TT_SORTBY) {
102 CQLSortNode sortnode = new CQLSortNode(node);
103 while (lexer.ttype != lexer.TT_EOF) {
104 String sortindex = matchSymbol("sort index");
105 ModifierSet ms = gatherModifiers(sortindex);
106 sortnode.addSortIndex(ms);
109 if (sortnode.keys.size() == 0) {
110 throw new CQLParseException("no sort keys");
119 private CQLNode parseQuery(String index, CQLRelation relation)
120 throws CQLParseException, IOException {
121 debug("in parseQuery()");
123 CQLNode term = parseTerm(index, relation);
124 while (lexer.ttype != lexer.TT_EOF &&
125 lexer.ttype != ')' &&
126 lexer.ttype != lexer.TT_SORTBY) {
127 if (lexer.ttype == lexer.TT_AND ||
128 lexer.ttype == lexer.TT_OR ||
129 lexer.ttype == lexer.TT_NOT ||
130 lexer.ttype == lexer.TT_PROX) {
131 int type = lexer.ttype;
132 String val = lexer.sval;
134 ModifierSet ms = gatherModifiers(val);
135 CQLNode term2 = parseTerm(index, relation);
136 term = ((type == lexer.TT_AND) ? new CQLAndNode(term, term2, ms) :
137 (type == lexer.TT_OR) ? new CQLOrNode (term, term2, ms) :
138 (type == lexer.TT_NOT) ? new CQLNotNode(term, term2, ms) :
139 new CQLProxNode(term, term2, ms));
141 throw new CQLParseException("expected boolean, got " +
146 debug("no more ops");
150 private ModifierSet gatherModifiers(String base)
151 throws CQLParseException, IOException {
152 debug("in gatherModifiers()");
154 ModifierSet ms = new ModifierSet(base);
155 while (lexer.ttype == '/') {
157 if (lexer.ttype != lexer.TT_WORD)
158 throw new CQLParseException("expected modifier, "
159 + "got " + lexer.render());
160 String type = lexer.sval.toLowerCase();
162 if (!isSymbolicRelation()) {
163 // It's a simple modifier consisting of type only
164 ms.addModifier(type);
166 // It's a complex modifier of the form type=value
167 String comparision = lexer.render(lexer.ttype, false);
169 String value = matchSymbol("modifier value");
170 ms.addModifier(type, comparision, value);
177 private CQLNode parseTerm(String index, CQLRelation relation)
178 throws CQLParseException, IOException {
179 debug("in parseTerm()");
183 if (lexer.ttype == '(') {
184 debug("parenthesised term");
186 CQLNode expr = parseQuery(index, relation);
189 } else if (lexer.ttype == '>') {
190 return parsePrefix(index, relation, false);
193 debug("non-parenthesised term");
194 word = matchSymbol("index or term");
195 if (!isSymbolicRelation() && lexer.ttype != lexer.TT_WORD)
199 String relstr = (lexer.ttype == lexer.TT_WORD ?
200 lexer.sval : lexer.render(lexer.ttype, false));
201 relation = new CQLRelation(relstr);
203 ModifierSet ms = gatherModifiers(relstr);
204 relation.setModifiers(ms);
205 debug("index='" + index + ", " +
206 "relation='" + relation.toCQL() + "'");
209 CQLTermNode node = new CQLTermNode(index, relation, word);
210 debug("made term node " + node.toCQL());
214 private CQLNode parsePrefix(String index, CQLRelation relation,
216 throws CQLParseException, IOException {
217 debug("prefix mapping");
221 String identifier = matchSymbol("prefix-name");
222 if (lexer.ttype == '=') {
225 identifier = matchSymbol("prefix-identifer");
227 CQLNode node = topLevel ?
228 parseTopLevelPrefixes(index, relation) :
229 parseQuery(index, relation);
231 return new CQLPrefixNode(name, identifier, node);
234 private boolean isRelation() {
235 debug("isRelation: checking ttype=" + lexer.ttype +
236 " (" + lexer.render() + ")");
237 if (lexer.ttype == lexer.TT_WORD &&
238 (lexer.sval == "exact" ||
239 lexer.sval == "any" ||
240 lexer.sval == "all" ||
241 (lexer.sval == "scr" && compat == V1POINT2)))
244 return isSymbolicRelation();
247 private boolean isSymbolicRelation() {
248 debug("isSymbolicRelation: checking ttype=" + lexer.ttype +
249 " (" + lexer.render() + ")");
250 return (lexer.ttype == '<' ||
251 lexer.ttype == '>' ||
252 lexer.ttype == '=' ||
253 lexer.ttype == lexer.TT_LE ||
254 lexer.ttype == lexer.TT_GE ||
255 lexer.ttype == lexer.TT_NE ||
256 lexer.ttype == lexer.TT_EQEQ);
259 private void match(int token)
260 throws CQLParseException, IOException {
261 debug("in match(" + lexer.render(token, true) + ")");
262 if (lexer.ttype != token)
263 throw new CQLParseException("expected " +
264 lexer.render(token, true) +
265 ", " + "got " + lexer.render());
266 int tmp = lexer.nextToken();
267 debug("match() got token=" + lexer.ttype + ", " +
268 "nval=" + lexer.nval + ", sval='" + lexer.sval + "'" +
269 " (tmp=" + tmp + ")");
272 private String matchSymbol(String expected)
273 throws CQLParseException, IOException {
275 debug("in matchSymbol()");
276 if (lexer.ttype == lexer.TT_WORD ||
277 lexer.ttype == lexer.TT_NUMBER ||
278 lexer.ttype == '"' ||
279 // The following is a complete list of keywords. Because
280 // they're listed here, they can be used unquoted as
281 // indexes, terms, prefix names and prefix identifiers.
282 // ### Instead, we should ask the lexer whether what we
283 // have is a keyword, and let the knowledge reside there.
284 lexer.ttype == lexer.TT_AND ||
285 lexer.ttype == lexer.TT_OR ||
286 lexer.ttype == lexer.TT_NOT ||
287 lexer.ttype == lexer.TT_PROX ||
288 lexer.ttype == lexer.TT_SORTBY) {
289 String symbol = (lexer.ttype == lexer.TT_NUMBER) ?
290 lexer.render() : lexer.sval;
295 throw new CQLParseException("expected " + expected + ", " +
296 "got " + lexer.render());
301 * Simple test-harness for the CQLParser class.
303 * Reads a CQL query either from its command-line argument, if
304 * there is one, or standard input otherwise. So these two
305 * invocations are equivalent:
307 * CQLParser 'au=(Kerninghan or Ritchie) and ti=Unix'
308 * echo au=(Kerninghan or Ritchie) and ti=Unix | CQLParser
310 * The test-harness parses the supplied query and renders is as
311 * XCQL, so that both of the invocations above produce the
316 * <value>and</value>
320 * <value>or</value>
322 * <searchClause>
323 * <index>au</index>
325 * <value>=</value>
327 * <term>Kerninghan</term>
328 * </searchClause>
329 * <searchClause>
330 * <index>au</index>
332 * <value>=</value>
334 * <term>Ritchie</term>
335 * </searchClause>
337 * <searchClause>
338 * <index>ti</index>
340 * <value>=</value>
342 * <term>Unix</term>
343 * </searchClause>
348 * CQL version 1.1 (default version 1.2)
350 * Debug mode: extra output written to stderr.
352 * Causes the output to be written in CQL rather than XCQL - that
353 * is, a query equivalent to that which was input, is output. In
354 * effect, the test harness acts as a query canonicaliser.
356 * The input query, either as XCQL [default] or CQL [if the
357 * <TT>-c</TT> option is supplied].
359 public static void main (String[] args) {
360 char mode = 'x'; // x=XCQL, c=CQL, p=PQF
363 Vector<String> argv = new Vector<String>();
364 for (int i = 0; i < args.length; i++) {
368 int compat = V1POINT2;
369 if (argv.size() > 0 && argv.get(0).equals("-1")) {
374 if (argv.size() > 0 && argv.get(0).equals("-d")) {
379 if (argv.size() > 0 && argv.get(0).equals("-c")) {
382 } else if (argv.size() > 1 && argv.get(0).equals("-p")) {
385 pfile = (String) argv.get(0);
389 if (argv.size() > 1) {
390 System.err.println("Usage: CQLParser [-1] [-d] [-c] " +
391 "[-p <pqf-properties> [<CQL-query>]");
392 System.err.println("If unspecified, query is read from stdin");
397 if (argv.size() == 1) {
398 cql = (String) argv.get(0);
400 byte[] bytes = new byte[10000];
402 // Read in the whole of standard input in one go
403 int nbytes = System.in.read(bytes);
404 } catch (IOException ex) {
405 System.err.println("Can't read query: " + ex.getMessage());
408 cql = new String(bytes);
411 CQLParser parser = new CQLParser(compat);
414 root = parser.parse(cql);
415 } catch (CQLParseException ex) {
416 System.err.println("Syntax error: " + ex.getMessage());
418 } catch (IOException ex) {
419 System.err.println("Can't compile query: " + ex.getMessage());
425 System.out.println(root.toCQL());
426 } else if (mode == 'p') {
427 InputStream f = new FileInputStream(pfile);
429 throw new FileNotFoundException(pfile);
431 Properties config = new Properties();
434 System.out.println(root.toPQF(config));
436 System.out.print(root.toXCQL(0));
438 } catch (IOException ex) {
439 System.err.println("Can't render query: " + ex.getMessage());
441 } catch (UnknownIndexException ex) {
442 System.err.println("Unknown index: " + ex.getMessage());
444 } catch (UnknownRelationException ex) {
445 System.err.println("Unknown relation: " + ex.getMessage());
447 } catch (UnknownRelationModifierException ex) {
448 System.err.println("Unknown relation modifier: " +
451 } catch (UnknownPositionException ex) {
452 System.err.println("Unknown position: " + ex.getMessage());
454 } catch (PQFTranslationException ex) {
455 // We catch all of this class's subclasses, so --
456 throw new Error("can't get a PQFTranslationException");