| 1 |
155
|
jones
|
/**
|
| 2 |
203
|
jones
|
* '$RCSfile$'
|
| 3 |
|
|
* Purpose: A Class that represents a structured query, and can be
|
| 4 |
|
|
* constructed from an XML serialization conforming to
|
| 5 |
|
|
* pathquery.dtd. The printSQL() method can be used to print
|
| 6 |
|
|
* a SQL serialization of the query.
|
| 7 |
|
|
* Copyright: 2000 Regents of the University of California and the
|
| 8 |
|
|
* National Center for Ecological Analysis and Synthesis
|
| 9 |
|
|
* Authors: Matt Jones
|
| 10 |
349
|
jones
|
* Release: @release@
|
| 11 |
155
|
jones
|
*
|
| 12 |
203
|
jones
|
* '$Author$'
|
| 13 |
|
|
* '$Date$'
|
| 14 |
|
|
* '$Revision$'
|
| 15 |
669
|
jones
|
*
|
| 16 |
|
|
* This program is free software; you can redistribute it and/or modify
|
| 17 |
|
|
* it under the terms of the GNU General Public License as published by
|
| 18 |
|
|
* the Free Software Foundation; either version 2 of the License, or
|
| 19 |
|
|
* (at your option) any later version.
|
| 20 |
|
|
*
|
| 21 |
|
|
* This program is distributed in the hope that it will be useful,
|
| 22 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
| 23 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
| 24 |
|
|
* GNU General Public License for more details.
|
| 25 |
|
|
*
|
| 26 |
|
|
* You should have received a copy of the GNU General Public License
|
| 27 |
|
|
* along with this program; if not, write to the Free Software
|
| 28 |
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
| 29 |
155
|
jones
|
*/
|
| 30 |
|
|
|
| 31 |
|
|
package edu.ucsb.nceas.metacat;
|
| 32 |
|
|
|
| 33 |
|
|
import java.io.*;
|
| 34 |
|
|
import java.util.Stack;
|
| 35 |
158
|
jones
|
import java.util.Vector;
|
| 36 |
159
|
jones
|
import java.util.Enumeration;
|
| 37 |
155
|
jones
|
|
| 38 |
185
|
jones
|
import org.xml.sax.Attributes;
|
| 39 |
158
|
jones
|
import org.xml.sax.InputSource;
|
| 40 |
|
|
import org.xml.sax.SAXException;
|
| 41 |
|
|
import org.xml.sax.SAXParseException;
|
| 42 |
185
|
jones
|
import org.xml.sax.XMLReader;
|
| 43 |
|
|
import org.xml.sax.helpers.XMLReaderFactory;
|
| 44 |
|
|
import org.xml.sax.helpers.DefaultHandler;
|
| 45 |
155
|
jones
|
|
| 46 |
402
|
berkley
|
/**
|
| 47 |
172
|
jones
|
* A Class that represents a structured query, and can be
|
| 48 |
|
|
* constructed from an XML serialization conforming to @see pathquery.dtd.
|
| 49 |
|
|
* The printSQL() method can be used to print a SQL serialization of the query.
|
| 50 |
155
|
jones
|
*/
|
| 51 |
185
|
jones
|
public class QuerySpecification extends DefaultHandler {
|
| 52 |
155
|
jones
|
|
| 53 |
402
|
berkley
|
private boolean containsExtendedSQL=false;
|
| 54 |
|
|
|
| 55 |
158
|
jones
|
// Query data structures
|
| 56 |
|
|
private String meta_file_id;
|
| 57 |
692
|
bojilova
|
// DOCTITLE attr cleared from the db
|
| 58 |
|
|
// private String querytitle;
|
| 59 |
172
|
jones
|
private Vector doctypeList;
|
| 60 |
402
|
berkley
|
private Vector returnFieldList;
|
| 61 |
535
|
jones
|
private Vector ownerList;
|
| 62 |
|
|
private Vector siteList;
|
| 63 |
158
|
jones
|
private QueryGroup query = null;
|
| 64 |
|
|
|
| 65 |
|
|
private Stack elementStack;
|
| 66 |
|
|
private Stack queryStack;
|
| 67 |
159
|
jones
|
private String currentValue;
|
| 68 |
|
|
private String currentPathexpr;
|
| 69 |
172
|
jones
|
private String parserName = null;
|
| 70 |
535
|
jones
|
private String accNumberSeparator = null;
|
| 71 |
158
|
jones
|
|
| 72 |
155
|
jones
|
/**
|
| 73 |
|
|
* construct an instance of the QuerySpecification class
|
| 74 |
|
|
*
|
| 75 |
172
|
jones
|
* @param queryspec the XML representation of the query (should conform
|
| 76 |
|
|
* to pathquery.dtd) as a Reader
|
| 77 |
|
|
* @param parserName the fully qualified name of a Java Class implementing
|
| 78 |
185
|
jones
|
* the org.xml.sax.XMLReader interface
|
| 79 |
155
|
jones
|
*/
|
| 80 |
535
|
jones
|
public QuerySpecification( Reader queryspec, String parserName,
|
| 81 |
|
|
String accNumberSeparator ) throws IOException {
|
| 82 |
155
|
jones
|
super();
|
| 83 |
402
|
berkley
|
|
| 84 |
172
|
jones
|
// Initialize the class variables
|
| 85 |
|
|
doctypeList = new Vector();
|
| 86 |
158
|
jones
|
elementStack = new Stack();
|
| 87 |
|
|
queryStack = new Stack();
|
| 88 |
402
|
berkley
|
returnFieldList = new Vector();
|
| 89 |
535
|
jones
|
ownerList = new Vector();
|
| 90 |
|
|
siteList = new Vector();
|
| 91 |
172
|
jones
|
this.parserName = parserName;
|
| 92 |
535
|
jones
|
this.accNumberSeparator = accNumberSeparator;
|
| 93 |
158
|
jones
|
|
| 94 |
|
|
// Initialize the parser and read the queryspec
|
| 95 |
185
|
jones
|
XMLReader parser = initializeParser();
|
| 96 |
181
|
jones
|
if (parser == null) {
|
| 97 |
|
|
System.err.println("SAX parser not instantiated properly.");
|
| 98 |
|
|
}
|
| 99 |
155
|
jones
|
try {
|
| 100 |
|
|
parser.parse(new InputSource(queryspec));
|
| 101 |
|
|
} catch (SAXException e) {
|
| 102 |
675
|
berkley
|
System.err.println("error parsing data in " +
|
| 103 |
|
|
"QuerySpecification.QuerySpecification");
|
| 104 |
180
|
jones
|
System.err.println(e.getMessage());
|
| 105 |
155
|
jones
|
}
|
| 106 |
|
|
}
|
| 107 |
|
|
|
| 108 |
|
|
/**
|
| 109 |
|
|
* construct an instance of the QuerySpecification class
|
| 110 |
|
|
*
|
| 111 |
172
|
jones
|
* @param queryspec the XML representation of the query (should conform
|
| 112 |
|
|
* to pathquery.dtd) as a String
|
| 113 |
|
|
* @param parserName the fully qualified name of a Java Class implementing
|
| 114 |
|
|
* the org.xml.sax.Parser interface
|
| 115 |
155
|
jones
|
*/
|
| 116 |
535
|
jones
|
public QuerySpecification( String queryspec, String parserName,
|
| 117 |
|
|
String accNumberSeparator) throws IOException {
|
| 118 |
|
|
this(new StringReader(queryspec), parserName, accNumberSeparator);
|
| 119 |
155
|
jones
|
}
|
| 120 |
|
|
|
| 121 |
|
|
/** Main routine for testing */
|
| 122 |
|
|
static public void main(String[] args) {
|
| 123 |
|
|
|
| 124 |
|
|
if (args.length < 1) {
|
| 125 |
|
|
System.err.println("Wrong number of arguments!!!");
|
| 126 |
|
|
System.err.println("USAGE: java QuerySpecification <xmlfile>");
|
| 127 |
|
|
return;
|
| 128 |
|
|
} else {
|
| 129 |
711
|
jones
|
int i = 0;
|
| 130 |
|
|
boolean useXMLIndex = true;
|
| 131 |
|
|
if ( args[i].equals( "-noindex" ) ) {
|
| 132 |
|
|
useXMLIndex = false;
|
| 133 |
|
|
i++;
|
| 134 |
|
|
}
|
| 135 |
|
|
String xmlfile = args[i];
|
| 136 |
|
|
|
| 137 |
155
|
jones
|
try {
|
| 138 |
203
|
jones
|
MetaCatUtil util = new MetaCatUtil();
|
| 139 |
155
|
jones
|
FileReader xml = new FileReader(new File(xmlfile));
|
| 140 |
203
|
jones
|
QuerySpecification qspec =
|
| 141 |
535
|
jones
|
new QuerySpecification(xml, util.getOption("saxparser"),
|
| 142 |
|
|
util.getOption("accNumberSeparator"));
|
| 143 |
706
|
bojilova
|
System.out.println(qspec.printSQL(useXMLIndex));
|
| 144 |
181
|
jones
|
|
| 145 |
155
|
jones
|
} catch (IOException e) {
|
| 146 |
|
|
System.err.println(e.getMessage());
|
| 147 |
|
|
}
|
| 148 |
|
|
|
| 149 |
|
|
}
|
| 150 |
|
|
}
|
| 151 |
402
|
berkley
|
|
| 152 |
|
|
/**
|
| 153 |
|
|
* Returns true if the parsed query contains and extended xml query
|
| 154 |
|
|
* (i.e. there is at least one <returnfield> in the pathquery document)
|
| 155 |
|
|
*/
|
| 156 |
|
|
public boolean containsExtendedSQL()
|
| 157 |
|
|
{
|
| 158 |
|
|
if(containsExtendedSQL)
|
| 159 |
|
|
{
|
| 160 |
|
|
return true;
|
| 161 |
|
|
}
|
| 162 |
|
|
else
|
| 163 |
|
|
{
|
| 164 |
|
|
return false;
|
| 165 |
|
|
}
|
| 166 |
|
|
}
|
| 167 |
|
|
|
| 168 |
|
|
/**
|
| 169 |
|
|
* Accessor method to return a vector of the extended return fields as
|
| 170 |
|
|
* defined in the <returnfield> tag in the pathquery dtd.
|
| 171 |
|
|
*/
|
| 172 |
|
|
public Vector getReturnFieldList()
|
| 173 |
|
|
{
|
| 174 |
|
|
return this.returnFieldList;
|
| 175 |
|
|
}
|
| 176 |
155
|
jones
|
|
| 177 |
172
|
jones
|
/**
|
| 178 |
|
|
* Set up the SAX parser for reading the XML serialized query
|
| 179 |
|
|
*/
|
| 180 |
185
|
jones
|
private XMLReader initializeParser() {
|
| 181 |
|
|
XMLReader parser = null;
|
| 182 |
172
|
jones
|
|
| 183 |
155
|
jones
|
// Set up the SAX document handlers for parsing
|
| 184 |
|
|
try {
|
| 185 |
|
|
|
| 186 |
|
|
// Get an instance of the parser
|
| 187 |
185
|
jones
|
parser = XMLReaderFactory.createXMLReader(parserName);
|
| 188 |
155
|
jones
|
|
| 189 |
185
|
jones
|
// Set the ContentHandler to this instance
|
| 190 |
|
|
parser.setContentHandler(this);
|
| 191 |
155
|
jones
|
|
| 192 |
185
|
jones
|
// Set the error Handler to this instance
|
| 193 |
158
|
jones
|
parser.setErrorHandler(this);
|
| 194 |
155
|
jones
|
|
| 195 |
|
|
} catch (Exception e) {
|
| 196 |
675
|
berkley
|
System.err.println("Error in QuerySpcecification.initializeParser " +
|
| 197 |
|
|
e.toString());
|
| 198 |
155
|
jones
|
}
|
| 199 |
|
|
|
| 200 |
|
|
return parser;
|
| 201 |
|
|
}
|
| 202 |
|
|
|
| 203 |
172
|
jones
|
/**
|
| 204 |
|
|
* callback method used by the SAX Parser when the start tag of an
|
| 205 |
|
|
* element is detected. Used in this context to parse and store
|
| 206 |
|
|
* the query information in class variables.
|
| 207 |
|
|
*/
|
| 208 |
185
|
jones
|
public void startElement (String uri, String localName,
|
| 209 |
|
|
String qName, Attributes atts)
|
| 210 |
155
|
jones
|
throws SAXException {
|
| 211 |
185
|
jones
|
BasicNode currentNode = new BasicNode(localName);
|
| 212 |
159
|
jones
|
// add attributes to BasicNode here
|
| 213 |
|
|
if (atts != null) {
|
| 214 |
|
|
int len = atts.getLength();
|
| 215 |
|
|
for (int i = 0; i < len; i++) {
|
| 216 |
185
|
jones
|
currentNode.setAttribute(atts.getLocalName(i), atts.getValue(i));
|
| 217 |
159
|
jones
|
}
|
| 218 |
|
|
}
|
| 219 |
|
|
|
| 220 |
158
|
jones
|
elementStack.push(currentNode);
|
| 221 |
159
|
jones
|
if (currentNode.getTagName().equals("querygroup")) {
|
| 222 |
158
|
jones
|
QueryGroup currentGroup = new QueryGroup(
|
| 223 |
178
|
jones
|
currentNode.getAttribute("operator"));
|
| 224 |
158
|
jones
|
if (query == null) {
|
| 225 |
|
|
query = currentGroup;
|
| 226 |
159
|
jones
|
} else {
|
| 227 |
|
|
QueryGroup parentGroup = (QueryGroup)queryStack.peek();
|
| 228 |
|
|
parentGroup.addChild(currentGroup);
|
| 229 |
158
|
jones
|
}
|
| 230 |
159
|
jones
|
queryStack.push(currentGroup);
|
| 231 |
158
|
jones
|
}
|
| 232 |
155
|
jones
|
}
|
| 233 |
|
|
|
| 234 |
172
|
jones
|
/**
|
| 235 |
|
|
* callback method used by the SAX Parser when the end tag of an
|
| 236 |
|
|
* element is detected. Used in this context to parse and store
|
| 237 |
|
|
* the query information in class variables.
|
| 238 |
|
|
*/
|
| 239 |
185
|
jones
|
public void endElement (String uri, String localName,
|
| 240 |
|
|
String qName) throws SAXException {
|
| 241 |
158
|
jones
|
BasicNode leaving = (BasicNode)elementStack.pop();
|
| 242 |
159
|
jones
|
if (leaving.getTagName().equals("queryterm")) {
|
| 243 |
|
|
boolean isCaseSensitive = (new Boolean(
|
| 244 |
|
|
leaving.getAttribute("casesensitive"))).booleanValue();
|
| 245 |
|
|
QueryTerm currentTerm = null;
|
| 246 |
|
|
if (currentPathexpr == null) {
|
| 247 |
|
|
currentTerm = new QueryTerm(isCaseSensitive,
|
| 248 |
|
|
leaving.getAttribute("searchmode"),currentValue);
|
| 249 |
|
|
} else {
|
| 250 |
|
|
currentTerm = new QueryTerm(isCaseSensitive,
|
| 251 |
|
|
leaving.getAttribute("searchmode"),currentValue,
|
| 252 |
|
|
currentPathexpr);
|
| 253 |
|
|
}
|
| 254 |
|
|
QueryGroup currentGroup = (QueryGroup)queryStack.peek();
|
| 255 |
|
|
currentGroup.addChild(currentTerm);
|
| 256 |
|
|
currentValue = null;
|
| 257 |
|
|
currentPathexpr = null;
|
| 258 |
|
|
} else if (leaving.getTagName().equals("querygroup")) {
|
| 259 |
|
|
QueryGroup leavingGroup = (QueryGroup)queryStack.pop();
|
| 260 |
158
|
jones
|
}
|
| 261 |
155
|
jones
|
}
|
| 262 |
158
|
jones
|
|
| 263 |
172
|
jones
|
/**
|
| 264 |
|
|
* callback method used by the SAX Parser when the text sequences of an
|
| 265 |
|
|
* xml stream are detected. Used in this context to parse and store
|
| 266 |
|
|
* the query information in class variables.
|
| 267 |
|
|
*/
|
| 268 |
158
|
jones
|
public void characters(char ch[], int start, int length) {
|
| 269 |
|
|
|
| 270 |
|
|
String inputString = new String(ch, start, length);
|
| 271 |
|
|
BasicNode currentNode = (BasicNode)elementStack.peek();
|
| 272 |
|
|
String currentTag = currentNode.getTagName();
|
| 273 |
|
|
if (currentTag.equals("meta_file_id")) {
|
| 274 |
|
|
meta_file_id = inputString;
|
| 275 |
692
|
bojilova
|
// DOCTITLE attr cleared from the db
|
| 276 |
|
|
// } else if (currentTag.equals("querytitle")) {
|
| 277 |
|
|
// querytitle = inputString;
|
| 278 |
159
|
jones
|
} else if (currentTag.equals("value")) {
|
| 279 |
|
|
currentValue = inputString;
|
| 280 |
|
|
} else if (currentTag.equals("pathexpr")) {
|
| 281 |
|
|
currentPathexpr = inputString;
|
| 282 |
172
|
jones
|
} else if (currentTag.equals("returndoctype")) {
|
| 283 |
|
|
doctypeList.add(inputString);
|
| 284 |
402
|
berkley
|
} else if (currentTag.equals("returnfield")) {
|
| 285 |
|
|
returnFieldList.add(inputString);
|
| 286 |
|
|
containsExtendedSQL = true;
|
| 287 |
535
|
jones
|
} else if (currentTag.equals("owner")) {
|
| 288 |
|
|
ownerList.add(inputString);
|
| 289 |
|
|
} else if (currentTag.equals("site")) {
|
| 290 |
|
|
siteList.add(inputString);
|
| 291 |
158
|
jones
|
}
|
| 292 |
|
|
}
|
| 293 |
|
|
|
| 294 |
172
|
jones
|
|
| 295 |
|
|
/**
|
| 296 |
|
|
* create a SQL serialization of the query that this instance represents
|
| 297 |
|
|
*/
|
| 298 |
706
|
bojilova
|
public String printSQL(boolean useXMLIndex) {
|
| 299 |
170
|
jones
|
StringBuffer self = new StringBuffer();
|
| 300 |
|
|
|
| 301 |
692
|
bojilova
|
self.append("SELECT docid,docname,doctype,");
|
| 302 |
626
|
berkley
|
self.append("date_created, date_updated, rev ");
|
| 303 |
172
|
jones
|
self.append("FROM xml_documents WHERE docid IN (");
|
| 304 |
170
|
jones
|
|
| 305 |
178
|
jones
|
// This determines the documents that meet the query conditions
|
| 306 |
706
|
bojilova
|
self.append(query.printSQL(useXMLIndex));
|
| 307 |
172
|
jones
|
|
| 308 |
|
|
self.append(") ");
|
| 309 |
402
|
berkley
|
|
| 310 |
172
|
jones
|
// Add SQL to filter for doctypes requested in the query
|
| 311 |
535
|
jones
|
// This is an implicit OR for the list of doctypes
|
| 312 |
172
|
jones
|
if (!doctypeList.isEmpty()) {
|
| 313 |
|
|
boolean firstdoctype = true;
|
| 314 |
|
|
self.append(" AND (");
|
| 315 |
|
|
Enumeration en = doctypeList.elements();
|
| 316 |
|
|
while (en.hasMoreElements()) {
|
| 317 |
|
|
String currentDoctype = (String)en.nextElement();
|
| 318 |
|
|
if (firstdoctype) {
|
| 319 |
402
|
berkley
|
firstdoctype = false;
|
| 320 |
|
|
self.append(" doctype = '" + currentDoctype + "'");
|
| 321 |
172
|
jones
|
} else {
|
| 322 |
|
|
self.append(" OR doctype = '" + currentDoctype + "'");
|
| 323 |
|
|
}
|
| 324 |
|
|
}
|
| 325 |
|
|
self.append(") ");
|
| 326 |
|
|
}
|
| 327 |
402
|
berkley
|
|
| 328 |
535
|
jones
|
// Add SQL to filter for owners requested in the query
|
| 329 |
|
|
// This is an implicit OR for the list of owners
|
| 330 |
|
|
if (!ownerList.isEmpty()) {
|
| 331 |
|
|
boolean first = true;
|
| 332 |
|
|
self.append(" AND (");
|
| 333 |
|
|
Enumeration en = ownerList.elements();
|
| 334 |
|
|
while (en.hasMoreElements()) {
|
| 335 |
|
|
String current = (String)en.nextElement();
|
| 336 |
|
|
if (first) {
|
| 337 |
|
|
first = false;
|
| 338 |
|
|
self.append(" user_owner = '" + current + "'");
|
| 339 |
|
|
} else {
|
| 340 |
|
|
self.append(" OR user_owner = '" + current + "'");
|
| 341 |
|
|
}
|
| 342 |
|
|
}
|
| 343 |
|
|
self.append(") ");
|
| 344 |
|
|
}
|
| 345 |
|
|
|
| 346 |
|
|
// Add SQL to filter for sites requested in the query
|
| 347 |
|
|
// This is an implicit OR for the list of sites
|
| 348 |
|
|
if (!siteList.isEmpty()) {
|
| 349 |
|
|
boolean first = true;
|
| 350 |
|
|
self.append(" AND (");
|
| 351 |
|
|
Enumeration en = siteList.elements();
|
| 352 |
|
|
while (en.hasMoreElements()) {
|
| 353 |
|
|
String current = (String)en.nextElement();
|
| 354 |
|
|
if (first) {
|
| 355 |
|
|
first = false;
|
| 356 |
|
|
self.append(" SUBSTR(docid, 1, INSTR(docid, '" +
|
| 357 |
|
|
accNumberSeparator + "')-1) = '" + current + "'");
|
| 358 |
|
|
} else {
|
| 359 |
|
|
self.append(" OR SUBSTR(docid, 1, INSTR(docid, '" +
|
| 360 |
|
|
accNumberSeparator + "')-1) = '" + current + "'");
|
| 361 |
|
|
}
|
| 362 |
|
|
}
|
| 363 |
|
|
self.append(") ");
|
| 364 |
|
|
}
|
| 365 |
710
|
berkley
|
//System.out.println(self.toString());
|
| 366 |
170
|
jones
|
return self.toString();
|
| 367 |
|
|
}
|
| 368 |
402
|
berkley
|
|
| 369 |
|
|
/**
|
| 370 |
|
|
* This method prints sql based upon the <returnfield> tag in the
|
| 371 |
|
|
* pathquery document. This allows for customization of the
|
| 372 |
|
|
* returned fields
|
| 373 |
465
|
berkley
|
* @param doclist the list of document ids to search by
|
| 374 |
402
|
berkley
|
*/
|
| 375 |
465
|
berkley
|
public String printExtendedSQL(String doclist)
|
| 376 |
402
|
berkley
|
{
|
| 377 |
|
|
StringBuffer self = new StringBuffer();
|
| 378 |
423
|
berkley
|
self.append("select xml_nodes.docid, xml_index.path, xml_nodes.nodedata ");
|
| 379 |
|
|
self.append("from xml_index, xml_nodes where xml_index.nodeid=");
|
| 380 |
|
|
self.append("xml_nodes.parentnodeid and (xml_index.path like '");
|
| 381 |
402
|
berkley
|
boolean firstfield = true;
|
| 382 |
|
|
//put the returnfields into the query
|
| 383 |
|
|
//the for loop allows for multiple fields
|
| 384 |
|
|
for(int i=0; i<returnFieldList.size(); i++)
|
| 385 |
|
|
{
|
| 386 |
|
|
if(firstfield)
|
| 387 |
|
|
{
|
| 388 |
|
|
firstfield = false;
|
| 389 |
405
|
berkley
|
self.append((String)returnFieldList.elementAt(i));
|
| 390 |
402
|
berkley
|
self.append("' ");
|
| 391 |
|
|
}
|
| 392 |
|
|
else
|
| 393 |
|
|
{
|
| 394 |
423
|
berkley
|
self.append("or xml_index.path like '");
|
| 395 |
405
|
berkley
|
self.append((String)returnFieldList.elementAt(i));
|
| 396 |
402
|
berkley
|
self.append("' ");
|
| 397 |
|
|
}
|
| 398 |
|
|
}
|
| 399 |
423
|
berkley
|
self.append(") AND xml_nodes.docid in (");
|
| 400 |
465
|
berkley
|
//self.append(query.printSQL());
|
| 401 |
|
|
self.append(doclist);
|
| 402 |
402
|
berkley
|
self.append(")");
|
| 403 |
423
|
berkley
|
self.append(" AND xml_nodes.nodetype = 'TEXT'");
|
| 404 |
405
|
berkley
|
|
| 405 |
423
|
berkley
|
//System.out.println(self.toString());
|
| 406 |
402
|
berkley
|
return self.toString();
|
| 407 |
|
|
}
|
| 408 |
465
|
berkley
|
|
| 409 |
|
|
public static String printRelationSQL(String docid)
|
| 410 |
|
|
{
|
| 411 |
|
|
StringBuffer self = new StringBuffer();
|
| 412 |
489
|
berkley
|
self.append("select subject, relationship, object, subdoctype, ");
|
| 413 |
|
|
self.append("objdoctype from xml_relation ");
|
| 414 |
465
|
berkley
|
self.append("where subject like '").append(docid).append("'");
|
| 415 |
|
|
return self.toString();
|
| 416 |
|
|
}
|
| 417 |
453
|
berkley
|
|
| 418 |
172
|
jones
|
/**
|
| 419 |
465
|
berkley
|
* Prints sql that returns all relations in the database.
|
| 420 |
453
|
berkley
|
*/
|
| 421 |
|
|
public static String printPackageSQL()
|
| 422 |
|
|
{
|
| 423 |
|
|
StringBuffer self = new StringBuffer();
|
| 424 |
|
|
self.append("select z.nodedata, x.nodedata, y.nodedata from ");
|
| 425 |
|
|
self.append("(select nodeid, parentnodeid from xml_index where path like ");
|
| 426 |
|
|
self.append("'package/relation/subject') s, (select nodeid, parentnodeid ");
|
| 427 |
|
|
self.append("from xml_index where path like ");
|
| 428 |
|
|
self.append("'package/relation/relationship') rel, ");
|
| 429 |
|
|
self.append("(select nodeid, parentnodeid from xml_index where path like ");
|
| 430 |
|
|
self.append("'package/relation/object') o, ");
|
| 431 |
|
|
self.append("xml_nodes x, xml_nodes y, xml_nodes z ");
|
| 432 |
|
|
self.append("where s.parentnodeid = rel.parentnodeid ");
|
| 433 |
|
|
self.append("and rel.parentnodeid = o.parentnodeid ");
|
| 434 |
|
|
self.append("and x.parentnodeid in rel.nodeid ");
|
| 435 |
|
|
self.append("and y.parentnodeid in o.nodeid ");
|
| 436 |
|
|
self.append("and z.parentnodeid in s.nodeid ");
|
| 437 |
|
|
//self.append("and z.nodedata like '%");
|
| 438 |
|
|
//self.append(docid);
|
| 439 |
|
|
//self.append("%'");
|
| 440 |
|
|
return self.toString();
|
| 441 |
|
|
}
|
| 442 |
|
|
|
| 443 |
|
|
/**
|
| 444 |
465
|
berkley
|
* Prints sql that returns all relations in the database that were input
|
| 445 |
|
|
* under a specific docid
|
| 446 |
|
|
* @param docid the docid to search for.
|
| 447 |
|
|
*/
|
| 448 |
|
|
public static String printPackageSQL(String docid)
|
| 449 |
|
|
{
|
| 450 |
|
|
StringBuffer self = new StringBuffer();
|
| 451 |
|
|
self.append("select z.nodedata, x.nodedata, y.nodedata from ");
|
| 452 |
|
|
self.append("(select nodeid, parentnodeid from xml_index where path like ");
|
| 453 |
|
|
self.append("'package/relation/subject') s, (select nodeid, parentnodeid ");
|
| 454 |
|
|
self.append("from xml_index where path like ");
|
| 455 |
|
|
self.append("'package/relation/relationship') rel, ");
|
| 456 |
|
|
self.append("(select nodeid, parentnodeid from xml_index where path like ");
|
| 457 |
|
|
self.append("'package/relation/object') o, ");
|
| 458 |
|
|
self.append("xml_nodes x, xml_nodes y, xml_nodes z ");
|
| 459 |
|
|
self.append("where s.parentnodeid = rel.parentnodeid ");
|
| 460 |
|
|
self.append("and rel.parentnodeid = o.parentnodeid ");
|
| 461 |
|
|
self.append("and x.parentnodeid in rel.nodeid ");
|
| 462 |
|
|
self.append("and y.parentnodeid in o.nodeid ");
|
| 463 |
|
|
self.append("and z.parentnodeid in s.nodeid ");
|
| 464 |
|
|
self.append("and z.docid like '").append(docid).append("'");
|
| 465 |
|
|
|
| 466 |
|
|
return self.toString();
|
| 467 |
|
|
}
|
| 468 |
|
|
|
| 469 |
|
|
/**
|
| 470 |
|
|
* Returns all of the relations that has a certain docid in the subject
|
| 471 |
|
|
* or the object.
|
| 472 |
|
|
*
|
| 473 |
|
|
* @param docid the docid to search for
|
| 474 |
|
|
*/
|
| 475 |
|
|
public static String printPackageSQL(String subDocidURL, String objDocidURL)
|
| 476 |
|
|
{
|
| 477 |
|
|
StringBuffer self = new StringBuffer();
|
| 478 |
|
|
self.append("select z.nodedata, x.nodedata, y.nodedata from ");
|
| 479 |
|
|
self.append("(select nodeid, parentnodeid from xml_index where path like ");
|
| 480 |
|
|
self.append("'package/relation/subject') s, (select nodeid, parentnodeid ");
|
| 481 |
|
|
self.append("from xml_index where path like ");
|
| 482 |
|
|
self.append("'package/relation/relationship') rel, ");
|
| 483 |
|
|
self.append("(select nodeid, parentnodeid from xml_index where path like ");
|
| 484 |
|
|
self.append("'package/relation/object') o, ");
|
| 485 |
|
|
self.append("xml_nodes x, xml_nodes y, xml_nodes z ");
|
| 486 |
|
|
self.append("where s.parentnodeid = rel.parentnodeid ");
|
| 487 |
|
|
self.append("and rel.parentnodeid = o.parentnodeid ");
|
| 488 |
|
|
self.append("and x.parentnodeid in rel.nodeid ");
|
| 489 |
|
|
self.append("and y.parentnodeid in o.nodeid ");
|
| 490 |
|
|
self.append("and z.parentnodeid in s.nodeid ");
|
| 491 |
|
|
self.append("and (z.nodedata like '");
|
| 492 |
|
|
self.append(subDocidURL);
|
| 493 |
|
|
self.append("' or y.nodedata like '");
|
| 494 |
|
|
self.append(objDocidURL);
|
| 495 |
|
|
self.append("')");
|
| 496 |
|
|
return self.toString();
|
| 497 |
|
|
}
|
| 498 |
|
|
|
| 499 |
|
|
public static String printGetDocByDoctypeSQL(String docid)
|
| 500 |
|
|
{
|
| 501 |
|
|
StringBuffer self = new StringBuffer();
|
| 502 |
|
|
|
| 503 |
692
|
bojilova
|
self.append("SELECT docid,docname,doctype,");
|
| 504 |
465
|
berkley
|
self.append("date_created, date_updated ");
|
| 505 |
|
|
self.append("FROM xml_documents WHERE docid IN (");
|
| 506 |
|
|
self.append(docid).append(")");
|
| 507 |
|
|
return self.toString();
|
| 508 |
|
|
}
|
| 509 |
|
|
|
| 510 |
|
|
/**
|
| 511 |
172
|
jones
|
* create a String description of the query that this instance represents.
|
| 512 |
|
|
* This should become a way to get the XML serialization of the query.
|
| 513 |
|
|
*/
|
| 514 |
159
|
jones
|
public String toString() {
|
| 515 |
692
|
bojilova
|
return "meta_file_id=" + meta_file_id + "\n" + query;
|
| 516 |
|
|
//DOCTITLE attr cleared from the db
|
| 517 |
|
|
// return "meta_file_id=" + meta_file_id + "\n" +
|
| 518 |
|
|
// "querytitle=" + querytitle + "\n" + query;
|
| 519 |
159
|
jones
|
}
|
| 520 |
|
|
|
| 521 |
158
|
jones
|
/** a utility class that represents a group of terms in a query */
|
| 522 |
|
|
private class QueryGroup {
|
| 523 |
178
|
jones
|
private String operator = null; // indicates how query terms are combined
|
| 524 |
|
|
private Vector children = null; // the list of query terms and groups
|
| 525 |
158
|
jones
|
|
| 526 |
172
|
jones
|
/**
|
| 527 |
|
|
* construct a new QueryGroup
|
| 528 |
|
|
*
|
| 529 |
178
|
jones
|
* @param operator the boolean conector used to connect query terms
|
| 530 |
172
|
jones
|
* in this query group
|
| 531 |
|
|
*/
|
| 532 |
178
|
jones
|
public QueryGroup(String operator) {
|
| 533 |
|
|
this.operator = operator;
|
| 534 |
158
|
jones
|
children = new Vector();
|
| 535 |
|
|
}
|
| 536 |
|
|
|
| 537 |
172
|
jones
|
/**
|
| 538 |
|
|
* Add a child QueryGroup to this QueryGroup
|
| 539 |
|
|
*
|
| 540 |
|
|
* @param qgroup the query group to be added to the list of terms
|
| 541 |
|
|
*/
|
| 542 |
158
|
jones
|
public void addChild(QueryGroup qgroup) {
|
| 543 |
|
|
children.add((Object)qgroup);
|
| 544 |
|
|
}
|
| 545 |
|
|
|
| 546 |
172
|
jones
|
/**
|
| 547 |
|
|
* Add a child QueryTerm to this QueryGroup
|
| 548 |
|
|
*
|
| 549 |
|
|
* @param qterm the query term to be added to the list of terms
|
| 550 |
|
|
*/
|
| 551 |
158
|
jones
|
public void addChild(QueryTerm qterm) {
|
| 552 |
|
|
children.add((Object)qterm);
|
| 553 |
|
|
}
|
| 554 |
|
|
|
| 555 |
172
|
jones
|
/**
|
| 556 |
|
|
* Retrieve an Enumeration of query terms for this QueryGroup
|
| 557 |
|
|
*/
|
| 558 |
158
|
jones
|
public Enumeration getChildren() {
|
| 559 |
|
|
return children.elements();
|
| 560 |
|
|
}
|
| 561 |
159
|
jones
|
|
| 562 |
172
|
jones
|
/**
|
| 563 |
|
|
* create a SQL serialization of the query that this instance represents
|
| 564 |
|
|
*/
|
| 565 |
706
|
bojilova
|
public String printSQL(boolean useXMLIndex) {
|
| 566 |
170
|
jones
|
StringBuffer self = new StringBuffer();
|
| 567 |
|
|
boolean first = true;
|
| 568 |
|
|
|
| 569 |
|
|
self.append("(");
|
| 570 |
|
|
|
| 571 |
|
|
Enumeration en= getChildren();
|
| 572 |
|
|
while (en.hasMoreElements()) {
|
| 573 |
|
|
Object qobject = en.nextElement();
|
| 574 |
|
|
if (first) {
|
| 575 |
|
|
first = false;
|
| 576 |
|
|
} else {
|
| 577 |
178
|
jones
|
self.append(" " + operator + " ");
|
| 578 |
170
|
jones
|
}
|
| 579 |
|
|
if (qobject instanceof QueryGroup) {
|
| 580 |
|
|
QueryGroup qg = (QueryGroup)qobject;
|
| 581 |
706
|
bojilova
|
self.append(qg.printSQL(useXMLIndex));
|
| 582 |
170
|
jones
|
} else if (qobject instanceof QueryTerm) {
|
| 583 |
|
|
QueryTerm qt = (QueryTerm)qobject;
|
| 584 |
706
|
bojilova
|
self.append(qt.printSQL(useXMLIndex));
|
| 585 |
170
|
jones
|
} else {
|
| 586 |
|
|
System.err.println("qobject wrong type: fatal error");
|
| 587 |
|
|
}
|
| 588 |
|
|
}
|
| 589 |
|
|
self.append(") \n");
|
| 590 |
|
|
return self.toString();
|
| 591 |
|
|
}
|
| 592 |
|
|
|
| 593 |
172
|
jones
|
/**
|
| 594 |
|
|
* create a String description of the query that this instance represents.
|
| 595 |
|
|
* This should become a way to get the XML serialization of the query.
|
| 596 |
|
|
*/
|
| 597 |
159
|
jones
|
public String toString() {
|
| 598 |
|
|
StringBuffer self = new StringBuffer();
|
| 599 |
|
|
|
| 600 |
178
|
jones
|
self.append(" (Query group operator=" + operator + "\n");
|
| 601 |
159
|
jones
|
Enumeration en= getChildren();
|
| 602 |
|
|
while (en.hasMoreElements()) {
|
| 603 |
|
|
Object qobject = en.nextElement();
|
| 604 |
|
|
self.append(qobject);
|
| 605 |
|
|
}
|
| 606 |
|
|
self.append(" )\n");
|
| 607 |
|
|
return self.toString();
|
| 608 |
|
|
}
|
| 609 |
158
|
jones
|
}
|
| 610 |
|
|
|
| 611 |
|
|
/** a utility class that represents a single term in a query */
|
| 612 |
|
|
private class QueryTerm {
|
| 613 |
|
|
private boolean casesensitive = false;
|
| 614 |
|
|
private String searchmode = null;
|
| 615 |
|
|
private String value = null;
|
| 616 |
|
|
private String pathexpr = null;
|
| 617 |
|
|
|
| 618 |
172
|
jones
|
/**
|
| 619 |
|
|
* Construct a new instance of a query term for a free text search
|
| 620 |
|
|
* (using the value only)
|
| 621 |
|
|
*
|
| 622 |
|
|
* @param casesensitive flag indicating whether case is used to match
|
| 623 |
|
|
* @param searchmode determines what kind of substring match is performed
|
| 624 |
|
|
* (one of starts-with|ends-with|contains|matches-exactly)
|
| 625 |
|
|
* @param value the text value to match
|
| 626 |
|
|
*/
|
| 627 |
158
|
jones
|
public QueryTerm(boolean casesensitive, String searchmode,
|
| 628 |
|
|
String value) {
|
| 629 |
|
|
this.casesensitive = casesensitive;
|
| 630 |
|
|
this.searchmode = searchmode;
|
| 631 |
|
|
this.value = value;
|
| 632 |
|
|
}
|
| 633 |
|
|
|
| 634 |
172
|
jones
|
/**
|
| 635 |
|
|
* Construct a new instance of a query term for a structured search
|
| 636 |
|
|
* (matching the value only for those nodes in the pathexpr)
|
| 637 |
|
|
*
|
| 638 |
|
|
* @param casesensitive flag indicating whether case is used to match
|
| 639 |
|
|
* @param searchmode determines what kind of substring match is performed
|
| 640 |
|
|
* (one of starts-with|ends-with|contains|matches-exactly)
|
| 641 |
|
|
* @param value the text value to match
|
| 642 |
|
|
* @param pathexpr the hierarchical path to the nodes to be searched
|
| 643 |
|
|
*/
|
| 644 |
158
|
jones
|
public QueryTerm(boolean casesensitive, String searchmode,
|
| 645 |
|
|
String value, String pathexpr) {
|
| 646 |
|
|
this(casesensitive, searchmode, value);
|
| 647 |
|
|
this.pathexpr = pathexpr;
|
| 648 |
|
|
}
|
| 649 |
|
|
|
| 650 |
172
|
jones
|
/** determine if the QueryTerm is case sensitive */
|
| 651 |
158
|
jones
|
public boolean isCaseSensitive() {
|
| 652 |
|
|
return casesensitive;
|
| 653 |
|
|
}
|
| 654 |
|
|
|
| 655 |
172
|
jones
|
/** get the searchmode parameter */
|
| 656 |
158
|
jones
|
public String getSearchMode() {
|
| 657 |
|
|
return searchmode;
|
| 658 |
|
|
}
|
| 659 |
|
|
|
| 660 |
172
|
jones
|
/** get the Value parameter */
|
| 661 |
158
|
jones
|
public String getValue() {
|
| 662 |
|
|
return value;
|
| 663 |
|
|
}
|
| 664 |
|
|
|
| 665 |
172
|
jones
|
/** get the path expression parameter */
|
| 666 |
158
|
jones
|
public String getPathExpression() {
|
| 667 |
|
|
return pathexpr;
|
| 668 |
|
|
}
|
| 669 |
159
|
jones
|
|
| 670 |
172
|
jones
|
/**
|
| 671 |
|
|
* create a SQL serialization of the query that this instance represents
|
| 672 |
|
|
*/
|
| 673 |
706
|
bojilova
|
public String printSQL(boolean useXMLIndex) {
|
| 674 |
170
|
jones
|
StringBuffer self = new StringBuffer();
|
| 675 |
|
|
|
| 676 |
|
|
// Uppercase the search string if case match is not important
|
| 677 |
|
|
String casevalue = null;
|
| 678 |
|
|
String nodedataterm = null;
|
| 679 |
|
|
|
| 680 |
|
|
if (casesensitive) {
|
| 681 |
|
|
nodedataterm = "nodedata";
|
| 682 |
|
|
casevalue = value;
|
| 683 |
|
|
} else {
|
| 684 |
|
|
nodedataterm = "UPPER(nodedata)";
|
| 685 |
|
|
casevalue = value.toUpperCase();
|
| 686 |
|
|
}
|
| 687 |
|
|
|
| 688 |
|
|
// Add appropriate wildcards to search string
|
| 689 |
|
|
String searchvalue = null;
|
| 690 |
|
|
if (searchmode.equals("starts-with")) {
|
| 691 |
|
|
searchvalue = casevalue + "%";
|
| 692 |
|
|
} else if (searchmode.equals("ends-with")) {
|
| 693 |
|
|
searchvalue = "%" + casevalue;
|
| 694 |
|
|
} else if (searchmode.equals("contains")) {
|
| 695 |
|
|
searchvalue = "%" + casevalue + "%";
|
| 696 |
|
|
} else {
|
| 697 |
|
|
searchvalue = casevalue;
|
| 698 |
|
|
}
|
| 699 |
|
|
|
| 700 |
178
|
jones
|
self.append("SELECT DISTINCT docid FROM xml_nodes WHERE \n");
|
| 701 |
170
|
jones
|
|
| 702 |
|
|
if (pathexpr != null) {
|
| 703 |
|
|
self.append(nodedataterm + " LIKE " + "'" + searchvalue + "' ");
|
| 704 |
|
|
self.append("AND parentnodeid IN ");
|
| 705 |
706
|
bojilova
|
// use XML Index
|
| 706 |
|
|
if ( useXMLIndex ) {
|
| 707 |
|
|
self.append("(SELECT nodeid FROM xml_index WHERE path LIKE " +
|
| 708 |
|
|
"'" + pathexpr + "') " );
|
| 709 |
|
|
// without using XML Index; using nested statements instead
|
| 710 |
|
|
} else {
|
| 711 |
|
|
self.append(useNestedStatements(pathexpr));
|
| 712 |
|
|
}
|
| 713 |
170
|
jones
|
} else {
|
| 714 |
|
|
self.append(nodedataterm + " LIKE " + "'" + searchvalue + "' ");
|
| 715 |
|
|
}
|
| 716 |
|
|
|
| 717 |
|
|
return self.toString();
|
| 718 |
|
|
}
|
| 719 |
|
|
|
| 720 |
706
|
bojilova
|
/*
|
| 721 |
|
|
* Constraint the query with @pathexp without using the XML Index,
|
| 722 |
|
|
* but nested SQL statements instead. The query migth be slower.
|
| 723 |
|
|
*/
|
| 724 |
|
|
private String useNestedStatements(String pathexpr)
|
| 725 |
|
|
{
|
| 726 |
|
|
StringBuffer nestedStmts = new StringBuffer();
|
| 727 |
|
|
Vector nodes = new Vector();
|
| 728 |
|
|
String path = pathexpr;
|
| 729 |
|
|
int inx = 0;
|
| 730 |
|
|
|
| 731 |
|
|
do {
|
| 732 |
|
|
inx = path.lastIndexOf("/");
|
| 733 |
|
|
//System.out.println(path.substring(inx+1));
|
| 734 |
|
|
nodes.addElement(path.substring(inx+1));
|
| 735 |
|
|
path = path.substring(0, Math.abs(inx));
|
| 736 |
|
|
} while ( inx > 0 );
|
| 737 |
|
|
|
| 738 |
|
|
// nested statements
|
| 739 |
|
|
int i = 0;
|
| 740 |
|
|
for (i = 0; i < nodes.size()-1; i++) {
|
| 741 |
|
|
nestedStmts.append("(SELECT nodeid FROM xml_nodes" +
|
| 742 |
|
|
" WHERE nodename LIKE '" +
|
| 743 |
|
|
(String)nodes.elementAt(i) + "'" +
|
| 744 |
|
|
" AND parentnodeid IN ");
|
| 745 |
|
|
}
|
| 746 |
|
|
// for the last statement: it is without " AND parentnodeid IN "
|
| 747 |
|
|
nestedStmts.append("(SELECT nodeid FROM xml_nodes" +
|
| 748 |
|
|
" WHERE nodename LIKE '" +
|
| 749 |
|
|
(String)nodes.elementAt(i) + "'" );
|
| 750 |
|
|
// node.size() number of closing brackets
|
| 751 |
|
|
for (i = 0; i < nodes.size(); i++) {
|
| 752 |
|
|
nestedStmts.append(")");
|
| 753 |
|
|
}
|
| 754 |
|
|
|
| 755 |
|
|
|
| 756 |
710
|
berkley
|
//System.out.println(nestedStmts.toString());
|
| 757 |
706
|
bojilova
|
return nestedStmts.toString();
|
| 758 |
|
|
}
|
| 759 |
|
|
|
| 760 |
172
|
jones
|
/**
|
| 761 |
|
|
* create a String description of the query that this instance represents.
|
| 762 |
|
|
* This should become a way to get the XML serialization of the query.
|
| 763 |
|
|
*/
|
| 764 |
709
|
bojilova
|
public String toString() {
|
| 765 |
159
|
jones
|
|
| 766 |
709
|
bojilova
|
return this.printSQL(true);
|
| 767 |
159
|
jones
|
}
|
| 768 |
158
|
jones
|
}
|
| 769 |
155
|
jones
|
}
|