1
|
/**
|
2
|
* '$RCSfile$'
|
3
|
* Purpose: A Class that handles the SAX XML events as they
|
4
|
* are generated from XML documents
|
5
|
* Copyright: 2000 Regents of the University of California and the
|
6
|
* National Center for Ecological Analysis and Synthesis
|
7
|
* Authors: Matt Jones, Jivka Bojilova
|
8
|
* Release: @release@
|
9
|
*
|
10
|
* '$Author: sgarg $'
|
11
|
* '$Date: 2005-11-02 10:31:32 -0800 (Wed, 02 Nov 2005) $'
|
12
|
* '$Revision: 2715 $'
|
13
|
*
|
14
|
* This program is free software; you can redistribute it and/or modify
|
15
|
* it under the terms of the GNU General Public License as published by
|
16
|
* the Free Software Foundation; either version 2 of the License, or
|
17
|
* (at your option) any later version.
|
18
|
*
|
19
|
* This program is distributed in the hope that it will be useful,
|
20
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
21
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
22
|
* GNU General Public License for more details.
|
23
|
*
|
24
|
* You should have received a copy of the GNU General Public License
|
25
|
* along with this program; if not, write to the Free Software
|
26
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
27
|
*/
|
28
|
|
29
|
package edu.ucsb.nceas.metacat;
|
30
|
|
31
|
import java.sql.PreparedStatement;
|
32
|
import java.sql.ResultSet;
|
33
|
import java.sql.SQLException;
|
34
|
import java.sql.Statement;
|
35
|
import java.util.EmptyStackException;
|
36
|
import java.util.Enumeration;
|
37
|
import java.util.Hashtable;
|
38
|
import java.util.Stack;
|
39
|
import java.util.Vector;
|
40
|
|
41
|
import edu.ucsb.nceas.morpho.datapackage.Triple;
|
42
|
import edu.ucsb.nceas.morpho.datapackage.TripleCollection;
|
43
|
|
44
|
import org.apache.log4j.Logger;
|
45
|
import org.xml.sax.Attributes;
|
46
|
import org.xml.sax.SAXException;
|
47
|
import org.xml.sax.SAXParseException;
|
48
|
import org.xml.sax.ext.DeclHandler;
|
49
|
import org.xml.sax.ext.LexicalHandler;
|
50
|
import org.xml.sax.helpers.DefaultHandler;
|
51
|
|
52
|
/**
|
53
|
* A database aware Class implementing callback bethods for the SAX parser to
|
54
|
* call when processing the XML stream and generating events.
|
55
|
*/
|
56
|
public class DBSAXHandler extends DefaultHandler implements LexicalHandler,
|
57
|
DeclHandler, Runnable
|
58
|
{
|
59
|
|
60
|
protected boolean atFirstElement;
|
61
|
|
62
|
protected boolean processingDTD;
|
63
|
|
64
|
protected String docname = null;
|
65
|
|
66
|
protected String doctype;
|
67
|
|
68
|
protected String catalogid = null;
|
69
|
|
70
|
protected String systemid;
|
71
|
|
72
|
private boolean stackCreated = false;
|
73
|
|
74
|
protected Stack nodeStack;
|
75
|
|
76
|
protected Vector nodeIndex;
|
77
|
|
78
|
protected DBConnection connection = null;
|
79
|
|
80
|
protected DocumentImpl currentDocument;
|
81
|
|
82
|
protected String createDate = null;
|
83
|
|
84
|
protected String updateDate = null;
|
85
|
|
86
|
protected DBSAXNode rootNode;
|
87
|
|
88
|
protected String action = null;
|
89
|
|
90
|
protected String docid = null;
|
91
|
|
92
|
protected String revision = null;
|
93
|
|
94
|
protected String user = null;
|
95
|
|
96
|
protected String[] groups = null;
|
97
|
|
98
|
protected String pub = null;
|
99
|
|
100
|
protected Thread xmlIndex;
|
101
|
|
102
|
private boolean endDocument = false;
|
103
|
|
104
|
protected int serverCode = 1;
|
105
|
|
106
|
protected Hashtable namespaces = new Hashtable();
|
107
|
|
108
|
protected boolean hitTextNode = false; // a flag to hit text node
|
109
|
|
110
|
// a buffer to keep all text nodes for same element
|
111
|
// it is for element was splited
|
112
|
protected StringBuffer textBuffer = new StringBuffer();
|
113
|
|
114
|
protected Stack textBufferStack = new Stack();
|
115
|
|
116
|
protected static final int MAXDATACHARS = 4000;
|
117
|
|
118
|
//protected static final int MAXDATACHARS = 50;
|
119
|
protected static final long INDEXDELAY = 10000;
|
120
|
|
121
|
// methods writeChildNodeToDB, setAttribute, setNamespace,
|
122
|
// writeTextForDBSAXNode will increase endNodeId.
|
123
|
protected long endNodeId = -1; // The end node id for a substree
|
124
|
// DOCTITLE attr cleared from the db
|
125
|
// private static final int MAXTITLELEN = 1000;
|
126
|
|
127
|
private boolean isRevisionDoc = false;
|
128
|
|
129
|
//HandlerTriple stuff
|
130
|
TripleCollection tripleList = new TripleCollection();
|
131
|
|
132
|
Triple currentTriple = new Triple();
|
133
|
|
134
|
boolean startParseTriple = false;
|
135
|
|
136
|
boolean hasTriple = false;
|
137
|
|
138
|
public static final String ECOGRID = "ecogrid://";
|
139
|
|
140
|
private Logger logMetacat = Logger.getLogger(DBSAXHandler.class);
|
141
|
|
142
|
/**
|
143
|
* Construct an instance of the handler class
|
144
|
*
|
145
|
* @param conn the JDBC connection to which information is written
|
146
|
*/
|
147
|
private DBSAXHandler(DBConnection conn, String createDate, String updateDate)
|
148
|
{
|
149
|
this.connection = conn;
|
150
|
this.atFirstElement = true;
|
151
|
this.processingDTD = false;
|
152
|
this.createDate = createDate;
|
153
|
this.updateDate = updateDate;
|
154
|
|
155
|
// Create the stack for keeping track of node context
|
156
|
// if it doesn't already exist
|
157
|
if (!stackCreated) {
|
158
|
nodeStack = new Stack();
|
159
|
nodeIndex = new Vector();
|
160
|
stackCreated = true;
|
161
|
}
|
162
|
}
|
163
|
|
164
|
/**
|
165
|
* Construct an instance of the handler class
|
166
|
*
|
167
|
* @param conn the JDBC connection to which information is written
|
168
|
* @param action - "INSERT" or "UPDATE"
|
169
|
* @param docid to be inserted or updated into JDBC connection
|
170
|
* @param user the user connected to MetaCat servlet and owns the document
|
171
|
* @param groups the groups to which user belongs
|
172
|
* @param pub flag for public "read" access on document
|
173
|
* @param serverCode the serverid from xml_replication on which this
|
174
|
* document resides.
|
175
|
*
|
176
|
*/
|
177
|
/* TODO excise this constructor because not used anywhere in project
|
178
|
public DBSAXHandler(DBConnection conn, String action, String docid,
|
179
|
String user, String[] groups, String pub, int serverCode)
|
180
|
{
|
181
|
this(conn);
|
182
|
this.action = action;
|
183
|
this.docid = docid;
|
184
|
this.user = user;
|
185
|
this.groups = groups;
|
186
|
this.pub = pub;
|
187
|
this.serverCode = serverCode;
|
188
|
this.xmlIndex = new Thread(this);
|
189
|
}
|
190
|
*/
|
191
|
/**
|
192
|
* Construct an instance of the handler class In this constructor, user can
|
193
|
* specify the version need to upadate
|
194
|
*
|
195
|
* @param conn the JDBC connection to which information is written
|
196
|
* @param action - "INSERT" or "UPDATE"
|
197
|
* @param docid to be inserted or updated into JDBC connection
|
198
|
* @param revision, the user specified the revision need to be update
|
199
|
* @param user the user connected to MetaCat servlet and owns the document
|
200
|
* @param groups the groups to which user belongs
|
201
|
* @param pub flag for public "read" access on document
|
202
|
* @param serverCode the serverid from xml_replication on which this
|
203
|
* document resides.
|
204
|
*
|
205
|
*/
|
206
|
public DBSAXHandler(DBConnection conn, String action, String docid,
|
207
|
String revision, String user, String[] groups, String pub,
|
208
|
int serverCode, String createDate, String updateDate)
|
209
|
{
|
210
|
this(conn, createDate, updateDate);
|
211
|
this.action = action;
|
212
|
this.docid = docid;
|
213
|
this.revision = revision;
|
214
|
this.user = user;
|
215
|
this.groups = groups;
|
216
|
this.pub = pub;
|
217
|
this.serverCode = serverCode;
|
218
|
this.xmlIndex = new Thread(this);
|
219
|
}
|
220
|
|
221
|
/** SAX Handler that receives notification of beginning of the document */
|
222
|
public void startDocument() throws SAXException
|
223
|
{
|
224
|
logMetacat.info("start Document");
|
225
|
|
226
|
// Create the document node representation as root
|
227
|
rootNode = new DBSAXNode(connection, this.docid);
|
228
|
// Add the node to the stack, so that any text data can be
|
229
|
// added as it is encountered
|
230
|
nodeStack.push(rootNode);
|
231
|
}
|
232
|
|
233
|
/** SAX Handler that receives notification of end of the document */
|
234
|
public void endDocument() throws SAXException
|
235
|
{
|
236
|
logMetacat.info("end Document");
|
237
|
// Starting new thread for writing XML Index.
|
238
|
// It calls the run method of the thread.
|
239
|
|
240
|
//if it is data package insert triple into relationtion table;
|
241
|
if (doctype != null
|
242
|
&& MetaCatUtil.getOptionList(
|
243
|
MetaCatUtil.getOption("packagedoctype")).contains(
|
244
|
doctype) && hasTriple && !isRevisionDoc) {
|
245
|
try {
|
246
|
//initial handler and write into relationdb only for xml-documents
|
247
|
if (!isRevisionDoc)
|
248
|
{
|
249
|
RelationHandler handler = new RelationHandler(docid, doctype,
|
250
|
connection, tripleList);
|
251
|
}
|
252
|
} catch (Exception e) {
|
253
|
logMetacat.error(
|
254
|
"Failed to write triples into relation table"
|
255
|
+ e.getMessage());
|
256
|
throw new SAXException(
|
257
|
"Failed to write triples into relation table "
|
258
|
+ e.getMessage());
|
259
|
}
|
260
|
}
|
261
|
}
|
262
|
|
263
|
/** SAX Handler that is called at the start of Namespace */
|
264
|
public void startPrefixMapping(String prefix, String uri)
|
265
|
throws SAXException
|
266
|
{
|
267
|
logMetacat.info("NAMESPACE");
|
268
|
|
269
|
namespaces.put(prefix, uri);
|
270
|
}
|
271
|
|
272
|
/** SAX Handler that is called at the start of each XML element */
|
273
|
public void startElement(String uri, String localName, String qName,
|
274
|
Attributes atts) throws SAXException
|
275
|
{
|
276
|
// for element <eml:eml...> qname is "eml:eml", local name is "eml"
|
277
|
// for element <acl....> both qname and local name is "eml"
|
278
|
// uri is namesapce
|
279
|
logMetacat.info("Start ELEMENT(qName) " + qName);
|
280
|
logMetacat.info("Start ELEMENT(localName) " + localName);
|
281
|
logMetacat.info("Start ELEMENT(uri) " + uri);
|
282
|
|
283
|
DBSAXNode parentNode = null;
|
284
|
DBSAXNode currentNode = null;
|
285
|
|
286
|
// Get a reference to the parent node for the id
|
287
|
try {
|
288
|
|
289
|
parentNode = (DBSAXNode) nodeStack.peek();
|
290
|
} catch (EmptyStackException e) {
|
291
|
parentNode = null;
|
292
|
}
|
293
|
|
294
|
// If hit a text node, we need write this text for current's parent
|
295
|
// node
|
296
|
// This will happend if the element is mixted
|
297
|
if (hitTextNode && parentNode != null) {
|
298
|
// write the textbuffer into db for parent node.
|
299
|
endNodeId = writeTextForDBSAXNode(endNodeId, textBuffer, parentNode);
|
300
|
// rest hitTextNode
|
301
|
hitTextNode = false;
|
302
|
// reset textbuffer
|
303
|
textBuffer = null;
|
304
|
textBuffer = new StringBuffer();
|
305
|
|
306
|
}
|
307
|
|
308
|
// Document representation that points to the root document node
|
309
|
if (atFirstElement) {
|
310
|
atFirstElement = false;
|
311
|
// If no DOCTYPE declaration: docname = root element
|
312
|
// doctype = root element name or name space
|
313
|
if (docname == null) {
|
314
|
docname = localName;
|
315
|
// if uri isn't null doctype = uri(namespace)
|
316
|
// othewise root element
|
317
|
if (uri != null && !(uri.trim()).equals("")) {
|
318
|
doctype = uri;
|
319
|
} else {
|
320
|
doctype = docname;
|
321
|
}
|
322
|
logMetacat.info("DOCNAME-a: " + docname);
|
323
|
logMetacat.info("DOCTYPE-a: " + doctype);
|
324
|
} else if (doctype == null) {
|
325
|
// because docname is not null and it is declared in dtd
|
326
|
// so could not be in schema, no namespace
|
327
|
doctype = docname;
|
328
|
logMetacat.info("DOCTYPE-b: " + doctype);
|
329
|
}
|
330
|
|
331
|
rootNode.writeNodename(docname);
|
332
|
|
333
|
try {
|
334
|
// for validated XML Documents store a reference to XML DB
|
335
|
// Catalog
|
336
|
// Because this is select statement and it needn't to roll back
|
337
|
// if
|
338
|
// insert document action fialed.
|
339
|
// In order to decrease DBConnection usage count, we get a new
|
340
|
// dbconnection from pool
|
341
|
|
342
|
DBConnection dbConn = null;
|
343
|
int serialNumber = -1;
|
344
|
|
345
|
if (systemid != null) {
|
346
|
try {
|
347
|
// Get dbconnection
|
348
|
dbConn = DBConnectionPool
|
349
|
.getDBConnection("DBSAXHandler.startElement");
|
350
|
serialNumber = dbConn.getCheckOutSerialNumber();
|
351
|
|
352
|
Statement stmt = dbConn.createStatement();
|
353
|
ResultSet rs = stmt
|
354
|
.executeQuery("SELECT catalog_id FROM xml_catalog "
|
355
|
+ "WHERE entry_type = 'DTD' "
|
356
|
+ "AND public_id = '" + doctype + "'");
|
357
|
boolean hasRow = rs.next();
|
358
|
if (hasRow) {
|
359
|
catalogid = rs.getString(1);
|
360
|
}
|
361
|
stmt.close();
|
362
|
}//try
|
363
|
finally {
|
364
|
// Return dbconnection
|
365
|
DBConnectionPool.returnDBConnection(dbConn,
|
366
|
serialNumber);
|
367
|
}//finally
|
368
|
}
|
369
|
|
370
|
//create documentImpl object by the constructor which can
|
371
|
// specify
|
372
|
//the revision
|
373
|
|
374
|
if (!isRevisionDoc)
|
375
|
{
|
376
|
//System.out.println("here!!!!!!!!!!!3");
|
377
|
currentDocument = new DocumentImpl(connection, rootNode
|
378
|
.getNodeID(), docname, doctype, docid, revision,
|
379
|
action, user, this.pub, catalogid, this.serverCode,
|
380
|
createDate, updateDate);
|
381
|
//System.out.println("here!!!!!!!!!!!4");
|
382
|
}
|
383
|
|
384
|
|
385
|
} catch (Exception ane) {
|
386
|
ane.printStackTrace();
|
387
|
throw (new SAXException("Error in DBSaxHandler.startElement "
|
388
|
+ action, ane));
|
389
|
}
|
390
|
}
|
391
|
|
392
|
// Create the current node representation
|
393
|
currentNode = new DBSAXNode(connection, qName, localName,
|
394
|
parentNode, rootNode.getNodeID(), docid, doctype);
|
395
|
|
396
|
// Add all of the namespaces
|
397
|
String prefix;
|
398
|
String nsuri;
|
399
|
Enumeration prefixes = namespaces.keys();
|
400
|
while (prefixes.hasMoreElements()) {
|
401
|
prefix = (String) prefixes.nextElement();
|
402
|
nsuri = (String) namespaces.get(prefix);
|
403
|
currentNode.setNamespace(prefix, nsuri, docid);
|
404
|
}
|
405
|
namespaces = null;
|
406
|
namespaces = new Hashtable();
|
407
|
|
408
|
// Add all of the attributes
|
409
|
for (int i = 0; i < atts.getLength(); i++) {
|
410
|
String attributeName = atts.getQName(i);
|
411
|
String attributeValue = atts.getValue(i);
|
412
|
endNodeId = currentNode.setAttribute(attributeName, attributeValue,
|
413
|
docid);
|
414
|
|
415
|
// To handle name space and schema location if the attribute name
|
416
|
// is
|
417
|
// xsi:schemaLocation. If the name space is in not in catalog table
|
418
|
// it will be regeistered.
|
419
|
if (attributeName != null
|
420
|
&& attributeName
|
421
|
.indexOf(MetaCatServlet.SCHEMALOCATIONKEYWORD) != -1) {
|
422
|
SchemaLocationResolver resolver = new SchemaLocationResolver(
|
423
|
attributeValue);
|
424
|
resolver.resolveNameSpace();
|
425
|
|
426
|
}
|
427
|
}
|
428
|
|
429
|
// Add the node to the stack, so that any text data can be
|
430
|
// added as it is encountered
|
431
|
nodeStack.push(currentNode);
|
432
|
// Add the node to the vector used by thread for writing XML Index
|
433
|
nodeIndex.addElement(currentNode);
|
434
|
// start parsing triple
|
435
|
if (doctype != null
|
436
|
&& MetaCatUtil.getOptionList(
|
437
|
MetaCatUtil.getOption("packagedoctype")).contains(
|
438
|
doctype) && localName.equals("triple")) {
|
439
|
startParseTriple = true;
|
440
|
hasTriple = true;
|
441
|
currentTriple = new Triple();
|
442
|
}
|
443
|
}
|
444
|
|
445
|
public void runIndexingThread(){
|
446
|
|
447
|
boolean useXMLIndex =
|
448
|
(new Boolean(MetaCatUtil.getOption("usexmlindex"))).booleanValue();
|
449
|
if (useXMLIndex && !isRevisionDoc) {
|
450
|
try {
|
451
|
xmlIndex.start();
|
452
|
} catch (NullPointerException e) {
|
453
|
xmlIndex = null;
|
454
|
logMetacat.error("Error in DBSAXHandler.runIndexingThread() "
|
455
|
+ e.getMessage());
|
456
|
}
|
457
|
}
|
458
|
}
|
459
|
|
460
|
/*
|
461
|
* Run a separate thread to build the XML index for this document. This
|
462
|
* thread is run asynchronously in order to more quickly return control to
|
463
|
* the submitting user. The run method checks to see if the document has
|
464
|
* been fully inserted before trying to update the xml_index table.
|
465
|
*/
|
466
|
public void run()
|
467
|
{
|
468
|
try {
|
469
|
if (!isRevisionDoc)
|
470
|
{
|
471
|
// stop 5 second
|
472
|
Thread.sleep(5000);
|
473
|
//make sure record is done
|
474
|
checkDocumentTable();
|
475
|
// Build the index for this document
|
476
|
currentDocument.buildIndex();
|
477
|
}
|
478
|
} catch (Exception e) {
|
479
|
logMetacat.error("Error in DBSAXHandler.run "
|
480
|
+ e.getMessage());
|
481
|
}
|
482
|
}
|
483
|
|
484
|
/*
|
485
|
* method to make sure insert is finished before create index table If new
|
486
|
* version of record is in xml_documents every thing will be fine
|
487
|
*/
|
488
|
private void checkDocumentTable() throws Exception
|
489
|
{
|
490
|
|
491
|
DBConnection dbConn = null;
|
492
|
int serialNumber = -1;
|
493
|
|
494
|
try {
|
495
|
// Opening separate db connection for writing XML Index
|
496
|
dbConn = DBConnectionPool
|
497
|
.getDBConnection("DBSAXHandler.checkDocumentTable");
|
498
|
serialNumber = dbConn.getCheckOutSerialNumber();
|
499
|
|
500
|
// the following while loop construct checks to make sure that
|
501
|
// the docid of the document that we are trying to index is already
|
502
|
// in the xml_documents table. if this is not the case, the foreign
|
503
|
// key relationship between xml_documents and xml_index is
|
504
|
// temporarily broken causing multiple problems.
|
505
|
boolean inxmldoc = false;
|
506
|
long startTime = System.currentTimeMillis();
|
507
|
while (!inxmldoc) {
|
508
|
String xmlDocumentsCheck = "select distinct docid from xml_documents"
|
509
|
+ " where docid ='"
|
510
|
+ docid
|
511
|
+ "' and "
|
512
|
+ " rev ='"
|
513
|
+ revision + "'";
|
514
|
|
515
|
PreparedStatement xmlDocCheck = dbConn
|
516
|
.prepareStatement(xmlDocumentsCheck);
|
517
|
// Increase usage count
|
518
|
dbConn.increaseUsageCount(1);
|
519
|
xmlDocCheck.execute();
|
520
|
ResultSet doccheckRS = xmlDocCheck.getResultSet();
|
521
|
boolean tableHasRows = doccheckRS.next();
|
522
|
if (tableHasRows) {
|
523
|
logMetacat.info(
|
524
|
"=========== found the correct document");
|
525
|
inxmldoc = true;
|
526
|
}
|
527
|
doccheckRS.close();
|
528
|
xmlDocCheck.close();
|
529
|
// make sure the while loop will be ended in reseaonable time
|
530
|
long stopTime = System.currentTimeMillis();
|
531
|
if ((stopTime - startTime) > INDEXDELAY) {
|
532
|
logMetacat.warn("Couldn't find the docid for index build in "
|
533
|
+ "reseaonable time!");
|
534
|
throw new Exception(
|
535
|
"Couldn't find the docid for index build in "
|
536
|
+ "reseaonable time!");
|
537
|
}
|
538
|
}//while
|
539
|
} catch (Exception e) {
|
540
|
try {
|
541
|
dbConn.rollback();
|
542
|
//dbconn.close();
|
543
|
} catch (SQLException sqle) {
|
544
|
}
|
545
|
logMetacat.error("Error in DBSAXHandler.checkDocumentTable "
|
546
|
+ e.getMessage());
|
547
|
|
548
|
} finally {
|
549
|
DBConnectionPool.returnDBConnection(dbConn, serialNumber);
|
550
|
}//finally
|
551
|
|
552
|
}
|
553
|
|
554
|
/** SAX Handler that is called for each XML text node */
|
555
|
public void characters(char[] cbuf, int start, int len) throws SAXException
|
556
|
{
|
557
|
logMetacat.info("CHARACTERS");
|
558
|
// buffer all text nodes for same element. This is for text was splited
|
559
|
// into different nodes
|
560
|
textBuffer.append(new String(cbuf, start, len));
|
561
|
// set hittextnode true
|
562
|
hitTextNode = true;
|
563
|
// if text buffer .size is greater than max, write it to db.
|
564
|
// so we can save memory
|
565
|
if (textBuffer.length() > MAXDATACHARS) {
|
566
|
logMetacat.info("Write text into DB in charaters"
|
567
|
+ " when text buffer size is greater than maxmum number");
|
568
|
DBSAXNode currentNode = (DBSAXNode) nodeStack.peek();
|
569
|
endNodeId = writeTextForDBSAXNode(endNodeId, textBuffer,
|
570
|
currentNode);
|
571
|
textBuffer = null;
|
572
|
textBuffer = new StringBuffer();
|
573
|
}
|
574
|
}
|
575
|
|
576
|
/**
|
577
|
* SAX Handler that is called for each XML text node that is Ignorable
|
578
|
* white space
|
579
|
*/
|
580
|
public void ignorableWhitespace(char[] cbuf, int start, int len)
|
581
|
throws SAXException
|
582
|
{
|
583
|
// When validation is turned "on", white spaces are reported here
|
584
|
// When validation is turned "off" white spaces are not reported here,
|
585
|
// but through characters() callback
|
586
|
logMetacat.info("IGNORABLEWHITESPACE");
|
587
|
|
588
|
DBSAXNode currentNode = (DBSAXNode) nodeStack.peek();
|
589
|
String data = null;
|
590
|
int leftover = len;
|
591
|
int offset = start;
|
592
|
boolean moredata = true;
|
593
|
|
594
|
// This loop deals with the case where there are more characters
|
595
|
// than can fit in a single database text field (limit is
|
596
|
// MAXDATACHARS). If the text to be inserted exceeds MAXDATACHARS,
|
597
|
// write a series of nodes that are MAXDATACHARS long, and then the
|
598
|
// final node contains the remainder
|
599
|
while (moredata) {
|
600
|
if (leftover > MAXDATACHARS) {
|
601
|
data = new String(cbuf, offset, MAXDATACHARS);
|
602
|
leftover -= MAXDATACHARS;
|
603
|
offset += MAXDATACHARS;
|
604
|
} else {
|
605
|
data = new String(cbuf, offset, leftover);
|
606
|
moredata = false;
|
607
|
}
|
608
|
|
609
|
// Write the content of the node to the database
|
610
|
endNodeId = currentNode.writeChildNodeToDB("TEXT", null, data,
|
611
|
docid);
|
612
|
}
|
613
|
}
|
614
|
|
615
|
/**
|
616
|
* SAX Handler called once for each processing instruction found: node that
|
617
|
* PI may occur before or after the root element.
|
618
|
*/
|
619
|
public void processingInstruction(String target, String data)
|
620
|
throws SAXException
|
621
|
{
|
622
|
logMetacat.info("PI");
|
623
|
DBSAXNode currentNode = (DBSAXNode) nodeStack.peek();
|
624
|
endNodeId = currentNode.writeChildNodeToDB("PI", target, data, docid);
|
625
|
}
|
626
|
|
627
|
/** SAX Handler that is called at the end of each XML element */
|
628
|
public void endElement(String uri, String localName, String qName)
|
629
|
throws SAXException
|
630
|
{
|
631
|
logMetacat.info("End ELEMENT " + qName);
|
632
|
|
633
|
// write buffered text nodes into db (so no splited)
|
634
|
DBSAXNode currentNode = (DBSAXNode) nodeStack.peek();
|
635
|
|
636
|
// If before the end element, the parser hit text nodes and store them
|
637
|
// into the buffer, write the buffer to data base. The reason we put
|
638
|
// write database here is for xerces some time split text node
|
639
|
if (hitTextNode) {
|
640
|
logMetacat.info("Write text into DB in End Element");
|
641
|
endNodeId = writeTextForDBSAXNode(endNodeId, textBuffer,
|
642
|
currentNode);
|
643
|
|
644
|
//if it is triple parsing process
|
645
|
if (startParseTriple) {
|
646
|
|
647
|
String content = textBuffer.toString().trim();
|
648
|
if (localName.equals("subject")) { //get the subject content
|
649
|
currentTriple.setSubject(content);
|
650
|
} else if (localName.equals("relationship")) { //get the
|
651
|
// relationship
|
652
|
// content
|
653
|
currentTriple.setRelationship(content);
|
654
|
} else if (localName.equals("object")) { //get the object
|
655
|
// content
|
656
|
currentTriple.setObject(content);
|
657
|
}
|
658
|
}
|
659
|
|
660
|
}//if
|
661
|
|
662
|
//set hitText false
|
663
|
hitTextNode = false;
|
664
|
// reset textbuff
|
665
|
textBuffer = null;
|
666
|
textBuffer = new StringBuffer();
|
667
|
|
668
|
// Get the node from the stack
|
669
|
currentNode = (DBSAXNode) nodeStack.pop();
|
670
|
//finishing parsing single triple
|
671
|
if (startParseTriple && localName.equals("triple")) {
|
672
|
// add trip to triple collection
|
673
|
tripleList.addTriple(currentTriple);
|
674
|
//rest variable
|
675
|
currentTriple = null;
|
676
|
startParseTriple = false;
|
677
|
}
|
678
|
}
|
679
|
|
680
|
//
|
681
|
// the next section implements the LexicalHandler interface
|
682
|
//
|
683
|
|
684
|
/** SAX Handler that receives notification of DOCTYPE. Sets the DTD */
|
685
|
public void startDTD(String name, String publicId, String systemId)
|
686
|
throws SAXException
|
687
|
{
|
688
|
docname = name;
|
689
|
doctype = publicId;
|
690
|
systemid = systemId;
|
691
|
|
692
|
processingDTD = true;
|
693
|
DBSAXNode currentNode = (DBSAXNode) nodeStack.peek();
|
694
|
//create a DTD node and write docname,publicid and system id into db
|
695
|
// we don't put the dtd node into node stack
|
696
|
DBSAXNode dtdNode = new DBSAXNode(connection, name, publicId, systemId,
|
697
|
currentNode, currentNode.getRootNodeID(), docid);
|
698
|
logMetacat.info("Start DTD");
|
699
|
logMetacat.info("Setting processingDTD to true");
|
700
|
logMetacat.info("DOCNAME: " + docname);
|
701
|
logMetacat.info("DOCTYPE: " + doctype);
|
702
|
logMetacat.info(" SYSID: " + systemid);
|
703
|
}
|
704
|
|
705
|
/**
|
706
|
* SAX Handler that receives notification of end of DTD
|
707
|
*/
|
708
|
public void endDTD() throws SAXException
|
709
|
{
|
710
|
|
711
|
processingDTD = false;
|
712
|
logMetacat.info("Setting processingDTD to false");
|
713
|
logMetacat.info("end DTD");
|
714
|
}
|
715
|
|
716
|
/**
|
717
|
* SAX Handler that receives notification of comments in the DTD
|
718
|
*/
|
719
|
public void comment(char[] ch, int start, int length) throws SAXException
|
720
|
{
|
721
|
logMetacat.info("COMMENT");
|
722
|
if (!processingDTD) {
|
723
|
DBSAXNode currentNode = (DBSAXNode) nodeStack.peek();
|
724
|
endNodeId = currentNode.writeChildNodeToDB("COMMENT", null,
|
725
|
new String(ch, start, length), docid);
|
726
|
}
|
727
|
}
|
728
|
|
729
|
/**
|
730
|
* SAX Handler that receives notification of the start of CDATA sections
|
731
|
*/
|
732
|
public void startCDATA() throws SAXException
|
733
|
{
|
734
|
logMetacat.info("start CDATA");
|
735
|
}
|
736
|
|
737
|
/**
|
738
|
* SAX Handler that receives notification of the end of CDATA sections
|
739
|
*/
|
740
|
public void endCDATA() throws SAXException
|
741
|
{
|
742
|
logMetacat.info("end CDATA");
|
743
|
}
|
744
|
|
745
|
/**
|
746
|
* SAX Handler that receives notification of the start of entities
|
747
|
*/
|
748
|
public void startEntity(String name) throws SAXException
|
749
|
{
|
750
|
logMetacat.info("start ENTITY: " + name);
|
751
|
//System.out.println("start ENTITY: " + name);
|
752
|
if (name.equals("[dtd]")) {
|
753
|
processingDTD = true;
|
754
|
}
|
755
|
}
|
756
|
|
757
|
/**
|
758
|
* SAX Handler that receives notification of the end of entities
|
759
|
*/
|
760
|
public void endEntity(String name) throws SAXException
|
761
|
{
|
762
|
logMetacat.info("end ENTITY: " + name);
|
763
|
//System.out.println("end ENTITY: " + name);
|
764
|
if (name.equals("[dtd]")) {
|
765
|
processingDTD = false;
|
766
|
}
|
767
|
}
|
768
|
|
769
|
/**
|
770
|
* SAX Handler that receives notification of element declarations
|
771
|
*/
|
772
|
public void elementDecl(String name, String model)
|
773
|
throws org.xml.sax.SAXException
|
774
|
{
|
775
|
//System.out.println("ELEMENTDECL: " + name + " " + model);
|
776
|
logMetacat.info("ELEMENTDECL: " + name + " " + model);
|
777
|
}
|
778
|
|
779
|
/**
|
780
|
* SAX Handler that receives notification of attribute declarations
|
781
|
*/
|
782
|
public void attributeDecl(String eName, String aName, String type,
|
783
|
String valueDefault, String value) throws org.xml.sax.SAXException
|
784
|
{
|
785
|
|
786
|
//System.out.println("ATTRIBUTEDECL: " + eName + " "
|
787
|
// + aName + " " + type + " " + valueDefault + " "
|
788
|
// + value);
|
789
|
logMetacat.info("ATTRIBUTEDECL: " + eName + " " + aName + " "
|
790
|
+ type + " " + valueDefault + " " + value);
|
791
|
}
|
792
|
|
793
|
/**
|
794
|
* SAX Handler that receives notification of internal entity declarations
|
795
|
*/
|
796
|
public void internalEntityDecl(String name, String value)
|
797
|
throws org.xml.sax.SAXException
|
798
|
{
|
799
|
//System.out.println("INTERNENTITYDECL: " + name + " " + value);
|
800
|
logMetacat.info("INTERNENTITYDECL: " + name + " " + value);
|
801
|
}
|
802
|
|
803
|
/**
|
804
|
* SAX Handler that receives notification of external entity declarations
|
805
|
*/
|
806
|
public void externalEntityDecl(String name, String publicId, String systemId)
|
807
|
throws org.xml.sax.SAXException
|
808
|
{
|
809
|
//System.out.println("EXTERNENTITYDECL: " + name + " " + publicId
|
810
|
// + " " + systemId);
|
811
|
logMetacat.info("EXTERNENTITYDECL: " + name + " " + publicId
|
812
|
+ " " + systemId);
|
813
|
// it processes other external entity, not the DTD;
|
814
|
// it doesn't signal for the DTD here
|
815
|
processingDTD = false;
|
816
|
}
|
817
|
|
818
|
//
|
819
|
// the next section implements the ErrorHandler interface
|
820
|
//
|
821
|
|
822
|
/**
|
823
|
* SAX Handler that receives notification of fatal parsing errors
|
824
|
*/
|
825
|
public void fatalError(SAXParseException exception) throws SAXException
|
826
|
{
|
827
|
logMetacat.fatal("FATALERROR: " + exception.getMessage());
|
828
|
throw (new SAXException("Fatal processing error.", exception));
|
829
|
}
|
830
|
|
831
|
/**
|
832
|
* SAX Handler that receives notification of recoverable parsing errors
|
833
|
*/
|
834
|
public void error(SAXParseException exception) throws SAXException
|
835
|
{
|
836
|
logMetacat.error("ERROR: " + exception.getMessage());
|
837
|
throw (new SAXException("Error in processing EML.", exception));
|
838
|
}
|
839
|
|
840
|
/**
|
841
|
* SAX Handler that receives notification of warnings
|
842
|
*/
|
843
|
public void warning(SAXParseException exception) throws SAXException
|
844
|
{
|
845
|
logMetacat.warn("WARNING: " + exception.getMessage());
|
846
|
throw (new SAXException("Warning.", exception));
|
847
|
}
|
848
|
|
849
|
//
|
850
|
// Helper, getter and setter methods
|
851
|
//
|
852
|
|
853
|
/**
|
854
|
* get the document name
|
855
|
*/
|
856
|
public String getDocname()
|
857
|
{
|
858
|
return docname;
|
859
|
}
|
860
|
|
861
|
/**
|
862
|
* get the document processing state
|
863
|
*/
|
864
|
public boolean processingDTD()
|
865
|
{
|
866
|
return processingDTD;
|
867
|
}
|
868
|
|
869
|
|
870
|
/**
|
871
|
* get the the is revision doc
|
872
|
* @return
|
873
|
*/
|
874
|
public boolean getIsRevisionDoc()
|
875
|
{
|
876
|
return isRevisionDoc;
|
877
|
}
|
878
|
|
879
|
/**
|
880
|
* Set the the handler is for revisionDoc
|
881
|
* @param isRevisionDoc
|
882
|
*/
|
883
|
public void setIsRevisionDoc(boolean isRevisionDoc)
|
884
|
{
|
885
|
this.isRevisionDoc = isRevisionDoc;
|
886
|
}
|
887
|
|
888
|
/* Method to write a text buffer for DBSAXNode */
|
889
|
protected long writeTextForDBSAXNode(long previousEndNodeId,
|
890
|
StringBuffer strBuffer, DBSAXNode node) throws SAXException
|
891
|
{
|
892
|
long nodeId = previousEndNodeId;
|
893
|
// Check parameter
|
894
|
if (strBuffer == null || node == null) { return nodeId; }
|
895
|
boolean moredata = true;
|
896
|
String data = null;
|
897
|
|
898
|
String normalizedData = strBuffer.toString();
|
899
|
strBuffer = new StringBuffer(MetaCatUtil.normalize(normalizedData));
|
900
|
|
901
|
int bufferSize = strBuffer.length();
|
902
|
int start = 0;
|
903
|
|
904
|
// if there are some cotent in buffer, write it
|
905
|
if (bufferSize > 0) {
|
906
|
logMetacat.info("Write text into DB");
|
907
|
// This loop deals with the case where there are more characters
|
908
|
// than can fit in a single database text field (limit is
|
909
|
// MAXDATACHARS). If the text to be inserted exceeds MAXDATACHARS,
|
910
|
// write a series of nodes that are MAXDATACHARS long, and then the
|
911
|
// final node contains the remainder
|
912
|
while (moredata) {
|
913
|
bufferSize = strBuffer.length();
|
914
|
if (bufferSize > MAXDATACHARS) {
|
915
|
data = strBuffer.substring(start, MAXDATACHARS);
|
916
|
// cut the stringbuffer part that already written into db
|
917
|
strBuffer = strBuffer.delete(start, MAXDATACHARS);
|
918
|
} else {
|
919
|
data = strBuffer.substring(start, bufferSize);
|
920
|
moredata = false;
|
921
|
}
|
922
|
|
923
|
// Write the content of the node to the database
|
924
|
nodeId = node.writeChildNodeToDB("TEXT", null, data, docid);
|
925
|
}//while
|
926
|
}//if
|
927
|
return nodeId;
|
928
|
}
|
929
|
|
930
|
public long getRootNodeId()
|
931
|
{
|
932
|
return rootNode.getNodeID();
|
933
|
}
|
934
|
|
935
|
public String getDocumentType()
|
936
|
{
|
937
|
return doctype;
|
938
|
}
|
939
|
|
940
|
public String getDocumentName()
|
941
|
{
|
942
|
return docname;
|
943
|
}
|
944
|
|
945
|
public String getCatalogId()
|
946
|
{
|
947
|
return catalogid;
|
948
|
}
|
949
|
}
|