1 | package org.greenstone.gsdl3.gs3build.metadata;
|
---|
2 |
|
---|
3 | import java.util.List;
|
---|
4 | import java.util.ArrayList;
|
---|
5 | import java.util.Iterator;
|
---|
6 |
|
---|
7 | import java.io.PrintWriter;
|
---|
8 |
|
---|
9 | import java.sql.SQLException;
|
---|
10 | import java.sql.Statement;
|
---|
11 | import java.sql.ResultSet;
|
---|
12 |
|
---|
13 | import org.w3c.dom.Document;
|
---|
14 | import org.w3c.dom.Element;
|
---|
15 | import org.w3c.dom.NamedNodeMap;
|
---|
16 | import org.w3c.dom.Node;
|
---|
17 | import org.w3c.dom.NodeList;
|
---|
18 | import org.w3c.dom.Text;
|
---|
19 |
|
---|
20 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentInterface;
|
---|
21 |
|
---|
22 | import org.greenstone.gsdl3.gs3build.util.XMLTools;
|
---|
23 | import org.greenstone.gsdl3.gs3build.database.*;
|
---|
24 |
|
---|
25 | /**
|
---|
26 | * Support the descriptive metadata of a given document
|
---|
27 | *
|
---|
28 | * The descriptive metadata can occur multiple times, so this 'wrapper' is
|
---|
29 | * in fact a holder for a number of individual metadata sets. Individual
|
---|
30 | * metadata groups are found in METSNamespace items.
|
---|
31 | *
|
---|
32 | *
|
---|
33 | */
|
---|
34 |
|
---|
35 | import org.greenstone.gsdl3.gs3build.util.MultiMap;
|
---|
36 |
|
---|
37 | public class METSDescriptive
|
---|
38 | {
|
---|
39 | MultiMap map;
|
---|
40 | String ID;
|
---|
41 | String name;
|
---|
42 |
|
---|
43 | public METSDescriptive(String ID, String name)
|
---|
44 | {
|
---|
45 | this.map = new MultiMap();
|
---|
46 | this.ID = ID;
|
---|
47 | this.name = name;
|
---|
48 | }
|
---|
49 |
|
---|
50 | public String getID()
|
---|
51 | {
|
---|
52 | return this.ID;
|
---|
53 | }
|
---|
54 |
|
---|
55 | public String getName()
|
---|
56 | {
|
---|
57 | return this.name;
|
---|
58 | }
|
---|
59 |
|
---|
60 | public void addNamespace(METSNamespace namespace)
|
---|
61 | {
|
---|
62 | this.map.put(namespace.getName(), namespace);
|
---|
63 | }
|
---|
64 |
|
---|
65 | public METSNamespace getNamespace(String namespace)
|
---|
66 | {
|
---|
67 | return (METSNamespace) this.map.get(namespace);
|
---|
68 | }
|
---|
69 |
|
---|
70 | /**
|
---|
71 | * Get the version of a metadata namespace that can be used for
|
---|
72 | * adding new values to the document. Remember, each namespace may
|
---|
73 | * occur more than once, but only one of the occurrences can be 'open'.
|
---|
74 | *
|
---|
75 | * @param <code>String</code> the name of the namespace
|
---|
76 | */
|
---|
77 | public METSNamespace getOpenNamespace(String namespace)
|
---|
78 | {
|
---|
79 | if (this.map.getAll(namespace) == null) {
|
---|
80 | return null;
|
---|
81 | }
|
---|
82 |
|
---|
83 | Iterator namespaces = ((List) this.map.getAll(namespace)).iterator();
|
---|
84 |
|
---|
85 | while (namespaces.hasNext()) {
|
---|
86 | METSNamespace namespaceData = (METSNamespace) namespaces.next();
|
---|
87 |
|
---|
88 | if (namespaceData.isEditable()){
|
---|
89 | return namespaceData;
|
---|
90 | }
|
---|
91 | }
|
---|
92 | return null;
|
---|
93 | }
|
---|
94 |
|
---|
95 | /**
|
---|
96 | * Add a piece of metadata to this document. Existing values are preserved.
|
---|
97 | *
|
---|
98 | * @param <code>String</code> the namespace in which the metadata occurs
|
---|
99 | * @param <code>String</code> the label of the metadata
|
---|
100 | * @param <code>String</code> the value of the metadata
|
---|
101 | */
|
---|
102 | public void addMetadata(String namespace, String label, String value)
|
---|
103 | {
|
---|
104 | METSNamespace namespaceData = this.getOpenNamespace(namespace);
|
---|
105 |
|
---|
106 | if (namespaceData == null ||
|
---|
107 | namespaceData.isEditable() == false) {
|
---|
108 | namespaceData = NamespaceFactory.initNamespace(namespace);
|
---|
109 | this.addNamespace(namespaceData);
|
---|
110 | }
|
---|
111 | namespaceData.addMetadata(label, value);
|
---|
112 | }
|
---|
113 |
|
---|
114 | /**
|
---|
115 | * Add a piece of metadata to this document. Existing values in editable
|
---|
116 | * parts of the document's metadata are destroyed.
|
---|
117 | *
|
---|
118 | * @param <code>String</code> the namespace in which the metadata occurs
|
---|
119 | * @param <code>String</code> the label of the metadata
|
---|
120 | * @param <code>String</code> the value of the metadata
|
---|
121 | */
|
---|
122 | public void setMetadata(String namespace, String label, String value)
|
---|
123 | {
|
---|
124 | METSNamespace namespaceData = this.getOpenNamespace(namespace);
|
---|
125 |
|
---|
126 | if (namespaceData == null ||
|
---|
127 | namespaceData.isEditable() == false) {
|
---|
128 | namespaceData = NamespaceFactory.initNamespace(namespace);
|
---|
129 | this.addNamespace(namespaceData);
|
---|
130 | }
|
---|
131 | namespaceData.setMetadata(label, value);
|
---|
132 | }
|
---|
133 |
|
---|
134 | /**
|
---|
135 | * Remove the metadata for a particular label.
|
---|
136 | *
|
---|
137 | * @param <code>String</code> the namespace in which the metadata occurs
|
---|
138 | * @param <code>String</code> the label of the metadata
|
---|
139 | */
|
---|
140 | public void removeMetadata(String namespace, String label)
|
---|
141 | {
|
---|
142 | METSNamespace namespaceData = this.getOpenNamespace(namespace);
|
---|
143 |
|
---|
144 | if (namespaceData != null && namespaceData.isEditable() == true) {
|
---|
145 | namespaceData.removeMetadata(label);
|
---|
146 | }
|
---|
147 | }
|
---|
148 |
|
---|
149 | /**
|
---|
150 | * Fetch a <code>List</code> of the values matching the given label in
|
---|
151 | * the requisite namespace.
|
---|
152 | *
|
---|
153 | * @param <code>String</code> namespace
|
---|
154 | * @param <code>String</code> label of metadata to match
|
---|
155 | *
|
---|
156 | * @return <code>List</code> the corresponding values found.
|
---|
157 | */
|
---|
158 | public List getMetadata(String namespace, String label)
|
---|
159 | {
|
---|
160 | // Simple case - no metadata
|
---|
161 | if (!this.map.containsKey(namespace)){
|
---|
162 | return null;
|
---|
163 | }
|
---|
164 |
|
---|
165 | // if there's just one instance of the requisite namespace,
|
---|
166 | // then again it is a simple case
|
---|
167 | if (this.map.getAll(namespace).size() == 1){
|
---|
168 | return this.getNamespace(namespace).getMetadata(label);
|
---|
169 | }
|
---|
170 |
|
---|
171 | // otherwise, step through and accumulate
|
---|
172 | List accumulatedValues = new ArrayList();
|
---|
173 | Iterator namespaceIterator = this.map.getAll(namespace).iterator();
|
---|
174 | while (namespaceIterator.hasNext()){
|
---|
175 | METSNamespace localNamespace = (METSNamespace) namespaceIterator.next();
|
---|
176 |
|
---|
177 | List localList = localNamespace.getMetadata(label);
|
---|
178 | if (localList != null && localList.size() > 0){
|
---|
179 | accumulatedValues.addAll(localList);
|
---|
180 | }
|
---|
181 | }
|
---|
182 |
|
---|
183 | // if the result of the accumulation is empty, return a null item
|
---|
184 | if (accumulatedValues.size() == 0){
|
---|
185 | return null;
|
---|
186 | }
|
---|
187 |
|
---|
188 | return accumulatedValues;
|
---|
189 | }
|
---|
190 |
|
---|
191 | protected void parse_xmlData(METSNamespace namespace, Element element)
|
---|
192 | {
|
---|
193 | if (namespace==null) {
|
---|
194 | System.err.println("namespace is null!!!!");
|
---|
195 | }
|
---|
196 | NodeList children = element.getChildNodes();
|
---|
197 |
|
---|
198 | for (int c = 0; c < children.getLength(); c++) {
|
---|
199 | if (children.item(c).getNodeType() != org.w3c.dom.Node.ELEMENT_NODE) {
|
---|
200 | continue;
|
---|
201 | }
|
---|
202 |
|
---|
203 | Element child = (Element) children.item(c);
|
---|
204 | String childName = child.getNodeName();
|
---|
205 |
|
---|
206 | // if (childName.equals("gsdl2:Metadata")) {
|
---|
207 | if (childName.equals("gsdl3:Metadata")) {
|
---|
208 | String metaName = child.getAttribute("name");
|
---|
209 | Text text = (Text) child.getFirstChild();
|
---|
210 | String metaValue = text.getData();
|
---|
211 |
|
---|
212 | addMetadata(namespace.getName(),metaName, metaValue);
|
---|
213 | }
|
---|
214 | }
|
---|
215 | }
|
---|
216 |
|
---|
217 | public void parse_mdWrap(Element element)
|
---|
218 | {
|
---|
219 | // deal with mets:mdWrap
|
---|
220 |
|
---|
221 | METSNamespace namespace = NamespaceFactory.parseXML(element);
|
---|
222 |
|
---|
223 | NodeList children = element.getChildNodes();
|
---|
224 |
|
---|
225 | for (int c = 0; c < children.getLength(); c ++) {
|
---|
226 | if (children.item(c).getNodeType() != org.w3c.dom.Node.ELEMENT_NODE) {
|
---|
227 | continue;
|
---|
228 | }
|
---|
229 |
|
---|
230 | Element child = (Element) children.item(c);
|
---|
231 | String childName = child.getNodeName();
|
---|
232 |
|
---|
233 | if (childName.equals("mets:xmlData")) {
|
---|
234 | parse_xmlData(namespace, child);
|
---|
235 | } else {
|
---|
236 | System.err.println("Warning: unrecognised tag " + childName
|
---|
237 | + "in mets:mdWrap");
|
---|
238 | }
|
---|
239 | }
|
---|
240 | }
|
---|
241 |
|
---|
242 | /**
|
---|
243 | * Parse an XML Element as a METS Descriptive Metadata section
|
---|
244 | *
|
---|
245 | * @param <code>Element</code> the XML element which represents the dmdSec itself
|
---|
246 | */
|
---|
247 | public static METSDescriptive parseXML(Element element)
|
---|
248 | {
|
---|
249 | // Note: no parsing of attributes required, we just move onto
|
---|
250 | // the metadata namespaces/sections themselves
|
---|
251 |
|
---|
252 | // deal with mets:dmdSec
|
---|
253 | String ID = element.getAttribute("ID");
|
---|
254 | String label = element.getAttribute("GROUPID");
|
---|
255 |
|
---|
256 | METSDescriptive thisDescriptive = new METSDescriptive(ID, label);
|
---|
257 |
|
---|
258 | //thisDescriptive = new METSDescriptive(ID, label);
|
---|
259 |
|
---|
260 | NodeList children = element.getChildNodes();
|
---|
261 |
|
---|
262 | for (int c = 0; c < children.getLength(); c ++) {
|
---|
263 | if (children.item(c).getNodeType() != org.w3c.dom.Node.ELEMENT_NODE) {
|
---|
264 | continue;
|
---|
265 | }
|
---|
266 |
|
---|
267 | Element child = (Element) children.item(c);
|
---|
268 | String childName = child.getNodeName();
|
---|
269 |
|
---|
270 | if (childName.equals("mets:mdRef")) {
|
---|
271 | //for External Descriptive Metadata
|
---|
272 | METSNamespace namespace = NamespaceFactory.parseXML(element);
|
---|
273 | System.err.println("*** mets::mdRef does not appear to be implemented yet");
|
---|
274 | } else if (childName.equals("mets:mdWrap")) {
|
---|
275 | thisDescriptive.parse_mdWrap(child);
|
---|
276 | } else {
|
---|
277 | // TODO: raise an error!
|
---|
278 | System.out.println("Error: METSDescriptive: unrecognised tag "+childName);
|
---|
279 | }
|
---|
280 | }
|
---|
281 | return thisDescriptive;
|
---|
282 | }
|
---|
283 | /**
|
---|
284 | * Write the document metadata to a <code>PrintWriter</code> in METS
|
---|
285 | * XML format.
|
---|
286 | *
|
---|
287 | * @param <code>PrintWriter</code> the writer to use for output
|
---|
288 | */
|
---|
289 | public void write(PrintWriter output)
|
---|
290 | {
|
---|
291 | String tag = XMLTools.getOpenTag("mets", "dmdSec");
|
---|
292 | tag = XMLTools.addAttribute(tag, "ID", this.ID);
|
---|
293 | tag = XMLTools.addAttribute(tag, "GROUPID", this.name);
|
---|
294 | output.println(tag);
|
---|
295 |
|
---|
296 | Iterator keys = this.map.keySet().iterator();
|
---|
297 |
|
---|
298 | // get the keys one by one
|
---|
299 | while (keys.hasNext()){
|
---|
300 | Object nextKey = keys.next();
|
---|
301 | if (nextKey == null){
|
---|
302 | continue;
|
---|
303 | }
|
---|
304 | String key = nextKey.toString();
|
---|
305 |
|
---|
306 | // get every copy of the current namespace name - namespaces may
|
---|
307 | // occur more than once, so this is a List
|
---|
308 | Iterator namespaces = this.map.getAll(key).iterator();
|
---|
309 |
|
---|
310 | // namespaces will write themselves...
|
---|
311 | while (namespaces.hasNext()){
|
---|
312 | METSNamespace namespace = (METSNamespace) namespaces.next();
|
---|
313 |
|
---|
314 | namespace.write(output);
|
---|
315 | }
|
---|
316 | }
|
---|
317 |
|
---|
318 | tag = XMLTools.getCloseTag("mets", "dmdSec");
|
---|
319 | output.println(tag);
|
---|
320 | }
|
---|
321 |
|
---|
322 | public void copy_metadata(METSDescriptive dst)
|
---|
323 | {
|
---|
324 | Iterator keys = this.map.keySet().iterator();
|
---|
325 |
|
---|
326 | // get the keys one by one
|
---|
327 | while (keys.hasNext()) {
|
---|
328 | Object nextKey = keys.next();
|
---|
329 | if (nextKey == null) {
|
---|
330 | continue;
|
---|
331 | }
|
---|
332 | String key = nextKey.toString();
|
---|
333 |
|
---|
334 | // get every copy of the current namespace name - namespaces may
|
---|
335 | // occur more than once, so this is a List
|
---|
336 | Iterator namespaces = this.map.getAll(key).iterator();
|
---|
337 |
|
---|
338 | while (namespaces.hasNext()) {
|
---|
339 | METSNamespace namespace = (METSNamespace) namespaces.next();
|
---|
340 |
|
---|
341 | // for each metadata name
|
---|
342 |
|
---|
343 | Iterator mdnIter = namespace.getMetadataNames();
|
---|
344 | while (mdnIter.hasNext()) {
|
---|
345 |
|
---|
346 | String md_name = mdnIter.next().toString();
|
---|
347 |
|
---|
348 | List values = getMetadata(key, md_name);
|
---|
349 |
|
---|
350 | if (values != null) {
|
---|
351 | Iterator valueIter = values.iterator();
|
---|
352 | while (valueIter.hasNext()) {
|
---|
353 | String value = valueIter.next().toString();
|
---|
354 | dst.addMetadata("gsdl3",md_name,value);
|
---|
355 | }
|
---|
356 | }
|
---|
357 | }
|
---|
358 | }
|
---|
359 | }
|
---|
360 | }
|
---|
361 |
|
---|
362 |
|
---|
363 | public boolean writeSQL(DocumentInterface document, GS3SQLConnection connection)
|
---|
364 | {
|
---|
365 | // check if this node is in the
|
---|
366 | int sqlId = -1;
|
---|
367 |
|
---|
368 | // System.out.println("Writing " + connection.toString());
|
---|
369 |
|
---|
370 | GS3SQLSelect select = new GS3SQLSelect("metadata");
|
---|
371 | select.addField("*");
|
---|
372 | GS3SQLWhereItem whereItem = new GS3SQLWhereItem("MetaID", "=", this.ID);
|
---|
373 | GS3SQLWhereItem item = new GS3SQLWhereItem("DocID", "=", document.getID().toString());
|
---|
374 | GS3SQLWhere where = new GS3SQLWhere(whereItem);
|
---|
375 | where.add(item);
|
---|
376 | select.setWhere(where);
|
---|
377 |
|
---|
378 | try {
|
---|
379 | Statement statement = connection.createStatement();
|
---|
380 | ResultSet resultSet = statement.executeQuery(select.toString());
|
---|
381 | GS3SQLAction action;
|
---|
382 |
|
---|
383 | if (resultSet.first()) {
|
---|
384 | // the node already exists - no need to do anything as such
|
---|
385 | GS3SQLUpdate update = new GS3SQLUpdate("metadata");
|
---|
386 | update.setWhere(where);
|
---|
387 | action = update;
|
---|
388 | }
|
---|
389 | else { // Set result set to null, just in case first() didn't work above...
|
---|
390 | // It is a new node and needs writing
|
---|
391 | action = new GS3SQLInsert("metadata");
|
---|
392 |
|
---|
393 | action.addValue("DocID", document.getID().toString());
|
---|
394 | action.addValue("MetaID", this.ID);
|
---|
395 | }
|
---|
396 |
|
---|
397 | // always set the group identifier
|
---|
398 | action.addValue("GroupID", this.name);
|
---|
399 |
|
---|
400 | // execute the action as required
|
---|
401 | statement.execute(action.toString());
|
---|
402 |
|
---|
403 | // get the resultSet again
|
---|
404 | // now execute the select statement again to get the new identifier
|
---|
405 | // 'MetadataRef'
|
---|
406 | // System.out.println(select.toString());
|
---|
407 | resultSet = statement.executeQuery(select.toString());
|
---|
408 |
|
---|
409 | if (resultSet.first()) {
|
---|
410 | // get the reference for this item...
|
---|
411 | sqlId = resultSet.getInt("MetadataRef");
|
---|
412 | }
|
---|
413 | statement.close();
|
---|
414 | }
|
---|
415 | catch (SQLException sql){
|
---|
416 | System.err.println("METSDescriptive.writeSQL(): "+sql);
|
---|
417 | }
|
---|
418 | if (sqlId == -1) {
|
---|
419 | return false;
|
---|
420 | }
|
---|
421 |
|
---|
422 | Iterator keys = this.map.keySet().iterator();
|
---|
423 |
|
---|
424 | // get the keys one by one
|
---|
425 | while (keys.hasNext()){
|
---|
426 | Object nextKey = keys.next();
|
---|
427 | if (nextKey == null){
|
---|
428 | continue;
|
---|
429 | }
|
---|
430 | String key = nextKey.toString();
|
---|
431 |
|
---|
432 | // get every current namespace - namespaces may
|
---|
433 | // occur more than once, so this is a List
|
---|
434 | Iterator namespaces = this.map.getAll(key).iterator();
|
---|
435 |
|
---|
436 | // namespaces will write themselves...
|
---|
437 | while (namespaces.hasNext()){
|
---|
438 | METSNamespace namespace = (METSNamespace) namespaces.next();
|
---|
439 |
|
---|
440 | if (!namespace.writeSQL(sqlId, connection)){
|
---|
441 | return false;
|
---|
442 | }
|
---|
443 | }
|
---|
444 | }
|
---|
445 | return true;
|
---|
446 | }
|
---|
447 |
|
---|
448 | public static METSDescriptive readSQL(DocumentInterface document, GS3SQLConnection connection,
|
---|
449 | ResultSet resultSet)
|
---|
450 | {
|
---|
451 | try {
|
---|
452 | String ID = resultSet.getString("MetaID");
|
---|
453 | String name = resultSet.getString("GroupID");
|
---|
454 |
|
---|
455 | // create the metadata block object
|
---|
456 | METSDescriptive descriptive = new METSDescriptive(ID, name);
|
---|
457 |
|
---|
458 | // get its metadata reference to retrieve namespaces
|
---|
459 | int metadataRef = resultSet.getInt("MetadataRef");
|
---|
460 |
|
---|
461 | // query the database for matching namespaces for this metadata block
|
---|
462 | GS3SQLSelect select = new GS3SQLSelect("namespaces");
|
---|
463 | select.addField("*");
|
---|
464 | GS3SQLWhereItem whereItem = new GS3SQLWhereItem("MetadataRef", "=", Integer.toString(metadataRef),
|
---|
465 | GS3SQLField.INTEGER_TYPE);
|
---|
466 | GS3SQLWhere where = new GS3SQLWhere(whereItem);
|
---|
467 | select.setWhere(where);
|
---|
468 |
|
---|
469 | Statement statement = connection.createStatement();
|
---|
470 | ResultSet namespaceSet = statement.executeQuery(select.toString());
|
---|
471 |
|
---|
472 | // parse through the namespaces, calling the namespace class to create instances as it will
|
---|
473 | if (namespaceSet.first()) {
|
---|
474 | do {
|
---|
475 | METSNamespace namespace = METSNamespace.readSQL(connection, namespaceSet);
|
---|
476 | if (namespace != null) {
|
---|
477 | descriptive.addNamespace(namespace);
|
---|
478 | }
|
---|
479 | else {
|
---|
480 | System.err.println("Null namespace output");
|
---|
481 | }
|
---|
482 | }
|
---|
483 | while (namespaceSet.next());
|
---|
484 | }
|
---|
485 | statement.close();
|
---|
486 | return descriptive;
|
---|
487 | }
|
---|
488 | catch (SQLException sqlEx){
|
---|
489 | System.err.println("METSDescriptive.readSQL(): "+sqlEx);
|
---|
490 | System.exit(1);
|
---|
491 | }
|
---|
492 | return null;
|
---|
493 | }
|
---|
494 | }
|
---|