1 | package org.greenstone.gsdl3.gs3build.classifier;
|
---|
2 |
|
---|
3 | import java.util.List;
|
---|
4 | import java.util.ArrayList;
|
---|
5 | import java.util.Iterator;
|
---|
6 |
|
---|
7 | import java.sql.ResultSet;
|
---|
8 | import java.sql.SQLException;
|
---|
9 | import java.sql.Statement;
|
---|
10 |
|
---|
11 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentID;
|
---|
12 |
|
---|
13 | import org.greenstone.gsdl3.gs3build.database.*;
|
---|
14 |
|
---|
15 | public abstract class AbstractHierarchyNode
|
---|
16 | {
|
---|
17 | String prefix; // a basic prefix used in this hierarchy
|
---|
18 | String descriptor; // the textual descriptor used on screen or long-hand
|
---|
19 | String name; // the index number, letter assignment or other item
|
---|
20 | // used to identify the position of the item in the
|
---|
21 | // hierarchy
|
---|
22 | String id; // an identifier used by the GLI for maintenance
|
---|
23 | // purposes; this plays no active role in the
|
---|
24 | // rebuilding process (at the moment)
|
---|
25 | List childNodes; // the child classification nodes of this node
|
---|
26 | List childDocs; // the child documents of this node
|
---|
27 | List matches; // the other metadata values that may be matched
|
---|
28 | // against the classifier
|
---|
29 | AbstractHierarchyNode parent; // the parent of the node
|
---|
30 |
|
---|
31 | class AbstractHierarchyDocument
|
---|
32 | { DocumentID documentId;
|
---|
33 | String sortKey;
|
---|
34 |
|
---|
35 | public AbstractHierarchyDocument(DocumentID documentId, String sortKey)
|
---|
36 | { this.documentId = documentId;
|
---|
37 | this.sortKey = sortKey;
|
---|
38 | }
|
---|
39 |
|
---|
40 | public DocumentID getID()
|
---|
41 | { return this.documentId;
|
---|
42 | }
|
---|
43 |
|
---|
44 | public String toString()
|
---|
45 | { return this.sortKey;
|
---|
46 | }
|
---|
47 | }
|
---|
48 |
|
---|
49 | /**
|
---|
50 | * Simple node
|
---|
51 | */
|
---|
52 | public AbstractHierarchyNode()
|
---|
53 | { this.descriptor = "";
|
---|
54 | this.name = "";
|
---|
55 | this.id = "";
|
---|
56 | this.prefix = "";
|
---|
57 | this.childNodes = new ArrayList();
|
---|
58 | this.childDocs = new ArrayList();
|
---|
59 | this.parent = null;
|
---|
60 | this.matches = new ArrayList();
|
---|
61 | }
|
---|
62 |
|
---|
63 | public AbstractHierarchyNode(String prefix, String name, String id, String descriptor)
|
---|
64 | { this.descriptor = descriptor;
|
---|
65 | this.name = name;
|
---|
66 | this.id = id;
|
---|
67 | this.prefix = prefix;
|
---|
68 | this.childNodes = new ArrayList();
|
---|
69 | this.childDocs = new ArrayList();
|
---|
70 | this.parent = null;
|
---|
71 | this.matches = new ArrayList();
|
---|
72 | this.matches.add(this.id);
|
---|
73 | this.matches.add(this.name);
|
---|
74 | }
|
---|
75 |
|
---|
76 | public void addChild(AbstractHierarchyNode child)
|
---|
77 | { this.childNodes.add(child);
|
---|
78 | child.setParent(this);
|
---|
79 |
|
---|
80 | /**
|
---|
81 | if (this.id == null) {
|
---|
82 | System.out.println(child.id.toString() + " added to root");
|
---|
83 | }
|
---|
84 | else {
|
---|
85 | System.out.println(child.id.toString() + " added to " + this.id);
|
---|
86 | }
|
---|
87 | */
|
---|
88 | }
|
---|
89 |
|
---|
90 | public boolean add(AbstractHierarchyNode child)
|
---|
91 | { if (this.id != null && this.id.length() > 0 &&
|
---|
92 | !child.id.startsWith(this.id + "."))
|
---|
93 | { return false;
|
---|
94 | }
|
---|
95 |
|
---|
96 | Iterator subNodes = this.childNodes.iterator();
|
---|
97 | while (subNodes.hasNext()) {
|
---|
98 | AbstractHierarchyNode subNode = (AbstractHierarchyNode) subNodes.next();
|
---|
99 |
|
---|
100 | if (subNode.add(child)) {
|
---|
101 | return true;
|
---|
102 | }
|
---|
103 | }
|
---|
104 |
|
---|
105 | this.addChild(child);
|
---|
106 | return true;
|
---|
107 | }
|
---|
108 |
|
---|
109 | public void addDocument(DocumentID document, String sortKey)
|
---|
110 | { if (sortKey == null && false) {
|
---|
111 | System.err.println("attempt to classify by an unset key");
|
---|
112 | return;
|
---|
113 | }
|
---|
114 | this.childDocs.add(new AbstractHierarchyDocument(document, sortKey));
|
---|
115 |
|
---|
116 | System.out.println("Adding " + document.toString() + " to " + this.name + " " + this.childDocs.size());
|
---|
117 | if (sortKey != null) {
|
---|
118 | int first = 0;
|
---|
119 | int last = this.childDocs.size() - 1;
|
---|
120 |
|
---|
121 | while (first != last) {
|
---|
122 | int at = (first + last) / 2;
|
---|
123 |
|
---|
124 | System.out.println(this.name + " " + at + " " + this.childDocs.size() + " " + this.childDocs.get(at) + " " + sortKey);
|
---|
125 | if (this.childDocs.get(at) == null ||
|
---|
126 | this.childDocs.get(at).toString() == null) {
|
---|
127 | last = at;
|
---|
128 | System.out.println("COntinuing");
|
---|
129 | continue;
|
---|
130 | }
|
---|
131 |
|
---|
132 | if (this.childDocs.get(at).toString().compareTo(sortKey) > 0) {
|
---|
133 | last = at;
|
---|
134 | }
|
---|
135 | else {
|
---|
136 | first = at + 1;
|
---|
137 | }
|
---|
138 | }
|
---|
139 |
|
---|
140 | Object newItem = this.childDocs.get(this.childDocs.size() - 1);
|
---|
141 | last = this.childDocs.size() - 1;
|
---|
142 | while (last > first) {
|
---|
143 | this.childDocs.set(last, this.childDocs.get(last-1));
|
---|
144 | last --;
|
---|
145 | }
|
---|
146 | this.childDocs.set(first, newItem);
|
---|
147 | }
|
---|
148 |
|
---|
149 | // System.out.println("Added " + document.toString() + " to " + this.name + " " + this.childDocs.size());
|
---|
150 | // for (int a = 0; a < this.childDocs.size(); a ++) {
|
---|
151 | //System.out.println(this.childDocs.get(a));
|
---|
152 | //}
|
---|
153 | }
|
---|
154 |
|
---|
155 | public void setParent(AbstractHierarchyNode parent)
|
---|
156 | { this.parent = parent;
|
---|
157 | }
|
---|
158 |
|
---|
159 | public AbstractHierarchyNode getParent()
|
---|
160 | { return this.parent;
|
---|
161 | }
|
---|
162 |
|
---|
163 | public String getParentId()
|
---|
164 | {
|
---|
165 | if (this.id == null) {
|
---|
166 | return "";
|
---|
167 | }
|
---|
168 |
|
---|
169 | int dotAt = this.id.lastIndexOf('.');
|
---|
170 | if (dotAt < 0) {
|
---|
171 | return "";
|
---|
172 | }
|
---|
173 | return this.id.substring(0, dotAt);
|
---|
174 | }
|
---|
175 |
|
---|
176 | public void setDescriptor(String descriptor)
|
---|
177 | { this.descriptor = descriptor;
|
---|
178 | }
|
---|
179 |
|
---|
180 | public void setID(String id)
|
---|
181 | { this.id = id;
|
---|
182 | this.matches.add(id);
|
---|
183 | }
|
---|
184 |
|
---|
185 | public String getID()
|
---|
186 | { return this.id;
|
---|
187 | }
|
---|
188 |
|
---|
189 | public int noOfChildDocs()
|
---|
190 | { return this.childDocs.size();
|
---|
191 | }
|
---|
192 |
|
---|
193 | public int noOfLeafDocs()
|
---|
194 | { int total = this.childDocs.size();
|
---|
195 |
|
---|
196 | Iterator iterator = this.childNodes.iterator();
|
---|
197 | while (iterator.hasNext()) {
|
---|
198 | AbstractHierarchyNode childNode = (AbstractHierarchyNode) iterator.next();
|
---|
199 |
|
---|
200 | total += childNode.noOfLeafDocs();
|
---|
201 | }
|
---|
202 |
|
---|
203 | return total;
|
---|
204 | }
|
---|
205 |
|
---|
206 | /**
|
---|
207 | * Set the name for this hierarchy node - i.e. its brief description
|
---|
208 | *
|
---|
209 | * @param <code>String</code> the new node name.
|
---|
210 | */
|
---|
211 | public void setName(String name)
|
---|
212 | { this.name = name;
|
---|
213 | this.matches.add(name);
|
---|
214 | }
|
---|
215 |
|
---|
216 | /**
|
---|
217 | * Get the name for this hierarchy node - i.e. its brief description
|
---|
218 | *
|
---|
219 | * @return <code>String</code> the node's name.
|
---|
220 | */
|
---|
221 | public String getName()
|
---|
222 | { return this.name;
|
---|
223 | }
|
---|
224 |
|
---|
225 | /**
|
---|
226 | * Add another string which can be used to describe this hierarchy node - i.e. one which if
|
---|
227 | * found in a document would indicate that that document belongs to this hierarchy node.
|
---|
228 | *
|
---|
229 | * @param <code>String</code> the string to match.
|
---|
230 | */
|
---|
231 | public void addMatch(String match)
|
---|
232 | { this.matches.add(match);
|
---|
233 | }
|
---|
234 |
|
---|
235 | /**
|
---|
236 | * Check if a given string matches any of the values given for this hierarchy node...
|
---|
237 | *
|
---|
238 | * @param <code>String</code> the descriptive string to be compared against the hierarchy
|
---|
239 | * @return <code>boolean</code> <code>true</code> whether the string matches this hierarchy
|
---|
240 | */
|
---|
241 | public boolean isMatch (String toMatch)
|
---|
242 | { Iterator thisMatch = this.matches.iterator();
|
---|
243 |
|
---|
244 | while (thisMatch.hasNext())
|
---|
245 | { String thisMatchText = thisMatch.next().toString();
|
---|
246 |
|
---|
247 | if (thisMatchText.equals(toMatch))
|
---|
248 | { return true;
|
---|
249 | }
|
---|
250 | }
|
---|
251 | return false;
|
---|
252 | }
|
---|
253 |
|
---|
254 | /**
|
---|
255 | * Take a document, and find the classifications that it matches against in
|
---|
256 | * the current hierarchy.
|
---|
257 | *
|
---|
258 | * @param <code>DocumentID</code> the id of the document being classified
|
---|
259 | * @param <code>List</code> the values against which the classifier should
|
---|
260 | * test for the document being a match - i.e. the pertinent document
|
---|
261 | * property values.
|
---|
262 | * @param <code>ClassifierObserverInterface</code>
|
---|
263 | * object modifies the document with information about the
|
---|
264 | * classifications that it fell within.
|
---|
265 | */
|
---|
266 | abstract public void getClassifications(DocumentID documentID, List values, String sortKey,
|
---|
267 | ClassifierObserverInterface observer);
|
---|
268 |
|
---|
269 | public boolean writeSQL(GS3SQLConnection connection)
|
---|
270 | {
|
---|
271 | int classifyRef;
|
---|
272 | GS3SQLAction action = null;
|
---|
273 | GS3SQLSelect select;
|
---|
274 | GS3SQLInsert insert;
|
---|
275 |
|
---|
276 | Statement statement;
|
---|
277 | // can we connect to the database?
|
---|
278 | try {
|
---|
279 | statement = connection.createStatement();
|
---|
280 | } catch (SQLException e) {
|
---|
281 | System.err.println("AbstractHierarchyNode.writeSQL(): "+e);
|
---|
282 | return false;
|
---|
283 | }
|
---|
284 |
|
---|
285 | // Get own full id
|
---|
286 | String fullId = this.id.length() > 0 ? this.prefix+"."+this.id : this.prefix;
|
---|
287 |
|
---|
288 | // check for an existing instance of this classifier
|
---|
289 | select = new GS3SQLSelect("classifiers");
|
---|
290 | select.addField("ClassifyRef");
|
---|
291 | GS3SQLWhereItem whereItem = new GS3SQLWhereItem("ClassifyID", "=", fullId);
|
---|
292 | GS3SQLWhere where = new GS3SQLWhere(whereItem);
|
---|
293 | select.setWhere(where);
|
---|
294 |
|
---|
295 |
|
---|
296 | // update or insert the classifier as required
|
---|
297 | try {
|
---|
298 | ResultSet results = statement.executeQuery(select.toString());
|
---|
299 | if (results.first()) {
|
---|
300 | GS3SQLUpdate update = new GS3SQLUpdate("classifiers");
|
---|
301 | update.setWhere(where);
|
---|
302 | action = update;
|
---|
303 |
|
---|
304 | classifyRef = results.getInt("ClassifyRef");
|
---|
305 | }
|
---|
306 | else {
|
---|
307 | insert = new GS3SQLInsert("classifiers");
|
---|
308 |
|
---|
309 | if (this.id == null || this.id.length() == 0) {
|
---|
310 | insert.addValue("ParentID", "");
|
---|
311 | }
|
---|
312 | else {
|
---|
313 | String parentId = this.getParentId();
|
---|
314 |
|
---|
315 | if (parentId.length() > 0) {
|
---|
316 | insert.addValue("ParentID", this.prefix+"."+this.getParentId());
|
---|
317 | }
|
---|
318 | else {
|
---|
319 | insert.addValue("ParentID", this.prefix);
|
---|
320 | }
|
---|
321 | }
|
---|
322 |
|
---|
323 | action = insert;
|
---|
324 | classifyRef = -1;
|
---|
325 | }
|
---|
326 | action.addValue("ClassifyID", fullId);
|
---|
327 | action.addValue("Name", this.name);
|
---|
328 | action.addValue("Description", this.descriptor);
|
---|
329 |
|
---|
330 | String tailId = "0";
|
---|
331 | if (this.id.length() > 0) {
|
---|
332 | int lastDot = this.id.lastIndexOf('.');
|
---|
333 | if (lastDot >= 0) {
|
---|
334 | tailId = this.id.substring(lastDot+1);
|
---|
335 | }
|
---|
336 | else {
|
---|
337 | tailId = this.id;
|
---|
338 | }
|
---|
339 | }
|
---|
340 | action.addValue("ClassifyOrder", tailId, GS3SQLField.INTEGER_TYPE);
|
---|
341 | action.addValue("NumLeafDocs", Integer.toString(this.noOfLeafDocs()), GS3SQLField.INTEGER_TYPE);
|
---|
342 |
|
---|
343 | statement.execute(action.toString());
|
---|
344 | }
|
---|
345 | catch (SQLException sqlEx) {
|
---|
346 | if (action == null) {
|
---|
347 | System.err.println("AbstractHierarchyNode.writeSQL(): "+sqlEx);
|
---|
348 | }
|
---|
349 | else {
|
---|
350 | System.err.println("AbstractHierarchyNode.writeSQL(): "+sqlEx + " " + action.toString());
|
---|
351 | }
|
---|
352 | return false;
|
---|
353 | }
|
---|
354 |
|
---|
355 | // get the ClassifyRef if we don't already have it (have done a
|
---|
356 | // insert action above)...
|
---|
357 | if (classifyRef == -1) {
|
---|
358 | try {
|
---|
359 | ResultSet results = statement.executeQuery(select.toString());
|
---|
360 | if (!results.first()) {
|
---|
361 | statement.close();
|
---|
362 | return false;
|
---|
363 | }
|
---|
364 |
|
---|
365 | classifyRef = results.getInt("ClassifyRef");
|
---|
366 |
|
---|
367 | }
|
---|
368 | catch (SQLException sqlEx) {
|
---|
369 | System.err.println("AbstractHierarchyNode.writeSQL(): "+sqlEx);
|
---|
370 | return false;
|
---|
371 | }
|
---|
372 | }
|
---|
373 | else {
|
---|
374 | // TODO: clear 'dead' child classifications
|
---|
375 |
|
---|
376 | // delete child documents
|
---|
377 | GS3SQLDelete delete = new GS3SQLDelete("classdocuments");
|
---|
378 | delete.setWhere(where);
|
---|
379 | try {
|
---|
380 | statement.execute(delete.toString());
|
---|
381 | } catch (SQLException e) {
|
---|
382 | System.err.println("AbstractHierarchyNode.writeSQL(): "+e);
|
---|
383 | return false;
|
---|
384 | }
|
---|
385 | }
|
---|
386 |
|
---|
387 | // post the child nodes...
|
---|
388 | Iterator iterator = this.childNodes.iterator();
|
---|
389 | while (iterator.hasNext()) {
|
---|
390 | AbstractHierarchyNode childNode = (AbstractHierarchyNode) iterator.next();
|
---|
391 |
|
---|
392 | if (!childNode.writeSQL(connection)) {
|
---|
393 | System.out.println("Failed to write " );
|
---|
394 | return false;
|
---|
395 | }
|
---|
396 | }
|
---|
397 |
|
---|
398 | // note the child documents...
|
---|
399 | iterator = this.childDocs.iterator();
|
---|
400 | int order = 1;
|
---|
401 | while (iterator.hasNext()) {
|
---|
402 | AbstractHierarchyDocument hierarchyDoc = (AbstractHierarchyDocument) iterator.next();
|
---|
403 | DocumentID docId = hierarchyDoc.getID();
|
---|
404 |
|
---|
405 | insert = new GS3SQLInsert("classdocuments");
|
---|
406 | insert.addValue("ClassifyRef", Integer.toString(classifyRef), GS3SQLField.INTEGER_TYPE);
|
---|
407 | insert.addValue("DocID", docId.toString());
|
---|
408 | insert.addValue("DocOrder", Integer.toString(order), GS3SQLField.INTEGER_TYPE);
|
---|
409 | try {
|
---|
410 | statement.execute(insert.toString());
|
---|
411 | } catch (SQLException e) {
|
---|
412 | System.err.println("AbstractHierarchyNode.writeSQL(): "+e);
|
---|
413 | return false;
|
---|
414 | }
|
---|
415 | order ++;
|
---|
416 | }
|
---|
417 | // close the statment
|
---|
418 | try {
|
---|
419 | statement.close();
|
---|
420 | } catch (SQLException e) {
|
---|
421 | System.err.println("AbstractHierarchyNode.writeSQL(): "+e);
|
---|
422 | }
|
---|
423 |
|
---|
424 | return true;
|
---|
425 | }
|
---|
426 | }
|
---|
427 |
|
---|
428 |
|
---|