1 | package org.greenstone.gsdl3.gs3build.classifier;
|
---|
2 |
|
---|
3 | import java.io.*;
|
---|
4 |
|
---|
5 | import java.util.List;
|
---|
6 | import java.util.ArrayList;
|
---|
7 | import java.util.Iterator;
|
---|
8 |
|
---|
9 | import java.sql.ResultSet;
|
---|
10 | import java.sql.SQLException;
|
---|
11 |
|
---|
12 | import org.xml.sax.XMLReader;
|
---|
13 | import org.xml.sax.InputSource;
|
---|
14 | import org.xml.sax.SAXException;
|
---|
15 | import org.xml.sax.Attributes;
|
---|
16 | import org.xml.sax.helpers.XMLReaderFactory;
|
---|
17 | import org.xml.sax.helpers.DefaultHandler;
|
---|
18 |
|
---|
19 | import org.w3c.dom.Element;
|
---|
20 |
|
---|
21 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentID;
|
---|
22 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentInterface;
|
---|
23 | import org.greenstone.gsdl3.gs3build.util.XMLTools;
|
---|
24 | import org.greenstone.gsdl3.gs3build.database.*;
|
---|
25 | import org.greenstone.gsdl3.gs3build.util.MultiMap;
|
---|
26 | import org.greenstone.gsdl3.util.GSXML;
|
---|
27 |
|
---|
28 | public class AZListClassifier extends AbstractClassifier
|
---|
29 | {
|
---|
30 | GS3SQLConnection database;
|
---|
31 | MultiMap alphaMap;
|
---|
32 | List fields;
|
---|
33 | String name;
|
---|
34 |
|
---|
35 | class AZDocumentItem
|
---|
36 | { public DocumentID documentId;
|
---|
37 | public String title;
|
---|
38 |
|
---|
39 | public AZDocumentItem(DocumentID documentId, String title)
|
---|
40 | { this.documentId = documentId;
|
---|
41 | this.title = title;
|
---|
42 | }
|
---|
43 |
|
---|
44 | public String getTitle()
|
---|
45 | { return this.title;
|
---|
46 | }
|
---|
47 |
|
---|
48 | public DocumentID getID()
|
---|
49 | { return this.documentId;
|
---|
50 | }
|
---|
51 |
|
---|
52 | public String toString()
|
---|
53 | { return this.title;
|
---|
54 | }
|
---|
55 | }
|
---|
56 |
|
---|
57 | private void sortList(String key)
|
---|
58 | { List list = this.alphaMap.getAll(key);
|
---|
59 | if (list.size() > 1) {
|
---|
60 | int first, last;
|
---|
61 |
|
---|
62 | first = 0;
|
---|
63 | last = list.size() - 1;
|
---|
64 | while (first != last)
|
---|
65 | { int at = (first + last) / 2;
|
---|
66 |
|
---|
67 | if (list.get(at).toString().compareTo(list.get(list.size() - 1).toString()) > 0) {
|
---|
68 | last = at;
|
---|
69 | }
|
---|
70 | else {
|
---|
71 | first = at + 1;
|
---|
72 | }
|
---|
73 | }
|
---|
74 |
|
---|
75 | Object newObject = list.get(list.size()-1);
|
---|
76 | last = list.size() - 1;
|
---|
77 | while (last > first) {
|
---|
78 | list.set(last, list.get(last-1));
|
---|
79 | last --;
|
---|
80 | }
|
---|
81 | list.set(first, newObject);
|
---|
82 | }
|
---|
83 | }
|
---|
84 |
|
---|
85 | public AZListClassifier(String name, List fields)
|
---|
86 | { this.fields = fields;
|
---|
87 | this.alphaMap = new MultiMap();
|
---|
88 | this.name = name;
|
---|
89 | // System.out.println("AZList :"+fields.size());
|
---|
90 | }
|
---|
91 |
|
---|
92 | public AZListClassifier()
|
---|
93 | { this.fields = new ArrayList();
|
---|
94 | }
|
---|
95 |
|
---|
96 | public void configure(List parameters)
|
---|
97 | {
|
---|
98 | Iterator iterator = parameters.iterator();
|
---|
99 | while (iterator.hasNext()) {
|
---|
100 | String param = iterator.next().toString();
|
---|
101 | if (param.equals("-metadata")) {
|
---|
102 | if (iterator.hasNext()) {
|
---|
103 | String field = iterator.next().toString();
|
---|
104 | if (field != null && field.length() > 0) {
|
---|
105 | this.fields.add(field);
|
---|
106 | }
|
---|
107 | }
|
---|
108 | }
|
---|
109 | }
|
---|
110 | }
|
---|
111 |
|
---|
112 | public void setDatabase(GS3SQLConnection connection)
|
---|
113 | { this.database = connection;
|
---|
114 | }
|
---|
115 |
|
---|
116 | public boolean classifyDocument(DocumentID documentID, DocumentInterface document)
|
---|
117 | { // get the metadata item from the document
|
---|
118 | Iterator thisField = this.fields.iterator();
|
---|
119 |
|
---|
120 | // for every field listed as being a scannable field...
|
---|
121 | while (thisField.hasNext())
|
---|
122 | { String fieldName = thisField.next().toString();
|
---|
123 |
|
---|
124 | // ...get the values for that field...
|
---|
125 | List values = document.getDocumentMetadataItem(fieldName);
|
---|
126 |
|
---|
127 | if (values == null)
|
---|
128 | { continue;
|
---|
129 | }
|
---|
130 |
|
---|
131 | // ...and send them to the classifier
|
---|
132 | Iterator thisValue = values.iterator();
|
---|
133 | while (thisValue.hasNext())
|
---|
134 | { String value = thisValue.next().toString();
|
---|
135 |
|
---|
136 | String classText = null;
|
---|
137 |
|
---|
138 | int c = 0;
|
---|
139 | while (c < value.length() && classText == null)
|
---|
140 | { if (Character.isDigit(value.charAt(c)))
|
---|
141 | { classText = "0-9";
|
---|
142 | }
|
---|
143 | else if (Character.isLetter(value.charAt(c)))
|
---|
144 | { classText = value.substring(c, c+1).toLowerCase();
|
---|
145 | }
|
---|
146 | c ++;
|
---|
147 | }
|
---|
148 |
|
---|
149 | if (classText != null) {
|
---|
150 | // System.out.println(classText + " " + documentID);
|
---|
151 | this.alphaMap.put(classText, new AZDocumentItem(documentID, value));
|
---|
152 | this.sortList(classText);
|
---|
153 | }
|
---|
154 | }
|
---|
155 | }
|
---|
156 |
|
---|
157 | return true;
|
---|
158 | }
|
---|
159 |
|
---|
160 | public void startClassifierPass(int pass)
|
---|
161 | {
|
---|
162 | }
|
---|
163 |
|
---|
164 | public int getClassifierPasses()
|
---|
165 | { return 1;
|
---|
166 | }
|
---|
167 |
|
---|
168 | public void endClassifierPass(int pass)
|
---|
169 | {
|
---|
170 | }
|
---|
171 |
|
---|
172 | public void completeClassification()
|
---|
173 | { this.writeSQL(this.database);
|
---|
174 | }
|
---|
175 |
|
---|
176 | public int writeSQLClassifyNode(GS3SQLConnection connection, String parentId, int orderRef,
|
---|
177 | String label, String name, String description, int noOfLeafDocs)
|
---|
178 | { GS3SQLAction action;
|
---|
179 | GS3SQLSelect select;
|
---|
180 | GS3SQLInsert insert;
|
---|
181 |
|
---|
182 | int classifyRef;
|
---|
183 |
|
---|
184 | // check for existing node
|
---|
185 | select = new GS3SQLSelect("classifiers");
|
---|
186 | select.addField("ClassifyRef");
|
---|
187 | GS3SQLWhereItem whereItem = new GS3SQLWhereItem("ClassifyID", "=", label);
|
---|
188 | GS3SQLWhere where = new GS3SQLWhere(whereItem);
|
---|
189 | select.setWhere(where);
|
---|
190 |
|
---|
191 | connection.execute(select.toString());
|
---|
192 |
|
---|
193 | try {
|
---|
194 | ResultSet results = connection.getResultSet();
|
---|
195 | if (results != null && results.first()) {
|
---|
196 | GS3SQLUpdate update = new GS3SQLUpdate("classifiers");
|
---|
197 | update.setWhere(where);
|
---|
198 | action = update;
|
---|
199 |
|
---|
200 | classifyRef = results.getInt("ClassifyRef");
|
---|
201 | }
|
---|
202 | else {
|
---|
203 | insert = new GS3SQLInsert("classifiers");
|
---|
204 |
|
---|
205 | insert.addValue("ParentID", parentId);
|
---|
206 |
|
---|
207 | action = insert;
|
---|
208 | }
|
---|
209 | action.addValue("ClassifyID", label);
|
---|
210 | action.addValue("Name", name);
|
---|
211 | action.addValue("Description", description);
|
---|
212 | action.addValue("ClassifyOrder", Integer.toString(orderRef), GS3SQLField.INTEGER_TYPE);
|
---|
213 | action.addValue("NumLeafDocs", Integer.toString(noOfLeafDocs), GS3SQLField.INTEGER_TYPE);
|
---|
214 |
|
---|
215 | connection.execute(action.toString());
|
---|
216 | classifyRef = -1;
|
---|
217 | }
|
---|
218 | catch (SQLException sqlEx) {
|
---|
219 | System.err.println(sqlEx);
|
---|
220 | return -1;
|
---|
221 | }
|
---|
222 |
|
---|
223 | // get the ClassifyRef if we don't already have it (have done a
|
---|
224 | // insert action above)...
|
---|
225 | if (classifyRef == -1) {
|
---|
226 | connection.execute(select.toString());
|
---|
227 |
|
---|
228 | try {
|
---|
229 | ResultSet results = connection.getResultSet();
|
---|
230 | if (results == null || !results.first()) {
|
---|
231 | return -1;
|
---|
232 | }
|
---|
233 |
|
---|
234 | classifyRef = results.getInt("ClassifyRef");
|
---|
235 | }
|
---|
236 | catch (SQLException sqlEx) {
|
---|
237 | System.err.println(sqlEx);
|
---|
238 | return -1;
|
---|
239 | }
|
---|
240 | }
|
---|
241 |
|
---|
242 | return classifyRef;
|
---|
243 | }
|
---|
244 |
|
---|
245 | public boolean writeSQL(GS3SQLConnection connection)
|
---|
246 | { GS3SQLAction action;
|
---|
247 | GS3SQLSelect select;
|
---|
248 | GS3SQLInsert insert;
|
---|
249 |
|
---|
250 | Iterator keys;
|
---|
251 |
|
---|
252 | String prefix = this.name;
|
---|
253 | if (prefix==null || prefix.equals("")) {
|
---|
254 | prefix = "CLAZ"+this.fields.get(0).toString();
|
---|
255 | // TODO: write this name back to collectionConfig.xml
|
---|
256 | this.name = prefix;
|
---|
257 | }
|
---|
258 |
|
---|
259 | int leafCount = 0;
|
---|
260 | keys = this.alphaMap.keySet().iterator();
|
---|
261 | while (keys.hasNext()) {
|
---|
262 | Object key = keys.next();
|
---|
263 |
|
---|
264 | List childDocs = (List) this.alphaMap.getAll(key);
|
---|
265 | if (childDocs != null) {
|
---|
266 | leafCount += childDocs.size();
|
---|
267 | }
|
---|
268 | }
|
---|
269 |
|
---|
270 | int parentClassify = this.writeSQLClassifyNode(connection, "", 0, prefix, "", "Classifier", leafCount);
|
---|
271 |
|
---|
272 | if (parentClassify < 0)
|
---|
273 | { return false;
|
---|
274 | }
|
---|
275 |
|
---|
276 | List children;
|
---|
277 |
|
---|
278 | // TODO: cope with change rather than create from scratch...
|
---|
279 | int nodeOrder = 1;
|
---|
280 | keys = this.alphaMap.keySet().iterator();
|
---|
281 | while (keys.hasNext()) {
|
---|
282 | Object key = keys.next();
|
---|
283 |
|
---|
284 | List childDocs = (List) this.alphaMap.getAll(key);
|
---|
285 |
|
---|
286 | if (childDocs != null) {
|
---|
287 | String className = prefix + "." + key.toString();
|
---|
288 |
|
---|
289 | int classifyRef = this.writeSQLClassifyNode(connection, prefix, nodeOrder, className, key.toString(), key.toString(), childDocs.size());
|
---|
290 |
|
---|
291 | // note the child documents...
|
---|
292 | Iterator iterator = childDocs.iterator();
|
---|
293 | int childOrder = 1;
|
---|
294 | while (iterator.hasNext()) {
|
---|
295 | AZDocumentItem documentItem = (AZDocumentItem) iterator.next();
|
---|
296 | DocumentID docId = documentItem.getID();
|
---|
297 |
|
---|
298 | insert = new GS3SQLInsert("classdocuments");
|
---|
299 | insert.addValue("ClassifyRef", Integer.toString(classifyRef), GS3SQLField.INTEGER_TYPE);
|
---|
300 | insert.addValue("DocID", docId.toString());
|
---|
301 | insert.addValue("DocOrder", Integer.toString(childOrder), GS3SQLField.INTEGER_TYPE);
|
---|
302 |
|
---|
303 | connection.execute(insert.toString());
|
---|
304 |
|
---|
305 | childOrder ++;
|
---|
306 | }
|
---|
307 |
|
---|
308 | nodeOrder ++;
|
---|
309 | }
|
---|
310 | }
|
---|
311 |
|
---|
312 | /*
|
---|
313 | else {
|
---|
314 | // TODO: clear 'dead' child classifications
|
---|
315 |
|
---|
316 | // delete child documents
|
---|
317 | GS3SQLDelete delete = new GS3SQLDelete("classdocuments");
|
---|
318 | delete.setWhere(where);
|
---|
319 |
|
---|
320 | connection.execute(delete.toString());
|
---|
321 | }
|
---|
322 |
|
---|
323 | // post the child nodes...
|
---|
324 | Iterator iterator = this.childNodes.iterator();
|
---|
325 | while (iterator.hasNext()) {
|
---|
326 | AbstractHierarchyNode childNode = (AbstractHierarchyNode) iterator.next();
|
---|
327 |
|
---|
328 | if (!childNode.writeSQL(connection)) {
|
---|
329 | return false;
|
---|
330 | }
|
---|
331 | }
|
---|
332 | */
|
---|
333 |
|
---|
334 | return true;
|
---|
335 | }
|
---|
336 | public boolean addClassifierDescription(Element classifier_list) {
|
---|
337 | // TODO check that there are some docs in the classification
|
---|
338 | Element classifier = classifier_list.getOwnerDocument().createElement(GSXML.CLASSIFIER_ELEM);
|
---|
339 | classifier.setAttribute(GSXML.NAME_ATT, this.name);
|
---|
340 | classifier_list.appendChild(classifier);
|
---|
341 | return true;
|
---|
342 | }
|
---|
343 |
|
---|
344 | }
|
---|
345 |
|
---|
346 |
|
---|
347 |
|
---|
348 |
|
---|
349 |
|
---|