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 org.xml.sax.XMLReader;
|
---|
10 | import org.xml.sax.InputSource;
|
---|
11 | import org.xml.sax.SAXException;
|
---|
12 | import org.xml.sax.Attributes;
|
---|
13 | import org.xml.sax.helpers.XMLReaderFactory;
|
---|
14 | import org.xml.sax.helpers.DefaultHandler;
|
---|
15 |
|
---|
16 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentID;
|
---|
17 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentInterface;
|
---|
18 | import org.greenstone.gsdl3.gs3build.doctypes.DocumentLoader;
|
---|
19 | import org.greenstone.gsdl3.gs3build.util.XMLTools;
|
---|
20 | import org.greenstone.gsdl3.gs3build.util.GS2TextFileHandler;
|
---|
21 | import org.greenstone.gsdl3.gs3build.util.GS3SQLConnection;
|
---|
22 |
|
---|
23 | public class GS2HierarchyClassifier implements ClassifierInterface
|
---|
24 | {
|
---|
25 | class GS2HierarchyClassifierObserver implements ClassifierObserverInterface
|
---|
26 | { List classifications;
|
---|
27 | DocumentInterface document;
|
---|
28 |
|
---|
29 | public GS2HierarchyClassifierObserver(DocumentInterface document)
|
---|
30 | { this.document = document;
|
---|
31 | }
|
---|
32 |
|
---|
33 | public void recordClassification(String label)
|
---|
34 | { this.document.addDocumentMetadata("gsdl3", "classified", label);
|
---|
35 | System.out.println("Assigned document " + this.document.getID().toString() + " to " + label);
|
---|
36 | }
|
---|
37 |
|
---|
38 | public List getClassifications()
|
---|
39 | { return this.classifications;
|
---|
40 | }
|
---|
41 | }
|
---|
42 |
|
---|
43 | class GS2HierarchyNode extends AbstractHierarchyNode
|
---|
44 | { /**
|
---|
45 | * Simple node
|
---|
46 | */
|
---|
47 | public GS2HierarchyNode()
|
---|
48 | { super();
|
---|
49 | }
|
---|
50 |
|
---|
51 | public GS2HierarchyNode(String name, String id, String descriptor)
|
---|
52 | { super(name, id, descriptor);
|
---|
53 | }
|
---|
54 |
|
---|
55 | /**
|
---|
56 | * Take a document, and find the classifications that it matches against in
|
---|
57 | * the current hierarchy.
|
---|
58 | *
|
---|
59 | * @param <code>DocumentID</code> the id of the document being classified
|
---|
60 | * @param <code>List</code> the values against which the classifier should
|
---|
61 | * test for the document being a match - i.e. the pertinent document
|
---|
62 | * property values.
|
---|
63 | * @param <code>ClassifierObserverInterface</code>
|
---|
64 | * object modifies the document with information about the
|
---|
65 | * classifications that it fell within.
|
---|
66 | */
|
---|
67 | public void getClassifications(DocumentID documentID, List values,
|
---|
68 | ClassifierObserverInterface observer)
|
---|
69 | { // Go through the list of values, and if this node matches, record
|
---|
70 | // the fact in the classifier observer...
|
---|
71 | Iterator valueList = values.iterator();
|
---|
72 |
|
---|
73 | while (valueList.hasNext())
|
---|
74 | { if (this.isMatch(valueList.next().toString()))
|
---|
75 | { observer.recordClassification(this.name);
|
---|
76 | this.addDocument(documentID);
|
---|
77 | }
|
---|
78 | }
|
---|
79 |
|
---|
80 | // recurse into the child nodes for them to do the same...
|
---|
81 | Iterator childList = this.childNodes.iterator();
|
---|
82 |
|
---|
83 | while (childList.hasNext())
|
---|
84 | { ((GS2HierarchyNode) childList.next()).getClassifications(documentID, values, observer);
|
---|
85 | }
|
---|
86 | }
|
---|
87 | }
|
---|
88 |
|
---|
89 | /**
|
---|
90 | * An inner class to handle Classifier files
|
---|
91 | */
|
---|
92 | class GS2HierarchyHandler extends GS2TextFileHandler
|
---|
93 | { GS2HierarchyNode rootNode = null;
|
---|
94 |
|
---|
95 | GS2HierarchyHandler(String hierarchyName, String content)
|
---|
96 | { super(content);
|
---|
97 |
|
---|
98 | this.rootNode = new GS2HierarchyNode();
|
---|
99 |
|
---|
100 | while (this.hasMoreLines()) {
|
---|
101 | String description;
|
---|
102 | String name, id;
|
---|
103 |
|
---|
104 | this.getLine();
|
---|
105 |
|
---|
106 | name = this.getEntry(true);
|
---|
107 | if (name == null || name.length() == 0) {
|
---|
108 | continue;
|
---|
109 | }
|
---|
110 |
|
---|
111 | id = this.getEntry(true);
|
---|
112 | if (id == null || id.length() == 0) {
|
---|
113 | continue;
|
---|
114 | }
|
---|
115 |
|
---|
116 | description = this.getEntry(true);
|
---|
117 | if (description == null || description.length() == 0) {
|
---|
118 | continue;
|
---|
119 | }
|
---|
120 |
|
---|
121 | // Initialise the tree
|
---|
122 | GS2HierarchyNode node = new GS2HierarchyNode(name, "CL"+hierarchyName+"."+id, description);
|
---|
123 |
|
---|
124 | this.rootNode.add(node);
|
---|
125 | }
|
---|
126 | }
|
---|
127 |
|
---|
128 | public GS2HierarchyNode getHierarchy()
|
---|
129 | { return this.rootNode;
|
---|
130 | }
|
---|
131 | }
|
---|
132 |
|
---|
133 | private GS2HierarchyNode hierarchy;
|
---|
134 | List fields;
|
---|
135 | String sortBy;
|
---|
136 | GS3SQLConnection database;
|
---|
137 |
|
---|
138 | public GS2HierarchyClassifier()
|
---|
139 | { // a 'do nothing' constructor which takes more information from configure...
|
---|
140 | this.fields = new ArrayList();
|
---|
141 | this.database = null;
|
---|
142 | }
|
---|
143 |
|
---|
144 | public GS2HierarchyClassifier(File file)
|
---|
145 | {
|
---|
146 | String documentText =
|
---|
147 | DocumentLoader.getAsString(file);
|
---|
148 |
|
---|
149 | if (documentText == null) {
|
---|
150 | System.err.println("GS2HierarchyClassifier: Unable to load any content for " + file);
|
---|
151 | return;
|
---|
152 | }
|
---|
153 |
|
---|
154 | GS2HierarchyHandler handler = new GS2HierarchyHandler("CL", documentText);
|
---|
155 |
|
---|
156 | this.hierarchy = handler.getHierarchy();
|
---|
157 | this.fields = new ArrayList();
|
---|
158 | System.out.println("loaded hierarchy for "+ file.toString());
|
---|
159 | this.database = null;
|
---|
160 | }
|
---|
161 |
|
---|
162 | public void setDatabase(GS3SQLConnection database)
|
---|
163 | { this.database = database;
|
---|
164 | }
|
---|
165 |
|
---|
166 | public void configure(List parameters)
|
---|
167 | {
|
---|
168 | File file = null;
|
---|
169 |
|
---|
170 | Iterator iterator = parameters.iterator();
|
---|
171 | while (iterator.hasNext()) {
|
---|
172 | String param = iterator.next().toString();
|
---|
173 |
|
---|
174 | if (param.equals("-metadata")) {
|
---|
175 | if (iterator.hasNext()) {
|
---|
176 | String field = iterator.next().toString();
|
---|
177 | if (field != null && field.length() > 0) {
|
---|
178 | this.fields.add(field);
|
---|
179 | }
|
---|
180 | }
|
---|
181 | }
|
---|
182 | else if (param.equals("-hfile")) {
|
---|
183 | // this parameter had better have been expanded previously to a full path,
|
---|
184 | // as we're not going to do that here!!!
|
---|
185 | if (iterator.hasNext()) {
|
---|
186 | file = new File(iterator.next().toString());
|
---|
187 | }
|
---|
188 | }
|
---|
189 | else if (param.equals("-sort")) {
|
---|
190 | if (iterator.hasNext()) {
|
---|
191 | String sort = iterator.next().toString();
|
---|
192 | if (sort != null && sort.length() > 0) {
|
---|
193 | this.sortBy = sort;
|
---|
194 | }
|
---|
195 | }
|
---|
196 | }
|
---|
197 | }
|
---|
198 |
|
---|
199 | if (file == null) {
|
---|
200 | return;
|
---|
201 | }
|
---|
202 |
|
---|
203 | String documentText =
|
---|
204 | DocumentLoader.getAsString(file);
|
---|
205 |
|
---|
206 | if (documentText == null) {
|
---|
207 | System.err.println("GS2HierarchyClassifier: Unable to load any content for " + file);
|
---|
208 | return;
|
---|
209 | }
|
---|
210 |
|
---|
211 | GS2HierarchyHandler handler = new GS2HierarchyHandler(this.fields.get(0).toString(), documentText);
|
---|
212 |
|
---|
213 | this.hierarchy = handler.getHierarchy();
|
---|
214 | }
|
---|
215 |
|
---|
216 | /**
|
---|
217 | * Classify a particular document - the document id will be used to record its
|
---|
218 | * membership of any classifier nodes, and the document interface will also be
|
---|
219 | * used to assign back knowledge of any classifiers to which the document is
|
---|
220 | * assigned.
|
---|
221 | *
|
---|
222 | * @param <code>DocumentID</code> the document identifier
|
---|
223 | * @param <code>DocumentInterface</code> the document interface
|
---|
224 | */
|
---|
225 | public boolean classifyDocument(DocumentID documentID, DocumentInterface document)
|
---|
226 | { // the observer records the assignment of documents to classifications
|
---|
227 | GS2HierarchyClassifierObserver classifyObserver = new GS2HierarchyClassifierObserver(document);
|
---|
228 |
|
---|
229 | // get the metadata item from the document
|
---|
230 | Iterator thisField = this.fields.iterator();
|
---|
231 |
|
---|
232 | // for every field listed as being a scannable field...
|
---|
233 | while (thisField.hasNext())
|
---|
234 | { String fieldName = thisField.next().toString();
|
---|
235 |
|
---|
236 | // ...get the values for that field...
|
---|
237 | List values = document.getDocumentMetadataItem(fieldName);
|
---|
238 |
|
---|
239 | if (values == null)
|
---|
240 | { continue;
|
---|
241 | }
|
---|
242 |
|
---|
243 | // ...and send them to the classifier
|
---|
244 | this.hierarchy.getClassifications(documentID, values, classifyObserver);
|
---|
245 | }
|
---|
246 | return true;
|
---|
247 | }
|
---|
248 |
|
---|
249 | public int getClassifierPasses()
|
---|
250 | { return 1;
|
---|
251 | }
|
---|
252 |
|
---|
253 | public void startClassifierPass(int pass)
|
---|
254 | {
|
---|
255 | }
|
---|
256 |
|
---|
257 | public void endClassifierPass(int pass)
|
---|
258 | {
|
---|
259 | }
|
---|
260 |
|
---|
261 | public void completeClassification()
|
---|
262 | { // TODO: store the classifications to file...
|
---|
263 |
|
---|
264 | // The hierarchy member may be null if the classifier did not initialise correctly...
|
---|
265 | if (this.hierarchy != null) {
|
---|
266 | this.hierarchy.writeSQL(this.database);
|
---|
267 | }
|
---|
268 | }
|
---|
269 | }
|
---|