|
1 |
// Copyright © 2010-2013 ENS de Lyon.
|
|
2 |
// Copyright © 2007-2010 ENS de Lyon, CNRS, INRP, University of
|
|
3 |
// Lyon 2, University of Franche-Comté, University of Nice
|
|
4 |
// Sophia Antipolis, University of Paris 3.
|
|
5 |
//
|
|
6 |
// The TXM platform is free software: you can redistribute it
|
|
7 |
// and/or modify it under the terms of the GNU General Public
|
|
8 |
// License as published by the Free Software Foundation,
|
|
9 |
// either version 2 of the License, or (at your option) any
|
|
10 |
// later version.
|
|
11 |
//
|
|
12 |
// The TXM platform is distributed in the hope that it will be
|
|
13 |
// useful, but WITHOUT ANY WARRANTY; without even the implied
|
|
14 |
// warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
|
|
15 |
// PURPOSE. See the GNU General Public License for more
|
|
16 |
// details.
|
|
17 |
//
|
|
18 |
// You should have received a copy of the GNU General
|
|
19 |
// Public License along with the TXM platform. If not, see
|
|
20 |
// http://www.gnu.org/licenses.
|
|
21 |
//
|
|
22 |
//
|
|
23 |
//
|
|
24 |
// $LastChangedDate: 2016-09-19 10:31:00 +0200 (Mon, 19 Sep 2016) $
|
|
25 |
// $LastChangedRevision: 3298 $
|
|
26 |
// $LastChangedBy: mdecorde $
|
|
27 |
//
|
|
28 |
package org.txm.information.core.functions;
|
|
29 |
|
|
30 |
import java.io.BufferedWriter;
|
|
31 |
import java.io.File;
|
|
32 |
import java.io.FileNotFoundException;
|
|
33 |
import java.io.FileOutputStream;
|
|
34 |
import java.io.IOException;
|
|
35 |
import java.io.OutputStreamWriter;
|
|
36 |
import java.io.UnsupportedEncodingException;
|
|
37 |
import java.util.ArrayList;
|
|
38 |
import java.util.Collections;
|
|
39 |
import java.util.HashMap;
|
|
40 |
import java.util.LinkedHashSet;
|
|
41 |
import java.util.List;
|
|
42 |
import java.util.Locale;
|
|
43 |
|
|
44 |
import org.eclipse.osgi.util.NLS;
|
|
45 |
import org.txm.core.preferences.TBXPreferences;
|
|
46 |
import org.txm.core.preferences.TXMPreferences;
|
|
47 |
import org.txm.core.results.Parameter;
|
|
48 |
import org.txm.core.results.TXMParameters;
|
|
49 |
import org.txm.core.results.TXMResult;
|
|
50 |
import org.txm.information.core.messages.InformationCoreMessages;
|
|
51 |
import org.txm.information.core.preferences.InformationPreferences;
|
|
52 |
import org.txm.objects.Base;
|
|
53 |
import org.txm.searchengine.cqp.AbstractCqiClient;
|
|
54 |
import org.txm.searchengine.cqp.CQPPreferences;
|
|
55 |
import org.txm.searchengine.cqp.clientExceptions.CqiClientException;
|
|
56 |
import org.txm.searchengine.cqp.corpus.Corpus;
|
|
57 |
import org.txm.searchengine.cqp.corpus.CorpusManager;
|
|
58 |
import org.txm.searchengine.cqp.corpus.Property;
|
|
59 |
import org.txm.searchengine.cqp.corpus.StructuralUnit;
|
|
60 |
import org.txm.searchengine.cqp.corpus.StructuralUnitProperty;
|
|
61 |
import org.txm.searchengine.cqp.corpus.WordProperty;
|
|
62 |
import org.txm.utils.AsciiUtils;
|
|
63 |
import org.txm.utils.logger.Log;
|
|
64 |
|
|
65 |
// TODO: Auto-generated Javadoc
|
|
66 |
/**
|
|
67 |
* compute a diagnostic of Corpus, sub corpus and partition @ author mdecorde.
|
|
68 |
*/
|
|
69 |
public class Information extends TXMResult {
|
|
70 |
// Base base;
|
|
71 |
/** The corpus. */
|
|
72 |
Corpus corpus;
|
|
73 |
|
|
74 |
// Numbers
|
|
75 |
/** The T. */
|
|
76 |
int T = 0; // number of word
|
|
77 |
|
|
78 |
/** The N properties. */
|
|
79 |
int NProperties = 0; // number of word properties
|
|
80 |
//HashMap<String, Integer> propertiesCounts = new HashMap<String, Integer>(); // properties counts
|
|
81 |
/** The properties values. */
|
|
82 |
HashMap<String, List<String>> propertiesValues = new HashMap<String, List<String>>(); // properties values
|
|
83 |
|
|
84 |
/** The N structures. */
|
|
85 |
int NStructures = 0; // number of structures
|
|
86 |
|
|
87 |
/** The structures counts. */
|
|
88 |
HashMap<String, Integer> structuresCounts = new HashMap<String, Integer>(); // structures counts
|
|
89 |
|
|
90 |
/** The internal architecture. */
|
|
91 |
List<String> internalArchitecture = new ArrayList<String>(); // wtc structures = propertiesCounts keys but ordered
|
|
92 |
|
|
93 |
/** The hierarchie counts. */
|
|
94 |
HashMap<String, Integer> hierarchieCounts = new HashMap<String, Integer>(); // hierarchic structures counts
|
|
95 |
|
|
96 |
/** The internal architecture properties. */
|
|
97 |
HashMap<String, HashMap<String, List<String>>> internalArchitectureProperties = new HashMap<String, HashMap<String, List<String>>>(); // hierarchic structures counts
|
|
98 |
|
|
99 |
/** The internal architecture properties counts. */
|
|
100 |
HashMap<String, HashMap<String, Integer>> internalArchitecturePropertiesCounts = new HashMap<String, HashMap<String, Integer>>(); // hierarchic structures counts
|
|
101 |
|
|
102 |
// Annotations description
|
|
103 |
/** The annotations types. */
|
|
104 |
HashMap<String, Integer> annotationsTypes = new HashMap<String, Integer>(); // for each annotation its description
|
|
105 |
|
|
106 |
/** The annotations origins. */
|
|
107 |
HashMap<String, Integer> annotationsOrigins = new HashMap<String, Integer>(); // for each annoation its origin description
|
|
108 |
|
|
109 |
// Import description
|
|
110 |
/** The import description. */
|
|
111 |
String importDescription = ""; //$NON-NLS-1$
|
|
112 |
|
|
113 |
/** The properties. */
|
|
114 |
List<WordProperty> properties;
|
|
115 |
|
|
116 |
/** The structures. */
|
|
117 |
List<StructuralUnit> structures;
|
|
118 |
|
|
119 |
/**
|
|
120 |
* The HTML file.
|
|
121 |
*/
|
|
122 |
protected File file;
|
|
123 |
|
|
124 |
|
|
125 |
|
|
126 |
/**
|
|
127 |
* The maximum number of word property values to display.
|
|
128 |
*/
|
|
129 |
@Parameter(key=InformationPreferences.MAX_VALUE)
|
|
130 |
protected int maxPropertiesToDisplay;
|
|
131 |
|
|
132 |
|
|
133 |
|
|
134 |
/**
|
|
135 |
* Instantiates a new Information.
|
|
136 |
*
|
|
137 |
* @param corpus a MainCorpus or a SubCorpus
|
|
138 |
*/
|
|
139 |
public Information(Corpus corpus) {
|
|
140 |
super(corpus);
|
|
141 |
this.corpus = corpus;
|
|
142 |
}
|
|
143 |
|
|
144 |
/**
|
|
145 |
* Sets the maximum properties to display.
|
|
146 |
* @param maxPropertiesToDisplay
|
|
147 |
*/
|
|
148 |
public void setParameters(int maxPropertiesToDisplay) {
|
|
149 |
this.maxPropertiesToDisplay = maxPropertiesToDisplay;
|
|
150 |
if (this.maxPropertiesToDisplay <= 0) {
|
|
151 |
this.maxPropertiesToDisplay = 10;
|
|
152 |
}
|
|
153 |
this.dirty = true;
|
|
154 |
}
|
|
155 |
|
|
156 |
|
|
157 |
// public boolean setParameters(TXMParameters parameters) {
|
|
158 |
// try {
|
|
159 |
// Integer max = null;
|
|
160 |
// Object o = parameters.get(CQPPreferences.MAX_VALUE);
|
|
161 |
// if (o instanceof Integer) max = (Integer)o;
|
|
162 |
// else if (o instanceof String) max = Integer.parseInt(o.toString());
|
|
163 |
//
|
|
164 |
// this.setParameters(max);
|
|
165 |
// } catch (Exception e) {
|
|
166 |
// System.out.println("Error while setting parameters of Informations: "+e.getLocalizedMessage());
|
|
167 |
// return false;
|
|
168 |
// }
|
|
169 |
// return true;
|
|
170 |
// }
|
|
171 |
|
|
172 |
/**
|
|
173 |
* Step general infos.
|
|
174 |
*/
|
|
175 |
public void stepGeneralInfos()
|
|
176 |
{
|
|
177 |
try {
|
|
178 |
properties = corpus.getProperties();
|
|
179 |
NProperties = properties.size();
|
|
180 |
} catch (CqiClientException e) {
|
|
181 |
Log.severe(InformationCoreMessages.Diagnostic_1 + corpus.getName() + InformationCoreMessages.Diagnostic_4);
|
|
182 |
Log.printStackTrace(e);
|
|
183 |
return;
|
|
184 |
}
|
|
185 |
|
|
186 |
try {
|
|
187 |
T = corpus.getSize();// corpus.getLexicon(corpus.getProperty("id")).nbrOfToken();
|
|
188 |
} catch (CqiClientException e) {
|
|
189 |
Log.severe(InformationCoreMessages.Diagnostic_6 + corpus.getName() + InformationCoreMessages.Diagnostic_8);
|
|
190 |
Log.printStackTrace(e);
|
|
191 |
return;
|
|
192 |
}
|
|
193 |
}
|
|
194 |
|
|
195 |
/**
|
|
196 |
* step2.
|
|
197 |
* randomly position are choose
|
|
198 |
*/
|
|
199 |
public void stepLexicalProperties()
|
|
200 |
{
|
|
201 |
AbstractCqiClient cqiClient = CorpusManager.getCorpusManager().getCqiClient();
|
|
202 |
for (Property p : properties) {
|
|
203 |
//int size;
|
|
204 |
try {
|
|
205 |
//size = CorpusManager.getCorpusManager().getCqiClient().lexiconSize(p.getQualifiedName());
|
|
206 |
//propertiesCounts.put(p.getName(), size);
|
|
207 |
//List<String> list = Arrays.asList(corpus.getLexicon(p).getForms(this.maxvalue));
|
|
208 |
|
|
209 |
int[] positions = new int[Math.min(maxPropertiesToDisplay, T)];
|
|
210 |
for(int i = 0 ; i < Math.min(maxPropertiesToDisplay, T) ; i++) {
|
|
211 |
positions[i] = i;
|
|
212 |
}
|
|
213 |
|
|
214 |
//ArrayList<String> values = new ArrayList<String>();
|
|
215 |
LinkedHashSet<String> values = new LinkedHashSet<String>(cqiClient.getSingleData(p, positions));
|
|
216 |
propertiesValues.put(p.getName(), new ArrayList<String>(values));
|
|
217 |
} catch (Exception e) {
|
|
218 |
Log.printStackTrace(e);
|
|
219 |
}
|
|
220 |
}
|
|
221 |
}
|
|
222 |
|
|
223 |
/**
|
|
224 |
* Step structural units.
|
|
225 |
*/
|
|
226 |
public void stepStructuralUnits()
|
|
227 |
{
|
|
228 |
try {
|
|
229 |
structures = corpus.getOrderedStructuralUnits();
|
|
230 |
} catch (CqiClientException e) {
|
|
231 |
Log.severe(InformationCoreMessages.Diagnostic_1 + corpus.getName() + InformationCoreMessages.Diagnostic_11);
|
|
232 |
Log.printStackTrace(e);
|
|
233 |
return;
|
|
234 |
}
|
|
235 |
NStructures = structures.size();
|
|
236 |
for (StructuralUnit su : structures) { // for each structural unit
|
|
237 |
internalArchitecture.add(su.getName());
|
|
238 |
hierarchieCounts.put(su.getName(), -1);
|
|
239 |
internalArchitectureProperties.put(su.getName(), new HashMap<String, List<String>>());
|
|
240 |
internalArchitecturePropertiesCounts.put(su.getName(), new HashMap<String, Integer>());
|
|
241 |
|
|
242 |
if (su.getProperties().size() > 0) {
|
|
243 |
for (StructuralUnitProperty sup : su.getProperties()) {// for each of its property
|
|
244 |
try {
|
|
245 |
List<String> allvalues = sup.getValues(corpus);
|
|
246 |
internalArchitectureProperties.get(su.getName()).put(sup.getName(), allvalues.subList(0, Math.min(allvalues.size(), maxPropertiesToDisplay)));
|
|
247 |
internalArchitecturePropertiesCounts.get(su.getName()).put(sup.getName(), allvalues.size());
|
|
248 |
} catch (Exception e) {
|
|
249 |
ArrayList<String> list = new ArrayList<String>(1);
|
|
250 |
list.add(InformationCoreMessages.Diagnostic_2);
|
|
251 |
internalArchitectureProperties.get(su.getName()).put(sup.getName(), list);
|
|
252 |
internalArchitecturePropertiesCounts.get(su.getName()).put(sup.getName(), -1);
|
|
253 |
}
|
|
254 |
}
|
|
255 |
}
|
|
256 |
}
|
|
257 |
}
|
|
258 |
|
|
259 |
// /**
|
|
260 |
// * Instantiates a new diagnostic.
|
|
261 |
// *
|
|
262 |
// * @param partition the partition
|
|
263 |
// * @throws CqiClientException the cqi client exception
|
|
264 |
// */
|
|
265 |
// public Information(Partition partition) throws CqiClientException {
|
|
266 |
// corpus = partition.getCorpus();
|
|
267 |
// this.partition = partition;
|
|
268 |
// int T = corpus.getSize();
|
|
269 |
// System.out.println(InformationCoreMessages.Diagnostic_12);
|
|
270 |
// int nbParts = partition.getNPart();
|
|
271 |
// List<String> partNames = partition.getPartNames();
|
|
272 |
// List<Integer> partSizes = new ArrayList<Integer>();
|
|
273 |
// List<List<Property>> partProperties = new ArrayList<List<Property>>();
|
|
274 |
// List<String> partQueries = new ArrayList<String>();
|
|
275 |
// for (Part part : partition.getParts()) {
|
|
276 |
// partSizes.add(part.getSize());
|
|
277 |
// partProperties.add(part.getProperties());
|
|
278 |
// partQueries.add(part.getQuery().getQueryString());
|
|
279 |
// }
|
|
280 |
// }
|
|
281 |
|
|
282 |
/**
|
|
283 |
* dump the results of the diagnostic in a String containing HTML code.
|
|
284 |
*
|
|
285 |
* @return the string
|
|
286 |
*/
|
|
287 |
public String htmlDump() {
|
|
288 |
StringBuffer buff = new StringBuffer();
|
|
289 |
|
|
290 |
buff.append("<? version=\"1.0\" encoding=\"UTF-8\"?>\n" + //$NON-NLS-1$
|
|
291 |
"<html>" + //$NON-NLS-1$
|
|
292 |
"<head>" + //$NON-NLS-1$
|
|
293 |
"<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\"/>" + //$NON-NLS-1$
|
|
294 |
"</head>" + //$NON-NLS-1$
|
|
295 |
"<body>\n"); //$NON-NLS-1$
|
|
296 |
buff.append("<h2 style'font-family:\"Arial\";'>"+InformationCoreMessages.Diagnostic_3 + this.corpus.getName() + "</h2>\n"); //$NON-NLS-2$ //$NON-NLS-1$
|
|
297 |
|
|
298 |
if (corpus.getDescription() != null && corpus.getDescription().length() > 0) {
|
|
299 |
buff.append("<p>"+corpus.getDescription()+"</p>"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
300 |
}
|
|
301 |
|
|
302 |
buff.append("<h3 style'font-family:\"Arial\";'>"+InformationCoreMessages.Diagnostic_5+ "</h3>\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
303 |
|
|
304 |
|
|
305 |
|
|
306 |
// counts
|
|
307 |
buff.append("<ul>\n"); //$NON-NLS-1$
|
|
308 |
buff.append("<li>" + InformationCoreMessages.Diagnostic_7 + T + "</li>\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
309 |
buff
|
|
310 |
.append("<li>" + InformationCoreMessages.Diagnostic_9 + (NProperties - 1) + "</li>\n"); //$NON-NLS-2$ //$NON-NLS-1$
|
|
311 |
buff.append("<li>" + InformationCoreMessages.Diagnostic_16 + NStructures + "</li>\n"); //$NON-NLS-2$ //$NON-NLS-1$
|
|
312 |
buff.append("</ul>\n"); //$NON-NLS-1$
|
|
313 |
|
|
314 |
// Propriétés d'occurrences
|
|
315 |
buff.append("<h3 style'font-family:\"Arial\";'>"+NLS.bind(InformationCoreMessages.Diagnostic_14, maxPropertiesToDisplay) + "</h3>\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
316 |
ArrayList<String> properties = new ArrayList<String>(propertiesValues.keySet());
|
|
317 |
Collections.sort(properties);
|
|
318 |
buff.append("<ul>\n"); //$NON-NLS-1$
|
|
319 |
for (String s : properties) {
|
|
320 |
if (!s.equals("id")) //$NON-NLS-1$
|
|
321 |
{
|
|
322 |
//buff.append("<li> " + s + " (" + propertiesCounts.get(s) + ") : "); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
323 |
buff.append("<li> " + s + " : "); //$NON-NLS-1$ //$NON-NLS-2$
|
|
324 |
for (String v : propertiesValues.get(s))
|
|
325 |
buff.append(v + ", "); //$NON-NLS-1$
|
|
326 |
buff.append("..."); //$NON-NLS-1$
|
|
327 |
buff.append("</li>\n"); //$NON-NLS-1$
|
|
328 |
}
|
|
329 |
}
|
|
330 |
buff.append("</ul>\n"); //$NON-NLS-1$
|
|
331 |
|
|
332 |
// Propriété de structures
|
|
333 |
buff.append("<h3 style'font-family:\"Arial\";'>"+NLS.bind(InformationCoreMessages.Diagnostic_22, maxPropertiesToDisplay)+"</h3>\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
334 |
buff.append("<ul>\n"); //$NON-NLS-1$
|
|
335 |
for (String s : internalArchitecture) {
|
|
336 |
StringBuffer subbuffer = new StringBuffer();
|
|
337 |
if (s.equals("txmcorpus")) //$NON-NLS-1$
|
|
338 |
continue;//ignore the txmcorpus structure
|
|
339 |
|
|
340 |
properties = new ArrayList<String>(internalArchitectureProperties.get(s).keySet());
|
|
341 |
|
|
342 |
subbuffer.append("<li> " + s + "\n<ul>\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
343 |
if (internalArchitectureProperties.get(s).keySet().size() == 0)
|
|
344 |
subbuffer.append("<li>"+InformationCoreMessages.Diagnostic_26+"</li>\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
345 |
|
|
346 |
|
|
347 |
//System.out.println("struct: "+s+" >> "+properties);
|
|
348 |
Collections.sort(properties);
|
|
349 |
//System.out.println("struct: "+s+" >sort> "+properties);
|
|
350 |
int nbOfValues = 0;
|
|
351 |
for (String ps : properties) {
|
|
352 |
if (!ps.equals("project") && !ps.equals("base") ) //$NON-NLS-1$ //$NON-NLS-2$
|
|
353 |
{
|
|
354 |
//System.out.println("struct: "+s+" >> "+ps);
|
|
355 |
int valuecount = internalArchitecturePropertiesCounts.get(s).get(ps);
|
|
356 |
ArrayList<String> values = new ArrayList<String>(internalArchitectureProperties.get(s).get(ps));
|
|
357 |
subbuffer.append("<li> " + ps + " (" + valuecount + ") = "); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
358 |
//System.out.println("VALUES: "+values);
|
|
359 |
Collections.sort(values);
|
|
360 |
for (int i = 0 ; i < values.size() ; i++) {
|
|
361 |
nbOfValues++;
|
|
362 |
String psv = values.get(i);
|
|
363 |
if (psv != null && psv.length() > 0)
|
|
364 |
subbuffer.append(psv.replace("<", "<")); //$NON-NLS-1$ //$NON-NLS-2$
|
|
365 |
else
|
|
366 |
subbuffer.append("\"\""); //$NON-NLS-1$
|
|
367 |
|
|
368 |
if (i == values.size() -1)
|
|
369 |
subbuffer.append("."); //$NON-NLS-1$
|
|
370 |
else
|
|
371 |
subbuffer.append(", "); //$NON-NLS-1$
|
|
372 |
|
|
373 |
if (i >= maxPropertiesToDisplay) {
|
|
374 |
subbuffer.append("..."); //$NON-NLS-1$
|
|
375 |
break;
|
|
376 |
}
|
|
377 |
}
|
|
378 |
subbuffer.append("</li>\n"); //$NON-NLS-1$
|
|
379 |
}
|
|
380 |
}
|
|
381 |
subbuffer.append("</ul>\n</li>\n"); //$NON-NLS-1$
|
|
382 |
if (nbOfValues > 0) {
|
|
383 |
buff.append(subbuffer);
|
|
384 |
}
|
|
385 |
}
|
|
386 |
buff.append("</ul>\n"); //$NON-NLS-1$
|
|
387 |
buff.append("</body>\n"); //$NON-NLS-1$
|
|
388 |
buff.append("</html>\n"); //$NON-NLS-1$
|
|
389 |
|
|
390 |
return buff.toString();
|
|
391 |
}
|
|
392 |
|
|
393 |
|
|
394 |
|
|
395 |
/**
|
|
396 |
* dump the result in the console.
|
|
397 |
*
|
|
398 |
* @param maxvalue the maxvalue
|
|
399 |
* @return the string
|
|
400 |
*/
|
|
401 |
public String dump(int maxvalue) {
|
|
402 |
StringBuffer buff = new StringBuffer();
|
|
403 |
|
|
404 |
buff.append(InformationCoreMessages.Diagnostic_40);
|
|
405 |
buff.append(InformationCoreMessages.Diagnostic_41 + String.format(Locale.FRANCE,"%,d", T) + "\n"); //$NON-NLS-1$
|
|
406 |
buff.append(InformationCoreMessages.Diagnostic_43 + String.format("%,d", NProperties) + "\n"); //$NON-NLS-1$
|
|
407 |
/*for (String s : propertiesCounts.keySet())
|
|
408 |
if (!s.equals("id")) //$NON-NLS-1$
|
|
409 |
buff.append(" - " + s + " " + propertiesCounts.get(s) + "\n"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
410 |
*/
|
|
411 |
buff.append(InformationCoreMessages.Diagnostic_49 + String.format("%,d", NStructures) + "\n"); //$NON-NLS-1$
|
|
412 |
for (String s : structuresCounts.keySet())
|
|
413 |
buff.append(" - " + s + "\n");//+structuresCounts.get(s)+"\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
414 |
|
|
415 |
/*
|
|
416 |
* buff.append("* Structural Units counts\n"); for(String s :
|
|
417 |
* internalArchitecture)
|
|
418 |
* buff.append(" - "+s+" "+hierarchieCounts.get(s)+"\n");
|
|
419 |
*/
|
|
420 |
buff.append(InformationCoreMessages.Diagnostic_0);
|
|
421 |
for (String s : internalArchitecture) {
|
|
422 |
buff.append(" - " + s + "\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
423 |
for (String ps : internalArchitectureProperties.get(s).keySet()) {
|
|
424 |
if (!(s.equals("text") && (ps.equals("project") || ps.equals("base") || ps.equals("id")))) //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
|
|
425 |
{
|
|
426 |
buff
|
|
427 |
.append(" - " + ps + "(" + internalArchitectureProperties.get(s).get(ps).size() + ") = "); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
428 |
|
|
429 |
int valuecount = 0;
|
|
430 |
for (String psv : internalArchitectureProperties.get(s)
|
|
431 |
.get(ps)) {
|
|
432 |
buff.append(psv);
|
|
433 |
if (valuecount++ >= maxvalue) {
|
|
434 |
buff.append("..."); //$NON-NLS-1$
|
|
435 |
break;
|
|
436 |
}
|
|
437 |
if (valuecount < internalArchitectureProperties.get(s)
|
|
438 |
.get(ps).size())
|
|
439 |
buff.append(", "); //$NON-NLS-1$
|
|
440 |
}
|
|
441 |
|
|
442 |
buff.append("\n"); //$NON-NLS-1$
|
|
443 |
}
|
|
444 |
}
|
|
445 |
}
|
|
446 |
|
|
447 |
return buff.toString();
|
|
448 |
}
|
|
449 |
|
|
450 |
/**
|
|
451 |
* dump the result in the console with maxvalue=30.
|
|
452 |
*
|
|
453 |
* @return the string
|
|
454 |
*/
|
|
455 |
public String dump() {
|
|
456 |
return dump(30);
|
|
457 |
}
|
|
458 |
|
|
459 |
/**
|
|
460 |
* Gets the name.
|
|
461 |
*
|
|
462 |
* @return the name
|
|
463 |
*/
|
|
464 |
public String getName() {
|
|
465 |
return corpus.getName();
|
|
466 |
}
|
|
467 |
|
|
468 |
/**
|
|
469 |
* Gets the base.
|
|
470 |
*
|
|
471 |
* @return the base
|
|
472 |
*/
|
|
473 |
public Base getBase() {
|
|
474 |
return corpus.getBase();
|
|
475 |
}
|
|
476 |
|
|
477 |
/**
|
|
478 |
* Gets the t.
|
|
479 |
*
|
|
480 |
* @return the t
|
|
481 |
*/
|
|
482 |
public int getT() {
|
|
483 |
return T;
|
|
484 |
}
|
|
485 |
|
|
486 |
/**
|
|
487 |
* Gets the n properties.
|
|
488 |
*
|
|
489 |
* @return the n properties
|
|
490 |
*/
|
|
491 |
public int getNProperties() {
|
|
492 |
return NProperties;
|
|
493 |
}
|
|
494 |
|
|
495 |
/**
|
|
496 |
* Gets the internal architecture.
|
|
497 |
*
|
|
498 |
* @return the internal architecture
|
|
499 |
*/
|
|
500 |
public List<String> getInternalArchitecture() {
|
|
501 |
return internalArchitecture;
|
|
502 |
}
|
|
503 |
|
|
504 |
/**
|
|
505 |
* Gets the n structures.
|
|
506 |
*
|
|
507 |
* @return the n structures
|
|
508 |
*/
|
|
509 |
public int getNStructures() {
|
|
510 |
return NStructures;
|
|
511 |
}
|
|
512 |
|
|
513 |
/**
|
|
514 |
* Gets the structures counts.
|
|
515 |
*
|
|
516 |
* @return the structures counts
|
|
517 |
*/
|
|
518 |
public HashMap<String, Integer> getStructuresCounts() {
|
|
519 |
return structuresCounts;
|
|
520 |
}
|
|
521 |
|
|
522 |
/**
|
|
523 |
* Gets the annotations types.
|
|
524 |
*
|
|
525 |
* @return the annotations types
|
|
526 |
*/
|
|
527 |
public HashMap<String, Integer> getAnnotationsTypes() {
|
|
528 |
return annotationsTypes;
|
|
529 |
}
|
|
530 |
|
|
531 |
/**
|
|
532 |
* Gets the annotations origins.
|
|
533 |
*
|
|
534 |
* @return the annotations origins
|
|
535 |
*/
|
|
536 |
public HashMap<String, Integer> getAnnotationsOrigins() {
|
|
537 |
return annotationsOrigins;
|
|
538 |
}
|
|
539 |
|
|
540 |
/**
|
|
541 |
* Gets the import description.
|
|
542 |
*
|
|
543 |
* @return the import description
|
|
544 |
*/
|
|
545 |
public String getImportDescription() {
|
|
546 |
return importDescription;
|
|
547 |
}
|
|
548 |
|
|
549 |
/**
|
|
550 |
* dump the diagnostic in a text file.
|
|
551 |
*
|
|
552 |
* @param outfile the outfile
|
|
553 |
* @param maxvalue the maxvalue
|
|
554 |
* @return true, if successful
|
|
555 |
*/
|
|
556 |
public boolean toTxt(File outfile, String encoding, int maxvalue) {
|
|
557 |
BufferedWriter writer = null;
|
|
558 |
try {
|
|
559 |
writer = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(outfile), encoding));
|
|
560 |
} catch (UnsupportedEncodingException e1) {
|
|
561 |
org.txm.utils.logger.Log.printStackTrace(e1);
|
|
562 |
return false;
|
|
563 |
} catch (FileNotFoundException e1) {
|
|
564 |
org.txm.utils.logger.Log.printStackTrace(e1);
|
|
565 |
return false;
|
|
566 |
}
|
|
567 |
|
|
568 |
try {
|
|
569 |
writer.write(this.dump(maxvalue));
|
|
570 |
writer.close();
|
|
571 |
} catch (IOException e) {
|
|
572 |
org.txm.utils.logger.Log.printStackTrace(e);
|
|
573 |
return false;
|
|
574 |
}
|
|
575 |
|
|
576 |
return true;
|
|
577 |
}
|
|
578 |
|
|
579 |
/**
|
|
580 |
* dump the diagnostic in a HTML file.
|
|
581 |
*
|
|
582 |
* @param outfile the outfile
|
|
583 |
* @return true, if successful
|
|
584 |
*/
|
|
585 |
public boolean toHTML(File outfile) {
|
|
586 |
OutputStreamWriter writer;
|
|
587 |
try {
|
|
588 |
writer = new OutputStreamWriter(new FileOutputStream(outfile), "UTF-8"); //$NON-NLS-1$
|
|
589 |
writer.write(this.htmlDump());
|
|
590 |
writer.close();
|
|
591 |
} catch (IOException e) {
|
|
592 |
org.txm.utils.logger.Log.printStackTrace(e);
|
|
593 |
return false;
|
|
594 |
}
|
|
595 |
|
|
596 |
return true;
|
|
597 |
}
|
|
598 |
|
|
599 |
/**
|
|
600 |
* dump the diagnostic in a text file, maxvalue = 100.
|
|
601 |
*
|
|
602 |
* @param outfile the outfile
|
|
603 |
* @return true, if successful
|
|
604 |
*/
|
|
605 |
public boolean toTxt(File outfile) {
|
|
606 |
return toTxt(outfile, "UTF-8", 100); //$NON-NLS-1$
|
|
607 |
}
|
|
608 |
|
|
609 |
@Override
|
|
610 |
public void clean() {
|
|
611 |
// TODO Auto-generated method stub
|
|
612 |
}
|
|
613 |
|
|
614 |
@Override
|
|
615 |
public boolean toTxt(File outfile, String encoding, String colseparator, String txtseparator) throws Exception {
|
|
616 |
return toTxt(outfile, encoding, 100);
|
|
617 |
}
|
|
618 |
|
|
619 |
@Override
|
|
620 |
public String getSimpleName() {
|
|
621 |
return "Informations";
|
|
622 |
}
|
|
623 |
|
|
624 |
@Override
|
|
625 |
public String getDetails() {
|
|
626 |
return "" + T;
|
|
627 |
}
|
|
628 |
|
|
629 |
@Override
|
|
630 |
protected boolean _compute() throws Exception {
|
|
631 |
|
|
632 |
this.beginTask("Computing informations...", 3);
|
|
633 |
this.stepGeneralInfos();
|
|
634 |
this.worked(1);
|
|
635 |
|
|
636 |
this.stepLexicalProperties();
|
|
637 |
this.worked(1);
|
|
638 |
|
|
639 |
this.stepStructuralUnits();
|
|
640 |
this.worked(1);
|
|
641 |
|
|
642 |
String txmhome = TXMPreferences.getString(TBXPreferences.USER_TXM_HOME, TBXPreferences.PREFERENCES_NODE);
|
|
643 |
String filename = corpus.getParent().getName() + "-" + corpus.getName(); //$NON-NLS-1$
|
|
644 |
filename = AsciiUtils.buildId(filename);
|
|
645 |
file = new File(txmhome, "results/" + filename + "-infos.html"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
646 |
|
|
647 |
this.toHTML(file);
|
|
648 |
return true;
|
|
649 |
}
|
|
650 |
|
|
651 |
/**
|
|
652 |
* Gets the HTML file.
|
|
653 |
* @return the HTML file
|
|
654 |
*/
|
|
655 |
public File getHTMLFile() {
|
|
656 |
return file;
|
|
657 |
}
|
|
658 |
|
|
659 |
@Override
|
|
660 |
public boolean canCompute() {
|
|
661 |
return corpus != null;
|
|
662 |
}
|
|
663 |
|
|
664 |
@Override
|
|
665 |
public boolean setParameters(TXMParameters parameters) {
|
|
666 |
this.maxPropertiesToDisplay = parameters.getInteger(InformationPreferences.MAX_VALUE);
|
|
667 |
return true;
|
|
668 |
}
|
|
669 |
|
|
670 |
@Override
|
|
671 |
public boolean loadParameters() {
|
|
672 |
// nothing to do
|
|
673 |
return true;
|
|
674 |
}
|
|
675 |
|
|
676 |
|
|
677 |
@Override
|
|
678 |
public boolean saveParameters() {
|
|
679 |
// nothing to do
|
|
680 |
return true;
|
|
681 |
}
|
|
682 |
}
|
0 |
683 |
|