root / tmp / org.txm.groovy.core / src / groovy / org / txm / scripts / importer / hyperbase / compiler.groovy @ 1688
History | View | Annotate | Download (6.5 kB)
1 | 321 | mdecorde | // Copyright © 2010-2013 ENS de Lyon.
|
---|---|---|---|
2 | 321 | mdecorde | // Copyright © 2007-2010 ENS de Lyon, CNRS, INRP, University of
|
3 | 321 | mdecorde | // Lyon 2, University of Franche-Comté, University of Nice
|
4 | 321 | mdecorde | // Sophia Antipolis, University of Paris 3.
|
5 | 321 | mdecorde | //
|
6 | 321 | mdecorde | // The TXM platform is free software: you can redistribute it
|
7 | 321 | mdecorde | // and/or modify it under the terms of the GNU General Public
|
8 | 321 | mdecorde | // License as published by the Free Software Foundation,
|
9 | 321 | mdecorde | // either version 2 of the License, or (at your option) any
|
10 | 321 | mdecorde | // later version.
|
11 | 321 | mdecorde | //
|
12 | 321 | mdecorde | // The TXM platform is distributed in the hope that it will be
|
13 | 321 | mdecorde | // useful, but WITHOUT ANY WARRANTY; without even the implied
|
14 | 321 | mdecorde | // warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
|
15 | 321 | mdecorde | // PURPOSE. See the GNU General Public License for more
|
16 | 321 | mdecorde | // details.
|
17 | 321 | mdecorde | //
|
18 | 321 | mdecorde | // You should have received a copy of the GNU General
|
19 | 321 | mdecorde | // Public License along with the TXM platform. If not, see
|
20 | 321 | mdecorde | // http://www.gnu.org/licenses.
|
21 | 321 | mdecorde | //
|
22 | 321 | mdecorde | //
|
23 | 321 | mdecorde | //
|
24 | 479 | mdecorde | // $LastChangedDate: 2016-05-26 17:42:36 +0200 (jeu. 26 mai 2016) $
|
25 | 321 | mdecorde | // $LastChangedRevision: 3219 $
|
26 | 321 | mdecorde | // $LastChangedBy: mdecorde $
|
27 | 321 | mdecorde | //
|
28 | 986 | mdecorde | package org.txm.scripts.importer.hyperbase;
|
29 | 321 | mdecorde | |
30 | 1000 | mdecorde | import org.txm.importer.cwb.CwbEncode |
31 | 1000 | mdecorde | import org.txm.importer.cwb.CwbMakeAll |
32 | 986 | mdecorde | import org.txm.scripts.importer.*; |
33 | 321 | mdecorde | import org.txm.scripts.*; |
34 | 1000 | mdecorde | import org.txm.importer.scripts.xmltxm.*; |
35 | 321 | mdecorde | import org.txm.utils.treetagger.TreeTagger; |
36 | 1115 | mdecorde | import org.txm.objects.* |
37 | 1115 | mdecorde | import org.txm.searchengine.cqp.corpus.* |
38 | 321 | mdecorde | import javax.xml.stream.*; |
39 | 321 | mdecorde | import java.net.URL; |
40 | 321 | mdecorde | import java.io.File; |
41 | 321 | mdecorde | import java.util.HashMap; |
42 | 321 | mdecorde | import java.util.List; |
43 | 321 | mdecorde | |
44 | 321 | mdecorde | // TODO: Auto-generated Javadoc
|
45 | 321 | mdecorde | /**
|
46 | 321 | mdecorde | * The Class compiler.
|
47 | 321 | mdecorde | */
|
48 | 321 | mdecorde | class compiler |
49 | 321 | mdecorde | { |
50 | 321 | mdecorde | |
51 | 321 | mdecorde | /** The debug. */
|
52 | 321 | mdecorde | private boolean debug= false; |
53 | 321 | mdecorde | |
54 | 321 | mdecorde | /** The input data. */
|
55 | 321 | mdecorde | private def inputData; |
56 | 321 | mdecorde | |
57 | 321 | mdecorde | /** The factory. */
|
58 | 321 | mdecorde | private def factory; |
59 | 321 | mdecorde | |
60 | 321 | mdecorde | /** The parser. */
|
61 | 321 | mdecorde | private XMLStreamReader parser;
|
62 | 321 | mdecorde | |
63 | 321 | mdecorde | /** The dir. */
|
64 | 321 | mdecorde | private def dir; |
65 | 321 | mdecorde | |
66 | 321 | mdecorde | /** The output. */
|
67 | 321 | mdecorde | private def output; |
68 | 321 | mdecorde | |
69 | 321 | mdecorde | /** The url. */
|
70 | 321 | mdecorde | private def url; |
71 | 321 | mdecorde | |
72 | 321 | mdecorde | /** The anahash. */
|
73 | 321 | mdecorde | private HashMap<String,String> anahash = new HashMap<String,String>() ; |
74 | 321 | mdecorde | |
75 | 321 | mdecorde | /** The text. */
|
76 | 321 | mdecorde | String text=""; |
77 | 321 | mdecorde | |
78 | 321 | mdecorde | /** The base. */
|
79 | 321 | mdecorde | String base=""; |
80 | 321 | mdecorde | |
81 | 321 | mdecorde | /** The text attributes. */
|
82 | 321 | mdecorde | String[] textAttributes = null; |
83 | 321 | mdecorde | |
84 | 321 | mdecorde | /** The lang. */
|
85 | 321 | mdecorde | private String lang ="fr"; |
86 | 321 | mdecorde | |
87 | 321 | mdecorde | /**
|
88 | 321 | mdecorde | * initialize.
|
89 | 321 | mdecorde | *
|
90 | 321 | mdecorde | */
|
91 | 321 | mdecorde | public compiler(){}
|
92 | 321 | mdecorde | |
93 | 321 | mdecorde | /**
|
94 | 321 | mdecorde | * set the language of the corpus.
|
95 | 321 | mdecorde | *
|
96 | 321 | mdecorde | * @param lang the lang
|
97 | 321 | mdecorde | * @return the java.lang. object
|
98 | 321 | mdecorde | */
|
99 | 321 | mdecorde | public setLang(String lang) |
100 | 321 | mdecorde | { |
101 | 321 | mdecorde | this.lang = lang;
|
102 | 321 | mdecorde | } |
103 | 321 | mdecorde | |
104 | 321 | mdecorde | /** The annotation success. */
|
105 | 321 | mdecorde | boolean annotationSuccess = false; |
106 | 321 | mdecorde | |
107 | 321 | mdecorde | /**
|
108 | 321 | mdecorde | * Sets the annotation success.
|
109 | 321 | mdecorde | *
|
110 | 321 | mdecorde | * @param value the value
|
111 | 321 | mdecorde | * @return the java.lang. object
|
112 | 321 | mdecorde | */
|
113 | 321 | mdecorde | public setAnnotationSuccess(boolean value) |
114 | 321 | mdecorde | { |
115 | 321 | mdecorde | this.annotationSuccess = value
|
116 | 321 | mdecorde | } |
117 | 321 | mdecorde | |
118 | 321 | mdecorde | /**
|
119 | 321 | mdecorde | * Creates the output.
|
120 | 321 | mdecorde | *
|
121 | 321 | mdecorde | * @param f the f
|
122 | 321 | mdecorde | * @return true, if successful
|
123 | 321 | mdecorde | */
|
124 | 321 | mdecorde | private boolean createOutput(File f){ |
125 | 321 | mdecorde | try {
|
126 | 321 | mdecorde | output = new OutputStreamWriter(new FileOutputStream(f,f.exists()) , "UTF-8"); |
127 | 321 | mdecorde | return true; |
128 | 321 | mdecorde | } catch (Exception e) { |
129 | 321 | mdecorde | System.err.println(e);
|
130 | 321 | mdecorde | |
131 | 321 | mdecorde | return false; |
132 | 321 | mdecorde | } |
133 | 321 | mdecorde | } |
134 | 321 | mdecorde | |
135 | 321 | mdecorde | /**
|
136 | 321 | mdecorde | * Go to text.
|
137 | 321 | mdecorde | */
|
138 | 321 | mdecorde | private void GoToText() |
139 | 321 | mdecorde | { |
140 | 321 | mdecorde | for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) |
141 | 321 | mdecorde | { |
142 | 321 | mdecorde | if(event == XMLStreamConstants.END_ELEMENT)
|
143 | 321 | mdecorde | if(parser.getLocalName().equals("teiHeader")) |
144 | 321 | mdecorde | return;
|
145 | 321 | mdecorde | } |
146 | 321 | mdecorde | } |
147 | 321 | mdecorde | |
148 | 321 | mdecorde | /**
|
149 | 321 | mdecorde | * Run.
|
150 | 321 | mdecorde | *
|
151 | 321 | mdecorde | * @param rootDirFile the root dir file
|
152 | 321 | mdecorde | * @param basename the basename
|
153 | 321 | mdecorde | * @return true, if successful
|
154 | 321 | mdecorde | */
|
155 | 1115 | mdecorde | public boolean run(Project project, File binDir, File txmDir, String corpusname) |
156 | 321 | mdecorde | { |
157 | 321 | mdecorde | String rootDir = binDir.getAbsolutePath();
|
158 | 321 | mdecorde | |
159 | 714 | mdecorde | if (!(CwbEncode.isExecutableAvailable() && CwbMakeAll.isExecutableAvailable())) {
|
160 | 714 | mdecorde | println ("Error: CWB executables not well set.")
|
161 | 321 | mdecorde | return false; |
162 | 321 | mdecorde | } |
163 | 1115 | mdecorde | CorpusBuild corpus = project.getCorpusBuild(project.getName()); |
164 | 1115 | mdecorde | if (corpus != null) { |
165 | 1115 | mdecorde | //println "CLEAN PREVIOUS CORPUS"
|
166 | 1115 | mdecorde | corpus.delete(); // remove old files
|
167 | 321 | mdecorde | } |
168 | 1115 | mdecorde | |
169 | 1115 | mdecorde | // make new one
|
170 | 1115 | mdecorde | corpus = new MainCorpus(project);
|
171 | 1115 | mdecorde | corpus.setID(project.getName()); |
172 | 1115 | mdecorde | corpus.setName(project.getName()); |
173 | 1615 | mdecorde | corpus.setDescription("Built with the Hyperbase import module");
|
174 | 1115 | mdecorde | |
175 | 803 | mdecorde | File cqpFile = new File(binDir,"cqp/"+corpusname+".cqp"); |
176 | 1395 | mdecorde | cqpFile.delete() |
177 | 1115 | mdecorde | new File(binDir,"cqp").mkdirs() |
178 | 1115 | mdecorde | new File(binDir,"data").mkdirs() |
179 | 1115 | mdecorde | new File(binDir,"registry").mkdirs() |
180 | 321 | mdecorde | |
181 | 321 | mdecorde | //start corpus
|
182 | 803 | mdecorde | if (createOutput(cqpFile)) {
|
183 | 321 | mdecorde | output.write("<txmcorpus lang=\""+lang+"\">\n"); |
184 | 321 | mdecorde | output.close(); |
185 | 321 | mdecorde | } |
186 | 321 | mdecorde | |
187 | 321 | mdecorde | String textid=""; |
188 | 321 | mdecorde | int counttext =0; |
189 | 321 | mdecorde | List<File> files = txmDir.listFiles(); |
190 | 321 | mdecorde | Collections.sort(files);
|
191 | 804 | mdecorde | //1- Transform into CQP file
|
192 | 804 | mdecorde | XMLTXM2CQP cqpbuilder = null;
|
193 | 321 | mdecorde | println("Compiling "+files.size()+" files") |
194 | 321 | mdecorde | for (File f : files) { |
195 | 321 | mdecorde | print "."
|
196 | 321 | mdecorde | counttext++; |
197 | 321 | mdecorde | if (!f.exists()) {
|
198 | 321 | mdecorde | println("file "+f+ " does not exists") |
199 | 321 | mdecorde | } |
200 | 321 | mdecorde | else {
|
201 | 804 | mdecorde | cqpbuilder = new XMLTXM2CQP(f.toURI().toURL());
|
202 | 321 | mdecorde | String txtname = f.getName().substring(0,f.getName().length()-4); |
203 | 803 | mdecorde | cqpbuilder.setTextInfo(txtname, corpusname, "project");
|
204 | 321 | mdecorde | |
205 | 803 | mdecorde | cqpbuilder.setBalisesToKeep(["text","p","s"]); |
206 | 803 | mdecorde | cqpbuilder.setSendToPAttributes(["s":["n"]]); |
207 | 803 | mdecorde | cqpbuilder.setLang(lang); |
208 | 803 | mdecorde | if (!cqpbuilder.transformFile(cqpFile)) {
|
209 | 321 | mdecorde | println("Failed to compile "+f)
|
210 | 321 | mdecorde | } |
211 | 321 | mdecorde | } |
212 | 321 | mdecorde | } |
213 | 321 | mdecorde | println ""
|
214 | 321 | mdecorde | |
215 | 321 | mdecorde | //end corpus
|
216 | 803 | mdecorde | if (createOutput(cqpFile)) {
|
217 | 321 | mdecorde | output.write("</txmcorpus>\n");
|
218 | 321 | mdecorde | output.close(); |
219 | 321 | mdecorde | } |
220 | 321 | mdecorde | |
221 | 803 | mdecorde | if (cqpbuilder == null) return false; |
222 | 321 | mdecorde | |
223 | 321 | mdecorde | //2- Import into CWB
|
224 | 321 | mdecorde | def outDir =rootDir;
|
225 | 321 | mdecorde | |
226 | 321 | mdecorde | CwbEncode cwbEn = new CwbEncode();
|
227 | 321 | mdecorde | cwbEn.setDebug(debug); |
228 | 321 | mdecorde | CwbMakeAll cwbMa = new CwbMakeAll();
|
229 | 321 | mdecorde | cwbMa.setDebug(debug); |
230 | 321 | mdecorde | |
231 | 803 | mdecorde | List<String> pAttributesList = cqpbuilder.getpAttributs(); |
232 | 803 | mdecorde | List<String> sAttributesList = cqpbuilder.getsAttributs(); |
233 | 321 | mdecorde | println "word properties : "+pAttributesList
|
234 | 321 | mdecorde | println "structures : "+sAttributesList
|
235 | 321 | mdecorde | String[] pAttributes = pAttributesList.toArray(new String[pAttributesList.size()]) |
236 | 321 | mdecorde | String[] sAttributes = sAttributesList.toArray(new String[sAttributesList.size()]) |
237 | 321 | mdecorde | |
238 | 321 | mdecorde | try {
|
239 | 321 | mdecorde | String regPath = outDir + "/registry/"+corpusname.toLowerCase(); |
240 | 803 | mdecorde | cwbEn.run(outDir + "/data/$corpusname", outDir + "/cqp/"+corpusname+".cqp", regPath,pAttributes, sAttributes); |
241 | 321 | mdecorde | if (!new File(regPath).exists()) { |
242 | 321 | mdecorde | println "Error: The registry file was not created: $regPath. See https://groupes.renater.fr/wiki/txm-users/public/faq"
|
243 | 321 | mdecorde | return false; |
244 | 321 | mdecorde | } |
245 | 714 | mdecorde | cwbMa.run(corpusname, outDir + "/registry");
|
246 | 321 | mdecorde | |
247 | 321 | mdecorde | } catch (Exception ex) {System.out.println(ex); return false;} |
248 | 321 | mdecorde | |
249 | 321 | mdecorde | return true; |
250 | 321 | mdecorde | } |
251 | 321 | mdecorde | |
252 | 321 | mdecorde | /**
|
253 | 321 | mdecorde | * Sets the debug.
|
254 | 321 | mdecorde | */
|
255 | 321 | mdecorde | public void setDebug() |
256 | 321 | mdecorde | { |
257 | 321 | mdecorde | this.debug = true; |
258 | 321 | mdecorde | } |
259 | 321 | mdecorde | |
260 | 321 | mdecorde | /**
|
261 | 321 | mdecorde | * The main method.
|
262 | 321 | mdecorde | *
|
263 | 321 | mdecorde | * @param args the arguments
|
264 | 321 | mdecorde | */
|
265 | 321 | mdecorde | public static void main(String[] args) |
266 | 321 | mdecorde | { |
267 | 321 | mdecorde | File dir = new File("~/xml/geo"); |
268 | 321 | mdecorde | def c = new compiler(); |
269 | 321 | mdecorde | c.setDebug(); |
270 | 321 | mdecorde | c.setCwbPath("~/TXM/cwb/bin");
|
271 | 321 | mdecorde | c.run(dir,"geo");
|
272 | 321 | mdecorde | } |
273 | 321 | mdecorde | } |