root / tmp / org.txm.groovy.core / src / groovy / org / txm / scripts / importer / discours / compiler.groovy @ 1000
History | View | Annotate | Download (6.8 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.discours
|
29 | 321 | mdecorde | |
30 | 1000 | mdecorde | import org.txm.importer.cwb.BuildCwbEncodeArgsFromTEITXM; |
31 | 1000 | mdecorde | import org.txm.importer.cwb.CwbEncode |
32 | 1000 | mdecorde | import org.txm.importer.cwb.CwbMakeAll |
33 | 986 | mdecorde | import org.txm.scripts.importer.*; |
34 | 321 | mdecorde | import org.txm.scripts.*; |
35 | 1000 | mdecorde | import org.txm.importer.scripts.xmltxm.*; |
36 | 321 | mdecorde | import org.txm.utils.treetagger.TreeTagger; |
37 | 321 | mdecorde | |
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 | /** The input data. */
|
51 | 321 | mdecorde | private def inputData; |
52 | 321 | mdecorde | |
53 | 321 | mdecorde | /** The factory. */
|
54 | 321 | mdecorde | private def factory; |
55 | 321 | mdecorde | |
56 | 321 | mdecorde | /** The parser. */
|
57 | 321 | mdecorde | private XMLStreamReader parser;
|
58 | 321 | mdecorde | |
59 | 321 | mdecorde | /** The dir. */
|
60 | 321 | mdecorde | private def dir; |
61 | 321 | mdecorde | |
62 | 321 | mdecorde | /** The output. */
|
63 | 321 | mdecorde | private def output; |
64 | 321 | mdecorde | |
65 | 321 | mdecorde | /** The url. */
|
66 | 321 | mdecorde | private def url; |
67 | 321 | mdecorde | |
68 | 321 | mdecorde | /** The anahash. */
|
69 | 321 | mdecorde | private HashMap<String,String> anahash = new HashMap<String,String>() ; |
70 | 321 | mdecorde | |
71 | 321 | mdecorde | /** The text. */
|
72 | 321 | mdecorde | String text=""; |
73 | 321 | mdecorde | |
74 | 321 | mdecorde | /** The base. */
|
75 | 321 | mdecorde | String base=""; |
76 | 321 | mdecorde | |
77 | 321 | mdecorde | /** The project. */
|
78 | 321 | mdecorde | String project=""; |
79 | 321 | mdecorde | |
80 | 321 | mdecorde | /** The lang. */
|
81 | 321 | mdecorde | private String lang ="fr"; |
82 | 321 | mdecorde | |
83 | 321 | mdecorde | /** The text attributes. */
|
84 | 321 | mdecorde | static String textAttributes = ""; |
85 | 321 | mdecorde | |
86 | 321 | mdecorde | /**
|
87 | 321 | mdecorde | * initialize.
|
88 | 321 | mdecorde | *
|
89 | 321 | mdecorde | */
|
90 | 321 | mdecorde | |
91 | 321 | mdecorde | public compiler(){
|
92 | 321 | mdecorde | } |
93 | 321 | mdecorde | |
94 | 321 | mdecorde | /**
|
95 | 321 | mdecorde | * Instantiates a new compiler.
|
96 | 321 | mdecorde | *
|
97 | 321 | mdecorde | * @param url the url
|
98 | 321 | mdecorde | * @param text the text
|
99 | 321 | mdecorde | * @param base the base
|
100 | 321 | mdecorde | * @param project the project
|
101 | 321 | mdecorde | */
|
102 | 321 | mdecorde | public compiler(URL url,String text,String base, String project) { |
103 | 321 | mdecorde | this.text = text
|
104 | 321 | mdecorde | this.base = base;
|
105 | 321 | mdecorde | this.project = project;
|
106 | 321 | mdecorde | try {
|
107 | 321 | mdecorde | this.url = url;
|
108 | 321 | mdecorde | inputData = url.openStream(); |
109 | 321 | mdecorde | |
110 | 321 | mdecorde | factory = XMLInputFactory.newInstance(); |
111 | 321 | mdecorde | parser = factory.createXMLStreamReader(inputData); |
112 | 321 | mdecorde | } catch (XMLStreamException ex) {
|
113 | 321 | mdecorde | System.out.println(ex);
|
114 | 321 | mdecorde | }catch (IOException ex) { |
115 | 321 | mdecorde | System.err.println("IOException while parsing "); |
116 | 321 | mdecorde | } |
117 | 321 | mdecorde | } |
118 | 321 | mdecorde | |
119 | 321 | mdecorde | /**
|
120 | 321 | mdecorde | * set the language of the corpus.
|
121 | 321 | mdecorde | *
|
122 | 321 | mdecorde | * @param lang the lang
|
123 | 321 | mdecorde | * @return the java.lang. object
|
124 | 321 | mdecorde | */
|
125 | 321 | mdecorde | public setLang(String lang) |
126 | 321 | mdecorde | { |
127 | 321 | mdecorde | this.lang = lang;
|
128 | 321 | mdecorde | } |
129 | 321 | mdecorde | |
130 | 321 | mdecorde | /**
|
131 | 321 | mdecorde | * Creates the output.
|
132 | 321 | mdecorde | *
|
133 | 321 | mdecorde | * @param f the f
|
134 | 321 | mdecorde | * @return true, if successful
|
135 | 321 | mdecorde | */
|
136 | 321 | mdecorde | private boolean createOutput(File f){ |
137 | 321 | mdecorde | try {
|
138 | 321 | mdecorde | output = new OutputStreamWriter(new FileOutputStream(f,f.exists()) , "UTF-8"); |
139 | 321 | mdecorde | return true; |
140 | 321 | mdecorde | } catch (Exception e) { |
141 | 321 | mdecorde | System.err.println(e);
|
142 | 321 | mdecorde | return false; |
143 | 321 | mdecorde | } |
144 | 321 | mdecorde | } |
145 | 321 | mdecorde | |
146 | 321 | mdecorde | /**
|
147 | 321 | mdecorde | * Go to text.
|
148 | 321 | mdecorde | */
|
149 | 321 | mdecorde | private void GoToText() { |
150 | 321 | mdecorde | for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) { |
151 | 321 | mdecorde | if(event == XMLStreamConstants.END_ELEMENT)
|
152 | 321 | mdecorde | if(parser.getLocalName().equals("teiHeader")) |
153 | 321 | mdecorde | return;
|
154 | 321 | mdecorde | } |
155 | 321 | mdecorde | } |
156 | 321 | mdecorde | |
157 | 321 | mdecorde | /** The debug. */
|
158 | 321 | mdecorde | boolean debug = false; |
159 | 321 | mdecorde | |
160 | 321 | mdecorde | /**
|
161 | 321 | mdecorde | * Sets the debug.
|
162 | 321 | mdecorde | */
|
163 | 321 | mdecorde | public void setDebug() |
164 | 321 | mdecorde | { |
165 | 321 | mdecorde | debug = true;
|
166 | 321 | mdecorde | } |
167 | 321 | mdecorde | |
168 | 321 | mdecorde | /**
|
169 | 321 | mdecorde | * Run.
|
170 | 321 | mdecorde | *
|
171 | 321 | mdecorde | * @param files the files
|
172 | 321 | mdecorde | * @param basename the basename
|
173 | 321 | mdecorde | * @return true, if successful
|
174 | 321 | mdecorde | */
|
175 | 321 | mdecorde | public boolean run(List<File> files, File binDir, File txmDir, String corpusname) { |
176 | 321 | mdecorde | String binDirPath = binDir.getAbsolutePath()+"/"; |
177 | 321 | mdecorde | |
178 | 714 | mdecorde | if (!(CwbEncode.isExecutableAvailable() && CwbMakeAll.isExecutableAvailable())) {
|
179 | 714 | mdecorde | println ("Error: CWB executables not well set.")
|
180 | 321 | mdecorde | return false; |
181 | 321 | mdecorde | } |
182 | 803 | mdecorde | File cqpFile = new File(binDir,"cqp/${corpusname}.cqp"); |
183 | 803 | mdecorde | new File(binDir, "cqp.").deleteDir(); |
184 | 803 | mdecorde | new File(binDir, "cqp").mkdirs(); |
185 | 321 | mdecorde | new File(binDir, "data/$corpusname").deleteDir(); |
186 | 321 | mdecorde | new File(binDir, "data/$corpusname").mkdir(); |
187 | 321 | mdecorde | new File(binDir, "registry").deleteDir(); |
188 | 321 | mdecorde | new File(binDir, "registry").mkdir(); |
189 | 321 | mdecorde | |
190 | 321 | mdecorde | //start corpus
|
191 | 803 | mdecorde | if (createOutput(cqpFile)) {
|
192 | 321 | mdecorde | output.write("<txmcorpus lang=\""+lang+"\">\n"); |
193 | 321 | mdecorde | output.close(); |
194 | 321 | mdecorde | } |
195 | 321 | mdecorde | |
196 | 321 | mdecorde | def second = 0 |
197 | 804 | mdecorde | //1- Transform into CQP file
|
198 | 321 | mdecorde | |
199 | 321 | mdecorde | Collections.sort(files);
|
200 | 804 | mdecorde | XMLTXM2CQP cqpbuilder = null;
|
201 | 321 | mdecorde | for (File f : files) { |
202 | 321 | mdecorde | if (second) { print(", ") } |
203 | 321 | mdecorde | if (second > 0 && (second % 5) == 0) println "" |
204 | 321 | mdecorde | print(f.getName().replaceFirst("\\.xml", "")); |
205 | 321 | mdecorde | second++ |
206 | 321 | mdecorde | |
207 | 804 | mdecorde | cqpbuilder = new XMLTXM2CQP(f.toURI().toURL());
|
208 | 321 | mdecorde | String txtname = f.getName().substring(0,f.getName().length()-4); |
209 | 803 | mdecorde | cqpbuilder.setTextInfo(txtname, corpusname, "project");
|
210 | 321 | mdecorde | |
211 | 803 | mdecorde | cqpbuilder.setBalisesToKeep(["text","p","s"]); |
212 | 803 | mdecorde | cqpbuilder.setSendToPAttributes(["s":["id"], "p":["id"]]); |
213 | 803 | mdecorde | cqpbuilder.setLang(lang); |
214 | 803 | mdecorde | if (!cqpbuilder.transformFile(cqpFile)) {
|
215 | 321 | mdecorde | println("Failed to compile "+f)
|
216 | 321 | mdecorde | } |
217 | 321 | mdecorde | |
218 | 321 | mdecorde | } |
219 | 321 | mdecorde | |
220 | 321 | mdecorde | //end corpus
|
221 | 803 | mdecorde | if (createOutput(cqpFile)) {
|
222 | 321 | mdecorde | output.write("</txmcorpus>\n");
|
223 | 321 | mdecorde | output.close(); |
224 | 321 | mdecorde | } |
225 | 321 | mdecorde | |
226 | 803 | mdecorde | if (cqpbuilder == null) { |
227 | 321 | mdecorde | println "there was no files to process: "+files
|
228 | 321 | mdecorde | return false; |
229 | 321 | mdecorde | } |
230 | 321 | mdecorde | |
231 | 321 | mdecorde | //2- Import into CWB
|
232 | 321 | mdecorde | def outDir = binDirPath;
|
233 | 321 | mdecorde | |
234 | 321 | mdecorde | CwbEncode cwbEn = new CwbEncode();
|
235 | 321 | mdecorde | CwbMakeAll cwbMa = new CwbMakeAll();
|
236 | 321 | mdecorde | |
237 | 803 | mdecorde | List<String> pAttributesList = cqpbuilder.getpAttributs(); |
238 | 803 | mdecorde | List<String> sAttributesList = cqpbuilder.getsAttributs(); |
239 | 321 | mdecorde | println "pAttrs : "+pAttributesList
|
240 | 321 | mdecorde | println "sAttrs : "+sAttributesList
|
241 | 321 | mdecorde | String[] pAttributes = pAttributesList.toArray(new String[pAttributesList.size()]) |
242 | 321 | mdecorde | String[] sAttributes = sAttributesList.toArray(new String[sAttributesList.size()]) |
243 | 321 | mdecorde | |
244 | 321 | mdecorde | try {
|
245 | 321 | mdecorde | //println "Indexing "+compiler.textAttributes
|
246 | 321 | mdecorde | cwbEn.setDebug(debug); |
247 | 321 | mdecorde | String regPath = outDir + "registry/"+corpusname.toLowerCase() |
248 | 321 | mdecorde | cwbEn.run( |
249 | 321 | mdecorde | outDir + "data/$corpusname",
|
250 | 803 | mdecorde | outDir + "/cqp/"+corpusname+".cqp", |
251 | 321 | mdecorde | regPath, |
252 | 321 | mdecorde | pAttributes, |
253 | 321 | mdecorde | sAttributes); |
254 | 321 | mdecorde | if (!new File(regPath).exists()) { |
255 | 321 | mdecorde | println "Error: The registry file was not created: $regPath. See https://groupes.renater.fr/wiki/txm-users/public/faq"
|
256 | 321 | mdecorde | return false; |
257 | 321 | mdecorde | } |
258 | 321 | mdecorde | //println "Binding all indexes"
|
259 | 321 | mdecorde | cwbMa.setDebug(debug); |
260 | 714 | mdecorde | cwbMa.run(corpusname, outDir + "registry");
|
261 | 321 | mdecorde | |
262 | 321 | mdecorde | } catch (Exception ex) { System.err.println(ex); return false;} |
263 | 321 | mdecorde | |
264 | 321 | mdecorde | return true; |
265 | 321 | mdecorde | } |
266 | 321 | mdecorde | |
267 | 321 | mdecorde | /**
|
268 | 321 | mdecorde | * The main method.
|
269 | 321 | mdecorde | *
|
270 | 321 | mdecorde | * @param args the arguments
|
271 | 321 | mdecorde | */
|
272 | 321 | mdecorde | public static void main(String[] args) { |
273 | 321 | mdecorde | File dir = new File(System.getProperty("user.home"),"xml/discours/txm/"); |
274 | 321 | mdecorde | List<File> files = dir.listFiles(); |
275 | 321 | mdecorde | def c = new compiler() |
276 | 321 | mdecorde | c.setCwbLoc(System.getProperty("user.home")+"/TXM/cwb/bin/") |
277 | 321 | mdecorde | if (!c.run(files,"discours")) { |
278 | 321 | mdecorde | println "Compiler failed"
|
279 | 321 | mdecorde | return;
|
280 | 321 | mdecorde | } |
281 | 321 | mdecorde | } |
282 | 321 | mdecorde | } |