root / tmp / org.txm.groovy.core / src / groovy / org / txm / scripts / importer / xtz / XTZImport.groovy @ 2126
History | View | Annotate | Download (3.4 kB)
1 | 986 | mdecorde | package org.txm.scripts.importer.xtz;
|
---|---|---|---|
2 | 321 | mdecorde | |
3 | 479 | mdecorde | import java.io.File; |
4 | 479 | mdecorde | import java.io.FileFilter; |
5 | 479 | mdecorde | import java.util.ArrayList; |
6 | 479 | mdecorde | import java.util.Arrays; |
7 | 479 | mdecorde | import java.util.Collections; |
8 | 479 | mdecorde | |
9 | 490 | mdecorde | import org.eclipse.core.runtime.IProgressMonitor; |
10 | 479 | mdecorde | import org.txm.utils.xml.DomUtils; |
11 | 321 | mdecorde | import org.txm.metadatas.Metadatas |
12 | 321 | mdecorde | import org.txm.utils.io.FileCopy; |
13 | 479 | mdecorde | import org.txm.* |
14 | 1135 | mdecorde | import org.txm.objects.* |
15 | 479 | mdecorde | import org.w3c.dom.Element |
16 | 1000 | mdecorde | import org.txm.importer.xtz.* |
17 | 321 | mdecorde | |
18 | 321 | mdecorde | public class XTZImport extends ImportModule { |
19 | 321 | mdecorde | |
20 | 321 | mdecorde | public XTZImport(File importParametersFile) { |
21 | 321 | mdecorde | super(importParametersFile);
|
22 | 321 | mdecorde | } |
23 | 321 | mdecorde | |
24 | 1135 | mdecorde | public XTZImport(Project p) {
|
25 | 1135 | mdecorde | super(p);
|
26 | 321 | mdecorde | } |
27 | 321 | mdecorde | |
28 | 321 | mdecorde | @Override
|
29 | 1135 | mdecorde | public void init(Project p) { |
30 | 321 | mdecorde | super.init(p);
|
31 | 321 | mdecorde | |
32 | 321 | mdecorde | importer = new XTZImporter(this) |
33 | 321 | mdecorde | compiler = new XTZCompiler(this) |
34 | 321 | mdecorde | annotater = new TTAnnotater(this); |
35 | 321 | mdecorde | pager = new XTZPager(this) |
36 | 321 | mdecorde | } |
37 | 321 | mdecorde | |
38 | 321 | mdecorde | @Override
|
39 | 321 | mdecorde | protected ArrayList<File> getTXMFilesOrder() { |
40 | 321 | mdecorde | //System.out.println("XTZ FILES ORDER");
|
41 | 321 | mdecorde | if (importer == null) { |
42 | 321 | mdecorde | println "no importer step, using default text order"
|
43 | 321 | mdecorde | return super.getTXMFilesOrder(); |
44 | 321 | mdecorde | } |
45 | 321 | mdecorde | Metadatas metadata = importer.getMetadata(); |
46 | 321 | mdecorde | if (metadata == null) { |
47 | 321 | mdecorde | println "no metadata, using default text order"
|
48 | 321 | mdecorde | return super.getTXMFilesOrder(); |
49 | 321 | mdecorde | } |
50 | 321 | mdecorde | File txmDirectory = new File(binaryDirectory, "txm/"+corpusName); |
51 | 321 | mdecorde | ArrayList<File> files = new ArrayList<File>(Arrays.asList(txmDirectory.listFiles(new FileFilter() { |
52 | 321 | mdecorde | @Override
|
53 | 321 | mdecorde | public boolean accept(File file) { |
54 | 321 | mdecorde | return file.isFile() && file.getName().endsWith(".xml"); |
55 | 321 | mdecorde | } |
56 | 321 | mdecorde | }))); |
57 | 321 | mdecorde | |
58 | 321 | mdecorde | final HashMap<File, String> textorder = new HashMap<File, String>(); |
59 | 321 | mdecorde | for (File f : files) { |
60 | 321 | mdecorde | HashMap<String, String> m = metadata.getTextMetadata(f); |
61 | 321 | mdecorde | if (m != null && m.containsKey("textorder")) textorder[f] = m["textorder"]; |
62 | 321 | mdecorde | } |
63 | 321 | mdecorde | println "Sorting texts using 'textorder' metadata values: "+textorder
|
64 | 321 | mdecorde | Collections.sort(files, new Comparator<File>() { |
65 | 321 | mdecorde | public int compare(File f1, File f2) { |
66 | 321 | mdecorde | String o1 = textorder[f1];
|
67 | 321 | mdecorde | String o2 = textorder[f2];
|
68 | 321 | mdecorde | if (o1 == null && o2 == null) { |
69 | 321 | mdecorde | return f1.getName().compareTo(f2.getName());
|
70 | 321 | mdecorde | } else if (o1 == null) { |
71 | 321 | mdecorde | return 1 |
72 | 321 | mdecorde | } else if (o2 == null) { |
73 | 321 | mdecorde | return -1 |
74 | 321 | mdecorde | } else {
|
75 | 321 | mdecorde | int c = o1.compareTo(o2);
|
76 | 321 | mdecorde | if (c == 0) return f1.getName().compareTo(f2.getName()); |
77 | 321 | mdecorde | else return c; |
78 | 321 | mdecorde | } |
79 | 321 | mdecorde | } |
80 | 321 | mdecorde | }); |
81 | 321 | mdecorde | //println files
|
82 | 321 | mdecorde | return files;
|
83 | 321 | mdecorde | } |
84 | 321 | mdecorde | |
85 | 321 | mdecorde | public void start() throws InterruptedException { |
86 | 321 | mdecorde | super.start();
|
87 | 321 | mdecorde | |
88 | 321 | mdecorde | if (isSuccessful) {
|
89 | 321 | mdecorde | |
90 | 321 | mdecorde | //declare a local KR
|
91 | 715 | mdecorde | //TODO find out how the annotation plugin may hook the import steps
|
92 | 715 | mdecorde | // List<String> krnames = importParameters.getKnowledgeRepositoryNames();
|
93 | 715 | mdecorde | // if (krnames.size() == 0) {
|
94 | 715 | mdecorde | // importParameters.createKnowledgeRepositoryElement("DEFAULT"); // set a default KR shared by all XTZ corpus
|
95 | 715 | mdecorde | // } else if (krnames.size() == 1 && krnames.get(0).equals("DEFAULT")) {
|
96 | 715 | mdecorde | // // nothing to do
|
97 | 715 | mdecorde | // } else {
|
98 | 715 | mdecorde | // println("Corpus is using custom Knowledge repositories: "+importParameters.getKnowledgeRepositoryNames());
|
99 | 715 | mdecorde | // }
|
100 | 321 | mdecorde | |
101 | 321 | mdecorde | //copy sub directories
|
102 | 321 | mdecorde | if (isUpdatingCorpus()) {
|
103 | 321 | mdecorde | |
104 | 321 | mdecorde | } else {
|
105 | 321 | mdecorde | def dirToCopy = ["xsl", "css", "dtd"] |
106 | 321 | mdecorde | println "--- Copying subdirectories $dirToCopy"
|
107 | 321 | mdecorde | for (String dir : dirToCopy) { |
108 | 321 | mdecorde | File origDirectory = new File(this.sourceDirectory, dir) |
109 | 321 | mdecorde | if (origDirectory.exists()) {
|
110 | 321 | mdecorde | print "."
|
111 | 321 | mdecorde | File copyDirectory = new File(this.binaryDirectory, dir) |
112 | 1971 | mdecorde | copyDirectory.deleteDir(); // clean before copying
|
113 | 321 | mdecorde | FileCopy.copyFiles(origDirectory, copyDirectory) |
114 | 321 | mdecorde | } |
115 | 321 | mdecorde | } |
116 | 321 | mdecorde | println ""
|
117 | 321 | mdecorde | } |
118 | 321 | mdecorde | } |
119 | 321 | mdecorde | } |
120 | 321 | mdecorde | } |