annotate software/mpdl-services/mpiwg-mpdl-lt/src/de/mpg/mpiwg/berlin/mpdl/lt/dict/app/Lexicon.java @ 23:e845310098ba

diverse Korrekturen
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Tue, 27 Nov 2012 12:35:19 +0100
parents 7d6d969b10cf
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1 package de.mpg.mpiwg.berlin.mpdl.lt.dict.app;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
2
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
3 import java.util.ArrayList;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
4 import java.util.Collections;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
5 import java.util.Enumeration;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
6 import java.util.Hashtable;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
7
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
8 import org.apache.commons.lang3.StringEscapeUtils;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
9
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
10 import de.mpg.mpiwg.berlin.mpdl.exception.ApplicationException;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
11 import de.mpg.mpiwg.berlin.mpdl.lt.general.Language;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
12 import de.mpg.mpiwg.berlin.mpdl.lt.text.transcode.Transcoder;
23
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
13 import de.mpg.mpiwg.berlin.mpdl.util.StringUtils;
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
14
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
15 public class Lexicon implements Comparable<Lexicon> {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
16 private String name;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
17 private String sourceLang;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
18 private String description;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
19 private String queryUrl;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
20 private String type; // local or remote
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
21 private Hashtable<String, LexiconEntry> entries;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
22
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
23 public Lexicon(String name, String sourceLanguage) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
24 this.name = name;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
25 this.sourceLang = sourceLanguage;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
26 this.type = "local"; // default is local
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
27 this.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
28 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
29
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
30 public int compareTo(Lexicon l) {
23
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
31 if (description != null)
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
32 return description.compareTo(l.description);
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
33 else
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
34 return name.compareTo(l.name);
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
35 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
36
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
37 public String getName() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
38 return name;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
39 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
40
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
41 public void setName(String name) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
42 this.name = name;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
43 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
44
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
45 public String getSourceLanguage() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
46 return sourceLang;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
47 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
48
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
49 public String getDescription() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
50 return description;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
51 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
52
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
53 public void setDescription(String description) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
54 this.description = description;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
55 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
56
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
57 public String getQueryUrl() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
58 return queryUrl;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
59 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
60
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
61 public void setQueryUrl(String queryUrl) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
62 this.queryUrl = queryUrl;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
63 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
64
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
65 public String getType() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
66 return type;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
67 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
68
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
69 public void setType(String type) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
70 this.type = type;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
71 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
72
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
73 public boolean isLocalLexicon() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
74 boolean isLocal = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
75 if (type != null && type.equals("local"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
76 isLocal = true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
77 return isLocal;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
78 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
79
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
80 public boolean isBetacodeLexicon() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
81 boolean isBetacode = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
82 if (name.equals("autenrieth") || name.equals("bonitz") || name.equals("lsj"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
83 isBetacode = true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
84 return isBetacode;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
85 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
86
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
87 public boolean isBuckwalterLexicon() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
88 boolean isBuckwalter = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
89 if (name.equals("salmone"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
90 isBuckwalter = true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
91 return isBuckwalter;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
92 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
93
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
94 public ArrayList<LexiconEntry> getEntries() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
95 ArrayList<LexiconEntry> result = new ArrayList<LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
96 if (entries != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
97 Enumeration<String> entryKeys = entries.keys();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
98 while(entryKeys.hasMoreElements()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
99 String entryKey = entryKeys.nextElement();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
100 LexiconEntry le = entries.get(entryKey);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
101 result.add(le);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
102 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
103 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
104 Collections.sort(result);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
105 if (result.isEmpty())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
106 return null;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
107 else
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
108 return result;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
109 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
110
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
111 public LexiconEntry getDynamicEntry(String formName) throws ApplicationException {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
112 LexiconEntry lexEntry = new LexiconEntry(name, formName, null);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
113 String linkForm = formName;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
114 if (Language.getInstance().isGreek(sourceLang)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
115 linkForm = Transcoder.getInstance().transcodeFromUnicode2BetaCode(formName);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
116 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
117 if (name.equals("linyutan")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
118 linkForm = Transcoder.getInstance().encodeBig5(formName);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
119 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
120 String remoteUrl = queryUrl + linkForm;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
121 lexEntry.setRemoteUrl(remoteUrl);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
122 return lexEntry;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
123 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
124
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
125 public boolean isEmpty() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
126 if (entries == null || entries.isEmpty())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
127 return true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
128 else
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
129 return false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
130 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
131
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
132 public void addEntry(LexiconEntry newEntry) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
133 if (entries == null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
134 this.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
135 entries.put(newEntry.getFormName(), newEntry);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
136 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
137
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
138 public void addEntries(ArrayList<LexiconEntry> newEntries) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
139 if (entries == null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
140 this.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
141 for (int i=0; i<newEntries.size(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
142 LexiconEntry newEntry = newEntries.get(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
143 entries.put(newEntry.getFormName(), newEntry);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
144 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
145 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
146
20
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
147 public LexiconEntry getEntry(String lexEntryName) {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
148 LexiconEntry retEntry = null;
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
149 if (entries == null) {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
150 entries.get(lexEntryName);
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
151 }
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
152 return retEntry;
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
153 }
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
154
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
155 /*
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
156 * without lexicon entries (non-Javadoc)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
157 * @see java.lang.Object#clone()
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
158 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
159 public Lexicon clone() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
160 Lexicon lex = new Lexicon(name, sourceLang);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
161 lex.description = description;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
162 lex.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
163 lex.queryUrl = queryUrl;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
164 lex.type = type;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
165 return lex;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
166 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
167
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
168 public String toXmlString() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
169 String result = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
170 result = result + "<dictionary>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
171 result = result + "<name>" + name + "</name>";
20
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
172 result = result + "<language>" + sourceLang + "</language>";
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
173 result = result + "<description>" + description + "</description>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
174 result = result + "<entries>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
175 for (int i=0; i<entries.size(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
176 result = result + "<entry>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
177 LexiconEntry entry = getEntries().get(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
178 result = result + "<form>" + entry.getFormName() + "</form>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
179 if (isLocalLexicon()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
180 result = result + "<xml-valid>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
181 String xmlValid = "false";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
182 if (entry.isXmlValid())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
183 xmlValid = "true";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
184 result = result + xmlValid;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
185 result = result + "</xml-valid>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
186 result = result + "<content>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
187 if (entry.isXmlValid()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
188 String repairedEntry = entry.getRepairedEntry();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
189 repairedEntry = repairedEntry.replaceAll("<repaired-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
190 repairedEntry = repairedEntry.replaceAll("</repaired-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
191 result = result + repairedEntry; // unicode content of the original entry
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
192 } else {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
193 result = result + "<remark>This dictionary entry has no valid XML/HTML content in database so a text version of this entry is shown</remark>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
194 String originalEntry = entry.getOriginalEntry(); // original content: not valid and e.g. in Betacode
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
195 originalEntry = originalEntry.replaceAll("<original-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
196 originalEntry = originalEntry.replaceAll("</original-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
197 originalEntry = StringEscapeUtils.escapeXml(originalEntry); // create text version of the invalid xml content
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
198 result = result + originalEntry;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
199 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
200 result = result + "</content>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
201 }
23
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
202 if (entry.getRemoteUrl() != null) {
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
203 String remoteUrl = entry.getRemoteUrl();
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
204 remoteUrl = StringEscapeUtils.escapeXml(remoteUrl);
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
205 result = result + "<remoteUrl>" + remoteUrl + "</remoteUrl>";
e845310098ba diverse Korrekturen
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 20
diff changeset
206 }
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
207 result = result + "</entry>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
208 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
209 result = result + "</entries>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
210 result = result + "</dictionary>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
211 return result;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
212 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
213
20
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
214 public String toXmlStringCompact() {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
215 String result = "";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
216 result = result + "<dictionary>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
217 result = result + "<name>" + name + "</name>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
218 result = result + "<language>" + sourceLang + "</language>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
219 result = result + "<description>" + description + "</description>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
220 result = result + "<entries>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
221 for (int i=0; i<entries.size(); i++) {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
222 result = result + "<entry>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
223 LexiconEntry entry = getEntries().get(i);
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
224 result = result + "<form>" + entry.getFormName() + "</form>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
225 result = result + "</entry>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
226 }
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
227 result = result + "</entries>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
228 result = result + "</dictionary>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
229 return result;
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
230 }
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
231 }