annotate software/mpdl-services/mpiwg-mpdl-lt/src/de/mpg/mpiwg/berlin/mpdl/lt/dict/app/Lexicon.java @ 20:7d6d969b10cf

little corrections
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Wed, 14 Dec 2011 12:48:43 +0100
parents 4a3641ae14d2
children e845310098ba
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;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
13
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
14 public class Lexicon implements Comparable<Lexicon> {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
15 private String name;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
16 private String sourceLang;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
17 private String description;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
18 private String queryUrl;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
19 private String type; // local or remote
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
20 private Hashtable<String, LexiconEntry> entries;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
21
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
22 public Lexicon(String name, String sourceLanguage) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
23 this.name = name;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
24 this.sourceLang = sourceLanguage;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
25 this.type = "local"; // default is local
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
26 this.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
27 }
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 public int compareTo(Lexicon l) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
30 return name.compareTo(l.name);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
31 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
32
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
33 public String getName() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
34 return name;
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 void setName(String name) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
38 this.name = 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 String getSourceLanguage() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
42 return sourceLang;
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 getDescription() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
46 return description;
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 void setDescription(String description) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
50 this.description = 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 String getQueryUrl() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
54 return queryUrl;
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 void setQueryUrl(String queryUrl) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
58 this.queryUrl = 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 String getType() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
62 return type;
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 void setType(String type) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
66 this.type = 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 boolean isLocalLexicon() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
70 boolean isLocal = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
71 if (type != null && type.equals("local"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
72 isLocal = true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
73 return isLocal;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
74 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
75
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
76 public boolean isBetacodeLexicon() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
77 boolean isBetacode = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
78 if (name.equals("autenrieth") || name.equals("bonitz") || name.equals("lsj"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
79 isBetacode = true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
80 return isBetacode;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
81 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
82
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
83 public boolean isBuckwalterLexicon() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
84 boolean isBuckwalter = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
85 if (name.equals("salmone"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
86 isBuckwalter = true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
87 return isBuckwalter;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
88 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
89
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
90 public ArrayList<LexiconEntry> getEntries() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
91 ArrayList<LexiconEntry> result = new ArrayList<LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
92 if (entries != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
93 Enumeration<String> entryKeys = entries.keys();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
94 while(entryKeys.hasMoreElements()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
95 String entryKey = entryKeys.nextElement();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
96 LexiconEntry le = entries.get(entryKey);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
97 result.add(le);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
98 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
99 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
100 Collections.sort(result);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
101 if (result.isEmpty())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
102 return null;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
103 else
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
104 return result;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
105 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
106
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
107 public LexiconEntry getDynamicEntry(String formName) throws ApplicationException {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
108 LexiconEntry lexEntry = new LexiconEntry(name, formName, null);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
109 String linkForm = formName;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
110 if (Language.getInstance().isGreek(sourceLang)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
111 linkForm = Transcoder.getInstance().transcodeFromUnicode2BetaCode(formName);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
112 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
113 if (name.equals("linyutan")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
114 linkForm = Transcoder.getInstance().encodeBig5(formName);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
115 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
116 String remoteUrl = queryUrl + linkForm;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
117 lexEntry.setRemoteUrl(remoteUrl);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
118 return lexEntry;
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
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
121 public boolean isEmpty() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
122 if (entries == null || entries.isEmpty())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
123 return true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
124 else
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
125 return false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
126 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
127
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
128 public void addEntry(LexiconEntry newEntry) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
129 if (entries == null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
130 this.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
131 entries.put(newEntry.getFormName(), newEntry);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
132 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
133
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
134 public void addEntries(ArrayList<LexiconEntry> newEntries) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
135 if (entries == null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
136 this.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
137 for (int i=0; i<newEntries.size(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
138 LexiconEntry newEntry = newEntries.get(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
139 entries.put(newEntry.getFormName(), newEntry);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
140 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
141 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
142
20
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
143 public LexiconEntry getEntry(String lexEntryName) {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
144 LexiconEntry retEntry = null;
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
145 if (entries == null) {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
146 entries.get(lexEntryName);
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
147 }
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
148 return retEntry;
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
149 }
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
150
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
151 /*
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
152 * without lexicon entries (non-Javadoc)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
153 * @see java.lang.Object#clone()
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
154 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
155 public Lexicon clone() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
156 Lexicon lex = new Lexicon(name, sourceLang);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
157 lex.description = description;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
158 lex.entries = new Hashtable<String, LexiconEntry>();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
159 lex.queryUrl = queryUrl;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
160 lex.type = type;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
161 return lex;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
162 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
163
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
164 public String toXmlString() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
165 String result = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
166 result = result + "<dictionary>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
167 result = result + "<name>" + name + "</name>";
20
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
168 result = result + "<language>" + sourceLang + "</language>";
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
169 result = result + "<description>" + description + "</description>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
170 result = result + "<entries>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
171 for (int i=0; i<entries.size(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
172 result = result + "<entry>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
173 LexiconEntry entry = getEntries().get(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
174 result = result + "<form>" + entry.getFormName() + "</form>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
175 if (isLocalLexicon()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
176 result = result + "<xml-valid>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
177 String xmlValid = "false";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
178 if (entry.isXmlValid())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
179 xmlValid = "true";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
180 result = result + xmlValid;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
181 result = result + "</xml-valid>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
182 result = result + "<content>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
183 if (entry.isXmlValid()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
184 String repairedEntry = entry.getRepairedEntry();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
185 repairedEntry = repairedEntry.replaceAll("<repaired-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
186 repairedEntry = repairedEntry.replaceAll("</repaired-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
187 result = result + repairedEntry; // unicode content of the original entry
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
188 } else {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
189 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
190 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
191 originalEntry = originalEntry.replaceAll("<original-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
192 originalEntry = originalEntry.replaceAll("</original-entry>", "");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
193 originalEntry = StringEscapeUtils.escapeXml(originalEntry); // create text version of the invalid xml content
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
194 result = result + originalEntry;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
195 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
196 result = result + "</content>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
197 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
198 if (entry.getRemoteUrl() != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
199 result = result + "<remoteUrl>" + entry.getRemoteUrl() + "</remoteUrl>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
200 result = result + "</entry>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
201 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
202 result = result + "</entries>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
203 result = result + "</dictionary>";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
204 return result;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
205 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
206
20
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
207 public String toXmlStringCompact() {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
208 String result = "";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
209 result = result + "<dictionary>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
210 result = result + "<name>" + name + "</name>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
211 result = result + "<language>" + sourceLang + "</language>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
212 result = result + "<description>" + description + "</description>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
213 result = result + "<entries>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
214 for (int i=0; i<entries.size(); i++) {
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
215 result = result + "<entry>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
216 LexiconEntry entry = getEntries().get(i);
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
217 result = result + "<form>" + entry.getFormName() + "</form>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
218 result = result + "</entry>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
219 }
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 result = result + "</dictionary>";
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
222 return result;
7d6d969b10cf little corrections
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents: 19
diff changeset
223 }
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
224 }