|
1
|
package eu.etaxonomy.cdm.io.taxonx2013;
|
|
2
|
|
|
3
|
import java.net.URI;
|
|
4
|
import java.net.URISyntaxException;
|
|
5
|
import java.util.ArrayList;
|
|
6
|
import java.util.HashMap;
|
|
7
|
import java.util.List;
|
|
8
|
import java.util.Map;
|
|
9
|
import java.util.UUID;
|
|
10
|
|
|
11
|
import org.apache.commons.lang.StringUtils;
|
|
12
|
import org.apache.log4j.Logger;
|
|
13
|
import org.w3c.dom.NamedNodeMap;
|
|
14
|
import org.w3c.dom.Node;
|
|
15
|
import org.w3c.dom.NodeList;
|
|
16
|
|
|
17
|
import eu.etaxonomy.cdm.model.agent.Person;
|
|
18
|
import eu.etaxonomy.cdm.model.agent.Team;
|
|
19
|
import eu.etaxonomy.cdm.model.common.TimePeriod;
|
|
20
|
import eu.etaxonomy.cdm.model.reference.IBook;
|
|
21
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
22
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
|
23
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
|
24
|
|
|
25
|
public class TaxonXModsExtractor extends TaxonXExtractor{
|
|
26
|
|
|
27
|
private final Map<String,UUID> personMap = new HashMap<String, UUID>();
|
|
28
|
|
|
29
|
Logger logger = Logger.getLogger(getClass());
|
|
30
|
|
|
31
|
/**
|
|
32
|
* @param agentService
|
|
33
|
*/
|
|
34
|
public TaxonXModsExtractor(TaxonXImport importer) {
|
|
35
|
this.importer = importer;
|
|
36
|
}
|
|
37
|
|
|
38
|
public Reference<?> extractMods(Node node){
|
|
39
|
// System.out.println("extractMods");
|
|
40
|
Map<String, String> modsMap = new HashMap<String, String>();
|
|
41
|
NodeList children = node.getChildNodes();
|
|
42
|
List<String> roleList = new ArrayList<String>();
|
|
43
|
String content="";
|
|
44
|
|
|
45
|
// int reftype = askQuestion("What kind of reference is it?\n 1: Generic\n 2: Book\n 3: Article\n" +
|
|
46
|
// " 4 : BookSection\n 5 : Journal\n 6 : Printseries\n 7: Thesis ");
|
|
47
|
int reftype=4;
|
|
48
|
Reference<?> ref= getReferenceType(reftype);
|
|
49
|
for (int i=0; i<children.getLength();i++){
|
|
50
|
|
|
51
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:titleinfo")){
|
|
52
|
NodeList tmp = children.item(i).getChildNodes();
|
|
53
|
for (int j=0;j<tmp.getLength();j++){
|
|
54
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:title")) {
|
|
55
|
content=tmp.item(j).getTextContent().trim();
|
|
56
|
if (!content.isEmpty()) {
|
|
57
|
modsMap.put("mainTitle",content);
|
|
58
|
// ref.setTitleCache(content,true);
|
|
59
|
ref.setTitle(content);
|
|
60
|
// ref.generateTitle();
|
|
61
|
}
|
|
62
|
}
|
|
63
|
}
|
|
64
|
}
|
|
65
|
|
|
66
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:name")){
|
|
67
|
Map<String,String> mapmap = getModsNames(children.item(i), ref);
|
|
68
|
if (!mapmap.isEmpty()) {
|
|
69
|
roleList.add(mapmap.toString());
|
|
70
|
}
|
|
71
|
}
|
|
72
|
|
|
73
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:typeofresource")){
|
|
74
|
content = children.item(i).getTextContent().trim();
|
|
75
|
if (!content.isEmpty()) {
|
|
76
|
modsMap.put("typeofresource",content);
|
|
77
|
}
|
|
78
|
}
|
|
79
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:identifier")){
|
|
80
|
content = children.item(i).getTextContent().trim();
|
|
81
|
if (!content.isEmpty()) {
|
|
82
|
modsMap.put(children.item(i).getAttributes().getNamedItem("type").getNodeValue(),content);
|
|
83
|
if (children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("isbn")) {
|
|
84
|
ref.setIsbn(content);
|
|
85
|
}
|
|
86
|
if (children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("issn")) {
|
|
87
|
ref.setIssn(content);
|
|
88
|
}
|
|
89
|
if (children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("GenericHash")) {
|
|
90
|
ref.setIssn("GenericHash: "+content);
|
|
91
|
try {
|
|
92
|
ref.setUri(new URI("http://plazi.cs.umb.edu/GgServer/search?MODS.ModsDocID="+content));
|
|
93
|
} catch (URISyntaxException e) {
|
|
94
|
// TODO Auto-generated catch block
|
|
95
|
e.printStackTrace();
|
|
96
|
}
|
|
97
|
}
|
|
98
|
}
|
|
99
|
}
|
|
100
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:location")){
|
|
101
|
NodeList tmp = children.item(i).getChildNodes();
|
|
102
|
for (int j=0;j<tmp.getLength();j++){
|
|
103
|
// System.out.println("Child of mods:location: "+tmp.item(j).getNodeName());
|
|
104
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:url")) {
|
|
105
|
content = tmp.item(j).getTextContent().trim();
|
|
106
|
if (!content.isEmpty() && (content != "http://un.availab.le")) {
|
|
107
|
modsMap.put("url",content);
|
|
108
|
ref.setUri(URI.create(content));
|
|
109
|
}
|
|
110
|
}
|
|
111
|
}
|
|
112
|
}
|
|
113
|
|
|
114
|
|
|
115
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:relatedItem")){
|
|
116
|
addRelatedMods(children.item(i), modsMap, ref);
|
|
117
|
}
|
|
118
|
|
|
119
|
|
|
120
|
}
|
|
121
|
modsMap.put("people",StringUtils.join(roleList.toArray(),SPLITTER));
|
|
122
|
|
|
123
|
List<Reference> references = importer.getReferenceService().list(Reference.class, 0, 0, null, null);
|
|
124
|
for(Reference<?> refe:references){
|
|
125
|
if (refe.getCitation().equalsIgnoreCase(ref.getCitation())) {
|
|
126
|
ref=refe;
|
|
127
|
}
|
|
128
|
}
|
|
129
|
// System.out.println(modsMap);
|
|
130
|
//
|
|
131
|
// System.out.println("REFERENCE "+ref.getCitation());
|
|
132
|
// System.out.println("REFERENCE "+ref.getTitle());
|
|
133
|
// System.out.println("REFERENCE "+ref.getTitleCache());
|
|
134
|
return ref;
|
|
135
|
}
|
|
136
|
|
|
137
|
private final String AUTHOR = "author";
|
|
138
|
private final String EDITOR = "editor";
|
|
139
|
/**
|
|
140
|
* @param item
|
|
141
|
* @return
|
|
142
|
*/
|
|
143
|
private Map<String, String> getModsNames(Node node, Reference<?> ref) {
|
|
144
|
NamedNodeMap attributeMap = node.getAttributes();
|
|
145
|
Map<String,String> mapmap = new HashMap<String, String>();
|
|
146
|
List<String> roleList = new ArrayList<String>();
|
|
147
|
boolean newRole=false;
|
|
148
|
String content="";
|
|
149
|
String role =null;
|
|
150
|
List<Person> persons = new ArrayList<Person>();
|
|
151
|
List<String> editors= new ArrayList<String>();
|
|
152
|
|
|
153
|
if ((attributeMap.getNamedItem("type") != null) && attributeMap.getNamedItem("type").getNodeValue().equalsIgnoreCase("personal")) {
|
|
154
|
|
|
155
|
NodeList tmp = node.getChildNodes();
|
|
156
|
for (int j=0;j<tmp.getLength();j++){
|
|
157
|
|
|
158
|
// System.out.println("Child of modsnametype: "+tmp.item(j).getNodeName());
|
|
159
|
Person p=null;
|
|
160
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:namePart")) {
|
|
161
|
content=tmp.item(j).getTextContent().trim();
|
|
162
|
if (!content.isEmpty()) {
|
|
163
|
mapmap.put("namePart",content);
|
|
164
|
p = Person.NewInstance();
|
|
165
|
p.setTitleCache(content, true);
|
|
166
|
}
|
|
167
|
}
|
|
168
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:role")) {
|
|
169
|
NodeList tmp2 = tmp.item(j).getChildNodes();
|
|
170
|
for (int k=0; k< tmp2.getLength();k++){
|
|
171
|
if (tmp2.item(k).getNodeName().equalsIgnoreCase("mods:roleTerm")){
|
|
172
|
content = tmp2.item(k).getTextContent().trim();
|
|
173
|
// System.out.println("ROLETERM!" +content);
|
|
174
|
if (!content.isEmpty()) {
|
|
175
|
roleList.add(content);
|
|
176
|
// p.setNomenclaturalTitle(content);
|
|
177
|
if (content.equalsIgnoreCase(EDITOR)) {
|
|
178
|
role=EDITOR;
|
|
179
|
}
|
|
180
|
if (content.equalsIgnoreCase(AUTHOR)) {
|
|
181
|
role=AUTHOR;
|
|
182
|
}
|
|
183
|
newRole=true;
|
|
184
|
}
|
|
185
|
}
|
|
186
|
}
|
|
187
|
}
|
|
188
|
if (newRole){
|
|
189
|
if ((p!=null) && role.equals(AUTHOR)) {
|
|
190
|
UUID uuid = null;
|
|
191
|
if (!personMap.containsKey(p.getTitleCache())){
|
|
192
|
uuid = importer.getAgentService().saveOrUpdate(p);
|
|
193
|
p = (Person) importer.getAgentService().find(uuid);
|
|
194
|
personMap.put(p.getTitleCache(),uuid);
|
|
195
|
}else{
|
|
196
|
uuid = personMap.get(p.getTitleCache());
|
|
197
|
p = (Person) importer.getAgentService().find(uuid);
|
|
198
|
}
|
|
199
|
// logger.info("ADD PERSON "+p);
|
|
200
|
persons.add(p);
|
|
201
|
}
|
|
202
|
if ((p!=null) && role.equals(EDITOR)) {
|
|
203
|
editors.add(p.getTitleCache());
|
|
204
|
}
|
|
205
|
}
|
|
206
|
}
|
|
207
|
}
|
|
208
|
if (persons.size()>0){
|
|
209
|
if (persons.size()==1){
|
|
210
|
ref.setAuthorTeam(persons.get(0));
|
|
211
|
}
|
|
212
|
else{
|
|
213
|
Team authorTeam = Team.NewInstance();
|
|
214
|
for (Person pers:persons){
|
|
215
|
authorTeam.addTeamMember(pers);
|
|
216
|
}
|
|
217
|
|
|
218
|
if (!personMap.containsKey(authorTeam.getTitleCache()) && (authorTeam.getTeamMembers().size()>0)){
|
|
219
|
UUID uuid = importer.getAgentService().saveOrUpdate(authorTeam);
|
|
220
|
personMap.put(authorTeam.getTitleCache(),uuid);
|
|
221
|
}else{
|
|
222
|
if(authorTeam.getTeamMembers().size()>1) {
|
|
223
|
authorTeam = (Team) importer.getAgentService().find(personMap.get(authorTeam.getTitleCache()));
|
|
224
|
}
|
|
225
|
}
|
|
226
|
|
|
227
|
ref.setAuthorTeam(authorTeam);
|
|
228
|
}
|
|
229
|
}
|
|
230
|
if (editors.size()>0) {
|
|
231
|
ref.setEditor(StringUtils.join(editors,", "));
|
|
232
|
}
|
|
233
|
mapmap.put("role",StringUtils.join(roleList.toArray(),SPLITTER));
|
|
234
|
return mapmap;
|
|
235
|
}
|
|
236
|
|
|
237
|
|
|
238
|
/**
|
|
239
|
* @param item
|
|
240
|
* @param modsMap
|
|
241
|
*/
|
|
242
|
private void addRelatedMods(Node node, Map<String, String> modsMap, Reference<?> ref) {
|
|
243
|
NodeList tmp =node.getChildNodes();
|
|
244
|
NodeList partNodes = null;
|
|
245
|
NodeList children = null;
|
|
246
|
|
|
247
|
List<String> originInfo = null;
|
|
248
|
List<String> partList = null;
|
|
249
|
|
|
250
|
TimePeriod date;
|
|
251
|
|
|
252
|
String publisher="";
|
|
253
|
String publishplace="";
|
|
254
|
String pstart="";
|
|
255
|
String pend="";
|
|
256
|
|
|
257
|
Map<String,String> mapmap=null;
|
|
258
|
|
|
259
|
Map<String, String> relatedInfoMap = new HashMap<String, String>();
|
|
260
|
List<String> roleList = new ArrayList<String>();
|
|
261
|
String content="";
|
|
262
|
|
|
263
|
relatedInfoMap.put("type",node.getAttributes().getNamedItem("type").getNodeValue());
|
|
264
|
|
|
265
|
for (int j=0;j<tmp.getLength();j++){
|
|
266
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:titleInfo")) {
|
|
267
|
content=tmp.item(j).getTextContent().trim();
|
|
268
|
if (!content.isEmpty()) {
|
|
269
|
relatedInfoMap.put("titleInfo",content);
|
|
270
|
if (node.getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("host")){
|
|
271
|
List<Reference> references = importer.getReferenceService().list(Reference.class, 0, 0, null, null);
|
|
272
|
boolean refFound = false;
|
|
273
|
IBook book = null;
|
|
274
|
for (Reference<?> refe:references){
|
|
275
|
if(refe.getTitleCache().equalsIgnoreCase(content)){
|
|
276
|
refFound=true;
|
|
277
|
book= refe;
|
|
278
|
}
|
|
279
|
}
|
|
280
|
if (!refFound){
|
|
281
|
book = ReferenceFactory.newBook();
|
|
282
|
// book.setTitleCache(content,true);
|
|
283
|
book.setTitle(content);
|
|
284
|
// book.generateTitle();
|
|
285
|
}
|
|
286
|
if ((ref.getInBook() == null) || !ref.getInBook().equals(book)) {
|
|
287
|
ref.setInBook(book);
|
|
288
|
// ref.generateTitle();
|
|
289
|
}
|
|
290
|
}
|
|
291
|
}
|
|
292
|
}
|
|
293
|
|
|
294
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:originInfo")) {
|
|
295
|
children = tmp.item(j).getChildNodes();
|
|
296
|
originInfo = new ArrayList<String>();
|
|
297
|
for (int i=0;i<children.getLength();i++){
|
|
298
|
content=children.item(i).getTextContent().trim();
|
|
299
|
if (!content.isEmpty()) {
|
|
300
|
originInfo.add(children.item(i).getNodeName()+":"+content);
|
|
301
|
if (children.item(i).getNodeName().contains("dateIssued")) {
|
|
302
|
ref.setDatePublished(TimePeriodParser.parseString(content));
|
|
303
|
}
|
|
304
|
}
|
|
305
|
publisher="";
|
|
306
|
publishplace="";
|
|
307
|
if (children.item(i).getNodeName().contains("publisher")) {
|
|
308
|
try{
|
|
309
|
publisher=children.item(i).getChildNodes().item(0).getTextContent().trim();
|
|
310
|
// System.out.println("PUBLISHER "+publisher);
|
|
311
|
}catch(Exception e){System.out.println("oups "+e);}
|
|
312
|
}
|
|
313
|
if (children.item(i).getNodeName().contains("place")) {
|
|
314
|
try{
|
|
315
|
publishplace=children.item(i).getTextContent().trim();
|
|
316
|
// System.out.println("PUBLISHED "+publishplace);
|
|
317
|
}catch(Exception e){System.out.println("oups "+e);}
|
|
318
|
}
|
|
319
|
if (publishplace.isEmpty() && !publisher.isEmpty()) {
|
|
320
|
ref.setPublisher(publisher);
|
|
321
|
}
|
|
322
|
if (!publishplace.isEmpty() && !publisher.isEmpty()) {
|
|
323
|
ref.setPublisher(publisher, publishplace);
|
|
324
|
}
|
|
325
|
}
|
|
326
|
relatedInfoMap.put("originInfo", StringUtils.join(originInfo.toArray(),SPLITTER));
|
|
327
|
}
|
|
328
|
|
|
329
|
|
|
330
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:name")){
|
|
331
|
mapmap = getModsNames(tmp.item(j),ref);
|
|
332
|
if (!mapmap.isEmpty()) {
|
|
333
|
roleList.add(mapmap.toString());
|
|
334
|
}
|
|
335
|
}
|
|
336
|
if (tmp.item(j).getNodeName().equalsIgnoreCase("mods:part")){
|
|
337
|
children = tmp.item(j).getChildNodes();
|
|
338
|
partList = new ArrayList<String>();
|
|
339
|
for (int i=0;i<children.getLength();i++){
|
|
340
|
mapmap = new HashMap<String, String>();
|
|
341
|
// System.out.println(children.item(i).getNodeName());
|
|
342
|
|
|
343
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:date")){
|
|
344
|
content = children.item(i).getTextContent().trim();
|
|
345
|
if (!content.isEmpty()){
|
|
346
|
date = TimePeriodParser.parseString(content);
|
|
347
|
ref.setDatePublished(date);
|
|
348
|
}
|
|
349
|
}
|
|
350
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:detail") &&
|
|
351
|
children.item(i).getAttributes().getNamedItem("type").getNodeValue().equalsIgnoreCase("volume")){
|
|
352
|
partNodes = children.item(i).getChildNodes();
|
|
353
|
for (int k=0; k<partNodes.getLength();k++){
|
|
354
|
if (partNodes.item(k).getNodeName().equalsIgnoreCase("mods:number")) {
|
|
355
|
content = partNodes.item(k).getTextContent().trim();
|
|
356
|
if (!content.isEmpty()) {
|
|
357
|
ref.setVolume(content);
|
|
358
|
}
|
|
359
|
}
|
|
360
|
}
|
|
361
|
}
|
|
362
|
if (children.item(i).getNodeName().equalsIgnoreCase("mods:extent")) {
|
|
363
|
mapmap.put("unit", children.item(i).getAttributes().getNamedItem("unit").getNodeValue());
|
|
364
|
partNodes = children.item(i).getChildNodes();
|
|
365
|
pstart="";
|
|
366
|
pend="";
|
|
367
|
for (int k=0; k<partNodes.getLength();k++){
|
|
368
|
if (partNodes.item(k).getNodeName().equalsIgnoreCase("mods:start")) {
|
|
369
|
content = partNodes.item(k).getTextContent().trim();
|
|
370
|
if (!content.isEmpty()) {
|
|
371
|
mapmap.put("start",content);
|
|
372
|
pstart=content;
|
|
373
|
}
|
|
374
|
}
|
|
375
|
if (partNodes.item(k).getNodeName().equalsIgnoreCase("mods:end")) {
|
|
376
|
content = partNodes.item(k).getTextContent().trim();
|
|
377
|
if (!content.isEmpty()) {
|
|
378
|
mapmap.put("end",content);
|
|
379
|
pend=content;
|
|
380
|
}
|
|
381
|
}
|
|
382
|
}
|
|
383
|
// System.out.println("SET PAGES "+pstart+"-"+pend);
|
|
384
|
ref.setPages(pstart+"-"+pend);
|
|
385
|
}
|
|
386
|
partList.add(mapmap.toString());
|
|
387
|
}
|
|
388
|
modsMap.put("part",StringUtils.join(partList.toArray(),SPLITTER));
|
|
389
|
}
|
|
390
|
}
|
|
391
|
relatedInfoMap.put("relatedRoles", StringUtils.join(roleList.toArray(),SPLITTER));
|
|
392
|
modsMap.put("relatedInfo",relatedInfoMap.toString());
|
|
393
|
}
|
|
394
|
|
|
395
|
}
|