|
1
|
/**
|
|
2
|
* Copyright (C) 2009 EDIT
|
|
3
|
* European Distributed Institute of Taxonomy
|
|
4
|
* http://www.e-taxonomy.eu
|
|
5
|
*
|
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
|
8
|
*/
|
|
9
|
|
|
10
|
package eu.etaxonomy.cdm.io.markup;
|
|
11
|
|
|
12
|
import javax.xml.stream.XMLEventReader;
|
|
13
|
import javax.xml.stream.XMLStreamException;
|
|
14
|
import javax.xml.stream.events.XMLEvent;
|
|
15
|
|
|
16
|
import org.apache.log4j.Logger;
|
|
17
|
|
|
18
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
19
|
import eu.etaxonomy.cdm.common.GeneralParser;
|
|
20
|
import eu.etaxonomy.cdm.model.agent.Institution;
|
|
21
|
import eu.etaxonomy.cdm.model.agent.Person;
|
|
22
|
import eu.etaxonomy.cdm.model.agent.Team;
|
|
23
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
|
24
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
25
|
import eu.etaxonomy.cdm.model.common.TimePeriod;
|
|
26
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
27
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
|
28
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
|
29
|
|
|
30
|
/**
|
|
31
|
* @author a.mueller
|
|
32
|
* @created 30.05.2012
|
|
33
|
*
|
|
34
|
*/
|
|
35
|
public class MarkupModsImport extends MarkupImportBase {
|
|
36
|
@SuppressWarnings("unused")
|
|
37
|
private static final Logger logger = Logger.getLogger(MarkupModsImport.class);
|
|
38
|
|
|
39
|
protected static final String MODS_TITLEINFO = "titleInfo";
|
|
40
|
protected static final String MODS_ABSTRACT = "abstract";
|
|
41
|
protected static final String MODS_TITLE = "title";
|
|
42
|
protected static final String MODS_SUBTITLE = "subTitle";
|
|
43
|
protected static final String MODS_PARTNUMBER = "partNumber";
|
|
44
|
protected static final String MODS_PARTNAME = "partName";
|
|
45
|
protected static final String MODS_NAME = "name";
|
|
46
|
protected static final String MODS_ORIGININFO = "originInfo";
|
|
47
|
protected static final String MODS_IDENTIFIER = "identifier";
|
|
48
|
protected static final String MODS_DESCRIPTION = "description";
|
|
49
|
protected static final String MODS_NAME_PART = "namePart";
|
|
50
|
protected static final String MODS_AFFILIATION = "affiliation";
|
|
51
|
protected static final String MODS_PUBLISHER ="publisher";
|
|
52
|
protected static final String MODS_DATE_ISSUED ="dateIssued";
|
|
53
|
protected static final String MODS_PLACE ="place";
|
|
54
|
protected static final String MODS_EDITION ="edition";
|
|
55
|
|
|
56
|
|
|
57
|
public MarkupModsImport(MarkupDocumentImport docImport) {
|
|
58
|
super(docImport);
|
|
59
|
}
|
|
60
|
|
|
61
|
public void handleMods(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent)
|
|
62
|
throws XMLStreamException {
|
|
63
|
checkNoAttributes(parentEvent);
|
|
64
|
|
|
65
|
Reference modsRef = ReferenceFactory.newGeneric();
|
|
66
|
while (reader.hasNext()) {
|
|
67
|
XMLEvent next = readNoWhitespace(reader);
|
|
68
|
if (isMyEndingElement(next, parentEvent)) {
|
|
69
|
//set the source reference
|
|
70
|
state.getConfig().setSourceReference(modsRef);
|
|
71
|
return;
|
|
72
|
}else if (isStartingElement(next, MODS_TITLEINFO)) {
|
|
73
|
handleTitleInfo(state, reader, next, modsRef);
|
|
74
|
}else if (isStartingElement(next, MODS_ABSTRACT)) {
|
|
75
|
String abstractStr = getCData(state, reader, next, true).trim();
|
|
76
|
if (abstractStr.startsWith("ABSTRACT")){
|
|
77
|
abstractStr = abstractStr.replaceFirst("ABSTRACT", "").trim();
|
|
78
|
}
|
|
79
|
modsRef.setReferenceAbstract(abstractStr);
|
|
80
|
} else if (isStartingElement(next, MODS_IDENTIFIER)) {
|
|
81
|
handleIdentifier(state, reader, next, modsRef);
|
|
82
|
} else if (isStartingElement(next, MODS_NAME)) {
|
|
83
|
handleName(state, reader, next, modsRef);
|
|
84
|
} else if (isStartingElement(next, MODS_ORIGININFO)) {
|
|
85
|
handleOriginInfo(state, reader, next, modsRef);
|
|
86
|
} else {
|
|
87
|
handleUnexpectedElement(next);
|
|
88
|
}
|
|
89
|
}
|
|
90
|
return;
|
|
91
|
}
|
|
92
|
|
|
93
|
private void handleOriginInfo(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, Reference modsRef) throws XMLStreamException {
|
|
94
|
checkNoAttributes(parentEvent);
|
|
95
|
while (reader.hasNext()) {
|
|
96
|
XMLEvent next = readNoWhitespace(reader);
|
|
97
|
|
|
98
|
if (isMyEndingElement(next, parentEvent)) {
|
|
99
|
return;
|
|
100
|
}else if (isStartingElement(next, MODS_PUBLISHER)) {
|
|
101
|
String publisher = this.getCData(state, reader, next);
|
|
102
|
if (modsRef.getPublisher() != null){
|
|
103
|
fireWarningEvent("Multiple publisher infos given. Concat by ;", next, 2);
|
|
104
|
}
|
|
105
|
modsRef.setPublisher(CdmUtils.concat(";", modsRef.getPublisher(), publisher));
|
|
106
|
}else if (isStartingElement(next, MODS_DATE_ISSUED)) {
|
|
107
|
String dateIssued = this.getCData(state, reader, next);
|
|
108
|
if (modsRef.getDatePublished() != null && ! modsRef.getDatePublished().isEmpty()){
|
|
109
|
fireWarningEvent("Multiple publish date infos given. I overwrite older information. Please check manually ;", next, 4);
|
|
110
|
}
|
|
111
|
TimePeriod timePeriod = TimePeriodParser.parseString(dateIssued);
|
|
112
|
modsRef.setDatePublished(timePeriod);
|
|
113
|
}else if (isStartingElement(next, MODS_PLACE)) {
|
|
114
|
String place = this.getCData(state, reader, next);
|
|
115
|
if (modsRef.getPlacePublished() != null){
|
|
116
|
fireWarningEvent("Multiple place published infos given. Concat by ;", next, 2);
|
|
117
|
}
|
|
118
|
modsRef.setPlacePublished(CdmUtils.concat(";", modsRef.getPlacePublished(), place));
|
|
119
|
}else if (isStartingElement(next, MODS_EDITION)) {
|
|
120
|
String edition = this.getCData(state, reader, next);
|
|
121
|
if (modsRef.getEdition() != null){
|
|
122
|
fireWarningEvent("Multiple edition infos given. Concat by ;", next, 2);
|
|
123
|
}
|
|
124
|
modsRef.setEdition(CdmUtils.concat(";", modsRef.getEdition(), edition));
|
|
125
|
} else {
|
|
126
|
handleUnexpectedElement(next);
|
|
127
|
}
|
|
128
|
}
|
|
129
|
return;
|
|
130
|
}
|
|
131
|
|
|
132
|
private void handleIdentifier(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, Reference modsRef) throws XMLStreamException {
|
|
133
|
checkNoAttributes(parentEvent);
|
|
134
|
|
|
135
|
|
|
136
|
String identifier = getCData(state, reader, parentEvent, true).trim();
|
|
137
|
|
|
138
|
if (GeneralParser.isIsbn(identifier)){
|
|
139
|
modsRef.setIsbn(identifier);
|
|
140
|
}else{
|
|
141
|
String message = "Identifier pattern not recognized: %s";
|
|
142
|
fireWarningEvent(String.format(message, identifier), parentEvent, 4);
|
|
143
|
}
|
|
144
|
|
|
145
|
return;
|
|
146
|
}
|
|
147
|
|
|
148
|
/**
|
|
149
|
* Reads all titleInfo information.
|
|
150
|
* ! Preliminary implementation !
|
|
151
|
*/
|
|
152
|
private void handleTitleInfo(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, Reference modsRef)
|
|
153
|
throws XMLStreamException {
|
|
154
|
checkNoAttributes(parentEvent);
|
|
155
|
|
|
156
|
String title = null;
|
|
157
|
String subTitle = null;
|
|
158
|
String partNumber = null;
|
|
159
|
String partName = null;
|
|
160
|
|
|
161
|
while (reader.hasNext()) {
|
|
162
|
XMLEvent next = readNoWhitespace(reader);
|
|
163
|
|
|
164
|
if (isMyEndingElement(next, parentEvent)) {
|
|
165
|
String all = CdmUtils.concat(" - ", title, subTitle);
|
|
166
|
//TODO according to http://library.princeton.edu/departments/tsd/metadoc/mods/titleinfo.html
|
|
167
|
//partNumber and partName can be repeated and the order should be kept
|
|
168
|
String part = CdmUtils.concat(" ", partNumber, partName);
|
|
169
|
all = CdmUtils.concat(", ", all, part);
|
|
170
|
modsRef.setTitle(all);
|
|
171
|
return;
|
|
172
|
}else if (isStartingElement(next, MODS_TITLE)) {
|
|
173
|
title = this.getCData(state, reader, next);
|
|
174
|
}else if (isStartingElement(next, MODS_SUBTITLE)) {
|
|
175
|
subTitle = this.getCData(state, reader, next);
|
|
176
|
}else if (isStartingElement(next, MODS_PARTNAME)) {
|
|
177
|
partName = this.getCData(state, reader, next);
|
|
178
|
}else if (isStartingElement(next, MODS_PARTNUMBER)) {
|
|
179
|
partNumber = this.getCData(state, reader, next);
|
|
180
|
} else {
|
|
181
|
handleUnexpectedElement(next);
|
|
182
|
}
|
|
183
|
}
|
|
184
|
return;
|
|
185
|
|
|
186
|
}
|
|
187
|
|
|
188
|
/**
|
|
189
|
* Reads all titleInfo information.
|
|
190
|
* ! Preliminary implementation !
|
|
191
|
*/
|
|
192
|
private void handleName(MarkupImportState state, XMLEventReader reader, XMLEvent parent, Reference modsRef)
|
|
193
|
throws XMLStreamException {
|
|
194
|
String type = getOnlyAttribute(parent, "type", true);
|
|
195
|
|
|
196
|
String description = null;
|
|
197
|
String namePart = null;
|
|
198
|
String partNumber = null;
|
|
199
|
String affiliation = null;
|
|
200
|
|
|
201
|
while (reader.hasNext()) {
|
|
202
|
XMLEvent next = readNoWhitespace(reader);
|
|
203
|
|
|
204
|
if (isMyEndingElement(next, parent)) {
|
|
205
|
if (! type.equals("personal")){
|
|
206
|
fireUnexpectedAttributeValue(parent, "type", type); //currently we handle only "personal"
|
|
207
|
}else{
|
|
208
|
Person person = Person.NewInstance();
|
|
209
|
TeamOrPersonBase<?> author = modsRef.getAuthorship();
|
|
210
|
if (author == null){
|
|
211
|
modsRef.setAuthorship(person);
|
|
212
|
}else if (author.isInstanceOf(Person.class)){
|
|
213
|
Team team = Team.NewInstance();
|
|
214
|
team.addTeamMember(person);
|
|
215
|
modsRef.setAuthorship(team);
|
|
216
|
}else {
|
|
217
|
CdmBase.deproxy(author, Team.class).addTeamMember(person);
|
|
218
|
}
|
|
219
|
if (isNotBlank(namePart)){
|
|
220
|
person.setTitleCache(namePart, true);
|
|
221
|
}
|
|
222
|
if (isNotBlank(description)){
|
|
223
|
fireWarningEvent("Mods:description needs to be handled manually",this.makeLocationStr(parent.getLocation()), 1);
|
|
224
|
}
|
|
225
|
if (isNotBlank(affiliation)){
|
|
226
|
Institution institution = Institution.NewInstance();
|
|
227
|
institution.setTitleCache(affiliation, true);
|
|
228
|
person.addInstitutionalMembership(institution, null, null, null);
|
|
229
|
}
|
|
230
|
|
|
231
|
}
|
|
232
|
|
|
233
|
return;
|
|
234
|
}else if (isStartingElement(next, MODS_DESCRIPTION)) {
|
|
235
|
description = this.getCData(state, reader, next);
|
|
236
|
}else if (isStartingElement(next, MODS_NAME_PART)) {
|
|
237
|
namePart = this.getCData(state, reader, next);
|
|
238
|
}else if (isStartingElement(next, MODS_AFFILIATION)) {
|
|
239
|
affiliation = this.getCData(state, reader, next);
|
|
240
|
} else {
|
|
241
|
handleUnexpectedElement(next);
|
|
242
|
}
|
|
243
|
}
|
|
244
|
return;
|
|
245
|
|
|
246
|
}
|
|
247
|
}
|