|
1
|
package eu.dnetlib.data.claims.migration.parser;
|
|
2
|
|
|
3
|
import eu.dnetlib.data.claims.migration.entity.Context;
|
|
4
|
import eu.dnetlib.data.claims.migration.entity.Result;
|
|
5
|
import eu.dnetlib.data.claims.migration.handler.DirectIndexHandler;
|
|
6
|
import eu.dnetlib.data.claimsDemo.ClaimUtils;
|
|
7
|
import eu.dnetlib.data.claimsDemo.ContextUtils;
|
|
8
|
import org.apache.log4j.Logger;
|
|
9
|
import org.w3c.dom.Document;
|
|
10
|
import org.w3c.dom.NodeList;
|
|
11
|
import org.xml.sax.InputSource;
|
|
12
|
import org.xml.sax.SAXException;
|
|
13
|
|
|
14
|
import javax.xml.parsers.DocumentBuilder;
|
|
15
|
import javax.xml.parsers.DocumentBuilderFactory;
|
|
16
|
import javax.xml.parsers.ParserConfigurationException;
|
|
17
|
import javax.xml.xpath.XPath;
|
|
18
|
import javax.xml.xpath.XPathConstants;
|
|
19
|
import javax.xml.xpath.XPathExpressionException;
|
|
20
|
import javax.xml.xpath.XPathFactory;
|
|
21
|
import java.io.IOException;
|
|
22
|
import java.io.StringReader;
|
|
23
|
|
|
24
|
/**
|
|
25
|
* Created by kiatrop on 5/2/2016.
|
|
26
|
*/
|
|
27
|
public class DMFParser {
|
|
28
|
private static final Logger logger = Logger.getLogger(DMFParser.class);
|
|
29
|
|
|
30
|
public static Result dmf2Result(String xml) throws ParserConfigurationException, IOException, SAXException {
|
|
31
|
Result r = new Result();
|
|
32
|
r.setMetadataRecord(xml);
|
|
33
|
r.setRecordFormat(ClaimUtils.FORMAT_XML);
|
|
34
|
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
|
|
35
|
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
|
|
36
|
InputSource inputSource = new InputSource(new StringReader(xml));
|
|
37
|
Document document = dBuilder.parse(inputSource);
|
|
38
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
|
39
|
XPath xpath = xPathfactory.newXPath();
|
|
40
|
try {
|
|
41
|
NodeList nl = (NodeList) xpath.compile("//*[local-name()=\"identifier\"][@identifierType='doi']/text()").evaluate(document, XPathConstants.NODESET);
|
|
42
|
if (nl.getLength() > 0) {
|
|
43
|
r.setDoi(nl.item(0).getNodeValue());
|
|
44
|
r.setExternalUrl("http://dx.doi.org/" + r.getDoi());
|
|
45
|
}
|
|
46
|
nl = (NodeList) xpath.compile("//*[local-name()=\"identifier\"][@identifierType='orcidworkid']/text()").evaluate(document, XPathConstants.NODESET);
|
|
47
|
if (nl.getLength() > 0) {
|
|
48
|
r.setOrcidworkid(nl.item(0).getNodeValue());
|
|
49
|
}
|
|
50
|
|
|
51
|
// DON'T DELETE following lines for orcidworkid!!
|
|
52
|
// @identiferType typo is made on purpose -> there are dmf xml with this typo.
|
|
53
|
if (r.getMetadataRecord().contains("@identiferType='orcidworkid'")) {
|
|
54
|
r.setMetadataRecord(r.getMetadataRecord().replace("@identiferType='orcidworkid'", "@identifierType='orcidworkid'"));
|
|
55
|
nl = (NodeList) xpath.compile("//*[local-name()=\"identifier\"][@identiferType='orcidworkid']/text()").evaluate(document, XPathConstants.NODESET);
|
|
56
|
if (nl.getLength() > 0) {
|
|
57
|
r.setOrcidworkid(nl.item(0).getNodeValue());
|
|
58
|
}
|
|
59
|
}
|
|
60
|
nl = (NodeList) xpath.compile("//*[local-name()='objIdentifier']/text()").evaluate(document, XPathConstants.NODESET);
|
|
61
|
if (nl.getLength() > 0) {
|
|
62
|
r.setOpenaireId(nl.item(0).getNodeValue());
|
|
63
|
}
|
|
64
|
if(r.getExternalUrl() == null) {
|
|
65
|
nl = (NodeList) xpath.compile("//*[local-name()='identifier']/text()").evaluate(document, XPathConstants.NODESET);
|
|
66
|
if (nl.getLength() > 0) {
|
|
67
|
r.setExternalUrl(nl.item(0).getNodeValue());
|
|
68
|
}
|
|
69
|
}
|
|
70
|
nl = (NodeList) xpath.compile("//*[local-name()='accessrights']/text()").evaluate(document, XPathConstants.NODESET);
|
|
71
|
if (nl.getLength() > 0) {
|
|
72
|
r.setAccessRights(nl.item(0).getNodeValue());
|
|
73
|
}
|
|
74
|
nl = (NodeList) xpath.compile("//embargoenddate/text()").evaluate(document, XPathConstants.NODESET);
|
|
75
|
if (nl.getLength() > 0) {
|
|
76
|
r.setEmbargoEndDate(nl.item(0).getNodeValue());
|
|
77
|
}
|
|
78
|
nl = (NodeList) xpath.compile("//*[local-name()='title']/text()").evaluate(document, XPathConstants.NODESET);
|
|
79
|
if (nl.getLength() > 0) {
|
|
80
|
r.setTitle(nl.item(0).getNodeValue());
|
|
81
|
}
|
|
82
|
nl = (NodeList) xpath.compile("//*[local-name()='collectedFrom']/@id").evaluate(document, XPathConstants.NODESET);
|
|
83
|
if (nl.getLength() > 0) {
|
|
84
|
r.setCollectedFrom(nl.item(0).getNodeValue());
|
|
85
|
}
|
|
86
|
|
|
87
|
} catch (Exception e) {
|
|
88
|
logger.error("Error Parsing dmf result:\n "+xml,e);
|
|
89
|
return null;
|
|
90
|
|
|
91
|
}
|
|
92
|
return r;
|
|
93
|
}
|
|
94
|
|
|
95
|
|
|
96
|
/**
|
|
97
|
*
|
|
98
|
* @param xml of a context DMF
|
|
99
|
* @return context id
|
|
100
|
* @throws ParserConfigurationException
|
|
101
|
* @throws IOException
|
|
102
|
* @throws SAXException
|
|
103
|
* @throws XPathExpressionException
|
|
104
|
*/
|
|
105
|
public static String getContextIdFromDMF(String xml) throws ParserConfigurationException, IOException, SAXException, XPathExpressionException {
|
|
106
|
String id=null;
|
|
107
|
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
|
|
108
|
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
|
|
109
|
InputSource inputSource = new InputSource(new StringReader(xml));
|
|
110
|
Document document = dBuilder.parse(inputSource);
|
|
111
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
|
112
|
XPath xpath = xPathfactory.newXPath();
|
|
113
|
|
|
114
|
NodeList nl = (NodeList) xpath.compile("//*[local-name()='concept']/@id").evaluate(document, XPathConstants.NODESET);
|
|
115
|
if (nl.getLength() > 0) {
|
|
116
|
id = (nl.item(0).getNodeValue());
|
|
117
|
}
|
|
118
|
System.out.println(id);
|
|
119
|
return id;
|
|
120
|
}
|
|
121
|
/**
|
|
122
|
*
|
|
123
|
* @param xml
|
|
124
|
* @return DOI
|
|
125
|
* @throws ParserConfigurationException
|
|
126
|
* @throws IOException
|
|
127
|
* @throws SAXException
|
|
128
|
* @throws XPathExpressionException
|
|
129
|
*/
|
|
130
|
public static String getDOIIdentifierFromDMF(String xml) throws ParserConfigurationException, IOException, SAXException, XPathExpressionException {
|
|
131
|
String id=null;
|
|
132
|
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
|
|
133
|
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
|
|
134
|
InputSource inputSource = new InputSource(new StringReader(xml));
|
|
135
|
Document document = dBuilder.parse(inputSource);
|
|
136
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
|
137
|
XPath xpath = xPathfactory.newXPath();
|
|
138
|
|
|
139
|
NodeList nl = (NodeList) xpath.compile("//*[local-name()=\"identifier\"][@identifierType='doi']/text()").evaluate(document, XPathConstants.NODESET);
|
|
140
|
if (nl.getLength() > 0) {
|
|
141
|
id = (nl.item(0).getNodeValue());
|
|
142
|
}
|
|
143
|
|
|
144
|
return id;
|
|
145
|
}
|
|
146
|
|
|
147
|
public static String getOrcidworkIdentifierFromDMF(String xml) throws ParserConfigurationException, IOException, SAXException, XPathExpressionException {
|
|
148
|
String id=null;
|
|
149
|
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
|
|
150
|
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
|
|
151
|
InputSource inputSource = new InputSource(new StringReader(xml));
|
|
152
|
Document document = dBuilder.parse(inputSource);
|
|
153
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
|
154
|
XPath xpath = xPathfactory.newXPath();
|
|
155
|
|
|
156
|
NodeList nl = (NodeList) xpath.compile("//*[local-name()=\"identifier\"][@identifierType='orcidworkid']/text()").evaluate(document, XPathConstants.NODESET);
|
|
157
|
if (nl.getLength() > 0) {
|
|
158
|
id = (nl.item(0).getNodeValue());
|
|
159
|
}
|
|
160
|
|
|
161
|
// DON'T DELETE following lines for orcidworkid!!
|
|
162
|
// @identiferType typo is made on purpose -> there are dmf xml with this typo.
|
|
163
|
if (id== null) {
|
|
164
|
nl = (NodeList) xpath.compile("//*[local-name()=\"identifier\"][@identiferType='orcidworkid']/text()").evaluate(document, XPathConstants.NODESET);
|
|
165
|
if (nl.getLength() > 0) {
|
|
166
|
id = nl.item(0).getNodeValue();
|
|
167
|
}
|
|
168
|
}
|
|
169
|
|
|
170
|
return id;
|
|
171
|
}
|
|
172
|
public static String getAccessRightsFromDMF(String xml) throws ParserConfigurationException, IOException, SAXException, XPathExpressionException {
|
|
173
|
String accessRights=null;
|
|
174
|
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
|
|
175
|
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
|
|
176
|
InputSource inputSource = new InputSource(new StringReader(xml));
|
|
177
|
Document document = dBuilder.parse(inputSource);
|
|
178
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
|
179
|
XPath xpath = xPathfactory.newXPath();
|
|
180
|
|
|
181
|
NodeList nl;
|
|
182
|
nl = (NodeList) xpath.compile("//*[local-name()='accessrights']/text()").evaluate(document, XPathConstants.NODESET);
|
|
183
|
if (nl.getLength() > 0) {
|
|
184
|
accessRights=nl.item(0).getNodeValue();
|
|
185
|
}
|
|
186
|
return accessRights;
|
|
187
|
}
|
|
188
|
|
|
189
|
public static String getEmbargoEndDateFromDMF(String xml) throws ParserConfigurationException, IOException, SAXException, XPathExpressionException {
|
|
190
|
String embargoDate=null;
|
|
191
|
DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance();
|
|
192
|
DocumentBuilder dBuilder = dbFactory.newDocumentBuilder();
|
|
193
|
InputSource inputSource = new InputSource(new StringReader(xml));
|
|
194
|
Document document = dBuilder.parse(inputSource);
|
|
195
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
|
196
|
XPath xpath = xPathfactory.newXPath();
|
|
197
|
|
|
198
|
NodeList nl;
|
|
199
|
nl = (NodeList) xpath.compile("//embargoenddate/text()").evaluate(document, XPathConstants.NODESET);
|
|
200
|
if (nl.getLength() > 0) {
|
|
201
|
embargoDate=nl.item(0).getNodeValue();
|
|
202
|
}
|
|
203
|
return embargoDate;
|
|
204
|
}
|
|
205
|
|
|
206
|
}
|