1
|
package eu.dnetlib.msro.workflows.nodes.download;
|
2
|
|
3
|
import java.io.ByteArrayInputStream;
|
4
|
import java.util.ArrayList;
|
5
|
import java.util.List;
|
6
|
|
7
|
import javax.xml.parsers.DocumentBuilder;
|
8
|
import javax.xml.parsers.DocumentBuilderFactory;
|
9
|
import javax.xml.xpath.XPath;
|
10
|
import javax.xml.xpath.XPathConstants;
|
11
|
import javax.xml.xpath.XPathExpression;
|
12
|
import javax.xml.xpath.XPathFactory;
|
13
|
|
14
|
import org.apache.commons.lang.StringUtils;
|
15
|
import org.apache.commons.logging.Log;
|
16
|
import org.apache.commons.logging.LogFactory;
|
17
|
import org.joda.time.DateTime;
|
18
|
import org.joda.time.format.DateTimeFormat;
|
19
|
import org.joda.time.format.DateTimeFormatter;
|
20
|
import org.w3c.dom.Document;
|
21
|
import org.w3c.dom.NodeList;
|
22
|
|
23
|
import com.google.common.base.Function;
|
24
|
import com.google.gson.Gson;
|
25
|
|
26
|
import eu.dnetlib.data.download.rmi.DownloadItem;
|
27
|
|
28
|
// TODO: Auto-generated Javadoc
|
29
|
/**
|
30
|
* The Class UrlExtractor.
|
31
|
*/
|
32
|
public class UrlExtractor implements Function<String, String> {
|
33
|
|
34
|
/** The Constant log. */
|
35
|
private static final Log log = LogFactory.getLog(UrlExtractor.class);
|
36
|
|
37
|
/** The xpath url. */
|
38
|
private String xpathURL;
|
39
|
|
40
|
/** The xpath. */
|
41
|
private String xpathMetadataID;
|
42
|
|
43
|
/** The xpath open access. */
|
44
|
private String xpathOpenAccess;
|
45
|
|
46
|
/** The xpath embargo date. */
|
47
|
private String xpathEmbargoDate;
|
48
|
|
49
|
/**
|
50
|
* Instantiates a new url extractor.
|
51
|
*
|
52
|
* @param xpath
|
53
|
* the xpath
|
54
|
* @param xpathMetadataID
|
55
|
* the xpath metadata id
|
56
|
*/
|
57
|
public UrlExtractor(final String xpath, final String xpathMetadataID, final String xpathOpenAccess, final String xpathEmbargoDate) {
|
58
|
this.xpathURL = xpath;
|
59
|
this.xpathMetadataID = xpathMetadataID;
|
60
|
this.xpathOpenAccess = xpathOpenAccess;
|
61
|
this.xpathEmbargoDate = xpathEmbargoDate;
|
62
|
}
|
63
|
|
64
|
/*
|
65
|
* (non-Javadoc)
|
66
|
*
|
67
|
* @see com.google.common.base.Function#apply(java.lang.Object)
|
68
|
*/
|
69
|
@Override
|
70
|
public String apply(final String input) {
|
71
|
try {
|
72
|
DownloadItem di = new DownloadItem();
|
73
|
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
74
|
DocumentBuilder builder;
|
75
|
builder = factory.newDocumentBuilder();
|
76
|
if (input == null) {
|
77
|
log.error("Metadata input is null");
|
78
|
return null;
|
79
|
}
|
80
|
Document doc = builder.parse(new ByteArrayInputStream(input.getBytes()));
|
81
|
XPathFactory xPathFactory = XPathFactory.newInstance();
|
82
|
XPath myXpath = xPathFactory.newXPath();
|
83
|
XPathExpression expression = myXpath.compile(xpathURL);
|
84
|
Object values = expression.evaluate(doc, XPathConstants.NODESET);
|
85
|
di.setUrl(getNodes((NodeList) values));
|
86
|
di.setOriginalUrl(getNodes((NodeList) values));
|
87
|
|
88
|
if (xpathOpenAccess != null) {
|
89
|
expression = myXpath.compile(xpathOpenAccess);
|
90
|
String openAccess = expression.evaluate(doc);
|
91
|
di.setOpenAccess(openAccess);
|
92
|
}
|
93
|
if (xpathEmbargoDate != null) {
|
94
|
expression = myXpath.compile(xpathEmbargoDate);
|
95
|
String embargoDate = expression.evaluate(doc);
|
96
|
if (!StringUtils.isEmpty(embargoDate)) {
|
97
|
try {
|
98
|
DateTimeFormatter fmt = DateTimeFormat.forPattern("yyyy-MM-dd");
|
99
|
DateTime dt = fmt.parseDateTime(embargoDate);
|
100
|
di.setEmbargoDate(dt.toDate());
|
101
|
} catch (Exception pe) {}
|
102
|
}
|
103
|
}
|
104
|
expression = myXpath.compile(xpathMetadataID);
|
105
|
String extracted_metadataId = expression.evaluate(doc);
|
106
|
di.setIdItemMetadata(extracted_metadataId);
|
107
|
return di.toJSON();
|
108
|
} catch (Exception e) {
|
109
|
log.error("OPSSS... Something bad happen on evaluating ", e);
|
110
|
return null;
|
111
|
}
|
112
|
|
113
|
}
|
114
|
|
115
|
/**
|
116
|
* Gets the nodes.
|
117
|
*
|
118
|
* @param nodes
|
119
|
* the nodes
|
120
|
* @return the nodes
|
121
|
*/
|
122
|
private String getNodes(final NodeList nodes) {
|
123
|
List<String> extracted_Url = new ArrayList<String>();
|
124
|
if (nodes != null) {
|
125
|
for (int i = 0; i < nodes.getLength(); i++) {
|
126
|
extracted_Url.add(nodes.item(i).getNodeValue());
|
127
|
}
|
128
|
}
|
129
|
return new Gson().toJson(extracted_Url);
|
130
|
}
|
131
|
|
132
|
/**
|
133
|
* Gets the xpath metadata id.
|
134
|
*
|
135
|
* @return the xpathMetadataID
|
136
|
*/
|
137
|
public String getXpathMetadataID() {
|
138
|
return xpathMetadataID;
|
139
|
}
|
140
|
|
141
|
/**
|
142
|
* Sets the xpath metadata id.
|
143
|
*
|
144
|
* @param xpathMetadataID
|
145
|
* the xpathMetadataID to set
|
146
|
*/
|
147
|
public void setXpathMetadataID(final String xpathMetadataID) {
|
148
|
this.xpathMetadataID = xpathMetadataID;
|
149
|
}
|
150
|
|
151
|
/**
|
152
|
* Gets the xpath url.
|
153
|
*
|
154
|
* @return the xpath url
|
155
|
*/
|
156
|
public String getXpathURL() {
|
157
|
return xpathURL;
|
158
|
}
|
159
|
|
160
|
/**
|
161
|
* Sets the xpath url.
|
162
|
*
|
163
|
* @param xpathURL
|
164
|
* the new xpath url
|
165
|
*/
|
166
|
public void setXpathURL(final String xpathURL) {
|
167
|
this.xpathURL = xpathURL;
|
168
|
}
|
169
|
|
170
|
/**
|
171
|
* Gets the xpath open access.
|
172
|
*
|
173
|
* @return the xpath open access
|
174
|
*/
|
175
|
public String getXpathOpenAccess() {
|
176
|
return xpathOpenAccess;
|
177
|
}
|
178
|
|
179
|
/**
|
180
|
* Sets the xpath open access.
|
181
|
*
|
182
|
* @param xpathOpenAccess
|
183
|
* the new xpath open access
|
184
|
*/
|
185
|
public void setXpathOpenAccess(final String xpathOpenAccess) {
|
186
|
this.xpathOpenAccess = xpathOpenAccess;
|
187
|
}
|
188
|
|
189
|
/**
|
190
|
* Gets the xpath embargo date.
|
191
|
*
|
192
|
* @return the xpath embargo date
|
193
|
*/
|
194
|
public String getXpathEmbargoDate() {
|
195
|
return xpathEmbargoDate;
|
196
|
}
|
197
|
|
198
|
/**
|
199
|
* Sets the xpath embargo date.
|
200
|
*
|
201
|
* @param xpathEmbargoDate
|
202
|
* the new xpath embargo date
|
203
|
*/
|
204
|
public void setXpathEmbargoDate(final String xpathEmbargoDate) {
|
205
|
this.xpathEmbargoDate = xpathEmbargoDate;
|
206
|
}
|
207
|
|
208
|
}
|