1
|
package eu.dnetlib.data.collector.plugins.datasources;
|
2
|
|
3
|
import java.io.InputStream;
|
4
|
import java.util.Iterator;
|
5
|
import java.util.NoSuchElementException;
|
6
|
import javax.xml.stream.XMLInputFactory;
|
7
|
import javax.xml.stream.XMLStreamConstants;
|
8
|
import javax.xml.stream.XMLStreamException;
|
9
|
import javax.xml.stream.XMLStreamReader;
|
10
|
|
11
|
import eu.dnetlib.data.collector.plugins.HttpConnector;
|
12
|
import eu.dnetlib.data.collector.rmi.CollectorServiceException;
|
13
|
import eu.dnetlib.data.collector.rmi.CollectorServiceRuntimeException;
|
14
|
import org.apache.commons.logging.Log;
|
15
|
import org.apache.commons.logging.LogFactory;
|
16
|
|
17
|
public class Re3DataRepositoriesIterator implements Iterator<String>, Iterable<String> {
|
18
|
|
19
|
private static final Log log = LogFactory.getLog(Re3DataRepositoriesIterator.class); // NOPMD by marko on 11/24/08 5:02 PM
|
20
|
|
21
|
private String baseURL;
|
22
|
private XMLStreamReader reader;
|
23
|
private int countedRepos = 0;
|
24
|
private String currentRepoPath = null;
|
25
|
|
26
|
private HttpConnector httpConnector;
|
27
|
|
28
|
@Override
|
29
|
public boolean hasNext() {
|
30
|
return currentRepoPath != null;
|
31
|
}
|
32
|
|
33
|
@Override
|
34
|
public String next() {
|
35
|
if (currentRepoPath == null) throw new NoSuchElementException();
|
36
|
|
37
|
try {
|
38
|
String repoInfo = getRepositoryInfo(currentRepoPath);
|
39
|
return repoInfo;
|
40
|
} finally {
|
41
|
currentRepoPath = moveToNextRepo();
|
42
|
}
|
43
|
}
|
44
|
|
45
|
@Override
|
46
|
public void remove() {
|
47
|
throw new UnsupportedOperationException();
|
48
|
}
|
49
|
|
50
|
@Override
|
51
|
public Iterator<String> iterator() {
|
52
|
return this;
|
53
|
}
|
54
|
|
55
|
public Re3DataRepositoriesIterator(final InputStream xmlInputStream, final String baseUrl, final HttpConnector httpConnector) throws CollectorServiceException {
|
56
|
this.httpConnector = httpConnector;
|
57
|
XMLInputFactory factory = XMLInputFactory.newInstance();
|
58
|
try {
|
59
|
reader = factory.createXMLStreamReader(xmlInputStream);
|
60
|
} catch (XMLStreamException e) {
|
61
|
throw new CollectorServiceException(e);
|
62
|
}
|
63
|
baseURL = baseUrl;
|
64
|
|
65
|
// try to fetch the 1st
|
66
|
currentRepoPath = moveToNextRepo();
|
67
|
}
|
68
|
|
69
|
private String getNextRepositoryPath() {
|
70
|
return reader.getAttributeValue(null, "href");
|
71
|
}
|
72
|
|
73
|
private String moveToNextRepo() {
|
74
|
try {
|
75
|
while (reader.hasNext()) {
|
76
|
int event = reader.next();
|
77
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
78
|
String elementName = reader.getLocalName();
|
79
|
if (elementName.equals("link")) {
|
80
|
String repoPath = getNextRepositoryPath();
|
81
|
log.debug(String.format("Found %s repositories. The last has link %s", ++countedRepos, repoPath));
|
82
|
return repoPath;
|
83
|
}
|
84
|
}
|
85
|
}
|
86
|
log.info("Seems there are no more repository to iterate on. Total: " + countedRepos);
|
87
|
return null;
|
88
|
} catch (XMLStreamException e) {
|
89
|
throw new CollectorServiceRuntimeException(e);
|
90
|
}
|
91
|
}
|
92
|
|
93
|
private String getRepositoryInfo(final String repositoryPath) throws CollectorServiceRuntimeException {
|
94
|
String targetURL = repositoryPath;
|
95
|
if(!repositoryPath.startsWith(baseURL))
|
96
|
targetURL = baseURL + repositoryPath;
|
97
|
try {
|
98
|
return getHttpConnector().getInputSource(targetURL);
|
99
|
} catch (CollectorServiceException e) {
|
100
|
throw new CollectorServiceRuntimeException("OOOPS something bad happen getting repo info from " + targetURL, e);
|
101
|
}
|
102
|
}
|
103
|
|
104
|
public String getBaseURL() {
|
105
|
return baseURL;
|
106
|
}
|
107
|
|
108
|
public void setBaseURL(final String baseURL) {
|
109
|
this.baseURL = baseURL;
|
110
|
}
|
111
|
|
112
|
public int getCountedRepos() {
|
113
|
return countedRepos;
|
114
|
}
|
115
|
|
116
|
public void setCountedRepos(final int countedRepos) {
|
117
|
this.countedRepos = countedRepos;
|
118
|
}
|
119
|
|
120
|
public XMLStreamReader getReader() {
|
121
|
return reader;
|
122
|
}
|
123
|
|
124
|
public void setReader(final XMLStreamReader reader) {
|
125
|
this.reader = reader;
|
126
|
}
|
127
|
|
128
|
public String getCurrentRepoPath() {
|
129
|
return currentRepoPath;
|
130
|
}
|
131
|
|
132
|
public void setCurrentRepoPath(final String currentRepoPath) {
|
133
|
this.currentRepoPath = currentRepoPath;
|
134
|
}
|
135
|
|
136
|
public HttpConnector getHttpConnector() {
|
137
|
return httpConnector;
|
138
|
}
|
139
|
|
140
|
public void setHttpConnector(final HttpConnector httpConnector) {
|
141
|
this.httpConnector = httpConnector;
|
142
|
}
|
143
|
}
|