|
1
|
package eu.dnetlib.oai.conf;
|
|
2
|
|
|
3
|
import java.io.IOException;
|
|
4
|
import java.io.StringReader;
|
|
5
|
import java.util.ArrayList;
|
|
6
|
import java.util.HashMap;
|
|
7
|
import java.util.List;
|
|
8
|
import java.util.Map;
|
|
9
|
import javax.xml.stream.XMLInputFactory;
|
|
10
|
import javax.xml.stream.XMLStreamConstants;
|
|
11
|
import javax.xml.stream.XMLStreamException;
|
|
12
|
import javax.xml.stream.XMLStreamReader;
|
|
13
|
import javax.xml.transform.stream.StreamSource;
|
|
14
|
|
|
15
|
import com.google.common.collect.ArrayListMultimap;
|
|
16
|
import com.google.common.collect.Multimap;
|
|
17
|
import eu.dnetlib.oai.PublisherField;
|
|
18
|
import eu.dnetlib.oai.info.SetInfo;
|
|
19
|
import eu.dnetlib.rmi.provision.MDFInfo;
|
|
20
|
import eu.dnetlib.rmi.provision.OaiPublisherRuntimeException;
|
|
21
|
import eu.dnetlib.utils.MetadataReference;
|
|
22
|
import org.apache.commons.logging.Log;
|
|
23
|
import org.apache.commons.logging.LogFactory;
|
|
24
|
|
|
25
|
/**
|
|
26
|
* Parses an XML document representing the OAI configuration profile and creates the corresponding OAIConfiguration object.
|
|
27
|
*
|
|
28
|
* @author alessia
|
|
29
|
*/
|
|
30
|
public class OAIConfigurationParser {
|
|
31
|
|
|
32
|
private static final Log log = LogFactory.getLog(OAIConfigurationParser.class); // NOPMD by marko on 11/24/08 5:02 PM
|
|
33
|
|
|
34
|
private final ThreadLocal<XMLInputFactory> factory = new ThreadLocal<XMLInputFactory>() {
|
|
35
|
|
|
36
|
@Override
|
|
37
|
protected XMLInputFactory initialValue() {
|
|
38
|
return XMLInputFactory.newInstance();
|
|
39
|
}
|
|
40
|
};
|
|
41
|
|
|
42
|
public OAIConfiguration getConfiguration(final String configurationProfile) throws IOException {
|
|
43
|
log.debug(configurationProfile);
|
|
44
|
final OAIConfiguration config = new OAIConfiguration();
|
|
45
|
final Map<String, SetInfo> setsMap = new HashMap<>();
|
|
46
|
final Map<String, MDFInfo> mdFormatsMap = new HashMap<>();
|
|
47
|
final List<String> indexNames = new ArrayList<>();
|
|
48
|
final List<PublisherField> fields = new ArrayList<>();
|
|
49
|
try {
|
|
50
|
final XMLStreamReader parser = this.factory.get().createXMLStreamReader(new StreamSource(new StringReader(configurationProfile)));
|
|
51
|
while (parser.hasNext()) {
|
|
52
|
int event = parser.next();
|
|
53
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
|
54
|
final String localName = parser.getLocalName();
|
|
55
|
if (localName.equals("IDSCHEME")) {
|
|
56
|
config.setIdScheme(parser.getElementText());
|
|
57
|
} else if (localName.equals("IDNAMESPACE")) {
|
|
58
|
config.setIdNamespace(parser.getElementText());
|
|
59
|
} else if (localName.equals("OAISET")) {
|
|
60
|
boolean inSet = true;
|
|
61
|
final SetInfo setInfo = new SetInfo();
|
|
62
|
final String enabled = parser.getAttributeValue(null, "enabled");
|
|
63
|
setInfo.setEnabled(Boolean.parseBoolean(enabled));
|
|
64
|
while (parser.hasNext() && inSet) {
|
|
65
|
event = parser.next();
|
|
66
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
|
67
|
final String setElementName = parser.getLocalName();
|
|
68
|
final String setElementValue = parser.getElementText();
|
|
69
|
this.handleSetInfo(setInfo, setElementName, setElementValue);
|
|
70
|
}
|
|
71
|
if ((event == XMLStreamConstants.END_ELEMENT) && parser.getLocalName().equals("OAISET")) {
|
|
72
|
inSet = false;
|
|
73
|
setsMap.put(setInfo.getSetSpec(), setInfo);
|
|
74
|
}
|
|
75
|
}
|
|
76
|
} else {
|
|
77
|
if (localName.equals("METADATAFORMAT")) {
|
|
78
|
boolean inMetadata = true;
|
|
79
|
final MDFInfo mdfInfo = new MDFInfo();
|
|
80
|
final String exportable = parser.getAttributeValue(null, "exportable");
|
|
81
|
mdfInfo.setEnabled(Boolean.parseBoolean(exportable));
|
|
82
|
final String mdPrefix = parser.getAttributeValue(null, "metadataPrefix");
|
|
83
|
mdfInfo.setPrefix(mdPrefix);
|
|
84
|
while (parser.hasNext() && inMetadata) {
|
|
85
|
event = parser.next();
|
|
86
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
|
87
|
final String mdfElementName = parser.getLocalName();
|
|
88
|
if (mdfElementName.equals("SOURCE_METADATA_FORMAT")) {
|
|
89
|
this.handleSourceMDF(mdfInfo, parser);
|
|
90
|
config.getSourcesMDF().add(mdfInfo.getSourceMetadataReference());
|
|
91
|
} else {
|
|
92
|
final String mdfElementValue = parser.getElementText();
|
|
93
|
this.handleMDFInfo(mdfInfo, mdfElementName, mdfElementValue);
|
|
94
|
}
|
|
95
|
}
|
|
96
|
if ((event == XMLStreamConstants.END_ELEMENT) && parser.getLocalName().equals("METADATAFORMAT")) {
|
|
97
|
inMetadata = false;
|
|
98
|
mdFormatsMap.put(mdPrefix, mdfInfo);
|
|
99
|
}
|
|
100
|
}
|
|
101
|
} else {
|
|
102
|
// INDICES
|
|
103
|
if (localName.equals("INDEX")) {
|
|
104
|
boolean inIndex = true;
|
|
105
|
final PublisherField publisherField = new PublisherField();
|
|
106
|
final String indexName = parser.getAttributeValue(null, "name");
|
|
107
|
final String repeatable = parser.getAttributeValue(null, "repeatable");
|
|
108
|
final boolean isRepeatable = Boolean.valueOf(repeatable);
|
|
109
|
indexNames.add(indexName);
|
|
110
|
publisherField.setFieldName(indexName);
|
|
111
|
publisherField.setRepeatable(isRepeatable);
|
|
112
|
// now let's set the SOURCES
|
|
113
|
final Multimap<String, String> fieldSources = ArrayListMultimap.create();
|
|
114
|
while (parser.hasNext() && inIndex) {
|
|
115
|
event = parser.next();
|
|
116
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
|
117
|
final String currentElementName = parser.getLocalName();
|
|
118
|
this.handleIndex(fieldSources, indexName, parser, currentElementName);
|
|
119
|
}
|
|
120
|
if ((event == XMLStreamConstants.END_ELEMENT) && parser.getLocalName().equals("INDEX")) {
|
|
121
|
inIndex = false;
|
|
122
|
}
|
|
123
|
}
|
|
124
|
publisherField.setSources(fieldSources);
|
|
125
|
fields.add(publisherField);
|
|
126
|
|
|
127
|
}
|
|
128
|
}
|
|
129
|
}
|
|
130
|
}
|
|
131
|
}
|
|
132
|
config.setFields(fields);
|
|
133
|
config.setFieldNames(indexNames);
|
|
134
|
config.setMdFormatsMap(mdFormatsMap);
|
|
135
|
config.setSetsMap(setsMap);
|
|
136
|
return config;
|
|
137
|
} catch (final XMLStreamException e) {
|
|
138
|
throw new OaiPublisherRuntimeException(e);
|
|
139
|
}
|
|
140
|
|
|
141
|
}
|
|
142
|
|
|
143
|
/**
|
|
144
|
* Sets information about the indices.
|
|
145
|
*
|
|
146
|
* @param fieldSources
|
|
147
|
* @param indexName
|
|
148
|
* @param parser
|
|
149
|
* @param currentLocalName
|
|
150
|
*/
|
|
151
|
private void handleIndex(final Multimap<String, String> fieldSources, final String indexName, final XMLStreamReader parser, final String currentLocalName) {
|
|
152
|
if (currentLocalName.equals("SOURCE")) {
|
|
153
|
final MDFInfo indexSource = new MDFInfo();
|
|
154
|
this.handleSourceMDF(indexSource, parser);
|
|
155
|
final String key =
|
|
156
|
indexSource.getSourceFormat() + "-" + indexSource.getSourceLayout() + "-" + indexSource.getSourceInterpretation();
|
|
157
|
fieldSources.put(key, parser.getAttributeValue(null, "path"));
|
|
158
|
} else {
|
|
159
|
log.warn("I do not know how to handle INDEX element with name " + currentLocalName);
|
|
160
|
}
|
|
161
|
}
|
|
162
|
|
|
163
|
/**
|
|
164
|
* Sets information about the source metadata format in the given instance of MDFInfo.
|
|
165
|
*
|
|
166
|
* @param mdfInfo
|
|
167
|
* @param parser
|
|
168
|
*/
|
|
169
|
private void handleSourceMDF(final MDFInfo mdfInfo, final XMLStreamReader parser) {
|
|
170
|
final String interpretation = parser.getAttributeValue(null, "interpretation");
|
|
171
|
final String layout = parser.getAttributeValue(null, "layout");
|
|
172
|
final String name = parser.getAttributeValue(null, "name");
|
|
173
|
MetadataReference mdref = new MetadataReference(name, layout, interpretation);
|
|
174
|
mdfInfo.setSourceMetadataReference(mdref);
|
|
175
|
}
|
|
176
|
|
|
177
|
/**
|
|
178
|
* Sets information in the MDFInfo instance based on the element name.
|
|
179
|
*
|
|
180
|
* @param mdfInfo
|
|
181
|
* @param elementValue
|
|
182
|
* @param elementName
|
|
183
|
*/
|
|
184
|
private void handleMDFInfo(final MDFInfo mdfInfo, final String elementName, final String elementValue) {
|
|
185
|
if (elementName.equals("NAMESPACE")) {
|
|
186
|
mdfInfo.setNamespace(elementValue);
|
|
187
|
} else {
|
|
188
|
if (elementName.equals("SCHEMA")) {
|
|
189
|
mdfInfo.setSchema(elementValue);
|
|
190
|
} else {
|
|
191
|
if (elementName.equals("TRANSFORMATION_RULE")) {
|
|
192
|
mdfInfo.setTransformationRuleID(elementValue);
|
|
193
|
} else {
|
|
194
|
if (elementName.equals("BASE_QUERY")) {
|
|
195
|
mdfInfo.setBaseQuery(elementValue);
|
|
196
|
} else {
|
|
197
|
log.warn("I do not know how to handle Metadata Format element with name " + elementName + " and value " + elementValue);
|
|
198
|
}
|
|
199
|
}
|
|
200
|
}
|
|
201
|
}
|
|
202
|
|
|
203
|
}
|
|
204
|
|
|
205
|
/**
|
|
206
|
* Sets information in the SetInfo instance based on the element name.
|
|
207
|
*
|
|
208
|
* @param setInfo
|
|
209
|
* @param elementName
|
|
210
|
* @param elementValue
|
|
211
|
*/
|
|
212
|
private void handleSetInfo(final SetInfo setInfo, final String elementName, final String elementValue) {
|
|
213
|
if (elementName.equals("spec")) {
|
|
214
|
setInfo.setSetSpec(elementValue);
|
|
215
|
} else {
|
|
216
|
if (elementName.equals("name")) {
|
|
217
|
setInfo.setSetName(elementValue);
|
|
218
|
} else {
|
|
219
|
if (elementName.equals("description")) {
|
|
220
|
setInfo.setSetDescription(elementValue);
|
|
221
|
} else {
|
|
222
|
if (elementName.equals("query")) {
|
|
223
|
setInfo.setQuery(elementValue);
|
|
224
|
} else {
|
|
225
|
log.warn("I do not know how to handle Set element with name " + elementName + " and value " + elementValue);
|
|
226
|
}
|
|
227
|
}
|
|
228
|
}
|
|
229
|
}
|
|
230
|
}
|
|
231
|
}
|