| 1 |
42181
|
sandro.lab
|
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 |
42184
|
michele.ar
|
|
| 10 |
42181
|
sandro.lab
|
import javax.xml.stream.XMLInputFactory;
|
| 11 |
|
|
import javax.xml.stream.XMLStreamConstants;
|
| 12 |
|
|
import javax.xml.stream.XMLStreamException;
|
| 13 |
|
|
import javax.xml.stream.XMLStreamReader;
|
| 14 |
|
|
import javax.xml.transform.stream.StreamSource;
|
| 15 |
|
|
|
| 16 |
42184
|
michele.ar
|
import org.apache.commons.logging.Log;
|
| 17 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 18 |
|
|
|
| 19 |
42181
|
sandro.lab
|
import com.google.common.collect.ArrayListMultimap;
|
| 20 |
|
|
import com.google.common.collect.Multimap;
|
| 21 |
42184
|
michele.ar
|
|
| 22 |
42181
|
sandro.lab
|
import eu.dnetlib.oai.PublisherField;
|
| 23 |
|
|
import eu.dnetlib.oai.info.SetInfo;
|
| 24 |
|
|
import eu.dnetlib.rmi.provision.MDFInfo;
|
| 25 |
42184
|
michele.ar
|
import eu.dnetlib.rmi.provision.OaiPublisherRuntimeException;
|
| 26 |
42181
|
sandro.lab
|
|
| 27 |
|
|
/**
|
| 28 |
|
|
* Parses an XML document representing the OAI configuration profile and creates the corresponding OAIConfiguration object.
|
| 29 |
|
|
*
|
| 30 |
|
|
* @author alessia
|
| 31 |
|
|
*/
|
| 32 |
|
|
public class OAIConfigurationParser {
|
| 33 |
|
|
|
| 34 |
|
|
private static final Log log = LogFactory.getLog(OAIConfigurationParser.class); // NOPMD by marko on 11/24/08 5:02 PM
|
| 35 |
|
|
|
| 36 |
|
|
private final ThreadLocal<XMLInputFactory> factory = new ThreadLocal<XMLInputFactory>() {
|
| 37 |
|
|
|
| 38 |
|
|
@Override
|
| 39 |
|
|
protected XMLInputFactory initialValue() {
|
| 40 |
|
|
return XMLInputFactory.newInstance();
|
| 41 |
|
|
}
|
| 42 |
|
|
};
|
| 43 |
|
|
|
| 44 |
|
|
public OAIConfiguration getConfiguration(final String configurationProfile) throws IOException {
|
| 45 |
|
|
log.debug(configurationProfile);
|
| 46 |
42184
|
michele.ar
|
final OAIConfiguration config = new OAIConfiguration();
|
| 47 |
|
|
final Map<String, SetInfo> setsMap = new HashMap<>();
|
| 48 |
|
|
final Map<String, MDFInfo> mdFormatsMap = new HashMap<>();
|
| 49 |
|
|
final List<String> indexNames = new ArrayList<>();
|
| 50 |
|
|
final List<PublisherField> fields = new ArrayList<>();
|
| 51 |
42181
|
sandro.lab
|
try {
|
| 52 |
42184
|
michele.ar
|
final XMLStreamReader parser = this.factory.get().createXMLStreamReader(new StreamSource(new StringReader(configurationProfile)));
|
| 53 |
42181
|
sandro.lab
|
while (parser.hasNext()) {
|
| 54 |
|
|
int event = parser.next();
|
| 55 |
|
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
| 56 |
|
|
final String localName = parser.getLocalName();
|
| 57 |
|
|
if (localName.equals("IDSCHEME")) {
|
| 58 |
|
|
config.setIdScheme(parser.getElementText());
|
| 59 |
|
|
} else if (localName.equals("IDNAMESPACE")) {
|
| 60 |
|
|
config.setIdNamespace(parser.getElementText());
|
| 61 |
|
|
} else if (localName.equals("OAISET")) {
|
| 62 |
|
|
boolean inSet = true;
|
| 63 |
42184
|
michele.ar
|
final SetInfo setInfo = new SetInfo();
|
| 64 |
|
|
final String enabled = parser.getAttributeValue(null, "enabled");
|
| 65 |
42181
|
sandro.lab
|
setInfo.setEnabled(Boolean.parseBoolean(enabled));
|
| 66 |
|
|
while (parser.hasNext() && inSet) {
|
| 67 |
|
|
event = parser.next();
|
| 68 |
|
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
| 69 |
42184
|
michele.ar
|
final String setElementName = parser.getLocalName();
|
| 70 |
|
|
final String setElementValue = parser.getElementText();
|
| 71 |
42181
|
sandro.lab
|
this.handleSetInfo(setInfo, setElementName, setElementValue);
|
| 72 |
|
|
}
|
| 73 |
|
|
if ((event == XMLStreamConstants.END_ELEMENT) && parser.getLocalName().equals("OAISET")) {
|
| 74 |
|
|
inSet = false;
|
| 75 |
|
|
setsMap.put(setInfo.getSetSpec(), setInfo);
|
| 76 |
|
|
}
|
| 77 |
|
|
}
|
| 78 |
|
|
} else {
|
| 79 |
|
|
if (localName.equals("METADATAFORMAT")) {
|
| 80 |
|
|
boolean inMetadata = true;
|
| 81 |
42184
|
michele.ar
|
final MDFInfo mdfInfo = new MDFInfo();
|
| 82 |
|
|
final String exportable = parser.getAttributeValue(null, "exportable");
|
| 83 |
42181
|
sandro.lab
|
mdfInfo.setEnabled(Boolean.parseBoolean(exportable));
|
| 84 |
42184
|
michele.ar
|
final String mdPrefix = parser.getAttributeValue(null, "metadataPrefix");
|
| 85 |
42181
|
sandro.lab
|
mdfInfo.setPrefix(mdPrefix);
|
| 86 |
|
|
while (parser.hasNext() && inMetadata) {
|
| 87 |
|
|
event = parser.next();
|
| 88 |
|
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
| 89 |
42184
|
michele.ar
|
final String mdfElementName = parser.getLocalName();
|
| 90 |
42181
|
sandro.lab
|
if (mdfElementName.equals("SOURCE_METADATA_FORMAT")) {
|
| 91 |
|
|
this.handleSourceMDF(mdfInfo, parser);
|
| 92 |
|
|
config.getSourcesMDF().add(mdfInfo);
|
| 93 |
|
|
} else {
|
| 94 |
42184
|
michele.ar
|
final String mdfElementValue = parser.getElementText();
|
| 95 |
42181
|
sandro.lab
|
this.handleMDFInfo(mdfInfo, mdfElementName, mdfElementValue);
|
| 96 |
|
|
}
|
| 97 |
|
|
}
|
| 98 |
|
|
if ((event == XMLStreamConstants.END_ELEMENT) && parser.getLocalName().equals("METADATAFORMAT")) {
|
| 99 |
|
|
inMetadata = false;
|
| 100 |
|
|
mdFormatsMap.put(mdPrefix, mdfInfo);
|
| 101 |
|
|
}
|
| 102 |
|
|
}
|
| 103 |
|
|
} else {
|
| 104 |
|
|
// INDICES
|
| 105 |
|
|
if (localName.equals("INDEX")) {
|
| 106 |
|
|
boolean inIndex = true;
|
| 107 |
42184
|
michele.ar
|
final PublisherField publisherField = new PublisherField();
|
| 108 |
|
|
final String indexName = parser.getAttributeValue(null, "name");
|
| 109 |
|
|
final String repeatable = parser.getAttributeValue(null, "repeatable");
|
| 110 |
|
|
final boolean isRepeatable = Boolean.valueOf(repeatable);
|
| 111 |
42181
|
sandro.lab
|
indexNames.add(indexName);
|
| 112 |
|
|
publisherField.setFieldName(indexName);
|
| 113 |
|
|
publisherField.setRepeatable(isRepeatable);
|
| 114 |
|
|
// now let's set the SOURCES
|
| 115 |
42184
|
michele.ar
|
final Multimap<String, String> fieldSources = ArrayListMultimap.create();
|
| 116 |
42181
|
sandro.lab
|
while (parser.hasNext() && inIndex) {
|
| 117 |
|
|
event = parser.next();
|
| 118 |
|
|
if (event == XMLStreamConstants.START_ELEMENT) {
|
| 119 |
42184
|
michele.ar
|
final String currentElementName = parser.getLocalName();
|
| 120 |
42181
|
sandro.lab
|
this.handleIndex(fieldSources, indexName, parser, currentElementName);
|
| 121 |
|
|
}
|
| 122 |
|
|
if ((event == XMLStreamConstants.END_ELEMENT) && parser.getLocalName().equals("INDEX")) {
|
| 123 |
|
|
inIndex = false;
|
| 124 |
|
|
}
|
| 125 |
|
|
}
|
| 126 |
|
|
publisherField.setSources(fieldSources);
|
| 127 |
|
|
fields.add(publisherField);
|
| 128 |
|
|
|
| 129 |
|
|
}
|
| 130 |
|
|
}
|
| 131 |
|
|
}
|
| 132 |
|
|
}
|
| 133 |
|
|
}
|
| 134 |
|
|
config.setFields(fields);
|
| 135 |
|
|
config.setFieldNames(indexNames);
|
| 136 |
|
|
config.setMdFormatsMap(mdFormatsMap);
|
| 137 |
|
|
config.setSetsMap(setsMap);
|
| 138 |
|
|
return config;
|
| 139 |
|
|
} catch (final XMLStreamException e) {
|
| 140 |
|
|
throw new OaiPublisherRuntimeException(e);
|
| 141 |
|
|
}
|
| 142 |
|
|
|
| 143 |
|
|
}
|
| 144 |
|
|
|
| 145 |
|
|
/**
|
| 146 |
|
|
* Sets information about the indices.
|
| 147 |
|
|
*
|
| 148 |
|
|
* @param fieldSources
|
| 149 |
|
|
* @param indexName
|
| 150 |
|
|
* @param parser
|
| 151 |
|
|
* @param currentLocalName
|
| 152 |
|
|
*/
|
| 153 |
|
|
private void handleIndex(final Multimap<String, String> fieldSources, final String indexName, final XMLStreamReader parser, final String currentLocalName) {
|
| 154 |
|
|
if (currentLocalName.equals("SOURCE")) {
|
| 155 |
42184
|
michele.ar
|
final MDFInfo indexSource = new MDFInfo();
|
| 156 |
42181
|
sandro.lab
|
this.handleSourceMDF(indexSource, parser);
|
| 157 |
42184
|
michele.ar
|
final String key =
|
| 158 |
|
|
indexSource.getSourceFormatName() + "-" + indexSource.getSourceFormatLayout() + "-" + indexSource.getSourceFormatInterpretation();
|
| 159 |
42181
|
sandro.lab
|
fieldSources.put(key, parser.getAttributeValue(null, "path"));
|
| 160 |
|
|
} else {
|
| 161 |
|
|
log.warn("I do not know how to handle INDEX element with name " + currentLocalName);
|
| 162 |
|
|
}
|
| 163 |
|
|
}
|
| 164 |
|
|
|
| 165 |
|
|
/**
|
| 166 |
|
|
* Sets information about the source metadata format in the given instance of MDFInfo.
|
| 167 |
|
|
*
|
| 168 |
|
|
* @param mdfInfo
|
| 169 |
|
|
* @param parser
|
| 170 |
|
|
*/
|
| 171 |
|
|
private void handleSourceMDF(final MDFInfo mdfInfo, final XMLStreamReader parser) {
|
| 172 |
42184
|
michele.ar
|
final String interpretation = parser.getAttributeValue(null, "interpretation");
|
| 173 |
|
|
final String layout = parser.getAttributeValue(null, "layout");
|
| 174 |
|
|
final String name = parser.getAttributeValue(null, "name");
|
| 175 |
42181
|
sandro.lab
|
mdfInfo.setSourceFormatInterpretation(interpretation);
|
| 176 |
|
|
mdfInfo.setSourceFormatLayout(layout);
|
| 177 |
|
|
mdfInfo.setSourceFormatName(name);
|
| 178 |
|
|
}
|
| 179 |
|
|
|
| 180 |
|
|
/**
|
| 181 |
|
|
* Sets information in the MDFInfo instance based on the element name.
|
| 182 |
|
|
*
|
| 183 |
|
|
* @param mdfInfo
|
| 184 |
|
|
* @param elementValue
|
| 185 |
|
|
* @param elementName
|
| 186 |
|
|
*/
|
| 187 |
|
|
private void handleMDFInfo(final MDFInfo mdfInfo, final String elementName, final String elementValue) {
|
| 188 |
|
|
if (elementName.equals("NAMESPACE")) {
|
| 189 |
|
|
mdfInfo.setNamespace(elementValue);
|
| 190 |
|
|
} else {
|
| 191 |
|
|
if (elementName.equals("SCHEMA")) {
|
| 192 |
|
|
mdfInfo.setSchema(elementValue);
|
| 193 |
|
|
} else {
|
| 194 |
|
|
if (elementName.equals("TRANSFORMATION_RULE")) {
|
| 195 |
|
|
mdfInfo.setTransformationRuleID(elementValue);
|
| 196 |
|
|
} else {
|
| 197 |
|
|
if (elementName.equals("BASE_QUERY")) {
|
| 198 |
|
|
mdfInfo.setBaseQuery(elementValue);
|
| 199 |
|
|
} else {
|
| 200 |
|
|
log.warn("I do not know how to handle Metadata Format element with name " + elementName + " and value " + elementValue);
|
| 201 |
|
|
}
|
| 202 |
|
|
}
|
| 203 |
|
|
}
|
| 204 |
|
|
}
|
| 205 |
|
|
|
| 206 |
|
|
}
|
| 207 |
|
|
|
| 208 |
|
|
/**
|
| 209 |
|
|
* Sets information in the SetInfo instance based on the element name.
|
| 210 |
|
|
*
|
| 211 |
|
|
* @param setInfo
|
| 212 |
|
|
* @param elementName
|
| 213 |
|
|
* @param elementValue
|
| 214 |
|
|
*/
|
| 215 |
|
|
private void handleSetInfo(final SetInfo setInfo, final String elementName, final String elementValue) {
|
| 216 |
|
|
if (elementName.equals("spec")) {
|
| 217 |
|
|
setInfo.setSetSpec(elementValue);
|
| 218 |
|
|
} else {
|
| 219 |
|
|
if (elementName.equals("name")) {
|
| 220 |
|
|
setInfo.setSetName(elementValue);
|
| 221 |
|
|
} else {
|
| 222 |
|
|
if (elementName.equals("description")) {
|
| 223 |
|
|
setInfo.setSetDescription(elementValue);
|
| 224 |
|
|
} else {
|
| 225 |
|
|
if (elementName.equals("query")) {
|
| 226 |
|
|
setInfo.setQuery(elementValue);
|
| 227 |
|
|
} else {
|
| 228 |
|
|
log.warn("I do not know how to handle Set element with name " + elementName + " and value " + elementValue);
|
| 229 |
|
|
}
|
| 230 |
|
|
}
|
| 231 |
|
|
}
|
| 232 |
|
|
}
|
| 233 |
|
|
}
|
| 234 |
|
|
}
|