Revision 52861
Added by Miriam Baglioni over 6 years ago
| modules/dnet-openaireplus-workflows/branches/solr7/src/main/java/eu/dnetlib/msro/openaireplus/workflows/nodes/bulktag/LoadBulkTaggingConfigurationJobNode.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.msro.openaireplus.workflows.nodes.bulktag; |
| 2 | 2 |
|
| 3 |
import java.io.StringReader; |
|
| 4 |
import java.util.ArrayList; |
|
| 3 | 5 |
import java.util.List; |
| 4 | 6 |
|
| 5 | 7 |
import com.googlecode.sarasvati.Arc; |
| ... | ... | |
| 10 | 12 |
import org.apache.commons.logging.Log; |
| 11 | 13 |
import org.apache.commons.logging.LogFactory; |
| 12 | 14 |
import org.springframework.beans.factory.annotation.Autowired; |
| 15 |
import org.dom4j.Document; |
|
| 16 |
import org.dom4j.DocumentException; |
|
| 17 |
import org.dom4j.Node; |
|
| 18 |
import org.dom4j.io.SAXReader; |
|
| 13 | 19 |
|
| 20 |
|
|
| 14 | 21 |
public class LoadBulkTaggingConfigurationJobNode extends SimpleJobNode {
|
| 15 | 22 |
|
| 16 | 23 |
private static final Log log = LogFactory.getLog(LoadBulkTaggingConfigurationJobNode.class); |
| ... | ... | |
| 24 | 31 |
|
| 25 | 32 |
@Override |
| 26 | 33 |
protected String execute(final NodeToken token) throws Exception {
|
| 27 |
|
|
| 34 |
xquery = "for $x in collection('/db/DRIVER/ContextDSResources/ContextDSResourceType') \n" +
|
|
| 35 |
"let $id := $x//CONFIGURATION/context/@id\n" + |
|
| 36 |
"let $subj := $x//CONFIGURATION/context/param[./@name='subject']/text()\n" + |
|
| 37 |
"let $datasources := $x//CONFIGURATION/context/category[./@id=concat($id,'::contentproviders')]/concept\n" + |
|
| 38 |
"let $communities := $x//CONFIGURATION/context/category[./@id=concat($id,'::zenodocommunities')]/concept\n" + |
|
| 39 |
"where $x//CONFIGURATION/context[./@type='community'] \n" + |
|
| 40 |
"return \n" + |
|
| 41 |
"<community>\n" + |
|
| 42 |
"{ $x//CONFIGURATION/context/@id}\n" +
|
|
| 43 |
"<subjects>\n" + |
|
| 44 |
"{for $y in tokenize($subj,',')\n" +
|
|
| 45 |
"return \n" + |
|
| 46 |
"<subj>{$y}</subj>}\n" +
|
|
| 47 |
"</subjects>\n" + |
|
| 48 |
"<datasources>\n" + |
|
| 49 |
"{for $d in $datasources\n" +
|
|
| 50 |
"return \n" + |
|
| 51 |
"<datasource>\n" + |
|
| 52 |
"<openaireId>\n" + |
|
| 53 |
"{$d//param[./@name='openaireId']/text()}\n" +
|
|
| 54 |
"</openaireId>\n" + |
|
| 55 |
"<selcriteria>\n" + |
|
| 56 |
"{$d//param[./@name='selcriteria']/text()}\n" +
|
|
| 57 |
"</selcriteria>\n" + |
|
| 58 |
"</datasource>\n" + |
|
| 59 |
"}\n" + |
|
| 60 |
"</datasources>\n" + |
|
| 61 |
"<zenodocommunities>\n" + |
|
| 62 |
"{$communities}\n" +
|
|
| 63 |
"</zenodocommunities>\n" + |
|
| 64 |
"</community>"; |
|
| 28 | 65 |
final List<String> res = serviceLocator.getService(ISLookUpService.class).quickSearchProfile(getXquery()); |
| 29 | 66 |
|
| 67 |
Configuration c = new Configuration(); |
|
| 30 | 68 |
|
| 31 |
//TODO manipulate res |
|
| 69 |
for(String r : res){
|
|
| 70 |
c.addCommunity(parseCommunity(r)); |
|
| 71 |
} |
|
| 32 | 72 |
|
| 33 |
token.getEnv().setAttribute(getBulkTaggingConfParam(), res.toString());
|
|
| 73 |
token.getEnv().setAttribute(getBulkTaggingConfParam(), c.getTaggingConfiguration());
|
|
| 34 | 74 |
|
| 75 |
log.info("tagging configuration " + c.getTaggingConfiguration());
|
|
| 35 | 76 |
return Arc.DEFAULT_ARC; |
| 36 | 77 |
} |
| 37 | 78 |
|
| 79 |
private Community parseCommunity(String r) {
|
|
| 80 |
|
|
| 81 |
try{
|
|
| 82 |
SAXReader reader = new SAXReader(); |
|
| 83 |
Document doc = reader.read(new StringReader(r)); |
|
| 84 |
Community c = new Community(); |
|
| 85 |
c.setId(doc.valueOf("./@id"));
|
|
| 86 |
|
|
| 87 |
List <Node> list = doc.selectNodes("//subject");
|
|
| 88 |
final List<String> sbj = new ArrayList<>(); |
|
| 89 |
for(Node n : list){
|
|
| 90 |
sbj.add(n.getText()); |
|
| 91 |
} |
|
| 92 |
if(sbj.size()>0) |
|
| 93 |
c.setSubjects(sbj); |
|
| 94 |
|
|
| 95 |
list = doc.selectNodes("//datasource");
|
|
| 96 |
final List<Datasource> datasourceList = new ArrayList<>(); |
|
| 97 |
for(Node n : list){
|
|
| 98 |
Datasource d = new Datasource(); |
|
| 99 |
Document dd = n.getDocument(); |
|
| 100 |
d.setOpenaireId(dd.valueOf(("./openaireId")));
|
|
| 101 |
d.setSelCriteria(dd.valueOf("./selcriteria"));
|
|
| 102 |
datasourceList.add(d); |
|
| 103 |
} |
|
| 104 |
if(datasourceList.size()>0) |
|
| 105 |
c.setDatasources(datasourceList); |
|
| 106 |
|
|
| 107 |
list = doc.selectNodes("//zenodocommunities");
|
|
| 108 |
final List<ZenodoCommunity> zenodoCommunityList = new ArrayList<>(); |
|
| 109 |
for(Node n : list){
|
|
| 110 |
ZenodoCommunity zc = new ZenodoCommunity(); |
|
| 111 |
Document dd = n.getDocument(); |
|
| 112 |
zc.setZenodoCommunityId(dd.valueOf("./zenodoid"));
|
|
| 113 |
zc.setSelCriteria(dd.valueOf("./selcriteria"));
|
|
| 114 |
zenodoCommunityList.add(zc); |
|
| 115 |
} |
|
| 116 |
if(zenodoCommunityList.size()>0) |
|
| 117 |
c.setZenodoCommunities(zenodoCommunityList); |
|
| 118 |
if(c.getSubjects() == null && c.getDatasources() == null && c.getZenodoCommunities() == null) |
|
| 119 |
return null; |
|
| 120 |
return c; |
|
| 121 |
}catch(DocumentException e){
|
|
| 122 |
log.error("Impossible to read the configuration for " + r);
|
|
| 123 |
return null; |
|
| 124 |
} |
|
| 125 |
|
|
| 126 |
} |
|
| 127 |
|
|
| 38 | 128 |
public String getXquery() {
|
| 39 | 129 |
return xquery; |
| 40 | 130 |
} |
Also available in: Unified diff
processing for bulk tagging configuration