1 |
36418
|
claudio.at
|
package eu.dnetlib.msro.workflows.dedup.conf;
|
2 |
|
|
|
3 |
|
|
import java.io.StringReader;
|
4 |
37127
|
claudio.at
|
import java.util.List;
|
5 |
36418
|
claudio.at
|
import java.util.Queue;
|
6 |
|
|
|
7 |
|
|
import javax.annotation.Resource;
|
8 |
|
|
|
9 |
|
|
import org.apache.commons.logging.Log;
|
10 |
|
|
import org.apache.commons.logging.LogFactory;
|
11 |
|
|
import org.dom4j.Document;
|
12 |
|
|
import org.dom4j.DocumentException;
|
13 |
|
|
import org.dom4j.Element;
|
14 |
|
|
import org.dom4j.io.SAXReader;
|
15 |
|
|
|
16 |
|
|
import com.google.common.collect.Lists;
|
17 |
|
|
|
18 |
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpDocumentNotFoundException;
|
19 |
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
|
20 |
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
21 |
|
|
import eu.dnetlib.enabling.locators.UniqueServiceLocator;
|
22 |
36623
|
claudio.at
|
import eu.dnetlib.pace.config.DedupConfig;
|
23 |
36418
|
claudio.at
|
|
24 |
|
|
/**
|
25 |
|
|
* The Class DedupConfigurationOrchestrationLoader.
|
26 |
|
|
*/
|
27 |
|
|
public class DedupConfigurationOrchestrationLoader {
|
28 |
|
|
|
29 |
|
|
/** The Constant log. */
|
30 |
|
|
private static final Log log = LogFactory.getLog(DedupConfigurationOrchestrationLoader.class);
|
31 |
|
|
|
32 |
|
|
/** The service locator. */
|
33 |
|
|
@Resource
|
34 |
|
|
private UniqueServiceLocator serviceLocator;
|
35 |
|
|
|
36 |
|
|
/**
|
37 |
|
|
* Load the dedup orchestration profile from the IS.
|
38 |
|
|
*
|
39 |
|
|
* @param id
|
40 |
|
|
* the id
|
41 |
|
|
* @return the dedup configuration orchestration
|
42 |
|
|
* @throws ISLookUpDocumentNotFoundException
|
43 |
|
|
* the IS look up document not found exception
|
44 |
|
|
* @throws ISLookUpException
|
45 |
|
|
* the IS look up exception
|
46 |
|
|
* @throws DocumentException
|
47 |
|
|
* the document exception
|
48 |
|
|
*/
|
49 |
37127
|
claudio.at
|
public DedupConfigurationOrchestration loadByActionSetId(final String id) throws Exception {
|
50 |
36418
|
claudio.at
|
|
51 |
|
|
final ISLookUpService isLookUpService = serviceLocator.getService(ISLookUpService.class);
|
52 |
|
|
|
53 |
36623
|
claudio.at
|
final String xquery = String.format("/RESOURCE_PROFILE[.//DEDUPLICATION/ACTION_SET/@id = '%s']", id);
|
54 |
|
|
log.info("loading dedup orchestration: " + xquery);
|
55 |
|
|
|
56 |
37127
|
claudio.at
|
return parseOrchestrationProfile(isLookUpService, isLookUpService.getResourceProfileByQuery(xquery));
|
57 |
|
|
}
|
58 |
|
|
|
59 |
|
|
public List<DedupConfigurationOrchestration> loadByEntityName(final String entityName) throws Exception {
|
60 |
|
|
|
61 |
|
|
final ISLookUpService isLookUpService = serviceLocator.getService(ISLookUpService.class);
|
62 |
|
|
|
63 |
|
|
final String xquery = String.format("/RESOURCE_PROFILE[.//DEDUPLICATION/ENTITY/@name = '%s']", entityName);
|
64 |
|
|
log.info("loading dedup orchestration: " + xquery);
|
65 |
|
|
|
66 |
|
|
final List<DedupConfigurationOrchestration> res = Lists.newArrayList();
|
67 |
|
|
|
68 |
|
|
for (final String profile : isLookUpService.quickSearchProfile(xquery)) {
|
69 |
|
|
res.add(parseOrchestrationProfile(isLookUpService, profile));
|
70 |
|
|
}
|
71 |
|
|
|
72 |
|
|
return res;
|
73 |
|
|
}
|
74 |
|
|
|
75 |
|
|
private DedupConfigurationOrchestration parseOrchestrationProfile(final ISLookUpService isLookUpService, final String dedupOrchestation)
|
76 |
|
|
throws DocumentException,
|
77 |
|
|
ISLookUpException, ISLookUpDocumentNotFoundException {
|
78 |
36418
|
claudio.at
|
final Document doc = new SAXReader().read(new StringReader(dedupOrchestation));
|
79 |
|
|
|
80 |
|
|
final Element e = (Element) doc.selectSingleNode("//DEDUPLICATION/ENTITY");
|
81 |
|
|
final Entity entity = new Entity(e.attributeValue("name"), e.attributeValue("code"), e.attributeValue("label"));
|
82 |
|
|
|
83 |
|
|
final String actionSetId = doc.valueOf("//DEDUPLICATION/ACTION_SET/@id");
|
84 |
|
|
final Queue<DedupConfig> configurations = Lists.newLinkedList();
|
85 |
|
|
|
86 |
|
|
for (final Object o : doc.selectNodes("//SCAN_SEQUENCE/SCAN")) {
|
87 |
37127
|
claudio.at
|
configurations.add(loadConfig(isLookUpService, actionSetId, o));
|
88 |
36418
|
claudio.at
|
}
|
89 |
|
|
|
90 |
|
|
final DedupConfigurationOrchestration dco = new DedupConfigurationOrchestration(entity, actionSetId, configurations);
|
91 |
|
|
|
92 |
36623
|
claudio.at
|
log.debug("loaded dedup configuration orchestration: " + dco.toString());
|
93 |
|
|
log.info("loaded dedup configuration orchestration, size: " + dco.getConfigurations().size());
|
94 |
36418
|
claudio.at
|
return dco;
|
95 |
|
|
}
|
96 |
37127
|
claudio.at
|
|
97 |
|
|
private DedupConfig loadConfig(final ISLookUpService isLookUpService, final String actionSetId, final Object o) throws ISLookUpException,
|
98 |
|
|
ISLookUpDocumentNotFoundException {
|
99 |
|
|
final Element s = (Element) o;
|
100 |
|
|
final String configProfileId = s.attributeValue("id");
|
101 |
|
|
final String conf =
|
102 |
|
|
isLookUpService.getResourceProfileByQuery(String.format(
|
103 |
|
|
"for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()",
|
104 |
|
|
configProfileId));
|
105 |
42015
|
claudio.at
|
log.debug("loaded dedup configuration from IS profile: " + conf);
|
106 |
37127
|
claudio.at
|
final DedupConfig dedupConfig = DedupConfig.load(conf);
|
107 |
|
|
dedupConfig.getWf().setConfigurationId(actionSetId);
|
108 |
|
|
return dedupConfig;
|
109 |
|
|
}
|
110 |
36418
|
claudio.at
|
}
|