|
1
|
package eu.dnetlib.msro.workflows.nodes.dedup.utils;
|
|
2
|
|
|
3
|
import java.io.StringReader;
|
|
4
|
import java.util.List;
|
|
5
|
import java.util.Queue;
|
|
6
|
import javax.annotation.Resource;
|
|
7
|
|
|
8
|
import com.google.common.collect.Lists;
|
|
9
|
import eu.dnetlib.enabling.locators.UniqueServiceLocator;
|
|
10
|
import eu.dnetlib.pace.config.DedupConfig;
|
|
11
|
import eu.dnetlib.rmi.enabling.ISLookUpDocumentNotFoundException;
|
|
12
|
import eu.dnetlib.rmi.enabling.ISLookUpException;
|
|
13
|
import eu.dnetlib.rmi.enabling.ISLookUpService;
|
|
14
|
import org.apache.commons.lang3.StringUtils;
|
|
15
|
import org.apache.commons.logging.Log;
|
|
16
|
import org.apache.commons.logging.LogFactory;
|
|
17
|
import org.dom4j.Document;
|
|
18
|
import org.dom4j.DocumentException;
|
|
19
|
import org.dom4j.Element;
|
|
20
|
import org.dom4j.io.SAXReader;
|
|
21
|
|
|
22
|
/**
|
|
23
|
* The Class DedupConfigurationOrchestrationLoader.
|
|
24
|
*/
|
|
25
|
public class DedupConfigurationOrchestrationLoader {
|
|
26
|
|
|
27
|
/**
|
|
28
|
* The Constant log.
|
|
29
|
*/
|
|
30
|
private static final Log log = LogFactory.getLog(DedupConfigurationOrchestrationLoader.class);
|
|
31
|
|
|
32
|
/**
|
|
33
|
* The service locator.
|
|
34
|
*/
|
|
35
|
@Resource
|
|
36
|
private UniqueServiceLocator serviceLocator;
|
|
37
|
|
|
38
|
/**
|
|
39
|
* Load the dedup orchestration profile from the IS.
|
|
40
|
*
|
|
41
|
* @param id the id
|
|
42
|
* @return the dedup configuration orchestration
|
|
43
|
* @throws ISLookUpDocumentNotFoundException the IS look up document not found exception
|
|
44
|
* @throws ISLookUpException the IS look up exception
|
|
45
|
* @throws DocumentException the document exception
|
|
46
|
*/
|
|
47
|
public DedupConfigurationOrchestration loadByActionSetId(final String id) throws Exception {
|
|
48
|
|
|
49
|
final ISLookUpService isLookUpService = serviceLocator.getService(ISLookUpService.class);
|
|
50
|
|
|
51
|
final String xquery = String.format("/RESOURCE_PROFILE[.//DEDUPLICATION/ACTION_SET/@id = '%s']", id);
|
|
52
|
log.info("loading dedup orchestration: " + xquery);
|
|
53
|
|
|
54
|
return parseOrchestrationProfile(isLookUpService, isLookUpService.getResourceProfileByQuery(xquery));
|
|
55
|
}
|
|
56
|
|
|
57
|
public List<DedupConfigurationOrchestration> loadByEntityName(final String entityName) throws Exception {
|
|
58
|
|
|
59
|
final ISLookUpService isLookUpService = serviceLocator.getService(ISLookUpService.class);
|
|
60
|
|
|
61
|
final String xquery = String.format("/RESOURCE_PROFILE[.//DEDUPLICATION/ENTITY/@name = '%s']", entityName);
|
|
62
|
log.info("loading dedup orchestration: " + xquery);
|
|
63
|
|
|
64
|
final List<DedupConfigurationOrchestration> res = Lists.newArrayList();
|
|
65
|
|
|
66
|
for (final String profile : isLookUpService.quickSearchProfile(xquery)) {
|
|
67
|
res.add(parseOrchestrationProfile(isLookUpService, profile));
|
|
68
|
}
|
|
69
|
|
|
70
|
return res;
|
|
71
|
}
|
|
72
|
|
|
73
|
private DedupConfigurationOrchestration parseOrchestrationProfile(final ISLookUpService isLookUpService, final String dedupOrchestation)
|
|
74
|
throws DocumentException, ISLookUpException {
|
|
75
|
|
|
76
|
final Document doc = new SAXReader().read(new StringReader(dedupOrchestation));
|
|
77
|
|
|
78
|
final Element e = (Element) doc.selectSingleNode("//DEDUPLICATION/ENTITY");
|
|
79
|
final Entity entity = new Entity(e.attributeValue("name"), e.attributeValue("code"), e.attributeValue("label"));
|
|
80
|
|
|
81
|
final String actionSetId = doc.valueOf("//DEDUPLICATION/ACTION_SET/@id");
|
|
82
|
final Queue<DedupConfig> configurations = Lists.newLinkedList();
|
|
83
|
|
|
84
|
for (final Object o : doc.selectNodes("//SCAN_SEQUENCE/SCAN")) {
|
|
85
|
configurations.add(loadConfig(isLookUpService, actionSetId, o));
|
|
86
|
}
|
|
87
|
|
|
88
|
final DedupConfigurationOrchestration dco = new DedupConfigurationOrchestration(entity, actionSetId, configurations);
|
|
89
|
|
|
90
|
log.debug("loaded dedup configuration orchestration: " + dco.toString());
|
|
91
|
log.info("loaded dedup configuration orchestration, size: " + dco.getConfigurations().size());
|
|
92
|
return dco;
|
|
93
|
}
|
|
94
|
|
|
95
|
private DedupConfig loadConfig(final ISLookUpService isLookUpService, final String actionSetId, final Object o) throws ISLookUpException {
|
|
96
|
final Element s = (Element) o;
|
|
97
|
final String configProfileId = s.attributeValue("id");
|
|
98
|
final String conf =
|
|
99
|
isLookUpService.getResourceProfileByQuery(String.format(
|
|
100
|
"for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()",
|
|
101
|
configProfileId));
|
|
102
|
log.debug("loaded dedup configuration from IS profile: " + conf);
|
|
103
|
if (StringUtils.isBlank(conf)) {
|
|
104
|
throw new IllegalStateException(String.format("got empty dedup configuration profile for '%s'", configProfileId));
|
|
105
|
}
|
|
106
|
final DedupConfig dedupConfig = DedupConfig.load(conf);
|
|
107
|
dedupConfig.getWf().setConfigurationId(actionSetId);
|
|
108
|
return dedupConfig;
|
|
109
|
}
|
|
110
|
}
|