1
|
package eu.dnetlib.msro.workers.aggregation.cleaner;
|
2
|
|
3
|
import java.io.StringReader;
|
4
|
import java.util.Set;
|
5
|
|
6
|
import org.codehaus.plexus.util.StringUtils;
|
7
|
import org.dom4j.Document;
|
8
|
import org.dom4j.Element;
|
9
|
import org.dom4j.io.SAXReader;
|
10
|
import org.springframework.beans.factory.annotation.Autowired;
|
11
|
import org.springframework.stereotype.Component;
|
12
|
|
13
|
import com.google.common.base.Splitter;
|
14
|
import com.google.common.collect.Sets;
|
15
|
|
16
|
import eu.dnetlib.clients.is.InformationServiceClient;
|
17
|
import eu.dnetlib.msro.exceptions.MSROException;
|
18
|
|
19
|
@Component
|
20
|
public class CleaningRuleFactory {
|
21
|
|
22
|
@Autowired
|
23
|
private InformationServiceClient isClient;
|
24
|
|
25
|
public CleaningRule obtainCleaningRule(final String ruleId) throws MSROException {
|
26
|
try {
|
27
|
final String prof =
|
28
|
isClient.findOne("/RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value='" + ruleId + "' or .//CLEANER_NAME='" + ruleId + "']//CONFIGURATION");
|
29
|
|
30
|
final SAXReader reader = new SAXReader();
|
31
|
final Document doc = reader.read(new StringReader(prof));
|
32
|
|
33
|
final CleaningRule rule = new CleaningRule();
|
34
|
|
35
|
for (final Object o : doc.selectNodes("//RULE")) {
|
36
|
final Element node = (Element) o;
|
37
|
|
38
|
final String xpath = node.valueOf("@xpath");
|
39
|
final String vocabularies = node.valueOf("@vocabularies");
|
40
|
final String groovyRule = node.valueOf("@groovy");
|
41
|
final String regexRule = node.valueOf("@regex");
|
42
|
final String strict = node.valueOf("@strict");
|
43
|
|
44
|
final XPATHCleaningRule xpathRule;
|
45
|
if (StringUtils.isNotBlank(vocabularies)) {
|
46
|
final Set<String> list = Sets.newHashSet(Splitter.on(",").omitEmptyStrings().trimResults().split(vocabularies));
|
47
|
xpathRule = new VocabularyRule(list, isClient);
|
48
|
xpathRule.setStrict("true".equals(strict));
|
49
|
} else if (StringUtils.isNotBlank(groovyRule)) {
|
50
|
xpathRule = new GroovyRule(groovyRule);
|
51
|
xpathRule.setStrict("true".equals(strict));
|
52
|
} else {
|
53
|
xpathRule = new TestRegexValueRule(regexRule);
|
54
|
xpathRule.setStrict(true);
|
55
|
}
|
56
|
|
57
|
xpathRule.setXpath(xpath);
|
58
|
rule.getXpathRules().add(xpathRule);
|
59
|
}
|
60
|
return rule;
|
61
|
} catch (final Exception e) {
|
62
|
throw new MSROException("Error obtaing cleaner rule " + ruleId, e);
|
63
|
}
|
64
|
}
|
65
|
|
66
|
}
|