Project

General

Profile

1
package eu.dnetlib.data.mdstore.plugins;
2

    
3
import java.io.StringReader;
4
import java.net.URI;
5
import java.net.URISyntaxException;
6
import java.util.ArrayList;
7
import java.util.Arrays;
8
import java.util.HashMap;
9
import java.util.List;
10
import java.util.Map;
11
import java.util.Set;
12
import java.util.stream.Collectors;
13

    
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.Node;
21
import org.dom4j.io.SAXReader;
22
import org.springframework.beans.factory.annotation.Value;
23

    
24
import com.google.common.base.Splitter;
25

    
26
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord;
27
import eu.dnetlib.data.mdstore.plugins.objects.MyURL;
28
import eu.dnetlib.data.mdstore.plugins.objects.Project;
29

    
30
public class EnrichOpenairePlugin extends GenericDoiMdstorePlugin {
31

    
32
	private static final Log log = LogFactory.getLog(EnrichOpenairePlugin.class);
33

    
34
	@Value("${plugin.enrich.publications.openaire.url}")
35
	private String baseUrl;
36

    
37
	@Value("${plugin.enrich.openaire.datasources.blacklist}")
38
	private String datasourceBlackList;
39

    
40
	private Map<String, Counter> counters = new HashMap<>();
41

    
42
	@Override
43
	protected URI prepareURI(final String doi) throws URISyntaxException {
44
		return new URI(String.format(baseUrl, doi));
45
	}
46

    
47
	@Override
48
	protected void reconfigure(final Map<String, String> params) {
49
		counters.clear();
50
		counters.put("subjects", new Counter());
51
		counters.put("citations", new Counter());
52
		counters.put("urls", new Counter());
53
		counters.put("projects", new Counter());
54
		counters.put("dois", new Counter());
55
	}
56

    
57
	@Override
58
	protected void resetConfiguration() {
59
		log.info("***** Openaire Enrichment - subjects  : " + counters.get("subjects"));
60
		log.info("***** Openaire Enrichment - citations : " + counters.get("citations"));
61
		log.info("***** Openaire Enrichment - urls      : " + counters.get("urls"));
62
		log.info("***** Openaire Enrichment - projects  : " + counters.get("projects"));
63
		log.info("***** Openaire Enrichment - dois      : " + counters.get("dois"));
64
		counters.clear();
65
	}
66

    
67
	@Override
68
	protected boolean updateDocument(final MdRecord doc, final String response) {
69
		counters.get("subjects").incrementBefore(doc.getSubjects().size());
70
		counters.get("citations").incrementBefore(doc.getCitations().size());
71
		counters.get("urls").incrementBefore(doc.getUrls().size());
72
		counters.get("projects").incrementBefore(doc.getProjects().size());
73
		counters.get("dois").incrementBefore(doc.getDois().size());
74

    
75
		try {
76
			final Document docRes = (new SAXReader()).read(new StringReader(response));
77

    
78
			final List<?> results = docRes.selectNodes("/response/results/result");
79

    
80
			if (results.size() == 1) {
81
				final Node n = (Node) results.get(0);
82
				updateSubjects(doc, n);
83
				updateCitations(doc, n);
84
				updateUrls(doc, n);
85
				updateProjects(doc, n);
86
				updateDois(doc, n);
87
				updateBestRights(doc);
88

    
89
				return true;
90
			} else if (results.size() == 1) {
91
				log.warn("Too many responses");
92
			}
93
		} catch (final DocumentException e) {
94
			log.warn("Invalid response", e);
95
		} finally {
96
			counters.get("subjects").incrementAfter(doc.getSubjects().size());
97
			counters.get("citations").incrementAfter(doc.getCitations().size());
98
			counters.get("urls").incrementAfter(doc.getUrls().size());
99
			counters.get("projects").incrementAfter(doc.getProjects().size());
100
			counters.get("dois").incrementAfter(doc.getDois().size());
101
		}
102

    
103
		return false;
104
	}
105

    
106
	private void updateSubjects(final MdRecord doc, final Node n) {
107
		final Set<String> subjects = doc.getSubjects()
108
				.stream()
109
				.map(EnrichOpenairePlugin::cleanSubject)
110
				.flatMap(List::stream)
111
				.collect(Collectors.toSet());
112

    
113
		for (final Object o : n.selectNodes(".//subject[@classid='keyword']")) {
114
			subjects.addAll(cleanSubject(((Node) o).getText().trim()));
115
		}
116

    
117
		doc.setSubjects(subjects);
118
	}
119

    
120
	public static List<String> cleanSubject(final String s) {
121
		if (s.isEmpty()) {
122
			return new ArrayList<>();
123
		} else if (s.startsWith("info:eu-repo/classification/msc/")) {
124
			return new ArrayList<>();
125
		} else if (s.startsWith("info:eu-repo/classification/acm/")) {
126
			return Arrays.asList(s.replaceFirst("info:eu-repo/classification/acm/", ""));
127
		} else if (s.contains(";")) {
128
			return Splitter.on(";").trimResults().omitEmptyStrings().splitToList(s);
129
		} else if (s.contains(",")) {
130
			return Splitter.on(",").trimResults().omitEmptyStrings().splitToList(s);
131
		} else {
132
			return Arrays.asList(s);
133
		}
134
	}
135

    
136
	private void updateCitations(final MdRecord doc, final Node n) {
137
		doc.getCitations().clear();
138

    
139
		for (final Object o : n.selectNodes(".//citations/citation/rawText")) {
140
			doc.getCitations().add(((Node) o).getText());
141
		}
142
	}
143

    
144
	private void updateUrls(final MdRecord doc, final Node n) {
145
		doc.getUrls().addAll(doc.getUrls());
146

    
147
		final Set<String> blacklist =
148
				Arrays.stream(datasourceBlackList.split(",")).map(String::trim).filter(StringUtils::isNotBlank).collect(Collectors.toSet());
149

    
150
		for (final Object oin : n.selectNodes(".//instance")) {
151

    
152
			final String hostedByid = ((Element) oin).valueOf("./hostedby/@id").trim();
153

    
154
			if (!blacklist.contains(hostedByid)) {
155
				final String hostedByName = ((Element) oin).valueOf("./hostedby/@name").trim();
156
				final String rights = ((Element) oin).valueOf("./accessright/@classname").trim();
157

    
158
				for (final Object ourl : ((Element) oin).selectNodes("./webresource/url")) {
159
					final MyURL u = new MyURL(((Node) ourl).getText().trim(), hostedByName, rights);
160
					doc.getUrls().remove(u);
161
					doc.getUrls().add(u);
162
				}
163
			}
164
		}
165
	}
166

    
167
	private void updateProjects(final MdRecord doc, final Node n) {
168

    
169
		for (final Object op : n.selectNodes(".//rels/rel[./to/@type='project']")) {
170
			final Node p = (Node) op;
171
			final String name = p.valueOf("./title").trim();
172

    
173
			if (StringUtils.isNotBlank(name) && !name.equalsIgnoreCase("null") && !name.equalsIgnoreCase("unidentified") && !name.equalsIgnoreCase("unknown")) {
174
				final Project np = new Project();
175
				np.setOpenaireId(p.valueOf("./to"));
176
				np.setCode(p.valueOf("./code"));
177
				np.setName(name);
178
				np.setAcronym(p.valueOf("./acronym"));
179
				np.setFunder(p.valueOf(".//funder/@shortname"));
180
				np.setProgram(p.valueOf(".//funding_level_0/@name"));
181
				np.setJurisdiction(p.valueOf(".//funder/@jurisdiction"));
182
				np.setInfoId(String.format("info:eu-repo/grantAgreement/%s/%s/%s/%s/%s/%s", np.getFunder(), np.getProgram(), np.getCode(), np.getJurisdiction(),
183
						np.getName(), np.getAcronym()));
184

    
185
				doc.getProjects().add(np);
186
			}
187
		}
188
	}
189

    
190
	private void updateDois(final MdRecord doc, final Node n) {
191
		for (final Object od : n.selectNodes(".//*[local-name()='result']/pid[@classid='doi']")) {
192
			final String doi = ((Node) od).getText().trim();
193
			if (StringUtils.isNotBlank(doi)) {
194
				doc.getDois().add(doi);
195
			}
196
		}
197
	}
198

    
199
	private void updateBestRights(final MdRecord doc) {
200
		final Set<String> availables = doc.getUrls().stream().map(MyURL::getRights).map(String::toUpperCase).collect(Collectors.toSet());
201
		if (availables.contains("OPEN ACCESS")) {
202
			doc.setBestRights("Open Access");
203
		} else if (availables.contains("EMBARGO")) {
204
			doc.setBestRights("Embargo");
205
		} else if (availables.contains("RESTRICTED")) {
206
			doc.setBestRights("Restricted");
207
		} else if (availables.contains("CLOSED ACCESS")) {
208
			doc.setBestRights("Closed Access");
209
		} else {
210
			doc.setBestRights("Unknown");
211
		}
212
	}
213

    
214
}
(7-7/11)