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.Autowired;
23
import org.springframework.beans.factory.annotation.Value;
24

    
25
import com.google.common.base.Splitter;
26
import com.mongodb.MongoClient;
27
import com.mongodb.client.MongoCollection;
28

    
29
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord;
30
import eu.dnetlib.data.mdstore.plugins.objects.MyURL;
31
import eu.dnetlib.data.mdstore.plugins.objects.Project;
32

    
33
public class EnrichOpenairePlugin extends GenericDoiMdstorePlugin {
34

    
35
	private static final Log log = LogFactory.getLog(EnrichOpenairePlugin.class);
36

    
37
	@Value("${plugin.enrich.publications.openaire.url}")
38
	private String baseUrl;
39

    
40
	@Value("${plugin.enrich.openaire.datasources.blacklist}")
41
	private String datasourceBlackList;
42

    
43
	@Autowired
44
	private MongoClient mongoClient;
45

    
46
	private Map<String, Counter> counters = new HashMap<>();
47

    
48
	@Override
49
	protected URI prepareURI(final String doi) throws URISyntaxException {
50
		return new URI(String.format(baseUrl, doi));
51
	}
52

    
53
	@Override
54
	protected MongoCollection<org.bson.Document> getCacheCollection() {
55
		return mongoClient.getDatabase("API_CACHES").getCollection("OPENAIRE_API_CACHE");
56
	}
57

    
58
	@Override
59
	protected void reconfigure(final Map<String, String> params) {
60
		counters.clear();
61
		counters.put("subjects", new Counter());
62
		counters.put("citations", new Counter());
63
		counters.put("urls", new Counter());
64
		counters.put("projects", new Counter());
65
		counters.put("dois", new Counter());
66
	}
67

    
68
	@Override
69
	protected void resetConfiguration() {
70
		log.info("***** Openaire Enrichment - subjects  : " + counters.get("subjects"));
71
		log.info("***** Openaire Enrichment - citations : " + counters.get("citations"));
72
		log.info("***** Openaire Enrichment - urls      : " + counters.get("urls"));
73
		log.info("***** Openaire Enrichment - projects  : " + counters.get("projects"));
74
		log.info("***** Openaire Enrichment - dois      : " + counters.get("dois"));
75
		counters.clear();
76
	}
77

    
78
	@Override
79
	protected boolean updateDocument(final MdRecord doc, final String response) {
80
		counters.get("subjects").incrementBefore(doc.getSubjects().size());
81
		counters.get("citations").incrementBefore(doc.getCitations().size());
82
		counters.get("urls").incrementBefore(doc.getUrls().size());
83
		counters.get("projects").incrementBefore(doc.getProjects().size());
84
		counters.get("dois").incrementBefore(doc.getDois().size());
85

    
86
		try {
87
			final Document docRes = (new SAXReader()).read(new StringReader(response));
88

    
89
			final List<?> results = docRes.selectNodes("/response/results/result");
90

    
91
			if (results.size() == 1) {
92
				final Node n = (Node) results.get(0);
93
				updateSubjects(doc, n);
94
				updateCitations(doc, n);
95
				updateUrls(doc, n);
96
				updateProjects(doc, n);
97
				updateDois(doc, n);
98
				updateBestRights(doc);
99

    
100
				return true;
101
			} else if (results.size() == 1) {
102
				log.warn("Too many responses");
103
			}
104
		} catch (final DocumentException e) {
105
			log.warn("Invalid response", e);
106
		} finally {
107
			counters.get("subjects").incrementAfter(doc.getSubjects().size());
108
			counters.get("citations").incrementAfter(doc.getCitations().size());
109
			counters.get("urls").incrementAfter(doc.getUrls().size());
110
			counters.get("projects").incrementAfter(doc.getProjects().size());
111
			counters.get("dois").incrementAfter(doc.getDois().size());
112
		}
113

    
114
		return false;
115
	}
116

    
117
	private void updateSubjects(final MdRecord doc, final Node n) {
118
		final Set<String> subjects = doc.getSubjects()
119
				.stream()
120
				.map(EnrichOpenairePlugin::cleanSubject)
121
				.flatMap(List::stream)
122
				.collect(Collectors.toSet());
123

    
124
		for (final Object o : n.selectNodes(".//subject[@classid='keyword']")) {
125
			subjects.addAll(cleanSubject(((Node) o).getText().trim()));
126
		}
127

    
128
		doc.setSubjects(subjects);
129
	}
130

    
131
	public static List<String> cleanSubject(final String s) {
132
		if (s.isEmpty()) {
133
			return new ArrayList<>();
134
		} else if (s.startsWith("info:eu-repo/classification/msc/")) {
135
			return new ArrayList<>();
136
		} else if (s.startsWith("info:eu-repo/classification/acm/")) {
137
			return Arrays.asList(s.replaceFirst("info:eu-repo/classification/acm/", ""));
138
		} else if (s.contains(";")) {
139
			return Splitter.on(";").trimResults().omitEmptyStrings().splitToList(s);
140
		} else if (s.contains(",")) {
141
			return Splitter.on(",").trimResults().omitEmptyStrings().splitToList(s);
142
		} else {
143
			return Arrays.asList(s);
144
		}
145
	}
146

    
147
	private void updateCitations(final MdRecord doc, final Node n) {
148
		doc.getCitations().clear();
149

    
150
		for (final Object o : n.selectNodes(".//citations/citation/rawText")) {
151
			doc.getCitations().add(((Node) o).getText());
152
		}
153
	}
154

    
155
	private void updateUrls(final MdRecord doc, final Node n) {
156
		doc.getUrls().addAll(doc.getUrls());
157

    
158
		final Set<String> blacklist =
159
				Arrays.stream(datasourceBlackList.split(",")).map(String::trim).filter(StringUtils::isNotBlank).collect(Collectors.toSet());
160

    
161
		for (final Object oin : n.selectNodes(".//instance")) {
162

    
163
			final String hostedByid = ((Element) oin).valueOf("./hostedby/@id").trim();
164

    
165
			if (!blacklist.contains(hostedByid)) {
166
				final String hostedByName = ((Element) oin).valueOf("./hostedby/@name").trim();
167
				final String rights = ((Element) oin).valueOf("./accessright/@classname").trim();
168

    
169
				for (final Object ourl : ((Element) oin).selectNodes("./webresource/url")) {
170
					final MyURL u = new MyURL(((Node) ourl).getText().trim(), hostedByName, rights);
171
					doc.getUrls().remove(u);
172
					doc.getUrls().add(u);
173
				}
174
			}
175
		}
176
	}
177

    
178
	private void updateProjects(final MdRecord doc, final Node n) {
179

    
180
		for (final Object op : n.selectNodes(".//rels/rel[./to/@type='project']")) {
181
			final Node p = (Node) op;
182
			final String name = p.valueOf("./title").trim();
183

    
184
			if (StringUtils.isNotBlank(name) && !name.equalsIgnoreCase("null") && !name.equalsIgnoreCase("unidentified") && !name.equalsIgnoreCase("unknown")) {
185
				final Project np = new Project();
186
				np.setOpenaireId(p.valueOf("./to"));
187
				np.setCode(p.valueOf("./code"));
188
				np.setName(name);
189
				np.setAcronym(p.valueOf("./acronym"));
190
				np.setFunder(p.valueOf(".//funder/@shortname"));
191
				np.setProgram(p.valueOf(".//funding_level_0/@name"));
192
				np.setJurisdiction(p.valueOf(".//funder/@jurisdiction"));
193
				np.setInfoId(String.format("info:eu-repo/grantAgreement/%s/%s/%s/%s/%s/%s", np.getFunder(), np.getProgram(), np.getCode(), np.getJurisdiction(),
194
						np.getName(), np.getAcronym()));
195

    
196
				doc.getProjects().add(np);
197
			}
198
		}
199
	}
200

    
201
	private void updateDois(final MdRecord doc, final Node n) {
202
		for (final Object od : n.selectNodes(".//*[local-name()='result']/pid[@classid='doi']")) {
203
			final String doi = ((Node) od).getText().trim();
204
			if (StringUtils.isNotBlank(doi)) {
205
				doc.getDois().add(doi);
206
			}
207
		}
208
	}
209

    
210
	private void updateBestRights(final MdRecord doc) {
211
		final Set<String> availables = doc.getUrls().stream().map(MyURL::getRights).map(String::toUpperCase).collect(Collectors.toSet());
212
		if (availables.contains("OPEN ACCESS")) {
213
			doc.setBestRights("Open Access");
214
		} else if (availables.contains("EMBARGO")) {
215
			doc.setBestRights("Embargo");
216
		} else if (availables.contains("RESTRICTED")) {
217
			doc.setBestRights("Restricted");
218
		} else if (availables.contains("CLOSED ACCESS")) {
219
			doc.setBestRights("Closed Access");
220
		} else {
221
			doc.setBestRights("Unknown");
222
		}
223
	}
224

    
225
}
(7-7/11)