Project

General

Profile

1
package eu.dnetlib.data.mapreduce.util;
2

    
3
import com.google.protobuf.GeneratedMessage;
4
import com.google.protobuf.InvalidProtocolBufferException;
5
import eu.dnetlib.data.graph.model.DNGFDecoder;
6
import eu.dnetlib.data.proto.DNGFProtos.DNGF;
7
import eu.dnetlib.data.proto.DNGFProtos.DNGFEntity;
8
import eu.dnetlib.data.proto.DNGFProtos.DNGFRel;
9
import eu.dnetlib.data.proto.DatasourceProtos.Datasource;
10
import eu.dnetlib.data.proto.FieldTypeProtos.*;
11
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty.Builder;
12
import eu.dnetlib.data.proto.KindProtos.Kind;
13
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
14
import eu.dnetlib.data.proto.PersonProtos.Person;
15
import eu.dnetlib.data.proto.ProjectProtos.Project;
16
import eu.dnetlib.data.proto.PublicationProtos.Publication;
17
import eu.dnetlib.data.proto.RelMetadataProtos.RelMetadata;
18
import eu.dnetlib.data.proto.TypeProtos.Type;
19
import org.apache.commons.logging.Log;
20
import org.apache.commons.logging.LogFactory;
21

    
22
public class DNGFTest {
23

    
24
	public static final String CITATION_JSON =
25
			"<citations>\n  <citation>\n    <rawText>[10] M. Foret et al., Phys. Rev. B 66, 024204 (2002).</rawText>\n  </citation>\n  <citation>\n    <rawText>[11] B. Ru\175404\264e et al., Phys. Rev. Lett. 90, 095502 (2003).</rawText>\n  </citation>\n  <citation>\n    <rawText>[12] U. Buchenau et al., Phys. Rev. B 34, 5665 (1986).</rawText>\n  </citation>\n  <citation>\n    <rawText>[13] S.N. Taraskin and S.R. Elliott, J. Phys.: Condens. Mat- ter 11, A219 (1999).</rawText>\n  </citation>\n  <citation>\n    <rawText>[14] B. Hehlen et al., Phys. Rev. Lett. 84, 5355 (2000).</rawText>\n  </citation>\n  <citation>\n    <rawText>[15] N.V. Surotsev et al., J. Phys.: Condens. Matter 10, L113 (1998).</rawText>\n  </citation>\n  <citation>\n    <rawText>[16] D.A. Parshin and C. Laermans, Phys. Rev. B 63, 132203 (2001).</rawText>\n  </citation>\n  <citation>\n    <rawText>[17] V.L. Gurevich et al., Phys. Rev. B 67, 094203 (2003).</rawText>\n  </citation>\n  <citation>\n    <rawText>[18] A. Matic et al., Phys. Rev. Lett. 86, 3803 (2001).</rawText>\n  </citation>\n  <citation>\n    <rawText>[19] E. Rat et al., arXiv:cond-mat/0505558, 23 May 2005.</rawText>\n  </citation>\n  <citation>\n    <rawText>[1] R.C. Zeller and R.O. Pohl, Phys. Rev. B 4, 2029 (1971).</rawText>\n  </citation>\n  <citation>\n    <rawText>[20] C.A. Angell, J. Non-Cryst. Solids 131\20023133, 13 (1991).</rawText>\n  </citation>\n  <citation>\n    <rawText>[21] A.P. Sokolov et al., Phys. Rev. Lett. 71, 2062 (1993).</rawText>\n  </citation>\n  <citation>\n    <rawText>[22] T. Matsuo et al., Solid State Ionics 154-155, 759 (2002).</rawText>\n  </citation>\n  <citation>\n    <rawText>[23] V.K. Malinovsky et al., Europhys. Lett. 11, 43 (1990).</rawText>\n  </citation>\n  <citation>\n    <rawText>[24] J. Lor\250osch et al., J. Non-Cryst. Solids 69, 1 (1984).</rawText>\n  </citation>\n  <citation>\n    <rawText>[25] U. Buchenau, Z. Phys. B 58, 181 (1985).</rawText>\n  </citation>\n  <citation>\n    <rawText>[26] A.F. Io\175400e and A.R. Regel, Prog. Semicond. 4, 237 (1960).</rawText>\n  </citation>\n  <citation>\n    <rawText>[27] R. Dell\20031Anna et al., Phys. Rev. Lett. 80, 1236 (1998).</rawText>\n  </citation>\n  <citation>\n    <rawText>[28] D. Fioretto et al., Phys. Rev. E 59, 4470 (1999).</rawText>\n  </citation>\n  <citation>\n    <rawText>[29] U. Buchenau et al., Phys. Rev. Lett. 77, 4035 (1996).</rawText>\n  </citation>\n  <citation>\n    <rawText>[2] M. Rothenfusser et al., Phys. Rev. B 27, 5196 (1983).</rawText>\n  </citation>\n  <citation>\n    <rawText>[30] J. Mattsson et al., J. Phys.: Condens. Matter 15, S1259 (2003).</rawText>\n  </citation>\n  <citation>\n    <rawText>[31] T. Scopigno et al., Phys. Rev. Lett. 92, 025503 (2004).</rawText>\n  </citation>\n  <citation>\n    <rawText>[32] M. Foret et al., Phys. Rev. Lett. 81, 2100 (1998).</rawText>\n  </citation>\n  <citation>\n    <rawText>[33] F. Sette et al., Science 280, 1550 (1998).</rawText>\n  </citation>\n  <citation>\n    <rawText>[34] J. Wuttke et al., Phys. Rev. E 52, 4026 (1995).</rawText>\n  </citation>\n  <citation>\n    <rawText>[35] M.A. Ramos et al., Phys. Rev. Lett. 78, 82 (1997).</rawText>\n  </citation>\n  <citation>\n    <rawText>[36] G. Monaco et al., Phys. Rev. Lett. 80, 2161 (1998).</rawText>\n  </citation>\n  <citation>\n    <rawText>[37] A. T\250olle, Rep. Prog. Phys. 64, 1473 (2001).</rawText>\n  </citation>\n  <citation>\n    <rawText>[38] As the straight lines do not cross the origin, this does not 2 imply \1623 \21035 \1651 .</rawText>\n  </citation>\n  <citation>\n    <rawText>[39] A. Matic et al., Europhys. Lett. 54, 77 (2001).</rawText>\n  </citation>\n  <citation>\n    <rawText>[3] S. Hunklinger and W. Arnold, in Physical Acoustics, Vol. XII, W.P. Mason and R.N. Thurston Eds. (Academic Press, N.Y. 1976), p. 155.</rawText>\n  </citation>\n  <citation>\n    <rawText>[40] IXS data are usually not available below \1651co, mostly for experimental reasons. E.g., that the rapid onset was not evidenced in vitreous silica [27], is not indicative of its absence but rather of a low qco \21074 1 nm\210221.</rawText>\n  </citation>\n  <citation>\n    <rawText>[41] G. Ruocco et al., Phys. Rev. Lett. 83, 5583 (1999).</rawText>\n  </citation>\n  <citation>\n    <rawText>[42] D. C\1307 iplys et al., J. Physique (Paris) 42, C6-184 (1981).</rawText>\n  </citation>\n  <citation>\n    <rawText>[43] R. Vacher et al., Rev. Sci. Instrum. 51, 288 (1980).</rawText>\n  </citation>\n  <citation>\n    <rawText>[44] R. Vacher et al., arXiv:cond-mat/0505560, 23 May 2005.</rawText>\n  </citation>\n  <citation>\n    <rawText>[45] T.N. Claytor et al., Phys. Rev. B 18, 5842 (1978).</rawText>\n  </citation>\n  <citation>\n    <rawText>[46] M. Arai et al., Physica B 263-264, 268 (1999).</rawText>\n  </citation>\n  <citation>\n    <rawText>[4] R. Vacher et al., J. Non-Cryst. Solids 45, 397 (1981); T.C. Zhu et al., Phys. Rev. B 44, 4281 (1991).</rawText>\n  </citation>\n  <citation>\n    <rawText>[5] J.E. Graebner et al., Phys. Rev. B 34, 5696 (1986).</rawText>\n  </citation>\n  <citation>\n    <rawText>[6] E. Duval and A. Mermet, Phys. Rev. B 58, 8159 (1998).</rawText>\n  </citation>\n  <citation>\n    <rawText>[7] A. Matic et al., Phys. Rev. Lett. 93, 145502 (2004).</rawText>\n  </citation>\n  <citation>\n    <rawText>[8] Often alluded to, e.g. in the Encyclopedia of Materials: Science and Technology, K.H.J. Buschow et al., Eds., Vol. 1 (Elsevier, Oxford, 2001), articles by S.R. Elliott on pp. 171-174 and U. Buchenau on pp. 212-215.</rawText>\n  </citation>\n  <citation>\n    <rawText>[9] E. Rat et al., Phys. Rev. Lett. 83, 1355 (1999).</rawText>\n  </citation>\n</citations>";
26
	public static final String STATISTICS_JSON =
27
			"[{ \"citationsPerYear\": \"many\", \"anotherCoolStatistic\": \"WoW\", \"nestedStat\": { \"firstNestedStat\" : \"value 1\", \"secondNestedStat\" : \"value 2\"}, \"listingStat\" : [ \"one\", \"two\" ] }]";
28
    private static final Log log = LogFactory.getLog(DNGFTest.class);
29
    private static String basePathProfiles = "/eu/dnetlib/test/profiles/OntologyDSResources/OntologyDSResourceType";
30

    
31
	public static StructuredProperty.Builder getStructuredproperty(final String value, final String classname, final String schemename) {
32
		return getStructuredproperty(value, classname, schemename, null);
33
	}
34

    
35
	public static StructuredProperty.Builder getStructuredproperty(final String value, final String classname, final String schemename, final DataInfo dataInfo) {
36
		final Builder sp = StructuredProperty.newBuilder().setValue(value).setQualifier(getQualifier(classname, schemename));
37
		if (dataInfo != null) {
38
			sp.setDataInfo(dataInfo);
39
		}
40
		return sp;
41
	}
42

    
43
	public static Qualifier.Builder getQualifier(final String classname, final String schemename) {
44
		return Qualifier.newBuilder().setClassid(classname).setClassname(classname).setSchemeid(schemename).setSchemename(schemename);
45
	}
46

    
47
	public static KeyValue getKV(final String id, final String name) {
48
		return KeyValue.newBuilder().setKey(id).setValue(name).build();
49
	}
50

    
51
	public static DNGFEntity getDatasource(final String datasourceId) {
52
		return DNGFEntity
53
				.newBuilder()
54
				.setType(Type.datasource)
55
				.setId(datasourceId)
56
				.setDatasource(
57
						Datasource.newBuilder().setMetadata(
58
								Datasource.Metadata.newBuilder().setOfficialname(sf("officialname")).setEnglishname(sf("englishname"))
59
										.setWebsiteurl(sf("websiteurl")).setContactemail(sf("contactemail")).addAccessinfopackage(sf("accessinforpackage"))
60
										.setNamespaceprefix(sf("namespaceprofix")).setDescription(sf("description")).setOdnumberofitems(sf("numberofitems"))
61
										.setOdnumberofitemsdate(sf("numberofitems date"))
62
										// .addOdsubjects("subjects")
63
										.setOdpolicies(sf("policies")).addOdlanguages(sf("languages")).addOdcontenttypes(sf("contenttypes"))
64
										.setDatasourcetype(getQualifier("type class", "type scheme")))).build();
65
	}
66

    
67
	public static DNGFEntity getResult(final String id) {
68
		return getResultBuilder(id).build();
69
	}
70

    
71
	public static DNGFEntity.Builder getResultBuilder(final String id) {
72
		return DNGFEntity
73
				.newBuilder()
74
				.setType(Type.publication)
75
				.setId(id)
76
				.setPublication(
77
						Publication.newBuilder()
78
								.setMetadata(
79
										Publication.Metadata
80
												.newBuilder()
81
												.addTitle(
82
														getStructuredproperty(
83
																"Analysis of cell viability in intervertebral disc: Effect of endplate permeability on cell population",
84
																"main title", "dnet:result_titles", getDataInfo()))
85
												.addTitle(getStructuredproperty("Another title", "alternative title", "dnet:result_titles", getDataInfo()))
86
												.addSubject(getStructuredproperty("Biophysics", "subject", "dnet:result_sujects"))
87
												.setDateofacceptance(sf("2010-01-01")).addSource(sf("sourceA")).addSource(sf("sourceB"))
88
												.addContext(Context.newBuilder().setId("egi::virtual::970"))
89
												.addContext(Context.newBuilder().setId("egi::classification::natsc::math::applied"))
90
												.addContext(Context.newBuilder().setId("egi::classification::natsc::math"))
91
												.addContext(Context.newBuilder().setId("egi::classification::natsc"))
92
												.addContext(Context.newBuilder().setId("egi::classification")).addContext(Context.newBuilder().setId("egi"))
93
												.addDescription(sf("Responsible for making and maintaining the extracellular matrix ..."))
94
												.addDescription(sf("Another description ...")).setPublisher(sf("ELSEVIER SCI LTD"))
95
												.setResulttype(getQualifier("publication", "dnet:result_types"))
96
												.setLanguage(getQualifier("eng", "dnet:languages"))).addInstance(getInstance("10|od__10", "Uk pubmed"))
97
								.addInstance(getInstance("10|od__10", "arxiv")))
98
				.addCollectedfrom(getKV("opendoar____::1064", "Oxford University Research Archive"))
99
				.addPid(getStructuredproperty("doi:74293", "doi", "dnet:pids")).addPid(getStructuredproperty("oai:74295", "oai", "dnet:pids"))
100
				.setDateofcollection("");
101
	}
102

    
103
	public static DataInfo getDataInfo() {
104
		return getDataInfo("0.4");
105
	}
106

    
107
	public static DataInfo getDataInfo(final String trust) {
108
		return DataInfo.newBuilder().setDeletedbyinference(false).setTrust("0.4").setInferenceprovenance("algo").setProvenanceaction(getQualifier("xx", "yy"))
109
				.build();
110
	}
111

    
112
	public static Instance.Builder getInstance(final String hostedbyId, final String hostedbyName) {
113
		return Instance.newBuilder().setHostedby(getKV(hostedbyId, hostedbyName)).setLicence(getQualifier("OpenAccess", "dnet:access_modes"))
114
				.setInstancetype(getQualifier("publication", "dnet:result_typologies")).addUrl("webresource url");
115

    
116
	}
117

    
118
	public static DNGFRel getDedupRel(final String source, final String target) {
119
		return DNGFRel.newBuilder().setSource(source).setTarget(target).setRelType(getQualifier("dedup", "dedupScheme"))
120
				.setChild(false).setCachedTarget(getResult(target)).build();
121
	}
122

    
123

    
124
	public static DNGFEntity getPerson() {
125
		return DNGFEntity
126
				.newBuilder()
127
				.setType(Type.person)
128
				.setId("WOS:000277866500014_A._Shirazi-Adl")
129
				.setPerson(
130
						Person.newBuilder().setMetadata(
131
								Person.Metadata.newBuilder().addSecondnames(sf("Shirazi-Adl")).setFullname(sf("A. Shirazi-Adl"))
132
										.setEmail(sf("name.surname@gmail.com")).setPhone(sf("12345")).setNationality(getQualifier("EN", "dnet:countries"))))
133
				.build();
134
	}
135

    
136

    
137
	public static RelMetadata.Builder relMetadata(final String classname, final String schemename) {
138
		return RelMetadata.newBuilder().setSemantics(getQualifier(classname, schemename));
139
	}
140

    
141
	public static DNGFEntity getOrganization(final String orgId) {
142
		return DNGFEntity
143
				.newBuilder()
144
				.setType(Type.organization)
145
				.setId(orgId)
146
				.addCollectedfrom(getKV("opendoar_1234", "UK pubmed"))
147
				.setOrganization(
148
						Organization.newBuilder().setMetadata(
149
								Organization.Metadata.newBuilder().setLegalname(sf("CENTRE D'APPUI A LA RECHERCHE ET A LA FORMATION GIE"))
150
										.setLegalshortname(sf("CAREF")).setWebsiteurl(sf("www.caref-mali.org"))
151
										.setCountry(getQualifier("ML", "dnet:countries")))).build();
152
	}
153

    
154
	public static DNGFEntity getProjectFP7(final String projectId, final String fundingProgram) throws InvalidProtocolBufferException {
155
		return DNGFEntity
156
				.newBuilder()
157
				.setType(Type.project)
158
				.setId(projectId)
159
				.addCollectedfrom(getKV("opendoar_1234", "UK pubmed"))
160
				.setProject(
161
						Project.newBuilder()
162
								.setMetadata(
163
										Project.Metadata
164
												.newBuilder()
165
												.setAcronym(sf("5CYRQOL"))
166
												.setTitle(sf("Cypriot Researchers Contribute to our Quality of Life"))
167
												.setStartdate(sf("2007-05-01"))
168
												.setEnddate(sf("2007-10-31"))
169
												.setEcsc39(sf("false"))
170
												.setContracttype(getQualifier("CSA", "ec:FP7contractTypes"))
171
												.addFundingtree(
172
														sf("<fundingtree><funder><id>ec__________::EC</id><shortname>EC</shortname><name>European Commission</name></funder><funding_level_2><id>ec__________::EC::FP7::"
173
																+ fundingProgram
174
																+ "::PEOPLE</id><description>Marie-Curie Actions</description><name>PEOPLE</name><class>ec:program</class><parent><funding_level_1><id>ec__________::EC::FP7::"
175
																+ fundingProgram
176
																+ "</id><description>"
177
																+ fundingProgram
178
																+ "-People</description><name>"
179
																+ fundingProgram
180
																+ "</name><class>ec:specificprogram</class><parent><funding_level_0><id>ec__________::EC::FP7</id><description>SEVENTH FRAMEWORK PROGRAMME</description><name>FP7</name><parent/><class>ec:frameworkprogram</class></funding_level_0></parent></funding_level_1></parent></funding_level_2></fundingtree>"))))
181
				.build();
182
	}
183

    
184
	public static DNGFEntity getProjectWT() throws InvalidProtocolBufferException {
185
		return DNGFEntity
186
				.newBuilder()
187
				.setType(Type.project)
188
				.setId("project|wt::087536")
189
				.addCollectedfrom(getKV("wellcomeTrust", "wellcome trust"))
190
				.setProject(
191
						Project.newBuilder()
192
								.setMetadata(
193
										Project.Metadata
194
												.newBuilder()
195
												.setAcronym(sf("UNKNOWN"))
196
												.setTitle(sf("Research Institute for Infectious Diseases of Poverty (IIDP)."))
197
												.setStartdate(sf("2007-05-01"))
198
												.setEnddate(sf("2007-10-31"))
199
												.setEcsc39(sf("false"))
200
												.setContracttype(getQualifier("UNKNOWN", "wt:contractTypes"))
201
												.addFundingtree(
202
														sf("<fundingtree><funder><id>wt__________::WT</id><shortname>WT</shortname><name>Wellcome Trust</name></funder><funding_level_0><id>wt__________::WT::UNKNOWN</id><description>UNKNOWN</description><name>UNKNOWN</name><class>wt:fundingStream</class><parent/></funding_level_0></fundingtree>"))
203
												.addFundingtree(
204
														sf("<fundingtree><funder><id>wt__________::WT</id><shortname>WT</shortname><name>Wellcome Trust</name></funder><funding_level_0><id>wt__________::WT::Technology Transfer</id><description>Technology Transfer</description><name>Technology Transfer</name><class>wt:fundingStream</class><parent/></funding_level_0></fundingtree>"))))
205
				.build();
206
	}
207

    
208
	public static ExtraInfo extraInfo(final String name, final String provenance, final String trust, final String typology, final String value) {
209
		final ExtraInfo.Builder e = ExtraInfo.newBuilder().setName(name).setProvenance(provenance).setTrust(trust).setTypology(typology).setValue(value);
210
		return e.build();
211
	}
212

    
213
	public static StringField sf(final String s) {
214
		return sf(s, null);
215
	}
216

    
217
	public static StringField sf(final String s, final DataInfo dataInfo) {
218
		final StringField.Builder sf = StringField.newBuilder().setValue(s);
219
		if (dataInfo != null) {
220
			sf.setDataInfo(dataInfo);
221
		}
222
		return sf.build();
223
	}
224

    
225
	public static DNGFDecoder embed(final GeneratedMessage msg,
226
			final Kind kind,
227
			final boolean deletedByInference,
228
			final boolean inferred,
229
			final String provenance,
230
			final String action) {
231

    
232
		final DNGF.Builder dngf = DNGF.newBuilder()
233
				.setKind(kind)
234
				.setLastupdatetimestamp(System.currentTimeMillis())
235
				.setDataInfo(
236
						DataInfo.newBuilder().setDeletedbyinference(deletedByInference).setInferred(inferred).setTrust("0.5")
237
								.setInferenceprovenance(provenance).setProvenanceaction(getQualifier(action, action)));
238
		switch (kind) {
239
		case entity:
240
			dngf.setEntity((DNGFEntity) msg);
241
			break;
242
		case relation:
243
			dngf.setRel((DNGFRel) msg);
244
			break;
245
		default:
246
			break;
247
		}
248

    
249
		return DNGFDecoder.decode(dngf.build());
250
	}
251

    
252
	public static DNGFDecoder embed(final GeneratedMessage msg, final Kind kind) {
253
		return embed(msg, kind, false, false, "inference_provenance", "provenance_action");
254
	}
255

    
256

    
257
}
(4-4/4)