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

    
21
public class DNGFTest {
22

    
23
	public static final String CITATION_JSON =
24
			"<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>";
25
	public static final String STATISTICS_JSON =
26
			"[{ \"citationsPerYear\": \"many\", \"anotherCoolStatistic\": \"WoW\", \"nestedStat\": { \"firstNestedStat\" : \"value 1\", \"secondNestedStat\" : \"value 2\"}, \"listingStat\" : [ \"one\", \"two\" ] }]";
27
    private static final Log log = LogFactory.getLog(DNGFTest.class);
28
    private static String basePathProfiles = "/eu/dnetlib/test/profiles/OntologyDSResources/OntologyDSResourceType";
29

    
30
	public static StructuredProperty.Builder getStructuredproperty(final String value, final String classname, final String schemename) {
31
		return getStructuredproperty(value, classname, schemename, null);
32
	}
33

    
34
	public static StructuredProperty.Builder getStructuredproperty(final String value, final String classname, final String schemename, final DataInfo dataInfo) {
35
		final Builder sp = StructuredProperty.newBuilder().setValue(value).setQualifier(getQualifier(classname, schemename));
36
		if (dataInfo != null) {
37
			sp.setDataInfo(dataInfo);
38
		}
39
		return sp;
40
	}
41

    
42
	public static Qualifier.Builder getQualifier(final String classname, final String schemename) {
43
		return Qualifier.newBuilder().setClassid(classname).setClassname(classname).setSchemeid(schemename).setSchemename(schemename);
44
	}
45

    
46
	public static KeyValue getKV(final String id, final String name) {
47
		return KeyValue.newBuilder().setKey(id).setValue(name).build();
48
	}
49

    
50
	public static DNGFEntity getDatasource(final String datasourceId) {
51
		return DNGFEntity
52
				.newBuilder()
53
				.setType(Type.datasource)
54
				.setId(datasourceId)
55
				.setDatasource(
56
						Datasource.newBuilder().setMetadata(
57
								Datasource.Metadata.newBuilder().setOfficialname(sf("officialname")).setEnglishname(sf("englishname"))
58
										.setWebsiteurl(sf("websiteurl")).setContactemail(sf("contactemail")).addAccessinfopackage(sf("accessinforpackage"))
59
										.setNamespaceprefix(sf("namespaceprofix")).setDescription(sf("description")).setOdnumberofitems(sf("numberofitems"))
60
										.setOdnumberofitemsdate(sf("numberofitems date"))
61
										// .addOdsubjects("subjects")
62
										.setOdpolicies(sf("policies")).addOdlanguages(sf("languages")).addOdcontenttypes(sf("contenttypes"))
63
										.setDatasourcetype(getQualifier("type class", "type scheme")))).build();
64
	}
65

    
66
	public static DNGFEntity getResult(final String id) {
67
		return getResultBuilder(id).build();
68
	}
69

    
70
	public static DNGFEntity.Builder getResultBuilder(final String id) {
71
		return DNGFEntity
72
				.newBuilder()
73
				.setType(Type.publication)
74
				.setId(id)
75
				.setPublication(
76
						Publication.newBuilder()
77
								.setMetadata(
78
										Publication.Metadata
79
												.newBuilder()
80
												.addTitle(
81
														getStructuredproperty(
82
																"Analysis of cell viability in intervertebral disc: Effect of endplate permeability on cell population",
83
																"main title", "dnet:result_titles", getDataInfo()))
84
												.addTitle(getStructuredproperty("Another title", "alternative title", "dnet:result_titles", getDataInfo()))
85
												.addSubject(getStructuredproperty("Biophysics", "subject", "dnet:result_sujects"))
86
												.setDateofacceptance(sf("2010-01-01")).addSource(sf("sourceA")).addSource(sf("sourceB"))
87
												.addContext(Context.newBuilder().setId("egi::virtual::970"))
88
												.addContext(Context.newBuilder().setId("egi::classification::natsc::math::applied"))
89
												.addContext(Context.newBuilder().setId("egi::classification::natsc::math"))
90
												.addContext(Context.newBuilder().setId("egi::classification::natsc"))
91
												.addContext(Context.newBuilder().setId("egi::classification")).addContext(Context.newBuilder().setId("egi"))
92
												.addDescription(sf("Responsible for making and maintaining the extracellular matrix ..."))
93
												.addDescription(sf("Another description ...")).setPublisher(sf("ELSEVIER SCI LTD"))
94
												.setResulttype(getQualifier("publication", "dnet:result_types"))
95
												.setLanguage(getQualifier("eng", "dnet:languages"))).addInstance(getInstance("10|od__10", "Uk pubmed"))
96
								.addInstance(getInstance("10|od__10", "arxiv")))
97
				.addCollectedfrom(getKV("opendoar____::1064", "Oxford University Research Archive"))
98
				.addPid(getStructuredproperty("doi:74293", "doi", "dnet:pids")).addPid(getStructuredproperty("oai:74295", "oai", "dnet:pids"))
99
				.setDateofcollection("");
100
	}
101

    
102
	public static DataInfo getDataInfo() {
103
		return getDataInfo("0.4");
104
	}
105

    
106
	public static DataInfo getDataInfo(final String trust) {
107
		return DataInfo.newBuilder().setDeletedbyinference(false).setTrust("0.4").setInferenceprovenance("algo").setProvenanceaction(getQualifier("xx", "yy"))
108
				.build();
109
	}
110

    
111
	public static Instance.Builder getInstance(final String hostedbyId, final String hostedbyName) {
112
		return Instance.newBuilder().setHostedby(getKV(hostedbyId, hostedbyName)).setLicence(getQualifier("OpenAccess", "dnet:access_modes"))
113
				.setInstancetype(getQualifier("publication", "dnet:result_typologies")).addUrl("webresource url");
114

    
115
	}
116

    
117
	public static DNGFRel getDedupRel(final String source, final String target) {
118
		return DNGFRel.newBuilder().setSource(source).setTarget(target).setRelType(getQualifier("dedup", "dedupScheme"))
119
				.setChild(false).setCachedTarget(getResult(target)).build();
120
	}
121

    
122

    
123
	public static DNGFEntity getPerson() {
124
		return DNGFEntity
125
				.newBuilder()
126
				.setType(Type.person)
127
				.setId("WOS:000277866500014_A._Shirazi-Adl")
128
				.setPerson(
129
						Person.newBuilder().setMetadata(
130
								Person.Metadata.newBuilder().addSecondnames(sf("Shirazi-Adl")).setFullname(sf("A. Shirazi-Adl"))
131
										.setEmail(sf("name.surname@gmail.com")).setPhone(sf("12345")).setNationality(getQualifier("EN", "dnet:countries"))))
132
				.build();
133
	}
134

    
135

    
136
	public static RelMetadata.Builder relMetadata(final String classname, final String schemename) {
137
		return RelMetadata.newBuilder().setSemantics(getQualifier(classname, schemename));
138
	}
139

    
140
	public static DNGFEntity getOrganization(final String orgId) {
141
		return DNGFEntity
142
				.newBuilder()
143
				.setType(Type.organization)
144
				.setId(orgId)
145
				.addCollectedfrom(getKV("opendoar_1234", "UK pubmed"))
146
				.setOrganization(
147
						Organization.newBuilder().setMetadata(
148
								Organization.Metadata.newBuilder().setLegalname(sf("CENTRE D'APPUI A LA RECHERCHE ET A LA FORMATION GIE"))
149
										.setLegalshortname(sf("CAREF")).setWebsiteurl(sf("www.caref-mali.org"))
150
										.setCountry(getQualifier("ML", "dnet:countries")))).build();
151
	}
152

    
153
	public static DNGFEntity getProjectFP7(final String projectId, final String fundingProgram) throws InvalidProtocolBufferException {
154
		return DNGFEntity
155
				.newBuilder()
156
				.setType(Type.project)
157
				.setId(projectId)
158
				.addCollectedfrom(getKV("opendoar_1234", "UK pubmed"))
159
				.setProject(
160
						Project.newBuilder()
161
								.setMetadata(
162
										Project.Metadata
163
												.newBuilder()
164
												.setAcronym(sf("5CYRQOL"))
165
												.setTitle(sf("Cypriot Researchers Contribute to our Quality of Life"))
166
												.setStartdate(sf("2007-05-01"))
167
												.setEnddate(sf("2007-10-31"))
168
												.setEcsc39(sf("false"))
169
												.setContracttype(getQualifier("CSA", "ec:FP7contractTypes"))
170
												.addFundingtree(
171
														sf("<fundingtree><funder><id>ec__________::EC</id><shortname>EC</shortname><name>European Commission</name></funder><funding_level_2><id>ec__________::EC::FP7::"
172
																+ fundingProgram
173
																+ "::PEOPLE</id><description>Marie-Curie Actions</description><name>PEOPLE</name><class>ec:program</class><parent><funding_level_1><id>ec__________::EC::FP7::"
174
																+ fundingProgram
175
																+ "</id><description>"
176
																+ fundingProgram
177
																+ "-People</description><name>"
178
																+ fundingProgram
179
																+ "</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>"))))
180
				.build();
181
	}
182

    
183
	public static DNGFEntity getProjectWT() throws InvalidProtocolBufferException {
184
		return DNGFEntity
185
				.newBuilder()
186
				.setType(Type.project)
187
				.setId("project|wt::087536")
188
				.addCollectedfrom(getKV("wellcomeTrust", "wellcome trust"))
189
				.setProject(
190
						Project.newBuilder()
191
								.setMetadata(
192
										Project.Metadata
193
												.newBuilder()
194
												.setAcronym(sf("UNKNOWN"))
195
												.setTitle(sf("Research Institute for Infectious Diseases of Poverty (IIDP)."))
196
												.setStartdate(sf("2007-05-01"))
197
												.setEnddate(sf("2007-10-31"))
198
												.setEcsc39(sf("false"))
199
												.setContracttype(getQualifier("UNKNOWN", "wt:contractTypes"))
200
												.addFundingtree(
201
														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>"))
202
												.addFundingtree(
203
														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>"))))
204
				.build();
205
	}
206

    
207
	public static ExtraInfo extraInfo(final String name, final String provenance, final String trust, final String typology, final String value) {
208
		final ExtraInfo.Builder e = ExtraInfo.newBuilder().setName(name).setProvenance(provenance).setTrust(trust).setTypology(typology).setValue(value);
209
		return e.build();
210
	}
211

    
212
	public static StringField sf(final String s) {
213
		return sf(s, null);
214
	}
215

    
216
	public static StringField sf(final String s, final DataInfo dataInfo) {
217
		final StringField.Builder sf = StringField.newBuilder().setValue(s);
218
		if (dataInfo != null) {
219
			sf.setDataInfo(dataInfo);
220
		}
221
		return sf.build();
222
	}
223

    
224
	public static DNGFDecoder embed(final GeneratedMessage msg,
225
			final Kind kind,
226
			final boolean deletedByInference,
227
			final boolean inferred,
228
			final String provenance,
229
			final String action) {
230

    
231
		final DNGF.Builder dngf = DNGF.newBuilder()
232
				.setKind(kind)
233
				.setLastupdatetimestamp(System.currentTimeMillis())
234
				.setDataInfo(
235
						DataInfo.newBuilder().setDeletedbyinference(deletedByInference).setInferred(inferred).setTrust("0.5")
236
								.setInferenceprovenance(provenance).setProvenanceaction(getQualifier(action, action)));
237
		switch (kind) {
238
		case entity:
239
			dngf.setEntity((DNGFEntity) msg);
240
			break;
241
		case relation:
242
			dngf.setRel((DNGFRel) msg);
243
			break;
244
		default:
245
			break;
246
		}
247

    
248
		return DNGFDecoder.decode(dngf.build());
249
	}
250

    
251
	public static DNGFDecoder embed(final GeneratedMessage msg, final Kind kind) {
252
		return embed(msg, kind, false, false, "inference_provenance", "provenance_action");
253
	}
254

    
255

    
256
}
(4-4/4)