Revision 39938
Added by Eri Katsari over 8 years ago
Serializer.java | ||
---|---|---|
7 | 7 |
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier; |
8 | 8 |
import eu.dnetlib.data.proto.FieldTypeProtos.StringField; |
9 | 9 |
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty; |
10 |
import eu.dnetlib.data.proto.OafProtos; |
|
10 | 11 |
import eu.dnetlib.data.proto.OafProtos.Oaf; |
11 | 12 |
import eu.dnetlib.data.proto.OafProtos.OafEntity; |
12 | 13 |
import eu.dnetlib.data.proto.OafProtos.OafRel; |
... | ... | |
14 | 15 |
import eu.dnetlib.data.proto.PersonProtos; |
15 | 16 |
import eu.dnetlib.data.proto.ProjectProtos.Project; |
16 | 17 |
import eu.dnetlib.data.proto.RelTypeProtos.RelType; |
18 |
import eu.dnetlib.data.proto.ResultProtos; |
|
17 | 19 |
import eu.dnetlib.data.proto.ResultProtos.Result; |
18 | 20 |
import eu.dnetlib.data.proto.ResultProtos.Result.Instance; |
19 | 21 |
import org.apache.log4j.Logger; |
... | ... | |
28 | 30 |
*/ |
29 | 31 |
public class Serializer { |
30 | 32 |
|
31 |
private static String DELIM;
|
|
32 |
private Logger log = Logger.getLogger(this.getClass());
|
|
33 |
private String NULL_STRING = null;
|
|
34 |
private String NULL_NUM = "0";
|
|
33 |
private static String DELIM;
|
|
34 |
private Logger log = Logger.getLogger(this.getClass());
|
|
35 |
private String NULL_STRING = null;
|
|
36 |
private String NULL_NUM = "0";
|
|
35 | 37 |
|
36 |
private static String ENCLOSED;
|
|
38 |
private static String ENCLOSED;
|
|
37 | 39 |
private FundingParser fundingParser = new FundingParser(); |
38 | 40 |
|
39 | 41 |
|
40 |
public Serializer() {
|
|
41 |
}
|
|
42 |
public Serializer() {
|
|
43 |
}
|
|
42 | 44 |
|
45 |
public String serialize(Oaf oaf) { |
|
43 | 46 |
|
44 |
public String serialize(Oaf oaf) { |
|
47 |
switch (oaf.getKind()) { |
|
48 |
case entity: |
|
49 |
OafEntity valueEntity = oaf.getEntity(); |
|
45 | 50 |
|
46 |
switch (oaf.getKind()) { |
|
47 |
case entity: |
|
48 |
OafEntity valueEntity = oaf.getEntity(); |
|
51 |
switch (valueEntity.getType()) { |
|
52 |
case datasource: |
|
49 | 53 |
|
50 |
switch (valueEntity.getType()) { |
|
51 |
case datasource: |
|
54 |
return buildDatasource(valueEntity); |
|
52 | 55 |
|
53 |
return buildDatasource(valueEntity);
|
|
56 |
case organization:
|
|
54 | 57 |
|
55 |
case organization:
|
|
58 |
return buildOrganization(valueEntity);
|
|
56 | 59 |
|
57 |
return buildOrganization(valueEntity);
|
|
60 |
case project:
|
|
58 | 61 |
|
59 |
case project: |
|
62 |
return buildProject(valueEntity); |
|
63 |
case result: |
|
60 | 64 |
|
61 |
return buildProject(valueEntity); |
|
62 |
case result: |
|
65 |
return buildResult(valueEntity); |
|
66 |
case person: |
|
67 |
return buildPerson(valueEntity); |
|
68 |
default: |
|
69 |
log.error("wrong type"); |
|
70 |
break; |
|
71 |
} |
|
72 |
break; |
|
73 |
case relation: |
|
74 |
OafRel valueRel = oaf.getRel(); |
|
63 | 75 |
|
64 |
return buildResult(valueEntity); |
|
65 |
case person: |
|
66 |
return buildPerson(valueEntity); |
|
67 |
default: |
|
68 |
log.error("wrong type"); |
|
69 |
break; |
|
70 |
} |
|
71 |
break; |
|
72 |
case relation: |
|
73 |
OafRel valueRel = oaf.getRel(); |
|
76 |
return buildRel(valueRel); |
|
74 | 77 |
|
75 |
return buildRel(valueRel);
|
|
78 |
}
|
|
76 | 79 |
|
77 |
}
|
|
80 |
return null;
|
|
78 | 81 |
|
79 |
return null;
|
|
82 |
}
|
|
80 | 83 |
|
81 |
}
|
|
84 |
public String serialize(OafRel oaf) {
|
|
82 | 85 |
|
83 |
public String serialize(OafRel oaf) { |
|
86 |
switch (oaf.getRelType()) { |
|
87 |
case resultProject: |
|
84 | 88 |
|
85 |
switch (oaf.getRelType()) { |
|
86 |
case resultProject: |
|
89 |
return getResultProject(oaf); |
|
87 | 90 |
|
88 |
return getResultProject(oaf); |
|
91 |
default: |
|
92 |
return buildRel(oaf); |
|
93 |
} |
|
94 |
} |
|
89 | 95 |
|
90 |
default: |
|
91 |
return buildRel(oaf); |
|
92 |
} |
|
93 |
} |
|
96 |
private String buildRel(OafRel Rel) { |
|
97 |
return getStringField(Rel.getTarget()); |
|
98 |
} |
|
94 | 99 |
|
95 |
private String buildRel(OafRel Rel) { |
|
96 |
return getStringField(Rel.getTarget()); |
|
97 |
} |
|
98 | 100 |
|
101 |
private String getResultProject(OafRel oaf) { |
|
102 |
String buff = new String(); |
|
103 |
String result = oaf.getTarget(); |
|
99 | 104 |
|
100 |
private String getResultProject(OafRel oaf) { |
|
101 |
String buff = new String(); |
|
102 |
String result = oaf.getTarget(); |
|
103 | 105 |
|
106 |
buff += getStringField(result); |
|
107 |
// is declared as int!!! |
|
108 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
109 |
if (diff < 0) { |
|
110 |
diff = 0; |
|
111 |
} |
|
112 |
buff += getNumericField(String.valueOf(diff)); |
|
104 | 113 |
|
105 |
buff += getStringField(result); |
|
106 |
// is declared as int!!! |
|
107 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
108 |
if (diff < 0) { |
|
109 |
diff = 0; |
|
110 |
} |
|
111 |
buff += getNumericField(String.valueOf(diff)); |
|
114 |
return buff; |
|
115 |
} |
|
112 | 116 |
|
113 | 117 |
|
114 |
return buff; |
|
115 |
} |
|
118 |
private String buildDatasource(OafEntity data) { |
|
116 | 119 |
|
120 |
String buff = new String(); |
|
117 | 121 |
|
118 |
public static boolean isNumeric(String str) {
|
|
122 |
Datasource d = data.getDatasource();
|
|
119 | 123 |
|
120 |
str = str.replaceAll("[^A-Za-z0-9 ]", ""); |
|
121 |
str = str.replaceAll(" ", ""); |
|
122 |
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional |
|
123 |
// '-' and decimal. |
|
124 |
} |
|
124 |
Metadata metadata = d.getMetadata(); |
|
125 | 125 |
|
126 |
// there are topics with "null" as value -> replace them
|
|
127 |
private boolean isValidTopic(String t) {
|
|
126 |
//`original Id`
|
|
127 |
String originalId = new String();
|
|
128 | 128 |
|
129 |
if (t == null || t.isEmpty()) {
|
|
130 |
return false;
|
|
131 |
}
|
|
129 |
for (String oid : data.getOriginalIdList()) {
|
|
130 |
originalId += oid + ";";
|
|
131 |
}
|
|
132 | 132 |
|
133 |
if (t.equals("") || t.equals(" ")) { |
|
134 |
return false; |
|
135 |
} |
|
136 |
if (t.equals("null") || t.equals("Null") || t.equals("NULL")) { |
|
137 |
return false; |
|
138 |
} |
|
133 |
buff += getStringField(originalId); |
|
139 | 134 |
|
140 |
if (t.equals(ENCLOSED + ENCLOSED + DELIM) || t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) { |
|
141 |
return false; |
|
142 |
} |
|
143 |
// skip dedups |
|
144 |
if (t.contains("ddc:")) { |
|
135 |
//dateOfCollection |
|
136 |
buff += getStringDateField(data.getDateofcollection()); |
|
145 | 137 |
|
146 |
return false; |
|
147 |
} |
|
148 |
return true; |
|
149 |
} |
|
138 |
//Provider Type |
|
139 |
if (metadata.hasDatasourcetype()) { |
|
140 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", "")); |
|
141 |
} else { |
|
142 |
buff += getStringField(null); |
|
143 |
} |
|
150 | 144 |
|
145 |
//Compatibility |
|
146 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname()); |
|
151 | 147 |
|
152 |
private String buildDatasource(OafEntity data) { |
|
148 |
//Official Name |
|
149 |
buff += getStringField(metadata.getOfficialname().getValue()); |
|
153 | 150 |
|
154 |
String buff = new String(); |
|
151 |
// English Name |
|
152 |
buff += getStringField(metadata.getEnglishname().getValue()); |
|
155 | 153 |
|
156 |
Datasource d = data.getDatasource(); |
|
154 |
//Website URL |
|
155 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
157 | 156 |
|
158 |
Metadata metadata = d.getMetadata(); |
|
157 |
//LogoURL |
|
158 |
buff += getStringField(metadata.getLogourl().getValue()); |
|
159 | 159 |
|
160 |
//`original Id`
|
|
161 |
String originalId = new String();
|
|
160 |
//Email
|
|
161 |
buff += getStringField(metadata.getContactemail().getValue());
|
|
162 | 162 |
|
163 |
for (String oid : data.getOriginalIdList()) { |
|
164 |
originalId += oid; |
|
165 |
break; |
|
166 |
} |
|
163 |
//Namespaceprefix |
|
164 |
buff += getStringField(metadata.getNamespaceprefix().getValue()); |
|
167 | 165 |
|
168 |
buff += getStringField(originalId); |
|
169 |
//dateOfCollection |
|
170 |
buff += getStringDateField(data.getDateofcollection()); |
|
166 |
// latitude |
|
167 |
buff += getStringField(metadata.getLatitude().getValue()); |
|
171 | 168 |
|
172 |
// name |
|
173 |
if (metadata.getOfficialname().getValue().equalsIgnoreCase("unknown")) { |
|
174 |
buff += getStringField("Unknown Repository"); |
|
175 |
} else { |
|
176 |
buff += getStringField(metadata.getOfficialname().getValue()); |
|
177 |
} |
|
178 |
// type |
|
169 |
// longtitude |
|
170 |
buff += getStringField(metadata.getLongitude().getValue()); |
|
179 | 171 |
|
180 |
if (metadata.hasDatasourcetype()) |
|
172 |
// dateofvalidation, |
|
173 |
buff += getStringField(metadata.getDateofvalidation().getValue()); |
|
181 | 174 |
|
182 |
{
|
|
183 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""));
|
|
175 |
//Description
|
|
176 |
buff += getStringField(metadata.getDescription().getValue());
|
|
184 | 177 |
|
185 |
} else { |
|
186 |
buff += getStringField(null); |
|
187 |
} |
|
188 |
//languages |
|
189 |
String languages = new String(); |
|
190 |
for (StringField lang : metadata.getOdlanguagesList()) { |
|
191 |
languages += lang.getValue() + ";"; |
|
192 |
} |
|
193 |
buff += getStringField(languages); |
|
194 |
// compatibility, |
|
195 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname()); |
|
178 |
//subjects |
|
179 |
String subj = new String(); |
|
180 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
181 |
subj += s.getValue() + ';'; |
|
196 | 182 |
|
197 |
// latitude |
|
198 |
buff += getStringField(metadata.getLatitude().getValue()); |
|
183 |
} |
|
199 | 184 |
|
200 |
// longtitude
|
|
201 |
buff += getStringField(metadata.getLongitude().getValue());
|
|
185 |
//subjects
|
|
186 |
buff += getStringField(subj);
|
|
202 | 187 |
|
203 |
// dateofvalidation,
|
|
204 |
buff += getStringField(metadata.getDateofvalidation().getValue());
|
|
188 |
//Number of items
|
|
189 |
buff += getStringField(metadata.getOdnumberofitems().getValue());
|
|
205 | 190 |
|
191 |
//Date of number of items |
|
192 |
buff += getStringField(metadata.getOdnumberofitemsdate().getValue()); |
|
206 | 193 |
|
207 |
// Website URL |
|
208 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
209 |
// certificates |
|
210 |
buff += getStringField(metadata.getCertificates().getValue()); |
|
211 |
// /citation guideline |
|
212 |
buff += getStringField(metadata.getCitationguidelineurl().getValue()); |
|
194 |
// Policies |
|
195 |
buff += getStringField(metadata.getOdpolicies().getValue()); |
|
213 | 196 |
|
214 |
// contact mail |
|
215 |
buff += getStringField(metadata.getContactemail().getValue()); |
|
197 |
//languages |
|
198 |
String languages = new String(); |
|
199 |
for (StringField lang : metadata.getOdlanguagesList()) { |
|
200 |
languages += lang.getValue() + ";"; |
|
201 |
} |
|
202 |
buff += getStringField(languages); |
|
216 | 203 |
|
217 |
// isDataProvider |
|
218 |
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue())); |
|
219 |
//Description |
|
220 |
buff += getStringField(metadata.getDescription().getValue()); |
|
221 |
//EnglishName |
|
222 |
buff += getStringField(metadata.getEnglishname().getValue()); |
|
223 |
//access restriction |
|
224 |
buff += getStringField(metadata.getDatabaseaccessrestriction().getValue()); |
|
225 |
//logoURL |
|
226 |
buff += getStringField(metadata.getLogourl().getValue()); |
|
227 |
//mission statament |
|
228 |
buff += getStringField(metadata.getMissionstatementurl().getValue()); |
|
229 |
//namespace prefix |
|
230 |
buff += getStringField(metadata.getNamespaceprefix().getValue()); |
|
231 |
//number of items |
|
232 |
buff += getStringField(metadata.getOdnumberofitems().getValue()); |
|
233 |
//OD policies |
|
234 |
buff += getStringField(metadata.getOdpolicies().getValue()); |
|
235 |
//subjects |
|
236 | 204 |
|
205 |
// Content type |
|
206 |
String contentType = new String(); |
|
207 |
for (StringField c : metadata.getOdcontenttypesList()) { |
|
208 |
contentType += c.getValue() + ";"; |
|
209 |
} |
|
210 |
buff += getStringField(contentType); |
|
237 | 211 |
|
238 |
String subj = new String(); |
|
239 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
240 |
subj += s.getValue() + ';'; |
|
212 |
//Access info package |
|
213 |
String accessInfo = new String(); |
|
214 |
for (StringField c : metadata.getAccessinfopackageList()) { |
|
215 |
accessInfo += c.getValue() + ";"; |
|
216 |
} |
|
217 |
buff += getStringField(accessInfo); |
|
241 | 218 |
|
242 |
} |
|
243 |
//subjects |
|
244 |
buff += getStringField(subj); |
|
219 |
//Release start date |
|
220 |
buff += getStringField(metadata.getReleasestartdate().getValue()); |
|
245 | 221 |
|
222 |
//Release end date |
|
223 |
buff += getStringField(metadata.getReleaseenddate().getValue()); |
|
246 | 224 |
|
247 |
return buff; |
|
248 |
} |
|
249 |
/*Legal short name |
|
250 |
Legal name |
|
251 |
Web site url  |
|
252 |
logo url |
|
253 |
*/ |
|
225 |
//Mission statement url |
|
226 |
buff += getStringField(metadata.getMissionstatementurl().getValue()); |
|
254 | 227 |
|
255 |
private String buildOrganization(OafEntity data) { |
|
228 |
//Data provider |
|
229 |
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue())); |
|
256 | 230 |
|
257 |
String buff = new String(); |
|
231 |
//Service provider |
|
232 |
buff += getStringField(String.valueOf(metadata.getServiceprovider().getValue())); |
|
258 | 233 |
|
259 |
Organization organization = data.getOrganization();
|
|
260 |
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata();
|
|
234 |
//Database access type
|
|
235 |
buff += getStringField(metadata.getDatabaseaccessrestriction().getValue());
|
|
261 | 236 |
|
262 |
//`original Id`
|
|
263 |
String originalId = new String();
|
|
237 |
//Data upload type
|
|
238 |
buff += getStringField(metadata.getDatauploadtype().getValue());
|
|
264 | 239 |
|
265 |
for (String oid : data.getOriginalIdList()) { |
|
266 |
originalId += oid; |
|
267 |
break; |
|
268 |
} |
|
240 |
//Data upload restrictions |
|
241 |
buff += getStringField(metadata.getDatauploadrestriction().getValue()); |
|
269 | 242 |
|
270 |
buff += getStringField(originalId); |
|
271 |
//dateOfCollection |
|
272 |
buff += getStringDateField(data.getDateofcollection()); |
|
243 |
//Versioning |
|
244 |
buff += getStringField(String.valueOf(metadata.getVersioning().getValue())); |
|
273 | 245 |
|
246 |
//Citation guideline url |
|
247 |
buff += getStringField(String.valueOf(metadata.getCitationguidelineurl().getValue())); |
|
274 | 248 |
|
275 |
//getLegalshortname |
|
276 |
buff += getStringField(metadata.getLegalshortname().getValue()); |
|
277 |
// `name`, |
|
278 |
buff += getStringField(metadata.getLegalname().getValue()); |
|
279 |
//website URL |
|
280 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
281 |
//logourl |
|
282 |
buff += getStringField(metadata.getLogourl().getValue()); |
|
283 |
// `country`, |
|
284 |
buff += getStringField(metadata.getCountry().getClassname()); |
|
249 |
//Quality management kind |
|
250 |
buff += getStringField(String.valueOf(metadata.getQualitymanagementkind().getValue())); |
|
285 | 251 |
|
252 |
//PID systems |
|
253 |
buff += getStringField(metadata.getPidsystems().getValue()); |
|
286 | 254 |
|
287 |
return buff; |
|
255 |
//Certificates |
|
256 |
buff += getStringField(metadata.getCertificates().getValue()); |
|
288 | 257 |
|
289 |
} |
|
290 |
/* |
|
258 |
//Policies |
|
259 |
String policies = new String(); |
|
260 |
for (FieldTypeProtos.KeyValue property : metadata.getPoliciesList()) { |
|
261 |
policies += property.getValue() + ";"; |
|
262 |
} |
|
263 |
buff += getStringField(policies); |
|
291 | 264 |
|
292 |
Title DateOfAcceptance |
|
265 |
return buff; |
|
266 |
} |
|
293 | 267 |
|
294 |
Publisher(optional) |
|
295 | 268 |
|
296 |
Description PID |
|
297 |
Language RelevantDate |
|
298 |
Subject Instance |
|
299 |
ExternalReference Source |
|
300 |
Format Context |
|
301 |
Country |
|
302 |
journalName     |
|
269 |
private String buildOrganization(OafEntity data) { |
|
303 | 270 |
|
304 |
Best License |
|
305 |
Description                |
|
306 |
Source                               |
|
307 |
Format      |
|
308 |
Journal   |
|
271 |
String buff = new String(); |
|
309 | 272 |
|
310 |
                 |
|
311 |
title                                   |
|
312 |
ISSN Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â |
|
313 |
embargoEndDate     |
|
314 |
hasAuthor |
|
315 |
isRelatedTo |
|
273 |
Organization organization = data.getOrganization(); |
|
274 |
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata(); |
|
316 | 275 |
|
317 |
Resource Type |
|
318 |
Device |
|
319 |
Size |
|
320 |
Version |
|
321 |
Last Metadata Update |
|
322 |
Metadata Version |
|
276 |
//`original Id` |
|
277 |
String originalId = new String(); |
|
278 |
for (String oid : data.getOriginalIdList()) { |
|
279 |
originalId += oid + ";"; |
|
280 |
} |
|
323 | 281 |
|
324 |
*/ |
|
325 | 282 |
|
326 |
private String buildResult(OafEntity data) { |
|
283 |
buff += getStringField(originalId); |
|
284 |
//dateOfCollection |
|
285 |
buff += getStringDateField(data.getDateofcollection()); |
|
327 | 286 |
|
328 | 287 |
|
329 |
String buff = new String(); |
|
288 |
//getLegalshortname |
|
289 |
buff += getStringField(metadata.getLegalshortname().getValue()); |
|
290 |
// `name`, |
|
291 |
buff += getStringField(metadata.getLegalname().getValue()); |
|
292 |
//website URL |
|
293 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
294 |
//logourl |
|
295 |
buff += getStringField(metadata.getLogourl().getValue()); |
|
296 |
// `country`, |
|
297 |
buff += getStringField(metadata.getCountry().getClassname()); |
|
330 | 298 |
|
331 |
Result result = data.getResult(); |
|
332 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata(); |
|
333 | 299 |
|
300 |
return buff; |
|
334 | 301 |
|
335 |
//`original Id` |
|
336 |
String originalId = new String(); |
|
302 |
} |
|
337 | 303 |
|
338 |
for (String oid : data.getOriginalIdList()) { |
|
339 |
originalId += oid; |
|
340 |
break; |
|
341 |
} |
|
304 |
private String buildResult(OafEntity data) { |
|
305 |
String buff = new String(); |
|
342 | 306 |
|
343 |
buff += getStringField(originalId); |
|
307 |
Result result = data.getResult(); |
|
308 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata(); |
|
344 | 309 |
|
345 |
//dateOfCollection
|
|
346 |
buff += getStringDateField(data.getDateofcollection());
|
|
310 |
//`original Id`
|
|
311 |
String originalId = new String();
|
|
347 | 312 |
|
348 |
// titleString |
|
313 |
for (String oid : data.getOriginalIdList()) { |
|
314 |
originalId += oid + ";"; |
|
315 |
} |
|
349 | 316 |
|
350 |
String titleString = new String(); |
|
351 |
String alternativeTitles = new String(); |
|
317 |
buff += getStringField(originalId); |
|
352 | 318 |
|
319 |
//dateOfCollection |
|
320 |
buff += getStringDateField(data.getDateofcollection()); |
|
321 |
// titleString |
|
322 |
String titleString = new String(); |
|
323 |
String alternativeTitles = new String(); |
|
353 | 324 |
|
354 |
for (int i = 0; i < metadata.getTitleList().size(); i++) { |
|
355 |
StructuredProperty title = metadata.getTitleList().get(i); |
|
356 |
if (i == 0) { |
|
357 |
titleString = title.getValue().replaceAll("\\s+", " "); |
|
358 |
titleString = titleString.replaceAll("\n", " "); |
|
359 |
} else { |
|
360 |
alternativeTitles += title.getValue().replaceAll("\\s+", " ") + " ; "; |
|
361 |
alternativeTitles = alternativeTitles.replaceAll("\n", " "); |
|
362 |
} |
|
363 |
break; |
|
364 |
} |
|
365 |
// pubtitle |
|
366 |
buff += getStringField(titleString); |
|
367 |
// alternative titles |
|
368 |
//buff += getStringField(alternativeTitles); // null#! |
|
325 |
for (int i = 0; i < metadata.getTitleList().size(); i++) { |
|
326 |
StructuredProperty title = metadata.getTitleList().get(i); |
|
327 |
if (i == 0) { |
|
328 |
titleString = title.getValue().replaceAll("\\s+", " "); |
|
329 |
titleString = titleString.replaceAll("\n", " "); |
|
330 |
} else { |
|
331 |
alternativeTitles += title.getValue().replaceAll("\\s+", " ") + " ; "; |
|
332 |
alternativeTitles = alternativeTitles.replaceAll("\n", " "); |
|
333 |
} |
|
334 |
break; |
|
335 |
} |
|
369 | 336 |
|
370 |
// format |
|
371 |
String formatString = new String(); |
|
372 |
for (StringField format : metadata.getFormatList()) { |
|
373 |
formatString = format.getValue(); |
|
374 |
break; |
|
375 |
} |
|
337 |
// pubtitle |
|
338 |
buff += getStringField(titleString); |
|
376 | 339 |
|
377 |
buff += getStringField(formatString); |
|
378 |
// publisher |
|
340 |
// alternative titles |
|
341 |
//buff += getStringField(alternativeTitles); // null#! |
|
342 |
// date of acceptance CHANGED THIS TO DATE FORMAT |
|
343 |
buff += getStringDateField(metadata.getDateofacceptance().getValue()); |
|
379 | 344 |
|
380 |
buff += getStringField(metadata.getPublisher().getValue());
|
|
381 |
// journal
|
|
345 |
// publisher
|
|
346 |
buff += getStringField(metadata.getPublisher().getValue());
|
|
382 | 347 |
|
383 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#! |
|
384 |
// isssnLinking |
|
385 |
buff += getStringField(metadata.getJournal().getIssnLinking()); //#null#! |
|
386 |
// year |
|
387 |
buff += getYearInt(metadata.getDateofacceptance().getValue()); |
|
388 | 348 |
|
389 |
// date CHANGED THIS TO DATE FORMAT |
|
390 |
buff += getStringDateField(metadata.getDateofacceptance().getValue()); |
|
349 |
//PID |
|
350 |
String pids = new String(); |
|
351 |
for (StructuredProperty p : data.getPidList()) { |
|
352 |
pids += p.getValue() + ";"; |
|
353 |
} |
|
354 |
buff += getStringField(pids); |
|
391 | 355 |
|
392 |
// access_mode,
|
|
393 |
buff += getStringField(getAccessMode(result));
|
|
356 |
//language
|
|
357 |
buff += getStringField(metadata.getLanguage().getClassname());
|
|
394 | 358 |
|
395 |
// bestlicense |
|
359 |
// RelevantDate |
|
360 |
String reldate = new String(); |
|
396 | 361 |
|
397 |
buff += getStringField(getBestLicense(result));
|
|
398 |
// type
|
|
399 |
buff += getStringField(metadata.getResulttype().getClassname());
|
|
400 |
// embargo_end_date
|
|
401 |
buff += getStringField(metadata.getEmbargoenddate().getValue());
|
|
362 |
for (StructuredProperty p : metadata.getRelevantdateList()) {
|
|
363 |
reldate += p.getValue();
|
|
364 |
break;
|
|
365 |
}
|
|
366 |
buff += getStringField(reldate);
|
|
402 | 367 |
|
403 |
//language |
|
368 |
//Subject |
|
369 |
String subjects = new String(); |
|
370 |
for (StructuredProperty subj : metadata.getSubjectList()) { |
|
404 | 371 |
|
405 |
buff += getStringField(metadata.getLanguage().getClassname()); |
|
406 |
//classification |
|
372 |
if (isValidTopic(subj.getValue())) { |
|
373 |
if (!isNumeric(subj.getValue())) { |
|
374 |
subjects += subj.getValue() + ";"; |
|
375 |
} |
|
376 |
} |
|
377 |
} |
|
407 | 378 |
|
408 |
String classification = new String(); |
|
409 |
for (Instance instance : (result.getInstanceList())) { |
|
379 |
buff += subjects; |
|
410 | 380 |
|
411 |
if (instance.getInstancetype().getClassname() != null && !instance.getInstancetype().getClassname().isEmpty()) { |
|
412 |
classification += instance.getInstancetype().getClassname() + ';'; |
|
413 |
} |
|
381 |
// TODO Instance |
|
382 |
// buff += getStringField(); |
|
414 | 383 |
|
415 |
} |
|
416 |
buff += getStringField(classification); |
|
384 |
//TODO ExternalReference |
|
417 | 385 |
|
418 | 386 |
|
419 |
String hostedBy = new String(); |
|
420 |
// hosted by |
|
421 |
for (Instance instance : (result.getInstanceList())) { |
|
387 |
//Source |
|
388 |
String source = new String(); |
|
389 |
for (StringField s : metadata.getSourceList()) { |
|
390 |
source += s.getValue() + ";"; |
|
391 |
} |
|
422 | 392 |
|
393 |
buff += getStringField(source); |
|
423 | 394 |
|
424 |
String host = instance.getHostedby().getKey(); |
|
425 |
if (host != null && !host.isEmpty()) { |
|
426 |
hostedBy += host + ";"; |
|
427 |
} |
|
428 |
} |
|
429 | 395 |
|
430 |
buff += getStringField(hostedBy); |
|
396 |
//TODO Format     |
|
397 |
buff += getStringField(""); |
|
398 |
//DOES NOT EXIST |
|
399 |
/*String formatString = new String(); |
|
400 |
for (StringField format : metadata.getFormatList()) { |
|
401 |
formatString = format.getValue(); |
|
402 |
break;} |
|
403 |
buff += getStringField(formatString); |
|
404 |
}*/ |
|
405 |
//Context |
|
406 |
String context = new String(); |
|
407 |
for (Result.Context s : metadata.getContextList()) { |
|
408 |
context += s.getId() + ";"; |
|
409 |
} |
|
410 |
buff += getStringField(context); |
|
431 | 411 |
|
432 |
String collectedFrom = new String(); |
|
433 |
//TODO collectedfrom |
|
412 |
//country |
|
413 |
String country = new String(); |
|
414 |
for (Qualifier s : metadata.getCountryList()) { |
|
415 |
country += s.getClassname() + ";"; |
|
416 |
} |
|
434 | 417 |
|
435 |
for (FieldTypeProtos.KeyValue collectedFromValue : (data.getCollectedfromList())) {
|
|
418 |
buff += getStringField(country);
|
|
436 | 419 |
|
437 |
String host = collectedFromValue.getKey(); |
|
438 |
if (host != null && !host.isEmpty()) { |
|
439 |
collectedFrom += host + ';'; |
|
420 |
// access_mode, |
|
421 |
buff += getStringField(getAccessMode(result)); |
|
422 |
//Best License |
|
423 |
buff += getStringField(getBestLicense(result)); |
|
424 |
//Description |
|
425 |
String description = new String(); |
|
440 | 426 |
|
441 |
} |
|
442 |
} |
|
427 |
for (StringField desc : metadata.getDescriptionList()) { |
|
428 |
description += desc; |
|
429 |
break; |
|
430 |
} |
|
431 |
buff += getStringField(description); |
|
432 |
//Journal  |
|
433 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#! |
|
443 | 434 |
|
444 |
buff += getStringField(collectedFrom); |
|
435 |
//journalName                     |
|
436 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#! |
|
445 | 437 |
|
446 |
String subjects = new String(); |
|
447 |
for (StructuredProperty subj : metadata.getSubjectList()) { |
|
438 |
// TODO ERI SOS : HERE IN GET JOUTNAL. GET DATA INFO I CAN FIND PROVENANCE AND SIMILARITY |
|
448 | 439 |
|
449 |
if (isValidTopic(subj.getValue())) { |
|
450 |
if (!isNumeric(subj.getValue())) { |
|
451 |
subjects += subj.getValue() + ";"; |
|
452 |
} |
|
453 |
} |
|
454 |
} |
|
440 |
//ISSNÂ Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â |
|
441 |
buff += getStringField(metadata.getJournal().getIssnLinking()); |
|
455 | 442 |
|
456 |
buff += subjects; |
|
443 |
//embargoEndDate    |
|
444 |
buff += getStringField(metadata.getEmbargoenddate().getValue()); |
|
457 | 445 |
|
458 |
// `authors`, |
|
459 |
int authors = 0; |
|
460 |
String delayed = "no"; |
|
461 | 446 |
|
462 |
for (OafRel rel : data.getCachedRelList()) { |
|
447 |
// `authors`, |
|
448 |
int authors = 0; |
|
449 |
String delayed = "no"; |
|
463 | 450 |
|
464 |
if (rel.getRelType().equals(RelType.personResult)) {
|
|
451 |
for (OafRel rel : data.getCachedRelList()) {
|
|
465 | 452 |
|
466 |
authors++; |
|
467 |
} else if (rel.getRelType().equals(RelType.resultProject)) |
|
468 |
// TODO remember : in result Project, first id is project, second is result. |
|
453 |
if (rel.getRelType().equals(RelType.personResult)) { |
|
469 | 454 |
|
470 |
{ |
|
471 |
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), rel.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
472 |
if (Integer.parseInt(daysfromend) > 0) { |
|
473 |
delayed = "yes"; |
|
474 |
} |
|
475 |
} |
|
476 |
} |
|
477 |
// `delayed`, |
|
478 |
buff += getStringField(delayed); |
|
455 |
authors++; |
|
456 |
} else if (rel.getRelType().equals(RelType.resultProject)) |
|
457 |
// TODO remember : in result Project, first id is project, second is result. |
|
479 | 458 |
|
480 |
// log.info("Result " + full_id +"Author count : " + authors ); |
|
481 |
buff += getNumericField(String.valueOf(authors)); |
|
459 |
{ |
|
460 |
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), rel.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
461 |
if (Integer.parseInt(daysfromend) > 0) { |
|
462 |
delayed = "yes"; |
|
463 |
} |
|
464 |
} |
|
465 |
} |
|
482 | 466 |
|
483 |
// `context `,
|
|
467 |
buff += getNumericField(String.valueOf(authors));
|
|
484 | 468 |
|
485 |
String context = new String(); |
|
486 | 469 |
|
487 |
for (Result.Context s : metadata.getContextList()) { |
|
488 |
context += s.getId() + ";"; |
|
470 |
// TODO isRelatedTo |
|
489 | 471 |
|
490 |
} |
|
472 |
// resource type |
|
473 |
buff += getStringField(metadata.getResourcetype().getClassname()); |
|
474 |
// device |
|
475 |
buff += getStringField(metadata.getDevice().getValue()); |
|
476 |
// size |
|
477 |
buff += getStringField(metadata.getSize().getValue()); |
|
478 |
// version |
|
479 |
buff += getStringField(metadata.getVersion().getValue()); |
|
480 |
// metadata update |
|
481 |
buff += getStringField(metadata.getLastmetadataupdate().getValue()); |
|
482 |
// metadata version |
|
483 |
buff += getStringField(metadata.getMetadataversionnumber().getValue()); |
|
491 | 484 |
|
492 |
buff += getStringField(context); |
|
485 |
// `delayed`, |
|
486 |
buff += getStringField(delayed); |
|
493 | 487 |
|
488 |
// year |
|
489 |
buff += getYearInt(metadata.getDateofacceptance().getValue()); |
|
494 | 490 |
|
495 |
String country = new String(); |
|
491 |
// type |
|
492 |
buff += getStringField(metadata.getResulttype().getClassname()); |
|
493 |
//classification |
|
494 |
String classification = new String(); |
|
495 |
for (Instance instance : (result.getInstanceList())) { |
|
496 | 496 |
|
497 |
for (Qualifier s : metadata.getCountryList()) { |
|
498 |
country += s.getClassname() + ";"; |
|
499 |
} |
|
497 |
if (instance.getInstancetype().getClassname() != null && !instance.getInstancetype().getClassname().isEmpty()) { |
|
498 |
classification += instance.getInstancetype().getClassname() + ';'; |
|
499 |
} |
|
500 |
} |
|
500 | 501 |
|
501 |
buff += getStringField(country);
|
|
502 |
buff += getStringField(classification);
|
|
502 | 503 |
|
503 |
String desc = new String(); |
|
504 |
for (StringField s : metadata.getDescriptionList()) { |
|
505 |
desc += s.getValue() + ";"; |
|
506 |
} |
|
504 |
// hosted by |
|
505 |
String hostedBy = new String(); |
|
506 |
for (Instance instance : (result.getInstanceList())) { |
|
507 |
String host = instance.getHostedby().getKey(); |
|
508 |
if (host != null && !host.isEmpty()) { |
|
509 |
hostedBy += host + ";"; |
|
510 |
} |
|
511 |
} |
|
507 | 512 |
|
508 |
buff += getStringField(desc); |
|
509 |
// embargo_end_date |
|
513 |
buff += getStringField(hostedBy); |
|
510 | 514 |
|
511 |
String relevantDate = new String(); |
|
512 |
for (StructuredProperty s : metadata.getRelevantdateList()) { |
|
513 |
relevantDate += s.getValue(); |
|
514 |
break; |
|
515 |
} |
|
516 |
buff += getStringField(relevantDate); |
|
515 |
// collectedfrom |
|
516 |
String collectedFrom = new String(); |
|
517 |
for (FieldTypeProtos.KeyValue collectedFromValue : (data.getCollectedfromList())) { |
|
517 | 518 |
|
519 |
String host = collectedFromValue.getKey(); |
|
520 |
if (host != null && !host.isEmpty()) { |
|
521 |
collectedFrom += host + ';'; |
|
518 | 522 |
|
519 |
// device |
|
520 |
buff += getStringField(metadata.getDevice().getValue()); |
|
521 |
// size |
|
522 |
buff += getStringField(metadata.getSize().getValue()); |
|
523 |
} |
|
524 |
} |
|
525 |
buff += getStringField(collectedFrom); |
|
526 |
return buff; |
|
527 |
} |
|
523 | 528 |
|
524 | 529 |
|
525 |
return buff; |
|
526 |
} |
|
530 |
private String buildProject(OafEntity data) { |
|
527 | 531 |
|
532 |
String buff = new String(); |
|
528 | 533 |
|
529 |
private boolean isValid(String buff, String id) { |
|
530 |
if (buff.endsWith(ENCLOSED)) { |
|
531 |
log.error("Empty Result with " + id + " with body: \n" + buff); |
|
532 |
return false; |
|
533 |
} |
|
534 |
return true; |
|
535 |
} |
|
534 |
Project project = data.getProject(); |
|
535 |
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata(); |
|
536 | 536 |
|
537 |
private String getBestLicense(Result result) { |
|
538 |
Qualifier bestLicense = null; |
|
539 |
LicenseComparator lc = new LicenseComparator(); |
|
540 |
for (Instance instance : (result.getInstanceList())) { |
|
541 |
if (lc.compare(bestLicense, instance.getLicence()) > 0) { |
|
542 |
bestLicense = instance.getLicence(); |
|
543 |
} |
|
544 |
} |
|
545 |
if (bestLicense != null) { |
|
546 |
return bestLicense.getClassname(); |
|
547 |
} else { |
|
548 |
return null; |
|
549 |
} |
|
550 |
} |
|
537 |
//`original Id` |
|
538 |
String originalId = new String(); |
|
551 | 539 |
|
552 |
// here iterate over all values |
|
553 |
private String getAccessMode(Result result) { |
|
554 |
String accessMode = NULL_STRING; |
|
555 |
for (Instance instance : (result.getInstanceList())) { |
|
556 |
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) { |
|
557 |
accessMode = instance.getLicence().getClassname(); |
|
558 |
break; |
|
559 |
} |
|
540 |
for (String oid : data.getOriginalIdList()) { |
|
541 |
originalId += oid + ";"; |
|
542 |
} |
|
560 | 543 |
|
561 |
}
|
|
544 |
buff += getStringField(originalId);
|
|
562 | 545 |
|
546 |
//dateOfCollection |
|
547 |
buff += getStringDateField(data.getDateofcollection()); |
|
563 | 548 |
|
564 |
return accessMode; |
|
565 |
} |
|
549 |
//Code |
|
550 |
buff += getStringField(metadata.getCode().getValue()); |
|
551 |
// `url`, |
|
552 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
566 | 553 |
|
554 |
// `acronym`, |
|
555 |
String acronym = metadata.getAcronym().getValue(); |
|
556 |
if (acronym.equalsIgnoreCase("UNKNOWN")) { |
|
557 |
acronym = metadata.getTitle().getValue(); |
|
558 |
} |
|
567 | 559 |
|
568 |
private String buildProject(OafEntity data) {
|
|
560 |
buff += getStringField(acronym);
|
|
569 | 561 |
|
570 |
String buff = new String(); |
|
562 |
//title! |
|
563 |
String title = getStringField(metadata.getTitle().getValue()); |
|
564 |
buff += getStringField(title); |
|
571 | 565 |
|
572 |
Project project = data.getProject();
|
|
573 |
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata();
|
|
566 |
// startdate
|
|
567 |
buff += getNumericField(metadata.getStartdate().getValue());
|
|
574 | 568 |
|
575 |
//`original Id`
|
|
576 |
String originalId = new String();
|
|
569 |
// enddate
|
|
570 |
buff += getNumericField(metadata.getEnddate().getValue());
|
|
577 | 571 |
|
578 |
for (String oid : data.getOriginalIdList()) { |
|
579 |
originalId += (oid); |
|
580 |
break; |
|
581 |
} |
|
582 |
buff += getStringField(originalId); |
|
572 |
//`Call identifer` |
|
573 |
buff += getStringField(metadata.getCallidentifier().getValue()); |
|
583 | 574 |
|
584 |
//dateOfCollection |
|
585 |
buff += getStringDateField(data.getDateofcollection()); |
|
586 |
//code |
|
587 |
buff += getStringField(metadata.getCode().getValue()); |
|
575 |
//`KeyWords` |
|
576 |
buff += getStringField(metadata.getKeywords().getValue()); |
|
588 | 577 |
|
589 |
// `acronym`, |
|
590 |
String acronym = metadata.getAcronym().getValue(); |
|
591 |
if (acronym.equalsIgnoreCase("UNKNOWN")) { |
|
592 |
acronym = metadata.getTitle().getValue(); |
|
593 |
} |
|
578 |
//`Duration` |
|
579 |
buff += getStringField(metadata.getDuration().getValue()); |
|
594 | 580 |
|
595 |
buff += getStringField(acronym); |
|
581 |
//esc39 |
|
582 |
buff += getStringField(metadata.getEcsc39().getValue().toString()); |
|
596 | 583 |
|
597 |
//title! |
|
598 |
String title = getStringField(metadata.getTitle().getValue()); |
|
599 |
buff += getStringField(title); |
|
584 |
//`Contracttype` |
|
585 |
buff += getStringField(metadata.getContracttype().getClassname()); |
|
600 | 586 |
|
601 |
List<StringField> fundList = metadata.getFundingtreeList(); |
|
587 |
//`OA mandate pubs` |
|
588 |
buff += getStringField(metadata.getOamandatepublications().getValue()); |
|
602 | 589 |
|
590 |
//`Subjects` |
|
591 |
String subjects = new String(); |
|
592 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
603 | 593 |
|
604 |
if (!fundList.isEmpty()) // `funding_lvl0`, |
|
605 |
{ |
|
606 |
//TODO funder + 3 funding levels |
|
607 |
/* funder text, |
|
608 |
funding_lvl0 text, |
|
594 |
subjects += s.getValue() + ';'; |
|
595 |
} |
|
596 |
buff += getStringField(subjects); |
|
597 |
|
|
598 |
|
|
599 |
//`EC293` |
|
600 |
buff += getStringField(metadata.getEcarticle293().getValue()); |
|
601 |
|
|
602 |
|
|
603 |
List<StringField> fundList = metadata.getFundingtreeList(); |
|
604 |
|
|
605 |
//TODO possible to have multiple funders per project? |
|
606 |
if (!fundList.isEmpty()) // `funding_lvl0`, |
|
607 |
{ |
|
608 |
//TODO funder + 3 funding levels |
|
609 |
/* funder text, |
|
610 |
funding_lvl0 text, |
|
609 | 611 |
funding_lvl1 text, |
610 | 612 |
funding_lvl2 text, |
611 | 613 |
funding_lvl3 text,*/ |
612 |
buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue());
|
|
614 |
buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue());
|
|
613 | 615 |
|
614 |
} else {
|
|
616 |
} else {
|
|
615 | 617 |
|
616 |
buff += this.fundingParser.getFundingInfo("");
|
|
617 |
}
|
|
618 |
buff += this.fundingParser.getFundingInfo("");
|
|
619 |
}
|
|
618 | 620 |
|
621 |
return buff; |
|
619 | 622 |
|
620 |
//esc39
|
|
623 |
}
|
|
621 | 624 |
|
622 |
buff += getStringField(metadata.getEcsc39().getValue().toString()); |
|
623 | 625 |
|
624 |
// `url`, |
|
625 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
626 |
private String buildPerson(OafEntity data) { |
|
626 | 627 |
|
627 |
// enddate |
|
628 |
buff += getNumericField(metadata.getEnddate().getValue()); |
|
629 |
// startdate |
|
630 |
buff += getNumericField(metadata.getStartdate().getValue()); |
|
631 |
//`Call identifer` |
|
632 |
buff += getStringField(metadata.getCallidentifier().getValue()); |
|
633 |
//`EC293` |
|
634 |
buff += getStringField(metadata.getEcarticle293().getValue()); |
|
635 |
//`KeyWords` |
|
636 |
buff += getStringField(metadata.getKeywords().getValue()); |
|
637 |
//`OA mandate pubs` |
|
638 |
buff += getStringField(metadata.getOamandatepublications().getValue()); |
|
639 |
//`Contracttype` |
|
640 |
buff += getStringField(metadata.getContracttype().getClassname()); |
|
641 |
//`Duration` |
|
642 |
buff += getStringField(metadata.getDuration().getValue()); |
|
628 |
String buff = new String(); |
|
643 | 629 |
|
644 |
//`Code` |
|
645 |
buff += getStringField(metadata.getCode().getValue()); |
|
646 |
//`Keywords` |
|
647 |
buff += getStringField(metadata.getKeywords().getValue()); |
|
630 |
PersonProtos.Person person = data.getPerson(); |
|
631 |
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata(); |
|
648 | 632 |
|
649 |
//`Subjects` |
|
650 | 633 |
|
651 |
String subjects = new String();
|
|
652 |
for (StructuredProperty s : metadata.getSubjectsList()) {
|
|
634 |
//`original Id`
|
|
635 |
String originalId = new String();
|
|
653 | 636 |
|
654 |
subjects += s.getValue() + ';';
|
|
655 |
}
|
|
656 |
buff += getStringField(subjects);
|
|
637 |
for (String oid : data.getOriginalIdList()) {
|
|
638 |
originalId += oid + ";";
|
|
639 |
}
|
|
657 | 640 |
|
641 |
buff += getStringField(originalId); |
|
642 |
//dateOfCollection |
|
643 |
buff += getStringDateField(data.getDateofcollection()); |
|
658 | 644 |
|
659 |
return buff; |
|
645 |
// `firstname`, |
|
646 |
buff += metadata.getFirstname(); |
|
660 | 647 |
|
661 |
}
|
|
648 |
// `secondNames`,
|
|
662 | 649 |
|
650 |
String secondNames = new String(); |
|
651 |
for (StringField s : metadata.getSecondnamesList()) { |
|
663 | 652 |
|
664 |
private String buildPerson(OafEntity data) { |
|
653 |
secondNames += s.getValue() + ' '; |
|
654 |
} |
|
665 | 655 |
|
666 |
String buff = new String();
|
|
656 |
buff += getStringField(secondNames);
|
|
667 | 657 |
|
668 |
PersonProtos.Person person = data.getPerson(); |
|
669 |
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata(); |
|
658 |
// `fullname`, |
|
659 |
buff += getStringField(metadata.getFullname().getValue()); |
|
660 |
// `Fax`, |
|
661 |
buff += getStringField(metadata.getFax().getValue()); |
|
662 |
// `Email`, |
|
663 |
buff += getStringField(metadata.getEmail().getValue()); |
|
664 |
// `Phone`, |
|
665 |
buff += getStringField(metadata.getPhone().getValue()); |
|
670 | 666 |
|
667 |
// `Nationality`, |
|
668 |
buff += getStringField(metadata.getNationality().getClassname()); |
|
671 | 669 |
|
672 |
//`original Id` |
|
673 |
String originalId = new String(); |
|
670 |
// `PIDS`, |
|
671 |
String pids = new String(); |
|
672 |
for (StructuredProperty s : data.getPidList()) { |
|
674 | 673 |
|
675 |
for (String oid : data.getOriginalIdList()) { |
|
676 |
originalId += oid; |
|
677 |
break; |
|
678 |
} |
|
674 |
pids += s.getValue() + ";"; |
|
679 | 675 |
|
680 |
buff += getStringField(originalId); |
|
681 |
//dateOfCollection |
|
682 |
buff += getStringDateField(data.getDateofcollection()); |
|
683 | 676 |
|
684 |
// `firstname`,
|
|
685 |
buff += metadata.getFirstname();
|
|
677 |
}
|
|
678 |
buff += getStringField(pids);
|
|
686 | 679 |
|
687 |
// `secondNames`, |
|
680 |
// `collected from`, |
|
681 |
String collectedFrom = new String(); |
|
682 |
for (FieldTypeProtos.KeyValue s : data.getCollectedfromList()) { |
|
683 |
collectedFrom += s.getValue() + ";"; |
|
684 |
} |
|
685 |
buff += getStringField(collectedFrom); |
|
686 |
return buff; |
|
688 | 687 |
|
689 |
String secondNames = new String(); |
|
690 |
for (StringField s : metadata.getSecondnamesList()) { |
|
688 |
} |
|
691 | 689 |
|
692 |
secondNames += s.getValue() + ' '; |
|
693 |
} |
|
690 |
public String cleanId(String value) { |
|
691 |
if (value != null) { |
|
692 |
// DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
694 | 693 |
|
695 |
buff += getStringField(secondNames); |
|
696 | 694 |
|
697 |
// `fullname`, |
|
698 |
buff += getStringField(metadata.getFullname().getValue()); |
|
695 |
// to datacite____:: ) |
|
696 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
697 |
value = value.replaceFirst(".*\\|", ""); |
|
698 |
value = value.replaceAll("\n", ""); |
|
699 |
value = value.replaceAll(DELIM, ""); |
|
700 |
value = value.replaceAll(ENCLOSED, ""); |
|
701 |
value = value.trim(); |
|
699 | 702 |
|
700 |
// `Nationality`, |
|
701 |
buff += getStringField(metadata.getNationality().getClassname()); |
|
702 |
// `Email`, |
|
703 |
buff += getStringField(metadata.getEmail().getValue()); |
|
704 |
// `Phone`, |
|
705 |
buff += getStringField(metadata.getPhone().getValue()); |
|
706 |
// `Fax`, |
|
707 |
buff += getStringField(metadata.getFax().getValue()); |
|
703 |
} |
|
704 |
if (value == null) { |
|
705 |
return null; |
|
706 |
} |
|
707 |
return ENCLOSED + value + ENCLOSED; |
|
708 | 708 |
|
709 |
// `collected from`, |
|
710 |
String collectedFrom = new String(); |
|
711 |
for (FieldTypeProtos.KeyValue s : data.getCollectedfromList()) { |
|
709 |
} |
|
712 | 710 |
|
713 |
collectedFrom += s.getValue(); |
|
714 |
break; |
|
715 | 711 |
|
716 |
} |
|
717 |
buff += getStringField(collectedFrom); |
|
712 |
private String getNumericField(String data) { |
|
713 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
714 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
715 |
} else { |
|
718 | 716 |
|
717 |
return ENCLOSED + data + ENCLOSED + DELIM; |
|
718 |
} |
|
719 |
} |
|
719 | 720 |
|
720 |
return buff; |
|
721 | 721 |
|
722 |
}
|
|
722 |
private String getYearDifferenceInteger(String enddate, String startdate) {
|
|
723 | 723 |
|
724 |
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) { |
|
724 | 725 |
|
725 |
private String getYearDifferenceInteger(String enddate, String startdate) {
|
|
726 |
String[] split = startdate.split("-");
|
|
726 | 727 |
|
727 |
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) { |
|
728 |
if (split == null || split.length == 0) { |
|
729 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
730 |
} |
|
728 | 731 |
|
729 |
String[] split = startdate.split("-");
|
|
732 |
int Startdate = Integer.parseInt(split[0]);
|
|
730 | 733 |
|
731 |
if (split == null || split.length == 0) { |
|
732 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
733 |
} |
|
734 |
split = enddate.split("-"); |
|
734 | 735 |
|
735 |
int Startdate = Integer.parseInt(split[0]); |
|
736 |
if (split == null || split.length == 0) { |
|
737 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
738 |
} |
|
736 | 739 |
|
737 |
split = enddate.split("-");
|
|
740 |
int Enddate = Integer.parseInt(split[0]);
|
|
738 | 741 |
|
739 |
if (split == null || split.length == 0) { |
|
740 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
741 |
} |
|
742 |
int diff = Enddate - Startdate; |
|
742 | 743 |
|
743 |
int Enddate = Integer.parseInt(split[0]);
|
|
744 |
return ENCLOSED + diff + ENCLOSED + DELIM;
|
|
744 | 745 |
|
745 |
int diff = Enddate - Startdate;
|
|
746 |
}
|
|
746 | 747 |
|
747 |
return ENCLOSED + diff + ENCLOSED + DELIM; |
|
748 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
749 |
} |
|
748 | 750 |
|
749 |
} |
|
751 |
private String getYearInt(String data) { |
|
752 |
if (data == null || data.isEmpty() || data.equals("-1")) { |
|
753 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
754 |
} |
|
750 | 755 |
|
751 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
752 |
} |
|
756 |
String[] split = data.split("-"); |
|
753 | 757 |
|
754 |
private String getYearInt(String data) { |
|
755 |
if (data == null || data.isEmpty() || data.equals("-1")) { |
|
756 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
757 |
} |
|
758 |
if (split == null || split.length == 0) { |
|
759 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
760 |
} |
|
758 | 761 |
|
759 |
String[] split = data.split("-");
|
|
762 |
String year = split[0];
|
|
760 | 763 |
|
761 |
if (split == null || split.length == 0) { |
|
762 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
763 |
} |
|
764 |
year = cleanNumber(year); |
|
764 | 765 |
|
765 |
String year = split[0];
|
|
766 |
return ENCLOSED + year + ENCLOSED + DELIM;
|
|
766 | 767 |
|
767 |
year = cleanNumber(year); |
|
768 | 768 |
|
769 |
return ENCLOSED + year + ENCLOSED + DELIM;
|
|
769 |
}
|
|
770 | 770 |
|
771 |
private static String cleanNumber(String number) { |
|
772 |
number = number.replaceAll("[^A-Za-z0-9:,____]", ""); |
|
771 | 773 |
|
772 |
} |
|
774 |
return number; |
|
775 |
} |
|
773 | 776 |
|
774 |
private String cleanNumber(String number) { |
|
775 |
number = number.replaceAll("[^A-Za-z0-9:,____]", ""); |
|
777 |
private String getStringField(String data) { |
|
776 | 778 |
|
777 |
return number; |
|
778 |
} |
|
779 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
779 | 780 |
|
780 |
private String getStringField(String data) { |
|
781 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
782 |
} else { |
|
781 | 783 |
|
782 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
784 |
String field = clean(data); |
|
785 |
if (field == null) { |
|
786 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
787 |
} else { |
|
788 |
return field + DELIM; |
|
789 |
} |
|
790 |
} |
|
791 |
} |
|
783 | 792 |
|
784 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
785 |
} else { |
|
793 |
private String getStringDateField(String data) { |
|
786 | 794 |
|
787 |
String field = clean(data); |
|
788 |
if (field == null) { |
|
789 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
790 |
} else { |
|
791 |
return field + DELIM; |
|
792 |
} |
|
793 |
} |
|
794 |
} |
|
795 |
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) { |
|
795 | 796 |
|
796 |
private String getStringDateField(String data) { |
|
797 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
798 |
} else { |
|
797 | 799 |
|
798 |
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) { |
|
800 |
String field = clean(data); |
|
801 |
if (field == null) { |
|
802 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
803 |
} else { |
|
804 |
return field + DELIM; |
|
805 |
} |
|
806 |
} |
|
807 |
} |
|
799 | 808 |
|
800 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
801 |
} else { |
|
802 | 809 |
|
803 |
String field = clean(data); |
|
804 |
if (field == null) { |
|
805 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
806 |
} else { |
|
807 |
return field + DELIM; |
|
808 |
} |
|
809 |
} |
|
810 |
} |
|
810 |
public String getId(OafProtos.Oaf oaf) { |
|
811 |
switch (oaf.getKind()) { |
|
812 |
case entity: |
|
811 | 813 |
|
812 |
private String getNumericField(String data) { |
|
813 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
814 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
815 |
} else { |
|
814 |
return cleanId(oaf.getEntity().getId()); |
|
815 |
case relation: |
|
816 | 816 |
|
817 |
return ENCLOSED + data + ENCLOSED + DELIM; |
|
818 |
} |
|
819 |
} |
|
817 |
return cleanId(oaf.getRel().getSource()); |
|
820 | 818 |
|
821 |
public String getId(Oaf oaf) { |
|
822 |
switch (oaf.getKind()) { |
|
823 |
case entity: |
|
819 |
} |
|
820 |
return null; |
|
824 | 821 |
|
825 |
return cleanId(oaf.getEntity().getId()); |
|
826 |
case relation: |
|
822 |
} |
|
827 | 823 |
|
828 |
return cleanId(oaf.getRel().getSource());
|
|
824 |
private boolean isNumeric(String str) {
|
|
829 | 825 |
|
830 |
} |
|
831 |
return null; |
|
826 |
str = str.replaceAll("[^A-Za-z0-9 ]", ""); |
|
827 |
str = str.replaceAll(" ", ""); |
|
828 |
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional |
|
829 |
// '-' and decimal. |
|
830 |
} |
|
832 | 831 |
|
833 |
} |
|
832 |
// there are topics with "null" as value -> replace them |
|
833 |
private boolean isValidTopic(String t) { |
|
834 | 834 |
|
835 |
public String getId(OafRel relOaf) {
|
|
836 |
return cleanId(relOaf.getSource());
|
|
837 |
}
|
|
835 |
if (t == null || t.isEmpty()) {
|
|
836 |
return false;
|
|
837 |
}
|
|
838 | 838 |
|
839 |
public static String clean(String value) { |
|
840 |
if (value != null) { |
|
841 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
842 |
// to datacite____:: ) |
|
843 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
844 |
value = value.replaceFirst(".*\\|", ""); |
|
845 |
value = value.replaceAll(DELIM, ""); |
|
846 |
value = value.replaceAll(",", ""); |
|
847 |
value = value.replaceAll("\"", ""); |
|
848 |
value = value.replaceAll("'", ""); |
|
849 |
value = value.replaceAll(ENCLOSED, ""); |
|
850 |
value = value.replaceAll("\\r\\n|\\r|\\n", " "); |
|
851 |
value = value.replaceAll("\\s+", " "); |
|
852 |
value = value.replaceAll("(\\r|\\n)", " "); |
|
853 |
value = value.replaceAll("\\t", " "); |
|
839 |
if (t.equals("") || t.equals(" ")) { |
|
840 |
return false; |
|
841 |
} |
|
842 |
if (t.equals("null") || t.equals("Null") || t.equals("NULL")) { |
|
843 |
return false; |
|
844 |
} |
|
854 | 845 |
|
855 |
// value = value.replaceAll("[^A-Za-z0-9:,____-;:]", " "); |
|
856 |
value = value.trim(); |
|
846 |
if (t.equals(ENCLOSED + ENCLOSED + DELIM) || t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) { |
|
847 |
return false; |
|
848 |
} |
|
849 |
// skip dedups |
|
850 |
if (t.contains("ddc:")) { |
|
857 | 851 |
|
858 |
} |
|
859 |
if (value == null) { |
|
860 |
return null; |
|
861 |
} |
|
862 |
return ENCLOSED + value + ENCLOSED; |
|
852 |
return false; |
|
853 |
} |
|
854 |
return true; |
|
855 |
} |
|
863 | 856 |
|
864 |
} |
|
865 | 857 |
|
866 |
public static String cleanId(String value) { |
|
867 |
if (value != null) { |
|
868 |
// DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
858 |
private String getBestLicense(ResultProtos.Result result) { |
|
859 |
FieldTypeProtos.Qualifier bestLicense = null; |
|
860 |
LicenseComparator lc = new LicenseComparator(); |
|
861 |
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) { |
|
862 |
if (lc.compare(bestLicense, instance.getLicence()) > 0) { |
|
863 |
bestLicense = instance.getLicence(); |
|
864 |
} |
|
865 |
} |
|
866 |
if (bestLicense != null) { |
|
867 |
return bestLicense.getClassname(); |
|
868 |
} else { |
|
869 |
return null; |
|
870 |
} |
|
871 |
} |
|
869 | 872 |
|
873 |
// here iterate over all values |
|
874 |
private String getAccessMode(ResultProtos.Result result) { |
|
875 |
String accessMode = null; |
|
876 |
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) { |
|
877 |
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) { |
|
878 |
accessMode = instance.getLicence().getClassname(); |
|
879 |
break; |
|
880 |
} |
|
870 | 881 |
|
871 |
// to datacite____:: ) |
|
872 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
873 |
value = value.replaceFirst(".*\\|", ""); |
|
874 |
value = value.replaceAll("\n", ""); |
|
875 |
value = value.replaceAll(DELIM, ""); |
|
876 |
value = value.replaceAll(ENCLOSED, ""); |
|
877 |
value = value.trim(); |
|
882 |
} |
|
878 | 883 |
|
879 |
} |
|
880 |
if (value == null) { |
|
881 |
return null; |
|
882 |
} |
|
883 |
return ENCLOSED + value + ENCLOSED; |
|
884 | 884 |
|
885 |
} |
|
885 |
return accessMode; |
|
886 |
} |
|
886 | 887 |
|
887 |
public long DATEDIFF(String startDate, String endDate) { |
|
888 |
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000; |
|
889 |
long days = 0l; |
|
890 |
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss"); |
|
891 |
// <startdate>2011-09-01</startdate> |
|
892 |
// <enddate>2015-08-31</enddate> |
|
893 |
Date dateIni = null; |
|
894 |
Date dateFin = null; |
|
895 | 888 |
|
896 |
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) { |
|
897 |
return 0; |
|
898 |
} |
|
899 |
try { |
Also available in: Unified diff
updates for lod