Revision 37693
Added by Eri Katsari almost 9 years ago
modules/dnet-openaire-stats/trunk/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/Serializer.java | ||
---|---|---|
30 | 30 |
*/ |
31 | 31 |
public class Serializer { |
32 | 32 |
|
33 |
private static String DELIM;
|
|
34 |
private Logger log = Logger.getLogger(this.getClass());
|
|
35 |
private String NULL_STRING;
|
|
36 |
private String NULL_NUM;
|
|
37 |
//TODO no longer used
|
|
38 |
private static String ENCLOSED;
|
|
33 |
private static String DELIM;
|
|
34 |
private Logger log = Logger.getLogger(this.getClass());
|
|
35 |
private String NULL_STRING;
|
|
36 |
private String NULL_NUM;
|
|
37 |
//TODO no longer used
|
|
38 |
private static String ENCLOSED;
|
|
39 | 39 |
|
40 |
private FundingParser fundingParser = new FundingParser(); |
|
41 | 40 |
|
41 |
private FundingParser fundingParser; |
|
42 | 42 |
|
43 |
public Serializer() { |
|
44 |
} |
|
43 |
public Serializer() { |
|
44 |
fundingParser = new FundingParser(); |
|
45 |
fundingParser.setDELIM(DELIM); |
|
46 |
fundingParser.setENCLOSED(ENCLOSED); |
|
47 |
fundingParser.setNULL_STRING(NULL_STRING); |
|
48 |
} |
|
45 | 49 |
|
50 |
; |
|
46 | 51 |
|
47 |
public String serialize(Oaf oaf) {
|
|
52 |
public String serialize(Oaf oaf) {
|
|
48 | 53 |
|
49 |
switch (oaf.getKind()) {
|
|
50 |
case entity:
|
|
51 |
OafEntity valueEntity = oaf.getEntity();
|
|
54 |
switch (oaf.getKind()) {
|
|
55 |
case entity:
|
|
56 |
OafEntity valueEntity = oaf.getEntity();
|
|
52 | 57 |
|
53 |
switch (valueEntity.getType()) {
|
|
54 |
case datasource:
|
|
58 |
switch (valueEntity.getType()) {
|
|
59 |
case datasource:
|
|
55 | 60 |
|
56 |
return buildDatasource(valueEntity);
|
|
61 |
return buildDatasource(valueEntity);
|
|
57 | 62 |
|
58 |
case organization:
|
|
63 |
case organization:
|
|
59 | 64 |
|
60 |
return buildOrganization(valueEntity);
|
|
65 |
return buildOrganization(valueEntity);
|
|
61 | 66 |
|
62 |
case project:
|
|
67 |
case project:
|
|
63 | 68 |
|
64 |
return buildProject(valueEntity);
|
|
65 |
case result:
|
|
69 |
return buildProject(valueEntity);
|
|
70 |
case result:
|
|
66 | 71 |
|
67 |
return buildResult(valueEntity);
|
|
72 |
return buildResult(valueEntity);
|
|
68 | 73 |
// case person: |
69 | 74 |
// |
70 | 75 |
// return buildPerson(valueEntity); |
71 |
default:
|
|
72 |
log.error("wrong type");
|
|
73 |
break;
|
|
74 |
}
|
|
75 |
break;
|
|
76 |
case relation:
|
|
77 |
OafRel valueRel = oaf.getRel();
|
|
76 |
default:
|
|
77 |
log.error("wrong type");
|
|
78 |
break;
|
|
79 |
}
|
|
80 |
break;
|
|
81 |
case relation:
|
|
82 |
OafRel valueRel = oaf.getRel();
|
|
78 | 83 |
|
79 |
return buildRel(valueRel);
|
|
84 |
return buildRel(valueRel);
|
|
80 | 85 |
|
81 |
}
|
|
86 |
}
|
|
82 | 87 |
|
83 |
return null;
|
|
88 |
return null;
|
|
84 | 89 |
|
85 |
}
|
|
90 |
}
|
|
86 | 91 |
|
87 |
public String serialize(OafRel oaf) {
|
|
92 |
public String serialize(OafRel oaf) {
|
|
88 | 93 |
|
89 |
switch (oaf.getRelType()) {
|
|
90 |
case resultProject:
|
|
94 |
switch (oaf.getRelType()) {
|
|
95 |
case resultProject:
|
|
91 | 96 |
|
92 |
return getResultProject(oaf);
|
|
97 |
return getResultProject(oaf);
|
|
93 | 98 |
|
94 |
default:
|
|
95 |
return buildRel(oaf);
|
|
96 |
}
|
|
97 |
}
|
|
99 |
default:
|
|
100 |
return buildRel(oaf);
|
|
101 |
}
|
|
102 |
}
|
|
98 | 103 |
|
99 |
private String buildRel(OafRel Rel) {
|
|
100 |
return getStringField(Rel.getTarget());
|
|
101 |
}
|
|
104 |
private String buildRel(OafRel Rel) {
|
|
105 |
return getStringField(Rel.getTarget());
|
|
106 |
}
|
|
102 | 107 |
|
103 | 108 |
|
104 |
public HashMap<String, List<String>> extractRelations(Oaf oaf) {
|
|
105 |
OafEntity valueEntity = oaf.getEntity();
|
|
106 |
switch (valueEntity.getType()) {
|
|
107 |
case datasource:
|
|
108 |
return getDatasourceLanguages(valueEntity);
|
|
109 |
case result:
|
|
110 |
HashMap<String, List<String>> relations = new HashMap<String, List<String>>();
|
|
111 |
relations.putAll(getResultLanguages(valueEntity));
|
|
112 |
//TODO no more resultTopics needed!!!
|
|
113 |
// relations.putAll(getResultTopics(valueEntity));
|
|
114 |
relations.putAll(getResultClassifications(valueEntity));
|
|
115 |
relations.putAll(getResultDatasources(valueEntity));
|
|
116 |
relations.putAll(getResultConcepts(valueEntity));
|
|
117 |
return relations;
|
|
118 |
default:
|
|
109 |
public HashMap<String, List<String>> extractRelations(Oaf oaf) {
|
|
110 |
OafEntity valueEntity = oaf.getEntity();
|
|
111 |
switch (valueEntity.getType()) {
|
|
112 |
case datasource:
|
|
113 |
return getDatasourceLanguages(valueEntity);
|
|
114 |
case result:
|
|
115 |
HashMap<String, List<String>> relations = new HashMap<String, List<String>>();
|
|
116 |
relations.putAll(getResultLanguages(valueEntity));
|
|
117 |
//TODO no more resultTopics needed!!!
|
|
118 |
// relations.putAll(getResultTopics(valueEntity));
|
|
119 |
relations.putAll(getResultClassifications(valueEntity));
|
|
120 |
relations.putAll(getResultDatasources(valueEntity));
|
|
121 |
relations.putAll(getResultConcepts(valueEntity));
|
|
122 |
return relations;
|
|
123 |
default:
|
|
119 | 124 |
|
120 |
return null;
|
|
121 |
}
|
|
125 |
return null;
|
|
126 |
}
|
|
122 | 127 |
|
123 |
}
|
|
128 |
}
|
|
124 | 129 |
|
125 |
private String getResultProject(OafRel oaf) {
|
|
126 |
String buff = new String();
|
|
127 |
String result = oaf.getTarget();
|
|
130 |
private String getResultProject(OafRel oaf) {
|
|
131 |
String buff = new String();
|
|
132 |
String result = oaf.getTarget();
|
|
128 | 133 |
|
129 | 134 |
|
130 |
buff += getStringField(result);
|
|
131 |
// TODO is declared as int!!!
|
|
132 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate());
|
|
133 |
if (diff < 0) {
|
|
134 |
diff = 0;
|
|
135 |
}
|
|
136 |
buff += getNumericField(String.valueOf(diff));
|
|
135 |
buff += getStringField(result);
|
|
136 |
// TODO is declared as int!!!
|
|
137 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate());
|
|
138 |
if (diff < 0) {
|
|
139 |
diff = 0;
|
|
140 |
}
|
|
141 |
buff += getNumericField(String.valueOf(diff));
|
|
137 | 142 |
|
138 | 143 |
|
139 |
return buff;
|
|
140 |
}
|
|
144 |
return buff;
|
|
145 |
}
|
|
141 | 146 |
|
142 | 147 |
|
143 |
private HashMap<String, List<String>> getDatasourceLanguages(OafEntity valueEntity) {
|
|
144 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
145 |
List<String> buffs = new ArrayList<String>();
|
|
148 |
private HashMap<String, List<String>> getDatasourceLanguages(OafEntity valueEntity) {
|
|
149 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
150 |
List<String> buffs = new ArrayList<String>();
|
|
146 | 151 |
|
147 | 152 |
|
148 |
Datasource d = valueEntity.getDatasource();
|
|
153 |
Datasource d = valueEntity.getDatasource();
|
|
149 | 154 |
|
150 |
Metadata metadata = d.getMetadata();
|
|
155 |
Metadata metadata = d.getMetadata();
|
|
151 | 156 |
|
152 |
for (StringField lang : metadata.getOdlanguagesList()) {
|
|
157 |
for (StringField lang : metadata.getOdlanguagesList()) {
|
|
153 | 158 |
|
154 |
buffs.add(getStringField(lang.getValue()));
|
|
155 |
}
|
|
156 |
rels.put("datasourceLanguage", buffs);
|
|
157 |
return rels;
|
|
158 |
}
|
|
159 |
buffs.add(getStringField(lang.getValue()));
|
|
160 |
}
|
|
161 |
rels.put("datasourceLanguage", buffs);
|
|
162 |
return rels;
|
|
163 |
}
|
|
159 | 164 |
|
160 |
private HashMap<String, List<String>> getResultLanguages(OafEntity valueEntity) {
|
|
161 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
162 |
List<String> buffs = new ArrayList<String>();
|
|
163 |
Result d = valueEntity.getResult();
|
|
165 |
private HashMap<String, List<String>> getResultLanguages(OafEntity valueEntity) {
|
|
166 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
167 |
List<String> buffs = new ArrayList<String>();
|
|
168 |
Result d = valueEntity.getResult();
|
|
164 | 169 |
|
165 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = d.getMetadata();
|
|
170 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = d.getMetadata();
|
|
166 | 171 |
|
167 |
if (metadata.getLanguage().getClassname() != null && !metadata.getLanguage().getClassname().isEmpty()) {
|
|
172 |
if (metadata.getLanguage().getClassname() != null && !metadata.getLanguage().getClassname().isEmpty()) {
|
|
168 | 173 |
|
169 |
buffs.add(getStringField(metadata.getLanguage().getClassname()));
|
|
170 |
}
|
|
171 |
rels.put("resultLanguage", buffs);
|
|
172 |
return rels;
|
|
174 |
buffs.add(getStringField(metadata.getLanguage().getClassname()));
|
|
175 |
}
|
|
176 |
rels.put("resultLanguage", buffs);
|
|
177 |
return rels;
|
|
173 | 178 |
|
174 |
}
|
|
179 |
}
|
|
175 | 180 |
|
176 |
private HashMap<String, List<String>> getResultClassifications(OafEntity valueEntity) {
|
|
181 |
private HashMap<String, List<String>> getResultClassifications(OafEntity valueEntity) {
|
|
177 | 182 |
|
178 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
179 |
List<String> buffs = new ArrayList<String>();
|
|
180 |
Result result = valueEntity.getResult();
|
|
183 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
184 |
List<String> buffs = new ArrayList<String>();
|
|
185 |
Result result = valueEntity.getResult();
|
|
181 | 186 |
|
182 |
for (Instance instance : (result.getInstanceList())) {
|
|
183 |
String classification = instance.getInstancetype().getClassname();
|
|
184 |
if (classification != null && !classification.isEmpty()) {
|
|
185 |
buffs.add(getStringField(instance.getInstancetype().getClassname()));
|
|
186 |
// TODO HERE KEEP ONLY ONE CLASSIFICATIONS PER RESULT
|
|
187 |
break;
|
|
188 |
}
|
|
189 |
}
|
|
190 |
rels.put("resultClassification", buffs);
|
|
191 |
return rels;
|
|
187 |
for (Instance instance : (result.getInstanceList())) {
|
|
188 |
String classification = instance.getInstancetype().getClassname();
|
|
189 |
if (classification != null && !classification.isEmpty()) {
|
|
190 |
buffs.add(getStringField(instance.getInstancetype().getClassname()));
|
|
191 |
// TODO HERE KEEP ONLY ONE CLASSIFICATIONS PER RESULT
|
|
192 |
break;
|
|
193 |
}
|
|
194 |
}
|
|
195 |
rels.put("resultClassification", buffs);
|
|
196 |
return rels;
|
|
192 | 197 |
|
193 |
}
|
|
198 |
}
|
|
194 | 199 |
|
195 |
private HashMap<String, List<String>> getResultConcepts(OafEntity valueEntity) {
|
|
196 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
197 |
List<String> buffs = new ArrayList<String>();
|
|
200 |
private HashMap<String, List<String>> getResultConcepts(OafEntity valueEntity) {
|
|
201 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
202 |
List<String> buffs = new ArrayList<String>();
|
|
198 | 203 |
|
199 |
Result result = valueEntity.getResult();
|
|
204 |
Result result = valueEntity.getResult();
|
|
200 | 205 |
|
201 |
for (eu.dnetlib.data.proto.ResultProtos.Result.Context context : result.getMetadata().getContextList()) {
|
|
206 |
for (eu.dnetlib.data.proto.ResultProtos.Result.Context context : result.getMetadata().getContextList()) {
|
|
202 | 207 |
|
203 |
buffs.add(getStringField(context.getId()));
|
|
204 |
}
|
|
205 |
rels.put("resultConcept", buffs);
|
|
206 |
return rels;
|
|
208 |
buffs.add(getStringField(context.getId()));
|
|
209 |
}
|
|
210 |
rels.put("resultConcept", buffs);
|
|
211 |
return rels;
|
|
207 | 212 |
|
208 |
}
|
|
213 |
}
|
|
209 | 214 |
|
210 |
private HashMap<String, List<String>> getResultDatasources(OafEntity valueEntity) {
|
|
215 |
private HashMap<String, List<String>> getResultDatasources(OafEntity valueEntity) {
|
|
211 | 216 |
|
212 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
213 |
List<String> buffs = new ArrayList<String>();
|
|
214 |
Result result = valueEntity.getResult();
|
|
217 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
218 |
List<String> buffs = new ArrayList<String>();
|
|
219 |
Result result = valueEntity.getResult();
|
|
215 | 220 |
|
216 | 221 |
//TODO hosted by |
217 |
for (Instance instance : (result.getInstanceList())) {
|
|
222 |
for (Instance instance : (result.getInstanceList())) {
|
|
218 | 223 |
|
219 | 224 |
|
220 |
String hostedBy = instance.getHostedby().getKey();
|
|
221 |
if (hostedBy != null && !hostedBy.isEmpty()) {
|
|
222 |
buffs.add((getStringField(hostedBy)));
|
|
223 |
}
|
|
224 |
}
|
|
225 |
String hostedBy = instance.getHostedby().getKey();
|
|
226 |
if (hostedBy != null && !hostedBy.isEmpty()) {
|
|
227 |
buffs.add((getStringField(hostedBy)));
|
|
228 |
}
|
|
229 |
}
|
|
225 | 230 |
|
226 | 231 |
//TODO collected froms |
227 |
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
|
232 |
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
|
228 | 233 |
|
229 |
String collectedFrom = collectedFromValue.getKey();
|
|
230 |
if (collectedFrom != null && !collectedFrom.isEmpty()) buffs.add((getStringField(collectedFrom)));
|
|
234 |
String collectedFrom = collectedFromValue.getKey();
|
|
235 |
if (collectedFrom != null && !collectedFrom.isEmpty()) buffs.add((getStringField(collectedFrom)));
|
|
231 | 236 |
|
232 |
}
|
|
233 |
rels.put("resultDatasource", buffs);
|
|
234 |
return rels;
|
|
237 |
}
|
|
238 |
rels.put("resultDatasource", buffs);
|
|
239 |
return rels;
|
|
235 | 240 |
|
236 |
}
|
|
241 |
}
|
|
237 | 242 |
|
238 |
public static boolean isNumeric(String str) {
|
|
243 |
public static boolean isNumeric(String str) {
|
|
239 | 244 |
|
240 |
str = str.replaceAll("[^A-Za-z0-9 ]", "");
|
|
241 |
str = str.replaceAll(" ", "");
|
|
242 |
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional
|
|
243 |
// '-' and decimal.
|
|
244 |
}
|
|
245 |
str = str.replaceAll("[^A-Za-z0-9 ]", "");
|
|
246 |
str = str.replaceAll(" ", "");
|
|
247 |
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional
|
|
248 |
// '-' and decimal.
|
|
249 |
}
|
|
245 | 250 |
|
246 |
// TODO there are topics with "null" as value -> replace them
|
|
247 |
private boolean isValidTopic(String t) {
|
|
251 |
// TODO there are topics with "null" as value -> replace them
|
|
252 |
private boolean isValidTopic(String t) {
|
|
248 | 253 |
|
249 |
if (t == null || t.isEmpty()) {
|
|
250 |
return false;
|
|
251 |
}
|
|
254 |
if (t == null || t.isEmpty()) {
|
|
255 |
return false;
|
|
256 |
}
|
|
252 | 257 |
|
253 |
if (t.equals("") || t.equals(" ")) {
|
|
254 |
return false;
|
|
255 |
}
|
|
256 |
if (t.equals("null") || t.equals("Null") || t.equals("NULL")) {
|
|
257 |
return false;
|
|
258 |
}
|
|
258 |
if (t.equals("") || t.equals(" ")) {
|
|
259 |
return false;
|
|
260 |
}
|
|
261 |
if (t.equals("null") || t.equals("Null") || t.equals("NULL")) {
|
|
262 |
return false;
|
|
263 |
}
|
|
259 | 264 |
|
260 |
if (t.equals(ENCLOSED + ENCLOSED + DELIM) || t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) {
|
|
261 |
return false;
|
|
262 |
}
|
|
263 |
// skip dedups
|
|
264 |
if (t.contains("ddc:")) {
|
|
265 |
return false;
|
|
266 |
}
|
|
267 |
return true;
|
|
268 |
}
|
|
265 |
if (t.equals(ENCLOSED + ENCLOSED + DELIM) || t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) {
|
|
266 |
return false;
|
|
267 |
}
|
|
268 |
// skip dedups
|
|
269 |
if (t.contains("ddc:")) {
|
|
270 |
return false;
|
|
271 |
}
|
|
272 |
return true;
|
|
273 |
}
|
|
269 | 274 |
|
270 |
private HashMap<String, List<String>> getResultTopics(OafEntity valueEntity) {
|
|
271 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
272 |
List<String> buffs = new ArrayList<String>();
|
|
273 |
Result d = valueEntity.getResult();
|
|
275 |
private HashMap<String, List<String>> getResultTopics(OafEntity valueEntity) {
|
|
276 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>();
|
|
277 |
List<String> buffs = new ArrayList<String>();
|
|
278 |
Result d = valueEntity.getResult();
|
|
274 | 279 |
|
275 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = d.getMetadata();
|
|
280 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = d.getMetadata();
|
|
276 | 281 |
|
277 |
List<StructuredProperty> Topics = metadata.getSubjectList();
|
|
278 |
String buff = new String();
|
|
279 |
for (StructuredProperty topic : Topics) {
|
|
280 |
// TODOs
|
|
281 |
if (isValidTopic(topic.getValue())) {
|
|
282 |
if (!isNumeric(topic.getValue())) {
|
|
283 |
String t = getStringField(topic.getValue());
|
|
284 |
if (isValidTopic(t)) {
|
|
285 |
buff += t + " ";
|
|
282 |
List<StructuredProperty> Topics = metadata.getSubjectList();
|
|
283 |
String buff = new String();
|
|
284 |
for (StructuredProperty topic : Topics) {
|
|
285 |
// TODOs
|
|
286 |
if (isValidTopic(topic.getValue())) {
|
|
287 |
if (!isNumeric(topic.getValue())) {
|
|
288 |
String t = getStringField(topic.getValue());
|
|
289 |
if (isValidTopic(t)) {
|
|
290 |
buff += t + " ";
|
|
286 | 291 |
|
287 |
}
|
|
292 |
}
|
|
288 | 293 |
|
289 |
}
|
|
290 |
}
|
|
294 |
}
|
|
295 |
}
|
|
291 | 296 |
|
292 |
}
|
|
293 |
if (!buff.isEmpty()) {
|
|
294 |
buff = getStringField(buff);
|
|
295 |
buffs.add(buff);
|
|
296 |
}
|
|
297 |
rels.put("resultTopic", buffs);
|
|
297 |
}
|
|
298 |
if (!buff.isEmpty()) {
|
|
299 |
buff = getStringField(buff);
|
|
300 |
buffs.add(buff);
|
|
301 |
}
|
|
302 |
rels.put("resultTopic", buffs);
|
|
298 | 303 |
|
299 |
return rels;
|
|
304 |
return rels;
|
|
300 | 305 |
|
301 |
}
|
|
306 |
}
|
|
302 | 307 |
|
303 |
private String buildDatasource(OafEntity data) {
|
|
308 |
private String buildDatasource(OafEntity data) {
|
|
304 | 309 |
|
305 |
String buff = new String();
|
|
310 |
String buff = new String();
|
|
306 | 311 |
|
307 |
Datasource d = data.getDatasource();
|
|
312 |
Datasource d = data.getDatasource();
|
|
308 | 313 |
|
309 |
Metadata metadata = d.getMetadata();
|
|
310 |
String full_id = getStringField(data.getId());
|
|
314 |
Metadata metadata = d.getMetadata();
|
|
315 |
String full_id = getStringField(data.getId());
|
|
311 | 316 |
|
312 |
buff += full_id;
|
|
313 |
buff += full_id;
|
|
314 |
buff += full_id;
|
|
315 |
buff += full_id;
|
|
317 |
buff += full_id;
|
|
318 |
buff += full_id;
|
|
319 |
buff += full_id;
|
|
320 |
buff += full_id;
|
|
316 | 321 |
|
317 | 322 |
|
318 |
// name
|
|
319 |
if (metadata.getOfficialname().getValue().equalsIgnoreCase("unknown")) {
|
|
320 |
buff += getStringField("Unknown Repository");
|
|
321 |
} else {
|
|
322 |
buff += getStringField(metadata.getOfficialname().getValue());
|
|
323 |
}
|
|
324 |
// type
|
|
323 |
// name
|
|
324 |
if (metadata.getOfficialname().getValue().equalsIgnoreCase("unknown")) {
|
|
325 |
buff += getStringField("Unknown Repository");
|
|
326 |
} else {
|
|
327 |
buff += getStringField(metadata.getOfficialname().getValue());
|
|
328 |
}
|
|
329 |
// type
|
|
325 | 330 |
|
326 |
if (metadata.hasDatasourcetype())
|
|
331 |
if (metadata.hasDatasourcetype())
|
|
327 | 332 |
|
328 |
{
|
|
329 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""));
|
|
333 |
{
|
|
334 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""));
|
|
330 | 335 |
|
331 |
} else {
|
|
332 |
buff += getStringField(null);
|
|
333 |
}
|
|
336 |
} else {
|
|
337 |
buff += getStringField(null);
|
|
338 |
}
|
|
334 | 339 |
|
335 |
// compatibility,
|
|
336 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname());
|
|
340 |
// compatibility,
|
|
341 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname());
|
|
337 | 342 |
|
338 |
// latitude
|
|
339 |
buff += getStringField(metadata.getLatitude().getValue());
|
|
343 |
// latitude
|
|
344 |
buff += getStringField(metadata.getLatitude().getValue());
|
|
340 | 345 |
|
341 |
// longtitude
|
|
342 |
buff += getStringField(metadata.getLongitude().getValue());
|
|
346 |
// longtitude
|
|
347 |
buff += getStringField(metadata.getLongitude().getValue());
|
|
343 | 348 |
|
344 |
// dateofvalidation,
|
|
345 |
buff += getStringField(metadata.getDateofvalidation().getValue());
|
|
349 |
// dateofvalidation,
|
|
350 |
buff += getStringField(metadata.getDateofvalidation().getValue());
|
|
346 | 351 |
|
347 |
// yearofvalidation,
|
|
352 |
// yearofvalidation,
|
|
348 | 353 |
|
349 |
// parse year of validation
|
|
350 |
buff += getYearInt(metadata.getDateofvalidation().getValue());
|
|
354 |
// parse year of validation
|
|
355 |
buff += getYearInt(metadata.getDateofvalidation().getValue());
|
|
351 | 356 |
|
352 |
// number??
|
|
357 |
// number??
|
|
353 | 358 |
|
354 |
buff += getStringField("1");
|
|
359 |
buff += getStringField("1");
|
|
355 | 360 |
|
356 |
return buff;
|
|
357 |
}
|
|
361 |
return buff;
|
|
362 |
}
|
|
358 | 363 |
|
359 |
private String buildOrganization(OafEntity data) {
|
|
364 |
private String buildOrganization(OafEntity data) {
|
|
360 | 365 |
|
361 |
String buff = new String();
|
|
366 |
String buff = new String();
|
|
362 | 367 |
|
363 |
Organization organization = data.getOrganization();
|
|
364 |
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata();
|
|
368 |
Organization organization = data.getOrganization();
|
|
369 |
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata();
|
|
365 | 370 |
|
366 |
// `organization_datasources`,
|
|
367 |
String full_id = getStringField(data.getId());
|
|
368 |
buff += full_id;
|
|
369 |
// organization_projects
|
|
370 |
buff += full_id;
|
|
371 |
// `name`,
|
|
372 |
buff += getStringField(metadata.getLegalname().getValue());
|
|
373 |
// `country`,
|
|
371 |
// `organization_datasources`,
|
|
372 |
String full_id = getStringField(data.getId());
|
|
373 |
buff += full_id;
|
|
374 |
// organization_projects
|
|
375 |
buff += full_id;
|
|
376 |
// `name`,
|
|
377 |
buff += getStringField(metadata.getLegalname().getValue());
|
|
378 |
// `country`,
|
|
374 | 379 |
|
375 | 380 |
|
376 |
buff += getStringField(metadata.getCountry().getClassname());
|
|
381 |
buff += getStringField(metadata.getCountry().getClassname());
|
|
377 | 382 |
|
378 | 383 |
|
379 |
// `number`,
|
|
384 |
// `number`,
|
|
380 | 385 |
|
381 |
buff += getStringField("1");
|
|
382 |
return buff;
|
|
386 |
buff += getStringField("1");
|
|
387 |
return buff;
|
|
383 | 388 |
|
384 |
}
|
|
389 |
}
|
|
385 | 390 |
|
386 |
private String buildResult(OafEntity data) {
|
|
391 |
private String buildResult(OafEntity data) {
|
|
387 | 392 |
|
388 |
String buff = new String();
|
|
393 |
String buff = new String();
|
|
389 | 394 |
|
390 |
Result result = data.getResult();
|
|
391 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata();
|
|
395 |
Result result = data.getResult();
|
|
396 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata();
|
|
392 | 397 |
|
393 |
// result_topics/
|
|
394 |
String full_id = getStringField(data.getId());
|
|
398 |
// result_topics/
|
|
399 |
String full_id = getStringField(data.getId());
|
|
395 | 400 |
|
396 |
buff += full_id;
|
|
401 |
buff += full_id;
|
|
397 | 402 |
|
398 |
// result_languages
|
|
399 |
buff += full_id;
|
|
403 |
// result_languages
|
|
404 |
buff += full_id;
|
|
400 | 405 |
|
401 |
// `result_projects`,
|
|
402 |
buff += full_id;
|
|
406 |
// `result_projects`,
|
|
407 |
buff += full_id;
|
|
403 | 408 |
|
404 |
// `result_datasources`,
|
|
405 |
buff += full_id;
|
|
409 |
// `result_datasources`,
|
|
410 |
buff += full_id;
|
|
406 | 411 |
|
407 |
// `result_classifications`,
|
|
408 |
buff += full_id;
|
|
412 |
// `result_classifications`,
|
|
413 |
buff += full_id;
|
|
409 | 414 |
|
410 |
// / `result_infrastructures`,
|
|
411 |
buff += full_id;
|
|
415 |
// / `result_infrastructures`,
|
|
416 |
buff += full_id;
|
|
412 | 417 |
|
413 |
// `result_claims`,
|
|
414 |
buff += full_id;
|
|
418 |
// `result_claims`,
|
|
419 |
buff += full_id;
|
|
415 | 420 |
|
416 |
// `result_results`,
|
|
417 |
buff += full_id;
|
|
421 |
// `result_results`,
|
|
422 |
buff += full_id;
|
|
418 | 423 |
|
419 |
//TODO pubtitle
|
|
424 |
//TODO pubtitle
|
|
420 | 425 |
|
421 |
String titleString = new String();
|
|
422 |
String alternativeTitles = new String();
|
|
426 |
String titleString = new String();
|
|
427 |
String alternativeTitles = new String();
|
|
423 | 428 |
|
424 |
// for (StructuredProperty title : metadata.getTitleList()) {
|
|
425 |
// titleString = title.getValue();
|
|
426 |
// break;
|
|
427 |
//}
|
|
429 |
// for (StructuredProperty title : metadata.getTitleList()) {
|
|
430 |
// titleString = title.getValue();
|
|
431 |
// break;
|
|
432 |
//}
|
|
428 | 433 |
|
429 |
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
|
430 |
StructuredProperty title = metadata.getTitleList().get(i);
|
|
431 |
if (i == 0) {
|
|
432 |
titleString = title.getValue().replaceAll("\\s+", " ");
|
|
433 |
titleString = titleString.replaceAll("\n", " ");
|
|
434 |
} else {
|
|
435 |
alternativeTitles += title.getValue().replaceAll("\\s+", " ") + " ; ";
|
|
436 |
alternativeTitles = alternativeTitles.replaceAll("\n", " ");
|
|
437 |
}
|
|
438 |
break;
|
|
439 |
}
|
|
440 |
//TODO pubtitle
|
|
441 |
buff += getStringField(titleString); //!#Osvaldo Raineri: Gli Atti etiopici del martire egiziano Giorgio il nuovo (β 978)#!#
|
|
442 |
//TODO alternative titles
|
|
443 |
//buff += getStringField(alternativeTitles); // null#!
|
|
434 |
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
|
435 |
StructuredProperty title = metadata.getTitleList().get(i);
|
|
436 |
if (i == 0) {
|
|
437 |
titleString = title.getValue().replaceAll("\\s+", " ");
|
|
438 |
titleString = titleString.replaceAll("\n", " ");
|
|
439 |
} else {
|
|
440 |
alternativeTitles += title.getValue().replaceAll("\\s+", " ") + " ; ";
|
|
441 |
alternativeTitles = alternativeTitles.replaceAll("\n", " ");
|
|
442 |
}
|
|
443 |
break;
|
|
444 |
}
|
|
445 |
//TODO pubtitle
|
|
446 |
buff += getStringField(titleString); //!#Osvaldo Raineri: Gli Atti etiopici del martire egiziano Giorgio il nuovo (β 978)#!#
|
|
447 |
//TODO alternative titles
|
|
448 |
//buff += getStringField(alternativeTitles); // null#!
|
|
444 | 449 |
|
445 |
// TODO format
|
|
446 |
String formatString = NULL_STRING;
|
|
447 |
//for (StringField format : metadata.getFormatList()) {
|
|
448 |
// formatString = format.getValue();
|
|
449 |
// break;
|
|
450 |
//}
|
|
450 |
// TODO format
|
|
451 |
String formatString = NULL_STRING;
|
|
452 |
//for (StringField format : metadata.getFormatList()) {
|
|
453 |
// formatString = format.getValue();
|
|
454 |
// break;
|
|
455 |
//}
|
|
451 | 456 |
|
452 |
buff += getStringField(formatString);
|
|
453 |
//TODO publisher
|
|
457 |
buff += getStringField(formatString);
|
|
458 |
//TODO publisher
|
|
454 | 459 |
|
455 |
buff += getStringField(metadata.getPublisher().getValue());
|
|
456 |
//TODO journal
|
|
460 |
buff += getStringField(metadata.getPublisher().getValue());
|
|
461 |
//TODO journal
|
|
457 | 462 |
|
458 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#!
|
|
463 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#!
|
|
459 | 464 |
|
460 |
// year
|
|
461 |
buff += getYearInt(metadata.getDateofacceptance().getValue());
|
|
465 |
// year
|
|
466 |
buff += getYearInt(metadata.getDateofacceptance().getValue());
|
|
462 | 467 |
|
463 |
// date CHANGED THIS TO DATE FORMAT
|
|
464 |
buff += getStringDateField(metadata.getDateofacceptance().getValue());
|
|
468 |
// date CHANGED THIS TO DATE FORMAT
|
|
469 |
buff += getStringDateField(metadata.getDateofacceptance().getValue());
|
|
465 | 470 |
|
466 |
// access_mode,
|
|
467 |
buff += getStringField(getAccessMode(result));
|
|
471 |
// access_mode,
|
|
472 |
buff += getStringField(getAccessMode(result));
|
|
468 | 473 |
|
469 |
// bestlicense
|
|
474 |
// bestlicense
|
|
470 | 475 |
|
471 |
buff += getStringField(getBestLicense(result));
|
|
472 |
// type
|
|
473 |
buff += getStringField(metadata.getResulttype().getClassname());
|
|
474 |
// embargo_end_date
|
|
475 |
buff += getStringField(metadata.getEmbargoenddate().getValue());
|
|
476 |
buff += getStringField(getBestLicense(result));
|
|
477 |
// type
|
|
478 |
buff += getStringField(metadata.getResulttype().getClassname());
|
|
479 |
// embargo_end_date
|
|
480 |
buff += getStringField(metadata.getEmbargoenddate().getValue());
|
|
476 | 481 |
|
477 |
// `authors`,
|
|
478 |
int authors = 0;
|
|
479 |
String delayed = "no";
|
|
482 |
// `authors`,
|
|
483 |
int authors = 0;
|
|
484 |
String delayed = "no";
|
|
480 | 485 |
|
481 |
for (OafRel rel : data.getCachedRelList()) {
|
|
486 |
for (OafRel rel : data.getCachedRelList()) {
|
|
482 | 487 |
|
483 |
if (rel.getRelType().equals(RelType.personResult)) {
|
|
488 |
if (rel.getRelType().equals(RelType.personResult)) {
|
|
484 | 489 |
|
485 |
authors++;
|
|
486 |
} else if (rel.getRelType().equals(RelType.resultProject))
|
|
487 |
// TODO remember : in result Project, first id is project, second is
|
|
488 |
// result.
|
|
489 |
{
|
|
490 |
authors++;
|
|
491 |
} else if (rel.getRelType().equals(RelType.resultProject))
|
|
492 |
// TODO remember : in result Project, first id is project, second is
|
|
493 |
// result.
|
|
494 |
{
|
|
490 | 495 |
|
491 |
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), rel.getResultProject().getOutcome().getRelMetadata().getStartdate());
|
|
492 |
if (Integer.parseInt(daysfromend) > 0) {
|
|
493 |
delayed = "yes";
|
|
494 |
}
|
|
495 |
}
|
|
496 |
}
|
|
497 |
// `delayed`,
|
|
498 |
buff += getStringField(delayed);
|
|
496 |
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), rel.getResultProject().getOutcome().getRelMetadata().getStartdate());
|
|
497 |
if (Integer.parseInt(daysfromend) > 0) {
|
|
498 |
delayed = "yes";
|
|
499 |
}
|
|
500 |
}
|
|
501 |
}
|
|
502 |
// `delayed`,
|
|
503 |
buff += getStringField(delayed);
|
|
499 | 504 |
|
500 |
// log.info("Result " + full_id +"Author count : " + authors );
|
|
501 |
buff += getNumericField(String.valueOf(authors));
|
|
505 |
// log.info("Result " + full_id +"Author count : " + authors );
|
|
506 |
buff += getNumericField(String.valueOf(authors));
|
|
502 | 507 |
|
503 |
// number??
|
|
508 |
// number??
|
|
504 | 509 |
|
505 |
buff += getStringField("1");
|
|
510 |
buff += getStringField("1");
|
|
506 | 511 |
|
507 |
if (isValid(buff, full_id)) {
|
|
508 |
return buff;
|
|
509 |
} else {
|
|
510 |
return null;
|
|
511 |
}
|
|
512 |
if (isValid(buff, full_id)) {
|
|
513 |
return buff;
|
|
514 |
} else {
|
|
515 |
return null;
|
|
516 |
}
|
|
512 | 517 |
|
513 |
}
|
|
518 |
}
|
|
514 | 519 |
|
515 |
//TODO here see if check is ok
|
|
516 |
private boolean isValid(String buff, String id) {
|
|
517 |
if (buff.endsWith(ENCLOSED)) {
|
|
518 |
log.error("Empty Result with " + id + " with body: \n" + buff);
|
|
519 |
return false;
|
|
520 |
}
|
|
521 |
return true;
|
|
522 |
}
|
|
520 |
//TODO here see if check is ok
|
|
521 |
private boolean isValid(String buff, String id) {
|
|
522 |
if (buff.endsWith(ENCLOSED)) {
|
|
523 |
log.error("Empty Result with " + id + " with body: \n" + buff);
|
|
524 |
return false;
|
|
525 |
}
|
|
526 |
return true;
|
|
527 |
}
|
|
523 | 528 |
|
524 |
private String getBestLicense(Result result) {
|
|
525 |
Qualifier bestLicense = null;
|
|
526 |
LicenseComparator lc = new LicenseComparator();
|
|
527 |
for (Instance instance : (result.getInstanceList())) {
|
|
528 |
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
|
529 |
bestLicense = instance.getLicence();
|
|
530 |
}
|
|
531 |
}
|
|
532 |
if (bestLicense != null) {
|
|
533 |
return bestLicense.getClassname();
|
|
534 |
} else {
|
|
535 |
return null;
|
|
536 |
}
|
|
537 |
}
|
|
529 |
private String getBestLicense(Result result) {
|
|
530 |
Qualifier bestLicense = null;
|
|
531 |
LicenseComparator lc = new LicenseComparator();
|
|
532 |
for (Instance instance : (result.getInstanceList())) {
|
|
533 |
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
|
534 |
bestLicense = instance.getLicence();
|
|
535 |
}
|
|
536 |
}
|
|
537 |
if (bestLicense != null) {
|
|
538 |
return bestLicense.getClassname();
|
|
539 |
} else {
|
|
540 |
return null;
|
|
541 |
}
|
|
542 |
}
|
|
538 | 543 |
|
539 |
// TODO here iterate over all values
|
|
540 |
private String getAccessMode(Result result) {
|
|
541 |
String accessMode = NULL_STRING;
|
|
542 |
for (Instance instance : (result.getInstanceList())) {
|
|
543 |
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) {
|
|
544 |
accessMode = instance.getLicence().getClassname();
|
|
545 |
break;
|
|
546 |
}
|
|
544 |
// TODO here iterate over all values
|
|
545 |
private String getAccessMode(Result result) {
|
|
546 |
String accessMode = NULL_STRING;
|
|
547 |
for (Instance instance : (result.getInstanceList())) {
|
|
548 |
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) {
|
|
549 |
accessMode = instance.getLicence().getClassname();
|
|
550 |
break;
|
|
551 |
}
|
|
547 | 552 |
|
548 |
}
|
|
553 |
}
|
|
549 | 554 |
|
550 |
return accessMode;
|
|
551 |
}
|
|
555 |
return accessMode;
|
|
556 |
}
|
|
552 | 557 |
|
553 |
private String buildProject(OafEntity data) {
|
|
558 |
private String buildProject(OafEntity data) {
|
|
554 | 559 |
|
555 |
String buff = new String();
|
|
560 |
String buff = new String();
|
|
556 | 561 |
|
557 |
Project project = data.getProject(); |
|
558 |
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata(); |
|
559 |
// project_organizations |
|
560 |
String full_id = getStringField(data.getId()); |
|
561 |
buff += full_id; |
|
562 |
// project_results |
|
563 |
buff += full_id; |
|
562 |
Project project = data.getProject(); |
|
563 |
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata(); |
|
564 |
// project_organizations |
|
564 | 565 |
|
565 |
String acronym = metadata.getAcronym().getValue(); |
|
566 |
if (acronym.equalsIgnoreCase("UNKNOWN")) { |
|
567 |
acronym = metadata.getTitle().getValue(); |
|
568 |
} |
|
566 |
String full_id = getStringField(data.getId()); |
|
567 |
buff += full_id; |
|
569 | 568 |
|
570 |
// `acronym`, |
|
571 |
buff += getStringField(acronym); |
|
569 |
// project_results |
|
570 |
buff += full_id; |
|
571 |
// `acronym`, |
|
572 |
String acronym = metadata.getAcronym().getValue(); |
|
573 |
if (acronym.equalsIgnoreCase("UNKNOWN")) { |
|
574 |
acronym = metadata.getTitle().getValue(); |
|
575 |
} |
|
572 | 576 |
|
573 |
//title! |
|
574 |
String title = getStringField(metadata.getTitle().getValue()); |
|
575 |
buff += getStringField(title); |
|
577 |
buff += getStringField(acronym); |
|
576 | 578 |
|
579 |
//title! |
|
580 |
String title = getStringField(metadata.getTitle().getValue()); |
|
581 |
buff += getStringField(title); |
|
577 | 582 |
|
578 |
List<StringField> fundList = metadata.getFundingtreeList();
|
|
583 |
List<StringField> fundList = metadata.getFundingtreeList();
|
|
579 | 584 |
|
580 | 585 |
|
581 |
if (!fundList.isEmpty()) // `funding_lvl0`, |
|
582 |
{ |
|
583 |
//TODO funder + 3 funding levels |
|
584 |
/* funder text, |
|
585 |
funding_lvl0 text, |
|
586 |
funding_lvl1 text, |
|
587 |
funding_lvl2 text, |
|
588 |
funding_lvl3 text,*/ |
|
589 |
buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue()); |
|
586 |
if (!fundList.isEmpty()) // `funding_lvl0`, |
|
587 |
{ |
|
588 |
//TODO funder + 3 funding levels |
|
590 | 589 |
|
591 |
} else { |
|
590 |
for (int i = 0; i < fundList.size(); i++) { |
|
591 |
buff += fundingParser.getFundingInfo(fundList.get(i).getValue()); |
|
592 |
} |
|
592 | 593 |
|
593 |
buff += this.fundingParser.getFundingInfo(""); |
|
594 |
} |
|
594 |
//buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue()); |
|
595 | 595 |
|
596 |
} |
|
596 | 597 |
|
597 |
String sc39 = metadata.getEcsc39().getValue().toString(); |
|
598 |
if (sc39.equalsIgnoreCase("true") || sc39.equalsIgnoreCase("t") || sc39.contains("yes")) { |
|
599 |
sc39 = "yes"; |
|
600 |
} else if (sc39.equalsIgnoreCase("false") || sc39.equalsIgnoreCase("f") || sc39.contains("no")) { |
|
601 |
sc39 = "no"; |
|
602 |
} |
|
603 | 598 |
|
604 |
// `sc39`, |
|
605 |
buff += getStringField(sc39); |
|
599 |
// if (funding_tree_2.equals(NULL_STRING)&&fundList.size() > 1) { |
|
600 |
// funding_tree_2 = (fundingParser.getFundingLevel(fundList.get(1).getValue(), 2)); |
|
601 |
// funding_tree_3 = (fundingParser.getFundingLevel(fundList.get(1).getValue(), 3)); |
|
602 |
//} |
|
606 | 603 |
|
607 |
// `url`, |
|
608 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
604 |
// `sc39`, |
|
609 | 605 |
|
610 |
// start_year |
|
606 |
String sc39 = metadata.getEcsc39().getValue().toString(); |
|
607 |
if (sc39.equalsIgnoreCase("true") || sc39.equalsIgnoreCase("t") || sc39.contains("yes")) { |
|
608 |
sc39 = "yes"; |
|
609 |
} else if (sc39.equalsIgnoreCase("false") || sc39.equalsIgnoreCase("f") || sc39.contains("no")) { |
|
610 |
sc39 = "no"; |
|
611 |
} |
|
611 | 612 |
|
612 |
buff += getYearInt(metadata.getStartdate().getValue());
|
|
613 |
buff += getStringField(sc39);
|
|
613 | 614 |
|
614 |
// end_year |
|
615 |
buff += getYearInt(metadata.getEnddate().getValue()); |
|
615 |
// `url`, |
|
616 | 616 |
|
617 |
// duration enddate-startdate
|
|
617 |
buff += getStringField(metadata.getWebsiteurl().getValue());
|
|
618 | 618 |
|
619 |
buff += getYearDifferenceInteger(metadata.getEnddate().getValue(), metadata.getStartdate().getValue());
|
|
619 |
// start_year
|
|
620 | 620 |
|
621 |
// haspubs |
|
622 |
buff += getStringField("no"); |
|
621 |
buff += getYearInt(metadata.getStartdate().getValue()); |
|
623 | 622 |
|
624 |
// numpubs |
|
625 |
buff += getNumericField("0"); |
|
626 |
// enddate |
|
627 |
buff += getNumericField(metadata.getEnddate().getValue()); |
|
628 |
// startdate |
|
629 |
buff += getNumericField(metadata.getStartdate().getValue()); |
|
623 |
// end_year |
|
624 |
buff += getYearInt(metadata.getEnddate().getValue()); |
|
630 | 625 |
|
631 |
// `daysforlastpub`, |
|
632 |
buff += getNumericField(""); |
|
633 |
// `delayedpubs`, |
|
634 |
buff += getNumericField(""); |
|
635 |
// `number` |
|
636 |
buff += getStringField("1"); |
|
637 |
return buff; |
|
626 |
// duration enddate-startdate |
|
638 | 627 |
|
639 |
}
|
|
628 |
buff += getYearDifferenceInteger(metadata.getEnddate().getValue(), metadata.getStartdate().getValue());
|
|
640 | 629 |
|
641 |
//TODO remove for production - not yet ready
|
|
642 |
private String buildPerson(OafEntity data) {
|
|
630 |
// haspubs
|
|
631 |
buff += getStringField("no");
|
|
643 | 632 |
|
644 |
String buff = new String(); |
|
633 |
// numpubs |
|
634 |
buff += getNumericField("0"); |
|
635 |
// enddate |
|
636 |
buff += getNumericField(metadata.getEnddate().getValue()); |
|
637 |
// startdate |
|
638 |
buff += getNumericField(metadata.getStartdate().getValue()); |
|
645 | 639 |
|
646 |
PersonProtos.Person person = data.getPerson(); |
|
647 |
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata(); |
|
640 |
// `daysforlastpub`, |
|
641 |
buff += getNumericField(""); |
|
642 |
// `delayedpubs`, |
|
643 |
buff += getNumericField(""); |
|
644 |
// `number` |
|
645 |
buff += getStringField("1"); |
|
646 |
return buff; |
|
648 | 647 |
|
649 |
person.getCoauthorsCount();
|
|
648 |
}
|
|
650 | 649 |
|
651 |
// `person_id`, |
|
652 |
String full_id = getStringField(data.getId()); |
|
653 |
buff += full_id; |
|
654 |
// person_result |
|
655 |
buff += full_id; |
|
650 |
//TODO remove for production - not yet ready |
|
651 |
private String buildPerson(OafEntity data) { |
|
656 | 652 |
|
657 |
// `fullname`, |
|
658 |
buff += metadata.getFullname(); |
|
653 |
String buff = new String(); |
|
659 | 654 |
|
660 |
// `Nationality`, |
|
661 |
buff += metadata.getNationality(); |
|
662 |
// `Email`, |
|
663 |
buff += metadata.getEmail(); |
|
664 |
// `Phone`, |
|
665 |
buff += metadata.getPhone(); |
|
666 |
// `CoauthorsCount` |
|
667 |
buff += person.getCoauthorsCount(); |
|
668 |
// `number`, |
|
669 |
buff += getStringField("1"); |
|
670 |
return buff; |
|
655 |
PersonProtos.Person person = data.getPerson(); |
|
656 |
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata(); |
|
671 | 657 |
|
672 |
}
|
|
658 |
person.getCoauthorsCount();
|
|
673 | 659 |
|
674 |
private String getFundingLevel(String funding_level, int level) { |
|
660 |
// `person_id`, |
|
661 |
String full_id = getStringField(data.getId()); |
|
662 |
buff += full_id; |
|
663 |
// person_result |
|
664 |
buff += full_id; |
|
675 | 665 |
|
676 |
if (funding_level.isEmpty()) {
|
|
677 |
return NULL_STRING;
|
|
666 |
// `fullname`,
|
|
667 |
buff += metadata.getFullname();
|
|
678 | 668 |
|
679 |
} |
|
669 |
// `Nationality`, |
|
670 |
buff += metadata.getNationality(); |
|
671 |
// `Email`, |
|
672 |
buff += metadata.getEmail(); |
|
673 |
// `Phone`, |
|
674 |
buff += metadata.getPhone(); |
|
675 |
// `CoauthorsCount` |
|
676 |
buff += person.getCoauthorsCount(); |
|
677 |
// `number`, |
|
678 |
buff += getStringField("1"); |
|
679 |
return buff; |
|
680 | 680 |
|
681 |
if (!funding_level.contains("funding_level_" + level)) { |
|
682 |
return NULL_STRING; |
|
683 |
} |
|
684 |
String[] split = funding_level.split("funding_level_" + level); |
|
681 |
} |
|
685 | 682 |
|
686 |
funding_level = split[1]; |
|
687 |
split = funding_level.split("name"); |
|
688 |
split = split[1].split(","); |
|
683 |
private String getFundingLevel(String funding_level, int level) { |
|
689 | 684 |
|
690 |
funding_level = split[0].replaceAll(".*:\"", ""); |
|
691 |
funding_level = funding_level.replaceFirst(ENCLOSED, ""); |
|
692 |
funding_level = funding_level.trim(); |
|
685 |
if (funding_level.isEmpty()) { |
|
686 |
return NULL_STRING; |
|
693 | 687 |
|
694 |
return funding_level; |
|
695 |
} |
|
688 |
} |
|
696 | 689 |
|
697 |
private String getYearDifferenceInteger(String enddate, String startdate) { |
|
690 |
if (!funding_level.contains("funding_level_" + level)) { |
|
691 |
return NULL_STRING; |
|
692 |
} |
|
693 |
String[] split = funding_level.split("funding_level_" + level); |
|
698 | 694 |
|
699 |
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) { |
|
695 |
funding_level = split[1]; |
|
696 |
split = funding_level.split("name"); |
|
697 |
split = split[1].split(","); |
|
700 | 698 |
|
701 |
String[] split = startdate.split("-"); |
|
699 |
funding_level = split[0].replaceAll(".*:\"", ""); |
|
700 |
funding_level = funding_level.replaceFirst(ENCLOSED, ""); |
|
701 |
funding_level = funding_level.trim(); |
|
702 | 702 |
|
703 |
int Startdate = Integer.parseInt(split[0]); |
|
703 |
return funding_level; |
|
704 |
} |
|
704 | 705 |
|
705 |
split = enddate.split("-");
|
|
706 |
private String getYearDifferenceInteger(String enddate, String startdate) {
|
|
706 | 707 |
|
707 |
int Enddate = Integer.parseInt(split[0]);
|
|
708 |
if (!enddate.isEmpty() && enddate != null && startdate != null && !startdate.isEmpty()) {
|
|
708 | 709 |
|
709 |
int diff = Enddate - Startdate; |
|
710 |
return ENCLOSED + diff + ENCLOSED + DELIM; |
|
711 |
} |
|
710 |
// if (enddate != null && startdate != null&&) { |
|
712 | 711 |
|
713 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
714 |
} |
|
712 |
String[] split = startdate.split("-"); |
|
715 | 713 |
|
716 |
private String getYearInt(String data) { |
|
717 |
if (data == null || data.isEmpty() || data.equals("-1")) { |
|
718 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
719 |
} |
|
714 |
int Startdate = Integer.parseInt(split[0]); |
|
720 | 715 |
|
721 |
String[] split = data.split("-");
|
|
716 |
split = enddate.split("-");
|
|
722 | 717 |
|
723 |
if (split != null) {
|
|
718 |
int Enddate = Integer.parseInt(split[0]);
|
|
724 | 719 |
|
725 |
String year = split[0]; |
|
726 |
year = cleanNumber(year); |
|
720 |
int diff = Enddate - Startdate; |
|
721 |
return ENCLOSED + diff + ENCLOSED + DELIM; |
|
722 |
} |
|
727 | 723 |
|
728 |
return ENCLOSED + year + ENCLOSED + DELIM; |
|
729 |
} else { |
|
730 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
731 |
} |
|
724 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
725 |
} |
|
732 | 726 |
|
733 |
} |
|
727 |
private String getYearInt(String data) { |
|
728 |
if (data == null || data.isEmpty() || data.equals("-1")) { |
|
729 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
730 |
} |
|
734 | 731 |
|
735 |
private String cleanNumber(String number) { |
|
736 |
number = number.replaceAll("[^A-Za-z0-9:,____]", ""); |
|
732 |
String[] split = data.split("-"); |
|
737 | 733 |
|
738 |
return number; |
|
739 |
} |
|
734 |
if (split != null) { |
|
740 | 735 |
|
741 |
private String getStringField(String data) { |
|
736 |
String year = split[0]; |
|
737 |
year = cleanNumber(year); |
|
742 | 738 |
|
743 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
739 |
return ENCLOSED + year + ENCLOSED + DELIM; |
|
740 |
} else { |
|
741 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
742 |
} |
|
744 | 743 |
|
745 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
746 |
} else { |
|
744 |
} |
|
747 | 745 |
|
748 |
String field = clean(data); |
|
749 |
if (field == null) { |
|
750 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
751 |
} else { |
|
752 |
return field + DELIM; |
|
753 |
} |
|
754 |
} |
|
755 |
} |
|
746 |
private String cleanNumber(String number) { |
|
747 |
number = number.replaceAll("[^A-Za-z0-9:,____]", ""); |
|
756 | 748 |
|
757 |
private String getStringDateField(String data) { |
|
749 |
return number; |
|
750 |
} |
|
758 | 751 |
|
759 |
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) {
|
|
752 |
private String getStringField(String data) {
|
|
760 | 753 |
|
761 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
762 |
} else { |
|
754 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
763 | 755 |
|
764 |
String field = clean(data); |
|
765 |
if (field == null) { |
|
766 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
767 |
} else { |
|
768 |
return field + DELIM; |
|
769 |
} |
|
770 |
} |
|
771 |
} |
|
756 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
757 |
} else { |
|
772 | 758 |
|
773 |
private String getNumericField(String data) { |
|
774 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
775 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
776 |
} else { |
|
759 |
String field = clean(data); |
|
760 |
if (field == null) { |
|
761 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
762 |
} else { |
|
763 |
return field + DELIM; |
|
764 |
} |
|
765 |
} |
|
766 |
} |
|
777 | 767 |
|
778 |
return ENCLOSED + data + ENCLOSED + DELIM; |
|
779 |
} |
|
780 |
} |
|
768 |
private String getStringDateField(String data) { |
|
781 | 769 |
|
782 |
public String getId(Oaf oaf) { |
|
783 |
switch (oaf.getKind()) { |
|
784 |
case entity: |
|
770 |
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) { |
|
785 | 771 |
|
786 |
return cleanId(oaf.getEntity().getId());
|
|
787 |
case relation:
|
|
772 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
|
773 |
} else {
|
|
788 | 774 |
|
789 |
return cleanId(oaf.getRel().getSource()); |
|
775 |
String field = clean(data); |
|
776 |
if (field == null) { |
|
777 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
778 |
} else { |
|
779 |
return field + DELIM; |
|
780 |
} |
|
781 |
} |
|
782 |
} |
|
790 | 783 |
|
791 |
} |
|
792 |
return null; |
|
784 |
private String getNumericField(String data) { |
|
785 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
786 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
787 |
} else { |
|
793 | 788 |
|
794 |
} |
|
789 |
return ENCLOSED + data + ENCLOSED + DELIM; |
|
790 |
} |
|
791 |
} |
|
795 | 792 |
|
796 |
public String getId(OafRel relOaf) {
|
|
797 |
return cleanId(relOaf.getSource());
|
|
798 |
}
|
|
793 |
public String getId(Oaf oaf) {
|
|
794 |
switch (oaf.getKind()) {
|
|
795 |
case entity:
|
|
799 | 796 |
|
800 |
public static String clean(String value) { |
|
801 |
if (value != null) { |
|
802 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
803 |
// to datacite____:: ) |
|
804 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
805 |
value = value.replaceFirst(".*\\|", ""); |
|
806 |
value = value.replaceAll(DELIM, ""); |
|
807 |
value = value.replaceAll(ENCLOSED, ""); |
|
808 |
value = value.replaceAll("\\r\\n|\\r|\\n", " "); |
|
797 |
return cleanId(oaf.getEntity().getId()); |
|
798 |
case relation: |
|
809 | 799 |
|
810 |
value = value.replaceAll("\\s+", " "); |
|
811 |
value = value.replaceAll("(\\r|\\n)", " "); |
|
812 |
value = value.replaceAll("\\t", " "); |
|
800 |
return cleanId(oaf.getRel().getSource()); |
|
813 | 801 |
|
802 |
} |
|
803 |
return null; |
|
814 | 804 |
|
815 |
// value = value.replaceAll("[^A-Za-z0-9:,____-;:]", " "); |
|
816 |
value = value.trim(); |
|
805 |
} |
|
817 | 806 |
|
818 |
} |
|
819 |
if (value == null) { |
|
820 |
return null; |
|
821 |
} |
|
822 |
return ENCLOSED + value + ENCLOSED; |
|
807 |
public String getId(OafRel relOaf) { |
|
808 |
return cleanId(relOaf.getSource()); |
|
809 |
} |
|
823 | 810 |
|
824 |
} |
|
811 |
public static String clean(String value) { |
|
812 |
if (value != null) { |
|
813 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
814 |
// to datacite____:: ) |
|
815 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
816 |
value = value.replaceFirst(".*\\|", ""); |
|
817 |
value = value.replaceAll(DELIM, ""); |
|
818 |
value = value.replaceAll(ENCLOSED, ""); |
|
819 |
value = value.replaceAll("\\r\\n|\\r|\\n", " "); |
|
825 | 820 |
|
826 |
public static String cleanId(String value) {
|
|
827 |
if (value != null) {
|
|
828 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
|
821 |
value = value.replaceAll("\\s+", " ");
|
|
822 |
value = value.replaceAll("(\\r|\\n)", " ");
|
|
823 |
value = value.replaceAll("\\t", " ");
|
|
829 | 824 |
|
830 | 825 |
|
831 |
// to datacite____:: ) |
|
832 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
833 |
value = value.replaceFirst(".*\\|", ""); |
|
834 |
value = value.replaceAll("\n", ""); |
|
835 |
value = value.replaceAll(DELIM, ""); |
|
836 |
value = value.replaceAll(ENCLOSED, ""); |
|
837 |
value = value.trim(); |
|
826 |
// value = value.replaceAll("[^A-Za-z0-9:,____-;:]", " "); |
|
827 |
value = value.trim(); |
|
838 | 828 |
|
839 |
}
|
|
840 |
if (value == null) {
|
|
841 |
return null;
|
|
842 |
}
|
|
843 |
return ENCLOSED + value + ENCLOSED;
|
|
829 |
}
|
|
830 |
if (value == null) {
|
|
831 |
return null;
|
|
832 |
}
|
|
833 |
return ENCLOSED + value + ENCLOSED;
|
|
844 | 834 |
|
845 |
}
|
|
835 |
}
|
|
846 | 836 |
|
847 |
public long DATEDIFF(String startDate, String endDate) { |
|
848 |
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000; |
|
849 |
long days = 0l; |
|
850 |
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss"); |
|
851 |
// <startdate>2011-09-01</startdate> |
|
852 |
// <enddate>2015-08-31</enddate> |
|
853 |
Date dateIni = null; |
|
854 |
Date dateFin = null; |
|
837 |
public static String cleanId(String value) { |
|
838 |
if (value != null) { |
|
839 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
855 | 840 |
|
856 |
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) { |
|
857 |
return 0; |
|
858 |
} |
|
859 |
try { |
|
860 |
dateIni = (Date) format.parse(startDate); |
|
861 |
dateFin = (Date) format.parse(endDate); |
|
862 |
days = (dateFin.getTime() - dateIni.getTime()) / MILLISECS_PER_DAY; |
|
863 |
} catch (Exception e) { |
|
864 |
log.error(e.toString()); |
|
865 |
return 0; |
|
866 |
} |
|
867 | 841 |
|
868 |
return days; |
|
869 |
} |
|
842 |
// to datacite____:: ) |
|
843 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
844 |
value = value.replaceFirst(".*\\|", ""); |
|
845 |
value = value.replaceAll("\n", ""); |
|
846 |
value = value.replaceAll(DELIM, ""); |
|
847 |
value = value.replaceAll(ENCLOSED, ""); |
|
848 |
value = value.trim(); |
|
870 | 849 |
|
871 |
public String getDELIM() { |
|
872 |
return DELIM; |
|
873 |
} |
|
850 |
} |
|
851 |
if (value == null) { |
|
852 |
return null; |
|
853 |
} |
|
854 |
return ENCLOSED + value + ENCLOSED; |
|
874 | 855 |
|
875 |
public void setDELIM(String dELIM) { |
|
876 |
DELIM = dELIM; |
|
877 |
} |
|
856 |
} |
|
878 | 857 |
|
879 |
public String getNULL_STRING() { |
|
880 |
return NULL_STRING; |
|
881 |
} |
|
858 |
public long DATEDIFF(String startDate, String endDate) { |
|
859 |
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000; |
|
860 |
long days = 0l; |
|
861 |
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss"); |
|
862 |
// <startdate>2011-09-01</startdate> |
|
863 |
// <enddate>2015-08-31</enddate> |
|
864 |
Date dateIni = null; |
|
865 |
Date dateFin = null; |
|
882 | 866 |
|
883 |
public void setNULL_STRING(String nULL_STRING) { |
|
884 |
NULL_STRING = nULL_STRING; |
|
885 |
} |
|
867 |
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) { |
|
868 |
return 0; |
|
869 |
} |
|
870 |
try { |
|
871 |
dateIni = (Date) format.parse(startDate); |
|
872 |
dateFin = (Date) format.parse(endDate); |
|
873 |
days = (dateFin.getTime() - dateIni.getTime()) / MILLISECS_PER_DAY; |
|
874 |
} catch (Exception e) { |
|
875 |
log.error(e.toString()); |
|
876 |
return 0; |
|
877 |
} |
|
886 | 878 |
|
887 |
public String getNULL_NUM() { |
|
888 |
return NULL_NUM; |
|
889 |
} |
|
879 |
return days; |
|
880 |
} |
|
890 | 881 |
|
891 |
public void setNULL_NUM(String nULL_NUM) {
|
|
892 |
NULL_NUM = nULL_NUM;
|
|
893 |
}
|
|
882 |
public String getDELIM() {
|
|
883 |
return DELIM;
|
|
884 |
}
|
|
894 | 885 |
|
895 |
public String getENCLOSED() {
|
|
896 |
return ENCLOSED;
|
|
897 |
}
|
|
886 |
public void setDELIM(String dELIM) {
|
|
887 |
DELIM = dELIM;
|
|
888 |
}
|
|
898 | 889 |
|
899 |
public void setENCLOSED(String eNCLOSED) {
|
|
900 |
ENCLOSED = eNCLOSED;
|
|
901 |
}
|
|
890 |
public String getNULL_STRING() {
|
|
891 |
return NULL_STRING;
|
|
892 |
}
|
|
902 | 893 |
|
894 |
public void setNULL_STRING(String nULL_STRING) { |
|
895 |
NULL_STRING = nULL_STRING; |
|
896 |
} |
|
897 |
|
|
898 |
public String getNULL_NUM() { |
|
899 |
return NULL_NUM; |
Also available in: Unified diff
funding tree change for beta