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