Revision 29380
Added by Eri Katsari over 10 years ago
| modules/dnet-openaire-stats/trunk/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/Serializer.java | ||
|---|---|---|
| 200 | 200 |
|
| 201 | 201 |
} |
| 202 | 202 |
|
| 203 |
public static boolean isNumeric(String str) {
|
|
| 204 |
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional
|
|
| 205 |
// '-' and decimal. |
|
| 206 |
} |
|
| 207 |
|
|
| 208 |
private boolean isValidTopic(String t) {
|
|
| 209 |
if (t != null && !t.isEmpty() && !t.equals(ENCLOSED + ENCLOSED + DELIM) && !t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) |
|
| 210 |
|
|
| 211 |
{
|
|
| 212 |
return true; |
|
| 213 |
} |
|
| 214 |
return false; |
|
| 215 |
} |
|
| 216 |
|
|
| 217 |
|
|
| 218 |
|
|
| 203 | 219 |
private HashMap<String, List<String>> getResultTopics(OafEntity valueEntity) {
|
| 204 | 220 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
| 205 | 221 |
List<String> buffs = new ArrayList<String>(); |
| ... | ... | |
| 209 | 225 |
List<StructuredProperty> Topics = metadata.getSubjectList(); |
| 210 | 226 |
for (StructuredProperty topic : Topics) {
|
| 211 | 227 |
// TODOs |
| 212 |
if (topic.getValue() != null && !topic.getValue().isEmpty() && !topic.getValue().equals("")) {
|
|
| 213 |
String t = getStringField(topic.getValue()); |
|
| 214 |
if (t != null && !t.isEmpty() && !t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) {
|
|
| 215 |
buffs.add(t); |
|
| 228 |
if (topic.getValue() != null && !topic.getValue().isEmpty() && !topic.getValue().equals("") && !topic.getValue().equals(" ")) {
|
|
| 229 |
|
|
| 230 |
if (!isNumeric(topic.getValue())) {
|
|
| 231 |
String t = getStringField(topic.getValue()); |
|
| 232 |
if (isValidTopic(t)) {
|
|
| 233 |
buffs.add(t); |
|
| 234 |
} |
|
| 235 |
|
|
| 216 | 236 |
} |
| 217 |
|
|
| 218 | 237 |
} |
| 219 | 238 |
} |
| 220 | 239 |
rels.put("resultTopic", buffs);
|
| ... | ... | |
| 327 | 346 |
|
| 328 | 347 |
// result_topics/ |
| 329 | 348 |
String full_id = getStringField(data.getId()); |
| 330 |
|
|
| 349 |
|
|
| 331 | 350 |
buff += full_id; |
| 332 | 351 |
|
| 333 | 352 |
// result_languages |
| ... | ... | |
| 386 | 405 |
// number?? |
| 387 | 406 |
|
| 388 | 407 |
buff += getStringField("1");
|
| 389 |
//TODO this is a rotten record.// |
|
| 390 |
if (full_id.contentEquals("datacite____::5903fa1fba477116aba4a922cc9eb56f")) {
|
|
| 391 |
log.error("ROTTEN RECORD " + buff);
|
|
| 392 |
return null;
|
|
| 393 |
}
|
|
| 408 |
// TODO this is a rotten record.//
|
|
| 409 |
if (full_id.contentEquals("datacite____::5903fa1fba477116aba4a922cc9eb56f")) {
|
|
| 410 |
log.error("ROTTEN RECORD " + buff);
|
|
| 411 |
return null; |
|
| 412 |
} |
|
| 394 | 413 |
// TODO check if valid or empty protobuff |
| 395 | 414 |
// TODO do it in all protos? |
| 396 | 415 |
if (isValid(buff, full_id)) {
|
Also available in: Unified diff