Revision 54274
Added by Claudio Atzori over 5 years ago
| modules/dnet-index-solr-common/trunk/src/main/java/eu/dnetlib/functionality/index/solr/feed/StreamingInputDocumentFactory.java | ||
|---|---|---|
| 13 | 13 |
import com.google.common.collect.Lists; |
| 14 | 14 |
import eu.dnetlib.functionality.index.solr.feed.ResultTransformer.Mode; |
| 15 | 15 |
import org.apache.solr.common.SolrInputDocument; |
| 16 |
import org.apache.solr.common.StringUtils; |
|
| 16 | 17 |
|
| 17 | 18 |
/** |
| 18 | 19 |
* Optimized version of the document parser, drop in replacement of InputDocumentFactory. |
| ... | ... | |
| 44 | 45 |
|
| 45 | 46 |
protected static final String ROOT_ELEMENT = "indexRecord"; |
| 46 | 47 |
|
| 48 |
protected static final int MAX_FIELD_LENGTH = 32765; |
|
| 49 |
|
|
| 47 | 50 |
protected ThreadLocal<XMLInputFactory> inputFactory = new ThreadLocal<XMLInputFactory>() {
|
| 48 | 51 |
|
| 49 | 52 |
@Override |
| ... | ... | |
| 257 | 260 |
if (text.isEndElement()) // log.warn("skipping because isEndOfElement " + text.asEndElement().getName().getLocalPart());
|
| 258 | 261 |
return ""; |
| 259 | 262 |
|
| 260 |
return text.asCharacters().getData(); |
|
| 263 |
final String data = text.asCharacters().getData(); |
|
| 264 |
if (data != null && data.length() > MAX_FIELD_LENGTH) {
|
|
| 265 |
return data.substring(0, MAX_FIELD_LENGTH); |
|
| 266 |
} |
|
| 267 |
|
|
| 268 |
return data; |
|
| 261 | 269 |
} |
| 262 | 270 |
|
| 263 | 271 |
} |
Also available in: Unified diff
field length capped to 32765 characters