forked from D-Net/dnet-hadoop
fix bug on oai iterator that skip record cleaned
This commit is contained in:
parent
d4dadf6d77
commit
3c6fc2096c
|
@ -161,7 +161,7 @@ public class OaiIterator implements Iterator<String> {
|
|||
report.put(e.getClass().getName(), e.getMessage());
|
||||
final String cleaned = XmlCleaner.cleanAllEntities(xml);
|
||||
try {
|
||||
doc = DocumentHelper.parseText(xml);
|
||||
doc = DocumentHelper.parseText(cleaned);
|
||||
} catch (final DocumentException e1) {
|
||||
final String resumptionToken = extractResumptionToken(xml);
|
||||
if (resumptionToken == null) {
|
||||
|
|
Loading…
Reference in New Issue