diff options
Diffstat (limited to 'mapreduce/extraction_cdx_grobid.py')
| -rwxr-xr-x | mapreduce/extraction_cdx_grobid.py | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/mapreduce/extraction_cdx_grobid.py b/mapreduce/extraction_cdx_grobid.py index f6c1ec1..6690f49 100755 --- a/mapreduce/extraction_cdx_grobid.py +++ b/mapreduce/extraction_cdx_grobid.py @@ -40,6 +40,7 @@ sentry_client = raven.Client() class MRExtractCdxGrobid(MRJob): # CDX lines in; JSON status out + HADOOP_INPUT_FORMAT = 'org.apache.hadoop.mapred.lib.NLineInputFormat' INPUT_PROTOCOL = mrjob.protocol.RawValueProtocol OUTPUT_PROTOCOL = mrjob.protocol.JSONValueProtocol |