diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2018-04-05 17:18:52 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2018-04-05 17:18:52 -0700 |
| commit | 77577da13afe07b5177452122f4cee77e3357b4e (patch) | |
| tree | d37e411121abb805c272f48a6f4579e7af650bb1 /mapreduce/tests/test_extraction_cdx_grobid.py | |
| parent | 37a775851b2d21f2afe7418a8628c50ae37edc5b (diff) | |
| download | sandcrawler-77577da13afe07b5177452122f4cee77e3357b4e.tar.gz sandcrawler-77577da13afe07b5177452122f4cee77e3357b4e.zip | |
improve test coverage
Diffstat (limited to 'mapreduce/tests/test_extraction_cdx_grobid.py')
| -rw-r--r-- | mapreduce/tests/test_extraction_cdx_grobid.py | 1 |
1 files changed, 0 insertions, 1 deletions
diff --git a/mapreduce/tests/test_extraction_cdx_grobid.py b/mapreduce/tests/test_extraction_cdx_grobid.py index 514a830..1d32c9f 100644 --- a/mapreduce/tests/test_extraction_cdx_grobid.py +++ b/mapreduce/tests/test_extraction_cdx_grobid.py @@ -121,7 +121,6 @@ def test_parse_cdx_skip(job): job.mapper_init() - print("warc format") raw = "com,sagepub,cep)/content/28/9/960.full.pdf 20170705062200 http://cep.sagepub.com/content/28/9/960.full.pdf application/pdf 200 3I42H3S6NNFQ2MSVX7XZKYAYSCX5QBYJ - - 401 313356621 CITESEERX-CRAWL-2017-06-20-20170705062052659-00043-31209~wbgrp-svc284.us.archive.org~8443.warc.gz" info, status = job.mapper(None, raw).__next__() |