diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2018-08-25 22:23:52 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2018-08-25 22:24:06 -0700 |
| commit | ec67bbe00efe04f120bb9c278da61545af436a4c (patch) | |
| tree | a28dd7cb738e72b0556d72dc5f57588c972aba94 /python/tests | |
| parent | a71d556763b4031bfa0e56abc72348d7f1d3d966 (diff) | |
| download | sandcrawler-ec67bbe00efe04f120bb9c278da61545af436a4c.tar.gz sandcrawler-ec67bbe00efe04f120bb9c278da61545af436a4c.zip | |
WIP: ungrobided doesn't inherit (copypasta)
Diffstat (limited to 'python/tests')
| -rw-r--r-- | python/tests/test_extraction_ungrobided.py | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/python/tests/test_extraction_ungrobided.py b/python/tests/test_extraction_ungrobided.py index 0ec47bc..366d392 100644 --- a/python/tests/test_extraction_ungrobided.py +++ b/python/tests/test_extraction_ungrobided.py @@ -139,28 +139,28 @@ def test_parse_ungrobided_invalid(job): print("space-prefixed line") 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-2017-06-20-20170705061647307-00039-00048-wbgrp-svc284/CITESEERX-CRAWL-2017-06-20-20170705062052659-00043-31209~wbgrp-svc284.us.archive.org~8443.warc.gz" - info, status = job.parse_line(raw) + info, status = job.parse_ungrobided_line(raw) assert info is None assert status['status'] == "invalid" assert 'prefix' in status['reason'] print("commented line") 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-2017-06-20-20170705061647307-00039-00048-wbgrp-svc284/CITESEERX-CRAWL-2017-06-20-20170705062052659-00043-31209~wbgrp-svc284.us.archive.org~8443.warc.gz" - info, status = job.parse_line(raw) + info, status = job.parse_ungrobided_line(raw) assert info is None assert status['status'] == "invalid" assert 'prefix' in status['reason'] print("wrong column count") raw = "a b c d e" - info, status = job.parse_line(raw) + info, status = job.parse_ungrobided_line(raw) assert info is None assert status['status'] == "invalid" assert 'parse' in status['reason'] print("CDX line, somehow") raw = "com,sagepub,cep)/content/28/9/960.full.pdf 20170705062200 http://cep.sagepub.com/content/28/9/960.full.pdf - 200 3I42H3S6NNFQ2MSVX7XZKYAYSCX5QBYJ - - 401 313356621 CITESEERX-CRAWL-2017-06-20-20170705061647307-00039-00048-wbgrp-svc284/CITESEERX-CRAWL-2017-06-20-20170705062052659-00043-31209~wbgrp-svc284.us.archive.org~8443.warc.gz" - info, status = job.parse_line(raw) + info, status = job.parse_ungrobided_line(raw) assert info is None print(status) assert status['status'] == "invalid" |