diff options
Diffstat (limited to 'python')
| -rwxr-xr-x | python/extraction_cdx_grobid.py | 8 | ||||
| -rwxr-xr-x | python/extraction_ungrobided.py | 8 |
2 files changed, 8 insertions, 8 deletions
diff --git a/python/extraction_cdx_grobid.py b/python/extraction_cdx_grobid.py index 040538c..b27d053 100755 --- a/python/extraction_cdx_grobid.py +++ b/python/extraction_cdx_grobid.py @@ -37,7 +37,7 @@ from grobid2json import teixml2json sentry_client = raven.Client() # Specific poison-pill rows we should skip -KEY_BLACKLIST = ( +KEY_DENYLIST = ( 'sha1:DLCCSMMVTCCIR6LRXHEQLZ4PWO6NG2YT', # "failed to guess ARC header format" ) @@ -210,9 +210,9 @@ class MRExtractCdxGrobid(MRJob): yield _, status return key = info['key'] - if key in KEY_BLACKLIST: - self.increment_counter('lines', 'blacklist') - yield _, dict(status='blacklist', key=key) + if key in KEY_DENYLIST: + self.increment_counter('lines', 'denylist') + yield _, dict(status='denylist', key=key) return # Note: this may not get "cleared" correctly diff --git a/python/extraction_ungrobided.py b/python/extraction_ungrobided.py index 4074112..972b8f9 100755 --- a/python/extraction_ungrobided.py +++ b/python/extraction_ungrobided.py @@ -37,7 +37,7 @@ from grobid2json import teixml2json sentry_client = raven.Client() # Specific poison-pill rows we should skip -KEY_BLACKLIST = ( +KEY_DENYLIST = ( 'sha1:DLCCSMMVTCCIR6LRXHEQLZ4PWO6NG2YT', # "failed to guess ARC header format" ) @@ -212,9 +212,9 @@ class MRExtractUnGrobided(MRJob): yield _, status return key = info['key'] - if key in KEY_BLACKLIST: - self.increment_counter('lines', 'blacklist') - yield _, dict(status='blacklist', key=key) + if key in KEY_DENYLIST: + self.increment_counter('lines', 'denylist') + yield _, dict(status='denylist', key=key) return # Note: this may not get "cleared" correctly |