diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2019-05-16 19:19:22 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2019-05-16 19:23:07 -0700 |
| commit | b888f2201f981a2e01026e53d5c6ba2d435e9506 (patch) | |
| tree | a22d0a3fbf7f23a72dca2f500eb46dc93121d05b | |
| parent | 7ed25a124e91f005b3347924be3783b51b79d197 (diff) | |
| download | fatcat-b888f2201f981a2e01026e53d5c6ba2d435e9506.tar.gz fatcat-b888f2201f981a2e01026e53d5c6ba2d435e9506.zip | |
fix elastic file pdf check
| -rw-r--r-- | python/fatcat_tools/transforms/elasticsearch.py | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/python/fatcat_tools/transforms/elasticsearch.py b/python/fatcat_tools/transforms/elasticsearch.py index 91086a0c..645142de 100644 --- a/python/fatcat_tools/transforms/elasticsearch.py +++ b/python/fatcat_tools/transforms/elasticsearch.py @@ -137,8 +137,10 @@ def release_to_elasticsearch(entity, force_bool=True): if f.extra and f.extra.get('shadows'): # TODO: shadow check goes here in_shadows = True - is_pdf = 'pdf' in (f.mimetype or '') or 'pdf' in url.lower() + is_pdf = 'pdf' in (f.mimetype or '') for url in (f.urls or []): + if not f.mimetype and 'pdf' in url.lower(): + is_pdf = True if url.url.lower().startswith('http'): in_web = True if url.rel in ('dweb', 'p2p', 'ipfs', 'dat', 'torrent'): |