diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2019-05-15 10:43:27 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2019-05-15 10:43:27 -0700 |
| commit | a4bc37289a74df90d6b095e8a3e1958e8be3147f (patch) | |
| tree | 179a05997a070412c5b0b0ca9f8f1012e042556f | |
| parent | 4d56e71598457489e9f71ef4ce5c9b0254a4cce1 (diff) | |
| download | fatcat-a4bc37289a74df90d6b095e8a3e1958e8be3147f.tar.gz fatcat-a4bc37289a74df90d6b095e8a3e1958e8be3147f.zip | |
elastic transforms: work around missing pdf mimetypes
| -rw-r--r-- | python/fatcat_tools/transforms/elasticsearch.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/python/fatcat_tools/transforms/elasticsearch.py b/python/fatcat_tools/transforms/elasticsearch.py index 971d1c11..91086a0c 100644 --- a/python/fatcat_tools/transforms/elasticsearch.py +++ b/python/fatcat_tools/transforms/elasticsearch.py @@ -137,7 +137,7 @@ 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 '') + is_pdf = 'pdf' in (f.mimetype or '') or 'pdf' in url.lower() for url in (f.urls or []): if url.url.lower().startswith('http'): in_web = True |