diff options
Diffstat (limited to 'python/fatcat_tools/transforms/access.py')
| -rw-r--r-- | python/fatcat_tools/transforms/access.py | 44 |
1 files changed, 24 insertions, 20 deletions
diff --git a/python/fatcat_tools/transforms/access.py b/python/fatcat_tools/transforms/access.py index ae9880e7..34212a6a 100644 --- a/python/fatcat_tools/transforms/access.py +++ b/python/fatcat_tools/transforms/access.py @@ -1,4 +1,3 @@ - from enum import Enum from typing import List, Optional @@ -16,6 +15,7 @@ class AccessType(str, Enum): openlibrary = "openlibrary" wikipedia = "wikipedia" + class AccessOption(BaseModel): access_type: AccessType @@ -40,27 +40,31 @@ def release_access_options(release: ReleaseEntity) -> List[AccessOption]: option found """ options = [] - for f in (release.files or []): + for f in release.files or []: thumbnail_url = None - if f.mimetype == 'application/pdf' and f.sha1 and f.urls: + if f.mimetype == "application/pdf" and f.sha1 and f.urls: # NOTE: scholar.archive.org does an actual database check before # generating these URLs, but we skip that for speed thumbnail_url = f"https://blobs.fatcat.wiki/thumbnail/pdf/{f.sha1[0:2]}/{f.sha1[2:4]}/{f.sha1}.180px.jpg" - for u in (f.urls or []): - if '://web.archive.org/' in u.url: - return [AccessOption( - access_type="wayback", - access_url=u.url, - mimetype=f.mimetype, - size_bytes=f.size, - thumbnail_url=thumbnail_url, - )] - elif '://archive.org/' in u.url: - return [AccessOption( - access_type="ia_file", - access_url=u.url, - mimetype=f.mimetype, - size_bytes=f.size, - thumbnail_url=thumbnail_url, - )] + for u in f.urls or []: + if "://web.archive.org/" in u.url: + return [ + AccessOption( + access_type="wayback", + access_url=u.url, + mimetype=f.mimetype, + size_bytes=f.size, + thumbnail_url=thumbnail_url, + ) + ] + elif "://archive.org/" in u.url: + return [ + AccessOption( + access_type="ia_file", + access_url=u.url, + mimetype=f.mimetype, + size_bytes=f.size, + thumbnail_url=thumbnail_url, + ) + ] return options |