diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2021-01-22 19:31:50 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2021-01-22 19:31:50 -0800 |
| commit | 2b5cfb73097f17f383457380e36bf977cd0da5ed (patch) | |
| tree | 67ab6bd2d53735f4394ad1ce6e98dd2e0c8e8ef3 /fatcat_scholar/work_pipeline.py | |
| parent | 0591b388a55771ed0581102989e23ba22a720a27 (diff) | |
| download | fatcat-scholar-2b5cfb73097f17f383457380e36bf977cd0da5ed.tar.gz fatcat-scholar-2b5cfb73097f17f383457380e36bf977cd0da5ed.zip | |
bug fix: actually fetch/include HTML fulltext
Diffstat (limited to 'fatcat_scholar/work_pipeline.py')
| -rw-r--r-- | fatcat_scholar/work_pipeline.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/fatcat_scholar/work_pipeline.py b/fatcat_scholar/work_pipeline.py index ea33a01..fd85089 100644 --- a/fatcat_scholar/work_pipeline.py +++ b/fatcat_scholar/work_pipeline.py @@ -317,7 +317,7 @@ class WorkPipeline: html_fulltext: Optional[Any] = None for ident in pref_idents: release = release_dict[ident] - if not release.files: + if not (release.files or release.webcaptures): continue for fe in release.files: if not fe.sha1 or fe.mimetype not in (None, "application/pdf"): |