diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-04-09 17:48:47 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-04-09 17:48:47 -0700 |
| commit | 042bd36c25206ff45e305d094028b6482a4c4074 (patch) | |
| tree | 5e25122ad6227855416fe964658e0581e048332c | |
| parent | 5b22495057fa9cb40271764c9e80166882ba3f21 (diff) | |
| download | fatcat-covid19-042bd36c25206ff45e305d094028b6482a4c4074.tar.gz fatcat-covid19-042bd36c25206ff45e305d094028b6482a4c4074.zip | |
bugfix: handle missing file mimetypes
| -rwxr-xr-x | bin/deliver_file2disk.py | 2 | ||||
| -rw-r--r-- | fatcat_covid19/common.py | 2 |
2 files changed, 2 insertions, 2 deletions
diff --git a/bin/deliver_file2disk.py b/bin/deliver_file2disk.py index 49e0c73..e5f13c0 100755 --- a/bin/deliver_file2disk.py +++ b/bin/deliver_file2disk.py @@ -123,7 +123,7 @@ class DeliverFatcatDisk: """ good = [] for f in files: - if f['mimetype'] and not 'pdf' in f['mimetype'].lower(): + if f.get('mimetype') and not 'pdf' in f['mimetype'].lower(): continue for url in f['urls']: if 'archive.org/' in url['url']: diff --git a/fatcat_covid19/common.py b/fatcat_covid19/common.py index 97bc675..7b88344 100644 --- a/fatcat_covid19/common.py +++ b/fatcat_covid19/common.py @@ -80,7 +80,7 @@ def find_local_file(files, base_dir="."): If found, returns the file entity; the path can be determined from the sha1hex field. """ for f in files: - if f['mimetype'] and not 'pdf' in f['mimetype'].lower(): + if f.get('mimetype') and not 'pdf' in f['mimetype'].lower(): continue pdf_path = blob_path(f['sha1'], directory="pdf/", file_suffix=".pdf", base_dir=base_dir) if os.path.isfile(pdf_path): |