diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2021-10-26 18:12:23 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2021-10-26 18:12:23 -0700 |
| commit | 485dd2cfd120c52bbc5cc7745e44176d1003b40d (patch) | |
| tree | 966bf78a4bd3cc1f6c94efb8fc3054a8a441dab0 /python/sandcrawler/misc.py | |
| parent | 7087e7f65d8b81e29af44a43c1067bb2ec618c4e (diff) | |
| download | sandcrawler-485dd2cfd120c52bbc5cc7745e44176d1003b40d.tar.gz sandcrawler-485dd2cfd120c52bbc5cc7745e44176d1003b40d.zip | |
lint collection membership (last lint for now)
Diffstat (limited to 'python/sandcrawler/misc.py')
| -rw-r--r-- | python/sandcrawler/misc.py | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/python/sandcrawler/misc.py b/python/sandcrawler/misc.py index 83a4626..1c779ce 100644 --- a/python/sandcrawler/misc.py +++ b/python/sandcrawler/misc.py @@ -58,7 +58,7 @@ def gen_file_metadata(blob: bytes, allow_empty: bool = False) -> dict: # crude checks for XHTML or JATS XML, using only first 1 kB of file if b"<htm" in blob[:1024] and b'xmlns="http://www.w3.org/1999/xhtml"' in blob[:1024]: mimetype = "application/xhtml+xml" - elif b"<article " in blob[:1024] and not b"<html" in blob[:1024]: + elif b"<article " in blob[:1024] and b"<html" not in blob[:1024]: mimetype = "application/jats+xml" hashes = [ hashlib.sha1(), @@ -88,7 +88,7 @@ def gen_file_metadata_path(path: str, allow_empty: bool = False) -> dict: # crude checks for XHTML or JATS XML, using only first 1 kB of file if b"<htm" in blob[:1024] and b'xmlns="http://www.w3.org/1999/xhtml"' in blob[:1024]: mimetype = "application/xhtml+xml" - elif b"<article " in blob[:1024] and not b"<html" in blob[:1024]: + elif b"<article " in blob[:1024] and b"<html" not in blob[:1024]: mimetype = "application/jats+xml" hashes = [ hashlib.sha1(), |