diff options
author | Bryan Newbold <bnewbold@archive.org> | 2020-11-08 13:20:24 -0800 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-11-08 13:20:26 -0800 |
commit | 3977afdff906367525cd6959221b6f3edf19793d (patch) | |
tree | c8284f463a65e47a854170ebf85c3f26a213643d /python/sandcrawler/misc.py | |
parent | e87f28d62e7e92de5502cd04d8c3effb19da19f8 (diff) | |
download | sandcrawler-3977afdff906367525cd6959221b6f3edf19793d.tar.gz sandcrawler-3977afdff906367525cd6959221b6f3edf19793d.zip |
gen_file_metadata: allow empty/null bodies (if flag set)
This is for HTML sub-resources, which can validly be empty (I think)
Diffstat (limited to 'python/sandcrawler/misc.py')
-rw-r--r-- | python/sandcrawler/misc.py | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/python/sandcrawler/misc.py b/python/sandcrawler/misc.py index 67e5c0b..1cdd038 100644 --- a/python/sandcrawler/misc.py +++ b/python/sandcrawler/misc.py @@ -18,13 +18,15 @@ def clean_url(s: str) -> str: parsed.colon_before_port = b'' return str(urlcanon.whatwg(parsed)) -def gen_file_metadata(blob: bytes) -> dict: +def gen_file_metadata(blob: bytes, allow_empty: bool = False) -> dict: """ Takes a file blob (bytestream) and returns hashes and other metadata. Returns a dict: size_bytes, md5hex, sha1hex, sha256hex, mimetype """ - assert blob + assert blob is not None + if not allow_empty: + assert blob mimetype = magic.Magic(mime=True).from_buffer(blob) if mimetype in ("application/xml", "text/xml"): # crude check for JATS XML, using only first 1 kB of file |