Repository navigation
Cannot open zip64 file if Zip64EOCD record has additional data #126834
Description
Bug report
Bug description:
"Zip64 end of central directory record" may have additional data in the "zip64 extensible data sector" field.
In that case, zipfile should use offset from "Zip64 end of central directory locator".
Now it's read but ignored.
def run_test():
with zipfile.ZipFile("file-with-extended-zip64eocd-record.zip", allowZip64=True) as special_file: # This fails
print("success")Problem lies in following code:
sig, diskno, reloff, disks = struct.unpack(structEndArchive64Locator, data)
if sig != stringEndArchive64Locator:
return endrec
if diskno != 0 or disks > 1:
raise BadZipFile("zipfiles that span multiple disks are not supported")
# Assume no 'zip64 extensible data'
fpin.seek(offset - sizeEndCentDir64Locator - sizeEndCentDir64, 2)
data = fpin.read(sizeEndCentDir64)While it should be:
sig, diskno, reloff, disks = struct.unpack(structEndArchive64Locator, data)
if sig != stringEndArchive64Locator:
return endrec
if diskno != 0 or disks > 1:
raise BadZipFile("zipfiles that span multiple disks are not supported")
fpin.seek(reloff, 0)
data = fpin.read(sizeEndCentDir64)CPython versions tested on:
CPython main branch
Operating systems tested on:
macOS
Activity
VladRassokhin commentedon Nov 14, 2024
Hex output of the end of such file:
VladRassokhin commentedon Nov 14, 2024
I can submit MR in a couple of days. Seems I've found a solution.
VladRassokhin commentedon Nov 14, 2024
Here's a python script which is able to generate a zip64 archive with some data in the zip64 end-of-archive record:
https://gist.github.com/VladRassokhin/9299bb8fbe3169b96e7bc31f91553815
I've copied and adjusted
zipfile.ZipFile._write_end_recordProperly read zip64 archives with non-empty zip64 extensible data sec…
thatch commentedon May 20, 2025
@serhiy-storchaka I'm reasonably happy with that PR, but it lacks tests. We currently can't create a zip with such data using just zipfile, so there's a little chicken-egg situation.
danifus commentedon Apr 27, 2026
Fixed by: #139702