ZipExtFile in zipfile can be seekable

[IridiumYang]

BPO	22908
Nosy	@gpshead, @serhiy-storchaka, @jjolly
PRs	bpo-22908: Add seek and tell functionality to ZipExtFile #4966
Files	zipfile.diff ziz.py: zip-in-zip test program

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/gpshead'
closed_at = <Date 2018-01-30.08:53:26.935>
created_at = <Date 2014-11-20.15:20:43.347>
labels = ['3.7', 'type-feature', 'library']
title = 'ZipExtFile in zipfile can be seekable'
updated_at = <Date 2018-01-30.08:53:26.933>
user = 'https://bugs.python.org/IridiumYang'

bugs.python.org fields:

activity = <Date 2018-01-30.08:53:26.933>
actor = 'gregory.p.smith'
assignee = 'gregory.p.smith'
closed = True
closed_date = <Date 2018-01-30.08:53:26.935>
closer = 'gregory.p.smith'
components = ['Library (Lib)']
creation = <Date 2014-11-20.15:20:43.347>
creator = 'Iridium.Yang'
dependencies = []
files = ['37237', '47345']
hgrepos = []
issue_num = 22908
keywords = ['patch']
message_count = 7.0
messages = ['231438', '231446', '231472', '256683', '268764', '308935', '311254']
nosy_count = 6.0
nosy_names = ['gregory.p.smith', 'jae', 'serhiy.storchaka', 'Iridium.Yang', 'dkessel', 'jjolly']
pr_nums = ['4966']
priority = 'normal'
resolution = 'fixed'
stage = 'commit review'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue22908'
versions = ['Python 3.7']

[IridiumYang]

The ZipExtFile class in zipfile module does not provide a seek method like GzipFile. As a result, it is hard to manipulate files without extract all the content.
For example, a very large tar file compressed with zip. The TarFile module can operate on file object, but need seek method. So the ZipExtFile instance return from ZipFile can not passed into TarFile.
This may seem strange but I encounter this on Samsung firmware.

In fact, the seek method in GzipFile or someother compressed format can be implemented in zipfile very easily. Here is my naive modification (nearly same as in GzipFile)

[serhiy-storchaka]

I'm -1 on adding the seek method with linear complexity. This looks as attractive nonsense to me. It would be better just make TarFile working with non-seekable streams.

[serhiy-storchaka]

Actually TarFile already works with non-seekable streams. Use TarFile.open() with mode='r|*' or like.

On other hand I'm not against the make non-compressed ZipExtFile seekable. It can be helpful in case when ZIP file is used just as a container for other files.

[dkessel]

It would be great to have the ZipFileExt class seekable.
This would help in implementing features in other projects.

For example, pydicom would gain the ability to read from ZIP files, as mentioned in pydicom/pydicom#219

[jae]

To add to this (without looking at the patch): I just to my astonishment learned that a ZipExtFile doesn't even support tell(). I can understand the seek being nontrivial... but the tell? It's a bytestream, and there is (isn't there?) a clear definition of what next byte a read(1) would deliver. It should be trivial to keep track of the (only ever increasing) file position.

[jjolly]

Please be gentle, this is my first submission to python.

The use case for me was a recursive zip-within-a-zip situation. I wanted to allow the creation of a zipfile.ZipFile from an existing zipfile.ZipExtFile, but the lack of seek prevented this.

I simply treated forward seeks as a read, and backward seeks as a reset-and-read. The reset was the tricky part as it required restoring several original values such as the remaining compressed length, remaining data length, and the running crc32.

I pushed this into the latest upstream branch, but as I am testing this in v3.4 it should be easy to backport if necessary (I suspect not).

I based my fix on a little program that I wrote to test the feasibility of this idea. I am attaching that test program here.

[gpshead]

New changeset 066df4f by Gregory P. Smith (John Jolly) in branch 'master':
bpo-22908: Add seek and tell functionality to ZipExtFile (GH-4966)
066df4f