How to create a zip archive of a directory?

S

30

885

How can I create a zip archive of a directory structure in Python?

Strop answered 6/12, 2009 at 11:12 Comment(5)

Don't use the solution suggested in the accepted answer but the one further down using make_archive from shutil (if you want to zip a single directory recursively). – Metameric 8/10, 2018 at 21:39

yes, agree with @Metameric - Martha Yi seems to be unregistered - so is there now way to change the accepted answer through a community process? – Solitary 20/9, 2021 at 14:22

one caveat with shutil.make_archive - it doesn't seem to follow symlinks – Employer 2/12, 2021 at 16:21

The accepted answer is the only one that is actually thread safe in regards to read/write access while creating the zipfile from a directory since each file is opened individually, locking read access to it until the file is closed. – Proceeding 16/6, 2022 at 17:3

@Proceeding It says in make_archive's documentation that "Changed in version 3.10.6: This function is now made thread-safe during creation of standard .zip and tar archives." – Anesthetic 29/3 at 23:16

T

718

As others have pointed out, you should use zipfile. The documentation tells you what functions are available, but doesn't really explain how you can use them to zip an entire directory. I think it's easiest to explain with some example code:

import os
import zipfile
    
def zipdir(path, ziph):
    # ziph is zipfile handle
    for root, dirs, files in os.walk(path):
        for file in files:
            ziph.write(os.path.join(root, file), 
                       os.path.relpath(os.path.join(root, file), 
                                       os.path.join(path, '..')))

with zipfile.ZipFile('Python.zip', 'w', zipfile.ZIP_DEFLATED) as zipf:
    zipdir('tmp/', zipf)

Tenpin answered 6/12, 2009 at 11:23 Comment(19)

I would add a second argument to the write call, passing os.path.relpath(os.path.join(root, file), os.path.join(path, '..')). That would let you zip a directory from any working directory, without getting the full absolute paths in the archive. – Elmer 29/6, 2013 at 14:35

There's a funny recursion going on when I try to zip a folder and output the resultant zip to the same folder. :-) – Charlatanism 23/3, 2017 at 4:22

shutil makes it really easy in just a single line. Please check the answer below.. – Neace 30/5, 2017 at 18:24

you may be more interested by doing ziph.write(os.path.join(path,file), arcname=file) so that the filenames inside the archive are not relative to the hard drive – Elka 14/6, 2017 at 7:42

I am getting "The compressed (zipped) folder is empty" on Windows 7 – Deka 22/1, 2018 at 20:38

Ah, I was missing the .close() call! – Deka 22/1, 2018 at 21:4

@RoachLord: file is not a keyword, even if your editor highlights it as one. – Aquitaine 2/11, 2018 at 17:17

May also be used with a context manager: with zipfile.ZipFile('Python.zip', 'w', zipfile.ZIP_DEFLATED) as zipf: zipdir('tmp/', zipfh) – Normally 11/11, 2020 at 11:11

The next answer if more straightforward, and should have been accepted – Vagus 18/3, 2021 at 13:20

What is ziph in this function? path is the directory. What if I have a directory containing other sub-directories? I added a question related to this. #67120176 – Trefor 16/4, 2021 at 11:12

The next answer does not handle anything related to compression, it just stores the results in a ZIP archive. If you are looking for actual compression, the correct answer is this one and not the next with shutil. – Carnify 11/7, 2021 at 16:13

I want to add that this option lets you make a zip archive with any file extension. i.e. zipfile.ZipFile('Python.meta_zip', 'w') At my work we have to group a metadata dir with files in it and the file they are associated with together. We don't need compression, we just want to easily have them all in one location when we upload them to the cloud. – Ashcroft 13/7, 2021 at 21:8

This answer creates a zip containing the directory. How would you change it so that the contents of the directory are added, rather than the directory itself? – Geer 11/8, 2021 at 1:39

Explaining with examples is the only way to explain properly, thanks :) – Flowerpot 29/10, 2021 at 14:55

Using the arcname parameter of the write function will solve the issue where the entire directory branch is getting zipped rather than just the contents. – Surgery 21/4, 2022 at 15:21

This will not write empty directories to the zip file. To fix this you can also write the root in every os.walk iteration. – Wirer 1/6, 2023 at 9:26

@Surgery and @Wirer do u have an example which accomplish these tasks i.e. writing empty directories and contents of the directory. because i am confused what should we assign to arcname – Rebate 7/6, 2023 at 20:9

ok the second argument in ziph.write should be modified to os.path.relpath(os.path.join(root, file), path) to add the contents of the directory not the entire directory. the only thing remaining is how to write empty directories. @Wirer do u have an example which shows to add empy directories – Rebate 7/6, 2023 at 21:15

@SajjadAli os.mkdir() – Wirer 27/6, 2023 at 14:12

G

2159

The easiest way is to use shutil.make_archive. It supports both zip and tar formats.

import shutil
shutil.make_archive(output_filename, 'zip', dir_name)

If you need to do something more complicated than zipping the whole directory (such as skipping certain files), then you'll need to dig into the zipfile module as others have suggested.

Geothermal answered 3/9, 2014 at 17:26 Comment(21)

shutil is part of the standard python library. This should be the top answer – Sidran 28/4, 2017 at 19:19

This is the most concise answer here and also has the advantage of adding all subdirectories and files to the archive directly, rather than having everything included in a top-level folder (which results in a redundant level in the folder structure when unzipping). – Affiant 22/6, 2017 at 8:5

@cmcginty could you please be a bit more specific as to what aspect of it is not thread-safe? Will running multiple threads while one calls this cause the interpreter to crash? – Josettejosey 4/11, 2017 at 22:49

@std''OrgnlDave No, it will not crash. The issue I've seen is that files from different threads will be copied into the same zip file. See bugs.python.org/issue30511 – Lefthand 6/11, 2017 at 7:57

Be warned that prior to Python 3.4, shutil.make_archive does not support ZIP64 and will fail on creating ZIP files larger than 2GB. – Eklund 18/1, 2018 at 23:42

Even though it's not thread-safe, it should still be safe if being used by different threads in different directories, right? – Giggle 25/5, 2018 at 21:30

@Giggle No. If you look at the bugs report (bugs.python.org/issue30511), you'll see that shutil.make_archive uses os.chdir(). From what I'm reading about os.chdir(), it operates globally. – Barthelemy 25/7, 2018 at 0:36

@azdev, what is the solution for shutil.make_archive for files more than 2GB? I had this error( – Narbada 3/9, 2018 at 9:8

@Narbada See the accepted answer or one of the other answers on this page that uses the zipfile library instead of shutil. – Eklund 4/9, 2018 at 12:56

shutil is very, very slow. So, for it is not good option for dirs with large files. – Stopgap 14/10, 2018 at 22:54

@azdev For python version 3.6 to 3.7 does shutil work safely for files larger than 2GB? – Jointer 17/1, 2020 at 8:28

If I want to zip a directory instead of the content of directory - is there an option for that? Aslo: can't find what's the equivalent of --remove-files? – Escorial 15/4, 2020 at 11:16

@Escorial see base_dir option – Supersonic 26/6, 2020 at 10:5

I'm trying to automate zipping my work directory which is ~200MB as daily backup, this code kept working and made +84GB and threw an error because my work PC ran out of storage! – Charry 7/11, 2020 at 10:57

Note this is NOT threadsafe! This does an explicit os.chdir which can lead to the wrong directory getting zipped. This led to my zips being gigantic and my disk running out of space. – Kee 1/3, 2021 at 15:57

for an actual example using both root_dir and base_dir, see: docs.python.org/3/library/… – Elodiaelodie 4/3, 2021 at 12:14

This is not safe,this fails for directories with huge files. – Durga 10/3, 2021 at 13:37

This answer is not correct. It doesn't compress anything, plus is slower. the accepted answer is correct and should be the reference. – Carnify 11/7, 2021 at 16:14

Does anyone know how to get files referenced by a symlink to copy into the zip archive? When I unzip the created archive a directory for the symlink files is there but the directory is empty. – Wheelhouse 5/10, 2021 at 18:46

can we get to 2023? – Endolymph 9/3, 2023 at 15:15

@Kee Looks like this is thread-safe now for normal use. From the docs: Changed in version 3.10.6: This function is now made thread-safe during creation of standard .zip and tar archives. – Rame 20/11, 2023 at 20:8

T

718

As others have pointed out, you should use zipfile. The documentation tells you what functions are available, but doesn't really explain how you can use them to zip an entire directory. I think it's easiest to explain with some example code:

import os
import zipfile
    
def zipdir(path, ziph):
    # ziph is zipfile handle
    for root, dirs, files in os.walk(path):
        for file in files:
            ziph.write(os.path.join(root, file), 
                       os.path.relpath(os.path.join(root, file), 
                                       os.path.join(path, '..')))

with zipfile.ZipFile('Python.zip', 'w', zipfile.ZIP_DEFLATED) as zipf:
    zipdir('tmp/', zipf)