Python pandas to_csv zip format

问题

I am having a peculiar problem when writing zip files through to_csv.

Using GZIP:

df.to_csv(path_or_buf = 'sample.csv.gz', compression="gzip", index = None, sep = ",", header=True, encoding='utf-8-sig')

gives a neat gzip file with name 'sample.csv.gz' and inside it I get my csv 'sample.csv'

However, things change when using ZIP

df.to_csv(path_or_buf = 'sample.csv.zip', compression="zip", index = None, sep = ",", header=True, encoding='utf-8-sig')

gives a zip file with name 'sample.csv.zip', but inside it the csv has been renamed to 'sample.csv.zip' as well. Removing the extra '.zip' from the file gives the csv back.

How can I implement zip extension without this issue? I need to have zip files as a requirement that I can't bypass. I am using python 2.7 on windows 10 machine.

Thanks in advance for help.

回答1:

It is pretty straightforward in pandas since version 1.0.0 using dict as compression options:

filename = 'sample'
compression_options = dict(method='zip', archive_name=f'{filename}.csv')
df.to_csv(f'{filename}.zip', compression=compression_options, ...)

https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_csv.html

回答2:

As the thread linked in the comment discusses, ZIP's directory-like nature makes it hard to do what you want without making a lot of assumptions or complicating the arguments for to_csv

If your goal is to write the data directly to a ZIP file, that's harder than you'd think.

If you can bear temporarily writing your data to the filesystem, you can use Python's zipfile module to put that file in a ZIP with the name you preferred, and then delete the file.

import zipfile
import os

df.to_csv('sample.csv',index=None,sep=",",header=True,encoding='utf-8-sig')

with zipfile.ZipFile('sample.zip', 'w') as zf:
    zf.write('sample.csv')
os.remove('sample.csv')

来源：https://stackoverflow.com/questions/55134716/python-pandas-to-csv-zip-format