Implement glob support for Windows shells (which don't do this themselves)

[tc-ying]

Is there a way to group " *.csv " in a directory by --filenames ? If I pass the CSVs into csvstack using stdin, then their group names become .

Forgive this newbie question.

[jpmckinney]

Can you share the complete command that you're running?

[tc-ying]

Since this is not permissible — csvstack --filenames "*.csv"

Let's do it in windows command : type "to_append*.csv" | csvstack --filenames "master.csv" -

Output is something like :

group,header1,header2
master.csv,col1,col2
,col1,col2
,col1,col2

Can't figure how to stack a batch of CSVs at one command.

[jpmckinney]

Why isn't csvstack --filenames "*.csv" permissible?

If the file comes from stdin, then it doesn't have a filename. That's just UNIX; csvkit can't fix that.

[tc-ying]

Using wildcard file names generates "error: You must specify at least two files to stack."

Its seems most csvkit commands do not accept wildcard file arguments. It makes batch csvstack less convenient coz I can't figure out a FOR... DO loop for csvstack.

[jpmckinney]

That error no longer exists in the current code base. Try the latest csvkit:

pip install --upgrade -e git+git://github.com/onyxfish/csvkit.git@master#egg=csvkit

I can re-open if it still doesn't work with the latest version.

[tc-ying]

It now works with ONE argument 👍 , but if that argument is :

INPUT – csvstack "*.csv"
OUTPUT – [Errno 22] Invalid argument: '*.csv'

Accepting wildcard argument seems to require fundamental program change.

[jpmckinney]

What operating system are you on?

[tc-ying]

UPDATE :

Windows command prompt seems not to support wildcard argument. An option to stack all CSVs under a directory is most handy for Windows...

[tc-ying]

users may switch to run .sh scripts in windows...

shopt -s nullglob
files=(*.csv)
csvstack "${files[@]}" > merged.csv

done.

[ilovefreesw]

2024 and still doesn't work

[jpmckinney]

We can do something like Click did: https://github.com/pallets/click/pull/1830/files