Guess the output format based on the --output argument

[Deuchnord]

(Feature request by @AmauryCarrade on Discord)

Is your feature request related to a problem? Please describe.
Currently, to export in a PDF file, we need to give two arguments to the command:

• --format=pdf
• --output=myfile.pdf
  Giving both arguments, even though acceptable, is redundant.

Describe the solution you'd like

• The --format argument becomes optional.
• If both --format and --output is given, then the behavior stays the same as today: the content is output in the given file using the given format, even though the extension does not match the format.
• If only --output is given, then guess the format to use from it:
  • --output=myfile.txt ou --output=myfile (without extension) will output a file containing the text usually returned in standard output
  • --output=myfile.json will output a file containing a JSON content
  • --output=myfile.pdf will output a file containing the PDF content
• If only --output is given, but the file format is not recognized, then an error is returned and nothing is done.

[github-actions]

This issue has not been active since a lot of time. If you think it is still valid, feel free to reply with a simple comment. Without any update, I will close it in 7 days.

[Deuchnord]

Nope, you don't want to close that. 👀

[github-actions]

This issue has not been active since a lot of time. If you think it is still valid, feel free to reply with a simple comment. Without any update, I will close it in 7 days.

[nicfb]

I see that you self-assigned this issue. Are you currently working on this? If not, I would love to give it a shot.

[Deuchnord]

Hello, no, I haven't started it yet, you can take it.
Thank you! 😄

[nicfb]

I am having some issues running using the different output types, specifically on this line. When I use json or text, I have to remove the "b" from the arguments so the file isn't opened in binary mode or else I see this error: TypeError: a bytes-like object is required, not 'str'. I have to add it back to use the pdf output format or else I see the same error. Am I doing something wrong?

[Deuchnord]

No, I think I see what you are facing. It is a bit tricky, since the PDF format is a binary format, but not the plain text and JOSN format.
I think the most elegant way to fix this would be to create a small function you would call in the second argument of open() and that would return the correct mode depending on the output format.
Something like this:

def get_opening_mode(format: str) -> str:
    if format == "pdf":
        return "wb"

    return "w"

WDYT?

[nicfb]

Ok, just wanted to make sure. I like the solution you suggested!

[nicfb]

I have something working now. I think all that is left is to add some tests. Would I add these in the end to end test script?

[Deuchnord]

I think you can add them just after the existing export ones :)

[Deuchnord]

Btw @nicfb, be careful, there is a license change, Kosmorro is rolling back to GNU AGPL instead of CeCILL (#196).
Your PR can stay under CeCILL if you want (AFAIK it is AGPL-compatible), just making sure you are aware about that :)