Wrong charset handling in MailSender

Looks like passing charset='utf-8' makes a plain text message with Content-Transfer-Encoding: base64 which then can’t be read.

At https://github.com/scrapy/scrapy/blob/5a75b14a5fbbbd37c14aa7317761655ac7706b70/scrapy/mail.py#L81 set_charset is called but as the payload is not set yet, the underlying class just sets the headers. When later set_payload is called, it doesn’t do any encoding, but the Content-Transfer-Encoding is already set. Looks like the fix should be passing the encoding to set_payload too, like was proposed in #3722 (and set_charset may be safe to be removed, not sure about this). Note that we have tests but they don’t catch this.

Note also, that all of this seems to be compat code according to the Python docs.

[Gallaecio]

Do you want me to do a pull request or maybe a draft pull request to see what I have done and give me feedback?

That would be great! Feel free to make it a regular pull request if it’s ready for review and discussion.

[Gallaecio]

Can you please give me some guidlines and tell me how to recreate this issue?

The documentation about using the MailSender class is at https://docs.scrapy.org/en/latest/topics/email.html . Then it should be a matter of following the instructions on the issue description (“passing charset=‘utf-8’”), and verify that indeed the resulting email message cannot be read.