Make Base64 decoding strict by default

[serhiy-storchaka]

Feature or enhancement

Earlier specifications of Base64 encoding (e.g. RFC 1421, RFC 2045) allowed to ignore any non-alphabetical characters, and that was the behavior of the base64 module. The later specifications (RFC 3548, RFC 4648) require that non-alphabetic characters should be errors, unless opposite is explicitly specified. The Base64 decoder in Python got the option to enable more strict behavior, and now it got option to specify what exactly characters can be ignored.

I think that we should make Base64 decoding strict by default. I am not sure when we should do this, in 3.15 or in future versions. There is no reasonable way to use warnings, they would do more harm.

cc @gpshead

[serhiy-storchaka]

standard_b64decode() and urlsafe_b64encode() do not have validate and ignorechars parameters. But decoding should be strict if it is standard. If you need something non-standard, you can use b64decode().