Would it be possible to add the value of c in this message as well ?

This error is raised in the case if an underscore or 0o is not followed by a digit. What error messages could be helpful for 0o+2, 0o + 2, (2+0o), 0or[]?

I don't know, I'd expect "invalid character '%c' in octal literal" to be useful in all cases.

It is easy to report only if an invalid digit (in the range 2-9 or 8-9) is occurred. In general case there are much subtle details, handling them will complicate the code too much:

• Not always an invalid character exists. This error can be raised at the end of the input.
• It can be non-ASCII. In this case we need to decode a multibyte UTF-8 for getting a character.
• It can be non-printable.
• Even if it is printable from the Unicode's point of view, it can look indistinguishably from other characters. For example, non-breakable space character looks like an ordinary space for humans, but not for the Python parser.
• Even in ASCII there are non-printable characters, or characters that need special handling: tab, newline, single quote, backslash, ...

It may be worth to produce more specialized error message for some cases, but just reporting the next invalid character is no a way.

Ho, indeed, I didn't think about all these issues...