Too strict markdown escape in PythonNews

[Bluenix2]

The bot can be seen escaping underscores in code in this message.

The regex (if possible) should be updated to not insert \ inside codelines or codeblocks such as (in this case) __eq__.

On the bright side, it did work for one ocasion I've screenshoted below!

[TizzySaurus]

The regex (if possible) should be updated not to insert \ inside codelines or codeblocks

It should just be a matter of update the regex here.

>>> string = "Hello world `__repr__` how are _you_? I'm good _today_, but __tomorrow__ may be different, `_User`."
>>> reg = r'(?:`[^`\\]*(?:\\.[^`\\]*)*`)|([_|])'  # will get underscores and pipes that aren't in backquotes
>>> to_escape = [m.span()[0] for m in re.finditer(reg, string) if m.group(1)]  # indexes of said underscores and pipes (the chars we need to escape)
>>> to_escape
[31, 35, 47, 53, 60, 61, 70, 71]
>>> new_str = ""
>>> for index, char in enumerate(string):
	if index in to_escape:
		new_str += "\\" + char
	else:
		new_str += char

		
>>> print(new_str)
Hello world `__repr__` how are \_you\_? I'm good \_today\_, but \_\_tomorrow\_\_ may be different, `_User`.

I believe this is the rough behaviour wanted, and from quick testing it does seem to also work with multi-line codeblocks.

Issues with this regex:

• If an underscore / pipe is already escaped then it'll be escaped again (same behaviour as currently, but may want to change)

>>> print(new_str)
Hello world `__repr__` how are \\_you\\_?

This issue could probably be fixed by changing the final if statement to something along the lines of if index in to_escape and string[index-1] != "\\": but don't have time to test this.

I'd be interested in making the PR for this once it's confirmed this is the wanted behaviour.

[TizzySaurus]

I've realised a look-behind can be used to fix the issue of escaping pre-escaped markdown and @Bluenix2 has confirmed this is the wanted behaviour on discord so will work on the PR.