Proposed Check: Unintentional str concat

[jamesstidard]

One of the common mistakes I can make is to miss out a comma in collections of literal strings, resulting in the strings being concatenated.

my_list = ["a", "b", "c" "d"]  # note the missing comma between 'c' and 'd'
print(my_list)  # ["a", "b", "cd"]

I've not come across a time I've ever used this intentionally, and will either explicitly write "c" + "d" or use a format string "{a} {b} when a and b are variables.

If you intend this using brackets might be a fine way. my_list = ["a", "b", ("c" "d")]. Similar to the trailing comma tuple mistake where flake8-comma flags my_int = 5, but allows the more explicit my_int = (5,).

The above would mean multiline strings could be concerted still as:

my_multiline = (
    "part one "
    "part two"
) 

As soon as a comma is introduced though it becomes a tuple, and I'd argue each str literal should have a comma between it or none should.

[cooperlees]

I have also personally hit this and would love this added to bugbear.

I like the using parens "()" to make this non + concatting explicit, although I would prefer people just always use '+'. I bet this will cause a lot of noise to our users but will find bugs.

If you have the cycles, I'd take a PR, but would request very good unit testing here to ensure we're covering all of our edge cases please!

[kkirsche]

+1 for me

In the past, I've used https://github.com/Instagram/Fixit to identify things like this. Reducing the tools necessary would be great, and they may have an example implementation that can be used as a foundation.

Legacy rule they had for preventing implicit concatenation: https://github.com/Instagram/Fixit/blob/c3f8ff9e6cfa2199c6154eefea24ea39e5278dc6/legacy/rules/no_implicit_concat.py