Do not register an offense for UnneededPercentQ when %q or %Q occurs in an interpolated string

[rrosenblum]

I came across a strange issue with UnneededPercentQ earlier. It would register an offense for a double quoted string that uses interpolation and has a section that begins with %q or %Q.

example offenses

"%q#{foo}"
"%Q#{foo}"
"#{foo}%q"
"#{foo}%Q"

[bbatsov]

Hmm, that's odd. A string with interpolation is actually a dstr and we have a check for it in on_dstr. I guess we can do more or less the same check here.

[rrosenblum]

I was surprised by this as well. Given the example string "%q(a)#{b}", this cop was called twice. Once for dstr with "%q(a)#{b}", and a second time for str with the substring "%q(a)". It seemed weird to me that this was the behavior. That is why I had to add in the check to see if the string is a substring.

[bbatsov]

This seems like some parser bug. You should check the generated AST.

[bbatsov]

Scratch this, it's not really a bug of course.

[bbatsov]

I'll think about this a bit more - at the very least standalone_string? should be string_literal? and there should probably be some explanation about why this check is needed. The cop could use a better class comment in general. Probably using the same check method for both cases wasn't a great idea either.

[rrosenblum]

I will make the change to the method name and add comments. I will also look into refactoring this cop to implement different checks.

[bbatsov]

Great.

[rrosenblum]

I think that the source code is about as clean as it is going to get. It turns out that my fix also handles the case for regex. I tried to extract the code for dstr and str but the necessary code was nearly identical. Let me know what you think of the changes.

[bbatsov]

Looks OK to me.