gsub -> tr cop messes up escaped string

[cheerfulstoic]

First off, thanks for rubocop!

So if I have a file like this:

"".gsub('/', '\\')

rubocop (0.33.0 as well as GitHub master) transforms it to this:

"".tr('/', '\')

This isn't good because Ruby sees the second single quote as being escaped and there syntax errors.

Here's the rubocop output:

$ rubocop -aD test.rb
Inspecting 1 file
F

Offenses:

test.rb:1:4: C: [Corrected] Performance/StringReplacement: Use tr instead of gsub.
"".gsub('/', '\\')
   ^^^^^^^^^^^^^^^
test.rb:1:12: F: unterminated string meets end of file
"".tr('/', '\')
           ^

1 file inspected, 2 offenses detected, 1 offense corrected

rubocop -V:

$ rubocop -V
0.33.0 (using Parser 2.2.2.6, running on ruby 2.0.0 x86_64-darwin14.3.0)

You can also see my .rubocop.yml and .rubocop_todo.yml here:

https://gist.github.com/cheerfulstoic/062fa3b79e8397b82c19

[bbatsov]

/cc @rrosenblum

[rrosenblum]

Thanks for the bug report. Escape characters tend to get nasty when dealing with auto-correct. I have a test case for "\n", but I guess that isn't good enough. I will look into a fix for this.

[cheerfulstoic]

Awesome, thanks ;)

[cheerfulstoic]

🤘 Thanks

[aevernon]

I see the behavior below on master:

$ git branch -v
* master a584c06
$ cat test.rb
''.gsub('\\', '')
$ rubocop -aD test.rb
Inspecting 1 file
F

Offenses:


test.rb:1:4: C: [Corrected] Performance/StringReplacement: Use delete instead of gsub.
''.gsub('\\', '')
   ^^^^^^^^^^^^^^
test.rb:1:11: F: unterminated string meets end of file
''.delete('\')
          ^

1 file inspected, 3 offenses detected, 2 offenses corrected

[rrosenblum]

Good catch. '\\' seems to be a special case compared to other escaped strings. I will find a fix.