Failing to parse "\\("

[JesseSchobben-TomTom]

Describe the bug
yq 4.43.1 introduces string interpolation, but it doesn't play nice with escaped braces in a regex.

Version of yq: 4.43.1
Operating system: mac
Installed via: homebrew

Command
The following throws an error:

$ yq -n '"test(" | match("test\\(")'
Error: unclosed interpolation string \(

With a single backslash that failure would be expected, but for the above command I'd expect a successful match reporting "test(".

Expected behavior

Maybe we should follow the behavior of jq here:

$ jq -n '"test(" | match("test\(")'
jq: error: syntax error, unexpected $end, expecting QQSTRING_TEXT or QQSTRING_INTERP_START or QQSTRING_END (Unix shell quoting issues?) at <top-level>, line 1:
"test(" | match("test\(")                        
jq: 1 compile error

$ jq -n '"test(" | match("test\\(")'
{
  "offset": 0,
  "length": 5,
  "string": "test(",
  "captures": []
}

Additional context
As side info, with yq 4.42.1 the single-backslash variant works, but double-backslash fails, so we'll have an incompatibility either way:

$ yq -n '"test(" | match("test\(")'
string: test(
offset: 0
length: 5
captures: []

$ yq -n '"test(" | match("test\\(")'
Error: error parsing regexp: missing closing ): `test\\(`

[mbenson]

I am also encountering issues with this feature, namely that I want to do function-like stuff by saving off eval strings as variables for later use, which means I want to be able to escape a \() construct for evaluation later. I hadn't realized the functionality was so new... is that right that it was just released in a point release? I have a patch that I think will address the issue, but it alters the behavior of one of the unit tests by interpreting \\ constructs in the "normal" way, i.e. by assuming the first backslash escapes the second. I'll send a PR for consideration after I test it a bit.

[mbenson]

"point release" I see there is no .0 😄

[mbenson]

#1997

[mbenson]

@mikefarah merged #1997 over the weekend... I actually never tried your original example, @JesseSchobben-TomTom , but I just have and it seems to work fine.

[mikefarah]

Got @mbenson 's fix in 4.44.1 :)