Multiline with replace (and passthru) duplicates lines from the input in the output

[janickr]

What version of ripgrep are you using?

ripgrep 13.0.0 -SIMD -AVX (compiled) +SIMD +AVX (runtime)

How did you install ripgrep?

homebrew: brew install ripgrep

What operating system are you using ripgrep on?

MacOS Big Sur

ProductName: macOS ProductVersion: 11.6 BuildVersion: 20G165

Describe your bug.

Give a high level description of the bug.

What are the steps to reproduce the behavior?

When performing a multiline search and replace with passthru, rg duplicates lines in the output

What is the actual behavior?

 ~$ seq 0 9 | rg -U  -e '1\n2' --passthru -r 'x' --debug                                                                      
DEBUG|globset|crates/globset/src/lib.rs:421: built glob set; 0 literals, 0 basenames, 12 extensions, 0 prefixes, 0 suffixes, 0 required extensions, 0 regexes
0
x
3
4
5
6
7
8
9
3
4
5
6
7
8
9

What is the expected behavior?

I expect only 1\n2 to be replaced by x and the rest of the input to be left unchanged

0
x
3
4
5
6
7
8
9

A clue, maybe

When invoking the same command without passthru and replace has the expected output:

~$ seq 0 9 | rg -U  -e '1\n2$'                                                                                                          
1
2

But invoking without passthru but with replace also outputs too many lines

~$ seq 0 9 | rg -U  -e '1\n2$' -r 'x'                                                                                                         
x
3
4
5
6
7
8
9

I would expect it to output only the replacement:

x

[balupton]

Seems the --passthru isn't necessary to reproduce this, as I'm getting it with just --multiline and --replace:

https://github.com/BurntSushi/ripgrep/issues/2095

[gthb]

A second, maybe more motivating example: I wanted to improve on jq's pretty-printing by postprocessing it to put single-property objects on one line:

rg --passthru -U '\{\n\s*("[^"]*":[^,\n]+)\n\s*\}' --replace '{ $1 }'

but that hits this problem, so with this input:

{
  "foo": [
    {
      "foo": [
        {
          "bar": "x"
        },
        {
          "bar": "y"
        }
      ]
    },
    {
      "bar": "z"
    }
  ]
}

I get:

{
  "foo": [
    {
      "foo": [
        { "bar": "x" },
        { "bar": "y" }
      ]
    },
    { "bar": "z" }
  ]
}
      ]
    },
    { "bar": "z" }
  ]
}
  ]
}

[BurntSushi]

Hmmm, I wonder if this is a duplicate of #2095? If so, it might be fixed on master?

[gthb]

Hmmm, I wonder if this is a duplicate of #2095? If so, it might be fixed on master?

Well, yep! Tried out:

git clone [email protected]:BurntSushi/ripgrep.git
cargo install --path ripgrep
~/.cargo/bin/rg --passthru -U '\{\n\s*("[^"]*":[^,\n]+)\n\s*\}' --replace '{ $1 }' rgbug.json | cat

Sure enough, that outputs:

{
  "foo": [
    {
      "foo": [
        { "bar": "x" },
        { "bar": "y" }
      ]
    },
    { "bar": "z" }
  ]
}

[BurntSushi]

Sweet!