When doing replacement, why does $1 www$3 yield expected results but $1www$3 does not?

[Goobyalus]

Here is a simplified example of what I'm confused by:

# Example input file
➜  ~/Desktop/temp cat rg_test.txt 
https://one.potato.com/123456
https://two.potato.com/891011

# This makes sense to me, we replace the line with match 1 + match 2 + match 3.
➜  ~/Desktop/temp rg -N "(https://)(.*)(\.potato.com/.*)" rg_test.txt -o -r '$1$2$3' 
https://one.potato.com/123456
https://two.potato.com/891011

# I am attempting to replace the subdomain here with "www", so match 1 + "www" + match 3.
# It seems that it's tokenized as `$1www`, and there's no match named "1www", so it becomes
# "" + match 3.
➜  ~/Desktop/temp rg -N "(https://)(.*)(\.potato.com/.*)" rg_test.txt -o -r '$1www$3'
.potato.com/123456
.potato.com/891011

# With a space it seems to delimit the token and interpret the " www" as a literal and not part of the $1.
➜  ~/Desktop/temp rg -N "(https://)(.*)(\.potato.com/.*)" rg_test.txt -o -r '$1 www$3'
https:// www.potato.com/123456
https:// www.potato.com/891011

Is there some escape to delimit the tokens so I can get $1 + www + $3?

[BurntSushi]

This is answered in the documentation. This case is specifically called out in the help for the -r/--replace flag:

-r REPLACEMENT, --replace=REPLACEMENT
    Replaces every match with the text given when printing results. Neither
    this flag nor any other ripgrep flag will modify your files.

    Capture  group  indices (e.g., $5) and names (e.g., $foo) are supported
    in the replacement string. Capture group indices are numbered based  on
    the  position  of the opening parenthesis of the group, where the left‐
    most such group is $1. The special $0 group corresponds to  the  entire
    match.

    The  name of a group is formed by taking the longest string of letters,
    numbers and underscores (i.e. [_0-9A-Za-z]) after the $.  For  example,
    $1a will be replaced with the group named 1a, not the group at index 1.
    If the group's name contains characters that aren't letters, numbers or
    underscores,  or  you want to immediately follow the group with another
    string, the name should be put inside braces. For example,  ${1}a  will
    take the content of the group at index 1 and append a to the end of it.

    If an index or name does not refer to a valid capture group, it will be
    replaced with an empty string.

    In  shells  such as Bash and zsh, you should wrap the pattern in single
    quotes instead of double quotes. Otherwise, capture group indices  will
    be  replaced  by  expanded  shell  variables  which will most likely be
    empty.

    To write a literal $, use $$.

    Note that the replacement by default replaces each match, and  not  the
    entire  line.  To  replace the entire line, you should match the entire
    line.

    This flag can be used with the -o/--only-matching flag.

Specifically, the third paragraph answers your question.

[Goobyalus]

Thank you.

For what it'sworth, I did check the help and the user guide. My help has this:

 -e, --regexp <PATTERN>...               
            A pattern to search for. This option can be provided multiple times, where
            all patterns given are searched. Lines matching at least one of the provided
            patterns are printed. This flag can also be used when searching for patterns
            that start with a dash.
            
            For example, to search for the literal '-foo', you can use this flag:
            
                rg -e -foo
            
            You can also use the special '--' delimiter to indicate that no more flags
            will be provided. Namely, the following is equivalent to the above:
            
                rg -- -foo
             
    -r, --replace <REPLACEMENT_TEXT>        
            Replace every match with the text given when printing results. Neither this
            flag nor any other ripgrep flag will modify your files.
            
            Capture group indices (e.g., $5) and names (e.g., $foo) are supported in the
            replacement string. In shells such as Bash and zsh, you should wrap the
            pattern in single quotes instead of double quotes. Otherwise, capture group
            indices will be replaced by expanded shell variables which will most likely
            be empty.
            
            To write a literal '$', use '$$'.
            
            Note that the replacement by default replaces each match, and NOT the entire
            line. To replace the entire line, you should match the entire line.
            
            This flag can be used with the -o/--only-matching flag.
             
    -z, --search-zip                        
            Search in compressed files. Currently gzip, bzip2, xz, LZ4, LZMA, Brotli and
            Zstd files are supported. This option expects the decompression binaries to be
            available in your PATH.
            
            This flag can be disabled with --no-search-zip.

and I read the user guide here: https://github.com/BurntSushi/ripgrep/blob/master/GUIDE.md#replacements

Neither mentions this.

Which documentation are you referring to? Is this because I'm on an oldler version (12.1.0), or am I just looking in the wrong place?

[BurntSushi]

Your version of ripgrep is many years old. A number of improvements and new features, including to the docs, have been released since then.

When filing issues or asking questions, please include the version of the tool you're using. And if possible, please update your version first. I don't provide support for anything other than the latest version of ripgrep.

[Goobyalus]

Thank you,, I updated and see the current documentation now!