dtx file support?

[OopsYao]

I'd love to use this for formatting my .dtx file, but it cannot handle the comment indent properly. And I could not find any docs about the comment indent nor the dtx support.
Suppose I have a demo.dtx file with content

%    \begin{macrocode}
\def\somecommand{
  do something
%    \end{macrocode}
% Some notes for the rest fragment
%    \begin{macrocode}
  do another thing
}
%    \end{macrocode}

After I run

latexindent -y="defaultIndent:'  '" demo.dtx | diff demo.dtx -

It gives

4,6c4,6
< %    \end{macrocode}
< % Some notes for the rest fragment
< %    \begin{macrocode}
---
>   %    \end{macrocode}
>   % Some notes for the rest fragment
>   %    \begin{macrocode}

But I don't want the doc comment to be indented. So I wonder if there is an option to ignore comment indent for .dtx file?

[cmhughes]

Take a look at the replacement switch, you should be able to remove leading space in front of comment symbols after indenting.

Let me know how it goes.

[OopsYao]

Thanks for your quick reply, it works, though not perfectly.
With localSettings.yaml

defaultIndent: "  "
replacements:
  - substitution: s/^ +%/%/gm
    when: after

then

latexindent -l -r demo.dtx | diff demo.dtx -

would be ok. The regex is simple, but also trims the leading space of comments inside macro code

% ^^A demo2.dtx
%    \begin{macrocode}
\def\somecommand{
  % comment inside macrocode
  do something
%    \end{macrocode}
% Some notes for the rest fragment
%    \begin{macrocode}
  do another thing
}
%    \end{macrocode}

4c4
<   % comment inside macrocode
---
> % comment inside macrocode

In most cases, the dtx doc code would be comment lines between % \end{macrocode} and % \begin{macrocode}. So dealing with those lines is properer. But I could not find a proper regex to only match leading spaces in such pattern (not familiar with perl regex, but I believe it exists. 👀)
And I also tried the specialBeginEnd option to specify indent rules, but it seems not working with lines with leading %.
And the noIndentBlock would be another approach, but it assumes the environment begins with % \begin{macrocode} and ends with % \end{macrocode}, not inversely.

It would be better if it can handle doc code separately, as if there is no leading %, thus supporting .dtx file almostly.

[cmhughes]

Your original post and your response seem to contradict each other.

[OopsYao]

You missed my point. I do not want to trim the leading space of all comment lines, just those between % \end{macrocode} and % \begin{macrocode}.
In the example below, line 4 and line 6-8 should be treated differently. As the former is comment in macrocode, which should be indented as normal, and the latter is the doc code (do no indent).

[cmhughes]

So, just to be completely clear, your desired result is

% ^^A demo2.dtx
%    \begin{macrocode}
\def\somecommand{
  % comment inside macrocode
  do something
%    \end{macrocode}
% Some notes for the rest fragment
%    \begin{macrocode}
  do another thing
}
%    \end{macrocode}

If not, please provide your desired output as text, not a screenshot.

[OopsYao]

Yes it is. (The screenshot was intended for showing line numbers 👀)

[cmhughes]

I don't have a way to do this in a single call, but it can be done in two calls. Starting with your original code, if you run

latexindent.pl -w myfile.tex
latexindent.pl -l oopsyao.yaml -rr myfile.tex

with oopsyao.yaml as follows:

defaultIndent: "  "
replacements:
  - 
    when: after
    substitution: |-
       s/\h*(\%\h*\\end\{macrocode\})
           (.*?)
         \h*(\%\h*\\begin\{macrocode\})/
         my $begin = $1;
         my $body  = $2;
         my $end   = $3;
         $body =~ s|^\h*\%|%|mg;
         $begin.$body.$end;/smgxe

then the output is

% ^^A demo2.dtx
%    \begin{macrocode}
\def\somecommand{
  % comment inside macrocode
  do something
%    \end{macrocode}
% Some notes for the rest fragment
% more comments
%    \begin{macrocode}
  do another thing
}
%    \end{macrocode}

The reason that two calls to latexindent.pl are necessary is because if the indentation mode is active (which it is when anything other than the -rr switch is active) then trailing comments are stored so that latexindent doesn't know what string is contained within the comment.

I hope this helps. For reference, there's also https://tex.stackexchange.com/a/528732/6621. Let me know.

[cmhughes]

I'll close this for the moment, but let me know if you have any other questions.