Text wrapping introduces spaces when trailing comments are present

[marcin-serwin]

original .tex code

\documentclass{article}

\begin{document}

Something%
\index{something}.

\end{document}

yaml settings

modifyLineBreaks:
  textWrapOptions:
    columns: 70

actual/given output

\documentclass{article}

\begin{document}

Something \index{something}.%

\end{document}

desired or expected output

\documentclass{article}

\begin{document}

Something\index{something}.%

\end{document}

or

\documentclass{article}

\begin{document}

Something%
\index{something}.

\end{document}

anything else

The -m switch with text wrapping introduces additional spaces when trailing comments are present. This may change the document in some cases. For example if pagebreak occurs between the Something and \index{something} then the index entry will point to the wrong page.

[cmhughes]

Apologies for the delay, I'll get to this soon hopefully.

[cmhughes]

The default behaviour is for the text wrapping routine of latexindent.pl to join lines using a space; this seems like the most sensible default.

For your desired output, there are at least two options.

option 1 (using blocksEndBefore)

modifyLineBreaks:
  textWrapOptions:
    columns: 70
    blocksEndBefore:
       other: |-
         (?x)
           \\begin\{
             |
           \\\[
             |
           \\end\{
             |
           \\index

and then calling

latexindent.pl -l -m myfile.tex

gives

\documentclass{article}

\begin{document}

Something%
\index{something}.

\end{document}

The above yaml instructs the text wrapping routine to stop at \index.

option 2 (using the replacement switch)

modifyLineBreaks:
  textWrapOptions:
    columns: 70

replacements:
  -
    when: after
    substitution: |-
        s/([a-zA-Z])\h+(\\index)/$1$2/sg

and then calling

latexindent.pl -rv -l -m myfile.tex

gives

\documentclass{article}

\begin{document}

Something\index{something}.%

\end{document}

The replacement switch is used here to remove spaces between letters ([a-zA-Z]) and the \index command.

[marcin-serwin]

Thanks for the quick answer and the proposed solutions. My point was a bit more general though. While it is true that replacing newlines by spaces is a reasonable behavior in most cases, this is not the case if the line is ended by comment. This tells LaTeX that next newline should be ignored, so the latexinndent transformation changes document, not just the source code formatting. While the proposed solutions work fine for lines starting with \index command they require to specify exceptions case by case. In general they wouldn't for example differentiate between these two situations:

foo%
bar

and

foo % this is a comment
bar

which communicate different intent on the part of the author.

[cmhughes]

So what about

\begin{myenv} 
foo%
bar
\end{myenv} 

Which changes to

begin{myenv} 
     foo%
     bar
\end{myenv} 

Again, this will change the output....

[marcin-serwin]

Spaces at the beginning of the line are ignored by LaTeX (unless it is verbatim environment, which can be controlled in localSettings) so it wouldn't change the output.

[cmhughes]

OK, understood. I'll explore this.

[cmhughes]

For my reference :split at comment regex, test for trailing space.

[cmhughes]

As of https://github.com/cmhughes/latexindent.pl/commit/079b219f42c17073325ca56ab8308f83201a2294 the text wrap routine has been updated so that starting with

\documentclass{article}

\begin{document}

Something%
\index{something}.

\end{document}

and using your YAML settings, gives

\documentclass{article}

\begin{document}

Something\index{something}.%

\end{document}

I'll be getting this released and uploaded to ctan.

Thanks for highlighting this!

[cmhughes]

I've released this at https://github.com/cmhughes/latexindent.pl/releases/tag/V3.17.3 and uploaded it to ctan :) Thanks again!

[cmhughes]

For reference, dedicated part of the documentation

https://latexindentpl.readthedocs.io/en/latest/sec-the-m-switch.html#text-wrap-trailing-comments-and-spaces