Possible regression with detecting and escaping raw latex in markdown when using \iffalse

[cderv]

Opening this after an answered question on pandoc-dicuss: https://groups.google.com/g/pandoc-discuss/c/nylIR9ZCSRg

Explain the problem.

I think this is a regression as it was supported before but is no more.

In our markdown doc, we have some LaTeX syntax using \iffalse{}\fi

\emph{Some text}\iffalse{Some ignored comment}\fi{}

With previous pandoc version, it was correctly seen as raw LaTeX and kept as is.

❯ cd C:\Users\chris\AppData\Local\r-pandoc\r-pandoc\2.13\

❯ echo "\emph{Some text}\iffalse{Some ignored comment}\fi{}" | ./pandoc -t latex
\emph{Some text}\iffalse{Some ignored comment}\fi{}

❯ echo "\emph{Some text}\iffalse{Some ignored comment}\fi{}" | ./pandoc -t native
[Para [RawInline (Format "tex") "\\emph{Some text}",RawInline (Format "tex") "\\iffalse{Some ignored comment}",RawInline (Format "tex") "\\fi{}"]]

but with newer pandoc it gets escaped (with nightly from today 1f1a30b) because not seen as Raw TeX I think.

❯ cd C:\Users\chris\AppData\Local\r-pandoc\r-pandoc\nightly\

❯ echo "\emph{Some text}\iffalse{Some ignored comment}\fi{}" | ./pandoc -t latex
\emph{Some text}\textbackslash iffalse\{Some ignored comment\}\fi{}

❯ echo "\emph{Some text}\iffalse{Some ignored comment}\fi{}" | ./pandoc -t native
[Para [RawInline (Format "tex") "\\emph{Some text}",Str "\\iffalse{Some",Space,Str "ignored",Space,Str "comment}",RawInline (Format "tex") "\\fi{}"]]

Second part which got escaped is not seen as RawInline anymore.

It seems to behave that way since Pandoc 2.14

❯ cd C:\Users\chris\AppData\Local\r-pandoc\r-pandoc\2.14

❯ echo "\emph{Some text}\iffalse{Some ignored comment}\fi{}" | ./pandoc -t latex
\emph{Some text}\textbackslash iffalse\{Some ignored comment\}\fi{}

Is this an expected change ? I did not find a related one in the Release notes.

Pandoc version?

• Found in Pandoc 2.14
• Still present in Pandoc nightly 1f1a30b
• Working in Pandoc 2.13

[cderv]

It seems there was a change in pandoc 2.14 that indeed changed this behavior. I found a related issue #6096 where expected behavior was ok in Pandoc 2.13 and not after

For the record here the native representation formatted

Pandoc 2.13

[Para 
    [RawInline (Format "tex") "\\emph{Some text}"
    ,RawInline (Format "tex") "\\iffalse{Some ignored comment}"
    ,RawInline (Format "tex") "\\fi{}"]
]

Current version (2.17.0.1)

[ Para
    [ RawInline (Format "tex") "\\emph{Some text}"
    , Str "\\iffalse{Some"
    , Space
    , Str "ignored"
    , Space
    , Str "comment}"
    , RawInline (Format "tex") "\\fi{}"
    ]
]

So the content is no more identified as RawInline it seems

[jgm]

This does seem like a bug, I'd have to look into it. But you can always work around this kind of thing using "raw attribute" syntax:

`\emph{Some text}\iffalse{Some ignored comment}\fi{}`{=latex}