[*nix] Convert man pages to EPUB?

[Kurler]

Originally Posted by ratinox

groff uses mandoc underneath with the eponymous '-mandoc' switch.

No, “mandoc” is not the same thing as “groff -mandoc”. Both tools can convert manpages to HTML, but they are different programs written by different authors and produce different HTML. “groff -mandoc” refers to the input filetypes it supports (the two common manpage languages “man” and “mdoc”), as does the program name “mandoc,” but they’re not otherwise related. mandoc’s HTML is better quality, making it the tool of choice for this job.

[Shohreh]

Originally Posted by Kurler

I recommend using the program “mandoc” (https://mandoc.bsd.lv/) to convert a manpage from source to HTML. It was designed from the start specifically as a manpage-to-HTML converter, so it does a better job of realizing semantics from the manpage, resulting in better quality HTML than pandoc or groff.

Thanks for the info. I'll try it next time.

mandoc -Thtml /usr/share/man/man1/ls.1.gz > ls.html

Turns out pandoc supports man pages too:

pandoc --from man --to html < input.1 > output.html

[ratinox]

Originally Posted by Kurler

No, “mandoc” is not the same thing as “groff -mandoc”. Both tools can convert manpages to HTML, but they are different programs written by different authors and produce different HTML. “groff -mandoc” refers to the input filetypes it supports (the two common manpage languages “man” and “mdoc”), as does the program name “mandoc,” but they’re not otherwise related. mandoc’s HTML is better quality, making it the tool of choice for this job.

Are you sure about that? I was under the impression that groff uses the mandoc(1) program or code when the -mandoc switch is used. I do know that standalone mandoc on BSD is designed specifically and only for man pages where groff and other roff-alikes are general purpose markup parsers, so if groff isn't using mandoc(1) then that would be a reason for the differences.

[Kurler]

Originally Posted by ratinox

Are you sure about that?

Yes, I am sure.

Originally Posted by ratinox

I was under the impression that groff uses the mandoc(1) program or code when the -mandoc switch is used.

That is incorrect. groff has been used for manpages since the early nineties, and the mandoc program was only created in 2008. But even current groff has never used mandoc in that way, or any way.

Originally Posted by ratinox

I do know that standalone mandoc on BSD is designed specifically and only for man pages where groff and other roff-alikes are general purpose markup parsers, so if groff isn't using mandoc(1) then that would be a reason for the differences.

Yes, this is reasonably accurate. To groff, the -mandoc switch is just like any of its other macro language switches like -ms or -mm; it loses track of the manpage semantics along the way which has a negative impact on its HTML.

[ratinox]

Cool. Thank you for the clarification.