Support "lhs" as name for literate Haskell

[ad-si]

When converting markdown to markdown+lhs following 2 variants should yield the same output:

``` {.literate .haskell}
main = putStr "Hello World!"
```

```lhs
main = putStr "Hello World!"
```

should both yield

> main = putStr "Hello World!"

[jgm]

The behavior is exactly as documented in the manual, isn't it?

In Markdown [+lhs] output, code blocks with classes haskell and literate will be rendered using bird tracks, and block quotations will be indented one space, so they will not be treated as Haskell code.

Nothing is said here about lhs. Your code block with lhs is just a regular code block.

[jgm]

I think it's actually quite useful to have a way to do both of these things in a Markdown document:

1. Insert some literate Haskell, which will be treated as code in markdown+lhs output.
2. Insert a regular code block which happens to display some literate Haskell code, and which will be highlighted as such in output formats that do highlighting.

[ad-si]

I think it's actually quite useful to have a way to do both of these things in a Markdown document:

Yeah, I guess you're right!

It just bugs me, that it's not really correct markdown, or rather markdown which other parsers can easily handle. E.g. Commonmark will do this:

```{.literate .haskell}
etst
```

<pre lang="{.literate"><code>etst
</code></pre>

Which is not really what anyone wants, who writes this code...

Other parsers completely mess it up: gettalong/kramdown#503

It would be cool if there was a more standard compliant way to differentiate between Haskell code which gets displayed and code which gets executed ...

[jgm]

You could use a very simple lua filter to change code blocks with class `lhs` to `literate haskell`, and use this when you're converting using pandoc...

[ad-si]

Well, for the time being I use this workaround to execute / compile the Haskell:

cat document.md \
| sed 's/```haskell/```{.literate .haskell}/g' \
| pandoc \
  --from markdown \
  --to markdown+lhs \
  --output temp.lhs \
| stack runhaskell \
  --resolver lts-11.1 \
  --package cmdargs \
  -- \
  temp.lhs \
; rm -f temp.lhs

Not glamorous, but get's the job done =P.
I'll switch to a Haskell pipeline for my blog in the future (shake + pandoc probably), then the lua filter might be a good idea. Thanks for the tip.