Should fenced and indented markdown codeblocks be parsing differently?

[solomon-b]

When using fenced codeblocks I'm seeing newline characters get stripped from my codeblocks and the codeblock is wrapped in a Para node. Indented codeblocks preserve the newline and don't wrap in a Para. Is there a way to make fenced codeblocks behave the same as indented ones?

Fenced

Example codeblock:

```
id :: forall a. a -> a
id x  = x
```

Pandoc (json serialized) AST:

{
    "blocks": [
      {
        "c": [
          {
            "c": [
              [
                "",
                [],
                []
              ],
              "id :: forall a. a -> a id x  = x"
            ],
            "t": "Code"
          }
        ],
        "t": "Para"
      }
    ],
    "meta": {},
    "pandoc-api-version": [
      1,
      23,
      1
    ]
  }

Space Indented

Example codebloack:

    id :: forall a. a -> a
    id x  = x

Pandoc (json serialized) AST:

{
    "blocks": [
      {
        "c": [
          [
            "",
            [],
            []
          ],
          "id :: forall a. a -> a\nid x  = x"
        ],
        "t": "CodeBlock"
      }
    ],
    "meta": {},
    "pandoc-api-version": [
      1,
      23,
      1
    ]
  }

[kysko]

Gives correct output for me (pandoc 3.6 Windows).

Only way I could reproduce your first output was using this as input:

`
id :: forall a. a -> a
id x  = x
`

ie using only one tick, which would be a Para with an inline Code.

[jgm]

I can't reproduce what you're reporting. Use https://pandoc.org/try and get a permalink if you can reproduce there.

[solomon-b]

I'm using pandoc as a library. This is the exact snippet I used to generate the code:

  pandocResult <- liftIO . runIO $ do
    let readerConfig = def { readerExtensions = extensionsFromList [Ext_fenced_code_blocks, Ext_hard_line_breaks] }
        writerConfig = def { writerExtensions =  extensionsFromList [Ext_fenced_code_blocks, Ext_hard_line_breaks] }
    pandocAst <- readMarkdown readerConfig md
    result <- writeHtml5String writerConfig pandocAst
    pure (pandocAst, result)

Here is the input Text:

"```\nid :: forall a => a -> a\nid x = x\n```\n\n    id :: forall a => a -> a\n    id x = x"

and here is the Pandoc AST

Pandoc (Meta {unMeta = fromList []}) [Para [Code ("",[],[]) "id :: forall a => a -> a id x = x"],CodeBlock ("",[],[]) "id :: forall a => a -> a\nid x = x"]

This is with pandoc-3.5. I'll try 3.6 and report back.

EDIT: Same results with pandoc-3.6

[jgm]

Check the documentations for the extensions you're using:
https://pandoc.org/MANUAL.html#extension-fenced_code_blocks
The fenced_code_blocks extension is for tilde-fenced code blocks (a pandoc innovation that predates GFM style fenced code blocks). You want backtick_code_blocks.

[solomon-b]

Thank you that solved it!

[solomon-b]

Where can I see what extensions are enabled in the pandoc binary?

[jgm]

See T.P.Extensions in the code.
And --list-extensions in the manual.