Warp math block into div block using pandoc filter

[Xmarmalade]

I am using pandoc to convert markdown files to docx files, but I need to apply a separate format to the math block in markdown.
I found that there is no standalone format for math block in pandoc's default reference docx file, so I added a custom formatting like the following:

<div custom-style="Equations">

$$\begin{aligned}
   V_{\mathrm{yFk}}&=(1.08-1)\times7.186\times0.4m_{\mathrm{r}}\\
   &=0.230m_{\mathrm{r}}
\end{aligned}$$

</div>

the AST should like this

[ Para
      [ Math
          DisplayMath
          "\\begin{aligned}\n   V_{\\mathrm{yFk}}&=(1.08-1)\\times7.186\\times0.4m_{\\mathrm{r}}\\\\\n   &=0.230m_{\\mathrm{r}}\n\\end{aligned}"
      ]
  , Div
      ( "" , [] , [ ( "custom-style" , "Equations" ) ] )
      [ Para
          [ Math
              DisplayMath
              "\\begin{aligned}\n   V_{\\mathrm{yFk}}&=(1.08-1)\\times7.186\\times0.4m_{\\mathrm{r}}\\\\\n   &=0.230m_{\\mathrm{r}}\n\\end{aligned}"
          ]
      ]
  ]

However, this seems to be a pain as I have hundreds of math blocks to add, is there any way to write a filter or other ways to add custom style div?

[cagix]

you could try something like

function Math(el)
    if el.mathtype:match 'DisplayMath' then
        local kv = {}
        kv['custom-style'] = 'Equations'
        return pandoc.Div(el, kv)
    end
end

(not tested)

[Xmarmalade]

on third tought, since Math is inline, we can't return a block ... so either you need to run over all blocks to find block math inside or you could try RawInline like在第三个 tought 上，由于 Math 是内联的，我们不能返回一个块......因此，要么您需要遍历所有块才能在里面找到块数学，要么您可以尝试 RawInline 像

function Math(el)
    if el.mathtype:match 'DisplayMath' then
        return {
            pandoc.RawInline("html", "<div custom-style='Equations'>"),
            el,
            pandoc.RawInline("html", "</div>")
        }
    end
end

This seems to work, but MS Word doesn't recognize them correctly as 'Equations', the AST looks like

Para
      [ RawInline (Format "html") "<div custom-style='Equations'>"
      , Math
          DisplayMath
          "\\begin{aligned}\n   V_{\\mathrm{yFk}}&=(1.08-1)\\times7.186\\times0.4m_{\\mathrm{r}}\\\\\n   &=0.230m_{\\mathrm{r}}\n\\end{aligned}"
      , RawInline (Format "html") "</div>"
      ]

in MS word, the 'Equations' blocks still have 'Body Text' style

and im trying using python filter to check every para, but i dont know how to put the result together

def warp_math_in_div(key, value, format, meta):
    if key == 'Para':
        if value[0]['t'] == 'Math':
          # dont know return <div custom-style="Equations">$$math$$</div>
          return Div(...)

[jgm]

Changing the last line of the first attempt above to something like this should give you what you want:

        return pandoc.Div(pandoc.Para({el}), kv)

[cagix]

Changing the last line of the first attempt above to something like this should give you what you want:

        return pandoc.Div(pandoc.Para({el}), kv)

Yeah, that was my second thought. Unfortunately, Math is an inline element, so when returning a Div instead, Lua gives me an error.

% pandoc -L math.lua mwe.md  
Error running filter math.lua:
Inline, list of Inlines, or string expected, got Block

[cagix]

Seems, we have to take the scenic route:

result = empty list
for all blocks:
  if block contains block math:
    block_wo_math, math = remove_block_math(block)
    result.append(math)
    result.append(block_wo_math)
  else:
    result.append(block)
return result

[jgm]

Ah right, you'll have to split the paragraph and create a new block where the Math element was...

[Xmarmalade]

Found this solution:

function Para(para)
    -- if para sub el is math, then create a div with custom-style = Equations and put the para in it
    if para.content[1].t == 'Math' then
        local kv = {}
        kv['custom-style'] = 'Equations'
        return pandoc.Div(para, kv)
    end
end

Before

Pandoc
  Meta { unMeta = fromList [] }
  [ Para
      [ Math
          DisplayMath
          "\\begin{aligned}\n   V_{\\mathrm{yFk}}&=(1.08-1)\\times7.186\\times0.4m_{\\mathrm{r}}\\\\\n   &=0.230m_{\\mathrm{r}}\n\\end{aligned}"
      ]
  ]

After

Pandoc
  Meta { unMeta = fromList [] }
  [ Div
      ( "" , [] , [ ( "custom-style" , "Equations" ) ] )
      [ Para
          [ Math
              DisplayMath
              "\\begin{aligned}\n   V_{\\mathrm{yFk}}&=(1.08-1)\\times7.186\\times0.4m_{\\mathrm{r}}\\\\\n   &=0.230m_{\\mathrm{r}}\n\\end{aligned}"
          ]
      ]
  ]

Works well on my markdown, thanks!

[cagix]

that's a nice solution! however, keep in mind that this only works if the block math is (always) the first element in a paragraph, i.e. if there is always an empty line before each block math element.

to make it work in the general case, you would have to implement as described above (here, here).