Markdown writer not escaping quotes within JSON

[choldgraf]

Pandoc version: 2.7.1
OS: ubuntu latest

---

I'm playing around with converting .ipynb format to markdown, and ran into a snag that I think is because Pandoc is embedding JSON in a string but is not escaping the quotes characteres.

When converting ipynb to markdown, I noticed that double-quotes were being used within the contents the metadata, like so:

::: {.cell .markdown tags="["heresatag"]"}

This was confusing the subsequent writer, which didn't correctly parse the contents of tags= because of the double-quotation marks.

Is there a way that we could escape the inner quotation marks so that they don't mess up the conversion?

[mb21]

Can you post a minimal sample ipynb file that is converted to the markdown you posted?

[choldgraf]

Sure thing 👍

Here's the contents of an .ipynb file:

ipynb file

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": [
     "one",
     "two"
    ]
   },
   "source": [
    "This cell has metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "tags": [
     "three",
     "four"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "This code cell has metadata\n"
     ]
    }
   ],
   "source": [
    "print(\"This code cell has metadata\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}

it has two cells, each of which has two tags. (above the metadata for each cell is in a "metadata" key.

when I convert it to markdown with

# First copy the above code into a file called `minimal.ipynb`
pandoc minimal.ipynb -o minimal.md -s

it outputs this:

---
jupyter:
  kernelspec:
    display_name: Python 3
    language: python
    name: python3
  language_info:
    codemirror_mode:
      name: ipython
      version: '3.0'
    file_extension: '.py'
    mimetype: 'text/x-python'
    name: python
    nbconvert_exporter: python
    pygments_lexer: ipython3
    version: '3.6.5'
  nbformat: 4
  nbformat_minor: 2
---

::: {.cell .markdown tags="["one","two"]"}
This cell has metadata
:::

::: {.cell .code execution_count="2" tags="["three","four"]"}
\``` {.python}
print("This code cell has metadata")
\```

::: {.output .stream .stdout}
    This code cell has metadata
:::
:::

(backslashes at the start of code fences are mine, I added them so github would render it)

[mb21]

Thanks! Surprisingly, this appears to be a problem with the markdown writer:

echo '[Div ("",[],[("tags","[\"one\",\"two\"]")]) [] ]' | pandoc -f native -t markdown

::: {tags="["one","two"]"}
:::