Pipe inside code block inside table cell doesn't get rendered properly

[MustafaHaddara]

Using a pipe character | inside of a code block inside of a table cell doesn't maintain the code block. eg.

Column 1 | Column 2 | Column 3
---------|----------|---------
Pipe     | `|`      | word

Should work, but doesn't. Instead, we see this:

A workaround for now is to replace the | with <code>&#124;</code> (ie. the html literal). This renders correctly.

See mkdocs/mkdocs#743 for more details

[waylan]

Hmm, so I'm not sure of the cleanest way to address this.

Currently I'm just doing row.split('|'). Obviously, this is a no-go when pipes are included in a code span. That means I need to parse the row accounting for quoted strings (quoted with backticks).

I thought maybe the csv lib could do this:

>>> list(csv.reader(['foo | `|` | bar | 42'], delimiter='|', quotechar='`', skipinitialspace=True))
[['foo ', '| ', 'bar ', '42']]

Unfortunately, csv.reader strips the quotechar from the output, which means we loose the codespan. In other words the output should be: [['foo', '|', 'bar ', '42']]. Also, skipinitialspace=True is required to make it work if the code span is padded with initial whitespace. But there is no way to have it strip trailing whitespace which is a little inconsistent (although certainly workable). Finally, on Python 2 there is the weirdness that Unicode input is not supported (you have to encode as UTF-8, read, then yield items, decoding each back to Unicode).

Looks like I would need to write my own parser. I don't have time for that right now. You'll have to use the workaround for now. Of course, patches (pull requests) are always welcome.

[MustafaHaddara]

Whelp, here's a patch #437

I'm not sure if it's the best approach, but it's one way of going about it.

[PabloLION]

Hi maintainers.
This issue re-appeared. I'm with py3.9 and Mac M1Max
I did pip install markdown, showing Successfully installed markdown-3.3.6 and getting /opt/homebrew/lib/python3.9/site-packages/Markdown-3.3.6.dist-info/

This code piece

import markdown
str= """
## test

| test              | test |
| ----------------- | ---- |
| <code> \| </code> | a    |

another table

| test            | test |
| --------------- | ---- |
| <pre> \| </pre> | a    |
"""

html = markdown.markdown(str)
print(html)

prints this result

<h2>test</h2>
<p>| test              | test |
| ----------------- | ---- |
| <code> \| </code> | a    |</p>
<p>another table</p>
<p>| test            | test |
| --------------- | ---- |
| <pre> \| </pre> | a    |</p>

So, is this function down or incorrectly implemented?
Should I create a new issue?

[facelessuser]

@PabloLION, you need to enable the tables extension.

[PabloLION]

@facelessuser Thanks. Just found it here. Thing is I'm using mkdocs.. I'll try to see how to combine them. Thanks again.

[facelessuser]

https://www.mkdocs.org/user-guide/configuration/#markdown_extensions. It's all in the docs 😉 .

[PabloLION]

Thanks again 😄
Although setting the markdown_extensions: doesn't work for me (I saw tables in the list of the link you posted)
I tried all these line by line and no one worked. I guess it's a mkdocs bug. (with python markdown.markdown(str, extensions=['markdown.extensions.tables']) worked fine)

markdown_extensions:
  - tables
  - markdown.extensions.tables
  - tables: {}
  - markdown.extensions.tables: {}
  - extra:
    - tables
    - markdown.extensions.tables
    - tables: {}
    - markdown.extensions.tables: {}

tables is used by default. This is not a common problem and I was confused here. See the solution and more details

[vedranmiletic]

I guess it is what it is now, but I wish we took a different approach here with mandatory escaping of pipes in tables, even under backticks.

We are not bound by what other implementations do, of course, but in the end it is nice to have some interoperability. FWIW, GFM requires escaping so linters such as markdownlint implement it, cf. issue 1206 and both HackMD and VS Code adhere to mandatory escaping like GFM describes it.