Table layout and carriage return

[ncarboni]

I am testing with this plugin and it is already fantastic, but I was wondering if there is a way to take into account carriage return within a cell. I would like a cell to display

→ P108i → E12 → P32 → E55 
→ P108i → E12 → P2 → E55

But if I tried to load a CSV with a carriage return I obtain this:

fair enough, it is a CSV. Using an excel file (using {{ read_excel('file'.xlsx) }} instead I obtain this result:

While nicer the result coming from excel do present a new line instead of a carriage return within the same cell. I was wondering if it would be possible to have a real carriage return, something like this (obtained using <br>):

[timvink]

Thanks for the feedback, glad to hear you like it! And a different use case than I imaged when I created it, nice!

Some technical context

This plugin uses df.to_markdown() to create markdown from a table. That's included in pandas>=1.03, but actually uses tabulate under the hood.

**kwargs These parameters will be passed to tabulate. docs for pandas.DataFrame.to_markdown

Currently, this is what the call looks like in this plugin:

https://github.com/timvink/mkdocs-table-reader-plugin/blob/1d75fbfc8cdb07c21b196065965a1e8e7845baff/mkdocs_table_reader_plugin/plugin.py#L10

There are other parameters supported by tabulate, see python-tabulate. Specifically the section on line-breaks is very relevant: "Note that some output formats (e.g. simple, or plain) do not represent row delimiters, so that the representation of multiline cells in such formats may be ambiguous to the reader" . The default is tablefmt="simple" so it might be useful to try a different format.

MkDocs uses python-markdown which implements this reference syntax. That one doesn't include tables, which is why python-markdown has a tables extension. That extension in turn uses the PHP Extra Markdown syntax for tables, which is supported by tabulate via tablefmt="pipe", which is included in the list of output formats that support multiline cells!

That was a fun bit of detective work :) With this knowledge, I think "pipe" should be the default tablefmt for this plugin. I created a new merge request with the change in #3 that we can test.

Possible solutions

1. The new default tablefmt="pipe" output setting. Can you test (by installing #3) if this solves your issue before I release a new version?
2. If you are on windows, make sure your line endings are in unix format (\n instead of \r\n). context

[ncarboni]

Thanks for the explanation, it is quite useful to understand the possibilities. I quickly tried yesterday using tablefmt="pipe", but I still did not get any multiline. I will do some other testing today (I will try other tablefmt options and see if something change) and come back with more info. In the meanwhile, thank you very much!

[ncarboni]

A bit of recap after some tests. I added tablefmt="pipe" (return df.to_markdown(showindex=False, tablefmt="pipe") to read_csv and read_excel, and re-installed the package. However, I do not see any differences in the creation of the table. I tried with other table format from python-tabulate, but they did not seems to work either. I tried to inserti a \n in order to signal a <br> in the cell of the CSV or excel file, but also there I did not obtained a real carriage return but only a new row. I am maybe doing something wrong from my side? Did you by any chance make it work?

[timvink]

Did some experiments (and found + fixed a different bug in #4).

This csv:

Renders as:

However inserting <br/> directly works as intended, and might already be the solution for your specific use-case:

It seems that the csv standard does not support escape characters like \n (source). Alternatively, specifying the newline differently in the CSV creates a new row but the alignment is proper:

---

I'm not sure what the proper behaviour should be, and where that should be addressed. Most likely it's in python-markdown tables extension. But the root cause is not in mkdocs-table-reader-plugin, so I suggest to close this issue. I might investigate further and open a new issue elsewhere.

[timvink]

FYI, I opened https://github.com/Python-Markdown/markdown/issues/975

[ncarboni]

Thanks for the testing! If will use
for now and let's see if something will change from python-markdown side. Feel free to close :-)

[timvink]

Long story short, for multiline cells the solution is to use <br/>. (see also the python-markdown issue). If you need anything more complex, you can always fall back to raw HTML tables.