gpoore / codebraid

Live code in Pandoc Markdown
BSD 3-Clause "New" or "Revised" License
367 stars 13 forks source link

Executing python code within $...$? [enhancement: add cb.sub command] #27

Open dataopt opened 4 years ago

dataopt commented 4 years ago

In knitr, one could have something like this

```{r include = FALSE}
var = 123

Note that $\sqrt{r var} =r sqrt(var)$.



The last line will get rendered to html as
``
Note that \(\sqrt{123} = 11.0905365\).
``

Does CodeBraid allow executing the python code within $...$ to accomplish the same thing?  My initial experiment suggests that it doesn't. Any help is greatly appreciated.
gpoore commented 4 years ago

It looks like Pandoc won't parse Markdown inside LaTeX math. So I'd suggest something like this, using a format string:

```{.python .cb.run}
from math import *
var = 123

Note that rf"$\sqrt{{{var}}} = {sqrt(var)}$"{.python .cb.expr}.

And then compile with something like

codebraid pandoc -f markdown -t html --mathjax -s -o temp.html temp.md


If this sort of situation comes up a lot, I suppose I might consider creating a string template command similar to the `pysub` I created in [PythonTeX](https://github.com/gpoore/pythontex/). Maybe something like

$\sqrt{!{var}} = !{sqrt(var)}${.python .cb.sub}


Basically, everything inside backticks could be treated as a template string, with substitution fields denoted by `!{<code>}`, and an option to use multiple braces as delimiters (similar to Markdown backticks) to escape any literal braces within `<code>`. This would eliminate string quoting, string backslash escapes for non-raw strings, and LaTeX brace escaping.
dataopt commented 4 years ago

Thank you for the helpful suggestion.

Any chance that the string template could become a command-line option for CodeBraid?

gpoore commented 4 years ago

I was thinking about a string template as another command that could be applied to code blocks. What are you thinking about in terms of a command-line option? How might that work?

dataopt commented 4 years ago

I was thinking of an option (say -s) that will make the following syntax

`$\sqrt{!{var}} = !{sqrt(var)}$`{.python .cb.sub}

become automatic for the entire file.

EDIT: I probably haven't articulated myself very well. I am new to CodeBraid and I was just hoping that inlining would be as simple as Knitr. I do hope to use CodeBraid whenever possible since it's so much more powerful.

gpoore commented 4 years ago

It sounds like you're interested in a more compact, less verbose syntax, especially for inline code. I'm interested in adding alternate syntax eventually. So far I've limited everything to Pandoc's built-in attribute syntax. That can be verbose and isn't ideal in some respects, but it does allow all of the Markdown processing to be delegated to Pandoc, which has a lot of advantages.

dataopt commented 4 years ago

That would be a fair assessment of my interest. But I now do understand the philosophy behind CodeBraid.

Perhaps one could just write a preprocessor that turns a compact syntax into the full-blown syntax that CodeBraid accepts. I should take a stab at this, perhaps mimicking what Knitr does to ease my own transition.

gpoore commented 4 years ago

You might look into Pandoc filters.

For example, create inline_cb.lua:

Code = function(elem)
    if string.sub(elem.text, 1, 3) == "py " then
        elem.text = string.sub(elem.text, 4, string.len(elem.text)):match("^%s*(.*)")
        elem.classes:insert("python")
        elem.classes:insert("cb.expr")
    end
    return elem
end

Create a file temp.md:

`py 1+1`

Run pandoc -L inline_cb.lua -f markdown -t markdown -o intermediate.md temp.md to create intermediate.md:

`1+1`{.python .cb.expr}

Then send that through Codebraid. Codebraid can use filters, but I don't yet have a way to run filters before code execution, which is what would be needed in this case. Also, I'm using an intermediate file instead of piping since I just realized that there isn't yet support for reading from stdin. Both of those limitations should be easy to fix in future releases, so that you could just use a filter when running Codebraid directly.

dataopt commented 4 years ago

Thank you. I think adding pysub and overcoming those limitations will be tremendously useful.