Improve support for LaTeX via MathML in documentation

[mmatera]

This PR contains the minimal changes needed to support LaTeX expressions in the documentation.

[mmatera]

[rocky]

LGTM - thanks.

This starts us toward improving the online doc so it has can have an impact larger than what is visible right now. I suspect our homegrown parsing limits what we can offer MathML. For example, right now we have to tag $x$^2 rather than $x^2$.

Some things to note:

While this works in a "dd" tag, I don't think it works in "dt". I tried putting $\sigma_k$ in the <dt> portion of DivisorSigma , and that came out untranslated.

Also, when I look at Cuboid, is don't see p_min (in the dd portion since dt doesn't work) getting rendered with a subscript. Know why?

At any rate, these can be done in a future PR.

[mmatera]

LGTM - thanks.

This starts us toward improving the online doc so it has can have an impact larger than what is visible right now. I suspect our homegrown parsing limits what we can offer MathML. For example, right now we have to tag $x$^2 rather than $x^2$.

I can check, but in principle now $x^2$ would work fine both in the PDF and the online documentation

Some things to note:

While this works in a "dd" tag, I don't think it works in "dt". I tried putting $\sigma_k$ in the <dt> portion of DivisorSigma , and that came out untranslated.

OK, but we use <dt> blocks for code, so we do not need LaTeX code there.

Also, when I look at Cuboid, is don't see p_min (in the dd portion since dt doesn't work) getting rendered with a subscript. Know why?

The reason seems that in these blocks, $...$ blocks are translated to <var>...</var>. I will check if we can get ride of this conversions.

At any rate, these can be done in a future PR.

[rocky]

OK, but we use <dt> blocks for code, so we do not need LaTeX code there.

That is not the way "dt" is supposed to work. A lot of our formatting and WMA's formatting use Mathematical notation in the definition term.

Also, when I look at Cuboid, is don't see p_min (in the dd portion since dt doesn't work) getting rendered with a subscript. Know why?

Maybe because p_min is not wrapped in $ ?

No. See https://github.com/Mathics3/mathics-core/blob/master/mathics/builtin/drawing/graphics3d.py#L257

[mmatera]

Yep. The problem is a concatenation of several small issues:

• This regular expression
  LATEX_RE = re.compile(r"(\s?)\$(\w+?)\$(\s?)") defined in mathics.doc.doc_entries is used in

  mathics-django/mathics_django/doc/utils.py

  Line 80 in 2ed4a3d

  text = LATEX_RE.sub(repl_latex, text)

  to replace $var$ by <VAR>var</VAR> . This makes sense if var is a sequence of alphabetic characters and numbers, but not for a LaTeX expression. To avoid using it for more complicated expressions, we should change the regular expression to
  LATEX_RE = re.compile(r"(\s?)\$([A-Za-z]+?)\$(\s?)")

• Due to the quotes, 'Cuboid[$p_min$]' is translated to HTML as <code>Cuboid[$p_min$]</code>, which makes the code to be handled as "code" i.e., verbatim. The way to allow MathML in the arguments would be to use the quotes around the HEAD of the expression, like this: 'Cuboid'[$p_{min}$]

• LaTeX interpreter understand p_min as p_{m}in, i.e., just the first character is considered as the subindex. To have min as subindex, we need to wrap it in curly brackets.