PDF Support

[mmcky]

The latex generated from rst sources around line 93

\chapter{Dynamics in One Dimension}
\label{\detokenize{scalar_dynam:dynamics-in-one-dimension}}\label{\detokenize{scalar_dynam::doc}}\index{Dynamics in One Dimension@\spxentry{Dynamics in One Dimension}}
\begin{sphinxShadowBox}
\sphinxstyletopictitle{Contents}
\begin{itemize}

compare to the latex generated from the myst sources around line 93

\chapter{\label{\detokenize{scalar_dynam-MyST:scalar-d}} Dynamics in One Dimension}
\label{\detokenize{scalar_dynam-MyST:dynamics-in-one-dimension}}\label{\detokenize{scalar_dynam-MyST::doc}}
\begin{sphinxShadowBox}
\sphinxstyletopictitle{Contents}
\begin{itemize}

It is interested as the myst output has included labels within the chapter name likely causing an issue.

[mmcky]

@chrisjsewell have we compared the AST produced from rst and myst source files that contains the same content (not markup clearly)? Just trying to figure out why sphinx is adding in a label within the chapter tags. I seem to recall there was a nice way to visualise the ast -- is this right?

I guess I am wondering if visit_title and visit_label method calls are in a different order.

[chrisjsewell]

In executablebooks/MyST-Parser#123 there are now test fixtures that test and clearly define the AST output from myst-parser: https://github.com/ExecutableBookProject/MyST-Parser/tree/markdown-it/tests/test_renderers/fixtures (in particular see syntax_elements.md).

You could compare this with the test fixtures in docutils, e.g. https://github.com/docutils-mirror/docutils/blob/master/test/test_parsers/test_rst/test_section_headers.py

Note you may want to try running this project with myst-parser==0.8.0a3, before the new implementation is merged, and have a quick check if any issues are inadvertently caused (hopefully not!)

[chrisjsewell]

Also in https://github.com/ExecutableBookProject/MyST-NB/blob/92757ba449130784d2642f7f8fae75f7dd79cfb7/tests/conftest.py#L42, is an example of loading the .doctree generated in the sphinx build folder, hwich you can then get the text representation of the AST the .pformat():

from pathlib import Path
import pickle

document = pickle.loads(Path("path/to/build/file.doctree").read_bytes())
print(document.pformat())

[mmcky]

thanks @chrisjsewell -- this is really helpful.

[mmcky]

So I have updated to myst-parser==0.8.0a3 and the ordering of the AST looks to be correct. There are still some edges cases but make latexpdf now runs. This repository is configured to use xelatex as there is unicode in the source files.

mystexampleproject.pdf

[mmcky]

Edge problems:

1. Titles are not just the title text: (scalar_d)= Dynamics in One Dimension
2. some inline math is not getting rendered and showing as raw latex in the pdf. ($S = \mathbb R_+ = [0, \infty)$.

The latex for the above issues is:

1. \chapter{(scalar\_d)= Dynamics in One Dimension}
2. \$S = \textbackslash{}mathbb R\_+ = {[}0, \textbackslash{}infty)\$

[mmcky]

So looking at sphinx warnings:

/Users/matthewmckay/repos-collab/ebp/myst-parser.example-project/source/ar1_processes-MyST.md:9: WARNING: None:any reference target not found: scalar_d
/Users/matthewmckay/repos-collab/ebp/myst-parser.example-project/source/finite_markov-MyST.md:741: WARNING: None:any reference target not found: ar1_processes
/Users/matthewmckay/repos-collab/ebp/myst-parser.example-project/source/finite_markov-MyST.md:970: WARNING: undefined label: mc_eg1-1 (if the link has no caption the label must precede a section header)

Might cause the title issues -- will fix this in the source.

[chrisjsewell]

(scalar_d)= Dynamics in One Dimension; You can't write this in one line, at least not at present, since it is a block level element, i.e. these should be:

(scalar_d)=
# Dynamics in One Dimension

don't know where you got the other approach from, that's not what's in the docs lol

Although let me know if you think it would be needed

[chrisjsewell]

I'll need more info on the 2nd point. When I run this in the test fixtures I correctly get

<document source="notset">
    <paragraph>
        <math>
            S = \mathbb R_+ = [0, \infty)
        .

does it work using the math directive?

[mmcky]

hey @chrisjsewell does the parser assume inline objects are contained on the same line.

It looks like the issue with inline math for pdf is associated with line splits

Consider again the linear model $x_{t+1} = a x_t + b$ with $a
\not=1$.

the first equation is rendered fine while the later is not.

This is the same for html output also

[mmcky]

Although let me know if you think it would be needed

Thanks @chrisjsewell I don't think so. I think an independent link as a placeholder in a document should be on its own line.

[chrisjsewell]

does the parser assume inline objects are contained on the same line

For inline math it does yes. This is a consequence of porting this markdown-it plugin, which does not support multi-lines, as mentioned in goessner/markdown-it-texmath#18.
I don't know the reason for this design decision, but it should be quite possible in the future (probably not right now) to make the plugin accept new lines.

[jstac]

I just tested the Jupyter markdown parser and it has no issues with rendering

$f
g$

so this is a likely source of errors.

[mmcky]

Yeah - we have a bunch of them across our projects as sphinx supports them and we are migrating from rst. I think if the notebook supports them - we will need to as well (as @jstac points out). I will raise an issue for tracking this on myst-parser. thanks @chrisjsewell