WIP: add support for ipywidgets #2
Conversation
changed "no kernel is connect" to "no kernel is connected"
README typo
TODO
|
jupyter_sphinx/execute.py
Outdated
| nb, resources = super(ExecutePreprocessor, ep).preprocess(nb, resources) | ||
| info_msg = ep._wait_for_reply(ep.kc.kernel_info()) | ||
| nb.metadata['language_info'] = info_msg['content']['language_info'] | ||
| return extract_widget_state(ep) |
There was a problem hiding this comment.
extract_widget_state will only work properly if:
- The kernel is IPython (or anything running Python more generally, e.g. xonsh).
- ipywidgets are importable
Therefore calling extract_widget_state should only be done when it's safe.
In absence of multi-language implementation of the Kernel side of jupyter widgets, I believe a relatively safe heuristic for doing this is to check that the outputs contain at least one widget view.
There was a problem hiding this comment.
What about just wrapping the code to execute in a try/catch block>
jupyter_sphinx/execute.py
Outdated
| nb, resources = super(ExecutePreprocessor, ep).preprocess(nb, resources) | ||
| info_msg = ep._wait_for_reply(ep.kc.kernel_info()) | ||
| nb.metadata['language_info'] = info_msg['content']['language_info'] | ||
| return extract_widget_state(ep) |
There was a problem hiding this comment.
We need to be more careful here. Widgets are only a thing for Python kernels
There was a problem hiding this comment.
I suppose the widget protocol could be used from kernels of other languages, but I've never seen this.
We can support only extracting widget state from IPython kernels for the moment, and implement others later if there is demand.
There was a problem hiding this comment.
minimally we would need to only extract if the kernel is a python one.
jupyter_sphinx/execute.py
Outdated
| nb, resources = super(ExecutePreprocessor, ep).preprocess(nb, resources) | ||
| info_msg = ep._wait_for_reply(ep.kc.kernel_info()) | ||
| nb.metadata['language_info'] = info_msg['content']['language_info'] | ||
| return extract_widget_state(ep) |
There was a problem hiding this comment.
What about just wrapping the code to execute in a try/catch block>
jupyter_sphinx/execute.py
Outdated
| nb, resources = super(ExecutePreprocessor, ep).preprocess(nb, resources) | ||
| info_msg = ep._wait_for_reply(ep.kc.kernel_info()) | ||
| nb.metadata['language_info'] = info_msg['content']['language_info'] | ||
| return extract_widget_state(ep) |
There was a problem hiding this comment.
Soon this will be moot: jupyter/nbconvert#900
There was a problem hiding this comment.
And this is targeted for nbconvert 5.5
|
The url of the external json is not correctly parsed I think: <script type="application/vnd.jupyter.widget-state+json" src="/home/docs/checkouts/readthedocs.org/user_builds/adaptive/checkouts/latest/docs/source/_build/jupyter_execute/tutorial/tutorial.Learner1D_widget-state.json"></script>from https://adaptive.readthedocs.io/en/latest/tutorial/tutorial.Learner1D.html |
|
@akhmerov unfortunately we can't put the widget state in a separate file because of this: https://github.com/jupyter-widgets/ipywidgets/blob/master/packages/html-manager/src/libembed.ts#L36 because they just take the "innerHTML" from the script tag we can't specify it as a separate file |
jbweston
force-pushed
the
feature/execute
branch
from
5d005d8
to
fb90706
Compare
jbweston
force-pushed
the
feature/execute
branch
from
fb90706
to
5af07f0
Compare
jbweston
force-pushed
the
feature/execute
branch
from
5af07f0
to
9e9cf80
Compare
add an extension for executing code in a jupyter kernel
|
I merged the main PR to jupyter-sphinx. Would you like to reopen this to master? |
Co-authored-by: Bas Nijholt <[email protected]>
|
Closing to reopen against the upstream repo |
No description provided.