Slight modification of main, styles, and script to automate notebook to HTML #1
Conversation
There was a problem hiding this comment.
Hi @rlphillips and thanks for your interest and your contribution!
I completely understand what you want to do, which is to make these tools usable in general. And that's a very good idea!
But at the same time, for now, the purpose of this repository is to serve as a reference for my blog post. So this repository should stay in sync with the post. For now, it means that the current behaviour should be preserved. I can also change the post later on, when we converge on the solution.
For this reason, I cannot accept the PR as it is, and would like to suggest a few changes.
Also, I think that we should try and get rid of the bash scripts, we can get the functionalities you need with python.
At some point, we could even think of a proper python package. I hadn't done it because I was not expecting this repo to start coming to life :-)
Finally, the docker solution is quite interesting, and I'm a big fan of docker myself. We can go further later on, and provide a docker file that does what we want.
Cheers!
| </pre> | ||
|
|
||
| {% include 'overfitting.html' %} | ||
| {% include FNAME %} |
There was a problem hiding this comment.
If used out of your scripts, FNAME is undefined. So the current behavior is not preserved. Could you use jinja2 in python instead of sed in a shell script to replace FNAME?
There was a problem hiding this comment.
Right: Here, I was using FNAME as a placeholder and replacing it in the bash Script: The "template/main.html" file was copied into "template/[name-of-the-notebook].html" file, and replaced FNAME with the name of the html created in the previous step.
| @@ -15,8 +16,8 @@ | |||
| loader = FileSystemLoader('templates'), | |||
| ) | |||
|
|
|||
| input_file = 'main.html' | |||
| output_file = 'index.html' | |||
| input_file = sys.argv[1] | |||
There was a problem hiding this comment.
The current behaviour is not preserved. Could you please do the following?
- update the README.md file to give the proper way to run
render.py - document the new arguments in render.py
- deal with incorrect numbers of arguments (exit and print help in this case)
| @@ -559,3 +559,33 @@ div.output_javascript:empty { | |||
| .th { | |||
| text-center: left; | |||
| } | |||
|
|
|||
|
|
|||
| .dataframe tbody tr { | |||
There was a problem hiding this comment.
interesting, I look forward to see the new style.
| @@ -0,0 +1,33 @@ | |||
| #!/usr/bin/bash | |||
There was a problem hiding this comment.
Isn't this specific to your workflow? if yes, could you remove this file?
|
|
||
| #First argument: File name of the notebook to work on: e.g., if the notebook is analysis.ipynb, argument is analysis. | ||
|
|
||
| cd notebooks/ |
There was a problem hiding this comment.
This is a good idea, but I think that it would be better to write this code in python instead of in a shell script.
I think that a good interface could be:
python render.py base.html my_nb_1.ipynb my_nb_2.ipynb ...
- The user could provide any number of notebooks to insert.
- In my_output.html, we would use jinja2 tags to specify where each notebook should be inserted, based on the name the notebook
- The script would run jupyter nbconvert (maybe there is a python API, or with os.system or something)
- Finally, the ouput file would be rendered
One could also think about the options. For example, the name of the output file (index.html by default)
There was a problem hiding this comment.
I'll use this interface, then. I'd have to research jinja2 a bit to see what templating options are available.
|
Hello! Alright, thanks for your feedback! Since I'm not so experienced with Python and Jinja2 yet, I added bash to the mix, to get what I was working with reliably. But I understand it's not really in line with the philosophy of the repo. It would probably be tidier to close this PR then, and if I get around to implementing this on Python I'll submit another one. I'll keep you posted, since I plan on regularly converting notebooks to Jekyll blog posts, based on this tool you introduce. I'll still comment and keep in mind your review requests for a future Python implementation. As for a Docker solution with every step of the process, I'll start going in that direction, but it probably won't be completely fleshed out until I use it several times and run into some issues in order to make the solution more general. We'll be in touch! Take care. |
Hello!
I've taken this project and slightly modified the main.html in order to parametrize the name of the html used as template. render.py now takes two parameters: name of the template to use, and name of the output file.
There's a bash script to run with the jupyter notebook you might be using, that deals with the nbconvert and calling render.py. I've tested with a few sample files, and I'm uploading one as a sample.
I've also added a bit of the styles for tables, that seemed to be missing.