Adding download links for the notebook files #104
Conversation
| @@ -0,0 +1,4 @@ | |||
| {% if site.use_download_button -%} | |||
| <a href="{{ page.interact_link | relative_url }}" download> | |||
There was a problem hiding this comment.
@TomDonoghue apparently the download> part here is what causes the browser to auto-download instead of trying to open the file
There was a problem hiding this comment.
That's cool - nice find!
|
an example page w/ a download link: https://640-137292803-gh.circle-artifacts.com/0/html/features/notebooks.html |
| @@ -0,0 +1,4 @@ | |||
| {% if site.use_download_button -%} | |||
| <a href="{{ page.interact_link | relative_url }}" download> | |||
There was a problem hiding this comment.
That's cool - nice find!
| @@ -1,5 +1,4 @@ | |||
| """A helper script to "clean up" all of your notebooks and | |||
| generated markdown files.""" | |||
| """A helper script to "clean up" all of your generated markdown, zip, and HTML files.""" | |||
There was a problem hiding this comment.
I had updated this comment, since at the time I was 'cleaning' out created zip files, but as is, the reference to zip files is outdated.
jupyter_book/build.py
Outdated
| @@ -9,6 +9,7 @@ | |||
| from tqdm import tqdm | |||
| import numpy as np | |||
| from glob import glob | |||
| from zipfile import ZipFile | |||
There was a problem hiding this comment.
This import is currently not used - can be dropped.
|
@choldgraf - thanks for porting this to the new organization! I like the update to just grabs the notebook file itself - nice and clean. Couple small comments on the code above. There do appear to be some os / browser specific characteristics. When I download the the file from the example you linked above, browsers seem to recognize the file as a txt file. This is broadly fine, except that for some reason Safari then appends a '.txt' to the downloaded file, which is annoying. I think Safari just in general has different approaches. It also, by default, opens zip files, so in the zip version, Safari ends up with the notebook file. On my system (MacOS), Firefox & Chrome are both more hands off - downloading the unzipped files normally (as .ipynb) and keeping the zip files as zip. ^I'm not sure that there's anything to be done about this, just worth knowing I guess. Possibly downloading zip files does have a more consistent effect, than possible variation of how .ipynb files are treated? Also - presumably, we have a similar desired workflow to how sphinx-gallery downloads notebooks and files. From a quick scan, they don't seem to zip by default for single files, but something is different which avoids the file extension problem in safari, for me at least. Haven't had a chance yet to figure out how they're doing it, but maybe worth looking into. |
|
Thanks for the feedback @TomDonoghue ! I cleaned up the extra text you flagged, and also added a short section in the FAQ about behavior on different browsers. Gonna merge this one and start iterating from there! |
|
All looks good! Thanks for finishing this up & merging @choldgraf |
This riffs off of #38 which @TomDonoghue pioneered getting a download link to work.
In that PR we were using a Zip file to download because it seemed like that was the only way to get the file to "auto" download. I found a blog post (https://www.philowen.co/blog/force-a-file-to-download-when-link-is-clicked/) that showed how to make the file automatically download even if it's an ipynb file. So, this PR does two things:
I still think that it could be a good idea to use zip files, but for now since we were only downloading the single notebook file anyway, this seems like a much simpler path that we can build upon. @TomDonoghue what do you think?