[jupyter] use zipping for stored json #19970
Conversation
|
Here one can compare new produced notebook code: https://github.com/linev/misc/blob/master/notebook/Simple.ipynb?short_path=26ba30e#L30-L35 with old stored notebooks: https://github.com/linev/misc/blob/master/notebook/Simple0.ipynb?short_path=58fd63c#L32-L36 |
Test Results 20 files 20 suites 3d 16h 23m 54s ⏱️ Results for commit 47e762b. ♻️ This comment has been updated with latest results. |
There was a problem hiding this comment.
Thank you very much for implementing this so quickly!
Also it's good that you used zlib for the compression, as ROOT is always guaranteed to include this compression library (either via builtin_zlib or not). So it should be possible to display notebooks produced with any build configuration of ROOT with any other build configuration.
|
Any supported compression algorithm can be used here. |
It idea to provide string which can be embed in any JavaScript code and then unzipped whith JSROOT.unzipJSON function. Primary aim is to improve json handling in jupyter
Use TBufferJSON.zipJSON method. dev version of JSROOT has to be used to be able unpack such string.
1. Introduce unzipJSON for processing zip strings 2. Better detect div layout when creating canvas svg element 3. Fix minor latex problem 4. Detect default histogram in TGraph to correctly set ranges then
linev
force-pushed
the
notebook_zipped
branch
from
53577c1 to
47e762b
Compare
Up to now plain json code embed into ipython notebooks.
This is very inefficient and can make notebook huge.
Especially because of fact that notebook stored as json - so any double-quote
"symbol should be escaped.Therefore use standard compression function of ROOT and then base64 encoding to produce string suitable for embeding. This done in
TBufferJSON::zipJSON()method.On JSROOT side just provide correspondent
unzipJSON()function.And modify jupyter
utils.pycode to use such functionality.