%blockrun button executes even after clicking the button on google colab

[DheerajKN]

Error reproducible at
cell 1
try:
import ipython_blocking
except ImportError:
!pip install ipython_blocking
import ipython_blocking

cell 2
from ipywidgets import Dropdown, Button, Output
from IPython.display import display, clear_output

outs = Output()
query_selector = Dropdown(
options=['Select', 'Q1', 'Q2'],
value='Select',
description='Query Type:'
)

process_button = Button(
description='Process',
disabled=False
)

def proc_button(bb):
with outs:
clear_output()
display(process_button)

display(query_selector)
query_selector.observe(proc_button, names='value')
display(outs)

cell 3
%blockrun process_button

cell 4
print(query_selector.value)

causing the cell 3 to run continuously even after the button is selected.

[kafonek]

@DheerajKN thanks for opening this issue. I'm able to reproduce the issue, and I see this in the javascript console -

Error evaluating Javascript output:  ReferenceError: Jupyter is not defined
    at eval (eval at <anonymous> (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:1), <anonymous>:1:1)
    at eval (<anonymous>)
    at qa.eval [as h] (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:99)
    at sa (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:15)
    at ua.next (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:15)
    at eval (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:16)
    at new Promise (<anonymous>)
    at wa (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:16)
    at z (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:16)
    at ie (output_binary.js?vrz=colab-20210322-060003-RC00_364283823:97)

I'm not too familiar with what front-end javascript changes Colab makes over generic classic Notebook. Classic Notebook has a Javascript object that can control the Notebook DOM. In ipython_blocking, %blockrun calls this function to execute everything below,

def run_all_below():
    js = 'Jupyter.notebook.select_next().execute_cells_below()'
    display(Javascript(js))

If you know the equivalent Colab front-end API to use, I'm happy to look at this more deeply or review a PR. Thanks.

[DheerajKN]

I am not much well-versed with using magic functions. Can you provide me the code run magic function from cloned GitHub project?

[DheerajKN]

Can you try to replace Jupyter with IPython as maybe colab doesn't support Jupyter and instead uses IPython
def run_all_below():
js = 'IPython.notebook.select_next().execute_cells_below()'
display(Javascript(js))

[kafonek]

@DheerajKN you can test that idea out by going to colab, opening up your browser developer tools, and trying to execute IPython.notebook.select_next().execute_cells_below() in a console window. Unfortunately IPython is not defined, just like Jupyter.

There are two things to research here:

• What is the equivalent to Jupyter.notebook.select_next().execute_cells_below() in Google Colab

• How could ipython_blocking identify that it is in a Google Colab environment and should use the alternate syntax

I see that there is a google-colaboratory tag on stackoverflow that might be a good way to seek help on those questions. Unfortunately it doesn't appear too active, for instance https://stackoverflow.com/questions/65984431/run-all-cells-command-in-google-colab-programmatically is very close to what we're asking but there's no answer.

[DheerajKN]

1. For the first I couldn't find any so dropped a StackOverflow Request: https://stackoverflow.com/questions/66786245/what-is-the-equivalent-to-jupyter-notebook-select-next-execute-cells-below-i
2. This did the trick:

if 'google.colab' in str(get_ipython()):
  print('Running on CoLab')
else:
  print('Not running on CoLab')

[saraswathykrk]

Error reproducible at
cell 1
try:
import ipython_blocking
except ImportError:
!pip install ipython_blocking
import ipython_blocking

cell 2
from ipywidgets import Dropdown, Button, Output
from IPython.display import display, clear_output

outs = Output()
query_selector = Dropdown(
options=['Select', 'Q1', 'Q2'],
value='Select',
description='Query Type:'
)

process_button = Button(
description='Process',
disabled=False
)

def proc_button(bb):
with outs:
clear_output()
display(process_button)

display(query_selector)
query_selector.observe(proc_button, names='value')
display(outs)

cell 3
%blockrun process_button

cell 4
print(query_selector.value)

causing the cell 3 to run continuously even after the button is selected.

Hi @DheerajKN,

Can you please let me know how you finally resolved this in colab?

Regards
Saraswathy.

[DheerajKN]

@saraswathykrk I couldn't resolve this issue. As we were not able to find an equivalent function for execute_cells_below that is supported in Google Colab. StackOverflow for the same was provided.
It was part of my assignment, so just submitted it in a local Jupyter notebook. Locally it works!!

[saraswathykrk]

Thanks for the reply @DheerajKN.

Did you also try the below code given for Pause Colab Outputs? Just thinking if we could leverage this and achieve the blocking.

display(IPython.display.Javascript('''
const promise = new Promise((resolve, reject) => {
const script = document.createElement('script');
script.src = 'data:,window.value = "hello world!"';
script.onload = resolve;
script.onerror = reject;
document.head.appendChild(script);
});
// Pause subsequent outputs until the script has been loaded.
google.colab.output.pauseOutputUntil(promise);
'''))

display(IPython.display.Javascript('''
// Can immediately reference scripts loaded earlier since
// output processing was blocked on them.
document.body.appendChild(document.createTextNode(window.value));
'''))

Regards
Saraswathy.