Home

Awesome

colab-convert

Converts ipython/Google Colab Notebooks into runable Python code and vice versa

Features

Atom/Hydrogen or VSCode/Python allows creating a python file split into cells with # %% separators having the ability to run cells via the backend of a Jupyter session and interactively show results back.

VSCode

Jupyter ipynb notebook

Install & Basic Usage

pip install colab-convert
Usage: colab-convert <input_file> <output_file> <flags>

<input_file>: input file to convert
<output_file>: output file to write to
<flags>: extra flags to pass to the converter

all flags are optional and have set defaults for best results
use flags to enable or disable certain functions on/off by default
colab-convert in.ipynb out.py -nc -rm -o

Default options and Flags

Default Flags Set (defaults are determined by input file)
  ipynb input file:
    [YES] convert magic , [YES] auto comment , [YES] imports , [NO] Outputs
  py input file:
    [NO] convert magic , [NO] auto comment , [NO] imports , [NO] Outputs

Available Flags
  toggle certain items on or off

  --retain-magic  (-rm)  : Keep magic commands in the output
      .py default    [ON]
      .ipynb default [OFF]
  --convert-magic  (-cm) : Convert magic commands to python code
      .py default    [OFF]
      .ipynb default [ON]
  --auto-comment  (-ac)  : Convert unsupported magic commands to comments
      .py default    [OFF]
      .ipynb default [ON]
  --no-comment  (-nc)    : Keep unsupported magic commands
      .py default    [ON]
      .ipynb default [OFF]
  --no-imports  (-ni)    : Do not add imports from converted magic commands
      .py default    [OFF]
      .ipynb default [OFF]
  --outputs  (-o)        : Outputs to console of conversions and commented lines.
      .py default    [OFF]
      .ipynb default [OFF]
  --lang=  (-l=)         : Language to change output messages to
       default [English]
      --lang=en_US
      en_US, en, english, eng, nl_NL, nl, dutch, dut, nlt, nederlands

Conversion Code used

<details> <summary>>click me to see code<</summary>

Magic commands using bang (!)

for this particular magic we send the command to the subprocess system and print the results

#   !git clone https://test.com/test/test.git

sub_p_res = subprocess.run(['git', 'clone' ,'https://test.com/test/test.git'], stdout=subprocess.PIPE).stdout.decode('utf-8')
print(sub_p_res)

Magic commands using percent (%)

%pwd - get current working directory

#   %pwd

os.getcwd()

%ls - list items in directory

#   %ls

os.listdir()
#   %ls folderName/subFolder

os.listdir('folderName/subFolder')

%cd - change directory

#   %cd test-directory

os.chdir('test-directory')

%mkdir - make a new directory

#   %mkdir test/newFolder

os.makedirs('test/newFolder')

%mv - move file from one location to another

#   %mv testFile.txt testFolder/

shutil.move('testFile.txt', 'testFolder/testFile.txt')

%cp - copy file from one location to another

#   %cp testFolder/testFile.txt newFolder/newTestFile.txt

shutil.copy('testFolder/testFile.txt', 'newFolder/newTestFile.txt')

%cat - show the output of a file in standard format

#   %cat testFolder/testFile.txt

cat_read_file = open('testFolder/testFile.txt', 'r')
cat_read_text = cat_read_file.read()
print(cat_read_text)
cat_read_file.close()

%env & %set_env - get, set or list environmental variables

this command actually has 5 ways to be used

%env
    lists all environment variables/values
%env var
    get value for var
[%env or %set_env] var val
    set value for var
[%env or %set_env] var=val
    set value for var
[%env or %set_env] var=$val
    set value for var, using python expansion if possible
#   %env

for k, v in os.environ.items():
    print(f'{k}={v}')
#   %env var

os.environ['var']
#   %env var value
#   %set_env var value

os.environ['var'] = 'value'
#   %env var=value
#   %set_env var=value

os.environ['var'] = 'value'
#   %env var=$value
#   %set_env var=$value

os.environ['var'] = '$value'

%pip - install a pip package or other pip functions

#   %pip install colab-convert

pip_sub_p_res = subprocess.run(['pip', 'install', 'colab-convert'], stdout=subprocess.PIPE).stdout.decode('utf-8')
print(pip_sub_p_res)

%conda - install a conda package or other conda functions

#   %conda install colab-convert

conda_sub_p_res = subprocess.run(['conda', 'install', 'colab-convert'], stdout=subprocess.PIPE).stdout.decode('utf-8')
print(conda_sub_p_res)

Unsupported Magic Commands

these will be commented out

#   %quickref

#<cc-cm> %quickref
</details>

Example

colab-convert examples/plot.py examples/plot.ipynb

or

colab-convert examples/plot.ipynb examples/plot.py

Markdown cells are converted to python multiline strings '''. Code cells are left as is.

eg. will render header section

"""
## Matplot example

** Run the cell below to import some packages and show a line plot **
"""

# %% is used by vscode as the cell marker on which 'Run Cell' action is available.

eg. will render a code cell

# %%
import matplotlib.pyplot as plt

Metadata is converted from notebooks into .py and vise versa using # !! to denote the meta data lines in the .py files

eg.

# %%
# !! {"metadata":{
# !!   "id": "PlotIt"
# !! }}
import matplotlib.pyplot as plt

eg. final code block must include atleast this

# %%
# !! {"main_metadata":{
# !!   "anaconda-cloud": {},
# !!   "kernelspec": {
# !!     "display_name": "Python 3",
# !!     "language": "python",
# !!     "name": "python3"
# !!   },
# !!   "language_info": {
# !!     "codemirror_mode": {
# !!       "name": "ipython",
# !!       "version": 3
# !!     },
# !!     "file_extension": ".py",
# !!     "mimetype": "text/x-python",
# !!     "name": "python",
# !!     "nbconvert_exporter": "python",
# !!     "pygments_lexer": "ipython3",
# !!     "version": "3.6.1"
# !!   }
# !! }}

Troubleshooting

export LANG=C.UTF-8
export PYTHONIOENCODING=utf-8
export PYTHONUTF8=1

should be enough. Also try setting default Bash settings to UTF-8: [Options] - [Text] - [Locale / Character set] - [C / UTF-8]. It might affect all Bash runs so there would be no need to setting encoding every time.

Credits

colab-convert is a fork of the ipynb-py-convert.