Notebook: Strip trailing whitespace

[jakirkham]

Would be nice if there was an option that automatically strips trailing whitespace.

[willingc]

@jakirkham Just to clarify... trailing cells at the end of a notebook or from markdown entered in a cell or something completely different.

[fperez]

Closing for now due to lack of info (trying to aggressively triage here...). Can reopen if more info comes along.

[jakirkham]

That's fine. It's a pretty minor point.

Sorry to not respond for so long, @willingc. I had to restructure my notifications around that time so I could keep track of things better. What I was meaning was trailing whitespace in notebook cells. PyCharm does something similar on saves and I was wondering if such a thing could be pursued with the notebook as well.

[gabrielboucher]

Could this feature be provided through the mean of an extension?

[takluyver]

Yes, probably. On the server side, you could set up a save hook. And there is probably some way to do it in the JS side as well, but I'm less sure of how that would look.

[casperdcl]

following @takluyver's suggestion:

jupyter notebook --generate-config

Then add to the generated file:

def stripWS(t):
    return '\n'.join([i.rstrip() for i in t.split('\n')])

def scrub_output_pre_save(model=None, **kwargs):
    """strip trailing space before saving"""
    if model['type'] == 'notebook':
        # only run on nbformat v4
        if model['content']['nbformat'] != 4:
            print("skipping WS stripping since `nbformat` != 4")
            return
        print("Stripping WS")
        for cell in model['content']['cells']:
            if cell['cell_type'] != 'code':
                continue
            cell['source'] = stripWS(cell['source'])
    elif model['type'] == 'file':
        if model['format'] == 'text':
            print("Stripping WS")
            model['content'] = stripWS(model['content'])

c.ContentsManager.pre_save_hook = scrub_output_pre_save

[casperdcl]

really odd that there isn't an editor option to do this

[xapple]

This should really be re-opened. The functionality of stripping trailing whitespace is pretty basic and expected in any tool that lets you write and edit code.

[xapple]

The ticket was originally closed for lack of information. Here is a precise description of what stripping trailing whitespace means:

Imagine the following python code in a Jupyter cell:

␣␣␣␣def␣foo(x):␣
␣␣␣␣␣␣␣␣print(x)␣␣␣
␣␣␣␣␣␣␣␣return␣1+x␣␣␣␣␣␣␣␣

Upon saving this code, it would automatically be reformatted to the following:

␣␣␣␣def␣foo(x):
␣␣␣␣␣␣␣␣print(x)
␣␣␣␣␣␣␣␣return␣1+x

To view whitespaces, they have been replaced with ␣

[xapple]

In the meantime, based on @casperdcl answer adding this to my ~/.jupyter/jupyter_notebook_config.py file seems to work:

#------------------------------------------------------------------------------
# Custom whitespace stripping
#------------------------------------------------------------------------------
def strip_white_space(text):
    return '\n'.join([line.rstrip() for line in text.split('\n')])

def scrub_output_pre_save(model=None, **kwargs):
    """Auto strip trailing white space before saving."""
    # If we are dealing with a notebook #
    if model['type'] == 'notebook':
        # Don't run on anything other than nbformat version 4 #
        if model['content']['nbformat'] != 4:
            print("Skipping white space stripping since `nbformat` != 4.")
            return
        # Apply function to every cell #
        print("Stripping white space on a notebook.")
        for cell in model['content']['cells']:
            if cell['cell_type'] != 'code': continue
            cell['source'] = strip_white_space(cell['source'])
    # If we are dealing with a file #
    if model['type'] == 'file':
        if model['format'] == 'text':
            print("Stripping white space on a file.")
            model['content'] = strip_white_space(model['content'])

c.ContentsManager.pre_save_hook = scrub_output_pre_save

[jasongrout]

In the meantime, based on @casperdcl answer adding this to my ~/.jupyter/jupyter_notebook_config.py file seems to work:

An advantage of doing it on the server is that it automatically works for JupyterLab too then.