Change Black linelength defaults?

[coolum001]

Hi

First of all, thank you for this extension.

I plan to use nb_black and Jupyter Notebooks to generate blog posts, where code blocks are indented. This means I would like to set the line length to 65 (ie not the default).

I would also like to preserve single quote strings (a black option).

Can you indicate where in your code it might be possible to change the default parameters for black?

Second: I am running Jupyter lab Version 0.35.3, in an environment as below

python version : 3.7.1 (default, Dec 10 2018, 22:54:23) [MSC v.1915 64 bit (AMD64)]
python environment : ac5-py37
pandas version : 0.23.4
current wkg dir: C:\Users\donrc\Documents\JupyterNotebooks\OSMNXNotebookProject\develop
Notebook name: Communities.ipynb
Notebook run at: 2019-08-03 18:45:44.954269 local time
Notebook run at: 2019-08-03 08:45:44.954269 UTC
Notebook run on: Windows-10-10.0.18362-SP0

<IPython.core.display.Javascript object>

You will notice that Jupyterlab displays after each cell execution (I assume this is unintended behaviour)

<IPython.core.display.Javascript object>

I am running nb_black as below

(ac5-py37) C:\Users\donrc>conda list nb_black
# packages in environment at D:\Anaconda3\envs\ac5-py37:
#
# Name                    Version                   Build  Channel
nb_black                  1.0.6                      py_0    conda-forge

Now I can't reproduce this behaviour in smaller Notebooks, so you may see fit to ignore this until I can :)

Third, invoking %load_ext lab_black seems to stick with the notebook. Even if you reload the notebook and delete the %load_ext cell, the black formatting is still applied to new cells. You may wish to mention that you turn it off by a %unload_ext lab_black call.

Happy to supply more info if needed

Don Cameron

[coolum001]

Hi Don Cameron here.

I think that the <IPython.core.display.Javascript object> display after each cell was because I had loaded nb_black into a JupyterLab, environment, and not lab_black. Unloading the correct extension fixed the problem.

Sorry to waste your time on this, Don

[vmarceau]

Hey Don,

Just curious if you have found a way to change the default parameters for black when using nb_black (especially max line length and single quotes)? I would also be interested in doing that.

Thanks!

[michaelaye]

configuring black itself doesn't work? https://black.readthedocs.io/en/stable/pyproject_toml.html

[coolum001]

Hi I didn't want to reconfigure black as a whole, because I use it in VSCODE, and I want it as it is.

My use-case is creating blog entries from JupyterLab notebooks, where because my blog-engine (Pelican, built-texts theme, ipynb.markup plugin) has quite a narrow column width. So I just want the code in some Notebooks (but not all) formatted as narrow. For an example, see https://coolum001.github.io/jupyterblog.html

I have hacked up a solution:

in lab_black.py, there is a function defined _format_code(code)

My version of this now reads

    def _format_code(code):
#        return format_str(src_contents=code, mode=FileMode())
        my_mode = FileMode()
        my_mode.line_length = 65
        my_mode.string_normalization = False
        return format_str(src_contents=code, mode=my_mode)

When I stop being lazy, I might work out how to read parameters from the %load_ext command :) , but this works for me.

Thanks again for the package Don Cameron

[coolum001]

Just to be completely accurate, the blog post referenced above was formatted with a lab_black.py locally in the same directory as the Notebook, with _format_code() defined as:

    def _format_code(code):
#        return format_str(src_contents=code, mode=FileMode())
        my_mode = FileMode()
        my_mode.line_length = 60
        my_mode.string_normalization = False
        return format_str(src_contents=code, mode=my_mode)

[michaelaye]

Note that the black configuration file works locally as well, so you can have completely different settings per folder by having that config file in there.

[coolum001]

Michael

Didn't know that (obviously). I will give it a try, as being cleaner than having private versions of lab_black Thanks Don

[coolum001]

Hi I have a directory with a pyproject.toml file that looks like:

[tool.black]
skip-string-normalization = true
line-length = 60

and a batch (command line) invocation of black works as expected, giving me the message:

Using configuration from C:\Users\donrc\Documents\JupyterNotebooks\SymAlgebraExamplesNotebookProject\develop\pyproject.toml.

However, %load_ext lab_black in a JupyterLab Notebook in the same directory, still converts single quotes to double. I suspect that the Black formatting is being done in a backend server or process, that can't see my directory to find my TOML file, because it has a different current working directory.

This happens even when I run the jupyter lab command that started that session, from inside that directory.

So I might have to go back to my handcrafted lab_black.py file after all :( Thanks Don Cameron

[michaelaye]

Ah that’s a pity. Maybe that could be changed in this package? It would be rather better if it worked exactly like called in a terminal. On Aug 8, 2019, 21:18 -0600, Don Cameron notifications@github.com, wrote:

Hi I have a directory with a pyproject.toml file that looks like: [tool.black] skip-string-normalization = true line-length = 60 and a batch (command line) invocation of black works as expected, giving me the message: Using configuration from C:\Users\donrc\Documents\JupyterNotebooks\SymAlgebraExamplesNotebookProject\develop\pyproject.toml. However, %load_ext lab_black in a JupyterLab Notebook in the same directory, still converts single quotes to double. I suspect that the Black formatting is being done in a backend server or process, that can't see my directory to find my TOML file, because it has a different current working directory. This happens even when I run the jupyter lab command that started that session, from inside that directory. So I might have to go back to my handcrafted lab_black.py file after all :( Thanks Don Cameron — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[dnanhkhoa]

Hi, sorry for late reply and thank Michael for solutions so far, I will try to figure out why it cannot detect local configuration file

[n8henrie]

I'd also love to see this extension respect pyproject.toml's black settings, but no luck so far.

I've gone the route above and manually edited the file in my site-packages which works.

Looking at the black code, I think the issue is that format_str doesn't call the functions related to finding and loading the config file; it relies on the passed-in FileMode, which has been set to defaults in the extension. Unfortunately I don't see an easy way to import just the config loading code from black (tried by importing click.testing.FileRunner or something to that effect but couldn't sort it out).

[n8henrie]

Here's one way: https://github.com/arokem/python-matlab-bridge/blob/master/pymatbridge/matlab_magic.py#L68 accept kwargs into load_ipython_extension and allow users desiring custom configuration to manually import the extension and run the function with their config.

The below is just a proof-of-concept patch, I don't use yapf so I don't know if it breaks anything there, and I only implemented configuration for line-length. The screenshot below shows how to load it with config, and shows that if loaded with %load_ext, it successfully loads but will use the black defaults.

diff --git a/lab_black.py b/lab_black.py
index 9dbf6f4..bec76e3 100644
--- a/lab_black.py
+++ b/lab_black.py
@@ -23,10 +23,11 @@ from IPython.display import Javascript, display
 __BF_SIGNATURE__ = "__BF_HIDDEN_VARIABLE_{}__"

 if sys.version_info >= (3, 6, 0):
-    from black import format_str, FileMode
+    import black

-    def _format_code(code):
-        return format_str(src_contents=code, mode=FileMode())
+    def _format_code(code, mode):
+
+        return black.format_str(src_contents=code, mode=mode)

 else:
@@ -170,9 +171,10 @@ def _recover_magic_commands(cell, hidden_variables):

 class BlackFormatter(object):
-    def __init__(self, ip, is_lab):
+    def __init__(self, ip, is_lab, mode):
         self.shell = ip
         self.is_lab = is_lab
+        self.mode = mode

     def __set_cell(self, unformatted_cell, cell, cell_id=None):
         if self.is_lab:
@@ -215,7 +217,7 @@ class BlackFormatter(object):
                 # Transform magic commands into special variables
                 cell = _transform_magic_commands(unformatted_cell, hidden_variables)

-                formatted_code = _format_code(cell)
+                formatted_code = _format_code(cell, self.mode)

                 # Recover magic commands
                 formatted_code = _recover_magic_commands(
@@ -230,10 +232,14 @@ class BlackFormatter(object):
 black_formatter = None

-def load_ipython_extension(ip):
+def load_ipython_extension(ip, line_length=79):
     global black_formatter
+    mode = black.FileMode(
+        target_versions={black.TargetVersion.PY38}, line_length=line_length,
+    )
+
     if black_formatter is None:
-        black_formatter = BlackFormatter(ip, is_lab=True)
+        black_formatter = BlackFormatter(ip, is_lab=True, mode=mode)
         ip.events.register("post_run_cell", black_formatter.format_cell)

[msm1089]

@n8henrie 's solution is best if you may ned to configure the line length. Shorter way, if you will only ever need to change the line length rarely or never:

replace

https://github.com/dnanhkhoa/nb_black/blob/d4bd99e048da05dc1e2389c2b8fdf55ed9c9feb0/lab_black.py#L26-L29

with

    from black import format_str, FileMode, TargetVersion

    def _format_code(code):
        mode = FileMode(target_versions={TargetVersion.PY38},
                              line_length=79)
        return format_str(src_contents=code, mode=mode)

setting line_length as desired.

[n8henrie]

Work seems to have slowed down, so I created a fork that addresses this issue and adds a few other features: https://github.com/n8henrie/jupyter-black