Smoother Python with automatic imports | pyflyby

2021-12-20T11:34:47Z

This is not a flaky works half the time kind of plugin, it’s a seriously smooth editing experience. I’ve just started using pyflyby, and it is solid so far. I have automatic imports on every save of a python file in neovim, and automatic imports on every command in ipython.

I can’t tell you how pumped I am for this, and how good its felt to use over the past few weeks. It’s glorious.

YouTube video #

Listen to me rant on how great pyflyby is

https://youtu.be/2QW5DJiEJH4

Give the video a watch, I did not have noise-cancelling on in obs. My apologies for the background hum and the mic stand bumps. I did my best to fix them up.

Installation #

How to install pyflyby for automatic python imports

pyflyby is hosted on pypi, so you can get it with pip. I have had no issues installing it on 3.8+ so far.

pip install pyflyby

Configuration setup with stow #

always stow your dotfiles

If you’re going to configure any of your tools the first thing you should do is set it up with stow, seriously don’t sleep on the stow. If you don’t have stow installed or choose not to use stow you can skip this part.

cd ~/dotfiles
mkdir ipython
touch ipython/.pyflyby
stow ipython

Seriously don’t sleep on the stow.

How to Configure pyflyby #

it’s just a file full of import statements

pyflyby is configured simply by putting all of your import statements that you want to automatically import into your ~/.pyflyby file. You can import pandas, from pandas import DataFrame, or even import pandas as pd, and all of these will work pretty much as expected.

# comments start with a #
# import your favorite libraries
import visidata as vd
import fsspec
import difflib
import s3fs
import seaborn as sns
import plotly

# also supports from imports
from rich.layout import Layout
from rich.live import Live

# duplicates are allowed
import plotly
import plotly

# duplicate names from different libraries are not allowed
import copy
from numpy import copy

Add all the things you would like to be imported automatically, just as you would import them. I went kinda crazy and added over 200 to mine based on packages that I use.

Commas are even supported #

yep all the import styles are supported

This following example will set up auto import for both DataFrame and Series, they will both work separately. I removed these from my config as I felt it was cleaner without, but it works with them.

from pandas import DataFrame, Series

Even imports with a comma will be treated separately.

Jupyter note #

Both work the same, use what your comfortable with

I only really mention ipython here, but the same all applies to Jupyter as well. I just really like ipython itself, c’mon its right there in the terminal integrating with the rest of your terminal experience so well.

Ipython setup #

Automatically import python libraries in ipython with pyflyby

The recommended way to set uppyflybyfrom the docs is to run the following magic command. This works well, but I want even less typing, I want pyflyby automatically installed and importing things without me even thinking about it.

%load_ext pyflyby

Ipython setup next level #

automatically import modules in python without %load_ext

I really want pyflyby to just work in every environment without me thinking much about it. I want it to load automatically, and even to attempt to install itself if it’s missing.

from IPython import get_ipython
import subprocess


ipython = get_ipython()

try:
    ipython.magic("load_ext pyflyby")
except ModuleNotFoundError:
    print("installing pyflyby")
    subprocess.Popen(
        ["pip", "install", "pyflyby"],
        stdout=subprocess.DEVNULL,
        stderr=subprocess.DEVNULL,
    ).wait()
    ipython.magic("load_ext pyflyby")

Note: if installation fails you will still make it into ipython, there will just be a traceback to the failed command as you enter.

I’ve had zero issues with this, but if there ever comes a time when it does not work in certain environments for you. I’d strongly suggest you to add this to a separate profile.

Ipython-Config

Check out this article for a bit more in depth ipython configuration

ipython auto import examples #

pyflyby can import all the various import types just fine.

import something
from module import something
import something as alias

df = pd.read_csv("https://waylonwalker.com/cars.csv")
[PYFLYBY] import pandas as pd

Getting Help #

Want help on something that you have in your pyflyby config, just give it the ?, ??, or help and pyflyby will import it for you.

Popen?

Just Ask Ipython for help

Autocomplete #

This is next level python auto-import

pyflyby even goes as far as helping tab completion. If you try to tab complete Pop it will complete to Popen without even adding Popen to your local namespace. If you ask for something inside of a module i.e. requests.<tab>, then it will import requests.

# does not populate the namespace
Pop<tab>

# !!does populate the local namespace
requests.<tab>

What happens when a module is not installed #

ModuleNotFoundError

When you are in an environment where you do not have a module installed that is in your pyflyby config, it will throw a ModuleNotFoundError when it tries to import, and it will not import or try to install for you. You will have to change environments or install that module.

❯ pd?
[PYFLYBY] import pandas as pd
[PYFLYBY] Error attempting to 'import pandas as pd': ModuleNotFoundError: No module named 'pandas'
[PYFLYBY] Traceback (most recent call last):
[PYFLYBY]   File "/home/u_walkews/.local/lib/python3.8/site-packages/pyflyby/_autoimp.py", line 1610, in _try_import
[PYFLYBY]     exec_(stmt, scratch_namespace)
[PYFLYBY]   File "<string>", line 1, in <module>
[PYFLYBY] ModuleNotFoundError: No module named 'pandas'
Object `pd` not found.

❯ df = pd.read_csv("https://waylonwalker.com/cars.csv")
<ipython-input-3-69b040434562>:1 in <module>

NameError: name 'pd' is not defined

nvim pyflyby setup #

automatically importing python modules in vim, neovim, nvim

This is by far the best part of this article. It makes development so fluid. It’s not necessarily all about the speed. It really helps you move at the speed of your thoughts, without needing to worry about imports. Remembering where deeply nested modules are does not need to be a thing.

function! s:PyPreSave()
    Black
endfunction

function! s:PyPostSave()
    execute "silent !tidy-imports --black --quiet --replace-star-imports --action REPLACE " . bufname("%")
    execute "e"
endfunction

:command! PyPreSave :call s:PyPreSave()
:command! PyPostSave :call s:PyPostSave()

augroup waylonwalker
    autocmd!
    autocmd bufwritepre *.py execute 'PyPreSave'
    autocmd bufwritepost *.py execute 'PyPostSave'
    autocmd bufwritepost .tmux.conf execute ':!tmux source-file %'
    autocmd bufwritepost .tmux.local.conf execute ':!tmux source-file %'
    autocmd bufwritepost *.vim execute ':source %'
augroup end

refactoring #

This is where it really shines

This setup really shines when you are refactoring. You can freely move modules and classes around without worrying about bringing imports with them. Often when refactoring some modules from one file to another the most tedious part is editing the imports. Often you can’t even grab whole lines because there are several imports and some are needed in both places but not all. pyflyby handles all this like a champ.

Where to install for vim #

just make sure the tidy-imports command is available to vim

pyflyby goes into the environment that you have active at the time that you start neovim. Typically, this is the virtual environment that I am using for the project I am editing.

What gets imported/removed #

only give me what I actually use

Anything within the base config of pyflyby or your own config specified in ~/.pyflyby will get automatically imported if it is used within the file/console. If you are working in a file, and stop using a module, it will automatically get removed.

Anything that is used, and found in the config is added
Anything that is unused gets removed

Where does it put imports #

after the last import

pyflyby does not sort imports into paragraphs or by category. When it needs to add new imports. It will find the last paragraph of imports in your file, add the new one, and sort that paragraph alphabetically.

from collections import Counter

import requests

from plugins.custom_seo import post_render
# <-- pyflyby will put the import here

What about isort #

put those imports where they go

I did not like that I was getting pre-commit issues when using pyflyby, so I added isort to my chain of autocommands to automatically run isort and make my pre-commit happy.

function! s:PyPostSave()
    execute "silent !tidy-imports --black --quiet --replace-star-imports --action REPLACE " . bufname("%")
    execute "silent !isort " . bufname("%")
    execute "e"
endfunction

Let’s write some code

def get():
    """
    Get all the posts from waylonwalker.com.

    Yes theres an rss feed, you should be subscribed if your not already.

    Oh, and we don't need no stinkin error handing because it's always live
    """
    r = requests.get("https://waylonwalker.com/rss.xml")
    return r.content

Save it and pyflyby will inject requests into our file automatically, no need to type that out anymore.

import requests

def get():
    """
    Get all the posts from waylonwalker.com.

    Yes theres an rss feed, you should be subscribed if your not already.

    Oh, and we don't need no stinkin error handing because it's always live
    """
    r = requests.get("https://waylonwalker.com/rss.xml")
    return r.content

What about init / api’s #

careful to fill in the __all__ like you are supposed to

Files such as init.py often import things they do not need, this is simply there for a convenience of the library user and to make the api cleaner. These type of modules should implement a __all__ list of all the unused things that are imported according to pep8. Pyflyby will remove any unused modules unless they are in the __all__ list.

# snippet from kedro.extras.datasets.pandas

__all__ = [
    "CSVDataSet",
    "ExcelDataSet",
    "FeatherDataSet",
    "GBQTableDataSet",
    "ExcelDataSet",
    "AppendableExcelDataSet",
    "HDFDataSet",
    "JSONDataSet",
    "ParquetDataSet",
    "SQLQueryDataSet",
    "SQLTableDataSet",
]

py command #

one liners that need imports

pyflyby also comes with a cli command to run one liners. It’s pretty genius, I’m sure I will find a use or two for it, but so far it’s been more of a novelty for me.

py help pd
py help pd.DataFrame

py pd.read_csv 'https://waylonwalker.com/cars.csv'

Links #

Ipython Ninjitsu

2020-12-14T00:00:00Z

?docstring
??sourcecode
%run
%debug
%autoreload
%history
autoformat
%reset
!shell commands

?docstring #

Stop going to google everytime your stuck and stay in your workflow. The ipython ? is a superhero for productivity and staying on task.

from kedro.pipeline import Pipeline
Pipeline?

Init signature:
Pipeline(
    nodes: Iterable[Union[kedro.pipeline.node.Node, ForwardRef('Pipeline')]],
    *,
    tags: Union[str, Iterable[str]] = None,
)
Docstring:
A ``Pipeline`` defined as a collection of ``Node`` objects. This class
treats nodes as part of a graph representation and provides inputs,
outputs and execution order.
Init docstring:
Initialise ``Pipeline`` with a list of ``Node`` instances.

Args:
    nodes: The iterable of nodes the ``Pipeline`` will be made of. If you
        provide pipelines among the list of nodes, those pipelines will
        be expanded and all their nodes will become part of this
        new pipeline.
    tags: Optional set of tags to be applied to all the pipeline nodes.

Raises:
    ValueError:
        When an empty list of nodes is provided, or when not all
        nodes have unique names.
    CircularDependencyError:
        When visiting all the nodes is not
        possible due to the existence of a circular dependency.
:

Note This does jump you into a pager, a j,k or up, down to navigate, q to quit.

??sourcecode #

Docstring not enough for you use case. I often run into cases where the docstring is not clear enough and I need to see the implementation for myself to see what a function does.

%run #

I turned my nose up at this one, prior to seeing the famous I don’t like notebooks by Joel Grus. My first snobby reaction was that developing modules and using autoreload was superior. I have since realized there is a place for %run, and it can cut down on some keystrokes to import, setup, and run even when developing in modules.

%debug #

ipython comes with a post-mortem debugger, and it can be a lifesaver. If we have a long running function that runs into an error it can be a complete buzzkill.

def long_func():
   import time
   time.sleep(12)
   n = 12
   df = pd.Data({'a': range(n)})
   return df

long_func()

%reset #

Reclaim memory usage in Jupyter

%autoreload #

Autoreload in Ipython

c.InteractiveShellApp.extensions = ["autoreload"]
c.InteractiveShellApp.exec_lines = ["%autoreload 2"]
c.InteractiveShellApp.exec_lines.append(
    'print("Warning: disable autoreload in ipython_config.py to improve performance.")'
)

place this in your ~/.ipython/profile_default/ipython_config.py to auto reload without needing to run the magic every time

autoformat #

This is a relatively new feature to ipython. I really enjoy it, as the time that I need the most help autoformatting my code is riffing on an ad hoc analysis at the command line.

c.TerminalInteractiveShell.autoformatter = "black"

place this in your ~/.ipython/profile_default/ipython_config.py to autoformat with black by default

new prompt #

reverse history search #

Control R

Want automatic imports?? #

Smoother Python with automatic imports | pyflyby

This article covers how I setup automatic imports in ipython

Posts tagged: ipython

Smoother Python with automatic imports | pyflyby

YouTube video #

Installation #

Configuration setup with stow #

How to Configure pyflyby #

Commas are even supported #

Jupyter note #

Ipython setup #

Ipython setup next level #

ipython auto import examples #

Getting Help #

Autocomplete #

What happens when a module is not installed #

nvim pyflyby setup #

refactoring #

Where to install for vim #

What gets imported/removed #

Where does it put imports #

What about isort #

What about init / api’s #

py command #

Links #

Ipython Ninjitsu

?docstring #

??sourcecode #

%run #

%debug #

%reset #

%autoreload #

autoformat #

new prompt #

reverse history search #

Want automatic imports?? #