nokogiri

This Python library is a sharp "nokogiri" that cuts through the problems that arise in the process of trial and error, such as machine learning.

pip install nokogiri

spylus

• Now you can draw image with stylus pen or drawing tablet such as iPad, Surface and Wacom in JupyterLab/Notebook.
• Of course it senses pressure so you can control the thickness of the line.
• Clipboard image copy/paste feature supported! this is super effective for AI painting with clip studio/photoshop.
• VSCode greatly limits functionality due to policies that prohibit the execution of javascript in notebooks. Please use JupyterLab/Notebook.

screenshot

Try it

Hit this button and launch demo via binder.

• launch demo via binder
• Just read usage on github

from nokogiri.spylus import spylus

---

anyparse

Drop-in replacement library for (config)argparse that works anywhere (Jupyter Notebook, JupyterLab, VSCode Python Interactive). Have you ever met an annying error message like this?

usage: ipykernel_launcher.py [-h]
ipykernel_launcher.py: error: unrecognized arguments: -f 
…
An exception has occurred, use %tb to see the full traceback.
SystemExit: 2

As shown in the following example. You can drop-in replace argparse with anyparse. You can also specify commands to be parsed only on jupyter.

from nokogiri import anyparse as argparse # Drop-in replacement
parser = argparse.ArgumentParser()
parser.add_argument('--readlink', type=str, default="")
parser.add_argument('--metrics', type=bool, default=False)
parser.jupyter_argument("--metrics True") # This command is parsed only on jupyter.
args = parser.parse_args()

If configargparse is already installed, all features are available!

working_dir

Enables import of any module not in the project tree from anywhere at any time

from nokogiri.working_dir import working_dir
with working_dir("/path/to/isolated_module"):
    import isolated_module
    with open("isolated_data") as f:
        isolated_data = f.read()

tqdm_load

Display a progress bar while reading a pickle file

from nokogiri.tqdm_load import tqdm_load
data = tqdm_load("/path/to/data.pkl")

defaultdotdict

defaultdict but also dotdict

from nokogiri.defaultdotdict import defaultdotdict
sounds = defaultdotdict(lambda: "meow")
print(sounds.dog) # "meow"
sounds["dog"] = "bow" # you can also access with []
print(sounds.dog) # "bow"

curry

currying function. add some syntax sugar for a lot of function arguments.

from nokogiri.curry import curry
# "pipe" iterable objects
@curry
def f(x):
    return x+1
@curry
def g(x):
    return 2*x
range(5)|f|g # [2,4,6,8,10]

# example of common use case
@curry
def f(a="a", b="b", c="c", d="d"):
    return f"/{a}/{b}/{c}/{d}"
f.hoge.fuga.piyo("root") # "/root/hoge/fuga/piyo"

# syntax
f.hoge.fuga.piyo == lambda x: f(x, 'hoge', 'fuga', 'piyo')
f[1][None][True] == lambda x: f(x, 1, None, True)
f._(key="value") == lambda x: f(x, key="value")

which_env

Detect notebook environment is Jupyter or Terminal or …

from nokogiri.which_env import which_env
if which_env() == which_env.JUPYTER:
    from tqdm.notebook import tqdm
else:
    from tqdm import tqdm

shared_map

share large size map/dict between notebok kernels

python -m nokogiri.shared_map /path/to/*.pkl

from nokogiri.shared_map import shared_map
sm = shared_map("/path/to")
for x in tqdm(range(100)):
    y = sm.hoge(x) # read from /path/to/hoge.pkl

cell

Bring Jupyter cell behavior to VSCode Python Interactive

from nokogiri.cell import cell
A = 1
B = 2
@cell()
def func(A, B):
    C = A+B
    D = A*B
    return C, D
print(C, D) # 3, 2

---

code embeding test (recently I noticed that Readme.md is code-embeddable like this.)

nokogiri/nokogiri/tqdm_load.py

Lines 9 to 17 in e0d438d

	self.fd = fd
	self.tqdm = tqdm(total=total)
	self.tqdm.set_description(desc)
	def read(self, size=-1):
	bytes = self.fd.read(size)
	self.tqdm.update(len(bytes))
	return bytes
	def readline(self):
	bytes = self.fd.readline()