Andrey Dugin

import pandas as pd

pd.read_csv('in.csv', sep=';', header=None) \
.drop_duplicates(4, keep='first') \
.to_csv('out.csv', header=False, index=False)

from functools import lru_cache

@lru_cache(maxsize=None)
def write(unique_field):
    fo.write(line)
    
with open('in.csv', 'r') as fi, open('out.csv', 'w') as fo:
    for line in fi:
        write(line.rstrip().split(';')[-1])

import numpy as np
from skimage.util import view_as_windows  # Библиотека scikit-image

tsize = 256
image = np.random.randint(0, 256, (8192, 8192, 3), np.uint8)

# Значения яркости - это канал L из пространства LAB
lab_l_channel = cv2.cvtColor(image, cv2.COLOR_BGR2LAB)[..., 0]

# Плитки 256x256
tiles = view_as_windows(lab_l_channel, (tsize, tsize), (tsize, tsize))

# Примечательно, что
assert np.shares_memory(lab_l_channel, tiles) == True

# Средняя яркость в виде матрицы 32х32
lightness = tiles.reshape(-1, tsize ** 2).mean(axis=-1).reshape(np.array(image.shape[:2]) // tsize)

import numpy as np

a, b, c = np.random.randint(0, 256, 3, dtype=np.uint8)

palette = [ [a, b, c], [a, a, b], [c, a, b], [b, a, c], [c, c, a] ]

rgb_img = [
  [ (a+3, b-2, c-1), (a, a+1, b-2), (c+2, a, b-3) ],
  [ (b, a-2, c+1), (c, c+2, a-1), (b-1, a, c-3) ],
  [ (a, a+1, b-2), (a+2, b-2, c-1), (c+1, c+1, a-2) ]
]

palette = np.array(palette).reshape(1, 1, 5, 3)
rgb_img = np.array(rgb_img).reshape(3, 3, 1, 3)
indices = np.linalg.norm(palette - rgb_img, axis=-1).argmin(axis=-1).ravel()

print(indices)

from sklearn.cluster import KMeans

a, b, c = np.random.randint(0, 256, 3, dtype=np.uint8)

palette = [ [a, b, c], [a, a, b], [c, a, b], [b, a, c], [c, c, a] ]

rgb_img = [
  [ (a+3, b-2, c-1), (a, a+1, b-2), (c+2, a, b-3) ],
  [ (b, a-2, c+1), (c, c+2, a-1), (b-1, a, c-3) ],
  [ (a, a+1, b-2), (a+2, b-2, c-1), (c+1, c+1, a-2) ]
]

result = (
    KMeans(n_clusters=len(palette), init=np.array(palette), n_init=1)
    .fit_transform(np.array(rgb_img).reshape(-1, 3))
    .argmin(axis=-1)
)

import numpy as np

n = 5
matrix = np.random.choice(range(100), (n, n), replace=False)
result = np.dot(matrix[matrix.argmax() // n, :], matrix[:, matrix.argmin() % n])

Run pip install -r requirements.txt to install the necessary packages. Additionally you will need PyTorch (>=1.0.1).

>>> from pathlib import Path
>>> Path(r'c:\somefolder\example.py').stem
'example'
>>> Path('/home/user/somefolder/example.py').stem
'example'

from ast import literal_eval as eval  # ast.literal_eval() безопасный, обычный eval() - нет

with open('input.txt', 'r') as fi, open('output.txt', 'w') as fo:
    cache = set()
    for line in fi:
        title = eval(line).get('title')
        if title not in cache:
            cache.add(title)
            fo.write(line)

from ast import literal_eval as eval
from functools import lru_cache

@lru_cache(None)
def process(title):
    print(record, file=fo)

with open('input.txt', 'r') as fi, open('output.txt', 'w') as fo:
    for record in map(eval, fi):
        process(record['title'])

>>> process.cache_info()
CacheInfo(hits=994960, misses=5040, maxsize=None, currsize=5040)

if not IsLanguageExistInWikipedia(language):
    print('no results')