import numpy as np
from sklearn.datasets import load_iris
from sklearn.preprocessing import Normalizer as skNormalizer

class Normalizer():
    def __init__(self, norm='l2'):
        self.norm = norm

    def fit(self, X):
        return self

    def transform(self, X):
        if norm == 'l1':
            norms = np.sum(np.abs(X), axis=1)
        elif norm == 'l2':
            norms = np.sqrt(np.sum(np.square(X), axis=1))
        elif norm == 'max':
            norms = np.max(X, axis=1)
        return X / norms[:, np.newaxis]

X, _ = load_iris(return_X_y=True)
for norm in ['l1', 'l2', 'max']:
    sc1 = Normalizer(norm=norm).fit(X)
    sc2 = skNormalizer(norm=norm).fit(X)
    Xt1 = sc1.transform(X)
    Xt2 = sc2.transform(X)
    assert np.allclose(Xt1, Xt2)