def test_invalid_n(self):
fhp = FieldHashingProperties(ngram=2, k=20, positional=True)
fhp.ngram = -6
with self.assertRaises(
ValueError,
msg='Expected raise ValueError on invalid n.'):
tok = get_tokenizer(fhp)
tok('prawn')
def stream_bloom_filters(
dataset, # type: Iterable[Sequence[Text]]
keys, # type: Sequence[Sequence[bytes]]
schema # type: Schema
):
# type: (...) -> Iterable[Tuple[bitarray, Text, int]]
""" Compute composite Bloom filters (CLKs) for every record in an
iterable dataset.
:param dataset: An iterable of indexable records.
:param schema: An instantiated Schema instance
:param keys: A tuple of two lists of secret keys used in the HMAC.
:return: Generator yielding bloom filters as 3-tuples
"""
tokenizers = [
tokenizer.get_tokenizer(field.hashing_properties)
for field in schema.fields
]
return (crypto_bloom_filter(s, tokenizers, schema, keys) for s in dataset)
import unittest
from clkhash.field_formats import FieldHashingProperties
from clkhash.tokenizer import get_tokenizer
__author__ = 'shardy'
# some tokenizers
p1_20 = get_tokenizer(
FieldHashingProperties(ngram=1, k=20)
)
p2_20 = get_tokenizer(
FieldHashingProperties(ngram=2, k=20)
)
p1_20_true = get_tokenizer(
FieldHashingProperties(ngram=1, k=20, positional=True)
)
dummy = get_tokenizer(None)
class TestTokenizer(unittest.TestCase):
def test_unigram_1(self):
self.assertEqual(list(p1_20("1/2/93", ignore='/')),
['1', '2', '9', '3'])
def test_unigram_2(self):
self.assertEqual(list(p1_20("1*2*93", ignore='*')),
def test_bigram_duplicate(self):
properties = FieldHashingProperties(ngram=2, positional=False)
self.assertEqual(list(get_tokenizer(properties)("abab")),
[' a', 'ab', 'ba', 'ab', 'b '])
def test_unigram_1(self):
properties = FieldHashingProperties(ngram=1, positional=False)
self.assertEqual(list(get_tokenizer(properties)("1/2/93", ignore='/')),
['1', '2', '9', '3'])
def test_bigram_2(self):
properties = FieldHashingProperties(ngram=2, positional=False)
self.assertEqual(list(get_tokenizer(properties)("steve", ignore='e')),
[' s', 'st', 'tv', 'v '])
def test_bigram_1(self):
properties = FieldHashingProperties(ngram=2, positional=False)
self.assertEqual(list(get_tokenizer(properties)("steve")),
[' s', 'st', 'te', 'ev', 've', 'e '])
def test_positional_unigram_duplicate(self):
properties = FieldHashingProperties(ngram=1, positional=True)
self.assertEqual(list(get_tokenizer(properties)("111")),
['1 1', '2 1', '3 1'])
def test_positional_unigram_2(self):
properties = FieldHashingProperties(ngram=1, positional=True)
self.assertEqual(list(get_tokenizer(properties)("1*2*")),
['1 1', '2 *', '3 2', '4 *'])
def test_unigram_duplicate(self):
properties = FieldHashingProperties(ngram=1, positional=False)
self.assertEqual(list(get_tokenizer(properties)("1212")),
['1', '2', '1', '2'])
