def __init__(self, tokenizer=None, cache_path='/tmp/'):
"""
:param tokenizer: A tokenizer override
:type tokenizer: function (optional)
:param cache_path: path to data storage
:type cache_path: str
"""
self.categories = BayesCategories()
self.tokenizer = tokenizer or SimpleBayes.tokenize_text
self.cache_path = cache_path
self.probabilities = {}
def test_persist_cache(self, dump_mock, open_mock):
open_mock.return_value = 'opened'
categories = BayesCategories()
categories.categories = {'foo': 'bar'}
sb = SimpleBayes()
sb.cache_path = '/tmp/'
sb.categories = categories
sb.cache_persist()
open_mock.assert_called_once_with('/tmp/_simplebayes.pickle', 'wb')
dump_mock.assert_called_once_with(categories, 'opened')
def test_cache_train(self, exists_mock, load_mock, open_mock, calc_mock):
categories = BayesCategories()
categories.categories = {'foo': 'bar'}
load_mock.return_value = categories
open_mock.return_value = 'opened'
exists_mock.return_value = True
sb = SimpleBayes(cache_path='foo')
sb.cache_train()
exists_mock.assert_called_once_with('foo/_simplebayes.pickle')
open_mock.assert_called_once_with('foo/_simplebayes.pickle', 'rb')
load_mock.assert_called_once_with('opened')
calc_mock.assert_called_once_with()
self.assertEqual(sb.categories, categories)
def __init__(self,
tokenizer=None,
cache_path='/tmp/',
cache_data=None,
cache_in_memory=False):
"""
:param tokenizer: A tokenizer override
:type tokenizer: function (optional)
:param cache_path: path to data storage
:type cache_path: str
:param cache_data: from an existing cache
:type cache_data: pickle.dumps object
:param cache_in_memory: True if the persistant cache is kept in memory
:type: boolean
"""
self.categories = BayesCategories()
self.tokenizer = tokenizer or SimpleBayes.tokenize_text
self.cache_path = cache_path
self.cache_data = cache_data
self.cache_in_memory = True if cache_data is not None else cache_in_memory
self.probabilities = {}
def test_get_categories(self):
bc = BayesCategories()
bc.add_category('foo')
self.assertEqual(bc.get_categories(), bc.categories)
def test_get_category(self):
bc = BayesCategories()
bc.add_category('foo')
self.assertIsInstance(bc.get_category('foo'), BayesCategory)
def test_add_category(self):
bc = BayesCategories()
bc.add_category('foo')
self.assertIn('foo', bc.categories)
self.assertIsInstance(bc.categories['foo'], BayesCategory)
def flush(self):
"""
Deletes all tokens & categories
"""
self.categories = BayesCategories()
