def _convertDocumentsToVector(self):
self.vectors, self.masks, self.docIds = [], [], []
dimensions = TwoWayMap()
for docId, document in self.documents:
for w in document.split():
if not dimensions.contains(Clustering.PHRASE_TO_DIMENSION, w): dimensions.set(Clustering.PHRASE_TO_DIMENSION, w, len(dimensions))
for docId, document in self.documents:
vector = zeros(len(dimensions))
for w in document.split(): vector[dimensions.get(Clustering.PHRASE_TO_DIMENSION, w)]+=1
self.vectors.append(vector)
self.masks.append(ones(len(dimensions)))
self.docIds.append(docId)
# self.vectors = whiten(self.vectors)
self.dimensions = dimensions
def _convertDocumentsToVector(self):
self.vectors, self.masks, self.docIds = [], [], []
dimensions = TwoWayMap()
for docId, document in self.documents:
for w in document.split():
if not dimensions.contains(Clustering.PHRASE_TO_DIMENSION, w):
dimensions.set(Clustering.PHRASE_TO_DIMENSION, w,
len(dimensions))
for docId, document in self.documents:
vector = zeros(len(dimensions))
for w in document.split():
vector[dimensions.get(Clustering.PHRASE_TO_DIMENSION, w)] += 1
self.vectors.append(vector)
self.masks.append(ones(len(dimensions)))
self.docIds.append(docId)
# self.vectors = whiten(self.vectors)
self.dimensions = dimensions
def setUp(self):
self.twoWayMap = TwoWayMap()
self.assertRaises(TypeError, self.twoWayMap.set, (5, 1, 2))
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'A')
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'B', 'b')
class TwoWayDictTests(unittest.TestCase):
def setUp(self):
self.twoWayMap = TwoWayMap()
self.assertRaises(TypeError, self.twoWayMap.set, (5, 1, 2))
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'A')
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'B', 'b')
def __isValidObject(self): return self.twoWayMap.getMap(TwoWayMap.MAP_REVERSE)==GeneralMethods.reverseDict(self.twoWayMap.getMap(TwoWayMap.MAP_FORWARD))
def test_basicOperation(self):
self.assertEqual('A', self.twoWayMap.get(TwoWayMap.MAP_FORWARD, 'a'))
self.assertEqual('a', self.twoWayMap.get(TwoWayMap.MAP_REVERSE, 'A'))
self.assertEqual('B', self.twoWayMap.get(TwoWayMap.MAP_FORWARD, 'b'))
self.assertEqual('b', self.twoWayMap.get(TwoWayMap.MAP_REVERSE, 'B'))
self.assertEqual({'b': 'B', 'a': 'A'}, self.twoWayMap.getMap(self.twoWayMap.MAP_FORWARD))
self.assertEqual({'B': 'b', 'A': 'a'}, self.twoWayMap.getMap(self.twoWayMap.MAP_REVERSE))
self.assertTrue(self.__isValidObject())
def test_set(self):
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'c', 'A'), self.assertTrue(self.__isValidObject())
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'C', 'b'), self.assertTrue(self.__isValidObject())
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'D'), self.assertTrue(self.__isValidObject())
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'B', 'd'), self.assertTrue(self.__isValidObject())
def test_delete(self):
self.twoWayMap.remove(TwoWayMap.MAP_FORWARD, 'a')
self.assertEqual({'b': 'B'}, self.twoWayMap.getMap(self.twoWayMap.MAP_FORWARD))
self.assertEqual({'B': 'b'}, self.twoWayMap.getMap(self.twoWayMap.MAP_REVERSE))
self.twoWayMap.remove(TwoWayMap.MAP_FORWARD, 'a')
self.assertTrue(self.__isValidObject())
def test_length(self):
self.assertEqual(2, len(self.twoWayMap))
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'C', 'c')
self.assertEqual(3, len(self.twoWayMap))
def test_contains(self):
self.assertTrue(self.twoWayMap.contains(TwoWayMap.MAP_FORWARD, 'a'))
self.assertFalse(self.twoWayMap.contains(TwoWayMap.MAP_FORWARD, 'A'))
self.assertTrue(self.twoWayMap.contains(TwoWayMap.MAP_REVERSE, 'B'))
def test_resetKey(self):
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'C')
self.assertTrue(self.__isValidObject())
def setUp(self):
self.twoWayMap = TwoWayMap()
self.assertRaises(TypeError, self.twoWayMap.set, (5, 1, 2))
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'A')
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'B', 'b')
class TwoWayDictTests(unittest.TestCase):
def setUp(self):
self.twoWayMap = TwoWayMap()
self.assertRaises(TypeError, self.twoWayMap.set, (5, 1, 2))
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'A')
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'B', 'b')
def __isValidObject(self):
return self.twoWayMap.getMap(
TwoWayMap.MAP_REVERSE) == GeneralMethods.reverseDict(
self.twoWayMap.getMap(TwoWayMap.MAP_FORWARD))
def test_basicOperation(self):
self.assertEqual('A', self.twoWayMap.get(TwoWayMap.MAP_FORWARD, 'a'))
self.assertEqual('a', self.twoWayMap.get(TwoWayMap.MAP_REVERSE, 'A'))
self.assertEqual('B', self.twoWayMap.get(TwoWayMap.MAP_FORWARD, 'b'))
self.assertEqual('b', self.twoWayMap.get(TwoWayMap.MAP_REVERSE, 'B'))
self.assertEqual({
'b': 'B',
'a': 'A'
}, self.twoWayMap.getMap(self.twoWayMap.MAP_FORWARD))
self.assertEqual({
'B': 'b',
'A': 'a'
}, self.twoWayMap.getMap(self.twoWayMap.MAP_REVERSE))
self.assertTrue(self.__isValidObject())
def test_set(self):
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'c',
'A'), self.assertTrue(self.__isValidObject())
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'C',
'b'), self.assertTrue(self.__isValidObject())
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a',
'D'), self.assertTrue(self.__isValidObject())
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'B',
'd'), self.assertTrue(self.__isValidObject())
def test_delete(self):
self.twoWayMap.remove(TwoWayMap.MAP_FORWARD, 'a')
self.assertEqual({'b': 'B'},
self.twoWayMap.getMap(self.twoWayMap.MAP_FORWARD))
self.assertEqual({'B': 'b'},
self.twoWayMap.getMap(self.twoWayMap.MAP_REVERSE))
self.twoWayMap.remove(TwoWayMap.MAP_FORWARD, 'a')
self.assertTrue(self.__isValidObject())
def test_length(self):
self.assertEqual(2, len(self.twoWayMap))
self.twoWayMap.set(TwoWayMap.MAP_REVERSE, 'C', 'c')
self.assertEqual(3, len(self.twoWayMap))
def test_contains(self):
self.assertTrue(self.twoWayMap.contains(TwoWayMap.MAP_FORWARD, 'a'))
self.assertFalse(self.twoWayMap.contains(TwoWayMap.MAP_FORWARD, 'A'))
self.assertTrue(self.twoWayMap.contains(TwoWayMap.MAP_REVERSE, 'B'))
def test_resetKey(self):
self.twoWayMap.set(TwoWayMap.MAP_FORWARD, 'a', 'C')
self.assertTrue(self.__isValidObject())