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')
