def testMask(self):
npDataFrame = NumpyDataFrame(dict(a=range(5), b='a b c d e'.split(' ')))
npDataFrame.mask = [False, False, True, False, True]
self.assertListsOrDicts(npDataFrame.asArrayDict(),
OrderedDict([('a', np.array([0, 1, 3])),
('b', np.array(['a', 'b', 'd']))]))
self.assertListsOrDicts(npDataFrame.getArrayNoMask('a'), np.array([0, 1, 2, 3, 4]))
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([0, 1, 3]))
self.assertEquals(len(npDataFrame), 5)
slicedDataFrame = npDataFrame[1:4]
self.assertListsOrDicts(slicedDataFrame.mask, np.array([False, True, False]))
self.assertListsOrDicts(slicedDataFrame.asArrayDict(),
OrderedDict([('a', np.array([1, 3])),
('b', np.array(['b', 'd']))]))
self.assertListsOrDicts(slicedDataFrame.getArrayNoMask('a'), np.array([1, 2, 3]))
self.assertListsOrDicts(slicedDataFrame.getArray('a'), np.array([1, 3]))
self.assertEquals(len(slicedDataFrame), 3)
npDataFrame.mask[0:2] = True
self.assertListsOrDicts(npDataFrame.asArrayDict(),
OrderedDict([('a', np.array([3])), ('b', np.array(['d']))]))
self.assertListsOrDicts(npDataFrame.getArrayNoMask('a'), np.array([0, 1, 2, 3, 4]))
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([3]))
self.assertEquals(len(npDataFrame), 5)
def _initDataFrame(self):
dataFrame = NumpyDataFrame()
listOfColNameToTrackBinArrayDicts = self._initializeRandAlgArrays()
arrayLengths = self._addConcatenatedTrackBinArraysToDataFrame(
dataFrame, listOfColNameToTrackBinArrayDicts)
self._addMandatoryArraysToDataFrame(dataFrame, arrayLengths)
if self._needsMask:
# TODO: is len(dataFrame) safe? Can a member array be multidimensional?
dataFrame.mask = np.zeros(len(dataFrame), dtype=bool)
return dataFrame
def testMaskSorting(self):
npDataFrame = NumpyDataFrame(dict(a=range(5), b='e d c b a'.split(' ')))
npDataFrame.mask = [False, False, True, False, True]
npDataFrame.sort(['b'])
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([3, 1, 0]))
npDataFrame.mask = [False, False, False, False, False]
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([4, 3, 2, 1, 0]))
def testInitMask(self):
npDataFrame = NumpyDataFrame(dict(a=range(5)),
mask=[False, False, True, False, True])
self.assertListsOrDicts(npDataFrame.mask, np.array([False, False, True, False, True]))
npDataFrame = NumpyDataFrame(dict(a=range(5)), mask=range(5))
self.assertListsOrDicts(npDataFrame.mask, np.array([False, True, True, True, True]))
npDataFrame = NumpyDataFrame(dict(a=range(5)))
self.assertIsNone(npDataFrame.mask)
npDataFrame.mask = [False, False, True, False, True]
self.assertListsOrDicts(npDataFrame.mask, np.array([False, False, True, False, True]))
npDataFrame.mask = np.array([False, False, True, False, True])
self.assertListsOrDicts(npDataFrame.mask, np.array([False, False, True, False, True]))
npDataFrame.mask = range(5)
self.assertListsOrDicts(npDataFrame.mask, np.array([False, True, True, True, True]))
self.assertRaises(ValueError, npDataFrame.__setattr__, 'mask', range(4))
npDataFrame.mask = None
self.assertIsNone(npDataFrame.mask)
def _createTestDataFrame():
npDataFrame = NumpyDataFrame()
npDataFrame.addArray('floats', [float(_) for _ in range(5)])
npDataFrame.addArray('strs', 'ab cd efg hi j'.split(' '))
npDataFrame.addArray('tuples', [(1, 2), (3, 4), (5, 6), (7, 8), (9, 10)])
return npDataFrame
def testArrayAddRemove(self):
npDataFrame = NumpyDataFrame(dict(a=range(5)))
npDataFrame.addArray('b', list('abcde'))
self.assertTrue(npDataFrame.hasArray('b'))
self.assertEquals(npDataFrame.arrayKeys(), ['a', 'b'])
self.assertListsOrDicts(npDataFrame.asArrayDict(),
OrderedDict([('a', np.array(range(5))),
('b', np.array(['a', 'b', 'c', 'd', 'e']))]))
self.assertEquals(len(npDataFrame), 5)
self.assertRaises(ValueError, npDataFrame.addArray, 'c', range(4))
self.assertFalse(npDataFrame.hasArray('c'))
npDataFrame.addArray('aa', [(1, 2), (3, 4), (5, 6), (7, 8), (9, 10)])
self.assertTrue(npDataFrame.hasArray('aa'))
self.assertEquals(npDataFrame.arrayKeys(), ['a', 'b', 'aa'])
self.assertListsOrDicts(npDataFrame.getArray('aa'),
np.array([(1, 2), (3, 4), (5, 6), (7, 8), (9, 10)]))
self.assertEquals(len(npDataFrame), 5)
self.assertRaises(KeyError, npDataFrame.removeArray, 'd')
npDataFrame.removeArray('a')
self.assertFalse(npDataFrame.hasArray('a'))
self.assertEquals(npDataFrame.arrayKeys(), ['b', 'aa'])
self.assertListsOrDicts(npDataFrame.asArrayDict(),
OrderedDict([
('b', np.array(['a', 'b', 'c', 'd', 'e'])),
('aa', np.array([(1, 2), (3, 4), (5, 6), (7, 8), (9, 10)]))
]))
self.assertEquals(len(npDataFrame), 5)
def testSort(self):
npDataFrame = NumpyDataFrame(dict(a=[2,3,2], b=[3,1,1], c=['c', 'b', 'a']))
npDataFrame.sort(order=['c'])
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([2,3,2]))
self.assertListsOrDicts(npDataFrame.getArray('b'), np.array([1,1,3]))
self.assertListsOrDicts(npDataFrame.getArray('c'), np.array(['a', 'b', 'c']))
npDataFrame = NumpyDataFrame(dict(a=[2, 3, 2], b=[3, 1, 1], c=['c', 'b', 'a']))
npDataFrame.sort(order=['a'])
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([2,2,3]))
self.assertListsOrDicts(npDataFrame.getArray('b'), np.array([3,1,1]))
self.assertListsOrDicts(npDataFrame.getArray('c'), np.array(['c', 'a', 'b']))
npDataFrame = NumpyDataFrame(dict(a=[2, 3, 2], b=[3, 1, 1], c=['c', 'b', 'a']))
npDataFrame.sort(order=['b'])
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([3,2,2]))
self.assertListsOrDicts(npDataFrame.getArray('b'), np.array([1,1,3]))
self.assertListsOrDicts(npDataFrame.getArray('c'), np.array(['b', 'a', 'c']))
npDataFrame = NumpyDataFrame(dict(a=[2, 3, 2], b=[3, 1, 1], c=['c', 'b', 'a']))
npDataFrame.sort(order=['b', 'a'])
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array([2,3,2]))
self.assertListsOrDicts(npDataFrame.getArray('b'), np.array([1,1,3]))
self.assertListsOrDicts(npDataFrame.getArray('c'), np.array(['a', 'b', 'c']))
self.assertRaises(AssertionError, npDataFrame.sort, order=[])
self.assertRaises(AssertionError, npDataFrame.sort, order=['a', 'd'])
def testInit(self):
npDataFrame = NumpyDataFrame()
self.assertEquals(npDataFrame.asArrayDict(), {})
self.assertFalse(npDataFrame.hasArray('a'))
self.assertEquals(npDataFrame.arrayKeys(), [])
self.assertEquals(len(npDataFrame), 0)
npDataFrame = NumpyDataFrame(dict(a=range(5)))
self.assertListsOrDicts(npDataFrame.asArrayDict(),
OrderedDict([('a', np.array(range(5)))]))
self.assertTrue(npDataFrame.hasArray('a'))
self.assertEquals(npDataFrame.arrayKeys(), ['a'])
self.assertListsOrDicts(npDataFrame.getArray('a'), np.array(range(5)))
self.assertEquals(len(npDataFrame), 5)
self.assertRaises(ValueError, NumpyDataFrame, dict(a=range(5), b=range(4)))
self.assertRaises(ValueError, NumpyDataFrame, dict(a=1))