#!/usr/bin/env python2.2 ######################################## # The contents of this file are subject to the MLX PUBLIC LICENSE version # 1.0 (the "License"); you may not use this file except in # compliance with the License. # # Software distributed under the License is distributed on an "AS IS" # basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See # the License for the specific language governing rights and limitations # under the License. # # The Original Source Code is "compClust", released 2003 September 03. # # The Original Source Code was developed by the California Institute of # Technology (Caltech). Portions created by Caltech are Copyright (C) # 2002-2003 California Institute of Technology. All Rights Reserved. ######################################## """ Test Suite for the current Dataset schema. """ import MA import Numeric import unittest import os import inspect from compClust.util.DistanceMetrics import * from compClust.mlx.datasets import Dataset from compClust.mlx.labelings import Labeling from compClust.mlx.labelings import subsetByLabeling from compClust.mlx.views import RowSubsetView, ColumnSubsetView from compClust.mlx.views import RowSupersetView, ColumnSupersetView from compClust.mlx.views import SortedView import compClust.mlx #def read_testset(filename): # dataset = Dataset(test_data_files_home + "/" + filename) # return dataset class Dataset2TestCases(unittest.TestCase): def setUp(self): self.datadir = os.path.split(inspect.getsourcefile(Dataset2TestCases))[0] dataset_spurious_filename = "dataset_tab_array_spurious_whitespace.txt" #self.original_dir = os.getcwd() #os.chdir(compClust.mlx.__path__[0]) #def tearDown(self): # os.chdir(self.original_dir) def assertPass(self, func, args): try: apply(func, args) flag = 1 except: flag = 0 if not flag: self.fail("Function did not pass") def assertFail(self, func, args): try: apply(func, args) flag = 0 except: flag = 1 if not flag: self.fail("Function did not fail") def test_hard_subset(self): # # Test the view-subset-of-another-view's-labeling # d1 = Dataset([[1,2,3],[4,5,6],[7,8,9],[10,11,12]]) d2 = Dataset([[13,14,15],[16,17,18],[19,20,21],[22,23,24]]) # # Make some supersets # s0 = RowSupersetView(d1,d1) s1 = ColumnSupersetView(d1,d2) s2 = ColumnSupersetView(d2,d1) s3 = RowSupersetView(d2,d2) # # Make some views # v1 = RowSubsetView(d1, [0,2,3]) v2 = RowSubsetView(d1, [0,2,3]) v3 = ColumnSubsetView(d2, [1,2]) v4 = ColumnSubsetView(d2, [0,1]) v5 = RowSubsetView(s0, [0,2,4,5,7]) v6 = ColumnSubsetView(s1, [1,2,3,4]) v7 = RowSubsetView(s2, [0,3]) v8 = ColumnSubsetView(s3, [0,2]) # # Try the simplest case: label between v1 and v2 # l1 = Labeling(v1) l1.addLabelToRows("v1", [0,1,2]) assert Numeric.sum( Numeric.sum(subsetByLabeling(v2, l1, "v1").getData() == Numeric.array([[7,8,9]]))) == 3 l3 = Labeling(v3) l3.addLabelToCols("v3", [0,1]) assert Numeric.sum( Numeric.sum(subsetByLabeling(v4, l3, "v3").getData() == Numeric.array([[14],[17],[20],[23]]))) == 4 def test_dataset_torture(self): # # Test every method of the dataset # d1 = Dataset([[1,2,3],[4,5,6]]) self.assertPass(d1.getRowData, [0]) self.assertFail(d1.getRowData, [-1]) self.assertFail(d1.getRowData, [2]) self.assertPass(d1.getColData, [0]) self.assertFail(d1.getColData, [-1]) self.assertFail(d1.getColData, [3]) self.assertPass(d1.getData, []) self.assertPass(d1.getData, [None]) self.assertPass(d1.getData, [4]) self.assertFail(d1.getData, [-1]) self.assertFail(d1.getData, [5]) self.assertPass(d1.getRowKey, [0]) self.assertFail(d1.getRowKey, [-1]) self.assertFail(d1.getRowKey, [2]) self.assertPass(d1.getColKey, [0]) self.assertFail(d1.getColKey, [-1]) self.assertFail(d1.getColKey, [3]) self.assertPass(d1.getRowKeys, []) self.assertPass(d1.getColKeys, []) self.assertPass(d1.getKeys, [0]) self.assertPass(d1.getKeys, [1]) self.assertPass(d1.getKeys, [-1]) self.assertPass(d1.getKeys, [100]) self.assertPass(d1.getViews, []) self.assertPass(d1.removeView, [None]) self.assertPass(RowSubsetView, [d1, [0,1]]) self.assertFail(RowSubsetView, [d1, [0,1,2]]) self.assertFail(RowSubsetView, [d1, [-1,0,1]]) self.assertPass(RowSubsetView, [d1, [0,1]]) self.assertFail(RowSubsetView, [d1, [0,1,4]]) self.assertFail(RowSubsetView, [d1, [-1,0,1]]) l1 = Labeling(d1) l2 = Labeling(d1) l3 = Labeling(d1) l4 = Labeling(d1) l2.addLabelToRows("foo", [0]) l3.addLabelToCols("foo", [0,2]) l4.addLabelToRows("foo", [1]) l4.addLabelToCols("foo", [1,2]) self.assertPass(subsetByLabeling, [d1, l1, "foo"]) self.assertPass(subsetByLabeling, [d1, l2, "foo"]) self.assertPass(subsetByLabeling, [d1, l3, "foo"]) self.assertPass(subsetByLabeling, [d1, l4, "foo"]) self.assertPass(RowSupersetView, [d1, d1]) self.assertPass(ColumnSupersetView, [d1, d1]) def test_superset(self): d1 = Dataset([[1,2],[3,4]]) d2 = Dataset([[5,6],[7,8]]) s1 = RowSupersetView(d1, d2) s2 = ColumnSupersetView(d1, d2) all1 = s1.getData() all2 = s2.getData() if Numeric.sum(Numeric.sum(all1 == Numeric.array([[1.,2.],[3.,4.],[5.,6.],[7.,8.]]))) != 8: self.fail("Row superset incorrect") if Numeric.sum(Numeric.sum(all2 == Numeric.array([[1.,2.,5.,6.],[3.,4.,7.,8.]]))) != 8: self.fail("Column superset incorrect") v = s1.getRowData(0) if Numeric.sum(v == Numeric.array([1., 2.])) != 2: self.fail() v = s1.getRowData(1) if Numeric.sum(v == Numeric.array([3., 4.])) != 2: self.fail() v = s1.getRowData(2) if Numeric.sum(v == Numeric.array([5., 6.])) != 2: self.fail() v = s1.getRowData(3) if Numeric.sum(v == Numeric.array([7., 8.])) != 2: self.fail() v = s1.getColData(0) if Numeric.sum(v == Numeric.array([1., 3., 5., 7.])) != 4: self.fail() v = s1.getColData(1) if Numeric.sum(v == Numeric.array([2., 4., 6., 8.])) != 4: self.fail() v = s2.getRowData(0) if Numeric.sum(v == Numeric.array([1., 2., 5., 6.])) != 4: self.fail() v = s2.getRowData(1) if Numeric.sum(v == Numeric.array([3., 4., 7., 8.])) != 4: self.fail() v = s2.getColData(0) if Numeric.sum(v == Numeric.array([1., 3.])) != 2: self.fail() v = s2.getColData(1) if Numeric.sum(v == Numeric.array([2., 4.])) != 2: self.fail() v = s2.getColData(2) if Numeric.sum(v == Numeric.array([5., 7.])) != 2: self.fail() v = s2.getColData(3) if Numeric.sum(v == Numeric.array([6., 8.])) != 2: self.fail() def test_subset(self): d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) l = Labeling(d); l.addLabelToRows("foo", [0,2]); l.addLabelToCols("foo", [1,2]); view = subsetByLabeling(d, l, "foo") all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[2.,3.],[8.,9.]]))) != 4: self.fail("Labeling subset getData() incorrect") r = view.getKeys() c = view.getKeys(1) v = view.getData(r[0]) if Numeric.sum(v == Numeric.array([2., 3.])) != 2: self.fail("View Row 0 incorrect") v = view.getData(r[1]) if Numeric.sum(v == Numeric.array([8., 9.])) != 2: self.fail("View Row 1 incorrect") v = view.getData(c[0]) if Numeric.sum(v == Numeric.array([2., 8.])) != 2: self.fail("View Col 0 incorrect") v = view.getData(c[1]) if Numeric.sum(v == Numeric.array([3., 9.])) != 2: self.fail("View Col 1 incorrect") def test_col_subset(self): d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) view = ColumnSubsetView(d, [2,1]) all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[3., 2.],[6., 5.], [9., 8.]]))) != 6: self.fail("Column subset getData() incorrect") r = view.getKeys() c = view.getKeys(1) v = view.getData(r[0]) if Numeric.sum(v == Numeric.array([3., 2.])) != 2: self.fail("View Row 0 incorrect") v = view.getData(r[1]) if Numeric.sum(v == Numeric.array([6., 5.])) != 2: self.fail("View Row 1 incorrect") v = view.getData(r[2]) if Numeric.sum(v == Numeric.array([9., 8.])) != 2: self.fail("View Row 2 incorrect") v = view.getData(c[0]) if Numeric.sum(v == Numeric.array([3., 6., 9.])) != 3: self.fail("View Col 0 incorrect") v = view.getData(c[1]) if Numeric.sum(v == Numeric.array([2., 5., 8.])) != 3: self.fail("View Col 1 incorrect") def test_multi_label_subset(self): d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9], [10, 11, 12], [13, 14, 15]]) l1 = Labeling(d) l1.addLabelToRows("x", [0,1]) l1.addLabelToRows("y", [1,2]) l1.addLabelToCols("z", [0,2]) l1.addLabelToCols(0.1, [0,2]) view = subsetByLabeling(d, l1, "x") all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[1.,2.,3.],[4.,5.,6.]]))) != 6: self.fail("Multi label failed first test") view = subsetByLabeling(d, l1, ["x", "y"]) all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[1.,2.,3.],[4.,5.,6.],[7.,8.,9.]]))) != 9: self.fail("Multi label failed second test") view = subsetByLabeling(d, l1, ["x", "z"]) all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[1.,3.],[4.,6.]]))) != 4: self.fail("Multi label failed third test") view = subsetByLabeling(d, l1, ["x", 0.1]) all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[1.,3.],[4.,6.]]))) != 4: self.fail("Multi label failed fourth test") def test_row_subset(self): d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) view = RowSubsetView(d, [2,1]) all = view.getData() if Numeric.sum(Numeric.sum(all == Numeric.array([[7.,8.,9.],[4.,5.,6.]]))) != 6: self.fail("Row subset getData() incorrect") r = view.getKeys() c = view.getKeys(1) v = view.getData(r[0]) if Numeric.sum(v == Numeric.array([7., 8., 9.])) != 3: self.fail("View Row 0 incorrect") v = view.getData(r[1]) if Numeric.sum(v == Numeric.array([4., 5., 6.])) != 3: self.fail("View Row 1 incorrect") v = view.getData(c[0]) if Numeric.sum(v == Numeric.array([7., 4.])) != 2: self.fail("View Col 0 incorrect") v = view.getData(c[1]) if Numeric.sum(v == Numeric.array([8., 5.])) != 2: self.fail("View Col 1 incorrect") v = view.getData(c[2]) if Numeric.sum(v == Numeric.array([9., 6.])) != 2: self.fail("View Col 2 incorrect") def test_sort(self): # Obsoleted by SortedViews d = Dataset([[1,2,3],[0,-1,0],[7,8,9],[0,1,0],[-1,3,2],[9,6,-3],[1,2,1],[5,6,7]]) v = SortedView(d) origin = Numeric.array([[0,0,0]]) l = Labeling(v) l.addLabelToRows("foo",[0,2,5]) d0 = v.getRowData(0) d1 = v.getRowData(5) d2 = v.getRowData(2) v.sort(lambda x : EuclideanDistance(x, origin)) h = l.getRowsByLabel("foo") self.failUnless(Numeric.sum(v.getRowData(h[0]) == d0)==3) self.failUnless(Numeric.sum(v.getRowData(h[1]) == d1)==3) self.failUnless(Numeric.sum(v.getRowData(h[2]) == d2)==3) def test_label_using(self): d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9], [10, 11, 12]]) v1 = RowSubsetView(d, [0,1,2]) v2 = RowSubsetView(d, [1,2,3]) v3 = RowSubsetView(d, [3]) l = Labeling(d) l.labelCols([1,2,3]) l1 = Labeling(v1) l1.addLabelToRows("foo",[0,1,2]) l2 = Labeling(v2) l2.labelFrom(l1) self.failUnless(l2.getRowsByLabel("foo") == [0,1]) l3 = Labeling(v3) l3.labelFrom(l1) self.failUnless(l3.getRowsByLabel("foo") == []) # Make sure the columns labels come through OK lu = Labeling(v1) lu.labelFrom(l) self.failUnless(lu.getLabels() == [1,2,3]) def test_keys(self): # # Checks for keys and how ranges are handled # d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) r = d.getKeys() c = d.getKeys(1) # # Try getting the data with the keys # v = d.getData(r[0]) if Numeric.sum(v == Numeric.array([1., 2., 3.])) != 3: self.fail("Row 0 incorrect") v = d.getData(r[1]) if Numeric.sum(v == Numeric.array([4., 5., 6.])) != 3: self.fail("Row 1 incorrect") v = d.getData(r[2]) if Numeric.sum(v == Numeric.array([7., 8., 9.])) != 3: self.fail("Row 2 incorrect") v = d.getData(c[0]) if Numeric.sum(v == Numeric.array([1., 4., 7.])) != 3: self.fail("Col 0 incorrect") v = d.getData(c[1]) if Numeric.sum(v == Numeric.array([2., 5., 8.])) != 3: self.fail("Col 1 incorrect") v = d.getData(c[2]) if Numeric.sum(v == Numeric.array([3., 6., 9.])) != 3: self.fail("Col 2 incorrect") try: v = d.getData(-1) self.fail("Dataset allowed negative key") except: pass try: v = d.getData(10) self.fail("Dataset allowed too large key") except: pass def test_is_row_col_key(self): """Does the isRowKey and isColKey work correctly?""" d = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) r = d.getKeys() c = d.getKeys(1) for k in d.getRowKeys(): self.failUnless(d.isRowKey(k)) self.failIf(d.isColKey(k)) for k in d.getColKeys(): self.failIf(d.isRowKey(k)) self.failUnless(d.isColKey(k)) self.failIf(d.isRowKey(-1)) self.failIf(d.isColKey(-1)) def test_labeling(self): # # Verifies that labeling do not get confused # d1 = Dataset([[1,2,3],[4,5,6],[7,8,9]]) d2 = Dataset([[10,11,12],[13,14,15],[16,17,18]]) s1 = RowSubsetView(d1, [1,2]) s2 = RowSubsetView(d1, [0,1]) s3 = RowSubsetView(d2, [1,2]) s4 = RowSubsetView(d2, [0,1]) s5 = RowSupersetView(s2, s3) l1 = Labeling(s1) l2 = Labeling(s2) l3 = Labeling(s3) l4 = Labeling(s4) l5 = Labeling(s5) # use strings in labeling 1 l1.addLabelToRows('X', [0,1]) # Use ints in labeling 2 l2.addLabelToRows(2, [0,1]) def test_constructor(self): # # Check the all the variations of args pass and others fail # ref = Numeric.array([[1, 0, 0], [0, 1, 0], [0, 0, 1]], Numeric.Float) d = Dataset([[1, 0, 0], [0, 1, 0], [0, 0, 1]]) if Numeric.sum(Numeric.sum(d.getData() == ref)) != 9: self.fail("Did not contruct from list of lists correctly") d = Dataset(((1, 0, 0), (0, 1, 0), (0, 0, 1))) if Numeric.sum(Numeric.sum(d.getData() == ref)) != 9: self.fail("Did not contruct from tuples of tuples correctly") d = Dataset(os.path.join(self.datadir,"DStest.txt")) if Numeric.sum(Numeric.sum(d.getData() == ref)) != 9: self.fail("Did not contruct from filename correctly") stream = open(os.path.join(self.datadir,"DStest.txt")) d = Dataset(stream) stream.close() if Numeric.sum(Numeric.sum(d.getData() == ref)) != 9: self.fail("Did not contruct from I/O stream correctly") d = Dataset(ref) if Numeric.sum(Numeric.sum(d.getData() == ref)) != 9: self.fail("Did not contruct from Numeric array correctly") d2 = Dataset(d) if Numeric.sum(Numeric.sum(d2.getData() == ref)) != 9: self.fail("Did not contruct from Dataset correctly") self.assertFail(Dataset, [{}]) self.assertFail(Dataset, [None]) # # Check the anomalous dataset handling (NaN, Tabs) # base = MA.array([[1,2,3],[4,5,6],[7,8,9],[10,11,12],[13,14,15],[16,17,18], [19,20,21],[22,23,24]]) ds3 = Dataset(os.path.join(self.datadir,"dataset-NaNs.txt")) ds4 = Dataset(os.path.join(self.datadir,"dataset-Tabs.txt")) ds3sum = MA.sum(MA.sum(ds3.getData() == base)) ds4sum = MA.sum(MA.sum(ds4.getData() == base)) ds3mask = MA.sum(MA.sum(ds3.getData().mask())) ds4mask = MA.sum(MA.sum(ds4.getData().mask())) if (ds3sum + ds3mask != 24): self.fail("Did not read NaN data correctly") if (ds4sum + ds4mask != 24): self.fail("Did not read NaN data correctly") def test_delimiter(self): ds_comma = Dataset(os.path.join(self.datadir,"dataset-full-comma.txt"), delimiter=",") ds_spaces = Dataset(os.path.join(self.datadir,"dataset-full-spaces.txt"), delimiter=" +") self.failUnless(Numeric.alltrue(ds_spaces.getData() == ds_comma.getData())) def test_mixin(self): class TestMixin: def testMixin(self): return "mixin" d1 = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) d2 = Dataset([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) self.failIf('testMixin' in dir(d1)) d1.addMixin(TestMixin) self.failUnless(d1.testMixin() == TestMixin().testMixin()) v1 = RowSubsetView(d1, [1]) v2 = RowSubsetView(d2, [2]) self.failUnless(v1.testMixin() == TestMixin().testMixin()) self.failIf('testMixin' in dir(v2)) d1.removeMixin(TestMixin) self.failIf('testMixin' in dir(d1)) def test_primary_labelings(self): """Make sure that the primary row and column labelings behave as expected They can only be labelings or none, and they must be unique in the axis specified. """ d1 = Dataset([[1,2,3],[4,5,6],[7,8,9],[10,11,12]]) # test row labelings row_label_duplicate = Labeling(d1) row_label_duplicate.labelRows(["row"] * 4) row_label_unique = Labeling(d1) row_label_unique.labelRows(["one", "two", "three", "four"]) # make sure we received a labeling self.failUnlessRaises(ValueError, d1.setPrimaryRowLabeling, 3) # make sure we are unqiue self.failUnlessRaises(ValueError, d1.setPrimaryRowLabeling, row_label_duplicate) d1.setPrimaryRowLabeling(row_label_unique) self.failUnless(d1.getPrimaryRowLabeling() == row_label_unique) d1.setPrimaryRowLabeling(None) # test column labelings column_label_duplicate = Labeling(d1) column_label_duplicate.labelCols(["col"]*3) column_label_unique = Labeling(d1) column_label_unique.labelCols(["one", "two", "three"]) # make sure we received a labeling self.failUnlessRaises(ValueError, d1.setPrimaryColumnLabeling, 3) # make sure we are unqiue self.failUnlessRaises(ValueError, d1.setPrimaryColumnLabeling, column_label_duplicate) d1.setPrimaryColumnLabeling(column_label_unique) self.failUnless(d1.getPrimaryColumnLabeling() == column_label_unique) d1.setPrimaryColumnLabeling(None) def test_setviewdefault(self): """Test to make sure that setViewDefault will return a previously created instance of an attached view. """ d1 = Dataset([[1,2,3],[4,5,6],[7,8,9],[10,11,12]]) # start with no views self.failUnless(len(d1.getViews()) == 0, "Start with no views") # add a couple of anonymous views s1 = RowSubsetView(d1, [1,2]) s2 = RowSubsetView(d1, [0,1]) self.failUnless(len(d1.getViews()) == 2, "Should now have 2 anonymous views") # add a named view via setViewDefault s3 = d1.addViewDefault('subset1', RowSubsetView, d1, [1,2]) self.failUnless(len(d1.getViews()) == 3, "Should now have 3 views") s4 = d1.addViewDefault('subset1', RowSubsetView, d1, [1,2]) self.failUnless(len(d1.getViews()) == 3, "Should still have 3 views") self.failUnless(s3 == s4, "and they should return the same instance") # of course now we have a new problem of potential name collision # so we should not allow names to be reused (This might break old code) # (well anonymous ones can get through) s5 = RowSubsetView(d1,[1,2]) self.failUnlessRaises(KeyError, s5.setName, 'subset1') # we should allow names to be repeated at different levels of the # hierarchy s6 = RowSubsetView(s5, [0]) s6.setName('subset1') s7 = s5.addViewDefault('subset2', RowSubsetView, s5, [0,1]) self.failUnless(s7 == s5.addViewDefault('subset2', RowSubsetView, s5, [3]), "addViewDefault should return the same instance") # make sure that even in subviews we're testing names correctly s8 = RowSubsetView(s5, [0,1]) self.failUnlessRaises(KeyError, s8.setName, 'subset2') def testSplitRowColKeylist(self): """Split row col keylist splits a list of keys into lists ofrow and column keys """ d1 = Dataset([[1,2,3],[4,5,6],[7,8,9],[10,11,12]]) d1_row_keys = d1.getRowKeys() d1_col_keys = d1.getColKeys() row_keys, col_keys = d1.splitRowColKeylist(d1_row_keys+d1_col_keys) self.failUnless(row_keys == d1_row_keys) self.failUnless(col_keys == d1_col_keys) row_keys, col_keys = d1.splitRowColKeylist(d1_row_keys) self.failUnless(row_keys == d1_row_keys) self.failUnless(col_keys == []) row_keys, col_keys = d1.splitRowColKeylist(d1_col_keys) self.failUnless(row_keys == []) self.failUnless(col_keys == d1_col_keys) row_keys, col_keys = d1.splitRowColKeylist([]) self.failUnless(row_keys == []) self.failUnless(col_keys == []) max_key = d1.getNumRows() + d1.getNumCols() self.failUnlessRaises(ValueError, d1.splitRowColKeylist, [max_key + 1]) def suite(**kw): suite = unittest.makeSuite(Dataset2TestCases) return suite if __name__ == "__main__": unittest.main(defaultTest="suite")