|
| 1 | +#!/usr/bin/env python |
| 2 | +# encoding: utf-8 |
| 3 | +""" |
| 4 | +Testing code for the FP-growth implementation. |
| 5 | +""" |
| 6 | + |
| 7 | +import unittest |
| 8 | +import fp_growth |
| 9 | +from itertools import izip |
| 10 | +from pprint import pprint |
| 11 | + |
| 12 | +class NodeTester(object): |
| 13 | + def __init__(self, case, node): |
| 14 | + self.case = case |
| 15 | + self.node = node |
| 16 | + |
| 17 | + def child(self, item, count=None): |
| 18 | + c = self.node.search(item) |
| 19 | + self.case.failIf(c is None, 'no child with item %s' % item) |
| 20 | + |
| 21 | + tester = NodeTester(self.case, c) |
| 22 | + if count is not None: |
| 23 | + tester.count(count) |
| 24 | + return tester |
| 25 | + |
| 26 | + def count(self, count): |
| 27 | + self.case.assertEqual(self.node.count, count, |
| 28 | + 'expected count to be %d; instead it was %d' % |
| 29 | + (count, self.node.count)) |
| 30 | + return self |
| 31 | + |
| 32 | + def leaf(self): |
| 33 | + self.case.assertTrue(self.node.leaf, 'node must be a leaf') |
| 34 | + return self |
| 35 | + |
| 36 | +class TreeTestCase(unittest.TestCase): |
| 37 | + def setUp(self): |
| 38 | + self.tree = fp_growth.FPTree() |
| 39 | + self.root = NodeTester(self, self.tree.root) |
| 40 | + |
| 41 | + def nodes(self, item): |
| 42 | + return list(self.tree.nodes(item)) |
| 43 | + |
| 44 | + def assertPathsEqual(self, expected, actual): |
| 45 | + actual = list(actual) |
| 46 | + self.assertEqual(len(expected), len(actual)) |
| 47 | + |
| 48 | + for items, path in izip(expected, actual): |
| 49 | + self.assertEqual(len(items), len(path)) |
| 50 | + for item, node in izip(items, path): |
| 51 | + self.assertEqual(item, node.item) |
| 52 | + |
| 53 | +class InsertionTests(TreeTestCase): |
| 54 | + def testOneBranch(self): |
| 55 | + self.tree.add('abc') |
| 56 | + self.root.child('a', 1).child('b', 1).child('c', 1) |
| 57 | + |
| 58 | + def testIndependentBranches(self): |
| 59 | + self.tree.add('abc') |
| 60 | + self.tree.add('def') |
| 61 | + |
| 62 | + self.root.child('a', 1).child('b', 1).child('c', 1) |
| 63 | + self.root.child('d', 1).child('e', 1).child('f', 1) |
| 64 | + |
| 65 | + def testCommonPrefix(self): |
| 66 | + self.tree.add('abcd') |
| 67 | + self.tree.add('abde') |
| 68 | + |
| 69 | + b = self.root.child('a', 2).child('b', 2) |
| 70 | + b.child('c', 1).child('d', 1) |
| 71 | + b.child('d', 1).child('e', 1) |
| 72 | + |
| 73 | +class RouteTests(TreeTestCase): |
| 74 | + def testRoutes(self): |
| 75 | + self.tree.add('abc') |
| 76 | + self.tree.add('bcd') |
| 77 | + self.tree.add('cde') |
| 78 | + |
| 79 | + self.assertEqual(1, len(self.nodes('a'))) |
| 80 | + self.assertEqual(2, len(self.nodes('b'))) |
| 81 | + self.assertEqual(3, len(self.nodes('c'))) |
| 82 | + self.assertEqual(2, len(self.nodes('d'))) |
| 83 | + self.assertEqual(1, len(self.nodes('e'))) |
| 84 | + |
| 85 | + def testRemoveOnly(self): |
| 86 | + self.tree.add('ab') |
| 87 | + |
| 88 | + self.assertEqual(1, len(self.nodes('b'))) |
| 89 | + b = self.root.child('a').child('b').node |
| 90 | + self.assertTrue(self.root.child('a').node is b.parent) |
| 91 | + b.parent.remove(b) |
| 92 | + self.assertTrue(b.parent is None) |
| 93 | + self.assertEqual(0, len(self.nodes('b'))) |
| 94 | + |
| 95 | + def testNeighbors(self): |
| 96 | + self.tree.add('abc') |
| 97 | + self.tree.add('bcd') |
| 98 | + self.tree.add('cde') |
| 99 | + |
| 100 | + left_c = self.root.child('a').child('b').child('c').node |
| 101 | + middle_c = self.root.child('b').child('c').node |
| 102 | + right_c = self.root.child('c').node |
| 103 | + |
| 104 | + self.assertTrue(left_c.neighbor is middle_c) |
| 105 | + self.assertTrue(middle_c.neighbor is right_c) |
| 106 | + |
| 107 | + def testRemoveMiddle(self): |
| 108 | + self.tree.add('abc') |
| 109 | + self.tree.add('bcd') |
| 110 | + self.tree.add('cde') |
| 111 | + |
| 112 | + left_c = self.root.child('a').child('b').child('c').node |
| 113 | + middle_c = self.root.child('b').child('c').node |
| 114 | + right_c = self.root.child('c').node |
| 115 | + |
| 116 | + self.assertEqual(3, len(self.nodes('c'))) |
| 117 | + |
| 118 | + middle_c.parent.remove(middle_c) |
| 119 | + self.assertEqual(2, len(self.nodes('c'))) |
| 120 | + self.assertTrue(left_c.neighbor is right_c) |
| 121 | + |
| 122 | + def testRemoveEnd(self): |
| 123 | + self.tree.add('abc') |
| 124 | + self.tree.add('bcd') |
| 125 | + self.tree.add('cde') |
| 126 | + |
| 127 | + left_c = self.root.child('a').child('b').child('c').node |
| 128 | + middle_c = self.root.child('b').child('c').node |
| 129 | + right_c = self.root.child('c').node |
| 130 | + |
| 131 | + self.assertEqual(3, len(self.nodes('c'))) |
| 132 | + |
| 133 | + right_c.parent.remove(right_c) |
| 134 | + self.assertEqual(2, len(self.nodes('c'))) |
| 135 | + self.assertTrue(left_c.neighbor is middle_c) |
| 136 | + self.assertTrue(middle_c.neighbor is None) |
| 137 | + |
| 138 | +class PrefixPathTests(TreeTestCase): |
| 139 | + def testPaths(self): |
| 140 | + self.tree.add('abc') |
| 141 | + self.tree.add('bcd') |
| 142 | + self.tree.add('cde') |
| 143 | + |
| 144 | + self.assertPathsEqual(['abc', 'bc', 'c'], self.tree.prefix_paths('c')) |
| 145 | + |
| 146 | +class RemovalTests(TreeTestCase): |
| 147 | + def testLeafRemoval(self): |
| 148 | + self.tree.add('abc') |
| 149 | + c = self.root.child('a').child('b').child('c').node |
| 150 | + b = c.parent |
| 151 | + b.remove(c) |
| 152 | + self.assertTrue(c.leaf) |
| 153 | + |
| 154 | + def testMiddleRemoval(self): |
| 155 | + self.tree.add('abc') |
| 156 | + c = self.root.child('a').child('b').child('c').node |
| 157 | + b = c.parent |
| 158 | + a = b.parent |
| 159 | + |
| 160 | + a.remove(b) |
| 161 | + self.failIf(a.leaf, "the 'a' node should not be a leaf") |
| 162 | + self.assertTrue(c.parent is a, |
| 163 | + "the 'c' node should now be a child of 'a'") |
| 164 | + |
| 165 | + def testMerging(self): |
| 166 | + self.tree.add('abc') |
| 167 | + self.tree.add('ac') |
| 168 | + |
| 169 | + b = self.root.child('a').child('b').node |
| 170 | + a = b.parent |
| 171 | + a.remove(b) |
| 172 | + self.root.child('a').child('c', 2) |
| 173 | + |
| 174 | +class ConditionalTreeTests(TreeTestCase): |
| 175 | + def testGeneration(self): |
| 176 | + self.tree.add('abc') |
| 177 | + self.tree.add('abd') |
| 178 | + self.tree.add('ade') |
| 179 | + |
| 180 | + # Test the tree's structure, just because I'm paranoid. |
| 181 | + b = self.root.child('a', 3).child('b', 2) |
| 182 | + b.child('c', 1) |
| 183 | + b.child('d', 1) |
| 184 | + self.root.child('a').child('d', 1).child('e', 1) |
| 185 | + |
| 186 | + paths = list(self.tree.prefix_paths('d')) |
| 187 | + ct = fp_growth.conditional_tree_from_paths(paths, 1) |
| 188 | + root = NodeTester(self, ct.root) |
| 189 | + |
| 190 | + a = root.child('a', 2) |
| 191 | + a.child('b', 1).leaf() |
| 192 | + self.assertEqual(1, len(a.node.children)) |
| 193 | + |
| 194 | + def testPruning(self): |
| 195 | + self.tree.add('abc') |
| 196 | + self.tree.add('abd') |
| 197 | + self.tree.add('ac') |
| 198 | + self.tree.add('abe') |
| 199 | + self.tree.add('dc') |
| 200 | + |
| 201 | + paths = list(self.tree.prefix_paths('c')) |
| 202 | + ct = fp_growth.conditional_tree_from_paths(paths, 2) |
| 203 | + root = NodeTester(self, ct.root) |
| 204 | + |
| 205 | + root.child('a', 2).leaf() |
| 206 | + self.assertEqual(1, len(root.node.children)) |
| 207 | + |
| 208 | +if __name__ == '__main__': |
| 209 | + unittest.main() |
0 commit comments