|
1 | | -""" |
2 | | -MAP Abstract Data Type |
3 | | -Map() Create a new, empty map. It returns an empty map collection. |
4 | | -put(key,val) Add a new key-value pair to the map. If the key is already in the map then replace the old value with the new value. |
5 | | -get(key) Given a key, return the value stored in the map or None otherwise. |
6 | | -del Delete the key-value pair from the map using a statement of the form del map[key]. |
7 | | -len() Return the number of key-value pairs stored in the map. |
8 | | -in Return True for a statement of the form key in map, if the given key is in the map, False otherwise. |
9 | | -""" |
| 1 | +from unittest import TestCase |
| 2 | + |
| 3 | + |
10 | 4 | class HashTable(object): |
11 | | - def __init__(self, size = 11): |
| 5 | + """ |
| 6 | + HashMap Data Type |
| 7 | + HashMap() Create a new, empty map. It returns an empty map collection. |
| 8 | + put(key, val) Add a new key-value pair to the map. If the key is already in the map then replace |
| 9 | + the old value with the new value. |
| 10 | + get(key) Given a key, return the value stored in the map or None otherwise. |
| 11 | + del_(key) or del map[key] Delete the key-value pair from the map using a statement of the form del map[key]. |
| 12 | + len() Return the number of key-value pairs stored in the map. |
| 13 | + in Return True for a statement of the form key in map, if the given key is in the map, False otherwise. |
| 14 | + """ |
| 15 | + |
| 16 | + _empty = object() |
| 17 | + _deleted = object() |
| 18 | + |
| 19 | + def __init__(self, size=11): |
12 | 20 | self.size = size |
13 | | - self.keys = [None] * size # keys |
14 | | - self.values = [None] * size # values |
| 21 | + self._len = 0 |
| 22 | + self._keys = [self._empty] * size # keys |
| 23 | + self._values = [self._empty] * size # values |
15 | 24 |
|
16 | 25 | def put(self, key, value): |
17 | | - hashval = self.hash(key) |
18 | | - |
19 | | - if hashval not in keys: |
20 | | - self.keys[hashval] = key |
21 | | - self.values[hashval] = value |
22 | | - else: |
23 | | - if self.keys[hashval] == key: # replace |
24 | | - self.values[hashval] = value |
25 | | - else: # probing |
26 | | - rehashval = self.rehash(key) |
27 | | - while self.keys[rehashval] is not None and \ |
28 | | - self.keys[rehashval] != key: |
29 | | - rehashval = self.rehash(rehashval) |
30 | | - if keys[rehashval] is None: |
31 | | - self.keys[rehashval] = key |
32 | | - self.values[rehashval] = value |
33 | | - else: |
34 | | - self.values[rehashval] = value # replacee |
| 26 | + initial_hash = hash_ = self.hash(key) |
| 27 | + |
| 28 | + while True: |
| 29 | + if self._keys[hash_] is self._empty or self._keys[hash_] is self._deleted: |
| 30 | + # can assign to hash_ index |
| 31 | + self._keys[hash_] = key |
| 32 | + self._values[hash_] = value |
| 33 | + self._len += 1 |
| 34 | + return |
| 35 | + elif self._keys[hash_] == key: |
| 36 | + # key already exists here, assign over |
| 37 | + self._keys[hash_] = key |
| 38 | + self._values[hash_] = value |
| 39 | + return |
| 40 | + |
| 41 | + hash_ = self._rehash(hash_) |
| 42 | + |
| 43 | + if initial_hash == hash_: |
| 44 | + # table is full |
| 45 | + raise ValueError("Table is full") |
35 | 46 |
|
36 | 47 | def get(self, key): |
37 | | - pass |
| 48 | + initial_hash = hash_ = self.hash(key) |
| 49 | + while True: |
| 50 | + if self._keys[hash_] is self._empty: |
| 51 | + # That key was never assigned |
| 52 | + return None |
| 53 | + elif self._keys[hash_] == key: |
| 54 | + # key found |
| 55 | + return self._values[hash_] |
| 56 | + |
| 57 | + hash_ = self._rehash(hash_) |
| 58 | + if initial_hash == hash_: |
| 59 | + # table is full and wrapped around |
| 60 | + return None |
| 61 | + |
| 62 | + def del_(self, key): |
| 63 | + initial_hash = hash_ = self.hash(key) |
| 64 | + while True: |
| 65 | + if self._keys[hash_] is self._empty: |
| 66 | + # That key was never assigned |
| 67 | + return None |
| 68 | + elif self._keys[hash_] == key: |
| 69 | + # key found, assign with deleted sentinel |
| 70 | + self._keys[hash_] = self._deleted |
| 71 | + self._values[hash_] = self._deleted |
| 72 | + self._len -= 1 |
| 73 | + return |
| 74 | + |
| 75 | + hash_ = self._rehash(hash_) |
| 76 | + if initial_hash == hash_: |
| 77 | + # table is full and wrapped around |
| 78 | + return None |
38 | 79 |
|
39 | 80 | def hash(self, key): |
40 | 81 | return key % self.size |
41 | 82 |
|
42 | | - def rehash(self, oldhash): |
| 83 | + def _rehash(self, old_hash): |
43 | 84 | """ |
44 | 85 | linear probing |
45 | 86 | """ |
46 | | - return (oldhash + 1) % self.size |
| 87 | + return (old_hash + 1) % self.size |
47 | 88 |
|
48 | | - def __getitem__(self,key): |
| 89 | + def __getitem__(self, key): |
49 | 90 | return self.get(key) |
50 | 91 |
|
| 92 | + def __delitem__(self, key): |
| 93 | + return self.del_(key) |
| 94 | + |
51 | 95 | def __setitem__(self, key, value): |
52 | 96 | self.put(key, value) |
53 | 97 |
|
| 98 | + def __len__(self): |
| 99 | + return self._len |
| 100 | + |
| 101 | + |
| 102 | +class ResizableHashTable(HashTable): |
| 103 | + MIN_SIZE = 8 |
| 104 | + |
| 105 | + def __init__(self): |
| 106 | + super().__init__(self.MIN_SIZE) |
| 107 | + |
| 108 | + def put(self, key, value): |
| 109 | + rv = super().put(key, value) |
| 110 | + # increase size of dict * 2 if filled >= 2/3 size (like python dict) |
| 111 | + if len(self) >= (self.size * 2) / 3: |
| 112 | + self.__resize() |
| 113 | + |
| 114 | + def __resize(self): |
| 115 | + keys, values = self._keys, self._values |
| 116 | + self.size *= 2 # this will be the new size |
| 117 | + self._len = 0 |
| 118 | + self._keys = [self._empty] * self.size |
| 119 | + self._values = [self._empty] * self.size |
| 120 | + for key, value in zip(keys, values): |
| 121 | + if key is not self._empty and key is not self._deleted: |
| 122 | + self.put(key, value) |
| 123 | + |
| 124 | + |
| 125 | +class TestHashTable(TestCase): |
| 126 | + def test_one_entry(self): |
| 127 | + m = HashTable(10) |
| 128 | + m.put(1, '1') |
| 129 | + self.assertEqual('1', m.get(1)) |
| 130 | + |
| 131 | + def test_add_entry_bigger_than_table_size(self): |
| 132 | + m = HashTable(10) |
| 133 | + m.put(11, '1') |
| 134 | + self.assertEqual('1', m.get(11)) |
| 135 | + |
| 136 | + def test_get_none_if_key_missing_and_hash_collision(self): |
| 137 | + m = HashTable(10) |
| 138 | + m.put(1, '1') |
| 139 | + self.assertEqual(None, m.get(11)) |
| 140 | + |
| 141 | + def test_two_entries_with_same_hash(self): |
| 142 | + m = HashTable(10) |
| 143 | + m.put(1, '1') |
| 144 | + m.put(11, '11') |
| 145 | + self.assertEqual('1', m.get(1)) |
| 146 | + self.assertEqual('11', m.get(11)) |
| 147 | + |
| 148 | + def test_get_on_full_table_does_halts(self): |
| 149 | + # and does not search forever |
| 150 | + m = HashTable(10) |
| 151 | + for i in range(10, 20): |
| 152 | + m.put(i, i) |
| 153 | + self.assertEqual(None, m.get(1)) |
| 154 | + |
| 155 | + def test_delete_key(self): |
| 156 | + m = HashTable(10) |
| 157 | + m.put(1, 1) |
| 158 | + m.del_(1) |
| 159 | + self.assertEqual(None, m.get(1)) |
| 160 | + |
| 161 | + def test_delete_key_and_reassign(self): |
| 162 | + m = HashTable(10) |
| 163 | + m.put(1, 1) |
| 164 | + del m[1] |
| 165 | + m.put(1, 2) |
| 166 | + self.assertEqual(2, m.get(1)) |
| 167 | + |
| 168 | + def test_assigning_to_full_table_throws_error(self): |
| 169 | + m = HashTable(3) |
| 170 | + m.put(1, 1) |
| 171 | + m.put(2, 2) |
| 172 | + m.put(3, 3) |
| 173 | + with self.assertRaises(ValueError): |
| 174 | + m.put(4, 4) |
| 175 | + |
| 176 | + def test_len_trivial(self): |
| 177 | + m = HashTable(10) |
| 178 | + self.assertEqual(0, len(m)) |
| 179 | + for i in range(10): |
| 180 | + m.put(i, i) |
| 181 | + self.assertEqual(i + 1, len(m)) |
| 182 | + |
| 183 | + def test_len_after_deletions(self): |
| 184 | + m = HashTable(10) |
| 185 | + m.put(1, 1) |
| 186 | + self.assertEqual(1, len(m)) |
| 187 | + m.del_(1) |
| 188 | + self.assertEqual(0, len(m)) |
| 189 | + m.put(11, 42) |
| 190 | + self.assertEqual(1, len(m)) |
54 | 191 |
|
| 192 | + def test_resizable_hash_table(self): |
| 193 | + m = ResizableHashTable() |
| 194 | + self.assertEqual(ResizableHashTable.MIN_SIZE, m.size) |
| 195 | + for i in range(ResizableHashTable.MIN_SIZE): |
| 196 | + m.put(i, 'foo') |
| 197 | + self.assertEqual(ResizableHashTable.MIN_SIZE * 2, m.size) |
| 198 | + self.assertEqual('foo', m.get(1)) |
| 199 | + self.assertEqual('foo', m.get(3)) |
| 200 | + self.assertEqual('foo', m.get(ResizableHashTable.MIN_SIZE - 1)) |
0 commit comments