X-Git-Url: http://nitlanguage.org diff --git a/lib/standard/collection/hash_collection.nit b/lib/standard/collection/hash_collection.nit index 13d5a7a..7c90189 100644 --- a/lib/standard/collection/hash_collection.nit +++ b/lib/standard/collection/hash_collection.nit @@ -11,27 +11,26 @@ # another product. # Introduce Hashmap and Hashset. -package hash_collection +module hash_collection import array -import hash # A HashCollection is an array of HashNode[K] indexed by the K hash value -private class HashCollection[K: Object, N: HashNode[K], E] - super Collection[E] +private abstract class HashCollection[K: Object, N: HashNode[Object]] super ArrayCapable[nullable N] - var _array: nullable NativeArray[nullable N] = null # Used to store items - var _capacity: Int = 0 # Size of _array - redef readable var _length: Int = 0 # Number of items in the map - readable var _first_item: nullable N = null # First added item (used to visit items in nice order) - var _last_item: nullable N = null # Last added item (same) + private var array: nullable NativeArray[nullable N] = null # Used to store items + private var capacity: Int = 0 # Size of _array + private var the_length: Int = 0 # Number of items in the map + + private var first_item: nullable N = null # First added item (used to visit items in nice order) + private var last_item: nullable N = null # Last added item (same) # The last key accessed (used for cache) - var _last_accessed_key: nullable K = null + private var last_accessed_key: nullable K = null # The last node accessed (used for cache) - var _last_accessed_node: nullable N = null + private var last_accessed_node: nullable N = null # Return the index of the key k fun index_at(k: K): Int @@ -44,8 +43,8 @@ private class HashCollection[K: Object, N: HashNode[K], E] # Return the node assosiated with the key fun node_at(k: K): nullable N do - # cache: `is' is used instead of `==' because it is a faster filter (even if not exact) - if k is _last_accessed_key then return _last_accessed_node + # cache: `is` is used instead of `==` because it is a faster filter (even if not exact) + if k.is_same_instance(_last_accessed_key) then return _last_accessed_node var res = node_at_idx(index_at(k), k) _last_accessed_key = k @@ -59,7 +58,7 @@ private class HashCollection[K: Object, N: HashNode[K], E] var c = _array[i] while c != null do var ck = c._key - if ck is k or ck == k then # prefilter with `is' because the compiler is not smart enought yet + if ck.is_same_instance(k) or ck == k then # FIXME prefilter because the compiler is not smart enought yet break end c = c._next_in_bucklet @@ -90,11 +89,15 @@ private class HashCollection[K: Object, N: HashNode[K], E] _last_accessed_node = node # Enlarge if needed - var l = _length - _length = l + 1 - l = (l + 5) * 3 / 2 + var l = _the_length + _the_length = l + 1 + + # Magic values determined empirically + # We do not want to enlarge too much + # We also want a odd capacity so that the modulo is more distributive + l = (l + 5) * 2 + 1 if l >= _capacity then - enlarge(l * 2) + enlarge(l * 3 / 2 + 1) end end @@ -120,7 +123,7 @@ private class HashCollection[K: Object, N: HashNode[K], E] end # Remove the item in the array - _length -= 1 + _the_length -= 1 prev = node._prev_in_bucklet next = node._next_in_bucklet if prev != null then @@ -135,6 +138,7 @@ private class HashCollection[K: Object, N: HashNode[K], E] _last_accessed_key = null end + # Clear the whole structure fun raz do var i = _capacity - 1 @@ -142,17 +146,18 @@ private class HashCollection[K: Object, N: HashNode[K], E] _array[i] = null i -= 1 end - _length = 0 + _the_length = 0 _first_item = null _last_item = null _last_accessed_key = null end + # Force a capacity fun enlarge(cap: Int) do var old_cap = _capacity # get a new capacity - if cap < _length + 1 then cap = _length + 1 + if cap < _the_length + 1 then cap = _the_length + 1 if cap <= _capacity then return _capacity = cap _last_accessed_key = null @@ -177,6 +182,7 @@ private class HashCollection[K: Object, N: HashNode[K], E] # Then store it in the array var next = new_array[index] new_array[index] = node + node._prev_in_bucklet = null node._next_in_bucklet = next if next != null then next._prev_in_bucklet = node node = node._next_item @@ -184,69 +190,108 @@ private class HashCollection[K: Object, N: HashNode[K], E] end end -private class HashNode[K: Object] - var _key: K +private abstract class HashNode[K: Object] + private var key: K type N: HashNode[K] - readable writable var _next_item: nullable N = null - readable writable var _prev_item: nullable N = null - var _prev_in_bucklet: nullable N = null - var _next_in_bucklet: nullable N = null + private var next_item: nullable N = null + private var prev_item: nullable N = null + private var prev_in_bucklet: nullable N = null + private var next_in_bucklet: nullable N = null init(k: K) do _key = k end end +# A map implemented with a hash table. +# Keys of such a map cannot be null and require a working `hash` method class HashMap[K: Object, V] super Map[K, V] - super HashCollection[K, HashMapNode[K, V], V] + super HashCollection[K, HashMapNode[K, V]] redef fun [](key) do var c = node_at(key) if c == null then - abort + return provide_default_value(key) else return c._value end end - redef fun has_key(key) do return node_at(key) != null - redef fun iterator: HashMapIterator[K, V] do return new HashMapIterator[K,V](self) - redef fun iterate - !each(e: V) + redef fun length do return _the_length + + redef fun is_empty do return _the_length == 0 + + redef fun []=(key, v) do - var c = _first_item - while c != null do - each(c._value) - c = c._next_item + var i = index_at(key) + var c = node_at_idx(i, key) + if c != null then + c._key = key + c._value = v + else + store(i, new HashMapNode[K, V](key, v)) end end - redef fun first + redef fun clear do raz + + init do - assert _length > 0 - return _first_item._value + _capacity = 0 + _the_length = 0 + enlarge(0) end - redef fun is_empty do return _length == 0 + redef var keys: RemovableCollection[K] = new HashMapKeys[K, V](self) + redef var values: RemovableCollection[V] = new HashMapValues[K, V](self) +end + +# View of the keys of a HashMap +private class HashMapKeys[K: Object, V] + super RemovableCollection[K] + # The original map + var map: HashMap[K, V] + + redef fun count(k) do if self.has(k) then return 1 else return 0 + redef fun first do return self.map._first_item._key + redef fun has(k) do return self.map.node_at(k) != null + redef fun has_only(k) do return (self.has(k) and self.length == 1) or self.is_empty + redef fun is_empty do return self.map.is_empty + redef fun length do return self.map.length + + redef fun iterator do return new MapKeysIterator[K, V](self.map.iterator) + + redef fun clear do self.map.clear + + redef fun remove(key) do self.map.remove_node(key) + redef fun remove_all(key) do self.map.remove_node(key) +end + +# View of the values of a Map +private class HashMapValues[K: Object, V] + super RemovableCollection[V] + # The original map + var map: HashMap[K, V] redef fun count(item) do var nb = 0 - var c = _first_item + var c = self.map._first_item while c != null do if c._value == item then nb += 1 c = c._next_item end return nb end + redef fun first do return self.map._first_item._value redef fun has(item) do - var c = _first_item + var c = self.map._first_item while c != null do if c._value == item then return true c = c._next_item @@ -256,7 +301,7 @@ class HashMap[K: Object, V] redef fun has_only(item) do - var c = _first_item + var c = self.map._first_item while c != null do if c._value != item then return false c = c._next_item @@ -264,46 +309,43 @@ class HashMap[K: Object, V] return true end - redef fun []=(key, v) - do - var i = index_at(key) - var c = node_at_idx(i, key) - if c != null then - c._key = key - c._value = v - else - store(i, new HashMapNode[K, V](key, v)) - end - end + redef fun is_empty do return self.map.is_empty + redef fun length do return self.map.length + + redef fun iterator do return new MapValuesIterator[K, V](self.map.iterator) + + redef fun clear do self.map.clear redef fun remove(item) do - var c = _first_item + var map = self.map + var c = map._first_item while c != null do if c._value == item then - remove_node(c._key) + map.remove_node(c._key) return end c = c._next_item end end - redef fun remove_at(key) do remove_node(key) - - redef fun clear do raz - - init + redef fun remove_all(item) do - _capacity = 0 - _length = 0 - enlarge(0) + var map = self.map + var c = map._first_item + while c != null do + if c._value == item then + map.remove_node(c._key) + end + c = c._next_item + end end end -class HashMapNode[K: Object, V] +private class HashMapNode[K: Object, V] super HashNode[K] redef type N: HashMapNode[K, V] - var _value: V + private var value: V init(k: K, v: V) do @@ -341,27 +383,31 @@ class HashMapIterator[K: Object, V] end # The map to iterate on - var _map: HashMap[K, V] + private var map: HashMap[K, V] # The current node - var _node: nullable HashMapNode[K, V] + private var node: nullable HashMapNode[K, V] init(map: HashMap[K, V]) do _map = map - _node = map.first_item + _node = map._first_item end end +# A `Set` implemented with a hash table. +# Keys of such a map cannot be null and require a working `hash` method class HashSet[E: Object] super Set[E] - super HashCollection[E, HashSetNode[E], E] + super HashCollection[E, HashSetNode[E]] + + redef fun length do return _the_length - redef fun is_empty do return _length == 0 + redef fun is_empty do return _the_length == 0 redef fun first do - assert _length > 0 + assert _the_length > 0 return _first_item._key end @@ -390,12 +436,20 @@ class HashSet[E: Object] init do _capacity = 0 - _length = 0 + _the_length = 0 enlarge(0) end + + # Build a list filled with the items of `coll`. + init from(coll: Collection[E]) do + init + add_all(coll) + end + + redef fun new_set do return new HashSet[E] end -class HashSetNode[E: Object] +private class HashSetNode[E: Object] super HashNode[E] redef type N: HashSetNode[E] @@ -405,7 +459,7 @@ class HashSetNode[E: Object] end end -class HashSetIterator[E: Object] +private class HashSetIterator[E: Object] super Iterator[E] redef fun is_ok do return _node != null @@ -422,10 +476,10 @@ class HashSetIterator[E: Object] end # The set to iterate on - var _set: HashSet[E] + private var set: HashSet[E] # The position in the internal map storage - var _node: nullable HashSetNode[E] + private var node: nullable HashSetNode[E] init(set: HashSet[E]) do