def test_to_int(value):
if isinstance(value, int):
assert XFastTrie._to_int(value, max_trie_entry_size) == value
elif isinstance(value, bytes):
value_int = unpack(
">Q", value.rjust((maxsize.bit_length() + 1) // 8, b'\x00'))[0]
assert XFastTrie._to_int(value, max_trie_entry_size) == value_int
def successor(self, value: Union[int, bytes]) -> Optional[int]:
"""
Find the smallest value in the trie strictly greater than the given value,
if it exists
:param value: The value to find the successor of
:return: The successor of the given value, or None if it doesn't exist
"""
value = XFastTrie._to_int(value, self._maxlen)
subtree, rep_node = self._get_value_subtree(value)
# subtree should be None only if the trie is empty
if subtree is None and self._count == 0:
raise ValueError("No values exist in trie")
elif value >= cast(int, self._max) or self._max is None:
return None
elif value < cast(int, self._min):
return self._min
subtree = cast(SortedList, subtree)
rep_node = cast(TrieNode, rep_node)
if max(subtree) <= value:
subtree = self._subtrees[rep_node.succ.value]
return cast(int, subtree[subtree.bisect_right(value)])
def predecessor(self, value: Union[int, bytes]) -> Optional[int]:
"""
Find the largest value in the trie strictly less than the given value,
if it exists
:param value: The value to find the predecessor of
:return: The predecessor of the given value, or None if it doesn't exist
"""
value = XFastTrie._to_int(value, self._maxlen)
subtree, rep_node = self._get_value_subtree(value)
# subtree should be None only if the trie is empty
if subtree is None and self._count == 0:
raise ValueError("No values exist in trie")
elif value <= cast(int, self._min) or self._min is None:
return None
elif value > cast(int, self._max):
return self._max
subtree = cast(SortedList, subtree)
rep_node = cast(TrieNode, rep_node)
if min(subtree) >= value:
subtree = self._subtrees[rep_node.pred.value]
return cast(int, subtree[subtree.bisect_left(value) - 1])
def insert(self, value: Union[int, bytes]) -> None:
"""
Insert a value into the trie
:param value: The value to insert into the trie
"""
value = XFastTrie._to_int(value, self._maxlen)
subtree, rep_node = self._get_value_subtree(value, True)
subtree = cast(SortedList, subtree)
rep_node = cast(TrieNode, rep_node)
# Do nothing if the value is already in the trie
if value in subtree:
return
if self._max is None or value > self._max:
self._max = value
if self._min is None or value < self._min:
self._min = value
subtree.add(value)
if len(subtree) > self._max_subtree_size:
# Out with the old
del self._subtrees[rep_node.value]
self._partitions -= rep_node.value
# In with the new
for tree in self._split_subtree(subtree, self._maxlen):
rep = self._calculate_representative(max(tree), self._maxlen)
self._partitions += rep
self._subtrees[rep] = tree
self._count += 1
def test_to_int_exceptions(value):
if isinstance(value, Integral):
with pytest.raises(ValueError):
XFastTrie._to_int(value, max_trie_entry_size)
elif isinstance(value, bytes):
with pytest.raises(ValueError):
XFastTrie._to_int(value, max_trie_entry_size)
else:
with pytest.raises(TypeError):
XFastTrie._to_int(value, max_trie_entry_size)
def test_get_closest_ancestor(entries, test_values):
t = XFastTrie(max_trie_entry_size)
for entry in entries:
t += entry
entries = [t._to_int(e, t._maxlen) for e in entries]
for val in test_values:
ancestor, level = t._get_closest_ancestor(val)
if val in entries:
assert ancestor.leaf
assert ancestor.value == val
else:
test_bits = format(val, 'b').zfill(t._maxlen)[:level + 2]
assert not ancestor.leaf
assert not ancestor.left.value_bits.startswith(test_bits)
assert not ancestor.right.value_bits.startswith(test_bits)
def test_get_closest_leaf(entries, test_values):
t = XFastTrie(max_trie_entry_size)
for entry in entries:
t += entry
entries = [t._to_int(e, t._maxlen) for e in entries]
for val in test_values:
neighbor = t._get_closest_leaf(val)
assert neighbor.leaf
if val in entries:
assert neighbor.value == val
else:
if neighbor.pred is not None:
assert abs(neighbor.value - val) <= abs(neighbor.pred.value -
val)
if neighbor.succ is not None:
assert abs(neighbor.value - val) <= abs(neighbor.succ.value -
val)
def __lt__(self, value: Union[int, bytes]) -> Optional[int]:
value = XFastTrie._to_int(value, self._maxlen)
return self.predecessor(value)
def __isub__(self, value: Union[int, bytes]) -> "YFastTrie":
value = XFastTrie._to_int(value, self._maxlen)
self.remove(value)
return self
def __iadd__(self, value: Union[int, bytes]) -> "YFastTrie":
value = XFastTrie._to_int(value, self._maxlen)
self.insert(value)
return self
def __contains__(self, value: Union[int, bytes]) -> bool:
value = XFastTrie._to_int(value, self._maxlen)
subtree, _ = self._get_value_subtree(value)
return subtree is not None and value in subtree
def remove(self, value: Union[int, bytes]) -> None:
"""
Remove the given value from the trie
:param value: The value to remove from the trie
"""
value = XFastTrie._to_int(value, self._maxlen)
subtree, rep_node = self._get_value_subtree(value)
# There should be no subtree only if the given value is not in the trie
if subtree is None or value not in subtree:
raise ValueError("Value does not exist in trie")
subtree = cast(SortedList, subtree)
rep_node = cast(TrieNode, rep_node)
if self._min == value:
if len(subtree) > 1:
min_succ = subtree[1]
else:
min_succ = self.successor(value)
else:
min_succ = -1
if self._max == value:
if len(subtree) > 1:
max_pred = subtree[-2]
else:
max_pred = self.predecessor(value)
else:
max_pred = -1
if min_succ != -1:
self._min = min_succ
if max_pred != -1:
self._max = max_pred
subtree.remove(value)
if len(subtree) == 0:
del self._subtrees[rep_node.value]
self._partitions -= rep_node.value
elif len(subtree) < self._min_subtree_size and len(
self._partitions) > 1:
if rep_node.pred is not None:
left_rep = rep_node.pred
right_rep = rep_node
else:
left_rep = rep_node
right_rep = rep_node.succ
left_tree = self._subtrees[left_rep.value]
right_tree = self._subtrees[right_rep.value]
# Out with the old
del self._subtrees[left_rep.value]
del self._subtrees[right_rep.value]
self._partitions -= left_rep.value
self._partitions -= right_rep.value
# In with the new
tree: SortedList
for tree in filter(
None,
self._merge_subtrees(left_tree, right_tree,
2 * self._maxlen)):
rep = self._calculate_representative(max(tree), self._maxlen)
self._partitions += rep
self._subtrees[rep] = tree
self._count -= 1