def test_delete_last():
heap = Heap()
items = [KeyedItem(key=i) for i in range(10)]
heap.heapify(items)
heap.delete(items[9])
for i in range(9):
assert heap.extract_root() is items[i]
assert heap.extract_root() is None
def test_construct_max_heap():
heap = Heap(heaptype='max')
items = [KeyedItem(key=i) for i in range(10)]
random_items = [item for item in items]
random.shuffle(random_items)
heap.heapify(random_items)
for i in range(10):
assert heap.extract_root() is items[10 - i - 1]
assert heap.extract_root() is None
def test_delete_root():
heap = Heap()
items = [KeyedItem(key=i) for i in range(10)]
random_items = [item for item in items]
random.shuffle(random_items)
heap.heapify(random_items)
heap.delete(items[0])
for i in range(1, 10):
assert heap.extract_root() is items[i]
assert heap.extract_root() is None
def test_insert_min_heap():
heap = Heap()
items = [KeyedItem(key=i) for i in range(10)]
random_items = [item for item in items]
random.shuffle(random_items)
for item in random_items:
heap.insert(item)
for i in range(10):
assert heap.extract_root() is items[i]
assert heap.extract_root() is None
def test_delete_with_bubble_down():
heap = Heap()
items_keys = list(range(5)) + list(range(100, 112))
items = [KeyedItem(key=i) for i in items_keys]
heap.heapify(items)
heap.delete(items[5])
for i in range(17):
if i != 5:
root = heap.extract_root()
assert root is items[i]
assert heap.extract_root() is None
def heapsort(items: Sequence[KeyedItem], order=None) -> List[KeyedItem]:
"""O(n*log n)"""
heap = Heap(heaptype=order)
heap.heapify(items)
sorted_items = []
while True:
item = heap.extract_root()
if not item:
break
sorted_items.append(item)
return sorted_items
def test_delete_with_bubble_up():
heap = Heap()
items_keys = (list(range(2)) + [100] + list(range(2, 4)) +
list(range(101, 103)) + list(range(4, 8)) +
list(range(103, 107)) + list(range(8, 10)))
items = [KeyedItem(key=i) for i in items_keys]
heap.heapify(items)
heap.delete(items[5])
for i in range(2):
assert heap.extract_root() is items[i]
for i in range(3, 5):
assert heap.extract_root() is items[i]
for i in range(7, 11):
assert heap.extract_root() is items[i]
for i in range(15, 17):
assert heap.extract_root() is items[i]
assert heap.extract_root() is items[2]
assert heap.extract_root() is items[6]
for i in range(11, 15):
assert heap.extract_root() is items[i]
assert heap.extract_root() is None
def test_delete_only_item():
heap = Heap()
item = KeyedItem(key=0)
heap.insert(item)
heap.delete(item)
assert heap.extract_root() is None
def test_construct_empty_heap():
heap = Heap()
items = []
heap.heapify(items)
assert heap.extract_root() is None
def run_dijkstra_algorithm(graph: Graph,
source: Vertex) -> NodeList:
nb_processed_vertices = 1
nb_vertices = graph.nb_vertices
shortest_paths = NodeList(vertices=graph.adjacency_lists)
# a list of vertices does the bookkeeping of keeping track
# of which vertices are processed/unprocessed
vertex_statuses = NodeList(vertices=graph.adjacency_lists,
default=False)
class ItemContent:
def __init__(self, vertex: Vertex, edge: Edge=None):
self.vertex = vertex
self.edge = edge
class Path:
def __init__(self, source: Vertex):
self.source = source
self.destination = None
self.path_to_last_hop = None
self.last_hop = None
self.distance = None
def build(self, path_to_last_hop, last_hop: Edge):
if self.source is not path_to_last_hop.source:
raise PathError
if path_to_last_hop.destination is not last_hop.head:
raise PathError
self.path_to_last_hop = path_to_last_hop
self.last_hop = last_hop
self.destination = last_hop.tail
self.distance = path_to_last_hop.distance + last_hop.weight
# a heap contains all vertices not in the path
# their keys are the min distance to the source vertex
# initialization in O(n + m)
heap = Heap(heaptype='min')
for vertex in graph.adjacency_lists:
shortest_paths[vertex] = Path(source=source)
shortest_paths[vertex].distance = inf
shortest_paths[source].destination = source
shortest_paths[source].distance = 0
for edgenode in graph.adjacency_lists[source].edgenodes:
tail = edgenode.tail
if edgenode.weight < 0:
raise GraphTypeError
if shortest_paths[tail].distance is not None and edgenode.weight < shortest_paths[tail].distance:
shortest_paths[tail].destination = tail
shortest_paths[tail].last_hop = edgenode.to_edge(head=source)
shortest_paths[tail].distance = edgenode.weight
vertex_items = []
for vertex in graph.adjacency_lists:
if vertex is not source:
vertex_item = KeyedItem(key=shortest_paths[vertex].distance,
content=ItemContent(vertex=vertex,
edge=shortest_paths[vertex].last_hop))
vertex_items.append(vertex_item)
vertex_statuses[vertex] = vertex_item
heap.heapify(vertex_items)
del vertex_items
def update_heap(deleted_vertex: Vertex, distance_to_deleted_vertex: float):
nonlocal graph, heap, vertex_statuses
# update status
vertex_statuses[deleted_vertex] = False
# remove from the heap all vertices connected to
# the deleted vertex
for edgenode in graph.adjacency_lists[deleted_vertex].edgenodes:
tail = edgenode.tail
# check if the tail is in unprocessed vertices set
vertex_item = vertex_statuses[tail]
if vertex_item:
distance = vertex_item.key
vertex = vertex_item.content.vertex
edge = vertex_item.content.edge
# delete it from the heap
heap.delete(vertex_item)
# recompute its distance
if edgenode.weight < 0:
raise GraphTypeError
new_distance = distance_to_deleted_vertex + edgenode.weight
if new_distance < distance:
distance = new_distance
edge = edgenode.to_edge(head=deleted_vertex)
vertex_item.key = distance
vertex_item.content = ItemContent(vertex=vertex, edge=edge)
# insert it back into the heap
heap.insert(vertex_item)
while nb_processed_vertices < nb_vertices:
# pick cheapest edge crossing the cut
# from the graph shortest_path_tree = (processed_vertices, paths)
# to the graph graph - shortest_path_tree
heap_item = heap.extract_root()
last_hop = heap_item.content.edge
v = heap_item.content.vertex
# add it to the path leading to v
shortest_paths[v].build(path_to_last_hop=shortest_paths[last_hop.head],
last_hop=last_hop)
# update nb_processed_vertices
nb_processed_vertices += 1
# update heap
update_heap(deleted_vertex=v, distance_to_deleted_vertex=shortest_paths[v].distance)
return shortest_paths