def get_similar_films(db, main_film_id, weights=[1,1,1,1,1,1]):
# Given a Film ID and feature weights, calculate distances of this film to all films in the database
# using a batch_size of 100 films at a time
# @return: Film instance of main film, list of Film instances of 100 nearest neighbours
num_films = 100
main_film = Film(main_film_id)
vectors = main_film.build_vectors(db)
if vectors is None:
return None
else:
main_film.set_vectors(vectors)
main_film.set_metadata(db)
similar_films = []
batch_size = 100
offset = 0
while True:
vector_matrix = build_vector_matrix(db, batch_size, offset)
if vector_matrix is None:
break
distance_matrix = calculate_distance_matrix(main_film.vectors, vector_matrix, weights)
distance_matrix = [film for film in distance_matrix if int(film[0]) != int(main_film_id)]
distance_matrix = [(distance_matrix[i] + (vector_matrix[i][1:],)) for i in range(0,len(distance_matrix))
if distance_matrix[i][0] == vector_matrix[i][0]]
distance_matrix = sorted(distance_matrix, key=lambda x:x[1])
similar_films += distance_matrix[:num_films]
offset += batch_size
similar_films = sorted(similar_films, key=lambda x:x[1])
similar_films = similar_films[:num_films]
similar_films_instances = []
for film_id, distance, vectors in similar_films:
similar_film = Film(film_id)
similar_film.set_distance(main_film_id, distance)
similar_film.set_metadata(db)
similar_film.set_vectors(vectors)
similar_films_instances.append(similar_film)
vector_matrix = [(film.id,) + film.vectors for film in similar_films_instances]
for film in vector_matrix:
distance_matrix = calculate_distance_matrix(film, vector_matrix, weights)
for i, (film_id, dist) in enumerate(distance_matrix):
if similar_films_instances[i].id != film_id:
print("error!")
similar_films_instances[i].set_distance(film[0], dist)
return main_film, similar_films_instances
def get_film(db, main_film_id):
# Get Film instance for given ID from database
# @return: Film instance
main_film = Film(main_film_id)
vectors = main_film.build_vectors(db)
if vectors is None:
return None
main_film.set_vectors(vectors)
main_film.set_metadata(db)
return main_film