def get_clean_graph_generator_memory_repos():
artists_ngram_repo = MemoryEntityNgrams(base_entity_uri=base_entities_URI,
type_of_entity_collection=ARTIST_COLLECTION)
songs_ngram_repo = MemoryEntityNgrams(base_entity_uri=base_entities_URI,
type_of_entity_collection=SONG_COLLECTION)
entity_counter_repo = MemoryEntityCounter()
songs_ngram_repo.reset_collection()
artists_ngram_repo.reset_collection()
entity_counter_repo.reset_count()
return GraphGenerator(repo_songs=songs_ngram_repo,
repo_artist=artists_ngram_repo,
repo_counter=entity_counter_repo)
set_of_artist = read_necessary_index_songs("files/artists_indexes_for_500000_discogs_songs.txt")
set_of_ids_aol = read_aol_discogs_ids("files/random_aol.txt")
set_of_musicbrainz_ids = read_mb_discogs_ids("files/random_musicbrainz.tsv")
ip_mongo = "127.0.0.1"
port_mongo = 27017
path_mongo = ip_mongo + ":" + str(port_mongo)
dataset = Dataset(title="selected_discogs_indexes")
artist_parser = DiscogsArtistParserFiltering("files/discogs_artists.xml",
dataset,
set_of_artist)
graph = MeraRdflibGraph(rdflib_graph=Graph())
artist_ngrams = MemoryEntityNgrams(base_entity_uri=base_entities_URI,
type_of_entity_collection=ARTIST_COLLECTION,
load_file=None)
song_ngrams = MemoryEntityNgrams(base_entity_uri=base_entities_URI,
type_of_entity_collection=SONG_COLLECTION,
load_file=None)
counter = MemoryEntityCounter()
# artist_ngrams = MongoEntityNgramsRepository(base_entity_uri=base_entities_URI,
# type_of_entity_collection=ARTIST_COLLECTION,
# url_root=path_mongo,
# host=ip_mongo,
# port=port_mongo)
# song_ngrams = MongoEntityNgramsRepository(base_entity_uri=base_entities_URI,
# type_of_entity_collection=SONG_COLLECTION,
# url_root=path_mongo,
# host=ip_mongo,
# port=port_mongo)