Пример #1
0
else:
    logging.disable(logging.NOTSET)




collection = EmailCollection()
collection.add_from_directory(cfg.PATH)

if hasattr(cfg,"LANG"):
    collection.keep_lang("fr")

engine = Clusterizer(collection)

engine.vectorizer            = getattr(cfg,"VECTORIZER", LogicVectorizer())
engine.algorithm             = getattr(cfg,"ALGORITHM",  HierarchicalAlgo())
engine.algorithm.n_clusters  = getattr(cfg,"N_CLUSTERS", 3)

engine.target                = getattr(cfg,"TARGET", "both")


if getattr(cfg,"ENABLE_TEST", False):
    assert hasattr(cfg,"TEST_CLUSTERING_RANGE"), "TEST_CLUSTERING_RANGE has not been defined"

    test_folder = getattr(cfg,"TEST_FOLDER", "results")
    if not os.path.exists(test_folder):
        os.makedirs(test_folder)

    engine.run_cleaner()
    engine.run_vectorizer()