recommendation_creation.py 文件源码

def import_data_from_psql(user_id):
    """Import data from psql; clean & merge dataframes."""
    library = pd.read_sql_table(
        'library',
        con='postgres:///nextbook',
        columns=['book_id', 'title', 'author', 'pub_year', 'original_pub_year', 'pages'])

    book_subjects = pd.read_sql_table(
        'book_subjects',
        con='postgres:///nextbook')

    subjects = pd.read_sql_table(
        'subjects', con='postgres:///nextbook',
        columns=['subject_id', 'subject'])

    user_ratings = pd.read_sql_query(
        sql=('SELECT book_id, user_id, status, rating FROM user_books WHERE user_id=%s' % user_id),
        con='postgres:///nextbook')

    library = library.merge(user_ratings, how='left', on='book_id')
    library['pages'].fillna(0, inplace=True)

    # merge subject names into book_subjects; drop uninteresting subjects from book_subjects table
    book_subjects = book_subjects.merge(subjects, how='left', on='subject_id')
    delete_values = ["protected daisy", "accessible book", "in library", "overdrive", "large type books", 'ficci\xc3\xb3n juvenil', 'ficci\xc3\xb3n', 'lending library']
    book_subjects = book_subjects[~book_subjects['subject'].isin(delete_values)]

    return [library, book_subjects, subjects]