Esempio n. 1
0
    def setUpClass(cls):
        mdb, datastore = run_environment('mysql', config)
        cls.mdb = mdb

        models = cls.mdb.get_models()
        models = [x['name'] for x in models]
        if TEST_PREDICTOR_NAME in models:
            cls.mdb.delete_model(TEST_PREDICTOR_NAME)

        test_csv_path = get_test_csv(TEST_CSV['name'], TEST_CSV['url'])

        query('create database if not exists test')
        test_tables = query('show tables from test')
        test_tables = [x[0] for x in test_tables]
        if TEST_DATA_TABLE not in test_tables:
            print('creating test data table...')
            query(f'''
                CREATE TABLE test.{TEST_DATA_TABLE} (
                    number_of_rooms int,
                    number_of_bathrooms int,
                    sqft int,
                    location varchar(100),
                    days_on_market int,
                    initial_price int,
                    neighborhood varchar(100),
                    rental_price int
                )
            ''')

            with open(test_csv_path) as f:
                csvf = csv.reader(f)
                i = 0
                for row in csvf:
                    if i > 0:
                        number_of_rooms = int(row[0])
                        number_of_bathrooms = int(row[1])
                        sqft = int(float(row[2].replace(',', '.')))
                        location = str(row[3])
                        days_on_market = int(row[4])
                        initial_price = int(row[5])
                        neighborhood = str(row[6])
                        rental_price = int(float(row[7]))
                        query(f'''INSERT INTO test.{TEST_DATA_TABLE} VALUES (
                            {number_of_rooms},
                            {number_of_bathrooms},
                            {sqft},
                            '{location}',
                            {days_on_market},
                            {initial_price},
                            '{neighborhood}',
                            {rental_price}
                        )''')
                    i += 1
            print('done')

        ds = datastore.get_datasource(EXTERNAL_DS_NAME)
        if ds is not None:
            datastore.delete_datasource(EXTERNAL_DS_NAME)
        short_csv_file_path = get_test_csv(f'{EXTERNAL_DS_NAME}.csv', TEST_CSV['url'], lines_count=300, rewrite=True)
        datastore.save_datasource(EXTERNAL_DS_NAME, 'file', 'test.csv', short_csv_file_path)
Esempio n. 2
0
    def setUpClass(cls):
        run_container('mongo-config')
        ready = wait_port(27000, DOCKER_TIMEOUT)
        assert ready

        run_container('mongo-instance')
        ready = wait_port(27001, DOCKER_TIMEOUT)
        assert ready

        cls.config_client = MongoClient('mongodb://localhost:27000/')
        cls.instance_client = MongoClient('mongodb://localhost:27001/')

        print('init replconf')
        try:
            r = cls.config_client.admin.command(
                'replSetInitiate', {
                    '_id': 'replconf',
                    'members': [{
                        '_id': 0,
                        'host': '127.0.0.1:27000'
                    }]
                })
        except Exception as e:
            print('already initialized')
            if str(e) == 'already initialized':
                r = {'ok': 1}

        if bool(r['ok']) is not True:
            assert False

        print('init replmain')
        try:
            r = cls.instance_client.admin.command(
                'replSetInitiate', {
                    '_id': 'replmain',
                    'members': [{
                        '_id': 0,
                        'host': '127.0.0.1:27001'
                    }]
                })
        except Exception as e:
            print('already initialized')
            if str(e) == 'already initialized':
                r = {'ok': 1}

        if bool(r['ok']) is not True:
            assert False

        mdb, datastore = run_environment('mongodb', config, run_apis='mongodb')
        cls.mdb = mdb

        models = cls.mdb.get_models()
        models = [x['name'] for x in models]
        if TEST_PREDICTOR_NAME in models:
            cls.mdb.delete_model(TEST_PREDICTOR_NAME)

        test_csv_path = get_test_csv(TEST_CSV['name'], TEST_CSV['url'])

        db = cls.instance_client['test_data']
        colls = db.list_collection_names()

        if 'home_rentals' not in colls:
            print('creatating test data')
            with open(test_csv_path) as f:
                csvf = csv.reader(f)
                data = []
                for i, row in enumerate(csvf):
                    if i > 0:
                        data.append(
                            dict(number_of_rooms=int(row[0]),
                                 number_of_bathrooms=int(row[1]),
                                 sqft=int(float(row[2].replace(',', '.'))),
                                 location=str(row[3]),
                                 days_on_market=int(row[4]),
                                 initial_price=int(row[5]),
                                 neighborhood=str(row[6]),
                                 rental_price=int(float(row[7]))))
                db['home_rentals'].insert_many(data)
            print('done')

        run_container('mongo-mongos')
        ready = wait_port(27002, DOCKER_TIMEOUT)
        assert ready
        cls.mongos_client = MongoClient('mongodb://localhost:27002/')

        cls.mongos_client.admin.command('addShard', 'replmain/127.0.0.1:27001')
        cls.mongos_client.admin.command(
            'addShard', f"127.0.0.1:{config['api']['mongodb']['port']}")
Esempio n. 3
0
    def setUpClass(cls):
        mdb, datastore = run_environment('mssql', config)
        cls.mdb = mdb

        models = cls.mdb.get_models()
        models = [x['name'] for x in models]
        if TEST_PREDICTOR_NAME in models:
            cls.mdb.delete_model(TEST_PREDICTOR_NAME)

        test_csv_path = get_test_csv(TEST_CSV['name'], TEST_CSV['url'])

        res = query(
            "SELECT name FROM master.dbo.sysdatabases where name = 'mindsdb_test'",
            fetch=True)
        if len(res) == 0:
            query("create database mindsdb_test")
        res = query('''
            select * from sys.schemas where name = 'mindsdb_schema';
        ''',
                    fetch=True)
        if len(res) == 0:
            query('''
                create schema [mindsdb_schema];
            ''')

        # query('create database if not exists test')
        # show tables from test
        test_tables = query(f'''
            select 1 from sysobjects where name='{TEST_DATA_TABLE}' and xtype='U';
        ''',
                            fetch=True)
        if len(test_tables) == 0:
            print('creating test data table...')
            query(f'''
                CREATE TABLE mindsdb_schema.{TEST_DATA_TABLE} (
                    number_of_rooms int,
                    number_of_bathrooms int,
                    sqft int,
                    location varchar(100),
                    days_on_market int,
                    initial_price int,
                    neighborhood varchar(100),
                    rental_price int
                )
            ''')

            with open(test_csv_path) as f:
                csvf = csv.reader(f)
                i = 0
                for row in csvf:
                    if i > 0:
                        number_of_rooms = int(row[0])
                        number_of_bathrooms = int(row[1])
                        sqft = int(float(row[2].replace(',', '.')))
                        location = str(row[3])
                        days_on_market = int(row[4])
                        initial_price = int(row[5])
                        neighborhood = str(row[6])
                        rental_price = int(float(row[7]))
                        query(f'''
                            INSERT INTO mindsdb_schema.{TEST_DATA_TABLE} VALUES (
                            {number_of_rooms},
                            {number_of_bathrooms},
                            {sqft},
                            '{location}',
                            {days_on_market},
                            {initial_price},
                            '{neighborhood}',
                            {rental_price}
                        )''')
                    i += 1
                    if i % 100 == 0:
                        print(i)
            print('done')

        ds = datastore.get_datasource(EXTERNAL_DS_NAME)
        if ds is not None:
            datastore.delete_datasource(EXTERNAL_DS_NAME)
        short_csv_file_path = get_test_csv(f'{EXTERNAL_DS_NAME}.csv',
                                           TEST_CSV['url'],
                                           lines_count=300,
                                           rewrite=True)
        datastore.save_datasource(EXTERNAL_DS_NAME, 'file', 'test.csv',
                                  short_csv_file_path)