测试 Databricks ODBC 驱动程序 （Simba）

本页介绍如何测试使用 Databricks ODBC 驱动程序的代码。

对 ODBC 兼容的语言使用任何测试框架。 以下示例使用 pyodbc、pytest 和 unittest.mock 测试 ODBC 驱动程序连接。 此代码基于 将 Python 和 pyodbc 连接到 Azure Databricks 中的示例。

帮助函数

该文件 helpers.py 包含用于处理 ODBC 连接的实用工具函数：

• connect_to_dsn：打开与 Azure Databricks 计算资源的连接。
• get_cursor_from_connection：获取用于执行查询的游标。
• select_from_nyctaxi_trips：从samples.nyctaxi.trips查询指定的行数。
• print_rows：将结果集内容打印到控制台。

# helpers.py

from pyodbc import connect, Connection, Cursor

def connect_to_dsn(
  connstring: str,
  autocommit: bool
) -> Connection:

  connection = connect(
    connstring,
    autocommit = autocommit
  )

  return connection

def get_cursor_from_connection(
  connection: Connection
) -> Cursor:

  cursor = connection.cursor()
  return cursor

def select_from_nyctaxi_trips(
  cursor: Cursor,
  num_rows: int
) -> Cursor:

  select_cursor = cursor.execute(f"SELECT * FROM samples.nyctaxi.trips LIMIT {num_rows}")
  return select_cursor

def print_rows(cursor: Cursor):
  for row in cursor.fetchall():
    print(row)

Main 类

该文件 main.py 调用帮助程序函数来连接和查询数据：

# main.py

from helpers import *

connection = connect_to_dsn(
  connstring = "DSN=<your-dsn-name>",
  autocommit = True
)

cursor = get_cursor_from_connection(
  connection = connection)

select_cursor = select_from_nyctaxi_trips(
  cursor = cursor,
  num_rows = 2
)

print_rows(
  cursor = select_cursor
)

使用模拟进行单元测试

该文件 test_helpers.py 使用 pytest 和 unittest.mock 测试 select_from_nyctaxi_trips 函数。 通过模拟（Mocking）来仿真数据库连接，无需使用实际的计算资源，因此测试可以在数秒内完成，而不会对您的 Azure Databricks 工作区造成影响。

# test_helpers.py

from pyodbc import SQL_DBMS_NAME
from helpers import *
from unittest.mock import patch
import datetime

@patch("helpers.connect_to_dsn")
def test_connect_to_dsn(mock_connection):
  mock_connection.return_value.getinfo.return_value = "Spark SQL"

  mock_connection = connect_to_dsn(
    connstring = "DSN=<your-dsn-name>",
    autocommit = True
  )

  assert mock_connection.getinfo(SQL_DBMS_NAME) == "Spark SQL"

@patch('helpers.get_cursor_from_connection')
def test_get_cursor_from_connection(mock_connection):
  mock_cursor = mock_connection.return_value.cursor
  mock_cursor.return_value.rowcount = -1

  mock_connection = connect_to_dsn(
    connstring = "DSN=<your-dsn-name>",
    autocommit = True
  )

  mock_cursor = get_cursor_from_connection(
    connection = mock_connection
  )

  assert mock_cursor.rowcount == -1

@patch('helpers.select_from_nyctaxi_trips')
def test_select_from_nyctaxi_trips(mock_connection):
  mock_cursor = mock_connection.return_value.cursor
  mock_get_cursor = mock_cursor.return_value.execute
  mock_select_cursor = mock_get_cursor.return_value.arraysize = 1

  mock_connection = connect_to_dsn(
    connstring = "DSN=<your-dsn-name>",
    autocommit = True
  )

  mock_get_cursor = get_cursor_from_connection(
    connection = mock_connection
  )

  mock_select_cursor = select_from_nyctaxi_trips(
    cursor = mock_get_cursor,
    num_rows = 2
  )

  assert mock_select_cursor.arraysize == 1

@patch('helpers.print_rows')
def test_print_rows(mock_connection, capsys):
  mock_cursor = mock_connection.return_value.cursor
  mock_get_cursor = mock_cursor.return_value.execute
  mock_select_cursor = mock_get_cursor.return_value.fetchall.return_value = [
    (datetime.datetime(2016, 2, 14, 16, 52, 13), datetime.datetime(2016, 2, 14, 17, 16, 4), 4.94, 19.0, 10282, 10171),
    (datetime.datetime(2016, 2, 4, 18, 44, 19), datetime.datetime(2016, 2, 4, 18, 46), 0.28, 3.5, 10110, 10110)
  ]

  mock_connection = connect_to_dsn(
    connstring = "DSN=<your-dsn-name>",
    autocommit = True
  )

  mock_get_cursor = get_cursor_from_connection(
    connection = mock_connection
  )

  mock_select_cursor = select_from_nyctaxi_trips(
    cursor = mock_get_cursor,
    num_rows = 2
  )

  print_rows(
    cursor = mock_select_cursor
  )

  captured = capsys.readouterr()
  assert captured.out == "(datetime.datetime(2016, 2, 14, 16, 52, 13), datetime.datetime(2016, 2, 14, 17, 16, 4), 4.94, 19.0, 10282, 10171)\n" \
                         "(datetime.datetime(2016, 2, 4, 18, 44, 19), datetime.datetime(2016, 2, 4, 18, 46), 0.28, 3.5, 10110, 10110)\n"

由于 select_from_nyctaxi_trips 只运行 SELECT 语句，因此这里不需要严格的模拟。 但是，模拟在测试修改数据的函数 (INSERT INTO，UPDATE，DELETE FROM) 时特别有用，因为可以重复运行测试，而不会影响表状态。