我正在对 Expense 模型的方法进行单元测试。模型的相关部分如下所示:
class Expense(ndb.Model):
user = ndb.KeyProperty(User, required=True)
timestamp = ndb.DateTimeProperty(required=True)
...
@staticmethod
def new(user, amount, comment, description, timestamp):
"""creates a new expense and stores it. It returns the Expense instance"""
... validation ...
exp = Expense(user=user, timestamp=timestamp, ...)
exp.put()
return exp
@staticmethod
def list(user, from_date=None, to_date=None):
...
query = Expense.query(Expense.user == user)
if from_date is not None:
if not isinstance(from_date, datetime.datetime):
raise ValueError("from_date must be a datetime.datetime instance")
query.filter(Expense.timestamp >= from_date)
if to_date is not None:
if not isinstance(to_date, datetime.datetime):
raise ValueError("to_date must be a datetime.datetime instance")
query.filter(Expense.timestamp <= to_date)
query.order(- Expense.timestamp)
return query
我有以下单元测试来测试 Expense.list:
测试的前 4 行创建了 Expense 模型实例的实例。
- 2013 年 1 月 3 日
- 2013 年 11 月 1 日
- 2013 年 11 月 2 日
- 2014 年 1 月 1 日
为您节省一些水平滚动的主页。
def test_list_expenses_filter_by_date(user):
expense1 = Expense.new(user=user, amount=200.12, description="a netbook", comment="", timestamp=datetime.datetime(year=2013, month=1, day=3, hour=12, minute=30))
expense2 = Expense.new(user=user, amount=1.0, description="A can of coke", comment="Really cheap!",timestamp=datetime.datetime(year=2013, month=11, day=1, hour=12, minute=30))
expense3 = Expense.new(user=user, amount=3.05, description="A burger", comment="Tasty!", timestamp=datetime.datetime(year=2013, month=11, day=2, hour=12))
expense4 = Expense.new(user=user, amount=0.30, description="Toilet visit", comment="I normally don't pay for these things, but it was urgent.", timestamp=datetime.datetime(year=2014, month=1, day=1, hour=1, minute=15))
from_date = datetime.datetime(year=2013, month=5, day=1)
to_date = datetime.datetime(year=2013, month=12, day=31)
exps = Expense.list(user=user, from_date=from_date, to_date=to_date)
for expense in exps:
assert from_date <= expense.timestamp <= to_date
但是当我从 pytest 运行这个测试时，我在下一行得到了一个 AssertionError，正如 pytest 所显示的:
assert datetime.datetime(2013, 5, 1, 0, 0) <= datetime.datetime(2013, 1, 3, 12, 30)
E + 其中 datetime.datetime(2013, 1, 3, 12, 30) = Expense(key=Key('Expense', 5066549580791808), amount=200.12, comment=u'', desc...tamp= datetime.datetime(2013, 1, 3, 12, 30), user=Key('User', 5629499534213120)).timestamp
那么，为什么我的 Expense.list 方法会返回数据超出范围的实例？
我已经在 App Engine 测试台和请求处理程序中尝试过此操作，在浏览器中查看输出，因此它不是数据存储 stub 中的错误。
我也尝试过以不同的方式构建查询。我没有重复调用 Query.filter，而是将过滤器表达式添加到列表中，将该列表作为 *args 传递给 ndb.AND 并将其传递到单个 Query.filter 但这也不起作用。
我已经排除了 ndb 中存在错误的可能性。以下测试通过:
def test_ndb_date():
from google.appengine.ext import ndb
class Time(ndb.Model):
timestamp = ndb.DateTimeProperty(required=True)
t1 = Time(timestamp=datetime.datetime.now() - datetime.timedelta(days=365))
t2 = Time(timestamp=datetime.datetime.now())
t3 = Time(timestamp=datetime.datetime.now() + datetime.timedelta(days=365))
ndb.put_multi([t1, t2, t3])
from_date = datetime.datetime.now() - datetime.timedelta(days=1)
to_date = datetime.datetime.now() + datetime.timedelta(days = 1)
result = Time.query(Time.timestamp <= from_date, Time.timestamp >= to_date).fetch()
for time in result:
assert from_date <= time.timestamp >= to_date
请您参考如下方法:
我误解了 Query.filter 的工作原理。它不会改变当前的 Query 实例，而是返回一个新实例。
