PyHamcrest
Introduction
PyHamcrest is a framework for writing matcher objects, allowing you to declaratively define "match" rules. There are a number of situations where matchers are invaluable, such as UI validation, or data filtering, but it is in the area of writing flexible tests that matchers are most commonly used. This tutorial shows you how to use PyHamcrest for unit testing.
When writing tests it is sometimes difficult to get the balance right between overspecifying the test (and making it brittle to changes), and not specifying enough (making the test less valuable since it continues to pass even when the thing being tested is broken). Having a tool that allows you to pick out precisely the aspect under test and describe the values it should have, to a controlled level of precision, helps greatly in writing tests that are "just right." Such tests fail when the behavior of the aspect under test deviates from the expected behavior, yet continue to pass when minor, unrelated changes to the behaviour are made.
Installation
Hamcrest can be installed using the usual Python packaging tools. It depends on distribute, but as long as you have a network connection when you install, the installation process will take care of that for you.
My first PyHamcrest test
We'll start by writing a very simple PyUnit test, but instead of using PyUnit's assertEqual
method, we'll use PyHamcrest's assert_that
construct and the standard set of matchers:
from hamcrest import *
import unittest
class BiscuitTest(unittest.TestCase):
def testEquals(self):
theBiscuit = Biscuit("Ginger")
myBiscuit = Biscuit("Ginger")
assert_that(theBiscuit, equal_to(myBiscuit))
if __name__ == "__main__":
unittest.main()
The assert_that
function is a stylized sentence for making a test assertion. In this example, the subject of the assertion is the object theBiscuit
, which is the first method parameter. The second method parameter is a matcher for Biscuit
objects, here a matcher that checks one object is equal to another using the Python ==
operator. The test passes since the Biscuit
class defines an __eq__
method.
If you have more than one assertion in your test you can include an identifier for the tested value in the assertion:
assert_that(theBiscuit.getChocolateChipCount(), equal_to(10), "chocolate chips")
assert_that(theBiscuit.getHazelnutCount(), equal_to(3), "hazelnuts")
As a convenience, assert_that can also be used to verify a boolean condition:
assert_that(theBiscuit.isCooked(), "cooked")
This is equivalent to the assert_
method of unittest.TestCase, but because it's a standalone function, it offers greater flexibility in test writing.
Predefined matchers
PyHamcrest comes with a library of useful matchers:
- Object
equal_to
- match equal objecthas_length
- matchlen()
has_property
- match value of property with given namehas_properties
- match an object that has all of the given properties.has_string
- matchstr()
instance_of
- match object typenone
,not_none
- matchNone
, or notNone
same_instance
- match same objectcalling, raises
- wrap a method call and assert that it raises an exception
- Number
close_to
- match number close to a given valuegreater_than
,greater_than_or_equal_to
,less_than
,less_than_or_equal_to
- match numeric ordering
- Text
contains_string
- match part of a stringends_with
- match the end of a stringequal_to_ignoring_case
- match the complete string but ignore caseequal_to_ignoring_whitespace
- match the complete string but ignore extra whitespacematches_regexp
- match a regular expression in a stringstarts_with
- match the beginning of a stringstring_contains_in_order
- match parts of a string, in relative order
- Logical
all_of
-and
together all matchersany_of
-or
together all matchersanything
- match anything, useful in composite matchers when you don't care about a particular valueis_not
,not_
- negate the matcher
- Sequence
contains
- exactly match the entire sequencecontains_inanyorder
- match the entire sequence, but in any orderhas_item
- match if given item appears in the sequencehas_items
- match if all given items appear in the sequence, in any orderis_in
- match if item appears in the given sequenceonly_contains
- match if sequence's items appear in given listempty
- match if the sequence is empty
- Dictionary
has_entries
- match dictionary with list of key-value pairshas_entry
- match dictionary containing a key-value pairhas_key
- match dictionary with a keyhas_value
- match dictionary with a value
- Decorator
calling
- wrap a callable in a deferred object, for subsequent matching on calling behaviourraises
- Ensure that a deferred callable raises as expecteddescribed_as
- give the matcher a custom failure descriptionis_
- decorator to improve readability - see Syntactic sugar below
The arguments for many of these matchers accept not just a matching value, but another matcher, so matchers can be composed for greater flexibility. For example, only_contains(less_than(5))
will match any sequence where every item is less than 5.
Syntactic sugar
PyHamcrest strives to make your tests as readable as possible. For example, the is_
matcher is a wrapper that doesn't add any extra behavior to the underlying matcher. The following assertions are all equivalent:
assert_that(theBiscuit, equal_to(myBiscuit))
assert_that(theBiscuit, is_(equal_to(myBiscuit)))
assert_that(theBiscuit, is_(myBiscuit))
The last form is allowed since is_(value)
wraps most non-matcher arguments with equal_to
. But if the argument is a type, it is wrapped with instance_of
, so the following are also equivalent:
assert_that(theBiscuit, instance_of(Biscuit))
assert_that(theBiscuit, is_(instance_of(Biscuit)))
assert_that(theBiscuit, is_(Biscuit))
Note that PyHamcrest's ``is_`` matcher is unrelated to Python's ``is`` operator. The matcher for object identity is ``same_instance``.
Writing custom matchers
PyHamcrest comes bundled with lots of useful matchers, but you'll probably find that you need to create your own from time to time to fit your testing needs. This commonly occurs when you find a fragment of code that tests the same set of properties over and over again (and in different tests), and you want to bundle the fragment into a single assertion. By writing your own matcher you'll eliminate code duplication and make your tests more readable!
Let's write our own matcher for testing if a calendar date falls on a Saturday. This is the test we want to write:
def testDateIsOnASaturday(self):
d = datetime.date(2008, 4, 26)
assert_that(d, is_(on_a_saturday()))
And here's the implementation:
from hamcrest.core.base_matcher import BaseMatcher
from hamcrest.core.helpers.hasmethod import hasmethod
class IsGivenDayOfWeek(BaseMatcher):
def __init__(self, day):
self.day = day # Monday is 0, Sunday is 6
def _matches(self, item):
if not hasmethod(item, "weekday"):
return False
return item.weekday() == self.day
def describe_to(self, description):
day_as_string = [
"Monday",
"Tuesday",
"Wednesday",
"Thursday",
"Friday",
"Saturday",
"Sunday",
]
description.append_text("calendar date falling on ").append_text(
day_as_string[self.day]
)
def on_a_saturday():
return IsGivenDayOfWeek(5)
For our Matcher implementation we implement the _matches
method - which calls the weekday
method after confirming that the argument (which may not be a date) has such a method - and the describe_to
method - which is used to produce a failure message when a test fails. Here's an example of how the failure message looks:
assert_that(datetime.date(2008, 4, 6), is_(on_a_saturday()))
fails with the message:
AssertionError: Expected: is calendar date falling on Saturday got: <2008-04-06>
Let's say this matcher is saved in a module named isgivendayofweek
. We could use it in our test by importing the factory function on_a_saturday
:
from hamcrest import *
import unittest
from isgivendayofweek import on_a_saturday
class DateTest(unittest.TestCase):
def testDateIsOnASaturday(self):
d = datetime.date(2008, 4, 26)
assert_that(d, is_(on_a_saturday()))
if __name__ == "__main__":
unittest.main()
Even though the on_a_saturday
function creates a new matcher each time it is called, you should not assume this is the only usage pattern for your matcher. Therefore you should make sure your matcher is stateless, so a single instance can be reused between matches.