2、pytest中文文档--使用和调用

使用和调用

通过python -m pytest调用pytest

你可以通过python的解释器来执行测试：

python -m pytest [...]

但是，这和直接执行pytest [...]命令的效果几乎是一模一样的；

pytest执行结束时返回的状态码

pytest命令执行结束，可能会返回以下六种状态码：

• 0（OK）：所有收集到的用例测试通过
• 1（TESTS_FAILED）：有用例测试失败
• 2（INTERRUPTED）：用户打断测试执行
• 3（INTERNAL_ERROR）：测试执行的过程中，发生内部错误
• 4（USAGE_ERROR）：pytest命令使用错误
• 5（NO_TESTS_COLLECTED）：没有收集到测试用例

它们在枚举类_pytest.main.ExitCode中声明。并且，其作为公开API的一部分，能够直接引入和访问。

from pytest import ExitCode

获取帮助信息

pytest --version  # 查看版本号和pytest的引入路径
pytest -h  # 查看帮助信息

最多允许失败的测试用例数

当达到最大上限时，退出执行；如未配置，则没有上限；

pytest -x  # 遇到第一个失败时，退出执行
pytest --maxfail==2  # 遇到第二个失败时，退出执行

执行指定的测试用例

pytest支持多种方式来执行特定的测试用例：

执行指定模块中的测试用例：

pytest test_mod.py

执行指定目录下的测试用例：

pytest testing/

执行文件名、类名或者函数名中包含特定关键字的测试用例：

yaomengdeMacBook-Air:src yaomeng$ pytest -k "_class and not two"
============================================================ test session starts =============================================================
platform darwin -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: /Users/yaomeng/Private/Projects/pytest-chinese-doc/src
plugins: repeat-0.8.0
collected 5 items / 4 deselected / 1 selected                                                                                                

test_class.py .                                                                                                                        [100%]

====================================================== 1 passed, 4 deselected in 0.03s =======================================================

执行src/目录下，名字包含_class但不包含two的测试用例，即执行test_class.py模块中的test_one用例。

注意：python的关键字不可以应用在-k选项中，例如，class、def等。

执行指定nodeid的测试用例：

pytest为每一个收集到的测试用例指定一个唯一的nodeid。其由模块名加说明符构成，中间以::间隔。其中，说明符可以是类名、函数名以及由parametrize标记赋予的参数；

# test_nodeid.py

import pytest


def test_one():
    print('test_one')
    assert 1


class TestNodeId:
    def test_one(self):
        print('TestNodeId::test_one')
        assert 1

    @pytest.mark.parametrize('x,y', [(1, 1), (3, 4)])
    def test_two(self, x, y):
        print(f'TestNodeId::test_two::{x} == {y}')
        assert x == y

执行test_one用例：

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest -s test_nodeid.py::test_one
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src
collected 1 item

test_nodeid.py test_one
.

================================================== 1 passed in 0.02s ==================================================

执行TestNodeId中的test_one用例：

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest -s test_nodeid.py::TestNodeId::test_one
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src
collected 1 item

test_nodeid.py TestNodeId::test_one
.

================================================== 1 passed in 0.02s ==================================================

执行TestNodeId中的test_two用例，并指定参数x、y的值：

λ pytest -s test_nodeid.py::TestNodeId::test_two[3-4]
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src
collected 1 item

test_nodeid.py TestNodeId::test_two::3 == 4
F

====================================================== FAILURES ======================================================= ______________________________________________ TestNodeId.test_two[3-4] _______________________________________________

self = <test_nodeid.TestNodeId object at 0x00000152613C49B0>, x = 3, y = 4

    @pytest.mark.parametrize('x,y', [(1, 1), (3, 4)])
    def test_two(self, x, y):
        print(f'TestNodeId::test_two::{x} == {y}')
>       assert x == y
E       assert 3 == 4

test_nodeid.py:37: AssertionError
================================================== 1 failed in 0.05s ==================================================

这里对参数x、y赋值的形式是[3-4]，中间以-间隔；单个或多个参数的赋值形式以此类比；并且，只能为[1-1]或者[3-4]，其它的会报错；

执行指定标记的用例

pytest -m slow

执行所有用@pytest.mark.slow装饰器标记的测试用例。

执行指定包中的测试用例

/d/Personal Files/Python/pytest-chinese-doc (5.1.2)
λ pytest --pyargs src

执行src包中所有的测试用例。

注意：src文件夹中要有一个__init__.py文件。

修改回溯信息的输出模式

pytest回溯信息的输出一共有六种模式：auto/long/short/line/native/no，用--tb选项指定；

pytest -l, --showlocals         # 打印本地变量
pytest --tb=auto                # 默认模式
pytest --tb=long                # 尽可能详细的输出
pytest --tb=short               # 更简短的输出
pytest --tb=line                # 每个失败信息总结在一行中
pytest --tb=native              # python的标准输出
pytest --tb=no                  # 不打印失败信息

--full-trace是一种比--tb=long更详细的输出模式。它甚至能观察到用户打断执行（Ctrl+C）时的回溯信息，而上述六种模式默认是不输出信息的。

总结报告

-r选项可以在执行结束后，打印一个简短的总结报告。在执行的测试用例很多时，可以让你对结果有个清晰的了解：

# test_report.py

import pytest


@pytest.fixture
def error_fixture():
    assert 0


def test_ok():
    print("ok")


def test_fail():
    assert 0


def test_error(error_fixture):
    pass


def test_skip():
    pytest.skip("skipping this test")


def test_xfail():
    pytest.xfail("xfailing this test")


@pytest.mark.xfail(reason="always xfail")
def test_xpass():
    pass

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest -ra test_report.py
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src
collected 6 items

test_report.py .FEsxX                                                                                            [100%]

======================================================= ERRORS ======================================================== ____________________________________________ ERROR at setup of test_error _____________________________________________

    @pytest.fixture
    def error_fixture():
>       assert 0
E       assert 0

test_report.py:27: AssertionError
====================================================== FAILURES ======================================================= ______________________________________________________ test_fail ______________________________________________________

    def test_fail():
>       assert 0
E       assert 0

test_report.py:35: AssertionError
=============================================== short test summary info ===============================================
SKIPPED [1] D:\Personal Files\Python\pytest-chinese-doc\src\test_report.py:44: skipping this test
XFAIL test_report.py::test_xfail
  reason: xfailing this test
XPASS test_report.py::test_xpass always xfail
ERROR test_report.py::test_error - assert 0
FAILED test_report.py::test_fail - assert 0
======================== 1 failed, 1 passed, 1 skipped, 1 xfailed, 1 xpassed, 1 error in 0.10s ========================

-r选项后面要紧接这一个参数，用于过滤显示测试用例的结果。a表示除了成功的其它所有的测试用例。

以下是所有有效的字符参数：

• f：失败的
• E：出错的
• s：跳过执行的
• x：跳过执行，并标记为xfailed的
• X：跳过执行，并标记为xpassed的
• p：测试通过的
• P：测试通过，并且有输出信息的；即用例中有print等
• a：除了测试通过的，其他所有的；即除了p和P的
• A：所有的

上述字符参数可以叠加使用，例如：我们期望过滤出失败的和未执行的：

pytest -rfs

失败时加载PDB(Python Debugger)环境

PDB是python内建的诊断器，pytest允许通过以下命令在执行失败时进入这个诊断器模式：

pytest --pdb

pytest会在测试用例失败（或者Ctrl+C）时，调用这个诊断器：

luyao@NJ-LUYAO-T460 /d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest --pdb test_class.py
================================================= test session starts ================================================= 
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src
collected 2 items

test_class.py .F
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> traceback >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

self = <src.test_class.TestClass object at 0x00000269030ECB70>

    def test_two(self):
        x = 'hello'
>       assert hasattr(x, 'check')
E       AssertionError: assert False
E        +  where False = hasattr('hello', 'check')

test_class.py:30: AssertionError
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> entering PDB >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> PDB post_mortem (IO-capturing turned off) >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> 
> d:\personal files\python\pytest-chinese-doc\src\test_class.py(30)test_two()
-> assert hasattr(x, 'check')
(Pdb) x
'hello'
(Pdb)

此时，你可以访问测试用例的本地变量x;

并且，失败的信息存储在sys.last_value, sys.last_type, sys.last_traceback变量中，你可以在交互环境中访问它们：

(Pdb) import sys
(Pdb) sys.last_value
AssertionError("assert False\n +  where False = hasattr('hello', 'check')")
(Pdb) sys.last_type
<class 'AssertionError'>
(Pdb) sys.last_traceback
<traceback object at 0x00000269030F0908>
(Pdb)

开始执行时就加载PDB环境

通过以下命令，pytest允许你在每个测试用例执行时，就加载PDB环境：

pytest --trace

设置断点

在测试用例代码中添加import pdb;pdb.set_trace()，当其被调用时，pytest会停止这条用例的输出：

• 其他用例不受影响；
• 通过continue命令，退出PDB环境，并继续执行用例；

使用内置的中断函数

python 3.7介绍了一个内置breakpoint()函数。pytest可以在以下场景中支持使用：

• 当breakpoint()被调用，并且PYTHONBREAKPOINT未设置时，pytest会使用内部自定义的PDB代替系统的PDB；
• 测试执行结束时，自动切回系统自带的PDB；
• 当加上--pdb选项时，breakpoint()和测试发生错误时，都会调用内部自定义的PDB；
• --pdbcls选项允许指定一个用户自定义的PDB类；

分析测试执行时间

获取执行最慢的10个测试用例：

pytest --durations=10

默认情况下，pytest不会显示执行时间<0.01s的测试用例，可以使用-vv选项查看它们；

错误句柄

5.0版本新增特性

在测试执行中发生段错误或者超时的情况下，faulthandler标准模块可以转储python的回溯信息；

它在pytest的执行中默认使能，使用-p no:faulthandler选项可以关闭它；

同样，faulthandler_timeout=X配置项，可用于当测试用例的完成时间超过X秒时，转储所有线程的python回溯信息；

# pytest.ini

[pytest]
faulthandler_timeout=5

配置测试执行的超时时间是5秒；

# test_fault_handler.py 

import time


def test_fault_handler():
    time.sleep(10)
    assert 1

测试用例中添加等待10秒的操作；

yaomengdeMacBook-Air:src yaomeng$ pytest test_fault_handler.py 
============================================================ test session starts =============================================================
platform darwin -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: /Users/yaomeng/Private/Projects/pytest-chinese-doc/src, inifile: pytest.ini
plugins: repeat-0.8.0
collected 1 item                                                                                                                             

test_fault_handler.py Timeout (0:00:05)!
Thread 0x000000010d5895c0 (most recent call first):
  File "/Users/yaomeng/Private/Projects/pytest-chinese-doc/src/test_fault_handler.py", line 26 in test_fault_handler
  File "/usr/local/lib/python3.7/site-packages/_pytest/python.py", line 170 in pytest_pyfunc_call
  File "/usr/local/lib/python3.7/site-packages/pluggy/callers.py", line 187 in _multicall
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 81 in <lambda>
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 87 in _hookexec
  File "/usr/local/lib/python3.7/site-packages/pluggy/hooks.py", line 289 in __call__
  File "/usr/local/lib/python3.7/site-packages/_pytest/python.py", line 1423 in runtest
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 117 in pytest_runtest_call
  File "/usr/local/lib/python3.7/site-packages/pluggy/callers.py", line 187 in _multicall
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 81 in <lambda>
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 87 in _hookexec
  File "/usr/local/lib/python3.7/site-packages/pluggy/hooks.py", line 289 in __call__
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 192 in <lambda>
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 220 in from_call
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 192 in call_runtest_hook
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 167 in call_and_report
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 87 in runtestprotocol
  File "/usr/local/lib/python3.7/site-packages/_pytest/runner.py", line 72 in pytest_runtest_protocol
  File "/usr/local/lib/python3.7/site-packages/pluggy/callers.py", line 187 in _multicall
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 81 in <lambda>
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 87 in _hookexec
  File "/usr/local/lib/python3.7/site-packages/pluggy/hooks.py", line 289 in __call__
  File "/usr/local/lib/python3.7/site-packages/_pytest/main.py", line 256 in pytest_runtestloop
  File "/usr/local/lib/python3.7/site-packages/pluggy/callers.py", line 187 in _multicall
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 81 in <lambda>
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 87 in _hookexec
  File "/usr/local/lib/python3.7/site-packages/pluggy/hooks.py", line 289 in __call__
  File "/usr/local/lib/python3.7/site-packages/_pytest/main.py", line 235 in _main
  File "/usr/local/lib/python3.7/site-packages/_pytest/main.py", line 191 in wrap_session
  File "/usr/local/lib/python3.7/site-packages/_pytest/main.py", line 228 in pytest_cmdline_main
  File "/usr/local/lib/python3.7/site-packages/pluggy/callers.py", line 187 in _multicall
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 81 in <lambda>
  File "/usr/local/lib/python3.7/site-packages/pluggy/manager.py", line 87 in _hookexec
  File "/usr/local/lib/python3.7/site-packages/pluggy/hooks.py", line 289 in __call__
  File "/usr/local/lib/python3.7/site-packages/_pytest/config/__init__.py", line 78 in main
  File "/usr/local/bin/pytest", line 10 in <module>
.                                                                                                                [100%]

============================================================= 1 passed in 10.02s =============================================================

在执行刚超过5秒的时候会打印出回溯信息。但不会中断测试的执行；

yaomengdeMacBook-Air:src yaomeng$ pytest -p no:faulthandler test_fault_handler.py 
============================================================ test session starts =============================================================
platform darwin -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: /Users/yaomeng/Private/Projects/pytest-chinese-doc/src, inifile: pytest.ini
plugins: repeat-0.8.0
collected 1 item 

test_fault_handler.py .                                                                                            [100%]

============================================================= 1 passed in 10.02s =============================================================

当去使能faulthandler的时，超时并不会触发回溯信息的打印；

注意：这个功能是从pytest-faulthandler插件合并而来的，但是有两点不同：

• 去使能时，使用-p no:faulthandler代替原来的--no-faulthandler;
• 使用faulthandler_timeout配置项代替--faulthandler-timeout命令行选项来配置超时时间。当然，你也可以使用-o faulthandler_timeout=X在命令行配置；

创建JUnitXML格式的结果报告

使用如下命令，可以创建一个能被Jenkins或者其他CI工具读取的测试结果文件：

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_class.py --junitxml=../report/test_class.xml

它会在指定目录下创建一个XML文件：

<!--test_class.xml-->
<?xml version="1.0" encoding="utf-8"?>
<testsuites>
    <testsuite errors="0" failures="1" hostname="NJ-LUYAO-T460" name="pytest" skipped="0" tests="2" time="0.058"
        timestamp="2019-09-10T09:53:07.554001">
        <testcase classname="test_class.TestClass" file="test_class.py" line="23" name="test_one" time="0.001">
        </testcase>
        <testcase classname="test_class.TestClass" file="test_class.py" line="27" name="test_two" time="0.002">
            <failure message="AssertionError: assert False
 +  where False = hasattr(&apos;hello&apos;, &apos;check&apos;)">self = &lt;src.test_class.TestClass object at
                0x000001773E688C50&gt;

                def test_two(self):
                x = &apos;hello&apos;
                &gt; assert hasattr(x, &apos;check&apos;)
                E AssertionError: assert False
                E + where False = hasattr(&apos;hello&apos;, &apos;check&apos;)

                test_class.py:30: AssertionError</failure>
        </testcase>
    </testsuite>
</testsuites>

你可以在pytest.ini文件中设置junit_suite_name，来修改xml文件中testsuite根节点的name信息：

# pytest.ini

[pytest]
junit_suite_name = pytest_chinese_doc

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
pytest test_class.py test_sample.py --junitxml=../report/test_mysuitename.xml

<!--test_mysuitename.xml-->
<?xml version="1.0" encoding="utf-8"?>
<testsuites>
    <testsuite errors="0" failures="2" hostname="NJ-LUYAO-T460" name="pytest_chinese_doc" skipped="0" tests="3"
        time="0.070" timestamp="2019-09-10T10:07:29.078541">
        <testcase classname="test_class.TestClass" file="test_class.py" line="23" name="test_one" time="0.000">
        </testcase>
        <testcase classname="test_class.TestClass" file="test_class.py" line="27" name="test_two" time="0.001">
            <failure message="AssertionError: assert False
 +  where False = hasattr(&apos;hello&apos;, &apos;check&apos;)">self = &lt;src.test_class.TestClass object at
                0x000001C7F1155550&gt;

                def test_two(self):
                x = &apos;hello&apos;
                &gt; assert hasattr(x, &apos;check&apos;)
                E AssertionError: assert False
                E + where False = hasattr(&apos;hello&apos;, &apos;check&apos;)

                test_class.py:30: AssertionError</failure>
        </testcase>
        <testcase classname="test_sample" file="test_sample.py" line="26" name="test_sample" time="0.001">
            <failure message="assert 4 == 5
 +  where 4 = func(3)">def test_sample():
                &gt; assert func(3) == 5
                E assert 4 == 5
                E + where 4 = func(3)

                test_sample.py:28: AssertionError</failure>
        </testcase>
    </testsuite>
</testsuites>

可以看到，<testsuite errors="0" failures="2" hostname="NJ-LUYAO-T460" name="pytest_chinese_doc"的name信息已经从默认的pytest变成我们期望的pytest_chinese_doc；

注意：这是4.0版本的新增配置项；

JUnit XML规定time属性应该表明测试用例执行的全部耗时，包含setup和teardown中的操作，这也是pytest的默认行为；如果你只想记录测试用例执行的时间，只需要做如下配置：

# pytest.ini

junit_duration_report = call

record_property

如果你想在一个测试用例的报告中添加一些额外的属性节点，可以使用record_property fixture：

# test_record_property.py

def test_record_property(record_property):
    record_property('example_key', 1)
    assert 1

我们期望添加一个example_key属性节点；

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_record_property.py --junitxml=../report/test_record_property.xml

<!--test_record_property.xml-->
<?xml version="1.0" encoding="utf-8"?>
<testsuites>
    <testsuite errors="0" failures="0" hostname="NJ-LUYAO-T460" name="pytest_chinese_doc" skipped="0" tests="1"
        time="0.025" timestamp="2019-09-10T10:24:20.796619">
        <testcase classname="test_record_property" file="test_record_property.py" line="22" name="test_record_property"
            time="0.001">
            <properties>
                <property name="example_key" value="1" />
            </properties>
        </testcase>
    </testsuite>
</testsuites>

这样，我们就为test_record_property用例添加了一个example_key="1"的新属性；

或者，你可以将这种功能集成到一个自定义的标记中，例如：允许为测试用例添加一个测试ID：

# conftest.py

def pytest_collection_modifyitems(session, config, items):
    for item in items:
        for marker in item.iter_markers(name='test_id'):
            test_id = marker.args[0]
            item.user_properties.append(('test_id', test_id))

为以下用例标记一个测试ID：

# test_testid.py
import pytest


@pytest.mark.test_id(10010)
def test_record_property(record_property):
    record_property('example_key', 1)
    assert 1

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_testid.py --junitxml=../report/test_testid.xml
================================================= test session starts ================================================= 
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src, inifile: pytest.ini
collected 1 item

test_testid.py .                                                                                                 [100%]

================================================== warnings summary =================================================== 
d:\program files\python\python37\lib\site-packages\_pytest\mark\structures.py:324
  d:\program files\python\python37\lib\site-packages\_pytest\mark\structures.py:324: PytestUnknownMarkWarning: Unknown pytest.mark.test_id - is this a typo?  You can register custom marks to avoid this warning - for details, see https://docs.pytest.org/en/latest/mark.html
    PytestUnknownMarkWarning,

-- Docs: https://docs.pytest.org/en/latest/warnings.html
----------- generated xml file: D:\Personal Files\Python\pytest-chinese-doc\report\test_record_property.xml ----------- ============================================ 1 passed, 1 warnings in 0.03s ============================================

这里有个告警，提示我们没有注册自定义的标记，但并不影响执行。我们只需要在pytest.ini中注册一下就可以了：

# pytest.ini

[pytest]
markers =
    test_id: 为测试用例添加ID

<!--test_testid.xml-->
<?xml version="1.0" encoding="utf-8"?>
<testsuites>
    <testsuite errors="0" failures="0" hostname="NJ-LUYAO-T460" name="pytest_chinese_doc" skipped="0" tests="1"
        time="0.025" timestamp="2019-09-10T10:46:50.694771">
        <testcase classname="test_testid" file="test_testid.py" line="24" name="test_record_property" time="0.002">
            <properties>
                <property name="test_id" value="10010" />
                <property name="example_key" value="1" />
            </properties>
        </testcase>
    </testsuite>
</testsuites>

可以看到，我们又为test_record_property用例添加了一个test_id="10010"的新属性；

注意：这样使用可能会使XML文件不符合最新的JUnitXML的模式检查规则，导致在某些CI工具可能会发生未知的错误；

record_xml_attribute

如果你想在testcase节点中添加一个新的属性，可以使用record_xml_attribute fixture。如果这个属性之前已经存在了，那么就会用新值覆盖掉原来的值：

# test_record_attribute.py

def test_record_property(record_xml_attribute):
    record_xml_attribute('example_key', 1)
    record_xml_attribute('classname', 'overwrite_classname')
    assert 1

添加一个新的example_key属性，并修改classname属性；

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_record_attribute.py --junitxml=../report/test_record_attribute.xml
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src, inifile: pytest.ini
collected 1 item

test_record_attribute.py .                                                                                       [100%]

================================================== warnings summary =================================================== test_record_attribute.py::test_record_attribute
  test_record_attribute.py:23: PytestExperimentalApiWarning: record_xml_attribute is an experimental feature
    def test_record_attribute(record_xml_attribute):

-- Docs: https://docs.pytest.org/en/latest/warnings.html
---------- generated xml file: D:\Personal Files\Python\pytest-chinese-doc\report\test_record_attribute.xml ----------- ============================================ 1 passed, 1 warnings in 0.03s ============================================

<!--test_record_attribute.xml-->
<?xml version="1.0" encoding="utf-8"?>
<testsuites>
    <testsuite errors="0" failures="0" hostname="NJ-LUYAO-T460" name="pytest_chinese_doc" skipped="0" tests="1"
        time="0.026" timestamp="2019-09-10T11:00:57.818851">
        <testcase classname="overwrite_classname" example_key="1" file="test_record_attribute.py" line="22"
            name="test_record_attribute" time="0.002"></testcase>
    </testsuite>
</testsuites>

可以看到，<testcase classname="overwrite_classname" example_key="1"节点的classname被替换，并且新增了example_key属性；

注意：record_xml_attribute目前是一个实验性的功能，未来可能被更强大的API所替代，但功能本身会被保留。

注意：这样使用可能会使XML文件不符合最新的JUnitXML的模式检查规则，导致在某些CI工具可能会发生未知的错误；

record_testsuite_property

4.5版本新增功能

如果你想在testsuite级别上添加属性节点，并且希望其作用于所有的测试用例，那么你可以在session作用域中使用record_testsuite_property fixture：

# conftest.py

@pytest.fixture(scope="session", autouse=True)
def log_global_env_facts(record_testsuite_property):
    record_testsuite_property("ARCH", "PPC")
    record_testsuite_property("STORAGE_TYPE", "CEPH")

record_testsuite_property接收两个参数name和value以构成<property>标签，其中，name必须为字符串，value会转换为字符串并进行XML转义；

# test_record_testsuite_property.py

class TestMe:
    def test_foo(self):
        assert True

log_global_env_facts为session级别的fixture，并且，autouse为True，所以并不需要事先引入；

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_record_testsuite_property.py --junitxml=../report/test_record_testsuite_property.xml

<!--test_record_testsuite_property.xml-->
<?xml version="1.0" encoding="utf-8"?>
<testsuites>
    <testsuite errors="0" failures="0" hostname="NJ-LUYAO-T460" name="pytest_chinese_doc" skipped="0" tests="1"
        time="0.029" timestamp="2019-09-10T11:20:46.123978">
        <properties>
            <property name="ARCH" value="PPC" />
            <property name="STORAGE_TYPE" value="CEPH" />
        </properties>
        <testcase classname="test_record_testsuite_property.TestMe" file="test_record_testsuite_property.py" line="23"
            name="test_foo" time="0.001"></testcase>
    </testsuite>
</testsuites>

可以看到，在testsuite节点中，新增了一个properties属性节点，包含所有新增的属性，而且，它和testcase节点是兄弟节点；

注意：这样生成的XML文件是符合最新的xunit标准的，这点和record_property、record_xml_attribute正好相反；

为测试报告提供在线的pastebin展示

目前，只实现了在http://bpaste.net上的展示功能；

为每一个失败用例创建URL

luyao@NJ-LUYAO-T460 /d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_class.py --pastebin=failed
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0 -- d:\program files\python\python37\python.exe
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src, inifile: pytest.ini
collected 2 items

test_class.py .F                                                                                                 [100%]

====================================================== FAILURES ======================================================= 
_________________________________________________ TestClass.test_two __________________________________________________

self = <src.test_class.TestClass object at 0x000001AB97F29208>

    def test_two(self):
        x = 'hello'
>       assert hasattr(x, 'check')
E       AssertionError: assert False
E        +  where False = hasattr('hello', 'check')

test_class.py:30: AssertionError
======================================== Sending information to Paste Service ========================================= 
test_class.py:30: AssertionError --> https://bpaste.net/show/HGeB
============================================= 1 failed, 1 passed in 5.64s =============================================

在浏览器中访问https://bpaste.net/show/HGeB，就能查看到失败的信息了。

为所有的测试用例创建URL

luyao@NJ-LUYAO-T460 /d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ pytest test_class.py --pastebin=all
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0 -- d:\program files\python\python37\python.exe
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src, inifile: pytest.ini
collected 2 items

test_class.py .F                                                                                                 [100%]

====================================================== FAILURES ======================================================= 
_________________________________________________ TestClass.test_two __________________________________________________

self = <src.test_class.TestClass object at 0x00000218030F91D0>

    def test_two(self):
        x = 'hello'
>       assert hasattr(x, 'check')
E       AssertionError: assert False
E        +  where False = hasattr('hello', 'check')

test_class.py:30: AssertionError
============================================= 1 failed, 1 passed in 0.06s ============================================= 
======================================== Sending information to Paste Service =========================================
pastebin session-log: https://bpaste.net/show/XdVT

在浏览器中访问https://bpaste.net/show/XdVT，就能查看到所有用例的执行信息了。

尽量早的加载插件

你可以在命令行中使用-p选项，来尽量早的加载某一个插件：

pytest -p mypluginmodule

-p选项接收一个name参数，这个参数可以为：

• 一个完整的本地插件引入，例如：myproject.plugins，其必须是可以import的。
• 一个公共插件的名称，这是其注册时在setuptools中赋予的名字，例如：pytest-cov

去使能插件

你可以在命令行中使用-p结合no:，来去使能一个插件的加载，例如：

pytest -p no:doctest

在python代码中调用pytest

你可以在代码中直接调用pytest：

pytest.main()

这和你在命令行中执行pytest是一样的，不过它不会触发SystemExit，而是返回exitcode：

# invoke_with_main.py

import time


def test_one():
    time.sleep(10)
    assert True


if __name__ == '__main__':
    import pytest
    ret = pytest.main([__file__])
    print(ret == pytest.ExitCode.INTERRUPTED)

用例中有等待10秒的操作，当我们在这期间，打断执行（Ctr+C），我们期望pytest.main()返回的状态码是INTERRUPTED，也就是2；

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ python invoke_with_main.py
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src, inifile: pytest.ini
collected 1 item

invoke_with_main.py

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! KeyboardInterrupt !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
d:\personal files\python\pytest-chinese-doc\src\invoke_with_main.py:26: KeyboardInterrupt
(to show a full traceback on KeyboardInterrupt use --full-trace)
================================================ no tests ran in 2.11s ================================================
True

可以看到，pytest.main()确实返回了INTERRUPTED的状态码；

你也可以传递选项和参数：

pytest.main(["-x", "mytestdir"])

同样，也可以为pytest.main()指定一个自定义的插件：

# invoke_with_plugins.py

class MyPlugin:
    def pytest_sessionfinish(self):
        print('*** test run reporting finishing')


if __name__ == '__main__':
    import pytest
    pytest.main(['test_class.py'], plugins=[MyPlugin()])

执行这个文件，会发现插件正常使能，并且它的钩子方法也被调用：

/d/Personal Files/Python/pytest-chinese-doc/src (5.1.2)
λ python invoke_with_plugins.py
================================================= test session starts =================================================
platform win32 -- Python 3.7.3, pytest-5.1.2, py-1.8.0, pluggy-0.12.0
rootdir: D:\Personal Files\Python\pytest-chinese-doc\src, inifile: pytest.ini
collected 2 items

test_class.py .F                                                                                                 [100%]
*** test run reporting finishing


====================================================== FAILURES ======================================================= 
_________________________________________________ TestClass.test_two __________________________________________________

self = <src.test_class.TestClass object at 0x00000236949A20B8>

    def test_two(self):
        x = 'hello'
>       assert hasattr(x, 'check')
E       AssertionError: assert False
E        +  where False = hasattr('hello', 'check')

test_class.py:30: AssertionError
============================================= 1 failed, 1 passed in 0.06s =============================================

注意：调用pytest.main()会引入你的测试文件以及其引用的所有模块。由于python引入机制的缓存特性，当这些文件发生变化时，后续再调用pytest.main()（在同一个程序执行过程中）时，并不会响应这些文件的变化。基于这个原因，我们不推荐在同一个程序中多次调用pytest.main()(例如：为了重新执行测试；如果你确实有这个需求，或许可以考虑pytest-repeat插件)；