document and refine/rename item **runtest** hook invocations

capture output separately for fixture and actual test run

--HG--
branch : trunk
This commit is contained in:
holger krekel 2009-06-09 16:08:34 +02:00
parent d16688a1e6
commit ed216e77d0
13 changed files with 236 additions and 127 deletions

View File

@ -32,10 +32,6 @@ or class with a leading ``Test`` name is collected.
.. _`collection process`: ext.html#collection-process
Rapidly write integration, functional, unit tests
===================================================
XXX
funcargs and xUnit style setups
===================================================
@ -95,17 +91,20 @@ debug with the ``print`` statement
----------------------------------
By default, ``py.test`` catches text written to stdout/stderr during
the execution of each individual test. This output will only be
the execution of each individual test. This output will only be
displayed however if the test fails; you will not see it
otherwise. This allows you to put debugging print statements in your
code without being overwhelmed by all the output that might be
generated by tests that do not fail.
Each failing test that produced output during the running of the test
will have its output displayed in the ``recorded stdout`` section.
function will have its output displayed in the ``recorded stdout`` section.
During Setup and Teardown ("Fixture") capturing is performed separately so
that you will only see this output if the actual fixture functions fail.
The catching of stdout/stderr output can be disabled using the
``--nocapture`` option to the ``py.test`` tool. Any output will
``--nocapture`` or ``-s`` option to the ``py.test`` tool. Any output will
in this case be displayed as soon as it is generated.
test execution order

View File

@ -2,16 +2,17 @@
hooks and plugins
==========================
py.test implements much of its functionality by calling so called
**hooks**. A hook is a function with a ``pytest_`` prefix and a list of
named arguments. Hook functions are usually defined in plugins.
A plugin is a module or package that makes hook functions available.
py.test implements much of its functionality by calling **hooks**.
A hook is a function with a ``pytest_`` prefixed name. Hook functions
are usually defined in plugins. A plugin is a regular python module or
package that makes hook functions available.
When loading a plugin module (which needs to have a ``pytest_`` prefix as well)
py.test performs strict checking on the function signature. Function
and argument names need to match exactly the `original definition of the hook`_.
This allows for early mismatch reporting and minimizes version incompatibilites.
.. _`original definition of the hook`: http://bitbucket.org/hpk42/py-trunk/src/tip/py/test/plugin/api.py
Loading plugins and specifying dependencies
============================================
@ -42,7 +43,13 @@ Included plugins
================
You can find the source code of all default plugins in
http://bitbucket.org/hpk42/py-trunk/src/tip/py/test/plugin/
http://bitbucket.org/hpk42/py-trunk/src/tip/py/test/plugin/
Additionally you can check out some more contributed plugins here
http://bitbucket.org/hpk42/py-trunk/src/tip/contrib/
Overview on available hooks
====================================
@ -50,16 +57,54 @@ Overview on available hooks
"runtest" hooks
-------------------
A number of hooks allow interacting with the running of a test.
A test can be many things - for example a python, javascript
or prolog test function or a doctest. The following hooks are
usually invoked on running a test item::
Each test item is usually executed by calling the following three hooks::
pytest_runtest_protocol(item) -> True # and invokes:
pytest_runtest_setup(item) -> None
pytest_runtest_call(item) -> (excinfo, when, outerr)
pytest_runtest_makereport(item, excinfo, when, outerr) -> report
pytest_runtest_logreport(report) -> None
pytest_runtest_teardown(item) -> None
pytest_runtest_setup(item)
pytest_runtest_call(item)
pytest_runtest_teardown(item)
For each of the three invocations a `call object`_ encapsulates
information about the outcome of the call and is subsequently used
to make a report object::
report = hook.pytest_runtest_makereport(item, call)
For example, the `pytest_pdb plugin`_ uses this hook to activate
interactive debugging on failures when ``--pdb`` is specified on the
command line.
Usually three reports will be generated for a single test item. However,
if the ``pytest_runtest_setup`` fails no call or teardown hooks
will be called and only one report will be created.
Each of the up to three reports is eventually fed to the logreport hook::
pytest_runtest_logreport(report)
A ``report`` object contains status and reporting information::
report.longrepr = string/lines/object to print
report.when = "setup", "call" or "teardown"
report.shortrepr = letter for progress-report
report.passed = True or False
report.failed = True or False
report.skipped = True or False
The `pytest_terminal plugin`_ uses this hook to print information
about a test run.
The protocol described here is implemented via this hook::
pytest_runtest_protocol(item) -> True
.. _`call object`:
The call object contains information about a performed call::
call.excinfo = ExceptionInfo object or None
call.when = "setup", "call" or "teardown"
call.outerr = None or tuple of strings representing captured stdout/stderr
.. _`pytest_pdb plugin`: http://bitbucket.org/hpk42/py-trunk/src/tip/py/test/plugin/pytest_pdb.py
.. _`pytest_terminal plugin`: http://bitbucket.org/hpk42/py-trunk/src/tip/py/test/plugin/pytest_terminal.py

View File

@ -238,9 +238,9 @@ class DSession(Session):
self.node2pending[node].remove(item)
def handle_crashitem(self, item, node):
longrepr = "!!! Node %r crashed during running of test %r" %(node, item)
rep = item.config.hook.pytest_runtest_makereport(
item=item, when="???", excinfo=longrepr, outerr=None)
runner = item.config.pluginmanager.getplugin("runner")
info = "!!! Node %r crashed during running of test %r" %(node, item)
rep = runner.ItemTestReport(item=item, excinfo=info, when="???")
rep.node = node
self.config.hook.pytest_runtest_logreport(rep=rep)

View File

@ -5,8 +5,9 @@ import py
XSpec = py.execnet.XSpec
def run(item, node, excinfo=None):
rep = item.config.hook.pytest_runtest_makereport(
item=item, excinfo=excinfo, when="call", outerr=("", ""))
runner = item.config.pluginmanager.getplugin("runner")
rep = runner.ItemTestReport(item=item,
excinfo=excinfo, when="call", outerr=("", ""))
rep.node = node
return rep

View File

@ -72,18 +72,14 @@ class PluginHooks:
# runtest related hooks
# ------------------------------------------------------------------------------
def pytest_runtest_setup(self, item):
""" called before pytest_runtest(). """
def pytest_runtest_teardown(self, item):
""" called after pytest_runtest_call. """
""" called before pytest_runtest_call(). """
def pytest_runtest_call(self, item):
""" called after pytest_runtest_call. """
""" execute test item. """
def pytest_runtest_teardown(self, item):
""" called after pytest_runtest_call(). """
def pytest_runtest_makereport(self, item, excinfo, when, outerr):
""" make ItemTestReport for the specified test outcome. """
pytest_runtest_makereport.firstresult = True
def pytest_runtest_protocol(self, item):
""" run given test item and return test report. """
pytest_runtest_protocol.firstresult = True
@ -92,6 +88,9 @@ class PluginHooks:
""" return True if we consumed/did the call to the python function item. """
pytest_pyfunc_call.firstresult = True
def pytest_runtest_makereport(self, item, call):
""" make ItemTestReport for the specified test outcome. """
pytest_runtest_makereport.firstresult = True
def pytest_runtest_logreport(self, rep):
""" process item test report. """

View File

@ -2,8 +2,9 @@
import py
def pytest_pyfunc_call(pyfuncitem, args, kwargs):
pyfuncitem.obj(*args, **kwargs)
def pytest_pyfunc_call(__call__, pyfuncitem, args, kwargs):
if not __call__.execute(firstresult=True):
pyfuncitem.obj(*args, **kwargs)
def pytest_collect_file(path, parent):
ext = path.ext

View File

@ -22,16 +22,24 @@ def pytest_funcarg__capfd(request):
request.addfinalizer(capture.finalize)
return capture
def pytest_pyfunc_call(pyfuncitem, args, kwargs):
for funcarg, value in kwargs.items():
if funcarg == "capsys" or funcarg == "capfd":
value.reset()
class Capture:
_capture = None
def __init__(self, captureclass):
self._captureclass = captureclass
self._capture = self._captureclass()
def finalize(self):
self._capture.reset()
if self._capture:
self._capture.reset()
def reset(self):
res = self._capture.reset()
res = None
if self._capture:
res = self._capture.reset()
self._capture = self._captureclass()
return res

View File

@ -22,12 +22,12 @@ def pytest_configure(config):
config.pluginmanager.register(PdbInvoke())
class PdbInvoke:
def pytest_runtest_makereport(self, item, excinfo, when, outerr):
if excinfo and not excinfo.errisinstance(Skipped):
def pytest_runtest_makereport(self, item, call):
if call.excinfo and not call.excinfo.errisinstance(Skipped):
tw = py.io.TerminalWriter()
repr = excinfo.getrepr()
repr = call.excinfo.getrepr()
repr.toterminal(tw)
post_mortem(excinfo._excinfo[2])
post_mortem(call.excinfo._excinfo[2])
class Pdb(py.std.pdb.Pdb):
def do_list(self, arg):

View File

@ -269,8 +269,12 @@ class TmpTestdir:
p1 = py.path.local("stdout")
p2 = py.path.local("stderr")
print "running", cmdargs, "curdir=", py.path.local()
popen = self.popen(cmdargs, stdout=p1.open("w"), stderr=p2.open("w"))
f1 = p1.open("w")
f2 = p2.open("w")
popen = self.popen(cmdargs, stdout=f1, stderr=f2, close_fds=True)
ret = popen.wait()
f1.close()
f2.close()
out, err = p1.readlines(cr=0), p2.readlines(cr=0)
if err:
for line in err:
@ -356,7 +360,8 @@ class ReportRecorder(object):
failed = []
for rep in self.getreports("pytest_runtest_logreport"):
if rep.passed:
passed.append(rep)
if rep.when == "call":
passed.append(rep)
elif rep.skipped:
skipped.append(rep)
elif rep.failed:
@ -384,19 +389,25 @@ def test_reportrecorder(testdir):
recorder = testdir.getreportrecorder(registry)
assert not recorder.getfailures()
item = testdir.getitem("def test_func(): pass")
rep = item.config.hook.pytest_runtest_makereport(
item=item, excinfo=None, when="call", outerr=None)
class rep:
excinfo = None
passed = False
failed = True
skipped = False
when = "call"
rep.passed = False
rep.failed = True
recorder.hook.pytest_runtest_logreport(rep=rep)
failures = recorder.getfailures()
assert failures == [rep]
failures = recorder.getfailures()
assert failures == [rep]
rep = item.config.hook.pytest_runtest_makereport(
item=item, excinfo=None, when="call", outerr=None)
class rep:
excinfo = None
passed = False
failed = False
skipped = True
when = "call"
rep.passed = False
rep.skipped = True
recorder.hook.pytest_runtest_logreport(rep=rep)

View File

@ -12,7 +12,6 @@ from py.__.test.outcome import Skipped
#
# pytest plugin hooks
#
def pytest_addoption(parser):
group = parser.getgroup("general")
@ -38,13 +37,22 @@ def pytest_make_collect_report(collector):
return report
def pytest_runtest_protocol(item):
if item.config.option.boxed:
report = forked_run_report(item)
if item.config.getvalue("boxed"):
reports = forked_run_report(item)
for rep in reports:
item.config.hook.pytest_runtest_logreport(rep=rep)
else:
report = basic_run_report(item)
item.config.hook.pytest_runtest_logreport(rep=report)
runtestprotocol(item)
return True
def runtestprotocol(item, log=True):
rep = call_and_report(item, "setup", log)
reports = [rep]
if rep.passed:
reports.append(call_and_report(item, "call", log))
reports.append(call_and_report(item, "teardown", log))
return reports
def pytest_runtest_setup(item):
item.config._setupstate.prepare(item)
@ -52,44 +60,45 @@ def pytest_runtest_call(item):
if not item._deprecated_testexecution():
item.runtest()
def pytest_runtest_makereport(item, excinfo, when, outerr):
return ItemTestReport(item, excinfo, when, outerr)
def pytest_runtest_makereport(item, call):
if isinstance(call, str):
# crashed item
return ItemTestReport(item, excinfo=call, when="???")
else:
return ItemTestReport(item, call.excinfo, call.when, call.outerr)
def pytest_runtest_teardown(item):
item.config._setupstate.teardown_exact(item)
#
# Implementation
#
class Call:
excinfo = None
def __init__(self, when, func):
self.when = when
try:
self.result = func()
except KeyboardInterrupt:
raise
except:
self.excinfo = py.code.ExceptionInfo()
def basic_run_report(item):
""" return report about setting up and running a test item. """
capture = item.config._getcapture()
def call_and_report(item, when, log=True):
call = RuntestHookCall(item, when)
hook = item.config.hook
try:
call = Call("setup", lambda: hook.pytest_runtest_setup(item=item))
if not call.excinfo:
call = Call("call", lambda: hook.pytest_runtest_call(item=item))
# in case of an error we defer teardown to not shadow the error
if not call.excinfo:
call = Call("teardown", lambda: hook.pytest_runtest_teardown(item=item))
finally:
outerr = capture.reset()
return item.config.hook.pytest_runtest_makereport(
item=item, excinfo=call.excinfo,
when=call.when, outerr=outerr)
report = hook.pytest_runtest_makereport(item=item, call=call)
if log and (when == "call" or not report.passed):
hook.pytest_runtest_logreport(rep=report)
return report
class RuntestHookCall:
excinfo = None
_prefix = "pytest_runtest_"
def __init__(self, item, when):
self.when = when
hookname = self._prefix + when
hook = getattr(item.config.hook, hookname)
capture = item.config._getcapture()
try:
try:
self.result = hook(item=item)
except KeyboardInterrupt:
raise
except:
self.excinfo = py.code.ExceptionInfo()
finally:
self.outerr = capture.reset()
def forked_run_report(item):
EXITSTATUS_TESTEXIT = 4
@ -98,10 +107,10 @@ def forked_run_report(item):
ipickle.selfmemoize(item.config)
def runforked():
try:
testrep = basic_run_report(item)
reports = runtestprotocol(item, log=False)
except KeyboardInterrupt:
py.std.os._exit(EXITSTATUS_TESTEXIT)
return ipickle.dumps(testrep)
return ipickle.dumps(reports)
ff = py.process.ForkedFunc(runforked)
result = ff.waitfinish()
@ -110,15 +119,13 @@ def forked_run_report(item):
else:
if result.exitstatus == EXITSTATUS_TESTEXIT:
py.test.exit("forked test item %s raised Exit" %(item,))
return report_process_crash(item, result)
return [report_process_crash(item, result)]
def report_process_crash(item, result):
path, lineno = item._getfslineno()
longrepr = [
("X", "CRASHED"),
("%s:%s: CRASHED with signal %d" %(path, lineno, result.signal)),
]
return ItemTestReport(item, excinfo=longrepr, when="???")
info = "%s:%s: running the test CRASHED with signal %d" %(
path, lineno, result.signal)
return ItemTestReport(item, excinfo=info, when="???")
class BaseReport(object):
def __repr__(self):
@ -138,6 +145,8 @@ class ItemTestReport(BaseReport):
def __init__(self, item, excinfo=None, when=None, outerr=None):
self.item = item
self.when = when
self.outerr = outerr
if item and when != "setup":
self.keywords = item.readkeywords()
else:
@ -151,7 +160,6 @@ class ItemTestReport(BaseReport):
self.passed = True
self.shortrepr = "."
else:
self.when = when
if not isinstance(excinfo, py.code.ExceptionInfo):
self.failed = True
shortrepr = "?"

View File

@ -162,6 +162,8 @@ class TerminalReporter:
self.write_fspath_result(fspath, "")
def pytest_runtest_logreport(self, rep):
if rep.passed and rep.when in ("setup", "teardown"):
return
fspath = rep.item.fspath
cat, letter, word = self.getcategoryletterword(rep)
if isinstance(word, tuple):
@ -399,6 +401,9 @@ def repr_pythonversion(v=None):
import pytest_runner as runner # XXX
def basic_run_report(item):
return runner.call_and_report(item, "call", log=False)
class TestTerminal:
def test_pass_skip_fail(self, testdir, linecomp):
modcol = testdir.getmodulecol("""
@ -415,7 +420,7 @@ class TestTerminal:
rep.config.hook.pytest_sessionstart(session=testdir.session)
for item in testdir.genitems([modcol]):
ev = runner.basic_run_report(item)
ev = basic_run_report(item)
rep.config.hook.pytest_runtest_logreport(rep=ev)
linecomp.assert_contains_lines([
"*test_pass_skip_fail.py .sF"
@ -446,7 +451,7 @@ class TestTerminal:
rep.config.hook.pytest_itemstart(item=item, node=None)
s = linecomp.stringio.getvalue().strip()
assert s.endswith(item.name)
rep.config.hook.pytest_runtest_logreport(rep=runner.basic_run_report(item))
rep.config.hook.pytest_runtest_logreport(rep=basic_run_report(item))
linecomp.assert_contains_lines([
"*test_pass_skip_fail_verbose.py:2: *test_ok*PASS*",
@ -537,7 +542,7 @@ class TestTerminal:
raise ValueError()
""")
rep = TerminalReporter(modcol.config, file=linecomp.stringio)
reports = [runner.basic_run_report(x) for x in modcol.collect()]
reports = [basic_run_report(x) for x in modcol.collect()]
rep.pytest_looponfailinfo(reports, [modcol.config.topdir])
linecomp.assert_contains_lines([
"*test_looponfailreport.py:2: assert 0",
@ -563,7 +568,7 @@ class TestTerminal:
rep.config.hook.pytest_sessionstart(session=testdir.session)
for item in testdir.genitems([modcol]):
rep.config.hook.pytest_runtest_logreport(
rep=runner.basic_run_report(item))
rep=basic_run_report(item))
rep.config.hook.pytest_sessionfinish(session=testdir.session, exitstatus=1)
s = linecomp.stringio.getvalue()
if tbopt == "long":
@ -644,7 +649,7 @@ class TestTerminal:
try:
for item in testdir.genitems([modcol]):
modcol.config.hook.pytest_runtest_logreport(
rep=runner.basic_run_report(item))
rep=basic_run_report(item))
except KeyboardInterrupt:
excinfo = py.code.ExceptionInfo()
else:

View File

@ -10,11 +10,13 @@ example:
"""
import py
def pytest_runtest_makereport(__call__, item, excinfo, when, outerr):
def pytest_runtest_makereport(__call__, item, call):
if call.when != "call":
return
if hasattr(item, 'obj') and hasattr(item.obj, 'func_dict'):
if 'xfail' in item.obj.func_dict:
res = __call__.execute(firstresult=True)
if excinfo:
if call.excinfo:
res.skipped = True
res.failed = res.passed = False
else:

View File

@ -30,29 +30,32 @@ class TestSetupState:
class BaseFunctionalTests:
def test_funcattr(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
@py.test.mark(xfail="needs refactoring")
def test_func():
raise Exit()
""")
rep = reports[1]
assert rep.keywords['xfail'] == "needs refactoring"
def test_passfunction(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
def test_func():
pass
""")
rep = reports[1]
assert rep.passed
assert not rep.failed
assert rep.shortrepr == "."
assert not hasattr(rep, 'longrepr')
def test_failfunction(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
def test_func():
assert 0
""")
rep = reports[1]
assert not rep.passed
assert not rep.skipped
assert rep.failed
@ -61,11 +64,12 @@ class BaseFunctionalTests:
assert str(rep.shortrepr) == "F"
def test_skipfunction(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
def test_func():
py.test.skip("hello")
""")
rep = reports[1]
assert not rep.failed
assert not rep.passed
assert rep.skipped
@ -77,44 +81,49 @@ class BaseFunctionalTests:
#assert not rep.skipped.failurerepr
def test_skip_in_setup_function(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
def setup_function(func):
py.test.skip("hello")
def test_func():
pass
""")
print rep
print reports
rep = reports[0]
assert not rep.failed
assert not rep.passed
assert rep.skipped
#assert rep.skipped.reason == "hello"
#assert rep.skipped.location.lineno == 3
#assert rep.skipped.location.lineno == 3
assert len(reports) == 1
def test_failure_in_setup_function(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
def setup_function(func):
raise ValueError(42)
def test_func():
pass
""")
print rep
rep = reports[0]
assert not rep.skipped
assert not rep.passed
assert rep.failed
assert rep.when == "setup"
assert len(reports) == 1
def test_failure_in_teardown_function(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
def teardown_function(func):
raise ValueError(42)
def test_func():
pass
""")
print rep
print reports
assert len(reports) == 3
rep = reports[2]
assert not rep.skipped
assert not rep.passed
assert rep.failed
@ -129,11 +138,12 @@ class BaseFunctionalTests:
def repr_failure(self, excinfo, outerr):
return "hello"
""")
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
def test_func():
assert 0
""")
rep = reports[1]
assert not rep.skipped
assert not rep.passed
assert rep.failed
@ -149,13 +159,15 @@ class BaseFunctionalTests:
def repr_failure(self, excinfo):
assert 0
""")
rep = testdir.runitem("""
reports = testdir.runitem("""
import py
def setup_function(func):
raise ValueError(42)
def test_func():
pass
""")
assert len(reports) == 1
rep = reports[0]
print rep
assert not rep.skipped
assert not rep.passed
@ -166,29 +178,29 @@ class BaseFunctionalTests:
#assert instanace(rep.failed.failurerepr, PythonFailureRepr)
def test_capture_in_func(self, testdir):
rep = testdir.runitem("""
import py
reports = testdir.runitem("""
import sys
def setup_function(func):
print >>py.std.sys.stderr, "in setup"
print "in setup"
def test_func():
print "in function"
assert 0
def teardown_func(func):
def teardown_function(func):
print "in teardown"
""")
assert rep.failed
# out, err = rep.failed.outerr
# assert out == ['in function\nin teardown\n']
# assert err == ['in setup\n']
assert reports[0].outerr[0] == "in setup\n"
assert reports[1].outerr[0] == "in function\n"
assert reports[2].outerr[0] == "in teardown\n"
def test_systemexit_does_not_bail_out(self, testdir):
try:
rep = testdir.runitem("""
reports = testdir.runitem("""
def test_func():
raise SystemExit(42)
""")
except SystemExit:
py.test.fail("runner did not catch SystemExit")
rep = reports[1]
assert rep.failed
assert rep.when == "call"
@ -208,7 +220,9 @@ class BaseFunctionalTests:
class TestExecutionNonForked(BaseFunctionalTests):
def getrunner(self):
return runner.basic_run_report
def f(item):
return runner.runtestprotocol(item, log=False)
return f
def test_keyboardinterrupt_propagates(self, testdir):
from py.__.test.outcome import Exit
@ -229,11 +243,12 @@ class TestExecutionForked(BaseFunctionalTests):
return runner.forked_run_report
def test_suicide(self, testdir):
rep = testdir.runitem("""
reports = testdir.runitem("""
def test_func():
import os
os.kill(os.getpid(), 15)
""")
rep = reports[0]
assert rep.failed
assert rep.when == "???"
@ -265,3 +280,18 @@ class TestCollectionReports:
assert not rep.failed
assert not rep.passed
assert rep.skipped
def test_functional_boxed(testdir):
if not hasattr(py.std.os, 'fork'):
py.test.skip("needs os.fork")
p1 = testdir.makepyfile("""
import os
def test_function():
os.kill(os.getpid(), 15)
""")
result = testdir.runpytest(p1, "--boxed")
assert result.stdout.fnmatch_lines([
"*CRASHED*",
"*1 failed*"
])