cpython/Lib/test/test_subprocess.py

1160 lines
46 KiB
Python

import unittest
from test import support
import subprocess
import sys
import signal
import os
import errno
import tempfile
import time
import re
import sysconfig
try:
import gc
except ImportError:
gc = None
mswindows = (sys.platform == "win32")
#
# Depends on the following external programs: Python
#
if mswindows:
SETBINARY = ('import msvcrt; msvcrt.setmode(sys.stdout.fileno(), '
'os.O_BINARY);')
else:
SETBINARY = ''
try:
mkstemp = tempfile.mkstemp
except AttributeError:
# tempfile.mkstemp is not available
def mkstemp():
"""Replacement for mkstemp, calling mktemp."""
fname = tempfile.mktemp()
return os.open(fname, os.O_RDWR|os.O_CREAT), fname
class BaseTestCase(unittest.TestCase):
def setUp(self):
# Try to minimize the number of children we have so this test
# doesn't crash on some buildbots (Alphas in particular).
support.reap_children()
def tearDown(self):
for inst in subprocess._active:
inst.wait()
subprocess._cleanup()
self.assertFalse(subprocess._active, "subprocess._active not empty")
def assertStderrEqual(self, stderr, expected, msg=None):
# In a debug build, stuff like "[6580 refs]" is printed to stderr at
# shutdown time. That frustrates tests trying to check stderr produced
# from a spawned Python process.
actual = support.strip_python_stderr(stderr)
self.assertEqual(actual, expected, msg)
class ProcessTestCase(BaseTestCase):
def test_call_seq(self):
# call() function with sequence argument
rc = subprocess.call([sys.executable, "-c",
"import sys; sys.exit(47)"])
self.assertEqual(rc, 47)
def test_check_call_zero(self):
# check_call() function with zero return code
rc = subprocess.check_call([sys.executable, "-c",
"import sys; sys.exit(0)"])
self.assertEqual(rc, 0)
def test_check_call_nonzero(self):
# check_call() function with non-zero return code
with self.assertRaises(subprocess.CalledProcessError) as c:
subprocess.check_call([sys.executable, "-c",
"import sys; sys.exit(47)"])
self.assertEqual(c.exception.returncode, 47)
def test_check_output(self):
# check_output() function with zero return code
output = subprocess.check_output(
[sys.executable, "-c", "print('BDFL')"])
self.assertIn(b'BDFL', output)
def test_check_output_nonzero(self):
# check_call() function with non-zero return code
with self.assertRaises(subprocess.CalledProcessError) as c:
subprocess.check_output(
[sys.executable, "-c", "import sys; sys.exit(5)"])
self.assertEqual(c.exception.returncode, 5)
def test_check_output_stderr(self):
# check_output() function stderr redirected to stdout
output = subprocess.check_output(
[sys.executable, "-c", "import sys; sys.stderr.write('BDFL')"],
stderr=subprocess.STDOUT)
self.assertIn(b'BDFL', output)
def test_check_output_stdout_arg(self):
# check_output() function stderr redirected to stdout
with self.assertRaises(ValueError) as c:
output = subprocess.check_output(
[sys.executable, "-c", "print('will not be run')"],
stdout=sys.stdout)
self.fail("Expected ValueError when stdout arg supplied.")
self.assertIn('stdout', c.exception.args[0])
def test_call_kwargs(self):
# call() function with keyword args
newenv = os.environ.copy()
newenv["FRUIT"] = "banana"
rc = subprocess.call([sys.executable, "-c",
'import sys, os;'
'sys.exit(os.getenv("FRUIT")=="banana")'],
env=newenv)
self.assertEqual(rc, 1)
def test_stdin_none(self):
# .stdin is None when not redirected
p = subprocess.Popen([sys.executable, "-c", 'print("banana")'],
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
p.wait()
self.assertEqual(p.stdin, None)
def test_stdout_none(self):
# .stdout is None when not redirected
p = subprocess.Popen([sys.executable, "-c",
'print(" this bit of output is from a '
'test of stdout in a different '
'process ...")'],
stdin=subprocess.PIPE, stderr=subprocess.PIPE)
p.wait()
self.assertEqual(p.stdout, None)
def test_stderr_none(self):
# .stderr is None when not redirected
p = subprocess.Popen([sys.executable, "-c", 'print("banana")'],
stdin=subprocess.PIPE, stdout=subprocess.PIPE)
p.wait()
self.assertEqual(p.stderr, None)
def test_executable_with_cwd(self):
python_dir = os.path.dirname(os.path.realpath(sys.executable))
p = subprocess.Popen(["somethingyoudonthave", "-c",
"import sys; sys.exit(47)"],
executable=sys.executable, cwd=python_dir)
p.wait()
self.assertEqual(p.returncode, 47)
@unittest.skipIf(sysconfig.is_python_build(),
"need an installed Python. See #7774")
def test_executable_without_cwd(self):
# For a normal installation, it should work without 'cwd'
# argument. For test runs in the build directory, see #7774.
p = subprocess.Popen(["somethingyoudonthave", "-c",
"import sys; sys.exit(47)"],
executable=sys.executable)
p.wait()
self.assertEqual(p.returncode, 47)
def test_stdin_pipe(self):
# stdin redirection
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.exit(sys.stdin.read() == "pear")'],
stdin=subprocess.PIPE)
p.stdin.write(b"pear")
p.stdin.close()
p.wait()
self.assertEqual(p.returncode, 1)
def test_stdin_filedes(self):
# stdin is set to open file descriptor
tf = tempfile.TemporaryFile()
d = tf.fileno()
os.write(d, b"pear")
os.lseek(d, 0, 0)
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.exit(sys.stdin.read() == "pear")'],
stdin=d)
p.wait()
self.assertEqual(p.returncode, 1)
def test_stdin_fileobj(self):
# stdin is set to open file object
tf = tempfile.TemporaryFile()
tf.write(b"pear")
tf.seek(0)
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.exit(sys.stdin.read() == "pear")'],
stdin=tf)
p.wait()
self.assertEqual(p.returncode, 1)
def test_stdout_pipe(self):
# stdout redirection
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stdout.write("orange")'],
stdout=subprocess.PIPE)
self.assertEqual(p.stdout.read(), b"orange")
def test_stdout_filedes(self):
# stdout is set to open file descriptor
tf = tempfile.TemporaryFile()
d = tf.fileno()
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stdout.write("orange")'],
stdout=d)
p.wait()
os.lseek(d, 0, 0)
self.assertEqual(os.read(d, 1024), b"orange")
def test_stdout_fileobj(self):
# stdout is set to open file object
tf = tempfile.TemporaryFile()
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stdout.write("orange")'],
stdout=tf)
p.wait()
tf.seek(0)
self.assertEqual(tf.read(), b"orange")
def test_stderr_pipe(self):
# stderr redirection
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stderr.write("strawberry")'],
stderr=subprocess.PIPE)
self.assertStderrEqual(p.stderr.read(), b"strawberry")
def test_stderr_filedes(self):
# stderr is set to open file descriptor
tf = tempfile.TemporaryFile()
d = tf.fileno()
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stderr.write("strawberry")'],
stderr=d)
p.wait()
os.lseek(d, 0, 0)
self.assertStderrEqual(os.read(d, 1024), b"strawberry")
def test_stderr_fileobj(self):
# stderr is set to open file object
tf = tempfile.TemporaryFile()
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stderr.write("strawberry")'],
stderr=tf)
p.wait()
tf.seek(0)
self.assertStderrEqual(tf.read(), b"strawberry")
def test_stdout_stderr_pipe(self):
# capture stdout and stderr to the same pipe
p = subprocess.Popen([sys.executable, "-c",
'import sys;'
'sys.stdout.write("apple");'
'sys.stdout.flush();'
'sys.stderr.write("orange")'],
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
self.assertStderrEqual(p.stdout.read(), b"appleorange")
def test_stdout_stderr_file(self):
# capture stdout and stderr to the same open file
tf = tempfile.TemporaryFile()
p = subprocess.Popen([sys.executable, "-c",
'import sys;'
'sys.stdout.write("apple");'
'sys.stdout.flush();'
'sys.stderr.write("orange")'],
stdout=tf,
stderr=tf)
p.wait()
tf.seek(0)
self.assertStderrEqual(tf.read(), b"appleorange")
def test_stdout_filedes_of_stdout(self):
# stdout is set to 1 (#1531862).
cmd = r"import sys, os; sys.exit(os.write(sys.stdout.fileno(), b'.\n'))"
rc = subprocess.call([sys.executable, "-c", cmd], stdout=1)
self.assertEqual(rc, 2)
def test_cwd(self):
tmpdir = tempfile.gettempdir()
# We cannot use os.path.realpath to canonicalize the path,
# since it doesn't expand Tru64 {memb} strings. See bug 1063571.
cwd = os.getcwd()
os.chdir(tmpdir)
tmpdir = os.getcwd()
os.chdir(cwd)
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;'
'sys.stdout.write(os.getcwd())'],
stdout=subprocess.PIPE,
cwd=tmpdir)
normcase = os.path.normcase
self.assertEqual(normcase(p.stdout.read().decode("utf-8")),
normcase(tmpdir))
def test_env(self):
newenv = os.environ.copy()
newenv["FRUIT"] = "orange"
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;'
'sys.stdout.write(os.getenv("FRUIT"))'],
stdout=subprocess.PIPE,
env=newenv)
self.assertEqual(p.stdout.read(), b"orange")
def test_communicate_stdin(self):
p = subprocess.Popen([sys.executable, "-c",
'import sys;'
'sys.exit(sys.stdin.read() == "pear")'],
stdin=subprocess.PIPE)
p.communicate(b"pear")
self.assertEqual(p.returncode, 1)
def test_communicate_stdout(self):
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stdout.write("pineapple")'],
stdout=subprocess.PIPE)
(stdout, stderr) = p.communicate()
self.assertEqual(stdout, b"pineapple")
self.assertEqual(stderr, None)
def test_communicate_stderr(self):
p = subprocess.Popen([sys.executable, "-c",
'import sys; sys.stderr.write("pineapple")'],
stderr=subprocess.PIPE)
(stdout, stderr) = p.communicate()
self.assertEqual(stdout, None)
self.assertStderrEqual(stderr, b"pineapple")
def test_communicate(self):
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;'
'sys.stderr.write("pineapple");'
'sys.stdout.write(sys.stdin.read())'],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
(stdout, stderr) = p.communicate(b"banana")
self.assertEqual(stdout, b"banana")
self.assertStderrEqual(stderr, b"pineapple")
# This test is Linux specific for simplicity to at least have
# some coverage. It is not a platform specific bug.
@unittest.skipUnless(os.path.isdir('/proc/%d/fd' % os.getpid()),
"Linux specific")
# Test for the fd leak reported in http://bugs.python.org/issue2791.
def test_communicate_pipe_fd_leak(self):
fd_directory = '/proc/%d/fd' % os.getpid()
num_fds_before_popen = len(os.listdir(fd_directory))
p = subprocess.Popen([sys.executable, "-c", "print()"],
stdout=subprocess.PIPE)
p.communicate()
num_fds_after_communicate = len(os.listdir(fd_directory))
del p
num_fds_after_destruction = len(os.listdir(fd_directory))
self.assertEqual(num_fds_before_popen, num_fds_after_destruction)
self.assertEqual(num_fds_before_popen, num_fds_after_communicate)
def test_communicate_returns(self):
# communicate() should return None if no redirection is active
p = subprocess.Popen([sys.executable, "-c",
"import sys; sys.exit(47)"])
(stdout, stderr) = p.communicate()
self.assertEqual(stdout, None)
self.assertEqual(stderr, None)
def test_communicate_pipe_buf(self):
# communicate() with writes larger than pipe_buf
# This test will probably deadlock rather than fail, if
# communicate() does not work properly.
x, y = os.pipe()
if mswindows:
pipe_buf = 512
else:
pipe_buf = os.fpathconf(x, "PC_PIPE_BUF")
os.close(x)
os.close(y)
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;'
'sys.stdout.write(sys.stdin.read(47));'
'sys.stderr.write("xyz"*%d);'
'sys.stdout.write(sys.stdin.read())' % pipe_buf],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
string_to_write = b"abc"*pipe_buf
(stdout, stderr) = p.communicate(string_to_write)
self.assertEqual(stdout, string_to_write)
def test_writes_before_communicate(self):
# stdin.write before communicate()
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;'
'sys.stdout.write(sys.stdin.read())'],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
p.stdin.write(b"banana")
(stdout, stderr) = p.communicate(b"split")
self.assertEqual(stdout, b"bananasplit")
self.assertStderrEqual(stderr, b"")
def test_universal_newlines(self):
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;' + SETBINARY +
'sys.stdout.write("line1\\n");'
'sys.stdout.flush();'
'sys.stdout.write("line2\\n");'
'sys.stdout.flush();'
'sys.stdout.write("line3\\r\\n");'
'sys.stdout.flush();'
'sys.stdout.write("line4\\r");'
'sys.stdout.flush();'
'sys.stdout.write("\\nline5");'
'sys.stdout.flush();'
'sys.stdout.write("\\nline6");'],
stdout=subprocess.PIPE,
universal_newlines=1)
stdout = p.stdout.read()
self.assertEqual(stdout, "line1\nline2\nline3\nline4\nline5\nline6")
def test_universal_newlines_communicate(self):
# universal newlines through communicate()
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;' + SETBINARY +
'sys.stdout.write("line1\\n");'
'sys.stdout.flush();'
'sys.stdout.write("line2\\n");'
'sys.stdout.flush();'
'sys.stdout.write("line3\\r\\n");'
'sys.stdout.flush();'
'sys.stdout.write("line4\\r");'
'sys.stdout.flush();'
'sys.stdout.write("\\nline5");'
'sys.stdout.flush();'
'sys.stdout.write("\\nline6");'],
stdout=subprocess.PIPE, stderr=subprocess.PIPE,
universal_newlines=1)
(stdout, stderr) = p.communicate()
self.assertEqual(stdout, "line1\nline2\nline3\nline4\nline5\nline6")
def test_no_leaking(self):
# Make sure we leak no resources
if not mswindows:
max_handles = 1026 # too much for most UNIX systems
else:
max_handles = 2050 # too much for (at least some) Windows setups
handles = []
try:
for i in range(max_handles):
try:
handles.append(os.open(support.TESTFN,
os.O_WRONLY | os.O_CREAT))
except OSError as e:
if e.errno != errno.EMFILE:
raise
break
else:
self.skipTest("failed to reach the file descriptor limit "
"(tried %d)" % max_handles)
# Close a couple of them (should be enough for a subprocess)
for i in range(10):
os.close(handles.pop())
# Loop creating some subprocesses. If one of them leaks some fds,
# the next loop iteration will fail by reaching the max fd limit.
for i in range(15):
p = subprocess.Popen([sys.executable, "-c",
"import sys;"
"sys.stdout.write(sys.stdin.read())"],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
data = p.communicate(b"lime")[0]
self.assertEqual(data, b"lime")
finally:
for h in handles:
os.close(h)
def test_list2cmdline(self):
self.assertEqual(subprocess.list2cmdline(['a b c', 'd', 'e']),
'"a b c" d e')
self.assertEqual(subprocess.list2cmdline(['ab"c', '\\', 'd']),
'ab\\"c \\ d')
self.assertEqual(subprocess.list2cmdline(['ab"c', ' \\', 'd']),
'ab\\"c " \\\\" d')
self.assertEqual(subprocess.list2cmdline(['a\\\\\\b', 'de fg', 'h']),
'a\\\\\\b "de fg" h')
self.assertEqual(subprocess.list2cmdline(['a\\"b', 'c', 'd']),
'a\\\\\\"b c d')
self.assertEqual(subprocess.list2cmdline(['a\\\\b c', 'd', 'e']),
'"a\\\\b c" d e')
self.assertEqual(subprocess.list2cmdline(['a\\\\b\\ c', 'd', 'e']),
'"a\\\\b\\ c" d e')
self.assertEqual(subprocess.list2cmdline(['ab', '']),
'ab ""')
def test_poll(self):
p = subprocess.Popen([sys.executable,
"-c", "import time; time.sleep(1)"])
count = 0
while p.poll() is None:
time.sleep(0.1)
count += 1
# We expect that the poll loop probably went around about 10 times,
# but, based on system scheduling we can't control, it's possible
# poll() never returned None. It "should be" very rare that it
# didn't go around at least twice.
self.assertGreaterEqual(count, 2)
# Subsequent invocations should just return the returncode
self.assertEqual(p.poll(), 0)
def test_wait(self):
p = subprocess.Popen([sys.executable,
"-c", "import time; time.sleep(2)"])
self.assertEqual(p.wait(), 0)
# Subsequent invocations should just return the returncode
self.assertEqual(p.wait(), 0)
def test_invalid_bufsize(self):
# an invalid type of the bufsize argument should raise
# TypeError.
with self.assertRaises(TypeError):
subprocess.Popen([sys.executable, "-c", "pass"], "orange")
def test_bufsize_is_none(self):
# bufsize=None should be the same as bufsize=0.
p = subprocess.Popen([sys.executable, "-c", "pass"], None)
self.assertEqual(p.wait(), 0)
# Again with keyword arg
p = subprocess.Popen([sys.executable, "-c", "pass"], bufsize=None)
self.assertEqual(p.wait(), 0)
def test_leaking_fds_on_error(self):
# see bug #5179: Popen leaks file descriptors to PIPEs if
# the child fails to execute; this will eventually exhaust
# the maximum number of open fds. 1024 seems a very common
# value for that limit, but Windows has 2048, so we loop
# 1024 times (each call leaked two fds).
for i in range(1024):
# Windows raises IOError. Others raise OSError.
with self.assertRaises(EnvironmentError) as c:
subprocess.Popen(['nonexisting_i_hope'],
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if c.exception.errno != errno.ENOENT: # ignore "no such file"
raise c.exception
def test_issue8780(self):
# Ensure that stdout is inherited from the parent
# if stdout=PIPE is not used
code = ';'.join((
'import subprocess, sys',
'retcode = subprocess.call('
"[sys.executable, '-c', 'print(\"Hello World!\")'])",
'assert retcode == 0'))
output = subprocess.check_output([sys.executable, '-c', code])
self.assert_(output.startswith(b'Hello World!'), ascii(output))
def test_handles_closed_on_exception(self):
# If CreateProcess exits with an error, ensure the
# duplicate output handles are released
ifhandle, ifname = mkstemp()
ofhandle, ofname = mkstemp()
efhandle, efname = mkstemp()
try:
subprocess.Popen (["*"], stdin=ifhandle, stdout=ofhandle,
stderr=efhandle)
except OSError:
os.close(ifhandle)
os.remove(ifname)
os.close(ofhandle)
os.remove(ofname)
os.close(efhandle)
os.remove(efname)
self.assertFalse(os.path.exists(ifname))
self.assertFalse(os.path.exists(ofname))
self.assertFalse(os.path.exists(efname))
# context manager
class _SuppressCoreFiles(object):
"""Try to prevent core files from being created."""
old_limit = None
def __enter__(self):
"""Try to save previous ulimit, then set it to (0, 0)."""
try:
import resource
self.old_limit = resource.getrlimit(resource.RLIMIT_CORE)
resource.setrlimit(resource.RLIMIT_CORE, (0, 0))
except (ImportError, ValueError, resource.error):
pass
if sys.platform == 'darwin':
# Check if the 'Crash Reporter' on OSX was configured
# in 'Developer' mode and warn that it will get triggered
# when it is.
#
# This assumes that this context manager is used in tests
# that might trigger the next manager.
value = subprocess.Popen(['/usr/bin/defaults', 'read',
'com.apple.CrashReporter', 'DialogType'],
stdout=subprocess.PIPE).communicate()[0]
if value.strip() == b'developer':
print("this tests triggers the Crash Reporter, "
"that is intentional", end='')
sys.stdout.flush()
def __exit__(self, *args):
"""Return core file behavior to default."""
if self.old_limit is None:
return
try:
import resource
resource.setrlimit(resource.RLIMIT_CORE, self.old_limit)
except (ImportError, ValueError, resource.error):
pass
@unittest.skipIf(mswindows, "POSIX specific tests")
class POSIXProcessTestCase(BaseTestCase):
def test_exceptions(self):
nonexistent_dir = "/_this/pa.th/does/not/exist"
try:
os.chdir(nonexistent_dir)
except OSError as e:
# This avoids hard coding the errno value or the OS perror()
# string and instead capture the exception that we want to see
# below for comparison.
desired_exception = e
else:
self.fail("chdir to nonexistant directory %s succeeded." %
nonexistent_dir)
# Error in the child re-raised in the parent.
try:
p = subprocess.Popen([sys.executable, "-c", ""],
cwd=nonexistent_dir)
except OSError as e:
# Test that the child process chdir failure actually makes
# it up to the parent process as the correct exception.
self.assertEqual(desired_exception.errno, e.errno)
self.assertEqual(desired_exception.strerror, e.strerror)
else:
self.fail("Expected OSError: %s" % desired_exception)
def test_restore_signals(self):
# Code coverage for both values of restore_signals to make sure it
# at least does not blow up.
# A test for behavior would be complex. Contributions welcome.
subprocess.call([sys.executable, "-c", ""], restore_signals=True)
subprocess.call([sys.executable, "-c", ""], restore_signals=False)
def test_start_new_session(self):
# For code coverage of calling setsid(). We don't care if we get an
# EPERM error from it depending on the test execution environment, that
# still indicates that it was called.
try:
output = subprocess.check_output(
[sys.executable, "-c",
"import os; print(os.getpgid(os.getpid()))"],
start_new_session=True)
except OSError as e:
if e.errno != errno.EPERM:
raise
else:
parent_pgid = os.getpgid(os.getpid())
child_pgid = int(output)
self.assertNotEqual(parent_pgid, child_pgid)
def test_run_abort(self):
# returncode handles signal termination
with _SuppressCoreFiles():
p = subprocess.Popen([sys.executable, "-c",
'import os; os.abort()'])
p.wait()
self.assertEqual(-p.returncode, signal.SIGABRT)
def test_preexec(self):
# DISCLAIMER: Setting environment variables is *not* a good use
# of a preexec_fn. This is merely a test.
p = subprocess.Popen([sys.executable, "-c",
'import sys,os;'
'sys.stdout.write(os.getenv("FRUIT"))'],
stdout=subprocess.PIPE,
preexec_fn=lambda: os.putenv("FRUIT", "apple"))
self.assertEqual(p.stdout.read(), b"apple")
def test_preexec_exception(self):
def raise_it():
raise ValueError("What if two swallows carried a coconut?")
try:
p = subprocess.Popen([sys.executable, "-c", ""],
preexec_fn=raise_it)
except RuntimeError as e:
self.assertTrue(
subprocess._posixsubprocess,
"Expected a ValueError from the preexec_fn")
except ValueError as e:
self.assertIn("coconut", e.args[0])
else:
self.fail("Exception raised by preexec_fn did not make it "
"to the parent process.")
@unittest.skipUnless(gc, "Requires a gc module.")
def test_preexec_gc_module_failure(self):
# This tests the code that disables garbage collection if the child
# process will execute any Python.
def raise_runtime_error():
raise RuntimeError("this shouldn't escape")
enabled = gc.isenabled()
orig_gc_disable = gc.disable
orig_gc_isenabled = gc.isenabled
try:
gc.disable()
self.assertFalse(gc.isenabled())
subprocess.call([sys.executable, '-c', ''],
preexec_fn=lambda: None)
self.assertFalse(gc.isenabled(),
"Popen enabled gc when it shouldn't.")
gc.enable()
self.assertTrue(gc.isenabled())
subprocess.call([sys.executable, '-c', ''],
preexec_fn=lambda: None)
self.assertTrue(gc.isenabled(), "Popen left gc disabled.")
gc.disable = raise_runtime_error
self.assertRaises(RuntimeError, subprocess.Popen,
[sys.executable, '-c', ''],
preexec_fn=lambda: None)
del gc.isenabled # force an AttributeError
self.assertRaises(AttributeError, subprocess.Popen,
[sys.executable, '-c', ''],
preexec_fn=lambda: None)
finally:
gc.disable = orig_gc_disable
gc.isenabled = orig_gc_isenabled
if not enabled:
gc.disable()
def test_args_string(self):
# args is a string
fd, fname = mkstemp()
# reopen in text mode
with open(fd, "w") as fobj:
fobj.write("#!/bin/sh\n")
fobj.write("exec '%s' -c 'import sys; sys.exit(47)'\n" %
sys.executable)
os.chmod(fname, 0o700)
p = subprocess.Popen(fname)
p.wait()
os.remove(fname)
self.assertEqual(p.returncode, 47)
def test_invalid_args(self):
# invalid arguments should raise ValueError
self.assertRaises(ValueError, subprocess.call,
[sys.executable, "-c",
"import sys; sys.exit(47)"],
startupinfo=47)
self.assertRaises(ValueError, subprocess.call,
[sys.executable, "-c",
"import sys; sys.exit(47)"],
creationflags=47)
def test_shell_sequence(self):
# Run command through the shell (sequence)
newenv = os.environ.copy()
newenv["FRUIT"] = "apple"
p = subprocess.Popen(["echo $FRUIT"], shell=1,
stdout=subprocess.PIPE,
env=newenv)
self.assertEqual(p.stdout.read().strip(b" \t\r\n\f"), b"apple")
def test_shell_string(self):
# Run command through the shell (string)
newenv = os.environ.copy()
newenv["FRUIT"] = "apple"
p = subprocess.Popen("echo $FRUIT", shell=1,
stdout=subprocess.PIPE,
env=newenv)
self.assertEqual(p.stdout.read().strip(b" \t\r\n\f"), b"apple")
def test_call_string(self):
# call() function with string argument on UNIX
fd, fname = mkstemp()
# reopen in text mode
with open(fd, "w") as fobj:
fobj.write("#!/bin/sh\n")
fobj.write("exec '%s' -c 'import sys; sys.exit(47)'\n" %
sys.executable)
os.chmod(fname, 0o700)
rc = subprocess.call(fname)
os.remove(fname)
self.assertEqual(rc, 47)
def test_specific_shell(self):
# Issue #9265: Incorrect name passed as arg[0].
shells = []
for prefix in ['/bin', '/usr/bin/', '/usr/local/bin']:
for name in ['bash', 'ksh']:
sh = os.path.join(prefix, name)
if os.path.isfile(sh):
shells.append(sh)
if not shells: # Will probably work for any shell but csh.
self.skipTest("bash or ksh required for this test")
sh = '/bin/sh'
if os.path.isfile(sh) and not os.path.islink(sh):
# Test will fail if /bin/sh is a symlink to csh.
shells.append(sh)
for sh in shells:
p = subprocess.Popen("echo $0", executable=sh, shell=True,
stdout=subprocess.PIPE)
self.assertEqual(p.stdout.read().strip(), bytes(sh, 'ascii'))
def _kill_process(self, method, *args):
# Do not inherit file handles from the parent.
# It should fix failures on some platforms.
p = subprocess.Popen([sys.executable, "-c", """if 1:
import sys, time
sys.stdout.write('x\\n')
sys.stdout.flush()
time.sleep(30)
"""],
close_fds=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
# Wait for the interpreter to be completely initialized before
# sending any signal.
p.stdout.read(1)
getattr(p, method)(*args)
return p
def test_send_signal(self):
p = self._kill_process('send_signal', signal.SIGINT)
_, stderr = p.communicate()
self.assertIn(b'KeyboardInterrupt', stderr)
self.assertNotEqual(p.wait(), 0)
def test_kill(self):
p = self._kill_process('kill')
_, stderr = p.communicate()
self.assertStderrEqual(stderr, b'')
self.assertEqual(p.wait(), -signal.SIGKILL)
def test_terminate(self):
p = self._kill_process('terminate')
_, stderr = p.communicate()
self.assertStderrEqual(stderr, b'')
self.assertEqual(p.wait(), -signal.SIGTERM)
def test_surrogates_error_message(self):
def prepare():
raise ValueError("surrogate:\uDCff")
try:
subprocess.call(
[sys.executable, "-c", "pass"],
preexec_fn=prepare)
except ValueError as err:
# Pure Python implementations keeps the message
self.assertIsNone(subprocess._posixsubprocess)
self.assertEqual(str(err), "surrogate:\uDCff")
except RuntimeError as err:
# _posixsubprocess uses a default message
self.assertIsNotNone(subprocess._posixsubprocess)
self.assertEqual(str(err), "Exception occurred in preexec_fn.")
else:
self.fail("Expected ValueError or RuntimeError")
def test_undecodable_env(self):
for key, value in (('test', 'abc\uDCFF'), ('test\uDCFF', '42')):
# test str with surrogates
script = "import os; print(ascii(os.getenv(%s)))" % repr(key)
env = os.environ.copy()
env[key] = value
# Force surrogate-escaping of \xFF in the child process;
# otherwise it can be decoded as-is if the default locale
# is latin-1.
env['PYTHONFSENCODING'] = 'ascii'
stdout = subprocess.check_output(
[sys.executable, "-c", script],
env=env)
stdout = stdout.rstrip(b'\n\r')
self.assertEquals(stdout.decode('ascii'), ascii(value))
# test bytes
key = key.encode("ascii", "surrogateescape")
value = value.encode("ascii", "surrogateescape")
script = "import os; print(ascii(os.getenvb(%s)))" % repr(key)
env = os.environ.copy()
env[key] = value
stdout = subprocess.check_output(
[sys.executable, "-c", script],
env=env)
stdout = stdout.rstrip(b'\n\r')
self.assertEquals(stdout.decode('ascii'), ascii(value))
def test_bytes_program(self):
abs_program = os.fsencode(sys.executable)
path, program = os.path.split(sys.executable)
program = os.fsencode(program)
# absolute bytes path
exitcode = subprocess.call([abs_program, "-c", "pass"])
self.assertEquals(exitcode, 0)
# bytes program, unicode PATH
env = os.environ.copy()
env["PATH"] = path
exitcode = subprocess.call([program, "-c", "pass"], env=env)
self.assertEquals(exitcode, 0)
# bytes program, bytes PATH
envb = os.environb.copy()
envb[b"PATH"] = os.fsencode(path)
exitcode = subprocess.call([program, "-c", "pass"], env=envb)
self.assertEquals(exitcode, 0)
@unittest.skipUnless(mswindows, "Windows specific tests")
class Win32ProcessTestCase(BaseTestCase):
def test_startupinfo(self):
# startupinfo argument
# We uses hardcoded constants, because we do not want to
# depend on win32all.
STARTF_USESHOWWINDOW = 1
SW_MAXIMIZE = 3
startupinfo = subprocess.STARTUPINFO()
startupinfo.dwFlags = STARTF_USESHOWWINDOW
startupinfo.wShowWindow = SW_MAXIMIZE
# Since Python is a console process, it won't be affected
# by wShowWindow, but the argument should be silently
# ignored
subprocess.call([sys.executable, "-c", "import sys; sys.exit(0)"],
startupinfo=startupinfo)
def test_creationflags(self):
# creationflags argument
CREATE_NEW_CONSOLE = 16
sys.stderr.write(" a DOS box should flash briefly ...\n")
subprocess.call(sys.executable +
' -c "import time; time.sleep(0.25)"',
creationflags=CREATE_NEW_CONSOLE)
def test_invalid_args(self):
# invalid arguments should raise ValueError
self.assertRaises(ValueError, subprocess.call,
[sys.executable, "-c",
"import sys; sys.exit(47)"],
preexec_fn=lambda: 1)
self.assertRaises(ValueError, subprocess.call,
[sys.executable, "-c",
"import sys; sys.exit(47)"],
stdout=subprocess.PIPE,
close_fds=True)
def test_close_fds(self):
# close file descriptors
rc = subprocess.call([sys.executable, "-c",
"import sys; sys.exit(47)"],
close_fds=True)
self.assertEqual(rc, 47)
def test_shell_sequence(self):
# Run command through the shell (sequence)
newenv = os.environ.copy()
newenv["FRUIT"] = "physalis"
p = subprocess.Popen(["set"], shell=1,
stdout=subprocess.PIPE,
env=newenv)
self.assertIn(b"physalis", p.stdout.read())
def test_shell_string(self):
# Run command through the shell (string)
newenv = os.environ.copy()
newenv["FRUIT"] = "physalis"
p = subprocess.Popen("set", shell=1,
stdout=subprocess.PIPE,
env=newenv)
self.assertIn(b"physalis", p.stdout.read())
def test_call_string(self):
# call() function with string argument on Windows
rc = subprocess.call(sys.executable +
' -c "import sys; sys.exit(47)"')
self.assertEqual(rc, 47)
def _kill_process(self, method, *args):
# Some win32 buildbot raises EOFError if stdin is inherited
p = subprocess.Popen([sys.executable, "-c", """if 1:
import sys, time
sys.stdout.write('x\\n')
sys.stdout.flush()
time.sleep(30)
"""],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
# Wait for the interpreter to be completely initialized before
# sending any signal.
p.stdout.read(1)
getattr(p, method)(*args)
_, stderr = p.communicate()
self.assertStderrEqual(stderr, b'')
returncode = p.wait()
self.assertNotEqual(returncode, 0)
def test_send_signal(self):
self._kill_process('send_signal', signal.SIGTERM)
def test_kill(self):
self._kill_process('kill')
def test_terminate(self):
self._kill_process('terminate')
# The module says:
# "NB This only works (and is only relevant) for UNIX."
#
# Actually, getoutput should work on any platform with an os.popen, but
# I'll take the comment as given, and skip this suite.
@unittest.skipUnless(os.name == 'posix', "only relevant for UNIX")
class CommandTests(unittest.TestCase):
def test_getoutput(self):
self.assertEqual(subprocess.getoutput('echo xyzzy'), 'xyzzy')
self.assertEqual(subprocess.getstatusoutput('echo xyzzy'),
(0, 'xyzzy'))
# we use mkdtemp in the next line to create an empty directory
# under our exclusive control; from that, we can invent a pathname
# that we _know_ won't exist. This is guaranteed to fail.
dir = None
try:
dir = tempfile.mkdtemp()
name = os.path.join(dir, "foo")
status, output = subprocess.getstatusoutput('cat ' + name)
self.assertNotEqual(status, 0)
finally:
if dir is not None:
os.rmdir(dir)
@unittest.skipUnless(getattr(subprocess, '_has_poll', False),
"poll system call not supported")
class ProcessTestCaseNoPoll(ProcessTestCase):
def setUp(self):
subprocess._has_poll = False
ProcessTestCase.setUp(self)
def tearDown(self):
subprocess._has_poll = True
ProcessTestCase.tearDown(self)
@unittest.skipUnless(getattr(subprocess, '_posixsubprocess', False),
"_posixsubprocess extension module not found.")
class ProcessTestCasePOSIXPurePython(ProcessTestCase, POSIXProcessTestCase):
def setUp(self):
subprocess._posixsubprocess = None
ProcessTestCase.setUp(self)
POSIXProcessTestCase.setUp(self)
def tearDown(self):
subprocess._posixsubprocess = sys.modules['_posixsubprocess']
POSIXProcessTestCase.tearDown(self)
ProcessTestCase.tearDown(self)
class HelperFunctionTests(unittest.TestCase):
@unittest.skipIf(mswindows, "errno and EINTR make no sense on windows")
def test_eintr_retry_call(self):
record_calls = []
def fake_os_func(*args):
record_calls.append(args)
if len(record_calls) == 2:
raise OSError(errno.EINTR, "fake interrupted system call")
return tuple(reversed(args))
self.assertEqual((999, 256),
subprocess._eintr_retry_call(fake_os_func, 256, 999))
self.assertEqual([(256, 999)], record_calls)
# This time there will be an EINTR so it will loop once.
self.assertEqual((666,),
subprocess._eintr_retry_call(fake_os_func, 666))
self.assertEqual([(256, 999), (666,), (666,)], record_calls)
@unittest.skipUnless(mswindows, "Windows-specific tests")
class CommandsWithSpaces (BaseTestCase):
def setUp(self):
super().setUp()
f, fname = mkstemp(".py", "te st")
self.fname = fname.lower ()
os.write(f, b"import sys;"
b"sys.stdout.write('%d %s' % (len(sys.argv), [a.lower () for a in sys.argv]))"
)
os.close(f)
def tearDown(self):
os.remove(self.fname)
super().tearDown()
def with_spaces(self, *args, **kwargs):
kwargs['stdout'] = subprocess.PIPE
p = subprocess.Popen(*args, **kwargs)
self.assertEqual(
p.stdout.read ().decode("mbcs"),
"2 [%r, 'ab cd']" % self.fname
)
def test_shell_string_with_spaces(self):
# call() function with string argument with spaces on Windows
self.with_spaces('"%s" "%s" "%s"' % (sys.executable, self.fname,
"ab cd"), shell=1)
def test_shell_sequence_with_spaces(self):
# call() function with sequence argument with spaces on Windows
self.with_spaces([sys.executable, self.fname, "ab cd"], shell=1)
def test_noshell_string_with_spaces(self):
# call() function with string argument with spaces on Windows
self.with_spaces('"%s" "%s" "%s"' % (sys.executable, self.fname,
"ab cd"))
def test_noshell_sequence_with_spaces(self):
# call() function with sequence argument with spaces on Windows
self.with_spaces([sys.executable, self.fname, "ab cd"])
def test_main():
unit_tests = (ProcessTestCase,
POSIXProcessTestCase,
Win32ProcessTestCase,
ProcessTestCasePOSIXPurePython,
CommandTests,
ProcessTestCaseNoPoll,
HelperFunctionTests,
CommandsWithSpaces)
support.run_unittest(*unit_tests)
support.reap_children()
if __name__ == "__main__":
test_main()