| # Copyright (C) 2010 Google Inc. All rights reserved. |
| # Copyright (C) 2009 Daniel Bates (dbates@intudata.com). All rights reserved. |
| # |
| # Redistribution and use in source and binary forms, with or without |
| # modification, are permitted provided that the following conditions are |
| # met: |
| # |
| # * Redistributions of source code must retain the above copyright |
| # notice, this list of conditions and the following disclaimer. |
| # * Redistributions in binary form must reproduce the above |
| # copyright notice, this list of conditions and the following disclaimer |
| # in the documentation and/or other materials provided with the |
| # distribution. |
| # * Neither the name of Google Inc. nor the names of its |
| # contributors may be used to endorse or promote products derived from |
| # this software without specific prior written permission. |
| # |
| # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
| # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
| # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
| # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
| # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| |
| import os |
| import errno |
| import signal |
| import subprocess |
| import sys |
| import time |
| import unittest |
| |
| # Since we execute this script directly as part of the unit tests, we need to ensure |
| # that Tools/Scripts is in sys.path for the next imports to work correctly. |
| script_dir = os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))) |
| if script_dir not in sys.path: |
| sys.path.append(script_dir) |
| third_party_py = os.path.join(script_dir, "webkitpy", "thirdparty", "autoinstalled") |
| if third_party_py not in sys.path: |
| sys.path.append(third_party_py) |
| |
| from webkitpy.common.system.executive import Executive, ScriptError |
| from webkitpy.common.system.filesystem_mock import MockFileSystem |
| from webkitpy.common import unicode_compatibility |
| |
| |
| class ScriptErrorTest(unittest.TestCase): |
| def test_message_with_output(self): |
| error = ScriptError('My custom message!', '', -1) |
| self.assertEqual(error.message_with_output(), 'My custom message!') |
| error = ScriptError('My custom message!', '', -1, 'My output.') |
| self.assertEqual(error.message_with_output(), 'My custom message!\n\nMy output.') |
| error = ScriptError('', 'my_command!', -1, 'My output.', '/Users/username/blah') |
| self.assertEqual(error.message_with_output(), 'Failed to run "\'my_command!\'" exit_code: -1 cwd: /Users/username/blah\n\nMy output.') |
| error = ScriptError('', 'my_command!', -1, 'ab' + '1' * 499) |
| self.assertEqual(error.message_with_output(), 'Failed to run "\'my_command!\'" exit_code: -1\n\nLast 500 characters of output:\nb' + '1' * 499) |
| |
| def test_message_with_tuple(self): |
| error = ScriptError('', ('my', 'command'), -1, 'My output.', '/Users/username/blah') |
| self.assertEqual(error.message_with_output(), 'Failed to run "(\'my\', \'command\')" exit_code: -1 cwd: /Users/username/blah\n\nMy output.') |
| |
| |
| def never_ending_command(): |
| """Arguments for a command that will never end (useful for testing process |
| killing). It should be a process that is unlikely to already be running |
| because all instances will be killed.""" |
| if sys.platform.startswith('win'): |
| return ['wmic'] |
| return ['yes'] |
| |
| |
| def command_line(cmd, *args): |
| return [sys.executable, __file__, '--' + cmd] + list(args) |
| |
| |
| class ExecutiveTest(unittest.TestCase): |
| def assert_interpreter_for_content(self, intepreter, content): |
| fs = MockFileSystem() |
| |
| tempfile, temp_name = fs.open_binary_tempfile('') |
| tempfile.write(content) |
| tempfile.close() |
| file_interpreter = Executive.interpreter_for_script(temp_name, fs) |
| |
| self.assertEqual(file_interpreter, intepreter) |
| |
| def test_interpreter_for_script(self): |
| self.assert_interpreter_for_content(None, '') |
| self.assert_interpreter_for_content(None, 'abcd\nefgh\nijklm') |
| self.assert_interpreter_for_content(None, '##/usr/bin/perl') |
| self.assert_interpreter_for_content('perl', '#!/usr/bin/env perl') |
| self.assert_interpreter_for_content('perl', '#!/usr/bin/env perl\nfirst\nsecond') |
| self.assert_interpreter_for_content('perl', '#!/usr/bin/perl') |
| self.assert_interpreter_for_content('perl', '#!/usr/bin/perl -w') |
| self.assert_interpreter_for_content(sys.executable, '#!/usr/bin/env python') |
| self.assert_interpreter_for_content(sys.executable, '#!/usr/bin/env python\nfirst\nsecond') |
| self.assert_interpreter_for_content(sys.executable, '#!/usr/bin/python') |
| self.assert_interpreter_for_content('ruby', '#!/usr/bin/env ruby') |
| self.assert_interpreter_for_content('ruby', '#!/usr/bin/env ruby\nfirst\nsecond') |
| self.assert_interpreter_for_content('ruby', '#!/usr/bin/ruby') |
| |
| def test_run_command_with_bad_command(self): |
| self.assertRaises(OSError, lambda: Executive().run_command(["foo_bar_command_blah"], ignore_errors=True, return_exit_code=True)) |
| self.assertRaises(OSError, lambda: Executive().run_and_throw_if_fail(["foo_bar_command_blah"], quiet=True)) |
| self.assertRaises(ScriptError, lambda: Executive().run_command(['python', '-c', 'import sys; sys.exit(1)'])) |
| self.assertRaises(ScriptError, lambda: Executive().run_and_throw_if_fail(['python', '-c', 'import sys; sys.exit(1)'], quiet=True)) |
| |
| def test_run_command_args_type(self): |
| executive = Executive() |
| self.assertRaises(AssertionError, executive.run_command, "echo") |
| self.assertRaises(AssertionError, executive.run_command, u"echo") |
| executive.run_command(command_line('echo', 'foo')) |
| executive.run_command(tuple(command_line('echo', 'foo'))) |
| |
| def test_auto_stringify_args(self): |
| executive = Executive() |
| executive.run_command(command_line('echo', 1)) |
| with executive.popen(command_line('echo', 1), stdout=executive.PIPE) as process: |
| process.wait() |
| self.assertEqual('echo 1', executive.command_for_printing(['echo', 1])) |
| |
| def test_popen_args(self): |
| executive = Executive() |
| # Explicitly naming the 'args' argument should not thow an exception. |
| with executive.popen(args=command_line('echo', 1), stdout=executive.PIPE) as process: |
| process.wait() |
| |
| def test_run_command_with_unicode(self): |
| """Validate that it is safe to pass unicode() objects |
| to Executive.run* methods, and they will return unicode() |
| objects by default unless decode_output=False""" |
| unicode_tor_input = u"WebKit \u2661 Tor Arne Vestb\u00F8!" |
| if sys.platform.startswith('win'): |
| encoding = 'mbcs' |
| else: |
| encoding = 'utf-8' |
| encoded_tor = unicode_compatibility.encode_if_necessary(unicode_tor_input, encoding) |
| # On Windows, we expect the unicode->mbcs->unicode roundtrip to be |
| # lossy. On other platforms, we expect a lossless roundtrip. |
| if sys.platform.startswith('win'): |
| unicode_tor_output = unicode_compatibility.decode_if_necessary(encoded_tor, encoding) |
| else: |
| unicode_tor_output = unicode_tor_input |
| |
| executive = Executive() |
| |
| output = executive.run_command(command_line('cat'), input=unicode_tor_input) |
| self.assertEqual(output, unicode_tor_output) |
| |
| output = executive.run_command(command_line('echo', unicode_tor_input)) |
| self.assertEqual(output, unicode_tor_output) |
| |
| output = executive.run_command(command_line('echo', unicode_tor_input), decode_output=False) |
| self.assertEqual(output, encoded_tor) |
| |
| # Make sure that str() input also works. |
| output = executive.run_command(command_line('cat'), input=encoded_tor, decode_output=False) |
| self.assertEqual(output, encoded_tor) |
| |
| # FIXME: We should only have one run* method to test |
| output = executive.run_and_throw_if_fail(command_line('echo', unicode_tor_input), quiet=True) |
| self.assertEqual(output, unicode_tor_output) |
| |
| output = executive.run_and_throw_if_fail(command_line('echo', unicode_tor_input), quiet=True, decode_output=False) |
| self.assertEqual(output, encoded_tor) |
| |
| def serial_test_kill_process(self): |
| executive = Executive() |
| with executive.popen(never_ending_command(), stdout=subprocess.PIPE) as process: |
| self.assertEqual(process.poll(), None) # Process is running |
| executive.kill_process(process.pid) |
| # Note: Can't use a ternary since signal.SIGKILL is undefined for sys.platform == "win32" |
| if sys.platform.startswith('win'): |
| # FIXME: https://bugs.webkit.org/show_bug.cgi?id=54790 |
| # We seem to get either 0 or 1 here for some reason. |
| self.assertIn(process.wait(), (0, 1)) |
| elif sys.platform == "cygwin": |
| # FIXME: https://bugs.webkit.org/show_bug.cgi?id=98196 |
| # cygwin seems to give us either SIGABRT or SIGKILL |
| # Native Windows (via Cygwin) returns ENOTBLK (-15) |
| self.assertIn(process.wait(), (-signal.SIGABRT, -signal.SIGKILL, -15)) |
| else: |
| expected_exit_code = -signal.SIGTERM |
| self.assertEqual(process.wait(), expected_exit_code) |
| |
| # Killing again should fail silently. |
| executive.kill_process(process.pid) |
| |
| def serial_test_kill_all(self): |
| executive = Executive() |
| with executive.popen(never_ending_command(), stdout=subprocess.PIPE) as process: |
| self.assertIsNone(process.poll()) # Process is running |
| executive.kill_all(never_ending_command()[0]) |
| # Note: Can't use a ternary since signal.SIGTERM is undefined for sys.platform == "win32" |
| if sys.platform == "cygwin": |
| expected_exit_code = 0 # os.kill results in exit(0) for this process. |
| self.assertEqual(process.wait(), expected_exit_code) |
| elif sys.platform.startswith('win'): |
| # FIXME: https://bugs.webkit.org/show_bug.cgi?id=54790 |
| # We seem to get either 0 or 1 here for some reason. |
| self.assertIn(process.wait(), (0, 1)) |
| else: |
| expected_exit_code = -signal.SIGTERM |
| self.assertEqual(process.wait(), expected_exit_code) |
| # Killing again should fail silently. |
| executive.kill_all(never_ending_command()[0]) |
| |
| def _assert_windows_image_name(self, name, expected_windows_name): |
| executive = Executive() |
| windows_name = executive._windows_image_name(name) |
| self.assertEqual(windows_name, expected_windows_name) |
| |
| def test_windows_image_name(self): |
| self._assert_windows_image_name("foo", "foo.exe") |
| self._assert_windows_image_name("foo.exe", "foo.exe") |
| self._assert_windows_image_name("foo.com", "foo.com") |
| # If the name looks like an extension, even if it isn't |
| # supposed to, we have no choice but to return the original name. |
| self._assert_windows_image_name("foo.baz", "foo.baz") |
| self._assert_windows_image_name("foo.baz.exe", "foo.baz.exe") |
| |
| def serial_test_check_running_pid(self): |
| executive = Executive() |
| self.assertTrue(executive.check_running_pid(os.getpid())) |
| # Maximum pid number on Linux is 32768 by default |
| self.assertFalse(executive.check_running_pid(100000)) |
| |
| def serial_test_running_pids(self): |
| if sys.platform.startswith('win') or sys.platform == "cygwin": |
| return # This function isn't implemented on Windows yet. |
| |
| executive = Executive() |
| pids = executive.running_pids() |
| self.assertIn(os.getpid(), pids) |
| |
| def serial_test_run_in_parallel(self): |
| # We run this test serially to avoid overloading the machine and throwing off the timing. |
| |
| if sys.platform.startswith('win') or sys.platform == "cygwin": |
| return # This function isn't implemented properly on windows yet. |
| import multiprocessing |
| |
| NUM_PROCESSES = 4 |
| DELAY_SECS = 0.25 |
| cmd_line = [sys.executable, '-c', 'import time; time.sleep(%f); print("hello")' % DELAY_SECS] |
| cwd = os.getcwd() |
| commands = [tuple([cmd_line, cwd])] * NUM_PROCESSES |
| start = time.time() |
| command_outputs = Executive().run_in_parallel(commands, processes=NUM_PROCESSES) |
| done = time.time() |
| self.assertTrue(done - start < NUM_PROCESSES * DELAY_SECS) |
| self.assertEqual([output[1] for output in command_outputs], [b'hello\n'] * NUM_PROCESSES) |
| self.assertEqual([], multiprocessing.active_children()) |
| |
| def test_run_in_parallel_assert_nonempty(self): |
| self.assertRaises(AssertionError, Executive().run_in_parallel, []) |
| |
| |
| def main(platform, stdin, stdout, cmd, args): |
| if sys.platform.startswith('win') and hasattr(stdout, 'fileno'): |
| import msvcrt |
| msvcrt.setmode(stdout.fileno(), os.O_BINARY) |
| if cmd == '--cat': |
| stdout.write(stdin.read()) |
| elif cmd == '--echo': |
| stdout.write(' '.join(args)) |
| return 0 |
| |
| if __name__ == '__main__' and len(sys.argv) > 1 and sys.argv[1] in ('--cat', '--echo'): |
| sys.exit(main(sys.platform, sys.stdin, sys.stdout, sys.argv[1], sys.argv[2:])) |