~launchpad-pqm/launchpad/devel

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
# Copyright 2011 Canonical Ltd.  This software is licensed under the
# GNU Affero General Public License version 3 (see the file LICENSE).

"""Execute commands in parallel sub-processes."""

__metaclass__ = type
__all__ = [
    'CommandSpawner',
    'OutputLineHandler',
    'ReturnCodeReceiver',
    ]

import errno
from fcntl import (
    F_GETFL,
    F_SETFL,
    fcntl,
    )
from os import O_NONBLOCK
import select
import subprocess


def get_process_output_files(process):
    """Return the files we watch for output coming from `process`."""
    return [
        process.stdout,
        process.stderr,
        ]


def make_files_nonblocking(files):
    """Put each of `files` in non-blocking mode.

    This allows the `CommandSpawner` to read all available output from a
    process without blocking until the process completes.
    """
    for this_file in files:
        fcntl(this_file, F_SETFL, fcntl(this_file, F_GETFL) | O_NONBLOCK)


def has_pending_output(poll_event):
    """Does the given event mask from `poll` indicate there's data to read?"""
    input_mask = (select.POLLIN | select.POLLPRI)
    return (poll_event & input_mask) != 0


def has_terminated(poll_event):
    """Does the given event mask from `poll` indicate process death?"""
    death_mask = (select.POLLERR | select.POLLHUP | select.POLLNVAL)
    return (poll_event & death_mask) != 0


STDOUT = 1
STDERR = 2
COMPLETION = 3


class CommandSpawner:
    """Simple manager to execute commands in parallel.

    Lets you run commands in sub-processes that will run simulaneously.
    The CommandSpawner looks for output from the running processes, and
    manages their cleanup.

    The typical usage pattern is:

    >>> spawner = CommandSpawner()
    >>> spawner.start(["echo", "One parallel process"])
    >>> spawner.start(["echo", "Another parallel process"])
    >>> spawner.complete()

    There are facilities for processing output and error output from the
    sub-processes, as well as dealing with success and failure.  You can
    pass callbacks to the `start` method, to be called when these events
    occur.

    As yet there is no facility for feeding input to the processes.
    """

    def __init__(self):
        self.running_processes = {}
        self.poll = select.poll()

    def start(self, command, stdout_handler=None, stderr_handler=None,
              completion_handler=None):
        """Run `command` in a sub-process.

        This starts the command, but does not wait for it to complete.
        Instead of waiting for completion, you can pass handlers that
        will be called when certain events occur.

        :param command: Command line to execute in a sub-process.  May be
            either a string (for a single executable name) or a list of
            strings (for an executable name plus arguments).
        :param stdout_handler: Callback to handle output received from the
            sub-process.  Must take the output as its sole argument.  May be
            called any number of times as output comes in.
        :param stderr_handler: Callback to handle error output received from
            the sub-process.  Must take the output as its sole argument.  May
            be called any number of times as output comes in.
        :param completion_handler: Callback to be invoked, exactly once, when
            the sub-process exits.  Must take `command`'s numeric return code
            as its sole argument.
        """
        process = self._spawn(command)
        handlers = {
            STDOUT: stdout_handler,
            STDERR: stderr_handler,
            COMPLETION: completion_handler,
        }
        self.running_processes[process] = handlers
        pipes = get_process_output_files(process)
        for pipe in pipes:
            self.poll.register(pipe, select.POLLIN | select.POLLPRI)
        make_files_nonblocking(pipes)

    def communicate(self):
        """Execute one iteration of the main event loop.  Blocks."""
        # Poll for output, but also wake up periodically to check for
        # completed processes.
        milliseconds = 1
        poll_result = self.poll.poll(milliseconds)

        # Map each file descriptor to its poll events.
        events_by_fd = dict(poll_result)

        # Iterate over a copy of the processes list: we may be removing
        # items from the original as processes complete.
        processes = self.running_processes.keys()
        for process in processes:
            self._service(process, events_by_fd)
            if process.returncode is not None:
                # Process has completed.  Remove it.
                try:
                    self._handle(process, COMPLETION, process.returncode)
                finally:
                    for pipe in get_process_output_files(process):
                        self.poll.unregister(pipe)
                    del self.running_processes[process]

    def complete(self):
        """Run `self.communicate` until all sub-processes have completed."""
        while len(self.running_processes) > 0:
            self.communicate()

    def kill(self):
        """Kill any remaining child processes.

        You'll still need to call `complete` to make sure that the child
        processes are cleaned up.  Until then, they will stay around as
        zombies.
        """
        for process in self.running_processes.iterkeys():
            process.terminate()

    def _spawn(self, command):
        """Spawn a sub-process for `command`.  Overridable in tests."""
        return subprocess.Popen(
            command, stdout=subprocess.PIPE, stderr=subprocess.PIPE,
            close_fds=True)

    def _handle(self, process, event, *args):
        """If we have a handler for `event` on `process`, call it."""
        process_handlers = self.running_processes[process]
        handler = process_handlers.get(event)
        if handler is not None:
            handler(*args)

    def _read(self, process, pipe_file, event):
        """Read output from `pipe_file`."""
        try:
            output = pipe_file.read()
        except IOError, e:
            # "Resource temporarily unavailable"--not an error really,
            # just means there's nothing to read.
            if e.errno != errno.EAGAIN:
                raise
        else:
            if len(output) > 0:
                self._handle(process, event, output)

    def _service(self, process, events_by_fd):
        """Service `process`."""
        stdout_events = events_by_fd.get(process.stdout.fileno(), 0)
        stderr_events = events_by_fd.get(process.stderr.fileno(), 0)
        if has_pending_output(stdout_events):
            self._read(process, process.stdout, STDOUT)
        if has_pending_output(stderr_events):
            self._read(process, process.stderr, STDERR)
        if has_terminated(stdout_events):
            process.wait()


class OutputLineHandler:
    """Collect and handle lines of output from a sub-process.

    Output received from a sub-process may not be neatly broken down by
    line.  This class collects them into lines and processes them one by
    one.  If desired, it can also add a prefix to each.
    """

    def __init__(self, line_processor, prefix=""):
        """Set up an output line handler.

        :param line_processor: A callback to be invoked for each line of
            output received.  Will receive exactly one argument: a single
            nonempty line of text, without the trailing newline.
        :param prefix: An optional string to be prefixed to each line of
            output before it is sent into the `line_processor`.
        """
        self.line_processor = line_processor
        self.prefix = prefix
        self.incomplete_buffer = ""

    def process_line(self, line):
        """Process a single line of output."""
        if line != "":
            self.line_processor("%s%s" % (self.prefix, line))

    def __call__(self, output):
        """Process multi-line output.

        Any trailing text not (yet) terminated with a newline is buffered.
        """
        lines = (self.incomplete_buffer + output).split("\n")
        if len(lines) > 0:
            self.incomplete_buffer = lines[-1]
            for line in lines[:-1]:
                self.process_line(line)

    def finalize(self):
        """Process the remaining incomplete line, if any."""
        if self.incomplete_buffer != "":
            self.process_line(self.incomplete_buffer)
            self.incomplete_buffer = ""


class ReturnCodeReceiver:
    """A minimal completion handler for `CommandSpawner` processes.

    Does nothing but collect the command's return code.

    :ivar returncode: The numerical return code retrieved from the
        process.  Stays None until the process completes.
    """

    returncode = None

    def __call__(self, returncode):
        self.returncode = returncode