maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1 | #!/usr/bin/env python |
| 2 | # Copyright (c) 2012 The Chromium Authors. All rights reserved. |
| 3 | # Use of this source code is governed by a BSD-style license that can be |
| 4 | # found in the LICENSE file. |
| 5 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 6 | """Reads a .isolated, creates a tree of hardlinks and runs the test. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 7 | |
| 8 | Keeps a local cache. |
| 9 | """ |
| 10 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 11 | import cookielib |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 12 | import ctypes |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 13 | import functools |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 14 | import hashlib |
csharp@chromium.org | a110d79 | 2013-01-07 16:16:16 +0000 | [diff] [blame] | 15 | import httplib |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 16 | import itertools |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 17 | import json |
csharp@chromium.org | bfb9874 | 2013-03-26 20:28:36 +0000 | [diff] [blame] | 18 | import locale |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 19 | import logging |
csharp@chromium.org | ff2a466 | 2012-11-21 20:49:32 +0000 | [diff] [blame] | 20 | import logging.handlers |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 21 | import math |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 22 | import optparse |
| 23 | import os |
| 24 | import Queue |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 25 | import random |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 26 | import re |
| 27 | import shutil |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 28 | import socket |
| 29 | import ssl |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 30 | import stat |
| 31 | import subprocess |
| 32 | import sys |
| 33 | import tempfile |
| 34 | import threading |
| 35 | import time |
| 36 | import urllib |
csharp@chromium.org | a92403f | 2012-11-20 15:13:59 +0000 | [diff] [blame] | 37 | import urllib2 |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 38 | import urlparse |
csharp@chromium.org | a92403f | 2012-11-20 15:13:59 +0000 | [diff] [blame] | 39 | import zlib |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 40 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 41 | from utils import lru |
vadimsh@chromium.org | b074b16 | 2013-08-22 17:55:46 +0000 | [diff] [blame] | 42 | from utils import threading_utils |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 43 | from utils import zip_package |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 44 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 45 | # Try to import 'upload' module used by AppEngineService for authentication. |
| 46 | # If it is not there, app engine authentication support will be disabled. |
| 47 | try: |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 48 | from third_party.rietveld import upload |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 49 | # Hack out upload logging.info() |
| 50 | upload.logging = logging.getLogger('upload') |
| 51 | # Mac pylint choke on this line. |
| 52 | upload.logging.setLevel(logging.WARNING) # pylint: disable=E1103 |
| 53 | except ImportError: |
| 54 | upload = None |
| 55 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 56 | |
vadimsh@chromium.org | 8507106 | 2013-08-21 23:37:45 +0000 | [diff] [blame] | 57 | # Absolute path to this file (can be None if running from zip on Mac). |
| 58 | THIS_FILE_PATH = os.path.abspath(__file__) if __file__ else None |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 59 | |
| 60 | # Directory that contains this file (might be inside zip package). |
vadimsh@chromium.org | 8507106 | 2013-08-21 23:37:45 +0000 | [diff] [blame] | 61 | BASE_DIR = os.path.dirname(THIS_FILE_PATH) if __file__ else None |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 62 | |
| 63 | # Directory that contains currently running script file. |
| 64 | MAIN_DIR = os.path.dirname(os.path.abspath(zip_package.get_main_script_path())) |
| 65 | |
maruel@chromium.org | 6b365dc | 2012-10-18 19:17:56 +0000 | [diff] [blame] | 66 | # Types of action accepted by link_file(). |
maruel@chromium.org | ba6489b | 2013-07-11 20:23:33 +0000 | [diff] [blame] | 67 | HARDLINK, HARDLINK_WITH_FALLBACK, SYMLINK, COPY = range(1, 5) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 68 | |
| 69 | RE_IS_SHA1 = re.compile(r'^[a-fA-F0-9]{40}$') |
| 70 | |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 71 | # The file size to be used when we don't know the correct file size, |
| 72 | # generally used for .isolated files. |
| 73 | UNKNOWN_FILE_SIZE = None |
| 74 | |
csharp@chromium.org | a92403f | 2012-11-20 15:13:59 +0000 | [diff] [blame] | 75 | # The size of each chunk to read when downloading and unzipping files. |
| 76 | ZIPPED_FILE_CHUNK = 16 * 1024 |
| 77 | |
csharp@chromium.org | ff2a466 | 2012-11-21 20:49:32 +0000 | [diff] [blame] | 78 | # The name of the log file to use. |
| 79 | RUN_ISOLATED_LOG_FILE = 'run_isolated.log' |
| 80 | |
csharp@chromium.org | e217f30 | 2012-11-22 16:51:53 +0000 | [diff] [blame] | 81 | # The name of the log to use for the run_test_cases.py command |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 82 | RUN_TEST_CASES_LOG = 'run_test_cases.log' |
csharp@chromium.org | e217f30 | 2012-11-22 16:51:53 +0000 | [diff] [blame] | 83 | |
csharp@chromium.org | 9c59ff1 | 2012-12-12 02:32:29 +0000 | [diff] [blame] | 84 | # The delay (in seconds) to wait between logging statements when retrieving |
| 85 | # the required files. This is intended to let the user (or buildbot) know that |
| 86 | # the program is still running. |
| 87 | DELAY_BETWEEN_UPDATES_IN_SECS = 30 |
| 88 | |
vadimsh@chromium.org | 5db0f4f | 2013-07-04 13:57:02 +0000 | [diff] [blame] | 89 | # Maximum expected delay (in seconds) between successive file fetches |
| 90 | # in run_tha_test. If it takes longer than that, a deadlock might be happening |
| 91 | # and all stack frames for all threads are dumped to log. |
| 92 | DEADLOCK_TIMEOUT = 5 * 60 |
| 93 | |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 94 | # The name of the key to store the count of url attempts. |
| 95 | COUNT_KEY = 'UrlOpenAttempt' |
| 96 | |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 97 | # Default maximum number of attempts to trying opening a url before aborting. |
| 98 | URL_OPEN_MAX_ATTEMPTS = 30 |
| 99 | # Default timeout when retrying. |
| 100 | URL_OPEN_TIMEOUT = 6*60. |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 101 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 102 | # Read timeout in seconds for downloads from isolate storage. If there's no |
| 103 | # response from the server within this timeout whole download will be aborted. |
| 104 | DOWNLOAD_READ_TIMEOUT = 60 |
| 105 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 106 | # Global (for now) map: server URL (http://example.com) -> HttpService instance. |
| 107 | # Used by get_http_service to cache HttpService instances. |
| 108 | _http_services = {} |
| 109 | _http_services_lock = threading.Lock() |
| 110 | |
maruel@chromium.org | 9e9ceaa | 2013-04-05 15:42:42 +0000 | [diff] [blame] | 111 | # Used by get_flavor(). |
| 112 | FLAVOR_MAPPING = { |
| 113 | 'cygwin': 'win', |
| 114 | 'win32': 'win', |
| 115 | 'darwin': 'mac', |
| 116 | 'sunos5': 'solaris', |
| 117 | 'freebsd7': 'freebsd', |
| 118 | 'freebsd8': 'freebsd', |
| 119 | } |
| 120 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 121 | |
| 122 | class ConfigError(ValueError): |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 123 | """Generic failure to load a .isolated file.""" |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 124 | pass |
| 125 | |
| 126 | |
| 127 | class MappingError(OSError): |
| 128 | """Failed to recreate the tree.""" |
| 129 | pass |
| 130 | |
| 131 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 132 | class TimeoutError(IOError): |
| 133 | """Timeout while reading HTTP response.""" |
| 134 | |
| 135 | def __init__(self, inner_exc=None): |
| 136 | super(TimeoutError, self).__init__(str(inner_exc or 'Timeout')) |
| 137 | self.inner_exc = inner_exc |
| 138 | |
| 139 | |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 140 | def get_as_zip_package(executable=True): |
| 141 | """Returns ZipPackage with this module and all its dependencies. |
| 142 | |
| 143 | If |executable| is True will store run_isolated.py as __main__.py so that |
| 144 | zip package is directly executable be python. |
| 145 | """ |
| 146 | # Building a zip package when running from another zip package is |
| 147 | # unsupported and probably unneeded. |
| 148 | assert not zip_package.is_zipped_module(sys.modules[__name__]) |
vadimsh@chromium.org | 8507106 | 2013-08-21 23:37:45 +0000 | [diff] [blame] | 149 | assert THIS_FILE_PATH |
| 150 | assert BASE_DIR |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 151 | package = zip_package.ZipPackage(root=BASE_DIR) |
| 152 | package.add_python_file(THIS_FILE_PATH, '__main__.py' if executable else None) |
| 153 | package.add_directory(os.path.join(BASE_DIR, 'third_party')) |
| 154 | package.add_directory(os.path.join(BASE_DIR, 'utils')) |
| 155 | return package |
| 156 | |
| 157 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 158 | def get_flavor(): |
| 159 | """Returns the system default flavor. Copied from gyp/pylib/gyp/common.py.""" |
maruel@chromium.org | 9e9ceaa | 2013-04-05 15:42:42 +0000 | [diff] [blame] | 160 | return FLAVOR_MAPPING.get(sys.platform, 'linux') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 161 | |
| 162 | |
csharp@chromium.org | bfb9874 | 2013-03-26 20:28:36 +0000 | [diff] [blame] | 163 | def fix_default_encoding(): |
| 164 | """Forces utf8 solidly on all platforms. |
| 165 | |
| 166 | By default python execution environment is lazy and defaults to ascii |
| 167 | encoding. |
| 168 | |
| 169 | http://uucode.com/blog/2007/03/23/shut-up-you-dummy-7-bit-python/ |
| 170 | """ |
| 171 | if sys.getdefaultencoding() == 'utf-8': |
| 172 | return False |
| 173 | |
| 174 | # Regenerate setdefaultencoding. |
| 175 | reload(sys) |
| 176 | # Module 'sys' has no 'setdefaultencoding' member |
| 177 | # pylint: disable=E1101 |
| 178 | sys.setdefaultencoding('utf-8') |
| 179 | for attr in dir(locale): |
| 180 | if attr[0:3] != 'LC_': |
| 181 | continue |
| 182 | aref = getattr(locale, attr) |
| 183 | try: |
| 184 | locale.setlocale(aref, '') |
| 185 | except locale.Error: |
| 186 | continue |
| 187 | try: |
| 188 | lang = locale.getlocale(aref)[0] |
| 189 | except (TypeError, ValueError): |
| 190 | continue |
| 191 | if lang: |
| 192 | try: |
| 193 | locale.setlocale(aref, (lang, 'UTF-8')) |
| 194 | except locale.Error: |
| 195 | os.environ[attr] = lang + '.UTF-8' |
| 196 | try: |
| 197 | locale.setlocale(locale.LC_ALL, '') |
| 198 | except locale.Error: |
| 199 | pass |
| 200 | return True |
| 201 | |
| 202 | |
maruel@chromium.org | 46e61cc | 2013-03-25 19:55:34 +0000 | [diff] [blame] | 203 | class Unbuffered(object): |
| 204 | """Disable buffering on a file object.""" |
| 205 | def __init__(self, stream): |
| 206 | self.stream = stream |
| 207 | |
| 208 | def write(self, data): |
| 209 | self.stream.write(data) |
| 210 | if '\n' in data: |
| 211 | self.stream.flush() |
| 212 | |
| 213 | def __getattr__(self, attr): |
| 214 | return getattr(self.stream, attr) |
| 215 | |
| 216 | |
| 217 | def disable_buffering(): |
| 218 | """Makes this process and child processes stdout unbuffered.""" |
| 219 | if not os.environ.get('PYTHONUNBUFFERED'): |
| 220 | # Since sys.stdout is a C++ object, it's impossible to do |
| 221 | # sys.stdout.write = lambda... |
| 222 | sys.stdout = Unbuffered(sys.stdout) |
| 223 | os.environ['PYTHONUNBUFFERED'] = 'x' |
| 224 | |
| 225 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 226 | def os_link(source, link_name): |
| 227 | """Add support for os.link() on Windows.""" |
| 228 | if sys.platform == 'win32': |
| 229 | if not ctypes.windll.kernel32.CreateHardLinkW( |
| 230 | unicode(link_name), unicode(source), 0): |
| 231 | raise OSError() |
| 232 | else: |
| 233 | os.link(source, link_name) |
| 234 | |
| 235 | |
| 236 | def readable_copy(outfile, infile): |
| 237 | """Makes a copy of the file that is readable by everyone.""" |
csharp@chromium.org | 59d116d | 2013-07-05 18:04:08 +0000 | [diff] [blame] | 238 | shutil.copy2(infile, outfile) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 239 | read_enabled_mode = (os.stat(outfile).st_mode | stat.S_IRUSR | |
| 240 | stat.S_IRGRP | stat.S_IROTH) |
| 241 | os.chmod(outfile, read_enabled_mode) |
| 242 | |
| 243 | |
| 244 | def link_file(outfile, infile, action): |
| 245 | """Links a file. The type of link depends on |action|.""" |
| 246 | logging.debug('Mapping %s to %s' % (infile, outfile)) |
maruel@chromium.org | ba6489b | 2013-07-11 20:23:33 +0000 | [diff] [blame] | 247 | if action not in (HARDLINK, HARDLINK_WITH_FALLBACK, SYMLINK, COPY): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 248 | raise ValueError('Unknown mapping action %s' % action) |
| 249 | if not os.path.isfile(infile): |
| 250 | raise MappingError('%s is missing' % infile) |
| 251 | if os.path.isfile(outfile): |
| 252 | raise MappingError( |
| 253 | '%s already exist; insize:%d; outsize:%d' % |
| 254 | (outfile, os.stat(infile).st_size, os.stat(outfile).st_size)) |
| 255 | |
| 256 | if action == COPY: |
| 257 | readable_copy(outfile, infile) |
| 258 | elif action == SYMLINK and sys.platform != 'win32': |
| 259 | # On windows, symlink are converted to hardlink and fails over to copy. |
maruel@chromium.org | f43e68b | 2012-10-15 20:23:10 +0000 | [diff] [blame] | 260 | os.symlink(infile, outfile) # pylint: disable=E1101 |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 261 | else: |
| 262 | try: |
| 263 | os_link(infile, outfile) |
maruel@chromium.org | ba6489b | 2013-07-11 20:23:33 +0000 | [diff] [blame] | 264 | except OSError as e: |
| 265 | if action == HARDLINK: |
| 266 | raise MappingError( |
| 267 | 'Failed to hardlink %s to %s: %s' % (infile, outfile, e)) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 268 | # Probably a different file system. |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 269 | logging.warning( |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 270 | 'Failed to hardlink, failing back to copy %s to %s' % ( |
| 271 | infile, outfile)) |
| 272 | readable_copy(outfile, infile) |
| 273 | |
| 274 | |
| 275 | def _set_write_bit(path, read_only): |
| 276 | """Sets or resets the executable bit on a file or directory.""" |
| 277 | mode = os.lstat(path).st_mode |
| 278 | if read_only: |
| 279 | mode = mode & 0500 |
| 280 | else: |
| 281 | mode = mode | 0200 |
| 282 | if hasattr(os, 'lchmod'): |
| 283 | os.lchmod(path, mode) # pylint: disable=E1101 |
| 284 | else: |
| 285 | if stat.S_ISLNK(mode): |
| 286 | # Skip symlink without lchmod() support. |
| 287 | logging.debug('Can\'t change +w bit on symlink %s' % path) |
| 288 | return |
| 289 | |
| 290 | # TODO(maruel): Implement proper DACL modification on Windows. |
| 291 | os.chmod(path, mode) |
| 292 | |
| 293 | |
| 294 | def make_writable(root, read_only): |
| 295 | """Toggle the writable bit on a directory tree.""" |
csharp@chromium.org | 837352f | 2013-01-17 21:17:03 +0000 | [diff] [blame] | 296 | assert os.path.isabs(root), root |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 297 | for dirpath, dirnames, filenames in os.walk(root, topdown=True): |
| 298 | for filename in filenames: |
| 299 | _set_write_bit(os.path.join(dirpath, filename), read_only) |
| 300 | |
| 301 | for dirname in dirnames: |
| 302 | _set_write_bit(os.path.join(dirpath, dirname), read_only) |
| 303 | |
| 304 | |
| 305 | def rmtree(root): |
| 306 | """Wrapper around shutil.rmtree() to retry automatically on Windows.""" |
| 307 | make_writable(root, False) |
| 308 | if sys.platform == 'win32': |
| 309 | for i in range(3): |
| 310 | try: |
| 311 | shutil.rmtree(root) |
| 312 | break |
| 313 | except WindowsError: # pylint: disable=E0602 |
| 314 | delay = (i+1)*2 |
| 315 | print >> sys.stderr, ( |
| 316 | 'The test has subprocess outliving it. Sleep %d seconds.' % delay) |
| 317 | time.sleep(delay) |
| 318 | else: |
| 319 | shutil.rmtree(root) |
| 320 | |
| 321 | |
| 322 | def is_same_filesystem(path1, path2): |
| 323 | """Returns True if both paths are on the same filesystem. |
| 324 | |
| 325 | This is required to enable the use of hardlinks. |
| 326 | """ |
| 327 | assert os.path.isabs(path1), path1 |
| 328 | assert os.path.isabs(path2), path2 |
| 329 | if sys.platform == 'win32': |
| 330 | # If the drive letter mismatches, assume it's a separate partition. |
| 331 | # TODO(maruel): It should look at the underlying drive, a drive letter could |
| 332 | # be a mount point to a directory on another drive. |
| 333 | assert re.match(r'^[a-zA-Z]\:\\.*', path1), path1 |
| 334 | assert re.match(r'^[a-zA-Z]\:\\.*', path2), path2 |
| 335 | if path1[0].lower() != path2[0].lower(): |
| 336 | return False |
| 337 | return os.stat(path1).st_dev == os.stat(path2).st_dev |
| 338 | |
| 339 | |
| 340 | def get_free_space(path): |
| 341 | """Returns the number of free bytes.""" |
| 342 | if sys.platform == 'win32': |
| 343 | free_bytes = ctypes.c_ulonglong(0) |
| 344 | ctypes.windll.kernel32.GetDiskFreeSpaceExW( |
| 345 | ctypes.c_wchar_p(path), None, None, ctypes.pointer(free_bytes)) |
| 346 | return free_bytes.value |
maruel@chromium.org | f43e68b | 2012-10-15 20:23:10 +0000 | [diff] [blame] | 347 | # For OSes other than Windows. |
| 348 | f = os.statvfs(path) # pylint: disable=E1101 |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 349 | return f.f_bfree * f.f_frsize |
| 350 | |
| 351 | |
| 352 | def make_temp_dir(prefix, root_dir): |
| 353 | """Returns a temporary directory on the same file system as root_dir.""" |
| 354 | base_temp_dir = None |
| 355 | if not is_same_filesystem(root_dir, tempfile.gettempdir()): |
| 356 | base_temp_dir = os.path.dirname(root_dir) |
| 357 | return tempfile.mkdtemp(prefix=prefix, dir=base_temp_dir) |
| 358 | |
| 359 | |
frankf@chromium.org | 3348ee0 | 2013-06-27 14:53:17 +0000 | [diff] [blame] | 360 | def load_isolated(content, os_flavor=None): |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 361 | """Verifies the .isolated file is valid and loads this object with the json |
| 362 | data. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 363 | """ |
| 364 | try: |
| 365 | data = json.loads(content) |
| 366 | except ValueError: |
| 367 | raise ConfigError('Failed to parse: %s...' % content[:100]) |
| 368 | |
| 369 | if not isinstance(data, dict): |
| 370 | raise ConfigError('Expected dict, got %r' % data) |
| 371 | |
| 372 | for key, value in data.iteritems(): |
| 373 | if key == 'command': |
| 374 | if not isinstance(value, list): |
| 375 | raise ConfigError('Expected list, got %r' % value) |
maruel@chromium.org | 89ad2db | 2012-12-12 14:29:22 +0000 | [diff] [blame] | 376 | if not value: |
| 377 | raise ConfigError('Expected non-empty command') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 378 | for subvalue in value: |
| 379 | if not isinstance(subvalue, basestring): |
| 380 | raise ConfigError('Expected string, got %r' % subvalue) |
| 381 | |
| 382 | elif key == 'files': |
| 383 | if not isinstance(value, dict): |
| 384 | raise ConfigError('Expected dict, got %r' % value) |
| 385 | for subkey, subvalue in value.iteritems(): |
| 386 | if not isinstance(subkey, basestring): |
| 387 | raise ConfigError('Expected string, got %r' % subkey) |
| 388 | if not isinstance(subvalue, dict): |
| 389 | raise ConfigError('Expected dict, got %r' % subvalue) |
| 390 | for subsubkey, subsubvalue in subvalue.iteritems(): |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 391 | if subsubkey == 'l': |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 392 | if not isinstance(subsubvalue, basestring): |
| 393 | raise ConfigError('Expected string, got %r' % subsubvalue) |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 394 | elif subsubkey == 'm': |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 395 | if not isinstance(subsubvalue, int): |
| 396 | raise ConfigError('Expected int, got %r' % subsubvalue) |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 397 | elif subsubkey == 'h': |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 398 | if not RE_IS_SHA1.match(subsubvalue): |
| 399 | raise ConfigError('Expected sha-1, got %r' % subsubvalue) |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 400 | elif subsubkey == 's': |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 401 | if not isinstance(subsubvalue, int): |
| 402 | raise ConfigError('Expected int, got %r' % subsubvalue) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 403 | else: |
| 404 | raise ConfigError('Unknown subsubkey %s' % subsubkey) |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 405 | if bool('h' in subvalue) and bool('l' in subvalue): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 406 | raise ConfigError( |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 407 | 'Did not expect both \'h\' (sha-1) and \'l\' (link), got: %r' % |
| 408 | subvalue) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 409 | |
| 410 | elif key == 'includes': |
| 411 | if not isinstance(value, list): |
| 412 | raise ConfigError('Expected list, got %r' % value) |
maruel@chromium.org | 89ad2db | 2012-12-12 14:29:22 +0000 | [diff] [blame] | 413 | if not value: |
| 414 | raise ConfigError('Expected non-empty includes list') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 415 | for subvalue in value: |
| 416 | if not RE_IS_SHA1.match(subvalue): |
| 417 | raise ConfigError('Expected sha-1, got %r' % subvalue) |
| 418 | |
| 419 | elif key == 'read_only': |
| 420 | if not isinstance(value, bool): |
| 421 | raise ConfigError('Expected bool, got %r' % value) |
| 422 | |
| 423 | elif key == 'relative_cwd': |
| 424 | if not isinstance(value, basestring): |
| 425 | raise ConfigError('Expected string, got %r' % value) |
| 426 | |
| 427 | elif key == 'os': |
frankf@chromium.org | 3348ee0 | 2013-06-27 14:53:17 +0000 | [diff] [blame] | 428 | expected_value = os_flavor or get_flavor() |
| 429 | if value != expected_value: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 430 | raise ConfigError( |
| 431 | 'Expected \'os\' to be \'%s\' but got \'%s\'' % |
frankf@chromium.org | 3348ee0 | 2013-06-27 14:53:17 +0000 | [diff] [blame] | 432 | (expected_value, value)) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 433 | |
| 434 | else: |
| 435 | raise ConfigError('Unknown key %s' % key) |
| 436 | |
| 437 | return data |
| 438 | |
| 439 | |
| 440 | def fix_python_path(cmd): |
| 441 | """Returns the fixed command line to call the right python executable.""" |
| 442 | out = cmd[:] |
| 443 | if out[0] == 'python': |
| 444 | out[0] = sys.executable |
| 445 | elif out[0].endswith('.py'): |
| 446 | out.insert(0, sys.executable) |
| 447 | return out |
| 448 | |
| 449 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 450 | def url_open(url, **kwargs): |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 451 | """Attempts to open the given url multiple times. |
| 452 | |
| 453 | |data| can be either: |
| 454 | -None for a GET request |
| 455 | -str for pre-encoded data |
| 456 | -list for data to be encoded |
| 457 | -dict for data to be encoded (COUNT_KEY will be added in this case) |
| 458 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 459 | Returns HttpResponse object, where the response may be read from, or None |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 460 | if it was unable to connect. |
| 461 | """ |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 462 | urlhost, urlpath = split_server_request_url(url) |
| 463 | service = get_http_service(urlhost) |
| 464 | return service.request(urlpath, **kwargs) |
| 465 | |
| 466 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 467 | def url_read(url, **kwargs): |
| 468 | """Attempts to open the given url multiple times and read all data from it. |
| 469 | |
| 470 | Accepts same arguments as url_open function. |
| 471 | |
| 472 | Returns all data read or None if it was unable to connect or read the data. |
| 473 | """ |
| 474 | response = url_open(url, **kwargs) |
| 475 | if not response: |
| 476 | return None |
| 477 | try: |
| 478 | return response.read() |
| 479 | except TimeoutError: |
| 480 | return None |
| 481 | |
| 482 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 483 | def split_server_request_url(url): |
| 484 | """Splits the url into scheme+netloc and path+params+query+fragment.""" |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 485 | url_parts = list(urlparse.urlparse(url)) |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 486 | urlhost = '%s://%s' % (url_parts[0], url_parts[1]) |
| 487 | urlpath = urlparse.urlunparse(['', ''] + url_parts[2:]) |
| 488 | return urlhost, urlpath |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 489 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 490 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 491 | def get_http_service(urlhost): |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 492 | """Returns existing or creates new instance of HttpService that can send |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 493 | requests to given base urlhost. |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 494 | """ |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 495 | # Ensure consistency. |
| 496 | urlhost = str(urlhost).lower().rstrip('/') |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 497 | with _http_services_lock: |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 498 | service = _http_services.get(urlhost) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 499 | if not service: |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 500 | service = AppEngineService(urlhost) |
| 501 | _http_services[urlhost] = service |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 502 | return service |
| 503 | |
| 504 | |
| 505 | class HttpService(object): |
| 506 | """Base class for a class that provides an API to HTTP based service: |
| 507 | - Provides 'request' method. |
| 508 | - Supports automatic request retries. |
| 509 | - Supports persistent cookies. |
| 510 | - Thread safe. |
| 511 | """ |
| 512 | |
| 513 | # File to use to store all auth cookies. |
maruel@chromium.org | bf2a02a | 2013-07-11 13:27:16 +0000 | [diff] [blame] | 514 | COOKIE_FILE = os.path.join(os.path.expanduser('~'), '.isolated_cookies') |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 515 | |
| 516 | # CookieJar reused by all services + lock that protects its instantiation. |
| 517 | _cookie_jar = None |
| 518 | _cookie_jar_lock = threading.Lock() |
| 519 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 520 | def __init__(self, urlhost): |
| 521 | self.urlhost = urlhost |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 522 | self.cookie_jar = self.load_cookie_jar() |
| 523 | self.opener = self.create_url_opener() |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 524 | |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 525 | def authenticate(self): # pylint: disable=R0201 |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 526 | """Called when HTTP server asks client to authenticate. |
| 527 | Can be implemented in subclasses. |
| 528 | """ |
| 529 | return False |
| 530 | |
| 531 | @staticmethod |
| 532 | def load_cookie_jar(): |
| 533 | """Returns global CoookieJar object that stores cookies in the file.""" |
| 534 | with HttpService._cookie_jar_lock: |
| 535 | if HttpService._cookie_jar is not None: |
| 536 | return HttpService._cookie_jar |
maruel@chromium.org | bf2a02a | 2013-07-11 13:27:16 +0000 | [diff] [blame] | 537 | jar = ThreadSafeCookieJar(HttpService.COOKIE_FILE) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 538 | jar.load() |
| 539 | HttpService._cookie_jar = jar |
| 540 | return jar |
| 541 | |
| 542 | @staticmethod |
| 543 | def save_cookie_jar(): |
| 544 | """Called when cookie jar needs to be flushed to disk.""" |
| 545 | with HttpService._cookie_jar_lock: |
| 546 | if HttpService._cookie_jar is not None: |
| 547 | HttpService._cookie_jar.save() |
| 548 | |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 549 | def create_url_opener(self): # pylint: disable=R0201 |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 550 | """Returns OpenerDirector that will be used when sending requests. |
| 551 | Can be reimplemented in subclasses.""" |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 552 | return urllib2.build_opener(urllib2.HTTPCookieProcessor(self.cookie_jar)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 553 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 554 | def request(self, urlpath, data=None, content_type=None, **kwargs): |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 555 | """Attempts to open the given url multiple times. |
| 556 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 557 | |urlpath| is relative to the server root, i.e. '/some/request?param=1'. |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 558 | |
| 559 | |data| can be either: |
| 560 | -None for a GET request |
| 561 | -str for pre-encoded data |
| 562 | -list for data to be encoded |
| 563 | -dict for data to be encoded (COUNT_KEY will be added in this case) |
| 564 | |
| 565 | Returns a file-like object, where the response may be read from, or None |
| 566 | if it was unable to connect. |
| 567 | """ |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 568 | assert urlpath and urlpath[0] == '/' |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 569 | |
| 570 | if isinstance(data, dict) and COUNT_KEY in data: |
| 571 | logging.error('%s already existed in the data passed into UlrOpen. It ' |
| 572 | 'would be overwritten. Aborting UrlOpen', COUNT_KEY) |
| 573 | return None |
| 574 | |
| 575 | method = 'GET' if data is None else 'POST' |
| 576 | assert not ((method != 'POST') and content_type), ( |
| 577 | 'Can\'t use content_type on GET') |
| 578 | |
| 579 | def make_request(extra): |
| 580 | """Returns a urllib2.Request instance for this specific retry.""" |
| 581 | if isinstance(data, str) or data is None: |
| 582 | payload = data |
| 583 | else: |
| 584 | if isinstance(data, dict): |
| 585 | payload = data.items() |
| 586 | else: |
| 587 | payload = data[:] |
| 588 | payload.extend(extra.iteritems()) |
| 589 | payload = urllib.urlencode(payload) |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 590 | new_url = urlparse.urljoin(self.urlhost, urlpath[1:]) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 591 | if isinstance(data, str) or data is None: |
| 592 | # In these cases, add the extra parameter to the query part of the url. |
| 593 | url_parts = list(urlparse.urlparse(new_url)) |
| 594 | # Append the query parameter. |
| 595 | if url_parts[4] and extra: |
| 596 | url_parts[4] += '&' |
| 597 | url_parts[4] += urllib.urlencode(extra) |
| 598 | new_url = urlparse.urlunparse(url_parts) |
| 599 | request = urllib2.Request(new_url, data=payload) |
| 600 | if payload is not None: |
| 601 | if content_type: |
| 602 | request.add_header('Content-Type', content_type) |
| 603 | request.add_header('Content-Length', len(payload)) |
| 604 | return request |
| 605 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 606 | return self._retry_loop(make_request, **kwargs) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 607 | |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 608 | def _retry_loop( |
| 609 | self, |
| 610 | make_request, |
| 611 | max_attempts=URL_OPEN_MAX_ATTEMPTS, |
| 612 | retry_404=False, |
| 613 | retry_50x=True, |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 614 | timeout=URL_OPEN_TIMEOUT, |
| 615 | read_timeout=None): |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 616 | """Runs internal request-retry loop. |
| 617 | |
| 618 | - Optionally retries HTTP 404 and 50x. |
| 619 | - Retries up to |max_attempts| times. If None or 0, there's no limit in the |
| 620 | number of retries. |
| 621 | - Retries up to |timeout| duration in seconds. If None or 0, there's no |
| 622 | limit in the time taken to do retries. |
| 623 | - If both |max_attempts| and |timeout| are None or 0, this functions retries |
| 624 | indefinitely. |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 625 | |
| 626 | If |read_timeout| is not None will configure underlying socket to |
| 627 | raise TimeoutError exception whenever there's no response from the server |
| 628 | for more than |read_timeout| seconds. It can happen during any read |
| 629 | operation so once you pass non-None |read_timeout| be prepared to handle |
| 630 | these exceptions in subsequent reads from the stream. |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 631 | """ |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 632 | authenticated = False |
| 633 | last_error = None |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 634 | attempt = 0 |
| 635 | start = self._now() |
| 636 | for attempt in itertools.count(): |
| 637 | if max_attempts and attempt >= max_attempts: |
| 638 | # Too many attempts. |
| 639 | break |
| 640 | if timeout and (self._now() - start) >= timeout: |
| 641 | # Retried for too long. |
| 642 | break |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 643 | extra = {COUNT_KEY: attempt} if attempt else {} |
| 644 | request = make_request(extra) |
| 645 | try: |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 646 | url_response = self._url_open(request, timeout=read_timeout) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 647 | logging.debug('url_open(%s) succeeded', request.get_full_url()) |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 648 | return HttpResponse(url_response, request.get_full_url()) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 649 | except urllib2.HTTPError as e: |
| 650 | # Unauthorized. Ask to authenticate and then try again. |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 651 | if e.code in (401, 403): |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 652 | # Try to authenticate only once. If it doesn't help, then server does |
| 653 | # not support app engine authentication. |
vadimsh@chromium.org | a169734 | 2013-04-10 22:57:09 +0000 | [diff] [blame] | 654 | logging.error( |
vadimsh@chromium.org | dde2d73 | 2013-04-10 21:12:52 +0000 | [diff] [blame] | 655 | 'Authentication is required for %s on attempt %d.\n%s', |
| 656 | request.get_full_url(), attempt, |
| 657 | self._format_exception(e, verbose=True)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 658 | if not authenticated and self.authenticate(): |
| 659 | authenticated = True |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 660 | # Do not sleep. |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 661 | continue |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 662 | # If authentication failed, return. |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 663 | logging.error( |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 664 | 'Unable to authenticate to %s.\n%s', |
| 665 | request.get_full_url(), self._format_exception(e, verbose=True)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 666 | return None |
| 667 | |
maruel@chromium.org | d58bf5b | 2013-04-26 17:57:42 +0000 | [diff] [blame] | 668 | if ((e.code < 500 and not (retry_404 and e.code == 404)) or |
| 669 | (e.code >= 500 and not retry_50x)): |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 670 | # This HTTPError means we reached the server and there was a problem |
| 671 | # with the request, so don't retry. |
| 672 | logging.error( |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 673 | 'Able to connect to %s but an exception was thrown.\n%s', |
| 674 | request.get_full_url(), self._format_exception(e, verbose=True)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 675 | return None |
| 676 | |
| 677 | # The HTTPError was due to a server error, so retry the attempt. |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 678 | logging.warning('Able to connect to %s on attempt %d.\n%s', |
| 679 | request.get_full_url(), attempt, |
| 680 | self._format_exception(e)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 681 | last_error = e |
| 682 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 683 | except (urllib2.URLError, httplib.HTTPException, |
| 684 | socket.timeout, ssl.SSLError) as e: |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 685 | logging.warning('Unable to open url %s on attempt %d.\n%s', |
| 686 | request.get_full_url(), attempt, |
| 687 | self._format_exception(e)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 688 | last_error = e |
| 689 | |
| 690 | # Only sleep if we are going to try again. |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 691 | if max_attempts and attempt != max_attempts: |
| 692 | remaining = None |
| 693 | if timeout: |
| 694 | remaining = timeout - (self._now() - start) |
| 695 | if remaining <= 0: |
| 696 | break |
| 697 | self.sleep_before_retry(attempt, remaining) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 698 | |
| 699 | logging.error('Unable to open given url, %s, after %d attempts.\n%s', |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 700 | request.get_full_url(), max_attempts, |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 701 | self._format_exception(last_error, verbose=True)) |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 702 | return None |
| 703 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 704 | def _url_open(self, request, timeout=None): |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 705 | """Low level method to execute urllib2.Request's. |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 706 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 707 | To be mocked in tests. |
| 708 | """ |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 709 | if timeout is not None: |
| 710 | return self.opener.open(request, timeout=timeout) |
| 711 | else: |
| 712 | # Leave original default value for |timeout|. It's nontrivial. |
| 713 | return self.opener.open(request) |
maruel@chromium.org | ef33312 | 2013-03-12 20:36:40 +0000 | [diff] [blame] | 714 | |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 715 | @staticmethod |
| 716 | def _now(): |
| 717 | """To be mocked in tests.""" |
| 718 | return time.time() |
| 719 | |
| 720 | @staticmethod |
| 721 | def calculate_sleep_before_retry(attempt, max_duration): |
| 722 | # Maximum sleeping time. We're hammering a cloud-distributed service, it'll |
| 723 | # survive. |
| 724 | MAX_SLEEP = 10. |
| 725 | # random.random() returns [0.0, 1.0). Starts with relatively short waiting |
| 726 | # time by starting with 1.5/2+1.5^-1 median offset. |
| 727 | duration = (random.random() * 1.5) + math.pow(1.5, (attempt - 1)) |
| 728 | assert duration > 0.1 |
| 729 | duration = min(MAX_SLEEP, duration) |
| 730 | if max_duration: |
| 731 | duration = min(max_duration, duration) |
| 732 | return duration |
| 733 | |
| 734 | @classmethod |
| 735 | def sleep_before_retry(cls, attempt, max_duration): |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 736 | """Sleeps for some amount of time when retrying the request. |
maruel@chromium.org | 2b2139a | 2013-04-30 20:14:58 +0000 | [diff] [blame] | 737 | |
| 738 | To be mocked in tests. |
| 739 | """ |
| 740 | time.sleep(cls.calculate_sleep_before_retry(attempt, max_duration)) |
maruel@chromium.org | ef33312 | 2013-03-12 20:36:40 +0000 | [diff] [blame] | 741 | |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 742 | @staticmethod |
| 743 | def _format_exception(exc, verbose=False): |
| 744 | """Given an instance of some exception raised by urlopen returns human |
| 745 | readable piece of text with detailed information about the error. |
| 746 | """ |
| 747 | out = ['Exception: %s' % (exc,)] |
| 748 | if verbose: |
| 749 | if isinstance(exc, urllib2.HTTPError): |
| 750 | out.append('-' * 10) |
| 751 | if exc.hdrs: |
| 752 | for header, value in exc.hdrs.items(): |
| 753 | if not header.startswith('x-'): |
| 754 | out.append('%s: %s' % (header.capitalize(), value)) |
| 755 | out.append('') |
| 756 | out.append(exc.read() or '<empty body>') |
| 757 | out.append('-' * 10) |
| 758 | return '\n'.join(out) |
| 759 | |
maruel@chromium.org | ef33312 | 2013-03-12 20:36:40 +0000 | [diff] [blame] | 760 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 761 | class HttpResponse(object): |
| 762 | """Response from HttpService.""" |
| 763 | |
| 764 | def __init__(self, url_response, url): |
| 765 | self._url_response = url_response |
| 766 | self._url = url |
| 767 | self._read = 0 |
| 768 | |
| 769 | @property |
| 770 | def content_length(self): |
| 771 | """Total length to the response or None if not known in advance.""" |
| 772 | length = self._url_response.headers.get('Content-Length') |
| 773 | return int(length) if length is not None else None |
| 774 | |
| 775 | def read(self, size=None): |
| 776 | """Reads up to |size| bytes from the stream and returns them. |
| 777 | |
| 778 | If |size| is None reads all available bytes. |
| 779 | |
| 780 | Raises TimeoutError on read timeout. |
| 781 | """ |
| 782 | try: |
| 783 | data = self._url_response.read(size) |
| 784 | self._read += len(data) |
| 785 | return data |
| 786 | except (socket.timeout, ssl.SSLError) as e: |
| 787 | logging.error('Timeout while reading from %s, read %d of %s: %s', |
| 788 | self._url, self._read, self.content_length, e) |
| 789 | raise TimeoutError(e) |
| 790 | |
| 791 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 792 | class AppEngineService(HttpService): |
| 793 | """This class implements authentication support for |
| 794 | an app engine based services. |
maruel@chromium.org | ef33312 | 2013-03-12 20:36:40 +0000 | [diff] [blame] | 795 | """ |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 796 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 797 | # This lock ensures that user won't be confused with multiple concurrent |
| 798 | # login prompts. |
| 799 | _auth_lock = threading.Lock() |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 800 | |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 801 | def __init__(self, urlhost, email=None, password=None): |
| 802 | super(AppEngineService, self).__init__(urlhost) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 803 | self.email = email |
| 804 | self.password = password |
| 805 | self._keyring = None |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 806 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 807 | def authenticate(self): |
| 808 | """Authenticates in the app engine application. |
| 809 | Returns True on success. |
| 810 | """ |
| 811 | if not upload: |
vadimsh@chromium.org | a169734 | 2013-04-10 22:57:09 +0000 | [diff] [blame] | 812 | logging.error('\'upload\' module is missing, ' |
| 813 | 'app engine authentication is disabled.') |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 814 | return False |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 815 | cookie_jar = self.cookie_jar |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 816 | save_cookie_jar = self.save_cookie_jar |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 817 | # RPC server that uses AuthenticationSupport's cookie jar. |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 818 | class AuthServer(upload.AbstractRpcServer): |
| 819 | def _GetOpener(self): |
vadimsh@chromium.org | 2edbe3f | 2013-04-05 19:44:54 +0000 | [diff] [blame] | 820 | # Authentication code needs to know about 302 response. |
| 821 | # So make OpenerDirector without HTTPRedirectHandler. |
| 822 | opener = urllib2.OpenerDirector() |
| 823 | opener.add_handler(urllib2.ProxyHandler()) |
| 824 | opener.add_handler(urllib2.UnknownHandler()) |
| 825 | opener.add_handler(urllib2.HTTPHandler()) |
| 826 | opener.add_handler(urllib2.HTTPDefaultErrorHandler()) |
| 827 | opener.add_handler(urllib2.HTTPSHandler()) |
| 828 | opener.add_handler(urllib2.HTTPErrorProcessor()) |
| 829 | opener.add_handler(urllib2.HTTPCookieProcessor(cookie_jar)) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 830 | return opener |
| 831 | def PerformAuthentication(self): |
| 832 | self._Authenticate() |
| 833 | save_cookie_jar() |
| 834 | return self.authenticated |
| 835 | with AppEngineService._auth_lock: |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 836 | rpc_server = AuthServer(self.urlhost, self.get_credentials) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 837 | return rpc_server.PerformAuthentication() |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 838 | |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 839 | def get_credentials(self): |
| 840 | """Called during authentication process to get the credentials. |
| 841 | May be called mutliple times if authentication fails. |
| 842 | Returns tuple (email, password). |
| 843 | """ |
| 844 | # 'authenticate' calls this only if 'upload' is present. |
| 845 | # Ensure other callers (if any) fail non-cryptically if 'upload' is missing. |
| 846 | assert upload, '\'upload\' module is required for this to work' |
| 847 | if self.email and self.password: |
| 848 | return (self.email, self.password) |
| 849 | if not self._keyring: |
maruel@chromium.org | 000bb4d | 2013-04-26 17:53:27 +0000 | [diff] [blame] | 850 | self._keyring = upload.KeyringCreds(self.urlhost, |
| 851 | self.urlhost, |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 852 | self.email) |
| 853 | return self._keyring.GetUserCredentials() |
| 854 | |
| 855 | |
| 856 | class ThreadSafeCookieJar(cookielib.MozillaCookieJar): |
| 857 | """MozillaCookieJar with thread safe load and save.""" |
| 858 | |
| 859 | def load(self, filename=None, ignore_discard=False, ignore_expires=False): |
| 860 | """Loads cookies from the file if it exists.""" |
maruel@chromium.org | 4e2676d | 2013-06-06 18:39:48 +0000 | [diff] [blame] | 861 | filename = os.path.expanduser(filename or self.filename) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 862 | with self._cookies_lock: |
| 863 | if os.path.exists(filename): |
| 864 | try: |
| 865 | cookielib.MozillaCookieJar.load(self, filename, |
| 866 | ignore_discard, |
| 867 | ignore_expires) |
| 868 | logging.debug('Loaded cookies from %s', filename) |
| 869 | except (cookielib.LoadError, IOError): |
| 870 | pass |
| 871 | else: |
maruel@chromium.org | 16452a3 | 2013-04-05 00:18:44 +0000 | [diff] [blame] | 872 | try: |
| 873 | fd = os.open(filename, os.O_CREAT, 0600) |
| 874 | os.close(fd) |
| 875 | except OSError: |
| 876 | logging.error('Failed to create %s', filename) |
| 877 | try: |
| 878 | os.chmod(filename, 0600) |
| 879 | except OSError: |
| 880 | logging.error('Failed to fix mode for %s', filename) |
vadimsh@chromium.org | 87d6326 | 2013-04-04 19:34:21 +0000 | [diff] [blame] | 881 | |
| 882 | def save(self, filename=None, ignore_discard=False, ignore_expires=False): |
| 883 | """Saves cookies to the file, completely overwriting it.""" |
| 884 | logging.debug('Saving cookies to %s', filename or self.filename) |
| 885 | with self._cookies_lock: |
maruel@chromium.org | 16452a3 | 2013-04-05 00:18:44 +0000 | [diff] [blame] | 886 | try: |
| 887 | cookielib.MozillaCookieJar.save(self, filename, |
| 888 | ignore_discard, |
| 889 | ignore_expires) |
| 890 | except OSError: |
| 891 | logging.error('Failed to save %s', filename) |
csharp@chromium.org | f13eec0 | 2013-03-11 18:22:56 +0000 | [diff] [blame] | 892 | |
| 893 | |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 894 | def valid_file(filepath, size): |
| 895 | """Determines if the given files appears valid (currently it just checks |
| 896 | the file's size).""" |
maruel@chromium.org | 770993b | 2012-12-11 17:16:48 +0000 | [diff] [blame] | 897 | if size == UNKNOWN_FILE_SIZE: |
| 898 | return True |
| 899 | actual_size = os.stat(filepath).st_size |
| 900 | if size != actual_size: |
| 901 | logging.warning( |
| 902 | 'Found invalid item %s; %d != %d', |
| 903 | os.path.basename(filepath), actual_size, size) |
| 904 | return False |
| 905 | return True |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 906 | |
| 907 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 908 | class Profiler(object): |
| 909 | def __init__(self, name): |
| 910 | self.name = name |
| 911 | self.start_time = None |
| 912 | |
| 913 | def __enter__(self): |
| 914 | self.start_time = time.time() |
| 915 | return self |
| 916 | |
| 917 | def __exit__(self, _exc_type, _exec_value, _traceback): |
| 918 | time_taken = time.time() - self.start_time |
| 919 | logging.info('Profiling: Section %s took %3.3f seconds', |
| 920 | self.name, time_taken) |
| 921 | |
| 922 | |
| 923 | class Remote(object): |
maruel@chromium.org | fb155e9 | 2012-09-28 20:36:54 +0000 | [diff] [blame] | 924 | """Priority based worker queue to fetch or upload files from a |
| 925 | content-address server. Any function may be given as the fetcher/upload, |
| 926 | as long as it takes two inputs (the item contents, and their relative |
| 927 | destination). |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 928 | |
| 929 | Supports local file system, CIFS or http remotes. |
| 930 | |
| 931 | When the priority of items is equals, works in strict FIFO mode. |
| 932 | """ |
| 933 | # Initial and maximum number of worker threads. |
| 934 | INITIAL_WORKERS = 2 |
| 935 | MAX_WORKERS = 16 |
| 936 | # Priorities. |
| 937 | LOW, MED, HIGH = (1<<8, 2<<8, 3<<8) |
| 938 | INTERNAL_PRIORITY_BITS = (1<<8) - 1 |
| 939 | RETRIES = 5 |
| 940 | |
maruel@chromium.org | fb155e9 | 2012-09-28 20:36:54 +0000 | [diff] [blame] | 941 | def __init__(self, destination_root): |
| 942 | # Function to fetch a remote object or upload to a remote location.. |
| 943 | self._do_item = self.get_file_handler(destination_root) |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 944 | # Contains tuple(priority, obj). |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 945 | self._done = Queue.PriorityQueue() |
vadimsh@chromium.org | b074b16 | 2013-08-22 17:55:46 +0000 | [diff] [blame] | 946 | self._pool = threading_utils.ThreadPool( |
| 947 | self.INITIAL_WORKERS, self.MAX_WORKERS, 0, 'remote') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 948 | |
maruel@chromium.org | fb155e9 | 2012-09-28 20:36:54 +0000 | [diff] [blame] | 949 | def join(self): |
| 950 | """Blocks until the queue is empty.""" |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 951 | return self._pool.join() |
maruel@chromium.org | fb155e9 | 2012-09-28 20:36:54 +0000 | [diff] [blame] | 952 | |
vadimsh@chromium.org | 53f8d5a | 2013-06-19 13:03:55 +0000 | [diff] [blame] | 953 | def close(self): |
| 954 | """Terminates all worker threads.""" |
| 955 | self._pool.close() |
| 956 | |
csharp@chromium.org | df2968f | 2012-11-16 20:25:37 +0000 | [diff] [blame] | 957 | def add_item(self, priority, obj, dest, size): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 958 | """Retrieves an object from the remote data store. |
| 959 | |
| 960 | The smaller |priority| gets fetched first. |
| 961 | |
| 962 | Thread-safe. |
| 963 | """ |
| 964 | assert (priority & self.INTERNAL_PRIORITY_BITS) == 0 |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 965 | return self._add_item(priority, obj, dest, size) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 966 | |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 967 | def _add_item(self, priority, obj, dest, size): |
| 968 | assert isinstance(obj, basestring), obj |
| 969 | assert isinstance(dest, basestring), dest |
| 970 | assert size is None or isinstance(size, int), size |
| 971 | return self._pool.add_task( |
| 972 | priority, self._task_executer, priority, obj, dest, size) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 973 | |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 974 | def get_one_result(self): |
| 975 | return self._pool.get_one_result() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 976 | |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 977 | def _task_executer(self, priority, obj, dest, size): |
| 978 | """Wraps self._do_item to trap and retry on IOError exceptions.""" |
| 979 | try: |
| 980 | self._do_item(obj, dest) |
| 981 | if size and not valid_file(dest, size): |
| 982 | download_size = os.stat(dest).st_size |
| 983 | os.remove(dest) |
| 984 | raise IOError('File incorrect size after download of %s. Got %s and ' |
| 985 | 'expected %s' % (obj, download_size, size)) |
| 986 | # TODO(maruel): Technically, we'd want to have an output queue to be a |
| 987 | # PriorityQueue. |
| 988 | return obj |
| 989 | except IOError as e: |
| 990 | logging.debug('Caught IOError: %s', e) |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 991 | # Remove unfinished download. |
| 992 | if os.path.exists(dest): |
| 993 | os.remove(dest) |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 994 | # Retry a few times, lowering the priority. |
| 995 | if (priority & self.INTERNAL_PRIORITY_BITS) < self.RETRIES: |
| 996 | self._add_item(priority + 1, obj, dest, size) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 997 | return |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 998 | raise |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 999 | |
csharp@chromium.org | 59c7bcf | 2012-11-21 21:13:18 +0000 | [diff] [blame] | 1000 | def get_file_handler(self, file_or_url): # pylint: disable=R0201 |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1001 | """Returns a object to retrieve objects from a remote.""" |
| 1002 | if re.match(r'^https?://.+$', file_or_url): |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1003 | return functools.partial(self._download_file, file_or_url) |
| 1004 | else: |
| 1005 | return functools.partial(self._copy_file, file_or_url) |
csharp@chromium.org | e9c8d94 | 2013-03-11 20:48:36 +0000 | [diff] [blame] | 1006 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1007 | @staticmethod |
| 1008 | def _download_file(base_url, item, dest): |
| 1009 | # TODO(maruel): Reuse HTTP connections. The stdlib doesn't make this |
| 1010 | # easy. |
| 1011 | try: |
| 1012 | zipped_source = base_url + item |
| 1013 | logging.debug('download_file(%s)', zipped_source) |
csharp@chromium.org | ec47775 | 2013-05-24 20:48:48 +0000 | [diff] [blame] | 1014 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1015 | # Because the app engine DB is only eventually consistent, retry |
| 1016 | # 404 errors because the file might just not be visible yet (even |
| 1017 | # though it has been uploaded). |
| 1018 | connection = url_open(zipped_source, retry_404=True, |
| 1019 | read_timeout=DOWNLOAD_READ_TIMEOUT) |
| 1020 | if not connection: |
| 1021 | raise IOError('Unable to open connection to %s' % zipped_source) |
csharp@chromium.org | ec47775 | 2013-05-24 20:48:48 +0000 | [diff] [blame] | 1022 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1023 | content_length = connection.content_length |
| 1024 | decompressor = zlib.decompressobj() |
| 1025 | size = 0 |
| 1026 | with open(dest, 'wb') as f: |
| 1027 | while True: |
| 1028 | chunk = connection.read(ZIPPED_FILE_CHUNK) |
| 1029 | if not chunk: |
| 1030 | break |
| 1031 | size += len(chunk) |
| 1032 | f.write(decompressor.decompress(chunk)) |
| 1033 | # Ensure that all the data was properly decompressed. |
| 1034 | uncompressed_data = decompressor.flush() |
| 1035 | assert not uncompressed_data |
| 1036 | except IOError as e: |
| 1037 | logging.error('Failed to download %s at %s.\n%s', item, dest, e) |
| 1038 | raise |
| 1039 | except httplib.HTTPException as e: |
| 1040 | msg = 'HTTPException while retrieving %s at %s.\n%s' % (item, dest, e) |
| 1041 | logging.error(msg) |
| 1042 | raise IOError(msg) |
| 1043 | except zlib.error as e: |
| 1044 | msg = 'Corrupted zlib for item %s. Processed %d of %s bytes.\n%s' % ( |
| 1045 | item, size, content_length, e) |
| 1046 | logging.error(msg) |
csharp@chromium.org | e3413b4 | 2013-05-24 17:56:56 +0000 | [diff] [blame] | 1047 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1048 | # Testing seems to show that if a few machines are trying to download |
| 1049 | # the same blob, they can cause each other to fail. So if we hit a |
| 1050 | # zip error, this is the most likely cause (it only downloads some of |
| 1051 | # the data). Randomly sleep for between 5 and 25 seconds to try and |
| 1052 | # spread out the downloads. |
| 1053 | # TODO(csharp): Switch from blobstorage to cloud storage and see if |
| 1054 | # that solves the issue. |
| 1055 | sleep_duration = (random.random() * 20) + 5 |
| 1056 | time.sleep(sleep_duration) |
csharp@chromium.org | a92403f | 2012-11-20 15:13:59 +0000 | [diff] [blame] | 1057 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1058 | raise IOError(msg) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1059 | |
vadimsh@chromium.org | 80f7300 | 2013-07-12 14:52:44 +0000 | [diff] [blame] | 1060 | @staticmethod |
| 1061 | def _copy_file(base_path, item, dest): |
| 1062 | source = os.path.join(base_path, item) |
| 1063 | if source == dest: |
| 1064 | logging.info('Source and destination are the same, no action required') |
| 1065 | return |
| 1066 | logging.debug('copy_file(%s, %s)', source, dest) |
| 1067 | shutil.copy(source, dest) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1068 | |
| 1069 | |
| 1070 | class CachePolicies(object): |
| 1071 | def __init__(self, max_cache_size, min_free_space, max_items): |
| 1072 | """ |
| 1073 | Arguments: |
| 1074 | - max_cache_size: Trim if the cache gets larger than this value. If 0, the |
| 1075 | cache is effectively a leak. |
| 1076 | - min_free_space: Trim if disk free space becomes lower than this value. If |
| 1077 | 0, it unconditionally fill the disk. |
| 1078 | - max_items: Maximum number of items to keep in the cache. If 0, do not |
| 1079 | enforce a limit. |
| 1080 | """ |
| 1081 | self.max_cache_size = max_cache_size |
| 1082 | self.min_free_space = min_free_space |
| 1083 | self.max_items = max_items |
| 1084 | |
| 1085 | |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1086 | class NoCache(object): |
| 1087 | """This class is intended to be usable everywhere the Cache class is. |
| 1088 | Instead of downloading to a cache, all files are downloaded to the target |
| 1089 | directory and then moved to where they are needed. |
| 1090 | """ |
| 1091 | |
| 1092 | def __init__(self, target_directory, remote): |
| 1093 | self.target_directory = target_directory |
| 1094 | self.remote = remote |
| 1095 | |
| 1096 | def retrieve(self, priority, item, size): |
| 1097 | """Get the request file.""" |
| 1098 | self.remote.add_item(priority, item, self.path(item), size) |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 1099 | self.remote.get_one_result() |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1100 | |
| 1101 | def wait_for(self, items): |
| 1102 | """Download the first item of the given list if it is missing.""" |
| 1103 | item = items.iterkeys().next() |
| 1104 | |
| 1105 | if not os.path.exists(self.path(item)): |
| 1106 | self.remote.add_item(Remote.MED, item, self.path(item), UNKNOWN_FILE_SIZE) |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 1107 | downloaded = self.remote.get_one_result() |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1108 | assert downloaded == item |
| 1109 | |
| 1110 | return item |
| 1111 | |
| 1112 | def path(self, item): |
| 1113 | return os.path.join(self.target_directory, item) |
| 1114 | |
| 1115 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1116 | class Cache(object): |
| 1117 | """Stateful LRU cache. |
| 1118 | |
| 1119 | Saves its state as json file. |
| 1120 | """ |
| 1121 | STATE_FILE = 'state.json' |
| 1122 | |
| 1123 | def __init__(self, cache_dir, remote, policies): |
| 1124 | """ |
| 1125 | Arguments: |
| 1126 | - cache_dir: Directory where to place the cache. |
| 1127 | - remote: Remote where to fetch items from. |
| 1128 | - policies: cache retention policies. |
| 1129 | """ |
| 1130 | self.cache_dir = cache_dir |
| 1131 | self.remote = remote |
| 1132 | self.policies = policies |
| 1133 | self.state_file = os.path.join(cache_dir, self.STATE_FILE) |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1134 | self.lru = lru.LRUDict() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1135 | |
| 1136 | # Items currently being fetched. Keep it local to reduce lock contention. |
| 1137 | self._pending_queue = set() |
| 1138 | |
| 1139 | # Profiling values. |
| 1140 | self._added = [] |
| 1141 | self._removed = [] |
| 1142 | self._free_disk = 0 |
| 1143 | |
maruel@chromium.org | 770993b | 2012-12-11 17:16:48 +0000 | [diff] [blame] | 1144 | with Profiler('Setup'): |
| 1145 | if not os.path.isdir(self.cache_dir): |
| 1146 | os.makedirs(self.cache_dir) |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1147 | |
| 1148 | # Load state of the cache. |
vadimsh@chromium.org | a40428e | 2013-07-04 15:43:14 +0000 | [diff] [blame] | 1149 | if os.path.isfile(self.state_file): |
| 1150 | try: |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1151 | self.lru = lru.LRUDict.load(self.state_file) |
| 1152 | except ValueError as err: |
| 1153 | logging.error('Failed to load cache state: %s' % (err,)) |
| 1154 | # Don't want to keep broken state file. |
| 1155 | os.remove(self.state_file) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1156 | |
maruel@chromium.org | 770993b | 2012-12-11 17:16:48 +0000 | [diff] [blame] | 1157 | # Ensure that all files listed in the state still exist and add new ones. |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1158 | previous = self.lru.keys_set() |
| 1159 | unknown = [] |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1160 | for filename in os.listdir(self.cache_dir): |
| 1161 | if filename == self.STATE_FILE: |
| 1162 | continue |
| 1163 | if filename in previous: |
| 1164 | previous.remove(filename) |
| 1165 | continue |
| 1166 | # An untracked file. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1167 | if not RE_IS_SHA1.match(filename): |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1168 | logging.warning('Removing unknown file %s from cache', filename) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1169 | os.remove(self.path(filename)) |
maruel@chromium.org | 770993b | 2012-12-11 17:16:48 +0000 | [diff] [blame] | 1170 | continue |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1171 | # File that's not referenced in 'state.json'. |
| 1172 | # TODO(vadimsh): Verify its SHA1 matches file name. |
| 1173 | logging.warning('Adding unknown file %s to cache', filename) |
| 1174 | unknown.append(filename) |
maruel@chromium.org | 770993b | 2012-12-11 17:16:48 +0000 | [diff] [blame] | 1175 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1176 | if unknown: |
| 1177 | # Add as oldest files. They will be deleted eventually if not accessed. |
| 1178 | self._add_oldest_list(unknown) |
| 1179 | logging.warning('Added back %d unknown files', len(unknown)) |
| 1180 | |
maruel@chromium.org | 770993b | 2012-12-11 17:16:48 +0000 | [diff] [blame] | 1181 | if previous: |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1182 | # Filter out entries that were not found. |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1183 | logging.warning('Removed %d lost files', len(previous)) |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1184 | for filename in previous: |
| 1185 | self.lru.pop(filename) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1186 | self.trim() |
| 1187 | |
| 1188 | def __enter__(self): |
| 1189 | return self |
| 1190 | |
| 1191 | def __exit__(self, _exc_type, _exec_value, _traceback): |
| 1192 | with Profiler('CleanupTrimming'): |
| 1193 | self.trim() |
| 1194 | |
| 1195 | logging.info( |
maruel@chromium.org | 5fd6f47 | 2012-12-11 00:26:08 +0000 | [diff] [blame] | 1196 | '%5d (%8dkb) added', len(self._added), sum(self._added) / 1024) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1197 | logging.info( |
maruel@chromium.org | 5fd6f47 | 2012-12-11 00:26:08 +0000 | [diff] [blame] | 1198 | '%5d (%8dkb) current', |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1199 | len(self.lru), |
| 1200 | sum(self.lru.itervalues()) / 1024) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1201 | logging.info( |
maruel@chromium.org | 5fd6f47 | 2012-12-11 00:26:08 +0000 | [diff] [blame] | 1202 | '%5d (%8dkb) removed', len(self._removed), sum(self._removed) / 1024) |
| 1203 | logging.info(' %8dkb free', self._free_disk / 1024) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1204 | |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1205 | def remove_lru_file(self): |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1206 | """Removes the last recently used file and returns its size.""" |
| 1207 | item, size = self.lru.pop_oldest() |
| 1208 | self._delete_file(item, size) |
| 1209 | return size |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1210 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1211 | def trim(self): |
| 1212 | """Trims anything we don't know, make sure enough free space exists.""" |
| 1213 | # Ensure maximum cache size. |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1214 | if self.policies.max_cache_size: |
| 1215 | total_size = sum(self.lru.itervalues()) |
| 1216 | while total_size > self.policies.max_cache_size: |
| 1217 | total_size -= self.remove_lru_file() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1218 | |
| 1219 | # Ensure maximum number of items in the cache. |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1220 | if self.policies.max_items and len(self.lru) > self.policies.max_items: |
| 1221 | for _ in xrange(len(self.lru) - self.policies.max_items): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1222 | self.remove_lru_file() |
| 1223 | |
| 1224 | # Ensure enough free space. |
| 1225 | self._free_disk = get_free_space(self.cache_dir) |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1226 | trimmed_due_to_space = False |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1227 | while ( |
| 1228 | self.policies.min_free_space and |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1229 | self.lru and |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1230 | self._free_disk < self.policies.min_free_space): |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1231 | trimmed_due_to_space = True |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1232 | self.remove_lru_file() |
| 1233 | self._free_disk = get_free_space(self.cache_dir) |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1234 | if trimmed_due_to_space: |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1235 | total = sum(self.lru.itervalues()) |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1236 | logging.warning( |
| 1237 | 'Trimmed due to not enough free disk space: %.1fkb free, %.1fkb ' |
| 1238 | 'cache (%.1f%% of its maximum capacity)', |
| 1239 | self._free_disk / 1024., |
| 1240 | total / 1024., |
| 1241 | 100. * self.policies.max_cache_size / float(total), |
| 1242 | ) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1243 | self.save() |
| 1244 | |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1245 | def retrieve(self, priority, item, size): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1246 | """Retrieves a file from the remote, if not already cached, and adds it to |
| 1247 | the cache. |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1248 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1249 | If the file is in the cache, verify that the file is valid (i.e. it is |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1250 | the correct size), retrieving it again if it isn't. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1251 | """ |
| 1252 | assert not '/' in item |
| 1253 | path = self.path(item) |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1254 | found = False |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1255 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1256 | if item in self.lru: |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1257 | if not valid_file(self.path(item), size): |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1258 | self.lru.pop(item) |
| 1259 | self._delete_file(item, size) |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1260 | else: |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1261 | # Was already in cache. Update it's LRU value by putting it at the end. |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1262 | self.lru.touch(item) |
| 1263 | found = True |
csharp@chromium.org | 8dc5254 | 2012-11-08 20:29:55 +0000 | [diff] [blame] | 1264 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1265 | if not found: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1266 | if item in self._pending_queue: |
| 1267 | # Already pending. The same object could be referenced multiple times. |
| 1268 | return |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1269 | # TODO(maruel): It should look at the free disk space, the current cache |
| 1270 | # size and the size of the new item on every new item: |
| 1271 | # - Trim the cache as more entries are listed when free disk space is low, |
| 1272 | # otherwise if the amount of data downloaded during the run > free disk |
| 1273 | # space, it'll crash. |
| 1274 | # - Make sure there's enough free disk space to fit all dependencies of |
| 1275 | # this run! If not, abort early. |
csharp@chromium.org | df2968f | 2012-11-16 20:25:37 +0000 | [diff] [blame] | 1276 | self.remote.add_item(priority, item, path, size) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1277 | self._pending_queue.add(item) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1278 | |
| 1279 | def add(self, filepath, obj): |
| 1280 | """Forcibly adds a file to the cache.""" |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1281 | if obj not in self.lru: |
| 1282 | link_file(self.path(obj), filepath, HARDLINK) |
| 1283 | self._add(obj) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1284 | |
| 1285 | def path(self, item): |
| 1286 | """Returns the path to one item.""" |
| 1287 | return os.path.join(self.cache_dir, item) |
| 1288 | |
| 1289 | def save(self): |
| 1290 | """Saves the LRU ordering.""" |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1291 | self.lru.save(self.state_file) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1292 | |
| 1293 | def wait_for(self, items): |
| 1294 | """Starts a loop that waits for at least one of |items| to be retrieved. |
| 1295 | |
| 1296 | Returns the first item retrieved. |
| 1297 | """ |
| 1298 | # Flush items already present. |
| 1299 | for item in items: |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1300 | if item in self.lru: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1301 | return item |
| 1302 | |
| 1303 | assert all(i in self._pending_queue for i in items), ( |
| 1304 | items, self._pending_queue) |
| 1305 | # Note that: |
| 1306 | # len(self._pending_queue) == |
| 1307 | # ( len(self.remote._workers) - self.remote._ready + |
| 1308 | # len(self._remote._queue) + len(self._remote.done)) |
| 1309 | # There is no lock-free way to verify that. |
| 1310 | while self._pending_queue: |
maruel@chromium.org | 13eca0b | 2013-01-22 16:42:21 +0000 | [diff] [blame] | 1311 | item = self.remote.get_one_result() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1312 | self._pending_queue.remove(item) |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1313 | self._add(item) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1314 | if item in items: |
| 1315 | return item |
| 1316 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1317 | def _add(self, item): |
| 1318 | """Adds an item into LRU cache marking it as a newest one.""" |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1319 | size = os.stat(self.path(item)).st_size |
| 1320 | self._added.append(size) |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1321 | self.lru.add(item, size) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1322 | |
vadimsh@chromium.org | 3e97deb | 2013-08-24 00:56:44 +0000 | [diff] [blame^] | 1323 | def _add_oldest_list(self, items): |
| 1324 | """Adds a bunch of items into LRU cache marking them as oldest ones.""" |
| 1325 | pairs = [] |
| 1326 | for item in items: |
| 1327 | size = os.stat(self.path(item)).st_size |
| 1328 | self._added.append(size) |
| 1329 | pairs.append((item, size)) |
| 1330 | self.lru.batch_insert_oldest(pairs) |
| 1331 | |
| 1332 | def _delete_file(self, item, size): |
| 1333 | """Deletes cache file from the file system.""" |
| 1334 | self._removed.append(size) |
| 1335 | try: |
| 1336 | os.remove(self.path(item)) |
| 1337 | except OSError as e: |
| 1338 | logging.error('Error attempting to delete a file\n%s' % e) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1339 | |
| 1340 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1341 | class IsolatedFile(object): |
| 1342 | """Represents a single parsed .isolated file.""" |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1343 | def __init__(self, obj_hash): |
| 1344 | """|obj_hash| is really the sha-1 of the file.""" |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1345 | logging.debug('IsolatedFile(%s)' % obj_hash) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1346 | self.obj_hash = obj_hash |
| 1347 | # Set once all the left-side of the tree is parsed. 'Tree' here means the |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1348 | # .isolate and all the .isolated files recursively included by it with |
| 1349 | # 'includes' key. The order of each sha-1 in 'includes', each representing a |
| 1350 | # .isolated file in the hash table, is important, as the later ones are not |
| 1351 | # processed until the firsts are retrieved and read. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1352 | self.can_fetch = False |
| 1353 | |
| 1354 | # Raw data. |
| 1355 | self.data = {} |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1356 | # A IsolatedFile instance, one per object in self.includes. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1357 | self.children = [] |
| 1358 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1359 | # Set once the .isolated file is loaded. |
| 1360 | self._is_parsed = False |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1361 | # Set once the files are fetched. |
| 1362 | self.files_fetched = False |
| 1363 | |
| 1364 | def load(self, content): |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1365 | """Verifies the .isolated file is valid and loads this object with the json |
| 1366 | data. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1367 | """ |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1368 | logging.debug('IsolatedFile.load(%s)' % self.obj_hash) |
| 1369 | assert not self._is_parsed |
| 1370 | self.data = load_isolated(content) |
| 1371 | self.children = [IsolatedFile(i) for i in self.data.get('includes', [])] |
| 1372 | self._is_parsed = True |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1373 | |
| 1374 | def fetch_files(self, cache, files): |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1375 | """Adds files in this .isolated file not present in |files| dictionary. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1376 | |
| 1377 | Preemptively request files. |
| 1378 | |
| 1379 | Note that |files| is modified by this function. |
| 1380 | """ |
| 1381 | assert self.can_fetch |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1382 | if not self._is_parsed or self.files_fetched: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1383 | return |
| 1384 | logging.debug('fetch_files(%s)' % self.obj_hash) |
| 1385 | for filepath, properties in self.data.get('files', {}).iteritems(): |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1386 | # Root isolated has priority on the files being mapped. In particular, |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1387 | # overriden files must not be fetched. |
| 1388 | if filepath not in files: |
| 1389 | files[filepath] = properties |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 1390 | if 'h' in properties: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1391 | # Preemptively request files. |
| 1392 | logging.debug('fetching %s' % filepath) |
maruel@chromium.org | e5c1713 | 2012-11-21 18:18:46 +0000 | [diff] [blame] | 1393 | cache.retrieve(Remote.MED, properties['h'], properties['s']) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1394 | self.files_fetched = True |
| 1395 | |
| 1396 | |
| 1397 | class Settings(object): |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1398 | """Results of a completely parsed .isolated file.""" |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1399 | def __init__(self): |
| 1400 | self.command = [] |
| 1401 | self.files = {} |
| 1402 | self.read_only = None |
| 1403 | self.relative_cwd = None |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1404 | # The main .isolated file, a IsolatedFile instance. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1405 | self.root = None |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1406 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1407 | def load(self, cache, root_isolated_hash): |
| 1408 | """Loads the .isolated and all the included .isolated asynchronously. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1409 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1410 | It enables support for "included" .isolated files. They are processed in |
| 1411 | strict order but fetched asynchronously from the cache. This is important so |
| 1412 | that a file in an included .isolated file that is overridden by an embedding |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1413 | .isolated file is not fetched needlessly. The includes are fetched in one |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1414 | pass and the files are fetched as soon as all the ones on the left-side |
| 1415 | of the tree were fetched. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1416 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1417 | The prioritization is very important here for nested .isolated files. |
| 1418 | 'includes' have the highest priority and the algorithm is optimized for both |
| 1419 | deep and wide trees. A deep one is a long link of .isolated files referenced |
| 1420 | one at a time by one item in 'includes'. A wide one has a large number of |
| 1421 | 'includes' in a single .isolated file. 'left' is defined as an included |
| 1422 | .isolated file earlier in the 'includes' list. So the order of the elements |
| 1423 | in 'includes' is important. |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1424 | """ |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1425 | self.root = IsolatedFile(root_isolated_hash) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1426 | |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1427 | # Isolated files being retrieved now: hash -> IsolatedFile instance. |
| 1428 | pending = {} |
| 1429 | # Set of hashes of already retrieved items to refuse recursive includes. |
| 1430 | seen = set() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1431 | |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1432 | def retrieve(isolated_file): |
| 1433 | h = isolated_file.obj_hash |
| 1434 | if h in seen: |
| 1435 | raise ConfigError('IsolatedFile %s is retrieved recursively' % h) |
| 1436 | assert h not in pending |
| 1437 | seen.add(h) |
| 1438 | pending[h] = isolated_file |
| 1439 | cache.retrieve(Remote.HIGH, h, UNKNOWN_FILE_SIZE) |
| 1440 | |
| 1441 | retrieve(self.root) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1442 | |
| 1443 | while pending: |
| 1444 | item_hash = cache.wait_for(pending) |
| 1445 | item = pending.pop(item_hash) |
| 1446 | item.load(open(cache.path(item_hash), 'r').read()) |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1447 | if item_hash == root_isolated_hash: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1448 | # It's the root item. |
| 1449 | item.can_fetch = True |
| 1450 | |
| 1451 | for new_child in item.children: |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1452 | retrieve(new_child) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1453 | |
| 1454 | # Traverse the whole tree to see if files can now be fetched. |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1455 | self._traverse_tree(cache, self.root) |
| 1456 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1457 | def check(n): |
| 1458 | return all(check(x) for x in n.children) and n.files_fetched |
| 1459 | assert check(self.root) |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1460 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1461 | self.relative_cwd = self.relative_cwd or '' |
| 1462 | self.read_only = self.read_only or False |
| 1463 | |
vadimsh@chromium.org | f4c063e | 2013-07-04 14:23:31 +0000 | [diff] [blame] | 1464 | def _traverse_tree(self, cache, node): |
| 1465 | if node.can_fetch: |
| 1466 | if not node.files_fetched: |
| 1467 | self._update_self(cache, node) |
| 1468 | will_break = False |
| 1469 | for i in node.children: |
| 1470 | if not i.can_fetch: |
| 1471 | if will_break: |
| 1472 | break |
| 1473 | # Automatically mark the first one as fetcheable. |
| 1474 | i.can_fetch = True |
| 1475 | will_break = True |
| 1476 | self._traverse_tree(cache, i) |
| 1477 | |
| 1478 | def _update_self(self, cache, node): |
| 1479 | node.fetch_files(cache, self.files) |
| 1480 | # Grabs properties. |
| 1481 | if not self.command and node.data.get('command'): |
| 1482 | self.command = node.data['command'] |
| 1483 | if self.read_only is None and node.data.get('read_only') is not None: |
| 1484 | self.read_only = node.data['read_only'] |
| 1485 | if (self.relative_cwd is None and |
| 1486 | node.data.get('relative_cwd') is not None): |
| 1487 | self.relative_cwd = node.data['relative_cwd'] |
| 1488 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1489 | |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1490 | def create_directories(base_directory, files): |
| 1491 | """Creates the directory structure needed by the given list of files.""" |
| 1492 | logging.debug('create_directories(%s, %d)', base_directory, len(files)) |
| 1493 | # Creates the tree of directories to create. |
| 1494 | directories = set(os.path.dirname(f) for f in files) |
| 1495 | for item in list(directories): |
| 1496 | while item: |
| 1497 | directories.add(item) |
| 1498 | item = os.path.dirname(item) |
| 1499 | for d in sorted(directories): |
| 1500 | if d: |
| 1501 | os.mkdir(os.path.join(base_directory, d)) |
| 1502 | |
| 1503 | |
| 1504 | def create_links(base_directory, files): |
| 1505 | """Creates any links needed by the given set of files.""" |
| 1506 | for filepath, properties in files: |
csharp@chromium.org | 89eaf08 | 2013-03-26 18:56:21 +0000 | [diff] [blame] | 1507 | if 'l' not in properties: |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1508 | continue |
maruel@chromium.org | 3320ee1 | 2013-03-28 13:23:31 +0000 | [diff] [blame] | 1509 | if sys.platform == 'win32': |
| 1510 | # TODO(maruel): Create junctions or empty text files similar to what |
| 1511 | # cygwin do? |
| 1512 | logging.warning('Ignoring symlink %s', filepath) |
| 1513 | continue |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1514 | outfile = os.path.join(base_directory, filepath) |
| 1515 | # symlink doesn't exist on Windows. So the 'link' property should |
| 1516 | # never be specified for windows .isolated file. |
| 1517 | os.symlink(properties['l'], outfile) # pylint: disable=E1101 |
| 1518 | if 'm' in properties: |
| 1519 | lchmod = getattr(os, 'lchmod', None) |
| 1520 | if lchmod: |
| 1521 | lchmod(outfile, properties['m']) |
| 1522 | |
| 1523 | |
| 1524 | def setup_commands(base_directory, cwd, cmd): |
| 1525 | """Correctly adjusts and then returns the required working directory |
| 1526 | and command needed to run the test. |
| 1527 | """ |
| 1528 | assert not os.path.isabs(cwd), 'The cwd must be a relative path, got %s' % cwd |
| 1529 | cwd = os.path.join(base_directory, cwd) |
| 1530 | if not os.path.isdir(cwd): |
| 1531 | os.makedirs(cwd) |
| 1532 | |
| 1533 | # Ensure paths are correctly separated on windows. |
| 1534 | cmd[0] = cmd[0].replace('/', os.path.sep) |
| 1535 | cmd = fix_python_path(cmd) |
| 1536 | |
| 1537 | return cwd, cmd |
| 1538 | |
| 1539 | |
| 1540 | def generate_remaining_files(files): |
| 1541 | """Generates a dictionary of all the remaining files to be downloaded.""" |
| 1542 | remaining = {} |
| 1543 | for filepath, props in files: |
| 1544 | if 'h' in props: |
| 1545 | remaining.setdefault(props['h'], []).append((filepath, props)) |
| 1546 | |
| 1547 | return remaining |
| 1548 | |
| 1549 | |
| 1550 | def download_test_data(isolated_hash, target_directory, remote): |
| 1551 | """Downloads the dependencies to the given directory.""" |
| 1552 | if not os.path.exists(target_directory): |
| 1553 | os.makedirs(target_directory) |
| 1554 | |
| 1555 | settings = Settings() |
| 1556 | no_cache = NoCache(target_directory, Remote(remote)) |
| 1557 | |
| 1558 | # Download all the isolated files. |
| 1559 | with Profiler('GetIsolateds') as _prof: |
| 1560 | settings.load(no_cache, isolated_hash) |
| 1561 | |
| 1562 | if not settings.command: |
| 1563 | print >> sys.stderr, 'No command to run' |
| 1564 | return 1 |
| 1565 | |
| 1566 | with Profiler('GetRest') as _prof: |
| 1567 | create_directories(target_directory, settings.files) |
| 1568 | create_links(target_directory, settings.files.iteritems()) |
| 1569 | |
| 1570 | cwd, cmd = setup_commands(target_directory, settings.relative_cwd, |
| 1571 | settings.command[:]) |
| 1572 | |
| 1573 | remaining = generate_remaining_files(settings.files.iteritems()) |
| 1574 | |
| 1575 | # Now block on the remaining files to be downloaded and mapped. |
| 1576 | logging.info('Retrieving remaining files') |
| 1577 | last_update = time.time() |
| 1578 | while remaining: |
| 1579 | obj = no_cache.wait_for(remaining) |
| 1580 | files = remaining.pop(obj) |
| 1581 | |
| 1582 | for i, (filepath, properties) in enumerate(files): |
| 1583 | outfile = os.path.join(target_directory, filepath) |
| 1584 | logging.info(no_cache.path(obj)) |
| 1585 | |
| 1586 | if i + 1 == len(files): |
| 1587 | os.rename(no_cache.path(obj), outfile) |
| 1588 | else: |
| 1589 | shutil.copyfile(no_cache.path(obj), outfile) |
| 1590 | |
maruel@chromium.org | baa108d | 2013-03-28 13:24:51 +0000 | [diff] [blame] | 1591 | if 'm' in properties and not sys.platform == 'win32': |
| 1592 | # It's not set on Windows. It could be set only in the case of |
| 1593 | # downloading content generated from another OS. Do not crash in that |
| 1594 | # case. |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1595 | os.chmod(outfile, properties['m']) |
| 1596 | |
| 1597 | if time.time() - last_update > DELAY_BETWEEN_UPDATES_IN_SECS: |
csharp@chromium.org | 5daba35 | 2013-07-03 17:29:27 +0000 | [diff] [blame] | 1598 | msg = '%d files remaining...' % len(remaining) |
| 1599 | print msg |
| 1600 | logging.info(msg) |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1601 | last_update = time.time() |
| 1602 | |
| 1603 | print('.isolated files successfully downloaded and setup in %s' % |
| 1604 | target_directory) |
| 1605 | print('To run this test please run the command %s from the directory %s' % |
| 1606 | (cmd, cwd)) |
| 1607 | |
| 1608 | return 0 |
| 1609 | |
| 1610 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1611 | def run_tha_test(isolated_hash, cache_dir, remote, policies): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1612 | """Downloads the dependencies in the cache, hardlinks them into a temporary |
| 1613 | directory and runs the executable. |
| 1614 | """ |
| 1615 | settings = Settings() |
| 1616 | with Cache(cache_dir, Remote(remote), policies) as cache: |
| 1617 | outdir = make_temp_dir('run_tha_test', cache_dir) |
| 1618 | try: |
| 1619 | # Initiate all the files download. |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1620 | with Profiler('GetIsolateds') as _prof: |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1621 | # Optionally support local files. |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1622 | if not RE_IS_SHA1.match(isolated_hash): |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1623 | # Adds it in the cache. While not strictly necessary, this simplifies |
| 1624 | # the rest. |
maruel@chromium.org | cb3c3d5 | 2013-03-14 18:55:30 +0000 | [diff] [blame] | 1625 | h = hashlib.sha1(open(isolated_hash, 'rb').read()).hexdigest() |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1626 | cache.add(isolated_hash, h) |
| 1627 | isolated_hash = h |
| 1628 | settings.load(cache, isolated_hash) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1629 | |
| 1630 | if not settings.command: |
| 1631 | print >> sys.stderr, 'No command to run' |
| 1632 | return 1 |
| 1633 | |
| 1634 | with Profiler('GetRest') as _prof: |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1635 | create_directories(outdir, settings.files) |
| 1636 | create_links(outdir, settings.files.iteritems()) |
| 1637 | remaining = generate_remaining_files(settings.files.iteritems()) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1638 | |
| 1639 | # Do bookkeeping while files are being downloaded in the background. |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1640 | cwd, cmd = setup_commands(outdir, settings.relative_cwd, |
| 1641 | settings.command[:]) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1642 | |
| 1643 | # Now block on the remaining files to be downloaded and mapped. |
csharp@chromium.org | 9c59ff1 | 2012-12-12 02:32:29 +0000 | [diff] [blame] | 1644 | logging.info('Retrieving remaining files') |
| 1645 | last_update = time.time() |
vadimsh@chromium.org | b074b16 | 2013-08-22 17:55:46 +0000 | [diff] [blame] | 1646 | with threading_utils.DeadlockDetector(DEADLOCK_TIMEOUT) as detector: |
vadimsh@chromium.org | 5db0f4f | 2013-07-04 13:57:02 +0000 | [diff] [blame] | 1647 | while remaining: |
| 1648 | detector.ping() |
| 1649 | obj = cache.wait_for(remaining) |
| 1650 | for filepath, properties in remaining.pop(obj): |
| 1651 | outfile = os.path.join(outdir, filepath) |
maruel@chromium.org | b7c003d | 2013-07-24 13:04:30 +0000 | [diff] [blame] | 1652 | link_file(outfile, cache.path(obj), HARDLINK) |
vadimsh@chromium.org | 5db0f4f | 2013-07-04 13:57:02 +0000 | [diff] [blame] | 1653 | if 'm' in properties: |
| 1654 | # It's not set on Windows. |
| 1655 | os.chmod(outfile, properties['m']) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1656 | |
vadimsh@chromium.org | 5db0f4f | 2013-07-04 13:57:02 +0000 | [diff] [blame] | 1657 | if time.time() - last_update > DELAY_BETWEEN_UPDATES_IN_SECS: |
| 1658 | msg = '%d files remaining...' % len(remaining) |
| 1659 | print msg |
| 1660 | logging.info(msg) |
| 1661 | last_update = time.time() |
csharp@chromium.org | 9c59ff1 | 2012-12-12 02:32:29 +0000 | [diff] [blame] | 1662 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1663 | if settings.read_only: |
vadimsh@chromium.org | 5db0f4f | 2013-07-04 13:57:02 +0000 | [diff] [blame] | 1664 | logging.info('Making files read only') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1665 | make_writable(outdir, True) |
| 1666 | logging.info('Running %s, cwd=%s' % (cmd, cwd)) |
csharp@chromium.org | e217f30 | 2012-11-22 16:51:53 +0000 | [diff] [blame] | 1667 | |
| 1668 | # TODO(csharp): This should be specified somewhere else. |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 1669 | # TODO(vadimsh): Pass it via 'env_vars' in manifest. |
csharp@chromium.org | e217f30 | 2012-11-22 16:51:53 +0000 | [diff] [blame] | 1670 | # Add a rotating log file if one doesn't already exist. |
| 1671 | env = os.environ.copy() |
vadimsh@chromium.org | 8b9d56b | 2013-08-21 22:24:35 +0000 | [diff] [blame] | 1672 | env.setdefault('RUN_TEST_CASES_LOG_FILE', |
| 1673 | os.path.join(MAIN_DIR, RUN_TEST_CASES_LOG)) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1674 | try: |
| 1675 | with Profiler('RunTest') as _prof: |
csharp@chromium.org | e217f30 | 2012-11-22 16:51:53 +0000 | [diff] [blame] | 1676 | return subprocess.call(cmd, cwd=cwd, env=env) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1677 | except OSError: |
| 1678 | print >> sys.stderr, 'Failed to run %s; cwd=%s' % (cmd, cwd) |
| 1679 | raise |
| 1680 | finally: |
| 1681 | rmtree(outdir) |
| 1682 | |
| 1683 | |
maruel@chromium.org | ea10198 | 2013-07-24 15:54:29 +0000 | [diff] [blame] | 1684 | class OptionParserWithLogging(optparse.OptionParser): |
| 1685 | """Adds --verbose option.""" |
| 1686 | def __init__(self, verbose=0, log_file=None, **kwargs): |
| 1687 | kwargs.setdefault('description', sys.modules['__main__'].__doc__) |
| 1688 | optparse.OptionParser.__init__(self, **kwargs) |
| 1689 | self.add_option( |
| 1690 | '-v', '--verbose', |
| 1691 | action='count', |
| 1692 | default=verbose, |
| 1693 | help='Use multiple times to increase verbosity') |
| 1694 | self.add_option( |
| 1695 | '-l', '--log_file', |
| 1696 | default=log_file, |
| 1697 | help='The name of the file to store rotating log details.') |
| 1698 | |
| 1699 | def parse_args(self, *args, **kwargs): |
| 1700 | options, args = optparse.OptionParser.parse_args(self, *args, **kwargs) |
| 1701 | levels = [logging.ERROR, logging.INFO, logging.DEBUG] |
| 1702 | level = levels[min(len(levels) - 1, options.verbose)] |
| 1703 | |
| 1704 | logging_console = logging.StreamHandler() |
| 1705 | logging_console.setFormatter(logging.Formatter( |
| 1706 | '%(levelname)5s %(module)15s(%(lineno)3d): %(message)s')) |
| 1707 | logging_console.setLevel(level) |
| 1708 | logging.getLogger().setLevel(level) |
| 1709 | logging.getLogger().addHandler(logging_console) |
| 1710 | |
| 1711 | if options.log_file: |
| 1712 | # This is necessary otherwise attached handler will miss the messages. |
| 1713 | logging.getLogger().setLevel(logging.DEBUG) |
| 1714 | |
| 1715 | logging_rotating_file = logging.handlers.RotatingFileHandler( |
| 1716 | options.log_file, |
| 1717 | maxBytes=10 * 1024 * 1024, |
| 1718 | backupCount=5, |
| 1719 | encoding='utf-8') |
| 1720 | # log files are always at DEBUG level. |
| 1721 | logging_rotating_file.setLevel(logging.DEBUG) |
| 1722 | logging_rotating_file.setFormatter(logging.Formatter( |
| 1723 | '%(asctime)s %(levelname)-8s %(module)15s(%(lineno)3d): %(message)s')) |
| 1724 | logging.getLogger().addHandler(logging_rotating_file) |
| 1725 | |
| 1726 | return options, args |
| 1727 | |
| 1728 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1729 | def main(): |
maruel@chromium.org | 46e61cc | 2013-03-25 19:55:34 +0000 | [diff] [blame] | 1730 | disable_buffering() |
maruel@chromium.org | ea10198 | 2013-07-24 15:54:29 +0000 | [diff] [blame] | 1731 | parser = OptionParserWithLogging( |
| 1732 | usage='%prog <options>', log_file=RUN_ISOLATED_LOG_FILE) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1733 | |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1734 | group = optparse.OptionGroup(parser, 'Download') |
| 1735 | group.add_option( |
| 1736 | '--download', metavar='DEST', |
| 1737 | help='Downloads files to DEST and returns without running, instead of ' |
| 1738 | 'downloading and then running from a temporary directory.') |
| 1739 | parser.add_option_group(group) |
| 1740 | |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1741 | group = optparse.OptionGroup(parser, 'Data source') |
| 1742 | group.add_option( |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1743 | '-s', '--isolated', |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1744 | metavar='FILE', |
| 1745 | help='File/url describing what to map or run') |
| 1746 | group.add_option( |
| 1747 | '-H', '--hash', |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1748 | help='Hash of the .isolated to grab from the hash table') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1749 | parser.add_option_group(group) |
| 1750 | |
| 1751 | group.add_option( |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1752 | '-r', '--remote', metavar='URL', |
| 1753 | default= |
| 1754 | 'https://isolateserver.appspot.com/content/retrieve/default-gzip/', |
| 1755 | help='Remote where to get the items. Defaults to %default') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1756 | group = optparse.OptionGroup(parser, 'Cache management') |
| 1757 | group.add_option( |
| 1758 | '--cache', |
| 1759 | default='cache', |
| 1760 | metavar='DIR', |
| 1761 | help='Cache directory, default=%default') |
| 1762 | group.add_option( |
| 1763 | '--max-cache-size', |
| 1764 | type='int', |
| 1765 | metavar='NNN', |
| 1766 | default=20*1024*1024*1024, |
| 1767 | help='Trim if the cache gets larger than this value, default=%default') |
| 1768 | group.add_option( |
| 1769 | '--min-free-space', |
| 1770 | type='int', |
| 1771 | metavar='NNN', |
maruel@chromium.org | 9e98e43 | 2013-05-31 17:06:51 +0000 | [diff] [blame] | 1772 | default=2*1024*1024*1024, |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1773 | help='Trim if disk free space becomes lower than this value, ' |
| 1774 | 'default=%default') |
| 1775 | group.add_option( |
| 1776 | '--max-items', |
| 1777 | type='int', |
| 1778 | metavar='NNN', |
| 1779 | default=100000, |
| 1780 | help='Trim if more than this number of items are in the cache ' |
| 1781 | 'default=%default') |
| 1782 | parser.add_option_group(group) |
| 1783 | |
| 1784 | options, args = parser.parse_args() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1785 | |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1786 | if bool(options.isolated) == bool(options.hash): |
maruel@chromium.org | 5dd75dd | 2012-12-03 15:11:32 +0000 | [diff] [blame] | 1787 | logging.debug('One and only one of --isolated or --hash is required.') |
maruel@chromium.org | 0cd0b18 | 2012-10-22 13:34:15 +0000 | [diff] [blame] | 1788 | parser.error('One and only one of --isolated or --hash is required.') |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1789 | if args: |
maruel@chromium.org | 5dd75dd | 2012-12-03 15:11:32 +0000 | [diff] [blame] | 1790 | logging.debug('Unsupported args %s' % ' '.join(args)) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1791 | parser.error('Unsupported args %s' % ' '.join(args)) |
| 1792 | |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1793 | options.cache = os.path.abspath(options.cache) |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1794 | policies = CachePolicies( |
| 1795 | options.max_cache_size, options.min_free_space, options.max_items) |
csharp@chromium.org | ffd8cf0 | 2013-01-09 21:57:38 +0000 | [diff] [blame] | 1796 | |
| 1797 | if options.download: |
| 1798 | return download_test_data(options.isolated or options.hash, |
| 1799 | options.download, options.remote) |
| 1800 | else: |
| 1801 | try: |
| 1802 | return run_tha_test( |
| 1803 | options.isolated or options.hash, |
| 1804 | options.cache, |
| 1805 | options.remote, |
| 1806 | policies) |
| 1807 | except Exception, e: |
| 1808 | # Make sure any exception is logged. |
| 1809 | logging.exception(e) |
| 1810 | return 1 |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1811 | |
| 1812 | |
| 1813 | if __name__ == '__main__': |
csharp@chromium.org | bfb9874 | 2013-03-26 20:28:36 +0000 | [diff] [blame] | 1814 | # Ensure that we are always running with the correct encoding. |
| 1815 | fix_default_encoding() |
maruel@chromium.org | 9c72d4e | 2012-09-28 19:20:25 +0000 | [diff] [blame] | 1816 | sys.exit(main()) |