Add a module to fetch builds from different types of builders.

tools/auto_bisect/fetch_build.py

Index: tools/auto_bisect/fetch_build.py
diff --git a/tools/auto_bisect/fetch_build.py b/tools/auto_bisect/fetch_build.py
new file mode 100644
index 0000000000000000000000000000000000000000..ffeb30dcc0d1e609cf33132d4e853c9ec72b1d0a
--- /dev/null
+++ b/tools/auto_bisect/fetch_build.py
@@ -0,0 +1,319 @@
+# Copyright 2014 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+"""This module contains functions for fetching and extracting archived builds.
+
+The builds may be stored in different places by different types builders;
+for example, builders on tryserver.chromium.perf stores builds in one place,
+while builders on chromium.linux store builds in another.
+
+This module can be either imported or run as a stand-alone script to download
+and extract a build.
+
+Usage: fetch_build.py <type> <revision> <output_dir> [options]
+"""
+
+import argparse
+import errno
+import os
+import sys
+import zipfile
+
+import bisect_utils
+import cloud_storage_wrapper as cloud_storage
+
+# Possible builder types.
+PERF_BUILDER = 'perf'
+FULL_BUILDER = 'full'
+
+
+def FetchBuild(builder_type, revision, output_dir, target_arch='x64',
+               target_platform='chromium', deps_patch_sha=None):
+  """Downloads and extracts a build for a particular revision.
+
+  If the build is successfully downloaded and extracted to |output_dir|, the
+  downloaded archive file is also deleted.
+
+  Args:
+    revision: Revision string, e.g. a git commit hash.

[prasadv, 2014/09/11 17:04:10]

Before Git migration builds were stored as SVN revisions, so it can be git
commit hash or svn revision.

[qyearsley, 2014/09/11 23:05:43]

Right, comment updated.

+    builder_type: Type of build archive.
+    target_arch: Architecture, e.g. "ia32".
+    target_platform: Platform name, e.g. "chromium" or "android".
+    deps_patch_sha: SHA1 hash of a DEPS file, if we want to fetch a build for
+        a Chromium revision with custom dependencies.
+
+  Returns:
+    The directory path where the build has been extracted to.

[prasadv, 2014/09/11 17:04:10]

What will this function return in case of any failure?

[qyearsley, 2014/09/11 23:05:43]

It could return None, but another approach would be to raise an exception. Then
the caller who's trying to get the build would catch this exception and possibly
skip this revision.

+  """
+  build_archive = BuildArchive.Create(
+      builder_type, target_arch=target_arch, target_platform=target_platform)
+  bucket = build_archive.BucketName()
+  remote_path = build_archive.FilePath(revision, deps_patch_sha=deps_patch_sha)
+  filename = FetchFromCloudStorage(bucket, remote_path, output_dir)

[prasadv, 2014/09/11 17:04:10]

Shouldn't we validate return values from FetchFromCloudStorage and Unzip?
FetchFromCloudStorage returns None when file is not found.

[qyearsley, 2014/09/11 23:05:43]

Good point. In the case of Unzip, it may raise various types of exceptions (even
ones that aren't explicitly listed. For example, zipfile.BadZipfile or KeyError
could be raised.

If we're using exceptions to indicate failure of FetchBuild, then we don't need
to check here, we can just let the exception be raised and caught further up the
call stack.

+  Unzip(filename, output_dir)

[prasadv, 2014/09/11 17:04:10]

This function should return directory path right?

[qyearsley, 2014/09/11 23:05:43]

That's what I was thinking, but the directory path is always going to be
output_dir here, no need to return it I think.

+
+
+class BuildArchive(object):
+  """Represents a place where builds of some type are stored.
+
+  Subclasses of this class contain specific logic about which directory names
+  and file names should be used when fetching builds.
+  """
+
+  @staticmethod
+  def Create(builder_type, target_arch='x64', target_platform='chromium'):
+    """Creates a BuildArchive instance for fetching builds of some type."""
+    if builder_type == PERF_BUILDER:
+      return PerfBuildArchive(target_arch, target_platform)
+    if builder_type == FULL_BUILDER:
+      return FullBuildArchive(target_arch, target_platform)
+    raise NotImplementedError('Builder type "%s" not supported.' % builder_type)
+
+  def __init__(self, target_arch='x86', target_platform='chromium'):
+    self._target_arch = target_arch
+    self._target_platform = target_platform
+
+  def BucketName(self):
+    raise NotImplementedError()
+
+  def FilePath(self, revision, deps_patch_sha=None):
+    raise NotImplementedError()
+
+  def _ZipFileName(self, revision, deps_patch_sha=None):
+    """Gets the file name of a zip archive for a particular revision.
+
+    This returns a file name of the form full-build-<platform>_<revision>.zip,
+    which is a format used by multiple types of builders that store archives.
+
+    Args:
+      revision: A git commit hash or other revision string.
+      deps_patch_sha: SHA1 hash of a DEPS file patch.
+
+    Returns:
+      The archive file name.
+    """
+    base_name = 'full-build-%s' % self._PlatformName()
+    if deps_patch_sha:
+      revision = '%s_%s' % (revision , deps_patch_sha)
+    return '%s_%s.zip' % (base_name, revision)
+
+  @staticmethod
+  def _PlatformName():
+    """Return a string to be used in paths for the platform."""
+    if bisect_utils.IsWindowsHost() or bisect_utils.Is64BitWindows():
+      # Build archive for x64 is still stored with "win32" in the name.
+      return 'win32'
+    if bisect_utils.IsLinuxHost():
+      # Android builds are also stored with "linux" in the name.
+      return 'linux'
+    if bisect_utils.IsMacHost():
+      return 'mac'
+    raise NotImplementedError('Unknown platform "%s".' % sys.platform)
+
+
+class PerfBuildArchive(BuildArchive):
+  """Provides information about where perf builds are stored."""
+
+  def BucketName(self):
+    return 'chrome-perf'
+
+  def FilePath(self, revision, deps_patch_sha=None):
+    return '%s/%s' % (self._Directory(),
+                      self._ZipFileName(revision, deps_patch_sha))
+
+  def _Directory(self):
+    """Returns the root folder name to download from for perf builds."""
+    if bisect_utils.Is64BitWindows() and self._target_arch == 'x64':
+      return 'Win x64 Builder'
+    if bisect_utils.IsWindowsHost():
+      return 'Win Builder'
+    if bisect_utils.IsLinuxHost() and self._target_platform == 'android':
+      return 'android_perf_rel'
+    if bisect_utils.IsLinuxHost():
+      return 'Linux Builder'
+    if bisect_utils.IsMacHost():
+      return 'Mac Builder'
+    raise NotImplementedError('Unsupported Platform: "%s".' % sys.platform)
+
+
+class FullBuildArchive(BuildArchive):
+  """Provides information about where perf builds are stored."""
+
+  def BucketName(self):
+    if bisect_utils.IsWindowsHost() or bisect_utils.Is64BitWindows():
+      return 'chromium-win-archive'
+    if bisect_utils.IsLinuxHost() and self._target_platform == 'android':
+      return 'chromium-android'
+    if bisect_utils.IsLinuxHost():
+      return 'chromium-linux-archive'
+    if bisect_utils.IsMacHost():
+      return 'chromium-mac-archive'
+    raise NotImplementedError('Unknown platform "%s".' % sys.platform)
+
+  def FilePath(self, revision, deps_patch_sha=None):
+    return '%s/%s' % (self._Directory(),
+                      self._ZipFileName(revision, deps_patch_sha))
+
+  def _Directory(self):

[prasadv, 2014/09/11 17:04:10]

s/_Directory/_ArchiveDirectory?

[qyearsley, 2014/09/11 23:05:43]

Done.

+    if bisect_utils.Is64BitWindows() and self._target_arch == 'x64':
+      return 'chromium.win/Win x64 Builder'
+    if bisect_utils.IsWindowsHost():
+      return 'chromium.win/Win Builder'
+    if bisect_utils.IsLinuxHost() and self._target_platform == 'android':
+      return 'android_main_rel'
+    if bisect_utils.IsLinuxHost():
+      return 'chromium.linux/Linux Builder'
+    if bisect_utils.Mac():
+      return 'chromium.mac/Mac Builder'
+    raise NotImplementedError('Unknown platform "%s".' % sys.platform)
+
+
+def FetchFromCloudStorage(bucket_name, source_path, destination_dir):
+  """Fetches file(s) from the Google Cloud Storage.
+
+  As a side-effect, this prints messages to stdout about what's happening.
+
+  Args:
+    bucket_name: Google Storage bucket name.
+    source_path: Source file path.
+    destination_dir: Destination file path.
+
+  Returns:
+    Downloaded file path if exists, otherwise None.
+  """
+  target_file = os.path.join(destination_dir, os.path.basename(source_path))
+  gs_url = 'gs://%s/%s' % (bucket_name, source_path)
+  try:
+    if cloud_storage.Exists(bucket_name, source_path):
+      print 'Fetching file from %s...' % gs_url
+      cloud_storage.Get(bucket_name, source_path, target_file)
+      if os.path.exists(target_file):
+        return target_file
+    else:
+      print 'File %s not found in cloud storage.' % gs_url
+  except Exception as e:
+    print 'Exception while fetching from cloud storage: %s' % e
+    if os.path.exists(target_file):
+      os.remove(target_file)
+  return None
+
+
+def Unzip(filename, output_dir, verbose=True):
+  """Extracts a zip archive's contents into the given output directory.
+
+  This was based on ExtractZip from build/scripts/common/chromium_utils.py.
+
+  Args:
+    filename: Name of the zip file to extract.
+    output_dir: Path to the destination directory.
+    verbose: Whether to print out what is being extracted.
+
+  Raises:
+    IOError: The unzip command had a non-zero exit code.
+    RuntimeError: Failed to create the output directory.
+  """
+  _MakeDirectory(output_dir)
+
+  # On Linux and Mac, we use the unzip command because it handles links and
+  # file permissions bits, so achieving this behavior is easier than with
+  # ZipInfo options.
+  #
+  # The Mac Version of unzip unfortunately does not support Zip64, whereas
+  # the python module does, so we have to fall back to the python zip module
+  # on Mac if the file size is greater than 4GB.
+  mac_zip_size_limit = 2 ** 32  # 4GB
+  if (bisect_utils.IsLinuxHost() or
+      (bisect_utils.IsMacHost()
+       and os.path.getsize(filename) < mac_zip_size_limit)):
+    unzip_command = ['unzip', '-o']
+    _UnzipUsingCommand(unzip_command, filename, output_dir)
+    return
+
+  # On Windows, try to use 7z if it is installed, otherwise fall back to the
+  # Python zipfile module. If 7z is not installed, then this may fail if the
+  # zip file is larger than 512MB.
+  sevenzip_path = r'C:\Program Files\7-Zip\7z.exe'
+  if (bisect_utils.IsWindowsHost() and os.path.exists(sevenzip_path)):
+    unzip_command = [sevenzip_path, 'x', '-y']
+    _UnzipUsingCommand(unzip_command, filename, output_dir)
+    return
+
+  _UnzipUsingZipFile(filename, output_dir, verbose)
+
+
+def _MakeDirectory(path):
+  """Creates a directory if it doesn't already exist."""
+  try:
+    os.makedirs(path)
+  except OSError as e:
+    # Reference: https://docs.python.org/2/library/errno.html
+    if e.errno != errno.EEXIST:
+      raise RuntimeError('Failed to make directory: %s' % path)
+
+
+def _UnzipUsingCommand(unzip_command, filename, output_dir):
+  """Extracts a zip file using an external command.
+
+  Args:
+    unzip_command: An unzipping command, as a string list, without the filename.
+    filename: Path to the zip file.
+    output_dir: The directory which the contents should be extracted to.
+
+  Raises:
+    IOError: The command had a non-zero exit code.
+  """
+  absolute_filepath = os.path.abspath(filename)
+  command = unzip_command + [absolute_filepath]
+  return_code = _RunCommandInDirectory(output_dir, command)
+  if return_code:
+    raise IOError('Unzip failed: %s => %s' % (str(command), return_code))
+
+
+def _RunCommandInDirectory(directory, command):
+  """Changes to a directory, runs a command, then changes back."""
+  saved_dir = os.getcwd()
+  os.chdir(directory)
+  return_code = bisect_utils.RunProcess(command)
+  os.chdir(saved_dir)
+  return return_code
+
+
+def _UnzipUsingZipFile(filename, output_dir, verbose=True):
+  """Extracts a zip file using the Python zipfile module."""
+  assert bisect_utils.IsWindowsHost() or bisect_utils.IsMacHost()
+  zf = zipfile.ZipFile(filename)
+  for name in zf.namelist():
+    if verbose:
+      print 'Extracting %s' % name
+    zf.extract(name, output_dir)
+    if bisect_utils.IsMacHost():
+      # Restore file permission bits.
+      mode = zf.getinfo(name).external_attr >> 16
+      os.chmod(os.path.join(output_dir, name), mode)
+
+
+def Main(argv):
+  """Downloads and extracts a build based on the command line arguments."""
+  parser = argparse.ArgumentParser()
+  parser.add_argument('builder_type')
+  parser.add_argument('revision')
+  parser.add_argument('output_dir')
+  parser.add_argument('--target-arch', default='x64')

[prasadv, 2014/09/11 17:04:10]

I think default should be ia32.

[qyearsley, 2014/09/11 23:05:43]

Done.

+  parser.add_argument('--target-platform', default='chromium')
+  parser.add_argument('--deps-patch-sha')
+  args = parser.parse_args(argv[1:])
+
+  FetchBuild(
+      args.builder_type, args.revision, args.output_dir,
+      target_arch=args.target_arch, target_platform=args.target_platform,
+      deps_patch_sha=args.deps_patch_sha)
+
+  print 'Build has been downloaded to and extracted in %s.' % args.output_dir
+
+  return 0
+
+
+if __name__ == '__main__':
+  sys.exit(Main(sys.argv))
+