tools/utils.py

# Copyright (c) 2016, the R8 project authors. Please see the AUTHORS file
# for details. All rights reserved. Use of this source code is governed by a
# BSD-style license that can be found in the LICENSE file.

# Different utility functions used accross scripts

import hashlib
import os
import re
import shutil
import subprocess
import sys
import tempfile

TOOLS_DIR = os.path.abspath(os.path.normpath(os.path.join(__file__, '..')))
REPO_ROOT = os.path.realpath(os.path.join(TOOLS_DIR, '..'))
MEMORY_USE_TMP_FILE = 'memory_use.tmp'
DEX_SEGMENTS_RESULT_PATTERN = re.compile('- ([^:]+): ([0-9]+)')
BUILD = os.path.join(REPO_ROOT, 'build')
LIBS = os.path.join(BUILD, 'libs')
GENERATED_LICENSE_DIR = os.path.join(BUILD, 'generatedLicense')
SRC_ROOT = os.path.join(REPO_ROOT, 'src', 'main', 'java')

D8 = 'd8'
R8 = 'r8'
R8_SRC = 'sourceJar'
COMPATDX = 'compatdx'
COMPATPROGUARD = 'compatproguard'

D8_JAR = os.path.join(LIBS, 'd8.jar')
R8_JAR = os.path.join(LIBS, 'r8.jar')
R8_SRC_JAR = os.path.join(LIBS, 'r8-src.jar')
R8_EXCLUDE_DEPS_JAR = os.path.join(LIBS, 'r8-exclude-deps.jar')
COMPATDX_JAR = os.path.join(LIBS, 'compatdx.jar')
COMPATPROGUARD_JAR = os.path.join(LIBS, 'compatproguard.jar')
MAVEN_ZIP = os.path.join(LIBS, 'r8.zip')
GENERATED_LICENSE = os.path.join(GENERATED_LICENSE_DIR, 'LICENSE')

def PrintCmd(s):
  if type(s) is list:
    s = ' '.join(s)
  print 'Running: %s' % s
  # I know this will hit os on windows eventually if we don't do this.
  sys.stdout.flush()

def IsWindows():
  return os.name == 'nt'

def DownloadFromX20(sha1_file):
  download_script = os.path.join(REPO_ROOT, 'tools', 'download_from_x20.py')
  cmd = [download_script, sha1_file]
  PrintCmd(cmd)
  subprocess.check_call(cmd)

def DownloadFromGoogleCloudStorage(sha1_file, bucket='r8-deps'):
  suffix = '.bat' if IsWindows() else ''
  download_script = 'download_from_google_storage%s' % suffix
  cmd = [download_script, '-n', '-b', bucket, '-u', '-s',
         sha1_file]
  PrintCmd(cmd)
  subprocess.check_call(cmd)

def get_sha1(filename):
  sha1 = hashlib.sha1()
  with open(filename, 'rb') as f:
    while True:
      chunk = f.read(1024*1024)
      if not chunk:
        break
      sha1.update(chunk)
  return sha1.hexdigest()

def makedirs_if_needed(path):
  try:
    os.makedirs(path)
  except OSError:
    if not os.path.isdir(path):
        raise

def upload_dir_to_cloud_storage(directory, destination, is_html=False):
  # Upload and make the content encoding right for viewing directly
  cmd = ['gsutil.py', 'cp']
  if is_html:
    cmd += ['-z', 'html']
  cmd += ['-a', 'public-read', '-R', directory, destination]
  PrintCmd(cmd)
  subprocess.check_call(cmd)

def upload_file_to_cloud_storage(source, destination):
  cmd = ['gsutil.py', 'cp', '-a', 'public-read', source, destination]
  PrintCmd(cmd)
  subprocess.check_call(cmd)

# Note that gcs is eventually consistent with regards to list operations.
# This is not a problem in our case, but don't ever use this method
# for synchronization.
def cloud_storage_exists(destination):
  cmd = ['gsutil.py', 'ls', destination]
  PrintCmd(cmd)
  exit_code = subprocess.call(cmd)
  return exit_code == 0

class TempDir(object):
 def __init__(self, prefix=''):
   self._temp_dir = None
   self._prefix = prefix

 def __enter__(self):
   self._temp_dir = tempfile.mkdtemp(self._prefix)
   return self._temp_dir

 def __exit__(self, *_):
   shutil.rmtree(self._temp_dir, ignore_errors=True)

class ChangedWorkingDirectory(object):
 def __init__(self, working_directory):
   self._working_directory = working_directory

 def __enter__(self):
   self._old_cwd = os.getcwd()
   print 'Enter directory = ', self._working_directory
   os.chdir(self._working_directory)

 def __exit__(self, *_):
   print 'Enter directory = ', self._old_cwd
   os.chdir(self._old_cwd)

# Reading Android CTS test_result.xml

class CtsModule(object):
  def __init__(self, module_name):
    self.name = module_name

class CtsTestCase(object):
  def __init__(self, test_case_name):
    self.name = test_case_name

class CtsTest(object):
  def __init__(self, test_name, outcome):
    self.name = test_name
    self.outcome = outcome

# Generator yielding CtsModule, CtsTestCase or CtsTest from
# reading through a CTS test_result.xml file.
def read_cts_test_result(file_xml):
  re_module = re.compile('<Module name="([^"]*)"')
  re_test_case = re.compile('<TestCase name="([^"]*)"')
  re_test = re.compile('<Test result="(pass|fail)" name="([^"]*)"')
  with open(file_xml) as f:
    for line in f:
      m = re_module.search(line)
      if m:
        yield CtsModule(m.groups()[0])
        continue
      m = re_test_case.search(line)
      if m:
        yield CtsTestCase(m.groups()[0])
        continue
      m = re_test.search(line)
      if m:
        outcome = m.groups()[0]
        assert outcome in ['fail', 'pass']
        yield CtsTest(m.groups()[1], outcome == 'pass')

def grep_memoryuse(logfile):
  re_vmhwm = re.compile('^VmHWM:[ \t]*([0-9]+)[ \t]*([a-zA-Z]*)')
  result = None
  with open(logfile) as f:
    for line in f:
      m = re_vmhwm.search(line)
      if m:
        groups = m.groups()
        s = len(groups)
        if s >= 1:
          result = int(groups[0])
          if s >= 2:
            unit = groups[1]
            if unit == 'kB':
              result *= 1024
            elif unit != '':
              raise Exception('Unrecognized unit in memory usage log: {}'
                  .format(unit))
  if result is None:
    raise Exception('No memory usage found in log: {}'.format(logfile))
  return result

# Return a dictionary: {segment_name -> segments_size}
def getDexSegmentSizes(dex_files):
  assert len(dex_files) > 0
  cmd = ['java', '-jar', R8_JAR, 'dexsegments']
  cmd.extend(dex_files)
  PrintCmd(cmd)
  output = subprocess.check_output(cmd)

  matches = DEX_SEGMENTS_RESULT_PATTERN.findall(output)

  if matches is None or len(matches) == 0:
    raise Exception('DexSegments failed to return any output for' \
        ' these files: {}'.format(dex_files))

  result = {}

  for match in matches:
    result[match[0]] = int(match[1])

  return result

def get_maven_path(version):
  return os.path.join('com', 'android', 'tools', 'r8', version)

def print_dexsegments(prefix, dex_files):
  for segment_name, size in getDexSegmentSizes(dex_files).items():
    print('{}-{}(CodeSize): {}'
        .format(prefix, segment_name, size))

# Ensure that we are not benchmarking with a google jvm.
def check_java_version():
  cmd= ['java', '-version']
  output = subprocess.check_output(cmd, stderr = subprocess.STDOUT)
  m = re.search('openjdk version "([^"]*)"', output)
  if m is None:
    raise Exception("Can't check java version: no version string in output"
        " of 'java -version': '{}'".format(output))
  version = m.groups(0)[0]
  m = re.search('google', version)
  if m is not None:
    raise Exception("Do not use google JVM for benchmarking: " + version)

def verify_with_dex2oat(dex_file):

  # dex2oat accepts non-existent dex files, check here instead
  if not os.path.exists(dex_file):
    raise Exception('Dex file not found: "{}"'.format(dex_file))

  android_root_dir = os.path.join(TOOLS_DIR, 'linux', 'art', 'product',
    'angler')
  boot_art = os.path.join(android_root_dir, 'system', 'framework', 'boot.art')
  dex2oat = os.path.join(TOOLS_DIR, 'linux', 'art', 'bin', 'dex2oat')

  with TempDir() as temp:
    oat_file = os.path.join(temp, 'all.oat')

    cmd = [
        dex2oat,
        '--android-root=' + android_root_dir,
        '--runtime-arg', '-Xnorelocate',
        '--boot-image=' + boot_art,
        '--dex-file=' + dex_file,
        '--oat-file=' + oat_file,
        '--instruction-set=arm64',
        '--compiler-filter=quicken'
    ]

    PrintCmd(cmd)
    subprocess.check_call(cmd,
      env = {"LD_LIBRARY_PATH":
        os.path.join(TOOLS_DIR, 'linux', 'art', 'lib')}
    )