[Android] Apk size tool.

build/android/apksize.py

Index: build/android/apksize.py
diff --git a/build/android/apksize.py b/build/android/apksize.py
new file mode 100644
index 0000000000000000000000000000000000000000..8b2294271c828be17dd158a97f175e2c0e8b8e0a
--- /dev/null
+++ b/build/android/apksize.py
@@ -0,0 +1,210 @@
+import argparse
+import collections
+import logging
+import os
+import sys
+import zipfile
+
+from pylib import constants
+
+sys.path.append(os.path.join(constants.DIR_SOURCE_ROOT, 'build', 'util', 'lib',
+                             'common'))
+import perf_tests_results_helper # pylint: disable=import-error
+
+
+# TODO(rnephew): Add support for split apks.
+class ApkSizeInfo(object):
+
+  def __init__(self, path):
+    """ ApkSizeInfo constructor.
+
+    Args:
+      path: Path to apk.
+    """
+    if not os.path.isfile(path):
+      raise IOError('Not a valid file path for apk.')
+    if not os.access(path, os.R_OK):
+      raise IOError('File is not readable.')
+    logging.info('APK: %s', path)

[mikecase (-- gone --), 2015/10/07 19:33:01]

nit: Slightly strangely placed logging statement. Maybe just move to start of
function or move to be after all three "if" statements. Would help organize
things a little bit.

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+    if not zipfile.is_zipfile(path):
+      raise TypeError('Not a valid apk')
+    self._apk_size = os.path.getsize(path)
+    self._zipfile = zipfile.ZipFile(path, 'r')
+    self._processed_files = None
+    self._compressed_size = 0
+    self._total_files = 0
+    self._uncompressed_size = 0
+
+  def ProcessFiles(self):
+    """ Uses zipinfo to process apk file information."""
+    if self._processed_files:
+      return self._processed_files
+    self._processed_files = {}
+    for f in self._zipfile.infolist():
+      _, file_ext = os.path.splitext(f.filename)
+      file_ext = file_ext[1:] # Drop . from extension.
+      if file_ext in self._processed_files:
+        self._compressed_size += f.compress_size
+        self._total_files += 1
+        self._uncompressed_size += f.file_size
+        self._processed_files[file_ext]['number'] += 1
+        self._processed_files[file_ext]['compressed'] += f.compress_size

[mikecase (-- gone --), 2015/10/07 19:33:01]

Maybe add units into the key name, like 'compressed_bytes'.

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+        self._processed_files[file_ext]['uncompressed'] += f.file_size
+      else:
+
+        self._compressed_size += f.compress_size
+        self._total_files += 1
+        self._uncompressed_size += f.file_size
+        self._processed_files[file_ext] = {
+            'number': 1,
+            'compressed': f.compress_size,
+            'uncompressed': f.file_size,
+        }
+    return self._processed_files
+
+  def Compare(self, old_apk):

[mikecase (-- gone --), 2015/10/07 19:33:01]

maybe rename old_apk to other_apk. Doesn't appear to have to be an older apk.

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+    """Compares size information of two apks.
+
+    Args:
+      old_apk: Apk to compare size against.
+    """
+    if not isinstance(old_apk, type(self)):
+      raise TypeError('Must pass it an ApkSizeInfo object')
+    apk_one = old_apk.ProcessFiles()
+    apk_two = self.ProcessFiles()
+    old_lib_compressed = apk_one['so']['compressed']
+    old_lib_uncompressed = apk_one['so']['uncompressed']
+    new_lib_compressed = apk_two['so']['compressed']
+    new_lib_uncompressed = apk_two['so']['uncompressed']
+    reduction_apk_size = old_apk.compressed_size - self.compressed_size
+    reduction_compressed_files = old_lib_compressed - new_lib_compressed
+    reduction_uncompressed_files = old_lib_uncompressed - new_lib_uncompressed

[mikecase (-- gone --), 2015/10/07 19:33:01]

nit: ^ for the above 10 or so lines, maybe just fill in a dict as you go along
so you don't have to make a ton of new variables.

[rnephew (Reviews Here), 2015/10/07 21:18:24]

I did that with some of them, but not all; it made readability bad.

+    reduction_install_legacy = (
+        reduction_compressed_files + reduction_uncompressed_files)
+    # TODO(rnephew) This will be made obsolete with modern and legacy apks being
+    # seperate, a new method to compare will be required eventually.
+    ret_dict = collections.OrderedDict([

[mikecase (-- gone --), 2015/10/07 19:33:01]

nit: Rename ret_dict to something more descriptive like comparison_results

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Just got rid of it, no need to make a variable when all its going to do is
return it next line.

+        ('APK_size_reduction', reduction_apk_size),
+        ('ARM32_Legacy_install_or_upgrade_reduction', reduction_install_legacy),

[mikecase (-- gone --), 2015/10/07 19:33:01]

Does this size comparison only work for ARM32?

[rnephew (Reviews Here), 2015/10/07 21:18:24]

I picked the names based off the document here:
https://docs.google.com/document/d/1b-99ysCyQxhaTCFdBjRhrtft7-NXXAwlwG0K951e9...

I added Andrew as a reviewer to this CL so maybe he can comment more on it.

+        ('ARM32_Legacy_system_image_reduction', reduction_compressed_files),
+        ('ARM32_Legacy_patch_size_reduction', reduction_compressed_files),
+        ('ARM32_Modern_ARM64_install_or_upgrade_reduction',
+            reduction_uncompressed_files),
+        ('ARM32_Modern_ARM64_system_image_reduction',
+            reduction_uncompressed_files),
+        ('ARM32_Modern_ARM64_patch_size_reduction', reduction_compressed_files)
+    ])
+    return ret_dict
+
+  @property
+  def apk_size(self):
+    return self._apk_size
+
+  @property
+  def compressed_size(self):
+    if self._compressed_size is None:
+      self.ProcessFiles()
+    return self._compressed_size
+
+  @property
+  def total_files(self):
+    if self._total_files is None:
+      self.ProcessFiles()
+    return self._total_files
+
+  @property
+  def uncompressed_size(self):
+    if self._uncompressed_size is None:
+      self.ProcessFiles()
+    return self._uncompressed_size
+
+def dashboard_readable_size_info(apk):
+  """Prints size information in dashboard readable format.
+
+  Args:
+    apk: ApkSizeInfo object
+  """
+  files = apk.ProcessFiles()
+  perf_tests_results_helper.PrintPerfResult(
+      'apk_size', 'total_files', [apk.total_files], 'files')
+  perf_tests_results_helper.PrintPerfResult(
+      'apk_size', 'total_size_compressed', [apk.compressed_size], 'bytes')
+  perf_tests_results_helper.PrintPerfResult(
+      'apk_size', 'total_size_uncompressed', [apk.uncompressed_size], 'bytes')
+  perf_tests_results_helper.PrintPerfResult(
+      'apk_size', 'apk_overhead', [apk.apk_size - apk.compressed_size], 'bytes')
+  for ext in files:
+    perf_tests_results_helper.PrintPerfResult(
+        'apk_size', '%s_files' % ext, [files[ext]['number']], 'files')
+    perf_tests_results_helper.PrintPerfResult(
+        'apk_size', '%s_compressed_size' % ext, [files[ext]['compressed']],
+        'bytes')
+    perf_tests_results_helper.PrintPerfResult(
+        'apk_size', '%s_uncompressed_size' % ext, [files[ext]['uncompressed']],
+        'bytes')
+
+def human_readable_size_info(apk):
+  """Prints size information in human readable format.
+
+  Args:
+    apk: ApkSizeInfo object
+  """
+  files = apk.ProcessFiles()
+  logging.critical('Stats for files as they exist within the apk:')
+  for ext in files:
+    logging.critical('  %-8s %s bytes in %s files', ext,
+                     files[ext]['compressed'], files[ext]['number'])
+  logging.critical('--------------------------------------')
+  logging.critical(
+      'All Files: %s bytes in %s files', apk.compressed_size, apk.total_files)
+  logging.critical('APK Size: %s', apk.apk_size)
+  logging.critical('APK overhead: %s', apk.apk_size - apk.compressed_size)
+  logging.critical('--------------------------------------')
+  logging.critical('Stats for files when extracted from the apk:')
+  for ext in files:
+    logging.critical('  %-8s %s bytes in %s files', ext,
+                     files[ext]['uncompressed'], files[ext]['number'])
+  logging.critical('--------------------------------------')
+  logging.critical(
+      'All Files: %s bytes in %s files', apk.uncompressed_size, apk.total_files)
+
+def human_readable_compare(results):
+  """Prints size comparison between two apks in human readable format.
+
+  Args:
+    apk: ApkSizeInfo object

[mikecase (-- gone --), 2015/10/07 19:33:01]

Fix this Args doc.

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+  """
+  for key, value in results.iteritems():
+    logging.critical('  %-50s %s bytes', key, value)
+
+def dashboard_readable_compare(results):

[mikecase (-- gone --), 2015/10/07 19:33:01]

nit: I would rename these functions to have "output" or "print" in there names.

like

output_compare_results
output_dashboard_compare_results

or

print_compare_results
print_dashboard_compare_results

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+  """Prints size comparison between two apks in dashboard readable format.
+
+  Args:

[mikecase (-- gone --), 2015/10/07 19:33:01]

Fix this Args doc.

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+    apk: ApkSizeInfo object
+  """
+  for key, value in results.iteritems():
+    perf_tests_results_helper.PrintPerfResult(
+        'apk_size_compare', key, [value], 'bytes')
+
+def main():
+  parser = argparse.ArgumentParser()
+  parser.add_argument("file_path")

[mikecase (-- gone --), 2015/10/07 19:33:01]

nit: single quotes probably

[rnephew (Reviews Here), 2015/10/07 21:18:23]

Done.

+  parser.add_argument('-b', '--bot-mode', action='store_true',

[mikecase (-- gone --), 2015/10/07 19:33:01]

maybe rename to --perf-dashboard-formatting or --perf-dashboard-output

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+                      help=('Sets it to return data in bot readable format'))
+  parser.add_argument('-c', '--compare', help=('APK to compare against.'))
+  args = parser.parse_args()
+
+  apk = ApkSizeInfo(args.file_path)
+  if args.compare:
+    if args.bot_mode:
+      dashboard_readable_compare(apk.Compare(ApkSizeInfo(args.compare)))
+    else:
+      human_readable_compare(apk.Compare(ApkSizeInfo(args.compare)))
+  elif args.bot_mode:
+    dashboard_readable_size_info(apk)
+  else:
+    human_readable_size_info(apk)

[mikecase (-- gone --), 2015/10/07 19:33:01]

nit: I think this would look better as...

if args.compare:
  if args.bot_mode:
    dashboard_readable_compare(apk.Compare(ApkSizeInfo(args.compare)))
  else:
    human_readable_compare(apk.Compare(ApkSizeInfo(args.compare)))
else:
  if args.bot_mode:
    dashboard_readable_size_info(apk)
  else:
    human_readable_size_info(apk)

because its just more symmetric. 

[rnephew (Reviews Here), 2015/10/07 21:18:24]

Done.

+
+if __name__ == '__main__':
+  sys.exit(main())