[android] Ignore extra files on the client when comparing md5sum output.

build/android/pylib/android_commands.py

 # Copyright (c) 2012 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.
 
 """Provides an interface to communicate with the device via the adb command.
 
 Assumes adb binary is currently on system path.
 """
 
 import collections
@@ skipping 154 matching lines @@
         utc_delta = datetime.timedelta(hours=int(utc_offset[1:3]),
                                        minutes=int(utc_offset[3:5]))
         if utc_offset[0:1] == '-':
           utc_delta = -utc_delta
         lastmod -= utc_delta
       files[filename] = (int(file_match.group('size')), lastmod)
   return files
 
 
 def _ComputeFileListHash(md5sum_output):
-  """Returns a list of MD5 strings from the provided md5sum output."""
-  return [line.split('  ')[0] for line in md5sum_output]
+  """Returns a list of tuples from the provided md5sum output.
+
+  Args:
+    md5sum_output: output directly from md5sum binary.
+
+  Returns:
+    List of namedtuples (hash, path).
+  """
+  HashAndPath = collections.namedtuple('HashAndPath', ['hash', 'path'])
+  split_lines = [line.split('  ') for line in md5sum_output]
+  assert all(len(s) == 2 for s in split_lines), 'Invalid md5sum output'
+  return [HashAndPath._make(s) for s in split_lines]
 
 
 def _HasAdbPushSucceeded(command_output):
   """Returns whether adb push has succeeded from the provided output."""
   # TODO(frankf): We should look at the return code instead of the command
   # output for many of the commands in this file.
   if not command_output:
     return True
   # Success looks like this: "3035 KB/s (12512056 bytes in 4.025s)"
   # Errors look like this: "failed to copy  ... "
@@ skipping 533 matching lines @@
       if not os.path.exists(md5sum_dist_path):
         build_dir = '%s/Release/' % cmd_helper.OutDirectory().get()
         md5sum_dist_path = '%s/md5sum_dist' % build_dir
         assert os.path.exists(md5sum_dist_path), 'Please build md5sum.'
       command = 'push %s %s' % (md5sum_dist_path, MD5SUM_DEVICE_FOLDER)
       assert _HasAdbPushSucceeded(self._adb.SendCommand(command))
       self._md5sum_build_dir = build_dir
 
     cmd = (MD5SUM_LD_LIBRARY_PATH + ' ' + self._util_wrapper + ' ' +
            MD5SUM_DEVICE_PATH + ' ' + device_path)
-    hashes_on_device = _ComputeFileListHash(
+    device_hash_tuples = _ComputeFileListHash(
         self.RunShellCommand(cmd, timeout_time=2 * 60))
     assert os.path.exists(local_path), 'Local path not found %s' % local_path
     md5sum_output = cmd_helper.GetCmdOutput(
         ['%s/md5sum_bin_host' % self._md5sum_build_dir, local_path])
-    hashes_on_host = _ComputeFileListHash(md5sum_output.splitlines())
+    host_hash_tuples = _ComputeFileListHash(md5sum_output.splitlines())
 
-    if ignore_paths:
-      hashes_on_device = [h.split()[0] for h in hashes_on_device]
-      hashes_on_host = [h.split()[0] for h in hashes_on_host]
+    # Ignore extra files on the device.
+    if len(device_hash_tuples) > len(host_hash_tuples):
+      host_files = [os.path.relpath(os.path.normpath(p.path),
+                    os.path.normpath(local_path)) for p in host_hash_tuples]
 
+      def _host_has(fname):
+        return any(path in fname for path in host_files)
+
+      hashes_on_device = [h.hash for h in device_hash_tuples if
+                         _host_has(h.path)]
+    else:
+      hashes_on_device = [h.hash for h in device_hash_tuples]
+
+    # Compare md5sums between host and device files.
+    hashes_on_host = [h.hash for h in host_hash_tuples]
+    hashes_on_device.sort()
+    hashes_on_host.sort()
     return hashes_on_device == hashes_on_host
 
   def PushIfNeeded(self, local_path, device_path):
     """Pushes |local_path| to |device_path|.
 
     Works for files and directories. This method skips copying any paths in
     |test_data_paths| that already exist on the device with the same hash.
 
     All pushed files can be removed by calling RemovePushedFiles().
     """
@@ skipping 617 matching lines @@
   """
   def __init__(self, output):
     self._output = output
 
   def write(self, data):
     data = data.replace('\r\r\n', '\n')
     self._output.write(data)
 
   def flush(self):
     self._output.flush()