Move Autotest to /usr/local/autotest and fix hard-coded autodir paths in client tests.

client/site_tests/realtimecomm_GTalkunittest/realtimecomm_GTalkunittest.py

 # Copyright (c) 2010 The Chromium OS Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 import logging, os, shutil
 
 from autotest_lib.client.bin import test, utils
 from autotest_lib.client.common_lib import error
 
 class realtimecomm_GTalkunittest(test.test):
     version = 1
     # { test_exe : number_of_testcases }
     unittests = {
         'base_unittest'         : '155',
         'call_unittest'         : '2'  ,
         'flute_unittest'        : '146',
         'flutetesting_unittest' : '3'  ,
         'media_unittest'        : '181',
         'p2p_unittest'          : '228',
         'plugin_unittest'       : '10' ,
         'xmllite_unittest'      : '49' ,
         'xmpp_unittest'         : '37' ,
     }
 
     def run_once(self):
         # Stop Google Talk Plugin
         utils.run('pkill GoogleTalkPlugin', ignore_status=True)
 
         # Setup as appropriate
-        talk_path = '/home/autotest/talk'
+        talk_path = os.path.join(self.autodir, 'talk')
         shutil.rmtree(talk_path, ignore_errors=True)
         shutil.copytree(os.path.join(self.bindir, 'talk'), talk_path)
         utils.run('chown chronos %s -R' % talk_path)
         utils.run('chown chronos \
             /tmp/.google-talk-plugin-theuser.lock.testlock', ignore_status=True)
 
         # Run all unittests
         for test_exe in self.unittests:
             # TODO(zhurunz): Support ARM once available.
             x86_talk_path = os.path.join(talk_path, 'i686')
             # The unittest has to be run in 'talk' folder
             test_cmd = "cd %s && su chronos -c \'./%s\'" % \
                 (x86_talk_path, test_exe)
             self.__run_one_test(test_cmd, self.unittests[test_exe])
 
         # Clean up
         shutil.rmtree(talk_path)
 
 
     def __run_one_test(self, test_cmd, number_of_testcases):
         logging.info("Running %s" % test_cmd)
         result = utils.run(test_cmd, ignore_status=True)
         if "[  FAILED  ]" in result.stdout:
             raise error.TestFail(result.stdout)
         expected = "[  PASSED  ] %s tests." % number_of_testcases
         if not expected in result.stdout:
             raise error.TestFail(result.stdout)
         logging.info("%s passed." % test_cmd)