Add support for parsing code coverage file path to runtest and related cleanup.

This change is to adapt runtest to the InstrumentationTestRunner change which
dumps coverage data to the app's local data folder, instead of a fixed file on
the sdcard.

Also fixed adb_interace.StartInstrumentationForPackage, added better handling
when generating coverage for tests with unknown coverage targets, and reduced
the duplicate definitions of the "coverage_targets.xml" definition file.

testrunner/runtest.py

@@ -278,16 +278,53 @@ class TestRunner(object):
           raw_mode=self._options.raw_mode,
           instrumentation_args=instrumentation_args)
       logger.Log(adb_cmd)
+    elif self._options.coverage:
+      # need to parse test output to determine path to coverage file
+      logger.Log("Running in coverage mode, suppressing test output")
+      try:
+        (test_results, status_map) = self._adb.StartInstrumentationForPackage(
+          package_name=test_suite.GetPackageName(),
+          runner_name=test_suite.GetRunnerName(),
+          timeout_time=60*60,
+          instrumentation_args=instrumentation_args)
+      except errors.InstrumentationError, errors.DeviceUnresponsiveError:
+        return
+      self._PrintTestResults(test_results)
+      device_coverage_path = status_map.get("coverageFilePath", None)
+      if device_coverage_path is None:
+        logger.Log("Error: could not find coverage data on device")
+        return
+      coverage_file = self._coverage_gen.ExtractReport(test_suite, device_coverage_path)
+      if coverage_file is not None:
+        logger.Log("Coverage report generated at %s" % coverage_file)
     else:
       self._adb.StartInstrumentationNoResults(
           package_name=test_suite.GetPackageName(),
           runner_name=test_suite.GetRunnerName(),
           raw_mode=self._options.raw_mode,
           instrumentation_args=instrumentation_args)
-      if self._options.coverage and test_suite.GetTargetName() is not None:
-        coverage_file = self._coverage_gen.ExtractReport(test_suite)
-        if coverage_file is not None:
-          logger.Log("Coverage report generated at %s" % coverage_file)
+
+  def _PrintTestResults(self, test_results):
+    """Prints a summary of test result data to stdout.
+
+    Args:
+      test_results: a list of am_instrument_parser.TestResult
+    """
+    total_count = 0
+    error_count = 0
+    fail_count = 0
+    for test_result in test_results:
+      if test_result.GetStatusCode() == -1: # error
+        logger.Log("Error in %s: %s" % (test_result.GetTestName(),
+                                        test_result.GetFailureReason()))
+        error_count+=1
+      elif test_result.GetStatusCode() == -2: # failure
+        logger.Log("Failure in %s: %s" % (test_result.GetTestName(),
+                                          test_result.GetFailureReason()))
+        fail_count+=1
+      total_count+=1
+    logger.Log("Tests run: %d, Failures: %d, Errors: %d" %
+               (total_count, fail_count, error_count))
 
   def _CollectTestSources(self, test_list, dirname, files):
     """For each directory, find tests source file and add them to the list.