Add a --json option to git cl try-results.

tests/git_cl_test.py

Index: tests/git_cl_test.py
diff --git a/tests/git_cl_test.py b/tests/git_cl_test.py
index 9796b2bcb39430896f1985deba111aa3dd13f0e6..f7aabca98051ddc25d07abf53c7be9c633b4d47b 100755
--- a/tests/git_cl_test.py
+++ b/tests/git_cl_test.py
@@ -1484,7 +1484,7 @@ class TestGitCl(TestCase):
     })
     self.calls.append(
         ((['ask_for_data', 'If you know what you are doing, '
-				   								 'press Enter to continue, Ctrl+C to abort.'],), ''))
+                           'press Enter to continue, Ctrl+C to abort.'],), ''))
     self.assertIsNone(cl.EnsureAuthenticated(force=False))
 
   def test_gerrit_ensure_authenticated_ok(self):
@@ -1850,6 +1850,50 @@ class TestGitCl(TestCase):
         out.getvalue(),
         'scheduled CQ Dry Run on https://codereview.chromium.org/123\n')
 
+  def test_write_try_results_json(self):
+    builds = {
+        '9000': {
+            'id': '9000',
+            'status': 'STARTED',
+            'url': 'http://build.cr.org/p/x.y/builders/my-builder/builds/2',
+            'result_details_json': '{"properties": {}}',
+            'bucket': 'master.x.y',
+            'created_by': 'user:someone@chromium.org',
+            'created_ts': '147200002222000',
+            'parameters_json': '{"builder_name": "my-builder"}',
+        },
+        '8000': {
+            'id': '8000',
+            'status': 'COMPLETED',
+            'result': 'FAILURE',
+            'failure_reason': 'BUILD_FAILURE',
+            'url': 'http://build.cr.org/p/x.y/builders/my-builder/builds/1',
+            'result_details_json': '{"properties": {}}',
+            'bucket': 'master.x.y',
+            'created_by': 'user:someone@chromium.org',
+            'created_ts': '147200001111000',
+            'parameters_json': '{"builder_name": "my-builder"}',
+        },
+    }
+    expected_output = [
+        {
+            'buildbucket_id': '8000',
+            'builder_name': 'my-builder',
+            'status': 'COMPLETED',
+            'result': 'FAILURE',
+            'failure_reason': 'BUILD_FAILURE',
+        },
+        {
+            'buildbucket_id': '9000',
+            'builder_name': 'my-builder',
+            'status': 'STARTED',
+            'result': None,
+            'failure_reason': None,
+        }
+    ]
+    self.calls = [(('write_json', 'output.json', expected_output), '')]
+    git_cl.write_try_results_json('output.json', builds)

[qyearsley, 2016/08/30 22:13:22]

I'm not sure whether it would be worth it to write a method that invokes the
top-level command function (i.e. git_cl.main(['try-results', '--json',
'output.json
])); I imagine such a method would assert that a series of calls is made (e.g.
['git', 'symbolic-ref', 'HEAD'], ...) and then assert that write_json got called
in the expected way.

But, the above more specific test method already tests what the expected format
should be like, and I manually verified that git cl try-results --json
output.json works on my workstation.

[tandrii(chromium), 2016/08/30 22:22:45]

I'd keep more specific here, but it'd still have a general one for top level to
be sure nobody breaks your tooling, for which you can mock
git_cl.write_try_results_json since it's tested here.

[qyearsley, 2016/08/30 23:41:25]

Alright, sounds reasonable ^^

+
   def _common_GerritCommitMsgHookCheck(self):
     self.mock(git_cl.sys, 'stdout', StringIO.StringIO())
     self.mock(git_cl.os.path, 'abspath',