Improve json_comment_eater.py performance.

PRESUBMIT.py

Index: PRESUBMIT.py
diff --git a/PRESUBMIT.py b/PRESUBMIT.py
index ce91a4383343b14bdfbf264eb72a845c9ffd2988..004e59acbc355f64a1fb0f83a0caac5ef99a85a0 100644
--- a/PRESUBMIT.py
+++ b/PRESUBMIT.py
@@ -1413,6 +1413,9 @@ def _CheckParseErrors(input_api, output_api):
   excluded_patterns = [
     r'test[\\\/]data[\\\/]',
     r'^components[\\\/]policy[\\\/]resources[\\\/]policy_templates\.json$',
+    # Contains valid JSON but is excluded for performance reasons. The
+    # well-formedness of this file is checked during the build.
+    r'^net[\\\/]http[\\\/]transport_security_state_static\.json$',

[Dirk Pranke, 2017/04/10 19:34:27]

Seems like we should fix the comment eater instead, there must be a pathological
bug in the parser for it to be this slow.

However, I'm not really okay with checking things in that call themselves json
but aren't actually json, so theoretically these should be .json5 or something
else instead.

But either of those things are more work.

Can you file a bug to follow up on this and cc me on it?

[martijnc, 2017/04/11 20:18:13]

I've updated the CL to fix the performance issue instead.
Filed https://crbug.com/710587 for the incorrect extensions but I don't have
editbug privileges so I can't CC you.

   ]
   # Most JSON files are preprocessed and support comments, but these do not.
   json_no_comments_patterns = [