🎽 Fix instrumentation test proguard parser when gmscore has annotations

build/android/pylib/utils/proguard.py

 # Copyright 2014 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.
 
 import os
 import re
 import tempfile
 
 from devil.utils import cmd_helper
 from pylib import constants
@@ skipping 266 matching lines @@
         if m:
           state.InitAnnotationField(
             m.group(2), element_type, _GetDepth(m.group(1)))
           break
       if m:
         continue
       m = _PROGUARD_ANNOTATION_VALUE_RE.match(line)
       if m:
         state.UpdateCurrentAnnotationFieldValue(
           m.group(2), _GetDepth(m.group(1)))
+      else:
+        state.InitMethod(None)

[jbudorick, 2016/04/08 00:59:09]

What does this line that gets here contain? It's not empty, doesn't match any of
the other regexes, and we just parsed an annotation...?

[dgn, 2016/04/08 16:54:52]

So that clears the annotation stack when we hit something that is not an
annotation field value. Shouldn't the stack be cleared when we hit a new section
anyway? "- Signature attribute" should reset the state, right? 

[agrieve, 2016/04/08 17:35:36]

That's what we want, but it's not what the code was doing. It doesn't actually
look at indentation as a part of the core logic. It was just saying:

1. Keep looking until we find the start of a method
2. Then, keep looking until we find "Annotations"
3. Then, keep parsing annotations until we hit another method or section.

Our "sections" are defined as:
Interfaces|Constant Pool|Fields|Methods|Class file attributes

So, it was missing logic to handle more siblings of - Annotations.

+
 
   return results