user activity: Normalize the pathname from pprof and CWP data.

I noticed that lots of files have .. symbols in the pathname.

BUG=None
TEST=None

Change-Id: I7dbdc84b36b9496f2d8fa21ec4fafe0f1417f256
Reviewed-on: https://chrome-internal-review.googlesource.com/282197
Tested-by: Evelina Dumitrescu <evelinad@google.com>
Commit-Queue: Evelina Dumitrescu <evelinad@google.com>
Reviewed-by: George Burgess <gbiv@google.com>
Reviewed-by: Ting-Yuan Huang <laszio@google.com>
Reviewed-on: https://chromium-review.googlesource.com/435905
Commit-Ready: Luis Lozano <llozano@chromium.org>
Tested-by: Luis Lozano <llozano@chromium.org>
Reviewed-by: Luis Lozano <llozano@chromium.org>

3 files changed, 24 insertions, 15 deletions

diff --git a/user_activity_benchmarks/process_hot_functions.py b/user_activity_benchmarks/process_hot_functions.py
index 3c7d65df..f021e8c7 100755
--- a/user_activity_benchmarks/process_hot_functions.py
+++ b/user_activity_benchmarks/process_hot_functions.py
@@ -110,7 +110,7 @@ class HotFunctionsProcessor(object):
 
       for statistic in statistics_reader:
         function_name = statistic['function']
-        file_name = statistic['file']
+        file_name = os.path.normpath(statistic['file'])
         dso_name = statistic['dso']
         inclusive_count = statistic['inclusive_count']
 
@@ -153,7 +153,10 @@ class HotFunctionsProcessor(object):
     pprof_files = os.listdir(pprof_path)
     cwp_statistics = self.ParseCWPStatistics(cwp_functions_file)
     function_statistic_regex = re.compile(r'\S+\s+\S+%\s+\S+%\s+\S+\s+\S+%')
-    function_regex = re.compile(r'[a-zA-Z0-9-/_:.~\[\]]+[ a-zA-Z0-9-/_~:.]*')
+    function_regex = re.compile(r'[a-zA-Z0-9-_:.~\[\]]+')
+    # TODO(evelinad): Consider the case where the file name can have other
+    # characters.
+    file_regex = re.compile(r'[a-zA-Z0-9-/_.]+')
 
     for pprof_file in pprof_files:
       # In the pprof output, the statistics of the functions start from the
@@ -168,15 +171,21 @@ class HotFunctionsProcessor(object):
           function_statistic_regex.search(pprof_statistic)
         function_statistic = \
           ','.join(function_statistic_match.group(0).split())
-        function_match = function_regex.search(pprof_statistic[
-            function_statistic_match.end():])
-        function = ','.join(function_match.group(0).split())
-
-        if function in cwp_statistics:
-          cwp_statistic = cwp_statistics[function]
-          output_lines.append(','.join([function, cwp_statistic[0],
+        lookup_index = function_statistic_match.end()
+        function_match = function_regex.search(pprof_statistic[lookup_index:])
+        function_name = function_match.group(0)
+        lookup_index += function_match.end()
+        file_match = file_regex.search(pprof_statistic[lookup_index:])
+        if file_match:
+          key = ",".join([function_name, os.path.normpath(file_match.group(0))])
+        else:
+          key = function_name
+
+        if key in cwp_statistics:
+          cwp_statistic = cwp_statistics[key]
+          output_lines.append(','.join([key, cwp_statistic[0],
                                         function_statistic]))
-          cwp_statistics[function] = (cwp_statistic[0], self.COMMON_FUNCTION)
+          cwp_statistics[key] = (cwp_statistic[0], self.COMMON_FUNCTION)
 
       with open(os.path.join(common_functions_path, pprof_file), 'w') \
         as output_file:
diff --git a/user_activity_benchmarks/testdata/input/parse_cwp_statistics.csv b/user_activity_benchmarks/testdata/input/parse_cwp_statistics.csv
index 833beeed..a4c7ced9 100644
--- a/user_activity_benchmarks/testdata/input/parse_cwp_statistics.csv
+++ b/user_activity_benchmarks/testdata/input/parse_cwp_statistics.csv
@@ -1,6 +1,6 @@
 function,file,dso,inclusive_count
-dummy_method1,dummy_file1,dummy_object1,1
-dummy_method2,dummy_file2,dummy_object2,2
+dummy_method1,dummy_file1/a/b/../../,dummy_object1,1
+dummy_method2,dummy_file2//,dummy_object2,2
 ,,321223321,1
-dummy_method3,dummy_file3,dummy_object3,3
-dummy_method4,dummy_file4,dummy_object4,4
+dummy_method3,dummy_file3/a/../,dummy_object3,3
+dummy_method4,dummy_file4/./,dummy_object4,4
diff --git a/user_activity_benchmarks/testdata/input/pprof/file2.pprof b/user_activity_benchmarks/testdata/input/pprof/file2.pprof
index c1e6331d..6d22bff3 100644
--- a/user_activity_benchmarks/testdata/input/pprof/file2.pprof
+++ b/user_activity_benchmarks/testdata/input/pprof/file2.pprof
@@ -4,7 +4,7 @@ Type: instructions_event
 Showing nodes accounting for 48939666671, 42.14% of 116136877744 total
 Dropped 35196 nodes (cum <= 580684388)
       flat  flat%   sum%        cum   cum%
-      4585860529  3.95%  3.95% 13583834527 11.70%  blink::InvalidationSet::invalidatesElement /home/chrome-bot/chrome_root/src/third_party/WebKit/Source/core/css/invalidation/InvalidationSet.cpp
+      4585860529  3.95%  3.95% 13583834527 11.70%  blink::InvalidationSet::invalidatesElement /home/chrome-bot/chrome_root/src/third_party/WebKit/Source/core/css/invalidation/a/b/../../InvalidationSet.cpp
       3791928512  3.27%  7.21% 35145646088 30.26%  blink::StyleInvalidator::invalidate /home/chrome-bot/chrome_root/src/third_party/WebKit/Source/core/css/invalidation/StyleInvalidator.cpp
       2871318565  2.47%  9.69% 2979878602  2.57%  blink::StyleInvalidator::RecursionCheckpoint::~RecursionCheckpoint /home/chrome-bot/chrome_root/src/third_party/WebKit/Source/core/css/invalidation/StyleInvalidator.h (inline)
       1914657964  1.65% 11.33% 2164475253  1.86%  WTF::StringImpl::lower /home/chrome-bot/chrome_root/src/third_party/WebKit/Source/wtf/text/StringImpl.cpp