From: George Karpenkov <ekarpenkov@apple.com>
Date: Tue, 23 Oct 2018 01:30:26 +0000 (+0000)
Subject: [analyzer] [testing] Compute data on path length, compute percentiles
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=251cc4f6acd8f94572e1db2e9a113695efbde145;p=clang

[analyzer] [testing] Compute data on path length, compute percentiles

Differential Revision: https://reviews.llvm.org/D52844

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@344990 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/utils/analyzer/CmpRuns.py b/utils/analyzer/CmpRuns.py
index 7c9744727e..ab8c48b023 100755
--- a/utils/analyzer/CmpRuns.py
+++ b/utils/analyzer/CmpRuns.py
@@ -281,9 +281,21 @@ def compareResults(A, B, opts):
 
     return res
 
+def computePercentile(l, percentile):
+    """
+    Return computed percentile.
+    """
+    return sorted(l)[int(round(percentile * len(l) + 0.5)) - 1]
+
 def deriveStats(results):
     # Assume all keys are the same in each statistics bucket.
     combined_data = defaultdict(list)
+
+    # Collect data on paths length.
+    for report in results.reports:
+        for diagnostic in report.diagnostics:
+            combined_data['PathsLength'].append(diagnostic.getPathLength())
+
     for stat in results.stats:
         for key, value in stat.iteritems():
             combined_data[key].append(value)
@@ -293,6 +305,8 @@ def deriveStats(results):
             "max": max(values),
             "min": min(values),
             "mean": sum(values) / len(values),
+            "90th %tile": computePercentile(values, 0.9),
+            "95th %tile": computePercentile(values, 0.95),
             "median": sorted(values)[len(values) / 2],
             "total": sum(values)
         }