From dd45a96f0a035d10cc12973e9931df3c9d95a781 Mon Sep 17 00:00:00 2001
From: Jonathan Corbet <corbet@lwn.net>
Date: Fri, 8 Feb 2013 17:21:25 -0700
Subject: Add percentages to the touched-files report

...oh, and while I'm at it, make that report actually work right so that a
patch touching three driver files only shows up once at the drivers/ level.
---
 gitdm      | 27 ++++++++++++++++++---------
 reports.py |  5 +++--
 2 files changed, 21 insertions(+), 11 deletions(-)

diff --git a/gitdm b/gitdm
index d794d2f..14dc88e 100755
--- a/gitdm
+++ b/gitdm
@@ -134,14 +134,24 @@ def AddAccess(path):
     except KeyError:
         FileAccesses[path] = 1
 
-def NoteFileAccess(path):
-    if path.startswith('a/') or path.startswith('b/'):
-        path = path[2:]
-    AddAccess(path)
-    path, last = os.path.split(path)
-    while path and path not in ['a', 'b', '/']:
+def NoteFileAccess(paths):
+    #
+    # Keep separate track of what we've noted in this set so that each level
+    # of the tree only gets a single note from one patch.
+    #
+    noted = [ ]
+    for path in paths:
+        if path.startswith('a/') or path.startswith('b/'):
+            path = path[2:]
         AddAccess(path)
+        noted.append(path)
         path, last = os.path.split(path)
+        while path and path not in ['a', 'b', '/']:
+            if path in noted:
+                break
+            noted.append(path)
+            AddAccess(path)
+            path, last = os.path.split(path)
 
 
 
@@ -491,8 +501,7 @@ for logpatch in patches:
     # Now note the file accesses if need be.
     #
     if FileReport:
-        for file in p.files:
-            NoteFileAccess(file)
+        NoteFileAccess(p.files)
     #
     # Record some global information - but only if this patch had
     # stuff which wasn't ignored.
@@ -567,4 +576,4 @@ if ReportByFileType and Numstat:
     reports.ReportByFileType(hlist)
 
 if FileReport:
-    reports.FileAccessReport(FileReport, FileAccesses)
+    reports.FileAccessReport(FileReport, FileAccesses, CSCount)
diff --git a/reports.py b/reports.py
index 78800bd..084089e 100644
--- a/reports.py
+++ b/reports.py
@@ -455,10 +455,11 @@ def ReportByFileType(hacker_list):
 #
 # The file access report is a special beast.
 #
-def FileAccessReport(name, accesses):
+def FileAccessReport(name, accesses, total):
     outf = open(name, 'w')
     files = accesses.keys()
     files.sort()
     for file in files:
-        outf.write('%6d %s\n' % (accesses[file], file))
+        a = accesses[file]
+        outf.write('%6d %6.1f%% %s\n' % (a, (100.0*a)/total, file))
     outf.close()
-- 
cgit v1.2.3