From 8528d27df64cc84163c98bd76cd4921165392ee7 Mon Sep 17 00:00:00 2001
From: "R. Tyler Ballance" <tyler@monkeypox.org>
Date: Thu, 8 Jan 2009 00:42:42 -0800
Subject: [PATCH] Support the statistics gathering for the churn graph

I'm still somewhat uncertain about how I want to create the graphs, I may need to add historgram support to CairoPlot to do precisely what I want:

			|
			|  |+++|
			|  |+++|
			|  |///|        ___
			|  |///|       |+++| <-- avg added per commit
			|  |///|       |+++|
			|__|///|_______|///|_<-- avg removed per commit
			  [tyler]     [other]

Uncertainly, BAH.

Signed-off-by: R. Tyler Ballance <tyler@monkeypox.org>
---
 git-plot.py     |  2 +-
 internal/log.py | 45 +++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/git-plot.py b/git-plot.py
index 449cc36..16c6f74 100755
--- a/git-plot.py
+++ b/git-plot.py
@@ -6,7 +6,7 @@ import internal
 import internal.log
 
 def main():
-	sub_commands = ['leaderboard', 'timeofday']
+	sub_commands = ['leaderboard', 'timeofday', 'churn']
 	usage = 'usage: %%prog [options] <%s>' % '|'.join(sub_commands)
 	_op = OptionParser(usage=usage)
 	_op.add_option('-d', '--directory', default='.', dest='directory', help='Directory of the Git repository to analyze')
diff --git a/internal/log.py b/internal/log.py
index a513fa4..2614cf0 100644
--- a/internal/log.py
+++ b/internal/log.py
@@ -68,15 +68,16 @@ class GitLog(object):
 				# OW OW OW OW 
 				numstat = [n.split('\t') for n in r[1].strip().split('\n')]
 				def _numstat_gen(line):
-					d = {'added' : 0, 'removed' : 0, 'filename' : line[2], 'newfile' : False}
+					d = {'added' : 0, 'removed' : 0, 'filename' : line[2], 'binary' : False}
 					if line[0] == '-' and line[1] == '-':
-						d['newfile'] = True
+						d['binary'] = True
 					else:
-						d['added'], d['removed'] = line[0],line[1]
+						d['added'], d['removed'] = int(line[0]),int(line[1])
 					return d
 				t['numstat'] = map(lambda n: len(n) == 3 and _numstat_gen(n), numstat)
+				t['numstat'] = [nm for nm in t['numstat'] if nm]
 			else:
-				t['numstat'] = None
+				t['numstat'] = []
 			rc.append(t)
 		self.results = rc
 		def timeofday(f):
@@ -162,3 +163,39 @@ class GitLog(object):
 		except KeyError:
 			print 'This function does not support chart type: %s' % chart
 
+	def churn(self, width=None, height=None, filename=None, chart=internal.ChartType.Bar):
+		assert width or height
+		if not self.results:
+			self.load()
+		results = copy.deepcopy(self.results)
+		results.reverse()
+		filename = '%s_churn.png' % (self.filename or time.time())
+
+		rc = {}
+		for commit in results:
+			if not commit['numstat']:
+				continue
+			churn = []
+			for file in commit['numstat']:
+				if file['binary']:
+					continue
+				churn.append( (file['added'], file['removed'], file['filename']) )
+
+			author = commit['committer_handle']
+			if not rc.get(author):
+				rc[author] = {'commits' : 0, 'files' : [], 'added' : 0, 'removed' : 0}
+
+			rc[author]['commits'] += 1
+			rc[author]['files'].extend([f[2] for f in churn])
+			added = map(lambda c: c[0], churn)
+			removed = map(lambda c: c[1], churn)
+			rc[author]['added'] += added and reduce(lambda x,y: x+y, added) or 0
+			rc[author]['removed'] += removed and reduce(lambda x,y: x+y, removed) or 0
+		
+		for k in rc.keys():
+			rc[k]['files'] = list(set(rc[k]['files']))
+			rc[k]['distinct_files'] = len(rc[k]['files'])
+		
+		for k,r in rc.iteritems():
+			print '%s touched %d files, +%s, -%s lines (total: %s, net: %s)' % (k, r['distinct_files'], r['added'], r['removed'], r['added']+r['removed'], r['added']-r['removed'])
+