Make seivot measure the size of the repository after each command.

1 files changed, 13 insertions, 0 deletions

diff --git a/seivot b/seivot
index f619e92..7b06105 100755
--- a/seivot
+++ b/seivot
@@ -33,6 +33,7 @@ class Measurement(object):
         self.real = fields[2]
         self.maxrss = fields[3]
         self.new_data = 0
+        self.repo_size_after = 0
 
 
 def runcmd(argv, **kwargs):
@@ -262,6 +263,7 @@ class Report(object):
                     cp.set(section, '%s.%s' % (op, field), 
                            '%.1f' % getattr(m, field))
                 cp.set(section, '%s.new-data' % op, m.new_data)
+                cp.set(section, '%s.repo-size-after' % op, m.repo_size_after)
         
         cp.write(fp)
 
@@ -369,12 +371,23 @@ class Seivot(cliapp.Application):
         drop_caches()
         measurement = func(nth_gen, **kwargs)
         measurement.new_data = new_data
+        measurement.repo_size_after = self.disk_usage(self.repo)
         self.report.add_measurement(func.__name__, nth_gen, measurement)
 
     def cleanup(self):
         logging.info('Removing temporary directory %s' % self.tempdir)
         shutil.rmtree(self.tempdir)
 
+    def disk_usage(self, pathname):
+        '''Simulate du(1). Return disk usage in bytes.'''
+        def getsize(filename):
+            return os.lstat(filename).st_blocks * 512
+        bytes = 0
+        for dirname, subdirs, basenames in os.walk(pathname):
+            bytes += getsize(dirname)
+            bytes += sum(getsize(os.path.join(dirname, x)) for x in basenames)
+        return bytes
+
 
 if __name__ == '__main__':
     Seivot().run()