Add feature to run benchmarks on existing directory, not generated data.

1 files changed, 20 insertions, 2 deletions

diff --git a/seivot b/seivot
index fdaf48d..26f9e53 100755
--- a/seivot
+++ b/seivot
@@ -324,6 +324,9 @@ class Seivot(cliapp.Application):
                                     '(%default)',
                                metavar='SIZE',
                                default=1024)
+        self.settings.string(['use-existing'],
+                             'use exiting DIR for initial generation',
+                             metavar='DIR')
         self.settings.bytesize(['file-size'], 'size of files to create',
                                default=16*1024)
         self.settings.string(['obnam-branch'],
@@ -377,8 +380,15 @@ class Seivot(cliapp.Application):
         
         self.report = Report(prog)
 
-        self.generate_live_data(self.live_data, self.settings['initial-data'])
-        self.measure(prog.backup, 0, self.settings['initial-data'])
+        if self.settings['use-existing']:
+            os.rmdir(self.live_data)
+            shutil.copytree(self.settings['use-existing'], self.live_data,
+                            symlinks=True)
+            amount = self.file_sizes(self.live_data)
+        else:
+            amount = self.settings['initial-data']
+            self.generate_live_data(self.live_data, amount)
+        self.measure(prog.backup, 0, amount)
 
         for i in range(1, generations):
             self.generate_live_data(self.live_data,
@@ -408,6 +418,14 @@ class Seivot(cliapp.Application):
         runcmd(['genbackupdata', where, '--create', str(size),
                 '--file-size', str(self.settings['file-size'])])
 
+    def file_sizes(self, dirname):
+        bytes = 0
+        for dirname, subdirs, basenames in os.walk(dirname):
+            for filename in [os.path.join(dirname, x) for x in basenames]:
+                if os.path.isfile(filename):
+                    bytes += os.path.getsize(filename)
+        return bytes
+
     def measure(self, func, nth_gen, new_data, **kwargs):
         logging.info('Measuring %s gen %d' % (func.__name__, nth_gen))
         print 'Measuring %s generation %s' % (func.__name__, nth_gen)