print-commitsize

#!/usr/bin/env python
#
# Print statistics on the objects for a given ref/commit in an OSTree repo
#
# Copyright 2015 Colin Walters <walters@verbum.org>
# Licensed under the new-BSD license (http://www.opensource.org/licenses/bsd-license.php)

import gi
gi.require_version('OSTree', '1.0')
import sys,os
from gi.repository import GLib, Gio, OSTree

repopath = sys.argv[1]
ref = sys.argv[2]

r = OSTree.Repo.new(Gio.File.new_for_path(repopath))
r.open(None)

[_,rev] = r.resolve_rev(ref, False)

print "%s => %s" % (ref, rev)
[_,reachable] = r.traverse_commit(rev, 0, None)
total = 0
n_meta = 0
n_content = 0
buckets = {}
for k,v in reachable.iteritems():
    csum,objtype = k.unpack()
    if objtype == OSTree.ObjectType.FILE:
        [_,_,finfo,_] = r.load_file(csum, None)
        n_content += 1
        sz = finfo.get_size()
    else:
        [_,sz] = r.query_object_storage_size(objtype, csum, None)
        n_meta += 1
    sz_mb = sz / (1000*1000)
    [n,s] = buckets.get(sz_mb, [0,0])
    buckets[sz_mb] = (n+1, s+sz)
    total += sz

print "Meta: {0} Content: {1}".format(n_meta, n_content)
for v in sorted(buckets.keys()):
    [n,s] = buckets[v]
    print "%s => %s (%s MB)" % (v, n, s/(1000*1000))
print "Total: {0} MB".format(total / (1000*1000))
os._exit(0)