def test_commit_serialization(self):
assert_commit_serialization(self.gitrwrepo, self.gitrwrepo.head, True)
rwrepo = self.gitrwrepo
make_object = rwrepo.odb.store
# direct serialization - deserialization can be tested afterwards
# serialization is probably limited on IO
hc = rwrepo.commit(rwrepo.head)
nc = 5000
st = time()
for i in xrange(nc):
cm = Commit(rwrepo, Commit.NULL_BIN_SHA, hc.tree,
hc.author, hc.authored_date, hc.author_tz_offset,
hc.committer, hc.committed_date, hc.committer_tz_offset,
str(i), parents=hc.parents, encoding=hc.encoding)
stream = BytesIO()
cm._serialize(stream)
slen = stream.tell()
stream.seek(0)
cm.binsha = make_object(IStream(Commit.type, slen, stream)).binsha
# END commit creation
elapsed = time() - st
print("Serialized %i commits to loose objects in %f s ( %f commits / s )"
% (nc, elapsed, nc / elapsed), file=sys.stderr)
def assert_commit_serialization(self,
rwrepo,
commit_id,
print_performance_info=False):
"""traverse all commits in the history of commit identified by commit_id and check
if the serialization works.
:param print_performance_info: if True, we will show how fast we are"""
ns = 0 # num serializations
nds = 0 # num deserializations
st = time.time()
for cm in rwrepo.commit(commit_id).traverse():
nds += 1
# assert that we deserialize commits correctly, hence we get the same
# sha on serialization
stream = BytesIO()
cm._serialize(stream)
ns += 1
streamlen = stream.tell()
stream.seek(0)
istream = rwrepo.odb.store(IStream(Commit.type, streamlen, stream))
self.assertEqual(istream.hexsha, cm.hexsha.encode('ascii'))
nc = Commit(rwrepo, Commit.NULL_BIN_SHA, cm.tree, cm.author,
cm.authored_date, cm.author_tz_offset, cm.committer,
cm.committed_date, cm.committer_tz_offset, cm.message,
cm.parents, cm.encoding)
self.assertEqual(nc.parents, cm.parents)
stream = BytesIO()
nc._serialize(stream)
ns += 1
streamlen = stream.tell()
stream.seek(0)
# reuse istream
istream.size = streamlen
istream.stream = stream
istream.binsha = None
nc.binsha = rwrepo.odb.store(istream).binsha
# if it worked, we have exactly the same contents !
self.assertEqual(nc.hexsha, cm.hexsha)
# END check commits
elapsed = time.time() - st
if print_performance_info:
print(
"Serialized %i and deserialized %i commits in %f s ( (%f, %f) commits / s"
% (ns, nds, elapsed, ns / elapsed, nds / elapsed),
file=sys.stderr)
def assert_commit_serialization(rwrepo, commit_id, print_performance_info=False):
"""traverse all commits in the history of commit identified by commit_id and check
if the serialization works.
:param print_performance_info: if True, we will show how fast we are"""
ns = 0 # num serializations
nds = 0 # num deserializations
st = time.time()
for cm in rwrepo.commit(commit_id).traverse():
nds += 1
# assert that we deserialize commits correctly, hence we get the same
# sha on serialization
stream = BytesIO()
cm._serialize(stream)
ns += 1
streamlen = stream.tell()
stream.seek(0)
istream = rwrepo.odb.store(IStream(Commit.type, streamlen, stream))
assert istream.hexsha == cm.hexsha.encode('ascii')
nc = Commit(rwrepo, Commit.NULL_BIN_SHA, cm.tree,
cm.author, cm.authored_date, cm.author_tz_offset,
cm.committer, cm.committed_date, cm.committer_tz_offset,
cm.message, cm.parents, cm.encoding)
assert nc.parents == cm.parents
stream = BytesIO()
nc._serialize(stream)
ns += 1
streamlen = stream.tell()
stream.seek(0)
# reuse istream
istream.size = streamlen
istream.stream = stream
istream.binsha = None
nc.binsha = rwrepo.odb.store(istream).binsha
# if it worked, we have exactly the same contents !
assert nc.hexsha == cm.hexsha
# END check commits
elapsed = time.time() - st
if print_performance_info:
print("Serialized %i and deserialized %i commits in %f s ( (%f, %f) commits / s"
% (ns, nds, elapsed, ns / elapsed, nds / elapsed), file=sys.stderr)
def test_commit_serialization(self):
self.assert_commit_serialization(self.gitrwrepo, '58c78e6', True)
rwrepo = self.gitrwrepo
make_object = rwrepo.odb.store
# direct serialization - deserialization can be tested afterwards
# serialization is probably limited on IO
hc = rwrepo.commit(rwrepo.head)
nc = 5000
st = time()
for i in range(nc):
cm = Commit(rwrepo,
Commit.NULL_BIN_SHA,
hc.tree,
hc.author,
hc.authored_date,
hc.author_tz_offset,
hc.committer,
hc.committed_date,
hc.committer_tz_offset,
str(i),
parents=hc.parents,
encoding=hc.encoding)
stream = BytesIO()
cm._serialize(stream)
slen = stream.tell()
stream.seek(0)
cm.binsha = make_object(IStream(Commit.type, slen, stream)).binsha
# END commit creation
elapsed = time() - st
print(
"Serialized %i commits to loose objects in %f s ( %f commits / s )"
% (nc, elapsed, nc / elapsed),
file=sys.stderr)
def newcommit(repo, message, date, name=None, email=None):
"""Creates a commit object with a custom date.
:param repo: Repo object the commit should be part of
:param tree: Tree object
:param message: Commit message. It may be an empty string
if no message is provided.
:param date: Date in seconds, as an Int
:return: Commit object representing the new commit
:note:
Additional information about the committer and Author are taken from
the git configuration
"""
tree = repo.index.write_tree()
try:
parents = [repo.head.commit]
except ValueError:
parents = []
# Committer and Author info
cr = repo.config_reader()
if name == None or email == None:
actor = Actor.committer(cr)
else:
actor = Actor(name, email)
committer = actor
author = actor
# Date
# offset = altzone # 3600*offsethours
offset = 0 # UTC
author_time, author_offset = date, offset
committer_time, committer_offset = date, offset
# assume utf8 encoding
enc_section, enc_option = Commit.conf_encoding.split('.')
conf_encoding = cr.get_value(enc_section,
enc_option,
Commit.default_encoding)
# Create New Commit
new_commit = Commit(repo, Commit.NULL_BIN_SHA, tree,
author, author_time, author_offset,
committer, committer_time, committer_offset,
message, parents, conf_encoding)
stream = StringIO()
new_commit._serialize(stream)
streamlen = stream.tell()
stream.seek(0)
istream = repo.odb.store(IStream(Commit.type, streamlen, stream))
new_commit.binsha = istream.binsha
# Advance HEAD to the new commit automatically.
# need late import here, importing git at the very beginning throws.
import git.refs
try:
repo.head.set_commit(new_commit, logmsg="commit: %s" % message)
except ValueError:
# head is not yet set to the ref our HEAD points to
# Happens on first commit
import git.refs
master = git.refs.Head.create(repo, repo.head.ref,
new_commit, logmsg="commit (initial): %s" % message)
repo.head.set_reference(master, logmsg='commit: Switching to %s' % master)
# END handle empty repositories
return new_commit
