commit.create_from_tree now uses pure python implementation, fixed message parsing which truncated newlines although it was ilegitimate. Its up to the reader to truncate therse, nowhere in the git code I could find anyone adding newlines to commits where it is written

Added performance tests for serialization, it does about 5k commits per second if writing to tmpfs

7 files changed, 125 insertions, 18 deletions

diff --git a/test/git/performance/lib.py b/test/git/performance/lib.py
index 4b552b20..650bea82 100644
--- a/test/git/performance/lib.py
+++ b/test/git/performance/lib.py
@@ -1,6 +1,8 @@
 """Contains library functions"""
 import os
 from test.testlib import *
+import shutil
+import tempfile
 
 from git import (
 	Repo
@@ -25,7 +27,7 @@ def resolve_or_fail(env_var):
 
 #{ Base Classes 
 
-class TestBigRepoReadOnly(TestBase):
+class TestBigRepoR(TestBase):
 	"""TestCase providing access to readonly 'big' repositories using the following 
 	member variables:
 	
@@ -40,7 +42,24 @@ class TestBigRepoReadOnly(TestBase):
 	
 	@classmethod
 	def setUpAll(cls):
-		super(TestBigRepoReadOnly, cls).setUpAll()
-		cls.gitrepo = Repo(resolve_or_fail(k_env_git_repo))
+		super(TestBigRepoR, cls).setUpAll()
+		cls.gitrorepo = Repo(resolve_or_fail(k_env_git_repo))
 
+
+class TestBigRepoRW(TestBigRepoR):
+	"""As above, but provides a big repository that we can write to.
+	
+	Provides ``self.gitrwrepo``"""
+	
+	@classmethod
+	def setUpAll(cls):
+		super(TestBigRepoRW, cls).setUpAll()
+		dirname = tempfile.mktemp()
+		os.mkdir(dirname)
+		cls.gitrwrepo = cls.gitrorepo.clone(dirname, shared=True, bare=True)
+	
+	@classmethod
+	def tearDownAll(cls):
+		shutil.rmtree(cls.gitrwrepo.working_tree_dir)
+		
 #} END base classes
diff --git a/test/git/performance/test_commit.py b/test/git/performance/test_commit.py
index b4a9d868..2398c93d 100644
--- a/test/git/performance/test_commit.py
+++ b/test/git/performance/test_commit.py
@@ -6,10 +6,12 @@
 
 from lib import *
 from git import *
+from test.git.test_commit import assert_commit_serialization
+from cStringIO import StringIO
 from time import time
 import sys
 
-class TestPerformance(TestBigRepoReadOnly):
+class TestPerformance(TestBigRepoRW):
 
 	# ref with about 100 commits in its history
 	ref_100 = '0.1.6'
@@ -48,7 +50,7 @@ class TestPerformance(TestBigRepoReadOnly):
 		# bound to cat-file parsing performance
 		nc = 0
 		st = time()
-		for c in self.gitrepo.commit(self.head_sha_2k).traverse(branch_first=False):
+		for c in self.gitrorepo.commit(self.head_sha_2k).traverse(branch_first=False):
 			nc += 1
 			self._query_commit_info(c)
 		# END for each traversed commit
@@ -59,10 +61,38 @@ class TestPerformance(TestBigRepoReadOnly):
 		# bound to stream parsing performance
 		nc = 0
 		st = time()
-		for c in Commit.iter_items(self.gitrepo, self.head_sha_2k):
+		for c in Commit.iter_items(self.gitrorepo, self.head_sha_2k):
 			nc += 1
 			self._query_commit_info(c)
 		# END for each traversed commit
 		elapsed_time = time() - st
 		print >> sys.stderr, "Iterated %i Commits in %s [s] ( %f commits/s )" % (nc, elapsed_time, nc/elapsed_time)
 		
+	def test_commit_serialization(self):
+		assert_commit_serialization(self.gitrwrepo, self.head_sha_2k, True)
+		
+		rwrepo = self.gitrwrepo
+		make_object = rwrepo.odb.to_object
+		# direct serialization - deserialization can be tested afterwards
+		# serialization is probably limited on IO
+		hc = rwrepo.commit(self.head_sha_2k)
+		
+		commits = list()
+		nc = 5000
+		st = time()
+		for i in xrange(nc):
+			cm = Commit(	rwrepo, Commit.NULL_HEX_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 = StringIO()
+			cm._serialize(stream)
+			slen = stream.tell()
+			stream.seek(0)
+			
+			cm.sha = make_object(Commit.type, slen, stream)
+		# END commit creation
+		elapsed = time() - st
+		
+		print >> sys.stderr, "Serialized %i commits to loose objects in %f s ( %f commits / s )" % (nc, elapsed, nc / elapsed)
diff --git a/test/git/performance/test_odb.py b/test/git/performance/test_odb.py
index 0ad2ce33..7b1ee838 100644
--- a/test/git/performance/test_odb.py
+++ b/test/git/performance/test_odb.py
@@ -5,18 +5,18 @@ import sys
 import stat
 
 from lib import (
-	TestBigRepoReadOnly
+	TestBigRepoR
 	)
 
 
-class TestObjDBPerformance(TestBigRepoReadOnly):
+class TestObjDBPerformance(TestBigRepoR):
 	
 	def test_random_access(self):
 		
 		# GET COMMITS
 		# TODO: use the actual db for this
 		st = time()
-		root_commit = self.gitrepo.commit(self.head_sha_2k)
+		root_commit = self.gitrorepo.commit(self.head_sha_2k)
 		commits = list(root_commit.traverse())
 		nc = len(commits)
 		elapsed = time() - st
diff --git a/test/git/performance/test_streams.py b/test/git/performance/test_streams.py
index 6c2834b3..d31bee14 100644
--- a/test/git/performance/test_streams.py
+++ b/test/git/performance/test_streams.py
@@ -14,7 +14,7 @@ import subprocess
 
 
 from lib import (
-	TestBigRepoReadOnly
+	TestBigRepoR
 	)
 
 
@@ -32,7 +32,7 @@ def make_memory_file(size_in_bytes, randomize=False):
 	return actual_size*4, StringIO(a.tostring())
 
 
-class TestObjDBPerformance(TestBigRepoReadOnly):
+class TestObjDBPerformance(TestBigRepoR):
 	
 	large_data_size_bytes = 1000*1000*10		# some MiB should do it
 	moderate_data_size_bytes = 1000*1000*1		# just 1 MiB
diff --git a/test/git/performance/test_utils.py b/test/git/performance/test_utils.py
index 381d7c8b..47366d34 100644
--- a/test/git/performance/test_utils.py
+++ b/test/git/performance/test_utils.py
@@ -4,11 +4,11 @@ import sys
 import stat
 
 from lib import (
-	TestBigRepoReadOnly
+	TestBigRepoR
 	)
 
 
-class TestUtilPerformance(TestBigRepoReadOnly):
+class TestUtilPerformance(TestBigRepoR):
 	
 	def test_access(self):
 		# compare dict vs. slot access
diff --git a/test/git/test_commit.py b/test/git/test_commit.py
index ad7a0082..a5f184e6 100644
--- a/test/git/test_commit.py
+++ b/test/git/test_commit.py
@@ -7,6 +7,56 @@
 from test.testlib import *
 from git import *
 
+from cStringIO import StringIO
+import time
+import sys
+
+
+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 = StringIO()
+		cm._serialize(stream)
+		ns += 1
+		streamlen = stream.tell()
+		stream.seek(0)
+		
+		csha = rwrepo.odb.to_object(Commit.type, streamlen, stream)
+		assert csha == cm.sha
+		
+		nc = Commit(rwrepo, Commit.NULL_HEX_SHA, cm.tree.sha,
+						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 = StringIO()
+		nc._serialize(stream)
+		ns += 1
+		streamlen = stream.tell()
+		stream.seek(0)
+		nc.sha = rwrepo.odb.to_object(Commit.type, streamlen, stream)
+		
+		# if it worked, we have exactly the same contents !
+		assert nc.sha == cm.sha
+	# END check commits
+	elapsed = time.time() - st
+	
+	if print_performance_info:
+		print >> sys.stderr, "Serialized %i and deserialized %i commits in %f s ( (%f, %f) commits / s" % (ns, nds, elapsed, ns/elapsed, nds/elapsed)
+	# END handle performance info
+	
+
 class TestCommit(TestBase):
 
 	def test_bake(self):
@@ -19,7 +69,7 @@ class TestCommit(TestBase):
 		assert commit.author == commit.committer
 		assert isinstance(commit.authored_date, int) and isinstance(commit.committed_date, int)
 		assert isinstance(commit.author_tz_offset, int) and isinstance(commit.committer_tz_offset, int)
-		assert commit.message == "Added missing information to docstrings of commit and stats module"
+		assert commit.message == "Added missing information to docstrings of commit and stats module\n"
 
 
 	def test_stats(self):
@@ -49,7 +99,7 @@ class TestCommit(TestBase):
 		assert commit.committed_date == 1210193388
 		assert commit.author_tz_offset == 14400, commit.author_tz_offset
 		assert commit.committer_tz_offset == 14400, commit.committer_tz_offset
-		assert commit.message == "initial project"
+		assert commit.message == "initial project\n"
 		
 	def test_traversal(self):
 		start = self.rorepo.commit("a4d06724202afccd2b5c54f81bcf2bf26dea7fff")
@@ -171,3 +221,8 @@ class TestCommit(TestBase):
 		name_rev = self.rorepo.head.commit.name_rev
 		assert isinstance(name_rev, basestring)
 		
+	@with_bare_rw_repo
+	def test_serialization(self, rwrepo):
+		# create all commits of our repo
+		assert_commit_serialization(rwrepo, '0.1.6')
+		
diff --git a/test/git/test_utils.py b/test/git/test_utils.py
index 2c3c392b..83ef7e4b 100644
--- a/test/git/test_utils.py
+++ b/test/git/test_utils.py
@@ -116,8 +116,6 @@ class TestUtils(TestCase):
 			os.remove(my_file)
 		# END final cleanup
 		
-	  
-		
 	def test_user_id(self):
 		assert '@' in get_user_id()
 		
@@ -127,7 +125,12 @@ class TestUtils(TestCase):
 			assert len(rval) == 2
 			assert isinstance(rval[0], int) and isinstance(rval[1], int)
 			assert rval[0] == veri_time
-			assert rval[1] == offset 
+			assert rval[1] == offset
+			
+			# now that we are here, test our conversion functions as well
+			utctz = altz_to_utctz_str(offset)
+			assert isinstance(utctz, basestring)
+			assert utctz_to_altz(verify_utctz(utctz)) == offset
 		# END assert rval utility
 		
 		rfc = ("Thu, 07 Apr 2005 22:13:11 +0000", 0)