dulwich/tests/test_pack.py

   1 # test_pack.py -- Tests for the handling of git packs.
   2 # Copyright (C) 2007 James Westby <jw+debian@jameswestby.net>
   3 # Copyright (C) 2008 Jelmer Vernooij <jelmer@samba.org>
   4 #
   5 # This program is free software; you can redistribute it and/or
   6 # modify it under the terms of the GNU General Public License
   7 # as published by the Free Software Foundation; version 2
   8 # of the License, or (at your option) any later version of the license.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with this program; if not, write to the Free Software
  17 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
  18 # MA  02110-1301, USA.
  19
  20 import os
  21 import unittest
  22
  23 from dulwich.objects import (
  24         Tree,
  25         )
  26 from dulwich.pack import (
  27         Pack,
  28         PackIndex,
  29         PackData,
  30         hex_to_sha,
  31         sha_to_hex,
  32         write_pack_index_v1,
  33         write_pack_index_v2,
  34         write_pack,
  35         apply_delta,
  36         create_delta,
  37         read_zlib,
  38         )
  39
  40 pack1_sha = 'bc63ddad95e7321ee734ea11a7a62d314e0d7481'
  41
  42 a_sha = '6f670c0fb53f9463760b7295fbb814e965fb20c8'
  43 tree_sha = 'b2a2766a2879c209ab1176e7e778b81ae422eeaa'
  44 commit_sha = 'f18faa16531ac570a3fdc8c7ca16682548dafd12'
  45
  46 class PackTests(unittest.TestCase):
  47   """Base class for testing packs"""
  48
  49   datadir = os.path.join(os.path.dirname(__file__), 'data/packs')
  50
  51   def get_pack_index(self, sha):
  52     """Returns a PackIndex from the datadir with the given sha"""
  53     return PackIndex(os.path.join(self.datadir, 'pack-%s.idx' % sha))
  54
  55   def get_pack_data(self, sha):
  56     """Returns a PackData object from the datadir with the given sha"""
  57     return PackData(os.path.join(self.datadir, 'pack-%s.pack' % sha))
  58
  59   def get_pack(self, sha):
  60     return Pack(os.path.join(self.datadir, 'pack-%s' % sha))
  61
  62
  63 class PackIndexTests(PackTests):
  64   """Class that tests the index of packfiles"""
  65
  66   def test_object_index(self):
  67     """Tests that the correct object offset is returned from the index."""
  68     p = self.get_pack_index(pack1_sha)
  69     self.assertEqual(p.object_index(pack1_sha), None)
  70     self.assertEqual(p.object_index(a_sha), 178)
  71     self.assertEqual(p.object_index(tree_sha), 138)
  72     self.assertEqual(p.object_index(commit_sha), 12)
  73
  74   def test_index_len(self):
  75     p = self.get_pack_index(pack1_sha)
  76     self.assertEquals(3, len(p))
  77
  78   def test_get_stored_checksum(self):
  79     p = self.get_pack_index(pack1_sha)
  80     self.assertEquals("\xf2\x84\x8e*\xd1o2\x9a\xe1\xc9.;\x95\xe9\x18\x88\xda\xa5\xbd\x01", str(p.get_stored_checksums()[1]))
  81     self.assertEquals( 'r\x19\x80\xe8f\xaf\x9a_\x93\xadgAD\xe1E\x9b\x8b\xa3\xe7\xb7' , str(p.get_stored_checksums()[0]))
  82
  83   def test_index_check(self):
  84     p = self.get_pack_index(pack1_sha)
  85     self.assertEquals(True, p.check())
  86
  87
  88   def test_iterentries(self):
  89     p = self.get_pack_index(pack1_sha)
  90     self.assertEquals([('og\x0c\x0f\xb5?\x94cv\x0br\x95\xfb\xb8\x14\xe9e\xfb \xc8', 178, None), ('\xb2\xa2vj(y\xc2\t\xab\x11v\xe7\xe7x\xb8\x1a\xe4"\xee\xaa', 138, None), ('\xf1\x8f\xaa\x16S\x1a\xc5p\xa3\xfd\xc8\xc7\xca\x16h%H\xda\xfd\x12', 12, None)], list(p.iterentries()))
  91
  92   def test_iter(self):
  93     p = self.get_pack_index(pack1_sha)
  94     self.assertEquals(set([tree_sha, commit_sha, a_sha]), set(p))
  95
  96
  97 class TestPackDeltas(unittest.TestCase):
  98
  99   test_string1 = "The answer was flailing in the wind"
 100   test_string2 = "The answer was falling down the pipe"
 101   test_string3 = "zzzzz"
 102
 103   test_string_empty = ""
 104   test_string_big = "Z" * 8192
 105
 106   def _test_roundtrip(self, base, target):
 107     self.assertEquals(target,
 108       apply_delta(base, create_delta(base, target)))
 109
 110   def test_nochange(self):
 111     self._test_roundtrip(self.test_string1, self.test_string1)
 112
 113   def test_change(self):
 114     self._test_roundtrip(self.test_string1, self.test_string2)
 115
 116   def test_rewrite(self):
 117     self._test_roundtrip(self.test_string1, self.test_string3)
 118
 119   def test_overflow(self):
 120     self._test_roundtrip(self.test_string_empty, self.test_string_big)
 121
 122
 123 class TestPackData(PackTests):
 124   """Tests getting the data from the packfile."""
 125
 126   def test_create_pack(self):
 127     p = self.get_pack_data(pack1_sha)
 128
 129   def test_pack_len(self):
 130     p = self.get_pack_data(pack1_sha)
 131     self.assertEquals(3, len(p))
 132
 133   def test_index_check(self):
 134     p = self.get_pack_data(pack1_sha)
 135     self.assertEquals(True, p.check())
 136
 137   def test_iterobjects(self):
 138     p = self.get_pack_data(pack1_sha)
 139     self.assertEquals([(12, 1, 'tree b2a2766a2879c209ab1176e7e778b81ae422eeaa\nauthor James Westby <jw+debian@jameswestby.net> 1174945067 +0100\ncommitter James Westby <jw+debian@jameswestby.net> 1174945067 +0100\n\nTest commit\n'), (138, 2, '100644 a\x00og\x0c\x0f\xb5?\x94cv\x0br\x95\xfb\xb8\x14\xe9e\xfb \xc8'), (178, 3, 'test 1\n')], list(p.iterobjects()))
 140
 141   def test_iterentries(self):
 142     p = self.get_pack_data(pack1_sha)
 143     self.assertEquals(set([('og\x0c\x0f\xb5?\x94cv\x0br\x95\xfb\xb8\x14\xe9e\xfb \xc8', 178, -1718046665), ('\xb2\xa2vj(y\xc2\t\xab\x11v\xe7\xe7x\xb8\x1a\xe4"\xee\xaa', 138, -901046474), ('\xf1\x8f\xaa\x16S\x1a\xc5p\xa3\xfd\xc8\xc7\xca\x16h%H\xda\xfd\x12', 12, 1185722901)]), set(p.iterentries()))
 144
 145   def test_create_index_v1(self):
 146     p = self.get_pack_data(pack1_sha)
 147     p.create_index_v1("v1test.idx")
 148     idx1 = PackIndex("v1test.idx")
 149     idx2 = self.get_pack_index(pack1_sha)
 150     self.assertEquals(idx1, idx2)
 151
 152   def test_create_index_v2(self):
 153     p = self.get_pack_data(pack1_sha)
 154     p.create_index_v2("v2test.idx")
 155     idx1 = PackIndex("v2test.idx")
 156     idx2 = self.get_pack_index(pack1_sha)
 157     self.assertEquals(idx1, idx2)
 158
 159
 160
 161 class TestPack(PackTests):
 162
 163     def test_len(self):
 164         p = self.get_pack(pack1_sha)
 165         self.assertEquals(3, len(p))
 166
 167     def test_contains(self):
 168         p = self.get_pack(pack1_sha)
 169         self.assertTrue(tree_sha in p)
 170
 171     def test_get(self):
 172         p = self.get_pack(pack1_sha)
 173         self.assertEquals(type(p[tree_sha]), Tree)
 174
 175     def test_iter(self):
 176         p = self.get_pack(pack1_sha)
 177         self.assertEquals(set([tree_sha, commit_sha, a_sha]), set(p))
 178
 179     def test_get_object_at(self):
 180         """Tests random access for non-delta objects"""
 181         p = self.get_pack(pack1_sha)
 182         obj = p[a_sha]
 183         self.assertEqual(obj._type, 'blob')
 184         self.assertEqual(obj.sha().hexdigest(), a_sha)
 185         obj = p[tree_sha]
 186         self.assertEqual(obj._type, 'tree')
 187         self.assertEqual(obj.sha().hexdigest(), tree_sha)
 188         obj = p[commit_sha]
 189         self.assertEqual(obj._type, 'commit')
 190         self.assertEqual(obj.sha().hexdigest(), commit_sha)
 191
 192     def test_copy(self):
 193         p = self.get_pack(pack1_sha)
 194         write_pack("Elch", [(x, "") for x in p.iterobjects()], len(p))
 195         self.assertEquals(p, Pack("Elch"))
 196
 197     def test_commit_obj(self):
 198         p = self.get_pack(pack1_sha)
 199         commit = p[commit_sha]
 200         self.assertEquals("James Westby <jw+debian@jameswestby.net>", commit.author)
 201         self.assertEquals([], commit.parents)
 202
 203     def test_name(self):
 204         p = self.get_pack(pack1_sha)
 205         self.assertEquals(pack1_sha, p.name())
 206
 207
 208 class TestHexToSha(unittest.TestCase):
 209
 210     def test_simple(self):
 211         self.assertEquals('\xab\xcd' * 10, hex_to_sha("abcd" * 10))
 212
 213     def test_reverse(self):
 214         self.assertEquals("abcd" * 10, sha_to_hex('\xab\xcd' * 10))
 215
 216
 217 class BaseTestPackIndexWriting(object):
 218
 219     def test_empty(self):
 220         pack_checksum = 'r\x19\x80\xe8f\xaf\x9a_\x93\xadgAD\xe1E\x9b\x8b\xa3\xe7\xb7'
 221         self._write_fn("empty.idx", [], pack_checksum)
 222         idx = PackIndex("empty.idx")
 223         self.assertTrue(idx.check())
 224         self.assertEquals(idx.get_stored_checksums()[0], pack_checksum)
 225         self.assertEquals(0, len(idx))
 226
 227     def test_single(self):
 228         pack_checksum = 'r\x19\x80\xe8f\xaf\x9a_\x93\xadgAD\xe1E\x9b\x8b\xa3\xe7\xb7'
 229         my_entries = [('og\x0c\x0f\xb5?\x94cv\x0br\x95\xfb\xb8\x14\xe9e\xfb \xc8', 178, 42)]
 230         my_entries.sort()
 231         self._write_fn("single.idx", my_entries, pack_checksum)
 232         idx = PackIndex("single.idx")
 233         self.assertEquals(idx.version, self._expected_version)
 234         self.assertTrue(idx.check())
 235         self.assertEquals(idx.get_stored_checksums()[0], pack_checksum)
 236         self.assertEquals(1, len(idx))
 237         actual_entries = list(idx.iterentries())
 238         self.assertEquals(len(my_entries), len(actual_entries))
 239         for a, b in zip(my_entries, actual_entries):
 240             self.assertEquals(a[0], b[0])
 241             self.assertEquals(a[1], b[1])
 242             if self._has_crc32_checksum:
 243                 self.assertEquals(a[2], b[2])
 244             else:
 245                 self.assertTrue(b[2] is None)
 246
 247
 248 class TestPackIndexWritingv1(unittest.TestCase, BaseTestPackIndexWriting):
 249
 250     def setUp(self):
 251         unittest.TestCase.setUp(self)
 252         self._has_crc32_checksum = False
 253         self._expected_version = 1
 254         self._write_fn = write_pack_index_v1
 255
 256
 257 class TestPackIndexWritingv2(unittest.TestCase, BaseTestPackIndexWriting):
 258
 259     def setUp(self):
 260         unittest.TestCase.setUp(self)
 261         self._has_crc32_checksum = True
 262         self._expected_version = 2
 263         self._write_fn = write_pack_index_v2
 264
 265 TEST_COMP1 = """\x78\x9c\x9d\x8e\xc1\x0a\xc2\x30\x10\x44\xef\xf9\x8a\xbd\xa9\x08\x92\x86\xb4\x26\x20\xe2\xd9\x83\x78\xf2\xbe\x49\x37\xb5\xa5\x69\xca\x36\xf5\xfb\x4d\xfd\x04\x67\x6e\x33\xcc\xf0\x32\x13\x81\xc6\x16\x8d\xa9\xbd\xad\x6c\xe3\x8a\x03\x4a\x73\xd6\xda\xd5\xa6\x51\x2e\x58\x65\x6c\x13\xbc\x94\x4a\xcc\xc8\x34\x65\x78\xa4\x89\x04\xae\xf9\x9d\x18\xee\x34\x46\x62\x78\x11\x4f\x29\xf5\x03\x5c\x86\x5f\x70\x5b\x30\x3a\x3c\x25\xee\xae\x50\xa9\xf2\x60\xa4\xaa\x34\x1c\x65\x91\xf0\x29\xc6\x3e\x67\xfa\x6f\x2d\x9e\x9c\x3e\x7d\x4b\xc0\x34\x8f\xe8\x29\x6e\x48\xa1\xa0\xc4\x88\xf3\xfe\xb0\x5b\x20\x85\xb0\x50\x06\xe4\x6e\xdd\xca\xd3\x17\x26\xfa\x49\x23"""
 266
 267
 268 class ZlibTests(unittest.TestCase):
 269
 270     def test_simple_decompress(self):
 271         self.assertEquals(("tree 4ada885c9196b6b6fa08744b5862bf92896fc002\nparent None\nauthor Jelmer Vernooij <jelmer@samba.org> 1228980214 +0000\ncommitter Jelmer Vernooij <jelmer@samba.org> 1228980214 +0000\n\nProvide replacement for mmap()'s offset argument.", 158),
 272         read_zlib(TEST_COMP1, 0, 229))
 273