In dulwich.tests.test_object_store, use real sha lengths.
[jelmer/dulwich.git] / dulwich / tests / test_object_store.py
1 # test_object_store.py -- tests for object_store.py
2 # Copyright (C) 2008 Jelmer Vernooij <jelmer@samba.org>
3 #
4 # This program is free software; you can redistribute it and/or
5 # modify it under the terms of the GNU General Public License
6 # as published by the Free Software Foundation; version 2
7 # or (at your option) any later version of the License.
8 #
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
12 # GNU General Public License for more details.
13 #
14 # You should have received a copy of the GNU General Public License
15 # along with this program; if not, write to the Free Software
16 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
17 # MA  02110-1301, USA.
18
19 """Tests for the object store interface."""
20
21
22 from io import BytesIO
23 import os
24 import shutil
25 import tempfile
26
27 from dulwich.index import (
28     commit_tree,
29     )
30 from dulwich.errors import (
31     NotTreeError,
32     )
33 from dulwich.objects import (
34     sha_to_hex,
35     object_class,
36     Blob,
37     Tag,
38     Tree,
39     TreeEntry,
40     )
41 from dulwich.object_store import (
42     DiskObjectStore,
43     MemoryObjectStore,
44     ObjectStoreGraphWalker,
45     tree_lookup_path,
46     )
47 from dulwich.pack import (
48     REF_DELTA,
49     write_pack_objects,
50     )
51 from dulwich.tests import (
52     TestCase,
53     )
54 from dulwich.tests.utils import (
55     make_object,
56     build_pack,
57     )
58
59
60 testobject = make_object(Blob, data="yummy data")
61
62
63 class ObjectStoreTests(object):
64
65     def test_determine_wants_all(self):
66         self.assertEqual(["1" * 40],
67             self.store.determine_wants_all({"refs/heads/foo": "1" * 40}))
68
69     def test_determine_wants_all_zero(self):
70         self.assertEqual([],
71             self.store.determine_wants_all({"refs/heads/foo": "0" * 40}))
72
73     def test_iter(self):
74         self.assertEqual([], list(self.store))
75
76     def test_get_nonexistant(self):
77         self.assertRaises(KeyError, lambda: self.store["a" * 40])
78
79     def test_contains_nonexistant(self):
80         self.assertFalse(("a" * 40) in self.store)
81
82     def test_add_objects_empty(self):
83         self.store.add_objects([])
84
85     def test_add_commit(self):
86         # TODO: Argh, no way to construct Git commit objects without 
87         # access to a serialized form.
88         self.store.add_objects([])
89
90     def test_add_object(self):
91         self.store.add_object(testobject)
92         self.assertEqual(set([testobject.id]), set(self.store))
93         self.assertTrue(testobject.id in self.store)
94         r = self.store[testobject.id]
95         self.assertEqual(r, testobject)
96
97     def test_add_objects(self):
98         data = [(testobject, "mypath")]
99         self.store.add_objects(data)
100         self.assertEqual(set([testobject.id]), set(self.store))
101         self.assertTrue(testobject.id in self.store)
102         r = self.store[testobject.id]
103         self.assertEqual(r, testobject)
104
105     def test_tree_changes(self):
106         blob_a1 = make_object(Blob, data='a1')
107         blob_a2 = make_object(Blob, data='a2')
108         blob_b = make_object(Blob, data='b')
109         for blob in [blob_a1, blob_a2, blob_b]:
110             self.store.add_object(blob)
111
112         blobs_1 = [('a', blob_a1.id, 0o100644), ('b', blob_b.id, 0o100644)]
113         tree1_id = commit_tree(self.store, blobs_1)
114         blobs_2 = [('a', blob_a2.id, 0o100644), ('b', blob_b.id, 0o100644)]
115         tree2_id = commit_tree(self.store, blobs_2)
116         change_a = (('a', 'a'), (0o100644, 0o100644), (blob_a1.id, blob_a2.id))
117         self.assertEqual([change_a],
118                           list(self.store.tree_changes(tree1_id, tree2_id)))
119         self.assertEqual(
120           [change_a, (('b', 'b'), (0o100644, 0o100644), (blob_b.id, blob_b.id))],
121           list(self.store.tree_changes(tree1_id, tree2_id,
122                                        want_unchanged=True)))
123
124     def test_iter_tree_contents(self):
125         blob_a = make_object(Blob, data='a')
126         blob_b = make_object(Blob, data='b')
127         blob_c = make_object(Blob, data='c')
128         for blob in [blob_a, blob_b, blob_c]:
129             self.store.add_object(blob)
130
131         blobs = [
132           ('a', blob_a.id, 0o100644),
133           ('ad/b', blob_b.id, 0o100644),
134           ('ad/bd/c', blob_c.id, 0o100755),
135           ('ad/c', blob_c.id, 0o100644),
136           ('c', blob_c.id, 0o100644),
137           ]
138         tree_id = commit_tree(self.store, blobs)
139         self.assertEqual([TreeEntry(p, m, h) for (p, h, m) in blobs],
140                           list(self.store.iter_tree_contents(tree_id)))
141
142     def test_iter_tree_contents_include_trees(self):
143         blob_a = make_object(Blob, data='a')
144         blob_b = make_object(Blob, data='b')
145         blob_c = make_object(Blob, data='c')
146         for blob in [blob_a, blob_b, blob_c]:
147             self.store.add_object(blob)
148
149         blobs = [
150           ('a', blob_a.id, 0o100644),
151           ('ad/b', blob_b.id, 0o100644),
152           ('ad/bd/c', blob_c.id, 0o100755),
153           ]
154         tree_id = commit_tree(self.store, blobs)
155         tree = self.store[tree_id]
156         tree_ad = self.store[tree['ad'][1]]
157         tree_bd = self.store[tree_ad['bd'][1]]
158
159         expected = [
160           TreeEntry('', 0o040000, tree_id),
161           TreeEntry('a', 0o100644, blob_a.id),
162           TreeEntry('ad', 0o040000, tree_ad.id),
163           TreeEntry('ad/b', 0o100644, blob_b.id),
164           TreeEntry('ad/bd', 0o040000, tree_bd.id),
165           TreeEntry('ad/bd/c', 0o100755, blob_c.id),
166           ]
167         actual = self.store.iter_tree_contents(tree_id, include_trees=True)
168         self.assertEqual(expected, list(actual))
169
170     def make_tag(self, name, obj):
171         tag = make_object(Tag, name=name, message='',
172                           tag_time=12345, tag_timezone=0,
173                           tagger='Test Tagger <test@example.com>',
174                           object=(object_class(obj.type_name), obj.id))
175         self.store.add_object(tag)
176         return tag
177
178     def test_peel_sha(self):
179         self.store.add_object(testobject)
180         tag1 = self.make_tag('1', testobject)
181         tag2 = self.make_tag('2', testobject)
182         tag3 = self.make_tag('3', testobject)
183         for obj in [testobject, tag1, tag2, tag3]:
184             self.assertEqual(testobject, self.store.peel_sha(obj.id))
185
186     def test_get_raw(self):
187         self.store.add_object(testobject)
188         self.assertEqual((Blob.type_num, 'yummy data'),
189                          self.store.get_raw(testobject.id))
190
191     def test_close(self):
192         # For now, just check that close doesn't barf.
193         self.store.add_object(testobject)
194         self.store.close()
195
196
197 class MemoryObjectStoreTests(ObjectStoreTests, TestCase):
198
199     def setUp(self):
200         TestCase.setUp(self)
201         self.store = MemoryObjectStore()
202
203     def test_add_pack(self):
204         o = MemoryObjectStore()
205         f, commit, abort = o.add_pack()
206         try:
207             b = make_object(Blob, data="more yummy data")
208             write_pack_objects(f, [(b, None)])
209         except:
210             abort()
211             raise
212         else:
213             commit()
214
215     def test_add_thin_pack(self):
216         o = MemoryObjectStore()
217         blob = make_object(Blob, data='yummy data')
218         o.add_object(blob)
219
220         f = BytesIO()
221         entries = build_pack(f, [
222           (REF_DELTA, (blob.id, 'more yummy data')),
223           ], store=o)
224         o.add_thin_pack(f.read, None)
225         packed_blob_sha = sha_to_hex(entries[0][3])
226         self.assertEqual((Blob.type_num, 'more yummy data'),
227                          o.get_raw(packed_blob_sha))
228
229
230     def test_add_thin_pack_empty(self):
231         o = MemoryObjectStore()
232
233         f = BytesIO()
234         entries = build_pack(f, [], store=o)
235         self.assertEquals([], entries)
236         o.add_thin_pack(f.read, None)
237
238
239 class PackBasedObjectStoreTests(ObjectStoreTests):
240
241     def tearDown(self):
242         for pack in self.store.packs:
243             pack.close()
244
245     def test_empty_packs(self):
246         self.assertEqual([], self.store.packs)
247
248     def test_pack_loose_objects(self):
249         b1 = make_object(Blob, data="yummy data")
250         self.store.add_object(b1)
251         b2 = make_object(Blob, data="more yummy data")
252         self.store.add_object(b2)
253         self.assertEqual([], self.store.packs)
254         self.assertEqual(2, self.store.pack_loose_objects())
255         self.assertNotEqual([], self.store.packs)
256         self.assertEqual(0, self.store.pack_loose_objects())
257
258
259 class DiskObjectStoreTests(PackBasedObjectStoreTests, TestCase):
260
261     def setUp(self):
262         TestCase.setUp(self)
263         self.store_dir = tempfile.mkdtemp()
264         self.addCleanup(shutil.rmtree, self.store_dir)
265         self.store = DiskObjectStore.init(self.store_dir)
266
267     def tearDown(self):
268         TestCase.tearDown(self)
269         PackBasedObjectStoreTests.tearDown(self)
270
271     def test_alternates(self):
272         alternate_dir = tempfile.mkdtemp()
273         self.addCleanup(shutil.rmtree, alternate_dir)
274         alternate_store = DiskObjectStore(alternate_dir)
275         b2 = make_object(Blob, data="yummy data")
276         alternate_store.add_object(b2)
277         store = DiskObjectStore(self.store_dir)
278         self.assertRaises(KeyError, store.__getitem__, b2.id)
279         store.add_alternate_path(alternate_dir)
280         self.assertIn(b2.id, store)
281         self.assertEqual(b2, store[b2.id])
282
283     def test_add_alternate_path(self):
284         store = DiskObjectStore(self.store_dir)
285         self.assertEqual([], store._read_alternate_paths())
286         store.add_alternate_path("/foo/path")
287         self.assertEqual(["/foo/path"], store._read_alternate_paths())
288         store.add_alternate_path("/bar/path")
289         self.assertEqual(
290             ["/foo/path", "/bar/path"],
291             store._read_alternate_paths())
292
293     def test_rel_alternative_path(self):
294         alternate_dir = tempfile.mkdtemp()
295         self.addCleanup(shutil.rmtree, alternate_dir)
296         alternate_store = DiskObjectStore(alternate_dir)
297         b2 = make_object(Blob, data="yummy data")
298         alternate_store.add_object(b2)
299         store = DiskObjectStore(self.store_dir)
300         self.assertRaises(KeyError, store.__getitem__, b2.id)
301         store.add_alternate_path(os.path.relpath(alternate_dir, self.store_dir))
302         self.assertEqual(list(alternate_store), list(store.alternates[0]))
303         self.assertIn(b2.id, store)
304         self.assertEqual(b2, store[b2.id])
305
306     def test_pack_dir(self):
307         o = DiskObjectStore(self.store_dir)
308         self.assertEqual(os.path.join(self.store_dir, "pack"), o.pack_dir)
309
310     def test_add_pack(self):
311         o = DiskObjectStore(self.store_dir)
312         f, commit, abort = o.add_pack()
313         try:
314             b = make_object(Blob, data="more yummy data")
315             write_pack_objects(f, [(b, None)])
316         except:
317             abort()
318             raise
319         else:
320             commit()
321
322     def test_add_thin_pack(self):
323         o = DiskObjectStore(self.store_dir)
324         try:
325             blob = make_object(Blob, data='yummy data')
326             o.add_object(blob)
327
328             f = BytesIO()
329             entries = build_pack(f, [
330               (REF_DELTA, (blob.id, 'more yummy data')),
331               ], store=o)
332
333             with o.add_thin_pack(f.read, None) as pack:
334                 packed_blob_sha = sha_to_hex(entries[0][3])
335                 pack.check_length_and_checksum()
336                 self.assertEqual(sorted([blob.id, packed_blob_sha]), list(pack))
337                 self.assertTrue(o.contains_packed(packed_blob_sha))
338                 self.assertTrue(o.contains_packed(blob.id))
339                 self.assertEqual((Blob.type_num, 'more yummy data'),
340                                  o.get_raw(packed_blob_sha))
341         finally:
342             o.close()
343
344     def test_add_thin_pack_empty(self):
345         o = DiskObjectStore(self.store_dir)
346
347         f = BytesIO()
348         entries = build_pack(f, [], store=o)
349         self.assertEquals([], entries)
350         o.add_thin_pack(f.read, None)
351
352
353 class TreeLookupPathTests(TestCase):
354
355     def setUp(self):
356         TestCase.setUp(self)
357         self.store = MemoryObjectStore()
358         blob_a = make_object(Blob, data='a')
359         blob_b = make_object(Blob, data='b')
360         blob_c = make_object(Blob, data='c')
361         for blob in [blob_a, blob_b, blob_c]:
362             self.store.add_object(blob)
363
364         blobs = [
365           ('a', blob_a.id, 0o100644),
366           ('ad/b', blob_b.id, 0o100644),
367           ('ad/bd/c', blob_c.id, 0o100755),
368           ('ad/c', blob_c.id, 0o100644),
369           ('c', blob_c.id, 0o100644),
370           ]
371         self.tree_id = commit_tree(self.store, blobs)
372
373     def get_object(self, sha):
374         return self.store[sha]
375
376     def test_lookup_blob(self):
377         o_id = tree_lookup_path(self.get_object, self.tree_id, 'a')[1]
378         self.assertTrue(isinstance(self.store[o_id], Blob))
379
380     def test_lookup_tree(self):
381         o_id = tree_lookup_path(self.get_object, self.tree_id, 'ad')[1]
382         self.assertTrue(isinstance(self.store[o_id], Tree))
383         o_id = tree_lookup_path(self.get_object, self.tree_id, 'ad/bd')[1]
384         self.assertTrue(isinstance(self.store[o_id], Tree))
385         o_id = tree_lookup_path(self.get_object, self.tree_id, 'ad/bd/')[1]
386         self.assertTrue(isinstance(self.store[o_id], Tree))
387
388     def test_lookup_nonexistent(self):
389         self.assertRaises(KeyError, tree_lookup_path, self.get_object, self.tree_id, 'j')
390
391     def test_lookup_not_tree(self):
392         self.assertRaises(NotTreeError, tree_lookup_path, self.get_object, self.tree_id, 'ad/b/j')
393
394 # TODO: MissingObjectFinderTests
395
396 class ObjectStoreGraphWalkerTests(TestCase):
397
398     def get_walker(self, heads, parent_map):
399         new_parent_map = dict([
400             (k * 40, [(p * 40) for p in ps]) for (k, ps) in parent_map.items()])
401         return ObjectStoreGraphWalker([x * 40 for x in heads],
402             new_parent_map.__getitem__)
403
404     def test_empty(self):
405         gw = self.get_walker([], {})
406         self.assertIs(None, next(gw))
407         gw.ack("a" * 40)
408         self.assertIs(None, next(gw))
409
410     def test_descends(self):
411         gw = self.get_walker(["a"], {"a": ["b"], "b": []})
412         self.assertEqual("a" * 40, next(gw))
413         self.assertEqual("b" * 40, next(gw))
414
415     def test_present(self):
416         gw = self.get_walker(["a"], {"a": ["b"], "b": []})
417         gw.ack("a" * 40)
418         self.assertIs(None, next(gw))
419
420     def test_parent_present(self):
421         gw = self.get_walker(["a"], {"a": ["b"], "b": []})
422         self.assertEqual("a" * 40, next(gw))
423         gw.ack("a" * 40)
424         self.assertIs(None, next(gw))
425
426     def test_child_ack_later(self):
427         gw = self.get_walker(["a"], {"a": ["b"], "b": ["c"], "c": []})
428         self.assertEqual("a" * 40, next(gw))
429         self.assertEqual("b" * 40, next(gw))
430         gw.ack("a" * 40)
431         self.assertIs(None, next(gw))
432
433     def test_only_once(self):
434         # a  b
435         # |  |
436         # c  d
437         # \ /
438         #  e
439         gw = self.get_walker(["a", "b"], {
440                 "a": ["c"],
441                 "b": ["d"],
442                 "c": ["e"],
443                 "d": ["e"],
444                 "e": [],
445                 })
446         walk = []
447         acked = False
448         walk.append(next(gw))
449         walk.append(next(gw))
450         # A branch (a, c) or (b, d) may be done after 2 steps or 3 depending on
451         # the order walked: 3-step walks include (a, b, c) and (b, a, d), etc.
452         if walk == ["a" * 40, "c" * 40] or walk == ["b" * 40, "d" * 40]:
453           gw.ack(walk[0])
454           acked = True
455
456         walk.append(next(gw))
457         if not acked and walk[2] == "c" * 40:
458           gw.ack("a" * 40)
459         elif not acked and walk[2] == "d" * 40:
460           gw.ack("b" * 40)
461         walk.append(next(gw))
462         self.assertIs(None, next(gw))
463
464         self.assertEqual(["a" * 40, "b" * 40, "c" * 40, "d" * 40], sorted(walk))
465         self.assertLess(walk.index("a" * 40), walk.index("c" * 40))
466         self.assertLess(walk.index("b" * 40), walk.index("d" * 40))