265
472
writer = pack.ContainerWriter(write_data)
267
474
access.set_writer(writer, index, (transport, packname))
268
memos = access.add_raw_records([10], '1234567890')
475
memos = access.add_raw_records([('key', 10)], '1234567890')
270
477
self.assertEqual(['1234567890'], list(access.get_raw_records(memos)))
479
def test_missing_index_raises_retry(self):
480
memos = self.make_pack_file()
481
transport = self.get_transport()
482
reload_called, reload_func = self.make_reload_func()
483
# Note that the index key has changed from 'foo' to 'bar'
484
access = _DirectPackAccess({'bar':(transport, 'packname')},
485
reload_func=reload_func)
486
e = self.assertListRaises(errors.RetryWithNewPacks,
487
access.get_raw_records, memos)
488
# Because a key was passed in which does not match our index list, we
489
# assume that the listing was already reloaded
490
self.assertTrue(e.reload_occurred)
491
self.assertIsInstance(e.exc_info, tuple)
492
self.assertIs(e.exc_info[0], KeyError)
493
self.assertIsInstance(e.exc_info[1], KeyError)
495
def test_missing_index_raises_key_error_with_no_reload(self):
496
memos = self.make_pack_file()
497
transport = self.get_transport()
498
# Note that the index key has changed from 'foo' to 'bar'
499
access = _DirectPackAccess({'bar':(transport, 'packname')})
500
e = self.assertListRaises(KeyError, access.get_raw_records, memos)
502
def test_missing_file_raises_retry(self):
503
memos = self.make_pack_file()
504
transport = self.get_transport()
505
reload_called, reload_func = self.make_reload_func()
506
# Note that the 'filename' has been changed to 'different-packname'
507
access = _DirectPackAccess({'foo':(transport, 'different-packname')},
508
reload_func=reload_func)
509
e = self.assertListRaises(errors.RetryWithNewPacks,
510
access.get_raw_records, memos)
511
# The file has gone missing, so we assume we need to reload
512
self.assertFalse(e.reload_occurred)
513
self.assertIsInstance(e.exc_info, tuple)
514
self.assertIs(e.exc_info[0], errors.NoSuchFile)
515
self.assertIsInstance(e.exc_info[1], errors.NoSuchFile)
516
self.assertEqual('different-packname', e.exc_info[1].path)
518
def test_missing_file_raises_no_such_file_with_no_reload(self):
519
memos = self.make_pack_file()
520
transport = self.get_transport()
521
# Note that the 'filename' has been changed to 'different-packname'
522
access = _DirectPackAccess({'foo':(transport, 'different-packname')})
523
e = self.assertListRaises(errors.NoSuchFile,
524
access.get_raw_records, memos)
526
def test_failing_readv_raises_retry(self):
527
memos = self.make_pack_file()
528
transport = self.get_transport()
529
failing_transport = MockReadvFailingTransport(
530
[transport.get_bytes('packname')])
531
reload_called, reload_func = self.make_reload_func()
532
access = _DirectPackAccess({'foo':(failing_transport, 'packname')},
533
reload_func=reload_func)
534
# Asking for a single record will not trigger the Mock failure
535
self.assertEqual(['1234567890'],
536
list(access.get_raw_records(memos[:1])))
537
self.assertEqual(['12345'],
538
list(access.get_raw_records(memos[1:2])))
539
# A multiple offset readv() will fail mid-way through
540
e = self.assertListRaises(errors.RetryWithNewPacks,
541
access.get_raw_records, memos)
542
# The file has gone missing, so we assume we need to reload
543
self.assertFalse(e.reload_occurred)
544
self.assertIsInstance(e.exc_info, tuple)
545
self.assertIs(e.exc_info[0], errors.NoSuchFile)
546
self.assertIsInstance(e.exc_info[1], errors.NoSuchFile)
547
self.assertEqual('packname', e.exc_info[1].path)
549
def test_failing_readv_raises_no_such_file_with_no_reload(self):
550
memos = self.make_pack_file()
551
transport = self.get_transport()
552
failing_transport = MockReadvFailingTransport(
553
[transport.get_bytes('packname')])
554
reload_called, reload_func = self.make_reload_func()
555
access = _DirectPackAccess({'foo':(failing_transport, 'packname')})
556
# Asking for a single record will not trigger the Mock failure
557
self.assertEqual(['1234567890'],
558
list(access.get_raw_records(memos[:1])))
559
self.assertEqual(['12345'],
560
list(access.get_raw_records(memos[1:2])))
561
# A multiple offset readv() will fail mid-way through
562
e = self.assertListRaises(errors.NoSuchFile,
563
access.get_raw_records, memos)
565
def test_reload_or_raise_no_reload(self):
566
access = _DirectPackAccess({}, reload_func=None)
567
retry_exc = self.make_retry_exception()
568
# Without a reload_func, we will just re-raise the original exception
569
self.assertRaises(_TestException, access.reload_or_raise, retry_exc)
571
def test_reload_or_raise_reload_changed(self):
572
reload_called, reload_func = self.make_reload_func(return_val=True)
573
access = _DirectPackAccess({}, reload_func=reload_func)
574
retry_exc = self.make_retry_exception()
575
access.reload_or_raise(retry_exc)
576
self.assertEqual([1], reload_called)
577
retry_exc.reload_occurred=True
578
access.reload_or_raise(retry_exc)
579
self.assertEqual([2], reload_called)
581
def test_reload_or_raise_reload_no_change(self):
582
reload_called, reload_func = self.make_reload_func(return_val=False)
583
access = _DirectPackAccess({}, reload_func=reload_func)
584
retry_exc = self.make_retry_exception()
585
# If reload_occurred is False, then we consider it an error to have
586
# reload_func() return False (no changes).
587
self.assertRaises(_TestException, access.reload_or_raise, retry_exc)
588
self.assertEqual([1], reload_called)
589
retry_exc.reload_occurred=True
590
# If reload_occurred is True, then we assume nothing changed because
591
# it had changed earlier, but didn't change again
592
access.reload_or_raise(retry_exc)
593
self.assertEqual([2], reload_called)
595
def test_annotate_retries(self):
596
vf, reload_counter = self.make_vf_for_retrying()
597
# It is a little bit bogus to annotate the Revision VF, but it works,
598
# as we have ancestry stored there
600
reload_lines = vf.annotate(key)
601
self.assertEqual([1, 1, 0], reload_counter)
602
plain_lines = vf.annotate(key)
603
self.assertEqual([1, 1, 0], reload_counter) # No extra reloading
604
if reload_lines != plain_lines:
605
self.fail('Annotation was not identical with reloading.')
606
# Now delete the packs-in-use, which should trigger another reload, but
607
# this time we just raise an exception because we can't recover
608
for trans, name in vf._access._indices.itervalues():
610
self.assertRaises(errors.NoSuchFile, vf.annotate, key)
611
self.assertEqual([2, 1, 1], reload_counter)
613
def test__get_record_map_retries(self):
614
vf, reload_counter = self.make_vf_for_retrying()
615
keys = [('rev-1',), ('rev-2',), ('rev-3',)]
616
records = vf._get_record_map(keys)
617
self.assertEqual(keys, sorted(records.keys()))
618
self.assertEqual([1, 1, 0], reload_counter)
619
# Now delete the packs-in-use, which should trigger another reload, but
620
# this time we just raise an exception because we can't recover
621
for trans, name in vf._access._indices.itervalues():
623
self.assertRaises(errors.NoSuchFile, vf._get_record_map, keys)
624
self.assertEqual([2, 1, 1], reload_counter)
626
def test_get_record_stream_retries(self):
627
vf, reload_counter = self.make_vf_for_retrying()
628
keys = [('rev-1',), ('rev-2',), ('rev-3',)]
629
record_stream = vf.get_record_stream(keys, 'topological', False)
630
record = record_stream.next()
631
self.assertEqual(('rev-1',), record.key)
632
self.assertEqual([0, 0, 0], reload_counter)
633
record = record_stream.next()
634
self.assertEqual(('rev-2',), record.key)
635
self.assertEqual([1, 1, 0], reload_counter)
636
record = record_stream.next()
637
self.assertEqual(('rev-3',), record.key)
638
self.assertEqual([1, 1, 0], reload_counter)
639
# Now delete all pack files, and see that we raise the right error
640
for trans, name in vf._access._indices.itervalues():
642
self.assertListRaises(errors.NoSuchFile,
643
vf.get_record_stream, keys, 'topological', False)
645
def test_iter_lines_added_or_present_in_keys_retries(self):
646
vf, reload_counter = self.make_vf_for_retrying()
647
keys = [('rev-1',), ('rev-2',), ('rev-3',)]
648
# Unfortunately, iter_lines_added_or_present_in_keys iterates the
649
# result in random order (determined by the iteration order from a
650
# set()), so we don't have any solid way to trigger whether data is
651
# read before or after. However we tried to delete the middle node to
652
# exercise the code well.
653
# What we care about is that all lines are always yielded, but not
656
reload_lines = sorted(vf.iter_lines_added_or_present_in_keys(keys))
657
self.assertEqual([1, 1, 0], reload_counter)
658
# Now do it again, to make sure the result is equivalent
659
plain_lines = sorted(vf.iter_lines_added_or_present_in_keys(keys))
660
self.assertEqual([1, 1, 0], reload_counter) # No extra reloading
661
self.assertEqual(plain_lines, reload_lines)
662
self.assertEqual(21, len(plain_lines))
663
# Now delete all pack files, and see that we raise the right error
664
for trans, name in vf._access._indices.itervalues():
666
self.assertListRaises(errors.NoSuchFile,
667
vf.iter_lines_added_or_present_in_keys, keys)
668
self.assertEqual([2, 1, 1], reload_counter)
670
def test_get_record_stream_yields_disk_sorted_order(self):
671
# if we get 'unordered' pick a semi-optimal order for reading. The
672
# order should be grouped by pack file, and then by position in file
673
repo = self.make_repository('test', format='pack-0.92')
675
self.addCleanup(repo.unlock)
676
repo.start_write_group()
678
vf.add_lines(('f-id', 'rev-5'), [('f-id', 'rev-4')], ['lines\n'])
679
vf.add_lines(('f-id', 'rev-1'), [], ['lines\n'])
680
vf.add_lines(('f-id', 'rev-2'), [('f-id', 'rev-1')], ['lines\n'])
681
repo.commit_write_group()
682
# We inserted them as rev-5, rev-1, rev-2, we should get them back in
684
stream = vf.get_record_stream([('f-id', 'rev-1'), ('f-id', 'rev-5'),
685
('f-id', 'rev-2')], 'unordered', False)
686
keys = [r.key for r in stream]
687
self.assertEqual([('f-id', 'rev-5'), ('f-id', 'rev-1'),
688
('f-id', 'rev-2')], keys)
689
repo.start_write_group()
690
vf.add_lines(('f-id', 'rev-4'), [('f-id', 'rev-3')], ['lines\n'])
691
vf.add_lines(('f-id', 'rev-3'), [('f-id', 'rev-2')], ['lines\n'])
692
vf.add_lines(('f-id', 'rev-6'), [('f-id', 'rev-5')], ['lines\n'])
693
repo.commit_write_group()
694
# Request in random order, to make sure the output order isn't based on
696
request_keys = set(('f-id', 'rev-%d' % i) for i in range(1, 7))
697
stream = vf.get_record_stream(request_keys, 'unordered', False)
698
keys = [r.key for r in stream]
699
# We want to get the keys back in disk order, but it doesn't matter
700
# which pack we read from first. So this can come back in 2 orders
701
alt1 = [('f-id', 'rev-%d' % i) for i in [4, 3, 6, 5, 1, 2]]
702
alt2 = [('f-id', 'rev-%d' % i) for i in [5, 1, 2, 4, 3, 6]]
703
if keys != alt1 and keys != alt2:
704
self.fail('Returned key order did not match either expected order.'
705
' expected %s or %s, not %s'
706
% (alt1, alt2, keys))
273
709
class LowLevelKnitDataTests(TestCase):
379
858
# Change 2 bytes in the middle to \xff
380
859
gz_txt = gz_txt[:10] + '\xff\xff' + gz_txt[12:]
381
860
transport = MockTransport([gz_txt])
382
access = _KnitAccess(transport, 'filename', None, None, False, False)
383
data = _KnitData(access=access)
384
records = [('rev-id-1', (None, 0, len(gz_txt)))]
386
self.assertRaises(errors.KnitCorrupt, data.read_records, records)
388
# read_records_iter_raw will notice if we request the wrong version.
389
self.assertRaises(errors.KnitCorrupt, list,
390
data.read_records_iter_raw(records))
861
access = _KnitKeyAccess(transport, ConstantMapper('filename'))
862
knit = KnitVersionedFiles(None, access)
863
records = [(('rev-id-1',), (('rev-id-1',), 0, len(gz_txt)))]
864
self.assertRaises(errors.KnitCorrupt, list,
865
knit._read_records_iter(records))
866
# read_records_iter_raw will barf on bad gz data
867
self.assertRaises(errors.KnitCorrupt, list,
868
knit._read_records_iter_raw(records))
393
871
class LowLevelKnitIndexTests(TestCase):
395
def get_knit_index(self, *args, **kwargs):
873
def get_knit_index(self, transport, name, mode):
874
mapper = ConstantMapper(name)
396
875
orig = knit._load_data
398
877
knit._load_data = orig
399
878
self.addCleanup(reset)
400
879
from bzrlib._knit_load_data_py import _load_data_py
401
880
knit._load_data = _load_data_py
402
return _KnitIndex(*args, **kwargs)
404
def test_no_such_file(self):
405
transport = MockTransport()
407
self.assertRaises(NoSuchFile, self.get_knit_index,
408
transport, "filename", "r")
409
self.assertRaises(NoSuchFile, self.get_knit_index,
410
transport, "filename", "w", create=False)
881
allow_writes = lambda: 'w' in mode
882
return _KndxIndex(transport, mapper, lambda:None, allow_writes, lambda:True)
412
884
def test_create_file(self):
413
885
transport = MockTransport()
415
index = self.get_knit_index(transport, "filename", "w",
416
file_mode="wb", create=True)
418
("put_bytes_non_atomic",
419
("filename", index.HEADER), {"mode": "wb"}),
420
transport.calls.pop(0))
422
def test_delay_create_file(self):
423
transport = MockTransport()
425
index = self.get_knit_index(transport, "filename", "w",
426
create=True, file_mode="wb", create_parent_dir=True,
427
delay_create=True, dir_mode=0777)
428
self.assertEqual([], transport.calls)
430
index.add_versions([])
431
name, (filename, f), kwargs = transport.calls.pop(0)
432
self.assertEqual("put_file_non_atomic", name)
434
{"dir_mode": 0777, "create_parent_dir": True, "mode": "wb"},
436
self.assertEqual("filename", filename)
437
self.assertEqual(index.HEADER, f.read())
439
index.add_versions([])
440
self.assertEqual(("append_bytes", ("filename", ""), {}),
441
transport.calls.pop(0))
886
index = self.get_knit_index(transport, "filename", "w")
888
call = transport.calls.pop(0)
889
# call[1][1] is a StringIO - we can't test it by simple equality.
890
self.assertEqual('put_file_non_atomic', call[0])
891
self.assertEqual('filename.kndx', call[1][0])
892
# With no history, _KndxIndex writes a new index:
893
self.assertEqual(_KndxIndex.HEADER,
894
call[1][1].getvalue())
895
self.assertEqual({'create_parent_dir': True}, call[2])
443
897
def test_read_utf8_version_id(self):
444
898
unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
445
899
utf8_revision_id = unicode_revision_id.encode('utf-8')
446
900
transport = MockTransport([
448
902
'%s option 0 1 :' % (utf8_revision_id,)
450
904
index = self.get_knit_index(transport, "filename", "r")
451
# _KnitIndex is a private class, and deals in utf8 revision_ids, not
905
# _KndxIndex is a private class, and deals in utf8 revision_ids, not
452
906
# Unicode revision_ids.
453
self.assertTrue(index.has_version(utf8_revision_id))
454
self.assertFalse(index.has_version(unicode_revision_id))
907
self.assertEqual({(utf8_revision_id,):()},
908
index.get_parent_map(index.keys()))
909
self.assertFalse((unicode_revision_id,) in index.keys())
456
911
def test_read_utf8_parents(self):
457
912
unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
458
913
utf8_revision_id = unicode_revision_id.encode('utf-8')
459
914
transport = MockTransport([
461
916
"version option 0 1 .%s :" % (utf8_revision_id,)
463
918
index = self.get_knit_index(transport, "filename", "r")
464
self.assertEqual([utf8_revision_id],
465
index.get_parents_with_ghosts("version"))
919
self.assertEqual({("version",):((utf8_revision_id,),)},
920
index.get_parent_map(index.keys()))
467
922
def test_read_ignore_corrupted_lines(self):
468
923
transport = MockTransport([
471
926
"corrupted options 0 1 .b .c ",
472
927
"version options 0 1 :"
474
929
index = self.get_knit_index(transport, "filename", "r")
475
self.assertEqual(1, index.num_versions())
476
self.assertTrue(index.has_version("version"))
930
self.assertEqual(1, len(index.keys()))
931
self.assertEqual(set([("version",)]), index.keys())
478
933
def test_read_corrupted_header(self):
479
934
transport = MockTransport(['not a bzr knit index header\n'])
480
self.assertRaises(KnitHeaderError,
481
self.get_knit_index, transport, "filename", "r")
935
index = self.get_knit_index(transport, "filename", "r")
936
self.assertRaises(KnitHeaderError, index.keys)
483
938
def test_read_duplicate_entries(self):
484
939
transport = MockTransport([
486
941
"parent options 0 1 :",
487
942
"version options1 0 1 0 :",
488
943
"version options2 1 2 .other :",
489
944
"version options3 3 4 0 .other :"
491
946
index = self.get_knit_index(transport, "filename", "r")
492
self.assertEqual(2, index.num_versions())
947
self.assertEqual(2, len(index.keys()))
493
948
# check that the index used is the first one written. (Specific
494
949
# to KnitIndex style indices.
495
self.assertEqual("1", index._version_list_to_index(["version"]))
496
self.assertEqual((None, 3, 4), index.get_position("version"))
497
self.assertEqual(["options3"], index.get_options("version"))
498
self.assertEqual(["parent", "other"],
499
index.get_parents_with_ghosts("version"))
950
self.assertEqual("1", index._dictionary_compress([("version",)]))
951
self.assertEqual((("version",), 3, 4), index.get_position(("version",)))
952
self.assertEqual(["options3"], index.get_options(("version",)))
953
self.assertEqual({("version",):(("parent",), ("other",))},
954
index.get_parent_map([("version",)]))
501
956
def test_read_compressed_parents(self):
502
957
transport = MockTransport([
504
959
"a option 0 1 :",
505
960
"b option 0 1 0 :",
506
961
"c option 0 1 1 0 :",
508
963
index = self.get_knit_index(transport, "filename", "r")
509
self.assertEqual(["a"], index.get_parents("b"))
510
self.assertEqual(["b", "a"], index.get_parents("c"))
964
self.assertEqual({("b",):(("a",),), ("c",):(("b",), ("a",))},
965
index.get_parent_map([("b",), ("c",)]))
512
967
def test_write_utf8_version_id(self):
513
968
unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
514
969
utf8_revision_id = unicode_revision_id.encode('utf-8')
515
970
transport = MockTransport([
518
973
index = self.get_knit_index(transport, "filename", "r")
519
index.add_version(utf8_revision_id, ["option"], (None, 0, 1), [])
520
self.assertEqual(("append_bytes", ("filename",
521
"\n%s option 0 1 :" % (utf8_revision_id,)),
523
transport.calls.pop(0))
975
((utf8_revision_id,), ["option"], ((utf8_revision_id,), 0, 1), [])])
976
call = transport.calls.pop(0)
977
# call[1][1] is a StringIO - we can't test it by simple equality.
978
self.assertEqual('put_file_non_atomic', call[0])
979
self.assertEqual('filename.kndx', call[1][0])
980
# With no history, _KndxIndex writes a new index:
981
self.assertEqual(_KndxIndex.HEADER +
982
"\n%s option 0 1 :" % (utf8_revision_id,),
983
call[1][1].getvalue())
984
self.assertEqual({'create_parent_dir': True}, call[2])
525
986
def test_write_utf8_parents(self):
526
987
unicode_revision_id = u"version-\N{CYRILLIC CAPITAL LETTER A}"
527
988
utf8_revision_id = unicode_revision_id.encode('utf-8')
528
989
transport = MockTransport([
531
index = self.get_knit_index(transport, "filename", "r")
532
index.add_version("version", ["option"], (None, 0, 1), [utf8_revision_id])
533
self.assertEqual(("append_bytes", ("filename",
534
"\nversion option 0 1 .%s :" % (utf8_revision_id,)),
536
transport.calls.pop(0))
538
def test_get_graph(self):
539
transport = MockTransport()
540
index = self.get_knit_index(transport, "filename", "w", create=True)
541
self.assertEqual([], index.get_graph())
543
index.add_version("a", ["option"], (None, 0, 1), ["b"])
544
self.assertEqual([("a", ["b"])], index.get_graph())
546
index.add_version("c", ["option"], (None, 0, 1), ["d"])
547
self.assertEqual([("a", ["b"]), ("c", ["d"])],
548
sorted(index.get_graph()))
550
def test_get_ancestry(self):
551
transport = MockTransport([
554
"b option 0 1 0 .e :",
555
"c option 0 1 1 0 :",
556
"d option 0 1 2 .f :"
558
index = self.get_knit_index(transport, "filename", "r")
560
self.assertEqual([], index.get_ancestry([]))
561
self.assertEqual(["a"], index.get_ancestry(["a"]))
562
self.assertEqual(["a", "b"], index.get_ancestry(["b"]))
563
self.assertEqual(["a", "b", "c"], index.get_ancestry(["c"]))
564
self.assertEqual(["a", "b", "c", "d"], index.get_ancestry(["d"]))
565
self.assertEqual(["a", "b"], index.get_ancestry(["a", "b"]))
566
self.assertEqual(["a", "b", "c"], index.get_ancestry(["a", "c"]))
568
self.assertRaises(RevisionNotPresent, index.get_ancestry, ["e"])
570
def test_get_ancestry_with_ghosts(self):
571
transport = MockTransport([
574
"b option 0 1 0 .e :",
575
"c option 0 1 0 .f .g :",
576
"d option 0 1 2 .h .j .k :"
578
index = self.get_knit_index(transport, "filename", "r")
580
self.assertEqual([], index.get_ancestry_with_ghosts([]))
581
self.assertEqual(["a"], index.get_ancestry_with_ghosts(["a"]))
582
self.assertEqual(["a", "e", "b"],
583
index.get_ancestry_with_ghosts(["b"]))
584
self.assertEqual(["a", "g", "f", "c"],
585
index.get_ancestry_with_ghosts(["c"]))
586
self.assertEqual(["a", "g", "f", "c", "k", "j", "h", "d"],
587
index.get_ancestry_with_ghosts(["d"]))
588
self.assertEqual(["a", "e", "b"],
589
index.get_ancestry_with_ghosts(["a", "b"]))
590
self.assertEqual(["a", "g", "f", "c"],
591
index.get_ancestry_with_ghosts(["a", "c"]))
992
index = self.get_knit_index(transport, "filename", "r")
994
(("version",), ["option"], (("version",), 0, 1), [(utf8_revision_id,)])])
995
call = transport.calls.pop(0)
996
# call[1][1] is a StringIO - we can't test it by simple equality.
997
self.assertEqual('put_file_non_atomic', call[0])
998
self.assertEqual('filename.kndx', call[1][0])
999
# With no history, _KndxIndex writes a new index:
1000
self.assertEqual(_KndxIndex.HEADER +
1001
"\nversion option 0 1 .%s :" % (utf8_revision_id,),
1002
call[1][1].getvalue())
1003
self.assertEqual({'create_parent_dir': True}, call[2])
1005
def test_keys(self):
1006
transport = MockTransport([
1009
index = self.get_knit_index(transport, "filename", "r")
1011
self.assertEqual(set(), index.keys())
1013
index.add_records([(("a",), ["option"], (("a",), 0, 1), [])])
1014
self.assertEqual(set([("a",)]), index.keys())
1016
index.add_records([(("a",), ["option"], (("a",), 0, 1), [])])
1017
self.assertEqual(set([("a",)]), index.keys())
1019
index.add_records([(("b",), ["option"], (("b",), 0, 1), [])])
1020
self.assertEqual(set([("a",), ("b",)]), index.keys())
1022
def add_a_b(self, index, random_id=None):
1024
if random_id is not None:
1025
kwargs["random_id"] = random_id
1027
(("a",), ["option"], (("a",), 0, 1), [("b",)]),
1028
(("a",), ["opt"], (("a",), 1, 2), [("c",)]),
1029
(("b",), ["option"], (("b",), 2, 3), [("a",)])
1032
def assertIndexIsAB(self, index):
1037
index.get_parent_map(index.keys()))
1038
self.assertEqual((("a",), 1, 2), index.get_position(("a",)))
1039
self.assertEqual((("b",), 2, 3), index.get_position(("b",)))
1040
self.assertEqual(["opt"], index.get_options(("a",)))
1042
def test_add_versions(self):
1043
transport = MockTransport([
1046
index = self.get_knit_index(transport, "filename", "r")
1049
call = transport.calls.pop(0)
1050
# call[1][1] is a StringIO - we can't test it by simple equality.
1051
self.assertEqual('put_file_non_atomic', call[0])
1052
self.assertEqual('filename.kndx', call[1][0])
1053
# With no history, _KndxIndex writes a new index:
592
1054
self.assertEqual(
593
["a", "g", "f", "c", "e", "b", "k", "j", "h", "d"],
594
index.get_ancestry_with_ghosts(["b", "d"]))
596
self.assertRaises(RevisionNotPresent,
597
index.get_ancestry_with_ghosts, ["e"])
599
def test_iter_parents(self):
600
transport = MockTransport()
601
index = self.get_knit_index(transport, "filename", "w", create=True)
603
index.add_version('r0', ['option'], (None, 0, 1), [])
605
index.add_version('r1', ['option'], (None, 0, 1), ['r0'])
607
index.add_version('r2', ['option'], (None, 0, 1), ['r1', 'r0'])
609
# cases: each sample data individually:
610
self.assertEqual(set([('r0', ())]),
611
set(index.iter_parents(['r0'])))
612
self.assertEqual(set([('r1', ('r0', ))]),
613
set(index.iter_parents(['r1'])))
614
self.assertEqual(set([('r2', ('r1', 'r0'))]),
615
set(index.iter_parents(['r2'])))
616
# no nodes returned for a missing node
617
self.assertEqual(set(),
618
set(index.iter_parents(['missing'])))
619
# 1 node returned with missing nodes skipped
620
self.assertEqual(set([('r1', ('r0', ))]),
621
set(index.iter_parents(['ghost1', 'r1', 'ghost'])))
623
self.assertEqual(set([('r0', ()), ('r1', ('r0', ))]),
624
set(index.iter_parents(['r0', 'r1'])))
625
# 2 nodes returned, missing skipped
626
self.assertEqual(set([('r0', ()), ('r1', ('r0', ))]),
627
set(index.iter_parents(['a', 'r0', 'b', 'r1', 'c'])))
629
def test_num_versions(self):
630
transport = MockTransport([
633
index = self.get_knit_index(transport, "filename", "r")
635
self.assertEqual(0, index.num_versions())
636
self.assertEqual(0, len(index))
638
index.add_version("a", ["option"], (None, 0, 1), [])
639
self.assertEqual(1, index.num_versions())
640
self.assertEqual(1, len(index))
642
index.add_version("a", ["option2"], (None, 1, 2), [])
643
self.assertEqual(1, index.num_versions())
644
self.assertEqual(1, len(index))
646
index.add_version("b", ["option"], (None, 0, 1), [])
647
self.assertEqual(2, index.num_versions())
648
self.assertEqual(2, len(index))
650
def test_get_versions(self):
651
transport = MockTransport([
654
index = self.get_knit_index(transport, "filename", "r")
656
self.assertEqual([], index.get_versions())
658
index.add_version("a", ["option"], (None, 0, 1), [])
659
self.assertEqual(["a"], index.get_versions())
661
index.add_version("a", ["option"], (None, 0, 1), [])
662
self.assertEqual(["a"], index.get_versions())
664
index.add_version("b", ["option"], (None, 0, 1), [])
665
self.assertEqual(["a", "b"], index.get_versions())
667
def test_add_version(self):
668
transport = MockTransport([
671
index = self.get_knit_index(transport, "filename", "r")
673
index.add_version("a", ["option"], (None, 0, 1), ["b"])
674
self.assertEqual(("append_bytes",
675
("filename", "\na option 0 1 .b :"),
676
{}), transport.calls.pop(0))
677
self.assertTrue(index.has_version("a"))
678
self.assertEqual(1, index.num_versions())
679
self.assertEqual((None, 0, 1), index.get_position("a"))
680
self.assertEqual(["option"], index.get_options("a"))
681
self.assertEqual(["b"], index.get_parents_with_ghosts("a"))
683
index.add_version("a", ["opt"], (None, 1, 2), ["c"])
684
self.assertEqual(("append_bytes",
685
("filename", "\na opt 1 2 .c :"),
686
{}), transport.calls.pop(0))
687
self.assertTrue(index.has_version("a"))
688
self.assertEqual(1, index.num_versions())
689
self.assertEqual((None, 1, 2), index.get_position("a"))
690
self.assertEqual(["opt"], index.get_options("a"))
691
self.assertEqual(["c"], index.get_parents_with_ghosts("a"))
693
index.add_version("b", ["option"], (None, 2, 3), ["a"])
694
self.assertEqual(("append_bytes",
695
("filename", "\nb option 2 3 0 :"),
696
{}), transport.calls.pop(0))
697
self.assertTrue(index.has_version("b"))
698
self.assertEqual(2, index.num_versions())
699
self.assertEqual((None, 2, 3), index.get_position("b"))
700
self.assertEqual(["option"], index.get_options("b"))
701
self.assertEqual(["a"], index.get_parents_with_ghosts("b"))
703
def test_add_versions(self):
704
transport = MockTransport([
707
index = self.get_knit_index(transport, "filename", "r")
710
("a", ["option"], (None, 0, 1), ["b"]),
711
("a", ["opt"], (None, 1, 2), ["c"]),
712
("b", ["option"], (None, 2, 3), ["a"])
714
self.assertEqual(("append_bytes", ("filename",
715
1056
"\na option 0 1 .b :"
716
1057
"\na opt 1 2 .c :"
718
), {}), transport.calls.pop(0))
719
self.assertTrue(index.has_version("a"))
720
self.assertTrue(index.has_version("b"))
721
self.assertEqual(2, index.num_versions())
722
self.assertEqual((None, 1, 2), index.get_position("a"))
723
self.assertEqual((None, 2, 3), index.get_position("b"))
724
self.assertEqual(["opt"], index.get_options("a"))
725
self.assertEqual(["option"], index.get_options("b"))
726
self.assertEqual(["c"], index.get_parents_with_ghosts("a"))
727
self.assertEqual(["a"], index.get_parents_with_ghosts("b"))
1058
"\nb option 2 3 0 :",
1059
call[1][1].getvalue())
1060
self.assertEqual({'create_parent_dir': True}, call[2])
1061
self.assertIndexIsAB(index)
1063
def test_add_versions_random_id_is_accepted(self):
1064
transport = MockTransport([
1067
index = self.get_knit_index(transport, "filename", "r")
1068
self.add_a_b(index, random_id=True)
729
1070
def test_delay_create_and_add_versions(self):
730
1071
transport = MockTransport()
732
index = self.get_knit_index(transport, "filename", "w",
733
create=True, file_mode="wb", create_parent_dir=True,
734
delay_create=True, dir_mode=0777)
1073
index = self.get_knit_index(transport, "filename", "w")
735
1075
self.assertEqual([], transport.calls)
738
("a", ["option"], (None, 0, 1), ["b"]),
739
("a", ["opt"], (None, 1, 2), ["c"]),
740
("b", ["option"], (None, 2, 3), ["a"])
742
name, (filename, f), kwargs = transport.calls.pop(0)
743
self.assertEqual("put_file_non_atomic", name)
745
{"dir_mode": 0777, "create_parent_dir": True, "mode": "wb"},
747
self.assertEqual("filename", filename)
1078
#[ {"dir_mode": 0777, "create_parent_dir": True, "mode": "wb"},
1080
# Two calls: one during which we load the existing index (and when its
1081
# missing create it), then a second where we write the contents out.
1082
self.assertEqual(2, len(transport.calls))
1083
call = transport.calls.pop(0)
1084
self.assertEqual('put_file_non_atomic', call[0])
1085
self.assertEqual('filename.kndx', call[1][0])
1086
# With no history, _KndxIndex writes a new index:
1087
self.assertEqual(_KndxIndex.HEADER, call[1][1].getvalue())
1088
self.assertEqual({'create_parent_dir': True}, call[2])
1089
call = transport.calls.pop(0)
1090
# call[1][1] is a StringIO - we can't test it by simple equality.
1091
self.assertEqual('put_file_non_atomic', call[0])
1092
self.assertEqual('filename.kndx', call[1][0])
1093
# With no history, _KndxIndex writes a new index:
750
1096
"\na option 0 1 .b :"
751
1097
"\na opt 1 2 .c :"
752
1098
"\nb option 2 3 0 :",
755
def test_has_version(self):
756
transport = MockTransport([
760
index = self.get_knit_index(transport, "filename", "r")
762
self.assertTrue(index.has_version("a"))
763
self.assertFalse(index.has_version("b"))
1099
call[1][1].getvalue())
1100
self.assertEqual({'create_parent_dir': True}, call[2])
1102
def assertTotalBuildSize(self, size, keys, positions):
1103
self.assertEqual(size,
1104
knit._get_total_build_size(None, keys, positions))
1106
def test__get_total_build_size(self):
1108
('a',): (('fulltext', False), (('a',), 0, 100), None),
1109
('b',): (('line-delta', False), (('b',), 100, 21), ('a',)),
1110
('c',): (('line-delta', False), (('c',), 121, 35), ('b',)),
1111
('d',): (('line-delta', False), (('d',), 156, 12), ('b',)),
1113
self.assertTotalBuildSize(100, [('a',)], positions)
1114
self.assertTotalBuildSize(121, [('b',)], positions)
1115
# c needs both a & b
1116
self.assertTotalBuildSize(156, [('c',)], positions)
1117
# we shouldn't count 'b' twice
1118
self.assertTotalBuildSize(156, [('b',), ('c',)], positions)
1119
self.assertTotalBuildSize(133, [('d',)], positions)
1120
self.assertTotalBuildSize(168, [('c',), ('d',)], positions)
765
1122
def test_get_position(self):
766
1123
transport = MockTransport([
768
1125
"a option 0 1 :",
769
1126
"b option 1 2 :"
771
1128
index = self.get_knit_index(transport, "filename", "r")
773
self.assertEqual((None, 0, 1), index.get_position("a"))
774
self.assertEqual((None, 1, 2), index.get_position("b"))
1130
self.assertEqual((("a",), 0, 1), index.get_position(("a",)))
1131
self.assertEqual((("b",), 1, 2), index.get_position(("b",)))
776
1133
def test_get_method(self):
777
1134
transport = MockTransport([
779
1136
"a fulltext,unknown 0 1 :",
780
1137
"b unknown,line-delta 1 2 :",
1268
def test_scan_unvalidated_index_not_implemented(self):
1269
transport = MockTransport()
1270
index = self.get_knit_index(transport, 'filename', 'r')
1272
NotImplementedError, index.scan_unvalidated_index,
1273
'dummy graph_index')
1275
NotImplementedError, index.get_missing_compression_parents)
940
1277
def test_short_line(self):
941
1278
transport = MockTransport([
943
1280
"a option 0 10 :",
944
1281
"b option 10 10 0", # This line isn't terminated, ignored
946
1283
index = self.get_knit_index(transport, "filename", "r")
947
self.assertEqual(['a'], index.get_versions())
1284
self.assertEqual(set([('a',)]), index.keys())
949
1286
def test_skip_incomplete_record(self):
950
1287
# A line with bogus data should just be skipped
951
1288
transport = MockTransport([
953
1290
"a option 0 10 :",
954
1291
"b option 10 10 0", # This line isn't terminated, ignored
955
1292
"c option 20 10 0 :", # Properly terminated, and starts with '\n'
957
1294
index = self.get_knit_index(transport, "filename", "r")
958
self.assertEqual(['a', 'c'], index.get_versions())
1295
self.assertEqual(set([('a',), ('c',)]), index.keys())
960
1297
def test_trailing_characters(self):
961
1298
# A line with bogus data should just be skipped
962
1299
transport = MockTransport([
964
1301
"a option 0 10 :",
965
1302
"b option 10 10 0 :a", # This line has extra trailing characters
966
1303
"c option 20 10 0 :", # Properly terminated, and starts with '\n'
968
1305
index = self.get_knit_index(transport, "filename", "r")
969
self.assertEqual(['a', 'c'], index.get_versions())
1306
self.assertEqual(set([('a',), ('c',)]), index.keys())
972
1309
class LowLevelKnitIndexTests_c(LowLevelKnitIndexTests):
974
1311
_test_needs_features = [CompiledKnitFeature]
976
def get_knit_index(self, *args, **kwargs):
1313
def get_knit_index(self, transport, name, mode):
1314
mapper = ConstantMapper(name)
977
1315
orig = knit._load_data
979
1317
knit._load_data = orig
980
1318
self.addCleanup(reset)
981
from bzrlib._knit_load_data_c import _load_data_c
1319
from bzrlib._knit_load_data_pyx import _load_data_c
982
1320
knit._load_data = _load_data_c
983
return _KnitIndex(*args, **kwargs)
1321
allow_writes = lambda: mode == 'w'
1322
return _KndxIndex(transport, mapper, lambda:None, allow_writes, lambda:True)
1325
class Test_KnitAnnotator(TestCaseWithMemoryTransport):
1327
def make_annotator(self):
1328
factory = knit.make_pack_factory(True, True, 1)
1329
vf = factory(self.get_transport())
1330
return knit._KnitAnnotator(vf)
1332
def test__expand_fulltext(self):
1333
ann = self.make_annotator()
1334
rev_key = ('rev-id',)
1335
ann._num_compression_children[rev_key] = 1
1336
res = ann._expand_record(rev_key, (('parent-id',),), None,
1337
['line1\n', 'line2\n'], ('fulltext', True))
1338
# The content object and text lines should be cached appropriately
1339
self.assertEqual(['line1\n', 'line2'], res)
1340
content_obj = ann._content_objects[rev_key]
1341
self.assertEqual(['line1\n', 'line2\n'], content_obj._lines)
1342
self.assertEqual(res, content_obj.text())
1343
self.assertEqual(res, ann._text_cache[rev_key])
1345
def test__expand_delta_comp_parent_not_available(self):
1346
# Parent isn't available yet, so we return nothing, but queue up this
1347
# node for later processing
1348
ann = self.make_annotator()
1349
rev_key = ('rev-id',)
1350
parent_key = ('parent-id',)
1351
record = ['0,1,1\n', 'new-line\n']
1352
details = ('line-delta', False)
1353
res = ann._expand_record(rev_key, (parent_key,), parent_key,
1355
self.assertEqual(None, res)
1356
self.assertTrue(parent_key in ann._pending_deltas)
1357
pending = ann._pending_deltas[parent_key]
1358
self.assertEqual(1, len(pending))
1359
self.assertEqual((rev_key, (parent_key,), record, details), pending[0])
1361
def test__expand_record_tracks_num_children(self):
1362
ann = self.make_annotator()
1363
rev_key = ('rev-id',)
1364
rev2_key = ('rev2-id',)
1365
parent_key = ('parent-id',)
1366
record = ['0,1,1\n', 'new-line\n']
1367
details = ('line-delta', False)
1368
ann._num_compression_children[parent_key] = 2
1369
ann._expand_record(parent_key, (), None, ['line1\n', 'line2\n'],
1370
('fulltext', False))
1371
res = ann._expand_record(rev_key, (parent_key,), parent_key,
1373
self.assertEqual({parent_key: 1}, ann._num_compression_children)
1374
# Expanding the second child should remove the content object, and the
1375
# num_compression_children entry
1376
res = ann._expand_record(rev2_key, (parent_key,), parent_key,
1378
self.assertFalse(parent_key in ann._content_objects)
1379
self.assertEqual({}, ann._num_compression_children)
1380
# We should not cache the content_objects for rev2 and rev, because
1381
# they do not have compression children of their own.
1382
self.assertEqual({}, ann._content_objects)
1384
def test__expand_delta_records_blocks(self):
1385
ann = self.make_annotator()
1386
rev_key = ('rev-id',)
1387
parent_key = ('parent-id',)
1388
record = ['0,1,1\n', 'new-line\n']
1389
details = ('line-delta', True)
1390
ann._num_compression_children[parent_key] = 2
1391
ann._expand_record(parent_key, (), None,
1392
['line1\n', 'line2\n', 'line3\n'],
1393
('fulltext', False))
1394
ann._expand_record(rev_key, (parent_key,), parent_key, record, details)
1395
self.assertEqual({(rev_key, parent_key): [(1, 1, 1), (3, 3, 0)]},
1396
ann._matching_blocks)
1397
rev2_key = ('rev2-id',)
1398
record = ['0,1,1\n', 'new-line\n']
1399
details = ('line-delta', False)
1400
ann._expand_record(rev2_key, (parent_key,), parent_key, record, details)
1401
self.assertEqual([(1, 1, 2), (3, 3, 0)],
1402
ann._matching_blocks[(rev2_key, parent_key)])
1404
def test__get_parent_ann_uses_matching_blocks(self):
1405
ann = self.make_annotator()
1406
rev_key = ('rev-id',)
1407
parent_key = ('parent-id',)
1408
parent_ann = [(parent_key,)]*3
1409
block_key = (rev_key, parent_key)
1410
ann._annotations_cache[parent_key] = parent_ann
1411
ann._matching_blocks[block_key] = [(0, 1, 1), (3, 3, 0)]
1412
# We should not try to access any parent_lines content, because we know
1413
# we already have the matching blocks
1414
par_ann, blocks = ann._get_parent_annotations_and_matches(rev_key,
1415
['1\n', '2\n', '3\n'], parent_key)
1416
self.assertEqual(parent_ann, par_ann)
1417
self.assertEqual([(0, 1, 1), (3, 3, 0)], blocks)
1418
self.assertEqual({}, ann._matching_blocks)
1420
def test__process_pending(self):
1421
ann = self.make_annotator()
1422
rev_key = ('rev-id',)
1425
record = ['0,1,1\n', 'new-line\n']
1426
details = ('line-delta', False)
1427
p1_record = ['line1\n', 'line2\n']
1428
ann._num_compression_children[p1_key] = 1
1429
res = ann._expand_record(rev_key, (p1_key,p2_key), p1_key,
1431
self.assertEqual(None, res)
1432
# self.assertTrue(p1_key in ann._pending_deltas)
1433
self.assertEqual({}, ann._pending_annotation)
1434
# Now insert p1, and we should be able to expand the delta
1435
res = ann._expand_record(p1_key, (), None, p1_record,
1436
('fulltext', False))
1437
self.assertEqual(p1_record, res)
1438
ann._annotations_cache[p1_key] = [(p1_key,)]*2
1439
res = ann._process_pending(p1_key)
1440
self.assertEqual([], res)
1441
self.assertFalse(p1_key in ann._pending_deltas)
1442
self.assertTrue(p2_key in ann._pending_annotation)
1443
self.assertEqual({p2_key: [(rev_key, (p1_key, p2_key))]},
1444
ann._pending_annotation)
1445
# Now fill in parent 2, and pending annotation should be satisfied
1446
res = ann._expand_record(p2_key, (), None, [], ('fulltext', False))
1447
ann._annotations_cache[p2_key] = []
1448
res = ann._process_pending(p2_key)
1449
self.assertEqual([rev_key], res)
1450
self.assertEqual({}, ann._pending_annotation)
1451
self.assertEqual({}, ann._pending_deltas)
1453
def test_record_delta_removes_basis(self):
1454
ann = self.make_annotator()
1455
ann._expand_record(('parent-id',), (), None,
1456
['line1\n', 'line2\n'], ('fulltext', False))
1457
ann._num_compression_children['parent-id'] = 2
1459
def test_annotate_special_text(self):
1460
ann = self.make_annotator()
1462
rev1_key = ('rev-1',)
1463
rev2_key = ('rev-2',)
1464
rev3_key = ('rev-3',)
1465
spec_key = ('special:',)
1466
vf.add_lines(rev1_key, [], ['initial content\n'])
1467
vf.add_lines(rev2_key, [rev1_key], ['initial content\n',
1470
vf.add_lines(rev3_key, [rev1_key], ['initial content\n',
1473
spec_text = ('initial content\n'
1477
ann.add_special_text(spec_key, [rev2_key, rev3_key], spec_text)
1478
anns, lines = ann.annotate(spec_key)
1479
self.assertEqual([(rev1_key,),
1480
(rev2_key, rev3_key),
1484
self.assertEqualDiff(spec_text, ''.join(lines))
987
1487
class KnitTests(TestCaseWithTransport):
988
1488
"""Class containing knit test helper routines."""
990
def make_test_knit(self, annotate=False, delay_create=False, index=None,
993
factory = KnitPlainFactory()
996
return KnitVersionedFile(name, get_transport('.'), access_mode='w',
997
factory=factory, create=True,
998
delay_create=delay_create, index=index)
1000
def assertRecordContentEqual(self, knit, version_id, candidate_content):
1001
"""Assert that some raw record content matches the raw record content
1002
for a particular version_id in the given knit.
1004
index_memo = knit._index.get_position(version_id)
1005
record = (version_id, index_memo)
1006
[(_, expected_content)] = list(knit._data.read_records_iter_raw([record]))
1007
self.assertEqual(expected_content, candidate_content)
1010
class BasicKnitTests(KnitTests):
1012
def add_stock_one_and_one_a(self, k):
1013
k.add_lines('text-1', [], split_lines(TEXT_1))
1014
k.add_lines('text-1a', ['text-1'], split_lines(TEXT_1A))
1016
def test_knit_constructor(self):
1017
"""Construct empty k"""
1018
self.make_test_knit()
1020
def test_make_explicit_index(self):
1021
"""We can supply an index to use."""
1022
knit = KnitVersionedFile('test', get_transport('.'),
1023
index='strangelove')
1024
self.assertEqual(knit._index, 'strangelove')
1026
def test_knit_add(self):
1027
"""Store one text in knit and retrieve"""
1028
k = self.make_test_knit()
1029
k.add_lines('text-1', [], split_lines(TEXT_1))
1030
self.assertTrue(k.has_version('text-1'))
1031
self.assertEqualDiff(''.join(k.get_lines('text-1')), TEXT_1)
1033
def test_knit_reload(self):
1034
# test that the content in a reloaded knit is correct
1035
k = self.make_test_knit()
1036
k.add_lines('text-1', [], split_lines(TEXT_1))
1038
k2 = KnitVersionedFile('test', get_transport('.'), access_mode='r', factory=KnitPlainFactory(), create=True)
1039
self.assertTrue(k2.has_version('text-1'))
1040
self.assertEqualDiff(''.join(k2.get_lines('text-1')), TEXT_1)
1042
def test_knit_several(self):
1043
"""Store several texts in a knit"""
1044
k = self.make_test_knit()
1045
k.add_lines('text-1', [], split_lines(TEXT_1))
1046
k.add_lines('text-2', [], split_lines(TEXT_2))
1047
self.assertEqualDiff(''.join(k.get_lines('text-1')), TEXT_1)
1048
self.assertEqualDiff(''.join(k.get_lines('text-2')), TEXT_2)
1050
def test_repeated_add(self):
1051
"""Knit traps attempt to replace existing version"""
1052
k = self.make_test_knit()
1053
k.add_lines('text-1', [], split_lines(TEXT_1))
1054
self.assertRaises(RevisionAlreadyPresent,
1056
'text-1', [], split_lines(TEXT_1))
1058
def test_empty(self):
1059
k = self.make_test_knit(True)
1060
k.add_lines('text-1', [], [])
1061
self.assertEquals(k.get_lines('text-1'), [])
1063
def test_incomplete(self):
1064
"""Test if texts without a ending line-end can be inserted and
1066
k = KnitVersionedFile('test', get_transport('.'), delta=False, create=True)
1067
k.add_lines('text-1', [], ['a\n', 'b' ])
1068
k.add_lines('text-2', ['text-1'], ['a\rb\n', 'b\n'])
1069
# reopening ensures maximum room for confusion
1070
k = KnitVersionedFile('test', get_transport('.'), delta=False, create=True)
1071
self.assertEquals(k.get_lines('text-1'), ['a\n', 'b' ])
1072
self.assertEquals(k.get_lines('text-2'), ['a\rb\n', 'b\n'])
1074
def test_delta(self):
1075
"""Expression of knit delta as lines"""
1076
k = self.make_test_knit()
1078
td = list(line_delta(TEXT_1.splitlines(True),
1079
TEXT_1A.splitlines(True)))
1080
self.assertEqualDiff(''.join(td), delta_1_1a)
1081
out = apply_line_delta(TEXT_1.splitlines(True), td)
1082
self.assertEqualDiff(''.join(out), TEXT_1A)
1084
def assertDerivedBlocksEqual(self, source, target, noeol=False):
1085
"""Assert that the derived matching blocks match real output"""
1086
source_lines = source.splitlines(True)
1087
target_lines = target.splitlines(True)
1089
if noeol and not line.endswith('\n'):
1093
source_content = KnitContent([(None, nl(l)) for l in source_lines])
1094
target_content = KnitContent([(None, nl(l)) for l in target_lines])
1095
line_delta = source_content.line_delta(target_content)
1096
delta_blocks = list(KnitContent.get_line_delta_blocks(line_delta,
1097
source_lines, target_lines))
1098
matcher = KnitSequenceMatcher(None, source_lines, target_lines)
1099
matcher_blocks = list(list(matcher.get_matching_blocks()))
1100
self.assertEqual(matcher_blocks, delta_blocks)
1102
def test_get_line_delta_blocks(self):
1103
self.assertDerivedBlocksEqual('a\nb\nc\n', 'q\nc\n')
1104
self.assertDerivedBlocksEqual(TEXT_1, TEXT_1)
1105
self.assertDerivedBlocksEqual(TEXT_1, TEXT_1A)
1106
self.assertDerivedBlocksEqual(TEXT_1, TEXT_1B)
1107
self.assertDerivedBlocksEqual(TEXT_1B, TEXT_1A)
1108
self.assertDerivedBlocksEqual(TEXT_1A, TEXT_1B)
1109
self.assertDerivedBlocksEqual(TEXT_1A, '')
1110
self.assertDerivedBlocksEqual('', TEXT_1A)
1111
self.assertDerivedBlocksEqual('', '')
1112
self.assertDerivedBlocksEqual('a\nb\nc', 'a\nb\nc\nd')
1114
def test_get_line_delta_blocks_noeol(self):
1115
"""Handle historical knit deltas safely
1117
Some existing knit deltas don't consider the last line to differ
1118
when the only difference whether it has a final newline.
1120
New knit deltas appear to always consider the last line to differ
1123
self.assertDerivedBlocksEqual('a\nb\nc', 'a\nb\nc\nd\n', noeol=True)
1124
self.assertDerivedBlocksEqual('a\nb\nc\nd\n', 'a\nb\nc', noeol=True)
1125
self.assertDerivedBlocksEqual('a\nb\nc\n', 'a\nb\nc', noeol=True)
1126
self.assertDerivedBlocksEqual('a\nb\nc', 'a\nb\nc\n', noeol=True)
1128
def test_add_with_parents(self):
1129
"""Store in knit with parents"""
1130
k = self.make_test_knit()
1131
self.add_stock_one_and_one_a(k)
1132
self.assertEquals(k.get_parents('text-1'), [])
1133
self.assertEquals(k.get_parents('text-1a'), ['text-1'])
1135
def test_ancestry(self):
1136
"""Store in knit with parents"""
1137
k = self.make_test_knit()
1138
self.add_stock_one_and_one_a(k)
1139
self.assertEquals(set(k.get_ancestry(['text-1a'])), set(['text-1a', 'text-1']))
1141
def test_add_delta(self):
1142
"""Store in knit with parents"""
1143
k = KnitVersionedFile('test', get_transport('.'), factory=KnitPlainFactory(),
1144
delta=True, create=True)
1145
self.add_stock_one_and_one_a(k)
1147
self.assertEqualDiff(''.join(k.get_lines('text-1a')), TEXT_1A)
1149
def test_add_delta_knit_graph_index(self):
1150
"""Does adding work with a KnitGraphIndex."""
1151
index = InMemoryGraphIndex(2)
1152
knit_index = KnitGraphIndex(index, add_callback=index.add_nodes,
1154
k = KnitVersionedFile('test', get_transport('.'),
1155
delta=True, create=True, index=knit_index)
1156
self.add_stock_one_and_one_a(k)
1158
self.assertEqualDiff(''.join(k.get_lines('text-1a')), TEXT_1A)
1159
# check the index had the right data added.
1160
self.assertEqual(set([
1161
(index, ('text-1', ), ' 0 127', ((), ())),
1162
(index, ('text-1a', ), ' 127 140', ((('text-1', ),), (('text-1', ),))),
1163
]), set(index.iter_all_entries()))
1164
# we should not have a .kndx file
1165
self.assertFalse(get_transport('.').has('test.kndx'))
1167
def test_annotate(self):
1169
k = KnitVersionedFile('knit', get_transport('.'), factory=KnitAnnotateFactory(),
1170
delta=True, create=True)
1171
self.insert_and_test_small_annotate(k)
1173
def insert_and_test_small_annotate(self, k):
1174
"""test annotation with k works correctly."""
1175
k.add_lines('text-1', [], ['a\n', 'b\n'])
1176
k.add_lines('text-2', ['text-1'], ['a\n', 'c\n'])
1178
origins = k.annotate('text-2')
1179
self.assertEquals(origins[0], ('text-1', 'a\n'))
1180
self.assertEquals(origins[1], ('text-2', 'c\n'))
1182
def test_annotate_fulltext(self):
1184
k = KnitVersionedFile('knit', get_transport('.'), factory=KnitAnnotateFactory(),
1185
delta=False, create=True)
1186
self.insert_and_test_small_annotate(k)
1188
def test_annotate_merge_1(self):
1189
k = self.make_test_knit(True)
1190
k.add_lines('text-a1', [], ['a\n', 'b\n'])
1191
k.add_lines('text-a2', [], ['d\n', 'c\n'])
1192
k.add_lines('text-am', ['text-a1', 'text-a2'], ['d\n', 'b\n'])
1193
origins = k.annotate('text-am')
1194
self.assertEquals(origins[0], ('text-a2', 'd\n'))
1195
self.assertEquals(origins[1], ('text-a1', 'b\n'))
1197
def test_annotate_merge_2(self):
1198
k = self.make_test_knit(True)
1199
k.add_lines('text-a1', [], ['a\n', 'b\n', 'c\n'])
1200
k.add_lines('text-a2', [], ['x\n', 'y\n', 'z\n'])
1201
k.add_lines('text-am', ['text-a1', 'text-a2'], ['a\n', 'y\n', 'c\n'])
1202
origins = k.annotate('text-am')
1203
self.assertEquals(origins[0], ('text-a1', 'a\n'))
1204
self.assertEquals(origins[1], ('text-a2', 'y\n'))
1205
self.assertEquals(origins[2], ('text-a1', 'c\n'))
1207
def test_annotate_merge_9(self):
1208
k = self.make_test_knit(True)
1209
k.add_lines('text-a1', [], ['a\n', 'b\n', 'c\n'])
1210
k.add_lines('text-a2', [], ['x\n', 'y\n', 'z\n'])
1211
k.add_lines('text-am', ['text-a1', 'text-a2'], ['k\n', 'y\n', 'c\n'])
1212
origins = k.annotate('text-am')
1213
self.assertEquals(origins[0], ('text-am', 'k\n'))
1214
self.assertEquals(origins[1], ('text-a2', 'y\n'))
1215
self.assertEquals(origins[2], ('text-a1', 'c\n'))
1217
def test_annotate_merge_3(self):
1218
k = self.make_test_knit(True)
1219
k.add_lines('text-a1', [], ['a\n', 'b\n', 'c\n'])
1220
k.add_lines('text-a2', [] ,['x\n', 'y\n', 'z\n'])
1221
k.add_lines('text-am', ['text-a1', 'text-a2'], ['k\n', 'y\n', 'z\n'])
1222
origins = k.annotate('text-am')
1223
self.assertEquals(origins[0], ('text-am', 'k\n'))
1224
self.assertEquals(origins[1], ('text-a2', 'y\n'))
1225
self.assertEquals(origins[2], ('text-a2', 'z\n'))
1227
def test_annotate_merge_4(self):
1228
k = self.make_test_knit(True)
1229
k.add_lines('text-a1', [], ['a\n', 'b\n', 'c\n'])
1230
k.add_lines('text-a2', [], ['x\n', 'y\n', 'z\n'])
1231
k.add_lines('text-a3', ['text-a1'], ['a\n', 'b\n', 'p\n'])
1232
k.add_lines('text-am', ['text-a2', 'text-a3'], ['a\n', 'b\n', 'z\n'])
1233
origins = k.annotate('text-am')
1234
self.assertEquals(origins[0], ('text-a1', 'a\n'))
1235
self.assertEquals(origins[1], ('text-a1', 'b\n'))
1236
self.assertEquals(origins[2], ('text-a2', 'z\n'))
1238
def test_annotate_merge_5(self):
1239
k = self.make_test_knit(True)
1240
k.add_lines('text-a1', [], ['a\n', 'b\n', 'c\n'])
1241
k.add_lines('text-a2', [], ['d\n', 'e\n', 'f\n'])
1242
k.add_lines('text-a3', [], ['x\n', 'y\n', 'z\n'])
1243
k.add_lines('text-am',
1244
['text-a1', 'text-a2', 'text-a3'],
1245
['a\n', 'e\n', 'z\n'])
1246
origins = k.annotate('text-am')
1247
self.assertEquals(origins[0], ('text-a1', 'a\n'))
1248
self.assertEquals(origins[1], ('text-a2', 'e\n'))
1249
self.assertEquals(origins[2], ('text-a3', 'z\n'))
1251
def test_annotate_file_cherry_pick(self):
1252
k = self.make_test_knit(True)
1253
k.add_lines('text-1', [], ['a\n', 'b\n', 'c\n'])
1254
k.add_lines('text-2', ['text-1'], ['d\n', 'e\n', 'f\n'])
1255
k.add_lines('text-3', ['text-2', 'text-1'], ['a\n', 'b\n', 'c\n'])
1256
origins = k.annotate('text-3')
1257
self.assertEquals(origins[0], ('text-1', 'a\n'))
1258
self.assertEquals(origins[1], ('text-1', 'b\n'))
1259
self.assertEquals(origins[2], ('text-1', 'c\n'))
1261
def test_knit_join(self):
1262
"""Store in knit with parents"""
1263
k1 = KnitVersionedFile('test1', get_transport('.'), factory=KnitPlainFactory(), create=True)
1264
k1.add_lines('text-a', [], split_lines(TEXT_1))
1265
k1.add_lines('text-b', ['text-a'], split_lines(TEXT_1))
1267
k1.add_lines('text-c', [], split_lines(TEXT_1))
1268
k1.add_lines('text-d', ['text-c'], split_lines(TEXT_1))
1270
k1.add_lines('text-m', ['text-b', 'text-d'], split_lines(TEXT_1))
1272
k2 = KnitVersionedFile('test2', get_transport('.'), factory=KnitPlainFactory(), create=True)
1273
count = k2.join(k1, version_ids=['text-m'])
1274
self.assertEquals(count, 5)
1275
self.assertTrue(k2.has_version('text-a'))
1276
self.assertTrue(k2.has_version('text-c'))
1278
def test_reannotate(self):
1279
k1 = KnitVersionedFile('knit1', get_transport('.'),
1280
factory=KnitAnnotateFactory(), create=True)
1282
k1.add_lines('text-a', [], ['a\n', 'b\n'])
1284
k1.add_lines('text-b', ['text-a'], ['a\n', 'c\n'])
1286
k2 = KnitVersionedFile('test2', get_transport('.'),
1287
factory=KnitAnnotateFactory(), create=True)
1288
k2.join(k1, version_ids=['text-b'])
1291
k1.add_lines('text-X', ['text-b'], ['a\n', 'b\n'])
1293
k2.add_lines('text-c', ['text-b'], ['z\n', 'c\n'])
1295
k2.add_lines('text-Y', ['text-b'], ['b\n', 'c\n'])
1297
# test-c will have index 3
1298
k1.join(k2, version_ids=['text-c'])
1300
lines = k1.get_lines('text-c')
1301
self.assertEquals(lines, ['z\n', 'c\n'])
1303
origins = k1.annotate('text-c')
1304
self.assertEquals(origins[0], ('text-c', 'z\n'))
1305
self.assertEquals(origins[1], ('text-b', 'c\n'))
1307
def test_get_line_delta_texts(self):
1308
"""Make sure we can call get_texts on text with reused line deltas"""
1309
k1 = KnitVersionedFile('test1', get_transport('.'),
1310
factory=KnitPlainFactory(), create=True)
1315
parents = ['%d' % (t-1)]
1316
k1.add_lines('%d' % t, parents, ['hello\n'] * t)
1317
k1.get_texts(('%d' % t) for t in range(3))
1319
def test_iter_lines_reads_in_order(self):
1320
t = MemoryTransport()
1321
instrumented_t = TransportLogger(t)
1322
k1 = KnitVersionedFile('id', instrumented_t, create=True, delta=True)
1323
self.assertEqual([('id.kndx',)], instrumented_t._calls)
1324
# add texts with no required ordering
1325
k1.add_lines('base', [], ['text\n'])
1326
k1.add_lines('base2', [], ['text2\n'])
1328
instrumented_t._calls = []
1329
# request a last-first iteration
1330
results = list(k1.iter_lines_added_or_present_in_versions(['base2', 'base']))
1331
self.assertEqual([('id.knit', [(0, 87), (87, 89)])], instrumented_t._calls)
1332
self.assertEqual(['text\n', 'text2\n'], results)
1334
def test_create_empty_annotated(self):
1335
k1 = self.make_test_knit(True)
1337
k1.add_lines('text-a', [], ['a\n', 'b\n'])
1338
k2 = k1.create_empty('t', MemoryTransport())
1339
self.assertTrue(isinstance(k2.factory, KnitAnnotateFactory))
1340
self.assertEqual(k1.delta, k2.delta)
1341
# the generic test checks for empty content and file class
1343
def test_knit_format(self):
1344
# this tests that a new knit index file has the expected content
1345
# and that is writes the data we expect as records are added.
1346
knit = self.make_test_knit(True)
1347
# Now knit files are not created until we first add data to them
1348
self.assertFileEqual("# bzr knit index 8\n", 'test.kndx')
1349
knit.add_lines_with_ghosts('revid', ['a_ghost'], ['a\n'])
1350
self.assertFileEqual(
1351
"# bzr knit index 8\n"
1353
"revid fulltext 0 84 .a_ghost :",
1355
knit.add_lines_with_ghosts('revid2', ['revid'], ['a\n'])
1356
self.assertFileEqual(
1357
"# bzr knit index 8\n"
1358
"\nrevid fulltext 0 84 .a_ghost :"
1359
"\nrevid2 line-delta 84 82 0 :",
1361
# we should be able to load this file again
1362
knit = KnitVersionedFile('test', get_transport('.'), access_mode='r')
1363
self.assertEqual(['revid', 'revid2'], knit.versions())
1364
# write a short write to the file and ensure that its ignored
1365
indexfile = file('test.kndx', 'ab')
1366
indexfile.write('\nrevid3 line-delta 166 82 1 2 3 4 5 .phwoar:demo ')
1368
# we should be able to load this file again
1369
knit = KnitVersionedFile('test', get_transport('.'), access_mode='w')
1370
self.assertEqual(['revid', 'revid2'], knit.versions())
1371
# and add a revision with the same id the failed write had
1372
knit.add_lines('revid3', ['revid2'], ['a\n'])
1373
# and when reading it revid3 should now appear.
1374
knit = KnitVersionedFile('test', get_transport('.'), access_mode='r')
1375
self.assertEqual(['revid', 'revid2', 'revid3'], knit.versions())
1376
self.assertEqual(['revid2'], knit.get_parents('revid3'))
1378
def test_delay_create(self):
1379
"""Test that passing delay_create=True creates files late"""
1380
knit = self.make_test_knit(annotate=True, delay_create=True)
1381
self.failIfExists('test.knit')
1382
self.failIfExists('test.kndx')
1383
knit.add_lines_with_ghosts('revid', ['a_ghost'], ['a\n'])
1384
self.failUnlessExists('test.knit')
1385
self.assertFileEqual(
1386
"# bzr knit index 8\n"
1388
"revid fulltext 0 84 .a_ghost :",
1391
def test_create_parent_dir(self):
1392
"""create_parent_dir can create knits in nonexistant dirs"""
1393
# Has no effect if we don't set 'delay_create'
1394
trans = get_transport('.')
1395
self.assertRaises(NoSuchFile, KnitVersionedFile, 'dir/test',
1396
trans, access_mode='w', factory=None,
1397
create=True, create_parent_dir=True)
1398
# Nothing should have changed yet
1399
knit = KnitVersionedFile('dir/test', trans, access_mode='w',
1400
factory=None, create=True,
1401
create_parent_dir=True,
1403
self.failIfExists('dir/test.knit')
1404
self.failIfExists('dir/test.kndx')
1405
self.failIfExists('dir')
1406
knit.add_lines('revid', [], ['a\n'])
1407
self.failUnlessExists('dir')
1408
self.failUnlessExists('dir/test.knit')
1409
self.assertFileEqual(
1410
"# bzr knit index 8\n"
1412
"revid fulltext 0 84 :",
1415
def test_create_mode_700(self):
1416
trans = get_transport('.')
1417
if not trans._can_roundtrip_unix_modebits():
1418
# Can't roundtrip, so no need to run this test
1420
knit = KnitVersionedFile('dir/test', trans, access_mode='w',
1421
factory=None, create=True,
1422
create_parent_dir=True,
1426
knit.add_lines('revid', [], ['a\n'])
1427
self.assertTransportMode(trans, 'dir', 0700)
1428
self.assertTransportMode(trans, 'dir/test.knit', 0600)
1429
self.assertTransportMode(trans, 'dir/test.kndx', 0600)
1431
def test_create_mode_770(self):
1432
trans = get_transport('.')
1433
if not trans._can_roundtrip_unix_modebits():
1434
# Can't roundtrip, so no need to run this test
1436
knit = KnitVersionedFile('dir/test', trans, access_mode='w',
1437
factory=None, create=True,
1438
create_parent_dir=True,
1442
knit.add_lines('revid', [], ['a\n'])
1443
self.assertTransportMode(trans, 'dir', 0770)
1444
self.assertTransportMode(trans, 'dir/test.knit', 0660)
1445
self.assertTransportMode(trans, 'dir/test.kndx', 0660)
1447
def test_create_mode_777(self):
1448
trans = get_transport('.')
1449
if not trans._can_roundtrip_unix_modebits():
1450
# Can't roundtrip, so no need to run this test
1452
knit = KnitVersionedFile('dir/test', trans, access_mode='w',
1453
factory=None, create=True,
1454
create_parent_dir=True,
1458
knit.add_lines('revid', [], ['a\n'])
1459
self.assertTransportMode(trans, 'dir', 0777)
1460
self.assertTransportMode(trans, 'dir/test.knit', 0666)
1461
self.assertTransportMode(trans, 'dir/test.kndx', 0666)
1463
def test_plan_merge(self):
1464
my_knit = self.make_test_knit(annotate=True)
1465
my_knit.add_lines('text1', [], split_lines(TEXT_1))
1466
my_knit.add_lines('text1a', ['text1'], split_lines(TEXT_1A))
1467
my_knit.add_lines('text1b', ['text1'], split_lines(TEXT_1B))
1468
plan = list(my_knit.plan_merge('text1a', 'text1b'))
1469
for plan_line, expected_line in zip(plan, AB_MERGE):
1470
self.assertEqual(plan_line, expected_line)
1472
def test_get_stream_empty(self):
1473
"""Get a data stream for an empty knit file."""
1474
k1 = self.make_test_knit()
1475
format, data_list, reader_callable = k1.get_data_stream([])
1476
self.assertEqual('knit-plain', format)
1477
self.assertEqual([], data_list)
1478
content = reader_callable(None)
1479
self.assertEqual('', content)
1480
self.assertIsInstance(content, str)
1482
def test_get_stream_one_version(self):
1483
"""Get a data stream for a single record out of a knit containing just
1486
k1 = self.make_test_knit()
1488
('text-a', [], TEXT_1),
1490
expected_data_list = [
1491
# version, options, length, parents
1492
('text-a', ['fulltext'], 122, []),
1494
for version_id, parents, lines in test_data:
1495
k1.add_lines(version_id, parents, split_lines(lines))
1497
format, data_list, reader_callable = k1.get_data_stream(['text-a'])
1498
self.assertEqual('knit-plain', format)
1499
self.assertEqual(expected_data_list, data_list)
1500
# There's only one record in the knit, so the content should be the
1501
# entire knit data file's contents.
1502
self.assertEqual(k1.transport.get_bytes(k1._data._access._filename),
1503
reader_callable(None))
1505
def test_get_stream_get_one_version_of_many(self):
1506
"""Get a data stream for just one version out of a knit containing many
1509
k1 = self.make_test_knit()
1510
# Insert the same data as test_knit_join, as they seem to cover a range
1511
# of cases (no parents, one parent, multiple parents).
1513
('text-a', [], TEXT_1),
1514
('text-b', ['text-a'], TEXT_1),
1515
('text-c', [], TEXT_1),
1516
('text-d', ['text-c'], TEXT_1),
1517
('text-m', ['text-b', 'text-d'], TEXT_1),
1519
expected_data_list = [
1520
# version, options, length, parents
1521
('text-m', ['line-delta'], 84, ['text-b', 'text-d']),
1523
for version_id, parents, lines in test_data:
1524
k1.add_lines(version_id, parents, split_lines(lines))
1526
format, data_list, reader_callable = k1.get_data_stream(['text-m'])
1527
self.assertEqual('knit-plain', format)
1528
self.assertEqual(expected_data_list, data_list)
1529
self.assertRecordContentEqual(k1, 'text-m', reader_callable(None))
1531
def test_get_stream_ghost_parent(self):
1532
"""Get a data stream for a version with a ghost parent."""
1533
k1 = self.make_test_knit()
1535
k1.add_lines('text-a', [], split_lines(TEXT_1))
1536
k1.add_lines_with_ghosts('text-b', ['text-a', 'text-ghost'],
1537
split_lines(TEXT_1))
1539
expected_data_list = [
1540
# version, options, length, parents
1541
('text-b', ['line-delta'], 84, ['text-a', 'text-ghost']),
1544
format, data_list, reader_callable = k1.get_data_stream(['text-b'])
1545
self.assertEqual('knit-plain', format)
1546
self.assertEqual(expected_data_list, data_list)
1547
self.assertRecordContentEqual(k1, 'text-b', reader_callable(None))
1549
def test_get_stream_get_multiple_records(self):
1550
"""Get a stream for multiple records of a knit."""
1551
k1 = self.make_test_knit()
1552
# Insert the same data as test_knit_join, as they seem to cover a range
1553
# of cases (no parents, one parent, multiple parents).
1555
('text-a', [], TEXT_1),
1556
('text-b', ['text-a'], TEXT_1),
1557
('text-c', [], TEXT_1),
1558
('text-d', ['text-c'], TEXT_1),
1559
('text-m', ['text-b', 'text-d'], TEXT_1),
1561
expected_data_list = [
1562
# version, options, length, parents
1563
('text-b', ['line-delta'], 84, ['text-a']),
1564
('text-d', ['line-delta'], 84, ['text-c']),
1566
for version_id, parents, lines in test_data:
1567
k1.add_lines(version_id, parents, split_lines(lines))
1569
# Note that even though we request the revision IDs in a particular
1570
# order, the data stream may return them in any order it likes. In this
1571
# case, they'll be in the order they were inserted into the knit.
1572
format, data_list, reader_callable = k1.get_data_stream(
1573
['text-d', 'text-b'])
1574
self.assertEqual('knit-plain', format)
1575
self.assertEqual(expected_data_list, data_list)
1576
self.assertRecordContentEqual(k1, 'text-b', reader_callable(84))
1577
self.assertRecordContentEqual(k1, 'text-d', reader_callable(84))
1578
self.assertEqual('', reader_callable(None),
1579
"There should be no more bytes left to read.")
1581
def test_get_stream_all(self):
1582
"""Get a data stream for all the records in a knit.
1584
This exercises fulltext records, line-delta records, records with
1585
various numbers of parents, and reading multiple records out of the
1586
callable. These cases ought to all be exercised individually by the
1587
other test_get_stream_* tests; this test is basically just paranoia.
1589
k1 = self.make_test_knit()
1590
# Insert the same data as test_knit_join, as they seem to cover a range
1591
# of cases (no parents, one parent, multiple parents).
1593
('text-a', [], TEXT_1),
1594
('text-b', ['text-a'], TEXT_1),
1595
('text-c', [], TEXT_1),
1596
('text-d', ['text-c'], TEXT_1),
1597
('text-m', ['text-b', 'text-d'], TEXT_1),
1599
expected_data_list = [
1600
# version, options, length, parents
1601
('text-a', ['fulltext'], 122, []),
1602
('text-b', ['line-delta'], 84, ['text-a']),
1603
('text-c', ['fulltext'], 121, []),
1604
('text-d', ['line-delta'], 84, ['text-c']),
1605
('text-m', ['line-delta'], 84, ['text-b', 'text-d']),
1607
for version_id, parents, lines in test_data:
1608
k1.add_lines(version_id, parents, split_lines(lines))
1610
format, data_list, reader_callable = k1.get_data_stream(
1611
['text-a', 'text-b', 'text-c', 'text-d', 'text-m'])
1612
self.assertEqual('knit-plain', format)
1613
self.assertEqual(expected_data_list, data_list)
1614
for version_id, options, length, parents in expected_data_list:
1615
bytes = reader_callable(length)
1616
self.assertRecordContentEqual(k1, version_id, bytes)
1618
def assertKnitFilesEqual(self, knit1, knit2):
1619
"""Assert that the contents of the index and data files of two knits are
1623
knit1.transport.get_bytes(knit1._data._access._filename),
1624
knit2.transport.get_bytes(knit2._data._access._filename))
1626
knit1.transport.get_bytes(knit1._index._filename),
1627
knit2.transport.get_bytes(knit2._index._filename))
1629
def test_insert_data_stream_empty(self):
1630
"""Inserting a data stream with no records should not put any data into
1633
k1 = self.make_test_knit()
1634
k1.insert_data_stream(
1635
(k1.get_format_signature(), [], lambda ignored: ''))
1636
self.assertEqual('', k1.transport.get_bytes(k1._data._access._filename),
1637
"The .knit should be completely empty.")
1638
self.assertEqual(k1._index.HEADER,
1639
k1.transport.get_bytes(k1._index._filename),
1640
"The .kndx should have nothing apart from the header.")
1642
def test_insert_data_stream_one_record(self):
1643
"""Inserting a data stream with one record from a knit with one record
1644
results in byte-identical files.
1646
source = self.make_test_knit(name='source')
1647
source.add_lines('text-a', [], split_lines(TEXT_1))
1648
data_stream = source.get_data_stream(['text-a'])
1650
target = self.make_test_knit(name='target')
1651
target.insert_data_stream(data_stream)
1653
self.assertKnitFilesEqual(source, target)
1655
def test_insert_data_stream_records_already_present(self):
1656
"""Insert a data stream where some records are alreday present in the
1657
target, and some not. Only the new records are inserted.
1659
source = self.make_test_knit(name='source')
1660
target = self.make_test_knit(name='target')
1661
# Insert 'text-a' into both source and target
1662
source.add_lines('text-a', [], split_lines(TEXT_1))
1663
target.insert_data_stream(source.get_data_stream(['text-a']))
1664
# Insert 'text-b' into just the source.
1665
source.add_lines('text-b', ['text-a'], split_lines(TEXT_1))
1666
# Get a data stream of both text-a and text-b, and insert it.
1667
data_stream = source.get_data_stream(['text-a', 'text-b'])
1668
target.insert_data_stream(data_stream)
1669
# The source and target will now be identical. This means the text-a
1670
# record was not added a second time.
1671
self.assertKnitFilesEqual(source, target)
1673
def test_insert_data_stream_multiple_records(self):
1674
"""Inserting a data stream of all records from a knit with multiple
1675
records results in byte-identical files.
1677
source = self.make_test_knit(name='source')
1678
source.add_lines('text-a', [], split_lines(TEXT_1))
1679
source.add_lines('text-b', ['text-a'], split_lines(TEXT_1))
1680
source.add_lines('text-c', [], split_lines(TEXT_1))
1681
data_stream = source.get_data_stream(['text-a', 'text-b', 'text-c'])
1683
target = self.make_test_knit(name='target')
1684
target.insert_data_stream(data_stream)
1686
self.assertKnitFilesEqual(source, target)
1688
def test_insert_data_stream_ghost_parent(self):
1689
"""Insert a data stream with a record that has a ghost parent."""
1690
# Make a knit with a record, text-a, that has a ghost parent.
1691
source = self.make_test_knit(name='source')
1692
source.add_lines_with_ghosts('text-a', ['text-ghost'],
1693
split_lines(TEXT_1))
1694
data_stream = source.get_data_stream(['text-a'])
1696
target = self.make_test_knit(name='target')
1697
target.insert_data_stream(data_stream)
1699
self.assertKnitFilesEqual(source, target)
1701
# The target knit object is in a consistent state, i.e. the record we
1702
# just added is immediately visible.
1703
self.assertTrue(target.has_version('text-a'))
1704
self.assertTrue(target.has_ghost('text-ghost'))
1705
self.assertEqual(split_lines(TEXT_1), target.get_lines('text-a'))
1707
def test_insert_data_stream_inconsistent_version_lines(self):
1708
"""Inserting a data stream which has different content for a version_id
1709
than already exists in the knit will raise KnitCorrupt.
1711
source = self.make_test_knit(name='source')
1712
target = self.make_test_knit(name='target')
1713
# Insert a different 'text-a' into both source and target
1714
source.add_lines('text-a', [], split_lines(TEXT_1))
1715
target.add_lines('text-a', [], split_lines(TEXT_2))
1716
# Insert a data stream with conflicting content into the target
1717
data_stream = source.get_data_stream(['text-a'])
1719
errors.KnitCorrupt, target.insert_data_stream, data_stream)
1721
def test_insert_data_stream_inconsistent_version_parents(self):
1722
"""Inserting a data stream which has different parents for a version_id
1723
than already exists in the knit will raise KnitCorrupt.
1725
source = self.make_test_knit(name='source')
1726
target = self.make_test_knit(name='target')
1727
# Insert a different 'text-a' into both source and target. They differ
1728
# only by the parents list, the content is the same.
1729
source.add_lines_with_ghosts('text-a', [], split_lines(TEXT_1))
1730
target.add_lines_with_ghosts('text-a', ['a-ghost'], split_lines(TEXT_1))
1731
# Insert a data stream with conflicting content into the target
1732
data_stream = source.get_data_stream(['text-a'])
1734
errors.KnitCorrupt, target.insert_data_stream, data_stream)
1736
def test_insert_data_stream_incompatible_format(self):
1737
"""A data stream in a different format to the target knit cannot be
1740
It will raise KnitDataStreamIncompatible.
1742
data_stream = ('fake-format-signature', [], lambda _: '')
1743
target = self.make_test_knit(name='target')
1745
errors.KnitDataStreamIncompatible,
1746
target.insert_data_stream, data_stream)
1748
# * test that a stream of "already present version, then new version"
1749
# inserts correctly.
1760
Banana cup cake recipe
1766
- self-raising flour
1770
Banana cup cake recipe
1772
- bananas (do not use plantains!!!)
1779
Banana cup cake recipe
1782
- self-raising flour
1795
AB_MERGE_TEXT="""unchanged|Banana cup cake recipe
1800
new-b|- bananas (do not use plantains!!!)
1801
unchanged|- broken tea cups
1802
new-a|- self-raising flour
1805
AB_MERGE=[tuple(l.split('|')) for l in AB_MERGE_TEXT.splitlines(True)]
1808
def line_delta(from_lines, to_lines):
1809
"""Generate line-based delta from one text to another"""
1810
s = difflib.SequenceMatcher(None, from_lines, to_lines)
1811
for op in s.get_opcodes():
1812
if op[0] == 'equal':
1814
yield '%d,%d,%d\n' % (op[1], op[2], op[4]-op[3])
1815
for i in range(op[3], op[4]):
1819
def apply_line_delta(basis_lines, delta_lines):
1820
"""Apply a line-based perfect diff
1822
basis_lines -- text to apply the patch to
1823
delta_lines -- diff instructions and content
1825
out = basis_lines[:]
1828
while i < len(delta_lines):
1830
a, b, c = map(long, l.split(','))
1832
out[offset+a:offset+b] = delta_lines[i:i+c]
1834
offset = offset + (b - a) + c
1838
class TestWeaveToKnit(KnitTests):
1840
def test_weave_to_knit_matches(self):
1841
# check that the WeaveToKnit is_compatible function
1842
# registers True for a Weave to a Knit.
1844
k = self.make_test_knit()
1845
self.failUnless(WeaveToKnit.is_compatible(w, k))
1846
self.failIf(WeaveToKnit.is_compatible(k, w))
1847
self.failIf(WeaveToKnit.is_compatible(w, w))
1848
self.failIf(WeaveToKnit.is_compatible(k, k))
1851
class TestKnitCaching(KnitTests):
1853
def create_knit(self, cache_add=False):
1854
k = self.make_test_knit(True)
1858
k.add_lines('text-1', [], split_lines(TEXT_1))
1859
k.add_lines('text-2', [], split_lines(TEXT_2))
1862
def test_no_caching(self):
1863
k = self.create_knit()
1864
# Nothing should be cached without setting 'enable_cache'
1865
self.assertEqual({}, k._data._cache)
1867
def test_cache_add_and_clear(self):
1868
k = self.create_knit(True)
1870
self.assertEqual(['text-1', 'text-2'], sorted(k._data._cache.keys()))
1873
self.assertEqual({}, k._data._cache)
1875
def test_cache_data_read_raw(self):
1876
k = self.create_knit()
1878
# Now cache and read
1881
def read_one_raw(version):
1882
pos_map = k._get_components_positions([version])
1883
method, index_memo, next = pos_map[version]
1884
lst = list(k._data.read_records_iter_raw([(version, index_memo)]))
1885
self.assertEqual(1, len(lst))
1888
val = read_one_raw('text-1')
1889
self.assertEqual({'text-1':val[1]}, k._data._cache)
1892
# After clear, new reads are not cached
1893
self.assertEqual({}, k._data._cache)
1895
val2 = read_one_raw('text-1')
1896
self.assertEqual(val, val2)
1897
self.assertEqual({}, k._data._cache)
1899
def test_cache_data_read(self):
1900
k = self.create_knit()
1902
def read_one(version):
1903
pos_map = k._get_components_positions([version])
1904
method, index_memo, next = pos_map[version]
1905
lst = list(k._data.read_records_iter([(version, index_memo)]))
1906
self.assertEqual(1, len(lst))
1909
# Now cache and read
1912
val = read_one('text-2')
1913
self.assertEqual(['text-2'], k._data._cache.keys())
1914
self.assertEqual('text-2', val[0])
1915
content, digest = k._data._parse_record('text-2',
1916
k._data._cache['text-2'])
1917
self.assertEqual(content, val[1])
1918
self.assertEqual(digest, val[2])
1921
self.assertEqual({}, k._data._cache)
1923
val2 = read_one('text-2')
1924
self.assertEqual(val, val2)
1925
self.assertEqual({}, k._data._cache)
1927
def test_cache_read(self):
1928
k = self.create_knit()
1931
text = k.get_text('text-1')
1932
self.assertEqual(TEXT_1, text)
1933
self.assertEqual(['text-1'], k._data._cache.keys())
1936
self.assertEqual({}, k._data._cache)
1938
text = k.get_text('text-1')
1939
self.assertEqual(TEXT_1, text)
1940
self.assertEqual({}, k._data._cache)
1490
def make_test_knit(self, annotate=False, name='test'):
1491
mapper = ConstantMapper(name)
1492
return make_file_factory(annotate, mapper)(self.get_transport())
1495
class TestBadShaError(KnitTests):
1496
"""Tests for handling of sha errors."""
1498
def test_sha_exception_has_text(self):
1499
# having the failed text included in the error allows for recovery.
1500
source = self.make_test_knit()
1501
target = self.make_test_knit(name="target")
1502
if not source._max_delta_chain:
1503
raise TestNotApplicable(
1504
"cannot get delta-caused sha failures without deltas.")
1507
broken = ('broken',)
1508
source.add_lines(basis, (), ['foo\n'])
1509
source.add_lines(broken, (basis,), ['foo\n', 'bar\n'])
1510
# Seed target with a bad basis text
1511
target.add_lines(basis, (), ['gam\n'])
1512
target.insert_record_stream(
1513
source.get_record_stream([broken], 'unordered', False))
1514
err = self.assertRaises(errors.KnitCorrupt,
1515
target.get_record_stream([broken], 'unordered', True
1516
).next().get_bytes_as, 'chunked')
1517
self.assertEqual(['gam\n', 'bar\n'], err.content)
1518
# Test for formatting with live data
1519
self.assertStartsWith(str(err), "Knit ")
1943
1522
class TestKnitIndex(KnitTests):
2062
1650
add_callback = self.catch_add
2064
1652
add_callback = None
2065
return KnitGraphIndex(combined_index, deltas=deltas,
1653
return _KnitGraphIndex(combined_index, lambda:True, deltas=deltas,
2066
1654
add_callback=add_callback)
2068
def test_get_graph(self):
2069
index = self.two_graph_index()
2070
self.assertEqual(set([
2071
('tip', ('parent', )),
2073
('parent', ('tail', 'ghost')),
2075
]), set(index.get_graph()))
2077
def test_get_ancestry(self):
2078
# get_ancestry is defined as eliding ghosts, not erroring.
2079
index = self.two_graph_index()
2080
self.assertEqual([], index.get_ancestry([]))
2081
self.assertEqual(['separate'], index.get_ancestry(['separate']))
2082
self.assertEqual(['tail'], index.get_ancestry(['tail']))
2083
self.assertEqual(['tail', 'parent'], index.get_ancestry(['parent']))
2084
self.assertEqual(['tail', 'parent', 'tip'], index.get_ancestry(['tip']))
2085
self.assertTrue(index.get_ancestry(['tip', 'separate']) in
2086
(['tail', 'parent', 'tip', 'separate'],
2087
['separate', 'tail', 'parent', 'tip'],
2089
# and without topo_sort
2090
self.assertEqual(set(['separate']),
2091
set(index.get_ancestry(['separate'], topo_sorted=False)))
2092
self.assertEqual(set(['tail']),
2093
set(index.get_ancestry(['tail'], topo_sorted=False)))
2094
self.assertEqual(set(['tail', 'parent']),
2095
set(index.get_ancestry(['parent'], topo_sorted=False)))
2096
self.assertEqual(set(['tail', 'parent', 'tip']),
2097
set(index.get_ancestry(['tip'], topo_sorted=False)))
2098
self.assertEqual(set(['separate', 'tail', 'parent', 'tip']),
2099
set(index.get_ancestry(['tip', 'separate'])))
2100
# asking for a ghost makes it go boom.
2101
self.assertRaises(errors.RevisionNotPresent, index.get_ancestry, ['ghost'])
2103
def test_get_ancestry_with_ghosts(self):
2104
index = self.two_graph_index()
2105
self.assertEqual([], index.get_ancestry_with_ghosts([]))
2106
self.assertEqual(['separate'], index.get_ancestry_with_ghosts(['separate']))
2107
self.assertEqual(['tail'], index.get_ancestry_with_ghosts(['tail']))
2108
self.assertTrue(index.get_ancestry_with_ghosts(['parent']) in
2109
(['tail', 'ghost', 'parent'],
2110
['ghost', 'tail', 'parent'],
2112
self.assertTrue(index.get_ancestry_with_ghosts(['tip']) in
2113
(['tail', 'ghost', 'parent', 'tip'],
2114
['ghost', 'tail', 'parent', 'tip'],
2116
self.assertTrue(index.get_ancestry_with_ghosts(['tip', 'separate']) in
2117
(['tail', 'ghost', 'parent', 'tip', 'separate'],
2118
['ghost', 'tail', 'parent', 'tip', 'separate'],
2119
['separate', 'tail', 'ghost', 'parent', 'tip'],
2120
['separate', 'ghost', 'tail', 'parent', 'tip'],
2122
# asking for a ghost makes it go boom.
2123
self.assertRaises(errors.RevisionNotPresent, index.get_ancestry_with_ghosts, ['ghost'])
2125
def test_num_versions(self):
2126
index = self.two_graph_index()
2127
self.assertEqual(4, index.num_versions())
2129
def test_get_versions(self):
2130
index = self.two_graph_index()
2131
self.assertEqual(set(['tail', 'tip', 'parent', 'separate']),
2132
set(index.get_versions()))
2134
def test_has_version(self):
2135
index = self.two_graph_index()
2136
self.assertTrue(index.has_version('tail'))
2137
self.assertFalse(index.has_version('ghost'))
1656
def test_keys(self):
1657
index = self.two_graph_index()
1658
self.assertEqual(set([('tail',), ('tip',), ('parent',), ('separate',)]),
2139
1661
def test_get_position(self):
2140
1662
index = self.two_graph_index()
2141
self.assertEqual((index._graph_index._indices[0], 0, 100), index.get_position('tip'))
2142
self.assertEqual((index._graph_index._indices[1], 100, 78), index.get_position('parent'))
1663
self.assertEqual((index._graph_index._indices[0], 0, 100), index.get_position(('tip',)))
1664
self.assertEqual((index._graph_index._indices[1], 100, 78), index.get_position(('parent',)))
2144
1666
def test_get_method_deltas(self):
2145
1667
index = self.two_graph_index(deltas=True)
2146
self.assertEqual('fulltext', index.get_method('tip'))
2147
self.assertEqual('line-delta', index.get_method('parent'))
1668
self.assertEqual('fulltext', index.get_method(('tip',)))
1669
self.assertEqual('line-delta', index.get_method(('parent',)))
2149
1671
def test_get_method_no_deltas(self):
2150
1672
# check that the parent-history lookup is ignored with deltas=False.
2151
1673
index = self.two_graph_index(deltas=False)
2152
self.assertEqual('fulltext', index.get_method('tip'))
2153
self.assertEqual('fulltext', index.get_method('parent'))
1674
self.assertEqual('fulltext', index.get_method(('tip',)))
1675
self.assertEqual('fulltext', index.get_method(('parent',)))
2155
1677
def test_get_options_deltas(self):
2156
1678
index = self.two_graph_index(deltas=True)
2157
self.assertEqual(['fulltext', 'no-eol'], index.get_options('tip'))
2158
self.assertEqual(['line-delta'], index.get_options('parent'))
1679
self.assertEqual(['fulltext', 'no-eol'], index.get_options(('tip',)))
1680
self.assertEqual(['line-delta'], index.get_options(('parent',)))
2160
1682
def test_get_options_no_deltas(self):
2161
1683
# check that the parent-history lookup is ignored with deltas=False.
2162
1684
index = self.two_graph_index(deltas=False)
2163
self.assertEqual(['fulltext', 'no-eol'], index.get_options('tip'))
2164
self.assertEqual(['fulltext'], index.get_options('parent'))
2166
def test_get_parents(self):
2167
# get_parents ignores ghosts
2168
index = self.two_graph_index()
2169
self.assertEqual(('tail', ), index.get_parents('parent'))
2170
# and errors on ghosts.
2171
self.assertRaises(errors.RevisionNotPresent,
2172
index.get_parents, 'ghost')
2174
def test_get_parents_with_ghosts(self):
2175
index = self.two_graph_index()
2176
self.assertEqual(('tail', 'ghost'), index.get_parents_with_ghosts('parent'))
2177
# and errors on ghosts.
2178
self.assertRaises(errors.RevisionNotPresent,
2179
index.get_parents_with_ghosts, 'ghost')
2181
def test_check_versions_present(self):
2182
# ghosts should not be considered present
2183
index = self.two_graph_index()
2184
self.assertRaises(RevisionNotPresent, index.check_versions_present,
2186
self.assertRaises(RevisionNotPresent, index.check_versions_present,
2188
index.check_versions_present(['tail', 'separate'])
1685
self.assertEqual(['fulltext', 'no-eol'], index.get_options(('tip',)))
1686
self.assertEqual(['fulltext'], index.get_options(('parent',)))
1688
def test_get_parent_map(self):
1689
index = self.two_graph_index()
1690
self.assertEqual({('parent',):(('tail',), ('ghost',))},
1691
index.get_parent_map([('parent',), ('ghost',)]))
2190
1693
def catch_add(self, entries):
2191
1694
self.caught_entries.append(entries)
2193
1696
def test_add_no_callback_errors(self):
2194
1697
index = self.two_graph_index()
2195
self.assertRaises(errors.ReadOnlyError, index.add_version,
2196
'new', 'fulltext,no-eol', (None, 50, 60), ['separate'])
1698
self.assertRaises(errors.ReadOnlyError, index.add_records,
1699
[(('new',), 'fulltext,no-eol', (None, 50, 60), ['separate'])])
2198
1701
def test_add_version_smoke(self):
2199
1702
index = self.two_graph_index(catch_adds=True)
2200
index.add_version('new', 'fulltext,no-eol', (None, 50, 60), ['separate'])
1703
index.add_records([(('new',), 'fulltext,no-eol', (None, 50, 60),
2201
1705
self.assertEqual([[(('new', ), 'N50 60', ((('separate',),),))]],
2202
1706
self.caught_entries)
2204
1708
def test_add_version_delta_not_delta_index(self):
2205
1709
index = self.two_graph_index(catch_adds=True)
2206
self.assertRaises(errors.KnitCorrupt, index.add_version,
2207
'new', 'no-eol,line-delta', (None, 0, 100), ['parent'])
1710
self.assertRaises(errors.KnitCorrupt, index.add_records,
1711
[(('new',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
2208
1712
self.assertEqual([], self.caught_entries)
2210
1714
def test_add_version_same_dup(self):
2211
1715
index = self.two_graph_index(catch_adds=True)
2212
1716
# options can be spelt two different ways
2213
index.add_version('tip', 'fulltext,no-eol', (None, 0, 100), ['parent'])
2214
index.add_version('tip', 'no-eol,fulltext', (None, 0, 100), ['parent'])
2215
# but neither should have added data.
2216
self.assertEqual([[], []], self.caught_entries)
1717
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
1718
index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100), [('parent',)])])
1719
# position/length are ignored (because each pack could have fulltext or
1720
# delta, and be at a different position.
1721
index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100),
1723
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000),
1725
# but neither should have added data:
1726
self.assertEqual([[], [], [], []], self.caught_entries)
2218
1728
def test_add_version_different_dup(self):
2219
1729
index = self.two_graph_index(deltas=True, catch_adds=True)
2220
1730
# change options
2221
self.assertRaises(errors.KnitCorrupt, index.add_version,
2222
'tip', 'no-eol,line-delta', (None, 0, 100), ['parent'])
2223
self.assertRaises(errors.KnitCorrupt, index.add_version,
2224
'tip', 'line-delta,no-eol', (None, 0, 100), ['parent'])
2225
self.assertRaises(errors.KnitCorrupt, index.add_version,
2226
'tip', 'fulltext', (None, 0, 100), ['parent'])
2228
self.assertRaises(errors.KnitCorrupt, index.add_version,
2229
'tip', 'fulltext,no-eol', (None, 50, 100), ['parent'])
2230
self.assertRaises(errors.KnitCorrupt, index.add_version,
2231
'tip', 'fulltext,no-eol', (None, 0, 1000), ['parent'])
1731
self.assertRaises(errors.KnitCorrupt, index.add_records,
1732
[(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
1733
self.assertRaises(errors.KnitCorrupt, index.add_records,
1734
[(('tip',), 'fulltext', (None, 0, 100), [('parent',)])])
2233
self.assertRaises(errors.KnitCorrupt, index.add_version,
2234
'tip', 'fulltext,no-eol', (None, 0, 100), [])
1736
self.assertRaises(errors.KnitCorrupt, index.add_records,
1737
[(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
2235
1738
self.assertEqual([], self.caught_entries)
2237
1740
def test_add_versions_nodeltas(self):
2238
1741
index = self.two_graph_index(catch_adds=True)
2239
index.add_versions([
2240
('new', 'fulltext,no-eol', (None, 50, 60), ['separate']),
2241
('new2', 'fulltext', (None, 0, 6), ['new']),
1743
(('new',), 'fulltext,no-eol', (None, 50, 60), [('separate',)]),
1744
(('new2',), 'fulltext', (None, 0, 6), [('new',)]),
2243
1746
self.assertEqual([(('new', ), 'N50 60', ((('separate',),),)),
2244
1747
(('new2', ), ' 0 6', ((('new',),),))],
2259
1762
def test_add_versions_delta_not_delta_index(self):
2260
1763
index = self.two_graph_index(catch_adds=True)
2261
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2262
[('new', 'no-eol,line-delta', (None, 0, 100), ['parent'])])
1764
self.assertRaises(errors.KnitCorrupt, index.add_records,
1765
[(('new',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
2263
1766
self.assertEqual([], self.caught_entries)
1768
def test_add_versions_random_id_accepted(self):
1769
index = self.two_graph_index(catch_adds=True)
1770
index.add_records([], random_id=True)
2265
1772
def test_add_versions_same_dup(self):
2266
1773
index = self.two_graph_index(catch_adds=True)
2267
1774
# options can be spelt two different ways
2268
index.add_versions([('tip', 'fulltext,no-eol', (None, 0, 100), ['parent'])])
2269
index.add_versions([('tip', 'no-eol,fulltext', (None, 0, 100), ['parent'])])
1775
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100),
1777
index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100),
1779
# position/length are ignored (because each pack could have fulltext or
1780
# delta, and be at a different position.
1781
index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100),
1783
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000),
2270
1785
# but neither should have added data.
2271
self.assertEqual([[], []], self.caught_entries)
1786
self.assertEqual([[], [], [], []], self.caught_entries)
2273
1788
def test_add_versions_different_dup(self):
2274
1789
index = self.two_graph_index(deltas=True, catch_adds=True)
2275
1790
# change options
2276
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2277
[('tip', 'no-eol,line-delta', (None, 0, 100), ['parent'])])
2278
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2279
[('tip', 'line-delta,no-eol', (None, 0, 100), ['parent'])])
2280
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2281
[('tip', 'fulltext', (None, 0, 100), ['parent'])])
2283
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2284
[('tip', 'fulltext,no-eol', (None, 50, 100), ['parent'])])
2285
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2286
[('tip', 'fulltext,no-eol', (None, 0, 1000), ['parent'])])
1791
self.assertRaises(errors.KnitCorrupt, index.add_records,
1792
[(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
1793
self.assertRaises(errors.KnitCorrupt, index.add_records,
1794
[(('tip',), 'fulltext', (None, 0, 100), [('parent',)])])
2288
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2289
[('tip', 'fulltext,no-eol', (None, 0, 100), [])])
1796
self.assertRaises(errors.KnitCorrupt, index.add_records,
1797
[(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
2290
1798
# change options in the second record
2291
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2292
[('tip', 'fulltext,no-eol', (None, 0, 100), ['parent']),
2293
('tip', 'no-eol,line-delta', (None, 0, 100), ['parent'])])
1799
self.assertRaises(errors.KnitCorrupt, index.add_records,
1800
[(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)]),
1801
(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
2294
1802
self.assertEqual([], self.caught_entries)
2296
def test_iter_parents(self):
2297
index1 = self.make_g_index('1', 1, [
2299
(('r0', ), 'N0 100', ([], )),
2301
(('r1', ), '', ([('r0', )], ))])
2302
index2 = self.make_g_index('2', 1, [
2304
(('r2', ), 'N0 100', ([('r1', ), ('r0', )], )),
2306
combined_index = CombinedGraphIndex([index1, index2])
2307
index = KnitGraphIndex(combined_index)
2309
# cases: each sample data individually:
2310
self.assertEqual(set([('r0', ())]),
2311
set(index.iter_parents(['r0'])))
2312
self.assertEqual(set([('r1', ('r0', ))]),
2313
set(index.iter_parents(['r1'])))
2314
self.assertEqual(set([('r2', ('r1', 'r0'))]),
2315
set(index.iter_parents(['r2'])))
2316
# no nodes returned for a missing node
2317
self.assertEqual(set(),
2318
set(index.iter_parents(['missing'])))
2319
# 1 node returned with missing nodes skipped
2320
self.assertEqual(set([('r1', ('r0', ))]),
2321
set(index.iter_parents(['ghost1', 'r1', 'ghost'])))
2323
self.assertEqual(set([('r0', ()), ('r1', ('r0', ))]),
2324
set(index.iter_parents(['r0', 'r1'])))
2325
# 2 nodes returned, missing skipped
2326
self.assertEqual(set([('r0', ()), ('r1', ('r0', ))]),
2327
set(index.iter_parents(['a', 'r0', 'b', 'r1', 'c'])))
1804
def make_g_index_missing_compression_parent(self):
1805
graph_index = self.make_g_index('missing_comp', 2,
1806
[(('tip', ), ' 100 78',
1807
([('missing-parent', ), ('ghost', )], [('missing-parent', )]))])
1810
def make_g_index_missing_parent(self):
1811
graph_index = self.make_g_index('missing_parent', 2,
1812
[(('parent', ), ' 100 78', ([], [])),
1813
(('tip', ), ' 100 78',
1814
([('parent', ), ('missing-parent', )], [('parent', )])),
1818
def make_g_index_no_external_refs(self):
1819
graph_index = self.make_g_index('no_external_refs', 2,
1820
[(('rev', ), ' 100 78',
1821
([('parent', ), ('ghost', )], []))])
1824
def test_add_good_unvalidated_index(self):
1825
unvalidated = self.make_g_index_no_external_refs()
1826
combined = CombinedGraphIndex([unvalidated])
1827
index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1828
index.scan_unvalidated_index(unvalidated)
1829
self.assertEqual(frozenset(), index.get_missing_compression_parents())
1831
def test_add_missing_compression_parent_unvalidated_index(self):
1832
unvalidated = self.make_g_index_missing_compression_parent()
1833
combined = CombinedGraphIndex([unvalidated])
1834
index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1835
index.scan_unvalidated_index(unvalidated)
1836
# This also checks that its only the compression parent that is
1837
# examined, otherwise 'ghost' would also be reported as a missing
1840
frozenset([('missing-parent',)]),
1841
index.get_missing_compression_parents())
1843
def test_add_missing_noncompression_parent_unvalidated_index(self):
1844
unvalidated = self.make_g_index_missing_parent()
1845
combined = CombinedGraphIndex([unvalidated])
1846
index = _KnitGraphIndex(combined, lambda: True, deltas=True,
1847
track_external_parent_refs=True)
1848
index.scan_unvalidated_index(unvalidated)
1850
frozenset([('missing-parent',)]), index.get_missing_parents())
1852
def test_track_external_parent_refs(self):
1853
g_index = self.make_g_index('empty', 2, [])
1854
combined = CombinedGraphIndex([g_index])
1855
index = _KnitGraphIndex(combined, lambda: True, deltas=True,
1856
add_callback=self.catch_add, track_external_parent_refs=True)
1857
self.caught_entries = []
1859
(('new-key',), 'fulltext,no-eol', (None, 50, 60),
1860
[('parent-1',), ('parent-2',)])])
1862
frozenset([('parent-1',), ('parent-2',)]),
1863
index.get_missing_parents())
1865
def test_add_unvalidated_index_with_present_external_references(self):
1866
index = self.two_graph_index(deltas=True)
1867
# Ugly hack to get at one of the underlying GraphIndex objects that
1868
# two_graph_index built.
1869
unvalidated = index._graph_index._indices[1]
1870
# 'parent' is an external ref of _indices[1] (unvalidated), but is
1871
# present in _indices[0].
1872
index.scan_unvalidated_index(unvalidated)
1873
self.assertEqual(frozenset(), index.get_missing_compression_parents())
1875
def make_new_missing_parent_g_index(self, name):
1876
missing_parent = name + '-missing-parent'
1877
graph_index = self.make_g_index(name, 2,
1878
[((name + 'tip', ), ' 100 78',
1879
([(missing_parent, ), ('ghost', )], [(missing_parent, )]))])
1882
def test_add_mulitiple_unvalidated_indices_with_missing_parents(self):
1883
g_index_1 = self.make_new_missing_parent_g_index('one')
1884
g_index_2 = self.make_new_missing_parent_g_index('two')
1885
combined = CombinedGraphIndex([g_index_1, g_index_2])
1886
index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1887
index.scan_unvalidated_index(g_index_1)
1888
index.scan_unvalidated_index(g_index_2)
1890
frozenset([('one-missing-parent',), ('two-missing-parent',)]),
1891
index.get_missing_compression_parents())
1893
def test_add_mulitiple_unvalidated_indices_with_mutual_dependencies(self):
1894
graph_index_a = self.make_g_index('one', 2,
1895
[(('parent-one', ), ' 100 78', ([('non-compression-parent',)], [])),
1896
(('child-of-two', ), ' 100 78',
1897
([('parent-two',)], [('parent-two',)]))])
1898
graph_index_b = self.make_g_index('two', 2,
1899
[(('parent-two', ), ' 100 78', ([('non-compression-parent',)], [])),
1900
(('child-of-one', ), ' 100 78',
1901
([('parent-one',)], [('parent-one',)]))])
1902
combined = CombinedGraphIndex([graph_index_a, graph_index_b])
1903
index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1904
index.scan_unvalidated_index(graph_index_a)
1905
index.scan_unvalidated_index(graph_index_b)
1907
frozenset([]), index.get_missing_compression_parents())
2330
1910
class TestNoParentsGraphIndexKnit(KnitTests):
2331
"""Tests for knits using KnitGraphIndex with no parents."""
1911
"""Tests for knits using _KnitGraphIndex with no parents."""
2333
1913
def make_g_index(self, name, ref_lists=0, nodes=[]):
2334
1914
builder = GraphIndexBuilder(ref_lists)
2364
1952
add_callback = self.catch_add
2366
1954
add_callback = None
2367
return KnitGraphIndex(combined_index, parents=False,
1955
return _KnitGraphIndex(combined_index, lambda:True, parents=False,
2368
1956
add_callback=add_callback)
2370
def test_get_graph(self):
2371
index = self.two_graph_index()
2372
self.assertEqual(set([
2377
]), set(index.get_graph()))
2379
def test_get_ancestry(self):
2380
# with no parents, ancestry is always just the key.
2381
index = self.two_graph_index()
2382
self.assertEqual([], index.get_ancestry([]))
2383
self.assertEqual(['separate'], index.get_ancestry(['separate']))
2384
self.assertEqual(['tail'], index.get_ancestry(['tail']))
2385
self.assertEqual(['parent'], index.get_ancestry(['parent']))
2386
self.assertEqual(['tip'], index.get_ancestry(['tip']))
2387
self.assertTrue(index.get_ancestry(['tip', 'separate']) in
2388
(['tip', 'separate'],
2389
['separate', 'tip'],
2391
# asking for a ghost makes it go boom.
2392
self.assertRaises(errors.RevisionNotPresent, index.get_ancestry, ['ghost'])
2394
def test_get_ancestry_with_ghosts(self):
2395
index = self.two_graph_index()
2396
self.assertEqual([], index.get_ancestry_with_ghosts([]))
2397
self.assertEqual(['separate'], index.get_ancestry_with_ghosts(['separate']))
2398
self.assertEqual(['tail'], index.get_ancestry_with_ghosts(['tail']))
2399
self.assertEqual(['parent'], index.get_ancestry_with_ghosts(['parent']))
2400
self.assertEqual(['tip'], index.get_ancestry_with_ghosts(['tip']))
2401
self.assertTrue(index.get_ancestry_with_ghosts(['tip', 'separate']) in
2402
(['tip', 'separate'],
2403
['separate', 'tip'],
2405
# asking for a ghost makes it go boom.
2406
self.assertRaises(errors.RevisionNotPresent, index.get_ancestry_with_ghosts, ['ghost'])
2408
def test_num_versions(self):
2409
index = self.two_graph_index()
2410
self.assertEqual(4, index.num_versions())
2412
def test_get_versions(self):
2413
index = self.two_graph_index()
2414
self.assertEqual(set(['tail', 'tip', 'parent', 'separate']),
2415
set(index.get_versions()))
2417
def test_has_version(self):
2418
index = self.two_graph_index()
2419
self.assertTrue(index.has_version('tail'))
2420
self.assertFalse(index.has_version('ghost'))
1958
def test_keys(self):
1959
index = self.two_graph_index()
1960
self.assertEqual(set([('tail',), ('tip',), ('parent',), ('separate',)]),
2422
1963
def test_get_position(self):
2423
1964
index = self.two_graph_index()
2424
self.assertEqual((index._graph_index._indices[0], 0, 100), index.get_position('tip'))
2425
self.assertEqual((index._graph_index._indices[1], 100, 78), index.get_position('parent'))
1965
self.assertEqual((index._graph_index._indices[0], 0, 100),
1966
index.get_position(('tip',)))
1967
self.assertEqual((index._graph_index._indices[1], 100, 78),
1968
index.get_position(('parent',)))
2427
1970
def test_get_method(self):
2428
1971
index = self.two_graph_index()
2429
self.assertEqual('fulltext', index.get_method('tip'))
2430
self.assertEqual(['fulltext'], index.get_options('parent'))
1972
self.assertEqual('fulltext', index.get_method(('tip',)))
1973
self.assertEqual(['fulltext'], index.get_options(('parent',)))
2432
1975
def test_get_options(self):
2433
1976
index = self.two_graph_index()
2434
self.assertEqual(['fulltext', 'no-eol'], index.get_options('tip'))
2435
self.assertEqual(['fulltext'], index.get_options('parent'))
2437
def test_get_parents(self):
2438
index = self.two_graph_index()
2439
self.assertEqual((), index.get_parents('parent'))
2440
# and errors on ghosts.
2441
self.assertRaises(errors.RevisionNotPresent,
2442
index.get_parents, 'ghost')
2444
def test_get_parents_with_ghosts(self):
2445
index = self.two_graph_index()
2446
self.assertEqual((), index.get_parents_with_ghosts('parent'))
2447
# and errors on ghosts.
2448
self.assertRaises(errors.RevisionNotPresent,
2449
index.get_parents_with_ghosts, 'ghost')
2451
def test_check_versions_present(self):
2452
index = self.two_graph_index()
2453
self.assertRaises(RevisionNotPresent, index.check_versions_present,
2455
self.assertRaises(RevisionNotPresent, index.check_versions_present,
2456
['tail', 'missing'])
2457
index.check_versions_present(['tail', 'separate'])
1977
self.assertEqual(['fulltext', 'no-eol'], index.get_options(('tip',)))
1978
self.assertEqual(['fulltext'], index.get_options(('parent',)))
1980
def test_get_parent_map(self):
1981
index = self.two_graph_index()
1982
self.assertEqual({('parent',):None},
1983
index.get_parent_map([('parent',), ('ghost',)]))
2459
1985
def catch_add(self, entries):
2460
1986
self.caught_entries.append(entries)
2462
1988
def test_add_no_callback_errors(self):
2463
1989
index = self.two_graph_index()
2464
self.assertRaises(errors.ReadOnlyError, index.add_version,
2465
'new', 'fulltext,no-eol', (None, 50, 60), ['separate'])
1990
self.assertRaises(errors.ReadOnlyError, index.add_records,
1991
[(('new',), 'fulltext,no-eol', (None, 50, 60), [('separate',)])])
2467
1993
def test_add_version_smoke(self):
2468
1994
index = self.two_graph_index(catch_adds=True)
2469
index.add_version('new', 'fulltext,no-eol', (None, 50, 60), [])
1995
index.add_records([(('new',), 'fulltext,no-eol', (None, 50, 60), [])])
2470
1996
self.assertEqual([[(('new', ), 'N50 60')]],
2471
1997
self.caught_entries)
2473
1999
def test_add_version_delta_not_delta_index(self):
2474
2000
index = self.two_graph_index(catch_adds=True)
2475
self.assertRaises(errors.KnitCorrupt, index.add_version,
2476
'new', 'no-eol,line-delta', (None, 0, 100), [])
2001
self.assertRaises(errors.KnitCorrupt, index.add_records,
2002
[(('new',), 'no-eol,line-delta', (None, 0, 100), [])])
2477
2003
self.assertEqual([], self.caught_entries)
2479
2005
def test_add_version_same_dup(self):
2480
2006
index = self.two_graph_index(catch_adds=True)
2481
2007
# options can be spelt two different ways
2482
index.add_version('tip', 'fulltext,no-eol', (None, 0, 100), [])
2483
index.add_version('tip', 'no-eol,fulltext', (None, 0, 100), [])
2008
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
2009
index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100), [])])
2010
# position/length are ignored (because each pack could have fulltext or
2011
# delta, and be at a different position.
2012
index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100), [])])
2013
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000), [])])
2484
2014
# but neither should have added data.
2485
self.assertEqual([[], []], self.caught_entries)
2015
self.assertEqual([[], [], [], []], self.caught_entries)
2487
2017
def test_add_version_different_dup(self):
2488
2018
index = self.two_graph_index(catch_adds=True)
2489
2019
# change options
2490
self.assertRaises(errors.KnitCorrupt, index.add_version,
2491
'tip', 'no-eol,line-delta', (None, 0, 100), [])
2492
self.assertRaises(errors.KnitCorrupt, index.add_version,
2493
'tip', 'line-delta,no-eol', (None, 0, 100), [])
2494
self.assertRaises(errors.KnitCorrupt, index.add_version,
2495
'tip', 'fulltext', (None, 0, 100), [])
2497
self.assertRaises(errors.KnitCorrupt, index.add_version,
2498
'tip', 'fulltext,no-eol', (None, 50, 100), [])
2499
self.assertRaises(errors.KnitCorrupt, index.add_version,
2500
'tip', 'fulltext,no-eol', (None, 0, 1000), [])
2020
self.assertRaises(errors.KnitCorrupt, index.add_records,
2021
[(('tip',), 'no-eol,line-delta', (None, 0, 100), [])])
2022
self.assertRaises(errors.KnitCorrupt, index.add_records,
2023
[(('tip',), 'line-delta,no-eol', (None, 0, 100), [])])
2024
self.assertRaises(errors.KnitCorrupt, index.add_records,
2025
[(('tip',), 'fulltext', (None, 0, 100), [])])
2502
self.assertRaises(errors.KnitCorrupt, index.add_version,
2503
'tip', 'fulltext,no-eol', (None, 0, 100), ['parent'])
2027
self.assertRaises(errors.KnitCorrupt, index.add_records,
2028
[(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
2504
2029
self.assertEqual([], self.caught_entries)
2506
2031
def test_add_versions(self):
2507
2032
index = self.two_graph_index(catch_adds=True)
2508
index.add_versions([
2509
('new', 'fulltext,no-eol', (None, 50, 60), []),
2510
('new2', 'fulltext', (None, 0, 6), []),
2034
(('new',), 'fulltext,no-eol', (None, 50, 60), []),
2035
(('new2',), 'fulltext', (None, 0, 6), []),
2512
2037
self.assertEqual([(('new', ), 'N50 60'), (('new2', ), ' 0 6')],
2513
2038
sorted(self.caught_entries[0]))
2516
2041
def test_add_versions_delta_not_delta_index(self):
2517
2042
index = self.two_graph_index(catch_adds=True)
2518
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2519
[('new', 'no-eol,line-delta', (None, 0, 100), ['parent'])])
2043
self.assertRaises(errors.KnitCorrupt, index.add_records,
2044
[(('new',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
2520
2045
self.assertEqual([], self.caught_entries)
2522
2047
def test_add_versions_parents_not_parents_index(self):
2523
2048
index = self.two_graph_index(catch_adds=True)
2524
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2525
[('new', 'no-eol,fulltext', (None, 0, 100), ['parent'])])
2049
self.assertRaises(errors.KnitCorrupt, index.add_records,
2050
[(('new',), 'no-eol,fulltext', (None, 0, 100), [('parent',)])])
2526
2051
self.assertEqual([], self.caught_entries)
2053
def test_add_versions_random_id_accepted(self):
2054
index = self.two_graph_index(catch_adds=True)
2055
index.add_records([], random_id=True)
2528
2057
def test_add_versions_same_dup(self):
2529
2058
index = self.two_graph_index(catch_adds=True)
2530
2059
# options can be spelt two different ways
2531
index.add_versions([('tip', 'fulltext,no-eol', (None, 0, 100), [])])
2532
index.add_versions([('tip', 'no-eol,fulltext', (None, 0, 100), [])])
2060
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
2061
index.add_records([(('tip',), 'no-eol,fulltext', (None, 0, 100), [])])
2062
# position/length are ignored (because each pack could have fulltext or
2063
# delta, and be at a different position.
2064
index.add_records([(('tip',), 'fulltext,no-eol', (None, 50, 100), [])])
2065
index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000), [])])
2533
2066
# but neither should have added data.
2534
self.assertEqual([[], []], self.caught_entries)
2067
self.assertEqual([[], [], [], []], self.caught_entries)
2536
2069
def test_add_versions_different_dup(self):
2537
2070
index = self.two_graph_index(catch_adds=True)
2538
2071
# change options
2539
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2540
[('tip', 'no-eol,line-delta', (None, 0, 100), [])])
2541
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2542
[('tip', 'line-delta,no-eol', (None, 0, 100), [])])
2543
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2544
[('tip', 'fulltext', (None, 0, 100), [])])
2546
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2547
[('tip', 'fulltext,no-eol', (None, 50, 100), [])])
2548
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2549
[('tip', 'fulltext,no-eol', (None, 0, 1000), [])])
2072
self.assertRaises(errors.KnitCorrupt, index.add_records,
2073
[(('tip',), 'no-eol,line-delta', (None, 0, 100), [])])
2074
self.assertRaises(errors.KnitCorrupt, index.add_records,
2075
[(('tip',), 'line-delta,no-eol', (None, 0, 100), [])])
2076
self.assertRaises(errors.KnitCorrupt, index.add_records,
2077
[(('tip',), 'fulltext', (None, 0, 100), [])])
2551
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2552
[('tip', 'fulltext,no-eol', (None, 0, 100), ['parent'])])
2079
self.assertRaises(errors.KnitCorrupt, index.add_records,
2080
[(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
2553
2081
# change options in the second record
2554
self.assertRaises(errors.KnitCorrupt, index.add_versions,
2555
[('tip', 'fulltext,no-eol', (None, 0, 100), []),
2556
('tip', 'no-eol,line-delta', (None, 0, 100), [])])
2082
self.assertRaises(errors.KnitCorrupt, index.add_records,
2083
[(('tip',), 'fulltext,no-eol', (None, 0, 100), []),
2084
(('tip',), 'no-eol,line-delta', (None, 0, 100), [])])
2557
2085
self.assertEqual([], self.caught_entries)
2559
def test_iter_parents(self):
2560
index = self.two_graph_index()
2561
self.assertEqual(set([
2562
('tip', ()), ('tail', ()), ('parent', ()), ('separate', ())
2564
set(index.iter_parents(['tip', 'tail', 'ghost', 'parent', 'separate'])))
2565
self.assertEqual(set([('tip', ())]),
2566
set(index.iter_parents(['tip'])))
2567
self.assertEqual(set(),
2568
set(index.iter_parents([])))
2088
class TestKnitVersionedFiles(KnitTests):
2090
def assertGroupKeysForIo(self, exp_groups, keys, non_local_keys,
2091
positions, _min_buffer_size=None):
2092
kvf = self.make_test_knit()
2093
if _min_buffer_size is None:
2094
_min_buffer_size = knit._STREAM_MIN_BUFFER_SIZE
2095
self.assertEqual(exp_groups, kvf._group_keys_for_io(keys,
2096
non_local_keys, positions,
2097
_min_buffer_size=_min_buffer_size))
2099
def assertSplitByPrefix(self, expected_map, expected_prefix_order,
2101
split, prefix_order = KnitVersionedFiles._split_by_prefix(keys)
2102
self.assertEqual(expected_map, split)
2103
self.assertEqual(expected_prefix_order, prefix_order)
2105
def test__group_keys_for_io(self):
2106
ft_detail = ('fulltext', False)
2107
ld_detail = ('line-delta', False)
2115
f_a: (ft_detail, (f_a, 0, 100), None),
2116
f_b: (ld_detail, (f_b, 100, 21), f_a),
2117
f_c: (ld_detail, (f_c, 180, 15), f_b),
2118
g_a: (ft_detail, (g_a, 121, 35), None),
2119
g_b: (ld_detail, (g_b, 156, 12), g_a),
2120
g_c: (ld_detail, (g_c, 195, 13), g_a),
2122
self.assertGroupKeysForIo([([f_a], set())],
2123
[f_a], [], positions)
2124
self.assertGroupKeysForIo([([f_a], set([f_a]))],
2125
[f_a], [f_a], positions)
2126
self.assertGroupKeysForIo([([f_a, f_b], set([]))],
2127
[f_a, f_b], [], positions)
2128
self.assertGroupKeysForIo([([f_a, f_b], set([f_b]))],
2129
[f_a, f_b], [f_b], positions)
2130
self.assertGroupKeysForIo([([f_a, f_b, g_a, g_b], set())],
2131
[f_a, g_a, f_b, g_b], [], positions)
2132
self.assertGroupKeysForIo([([f_a, f_b, g_a, g_b], set())],
2133
[f_a, g_a, f_b, g_b], [], positions,
2134
_min_buffer_size=150)
2135
self.assertGroupKeysForIo([([f_a, f_b], set()), ([g_a, g_b], set())],
2136
[f_a, g_a, f_b, g_b], [], positions,
2137
_min_buffer_size=100)
2138
self.assertGroupKeysForIo([([f_c], set()), ([g_b], set())],
2139
[f_c, g_b], [], positions,
2140
_min_buffer_size=125)
2141
self.assertGroupKeysForIo([([g_b, f_c], set())],
2142
[g_b, f_c], [], positions,
2143
_min_buffer_size=125)
2145
def test__split_by_prefix(self):
2146
self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2147
'g': [('g', 'b'), ('g', 'a')],
2149
[('f', 'a'), ('g', 'b'),
2150
('g', 'a'), ('f', 'b')])
2152
self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2153
'g': [('g', 'b'), ('g', 'a')],
2155
[('f', 'a'), ('f', 'b'),
2156
('g', 'b'), ('g', 'a')])
2158
self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2159
'g': [('g', 'b'), ('g', 'a')],
2161
[('f', 'a'), ('f', 'b'),
2162
('g', 'b'), ('g', 'a')])
2164
self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2165
'g': [('g', 'b'), ('g', 'a')],
2166
'': [('a',), ('b',)]
2168
[('f', 'a'), ('g', 'b'),
2170
('g', 'a'), ('f', 'b')])
2173
class TestStacking(KnitTests):
2175
def get_basis_and_test_knit(self):
2176
basis = self.make_test_knit(name='basis')
2177
basis = RecordingVersionedFilesDecorator(basis)
2178
test = self.make_test_knit(name='test')
2179
test.add_fallback_versioned_files(basis)
2182
def test_add_fallback_versioned_files(self):
2183
basis = self.make_test_knit(name='basis')
2184
test = self.make_test_knit(name='test')
2185
# It must not error; other tests test that the fallback is referred to
2186
# when accessing data.
2187
test.add_fallback_versioned_files(basis)
2189
def test_add_lines(self):
2190
# lines added to the test are not added to the basis
2191
basis, test = self.get_basis_and_test_knit()
2193
key_basis = ('bar',)
2194
key_cross_border = ('quux',)
2195
key_delta = ('zaphod',)
2196
test.add_lines(key, (), ['foo\n'])
2197
self.assertEqual({}, basis.get_parent_map([key]))
2198
# lines added to the test that reference across the stack do a
2200
basis.add_lines(key_basis, (), ['foo\n'])
2202
test.add_lines(key_cross_border, (key_basis,), ['foo\n'])
2203
self.assertEqual('fulltext', test._index.get_method(key_cross_border))
2204
# we don't even need to look at the basis to see that this should be
2205
# stored as a fulltext
2206
self.assertEqual([], basis.calls)
2207
# Subsequent adds do delta.
2209
test.add_lines(key_delta, (key_cross_border,), ['foo\n'])
2210
self.assertEqual('line-delta', test._index.get_method(key_delta))
2211
self.assertEqual([], basis.calls)
2213
def test_annotate(self):
2214
# annotations from the test knit are answered without asking the basis
2215
basis, test = self.get_basis_and_test_knit()
2217
key_basis = ('bar',)
2218
key_missing = ('missing',)
2219
test.add_lines(key, (), ['foo\n'])
2220
details = test.annotate(key)
2221
self.assertEqual([(key, 'foo\n')], details)
2222
self.assertEqual([], basis.calls)
2223
# But texts that are not in the test knit are looked for in the basis
2225
basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
2227
details = test.annotate(key_basis)
2228
self.assertEqual([(key_basis, 'foo\n'), (key_basis, 'bar\n')], details)
2229
# Not optimised to date:
2230
# self.assertEqual([("annotate", key_basis)], basis.calls)
2231
self.assertEqual([('get_parent_map', set([key_basis])),
2232
('get_parent_map', set([key_basis])),
2233
('get_record_stream', [key_basis], 'topological', True)],
2236
def test_check(self):
2237
# At the moment checking a stacked knit does implicitly check the
2239
basis, test = self.get_basis_and_test_knit()
2242
def test_get_parent_map(self):
2243
# parents in the test knit are answered without asking the basis
2244
basis, test = self.get_basis_and_test_knit()
2246
key_basis = ('bar',)
2247
key_missing = ('missing',)
2248
test.add_lines(key, (), [])
2249
parent_map = test.get_parent_map([key])
2250
self.assertEqual({key: ()}, parent_map)
2251
self.assertEqual([], basis.calls)
2252
# But parents that are not in the test knit are looked for in the basis
2253
basis.add_lines(key_basis, (), [])
2255
parent_map = test.get_parent_map([key, key_basis, key_missing])
2256
self.assertEqual({key: (),
2257
key_basis: ()}, parent_map)
2258
self.assertEqual([("get_parent_map", set([key_basis, key_missing]))],
2261
def test_get_record_stream_unordered_fulltexts(self):
2262
# records from the test knit are answered without asking the basis:
2263
basis, test = self.get_basis_and_test_knit()
2265
key_basis = ('bar',)
2266
key_missing = ('missing',)
2267
test.add_lines(key, (), ['foo\n'])
2268
records = list(test.get_record_stream([key], 'unordered', True))
2269
self.assertEqual(1, len(records))
2270
self.assertEqual([], basis.calls)
2271
# Missing (from test knit) objects are retrieved from the basis:
2272
basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
2274
records = list(test.get_record_stream([key_basis, key_missing],
2276
self.assertEqual(2, len(records))
2277
calls = list(basis.calls)
2278
for record in records:
2279
self.assertSubset([record.key], (key_basis, key_missing))
2280
if record.key == key_missing:
2281
self.assertIsInstance(record, AbsentContentFactory)
2283
reference = list(basis.get_record_stream([key_basis],
2284
'unordered', True))[0]
2285
self.assertEqual(reference.key, record.key)
2286
self.assertEqual(reference.sha1, record.sha1)
2287
self.assertEqual(reference.storage_kind, record.storage_kind)
2288
self.assertEqual(reference.get_bytes_as(reference.storage_kind),
2289
record.get_bytes_as(record.storage_kind))
2290
self.assertEqual(reference.get_bytes_as('fulltext'),
2291
record.get_bytes_as('fulltext'))
2292
# It's not strictly minimal, but it seems reasonable for now for it to
2293
# ask which fallbacks have which parents.
2295
("get_parent_map", set([key_basis, key_missing])),
2296
("get_record_stream", [key_basis], 'unordered', True)],
2299
def test_get_record_stream_ordered_fulltexts(self):
2300
# ordering is preserved down into the fallback store.
2301
basis, test = self.get_basis_and_test_knit()
2303
key_basis = ('bar',)
2304
key_basis_2 = ('quux',)
2305
key_missing = ('missing',)
2306
test.add_lines(key, (key_basis,), ['foo\n'])
2307
# Missing (from test knit) objects are retrieved from the basis:
2308
basis.add_lines(key_basis, (key_basis_2,), ['foo\n', 'bar\n'])
2309
basis.add_lines(key_basis_2, (), ['quux\n'])
2311
# ask for in non-topological order
2312
records = list(test.get_record_stream(
2313
[key, key_basis, key_missing, key_basis_2], 'topological', True))
2314
self.assertEqual(4, len(records))
2316
for record in records:
2317
self.assertSubset([record.key],
2318
(key_basis, key_missing, key_basis_2, key))
2319
if record.key == key_missing:
2320
self.assertIsInstance(record, AbsentContentFactory)
2322
results.append((record.key, record.sha1, record.storage_kind,
2323
record.get_bytes_as('fulltext')))
2324
calls = list(basis.calls)
2325
order = [record[0] for record in results]
2326
self.assertEqual([key_basis_2, key_basis, key], order)
2327
for result in results:
2328
if result[0] == key:
2332
record = source.get_record_stream([result[0]], 'unordered',
2334
self.assertEqual(record.key, result[0])
2335
self.assertEqual(record.sha1, result[1])
2336
# We used to check that the storage kind matched, but actually it
2337
# depends on whether it was sourced from the basis, or in a single
2338
# group, because asking for full texts returns proxy objects to a
2339
# _ContentMapGenerator object; so checking the kind is unneeded.
2340
self.assertEqual(record.get_bytes_as('fulltext'), result[3])
2341
# It's not strictly minimal, but it seems reasonable for now for it to
2342
# ask which fallbacks have which parents.
2344
("get_parent_map", set([key_basis, key_basis_2, key_missing])),
2345
# topological is requested from the fallback, because that is what
2346
# was requested at the top level.
2347
("get_record_stream", [key_basis_2, key_basis], 'topological', True)],
2350
def test_get_record_stream_unordered_deltas(self):
2351
# records from the test knit are answered without asking the basis:
2352
basis, test = self.get_basis_and_test_knit()
2354
key_basis = ('bar',)
2355
key_missing = ('missing',)
2356
test.add_lines(key, (), ['foo\n'])
2357
records = list(test.get_record_stream([key], 'unordered', False))
2358
self.assertEqual(1, len(records))
2359
self.assertEqual([], basis.calls)
2360
# Missing (from test knit) objects are retrieved from the basis:
2361
basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
2363
records = list(test.get_record_stream([key_basis, key_missing],
2364
'unordered', False))
2365
self.assertEqual(2, len(records))
2366
calls = list(basis.calls)
2367
for record in records:
2368
self.assertSubset([record.key], (key_basis, key_missing))
2369
if record.key == key_missing:
2370
self.assertIsInstance(record, AbsentContentFactory)
2372
reference = list(basis.get_record_stream([key_basis],
2373
'unordered', False))[0]
2374
self.assertEqual(reference.key, record.key)
2375
self.assertEqual(reference.sha1, record.sha1)
2376
self.assertEqual(reference.storage_kind, record.storage_kind)
2377
self.assertEqual(reference.get_bytes_as(reference.storage_kind),
2378
record.get_bytes_as(record.storage_kind))
2379
# It's not strictly minimal, but it seems reasonable for now for it to
2380
# ask which fallbacks have which parents.
2382
("get_parent_map", set([key_basis, key_missing])),
2383
("get_record_stream", [key_basis], 'unordered', False)],
2386
def test_get_record_stream_ordered_deltas(self):
2387
# ordering is preserved down into the fallback store.
2388
basis, test = self.get_basis_and_test_knit()
2390
key_basis = ('bar',)
2391
key_basis_2 = ('quux',)
2392
key_missing = ('missing',)
2393
test.add_lines(key, (key_basis,), ['foo\n'])
2394
# Missing (from test knit) objects are retrieved from the basis:
2395
basis.add_lines(key_basis, (key_basis_2,), ['foo\n', 'bar\n'])
2396
basis.add_lines(key_basis_2, (), ['quux\n'])
2398
# ask for in non-topological order
2399
records = list(test.get_record_stream(
2400
[key, key_basis, key_missing, key_basis_2], 'topological', False))
2401
self.assertEqual(4, len(records))
2403
for record in records:
2404
self.assertSubset([record.key],
2405
(key_basis, key_missing, key_basis_2, key))
2406
if record.key == key_missing:
2407
self.assertIsInstance(record, AbsentContentFactory)
2409
results.append((record.key, record.sha1, record.storage_kind,
2410
record.get_bytes_as(record.storage_kind)))
2411
calls = list(basis.calls)
2412
order = [record[0] for record in results]
2413
self.assertEqual([key_basis_2, key_basis, key], order)
2414
for result in results:
2415
if result[0] == key:
2419
record = source.get_record_stream([result[0]], 'unordered',
2421
self.assertEqual(record.key, result[0])
2422
self.assertEqual(record.sha1, result[1])
2423
self.assertEqual(record.storage_kind, result[2])
2424
self.assertEqual(record.get_bytes_as(record.storage_kind), result[3])
2425
# It's not strictly minimal, but it seems reasonable for now for it to
2426
# ask which fallbacks have which parents.
2428
("get_parent_map", set([key_basis, key_basis_2, key_missing])),
2429
("get_record_stream", [key_basis_2, key_basis], 'topological', False)],
2432
def test_get_sha1s(self):
2433
# sha1's in the test knit are answered without asking the basis
2434
basis, test = self.get_basis_and_test_knit()
2436
key_basis = ('bar',)
2437
key_missing = ('missing',)
2438
test.add_lines(key, (), ['foo\n'])
2439
key_sha1sum = osutils.sha('foo\n').hexdigest()
2440
sha1s = test.get_sha1s([key])
2441
self.assertEqual({key: key_sha1sum}, sha1s)
2442
self.assertEqual([], basis.calls)
2443
# But texts that are not in the test knit are looked for in the basis
2444
# directly (rather than via text reconstruction) so that remote servers
2445
# etc don't have to answer with full content.
2446
basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
2447
basis_sha1sum = osutils.sha('foo\nbar\n').hexdigest()
2449
sha1s = test.get_sha1s([key, key_missing, key_basis])
2450
self.assertEqual({key: key_sha1sum,
2451
key_basis: basis_sha1sum}, sha1s)
2452
self.assertEqual([("get_sha1s", set([key_basis, key_missing]))],
2455
def test_insert_record_stream(self):
2456
# records are inserted as normal; insert_record_stream builds on
2457
# add_lines, so a smoke test should be all that's needed:
2459
key_basis = ('bar',)
2460
key_delta = ('zaphod',)
2461
basis, test = self.get_basis_and_test_knit()
2462
source = self.make_test_knit(name='source')
2463
basis.add_lines(key_basis, (), ['foo\n'])
2465
source.add_lines(key_basis, (), ['foo\n'])
2466
source.add_lines(key_delta, (key_basis,), ['bar\n'])
2467
stream = source.get_record_stream([key_delta], 'unordered', False)
2468
test.insert_record_stream(stream)
2469
# XXX: this does somewhat too many calls in making sure of whether it
2470
# has to recreate the full text.
2471
self.assertEqual([("get_parent_map", set([key_basis])),
2472
('get_parent_map', set([key_basis])),
2473
('get_record_stream', [key_basis], 'unordered', True)],
2475
self.assertEqual({key_delta:(key_basis,)},
2476
test.get_parent_map([key_delta]))
2477
self.assertEqual('bar\n', test.get_record_stream([key_delta],
2478
'unordered', True).next().get_bytes_as('fulltext'))
2480
def test_iter_lines_added_or_present_in_keys(self):
2481
# Lines from the basis are returned, and lines for a given key are only
2485
# all sources are asked for keys:
2486
basis, test = self.get_basis_and_test_knit()
2487
basis.add_lines(key1, (), ["foo"])
2489
lines = list(test.iter_lines_added_or_present_in_keys([key1]))
2490
self.assertEqual([("foo\n", key1)], lines)
2491
self.assertEqual([("iter_lines_added_or_present_in_keys", set([key1]))],
2493
# keys in both are not duplicated:
2494
test.add_lines(key2, (), ["bar\n"])
2495
basis.add_lines(key2, (), ["bar\n"])
2497
lines = list(test.iter_lines_added_or_present_in_keys([key2]))
2498
self.assertEqual([("bar\n", key2)], lines)
2499
self.assertEqual([], basis.calls)
2501
def test_keys(self):
2504
# all sources are asked for keys:
2505
basis, test = self.get_basis_and_test_knit()
2507
self.assertEqual(set(), set(keys))
2508
self.assertEqual([("keys",)], basis.calls)
2509
# keys from a basis are returned:
2510
basis.add_lines(key1, (), [])
2513
self.assertEqual(set([key1]), set(keys))
2514
self.assertEqual([("keys",)], basis.calls)
2515
# keys in both are not duplicated:
2516
test.add_lines(key2, (), [])
2517
basis.add_lines(key2, (), [])
2520
self.assertEqual(2, len(keys))
2521
self.assertEqual(set([key1, key2]), set(keys))
2522
self.assertEqual([("keys",)], basis.calls)
2524
def test_add_mpdiffs(self):
2525
# records are inserted as normal; add_mpdiff builds on
2526
# add_lines, so a smoke test should be all that's needed:
2528
key_basis = ('bar',)
2529
key_delta = ('zaphod',)
2530
basis, test = self.get_basis_and_test_knit()
2531
source = self.make_test_knit(name='source')
2532
basis.add_lines(key_basis, (), ['foo\n'])
2534
source.add_lines(key_basis, (), ['foo\n'])
2535
source.add_lines(key_delta, (key_basis,), ['bar\n'])
2536
diffs = source.make_mpdiffs([key_delta])
2537
test.add_mpdiffs([(key_delta, (key_basis,),
2538
source.get_sha1s([key_delta])[key_delta], diffs[0])])
2539
self.assertEqual([("get_parent_map", set([key_basis])),
2540
('get_record_stream', [key_basis], 'unordered', True),],
2542
self.assertEqual({key_delta:(key_basis,)},
2543
test.get_parent_map([key_delta]))
2544
self.assertEqual('bar\n', test.get_record_stream([key_delta],
2545
'unordered', True).next().get_bytes_as('fulltext'))
2547
def test_make_mpdiffs(self):
2548
# Generating an mpdiff across a stacking boundary should detect parent
2552
key_right = ('zaphod',)
2553
basis, test = self.get_basis_and_test_knit()
2554
basis.add_lines(key_left, (), ['bar\n'])
2555
basis.add_lines(key_right, (), ['zaphod\n'])
2557
test.add_lines(key, (key_left, key_right),
2558
['bar\n', 'foo\n', 'zaphod\n'])
2559
diffs = test.make_mpdiffs([key])
2561
multiparent.MultiParent([multiparent.ParentText(0, 0, 0, 1),
2562
multiparent.NewText(['foo\n']),
2563
multiparent.ParentText(1, 0, 2, 1)])],
2565
self.assertEqual(3, len(basis.calls))
2567
("get_parent_map", set([key_left, key_right])),
2568
("get_parent_map", set([key_left, key_right])),
2571
last_call = basis.calls[-1]
2572
self.assertEqual('get_record_stream', last_call[0])
2573
self.assertEqual(set([key_left, key_right]), set(last_call[1]))
2574
self.assertEqual('topological', last_call[2])
2575
self.assertEqual(True, last_call[3])
2578
class TestNetworkBehaviour(KnitTests):
2579
"""Tests for getting data out of/into knits over the network."""
2581
def test_include_delta_closure_generates_a_knit_delta_closure(self):
2582
vf = self.make_test_knit(name='test')
2583
# put in three texts, giving ft, delta, delta
2584
vf.add_lines(('base',), (), ['base\n', 'content\n'])
2585
vf.add_lines(('d1',), (('base',),), ['d1\n'])
2586
vf.add_lines(('d2',), (('d1',),), ['d2\n'])
2587
# But heuristics could interfere, so check what happened:
2588
self.assertEqual(['knit-ft-gz', 'knit-delta-gz', 'knit-delta-gz'],
2589
[record.storage_kind for record in
2590
vf.get_record_stream([('base',), ('d1',), ('d2',)],
2591
'topological', False)])
2592
# generate a stream of just the deltas include_delta_closure=True,
2593
# serialise to the network, and check that we get a delta closure on the wire.
2594
stream = vf.get_record_stream([('d1',), ('d2',)], 'topological', True)
2595
netb = [record.get_bytes_as(record.storage_kind) for record in stream]
2596
# The first bytes should be a memo from _ContentMapGenerator, and the
2597
# second bytes should be empty (because its a API proxy not something
2598
# for wire serialisation.
2599
self.assertEqual('', netb[1])
2601
kind, line_end = network_bytes_to_kind_and_offset(bytes)
2602
self.assertEqual('knit-delta-closure', kind)
2605
class TestContentMapGenerator(KnitTests):
2606
"""Tests for ContentMapGenerator"""
2608
def test_get_record_stream_gives_records(self):
2609
vf = self.make_test_knit(name='test')
2610
# put in three texts, giving ft, delta, delta
2611
vf.add_lines(('base',), (), ['base\n', 'content\n'])
2612
vf.add_lines(('d1',), (('base',),), ['d1\n'])
2613
vf.add_lines(('d2',), (('d1',),), ['d2\n'])
2614
keys = [('d1',), ('d2',)]
2615
generator = _VFContentMapGenerator(vf, keys,
2616
global_map=vf.get_parent_map(keys))
2617
for record in generator.get_record_stream():
2618
if record.key == ('d1',):
2619
self.assertEqual('d1\n', record.get_bytes_as('fulltext'))
2621
self.assertEqual('d2\n', record.get_bytes_as('fulltext'))
2623
def test_get_record_stream_kinds_are_raw(self):
2624
vf = self.make_test_knit(name='test')
2625
# put in three texts, giving ft, delta, delta
2626
vf.add_lines(('base',), (), ['base\n', 'content\n'])
2627
vf.add_lines(('d1',), (('base',),), ['d1\n'])
2628
vf.add_lines(('d2',), (('d1',),), ['d2\n'])
2629
keys = [('base',), ('d1',), ('d2',)]
2630
generator = _VFContentMapGenerator(vf, keys,
2631
global_map=vf.get_parent_map(keys))
2632
kinds = {('base',): 'knit-delta-closure',
2633
('d1',): 'knit-delta-closure-ref',
2634
('d2',): 'knit-delta-closure-ref',
2636
for record in generator.get_record_stream():
2637
self.assertEqual(kinds[record.key], record.storage_kind)
added
removed
bzrlib/tests/test_knit.py
