~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_knit.py

  • Committer: John Arbash Meinel
  • Date: 2008-10-04 14:10:13 UTC
  • mto: This revision was merged to the branch mainline in revision 3805.
  • Revision ID: john@arbash-meinel.com-20081004141013-yskxjlwtuy2k18ue
Playing around with expanding requests for btree index nodes into neighboring nodes.

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2006-2011 Canonical Ltd
 
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
17
17
"""Tests for Knit data structure"""
18
18
 
19
19
from cStringIO import StringIO
 
20
import difflib
20
21
import gzip
21
22
import sys
22
23
 
23
24
from bzrlib import (
24
25
    errors,
 
26
    generate_ids,
25
27
    knit,
26
28
    multiparent,
27
29
    osutils,
28
30
    pack,
29
 
    tests,
30
 
    transport,
31
31
    )
32
32
from bzrlib.errors import (
 
33
    RevisionAlreadyPresent,
33
34
    KnitHeaderError,
 
35
    RevisionNotPresent,
34
36
    NoSuchFile,
35
37
    )
36
38
from bzrlib.index import *
37
39
from bzrlib.knit import (
38
40
    AnnotatedKnitContent,
39
41
    KnitContent,
 
42
    KnitSequenceMatcher,
40
43
    KnitVersionedFiles,
41
44
    PlainKnitContent,
42
 
    _VFContentMapGenerator,
 
45
    _DirectPackAccess,
43
46
    _KndxIndex,
44
47
    _KnitGraphIndex,
45
48
    _KnitKeyAccess,
46
49
    make_file_factory,
47
50
    )
48
 
from bzrlib.patiencediff import PatienceSequenceMatcher
49
 
from bzrlib.repofmt import (
50
 
    knitpack_repo,
51
 
    pack_repo,
52
 
    )
53
51
from bzrlib.tests import (
 
52
    Feature,
 
53
    KnownFailure,
54
54
    TestCase,
55
55
    TestCaseWithMemoryTransport,
56
56
    TestCaseWithTransport,
57
 
    TestNotApplicable,
58
57
    )
 
58
from bzrlib.transport import get_transport
 
59
from bzrlib.transport.memory import MemoryTransport
 
60
from bzrlib.tuned_gzip import GzipFile
59
61
from bzrlib.versionedfile import (
60
62
    AbsentContentFactory,
61
63
    ConstantMapper,
62
 
    network_bytes_to_kind_and_offset,
63
64
    RecordingVersionedFilesDecorator,
64
65
    )
65
 
from bzrlib.tests import (
66
 
    features,
67
 
    )
68
 
 
69
 
 
70
 
compiled_knit_feature = features.ModuleAvailableFeature(
71
 
    'bzrlib._knit_load_data_pyx')
 
66
 
 
67
 
 
68
class _CompiledKnitFeature(Feature):
 
69
 
 
70
    def _probe(self):
 
71
        try:
 
72
            import bzrlib._knit_load_data_c
 
73
        except ImportError:
 
74
            return False
 
75
        return True
 
76
 
 
77
    def feature_name(self):
 
78
        return 'bzrlib._knit_load_data_c'
 
79
 
 
80
CompiledKnitFeature = _CompiledKnitFeature()
72
81
 
73
82
 
74
83
class KnitContentTestsMixin(object):
103
112
        line_delta = source_content.line_delta(target_content)
104
113
        delta_blocks = list(KnitContent.get_line_delta_blocks(line_delta,
105
114
            source_lines, target_lines))
106
 
        matcher = PatienceSequenceMatcher(None, source_lines, target_lines)
107
 
        matcher_blocks = list(matcher.get_matching_blocks())
 
115
        matcher = KnitSequenceMatcher(None, source_lines, target_lines)
 
116
        matcher_blocks = list(list(matcher.get_matching_blocks()))
108
117
        self.assertEqual(matcher_blocks, delta_blocks)
109
118
 
110
119
    def test_get_line_delta_blocks(self):
260
269
        return queue_call
261
270
 
262
271
 
263
 
class MockReadvFailingTransport(MockTransport):
264
 
    """Fail in the middle of a readv() result.
265
 
 
266
 
    This Transport will successfully yield the first two requested hunks, but
267
 
    raise NoSuchFile for the rest.
268
 
    """
269
 
 
270
 
    def readv(self, relpath, offsets):
271
 
        count = 0
272
 
        for result in MockTransport.readv(self, relpath, offsets):
273
 
            count += 1
274
 
            # we use 2 because the first offset is the pack header, the second
275
 
            # is the first actual content requset
276
 
            if count > 2:
277
 
                raise errors.NoSuchFile(relpath)
278
 
            yield result
279
 
 
280
 
 
281
272
class KnitRecordAccessTestsMixin(object):
282
273
    """Tests for getting and putting knit records."""
283
274
 
286
277
        access = self.get_access()
287
278
        memos = access.add_raw_records([('key', 10)], '1234567890')
288
279
        self.assertEqual(['1234567890'], list(access.get_raw_records(memos)))
289
 
 
 
280
 
290
281
    def test_add_several_raw_records(self):
291
282
        """add_raw_records with many records and read some back."""
292
283
        access = self.get_access()
312
303
        mapper = ConstantMapper("foo")
313
304
        access = _KnitKeyAccess(self.get_transport(), mapper)
314
305
        return access
315
 
 
316
 
 
317
 
class _TestException(Exception):
318
 
    """Just an exception for local tests to use."""
319
 
 
 
306
    
320
307
 
321
308
class TestPackKnitAccess(TestCaseWithMemoryTransport, KnitRecordAccessTestsMixin):
322
309
    """Tests for the pack based access."""
330
317
            transport.append_bytes(packname, bytes)
331
318
        writer = pack.ContainerWriter(write_data)
332
319
        writer.begin()
333
 
        access = pack_repo._DirectPackAccess({})
 
320
        access = _DirectPackAccess({})
334
321
        access.set_writer(writer, index, (transport, packname))
335
322
        return access, writer
336
323
 
337
 
    def make_pack_file(self):
338
 
        """Create a pack file with 2 records."""
339
 
        access, writer = self._get_access(packname='packname', index='foo')
340
 
        memos = []
341
 
        memos.extend(access.add_raw_records([('key1', 10)], '1234567890'))
342
 
        memos.extend(access.add_raw_records([('key2', 5)], '12345'))
343
 
        writer.end()
344
 
        return memos
345
 
 
346
 
    def test_pack_collection_pack_retries(self):
347
 
        """An explicit pack of a pack collection succeeds even when a
348
 
        concurrent pack happens.
349
 
        """
350
 
        builder = self.make_branch_builder('.')
351
 
        builder.start_series()
352
 
        builder.build_snapshot('rev-1', None, [
353
 
            ('add', ('', 'root-id', 'directory', None)),
354
 
            ('add', ('file', 'file-id', 'file', 'content\nrev 1\n')),
355
 
            ])
356
 
        builder.build_snapshot('rev-2', ['rev-1'], [
357
 
            ('modify', ('file-id', 'content\nrev 2\n')),
358
 
            ])
359
 
        builder.build_snapshot('rev-3', ['rev-2'], [
360
 
            ('modify', ('file-id', 'content\nrev 3\n')),
361
 
            ])
362
 
        self.addCleanup(builder.finish_series)
363
 
        b = builder.get_branch()
364
 
        self.addCleanup(b.lock_write().unlock)
365
 
        repo = b.repository
366
 
        collection = repo._pack_collection
367
 
        # Concurrently repack the repo.
368
 
        reopened_repo = repo.bzrdir.open_repository()
369
 
        reopened_repo.pack()
370
 
        # Pack the new pack.
371
 
        collection.pack()
372
 
 
373
 
    def make_vf_for_retrying(self):
374
 
        """Create 3 packs and a reload function.
375
 
 
376
 
        Originally, 2 pack files will have the data, but one will be missing.
377
 
        And then the third will be used in place of the first two if reload()
378
 
        is called.
379
 
 
380
 
        :return: (versioned_file, reload_counter)
381
 
            versioned_file  a KnitVersionedFiles using the packs for access
382
 
        """
383
 
        builder = self.make_branch_builder('.', format="1.9")
384
 
        builder.start_series()
385
 
        builder.build_snapshot('rev-1', None, [
386
 
            ('add', ('', 'root-id', 'directory', None)),
387
 
            ('add', ('file', 'file-id', 'file', 'content\nrev 1\n')),
388
 
            ])
389
 
        builder.build_snapshot('rev-2', ['rev-1'], [
390
 
            ('modify', ('file-id', 'content\nrev 2\n')),
391
 
            ])
392
 
        builder.build_snapshot('rev-3', ['rev-2'], [
393
 
            ('modify', ('file-id', 'content\nrev 3\n')),
394
 
            ])
395
 
        builder.finish_series()
396
 
        b = builder.get_branch()
397
 
        b.lock_write()
398
 
        self.addCleanup(b.unlock)
399
 
        # Pack these three revisions into another pack file, but don't remove
400
 
        # the originals
401
 
        repo = b.repository
402
 
        collection = repo._pack_collection
403
 
        collection.ensure_loaded()
404
 
        orig_packs = collection.packs
405
 
        packer = knitpack_repo.KnitPacker(collection, orig_packs, '.testpack')
406
 
        new_pack = packer.pack()
407
 
        # forget about the new pack
408
 
        collection.reset()
409
 
        repo.refresh_data()
410
 
        vf = repo.revisions
411
 
        # Set up a reload() function that switches to using the new pack file
412
 
        new_index = new_pack.revision_index
413
 
        access_tuple = new_pack.access_tuple()
414
 
        reload_counter = [0, 0, 0]
415
 
        def reload():
416
 
            reload_counter[0] += 1
417
 
            if reload_counter[1] > 0:
418
 
                # We already reloaded, nothing more to do
419
 
                reload_counter[2] += 1
420
 
                return False
421
 
            reload_counter[1] += 1
422
 
            vf._index._graph_index._indices[:] = [new_index]
423
 
            vf._access._indices.clear()
424
 
            vf._access._indices[new_index] = access_tuple
425
 
            return True
426
 
        # Delete one of the pack files so the data will need to be reloaded. We
427
 
        # will delete the file with 'rev-2' in it
428
 
        trans, name = orig_packs[1].access_tuple()
429
 
        trans.delete(name)
430
 
        # We don't have the index trigger reloading because we want to test
431
 
        # that we reload when the .pack disappears
432
 
        vf._access._reload_func = reload
433
 
        return vf, reload_counter
434
 
 
435
 
    def make_reload_func(self, return_val=True):
436
 
        reload_called = [0]
437
 
        def reload():
438
 
            reload_called[0] += 1
439
 
            return return_val
440
 
        return reload_called, reload
441
 
 
442
 
    def make_retry_exception(self):
443
 
        # We raise a real exception so that sys.exc_info() is properly
444
 
        # populated
445
 
        try:
446
 
            raise _TestException('foobar')
447
 
        except _TestException, e:
448
 
            retry_exc = errors.RetryWithNewPacks(None, reload_occurred=False,
449
 
                                                 exc_info=sys.exc_info())
450
 
        # GZ 2010-08-10: Cycle with exc_info affects 3 tests
451
 
        return retry_exc
452
 
 
453
324
    def test_read_from_several_packs(self):
454
325
        access, writer = self._get_access()
455
326
        memos = []
462
333
        memos.extend(access.add_raw_records([('key', 5)], 'alpha'))
463
334
        writer.end()
464
335
        transport = self.get_transport()
465
 
        access = pack_repo._DirectPackAccess({"FOO":(transport, 'packfile'),
 
336
        access = _DirectPackAccess({"FOO":(transport, 'packfile'),
466
337
            "FOOBAR":(transport, 'pack2'),
467
338
            "BAZ":(transport, 'pack3')})
468
339
        self.assertEqual(['1234567890', '12345', 'alpha'],
478
349
 
479
350
    def test_set_writer(self):
480
351
        """The writer should be settable post construction."""
481
 
        access = pack_repo._DirectPackAccess({})
 
352
        access = _DirectPackAccess({})
482
353
        transport = self.get_transport()
483
354
        packname = 'packfile'
484
355
        index = 'foo'
491
362
        writer.end()
492
363
        self.assertEqual(['1234567890'], list(access.get_raw_records(memos)))
493
364
 
494
 
    def test_missing_index_raises_retry(self):
495
 
        memos = self.make_pack_file()
496
 
        transport = self.get_transport()
497
 
        reload_called, reload_func = self.make_reload_func()
498
 
        # Note that the index key has changed from 'foo' to 'bar'
499
 
        access = pack_repo._DirectPackAccess({'bar':(transport, 'packname')},
500
 
                                   reload_func=reload_func)
501
 
        e = self.assertListRaises(errors.RetryWithNewPacks,
502
 
                                  access.get_raw_records, memos)
503
 
        # Because a key was passed in which does not match our index list, we
504
 
        # assume that the listing was already reloaded
505
 
        self.assertTrue(e.reload_occurred)
506
 
        self.assertIsInstance(e.exc_info, tuple)
507
 
        self.assertIs(e.exc_info[0], KeyError)
508
 
        self.assertIsInstance(e.exc_info[1], KeyError)
509
 
 
510
 
    def test_missing_index_raises_key_error_with_no_reload(self):
511
 
        memos = self.make_pack_file()
512
 
        transport = self.get_transport()
513
 
        # Note that the index key has changed from 'foo' to 'bar'
514
 
        access = pack_repo._DirectPackAccess({'bar':(transport, 'packname')})
515
 
        e = self.assertListRaises(KeyError, access.get_raw_records, memos)
516
 
 
517
 
    def test_missing_file_raises_retry(self):
518
 
        memos = self.make_pack_file()
519
 
        transport = self.get_transport()
520
 
        reload_called, reload_func = self.make_reload_func()
521
 
        # Note that the 'filename' has been changed to 'different-packname'
522
 
        access = pack_repo._DirectPackAccess(
523
 
            {'foo':(transport, 'different-packname')},
524
 
            reload_func=reload_func)
525
 
        e = self.assertListRaises(errors.RetryWithNewPacks,
526
 
                                  access.get_raw_records, memos)
527
 
        # The file has gone missing, so we assume we need to reload
528
 
        self.assertFalse(e.reload_occurred)
529
 
        self.assertIsInstance(e.exc_info, tuple)
530
 
        self.assertIs(e.exc_info[0], errors.NoSuchFile)
531
 
        self.assertIsInstance(e.exc_info[1], errors.NoSuchFile)
532
 
        self.assertEqual('different-packname', e.exc_info[1].path)
533
 
 
534
 
    def test_missing_file_raises_no_such_file_with_no_reload(self):
535
 
        memos = self.make_pack_file()
536
 
        transport = self.get_transport()
537
 
        # Note that the 'filename' has been changed to 'different-packname'
538
 
        access = pack_repo._DirectPackAccess(
539
 
            {'foo': (transport, 'different-packname')})
540
 
        e = self.assertListRaises(errors.NoSuchFile,
541
 
                                  access.get_raw_records, memos)
542
 
 
543
 
    def test_failing_readv_raises_retry(self):
544
 
        memos = self.make_pack_file()
545
 
        transport = self.get_transport()
546
 
        failing_transport = MockReadvFailingTransport(
547
 
                                [transport.get_bytes('packname')])
548
 
        reload_called, reload_func = self.make_reload_func()
549
 
        access = pack_repo._DirectPackAccess(
550
 
            {'foo': (failing_transport, 'packname')},
551
 
            reload_func=reload_func)
552
 
        # Asking for a single record will not trigger the Mock failure
553
 
        self.assertEqual(['1234567890'],
554
 
            list(access.get_raw_records(memos[:1])))
555
 
        self.assertEqual(['12345'],
556
 
            list(access.get_raw_records(memos[1:2])))
557
 
        # A multiple offset readv() will fail mid-way through
558
 
        e = self.assertListRaises(errors.RetryWithNewPacks,
559
 
                                  access.get_raw_records, memos)
560
 
        # The file has gone missing, so we assume we need to reload
561
 
        self.assertFalse(e.reload_occurred)
562
 
        self.assertIsInstance(e.exc_info, tuple)
563
 
        self.assertIs(e.exc_info[0], errors.NoSuchFile)
564
 
        self.assertIsInstance(e.exc_info[1], errors.NoSuchFile)
565
 
        self.assertEqual('packname', e.exc_info[1].path)
566
 
 
567
 
    def test_failing_readv_raises_no_such_file_with_no_reload(self):
568
 
        memos = self.make_pack_file()
569
 
        transport = self.get_transport()
570
 
        failing_transport = MockReadvFailingTransport(
571
 
                                [transport.get_bytes('packname')])
572
 
        reload_called, reload_func = self.make_reload_func()
573
 
        access = pack_repo._DirectPackAccess(
574
 
            {'foo':(failing_transport, 'packname')})
575
 
        # Asking for a single record will not trigger the Mock failure
576
 
        self.assertEqual(['1234567890'],
577
 
            list(access.get_raw_records(memos[:1])))
578
 
        self.assertEqual(['12345'],
579
 
            list(access.get_raw_records(memos[1:2])))
580
 
        # A multiple offset readv() will fail mid-way through
581
 
        e = self.assertListRaises(errors.NoSuchFile,
582
 
                                  access.get_raw_records, memos)
583
 
 
584
 
    def test_reload_or_raise_no_reload(self):
585
 
        access = pack_repo._DirectPackAccess({}, reload_func=None)
586
 
        retry_exc = self.make_retry_exception()
587
 
        # Without a reload_func, we will just re-raise the original exception
588
 
        self.assertRaises(_TestException, access.reload_or_raise, retry_exc)
589
 
 
590
 
    def test_reload_or_raise_reload_changed(self):
591
 
        reload_called, reload_func = self.make_reload_func(return_val=True)
592
 
        access = pack_repo._DirectPackAccess({}, reload_func=reload_func)
593
 
        retry_exc = self.make_retry_exception()
594
 
        access.reload_or_raise(retry_exc)
595
 
        self.assertEqual([1], reload_called)
596
 
        retry_exc.reload_occurred=True
597
 
        access.reload_or_raise(retry_exc)
598
 
        self.assertEqual([2], reload_called)
599
 
 
600
 
    def test_reload_or_raise_reload_no_change(self):
601
 
        reload_called, reload_func = self.make_reload_func(return_val=False)
602
 
        access = pack_repo._DirectPackAccess({}, reload_func=reload_func)
603
 
        retry_exc = self.make_retry_exception()
604
 
        # If reload_occurred is False, then we consider it an error to have
605
 
        # reload_func() return False (no changes).
606
 
        self.assertRaises(_TestException, access.reload_or_raise, retry_exc)
607
 
        self.assertEqual([1], reload_called)
608
 
        retry_exc.reload_occurred=True
609
 
        # If reload_occurred is True, then we assume nothing changed because
610
 
        # it had changed earlier, but didn't change again
611
 
        access.reload_or_raise(retry_exc)
612
 
        self.assertEqual([2], reload_called)
613
 
 
614
 
    def test_annotate_retries(self):
615
 
        vf, reload_counter = self.make_vf_for_retrying()
616
 
        # It is a little bit bogus to annotate the Revision VF, but it works,
617
 
        # as we have ancestry stored there
618
 
        key = ('rev-3',)
619
 
        reload_lines = vf.annotate(key)
620
 
        self.assertEqual([1, 1, 0], reload_counter)
621
 
        plain_lines = vf.annotate(key)
622
 
        self.assertEqual([1, 1, 0], reload_counter) # No extra reloading
623
 
        if reload_lines != plain_lines:
624
 
            self.fail('Annotation was not identical with reloading.')
625
 
        # Now delete the packs-in-use, which should trigger another reload, but
626
 
        # this time we just raise an exception because we can't recover
627
 
        for trans, name in vf._access._indices.itervalues():
628
 
            trans.delete(name)
629
 
        self.assertRaises(errors.NoSuchFile, vf.annotate, key)
630
 
        self.assertEqual([2, 1, 1], reload_counter)
631
 
 
632
 
    def test__get_record_map_retries(self):
633
 
        vf, reload_counter = self.make_vf_for_retrying()
634
 
        keys = [('rev-1',), ('rev-2',), ('rev-3',)]
635
 
        records = vf._get_record_map(keys)
636
 
        self.assertEqual(keys, sorted(records.keys()))
637
 
        self.assertEqual([1, 1, 0], reload_counter)
638
 
        # Now delete the packs-in-use, which should trigger another reload, but
639
 
        # this time we just raise an exception because we can't recover
640
 
        for trans, name in vf._access._indices.itervalues():
641
 
            trans.delete(name)
642
 
        self.assertRaises(errors.NoSuchFile, vf._get_record_map, keys)
643
 
        self.assertEqual([2, 1, 1], reload_counter)
644
 
 
645
 
    def test_get_record_stream_retries(self):
646
 
        vf, reload_counter = self.make_vf_for_retrying()
647
 
        keys = [('rev-1',), ('rev-2',), ('rev-3',)]
648
 
        record_stream = vf.get_record_stream(keys, 'topological', False)
649
 
        record = record_stream.next()
650
 
        self.assertEqual(('rev-1',), record.key)
651
 
        self.assertEqual([0, 0, 0], reload_counter)
652
 
        record = record_stream.next()
653
 
        self.assertEqual(('rev-2',), record.key)
654
 
        self.assertEqual([1, 1, 0], reload_counter)
655
 
        record = record_stream.next()
656
 
        self.assertEqual(('rev-3',), record.key)
657
 
        self.assertEqual([1, 1, 0], reload_counter)
658
 
        # Now delete all pack files, and see that we raise the right error
659
 
        for trans, name in vf._access._indices.itervalues():
660
 
            trans.delete(name)
661
 
        self.assertListRaises(errors.NoSuchFile,
662
 
            vf.get_record_stream, keys, 'topological', False)
663
 
 
664
 
    def test_iter_lines_added_or_present_in_keys_retries(self):
665
 
        vf, reload_counter = self.make_vf_for_retrying()
666
 
        keys = [('rev-1',), ('rev-2',), ('rev-3',)]
667
 
        # Unfortunately, iter_lines_added_or_present_in_keys iterates the
668
 
        # result in random order (determined by the iteration order from a
669
 
        # set()), so we don't have any solid way to trigger whether data is
670
 
        # read before or after. However we tried to delete the middle node to
671
 
        # exercise the code well.
672
 
        # What we care about is that all lines are always yielded, but not
673
 
        # duplicated
674
 
        count = 0
675
 
        reload_lines = sorted(vf.iter_lines_added_or_present_in_keys(keys))
676
 
        self.assertEqual([1, 1, 0], reload_counter)
677
 
        # Now do it again, to make sure the result is equivalent
678
 
        plain_lines = sorted(vf.iter_lines_added_or_present_in_keys(keys))
679
 
        self.assertEqual([1, 1, 0], reload_counter) # No extra reloading
680
 
        self.assertEqual(plain_lines, reload_lines)
681
 
        self.assertEqual(21, len(plain_lines))
682
 
        # Now delete all pack files, and see that we raise the right error
683
 
        for trans, name in vf._access._indices.itervalues():
684
 
            trans.delete(name)
685
 
        self.assertListRaises(errors.NoSuchFile,
686
 
            vf.iter_lines_added_or_present_in_keys, keys)
687
 
        self.assertEqual([2, 1, 1], reload_counter)
688
 
 
689
 
    def test_get_record_stream_yields_disk_sorted_order(self):
690
 
        # if we get 'unordered' pick a semi-optimal order for reading. The
691
 
        # order should be grouped by pack file, and then by position in file
692
 
        repo = self.make_repository('test', format='pack-0.92')
693
 
        repo.lock_write()
694
 
        self.addCleanup(repo.unlock)
695
 
        repo.start_write_group()
696
 
        vf = repo.texts
697
 
        vf.add_lines(('f-id', 'rev-5'), [('f-id', 'rev-4')], ['lines\n'])
698
 
        vf.add_lines(('f-id', 'rev-1'), [], ['lines\n'])
699
 
        vf.add_lines(('f-id', 'rev-2'), [('f-id', 'rev-1')], ['lines\n'])
700
 
        repo.commit_write_group()
701
 
        # We inserted them as rev-5, rev-1, rev-2, we should get them back in
702
 
        # the same order
703
 
        stream = vf.get_record_stream([('f-id', 'rev-1'), ('f-id', 'rev-5'),
704
 
                                       ('f-id', 'rev-2')], 'unordered', False)
705
 
        keys = [r.key for r in stream]
706
 
        self.assertEqual([('f-id', 'rev-5'), ('f-id', 'rev-1'),
707
 
                          ('f-id', 'rev-2')], keys)
708
 
        repo.start_write_group()
709
 
        vf.add_lines(('f-id', 'rev-4'), [('f-id', 'rev-3')], ['lines\n'])
710
 
        vf.add_lines(('f-id', 'rev-3'), [('f-id', 'rev-2')], ['lines\n'])
711
 
        vf.add_lines(('f-id', 'rev-6'), [('f-id', 'rev-5')], ['lines\n'])
712
 
        repo.commit_write_group()
713
 
        # Request in random order, to make sure the output order isn't based on
714
 
        # the request
715
 
        request_keys = set(('f-id', 'rev-%d' % i) for i in range(1, 7))
716
 
        stream = vf.get_record_stream(request_keys, 'unordered', False)
717
 
        keys = [r.key for r in stream]
718
 
        # We want to get the keys back in disk order, but it doesn't matter
719
 
        # which pack we read from first. So this can come back in 2 orders
720
 
        alt1 = [('f-id', 'rev-%d' % i) for i in [4, 3, 6, 5, 1, 2]]
721
 
        alt2 = [('f-id', 'rev-%d' % i) for i in [5, 1, 2, 4, 3, 6]]
722
 
        if keys != alt1 and keys != alt2:
723
 
            self.fail('Returned key order did not match either expected order.'
724
 
                      ' expected %s or %s, not %s'
725
 
                      % (alt1, alt2, keys))
726
 
 
727
365
 
728
366
class LowLevelKnitDataTests(TestCase):
729
367
 
734
372
        gz_file.close()
735
373
        return sio.getvalue()
736
374
 
737
 
    def make_multiple_records(self):
738
 
        """Create the content for multiple records."""
739
 
        sha1sum = osutils.sha_string('foo\nbar\n')
740
 
        total_txt = []
741
 
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
742
 
                                        'foo\n'
743
 
                                        'bar\n'
744
 
                                        'end rev-id-1\n'
745
 
                                        % (sha1sum,))
746
 
        record_1 = (0, len(gz_txt), sha1sum)
747
 
        total_txt.append(gz_txt)
748
 
        sha1sum = osutils.sha_string('baz\n')
749
 
        gz_txt = self.create_gz_content('version rev-id-2 1 %s\n'
750
 
                                        'baz\n'
751
 
                                        'end rev-id-2\n'
752
 
                                        % (sha1sum,))
753
 
        record_2 = (record_1[1], len(gz_txt), sha1sum)
754
 
        total_txt.append(gz_txt)
755
 
        return total_txt, record_1, record_2
756
 
 
757
375
    def test_valid_knit_data(self):
758
 
        sha1sum = osutils.sha_string('foo\nbar\n')
 
376
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
759
377
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
760
378
                                        'foo\n'
761
379
                                        'bar\n'
773
391
        raw_contents = list(knit._read_records_iter_raw(records))
774
392
        self.assertEqual([(('rev-id-1',), gz_txt, sha1sum)], raw_contents)
775
393
 
776
 
    def test_multiple_records_valid(self):
777
 
        total_txt, record_1, record_2 = self.make_multiple_records()
778
 
        transport = MockTransport([''.join(total_txt)])
779
 
        access = _KnitKeyAccess(transport, ConstantMapper('filename'))
780
 
        knit = KnitVersionedFiles(None, access)
781
 
        records = [(('rev-id-1',), (('rev-id-1',), record_1[0], record_1[1])),
782
 
                   (('rev-id-2',), (('rev-id-2',), record_2[0], record_2[1]))]
783
 
 
784
 
        contents = list(knit._read_records_iter(records))
785
 
        self.assertEqual([(('rev-id-1',), ['foo\n', 'bar\n'], record_1[2]),
786
 
                          (('rev-id-2',), ['baz\n'], record_2[2])],
787
 
                         contents)
788
 
 
789
 
        raw_contents = list(knit._read_records_iter_raw(records))
790
 
        self.assertEqual([(('rev-id-1',), total_txt[0], record_1[2]),
791
 
                          (('rev-id-2',), total_txt[1], record_2[2])],
792
 
                         raw_contents)
793
 
 
794
394
    def test_not_enough_lines(self):
795
 
        sha1sum = osutils.sha_string('foo\n')
 
395
        sha1sum = osutils.sha('foo\n').hexdigest()
796
396
        # record says 2 lines data says 1
797
397
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
798
398
                                        'foo\n'
810
410
        self.assertEqual([(('rev-id-1',),  gz_txt, sha1sum)], raw_contents)
811
411
 
812
412
    def test_too_many_lines(self):
813
 
        sha1sum = osutils.sha_string('foo\nbar\n')
 
413
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
814
414
        # record says 1 lines data says 2
815
415
        gz_txt = self.create_gz_content('version rev-id-1 1 %s\n'
816
416
                                        'foo\n'
829
429
        self.assertEqual([(('rev-id-1',), gz_txt, sha1sum)], raw_contents)
830
430
 
831
431
    def test_mismatched_version_id(self):
832
 
        sha1sum = osutils.sha_string('foo\nbar\n')
 
432
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
833
433
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
834
434
                                        'foo\n'
835
435
                                        'bar\n'
848
448
            knit._read_records_iter_raw(records))
849
449
 
850
450
    def test_uncompressed_data(self):
851
 
        sha1sum = osutils.sha_string('foo\nbar\n')
 
451
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
852
452
        txt = ('version rev-id-1 2 %s\n'
853
453
               'foo\n'
854
454
               'bar\n'
868
468
            knit._read_records_iter_raw(records))
869
469
 
870
470
    def test_corrupted_data(self):
871
 
        sha1sum = osutils.sha_string('foo\nbar\n')
 
471
        sha1sum = osutils.sha('foo\nbar\n').hexdigest()
872
472
        gz_txt = self.create_gz_content('version rev-id-1 2 %s\n'
873
473
                                        'foo\n'
874
474
                                        'bar\n'
891
491
 
892
492
    def get_knit_index(self, transport, name, mode):
893
493
        mapper = ConstantMapper(name)
 
494
        orig = knit._load_data
 
495
        def reset():
 
496
            knit._load_data = orig
 
497
        self.addCleanup(reset)
894
498
        from bzrlib._knit_load_data_py import _load_data_py
895
 
        self.overrideAttr(knit, '_load_data', _load_data_py)
 
499
        knit._load_data = _load_data_py
896
500
        allow_writes = lambda: 'w' in mode
897
501
        return _KndxIndex(transport, mapper, lambda:None, allow_writes, lambda:True)
898
502
 
1114
718
            call[1][1].getvalue())
1115
719
        self.assertEqual({'create_parent_dir': True}, call[2])
1116
720
 
1117
 
    def assertTotalBuildSize(self, size, keys, positions):
1118
 
        self.assertEqual(size,
1119
 
                         knit._get_total_build_size(None, keys, positions))
1120
 
 
1121
 
    def test__get_total_build_size(self):
1122
 
        positions = {
1123
 
            ('a',): (('fulltext', False), (('a',), 0, 100), None),
1124
 
            ('b',): (('line-delta', False), (('b',), 100, 21), ('a',)),
1125
 
            ('c',): (('line-delta', False), (('c',), 121, 35), ('b',)),
1126
 
            ('d',): (('line-delta', False), (('d',), 156, 12), ('b',)),
1127
 
            }
1128
 
        self.assertTotalBuildSize(100, [('a',)], positions)
1129
 
        self.assertTotalBuildSize(121, [('b',)], positions)
1130
 
        # c needs both a & b
1131
 
        self.assertTotalBuildSize(156, [('c',)], positions)
1132
 
        # we shouldn't count 'b' twice
1133
 
        self.assertTotalBuildSize(156, [('b',), ('c',)], positions)
1134
 
        self.assertTotalBuildSize(133, [('d',)], positions)
1135
 
        self.assertTotalBuildSize(168, [('c',), ('d',)], positions)
1136
 
 
1137
721
    def test_get_position(self):
1138
722
        transport = MockTransport([
1139
723
            _KndxIndex.HEADER,
1196
780
            self.assertRaises(errors.KnitCorrupt, index.keys)
1197
781
        except TypeError, e:
1198
782
            if (str(e) == ('exceptions must be strings, classes, or instances,'
1199
 
                           ' not exceptions.IndexError')):
 
783
                           ' not exceptions.IndexError')
 
784
                and sys.version_info[0:2] >= (2,5)):
1200
785
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
1201
786
                                  ' raising new style exceptions with python'
1202
787
                                  ' >=2.5')
1215
800
            self.assertRaises(errors.KnitCorrupt, index.keys)
1216
801
        except TypeError, e:
1217
802
            if (str(e) == ('exceptions must be strings, classes, or instances,'
1218
 
                           ' not exceptions.ValueError')):
 
803
                           ' not exceptions.ValueError')
 
804
                and sys.version_info[0:2] >= (2,5)):
1219
805
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
1220
806
                                  ' raising new style exceptions with python'
1221
807
                                  ' >=2.5')
1234
820
            self.assertRaises(errors.KnitCorrupt, index.keys)
1235
821
        except TypeError, e:
1236
822
            if (str(e) == ('exceptions must be strings, classes, or instances,'
1237
 
                           ' not exceptions.ValueError')):
 
823
                           ' not exceptions.ValueError')
 
824
                and sys.version_info[0:2] >= (2,5)):
1238
825
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
1239
826
                                  ' raising new style exceptions with python'
1240
827
                                  ' >=2.5')
1251
838
            self.assertRaises(errors.KnitCorrupt, index.keys)
1252
839
        except TypeError, e:
1253
840
            if (str(e) == ('exceptions must be strings, classes, or instances,'
1254
 
                           ' not exceptions.ValueError')):
 
841
                           ' not exceptions.ValueError')
 
842
                and sys.version_info[0:2] >= (2,5)):
1255
843
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
1256
844
                                  ' raising new style exceptions with python'
1257
845
                                  ' >=2.5')
1268
856
            self.assertRaises(errors.KnitCorrupt, index.keys)
1269
857
        except TypeError, e:
1270
858
            if (str(e) == ('exceptions must be strings, classes, or instances,'
1271
 
                           ' not exceptions.ValueError')):
 
859
                           ' not exceptions.ValueError')
 
860
                and sys.version_info[0:2] >= (2,5)):
1272
861
                self.knownFailure('Pyrex <0.9.5 fails with TypeError when'
1273
862
                                  ' raising new style exceptions with python'
1274
863
                                  ' >=2.5')
1275
864
            else:
1276
865
                raise
1277
866
 
1278
 
    def test_scan_unvalidated_index_not_implemented(self):
1279
 
        transport = MockTransport()
1280
 
        index = self.get_knit_index(transport, 'filename', 'r')
1281
 
        self.assertRaises(
1282
 
            NotImplementedError, index.scan_unvalidated_index,
1283
 
            'dummy graph_index')
1284
 
        self.assertRaises(
1285
 
            NotImplementedError, index.get_missing_compression_parents)
1286
 
 
1287
867
    def test_short_line(self):
1288
868
        transport = MockTransport([
1289
869
            _KndxIndex.HEADER,
1318
898
 
1319
899
class LowLevelKnitIndexTests_c(LowLevelKnitIndexTests):
1320
900
 
1321
 
    _test_needs_features = [compiled_knit_feature]
 
901
    _test_needs_features = [CompiledKnitFeature]
1322
902
 
1323
903
    def get_knit_index(self, transport, name, mode):
1324
904
        mapper = ConstantMapper(name)
1325
 
        from bzrlib._knit_load_data_pyx import _load_data_c
1326
 
        self.overrideAttr(knit, '_load_data', _load_data_c)
 
905
        orig = knit._load_data
 
906
        def reset():
 
907
            knit._load_data = orig
 
908
        self.addCleanup(reset)
 
909
        from bzrlib._knit_load_data_c import _load_data_c
 
910
        knit._load_data = _load_data_c
1327
911
        allow_writes = lambda: mode == 'w'
1328
 
        return _KndxIndex(transport, mapper, lambda:None,
1329
 
                          allow_writes, lambda:True)
1330
 
 
1331
 
 
1332
 
class Test_KnitAnnotator(TestCaseWithMemoryTransport):
1333
 
 
1334
 
    def make_annotator(self):
1335
 
        factory = knit.make_pack_factory(True, True, 1)
1336
 
        vf = factory(self.get_transport())
1337
 
        return knit._KnitAnnotator(vf)
1338
 
 
1339
 
    def test__expand_fulltext(self):
1340
 
        ann = self.make_annotator()
1341
 
        rev_key = ('rev-id',)
1342
 
        ann._num_compression_children[rev_key] = 1
1343
 
        res = ann._expand_record(rev_key, (('parent-id',),), None,
1344
 
                           ['line1\n', 'line2\n'], ('fulltext', True))
1345
 
        # The content object and text lines should be cached appropriately
1346
 
        self.assertEqual(['line1\n', 'line2'], res)
1347
 
        content_obj = ann._content_objects[rev_key]
1348
 
        self.assertEqual(['line1\n', 'line2\n'], content_obj._lines)
1349
 
        self.assertEqual(res, content_obj.text())
1350
 
        self.assertEqual(res, ann._text_cache[rev_key])
1351
 
 
1352
 
    def test__expand_delta_comp_parent_not_available(self):
1353
 
        # Parent isn't available yet, so we return nothing, but queue up this
1354
 
        # node for later processing
1355
 
        ann = self.make_annotator()
1356
 
        rev_key = ('rev-id',)
1357
 
        parent_key = ('parent-id',)
1358
 
        record = ['0,1,1\n', 'new-line\n']
1359
 
        details = ('line-delta', False)
1360
 
        res = ann._expand_record(rev_key, (parent_key,), parent_key,
1361
 
                                 record, details)
1362
 
        self.assertEqual(None, res)
1363
 
        self.assertTrue(parent_key in ann._pending_deltas)
1364
 
        pending = ann._pending_deltas[parent_key]
1365
 
        self.assertEqual(1, len(pending))
1366
 
        self.assertEqual((rev_key, (parent_key,), record, details), pending[0])
1367
 
 
1368
 
    def test__expand_record_tracks_num_children(self):
1369
 
        ann = self.make_annotator()
1370
 
        rev_key = ('rev-id',)
1371
 
        rev2_key = ('rev2-id',)
1372
 
        parent_key = ('parent-id',)
1373
 
        record = ['0,1,1\n', 'new-line\n']
1374
 
        details = ('line-delta', False)
1375
 
        ann._num_compression_children[parent_key] = 2
1376
 
        ann._expand_record(parent_key, (), None, ['line1\n', 'line2\n'],
1377
 
                           ('fulltext', False))
1378
 
        res = ann._expand_record(rev_key, (parent_key,), parent_key,
1379
 
                                 record, details)
1380
 
        self.assertEqual({parent_key: 1}, ann._num_compression_children)
1381
 
        # Expanding the second child should remove the content object, and the
1382
 
        # num_compression_children entry
1383
 
        res = ann._expand_record(rev2_key, (parent_key,), parent_key,
1384
 
                                 record, details)
1385
 
        self.assertFalse(parent_key in ann._content_objects)
1386
 
        self.assertEqual({}, ann._num_compression_children)
1387
 
        # We should not cache the content_objects for rev2 and rev, because
1388
 
        # they do not have compression children of their own.
1389
 
        self.assertEqual({}, ann._content_objects)
1390
 
 
1391
 
    def test__expand_delta_records_blocks(self):
1392
 
        ann = self.make_annotator()
1393
 
        rev_key = ('rev-id',)
1394
 
        parent_key = ('parent-id',)
1395
 
        record = ['0,1,1\n', 'new-line\n']
1396
 
        details = ('line-delta', True)
1397
 
        ann._num_compression_children[parent_key] = 2
1398
 
        ann._expand_record(parent_key, (), None,
1399
 
                           ['line1\n', 'line2\n', 'line3\n'],
1400
 
                           ('fulltext', False))
1401
 
        ann._expand_record(rev_key, (parent_key,), parent_key, record, details)
1402
 
        self.assertEqual({(rev_key, parent_key): [(1, 1, 1), (3, 3, 0)]},
1403
 
                         ann._matching_blocks)
1404
 
        rev2_key = ('rev2-id',)
1405
 
        record = ['0,1,1\n', 'new-line\n']
1406
 
        details = ('line-delta', False)
1407
 
        ann._expand_record(rev2_key, (parent_key,), parent_key, record, details)
1408
 
        self.assertEqual([(1, 1, 2), (3, 3, 0)],
1409
 
                         ann._matching_blocks[(rev2_key, parent_key)])
1410
 
 
1411
 
    def test__get_parent_ann_uses_matching_blocks(self):
1412
 
        ann = self.make_annotator()
1413
 
        rev_key = ('rev-id',)
1414
 
        parent_key = ('parent-id',)
1415
 
        parent_ann = [(parent_key,)]*3
1416
 
        block_key = (rev_key, parent_key)
1417
 
        ann._annotations_cache[parent_key] = parent_ann
1418
 
        ann._matching_blocks[block_key] = [(0, 1, 1), (3, 3, 0)]
1419
 
        # We should not try to access any parent_lines content, because we know
1420
 
        # we already have the matching blocks
1421
 
        par_ann, blocks = ann._get_parent_annotations_and_matches(rev_key,
1422
 
                                        ['1\n', '2\n', '3\n'], parent_key)
1423
 
        self.assertEqual(parent_ann, par_ann)
1424
 
        self.assertEqual([(0, 1, 1), (3, 3, 0)], blocks)
1425
 
        self.assertEqual({}, ann._matching_blocks)
1426
 
 
1427
 
    def test__process_pending(self):
1428
 
        ann = self.make_annotator()
1429
 
        rev_key = ('rev-id',)
1430
 
        p1_key = ('p1-id',)
1431
 
        p2_key = ('p2-id',)
1432
 
        record = ['0,1,1\n', 'new-line\n']
1433
 
        details = ('line-delta', False)
1434
 
        p1_record = ['line1\n', 'line2\n']
1435
 
        ann._num_compression_children[p1_key] = 1
1436
 
        res = ann._expand_record(rev_key, (p1_key,p2_key), p1_key,
1437
 
                                 record, details)
1438
 
        self.assertEqual(None, res)
1439
 
        # self.assertTrue(p1_key in ann._pending_deltas)
1440
 
        self.assertEqual({}, ann._pending_annotation)
1441
 
        # Now insert p1, and we should be able to expand the delta
1442
 
        res = ann._expand_record(p1_key, (), None, p1_record,
1443
 
                                 ('fulltext', False))
1444
 
        self.assertEqual(p1_record, res)
1445
 
        ann._annotations_cache[p1_key] = [(p1_key,)]*2
1446
 
        res = ann._process_pending(p1_key)
1447
 
        self.assertEqual([], res)
1448
 
        self.assertFalse(p1_key in ann._pending_deltas)
1449
 
        self.assertTrue(p2_key in ann._pending_annotation)
1450
 
        self.assertEqual({p2_key: [(rev_key, (p1_key, p2_key))]},
1451
 
                         ann._pending_annotation)
1452
 
        # Now fill in parent 2, and pending annotation should be satisfied
1453
 
        res = ann._expand_record(p2_key, (), None, [], ('fulltext', False))
1454
 
        ann._annotations_cache[p2_key] = []
1455
 
        res = ann._process_pending(p2_key)
1456
 
        self.assertEqual([rev_key], res)
1457
 
        self.assertEqual({}, ann._pending_annotation)
1458
 
        self.assertEqual({}, ann._pending_deltas)
1459
 
 
1460
 
    def test_record_delta_removes_basis(self):
1461
 
        ann = self.make_annotator()
1462
 
        ann._expand_record(('parent-id',), (), None,
1463
 
                           ['line1\n', 'line2\n'], ('fulltext', False))
1464
 
        ann._num_compression_children['parent-id'] = 2
1465
 
 
1466
 
    def test_annotate_special_text(self):
1467
 
        ann = self.make_annotator()
1468
 
        vf = ann._vf
1469
 
        rev1_key = ('rev-1',)
1470
 
        rev2_key = ('rev-2',)
1471
 
        rev3_key = ('rev-3',)
1472
 
        spec_key = ('special:',)
1473
 
        vf.add_lines(rev1_key, [], ['initial content\n'])
1474
 
        vf.add_lines(rev2_key, [rev1_key], ['initial content\n',
1475
 
                                            'common content\n',
1476
 
                                            'content in 2\n'])
1477
 
        vf.add_lines(rev3_key, [rev1_key], ['initial content\n',
1478
 
                                            'common content\n',
1479
 
                                            'content in 3\n'])
1480
 
        spec_text = ('initial content\n'
1481
 
                     'common content\n'
1482
 
                     'content in 2\n'
1483
 
                     'content in 3\n')
1484
 
        ann.add_special_text(spec_key, [rev2_key, rev3_key], spec_text)
1485
 
        anns, lines = ann.annotate(spec_key)
1486
 
        self.assertEqual([(rev1_key,),
1487
 
                          (rev2_key, rev3_key),
1488
 
                          (rev2_key,),
1489
 
                          (rev3_key,),
1490
 
                         ], anns)
1491
 
        self.assertEqualDiff(spec_text, ''.join(lines))
 
912
        return _KndxIndex(transport, mapper, lambda:None, allow_writes, lambda:True)
1492
913
 
1493
914
 
1494
915
class KnitTests(TestCaseWithTransport):
1499
920
        return make_file_factory(annotate, mapper)(self.get_transport())
1500
921
 
1501
922
 
1502
 
class TestBadShaError(KnitTests):
1503
 
    """Tests for handling of sha errors."""
1504
 
 
1505
 
    def test_sha_exception_has_text(self):
1506
 
        # having the failed text included in the error allows for recovery.
1507
 
        source = self.make_test_knit()
1508
 
        target = self.make_test_knit(name="target")
1509
 
        if not source._max_delta_chain:
1510
 
            raise TestNotApplicable(
1511
 
                "cannot get delta-caused sha failures without deltas.")
1512
 
        # create a basis
1513
 
        basis = ('basis',)
1514
 
        broken = ('broken',)
1515
 
        source.add_lines(basis, (), ['foo\n'])
1516
 
        source.add_lines(broken, (basis,), ['foo\n', 'bar\n'])
1517
 
        # Seed target with a bad basis text
1518
 
        target.add_lines(basis, (), ['gam\n'])
1519
 
        target.insert_record_stream(
1520
 
            source.get_record_stream([broken], 'unordered', False))
1521
 
        err = self.assertRaises(errors.KnitCorrupt,
1522
 
            target.get_record_stream([broken], 'unordered', True
1523
 
            ).next().get_bytes_as, 'chunked')
1524
 
        self.assertEqual(['gam\n', 'bar\n'], err.content)
1525
 
        # Test for formatting with live data
1526
 
        self.assertStartsWith(str(err), "Knit ")
1527
 
 
1528
 
 
1529
923
class TestKnitIndex(KnitTests):
1530
924
 
1531
925
    def test_add_versions_dictionary_compresses(self):
1603
997
        # could leave an empty .kndx file, which bzr would later claim was a
1604
998
        # corrupted file since the header was not present. In reality, the file
1605
999
        # just wasn't created, so it should be ignored.
1606
 
        t = transport.get_transport('.')
 
1000
        t = get_transport('.')
1607
1001
        t.put_bytes('test.kndx', '')
1608
1002
 
1609
1003
        knit = self.make_test_knit()
1610
1004
 
1611
1005
    def test_knit_index_checks_header(self):
1612
 
        t = transport.get_transport('.')
 
1006
        t = get_transport('.')
1613
1007
        t.put_bytes('test.kndx', '# not really a knit header\n\n')
1614
1008
        k = self.make_test_knit()
1615
1009
        self.assertRaises(KnitHeaderError, k.keys)
1731
1125
            [('parent',)])])
1732
1126
        # but neither should have added data:
1733
1127
        self.assertEqual([[], [], [], []], self.caught_entries)
1734
 
 
 
1128
        
1735
1129
    def test_add_version_different_dup(self):
1736
1130
        index = self.two_graph_index(deltas=True, catch_adds=True)
1737
1131
        # change options
1738
1132
        self.assertRaises(errors.KnitCorrupt, index.add_records,
1739
 
            [(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
 
1133
            [(('tip',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
 
1134
        self.assertRaises(errors.KnitCorrupt, index.add_records,
 
1135
            [(('tip',), 'line-delta,no-eol', (None, 0, 100), [('parent',)])])
1740
1136
        self.assertRaises(errors.KnitCorrupt, index.add_records,
1741
1137
            [(('tip',), 'fulltext', (None, 0, 100), [('parent',)])])
1742
1138
        # parents
1743
1139
        self.assertRaises(errors.KnitCorrupt, index.add_records,
1744
1140
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [])])
1745
1141
        self.assertEqual([], self.caught_entries)
1746
 
 
 
1142
        
1747
1143
    def test_add_versions_nodeltas(self):
1748
1144
        index = self.two_graph_index(catch_adds=True)
1749
1145
        index.add_records([
1791
1187
            [('parent',)])])
1792
1188
        # but neither should have added data.
1793
1189
        self.assertEqual([[], [], [], []], self.caught_entries)
1794
 
 
 
1190
        
1795
1191
    def test_add_versions_different_dup(self):
1796
1192
        index = self.two_graph_index(deltas=True, catch_adds=True)
1797
1193
        # change options
1798
1194
        self.assertRaises(errors.KnitCorrupt, index.add_records,
1799
 
            [(('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
 
1195
            [(('tip',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
 
1196
        self.assertRaises(errors.KnitCorrupt, index.add_records,
 
1197
            [(('tip',), 'line-delta,no-eol', (None, 0, 100), [('parent',)])])
1800
1198
        self.assertRaises(errors.KnitCorrupt, index.add_records,
1801
1199
            [(('tip',), 'fulltext', (None, 0, 100), [('parent',)])])
1802
1200
        # parents
1805
1203
        # change options in the second record
1806
1204
        self.assertRaises(errors.KnitCorrupt, index.add_records,
1807
1205
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)]),
1808
 
             (('tip',), 'line-delta', (None, 0, 100), [('parent',)])])
 
1206
             (('tip',), 'no-eol,line-delta', (None, 0, 100), [('parent',)])])
1809
1207
        self.assertEqual([], self.caught_entries)
1810
1208
 
1811
 
    def make_g_index_missing_compression_parent(self):
1812
 
        graph_index = self.make_g_index('missing_comp', 2,
1813
 
            [(('tip', ), ' 100 78',
1814
 
              ([('missing-parent', ), ('ghost', )], [('missing-parent', )]))])
1815
 
        return graph_index
1816
 
 
1817
 
    def make_g_index_missing_parent(self):
1818
 
        graph_index = self.make_g_index('missing_parent', 2,
1819
 
            [(('parent', ), ' 100 78', ([], [])),
1820
 
             (('tip', ), ' 100 78',
1821
 
              ([('parent', ), ('missing-parent', )], [('parent', )])),
1822
 
              ])
1823
 
        return graph_index
1824
 
 
1825
 
    def make_g_index_no_external_refs(self):
1826
 
        graph_index = self.make_g_index('no_external_refs', 2,
1827
 
            [(('rev', ), ' 100 78',
1828
 
              ([('parent', ), ('ghost', )], []))])
1829
 
        return graph_index
1830
 
 
1831
 
    def test_add_good_unvalidated_index(self):
1832
 
        unvalidated = self.make_g_index_no_external_refs()
1833
 
        combined = CombinedGraphIndex([unvalidated])
1834
 
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1835
 
        index.scan_unvalidated_index(unvalidated)
1836
 
        self.assertEqual(frozenset(), index.get_missing_compression_parents())
1837
 
 
1838
 
    def test_add_missing_compression_parent_unvalidated_index(self):
1839
 
        unvalidated = self.make_g_index_missing_compression_parent()
1840
 
        combined = CombinedGraphIndex([unvalidated])
1841
 
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1842
 
        index.scan_unvalidated_index(unvalidated)
1843
 
        # This also checks that its only the compression parent that is
1844
 
        # examined, otherwise 'ghost' would also be reported as a missing
1845
 
        # parent.
1846
 
        self.assertEqual(
1847
 
            frozenset([('missing-parent',)]),
1848
 
            index.get_missing_compression_parents())
1849
 
 
1850
 
    def test_add_missing_noncompression_parent_unvalidated_index(self):
1851
 
        unvalidated = self.make_g_index_missing_parent()
1852
 
        combined = CombinedGraphIndex([unvalidated])
1853
 
        index = _KnitGraphIndex(combined, lambda: True, deltas=True,
1854
 
            track_external_parent_refs=True)
1855
 
        index.scan_unvalidated_index(unvalidated)
1856
 
        self.assertEqual(
1857
 
            frozenset([('missing-parent',)]), index.get_missing_parents())
1858
 
 
1859
 
    def test_track_external_parent_refs(self):
1860
 
        g_index = self.make_g_index('empty', 2, [])
1861
 
        combined = CombinedGraphIndex([g_index])
1862
 
        index = _KnitGraphIndex(combined, lambda: True, deltas=True,
1863
 
            add_callback=self.catch_add, track_external_parent_refs=True)
1864
 
        self.caught_entries = []
1865
 
        index.add_records([
1866
 
            (('new-key',), 'fulltext,no-eol', (None, 50, 60),
1867
 
             [('parent-1',), ('parent-2',)])])
1868
 
        self.assertEqual(
1869
 
            frozenset([('parent-1',), ('parent-2',)]),
1870
 
            index.get_missing_parents())
1871
 
 
1872
 
    def test_add_unvalidated_index_with_present_external_references(self):
1873
 
        index = self.two_graph_index(deltas=True)
1874
 
        # Ugly hack to get at one of the underlying GraphIndex objects that
1875
 
        # two_graph_index built.
1876
 
        unvalidated = index._graph_index._indices[1]
1877
 
        # 'parent' is an external ref of _indices[1] (unvalidated), but is
1878
 
        # present in _indices[0].
1879
 
        index.scan_unvalidated_index(unvalidated)
1880
 
        self.assertEqual(frozenset(), index.get_missing_compression_parents())
1881
 
 
1882
 
    def make_new_missing_parent_g_index(self, name):
1883
 
        missing_parent = name + '-missing-parent'
1884
 
        graph_index = self.make_g_index(name, 2,
1885
 
            [((name + 'tip', ), ' 100 78',
1886
 
              ([(missing_parent, ), ('ghost', )], [(missing_parent, )]))])
1887
 
        return graph_index
1888
 
 
1889
 
    def test_add_mulitiple_unvalidated_indices_with_missing_parents(self):
1890
 
        g_index_1 = self.make_new_missing_parent_g_index('one')
1891
 
        g_index_2 = self.make_new_missing_parent_g_index('two')
1892
 
        combined = CombinedGraphIndex([g_index_1, g_index_2])
1893
 
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1894
 
        index.scan_unvalidated_index(g_index_1)
1895
 
        index.scan_unvalidated_index(g_index_2)
1896
 
        self.assertEqual(
1897
 
            frozenset([('one-missing-parent',), ('two-missing-parent',)]),
1898
 
            index.get_missing_compression_parents())
1899
 
 
1900
 
    def test_add_mulitiple_unvalidated_indices_with_mutual_dependencies(self):
1901
 
        graph_index_a = self.make_g_index('one', 2,
1902
 
            [(('parent-one', ), ' 100 78', ([('non-compression-parent',)], [])),
1903
 
             (('child-of-two', ), ' 100 78',
1904
 
              ([('parent-two',)], [('parent-two',)]))])
1905
 
        graph_index_b = self.make_g_index('two', 2,
1906
 
            [(('parent-two', ), ' 100 78', ([('non-compression-parent',)], [])),
1907
 
             (('child-of-one', ), ' 100 78',
1908
 
              ([('parent-one',)], [('parent-one',)]))])
1909
 
        combined = CombinedGraphIndex([graph_index_a, graph_index_b])
1910
 
        index = _KnitGraphIndex(combined, lambda: True, deltas=True)
1911
 
        index.scan_unvalidated_index(graph_index_a)
1912
 
        index.scan_unvalidated_index(graph_index_b)
1913
 
        self.assertEqual(
1914
 
            frozenset([]), index.get_missing_compression_parents())
1915
 
 
1916
1209
 
1917
1210
class TestNoParentsGraphIndexKnit(KnitTests):
1918
1211
    """Tests for knits using _KnitGraphIndex with no parents."""
1926
1219
        size = trans.put_file(name, stream)
1927
1220
        return GraphIndex(trans, name, size)
1928
1221
 
1929
 
    def test_add_good_unvalidated_index(self):
1930
 
        unvalidated = self.make_g_index('unvalidated')
1931
 
        combined = CombinedGraphIndex([unvalidated])
1932
 
        index = _KnitGraphIndex(combined, lambda: True, parents=False)
1933
 
        index.scan_unvalidated_index(unvalidated)
1934
 
        self.assertEqual(frozenset(),
1935
 
            index.get_missing_compression_parents())
1936
 
 
1937
1222
    def test_parents_deltas_incompatible(self):
1938
1223
        index = CombinedGraphIndex([])
1939
1224
        self.assertRaises(errors.KnitError, _KnitGraphIndex, lambda:True,
2020
1305
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000), [])])
2021
1306
        # but neither should have added data.
2022
1307
        self.assertEqual([[], [], [], []], self.caught_entries)
2023
 
 
 
1308
        
2024
1309
    def test_add_version_different_dup(self):
2025
1310
        index = self.two_graph_index(catch_adds=True)
2026
1311
        # change options
2034
1319
        self.assertRaises(errors.KnitCorrupt, index.add_records,
2035
1320
            [(('tip',), 'fulltext,no-eol', (None, 0, 100), [('parent',)])])
2036
1321
        self.assertEqual([], self.caught_entries)
2037
 
 
 
1322
        
2038
1323
    def test_add_versions(self):
2039
1324
        index = self.two_graph_index(catch_adds=True)
2040
1325
        index.add_records([
2072
1357
        index.add_records([(('tip',), 'fulltext,no-eol', (None, 0, 1000), [])])
2073
1358
        # but neither should have added data.
2074
1359
        self.assertEqual([[], [], [], []], self.caught_entries)
2075
 
 
 
1360
        
2076
1361
    def test_add_versions_different_dup(self):
2077
1362
        index = self.two_graph_index(catch_adds=True)
2078
1363
        # change options
2092
1377
        self.assertEqual([], self.caught_entries)
2093
1378
 
2094
1379
 
2095
 
class TestKnitVersionedFiles(KnitTests):
2096
 
 
2097
 
    def assertGroupKeysForIo(self, exp_groups, keys, non_local_keys,
2098
 
                             positions, _min_buffer_size=None):
2099
 
        kvf = self.make_test_knit()
2100
 
        if _min_buffer_size is None:
2101
 
            _min_buffer_size = knit._STREAM_MIN_BUFFER_SIZE
2102
 
        self.assertEqual(exp_groups, kvf._group_keys_for_io(keys,
2103
 
                                        non_local_keys, positions,
2104
 
                                        _min_buffer_size=_min_buffer_size))
2105
 
 
2106
 
    def assertSplitByPrefix(self, expected_map, expected_prefix_order,
2107
 
                            keys):
2108
 
        split, prefix_order = KnitVersionedFiles._split_by_prefix(keys)
2109
 
        self.assertEqual(expected_map, split)
2110
 
        self.assertEqual(expected_prefix_order, prefix_order)
2111
 
 
2112
 
    def test__group_keys_for_io(self):
2113
 
        ft_detail = ('fulltext', False)
2114
 
        ld_detail = ('line-delta', False)
2115
 
        f_a = ('f', 'a')
2116
 
        f_b = ('f', 'b')
2117
 
        f_c = ('f', 'c')
2118
 
        g_a = ('g', 'a')
2119
 
        g_b = ('g', 'b')
2120
 
        g_c = ('g', 'c')
2121
 
        positions = {
2122
 
            f_a: (ft_detail, (f_a, 0, 100), None),
2123
 
            f_b: (ld_detail, (f_b, 100, 21), f_a),
2124
 
            f_c: (ld_detail, (f_c, 180, 15), f_b),
2125
 
            g_a: (ft_detail, (g_a, 121, 35), None),
2126
 
            g_b: (ld_detail, (g_b, 156, 12), g_a),
2127
 
            g_c: (ld_detail, (g_c, 195, 13), g_a),
2128
 
            }
2129
 
        self.assertGroupKeysForIo([([f_a], set())],
2130
 
                                  [f_a], [], positions)
2131
 
        self.assertGroupKeysForIo([([f_a], set([f_a]))],
2132
 
                                  [f_a], [f_a], positions)
2133
 
        self.assertGroupKeysForIo([([f_a, f_b], set([]))],
2134
 
                                  [f_a, f_b], [], positions)
2135
 
        self.assertGroupKeysForIo([([f_a, f_b], set([f_b]))],
2136
 
                                  [f_a, f_b], [f_b], positions)
2137
 
        self.assertGroupKeysForIo([([f_a, f_b, g_a, g_b], set())],
2138
 
                                  [f_a, g_a, f_b, g_b], [], positions)
2139
 
        self.assertGroupKeysForIo([([f_a, f_b, g_a, g_b], set())],
2140
 
                                  [f_a, g_a, f_b, g_b], [], positions,
2141
 
                                  _min_buffer_size=150)
2142
 
        self.assertGroupKeysForIo([([f_a, f_b], set()), ([g_a, g_b], set())],
2143
 
                                  [f_a, g_a, f_b, g_b], [], positions,
2144
 
                                  _min_buffer_size=100)
2145
 
        self.assertGroupKeysForIo([([f_c], set()), ([g_b], set())],
2146
 
                                  [f_c, g_b], [], positions,
2147
 
                                  _min_buffer_size=125)
2148
 
        self.assertGroupKeysForIo([([g_b, f_c], set())],
2149
 
                                  [g_b, f_c], [], positions,
2150
 
                                  _min_buffer_size=125)
2151
 
 
2152
 
    def test__split_by_prefix(self):
2153
 
        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2154
 
                                  'g': [('g', 'b'), ('g', 'a')],
2155
 
                                 }, ['f', 'g'],
2156
 
                                 [('f', 'a'), ('g', 'b'),
2157
 
                                  ('g', 'a'), ('f', 'b')])
2158
 
 
2159
 
        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2160
 
                                  'g': [('g', 'b'), ('g', 'a')],
2161
 
                                 }, ['f', 'g'],
2162
 
                                 [('f', 'a'), ('f', 'b'),
2163
 
                                  ('g', 'b'), ('g', 'a')])
2164
 
 
2165
 
        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2166
 
                                  'g': [('g', 'b'), ('g', 'a')],
2167
 
                                 }, ['f', 'g'],
2168
 
                                 [('f', 'a'), ('f', 'b'),
2169
 
                                  ('g', 'b'), ('g', 'a')])
2170
 
 
2171
 
        self.assertSplitByPrefix({'f': [('f', 'a'), ('f', 'b')],
2172
 
                                  'g': [('g', 'b'), ('g', 'a')],
2173
 
                                  '': [('a',), ('b',)]
2174
 
                                 }, ['f', 'g', ''],
2175
 
                                 [('f', 'a'), ('g', 'b'),
2176
 
                                  ('a',), ('b',),
2177
 
                                  ('g', 'a'), ('f', 'b')])
2178
 
 
2179
 
 
2180
1380
class TestStacking(KnitTests):
2181
1381
 
2182
1382
    def get_basis_and_test_knit(self):
2208
1408
        basis.calls = []
2209
1409
        test.add_lines(key_cross_border, (key_basis,), ['foo\n'])
2210
1410
        self.assertEqual('fulltext', test._index.get_method(key_cross_border))
2211
 
        # we don't even need to look at the basis to see that this should be
2212
 
        # stored as a fulltext
2213
 
        self.assertEqual([], basis.calls)
 
1411
        self.assertEqual([("get_parent_map", set([key_basis]))], basis.calls)
2214
1412
        # Subsequent adds do delta.
2215
1413
        basis.calls = []
2216
1414
        test.add_lines(key_delta, (key_cross_border,), ['foo\n'])
2237
1435
        # self.assertEqual([("annotate", key_basis)], basis.calls)
2238
1436
        self.assertEqual([('get_parent_map', set([key_basis])),
2239
1437
            ('get_parent_map', set([key_basis])),
2240
 
            ('get_record_stream', [key_basis], 'topological', True)],
 
1438
            ('get_parent_map', set([key_basis])),
 
1439
            ('get_record_stream', [key_basis], 'unordered', True)],
2241
1440
            basis.calls)
2242
1441
 
2243
1442
    def test_check(self):
2244
1443
        # At the moment checking a stacked knit does implicitly check the
2245
 
        # fallback files.
 
1444
        # fallback files.  
2246
1445
        basis, test = self.get_basis_and_test_knit()
2247
1446
        test.check()
2248
1447
 
2340
1539
                True).next()
2341
1540
            self.assertEqual(record.key, result[0])
2342
1541
            self.assertEqual(record.sha1, result[1])
2343
 
            # We used to check that the storage kind matched, but actually it
2344
 
            # depends on whether it was sourced from the basis, or in a single
2345
 
            # group, because asking for full texts returns proxy objects to a
2346
 
            # _ContentMapGenerator object; so checking the kind is unneeded.
 
1542
            self.assertEqual(record.storage_kind, result[2])
2347
1543
            self.assertEqual(record.get_bytes_as('fulltext'), result[3])
2348
1544
        # It's not strictly minimal, but it seems reasonable for now for it to
2349
1545
        # ask which fallbacks have which parents.
2350
1546
        self.assertEqual([
2351
1547
            ("get_parent_map", set([key_basis, key_basis_2, key_missing])),
2352
 
            # topological is requested from the fallback, because that is what
2353
 
            # was requested at the top level.
2354
 
            ("get_record_stream", [key_basis_2, key_basis], 'topological', True)],
 
1548
            # unordered is asked for by the underlying worker as it still
 
1549
            # buffers everything while answering - which is a problem!
 
1550
            ("get_record_stream", [key_basis_2, key_basis], 'unordered', True)],
2355
1551
            calls)
2356
1552
 
2357
1553
    def test_get_record_stream_unordered_deltas(self):
2443
1639
        key_basis = ('bar',)
2444
1640
        key_missing = ('missing',)
2445
1641
        test.add_lines(key, (), ['foo\n'])
2446
 
        key_sha1sum = osutils.sha_string('foo\n')
 
1642
        key_sha1sum = osutils.sha('foo\n').hexdigest()
2447
1643
        sha1s = test.get_sha1s([key])
2448
1644
        self.assertEqual({key: key_sha1sum}, sha1s)
2449
1645
        self.assertEqual([], basis.calls)
2451
1647
        # directly (rather than via text reconstruction) so that remote servers
2452
1648
        # etc don't have to answer with full content.
2453
1649
        basis.add_lines(key_basis, (), ['foo\n', 'bar\n'])
2454
 
        basis_sha1sum = osutils.sha_string('foo\nbar\n')
 
1650
        basis_sha1sum = osutils.sha('foo\nbar\n').hexdigest()
2455
1651
        basis.calls = []
2456
1652
        sha1s = test.get_sha1s([key, key_missing, key_basis])
2457
1653
        self.assertEqual({key: key_sha1sum,
2473
1669
        source.add_lines(key_delta, (key_basis,), ['bar\n'])
2474
1670
        stream = source.get_record_stream([key_delta], 'unordered', False)
2475
1671
        test.insert_record_stream(stream)
2476
 
        # XXX: this does somewhat too many calls in making sure of whether it
2477
 
        # has to recreate the full text.
2478
 
        self.assertEqual([("get_parent_map", set([key_basis])),
2479
 
             ('get_parent_map', set([key_basis])),
2480
 
             ('get_record_stream', [key_basis], 'unordered', True)],
 
1672
        self.assertEqual([("get_parent_map", set([key_basis]))],
2481
1673
            basis.calls)
2482
1674
        self.assertEqual({key_delta:(key_basis,)},
2483
1675
            test.get_parent_map([key_delta]))
2486
1678
 
2487
1679
    def test_iter_lines_added_or_present_in_keys(self):
2488
1680
        # Lines from the basis are returned, and lines for a given key are only
2489
 
        # returned once.
 
1681
        # returned once. 
2490
1682
        key1 = ('foo1',)
2491
1683
        key2 = ('foo2',)
2492
1684
        # all sources are asked for keys:
2544
1736
        test.add_mpdiffs([(key_delta, (key_basis,),
2545
1737
            source.get_sha1s([key_delta])[key_delta], diffs[0])])
2546
1738
        self.assertEqual([("get_parent_map", set([key_basis])),
2547
 
            ('get_record_stream', [key_basis], 'unordered', True),],
 
1739
            ('get_record_stream', [key_basis], 'unordered', True),
 
1740
            ('get_parent_map', set([key_basis]))],
2548
1741
            basis.calls)
2549
1742
        self.assertEqual({key_delta:(key_basis,)},
2550
1743
            test.get_parent_map([key_delta]))
2569
1762
                multiparent.NewText(['foo\n']),
2570
1763
                multiparent.ParentText(1, 0, 2, 1)])],
2571
1764
            diffs)
2572
 
        self.assertEqual(3, len(basis.calls))
 
1765
        self.assertEqual(4, len(basis.calls))
2573
1766
        self.assertEqual([
2574
1767
            ("get_parent_map", set([key_left, key_right])),
2575
1768
            ("get_parent_map", set([key_left, key_right])),
 
1769
            ("get_parent_map", set([key_left, key_right])),
2576
1770
            ],
2577
 
            basis.calls[:-1])
2578
 
        last_call = basis.calls[-1]
 
1771
            basis.calls[:3])
 
1772
        last_call = basis.calls[3]
2579
1773
        self.assertEqual('get_record_stream', last_call[0])
2580
1774
        self.assertEqual(set([key_left, key_right]), set(last_call[1]))
2581
 
        self.assertEqual('topological', last_call[2])
 
1775
        self.assertEqual('unordered', last_call[2])
2582
1776
        self.assertEqual(True, last_call[3])
2583
 
 
2584
 
 
2585
 
class TestNetworkBehaviour(KnitTests):
2586
 
    """Tests for getting data out of/into knits over the network."""
2587
 
 
2588
 
    def test_include_delta_closure_generates_a_knit_delta_closure(self):
2589
 
        vf = self.make_test_knit(name='test')
2590
 
        # put in three texts, giving ft, delta, delta
2591
 
        vf.add_lines(('base',), (), ['base\n', 'content\n'])
2592
 
        vf.add_lines(('d1',), (('base',),), ['d1\n'])
2593
 
        vf.add_lines(('d2',), (('d1',),), ['d2\n'])
2594
 
        # But heuristics could interfere, so check what happened:
2595
 
        self.assertEqual(['knit-ft-gz', 'knit-delta-gz', 'knit-delta-gz'],
2596
 
            [record.storage_kind for record in
2597
 
             vf.get_record_stream([('base',), ('d1',), ('d2',)],
2598
 
                'topological', False)])
2599
 
        # generate a stream of just the deltas include_delta_closure=True,
2600
 
        # serialise to the network, and check that we get a delta closure on the wire.
2601
 
        stream = vf.get_record_stream([('d1',), ('d2',)], 'topological', True)
2602
 
        netb = [record.get_bytes_as(record.storage_kind) for record in stream]
2603
 
        # The first bytes should be a memo from _ContentMapGenerator, and the
2604
 
        # second bytes should be empty (because its a API proxy not something
2605
 
        # for wire serialisation.
2606
 
        self.assertEqual('', netb[1])
2607
 
        bytes = netb[0]
2608
 
        kind, line_end = network_bytes_to_kind_and_offset(bytes)
2609
 
        self.assertEqual('knit-delta-closure', kind)
2610
 
 
2611
 
 
2612
 
class TestContentMapGenerator(KnitTests):
2613
 
    """Tests for ContentMapGenerator"""
2614
 
 
2615
 
    def test_get_record_stream_gives_records(self):
2616
 
        vf = self.make_test_knit(name='test')
2617
 
        # put in three texts, giving ft, delta, delta
2618
 
        vf.add_lines(('base',), (), ['base\n', 'content\n'])
2619
 
        vf.add_lines(('d1',), (('base',),), ['d1\n'])
2620
 
        vf.add_lines(('d2',), (('d1',),), ['d2\n'])
2621
 
        keys = [('d1',), ('d2',)]
2622
 
        generator = _VFContentMapGenerator(vf, keys,
2623
 
            global_map=vf.get_parent_map(keys))
2624
 
        for record in generator.get_record_stream():
2625
 
            if record.key == ('d1',):
2626
 
                self.assertEqual('d1\n', record.get_bytes_as('fulltext'))
2627
 
            else:
2628
 
                self.assertEqual('d2\n', record.get_bytes_as('fulltext'))
2629
 
 
2630
 
    def test_get_record_stream_kinds_are_raw(self):
2631
 
        vf = self.make_test_knit(name='test')
2632
 
        # put in three texts, giving ft, delta, delta
2633
 
        vf.add_lines(('base',), (), ['base\n', 'content\n'])
2634
 
        vf.add_lines(('d1',), (('base',),), ['d1\n'])
2635
 
        vf.add_lines(('d2',), (('d1',),), ['d2\n'])
2636
 
        keys = [('base',), ('d1',), ('d2',)]
2637
 
        generator = _VFContentMapGenerator(vf, keys,
2638
 
            global_map=vf.get_parent_map(keys))
2639
 
        kinds = {('base',): 'knit-delta-closure',
2640
 
            ('d1',): 'knit-delta-closure-ref',
2641
 
            ('d2',): 'knit-delta-closure-ref',
2642
 
            }
2643
 
        for record in generator.get_record_stream():
2644
 
            self.assertEqual(kinds[record.key], record.storage_kind)