← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_btree_index.py

  • Committer: Ian Clatworthy
  • Date: 2009-09-02 16:03:51 UTC
  • mto: (4634.39.1 pdf-chm-docs)
  • mto: This revision was merged to the branch mainline in revision 4689.
  • Revision ID: ian.clatworthy@canonical.com-20090902160351-sxptcz3ttc1aencw
first cut at pdf docs via sphinx

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/tests/test_btree_index.py
23
23
from bzrlib import (
24
24
    btree_index,
25
25
    errors,
26
 
    fifo_cache,
27
 
    lru_cache,
28
26
    osutils,
29
27
    tests,
30
28
    )
43
41
        condition_isinstance(TestBTreeNodes))
44
42
    import bzrlib._btree_serializer_py as py_module
45
43
    scenarios = [('python', {'parse_btree': py_module})]
46
 
    if compiled_btreeparser_feature.available():
47
 
        scenarios.append(('C', {'parse_btree':
48
 
                                compiled_btreeparser_feature.module}))
 
44
    if CompiledBtreeParserFeature.available():
 
45
        # Is there a way to do this that gets missing feature failures rather
 
46
        # than no indication to the user?
 
47
        import bzrlib._btree_serializer_pyx as c_module
 
48
        scenarios.append(('C', {'parse_btree': c_module}))
49
49
    return multiply_tests(node_tests, scenarios, others)
50
50
 
51
51
 
52
 
compiled_btreeparser_feature = tests.ModuleAvailableFeature(
53
 
                                'bzrlib._btree_serializer_pyx')
 
52
class _CompiledBtreeParserFeature(tests.Feature):
 
53
    def _probe(self):
 
54
        try:
 
55
            import bzrlib._btree_serializer_pyx
 
56
        except ImportError:
 
57
            return False
 
58
        return True
 
59
 
 
60
    def feature_name(self):
 
61
        return 'bzrlib._btree_serializer_pyx'
 
62
 
 
63
CompiledBtreeParserFeature = _CompiledBtreeParserFeature()
54
64
 
55
65
 
56
66
class BTreeTestCase(TestCaseWithTransport):
59
69
 
60
70
    def setUp(self):
61
71
        TestCaseWithTransport.setUp(self)
62
 
        self.overrideAttr(btree_index, '_RESERVED_HEADER_BYTES', 100)
 
72
        self._original_header = btree_index._RESERVED_HEADER_BYTES
 
73
        def restore():
 
74
            btree_index._RESERVED_HEADER_BYTES = self._original_header
 
75
        self.addCleanup(restore)
 
76
        btree_index._RESERVED_HEADER_BYTES = 100
63
77
 
64
78
    def make_nodes(self, count, key_elements, reference_lists):
65
79
        """Generate count*key_elements sample nodes."""
99
113
 
100
114
    def shrink_page_size(self):
101
115
        """Shrink the default page size so that less fits in a page."""
102
 
        self.overrideAttr(btree_index, '_PAGE_SIZE')
 
116
        old_page_size = btree_index._PAGE_SIZE
 
117
        def cleanup():
 
118
            btree_index._PAGE_SIZE = old_page_size
 
119
        self.addCleanup(cleanup)
103
120
        btree_index._PAGE_SIZE = 2048
104
121
 
105
122
 
106
123
class TestBTreeBuilder(BTreeTestCase):
107
124
 
108
 
    def test_clear_cache(self):
109
 
        builder = btree_index.BTreeBuilder(reference_lists=0, key_elements=1)
110
 
        # This is a no-op, but we need the api to be consistent with other
111
 
        # BTreeGraphIndex apis.
112
 
        builder.clear_cache()
113
 
 
114
125
    def test_empty_1_0(self):
115
126
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
116
127
        # NamedTemporaryFile dies on builder.finish().read(). weird.
142
153
        temp_file = builder.finish()
143
154
        content = temp_file.read()
144
155
        del temp_file
145
 
        self.assertEqual(131, len(content))
 
156
        self.assertEqual(158, len(content))
146
157
        self.assertEqual(
147
158
            "B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=5\n"
148
159
            "row_lengths=1\n",
166
177
        temp_file = builder.finish()
167
178
        content = temp_file.read()
168
179
        del temp_file
169
 
        self.assertEqual(238, len(content))
 
180
        self.assertEqual(264, len(content))
170
181
        self.assertEqual(
171
182
            "B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=10\n"
172
183
            "row_lengths=1\n",
232
243
        temp_file = builder.finish()
233
244
        content = temp_file.read()
234
245
        del temp_file
235
 
        self.assertEqual(155, len(content))
 
246
        self.assertEqual(181, len(content))
236
247
        self.assertEqual(
237
248
            "B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=10\n"
238
249
            "row_lengths=1\n",
340
351
        # Test the parts of the index that take up memory are doing so
341
352
        # predictably.
342
353
        self.assertEqual(1, len(builder._nodes))
 
354
        self.assertEqual(1, len(builder._keys))
343
355
        self.assertIs(None, builder._nodes_by_key)
344
356
        builder.add_node(*nodes[1])
345
357
        self.assertEqual(0, len(builder._nodes))
 
358
        self.assertEqual(0, len(builder._keys))
346
359
        self.assertIs(None, builder._nodes_by_key)
347
360
        self.assertEqual(1, len(builder._backing_indices))
348
361
        self.assertEqual(2, builder._backing_indices[0].key_count())
349
362
        # now back to memory
350
363
        builder.add_node(*nodes[2])
351
364
        self.assertEqual(1, len(builder._nodes))
 
365
        self.assertEqual(1, len(builder._keys))
352
366
        self.assertIs(None, builder._nodes_by_key)
353
367
        # And spills to a second backing index combing all
354
368
        builder.add_node(*nodes[3])
355
369
        self.assertEqual(0, len(builder._nodes))
 
370
        self.assertEqual(0, len(builder._keys))
356
371
        self.assertIs(None, builder._nodes_by_key)
357
372
        self.assertEqual(2, len(builder._backing_indices))
358
373
        self.assertEqual(None, builder._backing_indices[0])
361
376
        builder.add_node(*nodes[4])
362
377
        builder.add_node(*nodes[5])
363
378
        self.assertEqual(0, len(builder._nodes))
 
379
        self.assertEqual(0, len(builder._keys))
364
380
        self.assertIs(None, builder._nodes_by_key)
365
381
        self.assertEqual(2, len(builder._backing_indices))
366
382
        self.assertEqual(2, builder._backing_indices[0].key_count())
424
440
        # Test the parts of the index that take up memory are doing so
425
441
        # predictably.
426
442
        self.assertEqual(1, len(builder._nodes))
 
443
        self.assertEqual(1, len(builder._keys))
427
444
        self.assertIs(None, builder._nodes_by_key)
428
445
        builder.add_node(*nodes[1])
429
446
        self.assertEqual(0, len(builder._nodes))
 
447
        self.assertEqual(0, len(builder._keys))
430
448
        self.assertIs(None, builder._nodes_by_key)
431
449
        self.assertEqual(1, len(builder._backing_indices))
432
450
        self.assertEqual(2, builder._backing_indices[0].key_count())
433
451
        # now back to memory
434
452
        builder.add_node(*nodes[2])
435
453
        self.assertEqual(1, len(builder._nodes))
 
454
        self.assertEqual(1, len(builder._keys))
436
455
        self.assertIs(None, builder._nodes_by_key)
437
456
        # And spills to a second backing index but doesn't combine
438
457
        builder.add_node(*nodes[3])
439
458
        self.assertEqual(0, len(builder._nodes))
 
459
        self.assertEqual(0, len(builder._keys))
440
460
        self.assertIs(None, builder._nodes_by_key)
441
461
        self.assertEqual(2, len(builder._backing_indices))
442
462
        for backing_index in builder._backing_indices:
445
465
        builder.add_node(*nodes[4])
446
466
        builder.add_node(*nodes[5])
447
467
        self.assertEqual(0, len(builder._nodes))
 
468
        self.assertEqual(0, len(builder._keys))
448
469
        self.assertIs(None, builder._nodes_by_key)
449
470
        self.assertEqual(3, len(builder._backing_indices))
450
471
        for backing_index in builder._backing_indices:
509
530
        builder.add_node(*nodes[0])
510
531
        # Test the parts of the index that take up memory are doing so
511
532
        # predictably.
 
533
        self.assertEqual(1, len(builder._keys))
512
534
        self.assertEqual(1, len(builder._nodes))
513
535
        self.assertIs(None, builder._nodes_by_key)
514
536
        builder.add_node(*nodes[1])
 
537
        self.assertEqual(0, len(builder._keys))
515
538
        self.assertEqual(0, len(builder._nodes))
516
539
        self.assertIs(None, builder._nodes_by_key)
517
540
        self.assertEqual(1, len(builder._backing_indices))
520
543
        old = dict(builder._get_nodes_by_key()) #Build up the nodes by key dict
521
544
        builder.add_node(*nodes[2])
522
545
        self.assertEqual(1, len(builder._nodes))
 
546
        self.assertEqual(1, len(builder._keys))
523
547
        self.assertIsNot(None, builder._nodes_by_key)
524
548
        self.assertNotEqual({}, builder._nodes_by_key)
525
549
        # We should have a new entry
527
551
        # And spills to a second backing index combing all
528
552
        builder.add_node(*nodes[3])
529
553
        self.assertEqual(0, len(builder._nodes))
 
554
        self.assertEqual(0, len(builder._keys))
530
555
        self.assertIs(None, builder._nodes_by_key)
531
556
        self.assertEqual(2, len(builder._backing_indices))
532
557
        self.assertEqual(None, builder._backing_indices[0])
535
560
        builder.add_node(*nodes[4])
536
561
        builder.add_node(*nodes[5])
537
562
        self.assertEqual(0, len(builder._nodes))
 
563
        self.assertEqual(0, len(builder._keys))
538
564
        self.assertIs(None, builder._nodes_by_key)
539
565
        self.assertEqual(2, len(builder._backing_indices))
540
566
        self.assertEqual(2, builder._backing_indices[0].key_count())
611
637
        size = trans.put_file('index', stream)
612
638
        return btree_index.BTreeGraphIndex(trans, 'index', size)
613
639
 
614
 
    def test_clear_cache(self):
615
 
        nodes = self.make_nodes(160, 2, 2)
616
 
        index = self.make_index(ref_lists=2, key_elements=2, nodes=nodes)
617
 
        self.assertEqual(1, len(list(index.iter_entries([nodes[30][0]]))))
618
 
        self.assertEqual([1, 4], index._row_lengths)
619
 
        self.assertIsNot(None, index._root_node)
620
 
        internal_node_pre_clear = index._internal_node_cache.keys()
621
 
        self.assertTrue(len(index._leaf_node_cache) > 0)
622
 
        index.clear_cache()
623
 
        # We don't touch _root_node or _internal_node_cache, both should be
624
 
        # small, and can save a round trip or two
625
 
        self.assertIsNot(None, index._root_node)
626
 
        # NOTE: We don't want to affect the _internal_node_cache, as we expect
627
 
        #       it will be small, and if we ever do touch this index again, it
628
 
        #       will save round-trips.  This assertion isn't very strong,
629
 
        #       becuase without a 3-level index, we don't have any internal
630
 
        #       nodes cached.
631
 
        self.assertEqual(internal_node_pre_clear,
632
 
                         index._internal_node_cache.keys())
633
 
        self.assertEqual(0, len(index._leaf_node_cache))
634
 
 
635
640
    def test_trivial_constructor(self):
636
641
        transport = get_transport('trace+' + self.get_url(''))
637
642
        index = btree_index.BTreeGraphIndex(transport, 'index', None)
684
689
        # The entire index should have been read, as it is one page long.
685
690
        self.assertEqual([('readv', 'index', [(0, size)], False, None)],
686
691
            transport._activity)
687
 
        self.assertEqual(1173, size)
 
692
        self.assertEqual(1199, size)
688
693
 
689
694
    def test__read_nodes_no_size_one_page_reads_once(self):
690
695
        self.make_index(nodes=[(('key',), 'value', ())])
738
743
        # The entire index should have been read linearly.
739
744
        self.assertEqual([('readv', 'index', [(0, size)], False, None)],
740
745
            transport._activity)
741
 
        self.assertEqual(1488, size)
 
746
        self.assertEqual(1514, size)
742
747
 
743
748
    def test_validate_two_pages(self):
744
749
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
1110
1115
        self.assertEqual({}, parent_map)
1111
1116
        self.assertEqual(set([('one',), ('two',)]), missing_keys)
1112
1117
 
1113
 
    def test_supports_unlimited_cache(self):
1114
 
        builder = btree_index.BTreeBuilder(reference_lists=0, key_elements=1)
1115
 
        # We need enough nodes to cause a page split (so we have both an
1116
 
        # internal node and a couple leaf nodes. 500 seems to be enough.)
1117
 
        nodes = self.make_nodes(500, 1, 0)
1118
 
        for node in nodes:
1119
 
            builder.add_node(*node)
1120
 
        stream = builder.finish()
1121
 
        trans = get_transport(self.get_url())
1122
 
        size = trans.put_file('index', stream)
1123
 
        index = btree_index.BTreeGraphIndex(trans, 'index', size)
1124
 
        self.assertEqual(500, index.key_count())
1125
 
        # We have an internal node
1126
 
        self.assertEqual(2, len(index._row_lengths))
1127
 
        # We have at least 2 leaf nodes
1128
 
        self.assertTrue(index._row_lengths[-1] >= 2)
1129
 
        self.assertIsInstance(index._leaf_node_cache, lru_cache.LRUCache)
1130
 
        self.assertEqual(btree_index._NODE_CACHE_SIZE,
1131
 
                         index._leaf_node_cache._max_cache)
1132
 
        self.assertIsInstance(index._internal_node_cache, fifo_cache.FIFOCache)
1133
 
        self.assertEqual(100, index._internal_node_cache._max_cache)
1134
 
        # No change if unlimited_cache=False is passed
1135
 
        index = btree_index.BTreeGraphIndex(trans, 'index', size,
1136
 
                                            unlimited_cache=False)
1137
 
        self.assertIsInstance(index._leaf_node_cache, lru_cache.LRUCache)
1138
 
        self.assertEqual(btree_index._NODE_CACHE_SIZE,
1139
 
                         index._leaf_node_cache._max_cache)
1140
 
        self.assertIsInstance(index._internal_node_cache, fifo_cache.FIFOCache)
1141
 
        self.assertEqual(100, index._internal_node_cache._max_cache)
1142
 
        index = btree_index.BTreeGraphIndex(trans, 'index', size,
1143
 
                                            unlimited_cache=True)
1144
 
        self.assertIsInstance(index._leaf_node_cache, dict)
1145
 
        self.assertIs(type(index._internal_node_cache), dict)
1146
 
        # Exercise the lookup code
1147
 
        entries = set(index.iter_entries([n[0] for n in nodes]))
1148
 
        self.assertEqual(500, len(entries))
1149
 
 
1150
1118
 
1151
1119
class TestBTreeNodes(BTreeTestCase):
1152
1120
 
 
1121
    def restore_parser(self):
 
1122
        btree_index._btree_serializer = self.saved_parser
 
1123
 
1153
1124
    def setUp(self):
1154
1125
        BTreeTestCase.setUp(self)
1155
 
        self.overrideAttr(btree_index, '_btree_serializer', self.parse_btree)
 
1126
        self.saved_parser = btree_index._btree_serializer
 
1127
        self.addCleanup(self.restore_parser)
 
1128
        btree_index._btree_serializer = self.parse_btree
1156
1129
 
1157
1130
    def test_LeafNode_1_0(self):
1158
1131
        node_bytes = ("type=leaf\n"
1269
1242
    def test_exists(self):
1270
1243
        # This is just to let the user know if they don't have the feature
1271
1244
        # available
1272
 
        self.requireFeature(compiled_btreeparser_feature)
 
1245
        self.requireFeature(CompiledBtreeParserFeature)
1273
1246
 
1274
1247
 
1275
1248
class TestMultiBisectRight(tests.TestCase):