← 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: John Arbash Meinel
  • Date: 2009-08-17 15:38:52 UTC
  • mfrom: (4615.1.1 2.0b1-merge-sort)
  • mto: This revision was merged to the branch mainline in revision 4629.
  • Revision ID: john@arbash-meinel.com-20090817153852-9127xf3aabmcoegt
Bring in the changes to merge_sort.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/tests/test_btree_index.py
1
 
# Copyright (C) 2008-2011 Canonical Ltd
 
1
# Copyright (C) 2008, 2009 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
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
 
    transport,
31
28
    )
32
29
from bzrlib.tests import (
33
30
    TestCaseWithTransport,
34
 
    scenarios,
 
31
    condition_isinstance,
 
32
    multiply_tests,
 
33
    split_suite_by_condition,
35
34
    )
36
 
 
37
 
 
38
 
load_tests = scenarios.load_tests_apply_scenarios
39
 
 
40
 
 
41
 
def btreeparser_scenarios():
 
35
from bzrlib.transport import get_transport
 
36
 
 
37
 
 
38
def load_tests(standard_tests, module, loader):
 
39
    # parameterise the TestBTreeNodes tests
 
40
    node_tests, others = split_suite_by_condition(standard_tests,
 
41
        condition_isinstance(TestBTreeNodes))
42
42
    import bzrlib._btree_serializer_py as py_module
43
43
    scenarios = [('python', {'parse_btree': py_module})]
44
 
    if compiled_btreeparser_feature.available():
45
 
        scenarios.append(('C', 
46
 
            {'parse_btree': compiled_btreeparser_feature.module}))
47
 
    return scenarios
48
 
 
49
 
 
50
 
compiled_btreeparser_feature = tests.ModuleAvailableFeature(
51
 
    'bzrlib._btree_serializer_pyx')
 
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
    return multiply_tests(node_tests, scenarios, others)
 
50
 
 
51
 
 
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()
52
64
 
53
65
 
54
66
class BTreeTestCase(TestCaseWithTransport):
57
69
 
58
70
    def setUp(self):
59
71
        TestCaseWithTransport.setUp(self)
60
 
        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
61
77
 
62
78
    def make_nodes(self, count, key_elements, reference_lists):
63
79
        """Generate count*key_elements sample nodes."""
97
113
 
98
114
    def shrink_page_size(self):
99
115
        """Shrink the default page size so that less fits in a page."""
100
 
        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)
101
120
        btree_index._PAGE_SIZE = 2048
102
121
 
103
122
 
104
123
class TestBTreeBuilder(BTreeTestCase):
105
124
 
106
 
    def test_clear_cache(self):
107
 
        builder = btree_index.BTreeBuilder(reference_lists=0, key_elements=1)
108
 
        # This is a no-op, but we need the api to be consistent with other
109
 
        # BTreeGraphIndex apis.
110
 
        builder.clear_cache()
111
 
 
112
125
    def test_empty_1_0(self):
113
126
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
114
127
        # NamedTemporaryFile dies on builder.finish().read(). weird.
140
153
        temp_file = builder.finish()
141
154
        content = temp_file.read()
142
155
        del temp_file
143
 
        self.assertEqual(131, len(content))
 
156
        self.assertEqual(158, len(content))
144
157
        self.assertEqual(
145
158
            "B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=5\n"
146
159
            "row_lengths=1\n",
164
177
        temp_file = builder.finish()
165
178
        content = temp_file.read()
166
179
        del temp_file
167
 
        self.assertEqual(238, len(content))
 
180
        self.assertEqual(264, len(content))
168
181
        self.assertEqual(
169
182
            "B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=10\n"
170
183
            "row_lengths=1\n",
214
227
        leaf1_bytes = zlib.decompress(leaf1)
215
228
        sorted_node_keys = sorted(node[0] for node in nodes)
216
229
        node = btree_index._LeafNode(leaf1_bytes, 1, 0)
217
 
        self.assertEqual(231, len(node))
218
 
        self.assertEqual(sorted_node_keys[:231], node.all_keys())
 
230
        self.assertEqual(231, len(node.keys))
 
231
        self.assertEqual(sorted_node_keys[:231], sorted(node.keys))
219
232
        leaf2_bytes = zlib.decompress(leaf2)
220
233
        node = btree_index._LeafNode(leaf2_bytes, 1, 0)
221
 
        self.assertEqual(400 - 231, len(node))
222
 
        self.assertEqual(sorted_node_keys[231:], node.all_keys())
 
234
        self.assertEqual(400 - 231, len(node.keys))
 
235
        self.assertEqual(sorted_node_keys[231:], sorted(node.keys))
223
236
 
224
237
    def test_last_page_rounded_1_layer(self):
225
238
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
230
243
        temp_file = builder.finish()
231
244
        content = temp_file.read()
232
245
        del temp_file
233
 
        self.assertEqual(155, len(content))
 
246
        self.assertEqual(181, len(content))
234
247
        self.assertEqual(
235
248
            "B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=10\n"
236
249
            "row_lengths=1\n",
239
252
        leaf2 = content[74:]
240
253
        leaf2_bytes = zlib.decompress(leaf2)
241
254
        node = btree_index._LeafNode(leaf2_bytes, 1, 0)
242
 
        self.assertEqual(10, len(node))
 
255
        self.assertEqual(10, len(node.keys))
243
256
        sorted_node_keys = sorted(node[0] for node in nodes)
244
 
        self.assertEqual(sorted_node_keys, node.all_keys())
 
257
        self.assertEqual(sorted_node_keys, sorted(node.keys))
245
258
 
246
259
    def test_last_page_not_rounded_2_layer(self):
247
260
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
261
274
        leaf2 = content[8192:]
262
275
        leaf2_bytes = zlib.decompress(leaf2)
263
276
        node = btree_index._LeafNode(leaf2_bytes, 1, 0)
264
 
        self.assertEqual(400 - 231, len(node))
 
277
        self.assertEqual(400 - 231, len(node.keys))
265
278
        sorted_node_keys = sorted(node[0] for node in nodes)
266
 
        self.assertEqual(sorted_node_keys[231:], node.all_keys())
 
279
        self.assertEqual(sorted_node_keys[231:], sorted(node.keys))
267
280
 
268
281
    def test_three_level_tree_details(self):
269
282
        # The left most pointer in the second internal node in a row should
278
291
 
279
292
        for node in nodes:
280
293
            builder.add_node(*node)
281
 
        t = transport.get_transport('trace+' + self.get_url(''))
282
 
        size = t.put_file('index', self.time(builder.finish))
 
294
        transport = get_transport('trace+' + self.get_url(''))
 
295
        size = transport.put_file('index', self.time(builder.finish))
283
296
        del builder
284
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
 
297
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
285
298
        # Seed the metadata, we're using internal calls now.
286
299
        index.key_count()
287
300
        self.assertEqual(3, len(index._row_lengths),
300
313
        # in the second node it points at
301
314
        pos = index._row_offsets[2] + internal_node2.offset + 1
302
315
        leaf = index._get_leaf_nodes([pos])[pos]
303
 
        self.assertTrue(internal_node2.keys[0] in leaf)
 
316
        self.assertTrue(internal_node2.keys[0] in leaf.keys)
304
317
 
305
318
    def test_2_leaves_2_2(self):
306
319
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
338
351
        # Test the parts of the index that take up memory are doing so
339
352
        # predictably.
340
353
        self.assertEqual(1, len(builder._nodes))
 
354
        self.assertEqual(1, len(builder._keys))
341
355
        self.assertIs(None, builder._nodes_by_key)
342
356
        builder.add_node(*nodes[1])
343
357
        self.assertEqual(0, len(builder._nodes))
 
358
        self.assertEqual(0, len(builder._keys))
344
359
        self.assertIs(None, builder._nodes_by_key)
345
360
        self.assertEqual(1, len(builder._backing_indices))
346
361
        self.assertEqual(2, builder._backing_indices[0].key_count())
347
362
        # now back to memory
348
363
        builder.add_node(*nodes[2])
349
364
        self.assertEqual(1, len(builder._nodes))
 
365
        self.assertEqual(1, len(builder._keys))
350
366
        self.assertIs(None, builder._nodes_by_key)
351
367
        # And spills to a second backing index combing all
352
368
        builder.add_node(*nodes[3])
353
369
        self.assertEqual(0, len(builder._nodes))
 
370
        self.assertEqual(0, len(builder._keys))
354
371
        self.assertIs(None, builder._nodes_by_key)
355
372
        self.assertEqual(2, len(builder._backing_indices))
356
373
        self.assertEqual(None, builder._backing_indices[0])
359
376
        builder.add_node(*nodes[4])
360
377
        builder.add_node(*nodes[5])
361
378
        self.assertEqual(0, len(builder._nodes))
 
379
        self.assertEqual(0, len(builder._keys))
362
380
        self.assertIs(None, builder._nodes_by_key)
363
381
        self.assertEqual(2, len(builder._backing_indices))
364
382
        self.assertEqual(2, builder._backing_indices[0].key_count())
407
425
        self.assertEqual(None, builder._backing_indices[2])
408
426
        self.assertEqual(16, builder._backing_indices[3].key_count())
409
427
        # Now finish, and check we got a correctly ordered tree
410
 
        t = self.get_transport('')
411
 
        size = t.put_file('index', builder.finish())
412
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
 
428
        transport = self.get_transport('')
 
429
        size = transport.put_file('index', builder.finish())
 
430
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
413
431
        nodes = list(index.iter_all_entries())
414
432
        self.assertEqual(sorted(nodes), nodes)
415
433
        self.assertEqual(16, len(nodes))
422
440
        # Test the parts of the index that take up memory are doing so
423
441
        # predictably.
424
442
        self.assertEqual(1, len(builder._nodes))
 
443
        self.assertEqual(1, len(builder._keys))
425
444
        self.assertIs(None, builder._nodes_by_key)
426
445
        builder.add_node(*nodes[1])
427
446
        self.assertEqual(0, len(builder._nodes))
 
447
        self.assertEqual(0, len(builder._keys))
428
448
        self.assertIs(None, builder._nodes_by_key)
429
449
        self.assertEqual(1, len(builder._backing_indices))
430
450
        self.assertEqual(2, builder._backing_indices[0].key_count())
431
451
        # now back to memory
432
452
        builder.add_node(*nodes[2])
433
453
        self.assertEqual(1, len(builder._nodes))
 
454
        self.assertEqual(1, len(builder._keys))
434
455
        self.assertIs(None, builder._nodes_by_key)
435
456
        # And spills to a second backing index but doesn't combine
436
457
        builder.add_node(*nodes[3])
437
458
        self.assertEqual(0, len(builder._nodes))
 
459
        self.assertEqual(0, len(builder._keys))
438
460
        self.assertIs(None, builder._nodes_by_key)
439
461
        self.assertEqual(2, len(builder._backing_indices))
440
462
        for backing_index in builder._backing_indices:
443
465
        builder.add_node(*nodes[4])
444
466
        builder.add_node(*nodes[5])
445
467
        self.assertEqual(0, len(builder._nodes))
 
468
        self.assertEqual(0, len(builder._keys))
446
469
        self.assertIs(None, builder._nodes_by_key)
447
470
        self.assertEqual(3, len(builder._backing_indices))
448
471
        for backing_index in builder._backing_indices:
507
530
        builder.add_node(*nodes[0])
508
531
        # Test the parts of the index that take up memory are doing so
509
532
        # predictably.
 
533
        self.assertEqual(1, len(builder._keys))
510
534
        self.assertEqual(1, len(builder._nodes))
511
535
        self.assertIs(None, builder._nodes_by_key)
512
536
        builder.add_node(*nodes[1])
 
537
        self.assertEqual(0, len(builder._keys))
513
538
        self.assertEqual(0, len(builder._nodes))
514
539
        self.assertIs(None, builder._nodes_by_key)
515
540
        self.assertEqual(1, len(builder._backing_indices))
518
543
        old = dict(builder._get_nodes_by_key()) #Build up the nodes by key dict
519
544
        builder.add_node(*nodes[2])
520
545
        self.assertEqual(1, len(builder._nodes))
 
546
        self.assertEqual(1, len(builder._keys))
521
547
        self.assertIsNot(None, builder._nodes_by_key)
522
548
        self.assertNotEqual({}, builder._nodes_by_key)
523
549
        # We should have a new entry
525
551
        # And spills to a second backing index combing all
526
552
        builder.add_node(*nodes[3])
527
553
        self.assertEqual(0, len(builder._nodes))
 
554
        self.assertEqual(0, len(builder._keys))
528
555
        self.assertIs(None, builder._nodes_by_key)
529
556
        self.assertEqual(2, len(builder._backing_indices))
530
557
        self.assertEqual(None, builder._backing_indices[0])
533
560
        builder.add_node(*nodes[4])
534
561
        builder.add_node(*nodes[5])
535
562
        self.assertEqual(0, len(builder._nodes))
 
563
        self.assertEqual(0, len(builder._keys))
536
564
        self.assertIs(None, builder._nodes_by_key)
537
565
        self.assertEqual(2, len(builder._backing_indices))
538
566
        self.assertEqual(2, builder._backing_indices[0].key_count())
605
633
        for key, value, references in nodes:
606
634
            builder.add_node(key, value, references)
607
635
        stream = builder.finish()
608
 
        trans = transport.get_transport('trace+' + self.get_url())
 
636
        trans = get_transport('trace+' + self.get_url())
609
637
        size = trans.put_file('index', stream)
610
638
        return btree_index.BTreeGraphIndex(trans, 'index', size)
611
639
 
612
 
    def make_index_with_offset(self, ref_lists=1, key_elements=1, nodes=[],
613
 
                               offset=0):
614
 
        builder = btree_index.BTreeBuilder(key_elements=key_elements,
615
 
                                           reference_lists=ref_lists)
616
 
        builder.add_nodes(nodes)
617
 
        transport = self.get_transport('')
618
 
        # NamedTemporaryFile dies on builder.finish().read(). weird.
619
 
        temp_file = builder.finish()
620
 
        content = temp_file.read()
621
 
        del temp_file
622
 
        size = len(content)
623
 
        transport.put_bytes('index', (' '*offset)+content)
624
 
        return btree_index.BTreeGraphIndex(transport, 'index', size=size,
625
 
                                           offset=offset)
626
 
 
627
 
    def test_clear_cache(self):
628
 
        nodes = self.make_nodes(160, 2, 2)
629
 
        index = self.make_index(ref_lists=2, key_elements=2, nodes=nodes)
630
 
        self.assertEqual(1, len(list(index.iter_entries([nodes[30][0]]))))
631
 
        self.assertEqual([1, 4], index._row_lengths)
632
 
        self.assertIsNot(None, index._root_node)
633
 
        internal_node_pre_clear = index._internal_node_cache.keys()
634
 
        self.assertTrue(len(index._leaf_node_cache) > 0)
635
 
        index.clear_cache()
636
 
        # We don't touch _root_node or _internal_node_cache, both should be
637
 
        # small, and can save a round trip or two
638
 
        self.assertIsNot(None, index._root_node)
639
 
        # NOTE: We don't want to affect the _internal_node_cache, as we expect
640
 
        #       it will be small, and if we ever do touch this index again, it
641
 
        #       will save round-trips.  This assertion isn't very strong,
642
 
        #       becuase without a 3-level index, we don't have any internal
643
 
        #       nodes cached.
644
 
        self.assertEqual(internal_node_pre_clear,
645
 
                         index._internal_node_cache.keys())
646
 
        self.assertEqual(0, len(index._leaf_node_cache))
647
 
 
648
640
    def test_trivial_constructor(self):
649
 
        t = transport.get_transport('trace+' + self.get_url(''))
650
 
        index = btree_index.BTreeGraphIndex(t, 'index', None)
 
641
        transport = get_transport('trace+' + self.get_url(''))
 
642
        index = btree_index.BTreeGraphIndex(transport, 'index', None)
651
643
        # Checks the page size at load, but that isn't logged yet.
652
 
        self.assertEqual([], t._activity)
 
644
        self.assertEqual([], transport._activity)
653
645
 
654
646
    def test_with_size_constructor(self):
655
 
        t = transport.get_transport('trace+' + self.get_url(''))
656
 
        index = btree_index.BTreeGraphIndex(t, 'index', 1)
 
647
        transport = get_transport('trace+' + self.get_url(''))
 
648
        index = btree_index.BTreeGraphIndex(transport, 'index', 1)
657
649
        # Checks the page size at load, but that isn't logged yet.
658
 
        self.assertEqual([], t._activity)
 
650
        self.assertEqual([], transport._activity)
659
651
 
660
652
    def test_empty_key_count_no_size(self):
661
653
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
662
 
        t = transport.get_transport('trace+' + self.get_url(''))
663
 
        t.put_file('index', builder.finish())
664
 
        index = btree_index.BTreeGraphIndex(t, 'index', None)
665
 
        del t._activity[:]
666
 
        self.assertEqual([], t._activity)
 
654
        transport = get_transport('trace+' + self.get_url(''))
 
655
        transport.put_file('index', builder.finish())
 
656
        index = btree_index.BTreeGraphIndex(transport, 'index', None)
 
657
        del transport._activity[:]
 
658
        self.assertEqual([], transport._activity)
667
659
        self.assertEqual(0, index.key_count())
668
660
        # The entire index should have been requested (as we generally have the
669
661
        # size available, and doing many small readvs is inappropriate).
670
662
        # We can't tell how much was actually read here, but - check the code.
671
 
        self.assertEqual([('get', 'index')], t._activity)
 
663
        self.assertEqual([('get', 'index')], transport._activity)
672
664
 
673
665
    def test_empty_key_count(self):
674
666
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
675
 
        t = transport.get_transport('trace+' + self.get_url(''))
676
 
        size = t.put_file('index', builder.finish())
 
667
        transport = get_transport('trace+' + self.get_url(''))
 
668
        size = transport.put_file('index', builder.finish())
677
669
        self.assertEqual(72, size)
678
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
679
 
        del t._activity[:]
680
 
        self.assertEqual([], t._activity)
 
670
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
671
        del transport._activity[:]
 
672
        self.assertEqual([], transport._activity)
681
673
        self.assertEqual(0, index.key_count())
682
674
        # The entire index should have been read, as 4K > size
683
675
        self.assertEqual([('readv', 'index', [(0, 72)], False, None)],
684
 
                         t._activity)
 
676
            transport._activity)
685
677
 
686
678
    def test_non_empty_key_count_2_2(self):
687
679
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
688
680
        nodes = self.make_nodes(35, 2, 2)
689
681
        for node in nodes:
690
682
            builder.add_node(*node)
691
 
        t = transport.get_transport('trace+' + self.get_url(''))
692
 
        size = t.put_file('index', builder.finish())
693
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
694
 
        del t._activity[:]
695
 
        self.assertEqual([], t._activity)
 
683
        transport = get_transport('trace+' + self.get_url(''))
 
684
        size = transport.put_file('index', builder.finish())
 
685
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
686
        del transport._activity[:]
 
687
        self.assertEqual([], transport._activity)
696
688
        self.assertEqual(70, index.key_count())
697
689
        # The entire index should have been read, as it is one page long.
698
690
        self.assertEqual([('readv', 'index', [(0, size)], False, None)],
699
 
            t._activity)
700
 
        self.assertEqual(1173, size)
701
 
 
702
 
    def test_with_offset_no_size(self):
703
 
        index = self.make_index_with_offset(key_elements=1, ref_lists=1,
704
 
                                            offset=1234,
705
 
                                            nodes=self.make_nodes(200, 1, 1))
706
 
        index._size = None # throw away the size info
707
 
        self.assertEqual(200, index.key_count())
708
 
 
709
 
    def test_with_small_offset(self):
710
 
        index = self.make_index_with_offset(key_elements=1, ref_lists=1,
711
 
                                            offset=1234,
712
 
                                            nodes=self.make_nodes(200, 1, 1))
713
 
        self.assertEqual(200, index.key_count())
714
 
 
715
 
    def test_with_large_offset(self):
716
 
        index = self.make_index_with_offset(key_elements=1, ref_lists=1,
717
 
                                            offset=123456,
718
 
                                            nodes=self.make_nodes(200, 1, 1))
719
 
        self.assertEqual(200, index.key_count())
 
691
            transport._activity)
 
692
        self.assertEqual(1199, size)
720
693
 
721
694
    def test__read_nodes_no_size_one_page_reads_once(self):
722
695
        self.make_index(nodes=[(('key',), 'value', ())])
723
 
        trans = transport.get_transport('trace+' + self.get_url())
 
696
        trans = get_transport('trace+' + self.get_url())
724
697
        index = btree_index.BTreeGraphIndex(trans, 'index', None)
725
698
        del trans._activity[:]
726
699
        nodes = dict(index._read_nodes([0]))
727
700
        self.assertEqual([0], nodes.keys())
728
701
        node = nodes[0]
729
 
        self.assertEqual([('key',)], node.all_keys())
 
702
        self.assertEqual([('key',)], node.keys.keys())
730
703
        self.assertEqual([('get', 'index')], trans._activity)
731
704
 
732
705
    def test__read_nodes_no_size_multiple_pages(self):
734
707
        index.key_count()
735
708
        num_pages = index._row_offsets[-1]
736
709
        # Reopen with a traced transport and no size
737
 
        trans = transport.get_transport('trace+' + self.get_url())
 
710
        trans = get_transport('trace+' + self.get_url())
738
711
        index = btree_index.BTreeGraphIndex(trans, 'index', None)
739
712
        del trans._activity[:]
740
713
        nodes = dict(index._read_nodes([0]))
745
718
        nodes = self.make_nodes(160, 2, 2)
746
719
        for node in nodes:
747
720
            builder.add_node(*node)
748
 
        t = transport.get_transport('trace+' + self.get_url(''))
749
 
        size = t.put_file('index', builder.finish())
 
721
        transport = get_transport('trace+' + self.get_url(''))
 
722
        size = transport.put_file('index', builder.finish())
750
723
        self.assertEqual(17692, size)
751
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
752
 
        del t._activity[:]
753
 
        self.assertEqual([], t._activity)
 
724
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
725
        del transport._activity[:]
 
726
        self.assertEqual([], transport._activity)
754
727
        self.assertEqual(320, index.key_count())
755
728
        # The entire index should not have been read.
756
729
        self.assertEqual([('readv', 'index', [(0, 4096)], False, None)],
757
 
                         t._activity)
 
730
            transport._activity)
758
731
 
759
732
    def test_validate_one_page(self):
760
733
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
761
734
        nodes = self.make_nodes(45, 2, 2)
762
735
        for node in nodes:
763
736
            builder.add_node(*node)
764
 
        t = transport.get_transport('trace+' + self.get_url(''))
765
 
        size = t.put_file('index', builder.finish())
766
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
767
 
        del t._activity[:]
768
 
        self.assertEqual([], t._activity)
 
737
        transport = get_transport('trace+' + self.get_url(''))
 
738
        size = transport.put_file('index', builder.finish())
 
739
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
740
        del transport._activity[:]
 
741
        self.assertEqual([], transport._activity)
769
742
        index.validate()
770
743
        # The entire index should have been read linearly.
771
744
        self.assertEqual([('readv', 'index', [(0, size)], False, None)],
772
 
                         t._activity)
773
 
        self.assertEqual(1488, size)
 
745
            transport._activity)
 
746
        self.assertEqual(1514, size)
774
747
 
775
748
    def test_validate_two_pages(self):
776
749
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
777
750
        nodes = self.make_nodes(80, 2, 2)
778
751
        for node in nodes:
779
752
            builder.add_node(*node)
780
 
        t = transport.get_transport('trace+' + self.get_url(''))
781
 
        size = t.put_file('index', builder.finish())
 
753
        transport = get_transport('trace+' + self.get_url(''))
 
754
        size = transport.put_file('index', builder.finish())
782
755
        # Root page, 2 leaf pages
783
756
        self.assertEqual(9339, size)
784
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
785
 
        del t._activity[:]
786
 
        self.assertEqual([], t._activity)
 
757
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
758
        del transport._activity[:]
 
759
        self.assertEqual([], transport._activity)
787
760
        index.validate()
788
761
        # The entire index should have been read linearly.
789
 
        self.assertEqual(
790
 
            [('readv', 'index', [(0, 4096)], False, None),
791
 
             ('readv', 'index', [(4096, 4096), (8192, 1147)], False, None)],
792
 
            t._activity)
 
762
        self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
 
763
            ('readv', 'index', [(4096, 4096), (8192, 1147)], False, None)],
 
764
            transport._activity)
793
765
        # XXX: TODO: write some badly-ordered nodes, and some pointers-to-wrong
794
766
        # node and make validate find them.
795
767
 
796
768
    def test_eq_ne(self):
797
769
        # two indices are equal when constructed with the same parameters:
798
 
        t1 = transport.get_transport('trace+' + self.get_url(''))
799
 
        t2 = self.get_transport()
800
 
        self.assertTrue(
801
 
            btree_index.BTreeGraphIndex(t1, 'index', None) ==
802
 
            btree_index.BTreeGraphIndex(t1, 'index', None))
803
 
        self.assertTrue(
804
 
            btree_index.BTreeGraphIndex(t1, 'index', 20) ==
805
 
            btree_index.BTreeGraphIndex(t1, 'index', 20))
806
 
        self.assertFalse(
807
 
            btree_index.BTreeGraphIndex(t1, 'index', 20) ==
808
 
            btree_index.BTreeGraphIndex(t2, 'index', 20))
809
 
        self.assertFalse(
810
 
            btree_index.BTreeGraphIndex(t1, 'inde1', 20) ==
811
 
            btree_index.BTreeGraphIndex(t1, 'inde2', 20))
812
 
        self.assertFalse(
813
 
            btree_index.BTreeGraphIndex(t1, 'index', 10) ==
814
 
            btree_index.BTreeGraphIndex(t1, 'index', 20))
815
 
        self.assertFalse(
816
 
            btree_index.BTreeGraphIndex(t1, 'index', None) !=
817
 
            btree_index.BTreeGraphIndex(t1, 'index', None))
818
 
        self.assertFalse(
819
 
            btree_index.BTreeGraphIndex(t1, 'index', 20) !=
820
 
            btree_index.BTreeGraphIndex(t1, 'index', 20))
821
 
        self.assertTrue(
822
 
            btree_index.BTreeGraphIndex(t1, 'index', 20) !=
823
 
            btree_index.BTreeGraphIndex(t2, 'index', 20))
824
 
        self.assertTrue(
825
 
            btree_index.BTreeGraphIndex(t1, 'inde1', 20) !=
826
 
            btree_index.BTreeGraphIndex(t1, 'inde2', 20))
827
 
        self.assertTrue(
828
 
            btree_index.BTreeGraphIndex(t1, 'index', 10) !=
829
 
            btree_index.BTreeGraphIndex(t1, 'index', 20))
 
770
        transport1 = get_transport('trace+' + self.get_url(''))
 
771
        transport2 = get_transport(self.get_url(''))
 
772
        self.assertTrue(
 
773
            btree_index.BTreeGraphIndex(transport1, 'index', None) ==
 
774
            btree_index.BTreeGraphIndex(transport1, 'index', None))
 
775
        self.assertTrue(
 
776
            btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
 
777
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
 
778
        self.assertFalse(
 
779
            btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
 
780
            btree_index.BTreeGraphIndex(transport2, 'index', 20))
 
781
        self.assertFalse(
 
782
            btree_index.BTreeGraphIndex(transport1, 'inde1', 20) ==
 
783
            btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
 
784
        self.assertFalse(
 
785
            btree_index.BTreeGraphIndex(transport1, 'index', 10) ==
 
786
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
 
787
        self.assertFalse(
 
788
            btree_index.BTreeGraphIndex(transport1, 'index', None) !=
 
789
            btree_index.BTreeGraphIndex(transport1, 'index', None))
 
790
        self.assertFalse(
 
791
            btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
 
792
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
 
793
        self.assertTrue(
 
794
            btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
 
795
            btree_index.BTreeGraphIndex(transport2, 'index', 20))
 
796
        self.assertTrue(
 
797
            btree_index.BTreeGraphIndex(transport1, 'inde1', 20) !=
 
798
            btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
 
799
        self.assertTrue(
 
800
            btree_index.BTreeGraphIndex(transport1, 'index', 10) !=
 
801
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
830
802
 
831
803
    def test_iter_all_only_root_no_size(self):
832
804
        self.make_index(nodes=[(('key',), 'value', ())])
833
 
        t = transport.get_transport('trace+' + self.get_url(''))
834
 
        index = btree_index.BTreeGraphIndex(t, 'index', None)
835
 
        del t._activity[:]
 
805
        trans = get_transport('trace+' + self.get_url(''))
 
806
        index = btree_index.BTreeGraphIndex(trans, 'index', None)
 
807
        del trans._activity[:]
836
808
        self.assertEqual([(('key',), 'value')],
837
809
                         [x[1:] for x in index.iter_all_entries()])
838
 
        self.assertEqual([('get', 'index')], t._activity)
 
810
        self.assertEqual([('get', 'index')], trans._activity)
839
811
 
840
812
    def test_iter_all_entries_reads(self):
841
813
        # iterating all entries reads the header, then does a linear
847
819
        nodes = self.make_nodes(10000, 2, 2)
848
820
        for node in nodes:
849
821
            builder.add_node(*node)
850
 
        t = transport.get_transport('trace+' + self.get_url(''))
851
 
        size = t.put_file('index', builder.finish())
 
822
        transport = get_transport('trace+' + self.get_url(''))
 
823
        size = transport.put_file('index', builder.finish())
852
824
        self.assertEqual(1303220, size, 'number of expected bytes in the'
853
825
                                        ' output changed')
854
826
        page_size = btree_index._PAGE_SIZE
855
827
        del builder
856
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
857
 
        del t._activity[:]
858
 
        self.assertEqual([], t._activity)
 
828
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
829
        del transport._activity[:]
 
830
        self.assertEqual([], transport._activity)
859
831
        found_nodes = self.time(list, index.iter_all_entries())
860
832
        bare_nodes = []
861
833
        for node in found_nodes:
882
854
        readv_request[-1] = (readv_request[-1][0], 1303220 % page_size)
883
855
        expected = [('readv', 'index', [(0, page_size)], False, None),
884
856
             ('readv',  'index', readv_request, False, None)]
885
 
        if expected != t._activity:
 
857
        if expected != transport._activity:
886
858
            self.assertEqualDiff(pprint.pformat(expected),
887
859
                                 pprint.pformat(transport._activity))
888
860
 
902
874
        nodes = self.make_nodes(160, 2, 2)
903
875
        for node in nodes:
904
876
            builder.add_node(*node)
905
 
        t = transport.get_transport('trace+' + self.get_url(''))
906
 
        size = t.put_file('index', builder.finish())
 
877
        transport = get_transport('trace+' + self.get_url(''))
 
878
        size = transport.put_file('index', builder.finish())
907
879
        del builder
908
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
909
 
        del t._activity[:]
910
 
        self.assertEqual([], t._activity)
 
880
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
881
        del transport._activity[:]
 
882
        self.assertEqual([], transport._activity)
911
883
        # search for one key
912
884
        found_nodes = list(index.iter_entries([nodes[30][0]]))
913
885
        bare_nodes = []
921
893
        # Should have read the root node, then one leaf page:
922
894
        self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
923
895
             ('readv',  'index', [(8192, 4096), ], False, None)],
924
 
            t._activity)
 
896
            transport._activity)
925
897
 
926
898
    def test_iter_key_prefix_1_element_key_None(self):
927
899
        index = self.make_index()
1110
1082
        min_l2_key = l2.min_key
1111
1083
        max_l1_key = l1.max_key
1112
1084
        self.assertTrue(max_l1_key < min_l2_key)
1113
 
        parents_min_l2_key = l2[min_l2_key][1][0]
 
1085
        parents_min_l2_key = l2.keys[min_l2_key][1][0]
1114
1086
        self.assertEqual((l1.max_key,), parents_min_l2_key)
1115
1087
        # Now, whatever key we select that would fall on the second page,
1116
1088
        # should give us all the parents until the page break
1129
1101
        parent_map = {}
1130
1102
        search_keys = index._find_ancestors([max_l1_key], 0, parent_map,
1131
1103
                                            missing_keys)
1132
 
        self.assertEqual(l1.all_keys(), sorted(parent_map))
 
1104
        self.assertEqual(sorted(l1.keys), sorted(parent_map))
1133
1105
        self.assertEqual(set(), missing_keys)
1134
1106
        self.assertEqual(set(), search_keys)
1135
1107
 
1143
1115
        self.assertEqual({}, parent_map)
1144
1116
        self.assertEqual(set([('one',), ('two',)]), missing_keys)
1145
1117
 
1146
 
    def test_supports_unlimited_cache(self):
1147
 
        builder = btree_index.BTreeBuilder(reference_lists=0, key_elements=1)
1148
 
        # We need enough nodes to cause a page split (so we have both an
1149
 
        # internal node and a couple leaf nodes. 500 seems to be enough.)
1150
 
        nodes = self.make_nodes(500, 1, 0)
1151
 
        for node in nodes:
1152
 
            builder.add_node(*node)
1153
 
        stream = builder.finish()
1154
 
        trans = self.get_transport()
1155
 
        size = trans.put_file('index', stream)
1156
 
        index = btree_index.BTreeGraphIndex(trans, 'index', size)
1157
 
        self.assertEqual(500, index.key_count())
1158
 
        # We have an internal node
1159
 
        self.assertEqual(2, len(index._row_lengths))
1160
 
        # We have at least 2 leaf nodes
1161
 
        self.assertTrue(index._row_lengths[-1] >= 2)
1162
 
        self.assertIsInstance(index._leaf_node_cache, lru_cache.LRUCache)
1163
 
        self.assertEqual(btree_index._NODE_CACHE_SIZE,
1164
 
                         index._leaf_node_cache._max_cache)
1165
 
        self.assertIsInstance(index._internal_node_cache, fifo_cache.FIFOCache)
1166
 
        self.assertEqual(100, index._internal_node_cache._max_cache)
1167
 
        # No change if unlimited_cache=False is passed
1168
 
        index = btree_index.BTreeGraphIndex(trans, 'index', size,
1169
 
                                            unlimited_cache=False)
1170
 
        self.assertIsInstance(index._leaf_node_cache, lru_cache.LRUCache)
1171
 
        self.assertEqual(btree_index._NODE_CACHE_SIZE,
1172
 
                         index._leaf_node_cache._max_cache)
1173
 
        self.assertIsInstance(index._internal_node_cache, fifo_cache.FIFOCache)
1174
 
        self.assertEqual(100, index._internal_node_cache._max_cache)
1175
 
        index = btree_index.BTreeGraphIndex(trans, 'index', size,
1176
 
                                            unlimited_cache=True)
1177
 
        self.assertIsInstance(index._leaf_node_cache, dict)
1178
 
        self.assertIs(type(index._internal_node_cache), dict)
1179
 
        # Exercise the lookup code
1180
 
        entries = set(index.iter_entries([n[0] for n in nodes]))
1181
 
        self.assertEqual(500, len(entries))
1182
 
 
1183
1118
 
1184
1119
class TestBTreeNodes(BTreeTestCase):
1185
1120
 
1186
 
    scenarios = btreeparser_scenarios()
 
1121
    def restore_parser(self):
 
1122
        btree_index._btree_serializer = self.saved_parser
1187
1123
 
1188
1124
    def setUp(self):
1189
1125
        BTreeTestCase.setUp(self)
1190
 
        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
1191
1129
 
1192
1130
    def test_LeafNode_1_0(self):
1193
1131
        node_bytes = ("type=leaf\n"
1205
1143
            ("2222222222222222222222222222222222222222",): ("value:2", ()),
1206
1144
            ("3333333333333333333333333333333333333333",): ("value:3", ()),
1207
1145
            ("4444444444444444444444444444444444444444",): ("value:4", ()),
1208
 
            }, dict(node.all_items()))
 
1146
            }, node.keys)
1209
1147
 
1210
1148
    def test_LeafNode_2_2(self):
1211
1149
        node_bytes = ("type=leaf\n"
1225
1163
            ('11', '33'): ('value:3',
1226
1164
                ((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
1227
1165
            ('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
1228
 
            }, dict(node.all_items()))
 
1166
            }, node.keys)
1229
1167
 
1230
1168
    def test_InternalNode_1(self):
1231
1169
        node_bytes = ("type=internal\n"
1266
1204
            ('11', '33'): ('value:3',
1267
1205
                ((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
1268
1206
            ('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
1269
 
            }, dict(node.all_items()))
 
1207
            }, node.keys)
1270
1208
 
1271
1209
    def assertFlattened(self, expected, key, value, refs):
1272
1210
        flat_key, flat_line = self.parse_btree._flatten_node(
1304
1242
    def test_exists(self):
1305
1243
        # This is just to let the user know if they don't have the feature
1306
1244
        # available
1307
 
        self.requireFeature(compiled_btreeparser_feature)
 
1245
        self.requireFeature(CompiledBtreeParserFeature)
1308
1246
 
1309
1247
 
1310
1248
class TestMultiBisectRight(tests.TestCase):
1350
1288
        This doesn't actually create anything on disk, it just primes a
1351
1289
        BTreeGraphIndex with the recommended information.
1352
1290
        """
1353
 
        index = btree_index.BTreeGraphIndex(
1354
 
            transport.get_transport('memory:///'), 'test-index', size=size)
 
1291
        index = btree_index.BTreeGraphIndex(get_transport('memory:///'),
 
1292
                                            'test-index', size=size)
1355
1293
        if recommended_pages is not None:
1356
1294
            index._recommended_pages = recommended_pages
1357
1295
        return index