← 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: Johan Walles
  • Date: 2009-05-07 05:08:46 UTC
  • mfrom: (4342 +trunk)
  • mto: This revision was merged to the branch mainline in revision 4343.
  • Revision ID: johan.walles@gmail.com-20090507050846-nkwvcyauf1eh653q
Merge from upstream.

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 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
 
    osutils,
29
26
    tests,
30
 
    transport,
31
27
    )
32
28
from bzrlib.tests import (
33
29
    TestCaseWithTransport,
34
 
    scenarios,
 
30
    condition_isinstance,
 
31
    multiply_tests,
 
32
    split_suite_by_condition,
35
33
    )
36
 
 
37
 
 
38
 
load_tests = scenarios.load_tests_apply_scenarios
39
 
 
40
 
 
41
 
def btreeparser_scenarios():
 
34
from bzrlib.transport import get_transport
 
35
 
 
36
 
 
37
def load_tests(standard_tests, module, loader):
 
38
    # parameterise the TestBTreeNodes tests
 
39
    node_tests, others = split_suite_by_condition(standard_tests,
 
40
        condition_isinstance(TestBTreeNodes))
42
41
    import bzrlib._btree_serializer_py as py_module
43
42
    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')
 
43
    if CompiledBtreeParserFeature.available():
 
44
        # Is there a way to do this that gets missing feature failures rather
 
45
        # than no indication to the user?
 
46
        import bzrlib._btree_serializer_c as c_module
 
47
        scenarios.append(('C', {'parse_btree': c_module}))
 
48
    return multiply_tests(node_tests, scenarios, others)
 
49
 
 
50
 
 
51
class _CompiledBtreeParserFeature(tests.Feature):
 
52
    def _probe(self):
 
53
        try:
 
54
            import bzrlib._btree_serializer_c
 
55
        except ImportError:
 
56
            return False
 
57
        return True
 
58
 
 
59
    def feature_name(self):
 
60
        return 'bzrlib._btree_serializer_c'
 
61
 
 
62
CompiledBtreeParserFeature = _CompiledBtreeParserFeature()
52
63
 
53
64
 
54
65
class BTreeTestCase(TestCaseWithTransport):
57
68
 
58
69
    def setUp(self):
59
70
        TestCaseWithTransport.setUp(self)
60
 
        self.overrideAttr(btree_index, '_RESERVED_HEADER_BYTES', 100)
 
71
        self._original_header = btree_index._RESERVED_HEADER_BYTES
 
72
        def restore():
 
73
            btree_index._RESERVED_HEADER_BYTES = self._original_header
 
74
        self.addCleanup(restore)
 
75
        btree_index._RESERVED_HEADER_BYTES = 100
61
76
 
62
77
    def make_nodes(self, count, key_elements, reference_lists):
63
78
        """Generate count*key_elements sample nodes."""
97
112
 
98
113
    def shrink_page_size(self):
99
114
        """Shrink the default page size so that less fits in a page."""
100
 
        self.overrideAttr(btree_index, '_PAGE_SIZE')
 
115
        old_page_size = btree_index._PAGE_SIZE
 
116
        def cleanup():
 
117
            btree_index._PAGE_SIZE = old_page_size
 
118
        self.addCleanup(cleanup)
101
119
        btree_index._PAGE_SIZE = 2048
102
120
 
103
121
 
104
122
class TestBTreeBuilder(BTreeTestCase):
105
123
 
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
124
    def test_empty_1_0(self):
113
125
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
114
126
        # NamedTemporaryFile dies on builder.finish().read(). weird.
140
152
        temp_file = builder.finish()
141
153
        content = temp_file.read()
142
154
        del temp_file
143
 
        self.assertEqual(131, len(content))
 
155
        self.assertEqual(158, len(content))
144
156
        self.assertEqual(
145
157
            "B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=5\n"
146
158
            "row_lengths=1\n",
164
176
        temp_file = builder.finish()
165
177
        content = temp_file.read()
166
178
        del temp_file
167
 
        self.assertEqual(238, len(content))
 
179
        self.assertEqual(264, len(content))
168
180
        self.assertEqual(
169
181
            "B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=10\n"
170
182
            "row_lengths=1\n",
214
226
        leaf1_bytes = zlib.decompress(leaf1)
215
227
        sorted_node_keys = sorted(node[0] for node in nodes)
216
228
        node = btree_index._LeafNode(leaf1_bytes, 1, 0)
217
 
        self.assertEqual(231, len(node))
218
 
        self.assertEqual(sorted_node_keys[:231], node.all_keys())
 
229
        self.assertEqual(231, len(node.keys))
 
230
        self.assertEqual(sorted_node_keys[:231], sorted(node.keys))
219
231
        leaf2_bytes = zlib.decompress(leaf2)
220
232
        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())
 
233
        self.assertEqual(400 - 231, len(node.keys))
 
234
        self.assertEqual(sorted_node_keys[231:], sorted(node.keys))
223
235
 
224
236
    def test_last_page_rounded_1_layer(self):
225
237
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
230
242
        temp_file = builder.finish()
231
243
        content = temp_file.read()
232
244
        del temp_file
233
 
        self.assertEqual(155, len(content))
 
245
        self.assertEqual(181, len(content))
234
246
        self.assertEqual(
235
247
            "B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=10\n"
236
248
            "row_lengths=1\n",
239
251
        leaf2 = content[74:]
240
252
        leaf2_bytes = zlib.decompress(leaf2)
241
253
        node = btree_index._LeafNode(leaf2_bytes, 1, 0)
242
 
        self.assertEqual(10, len(node))
 
254
        self.assertEqual(10, len(node.keys))
243
255
        sorted_node_keys = sorted(node[0] for node in nodes)
244
 
        self.assertEqual(sorted_node_keys, node.all_keys())
 
256
        self.assertEqual(sorted_node_keys, sorted(node.keys))
245
257
 
246
258
    def test_last_page_not_rounded_2_layer(self):
247
259
        builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
261
273
        leaf2 = content[8192:]
262
274
        leaf2_bytes = zlib.decompress(leaf2)
263
275
        node = btree_index._LeafNode(leaf2_bytes, 1, 0)
264
 
        self.assertEqual(400 - 231, len(node))
 
276
        self.assertEqual(400 - 231, len(node.keys))
265
277
        sorted_node_keys = sorted(node[0] for node in nodes)
266
 
        self.assertEqual(sorted_node_keys[231:], node.all_keys())
 
278
        self.assertEqual(sorted_node_keys[231:], sorted(node.keys))
267
279
 
268
280
    def test_three_level_tree_details(self):
269
281
        # The left most pointer in the second internal node in a row should
278
290
 
279
291
        for node in nodes:
280
292
            builder.add_node(*node)
281
 
        t = transport.get_transport('trace+' + self.get_url(''))
282
 
        size = t.put_file('index', self.time(builder.finish))
 
293
        transport = get_transport('trace+' + self.get_url(''))
 
294
        size = transport.put_file('index', self.time(builder.finish))
283
295
        del builder
284
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
 
296
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
285
297
        # Seed the metadata, we're using internal calls now.
286
298
        index.key_count()
287
299
        self.assertEqual(3, len(index._row_lengths),
300
312
        # in the second node it points at
301
313
        pos = index._row_offsets[2] + internal_node2.offset + 1
302
314
        leaf = index._get_leaf_nodes([pos])[pos]
303
 
        self.assertTrue(internal_node2.keys[0] in leaf)
 
315
        self.assertTrue(internal_node2.keys[0] in leaf.keys)
304
316
 
305
317
    def test_2_leaves_2_2(self):
306
318
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
338
350
        # Test the parts of the index that take up memory are doing so
339
351
        # predictably.
340
352
        self.assertEqual(1, len(builder._nodes))
 
353
        self.assertEqual(1, len(builder._keys))
341
354
        self.assertIs(None, builder._nodes_by_key)
342
355
        builder.add_node(*nodes[1])
343
356
        self.assertEqual(0, len(builder._nodes))
 
357
        self.assertEqual(0, len(builder._keys))
344
358
        self.assertIs(None, builder._nodes_by_key)
345
359
        self.assertEqual(1, len(builder._backing_indices))
346
360
        self.assertEqual(2, builder._backing_indices[0].key_count())
347
361
        # now back to memory
348
362
        builder.add_node(*nodes[2])
349
363
        self.assertEqual(1, len(builder._nodes))
 
364
        self.assertEqual(1, len(builder._keys))
350
365
        self.assertIs(None, builder._nodes_by_key)
351
366
        # And spills to a second backing index combing all
352
367
        builder.add_node(*nodes[3])
353
368
        self.assertEqual(0, len(builder._nodes))
 
369
        self.assertEqual(0, len(builder._keys))
354
370
        self.assertIs(None, builder._nodes_by_key)
355
371
        self.assertEqual(2, len(builder._backing_indices))
356
372
        self.assertEqual(None, builder._backing_indices[0])
359
375
        builder.add_node(*nodes[4])
360
376
        builder.add_node(*nodes[5])
361
377
        self.assertEqual(0, len(builder._nodes))
 
378
        self.assertEqual(0, len(builder._keys))
362
379
        self.assertIs(None, builder._nodes_by_key)
363
380
        self.assertEqual(2, len(builder._backing_indices))
364
381
        self.assertEqual(2, builder._backing_indices[0].key_count())
407
424
        self.assertEqual(None, builder._backing_indices[2])
408
425
        self.assertEqual(16, builder._backing_indices[3].key_count())
409
426
        # 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)
 
427
        transport = self.get_transport('')
 
428
        size = transport.put_file('index', builder.finish())
 
429
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
413
430
        nodes = list(index.iter_all_entries())
414
431
        self.assertEqual(sorted(nodes), nodes)
415
432
        self.assertEqual(16, len(nodes))
422
439
        # Test the parts of the index that take up memory are doing so
423
440
        # predictably.
424
441
        self.assertEqual(1, len(builder._nodes))
 
442
        self.assertEqual(1, len(builder._keys))
425
443
        self.assertIs(None, builder._nodes_by_key)
426
444
        builder.add_node(*nodes[1])
427
445
        self.assertEqual(0, len(builder._nodes))
 
446
        self.assertEqual(0, len(builder._keys))
428
447
        self.assertIs(None, builder._nodes_by_key)
429
448
        self.assertEqual(1, len(builder._backing_indices))
430
449
        self.assertEqual(2, builder._backing_indices[0].key_count())
431
450
        # now back to memory
432
451
        builder.add_node(*nodes[2])
433
452
        self.assertEqual(1, len(builder._nodes))
 
453
        self.assertEqual(1, len(builder._keys))
434
454
        self.assertIs(None, builder._nodes_by_key)
435
455
        # And spills to a second backing index but doesn't combine
436
456
        builder.add_node(*nodes[3])
437
457
        self.assertEqual(0, len(builder._nodes))
 
458
        self.assertEqual(0, len(builder._keys))
438
459
        self.assertIs(None, builder._nodes_by_key)
439
460
        self.assertEqual(2, len(builder._backing_indices))
440
461
        for backing_index in builder._backing_indices:
443
464
        builder.add_node(*nodes[4])
444
465
        builder.add_node(*nodes[5])
445
466
        self.assertEqual(0, len(builder._nodes))
 
467
        self.assertEqual(0, len(builder._keys))
446
468
        self.assertIs(None, builder._nodes_by_key)
447
469
        self.assertEqual(3, len(builder._backing_indices))
448
470
        for backing_index in builder._backing_indices:
507
529
        builder.add_node(*nodes[0])
508
530
        # Test the parts of the index that take up memory are doing so
509
531
        # predictably.
 
532
        self.assertEqual(1, len(builder._keys))
510
533
        self.assertEqual(1, len(builder._nodes))
511
534
        self.assertIs(None, builder._nodes_by_key)
512
535
        builder.add_node(*nodes[1])
 
536
        self.assertEqual(0, len(builder._keys))
513
537
        self.assertEqual(0, len(builder._nodes))
514
538
        self.assertIs(None, builder._nodes_by_key)
515
539
        self.assertEqual(1, len(builder._backing_indices))
518
542
        old = dict(builder._get_nodes_by_key()) #Build up the nodes by key dict
519
543
        builder.add_node(*nodes[2])
520
544
        self.assertEqual(1, len(builder._nodes))
 
545
        self.assertEqual(1, len(builder._keys))
521
546
        self.assertIsNot(None, builder._nodes_by_key)
522
547
        self.assertNotEqual({}, builder._nodes_by_key)
523
548
        # We should have a new entry
525
550
        # And spills to a second backing index combing all
526
551
        builder.add_node(*nodes[3])
527
552
        self.assertEqual(0, len(builder._nodes))
 
553
        self.assertEqual(0, len(builder._keys))
528
554
        self.assertIs(None, builder._nodes_by_key)
529
555
        self.assertEqual(2, len(builder._backing_indices))
530
556
        self.assertEqual(None, builder._backing_indices[0])
533
559
        builder.add_node(*nodes[4])
534
560
        builder.add_node(*nodes[5])
535
561
        self.assertEqual(0, len(builder._nodes))
 
562
        self.assertEqual(0, len(builder._keys))
536
563
        self.assertIs(None, builder._nodes_by_key)
537
564
        self.assertEqual(2, len(builder._backing_indices))
538
565
        self.assertEqual(2, builder._backing_indices[0].key_count())
605
632
        for key, value, references in nodes:
606
633
            builder.add_node(key, value, references)
607
634
        stream = builder.finish()
608
 
        trans = transport.get_transport('trace+' + self.get_url())
 
635
        trans = get_transport('trace+' + self.get_url())
609
636
        size = trans.put_file('index', stream)
610
637
        return btree_index.BTreeGraphIndex(trans, 'index', size)
611
638
 
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
639
    def test_trivial_constructor(self):
649
 
        t = transport.get_transport('trace+' + self.get_url(''))
650
 
        index = btree_index.BTreeGraphIndex(t, 'index', None)
 
640
        transport = get_transport('trace+' + self.get_url(''))
 
641
        index = btree_index.BTreeGraphIndex(transport, 'index', None)
651
642
        # Checks the page size at load, but that isn't logged yet.
652
 
        self.assertEqual([], t._activity)
 
643
        self.assertEqual([], transport._activity)
653
644
 
654
645
    def test_with_size_constructor(self):
655
 
        t = transport.get_transport('trace+' + self.get_url(''))
656
 
        index = btree_index.BTreeGraphIndex(t, 'index', 1)
 
646
        transport = get_transport('trace+' + self.get_url(''))
 
647
        index = btree_index.BTreeGraphIndex(transport, 'index', 1)
657
648
        # Checks the page size at load, but that isn't logged yet.
658
 
        self.assertEqual([], t._activity)
 
649
        self.assertEqual([], transport._activity)
659
650
 
660
651
    def test_empty_key_count_no_size(self):
661
652
        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)
 
653
        transport = get_transport('trace+' + self.get_url(''))
 
654
        transport.put_file('index', builder.finish())
 
655
        index = btree_index.BTreeGraphIndex(transport, 'index', None)
 
656
        del transport._activity[:]
 
657
        self.assertEqual([], transport._activity)
667
658
        self.assertEqual(0, index.key_count())
668
659
        # The entire index should have been requested (as we generally have the
669
660
        # size available, and doing many small readvs is inappropriate).
670
661
        # We can't tell how much was actually read here, but - check the code.
671
 
        self.assertEqual([('get', 'index')], t._activity)
 
662
        self.assertEqual([('get', 'index')], transport._activity)
672
663
 
673
664
    def test_empty_key_count(self):
674
665
        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())
 
666
        transport = get_transport('trace+' + self.get_url(''))
 
667
        size = transport.put_file('index', builder.finish())
677
668
        self.assertEqual(72, size)
678
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
679
 
        del t._activity[:]
680
 
        self.assertEqual([], t._activity)
 
669
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
670
        del transport._activity[:]
 
671
        self.assertEqual([], transport._activity)
681
672
        self.assertEqual(0, index.key_count())
682
673
        # The entire index should have been read, as 4K > size
683
674
        self.assertEqual([('readv', 'index', [(0, 72)], False, None)],
684
 
                         t._activity)
 
675
            transport._activity)
685
676
 
686
677
    def test_non_empty_key_count_2_2(self):
687
678
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
688
679
        nodes = self.make_nodes(35, 2, 2)
689
680
        for node in nodes:
690
681
            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)
 
682
        transport = get_transport('trace+' + self.get_url(''))
 
683
        size = transport.put_file('index', builder.finish())
 
684
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
685
        del transport._activity[:]
 
686
        self.assertEqual([], transport._activity)
696
687
        self.assertEqual(70, index.key_count())
697
688
        # The entire index should have been read, as it is one page long.
698
689
        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())
 
690
            transport._activity)
 
691
        self.assertEqual(1199, size)
720
692
 
721
693
    def test__read_nodes_no_size_one_page_reads_once(self):
722
694
        self.make_index(nodes=[(('key',), 'value', ())])
723
 
        trans = transport.get_transport('trace+' + self.get_url())
 
695
        trans = get_transport('trace+' + self.get_url())
724
696
        index = btree_index.BTreeGraphIndex(trans, 'index', None)
725
697
        del trans._activity[:]
726
698
        nodes = dict(index._read_nodes([0]))
727
699
        self.assertEqual([0], nodes.keys())
728
700
        node = nodes[0]
729
 
        self.assertEqual([('key',)], node.all_keys())
 
701
        self.assertEqual([('key',)], node.keys.keys())
730
702
        self.assertEqual([('get', 'index')], trans._activity)
731
703
 
732
704
    def test__read_nodes_no_size_multiple_pages(self):
734
706
        index.key_count()
735
707
        num_pages = index._row_offsets[-1]
736
708
        # Reopen with a traced transport and no size
737
 
        trans = transport.get_transport('trace+' + self.get_url())
 
709
        trans = get_transport('trace+' + self.get_url())
738
710
        index = btree_index.BTreeGraphIndex(trans, 'index', None)
739
711
        del trans._activity[:]
740
712
        nodes = dict(index._read_nodes([0]))
745
717
        nodes = self.make_nodes(160, 2, 2)
746
718
        for node in nodes:
747
719
            builder.add_node(*node)
748
 
        t = transport.get_transport('trace+' + self.get_url(''))
749
 
        size = t.put_file('index', builder.finish())
 
720
        transport = get_transport('trace+' + self.get_url(''))
 
721
        size = transport.put_file('index', builder.finish())
750
722
        self.assertEqual(17692, size)
751
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
752
 
        del t._activity[:]
753
 
        self.assertEqual([], t._activity)
 
723
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
724
        del transport._activity[:]
 
725
        self.assertEqual([], transport._activity)
754
726
        self.assertEqual(320, index.key_count())
755
727
        # The entire index should not have been read.
756
728
        self.assertEqual([('readv', 'index', [(0, 4096)], False, None)],
757
 
                         t._activity)
 
729
            transport._activity)
758
730
 
759
731
    def test_validate_one_page(self):
760
732
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
761
733
        nodes = self.make_nodes(45, 2, 2)
762
734
        for node in nodes:
763
735
            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)
 
736
        transport = get_transport('trace+' + self.get_url(''))
 
737
        size = transport.put_file('index', builder.finish())
 
738
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
739
        del transport._activity[:]
 
740
        self.assertEqual([], transport._activity)
769
741
        index.validate()
770
742
        # The entire index should have been read linearly.
771
743
        self.assertEqual([('readv', 'index', [(0, size)], False, None)],
772
 
                         t._activity)
773
 
        self.assertEqual(1488, size)
 
744
            transport._activity)
 
745
        self.assertEqual(1514, size)
774
746
 
775
747
    def test_validate_two_pages(self):
776
748
        builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
777
749
        nodes = self.make_nodes(80, 2, 2)
778
750
        for node in nodes:
779
751
            builder.add_node(*node)
780
 
        t = transport.get_transport('trace+' + self.get_url(''))
781
 
        size = t.put_file('index', builder.finish())
 
752
        transport = get_transport('trace+' + self.get_url(''))
 
753
        size = transport.put_file('index', builder.finish())
782
754
        # Root page, 2 leaf pages
783
755
        self.assertEqual(9339, size)
784
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
785
 
        del t._activity[:]
786
 
        self.assertEqual([], t._activity)
 
756
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
757
        del transport._activity[:]
 
758
        self.assertEqual([], transport._activity)
787
759
        index.validate()
788
760
        # 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)
 
761
        self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
 
762
            ('readv', 'index', [(4096, 4096), (8192, 1147)], False, None)],
 
763
            transport._activity)
793
764
        # XXX: TODO: write some badly-ordered nodes, and some pointers-to-wrong
794
765
        # node and make validate find them.
795
766
 
796
767
    def test_eq_ne(self):
797
768
        # 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))
 
769
        transport1 = get_transport('trace+' + self.get_url(''))
 
770
        transport2 = get_transport(self.get_url(''))
 
771
        self.assertTrue(
 
772
            btree_index.BTreeGraphIndex(transport1, 'index', None) ==
 
773
            btree_index.BTreeGraphIndex(transport1, 'index', None))
 
774
        self.assertTrue(
 
775
            btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
 
776
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
 
777
        self.assertFalse(
 
778
            btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
 
779
            btree_index.BTreeGraphIndex(transport2, 'index', 20))
 
780
        self.assertFalse(
 
781
            btree_index.BTreeGraphIndex(transport1, 'inde1', 20) ==
 
782
            btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
 
783
        self.assertFalse(
 
784
            btree_index.BTreeGraphIndex(transport1, 'index', 10) ==
 
785
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
 
786
        self.assertFalse(
 
787
            btree_index.BTreeGraphIndex(transport1, 'index', None) !=
 
788
            btree_index.BTreeGraphIndex(transport1, 'index', None))
 
789
        self.assertFalse(
 
790
            btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
 
791
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
 
792
        self.assertTrue(
 
793
            btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
 
794
            btree_index.BTreeGraphIndex(transport2, 'index', 20))
 
795
        self.assertTrue(
 
796
            btree_index.BTreeGraphIndex(transport1, 'inde1', 20) !=
 
797
            btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
 
798
        self.assertTrue(
 
799
            btree_index.BTreeGraphIndex(transport1, 'index', 10) !=
 
800
            btree_index.BTreeGraphIndex(transport1, 'index', 20))
830
801
 
831
802
    def test_iter_all_only_root_no_size(self):
832
803
        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[:]
 
804
        trans = get_transport('trace+' + self.get_url(''))
 
805
        index = btree_index.BTreeGraphIndex(trans, 'index', None)
 
806
        del trans._activity[:]
836
807
        self.assertEqual([(('key',), 'value')],
837
808
                         [x[1:] for x in index.iter_all_entries()])
838
 
        self.assertEqual([('get', 'index')], t._activity)
 
809
        self.assertEqual([('get', 'index')], trans._activity)
839
810
 
840
811
    def test_iter_all_entries_reads(self):
841
812
        # iterating all entries reads the header, then does a linear
847
818
        nodes = self.make_nodes(10000, 2, 2)
848
819
        for node in nodes:
849
820
            builder.add_node(*node)
850
 
        t = transport.get_transport('trace+' + self.get_url(''))
851
 
        size = t.put_file('index', builder.finish())
 
821
        transport = get_transport('trace+' + self.get_url(''))
 
822
        size = transport.put_file('index', builder.finish())
852
823
        self.assertEqual(1303220, size, 'number of expected bytes in the'
853
824
                                        ' output changed')
854
825
        page_size = btree_index._PAGE_SIZE
855
826
        del builder
856
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
857
 
        del t._activity[:]
858
 
        self.assertEqual([], t._activity)
 
827
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
828
        del transport._activity[:]
 
829
        self.assertEqual([], transport._activity)
859
830
        found_nodes = self.time(list, index.iter_all_entries())
860
831
        bare_nodes = []
861
832
        for node in found_nodes:
882
853
        readv_request[-1] = (readv_request[-1][0], 1303220 % page_size)
883
854
        expected = [('readv', 'index', [(0, page_size)], False, None),
884
855
             ('readv',  'index', readv_request, False, None)]
885
 
        if expected != t._activity:
 
856
        if expected != transport._activity:
886
857
            self.assertEqualDiff(pprint.pformat(expected),
887
858
                                 pprint.pformat(transport._activity))
888
859
 
902
873
        nodes = self.make_nodes(160, 2, 2)
903
874
        for node in nodes:
904
875
            builder.add_node(*node)
905
 
        t = transport.get_transport('trace+' + self.get_url(''))
906
 
        size = t.put_file('index', builder.finish())
 
876
        transport = get_transport('trace+' + self.get_url(''))
 
877
        size = transport.put_file('index', builder.finish())
907
878
        del builder
908
 
        index = btree_index.BTreeGraphIndex(t, 'index', size)
909
 
        del t._activity[:]
910
 
        self.assertEqual([], t._activity)
 
879
        index = btree_index.BTreeGraphIndex(transport, 'index', size)
 
880
        del transport._activity[:]
 
881
        self.assertEqual([], transport._activity)
911
882
        # search for one key
912
883
        found_nodes = list(index.iter_entries([nodes[30][0]]))
913
884
        bare_nodes = []
921
892
        # Should have read the root node, then one leaf page:
922
893
        self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
923
894
             ('readv',  'index', [(8192, 4096), ], False, None)],
924
 
            t._activity)
 
895
            transport._activity)
925
896
 
926
897
    def test_iter_key_prefix_1_element_key_None(self):
927
898
        index = self.make_index()
1009
980
            ])
1010
981
        self.assertEqual(set([]), index.external_references(0))
1011
982
 
1012
 
    def test__find_ancestors_one_page(self):
1013
 
        key1 = ('key-1',)
1014
 
        key2 = ('key-2',)
1015
 
        index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1016
 
            (key1, 'value', ([key2],)),
1017
 
            (key2, 'value', ([],)),
1018
 
            ])
1019
 
        parent_map = {}
1020
 
        missing_keys = set()
1021
 
        search_keys = index._find_ancestors([key1], 0, parent_map, missing_keys)
1022
 
        self.assertEqual({key1: (key2,), key2: ()}, parent_map)
1023
 
        self.assertEqual(set(), missing_keys)
1024
 
        self.assertEqual(set(), search_keys)
1025
 
 
1026
 
    def test__find_ancestors_one_page_w_missing(self):
1027
 
        key1 = ('key-1',)
1028
 
        key2 = ('key-2',)
1029
 
        key3 = ('key-3',)
1030
 
        index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1031
 
            (key1, 'value', ([key2],)),
1032
 
            (key2, 'value', ([],)),
1033
 
            ])
1034
 
        parent_map = {}
1035
 
        missing_keys = set()
1036
 
        search_keys = index._find_ancestors([key2, key3], 0, parent_map,
1037
 
                                            missing_keys)
1038
 
        self.assertEqual({key2: ()}, parent_map)
1039
 
        # we know that key3 is missing because we read the page that it would
1040
 
        # otherwise be on
1041
 
        self.assertEqual(set([key3]), missing_keys)
1042
 
        self.assertEqual(set(), search_keys)
1043
 
 
1044
 
    def test__find_ancestors_one_parent_missing(self):
1045
 
        key1 = ('key-1',)
1046
 
        key2 = ('key-2',)
1047
 
        key3 = ('key-3',)
1048
 
        index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1049
 
            (key1, 'value', ([key2],)),
1050
 
            (key2, 'value', ([key3],)),
1051
 
            ])
1052
 
        parent_map = {}
1053
 
        missing_keys = set()
1054
 
        search_keys = index._find_ancestors([key1], 0, parent_map,
1055
 
                                            missing_keys)
1056
 
        self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1057
 
        self.assertEqual(set(), missing_keys)
1058
 
        # all we know is that key3 wasn't present on the page we were reading
1059
 
        # but if you look, the last key is key2 which comes before key3, so we
1060
 
        # don't know whether key3 would land on this page or not.
1061
 
        self.assertEqual(set([key3]), search_keys)
1062
 
        search_keys = index._find_ancestors(search_keys, 0, parent_map,
1063
 
                                            missing_keys)
1064
 
        # passing it back in, we are sure it is 'missing'
1065
 
        self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1066
 
        self.assertEqual(set([key3]), missing_keys)
1067
 
        self.assertEqual(set([]), search_keys)
1068
 
 
1069
 
    def test__find_ancestors_dont_search_known(self):
1070
 
        key1 = ('key-1',)
1071
 
        key2 = ('key-2',)
1072
 
        key3 = ('key-3',)
1073
 
        index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1074
 
            (key1, 'value', ([key2],)),
1075
 
            (key2, 'value', ([key3],)),
1076
 
            (key3, 'value', ([],)),
1077
 
            ])
1078
 
        # We already know about key2, so we won't try to search for key3
1079
 
        parent_map = {key2: (key3,)}
1080
 
        missing_keys = set()
1081
 
        search_keys = index._find_ancestors([key1], 0, parent_map,
1082
 
                                            missing_keys)
1083
 
        self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1084
 
        self.assertEqual(set(), missing_keys)
1085
 
        self.assertEqual(set(), search_keys)
1086
 
 
1087
 
    def test__find_ancestors_multiple_pages(self):
1088
 
        # We need to use enough keys that we actually cause a split
1089
 
        start_time = 1249671539
1090
 
        email = "joebob@example.com"
1091
 
        nodes = []
1092
 
        ref_lists = ((),)
1093
 
        rev_keys = []
1094
 
        for i in xrange(400):
1095
 
            rev_id = '%s-%s-%s' % (email,
1096
 
                                   osutils.compact_date(start_time + i),
1097
 
                                   osutils.rand_chars(16))
1098
 
            rev_key = (rev_id,)
1099
 
            nodes.append((rev_key, 'value', ref_lists))
1100
 
            # We have a ref 'list' of length 1, with a list of parents, with 1
1101
 
            # parent which is a key
1102
 
            ref_lists = ((rev_key,),)
1103
 
            rev_keys.append(rev_key)
1104
 
        index = self.make_index(ref_lists=1, key_elements=1, nodes=nodes)
1105
 
        self.assertEqual(400, index.key_count())
1106
 
        self.assertEqual(3, len(index._row_offsets))
1107
 
        nodes = dict(index._read_nodes([1, 2]))
1108
 
        l1 = nodes[1]
1109
 
        l2 = nodes[2]
1110
 
        min_l2_key = l2.min_key
1111
 
        max_l1_key = l1.max_key
1112
 
        self.assertTrue(max_l1_key < min_l2_key)
1113
 
        parents_min_l2_key = l2[min_l2_key][1][0]
1114
 
        self.assertEqual((l1.max_key,), parents_min_l2_key)
1115
 
        # Now, whatever key we select that would fall on the second page,
1116
 
        # should give us all the parents until the page break
1117
 
        key_idx = rev_keys.index(min_l2_key)
1118
 
        next_key = rev_keys[key_idx+1]
1119
 
        # So now when we get the parent map, we should get the key we are
1120
 
        # looking for, min_l2_key, and then a reference to go look for the
1121
 
        # parent of that key
1122
 
        parent_map = {}
1123
 
        missing_keys = set()
1124
 
        search_keys = index._find_ancestors([next_key], 0, parent_map,
1125
 
                                            missing_keys)
1126
 
        self.assertEqual([min_l2_key, next_key], sorted(parent_map))
1127
 
        self.assertEqual(set(), missing_keys)
1128
 
        self.assertEqual(set([max_l1_key]), search_keys)
1129
 
        parent_map = {}
1130
 
        search_keys = index._find_ancestors([max_l1_key], 0, parent_map,
1131
 
                                            missing_keys)
1132
 
        self.assertEqual(l1.all_keys(), sorted(parent_map))
1133
 
        self.assertEqual(set(), missing_keys)
1134
 
        self.assertEqual(set(), search_keys)
1135
 
 
1136
 
    def test__find_ancestors_empty_index(self):
1137
 
        index = self.make_index(ref_lists=1, key_elements=1, nodes=[])
1138
 
        parent_map = {}
1139
 
        missing_keys = set()
1140
 
        search_keys = index._find_ancestors([('one',), ('two',)], 0, parent_map,
1141
 
                                            missing_keys)
1142
 
        self.assertEqual(set(), search_keys)
1143
 
        self.assertEqual({}, parent_map)
1144
 
        self.assertEqual(set([('one',), ('two',)]), missing_keys)
1145
 
 
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
983
 
1184
984
class TestBTreeNodes(BTreeTestCase):
1185
985
 
1186
 
    scenarios = btreeparser_scenarios()
 
986
    def restore_parser(self):
 
987
        btree_index._btree_serializer = self.saved_parser
1187
988
 
1188
989
    def setUp(self):
1189
990
        BTreeTestCase.setUp(self)
1190
 
        self.overrideAttr(btree_index, '_btree_serializer', self.parse_btree)
 
991
        self.saved_parser = btree_index._btree_serializer
 
992
        self.addCleanup(self.restore_parser)
 
993
        btree_index._btree_serializer = self.parse_btree
1191
994
 
1192
995
    def test_LeafNode_1_0(self):
1193
996
        node_bytes = ("type=leaf\n"
1205
1008
            ("2222222222222222222222222222222222222222",): ("value:2", ()),
1206
1009
            ("3333333333333333333333333333333333333333",): ("value:3", ()),
1207
1010
            ("4444444444444444444444444444444444444444",): ("value:4", ()),
1208
 
            }, dict(node.all_items()))
 
1011
            }, node.keys)
1209
1012
 
1210
1013
    def test_LeafNode_2_2(self):
1211
1014
        node_bytes = ("type=leaf\n"
1225
1028
            ('11', '33'): ('value:3',
1226
1029
                ((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
1227
1030
            ('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
1228
 
            }, dict(node.all_items()))
 
1031
            }, node.keys)
1229
1032
 
1230
1033
    def test_InternalNode_1(self):
1231
1034
        node_bytes = ("type=internal\n"
1266
1069
            ('11', '33'): ('value:3',
1267
1070
                ((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
1268
1071
            ('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
1269
 
            }, dict(node.all_items()))
 
1072
            }, node.keys)
1270
1073
 
1271
1074
    def assertFlattened(self, expected, key, value, refs):
1272
1075
        flat_key, flat_line = self.parse_btree._flatten_node(
1304
1107
    def test_exists(self):
1305
1108
        # This is just to let the user know if they don't have the feature
1306
1109
        # available
1307
 
        self.requireFeature(compiled_btreeparser_feature)
 
1110
        self.requireFeature(CompiledBtreeParserFeature)
1308
1111
 
1309
1112
 
1310
1113
class TestMultiBisectRight(tests.TestCase):
1350
1153
        This doesn't actually create anything on disk, it just primes a
1351
1154
        BTreeGraphIndex with the recommended information.
1352
1155
        """
1353
 
        index = btree_index.BTreeGraphIndex(
1354
 
            transport.get_transport('memory:///'), 'test-index', size=size)
 
1156
        index = btree_index.BTreeGraphIndex(get_transport('memory:///'),
 
1157
                                            'test-index', size=size)
1355
1158
        if recommended_pages is not None:
1356
1159
            index._recommended_pages = recommended_pages
1357
1160
        return index