1
# Copyright (C) 2008, 2009 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18
"""Tests for btree indices."""
31
from bzrlib.tests import (
32
TestCaseWithTransport,
35
split_suite_by_condition,
37
from bzrlib.transport import get_transport
40
def load_tests(standard_tests, module, loader):
41
# parameterise the TestBTreeNodes tests
42
node_tests, others = split_suite_by_condition(standard_tests,
43
condition_isinstance(TestBTreeNodes))
44
import bzrlib._btree_serializer_py as py_module
45
scenarios = [('python', {'parse_btree': py_module})]
46
if CompiledBtreeParserFeature.available():
47
# Is there a way to do this that gets missing feature failures rather
48
# than no indication to the user?
49
import bzrlib._btree_serializer_pyx as c_module
50
scenarios.append(('C', {'parse_btree': c_module}))
51
return multiply_tests(node_tests, scenarios, others)
54
class _CompiledBtreeParserFeature(tests.Feature):
57
import bzrlib._btree_serializer_pyx
62
def feature_name(self):
63
return 'bzrlib._btree_serializer_pyx'
65
CompiledBtreeParserFeature = _CompiledBtreeParserFeature()
68
class BTreeTestCase(TestCaseWithTransport):
69
# test names here are suffixed by the key length and reference list count
73
TestCaseWithTransport.setUp(self)
74
self._original_header = btree_index._RESERVED_HEADER_BYTES
76
btree_index._RESERVED_HEADER_BYTES = self._original_header
77
self.addCleanup(restore)
78
btree_index._RESERVED_HEADER_BYTES = 100
80
def make_nodes(self, count, key_elements, reference_lists):
81
"""Generate count*key_elements sample nodes."""
83
for prefix_pos in range(key_elements):
85
prefix = (str(prefix_pos) * 40,)
88
for pos in xrange(count):
89
# TODO: This creates odd keys. When count == 100,000, it
90
# creates a 240 byte key
91
key = prefix + (str(pos) * 40,)
92
value = "value:%s" % pos
94
# generate some references
96
for list_pos in range(reference_lists):
97
# as many keys in each list as its index + the key depth
98
# mod 2 - this generates both 0 length lists and
99
# ones slightly longer than the number of lists.
100
# It also ensures we have non homogeneous lists.
102
for ref_pos in range(list_pos + pos % 2):
104
# refer to a nearby key
105
refs[-1].append(prefix + ("ref" + str(pos - 1) * 40,))
107
# serial of this ref in the ref list
108
refs[-1].append(prefix + ("ref" + str(ref_pos) * 40,))
109
refs[-1] = tuple(refs[-1])
113
keys.append((key, value, refs))
116
def shrink_page_size(self):
117
"""Shrink the default page size so that less fits in a page."""
118
old_page_size = btree_index._PAGE_SIZE
120
btree_index._PAGE_SIZE = old_page_size
121
self.addCleanup(cleanup)
122
btree_index._PAGE_SIZE = 2048
125
class TestBTreeBuilder(BTreeTestCase):
127
def test_clear_cache(self):
128
builder = btree_index.BTreeBuilder(reference_lists=0, key_elements=1)
129
# This is a no-op, but we need the api to be consistent with other
130
# BTreeGraphIndex apis.
131
builder.clear_cache()
133
def test_empty_1_0(self):
134
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
135
# NamedTemporaryFile dies on builder.finish().read(). weird.
136
temp_file = builder.finish()
137
content = temp_file.read()
140
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=0\n"
144
def test_empty_2_1(self):
145
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=1)
146
# NamedTemporaryFile dies on builder.finish().read(). weird.
147
temp_file = builder.finish()
148
content = temp_file.read()
151
"B+Tree Graph Index 2\nnode_ref_lists=1\nkey_elements=2\nlen=0\n"
155
def test_root_leaf_1_0(self):
156
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
157
nodes = self.make_nodes(5, 1, 0)
159
builder.add_node(*node)
160
# NamedTemporaryFile dies on builder.finish().read(). weird.
161
temp_file = builder.finish()
162
content = temp_file.read()
164
self.assertEqual(131, len(content))
166
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=5\n"
169
node_content = content[73:]
170
node_bytes = zlib.decompress(node_content)
171
expected_node = ("type=leaf\n"
172
"0000000000000000000000000000000000000000\x00\x00value:0\n"
173
"1111111111111111111111111111111111111111\x00\x00value:1\n"
174
"2222222222222222222222222222222222222222\x00\x00value:2\n"
175
"3333333333333333333333333333333333333333\x00\x00value:3\n"
176
"4444444444444444444444444444444444444444\x00\x00value:4\n")
177
self.assertEqual(expected_node, node_bytes)
179
def test_root_leaf_2_2(self):
180
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
181
nodes = self.make_nodes(5, 2, 2)
183
builder.add_node(*node)
184
# NamedTemporaryFile dies on builder.finish().read(). weird.
185
temp_file = builder.finish()
186
content = temp_file.read()
188
self.assertEqual(238, len(content))
190
"B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=10\n"
193
node_content = content[74:]
194
node_bytes = zlib.decompress(node_content)
197
"0000000000000000000000000000000000000000\x000000000000000000000000000000000000000000\x00\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:0\n"
198
"0000000000000000000000000000000000000000\x001111111111111111111111111111111111111111\x000000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\r0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:1\n"
199
"0000000000000000000000000000000000000000\x002222222222222222222222222222222222222222\x00\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:2\n"
200
"0000000000000000000000000000000000000000\x003333333333333333333333333333333333333333\x000000000000000000000000000000000000000000\x00ref2222222222222222222222222222222222222222\t0000000000000000000000000000000000000000\x00ref2222222222222222222222222222222222222222\r0000000000000000000000000000000000000000\x00ref2222222222222222222222222222222222222222\x00value:3\n"
201
"0000000000000000000000000000000000000000\x004444444444444444444444444444444444444444\x00\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:4\n"
202
"1111111111111111111111111111111111111111\x000000000000000000000000000000000000000000\x00\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:0\n"
203
"1111111111111111111111111111111111111111\x001111111111111111111111111111111111111111\x001111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\r1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:1\n"
204
"1111111111111111111111111111111111111111\x002222222222222222222222222222222222222222\x00\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:2\n"
205
"1111111111111111111111111111111111111111\x003333333333333333333333333333333333333333\x001111111111111111111111111111111111111111\x00ref2222222222222222222222222222222222222222\t1111111111111111111111111111111111111111\x00ref2222222222222222222222222222222222222222\r1111111111111111111111111111111111111111\x00ref2222222222222222222222222222222222222222\x00value:3\n"
206
"1111111111111111111111111111111111111111\x004444444444444444444444444444444444444444\x00\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:4\n"
209
self.assertEqual(expected_node, node_bytes)
211
def test_2_leaves_1_0(self):
212
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
213
nodes = self.make_nodes(400, 1, 0)
215
builder.add_node(*node)
216
# NamedTemporaryFile dies on builder.finish().read(). weird.
217
temp_file = builder.finish()
218
content = temp_file.read()
220
self.assertEqual(9283, len(content))
222
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=400\n"
225
root = content[77:4096]
226
leaf1 = content[4096:8192]
227
leaf2 = content[8192:]
228
root_bytes = zlib.decompress(root)
232
) + ("307" * 40) + "\n"
233
self.assertEqual(expected_root, root_bytes)
234
# We already know serialisation works for leaves, check key selection:
235
leaf1_bytes = zlib.decompress(leaf1)
236
sorted_node_keys = sorted(node[0] for node in nodes)
237
node = btree_index._LeafNode(leaf1_bytes, 1, 0)
238
self.assertEqual(231, len(node.keys))
239
self.assertEqual(sorted_node_keys[:231], sorted(node.keys))
240
leaf2_bytes = zlib.decompress(leaf2)
241
node = btree_index._LeafNode(leaf2_bytes, 1, 0)
242
self.assertEqual(400 - 231, len(node.keys))
243
self.assertEqual(sorted_node_keys[231:], sorted(node.keys))
245
def test_last_page_rounded_1_layer(self):
246
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
247
nodes = self.make_nodes(10, 1, 0)
249
builder.add_node(*node)
250
# NamedTemporaryFile dies on builder.finish().read(). weird.
251
temp_file = builder.finish()
252
content = temp_file.read()
254
self.assertEqual(155, len(content))
256
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=10\n"
259
# Check thelast page is well formed
261
leaf2_bytes = zlib.decompress(leaf2)
262
node = btree_index._LeafNode(leaf2_bytes, 1, 0)
263
self.assertEqual(10, len(node.keys))
264
sorted_node_keys = sorted(node[0] for node in nodes)
265
self.assertEqual(sorted_node_keys, sorted(node.keys))
267
def test_last_page_not_rounded_2_layer(self):
268
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
269
nodes = self.make_nodes(400, 1, 0)
271
builder.add_node(*node)
272
# NamedTemporaryFile dies on builder.finish().read(). weird.
273
temp_file = builder.finish()
274
content = temp_file.read()
276
self.assertEqual(9283, len(content))
278
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=400\n"
281
# Check the last page is well formed
282
leaf2 = content[8192:]
283
leaf2_bytes = zlib.decompress(leaf2)
284
node = btree_index._LeafNode(leaf2_bytes, 1, 0)
285
self.assertEqual(400 - 231, len(node.keys))
286
sorted_node_keys = sorted(node[0] for node in nodes)
287
self.assertEqual(sorted_node_keys[231:], sorted(node.keys))
289
def test_three_level_tree_details(self):
290
# The left most pointer in the second internal node in a row should
291
# pointer to the second node that the internal node is for, _not_
292
# the first, otherwise the first node overlaps with the last node of
293
# the prior internal node on that row.
294
self.shrink_page_size()
295
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
296
# 40K nodes is enough to create a two internal nodes on the second
297
# level, with a 2K page size
298
nodes = self.make_nodes(20000, 2, 2)
301
builder.add_node(*node)
302
transport = get_transport('trace+' + self.get_url(''))
303
size = transport.put_file('index', self.time(builder.finish))
305
index = btree_index.BTreeGraphIndex(transport, 'index', size)
306
# Seed the metadata, we're using internal calls now.
308
self.assertEqual(3, len(index._row_lengths),
309
"Not enough rows: %r" % index._row_lengths)
310
self.assertEqual(4, len(index._row_offsets))
311
self.assertEqual(sum(index._row_lengths), index._row_offsets[-1])
312
internal_nodes = index._get_internal_nodes([0, 1, 2])
313
root_node = internal_nodes[0]
314
internal_node1 = internal_nodes[1]
315
internal_node2 = internal_nodes[2]
316
# The left most node node2 points at should be one after the right most
317
# node pointed at by node1.
318
self.assertEqual(internal_node2.offset, 1 + len(internal_node1.keys))
319
# The left most key of the second node pointed at by internal_node2
320
# should be its first key. We can check this by looking for its first key
321
# in the second node it points at
322
pos = index._row_offsets[2] + internal_node2.offset + 1
323
leaf = index._get_leaf_nodes([pos])[pos]
324
self.assertTrue(internal_node2.keys[0] in leaf.keys)
326
def test_2_leaves_2_2(self):
327
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
328
nodes = self.make_nodes(100, 2, 2)
330
builder.add_node(*node)
331
# NamedTemporaryFile dies on builder.finish().read(). weird.
332
temp_file = builder.finish()
333
content = temp_file.read()
335
self.assertEqual(12643, len(content))
337
"B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=200\n"
340
root = content[77:4096]
341
leaf1 = content[4096:8192]
342
leaf2 = content[8192:12288]
343
leaf3 = content[12288:]
344
root_bytes = zlib.decompress(root)
348
+ ("0" * 40) + "\x00" + ("91" * 40) + "\n"
349
+ ("1" * 40) + "\x00" + ("81" * 40) + "\n"
351
self.assertEqual(expected_root, root_bytes)
352
# We assume the other leaf nodes have been written correctly - layering
355
def test_spill_index_stress_1_1(self):
356
builder = btree_index.BTreeBuilder(key_elements=1, spill_at=2)
357
nodes = [node[0:2] for node in self.make_nodes(16, 1, 0)]
358
builder.add_node(*nodes[0])
359
# Test the parts of the index that take up memory are doing so
361
self.assertEqual(1, len(builder._nodes))
362
self.assertIs(None, builder._nodes_by_key)
363
builder.add_node(*nodes[1])
364
self.assertEqual(0, len(builder._nodes))
365
self.assertIs(None, builder._nodes_by_key)
366
self.assertEqual(1, len(builder._backing_indices))
367
self.assertEqual(2, builder._backing_indices[0].key_count())
369
builder.add_node(*nodes[2])
370
self.assertEqual(1, len(builder._nodes))
371
self.assertIs(None, builder._nodes_by_key)
372
# And spills to a second backing index combing all
373
builder.add_node(*nodes[3])
374
self.assertEqual(0, len(builder._nodes))
375
self.assertIs(None, builder._nodes_by_key)
376
self.assertEqual(2, len(builder._backing_indices))
377
self.assertEqual(None, builder._backing_indices[0])
378
self.assertEqual(4, builder._backing_indices[1].key_count())
379
# The next spills to the 2-len slot
380
builder.add_node(*nodes[4])
381
builder.add_node(*nodes[5])
382
self.assertEqual(0, len(builder._nodes))
383
self.assertIs(None, builder._nodes_by_key)
384
self.assertEqual(2, len(builder._backing_indices))
385
self.assertEqual(2, builder._backing_indices[0].key_count())
386
self.assertEqual(4, builder._backing_indices[1].key_count())
387
# Next spill combines
388
builder.add_node(*nodes[6])
389
builder.add_node(*nodes[7])
390
self.assertEqual(3, len(builder._backing_indices))
391
self.assertEqual(None, builder._backing_indices[0])
392
self.assertEqual(None, builder._backing_indices[1])
393
self.assertEqual(8, builder._backing_indices[2].key_count())
394
# And so forth - counting up in binary.
395
builder.add_node(*nodes[8])
396
builder.add_node(*nodes[9])
397
self.assertEqual(3, len(builder._backing_indices))
398
self.assertEqual(2, builder._backing_indices[0].key_count())
399
self.assertEqual(None, builder._backing_indices[1])
400
self.assertEqual(8, builder._backing_indices[2].key_count())
401
builder.add_node(*nodes[10])
402
builder.add_node(*nodes[11])
403
self.assertEqual(3, len(builder._backing_indices))
404
self.assertEqual(None, builder._backing_indices[0])
405
self.assertEqual(4, builder._backing_indices[1].key_count())
406
self.assertEqual(8, builder._backing_indices[2].key_count())
407
builder.add_node(*nodes[12])
408
# Test that memory and disk are both used for query methods; and that
409
# None is skipped over happily.
410
self.assertEqual([(builder,) + node for node in sorted(nodes[:13])],
411
list(builder.iter_all_entries()))
412
# Two nodes - one memory one disk
413
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
414
set(builder.iter_entries([nodes[12][0], nodes[11][0]])))
415
self.assertEqual(13, builder.key_count())
416
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
417
set(builder.iter_entries_prefix([nodes[12][0], nodes[11][0]])))
418
builder.add_node(*nodes[13])
419
self.assertEqual(3, len(builder._backing_indices))
420
self.assertEqual(2, builder._backing_indices[0].key_count())
421
self.assertEqual(4, builder._backing_indices[1].key_count())
422
self.assertEqual(8, builder._backing_indices[2].key_count())
423
builder.add_node(*nodes[14])
424
builder.add_node(*nodes[15])
425
self.assertEqual(4, len(builder._backing_indices))
426
self.assertEqual(None, builder._backing_indices[0])
427
self.assertEqual(None, builder._backing_indices[1])
428
self.assertEqual(None, builder._backing_indices[2])
429
self.assertEqual(16, builder._backing_indices[3].key_count())
430
# Now finish, and check we got a correctly ordered tree
431
transport = self.get_transport('')
432
size = transport.put_file('index', builder.finish())
433
index = btree_index.BTreeGraphIndex(transport, 'index', size)
434
nodes = list(index.iter_all_entries())
435
self.assertEqual(sorted(nodes), nodes)
436
self.assertEqual(16, len(nodes))
438
def test_spill_index_stress_1_1_no_combine(self):
439
builder = btree_index.BTreeBuilder(key_elements=1, spill_at=2)
440
builder.set_optimize(for_size=False, combine_backing_indices=False)
441
nodes = [node[0:2] for node in self.make_nodes(16, 1, 0)]
442
builder.add_node(*nodes[0])
443
# Test the parts of the index that take up memory are doing so
445
self.assertEqual(1, len(builder._nodes))
446
self.assertIs(None, builder._nodes_by_key)
447
builder.add_node(*nodes[1])
448
self.assertEqual(0, len(builder._nodes))
449
self.assertIs(None, builder._nodes_by_key)
450
self.assertEqual(1, len(builder._backing_indices))
451
self.assertEqual(2, builder._backing_indices[0].key_count())
453
builder.add_node(*nodes[2])
454
self.assertEqual(1, len(builder._nodes))
455
self.assertIs(None, builder._nodes_by_key)
456
# And spills to a second backing index but doesn't combine
457
builder.add_node(*nodes[3])
458
self.assertEqual(0, len(builder._nodes))
459
self.assertIs(None, builder._nodes_by_key)
460
self.assertEqual(2, len(builder._backing_indices))
461
for backing_index in builder._backing_indices:
462
self.assertEqual(2, backing_index.key_count())
463
# The next spills to the 3rd slot
464
builder.add_node(*nodes[4])
465
builder.add_node(*nodes[5])
466
self.assertEqual(0, len(builder._nodes))
467
self.assertIs(None, builder._nodes_by_key)
468
self.assertEqual(3, len(builder._backing_indices))
469
for backing_index in builder._backing_indices:
470
self.assertEqual(2, backing_index.key_count())
471
# Now spill a few more, and check that we don't combine
472
builder.add_node(*nodes[6])
473
builder.add_node(*nodes[7])
474
builder.add_node(*nodes[8])
475
builder.add_node(*nodes[9])
476
builder.add_node(*nodes[10])
477
builder.add_node(*nodes[11])
478
builder.add_node(*nodes[12])
479
self.assertEqual(6, len(builder._backing_indices))
480
for backing_index in builder._backing_indices:
481
self.assertEqual(2, backing_index.key_count())
482
# Test that memory and disk are both used for query methods; and that
483
# None is skipped over happily.
484
self.assertEqual([(builder,) + node for node in sorted(nodes[:13])],
485
list(builder.iter_all_entries()))
486
# Two nodes - one memory one disk
487
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
488
set(builder.iter_entries([nodes[12][0], nodes[11][0]])))
489
self.assertEqual(13, builder.key_count())
490
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
491
set(builder.iter_entries_prefix([nodes[12][0], nodes[11][0]])))
492
builder.add_node(*nodes[13])
493
builder.add_node(*nodes[14])
494
builder.add_node(*nodes[15])
495
self.assertEqual(8, len(builder._backing_indices))
496
for backing_index in builder._backing_indices:
497
self.assertEqual(2, backing_index.key_count())
498
# Now finish, and check we got a correctly ordered tree
499
transport = self.get_transport('')
500
size = transport.put_file('index', builder.finish())
501
index = btree_index.BTreeGraphIndex(transport, 'index', size)
502
nodes = list(index.iter_all_entries())
503
self.assertEqual(sorted(nodes), nodes)
504
self.assertEqual(16, len(nodes))
506
def test_set_optimize(self):
507
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
508
builder.set_optimize(for_size=True)
509
self.assertTrue(builder._optimize_for_size)
510
builder.set_optimize(for_size=False)
511
self.assertFalse(builder._optimize_for_size)
512
# test that we can set combine_backing_indices without effecting
515
builder._optimize_for_size = obj
516
builder.set_optimize(combine_backing_indices=False)
517
self.assertFalse(builder._combine_backing_indices)
518
self.assertIs(obj, builder._optimize_for_size)
519
builder.set_optimize(combine_backing_indices=True)
520
self.assertTrue(builder._combine_backing_indices)
521
self.assertIs(obj, builder._optimize_for_size)
523
def test_spill_index_stress_2_2(self):
524
# test that references and longer keys don't confuse things.
525
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2,
527
nodes = self.make_nodes(16, 2, 2)
528
builder.add_node(*nodes[0])
529
# Test the parts of the index that take up memory are doing so
531
self.assertEqual(1, len(builder._nodes))
532
self.assertIs(None, builder._nodes_by_key)
533
builder.add_node(*nodes[1])
534
self.assertEqual(0, len(builder._nodes))
535
self.assertIs(None, builder._nodes_by_key)
536
self.assertEqual(1, len(builder._backing_indices))
537
self.assertEqual(2, builder._backing_indices[0].key_count())
539
old = dict(builder._get_nodes_by_key()) #Build up the nodes by key dict
540
builder.add_node(*nodes[2])
541
self.assertEqual(1, len(builder._nodes))
542
self.assertIsNot(None, builder._nodes_by_key)
543
self.assertNotEqual({}, builder._nodes_by_key)
544
# We should have a new entry
545
self.assertNotEqual(old, builder._nodes_by_key)
546
# And spills to a second backing index combing all
547
builder.add_node(*nodes[3])
548
self.assertEqual(0, len(builder._nodes))
549
self.assertIs(None, builder._nodes_by_key)
550
self.assertEqual(2, len(builder._backing_indices))
551
self.assertEqual(None, builder._backing_indices[0])
552
self.assertEqual(4, builder._backing_indices[1].key_count())
553
# The next spills to the 2-len slot
554
builder.add_node(*nodes[4])
555
builder.add_node(*nodes[5])
556
self.assertEqual(0, len(builder._nodes))
557
self.assertIs(None, builder._nodes_by_key)
558
self.assertEqual(2, len(builder._backing_indices))
559
self.assertEqual(2, builder._backing_indices[0].key_count())
560
self.assertEqual(4, builder._backing_indices[1].key_count())
561
# Next spill combines
562
builder.add_node(*nodes[6])
563
builder.add_node(*nodes[7])
564
self.assertEqual(3, len(builder._backing_indices))
565
self.assertEqual(None, builder._backing_indices[0])
566
self.assertEqual(None, builder._backing_indices[1])
567
self.assertEqual(8, builder._backing_indices[2].key_count())
568
# And so forth - counting up in binary.
569
builder.add_node(*nodes[8])
570
builder.add_node(*nodes[9])
571
self.assertEqual(3, len(builder._backing_indices))
572
self.assertEqual(2, builder._backing_indices[0].key_count())
573
self.assertEqual(None, builder._backing_indices[1])
574
self.assertEqual(8, builder._backing_indices[2].key_count())
575
builder.add_node(*nodes[10])
576
builder.add_node(*nodes[11])
577
self.assertEqual(3, len(builder._backing_indices))
578
self.assertEqual(None, builder._backing_indices[0])
579
self.assertEqual(4, builder._backing_indices[1].key_count())
580
self.assertEqual(8, builder._backing_indices[2].key_count())
581
builder.add_node(*nodes[12])
582
# Test that memory and disk are both used for query methods; and that
583
# None is skipped over happily.
584
self.assertEqual([(builder,) + node for node in sorted(nodes[:13])],
585
list(builder.iter_all_entries()))
586
# Two nodes - one memory one disk
587
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
588
set(builder.iter_entries([nodes[12][0], nodes[11][0]])))
589
self.assertEqual(13, builder.key_count())
590
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
591
set(builder.iter_entries_prefix([nodes[12][0], nodes[11][0]])))
592
builder.add_node(*nodes[13])
593
self.assertEqual(3, len(builder._backing_indices))
594
self.assertEqual(2, builder._backing_indices[0].key_count())
595
self.assertEqual(4, builder._backing_indices[1].key_count())
596
self.assertEqual(8, builder._backing_indices[2].key_count())
597
builder.add_node(*nodes[14])
598
builder.add_node(*nodes[15])
599
self.assertEqual(4, len(builder._backing_indices))
600
self.assertEqual(None, builder._backing_indices[0])
601
self.assertEqual(None, builder._backing_indices[1])
602
self.assertEqual(None, builder._backing_indices[2])
603
self.assertEqual(16, builder._backing_indices[3].key_count())
604
# Now finish, and check we got a correctly ordered tree
605
transport = self.get_transport('')
606
size = transport.put_file('index', builder.finish())
607
index = btree_index.BTreeGraphIndex(transport, 'index', size)
608
nodes = list(index.iter_all_entries())
609
self.assertEqual(sorted(nodes), nodes)
610
self.assertEqual(16, len(nodes))
612
def test_spill_index_duplicate_key_caught_on_finish(self):
613
builder = btree_index.BTreeBuilder(key_elements=1, spill_at=2)
614
nodes = [node[0:2] for node in self.make_nodes(16, 1, 0)]
615
builder.add_node(*nodes[0])
616
builder.add_node(*nodes[1])
617
builder.add_node(*nodes[0])
618
self.assertRaises(errors.BadIndexDuplicateKey, builder.finish)
621
class TestBTreeIndex(BTreeTestCase):
623
def make_index(self, ref_lists=0, key_elements=1, nodes=[]):
624
builder = btree_index.BTreeBuilder(reference_lists=ref_lists,
625
key_elements=key_elements)
626
for key, value, references in nodes:
627
builder.add_node(key, value, references)
628
stream = builder.finish()
629
trans = get_transport('trace+' + self.get_url())
630
size = trans.put_file('index', stream)
631
return btree_index.BTreeGraphIndex(trans, 'index', size)
633
def test_clear_cache(self):
634
nodes = self.make_nodes(160, 2, 2)
635
index = self.make_index(ref_lists=2, key_elements=2, nodes=nodes)
636
self.assertEqual(1, len(list(index.iter_entries([nodes[30][0]]))))
637
self.assertEqual([1, 4], index._row_lengths)
638
self.assertIsNot(None, index._root_node)
639
internal_node_pre_clear = index._internal_node_cache.keys()
640
self.assertTrue(len(index._leaf_node_cache) > 0)
642
# We don't touch _root_node or _internal_node_cache, both should be
643
# small, and can save a round trip or two
644
self.assertIsNot(None, index._root_node)
645
# NOTE: We don't want to affect the _internal_node_cache, as we expect
646
# it will be small, and if we ever do touch this index again, it
647
# will save round-trips. This assertion isn't very strong,
648
# becuase without a 3-level index, we don't have any internal
650
self.assertEqual(internal_node_pre_clear,
651
index._internal_node_cache.keys())
652
self.assertEqual(0, len(index._leaf_node_cache))
654
def test_trivial_constructor(self):
655
transport = get_transport('trace+' + self.get_url(''))
656
index = btree_index.BTreeGraphIndex(transport, 'index', None)
657
# Checks the page size at load, but that isn't logged yet.
658
self.assertEqual([], transport._activity)
660
def test_with_size_constructor(self):
661
transport = get_transport('trace+' + self.get_url(''))
662
index = btree_index.BTreeGraphIndex(transport, 'index', 1)
663
# Checks the page size at load, but that isn't logged yet.
664
self.assertEqual([], transport._activity)
666
def test_empty_key_count_no_size(self):
667
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
668
transport = get_transport('trace+' + self.get_url(''))
669
transport.put_file('index', builder.finish())
670
index = btree_index.BTreeGraphIndex(transport, 'index', None)
671
del transport._activity[:]
672
self.assertEqual([], transport._activity)
673
self.assertEqual(0, index.key_count())
674
# The entire index should have been requested (as we generally have the
675
# size available, and doing many small readvs is inappropriate).
676
# We can't tell how much was actually read here, but - check the code.
677
self.assertEqual([('get', 'index')], transport._activity)
679
def test_empty_key_count(self):
680
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
681
transport = get_transport('trace+' + self.get_url(''))
682
size = transport.put_file('index', builder.finish())
683
self.assertEqual(72, size)
684
index = btree_index.BTreeGraphIndex(transport, 'index', size)
685
del transport._activity[:]
686
self.assertEqual([], transport._activity)
687
self.assertEqual(0, index.key_count())
688
# The entire index should have been read, as 4K > size
689
self.assertEqual([('readv', 'index', [(0, 72)], False, None)],
692
def test_non_empty_key_count_2_2(self):
693
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
694
nodes = self.make_nodes(35, 2, 2)
696
builder.add_node(*node)
697
transport = get_transport('trace+' + self.get_url(''))
698
size = transport.put_file('index', builder.finish())
699
index = btree_index.BTreeGraphIndex(transport, 'index', size)
700
del transport._activity[:]
701
self.assertEqual([], transport._activity)
702
self.assertEqual(70, index.key_count())
703
# The entire index should have been read, as it is one page long.
704
self.assertEqual([('readv', 'index', [(0, size)], False, None)],
706
self.assertEqual(1173, size)
708
def test__read_nodes_no_size_one_page_reads_once(self):
709
self.make_index(nodes=[(('key',), 'value', ())])
710
trans = get_transport('trace+' + self.get_url())
711
index = btree_index.BTreeGraphIndex(trans, 'index', None)
712
del trans._activity[:]
713
nodes = dict(index._read_nodes([0]))
714
self.assertEqual([0], nodes.keys())
716
self.assertEqual([('key',)], node.keys.keys())
717
self.assertEqual([('get', 'index')], trans._activity)
719
def test__read_nodes_no_size_multiple_pages(self):
720
index = self.make_index(2, 2, nodes=self.make_nodes(160, 2, 2))
722
num_pages = index._row_offsets[-1]
723
# Reopen with a traced transport and no size
724
trans = get_transport('trace+' + self.get_url())
725
index = btree_index.BTreeGraphIndex(trans, 'index', None)
726
del trans._activity[:]
727
nodes = dict(index._read_nodes([0]))
728
self.assertEqual(range(num_pages), nodes.keys())
730
def test_2_levels_key_count_2_2(self):
731
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
732
nodes = self.make_nodes(160, 2, 2)
734
builder.add_node(*node)
735
transport = get_transport('trace+' + self.get_url(''))
736
size = transport.put_file('index', builder.finish())
737
self.assertEqual(17692, size)
738
index = btree_index.BTreeGraphIndex(transport, 'index', size)
739
del transport._activity[:]
740
self.assertEqual([], transport._activity)
741
self.assertEqual(320, index.key_count())
742
# The entire index should not have been read.
743
self.assertEqual([('readv', 'index', [(0, 4096)], False, None)],
746
def test_validate_one_page(self):
747
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
748
nodes = self.make_nodes(45, 2, 2)
750
builder.add_node(*node)
751
transport = get_transport('trace+' + self.get_url(''))
752
size = transport.put_file('index', builder.finish())
753
index = btree_index.BTreeGraphIndex(transport, 'index', size)
754
del transport._activity[:]
755
self.assertEqual([], transport._activity)
757
# The entire index should have been read linearly.
758
self.assertEqual([('readv', 'index', [(0, size)], False, None)],
760
self.assertEqual(1488, size)
762
def test_validate_two_pages(self):
763
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
764
nodes = self.make_nodes(80, 2, 2)
766
builder.add_node(*node)
767
transport = get_transport('trace+' + self.get_url(''))
768
size = transport.put_file('index', builder.finish())
769
# Root page, 2 leaf pages
770
self.assertEqual(9339, size)
771
index = btree_index.BTreeGraphIndex(transport, 'index', size)
772
del transport._activity[:]
773
self.assertEqual([], transport._activity)
775
# The entire index should have been read linearly.
776
self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
777
('readv', 'index', [(4096, 4096), (8192, 1147)], False, None)],
779
# XXX: TODO: write some badly-ordered nodes, and some pointers-to-wrong
780
# node and make validate find them.
782
def test_eq_ne(self):
783
# two indices are equal when constructed with the same parameters:
784
transport1 = get_transport('trace+' + self.get_url(''))
785
transport2 = get_transport(self.get_url(''))
787
btree_index.BTreeGraphIndex(transport1, 'index', None) ==
788
btree_index.BTreeGraphIndex(transport1, 'index', None))
790
btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
791
btree_index.BTreeGraphIndex(transport1, 'index', 20))
793
btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
794
btree_index.BTreeGraphIndex(transport2, 'index', 20))
796
btree_index.BTreeGraphIndex(transport1, 'inde1', 20) ==
797
btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
799
btree_index.BTreeGraphIndex(transport1, 'index', 10) ==
800
btree_index.BTreeGraphIndex(transport1, 'index', 20))
802
btree_index.BTreeGraphIndex(transport1, 'index', None) !=
803
btree_index.BTreeGraphIndex(transport1, 'index', None))
805
btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
806
btree_index.BTreeGraphIndex(transport1, 'index', 20))
808
btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
809
btree_index.BTreeGraphIndex(transport2, 'index', 20))
811
btree_index.BTreeGraphIndex(transport1, 'inde1', 20) !=
812
btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
814
btree_index.BTreeGraphIndex(transport1, 'index', 10) !=
815
btree_index.BTreeGraphIndex(transport1, 'index', 20))
817
def test_iter_all_only_root_no_size(self):
818
self.make_index(nodes=[(('key',), 'value', ())])
819
trans = get_transport('trace+' + self.get_url(''))
820
index = btree_index.BTreeGraphIndex(trans, 'index', None)
821
del trans._activity[:]
822
self.assertEqual([(('key',), 'value')],
823
[x[1:] for x in index.iter_all_entries()])
824
self.assertEqual([('get', 'index')], trans._activity)
826
def test_iter_all_entries_reads(self):
827
# iterating all entries reads the header, then does a linear
829
self.shrink_page_size()
830
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
831
# 20k nodes is enough to create a two internal nodes on the second
832
# level, with a 2K page size
833
nodes = self.make_nodes(10000, 2, 2)
835
builder.add_node(*node)
836
transport = get_transport('trace+' + self.get_url(''))
837
size = transport.put_file('index', builder.finish())
838
self.assertEqual(1303220, size, 'number of expected bytes in the'
840
page_size = btree_index._PAGE_SIZE
842
index = btree_index.BTreeGraphIndex(transport, 'index', size)
843
del transport._activity[:]
844
self.assertEqual([], transport._activity)
845
found_nodes = self.time(list, index.iter_all_entries())
847
for node in found_nodes:
848
self.assertTrue(node[0] is index)
849
bare_nodes.append(node[1:])
850
self.assertEqual(3, len(index._row_lengths),
851
"Not enough rows: %r" % index._row_lengths)
852
# Should be as long as the nodes we supplied
853
self.assertEqual(20000, len(found_nodes))
854
# Should have the same content
855
self.assertEqual(set(nodes), set(bare_nodes))
856
# Should have done linear scan IO up the index, ignoring
857
# the internal nodes:
858
# The entire index should have been read
859
total_pages = sum(index._row_lengths)
860
self.assertEqual(total_pages, index._row_offsets[-1])
861
self.assertEqual(1303220, size)
862
# The start of the leaves
863
first_byte = index._row_offsets[-2] * page_size
865
for offset in range(first_byte, size, page_size):
866
readv_request.append((offset, page_size))
867
# The last page is truncated
868
readv_request[-1] = (readv_request[-1][0], 1303220 % page_size)
869
expected = [('readv', 'index', [(0, page_size)], False, None),
870
('readv', 'index', readv_request, False, None)]
871
if expected != transport._activity:
872
self.assertEqualDiff(pprint.pformat(expected),
873
pprint.pformat(transport._activity))
875
def _test_iter_entries_references_resolved(self):
876
index = self.make_index(1, nodes=[
877
(('name', ), 'data', ([('ref', ), ('ref', )], )),
878
(('ref', ), 'refdata', ([], ))])
879
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),('ref',)),)),
880
(index, ('ref', ), 'refdata', ((), ))]),
881
set(index.iter_entries([('name',), ('ref',)])))
883
def test_iter_entries_references_2_refs_resolved(self):
884
# iterating some entries reads just the pages needed. For now, to
885
# get it working and start measuring, only 4K pages are read.
886
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
887
# 80 nodes is enough to create a two-level index.
888
nodes = self.make_nodes(160, 2, 2)
890
builder.add_node(*node)
891
transport = get_transport('trace+' + self.get_url(''))
892
size = transport.put_file('index', builder.finish())
894
index = btree_index.BTreeGraphIndex(transport, 'index', size)
895
del transport._activity[:]
896
self.assertEqual([], transport._activity)
898
found_nodes = list(index.iter_entries([nodes[30][0]]))
900
for node in found_nodes:
901
self.assertTrue(node[0] is index)
902
bare_nodes.append(node[1:])
903
# Should be as long as the nodes we supplied
904
self.assertEqual(1, len(found_nodes))
905
# Should have the same content
906
self.assertEqual(nodes[30], bare_nodes[0])
907
# Should have read the root node, then one leaf page:
908
self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
909
('readv', 'index', [(8192, 4096), ], False, None)],
912
def test_iter_key_prefix_1_element_key_None(self):
913
index = self.make_index()
914
self.assertRaises(errors.BadIndexKey, list,
915
index.iter_entries_prefix([(None, )]))
917
def test_iter_key_prefix_wrong_length(self):
918
index = self.make_index()
919
self.assertRaises(errors.BadIndexKey, list,
920
index.iter_entries_prefix([('foo', None)]))
921
index = self.make_index(key_elements=2)
922
self.assertRaises(errors.BadIndexKey, list,
923
index.iter_entries_prefix([('foo', )]))
924
self.assertRaises(errors.BadIndexKey, list,
925
index.iter_entries_prefix([('foo', None, None)]))
927
def test_iter_key_prefix_1_key_element_no_refs(self):
928
index = self.make_index( nodes=[
929
(('name', ), 'data', ()),
930
(('ref', ), 'refdata', ())])
931
self.assertEqual(set([(index, ('name', ), 'data'),
932
(index, ('ref', ), 'refdata')]),
933
set(index.iter_entries_prefix([('name', ), ('ref', )])))
935
def test_iter_key_prefix_1_key_element_refs(self):
936
index = self.make_index(1, nodes=[
937
(('name', ), 'data', ([('ref', )], )),
938
(('ref', ), 'refdata', ([], ))])
939
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
940
(index, ('ref', ), 'refdata', ((), ))]),
941
set(index.iter_entries_prefix([('name', ), ('ref', )])))
943
def test_iter_key_prefix_2_key_element_no_refs(self):
944
index = self.make_index(key_elements=2, nodes=[
945
(('name', 'fin1'), 'data', ()),
946
(('name', 'fin2'), 'beta', ()),
947
(('ref', 'erence'), 'refdata', ())])
948
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
949
(index, ('ref', 'erence'), 'refdata')]),
950
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
951
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
952
(index, ('name', 'fin2'), 'beta')]),
953
set(index.iter_entries_prefix([('name', None)])))
955
def test_iter_key_prefix_2_key_element_refs(self):
956
index = self.make_index(1, key_elements=2, nodes=[
957
(('name', 'fin1'), 'data', ([('ref', 'erence')], )),
958
(('name', 'fin2'), 'beta', ([], )),
959
(('ref', 'erence'), 'refdata', ([], ))])
960
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
961
(index, ('ref', 'erence'), 'refdata', ((), ))]),
962
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
963
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
964
(index, ('name', 'fin2'), 'beta', ((), ))]),
965
set(index.iter_entries_prefix([('name', None)])))
967
# XXX: external_references tests are duplicated in test_index. We
968
# probably should have per_graph_index tests...
969
def test_external_references_no_refs(self):
970
index = self.make_index(ref_lists=0, nodes=[])
971
self.assertRaises(ValueError, index.external_references, 0)
973
def test_external_references_no_results(self):
974
index = self.make_index(ref_lists=1, nodes=[
975
(('key',), 'value', ([],))])
976
self.assertEqual(set(), index.external_references(0))
978
def test_external_references_missing_ref(self):
979
missing_key = ('missing',)
980
index = self.make_index(ref_lists=1, nodes=[
981
(('key',), 'value', ([missing_key],))])
982
self.assertEqual(set([missing_key]), index.external_references(0))
984
def test_external_references_multiple_ref_lists(self):
985
missing_key = ('missing',)
986
index = self.make_index(ref_lists=2, nodes=[
987
(('key',), 'value', ([], [missing_key]))])
988
self.assertEqual(set([]), index.external_references(0))
989
self.assertEqual(set([missing_key]), index.external_references(1))
991
def test_external_references_two_records(self):
992
index = self.make_index(ref_lists=1, nodes=[
993
(('key-1',), 'value', ([('key-2',)],)),
994
(('key-2',), 'value', ([],)),
996
self.assertEqual(set([]), index.external_references(0))
998
def test__find_ancestors_one_page(self):
1001
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1002
(key1, 'value', ([key2],)),
1003
(key2, 'value', ([],)),
1006
missing_keys = set()
1007
search_keys = index._find_ancestors([key1], 0, parent_map, missing_keys)
1008
self.assertEqual({key1: (key2,), key2: ()}, parent_map)
1009
self.assertEqual(set(), missing_keys)
1010
self.assertEqual(set(), search_keys)
1012
def test__find_ancestors_one_page_w_missing(self):
1016
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1017
(key1, 'value', ([key2],)),
1018
(key2, 'value', ([],)),
1021
missing_keys = set()
1022
search_keys = index._find_ancestors([key2, key3], 0, parent_map,
1024
self.assertEqual({key2: ()}, parent_map)
1025
# we know that key3 is missing because we read the page that it would
1027
self.assertEqual(set([key3]), missing_keys)
1028
self.assertEqual(set(), search_keys)
1030
def test__find_ancestors_one_parent_missing(self):
1034
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1035
(key1, 'value', ([key2],)),
1036
(key2, 'value', ([key3],)),
1039
missing_keys = set()
1040
search_keys = index._find_ancestors([key1], 0, parent_map,
1042
self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1043
self.assertEqual(set(), missing_keys)
1044
# all we know is that key3 wasn't present on the page we were reading
1045
# but if you look, the last key is key2 which comes before key3, so we
1046
# don't know whether key3 would land on this page or not.
1047
self.assertEqual(set([key3]), search_keys)
1048
search_keys = index._find_ancestors(search_keys, 0, parent_map,
1050
# passing it back in, we are sure it is 'missing'
1051
self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1052
self.assertEqual(set([key3]), missing_keys)
1053
self.assertEqual(set([]), search_keys)
1055
def test__find_ancestors_dont_search_known(self):
1059
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
1060
(key1, 'value', ([key2],)),
1061
(key2, 'value', ([key3],)),
1062
(key3, 'value', ([],)),
1064
# We already know about key2, so we won't try to search for key3
1065
parent_map = {key2: (key3,)}
1066
missing_keys = set()
1067
search_keys = index._find_ancestors([key1], 0, parent_map,
1069
self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1070
self.assertEqual(set(), missing_keys)
1071
self.assertEqual(set(), search_keys)
1073
def test__find_ancestors_multiple_pages(self):
1074
# We need to use enough keys that we actually cause a split
1075
start_time = 1249671539
1076
email = "joebob@example.com"
1080
for i in xrange(400):
1081
rev_id = '%s-%s-%s' % (email,
1082
osutils.compact_date(start_time + i),
1083
osutils.rand_chars(16))
1085
nodes.append((rev_key, 'value', ref_lists))
1086
# We have a ref 'list' of length 1, with a list of parents, with 1
1087
# parent which is a key
1088
ref_lists = ((rev_key,),)
1089
rev_keys.append(rev_key)
1090
index = self.make_index(ref_lists=1, key_elements=1, nodes=nodes)
1091
self.assertEqual(400, index.key_count())
1092
self.assertEqual(3, len(index._row_offsets))
1093
nodes = dict(index._read_nodes([1, 2]))
1096
min_l2_key = l2.min_key
1097
max_l1_key = l1.max_key
1098
self.assertTrue(max_l1_key < min_l2_key)
1099
parents_min_l2_key = l2.keys[min_l2_key][1][0]
1100
self.assertEqual((l1.max_key,), parents_min_l2_key)
1101
# Now, whatever key we select that would fall on the second page,
1102
# should give us all the parents until the page break
1103
key_idx = rev_keys.index(min_l2_key)
1104
next_key = rev_keys[key_idx+1]
1105
# So now when we get the parent map, we should get the key we are
1106
# looking for, min_l2_key, and then a reference to go look for the
1107
# parent of that key
1109
missing_keys = set()
1110
search_keys = index._find_ancestors([next_key], 0, parent_map,
1112
self.assertEqual([min_l2_key, next_key], sorted(parent_map))
1113
self.assertEqual(set(), missing_keys)
1114
self.assertEqual(set([max_l1_key]), search_keys)
1116
search_keys = index._find_ancestors([max_l1_key], 0, parent_map,
1118
self.assertEqual(sorted(l1.keys), sorted(parent_map))
1119
self.assertEqual(set(), missing_keys)
1120
self.assertEqual(set(), search_keys)
1122
def test__find_ancestors_empty_index(self):
1123
index = self.make_index(ref_lists=1, key_elements=1, nodes=[])
1125
missing_keys = set()
1126
search_keys = index._find_ancestors([('one',), ('two',)], 0, parent_map,
1128
self.assertEqual(set(), search_keys)
1129
self.assertEqual({}, parent_map)
1130
self.assertEqual(set([('one',), ('two',)]), missing_keys)
1132
def test_supports_unlimited_cache(self):
1133
builder = btree_index.BTreeBuilder(reference_lists=0, key_elements=1)
1134
# We need enough nodes to cause a page split (so we have both an
1135
# internal node and a couple leaf nodes. 500 seems to be enough.)
1136
nodes = self.make_nodes(500, 1, 0)
1138
builder.add_node(*node)
1139
stream = builder.finish()
1140
trans = get_transport(self.get_url())
1141
size = trans.put_file('index', stream)
1142
index = btree_index.BTreeGraphIndex(trans, 'index', size)
1143
self.assertEqual(500, index.key_count())
1144
# We have an internal node
1145
self.assertEqual(2, len(index._row_lengths))
1146
# We have at least 2 leaf nodes
1147
self.assertTrue(index._row_lengths[-1] >= 2)
1148
self.assertIsInstance(index._leaf_node_cache, lru_cache.LRUCache)
1149
self.assertEqual(btree_index._NODE_CACHE_SIZE,
1150
index._leaf_node_cache._max_cache)
1151
self.assertIsInstance(index._internal_node_cache, fifo_cache.FIFOCache)
1152
self.assertEqual(100, index._internal_node_cache._max_cache)
1153
# No change if unlimited_cache=False is passed
1154
index = btree_index.BTreeGraphIndex(trans, 'index', size,
1155
unlimited_cache=False)
1156
self.assertIsInstance(index._leaf_node_cache, lru_cache.LRUCache)
1157
self.assertEqual(btree_index._NODE_CACHE_SIZE,
1158
index._leaf_node_cache._max_cache)
1159
self.assertIsInstance(index._internal_node_cache, fifo_cache.FIFOCache)
1160
self.assertEqual(100, index._internal_node_cache._max_cache)
1161
index = btree_index.BTreeGraphIndex(trans, 'index', size,
1162
unlimited_cache=True)
1163
self.assertIsInstance(index._leaf_node_cache, dict)
1164
self.assertIs(type(index._internal_node_cache), dict)
1165
# Exercise the lookup code
1166
entries = set(index.iter_entries([n[0] for n in nodes]))
1167
self.assertEqual(500, len(entries))
1170
class TestBTreeNodes(BTreeTestCase):
1172
def restore_parser(self):
1173
btree_index._btree_serializer = self.saved_parser
1176
BTreeTestCase.setUp(self)
1177
self.saved_parser = btree_index._btree_serializer
1178
self.addCleanup(self.restore_parser)
1179
btree_index._btree_serializer = self.parse_btree
1181
def test_LeafNode_1_0(self):
1182
node_bytes = ("type=leaf\n"
1183
"0000000000000000000000000000000000000000\x00\x00value:0\n"
1184
"1111111111111111111111111111111111111111\x00\x00value:1\n"
1185
"2222222222222222222222222222222222222222\x00\x00value:2\n"
1186
"3333333333333333333333333333333333333333\x00\x00value:3\n"
1187
"4444444444444444444444444444444444444444\x00\x00value:4\n")
1188
node = btree_index._LeafNode(node_bytes, 1, 0)
1189
# We do direct access, or don't care about order, to leaf nodes most of
1190
# the time, so a dict is useful:
1192
("0000000000000000000000000000000000000000",): ("value:0", ()),
1193
("1111111111111111111111111111111111111111",): ("value:1", ()),
1194
("2222222222222222222222222222222222222222",): ("value:2", ()),
1195
("3333333333333333333333333333333333333333",): ("value:3", ()),
1196
("4444444444444444444444444444444444444444",): ("value:4", ()),
1199
def test_LeafNode_2_2(self):
1200
node_bytes = ("type=leaf\n"
1201
"00\x0000\x00\t00\x00ref00\x00value:0\n"
1202
"00\x0011\x0000\x00ref00\t00\x00ref00\r01\x00ref01\x00value:1\n"
1203
"11\x0033\x0011\x00ref22\t11\x00ref22\r11\x00ref22\x00value:3\n"
1204
"11\x0044\x00\t11\x00ref00\x00value:4\n"
1207
node = btree_index._LeafNode(node_bytes, 2, 2)
1208
# We do direct access, or don't care about order, to leaf nodes most of
1209
# the time, so a dict is useful:
1211
('00', '00'): ('value:0', ((), (('00', 'ref00'),))),
1212
('00', '11'): ('value:1',
1213
((('00', 'ref00'),), (('00', 'ref00'), ('01', 'ref01')))),
1214
('11', '33'): ('value:3',
1215
((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
1216
('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
1219
def test_InternalNode_1(self):
1220
node_bytes = ("type=internal\n"
1222
"0000000000000000000000000000000000000000\n"
1223
"1111111111111111111111111111111111111111\n"
1224
"2222222222222222222222222222222222222222\n"
1225
"3333333333333333333333333333333333333333\n"
1226
"4444444444444444444444444444444444444444\n"
1228
node = btree_index._InternalNode(node_bytes)
1229
# We want to bisect to find the right children from this node, so a
1230
# vector is most useful.
1232
("0000000000000000000000000000000000000000",),
1233
("1111111111111111111111111111111111111111",),
1234
("2222222222222222222222222222222222222222",),
1235
("3333333333333333333333333333333333333333",),
1236
("4444444444444444444444444444444444444444",),
1238
self.assertEqual(1, node.offset)
1240
def test_LeafNode_2_2(self):
1241
node_bytes = ("type=leaf\n"
1242
"00\x0000\x00\t00\x00ref00\x00value:0\n"
1243
"00\x0011\x0000\x00ref00\t00\x00ref00\r01\x00ref01\x00value:1\n"
1244
"11\x0033\x0011\x00ref22\t11\x00ref22\r11\x00ref22\x00value:3\n"
1245
"11\x0044\x00\t11\x00ref00\x00value:4\n"
1248
node = btree_index._LeafNode(node_bytes, 2, 2)
1249
# We do direct access, or don't care about order, to leaf nodes most of
1250
# the time, so a dict is useful:
1252
('00', '00'): ('value:0', ((), (('00', 'ref00'),))),
1253
('00', '11'): ('value:1',
1254
((('00', 'ref00'),), (('00', 'ref00'), ('01', 'ref01')))),
1255
('11', '33'): ('value:3',
1256
((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
1257
('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
1260
def assertFlattened(self, expected, key, value, refs):
1261
flat_key, flat_line = self.parse_btree._flatten_node(
1262
(None, key, value, refs), bool(refs))
1263
self.assertEqual('\x00'.join(key), flat_key)
1264
self.assertEqual(expected, flat_line)
1266
def test__flatten_node(self):
1267
self.assertFlattened('key\0\0value\n', ('key',), 'value', [])
1268
self.assertFlattened('key\0tuple\0\0value str\n',
1269
('key', 'tuple'), 'value str', [])
1270
self.assertFlattened('key\0tuple\0triple\0\0value str\n',
1271
('key', 'tuple', 'triple'), 'value str', [])
1272
self.assertFlattened('k\0t\0s\0ref\0value str\n',
1273
('k', 't', 's'), 'value str', [[('ref',)]])
1274
self.assertFlattened('key\0tuple\0ref\0key\0value str\n',
1275
('key', 'tuple'), 'value str', [[('ref', 'key')]])
1276
self.assertFlattened("00\x0000\x00\t00\x00ref00\x00value:0\n",
1277
('00', '00'), 'value:0', ((), (('00', 'ref00'),)))
1278
self.assertFlattened(
1279
"00\x0011\x0000\x00ref00\t00\x00ref00\r01\x00ref01\x00value:1\n",
1280
('00', '11'), 'value:1',
1281
((('00', 'ref00'),), (('00', 'ref00'), ('01', 'ref01'))))
1282
self.assertFlattened(
1283
"11\x0033\x0011\x00ref22\t11\x00ref22\r11\x00ref22\x00value:3\n",
1284
('11', '33'), 'value:3',
1285
((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22'))))
1286
self.assertFlattened(
1287
"11\x0044\x00\t11\x00ref00\x00value:4\n",
1288
('11', '44'), 'value:4', ((), (('11', 'ref00'),)))
1291
class TestCompiledBtree(tests.TestCase):
1293
def test_exists(self):
1294
# This is just to let the user know if they don't have the feature
1296
self.requireFeature(CompiledBtreeParserFeature)
1299
class TestMultiBisectRight(tests.TestCase):
1301
def assertMultiBisectRight(self, offsets, search_keys, fixed_keys):
1302
self.assertEqual(offsets,
1303
btree_index.BTreeGraphIndex._multi_bisect_right(
1304
search_keys, fixed_keys))
1306
def test_after(self):
1307
self.assertMultiBisectRight([(1, ['b'])], ['b'], ['a'])
1308
self.assertMultiBisectRight([(3, ['e', 'f', 'g'])],
1309
['e', 'f', 'g'], ['a', 'b', 'c'])
1311
def test_before(self):
1312
self.assertMultiBisectRight([(0, ['a'])], ['a'], ['b'])
1313
self.assertMultiBisectRight([(0, ['a', 'b', 'c', 'd'])],
1314
['a', 'b', 'c', 'd'], ['e', 'f', 'g'])
1316
def test_exact(self):
1317
self.assertMultiBisectRight([(1, ['a'])], ['a'], ['a'])
1318
self.assertMultiBisectRight([(1, ['a']), (2, ['b'])], ['a', 'b'], ['a', 'b'])
1319
self.assertMultiBisectRight([(1, ['a']), (3, ['c'])],
1320
['a', 'c'], ['a', 'b', 'c'])
1322
def test_inbetween(self):
1323
self.assertMultiBisectRight([(1, ['b'])], ['b'], ['a', 'c'])
1324
self.assertMultiBisectRight([(1, ['b', 'c', 'd']), (2, ['f', 'g'])],
1325
['b', 'c', 'd', 'f', 'g'], ['a', 'e', 'h'])
1327
def test_mixed(self):
1328
self.assertMultiBisectRight([(0, ['a', 'b']), (2, ['d', 'e']),
1330
['a', 'b', 'd', 'e', 'g', 'h'],
1331
['c', 'd', 'f', 'g'])
1334
class TestExpandOffsets(tests.TestCase):
1336
def make_index(self, size, recommended_pages=None):
1337
"""Make an index with a generic size.
1339
This doesn't actually create anything on disk, it just primes a
1340
BTreeGraphIndex with the recommended information.
1342
index = btree_index.BTreeGraphIndex(get_transport('memory:///'),
1343
'test-index', size=size)
1344
if recommended_pages is not None:
1345
index._recommended_pages = recommended_pages
1348
def set_cached_offsets(self, index, cached_offsets):
1349
"""Monkeypatch to give a canned answer for _get_offsets_for...()."""
1350
def _get_offsets_to_cached_pages():
1351
cached = set(cached_offsets)
1353
index._get_offsets_to_cached_pages = _get_offsets_to_cached_pages
1355
def prepare_index(self, index, node_ref_lists, key_length, key_count,
1356
row_lengths, cached_offsets):
1357
"""Setup the BTreeGraphIndex with some pre-canned information."""
1358
index.node_ref_lists = node_ref_lists
1359
index._key_length = key_length
1360
index._key_count = key_count
1361
index._row_lengths = row_lengths
1362
index._compute_row_offsets()
1363
index._root_node = btree_index._InternalNode('internal\noffset=0\n')
1364
self.set_cached_offsets(index, cached_offsets)
1366
def make_100_node_index(self):
1367
index = self.make_index(4096*100, 6)
1368
# Consider we've already made a single request at the middle
1369
self.prepare_index(index, node_ref_lists=0, key_length=1,
1370
key_count=1000, row_lengths=[1, 99],
1371
cached_offsets=[0, 50])
1374
def make_1000_node_index(self):
1375
index = self.make_index(4096*1000, 6)
1376
# Pretend we've already made a single request in the middle
1377
self.prepare_index(index, node_ref_lists=0, key_length=1,
1378
key_count=90000, row_lengths=[1, 9, 990],
1379
cached_offsets=[0, 5, 500])
1382
def assertNumPages(self, expected_pages, index, size):
1384
self.assertEqual(expected_pages, index._compute_total_pages_in_index())
1386
def assertExpandOffsets(self, expected, index, offsets):
1387
self.assertEqual(expected, index._expand_offsets(offsets),
1388
'We did not get the expected value after expanding'
1391
def test_default_recommended_pages(self):
1392
index = self.make_index(None)
1393
# local transport recommends 4096 byte reads, which is 1 page
1394
self.assertEqual(1, index._recommended_pages)
1396
def test__compute_total_pages_in_index(self):
1397
index = self.make_index(None)
1398
self.assertNumPages(1, index, 1024)
1399
self.assertNumPages(1, index, 4095)
1400
self.assertNumPages(1, index, 4096)
1401
self.assertNumPages(2, index, 4097)
1402
self.assertNumPages(2, index, 8192)
1403
self.assertNumPages(76, index, 4096*75 + 10)
1405
def test__find_layer_start_and_stop(self):
1406
index = self.make_1000_node_index()
1407
self.assertEqual((0, 1), index._find_layer_first_and_end(0))
1408
self.assertEqual((1, 10), index._find_layer_first_and_end(1))
1409
self.assertEqual((1, 10), index._find_layer_first_and_end(9))
1410
self.assertEqual((10, 1000), index._find_layer_first_and_end(10))
1411
self.assertEqual((10, 1000), index._find_layer_first_and_end(99))
1412
self.assertEqual((10, 1000), index._find_layer_first_and_end(999))
1414
def test_unknown_size(self):
1415
# We should not expand if we don't know the file size
1416
index = self.make_index(None, 10)
1417
self.assertExpandOffsets([0], index, [0])
1418
self.assertExpandOffsets([1, 4, 9], index, [1, 4, 9])
1420
def test_more_than_recommended(self):
1421
index = self.make_index(4096*100, 2)
1422
self.assertExpandOffsets([1, 10], index, [1, 10])
1423
self.assertExpandOffsets([1, 10, 20], index, [1, 10, 20])
1425
def test_read_all_from_root(self):
1426
index = self.make_index(4096*10, 20)
1427
self.assertExpandOffsets(range(10), index, [0])
1429
def test_read_all_when_cached(self):
1430
# We've read enough that we can grab all the rest in a single request
1431
index = self.make_index(4096*10, 5)
1432
self.prepare_index(index, node_ref_lists=0, key_length=1,
1433
key_count=1000, row_lengths=[1, 9],
1434
cached_offsets=[0, 1, 2, 5, 6])
1435
# It should fill the remaining nodes, regardless of the one requested
1436
self.assertExpandOffsets([3, 4, 7, 8, 9], index, [3])
1437
self.assertExpandOffsets([3, 4, 7, 8, 9], index, [8])
1438
self.assertExpandOffsets([3, 4, 7, 8, 9], index, [9])
1440
def test_no_root_node(self):
1441
index = self.make_index(4096*10, 5)
1442
self.assertExpandOffsets([0], index, [0])
1444
def test_include_neighbors(self):
1445
index = self.make_100_node_index()
1446
# We expand in both directions, until we have at least 'recommended'
1448
self.assertExpandOffsets([9, 10, 11, 12, 13, 14, 15], index, [12])
1449
self.assertExpandOffsets([88, 89, 90, 91, 92, 93, 94], index, [91])
1450
# If we hit an 'edge' we continue in the other direction
1451
self.assertExpandOffsets([1, 2, 3, 4, 5, 6], index, [2])
1452
self.assertExpandOffsets([94, 95, 96, 97, 98, 99], index, [98])
1454
# Requesting many nodes will expand all locations equally
1455
self.assertExpandOffsets([1, 2, 3, 80, 81, 82], index, [2, 81])
1456
self.assertExpandOffsets([1, 2, 3, 9, 10, 11, 80, 81, 82], index,
1459
def test_stop_at_cached(self):
1460
index = self.make_100_node_index()
1461
self.set_cached_offsets(index, [0, 10, 19])
1462
self.assertExpandOffsets([11, 12, 13, 14, 15, 16], index, [11])
1463
self.assertExpandOffsets([11, 12, 13, 14, 15, 16], index, [12])
1464
self.assertExpandOffsets([12, 13, 14, 15, 16, 17, 18], index, [15])
1465
self.assertExpandOffsets([13, 14, 15, 16, 17, 18], index, [16])
1466
self.assertExpandOffsets([13, 14, 15, 16, 17, 18], index, [17])
1467
self.assertExpandOffsets([13, 14, 15, 16, 17, 18], index, [18])
1469
def test_cannot_fully_expand(self):
1470
index = self.make_100_node_index()
1471
self.set_cached_offsets(index, [0, 10, 12])
1472
# We don't go into an endless loop if we are bound by cached nodes
1473
self.assertExpandOffsets([11], index, [11])
1475
def test_overlap(self):
1476
index = self.make_100_node_index()
1477
self.assertExpandOffsets([10, 11, 12, 13, 14, 15], index, [12, 13])
1478
self.assertExpandOffsets([10, 11, 12, 13, 14, 15], index, [11, 14])
1480
def test_stay_within_layer(self):
1481
index = self.make_1000_node_index()
1482
# When expanding a request, we won't read nodes from the next layer
1483
self.assertExpandOffsets([1, 2, 3, 4], index, [2])
1484
self.assertExpandOffsets([6, 7, 8, 9], index, [6])
1485
self.assertExpandOffsets([6, 7, 8, 9], index, [9])
1486
self.assertExpandOffsets([10, 11, 12, 13, 14, 15], index, [10])
1487
self.assertExpandOffsets([10, 11, 12, 13, 14, 15, 16], index, [13])
1489
self.set_cached_offsets(index, [0, 4, 12])
1490
self.assertExpandOffsets([5, 6, 7, 8, 9], index, [7])
1491
self.assertExpandOffsets([10, 11], index, [11])
1493
def test_small_requests_unexpanded(self):
1494
index = self.make_100_node_index()
1495
self.set_cached_offsets(index, [0])
1496
self.assertExpandOffsets([1], index, [1])
1497
self.assertExpandOffsets([50], index, [50])
1498
# If we request more than one node, then we'll expand
1499
self.assertExpandOffsets([49, 50, 51, 59, 60, 61], index, [50, 60])
1501
# The first pass does not expand
1502
index = self.make_1000_node_index()
1503
self.set_cached_offsets(index, [0])
1504
self.assertExpandOffsets([1], index, [1])
1505
self.set_cached_offsets(index, [0, 1])
1506
self.assertExpandOffsets([100], index, [100])
1507
self.set_cached_offsets(index, [0, 1, 100])
1508
# But after the first depth, we will expand
1509
self.assertExpandOffsets([2, 3, 4, 5, 6, 7], index, [2])
1510
self.assertExpandOffsets([2, 3, 4, 5, 6, 7], index, [4])
1511
self.set_cached_offsets(index, [0, 1, 2, 3, 4, 5, 6, 7, 100])
1512
self.assertExpandOffsets([102, 103, 104, 105, 106, 107, 108], index,
added
removed
bzrlib/tests/test_btree_index.py
