1
# Copyright (C) 2008 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
"""Tests for btree indices."""
28
from bzrlib.tests import (
29
TestCaseWithTransport,
33
split_suite_by_condition,
35
from bzrlib.transport import get_transport
38
def load_tests(standard_tests, module, loader):
39
# parameterise the TestBTreeNodes tests
40
node_tests, others = split_suite_by_condition(standard_tests,
41
condition_isinstance(TestBTreeNodes))
42
applier = TestScenarioApplier()
43
import bzrlib._btree_serializer_py as py_module
44
applier.scenarios = [('python', {'parse_btree': py_module})]
45
if CompiledBtreeParserFeature.available():
46
# Is there a way to do this that gets missing feature failures rather
47
# than no indication to the user?
48
import bzrlib._btree_serializer_c as c_module
49
applier.scenarios.append(('C', {'parse_btree': c_module}))
50
adapt_tests(node_tests, applier, others)
54
class _CompiledBtreeParserFeature(tests.Feature):
57
import bzrlib._btree_serializer_c
62
def feature_name(self):
63
return 'bzrlib._btree_serializer_c'
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))
117
class TestBTreeBuilder(BTreeTestCase):
119
def test_empty_1_0(self):
120
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
121
# NamedTemporaryFile dies on builder.finish().read(). weird.
122
temp_file = builder.finish()
123
content = temp_file.read()
126
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=0\n"
130
def test_empty_2_1(self):
131
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=1)
132
# NamedTemporaryFile dies on builder.finish().read(). weird.
133
temp_file = builder.finish()
134
content = temp_file.read()
137
"B+Tree Graph Index 2\nnode_ref_lists=1\nkey_elements=2\nlen=0\n"
141
def test_root_leaf_1_0(self):
142
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
143
nodes = self.make_nodes(5, 1, 0)
145
builder.add_node(*node)
146
# NamedTemporaryFile dies on builder.finish().read(). weird.
147
temp_file = builder.finish()
148
content = temp_file.read()
150
self.assertEqual(158, len(content))
152
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=5\n"
155
node_content = content[73:]
156
node_bytes = zlib.decompress(node_content)
157
expected_node = ("type=leaf\n"
158
"0000000000000000000000000000000000000000\x00\x00value:0\n"
159
"1111111111111111111111111111111111111111\x00\x00value:1\n"
160
"2222222222222222222222222222222222222222\x00\x00value:2\n"
161
"3333333333333333333333333333333333333333\x00\x00value:3\n"
162
"4444444444444444444444444444444444444444\x00\x00value:4\n")
163
self.assertEqual(expected_node, node_bytes)
165
def test_root_leaf_2_2(self):
166
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
167
nodes = self.make_nodes(5, 2, 2)
169
builder.add_node(*node)
170
# NamedTemporaryFile dies on builder.finish().read(). weird.
171
temp_file = builder.finish()
172
content = temp_file.read()
174
self.assertEqual(264, len(content))
176
"B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=10\n"
179
node_content = content[74:]
180
node_bytes = zlib.decompress(node_content)
183
"0000000000000000000000000000000000000000\x000000000000000000000000000000000000000000\x00\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:0\n"
184
"0000000000000000000000000000000000000000\x001111111111111111111111111111111111111111\x000000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\r0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:1\n"
185
"0000000000000000000000000000000000000000\x002222222222222222222222222222222222222222\x00\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:2\n"
186
"0000000000000000000000000000000000000000\x003333333333333333333333333333333333333333\x000000000000000000000000000000000000000000\x00ref2222222222222222222222222222222222222222\t0000000000000000000000000000000000000000\x00ref2222222222222222222222222222222222222222\r0000000000000000000000000000000000000000\x00ref2222222222222222222222222222222222222222\x00value:3\n"
187
"0000000000000000000000000000000000000000\x004444444444444444444444444444444444444444\x00\t0000000000000000000000000000000000000000\x00ref0000000000000000000000000000000000000000\x00value:4\n"
188
"1111111111111111111111111111111111111111\x000000000000000000000000000000000000000000\x00\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:0\n"
189
"1111111111111111111111111111111111111111\x001111111111111111111111111111111111111111\x001111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\r1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:1\n"
190
"1111111111111111111111111111111111111111\x002222222222222222222222222222222222222222\x00\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:2\n"
191
"1111111111111111111111111111111111111111\x003333333333333333333333333333333333333333\x001111111111111111111111111111111111111111\x00ref2222222222222222222222222222222222222222\t1111111111111111111111111111111111111111\x00ref2222222222222222222222222222222222222222\r1111111111111111111111111111111111111111\x00ref2222222222222222222222222222222222222222\x00value:3\n"
192
"1111111111111111111111111111111111111111\x004444444444444444444444444444444444444444\x00\t1111111111111111111111111111111111111111\x00ref0000000000000000000000000000000000000000\x00value:4\n"
195
self.assertEqual(expected_node, node_bytes)
197
def test_2_leaves_1_0(self):
198
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
199
nodes = self.make_nodes(800, 1, 0)
201
builder.add_node(*node)
202
# NamedTemporaryFile dies on builder.finish().read(). weird.
203
temp_file = builder.finish()
204
content = temp_file.read()
206
self.assertEqual(10646, len(content))
208
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=800\n"
211
root = content[77:4096]
212
leaf1 = content[4096:8192]
213
leaf2 = content[8192:]
214
root_bytes = zlib.decompress(root)
218
"503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503503\n"
220
self.assertEqual(expected_root, root_bytes)
221
# We already know serialisation works for leaves, check key selection:
222
leaf1_bytes = zlib.decompress(leaf1)
223
sorted_node_keys = sorted(node[0] for node in nodes)
224
node = btree_index._LeafNode(leaf1_bytes, 1, 0)
225
self.assertEqual(448, len(node.keys))
226
self.assertEqual(sorted_node_keys[:448], sorted(node.keys))
227
leaf2_bytes = zlib.decompress(leaf2)
228
node = btree_index._LeafNode(leaf2_bytes, 1, 0)
229
self.assertEqual(800 - 448, len(node.keys))
230
self.assertEqual(sorted_node_keys[448:], sorted(node.keys))
232
def test_last_page_rounded_1_layer(self):
233
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
234
nodes = self.make_nodes(10, 1, 0)
236
builder.add_node(*node)
237
# NamedTemporaryFile dies on builder.finish().read(). weird.
238
temp_file = builder.finish()
239
content = temp_file.read()
241
self.assertEqual(181, len(content))
243
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=10\n"
246
# Check thelast page is well formed
248
leaf2_bytes = zlib.decompress(leaf2)
249
node = btree_index._LeafNode(leaf2_bytes, 1, 0)
250
self.assertEqual(10, len(node.keys))
251
sorted_node_keys = sorted(node[0] for node in nodes)
252
self.assertEqual(sorted_node_keys, sorted(node.keys))
254
def test_last_page_not_rounded_2_layer(self):
255
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
256
nodes = self.make_nodes(800, 1, 0)
258
builder.add_node(*node)
259
# NamedTemporaryFile dies on builder.finish().read(). weird.
260
temp_file = builder.finish()
261
content = temp_file.read()
263
self.assertEqual(10646, len(content))
265
"B+Tree Graph Index 2\nnode_ref_lists=0\nkey_elements=1\nlen=800\n"
268
# Check the last page is well formed
269
leaf2 = content[8192:]
270
leaf2_bytes = zlib.decompress(leaf2)
271
node = btree_index._LeafNode(leaf2_bytes, 1, 0)
272
self.assertEqual(800 - 448, len(node.keys))
273
sorted_node_keys = sorted(node[0] for node in nodes)
274
self.assertEqual(sorted_node_keys[448:], sorted(node.keys))
276
def test_three_level_tree_details(self):
277
# The left most pointer in the second internal node in a row should
278
# pointer to the second node that the internal node is for, _not_
279
# the first, otherwise the first node overlaps with the last node of
280
# the prior internal node on that row.
281
# We will be adding 100,000 nodes, so spill at 100,001 to prevent
282
# having to flush anything out to disk.
283
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2,
285
# 100K nodes is *just* enough to create a two internal nodes on the
287
nodes = self.make_nodes(50000, 2, 2)
290
builder.add_node(*node)
291
transport = get_transport('trace+' + self.get_url(''))
292
size = transport.put_file('index', self.time(builder.finish))
294
index = btree_index.BTreeGraphIndex(transport, 'index', size)
295
# Seed the metadata, we're using internal calls now.
297
self.assertEqual(3, len(index._row_lengths),
298
"Not enough rows: %r" % index._row_lengths)
299
self.assertEqual(4, len(index._row_offsets))
300
self.assertEqual(sum(index._row_lengths), index._row_offsets[-1])
301
internal_nodes = index._get_internal_nodes([0, 1, 2])
302
root_node = internal_nodes[0]
303
internal_node1 = internal_nodes[1]
304
internal_node2 = internal_nodes[2]
305
# The left most node node2 points at should be one after the right most
306
# node pointed at by node1.
307
self.assertEqual(internal_node2.offset, 1 + len(internal_node1.keys))
308
# The left most key of the second node pointed at by internal_node2
309
# should be its first key. We can check this by looking for its first key
310
# in the second node it points at
311
pos = index._row_offsets[2] + internal_node2.offset + 1
312
leaf = index._get_leaf_nodes([pos])[pos]
313
self.assertTrue(internal_node2.keys[0] in leaf.keys)
315
def test_2_leaves_2_2(self):
316
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
317
nodes = self.make_nodes(200, 2, 2)
319
builder.add_node(*node)
320
# NamedTemporaryFile dies on builder.finish().read(). weird.
321
temp_file = builder.finish()
322
content = temp_file.read()
324
self.assertEqual(10574, len(content))
326
"B+Tree Graph Index 2\nnode_ref_lists=2\nkey_elements=2\nlen=400\n"
329
root = content[77:4096]
330
leaf1 = content[4096:8192]
331
leaf2 = content[8192:]
332
root_bytes = zlib.decompress(root)
336
"1111111111111111111111111111111111111111\x00"
337
"126126126126126126126126126126126126126126126126126126126"
338
"126126126126126126126126126126126126126126126126126126126126126\n"
340
self.assertEqual(expected_root, root_bytes)
341
# We assume the other leaf nodes have been written correctly - layering
344
def test_spill_index_stress_1_1(self):
345
builder = btree_index.BTreeBuilder(key_elements=1, spill_at=2)
346
nodes = [node[0:2] for node in self.make_nodes(16, 1, 0)]
347
builder.add_node(*nodes[0])
348
# Test the parts of the index that take up memory are doing so
350
self.assertEqual(1, len(builder._nodes))
351
self.assertEqual(1, len(builder._keys))
352
self.assertEqual({}, builder._nodes_by_key)
353
builder.add_node(*nodes[1])
354
self.assertEqual(0, len(builder._nodes))
355
self.assertEqual(0, len(builder._keys))
356
self.assertEqual({}, builder._nodes_by_key)
357
self.assertEqual(1, len(builder._backing_indices))
358
self.assertEqual(2, builder._backing_indices[0].key_count())
360
builder.add_node(*nodes[2])
361
self.assertEqual(1, len(builder._nodes))
362
self.assertEqual(1, len(builder._keys))
363
self.assertEqual({}, builder._nodes_by_key)
364
# And spills to a second backing index combing all
365
builder.add_node(*nodes[3])
366
self.assertEqual(0, len(builder._nodes))
367
self.assertEqual(0, len(builder._keys))
368
self.assertEqual({}, builder._nodes_by_key)
369
self.assertEqual(2, len(builder._backing_indices))
370
self.assertEqual(None, builder._backing_indices[0])
371
self.assertEqual(4, builder._backing_indices[1].key_count())
372
# The next spills to the 2-len slot
373
builder.add_node(*nodes[4])
374
builder.add_node(*nodes[5])
375
self.assertEqual(0, len(builder._nodes))
376
self.assertEqual(0, len(builder._keys))
377
self.assertEqual({}, builder._nodes_by_key)
378
self.assertEqual(2, len(builder._backing_indices))
379
self.assertEqual(2, builder._backing_indices[0].key_count())
380
self.assertEqual(4, builder._backing_indices[1].key_count())
381
# Next spill combines
382
builder.add_node(*nodes[6])
383
builder.add_node(*nodes[7])
384
self.assertEqual(3, len(builder._backing_indices))
385
self.assertEqual(None, builder._backing_indices[0])
386
self.assertEqual(None, builder._backing_indices[1])
387
self.assertEqual(8, builder._backing_indices[2].key_count())
388
# And so forth - counting up in binary.
389
builder.add_node(*nodes[8])
390
builder.add_node(*nodes[9])
391
self.assertEqual(3, len(builder._backing_indices))
392
self.assertEqual(2, builder._backing_indices[0].key_count())
393
self.assertEqual(None, builder._backing_indices[1])
394
self.assertEqual(8, builder._backing_indices[2].key_count())
395
builder.add_node(*nodes[10])
396
builder.add_node(*nodes[11])
397
self.assertEqual(3, len(builder._backing_indices))
398
self.assertEqual(None, builder._backing_indices[0])
399
self.assertEqual(4, builder._backing_indices[1].key_count())
400
self.assertEqual(8, builder._backing_indices[2].key_count())
401
builder.add_node(*nodes[12])
402
# Test that memory and disk are both used for query methods; and that
403
# None is skipped over happily.
404
self.assertEqual([(builder,) + node for node in sorted(nodes[:13])],
405
list(builder.iter_all_entries()))
406
# Two nodes - one memory one disk
407
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
408
set(builder.iter_entries([nodes[12][0], nodes[11][0]])))
409
self.assertEqual(13, builder.key_count())
410
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
411
set(builder.iter_entries_prefix([nodes[12][0], nodes[11][0]])))
412
builder.add_node(*nodes[13])
413
self.assertEqual(3, len(builder._backing_indices))
414
self.assertEqual(2, builder._backing_indices[0].key_count())
415
self.assertEqual(4, builder._backing_indices[1].key_count())
416
self.assertEqual(8, builder._backing_indices[2].key_count())
417
builder.add_node(*nodes[14])
418
builder.add_node(*nodes[15])
419
self.assertEqual(4, len(builder._backing_indices))
420
self.assertEqual(None, builder._backing_indices[0])
421
self.assertEqual(None, builder._backing_indices[1])
422
self.assertEqual(None, builder._backing_indices[2])
423
self.assertEqual(16, builder._backing_indices[3].key_count())
424
# Now finish, and check we got a correctly ordered tree
425
transport = self.get_transport('')
426
size = transport.put_file('index', builder.finish())
427
index = btree_index.BTreeGraphIndex(transport, 'index', size)
428
nodes = list(index.iter_all_entries())
429
self.assertEqual(sorted(nodes), nodes)
430
self.assertEqual(16, len(nodes))
432
def test_spill_index_stress_2_2(self):
433
# test that references and longer keys don't confuse things.
434
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2,
436
nodes = self.make_nodes(16, 2, 2)
437
builder.add_node(*nodes[0])
438
# Test the parts of the index that take up memory are doing so
440
self.assertEqual(1, len(builder._keys))
441
self.assertEqual(2, len(builder._nodes))
442
self.assertNotEqual({}, builder._nodes_by_key)
443
builder.add_node(*nodes[1])
444
self.assertEqual(0, len(builder._keys))
445
self.assertEqual(0, len(builder._nodes))
446
self.assertEqual({}, builder._nodes_by_key)
447
self.assertEqual(1, len(builder._backing_indices))
448
self.assertEqual(2, builder._backing_indices[0].key_count())
450
builder.add_node(*nodes[2])
451
self.assertEqual(2, len(builder._nodes))
452
self.assertEqual(1, len(builder._keys))
453
self.assertNotEqual({}, builder._nodes_by_key)
454
# And spills to a second backing index combing all
455
builder.add_node(*nodes[3])
456
self.assertEqual(0, len(builder._nodes))
457
self.assertEqual(0, len(builder._keys))
458
self.assertEqual({}, builder._nodes_by_key)
459
self.assertEqual(2, len(builder._backing_indices))
460
self.assertEqual(None, builder._backing_indices[0])
461
self.assertEqual(4, builder._backing_indices[1].key_count())
462
# The next spills to the 2-len slot
463
builder.add_node(*nodes[4])
464
builder.add_node(*nodes[5])
465
self.assertEqual(0, len(builder._nodes))
466
self.assertEqual(0, len(builder._keys))
467
self.assertEqual({}, builder._nodes_by_key)
468
self.assertEqual(2, len(builder._backing_indices))
469
self.assertEqual(2, builder._backing_indices[0].key_count())
470
self.assertEqual(4, builder._backing_indices[1].key_count())
471
# Next spill combines
472
builder.add_node(*nodes[6])
473
builder.add_node(*nodes[7])
474
self.assertEqual(3, len(builder._backing_indices))
475
self.assertEqual(None, builder._backing_indices[0])
476
self.assertEqual(None, builder._backing_indices[1])
477
self.assertEqual(8, builder._backing_indices[2].key_count())
478
# And so forth - counting up in binary.
479
builder.add_node(*nodes[8])
480
builder.add_node(*nodes[9])
481
self.assertEqual(3, len(builder._backing_indices))
482
self.assertEqual(2, builder._backing_indices[0].key_count())
483
self.assertEqual(None, builder._backing_indices[1])
484
self.assertEqual(8, builder._backing_indices[2].key_count())
485
builder.add_node(*nodes[10])
486
builder.add_node(*nodes[11])
487
self.assertEqual(3, len(builder._backing_indices))
488
self.assertEqual(None, builder._backing_indices[0])
489
self.assertEqual(4, builder._backing_indices[1].key_count())
490
self.assertEqual(8, builder._backing_indices[2].key_count())
491
builder.add_node(*nodes[12])
492
# Test that memory and disk are both used for query methods; and that
493
# None is skipped over happily.
494
self.assertEqual([(builder,) + node for node in sorted(nodes[:13])],
495
list(builder.iter_all_entries()))
496
# Two nodes - one memory one disk
497
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
498
set(builder.iter_entries([nodes[12][0], nodes[11][0]])))
499
self.assertEqual(13, builder.key_count())
500
self.assertEqual(set([(builder,) + node for node in nodes[11:13]]),
501
set(builder.iter_entries_prefix([nodes[12][0], nodes[11][0]])))
502
builder.add_node(*nodes[13])
503
self.assertEqual(3, len(builder._backing_indices))
504
self.assertEqual(2, builder._backing_indices[0].key_count())
505
self.assertEqual(4, builder._backing_indices[1].key_count())
506
self.assertEqual(8, builder._backing_indices[2].key_count())
507
builder.add_node(*nodes[14])
508
builder.add_node(*nodes[15])
509
self.assertEqual(4, len(builder._backing_indices))
510
self.assertEqual(None, builder._backing_indices[0])
511
self.assertEqual(None, builder._backing_indices[1])
512
self.assertEqual(None, builder._backing_indices[2])
513
self.assertEqual(16, builder._backing_indices[3].key_count())
514
# Now finish, and check we got a correctly ordered tree
515
transport = self.get_transport('')
516
size = transport.put_file('index', builder.finish())
517
index = btree_index.BTreeGraphIndex(transport, 'index', size)
518
nodes = list(index.iter_all_entries())
519
self.assertEqual(sorted(nodes), nodes)
520
self.assertEqual(16, len(nodes))
522
def test_spill_index_duplicate_key_caught_on_finish(self):
523
builder = btree_index.BTreeBuilder(key_elements=1, spill_at=2)
524
nodes = [node[0:2] for node in self.make_nodes(16, 1, 0)]
525
builder.add_node(*nodes[0])
526
builder.add_node(*nodes[1])
527
builder.add_node(*nodes[0])
528
self.assertRaises(errors.BadIndexDuplicateKey, builder.finish)
531
class TestBTreeIndex(BTreeTestCase):
533
def make_index(self, ref_lists=0, key_elements=1, nodes=[]):
534
builder = btree_index.BTreeBuilder(reference_lists=ref_lists,
535
key_elements=key_elements)
536
for key, value, references in nodes:
537
builder.add_node(key, value, references)
538
stream = builder.finish()
539
trans = get_transport('trace+' + self.get_url())
540
size = trans.put_file('index', stream)
541
return btree_index.BTreeGraphIndex(trans, 'index', size)
543
def test_trivial_constructor(self):
544
transport = get_transport('trace+' + self.get_url(''))
545
index = btree_index.BTreeGraphIndex(transport, 'index', None)
546
# Checks the page size at load, but that isn't logged yet.
547
self.assertEqual([], transport._activity)
549
def test_with_size_constructor(self):
550
transport = get_transport('trace+' + self.get_url(''))
551
index = btree_index.BTreeGraphIndex(transport, 'index', 1)
552
# Checks the page size at load, but that isn't logged yet.
553
self.assertEqual([], transport._activity)
555
def test_empty_key_count_no_size(self):
556
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
557
transport = get_transport('trace+' + self.get_url(''))
558
transport.put_file('index', builder.finish())
559
index = btree_index.BTreeGraphIndex(transport, 'index', None)
560
del transport._activity[:]
561
self.assertEqual([], transport._activity)
562
self.assertEqual(0, index.key_count())
563
# The entire index should have been requested (as we generally have the
564
# size available, and doing many small readvs is inappropriate).
565
# We can't tell how much was actually read here, but - check the code.
566
self.assertEqual([('get', 'index'),
567
('readv', 'index', [(0, 72)], False, None)],
570
def test_empty_key_count(self):
571
builder = btree_index.BTreeBuilder(key_elements=1, reference_lists=0)
572
transport = get_transport('trace+' + self.get_url(''))
573
size = transport.put_file('index', builder.finish())
574
self.assertEqual(72, size)
575
index = btree_index.BTreeGraphIndex(transport, 'index', size)
576
del transport._activity[:]
577
self.assertEqual([], transport._activity)
578
self.assertEqual(0, index.key_count())
579
# The entire index should have been read, as 4K > size
580
self.assertEqual([('readv', 'index', [(0, 72)], False, None)],
583
def test_non_empty_key_count_2_2(self):
584
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
585
nodes = self.make_nodes(35, 2, 2)
587
builder.add_node(*node)
588
transport = get_transport('trace+' + self.get_url(''))
589
size = transport.put_file('index', builder.finish())
590
index = btree_index.BTreeGraphIndex(transport, 'index', size)
591
del transport._activity[:]
592
self.assertEqual([], transport._activity)
593
self.assertEqual(70, index.key_count())
594
# The entire index should have been read, as it is one page long.
595
self.assertEqual([('readv', 'index', [(0, size)], False, None)],
597
self.assertEqual(1593, size)
599
def test_2_levels_key_count_2_2(self):
600
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
601
nodes = self.make_nodes(160, 2, 2)
603
builder.add_node(*node)
604
transport = get_transport('trace+' + self.get_url(''))
605
size = transport.put_file('index', builder.finish())
606
self.assertEqual(10242, size)
607
index = btree_index.BTreeGraphIndex(transport, 'index', size)
608
del transport._activity[:]
609
self.assertEqual([], transport._activity)
610
self.assertEqual(320, index.key_count())
611
# The entire index should not have been read.
612
self.assertEqual([('readv', 'index', [(0, 4096)], False, None)],
615
def test_validate_one_page(self):
616
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
617
nodes = self.make_nodes(80, 2, 2)
619
builder.add_node(*node)
620
transport = get_transport('trace+' + self.get_url(''))
621
size = transport.put_file('index', builder.finish())
622
index = btree_index.BTreeGraphIndex(transport, 'index', size)
623
del transport._activity[:]
624
self.assertEqual([], transport._activity)
626
# The entire index should have been read linearly.
627
self.assertEqual([('readv', 'index', [(0, size)], False, None)],
629
self.assertEqual(3846, size)
631
def test_validate_two_pages(self):
632
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
633
nodes = self.make_nodes(160, 2, 2)
635
builder.add_node(*node)
636
transport = get_transport('trace+' + self.get_url(''))
637
size = transport.put_file('index', builder.finish())
638
# Root page, 2 leaf pages
639
self.assertEqual(10242, size)
640
index = btree_index.BTreeGraphIndex(transport, 'index', size)
641
del transport._activity[:]
642
self.assertEqual([], transport._activity)
644
# The entire index should have been read linearly.
645
self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
646
('readv', 'index', [(4096, 4096), (8192, 2050)], False, None)],
648
# XXX: TODO: write some badly-ordered nodes, and some pointers-to-wrong
649
# node and make validate find them.
651
def test_eq_ne(self):
652
# two indices are equal when constructed with the same parameters:
653
transport1 = get_transport('trace+' + self.get_url(''))
654
transport2 = get_transport(self.get_url(''))
656
btree_index.BTreeGraphIndex(transport1, 'index', None) ==
657
btree_index.BTreeGraphIndex(transport1, 'index', None))
659
btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
660
btree_index.BTreeGraphIndex(transport1, 'index', 20))
662
btree_index.BTreeGraphIndex(transport1, 'index', 20) ==
663
btree_index.BTreeGraphIndex(transport2, 'index', 20))
665
btree_index.BTreeGraphIndex(transport1, 'inde1', 20) ==
666
btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
668
btree_index.BTreeGraphIndex(transport1, 'index', 10) ==
669
btree_index.BTreeGraphIndex(transport1, 'index', 20))
671
btree_index.BTreeGraphIndex(transport1, 'index', None) !=
672
btree_index.BTreeGraphIndex(transport1, 'index', None))
674
btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
675
btree_index.BTreeGraphIndex(transport1, 'index', 20))
677
btree_index.BTreeGraphIndex(transport1, 'index', 20) !=
678
btree_index.BTreeGraphIndex(transport2, 'index', 20))
680
btree_index.BTreeGraphIndex(transport1, 'inde1', 20) !=
681
btree_index.BTreeGraphIndex(transport1, 'inde2', 20))
683
btree_index.BTreeGraphIndex(transport1, 'index', 10) !=
684
btree_index.BTreeGraphIndex(transport1, 'index', 20))
686
def test_iter_all_entries_reads(self):
687
# iterating all entries reads the header, then does a linear
689
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2,
691
# 100k nodes is enough to create a three-level index, which shows that
692
# we skip the internal nodes and just read the leaf nodes.
693
nodes = self.make_nodes(50000, 2, 2)
695
builder.add_node(*node)
696
transport = get_transport('trace+' + self.get_url(''))
697
size = transport.put_file('index', builder.finish())
698
self.assertEqual(2026722, size, 'number of expected bytes in the'
701
index = btree_index.BTreeGraphIndex(transport, 'index', size)
702
del transport._activity[:]
703
self.assertEqual([], transport._activity)
704
found_nodes = list(index.iter_all_entries())
706
for node in found_nodes:
707
self.assertTrue(node[0] is index)
708
bare_nodes.append(node[1:])
709
self.assertEqual(3, len(index._row_lengths),
710
"Not enough rows: %r" % index._row_lengths)
711
# Should be as long as the nodes we supplied
712
self.assertEqual(100000, len(found_nodes))
713
# Should have the same content
714
self.assertEqual(set(nodes), set(bare_nodes))
715
# Should have done linear scan IO up the index, ignoring
716
# the internal nodes:
717
# The entire index should have been read
718
total_pages = sum(index._row_lengths)
719
self.assertEqual(total_pages, index._row_offsets[-1])
720
self.assertEqual(2026722, size)
721
# The start of the leaves
722
first_byte = index._row_offsets[-2] * btree_index._PAGE_SIZE
724
for offset in range(first_byte, size, 4096):
725
readv_request.append((offset, 4096))
726
# The last page is truncated
727
readv_request[-1] = (readv_request[-1][0], 3298)
728
expected = [('readv', 'index', [(0, 4096)], False, None),
729
('readv', 'index', readv_request, False, None)]
730
if expected != transport._activity:
731
self.assertEqualDiff(pprint.pformat(expected),
732
pprint.pformat(transport._activity))
734
def _test_iter_entries_references_resolved(self):
735
index = self.make_index(1, nodes=[
736
(('name', ), 'data', ([('ref', ), ('ref', )], )),
737
(('ref', ), 'refdata', ([], ))])
738
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),('ref',)),)),
739
(index, ('ref', ), 'refdata', ((), ))]),
740
set(index.iter_entries([('name',), ('ref',)])))
742
def test_iter_entries_references_2_refs_resolved(self):
743
# iterating some entries reads just the pages needed. For now, to
744
# get it working and start measuring, only 4K pages are read.
745
builder = btree_index.BTreeBuilder(key_elements=2, reference_lists=2)
746
# 80 nodes is enough to create a two-level index.
747
nodes = self.make_nodes(160, 2, 2)
749
builder.add_node(*node)
750
transport = get_transport('trace+' + self.get_url(''))
751
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
found_nodes = list(index.iter_entries([nodes[30][0]]))
759
for node in found_nodes:
760
self.assertTrue(node[0] is index)
761
bare_nodes.append(node[1:])
762
# Should be as long as the nodes we supplied
763
self.assertEqual(1, len(found_nodes))
764
# Should have the same content
765
self.assertEqual(nodes[30], bare_nodes[0])
766
# Should have read the root node, then one leaf page:
767
self.assertEqual([('readv', 'index', [(0, 4096)], False, None),
768
('readv', 'index', [(4096, 4096), ], False, None)],
771
def test_iter_key_prefix_1_element_key_None(self):
772
index = self.make_index()
773
self.assertRaises(errors.BadIndexKey, list,
774
index.iter_entries_prefix([(None, )]))
776
def test_iter_key_prefix_wrong_length(self):
777
index = self.make_index()
778
self.assertRaises(errors.BadIndexKey, list,
779
index.iter_entries_prefix([('foo', None)]))
780
index = self.make_index(key_elements=2)
781
self.assertRaises(errors.BadIndexKey, list,
782
index.iter_entries_prefix([('foo', )]))
783
self.assertRaises(errors.BadIndexKey, list,
784
index.iter_entries_prefix([('foo', None, None)]))
786
def test_iter_key_prefix_1_key_element_no_refs(self):
787
index = self.make_index( nodes=[
788
(('name', ), 'data', ()),
789
(('ref', ), 'refdata', ())])
790
self.assertEqual(set([(index, ('name', ), 'data'),
791
(index, ('ref', ), 'refdata')]),
792
set(index.iter_entries_prefix([('name', ), ('ref', )])))
794
def test_iter_key_prefix_1_key_element_refs(self):
795
index = self.make_index(1, nodes=[
796
(('name', ), 'data', ([('ref', )], )),
797
(('ref', ), 'refdata', ([], ))])
798
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
799
(index, ('ref', ), 'refdata', ((), ))]),
800
set(index.iter_entries_prefix([('name', ), ('ref', )])))
802
def test_iter_key_prefix_2_key_element_no_refs(self):
803
index = self.make_index(key_elements=2, nodes=[
804
(('name', 'fin1'), 'data', ()),
805
(('name', 'fin2'), 'beta', ()),
806
(('ref', 'erence'), 'refdata', ())])
807
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
808
(index, ('ref', 'erence'), 'refdata')]),
809
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
810
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
811
(index, ('name', 'fin2'), 'beta')]),
812
set(index.iter_entries_prefix([('name', None)])))
814
def test_iter_key_prefix_2_key_element_refs(self):
815
index = self.make_index(1, key_elements=2, nodes=[
816
(('name', 'fin1'), 'data', ([('ref', 'erence')], )),
817
(('name', 'fin2'), 'beta', ([], )),
818
(('ref', 'erence'), 'refdata', ([], ))])
819
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
820
(index, ('ref', 'erence'), 'refdata', ((), ))]),
821
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
822
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
823
(index, ('name', 'fin2'), 'beta', ((), ))]),
824
set(index.iter_entries_prefix([('name', None)])))
827
class TestBTreeNodes(BTreeTestCase):
829
def restore_parser(self):
830
btree_index._btree_serializer = self.saved_parser
833
BTreeTestCase.setUp(self)
834
self.saved_parser = btree_index._btree_serializer
835
self.addCleanup(self.restore_parser)
836
btree_index._btree_serializer = self.parse_btree
838
def test_LeafNode_1_0(self):
839
node_bytes = ("type=leaf\n"
840
"0000000000000000000000000000000000000000\x00\x00value:0\n"
841
"1111111111111111111111111111111111111111\x00\x00value:1\n"
842
"2222222222222222222222222222222222222222\x00\x00value:2\n"
843
"3333333333333333333333333333333333333333\x00\x00value:3\n"
844
"4444444444444444444444444444444444444444\x00\x00value:4\n")
845
node = btree_index._LeafNode(node_bytes, 1, 0)
846
# We do direct access, or don't care about order, to leaf nodes most of
847
# the time, so a dict is useful:
849
("0000000000000000000000000000000000000000",): ("value:0", ()),
850
("1111111111111111111111111111111111111111",): ("value:1", ()),
851
("2222222222222222222222222222222222222222",): ("value:2", ()),
852
("3333333333333333333333333333333333333333",): ("value:3", ()),
853
("4444444444444444444444444444444444444444",): ("value:4", ()),
856
def test_LeafNode_2_2(self):
857
node_bytes = ("type=leaf\n"
858
"00\x0000\x00\t00\x00ref00\x00value:0\n"
859
"00\x0011\x0000\x00ref00\t00\x00ref00\r01\x00ref01\x00value:1\n"
860
"11\x0033\x0011\x00ref22\t11\x00ref22\r11\x00ref22\x00value:3\n"
861
"11\x0044\x00\t11\x00ref00\x00value:4\n"
864
node = btree_index._LeafNode(node_bytes, 2, 2)
865
# We do direct access, or don't care about order, to leaf nodes most of
866
# the time, so a dict is useful:
868
('00', '00'): ('value:0', ((), (('00', 'ref00'),))),
869
('00', '11'): ('value:1',
870
((('00', 'ref00'),), (('00', 'ref00'), ('01', 'ref01')))),
871
('11', '33'): ('value:3',
872
((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
873
('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
876
def test_InternalNode_1(self):
877
node_bytes = ("type=internal\n"
879
"0000000000000000000000000000000000000000\n"
880
"1111111111111111111111111111111111111111\n"
881
"2222222222222222222222222222222222222222\n"
882
"3333333333333333333333333333333333333333\n"
883
"4444444444444444444444444444444444444444\n"
885
node = btree_index._InternalNode(node_bytes)
886
# We want to bisect to find the right children from this node, so a
887
# vector is most useful.
889
("0000000000000000000000000000000000000000",),
890
("1111111111111111111111111111111111111111",),
891
("2222222222222222222222222222222222222222",),
892
("3333333333333333333333333333333333333333",),
893
("4444444444444444444444444444444444444444",),
895
self.assertEqual(1, node.offset)
897
def test_LeafNode_2_2(self):
898
node_bytes = ("type=leaf\n"
899
"00\x0000\x00\t00\x00ref00\x00value:0\n"
900
"00\x0011\x0000\x00ref00\t00\x00ref00\r01\x00ref01\x00value:1\n"
901
"11\x0033\x0011\x00ref22\t11\x00ref22\r11\x00ref22\x00value:3\n"
902
"11\x0044\x00\t11\x00ref00\x00value:4\n"
905
node = btree_index._LeafNode(node_bytes, 2, 2)
906
# We do direct access, or don't care about order, to leaf nodes most of
907
# the time, so a dict is useful:
909
('00', '00'): ('value:0', ((), (('00', 'ref00'),))),
910
('00', '11'): ('value:1',
911
((('00', 'ref00'),), (('00', 'ref00'), ('01', 'ref01')))),
912
('11', '33'): ('value:3',
913
((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22')))),
914
('11', '44'): ('value:4', ((), (('11', 'ref00'),)))
917
def assertFlattened(self, expected, key, value, refs):
918
flat_key, flat_line = self.parse_btree._flatten_node(
919
(None, key, value, refs), bool(refs))
920
self.assertEqual('\x00'.join(key), flat_key)
921
self.assertEqual(expected, flat_line)
923
def test__flatten_node(self):
924
self.assertFlattened('key\0\0value\n', ('key',), 'value', [])
925
self.assertFlattened('key\0tuple\0\0value str\n',
926
('key', 'tuple'), 'value str', [])
927
self.assertFlattened('key\0tuple\0triple\0\0value str\n',
928
('key', 'tuple', 'triple'), 'value str', [])
929
self.assertFlattened('k\0t\0s\0ref\0value str\n',
930
('k', 't', 's'), 'value str', [[('ref',)]])
931
self.assertFlattened('key\0tuple\0ref\0key\0value str\n',
932
('key', 'tuple'), 'value str', [[('ref', 'key')]])
933
self.assertFlattened("00\x0000\x00\t00\x00ref00\x00value:0\n",
934
('00', '00'), 'value:0', ((), (('00', 'ref00'),)))
935
self.assertFlattened(
936
"00\x0011\x0000\x00ref00\t00\x00ref00\r01\x00ref01\x00value:1\n",
937
('00', '11'), 'value:1',
938
((('00', 'ref00'),), (('00', 'ref00'), ('01', 'ref01'))))
939
self.assertFlattened(
940
"11\x0033\x0011\x00ref22\t11\x00ref22\r11\x00ref22\x00value:3\n",
941
('11', '33'), 'value:3',
942
((('11', 'ref22'),), (('11', 'ref22'), ('11', 'ref22'))))
943
self.assertFlattened(
944
"11\x0044\x00\t11\x00ref00\x00value:4\n",
945
('11', '44'), 'value:4', ((), (('11', 'ref00'),)))
948
class TestCompiledBtree(tests.TestCase):
950
def test_exists(self):
951
# This is just to let the user know if they don't have the feature
953
self.requireFeature(CompiledBtreeParserFeature)
956
class TestMultiBisectRight(tests.TestCase):
958
def assertMultiBisectRight(self, offsets, search_keys, fixed_keys):
959
self.assertEqual(offsets,
960
btree_index.BTreeGraphIndex._multi_bisect_right(
961
search_keys, fixed_keys))
963
def test_after(self):
964
self.assertMultiBisectRight([(1, ['b'])], ['b'], ['a'])
965
self.assertMultiBisectRight([(3, ['e', 'f', 'g'])],
966
['e', 'f', 'g'], ['a', 'b', 'c'])
968
def test_before(self):
969
self.assertMultiBisectRight([(0, ['a'])], ['a'], ['b'])
970
self.assertMultiBisectRight([(0, ['a', 'b', 'c', 'd'])],
971
['a', 'b', 'c', 'd'], ['e', 'f', 'g'])
973
def test_exact(self):
974
self.assertMultiBisectRight([(1, ['a'])], ['a'], ['a'])
975
self.assertMultiBisectRight([(1, ['a']), (2, ['b'])], ['a', 'b'], ['a', 'b'])
976
self.assertMultiBisectRight([(1, ['a']), (3, ['c'])],
977
['a', 'c'], ['a', 'b', 'c'])
979
def test_inbetween(self):
980
self.assertMultiBisectRight([(1, ['b'])], ['b'], ['a', 'c'])
981
self.assertMultiBisectRight([(1, ['b', 'c', 'd']), (2, ['f', 'g'])],
982
['b', 'c', 'd', 'f', 'g'], ['a', 'e', 'h'])
984
def test_mixed(self):
985
self.assertMultiBisectRight([(0, ['a', 'b']), (2, ['d', 'e']),
987
['a', 'b', 'd', 'e', 'g', 'h'],
988
['c', 'd', 'f', 'g'])
added
removed
bzrlib/tests/test_btree_index.py
