1
# Copyright (C) 2007 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
17
"""Tests for indices."""
19
from bzrlib import errors
20
from bzrlib.index import *
21
from bzrlib.tests import TestCaseWithMemoryTransport
22
from bzrlib.transport import get_transport
25
class TestGraphIndexBuilder(TestCaseWithMemoryTransport):
27
def test_build_index_empty(self):
28
builder = GraphIndexBuilder()
29
stream = builder.finish()
30
contents = stream.read()
32
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=1\nlen=0\n\n",
35
def test_build_index_empty_two_element_keys(self):
36
builder = GraphIndexBuilder(key_elements=2)
37
stream = builder.finish()
38
contents = stream.read()
40
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=2\nlen=0\n\n",
43
def test_build_index_one_reference_list_empty(self):
44
builder = GraphIndexBuilder(reference_lists=1)
45
stream = builder.finish()
46
contents = stream.read()
48
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=1\nlen=0\n\n",
51
def test_build_index_two_reference_list_empty(self):
52
builder = GraphIndexBuilder(reference_lists=2)
53
stream = builder.finish()
54
contents = stream.read()
56
"Bazaar Graph Index 1\nnode_ref_lists=2\nkey_elements=1\nlen=0\n\n",
59
def test_build_index_one_node_no_refs(self):
60
builder = GraphIndexBuilder()
61
builder.add_node(('akey', ), 'data')
62
stream = builder.finish()
63
contents = stream.read()
65
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=1\nlen=1\n"
66
"akey\x00\x00\x00data\n\n", contents)
68
def test_build_index_one_node_no_refs_accepts_empty_reflist(self):
69
builder = GraphIndexBuilder()
70
builder.add_node(('akey', ), 'data', ())
71
stream = builder.finish()
72
contents = stream.read()
74
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=1\nlen=1\n"
75
"akey\x00\x00\x00data\n\n", contents)
77
def test_build_index_one_node_2_element_keys(self):
78
# multipart keys are separated by \x00 - because they are fixed length,
79
# not variable this does not cause any issues, and seems clearer to the
81
builder = GraphIndexBuilder(key_elements=2)
82
builder.add_node(('akey', 'secondpart'), 'data')
83
stream = builder.finish()
84
contents = stream.read()
86
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=2\nlen=1\n"
87
"akey\x00secondpart\x00\x00\x00data\n\n", contents)
89
def test_add_node_empty_value(self):
90
builder = GraphIndexBuilder()
91
builder.add_node(('akey', ), '')
92
stream = builder.finish()
93
contents = stream.read()
95
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=1\nlen=1\n"
96
"akey\x00\x00\x00\n\n", contents)
98
def test_build_index_nodes_sorted(self):
99
# the highest sorted node comes first.
100
builder = GraphIndexBuilder()
101
# use three to have a good chance of glitching dictionary hash
102
# lookups etc. Insert in randomish order that is not correct
103
# and not the reverse of the correct order.
104
builder.add_node(('2002', ), 'data')
105
builder.add_node(('2000', ), 'data')
106
builder.add_node(('2001', ), 'data')
107
stream = builder.finish()
108
contents = stream.read()
110
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=1\nlen=3\n"
111
"2000\x00\x00\x00data\n"
112
"2001\x00\x00\x00data\n"
113
"2002\x00\x00\x00data\n"
116
def test_build_index_2_element_key_nodes_sorted(self):
117
# multiple element keys are sorted first-key, second-key.
118
builder = GraphIndexBuilder(key_elements=2)
119
# use three values of each key element, to have a good chance of
120
# glitching dictionary hash lookups etc. Insert in randomish order that
121
# is not correct and not the reverse of the correct order.
122
builder.add_node(('2002', '2002'), 'data')
123
builder.add_node(('2002', '2000'), 'data')
124
builder.add_node(('2002', '2001'), 'data')
125
builder.add_node(('2000', '2002'), 'data')
126
builder.add_node(('2000', '2000'), 'data')
127
builder.add_node(('2000', '2001'), 'data')
128
builder.add_node(('2001', '2002'), 'data')
129
builder.add_node(('2001', '2000'), 'data')
130
builder.add_node(('2001', '2001'), 'data')
131
stream = builder.finish()
132
contents = stream.read()
134
"Bazaar Graph Index 1\nnode_ref_lists=0\nkey_elements=2\nlen=9\n"
135
"2000\x002000\x00\x00\x00data\n"
136
"2000\x002001\x00\x00\x00data\n"
137
"2000\x002002\x00\x00\x00data\n"
138
"2001\x002000\x00\x00\x00data\n"
139
"2001\x002001\x00\x00\x00data\n"
140
"2001\x002002\x00\x00\x00data\n"
141
"2002\x002000\x00\x00\x00data\n"
142
"2002\x002001\x00\x00\x00data\n"
143
"2002\x002002\x00\x00\x00data\n"
146
def test_build_index_reference_lists_are_included_one(self):
147
builder = GraphIndexBuilder(reference_lists=1)
148
builder.add_node(('key', ), 'data', ([], ))
149
stream = builder.finish()
150
contents = stream.read()
152
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=1\nlen=1\n"
153
"key\x00\x00\x00data\n"
156
def test_build_index_reference_lists_with_2_element_keys(self):
157
builder = GraphIndexBuilder(reference_lists=1, key_elements=2)
158
builder.add_node(('key', 'key2'), 'data', ([], ))
159
stream = builder.finish()
160
contents = stream.read()
162
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=2\nlen=1\n"
163
"key\x00key2\x00\x00\x00data\n"
166
def test_build_index_reference_lists_are_included_two(self):
167
builder = GraphIndexBuilder(reference_lists=2)
168
builder.add_node(('key', ), 'data', ([], []))
169
stream = builder.finish()
170
contents = stream.read()
172
"Bazaar Graph Index 1\nnode_ref_lists=2\nkey_elements=1\nlen=1\n"
173
"key\x00\x00\t\x00data\n"
176
def test_node_references_are_byte_offsets(self):
177
builder = GraphIndexBuilder(reference_lists=1)
178
builder.add_node(('reference', ), 'data', ([], ))
179
builder.add_node(('key', ), 'data', ([('reference', )], ))
180
stream = builder.finish()
181
contents = stream.read()
183
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=1\nlen=2\n"
184
"key\x00\x0072\x00data\n"
185
"reference\x00\x00\x00data\n"
188
def test_node_references_are_cr_delimited(self):
189
builder = GraphIndexBuilder(reference_lists=1)
190
builder.add_node(('reference', ), 'data', ([], ))
191
builder.add_node(('reference2', ), 'data', ([], ))
192
builder.add_node(('key', ), 'data', ([('reference', ), ('reference2', )], ))
193
stream = builder.finish()
194
contents = stream.read()
196
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=1\nlen=3\n"
197
"key\x00\x00077\r094\x00data\n"
198
"reference\x00\x00\x00data\n"
199
"reference2\x00\x00\x00data\n"
202
def test_multiple_reference_lists_are_tab_delimited(self):
203
builder = GraphIndexBuilder(reference_lists=2)
204
builder.add_node(('keference', ), 'data', ([], []))
205
builder.add_node(('rey', ), 'data', ([('keference', )], [('keference', )]))
206
stream = builder.finish()
207
contents = stream.read()
209
"Bazaar Graph Index 1\nnode_ref_lists=2\nkey_elements=1\nlen=2\n"
210
"keference\x00\x00\t\x00data\n"
211
"rey\x00\x0059\t59\x00data\n"
214
def test_add_node_referencing_missing_key_makes_absent(self):
215
builder = GraphIndexBuilder(reference_lists=1)
216
builder.add_node(('rey', ), 'data', ([('beference', ), ('aeference2', )], ))
217
stream = builder.finish()
218
contents = stream.read()
220
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=1\nlen=1\n"
221
"aeference2\x00a\x00\x00\n"
222
"beference\x00a\x00\x00\n"
223
"rey\x00\x00074\r059\x00data\n"
226
def test_node_references_three_digits(self):
227
# test the node digit expands as needed.
228
builder = GraphIndexBuilder(reference_lists=1)
229
references = [(str(val), ) for val in reversed(range(9))]
230
builder.add_node(('2-key', ), '', (references, ))
231
stream = builder.finish()
232
contents = stream.read()
234
"Bazaar Graph Index 1\nnode_ref_lists=1\nkey_elements=1\nlen=1\n"
238
"2-key\x00\x00151\r145\r139\r133\r127\r121\r071\r065\r059\x00\n"
247
def test_absent_has_no_reference_overhead(self):
248
# the offsets after an absent record should be correct when there are
249
# >1 reference lists.
250
builder = GraphIndexBuilder(reference_lists=2)
251
builder.add_node(('parent', ), '', ([('aail', ), ('zther', )], []))
252
stream = builder.finish()
253
contents = stream.read()
255
"Bazaar Graph Index 1\nnode_ref_lists=2\nkey_elements=1\nlen=1\n"
256
"aail\x00a\x00\x00\n"
257
"parent\x00\x0059\r84\t\x00\n"
258
"zther\x00a\x00\x00\n"
261
def test_add_node_bad_key(self):
262
builder = GraphIndexBuilder()
263
for bad_char in '\t\n\x0b\x0c\r\x00 ':
264
self.assertRaises(errors.BadIndexKey, builder.add_node,
265
('a%skey' % bad_char, ), 'data')
266
self.assertRaises(errors.BadIndexKey, builder.add_node,
268
self.assertRaises(errors.BadIndexKey, builder.add_node,
269
'not-a-tuple', 'data')
271
self.assertRaises(errors.BadIndexKey, builder.add_node,
274
self.assertRaises(errors.BadIndexKey, builder.add_node,
275
('primary', 'secondary'), 'data')
276
# secondary key elements get checked too:
277
builder = GraphIndexBuilder(key_elements=2)
278
for bad_char in '\t\n\x0b\x0c\r\x00 ':
279
self.assertRaises(errors.BadIndexKey, builder.add_node,
280
('prefix', 'a%skey' % bad_char), 'data')
282
def test_add_node_bad_data(self):
283
builder = GraphIndexBuilder()
284
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
286
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
289
def test_add_node_bad_mismatched_ref_lists_length(self):
290
builder = GraphIndexBuilder()
291
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
293
builder = GraphIndexBuilder(reference_lists=1)
294
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
296
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
298
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
300
builder = GraphIndexBuilder(reference_lists=2)
301
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
303
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
305
self.assertRaises(errors.BadIndexValue, builder.add_node, ('akey', ),
306
'data aa', ([], [], []))
308
def test_add_node_bad_key_in_reference_lists(self):
309
# first list, first key - trivial
310
builder = GraphIndexBuilder(reference_lists=1)
311
self.assertRaises(errors.BadIndexKey, builder.add_node, ('akey', ),
312
'data aa', ([('a key', )], ))
313
# references keys must be tuples too
314
self.assertRaises(errors.BadIndexKey, builder.add_node, ('akey', ),
315
'data aa', (['not-a-tuple'], ))
317
self.assertRaises(errors.BadIndexKey, builder.add_node, ('akey', ),
320
self.assertRaises(errors.BadIndexKey, builder.add_node, ('akey', ),
321
'data aa', ([('primary', 'secondary')], ))
322
# need to check more than the first key in the list
323
self.assertRaises(errors.BadIndexKey, builder.add_node, ('akey', ),
324
'data aa', ([('agoodkey', ), ('that is a bad key', )], ))
325
# and if there is more than one list it should be getting checked
327
builder = GraphIndexBuilder(reference_lists=2)
328
self.assertRaises(errors.BadIndexKey, builder.add_node, ('akey', ),
329
'data aa', ([], ['a bad key']))
331
def test_add_duplicate_key(self):
332
builder = GraphIndexBuilder()
333
builder.add_node(('key', ), 'data')
334
self.assertRaises(errors.BadIndexDuplicateKey, builder.add_node, ('key', ),
337
def test_add_duplicate_key_2_elements(self):
338
builder = GraphIndexBuilder(key_elements=2)
339
builder.add_node(('key', 'key'), 'data')
340
self.assertRaises(errors.BadIndexDuplicateKey, builder.add_node,
341
('key', 'key'), 'data')
343
def test_add_key_after_referencing_key(self):
344
builder = GraphIndexBuilder(reference_lists=1)
345
builder.add_node(('key', ), 'data', ([('reference', )], ))
346
builder.add_node(('reference', ), 'data', ([],))
348
def test_add_key_after_referencing_key_2_elements(self):
349
builder = GraphIndexBuilder(reference_lists=1, key_elements=2)
350
builder.add_node(('k', 'ey'), 'data', ([('reference', 'tokey')], ))
351
builder.add_node(('reference', 'tokey'), 'data', ([],))
353
def test_set_optimize(self):
354
builder = GraphIndexBuilder(reference_lists=1, key_elements=2)
355
builder.set_optimize(for_size=True)
356
self.assertTrue(builder._optimize_for_size)
357
builder.set_optimize(for_size=False)
358
self.assertFalse(builder._optimize_for_size)
361
class TestGraphIndex(TestCaseWithMemoryTransport):
363
def make_key(self, number):
364
return (str(number) + 'X'*100,)
366
def make_value(self, number):
367
return str(number) + 'Y'*100
369
def make_nodes(self, count=64):
370
# generate a big enough index that we only read some of it on a typical
373
for counter in range(count):
374
nodes.append((self.make_key(counter), self.make_value(counter), ()))
377
def make_index(self, ref_lists=0, key_elements=1, nodes=[]):
378
builder = GraphIndexBuilder(ref_lists, key_elements=key_elements)
379
for key, value, references in nodes:
380
builder.add_node(key, value, references)
381
stream = builder.finish()
382
trans = get_transport('trace+' + self.get_url())
383
size = trans.put_file('index', stream)
384
return GraphIndex(trans, 'index', size)
386
def test_open_bad_index_no_error(self):
387
trans = self.get_transport()
388
trans.put_bytes('name', "not an index\n")
389
index = GraphIndex(trans, 'name', 13)
391
def test_open_sets_parsed_map_empty(self):
392
index = self.make_index()
393
self.assertEqual([], index._parsed_byte_map)
394
self.assertEqual([], index._parsed_key_map)
396
def test_key_count_buffers(self):
397
index = self.make_index(nodes=self.make_nodes(2))
398
# reset the transport log
399
del index._transport._activity[:]
400
self.assertEqual(2, index.key_count())
401
# We should have requested reading the header bytes
403
('readv', 'index', [(0, 200)], True, index._size),
405
index._transport._activity)
406
# And that should have been enough to trigger reading the whole index
408
self.assertIsNot(None, index._nodes)
410
def test_lookup_key_via_location_buffers(self):
411
index = self.make_index()
412
# reset the transport log
413
del index._transport._activity[:]
414
# do a _lookup_keys_via_location call for the middle of the file, which
415
# is what bisection uses.
416
result = index._lookup_keys_via_location(
417
[(index._size // 2, ('missing', ))])
418
# this should have asked for a readv request, with adjust_for_latency,
419
# and two regions: the header, and half-way into the file.
421
('readv', 'index', [(30, 30), (0, 200)], True, 60),
423
index._transport._activity)
424
# and the result should be that the key cannot be present, because this
425
# is a trivial index.
426
self.assertEqual([((index._size // 2, ('missing', )), False)],
428
# And this should have caused the file to be fully buffered
429
self.assertIsNot(None, index._nodes)
430
self.assertEqual([], index._parsed_byte_map)
432
def test_first_lookup_key_via_location(self):
433
# We need enough data so that the _HEADER_READV doesn't consume the
434
# whole file. We always read 800 bytes for every key, and the local
435
# transport natural expansion is 4096 bytes. So we have to have >8192
436
# bytes or we will trigger "buffer_all".
437
# We also want the 'missing' key to fall within the range that *did*
440
index = self.make_index(nodes=self.make_nodes(64))
441
# reset the transport log
442
del index._transport._activity[:]
443
# do a _lookup_keys_via_location call for the middle of the file, which
444
# is what bisection uses.
445
start_lookup = index._size // 2
446
result = index._lookup_keys_via_location(
447
[(start_lookup, ('40missing', ))])
448
# this should have asked for a readv request, with adjust_for_latency,
449
# and two regions: the header, and half-way into the file.
452
[(start_lookup, 800), (0, 200)], True, index._size),
454
index._transport._activity)
455
# and the result should be that the key cannot be present, because this
456
# is a trivial index.
457
self.assertEqual([((start_lookup, ('40missing', )), False)],
459
# And this should not have caused the file to be fully buffered
460
self.assertIs(None, index._nodes)
461
# And the regions of the file that have been parsed should be in the
462
# parsed_byte_map and the parsed_key_map
463
self.assertEqual([(0, 4008), (5046, 8996)], index._parsed_byte_map)
464
self.assertEqual([(None, self.make_key(26)),
465
(self.make_key(31), self.make_key(48))],
466
index._parsed_key_map)
468
def test_parsing_non_adjacent_data_trims(self):
469
index = self.make_index(nodes=self.make_nodes(64))
470
result = index._lookup_keys_via_location(
471
[(index._size // 2, ('40', ))])
472
# and the result should be that the key cannot be present, because key is
473
# in the middle of the observed data from a 4K read - the smallest transport
474
# will do today with this api.
475
self.assertEqual([((index._size // 2, ('40', )), False)],
477
# and we should have a parse map that includes the header and the
478
# region that was parsed after trimming.
479
self.assertEqual([(0, 4008), (5046, 8996)], index._parsed_byte_map)
480
self.assertEqual([(None, self.make_key(26)),
481
(self.make_key(31), self.make_key(48))],
482
index._parsed_key_map)
484
def test_parsing_data_handles_parsed_contained_regions(self):
485
# the following patten creates a parsed region that is wholly within a
486
# single result from the readv layer:
487
# .... single-read (readv-minimum-size) ...
488
# which then trims the start and end so the parsed size is < readv
490
# then a dual lookup (or a reference lookup for that matter) which
491
# abuts or overlaps the parsed region on both sides will need to
492
# discard the data in the middle, but parse the end as well.
494
# we test this by doing a single lookup to seed the data, then
495
# a lookup for two keys that are present, and adjacent -
496
# we except both to be found, and the parsed byte map to include the
497
# locations of both keys.
498
index = self.make_index(nodes=self.make_nodes(128))
499
result = index._lookup_keys_via_location(
500
[(index._size // 2, ('40', ))])
501
# and we should have a parse map that includes the header and the
502
# region that was parsed after trimming.
503
self.assertEqual([(0, 4045), (11759, 15707)], index._parsed_byte_map)
504
self.assertEqual([(None, self.make_key(116)),
505
(self.make_key(35), self.make_key(51))],
506
index._parsed_key_map)
507
# now ask for two keys, right before and after the parsed region
508
result = index._lookup_keys_via_location(
509
[(11450, self.make_key(34)), (15707, self.make_key(52))])
511
((11450, self.make_key(34)),
512
(index, self.make_key(34), self.make_value(34))),
513
((15707, self.make_key(52)),
514
(index, self.make_key(52), self.make_value(52))),
517
self.assertEqual([(0, 4045), (9889, 17993)], index._parsed_byte_map)
519
def test_lookup_missing_key_answers_without_io_when_map_permits(self):
520
# generate a big enough index that we only read some of it on a typical
522
index = self.make_index(nodes=self.make_nodes(64))
523
# lookup the keys in the middle of the file
524
result =index._lookup_keys_via_location(
525
[(index._size // 2, ('40', ))])
526
# check the parse map, this determines the test validity
527
self.assertEqual([(0, 4008), (5046, 8996)], index._parsed_byte_map)
528
self.assertEqual([(None, self.make_key(26)),
529
(self.make_key(31), self.make_key(48))],
530
index._parsed_key_map)
531
# reset the transport log
532
del index._transport._activity[:]
533
# now looking up a key in the portion of the file already parsed should
534
# not create a new transport request, and should return False (cannot
535
# be in the index) - even when the byte location we ask for is outside
537
result = index._lookup_keys_via_location(
539
self.assertEqual([((4000, ('40', )), False)],
541
self.assertEqual([], index._transport._activity)
543
def test_lookup_present_key_answers_without_io_when_map_permits(self):
544
# generate a big enough index that we only read some of it on a typical
546
index = self.make_index(nodes=self.make_nodes(64))
547
# lookup the keys in the middle of the file
548
result =index._lookup_keys_via_location(
549
[(index._size // 2, ('40', ))])
550
# check the parse map, this determines the test validity
551
self.assertEqual([(0, 4008), (5046, 8996)], index._parsed_byte_map)
552
self.assertEqual([(None, self.make_key(26)),
553
(self.make_key(31), self.make_key(48))],
554
index._parsed_key_map)
555
# reset the transport log
556
del index._transport._activity[:]
557
# now looking up a key in the portion of the file already parsed should
558
# not create a new transport request, and should return False (cannot
559
# be in the index) - even when the byte location we ask for is outside
562
result = index._lookup_keys_via_location([(4000, self.make_key(40))])
564
[((4000, self.make_key(40)),
565
(index, self.make_key(40), self.make_value(40)))],
567
self.assertEqual([], index._transport._activity)
569
def test_lookup_key_below_probed_area(self):
570
# generate a big enough index that we only read some of it on a typical
572
index = self.make_index(nodes=self.make_nodes(64))
573
# ask for the key in the middle, but a key that is located in the
574
# unparsed region before the middle.
575
result =index._lookup_keys_via_location(
576
[(index._size // 2, ('30', ))])
577
# check the parse map, this determines the test validity
578
self.assertEqual([(0, 4008), (5046, 8996)], index._parsed_byte_map)
579
self.assertEqual([(None, self.make_key(26)),
580
(self.make_key(31), self.make_key(48))],
581
index._parsed_key_map)
582
self.assertEqual([((index._size // 2, ('30', )), -1)],
585
def test_lookup_key_above_probed_area(self):
586
# generate a big enough index that we only read some of it on a typical
588
index = self.make_index(nodes=self.make_nodes(64))
589
# ask for the key in the middle, but a key that is located in the
590
# unparsed region after the middle.
591
result =index._lookup_keys_via_location(
592
[(index._size // 2, ('50', ))])
593
# check the parse map, this determines the test validity
594
self.assertEqual([(0, 4008), (5046, 8996)], index._parsed_byte_map)
595
self.assertEqual([(None, self.make_key(26)),
596
(self.make_key(31), self.make_key(48))],
597
index._parsed_key_map)
598
self.assertEqual([((index._size // 2, ('50', )), +1)],
601
def test_lookup_key_resolves_references(self):
602
# generate a big enough index that we only read some of it on a typical
605
for counter in range(99):
606
nodes.append((self.make_key(counter), self.make_value(counter),
607
((self.make_key(counter + 20),),) ))
608
index = self.make_index(ref_lists=1, nodes=nodes)
609
# lookup a key in the middle that does not exist, so that when we can
610
# check that the referred-to-keys are not accessed automatically.
611
index_size = index._size
612
index_center = index_size // 2
613
result = index._lookup_keys_via_location(
614
[(index_center, ('40', ))])
615
# check the parse map - only the start and middle should have been
617
self.assertEqual([(0, 4027), (10198, 14028)], index._parsed_byte_map)
618
self.assertEqual([(None, self.make_key(17)),
619
(self.make_key(44), self.make_key(5))],
620
index._parsed_key_map)
621
# and check the transport activity likewise.
623
[('readv', 'index', [(index_center, 800), (0, 200)], True,
625
index._transport._activity)
626
# reset the transport log for testing the reference lookup
627
del index._transport._activity[:]
628
# now looking up a key in the portion of the file already parsed should
629
# only perform IO to resolve its key references.
630
result = index._lookup_keys_via_location([(11000, self.make_key(45))])
632
[((11000, self.make_key(45)),
633
(index, self.make_key(45), self.make_value(45),
634
((self.make_key(65),),)))],
636
self.assertEqual([('readv', 'index', [(15093, 800)], True, index_size)],
637
index._transport._activity)
639
def test_lookup_key_can_buffer_all(self):
641
for counter in range(64):
642
nodes.append((self.make_key(counter), self.make_value(counter),
643
((self.make_key(counter + 20),),) ))
644
index = self.make_index(ref_lists=1, nodes=nodes)
645
# lookup a key in the middle that does not exist, so that when we can
646
# check that the referred-to-keys are not accessed automatically.
647
index_size = index._size
648
index_center = index_size // 2
649
result = index._lookup_keys_via_location([(index_center, ('40', ))])
650
# check the parse map - only the start and middle should have been
652
self.assertEqual([(0, 3890), (6444, 10274)], index._parsed_byte_map)
653
self.assertEqual([(None, self.make_key(25)),
654
(self.make_key(37), self.make_key(52))],
655
index._parsed_key_map)
656
# and check the transport activity likewise.
658
[('readv', 'index', [(index_center, 800), (0, 200)], True,
660
index._transport._activity)
661
# reset the transport log for testing the reference lookup
662
del index._transport._activity[:]
663
# now looking up a key in the portion of the file already parsed should
664
# only perform IO to resolve its key references.
665
result = index._lookup_keys_via_location([(7000, self.make_key(40))])
667
[((7000, self.make_key(40)),
668
(index, self.make_key(40), self.make_value(40),
669
((self.make_key(60),),)))],
671
# Resolving the references would have required more data read, and we
672
# are already above the 50% threshold, so it triggered a _buffer_all
673
self.assertEqual([('get', 'index')], index._transport._activity)
675
def test_iter_all_entries_empty(self):
676
index = self.make_index()
677
self.assertEqual([], list(index.iter_all_entries()))
679
def test_iter_all_entries_simple(self):
680
index = self.make_index(nodes=[(('name', ), 'data', ())])
681
self.assertEqual([(index, ('name', ), 'data')],
682
list(index.iter_all_entries()))
684
def test_iter_all_entries_simple_2_elements(self):
685
index = self.make_index(key_elements=2,
686
nodes=[(('name', 'surname'), 'data', ())])
687
self.assertEqual([(index, ('name', 'surname'), 'data')],
688
list(index.iter_all_entries()))
690
def test_iter_all_entries_references_resolved(self):
691
index = self.make_index(1, nodes=[
692
(('name', ), 'data', ([('ref', )], )),
693
(('ref', ), 'refdata', ([], ))])
694
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
695
(index, ('ref', ), 'refdata', ((), ))]),
696
set(index.iter_all_entries()))
698
def test_iter_entries_buffers_once(self):
699
index = self.make_index(nodes=self.make_nodes(2))
700
# reset the transport log
701
del index._transport._activity[:]
702
self.assertEqual(set([(index, self.make_key(1), self.make_value(1))]),
703
set(index.iter_entries([self.make_key(1)])))
704
# We should have requested reading the header bytes
705
# But not needed any more than that because it would have triggered a
708
('readv', 'index', [(0, 200)], True, index._size),
710
index._transport._activity)
711
# And that should have been enough to trigger reading the whole index
713
self.assertIsNot(None, index._nodes)
715
def test_iter_entries_buffers_by_bytes_read(self):
716
index = self.make_index(nodes=self.make_nodes(64))
717
list(index.iter_entries([self.make_key(10)]))
718
# The first time through isn't enough to trigger a buffer all
719
self.assertIs(None, index._nodes)
720
self.assertEqual(4096, index._bytes_read)
721
# Grabbing a key in that same page won't trigger a buffer all, as we
722
# still haven't read 50% of the file
723
list(index.iter_entries([self.make_key(11)]))
724
self.assertIs(None, index._nodes)
725
self.assertEqual(4096, index._bytes_read)
726
# We haven't read more data, so reading outside the range won't trigger
727
# a buffer all right away
728
list(index.iter_entries([self.make_key(40)]))
729
self.assertIs(None, index._nodes)
730
self.assertEqual(8192, index._bytes_read)
731
# On the next pass, we will not trigger buffer all if the key is
732
# available without reading more
733
list(index.iter_entries([self.make_key(32)]))
734
self.assertIs(None, index._nodes)
735
# But if we *would* need to read more to resolve it, then we will
737
list(index.iter_entries([self.make_key(60)]))
738
self.assertIsNot(None, index._nodes)
740
def test_iter_entries_references_resolved(self):
741
index = self.make_index(1, nodes=[
742
(('name', ), 'data', ([('ref', ), ('ref', )], )),
743
(('ref', ), 'refdata', ([], ))])
744
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),('ref',)),)),
745
(index, ('ref', ), 'refdata', ((), ))]),
746
set(index.iter_entries([('name',), ('ref',)])))
748
def test_iter_entries_references_2_refs_resolved(self):
749
index = self.make_index(2, nodes=[
750
(('name', ), 'data', ([('ref', )], [('ref', )])),
751
(('ref', ), 'refdata', ([], []))])
752
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),), (('ref',),))),
753
(index, ('ref', ), 'refdata', ((), ()))]),
754
set(index.iter_entries([('name',), ('ref',)])))
756
def test_iteration_absent_skipped(self):
757
index = self.make_index(1, nodes=[
758
(('name', ), 'data', ([('ref', )], ))])
759
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),))]),
760
set(index.iter_all_entries()))
761
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),))]),
762
set(index.iter_entries([('name', )])))
763
self.assertEqual([], list(index.iter_entries([('ref', )])))
765
def test_iteration_absent_skipped_2_element_keys(self):
766
index = self.make_index(1, key_elements=2, nodes=[
767
(('name', 'fin'), 'data', ([('ref', 'erence')], ))])
768
self.assertEqual(set([(index, ('name', 'fin'), 'data', ((('ref', 'erence'),),))]),
769
set(index.iter_all_entries()))
770
self.assertEqual(set([(index, ('name', 'fin'), 'data', ((('ref', 'erence'),),))]),
771
set(index.iter_entries([('name', 'fin')])))
772
self.assertEqual([], list(index.iter_entries([('ref', 'erence')])))
774
def test_iter_all_keys(self):
775
index = self.make_index(1, nodes=[
776
(('name', ), 'data', ([('ref', )], )),
777
(('ref', ), 'refdata', ([], ))])
778
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
779
(index, ('ref', ), 'refdata', ((), ))]),
780
set(index.iter_entries([('name', ), ('ref', )])))
782
def test_iter_nothing_empty(self):
783
index = self.make_index()
784
self.assertEqual([], list(index.iter_entries([])))
786
def test_iter_missing_entry_empty(self):
787
index = self.make_index()
788
self.assertEqual([], list(index.iter_entries([('a', )])))
790
def test_iter_missing_entry_empty_no_size(self):
791
index = self.make_index()
792
index = GraphIndex(index._transport, 'index', None)
793
self.assertEqual([], list(index.iter_entries([('a', )])))
795
def test_iter_key_prefix_1_element_key_None(self):
796
index = self.make_index()
797
self.assertRaises(errors.BadIndexKey, list,
798
index.iter_entries_prefix([(None, )]))
800
def test_iter_key_prefix_wrong_length(self):
801
index = self.make_index()
802
self.assertRaises(errors.BadIndexKey, list,
803
index.iter_entries_prefix([('foo', None)]))
804
index = self.make_index(key_elements=2)
805
self.assertRaises(errors.BadIndexKey, list,
806
index.iter_entries_prefix([('foo', )]))
807
self.assertRaises(errors.BadIndexKey, list,
808
index.iter_entries_prefix([('foo', None, None)]))
810
def test_iter_key_prefix_1_key_element_no_refs(self):
811
index = self.make_index( nodes=[
812
(('name', ), 'data', ()),
813
(('ref', ), 'refdata', ())])
814
self.assertEqual(set([(index, ('name', ), 'data'),
815
(index, ('ref', ), 'refdata')]),
816
set(index.iter_entries_prefix([('name', ), ('ref', )])))
818
def test_iter_key_prefix_1_key_element_refs(self):
819
index = self.make_index(1, nodes=[
820
(('name', ), 'data', ([('ref', )], )),
821
(('ref', ), 'refdata', ([], ))])
822
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
823
(index, ('ref', ), 'refdata', ((), ))]),
824
set(index.iter_entries_prefix([('name', ), ('ref', )])))
826
def test_iter_key_prefix_2_key_element_no_refs(self):
827
index = self.make_index(key_elements=2, nodes=[
828
(('name', 'fin1'), 'data', ()),
829
(('name', 'fin2'), 'beta', ()),
830
(('ref', 'erence'), 'refdata', ())])
831
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
832
(index, ('ref', 'erence'), 'refdata')]),
833
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
834
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
835
(index, ('name', 'fin2'), 'beta')]),
836
set(index.iter_entries_prefix([('name', None)])))
838
def test_iter_key_prefix_2_key_element_refs(self):
839
index = self.make_index(1, key_elements=2, nodes=[
840
(('name', 'fin1'), 'data', ([('ref', 'erence')], )),
841
(('name', 'fin2'), 'beta', ([], )),
842
(('ref', 'erence'), 'refdata', ([], ))])
843
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
844
(index, ('ref', 'erence'), 'refdata', ((), ))]),
845
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
846
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
847
(index, ('name', 'fin2'), 'beta', ((), ))]),
848
set(index.iter_entries_prefix([('name', None)])))
850
def test_key_count_empty(self):
851
index = self.make_index()
852
self.assertEqual(0, index.key_count())
854
def test_key_count_one(self):
855
index = self.make_index(nodes=[(('name', ), '', ())])
856
self.assertEqual(1, index.key_count())
858
def test_key_count_two(self):
859
index = self.make_index(nodes=[
860
(('name', ), '', ()), (('foo', ), '', ())])
861
self.assertEqual(2, index.key_count())
863
def test_read_and_parse_tracks_real_read_value(self):
864
index = self.make_index(nodes=self.make_nodes(10))
865
del index._transport._activity[:]
866
index._read_and_parse([(0, 200)])
868
('readv', 'index', [(0, 200)], True, index._size),
870
index._transport._activity)
871
# The readv expansion code will expand the initial request to 4096
872
# bytes, which is more than enough to read the entire index, and we
873
# will track the fact that we read that many bytes.
874
self.assertEqual(index._size, index._bytes_read)
876
def test_read_and_parse_triggers_buffer_all(self):
877
index = self.make_index(key_elements=2, nodes=[
878
(('name', 'fin1'), 'data', ()),
879
(('name', 'fin2'), 'beta', ()),
880
(('ref', 'erence'), 'refdata', ())])
881
self.assertTrue(index._size > 0)
882
self.assertIs(None, index._nodes)
883
index._read_and_parse([(0, index._size)])
884
self.assertIsNot(None, index._nodes)
886
def test_validate_bad_index_errors(self):
887
trans = self.get_transport()
888
trans.put_bytes('name', "not an index\n")
889
index = GraphIndex(trans, 'name', 13)
890
self.assertRaises(errors.BadIndexFormatSignature, index.validate)
892
def test_validate_bad_node_refs(self):
893
index = self.make_index(2)
894
trans = self.get_transport()
895
content = trans.get_bytes('index')
896
# change the options line to end with a rather than a parseable number
897
new_content = content[:-2] + 'a\n\n'
898
trans.put_bytes('index', new_content)
899
self.assertRaises(errors.BadIndexOptions, index.validate)
901
def test_validate_missing_end_line_empty(self):
902
index = self.make_index(2)
903
trans = self.get_transport()
904
content = trans.get_bytes('index')
905
# truncate the last byte
906
trans.put_bytes('index', content[:-1])
907
self.assertRaises(errors.BadIndexData, index.validate)
909
def test_validate_missing_end_line_nonempty(self):
910
index = self.make_index(2, nodes=[(('key', ), '', ([], []))])
911
trans = self.get_transport()
912
content = trans.get_bytes('index')
913
# truncate the last byte
914
trans.put_bytes('index', content[:-1])
915
self.assertRaises(errors.BadIndexData, index.validate)
917
def test_validate_empty(self):
918
index = self.make_index()
921
def test_validate_no_refs_content(self):
922
index = self.make_index(nodes=[(('key', ), 'value', ())])
925
# XXX: external_references tests are duplicated in test_btree_index. We
926
# probably should have per_graph_index tests...
927
def test_external_references_no_refs(self):
928
index = self.make_index(ref_lists=0, nodes=[])
929
self.assertRaises(ValueError, index.external_references, 0)
931
def test_external_references_no_results(self):
932
index = self.make_index(ref_lists=1, nodes=[
933
(('key',), 'value', ([],))])
934
self.assertEqual(set(), index.external_references(0))
936
def test_external_references_missing_ref(self):
937
missing_key = ('missing',)
938
index = self.make_index(ref_lists=1, nodes=[
939
(('key',), 'value', ([missing_key],))])
940
self.assertEqual(set([missing_key]), index.external_references(0))
942
def test_external_references_multiple_ref_lists(self):
943
missing_key = ('missing',)
944
index = self.make_index(ref_lists=2, nodes=[
945
(('key',), 'value', ([], [missing_key]))])
946
self.assertEqual(set([]), index.external_references(0))
947
self.assertEqual(set([missing_key]), index.external_references(1))
949
def test_external_references_two_records(self):
950
index = self.make_index(ref_lists=1, nodes=[
951
(('key-1',), 'value', ([('key-2',)],)),
952
(('key-2',), 'value', ([],)),
954
self.assertEqual(set([]), index.external_references(0))
956
def test__find_ancestors(self):
959
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
960
(key1, 'value', ([key2],)),
961
(key2, 'value', ([],)),
965
search_keys = index._find_ancestors([key1], 0, parent_map, missing_keys)
966
self.assertEqual({key1: (key2,)}, parent_map)
967
self.assertEqual(set(), missing_keys)
968
self.assertEqual(set([key2]), search_keys)
969
search_keys = index._find_ancestors(search_keys, 0, parent_map,
971
self.assertEqual({key1: (key2,), key2: ()}, parent_map)
972
self.assertEqual(set(), missing_keys)
973
self.assertEqual(set(), search_keys)
975
def test__find_ancestors_w_missing(self):
979
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
980
(key1, 'value', ([key2],)),
981
(key2, 'value', ([],)),
985
search_keys = index._find_ancestors([key2, key3], 0, parent_map,
987
self.assertEqual({key2: ()}, parent_map)
988
self.assertEqual(set([key3]), missing_keys)
989
self.assertEqual(set(), search_keys)
991
def test__find_ancestors_dont_search_known(self):
995
index = self.make_index(ref_lists=1, key_elements=1, nodes=[
996
(key1, 'value', ([key2],)),
997
(key2, 'value', ([key3],)),
998
(key3, 'value', ([],)),
1000
# We already know about key2, so we won't try to search for key3
1001
parent_map = {key2: (key3,)}
1002
missing_keys = set()
1003
search_keys = index._find_ancestors([key1], 0, parent_map,
1005
self.assertEqual({key1: (key2,), key2: (key3,)}, parent_map)
1006
self.assertEqual(set(), missing_keys)
1007
self.assertEqual(set(), search_keys)
1010
class TestCombinedGraphIndex(TestCaseWithMemoryTransport):
1012
def make_index(self, name, ref_lists=0, key_elements=1, nodes=[]):
1013
builder = GraphIndexBuilder(ref_lists, key_elements=key_elements)
1014
for key, value, references in nodes:
1015
builder.add_node(key, value, references)
1016
stream = builder.finish()
1017
trans = self.get_transport()
1018
size = trans.put_file(name, stream)
1019
return GraphIndex(trans, name, size)
1021
def make_combined_index_with_missing(self, missing=['1', '2']):
1022
"""Create a CombinedGraphIndex which will have missing indexes.
1024
This creates a CGI which thinks it has 2 indexes, however they have
1025
been deleted. If CGI._reload_func() is called, then it will repopulate
1028
:param missing: The underlying indexes to delete
1029
:return: (CombinedGraphIndex, reload_counter)
1031
index1 = self.make_index('1', nodes=[(('1',), '', ())])
1032
index2 = self.make_index('2', nodes=[(('2',), '', ())])
1033
index3 = self.make_index('3', nodes=[
1037
# total_reloads, num_changed, num_unchanged
1038
reload_counter = [0, 0, 0]
1040
reload_counter[0] += 1
1041
new_indices = [index3]
1042
if index._indices == new_indices:
1043
reload_counter[2] += 1
1045
reload_counter[1] += 1
1046
index._indices[:] = new_indices
1048
index = CombinedGraphIndex([index1, index2], reload_func=reload)
1049
trans = self.get_transport()
1050
for fname in missing:
1052
return index, reload_counter
1054
def test_open_missing_index_no_error(self):
1055
trans = self.get_transport()
1056
index1 = GraphIndex(trans, 'missing', 100)
1057
index = CombinedGraphIndex([index1])
1059
def test_add_index(self):
1060
index = CombinedGraphIndex([])
1061
index1 = self.make_index('name', 0, nodes=[(('key', ), '', ())])
1062
index.insert_index(0, index1)
1063
self.assertEqual([(index1, ('key', ), '')], list(index.iter_all_entries()))
1065
def test_iter_all_entries_empty(self):
1066
index = CombinedGraphIndex([])
1067
self.assertEqual([], list(index.iter_all_entries()))
1069
def test_iter_all_entries_children_empty(self):
1070
index1 = self.make_index('name')
1071
index = CombinedGraphIndex([index1])
1072
self.assertEqual([], list(index.iter_all_entries()))
1074
def test_iter_all_entries_simple(self):
1075
index1 = self.make_index('name', nodes=[(('name', ), 'data', ())])
1076
index = CombinedGraphIndex([index1])
1077
self.assertEqual([(index1, ('name', ), 'data')],
1078
list(index.iter_all_entries()))
1080
def test_iter_all_entries_two_indices(self):
1081
index1 = self.make_index('name1', nodes=[(('name', ), 'data', ())])
1082
index2 = self.make_index('name2', nodes=[(('2', ), '', ())])
1083
index = CombinedGraphIndex([index1, index2])
1084
self.assertEqual([(index1, ('name', ), 'data'),
1085
(index2, ('2', ), '')],
1086
list(index.iter_all_entries()))
1088
def test_iter_entries_two_indices_dup_key(self):
1089
index1 = self.make_index('name1', nodes=[(('name', ), 'data', ())])
1090
index2 = self.make_index('name2', nodes=[(('name', ), 'data', ())])
1091
index = CombinedGraphIndex([index1, index2])
1092
self.assertEqual([(index1, ('name', ), 'data')],
1093
list(index.iter_entries([('name', )])))
1095
def test_iter_all_entries_two_indices_dup_key(self):
1096
index1 = self.make_index('name1', nodes=[(('name', ), 'data', ())])
1097
index2 = self.make_index('name2', nodes=[(('name', ), 'data', ())])
1098
index = CombinedGraphIndex([index1, index2])
1099
self.assertEqual([(index1, ('name', ), 'data')],
1100
list(index.iter_all_entries()))
1102
def test_iter_key_prefix_2_key_element_refs(self):
1103
index1 = self.make_index('1', 1, key_elements=2, nodes=[
1104
(('name', 'fin1'), 'data', ([('ref', 'erence')], ))])
1105
index2 = self.make_index('2', 1, key_elements=2, nodes=[
1106
(('name', 'fin2'), 'beta', ([], )),
1107
(('ref', 'erence'), 'refdata', ([], ))])
1108
index = CombinedGraphIndex([index1, index2])
1109
self.assertEqual(set([(index1, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
1110
(index2, ('ref', 'erence'), 'refdata', ((), ))]),
1111
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
1112
self.assertEqual(set([(index1, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
1113
(index2, ('name', 'fin2'), 'beta', ((), ))]),
1114
set(index.iter_entries_prefix([('name', None)])))
1116
def test_iter_nothing_empty(self):
1117
index = CombinedGraphIndex([])
1118
self.assertEqual([], list(index.iter_entries([])))
1120
def test_iter_nothing_children_empty(self):
1121
index1 = self.make_index('name')
1122
index = CombinedGraphIndex([index1])
1123
self.assertEqual([], list(index.iter_entries([])))
1125
def test_iter_all_keys(self):
1126
index1 = self.make_index('1', 1, nodes=[
1127
(('name', ), 'data', ([('ref', )], ))])
1128
index2 = self.make_index('2', 1, nodes=[
1129
(('ref', ), 'refdata', ((), ))])
1130
index = CombinedGraphIndex([index1, index2])
1131
self.assertEqual(set([(index1, ('name', ), 'data', ((('ref', ), ), )),
1132
(index2, ('ref', ), 'refdata', ((), ))]),
1133
set(index.iter_entries([('name', ), ('ref', )])))
1135
def test_iter_all_keys_dup_entry(self):
1136
index1 = self.make_index('1', 1, nodes=[
1137
(('name', ), 'data', ([('ref', )], )),
1138
(('ref', ), 'refdata', ([], ))])
1139
index2 = self.make_index('2', 1, nodes=[
1140
(('ref', ), 'refdata', ([], ))])
1141
index = CombinedGraphIndex([index1, index2])
1142
self.assertEqual(set([(index1, ('name', ), 'data', ((('ref',),),)),
1143
(index1, ('ref', ), 'refdata', ((), ))]),
1144
set(index.iter_entries([('name', ), ('ref', )])))
1146
def test_iter_missing_entry_empty(self):
1147
index = CombinedGraphIndex([])
1148
self.assertEqual([], list(index.iter_entries([('a', )])))
1150
def test_iter_missing_entry_one_index(self):
1151
index1 = self.make_index('1')
1152
index = CombinedGraphIndex([index1])
1153
self.assertEqual([], list(index.iter_entries([('a', )])))
1155
def test_iter_missing_entry_two_index(self):
1156
index1 = self.make_index('1')
1157
index2 = self.make_index('2')
1158
index = CombinedGraphIndex([index1, index2])
1159
self.assertEqual([], list(index.iter_entries([('a', )])))
1161
def test_iter_entry_present_one_index_only(self):
1162
index1 = self.make_index('1', nodes=[(('key', ), '', ())])
1163
index2 = self.make_index('2', nodes=[])
1164
index = CombinedGraphIndex([index1, index2])
1165
self.assertEqual([(index1, ('key', ), '')],
1166
list(index.iter_entries([('key', )])))
1167
# and in the other direction
1168
index = CombinedGraphIndex([index2, index1])
1169
self.assertEqual([(index1, ('key', ), '')],
1170
list(index.iter_entries([('key', )])))
1172
def test_key_count_empty(self):
1173
index1 = self.make_index('1', nodes=[])
1174
index2 = self.make_index('2', nodes=[])
1175
index = CombinedGraphIndex([index1, index2])
1176
self.assertEqual(0, index.key_count())
1178
def test_key_count_sums_index_keys(self):
1179
index1 = self.make_index('1', nodes=[
1182
index2 = self.make_index('2', nodes=[(('1',), '', ())])
1183
index = CombinedGraphIndex([index1, index2])
1184
self.assertEqual(3, index.key_count())
1186
def test_validate_bad_child_index_errors(self):
1187
trans = self.get_transport()
1188
trans.put_bytes('name', "not an index\n")
1189
index1 = GraphIndex(trans, 'name', 13)
1190
index = CombinedGraphIndex([index1])
1191
self.assertRaises(errors.BadIndexFormatSignature, index.validate)
1193
def test_validate_empty(self):
1194
index = CombinedGraphIndex([])
1197
def test_key_count_reloads(self):
1198
index, reload_counter = self.make_combined_index_with_missing()
1199
self.assertEqual(2, index.key_count())
1200
self.assertEqual([1, 1, 0], reload_counter)
1202
def test_key_count_no_reload(self):
1203
index, reload_counter = self.make_combined_index_with_missing()
1204
index._reload_func = None
1205
# Without a _reload_func we just raise the exception
1206
self.assertRaises(errors.NoSuchFile, index.key_count)
1208
def test_key_count_reloads_and_fails(self):
1209
# We have deleted all underlying indexes, so we will try to reload, but
1210
# still fail. This is mostly to test we don't get stuck in an infinite
1211
# loop trying to reload
1212
index, reload_counter = self.make_combined_index_with_missing(
1214
self.assertRaises(errors.NoSuchFile, index.key_count)
1215
self.assertEqual([2, 1, 1], reload_counter)
1217
def test_iter_entries_reloads(self):
1218
index, reload_counter = self.make_combined_index_with_missing()
1219
result = list(index.iter_entries([('1',), ('2',), ('3',)]))
1220
index3 = index._indices[0]
1221
self.assertEqual([(index3, ('1',), ''), (index3, ('2',), '')],
1223
self.assertEqual([1, 1, 0], reload_counter)
1225
def test_iter_entries_reloads_midway(self):
1226
# The first index still looks present, so we get interrupted mid-way
1228
index, reload_counter = self.make_combined_index_with_missing(['2'])
1229
index1, index2 = index._indices
1230
result = list(index.iter_entries([('1',), ('2',), ('3',)]))
1231
index3 = index._indices[0]
1232
# We had already yielded '1', so we just go on to the next, we should
1233
# not yield '1' twice.
1234
self.assertEqual([(index1, ('1',), ''), (index3, ('2',), '')],
1236
self.assertEqual([1, 1, 0], reload_counter)
1238
def test_iter_entries_no_reload(self):
1239
index, reload_counter = self.make_combined_index_with_missing()
1240
index._reload_func = None
1241
# Without a _reload_func we just raise the exception
1242
self.assertListRaises(errors.NoSuchFile, index.iter_entries, [('3',)])
1244
def test_iter_entries_reloads_and_fails(self):
1245
index, reload_counter = self.make_combined_index_with_missing(
1247
self.assertListRaises(errors.NoSuchFile, index.iter_entries, [('3',)])
1248
self.assertEqual([2, 1, 1], reload_counter)
1250
def test_iter_all_entries_reloads(self):
1251
index, reload_counter = self.make_combined_index_with_missing()
1252
result = list(index.iter_all_entries())
1253
index3 = index._indices[0]
1254
self.assertEqual([(index3, ('1',), ''), (index3, ('2',), '')],
1256
self.assertEqual([1, 1, 0], reload_counter)
1258
def test_iter_all_entries_reloads_midway(self):
1259
index, reload_counter = self.make_combined_index_with_missing(['2'])
1260
index1, index2 = index._indices
1261
result = list(index.iter_all_entries())
1262
index3 = index._indices[0]
1263
# We had already yielded '1', so we just go on to the next, we should
1264
# not yield '1' twice.
1265
self.assertEqual([(index1, ('1',), ''), (index3, ('2',), '')],
1267
self.assertEqual([1, 1, 0], reload_counter)
1269
def test_iter_all_entries_no_reload(self):
1270
index, reload_counter = self.make_combined_index_with_missing()
1271
index._reload_func = None
1272
self.assertListRaises(errors.NoSuchFile, index.iter_all_entries)
1274
def test_iter_all_entries_reloads_and_fails(self):
1275
index, reload_counter = self.make_combined_index_with_missing(
1277
self.assertListRaises(errors.NoSuchFile, index.iter_all_entries)
1279
def test_iter_entries_prefix_reloads(self):
1280
index, reload_counter = self.make_combined_index_with_missing()
1281
result = list(index.iter_entries_prefix([('1',)]))
1282
index3 = index._indices[0]
1283
self.assertEqual([(index3, ('1',), '')], result)
1284
self.assertEqual([1, 1, 0], reload_counter)
1286
def test_iter_entries_prefix_reloads_midway(self):
1287
index, reload_counter = self.make_combined_index_with_missing(['2'])
1288
index1, index2 = index._indices
1289
result = list(index.iter_entries_prefix([('1',)]))
1290
index3 = index._indices[0]
1291
# We had already yielded '1', so we just go on to the next, we should
1292
# not yield '1' twice.
1293
self.assertEqual([(index1, ('1',), '')], result)
1294
self.assertEqual([1, 1, 0], reload_counter)
1296
def test_iter_entries_prefix_no_reload(self):
1297
index, reload_counter = self.make_combined_index_with_missing()
1298
index._reload_func = None
1299
self.assertListRaises(errors.NoSuchFile, index.iter_entries_prefix,
1302
def test_iter_entries_prefix_reloads_and_fails(self):
1303
index, reload_counter = self.make_combined_index_with_missing(
1305
self.assertListRaises(errors.NoSuchFile, index.iter_entries_prefix,
1308
def test_validate_reloads(self):
1309
index, reload_counter = self.make_combined_index_with_missing()
1311
self.assertEqual([1, 1, 0], reload_counter)
1313
def test_validate_reloads_midway(self):
1314
index, reload_counter = self.make_combined_index_with_missing(['2'])
1317
def test_validate_no_reload(self):
1318
index, reload_counter = self.make_combined_index_with_missing()
1319
index._reload_func = None
1320
self.assertRaises(errors.NoSuchFile, index.validate)
1322
def test_validate_reloads_and_fails(self):
1323
index, reload_counter = self.make_combined_index_with_missing(
1325
self.assertRaises(errors.NoSuchFile, index.validate)
1327
def test_find_ancestors_across_indexes(self):
1332
index1 = self.make_index('12', ref_lists=1, nodes=[
1333
(key1, 'value', ([],)),
1334
(key2, 'value', ([key1],)),
1336
index2 = self.make_index('34', ref_lists=1, nodes=[
1337
(key3, 'value', ([key2],)),
1338
(key4, 'value', ([key3],)),
1340
c_index = CombinedGraphIndex([index1, index2])
1341
parent_map, missing_keys = c_index.find_ancestry([key1], 0)
1342
self.assertEqual({key1: ()}, parent_map)
1343
self.assertEqual(set(), missing_keys)
1344
# Now look for a key from index2 which requires us to find the key in
1345
# the second index, and then continue searching for parents in the
1347
parent_map, missing_keys = c_index.find_ancestry([key3], 0)
1348
self.assertEqual({key1: (), key2: (key1,), key3: (key2,)}, parent_map)
1349
self.assertEqual(set(), missing_keys)
1351
def test_find_ancestors_missing_keys(self):
1356
index1 = self.make_index('12', ref_lists=1, nodes=[
1357
(key1, 'value', ([],)),
1358
(key2, 'value', ([key1],)),
1360
index2 = self.make_index('34', ref_lists=1, nodes=[
1361
(key3, 'value', ([key2],)),
1363
c_index = CombinedGraphIndex([index1, index2])
1364
# Searching for a key which is actually not present at all should
1365
# eventually converge
1366
parent_map, missing_keys = c_index.find_ancestry([key4], 0)
1367
self.assertEqual({}, parent_map)
1368
self.assertEqual(set([key4]), missing_keys)
1370
def test_find_ancestors_no_indexes(self):
1371
c_index = CombinedGraphIndex([])
1373
parent_map, missing_keys = c_index.find_ancestry([key1], 0)
1374
self.assertEqual({}, parent_map)
1375
self.assertEqual(set([key1]), missing_keys)
1377
def test_find_ancestors_ghost_parent(self):
1382
index1 = self.make_index('12', ref_lists=1, nodes=[
1383
(key1, 'value', ([],)),
1384
(key2, 'value', ([key1],)),
1386
index2 = self.make_index('34', ref_lists=1, nodes=[
1387
(key4, 'value', ([key2, key3],)),
1389
c_index = CombinedGraphIndex([index1, index2])
1390
# Searching for a key which is actually not present at all should
1391
# eventually converge
1392
parent_map, missing_keys = c_index.find_ancestry([key4], 0)
1393
self.assertEqual({key4: (key2, key3), key2: (key1,), key1: ()},
1395
self.assertEqual(set([key3]), missing_keys)
1397
def test__find_ancestors_empty_index(self):
1398
index = self.make_index('test', ref_lists=1, key_elements=1, nodes=[])
1400
missing_keys = set()
1401
search_keys = index._find_ancestors([('one',), ('two',)], 0, parent_map,
1403
self.assertEqual(set(), search_keys)
1404
self.assertEqual({}, parent_map)
1405
self.assertEqual(set([('one',), ('two',)]), missing_keys)
1408
class TestInMemoryGraphIndex(TestCaseWithMemoryTransport):
1410
def make_index(self, ref_lists=0, key_elements=1, nodes=[]):
1411
result = InMemoryGraphIndex(ref_lists, key_elements=key_elements)
1412
result.add_nodes(nodes)
1415
def test_add_nodes_no_refs(self):
1416
index = self.make_index(0)
1417
index.add_nodes([(('name', ), 'data')])
1418
index.add_nodes([(('name2', ), ''), (('name3', ), '')])
1419
self.assertEqual(set([
1420
(index, ('name', ), 'data'),
1421
(index, ('name2', ), ''),
1422
(index, ('name3', ), ''),
1423
]), set(index.iter_all_entries()))
1425
def test_add_nodes(self):
1426
index = self.make_index(1)
1427
index.add_nodes([(('name', ), 'data', ([],))])
1428
index.add_nodes([(('name2', ), '', ([],)), (('name3', ), '', ([('r', )],))])
1429
self.assertEqual(set([
1430
(index, ('name', ), 'data', ((),)),
1431
(index, ('name2', ), '', ((),)),
1432
(index, ('name3', ), '', ((('r', ), ), )),
1433
]), set(index.iter_all_entries()))
1435
def test_iter_all_entries_empty(self):
1436
index = self.make_index()
1437
self.assertEqual([], list(index.iter_all_entries()))
1439
def test_iter_all_entries_simple(self):
1440
index = self.make_index(nodes=[(('name', ), 'data')])
1441
self.assertEqual([(index, ('name', ), 'data')],
1442
list(index.iter_all_entries()))
1444
def test_iter_all_entries_references(self):
1445
index = self.make_index(1, nodes=[
1446
(('name', ), 'data', ([('ref', )], )),
1447
(('ref', ), 'refdata', ([], ))])
1448
self.assertEqual(set([(index, ('name', ), 'data', ((('ref', ),),)),
1449
(index, ('ref', ), 'refdata', ((), ))]),
1450
set(index.iter_all_entries()))
1452
def test_iteration_absent_skipped(self):
1453
index = self.make_index(1, nodes=[
1454
(('name', ), 'data', ([('ref', )], ))])
1455
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),))]),
1456
set(index.iter_all_entries()))
1457
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),))]),
1458
set(index.iter_entries([('name', )])))
1459
self.assertEqual([], list(index.iter_entries([('ref', )])))
1461
def test_iter_all_keys(self):
1462
index = self.make_index(1, nodes=[
1463
(('name', ), 'data', ([('ref', )], )),
1464
(('ref', ), 'refdata', ([], ))])
1465
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
1466
(index, ('ref', ), 'refdata', ((), ))]),
1467
set(index.iter_entries([('name', ), ('ref', )])))
1469
def test_iter_key_prefix_1_key_element_no_refs(self):
1470
index = self.make_index( nodes=[
1471
(('name', ), 'data'),
1472
(('ref', ), 'refdata')])
1473
self.assertEqual(set([(index, ('name', ), 'data'),
1474
(index, ('ref', ), 'refdata')]),
1475
set(index.iter_entries_prefix([('name', ), ('ref', )])))
1477
def test_iter_key_prefix_1_key_element_refs(self):
1478
index = self.make_index(1, nodes=[
1479
(('name', ), 'data', ([('ref', )], )),
1480
(('ref', ), 'refdata', ([], ))])
1481
self.assertEqual(set([(index, ('name', ), 'data', ((('ref',),),)),
1482
(index, ('ref', ), 'refdata', ((), ))]),
1483
set(index.iter_entries_prefix([('name', ), ('ref', )])))
1485
def test_iter_key_prefix_2_key_element_no_refs(self):
1486
index = self.make_index(key_elements=2, nodes=[
1487
(('name', 'fin1'), 'data'),
1488
(('name', 'fin2'), 'beta'),
1489
(('ref', 'erence'), 'refdata')])
1490
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
1491
(index, ('ref', 'erence'), 'refdata')]),
1492
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
1493
self.assertEqual(set([(index, ('name', 'fin1'), 'data'),
1494
(index, ('name', 'fin2'), 'beta')]),
1495
set(index.iter_entries_prefix([('name', None)])))
1497
def test_iter_key_prefix_2_key_element_refs(self):
1498
index = self.make_index(1, key_elements=2, nodes=[
1499
(('name', 'fin1'), 'data', ([('ref', 'erence')], )),
1500
(('name', 'fin2'), 'beta', ([], )),
1501
(('ref', 'erence'), 'refdata', ([], ))])
1502
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
1503
(index, ('ref', 'erence'), 'refdata', ((), ))]),
1504
set(index.iter_entries_prefix([('name', 'fin1'), ('ref', 'erence')])))
1505
self.assertEqual(set([(index, ('name', 'fin1'), 'data', ((('ref', 'erence'),),)),
1506
(index, ('name', 'fin2'), 'beta', ((), ))]),
1507
set(index.iter_entries_prefix([('name', None)])))
1509
def test_iter_nothing_empty(self):
1510
index = self.make_index()
1511
self.assertEqual([], list(index.iter_entries([])))
1513
def test_iter_missing_entry_empty(self):
1514
index = self.make_index()
1515
self.assertEqual([], list(index.iter_entries(['a'])))
1517
def test_key_count_empty(self):
1518
index = self.make_index()
1519
self.assertEqual(0, index.key_count())
1521
def test_key_count_one(self):
1522
index = self.make_index(nodes=[(('name', ), '')])
1523
self.assertEqual(1, index.key_count())
1525
def test_key_count_two(self):
1526
index = self.make_index(nodes=[(('name', ), ''), (('foo', ), '')])
1527
self.assertEqual(2, index.key_count())
1529
def test_validate_empty(self):
1530
index = self.make_index()
1533
def test_validate_no_refs_content(self):
1534
index = self.make_index(nodes=[(('key', ), 'value')])
1538
class TestGraphIndexPrefixAdapter(TestCaseWithMemoryTransport):
1540
def make_index(self, ref_lists=1, key_elements=2, nodes=[], add_callback=False):
1541
result = InMemoryGraphIndex(ref_lists, key_elements=key_elements)
1542
result.add_nodes(nodes)
1544
add_nodes_callback = result.add_nodes
1546
add_nodes_callback = None
1547
adapter = GraphIndexPrefixAdapter(result, ('prefix', ), key_elements - 1,
1548
add_nodes_callback=add_nodes_callback)
1549
return result, adapter
1551
def test_add_node(self):
1552
index, adapter = self.make_index(add_callback=True)
1553
adapter.add_node(('key',), 'value', ((('ref',),),))
1554
self.assertEqual(set([(index, ('prefix', 'key'), 'value', ((('prefix', 'ref'),),))]),
1555
set(index.iter_all_entries()))
1557
def test_add_nodes(self):
1558
index, adapter = self.make_index(add_callback=True)
1560
(('key',), 'value', ((('ref',),),)),
1561
(('key2',), 'value2', ((),)),
1563
self.assertEqual(set([
1564
(index, ('prefix', 'key2'), 'value2', ((),)),
1565
(index, ('prefix', 'key'), 'value', ((('prefix', 'ref'),),))
1567
set(index.iter_all_entries()))
1569
def test_construct(self):
1570
index = InMemoryGraphIndex()
1571
adapter = GraphIndexPrefixAdapter(index, ('prefix', ), 1)
1573
def test_construct_with_callback(self):
1574
index = InMemoryGraphIndex()
1575
adapter = GraphIndexPrefixAdapter(index, ('prefix', ), 1, index.add_nodes)
1577
def test_iter_all_entries_cross_prefix_map_errors(self):
1578
index, adapter = self.make_index(nodes=[
1579
(('prefix', 'key1'), 'data1', ((('prefixaltered', 'key2'),),))])
1580
self.assertRaises(errors.BadIndexData, list, adapter.iter_all_entries())
1582
def test_iter_all_entries(self):
1583
index, adapter = self.make_index(nodes=[
1584
(('notprefix', 'key1'), 'data', ((), )),
1585
(('prefix', 'key1'), 'data1', ((), )),
1586
(('prefix', 'key2'), 'data2', ((('prefix', 'key1'),),))])
1587
self.assertEqual(set([(index, ('key1', ), 'data1', ((),)),
1588
(index, ('key2', ), 'data2', ((('key1',),),))]),
1589
set(adapter.iter_all_entries()))
1591
def test_iter_entries(self):
1592
index, adapter = self.make_index(nodes=[
1593
(('notprefix', 'key1'), 'data', ((), )),
1594
(('prefix', 'key1'), 'data1', ((), )),
1595
(('prefix', 'key2'), 'data2', ((('prefix', 'key1'),),))])
1596
# ask for many - get all
1597
self.assertEqual(set([(index, ('key1', ), 'data1', ((),)),
1598
(index, ('key2', ), 'data2', ((('key1', ),),))]),
1599
set(adapter.iter_entries([('key1', ), ('key2', )])))
1600
# ask for one, get one
1601
self.assertEqual(set([(index, ('key1', ), 'data1', ((),))]),
1602
set(adapter.iter_entries([('key1', )])))
1603
# ask for missing, get none
1604
self.assertEqual(set(),
1605
set(adapter.iter_entries([('key3', )])))
1607
def test_iter_entries_prefix(self):
1608
index, adapter = self.make_index(key_elements=3, nodes=[
1609
(('notprefix', 'foo', 'key1'), 'data', ((), )),
1610
(('prefix', 'prefix2', 'key1'), 'data1', ((), )),
1611
(('prefix', 'prefix2', 'key2'), 'data2', ((('prefix', 'prefix2', 'key1'),),))])
1612
# ask for a prefix, get the results for just that prefix, adjusted.
1613
self.assertEqual(set([(index, ('prefix2', 'key1', ), 'data1', ((),)),
1614
(index, ('prefix2', 'key2', ), 'data2', ((('prefix2', 'key1', ),),))]),
1615
set(adapter.iter_entries_prefix([('prefix2', None)])))
1617
def test_key_count_no_matching_keys(self):
1618
index, adapter = self.make_index(nodes=[
1619
(('notprefix', 'key1'), 'data', ((), ))])
1620
self.assertEqual(0, adapter.key_count())
1622
def test_key_count_some_keys(self):
1623
index, adapter = self.make_index(nodes=[
1624
(('notprefix', 'key1'), 'data', ((), )),
1625
(('prefix', 'key1'), 'data1', ((), )),
1626
(('prefix', 'key2'), 'data2', ((('prefix', 'key1'),),))])
1627
self.assertEqual(2, adapter.key_count())
1629
def test_validate(self):
1630
index, adapter = self.make_index()
1633
calls.append('called')
1634
index.validate = validate
1636
self.assertEqual(['called'], calls)
added
removed
bzrlib/tests/test_index.py
