1
# Copyright (C) 2006, 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 2", NL;
33
full checksum = "adler32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", digit, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but its a file. The fingerprint is a
87
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
91
--- Format 1 had the following different definition: ---
92
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
93
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
95
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
96
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
99
PARENT ROW's are emitted for every parent that is not in the ghosts details
100
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
101
each row will have a PARENT ROW for foo and baz, but not for bar.
104
In any tree, a kind of 'moved' indicates that the fingerprint field
105
(which we treat as opaque data specific to the 'kind' anyway) has the
106
details for the id of this row in that tree.
108
I'm strongly tempted to add a id->path index as well, but I think that
109
where we need id->path mapping; we also usually read the whole file, so
110
I'm going to skip that for the moment, as we have the ability to locate
111
via bisect any path in any tree, and if we lookup things by path, we can
112
accumulate a id->path mapping as we go, which will tend to match what we
115
I plan to implement this asap, so please speak up now to alter/tweak the
116
design - and once we stabilise on this, I'll update the wiki page for
119
The rationale for all this is that we want fast operations for the
120
common case (diff/status/commit/merge on all files) and extremely fast
121
operations for the less common but still occurs a lot status/diff/commit
122
on specific files). Operations on specific files involve a scan for all
123
the children of a path, *in every involved tree*, which the current
124
format did not accommodate.
128
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
129
2) fall back current object model as needed.
130
3) scale usably to the largest trees known today - say 50K entries. (mozilla
131
is an example of this)
135
Eventually reuse dirstate objects across locks IFF the dirstate file has not
136
been modified, but will require that we flush/ignore cached stat-hit data
137
because we wont want to restat all files on disk just because a lock was
138
acquired, yet we cannot trust the data after the previous lock was released.
140
Memory representation:
141
vector of all directories, and vector of the childen ?
143
root_entrie = (direntry for root, [parent_direntries_for_root]),
145
('', ['data for achild', 'data for bchild', 'data for cchild'])
146
('dir', ['achild', 'cchild', 'echild'])
148
- single bisect to find N subtrees from a path spec
149
- in-order for serialisation - this is 'dirblock' grouping.
150
- insertion of a file '/a' affects only the '/' child-vector, that is, to
151
insert 10K elements from scratch does not generates O(N^2) memoves of a
152
single vector, rather each individual, which tends to be limited to a
153
manageable number. Will scale badly on trees with 10K entries in a
154
single directory. compare with Inventory.InventoryDirectory which has
155
a dictionary for the children. No bisect capability, can only probe for
156
exact matches, or grab all elements and sorta.
157
- Whats the risk of error here? Once we have the base format being processed
158
we should have a net win regardless of optimality. So we are going to
159
go with what seems reasonably.
162
maybe we should do a test profile of these core structure - 10K simulated searches/lookups/etc?
164
Objects for each row?
165
The lifetime of Dirstate objects is current per lock, but see above for
166
possible extensions. The lifetime of a row from a dirstate is expected to be
167
very short in the optimistic case: which we are optimising for. For instance,
168
subtree status will determine from analysis of the disk data what rows need to
169
be examined at all, and will be able to determine from a single row whether
170
that file has altered or not, so we are aiming to process tens of thousands of
171
entries each second within the dirstate context, before exposing anything to
172
the larger codebase. This suggests we want the time for a single file
173
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
174
processed, and to scale to 100 thousand we'll another order of magnitude to do
175
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
176
the file on disk, and then immediately discard, the overhead of object creation
177
becomes a significant cost.
179
Figures: Creating a tuple from from 3 elements was profiled at 0.0625
180
microseconds, whereas creating a object which is subclassed from tuple was
181
0.500 microseconds, and creating an object with 3 elements and slots was 3
182
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
183
down to 10 microseconds for the total processing - having 33% of that be object
184
creation is a huge overhead. There is a potential cost in using tuples within
185
each row which is that the conditional code to do comparisons may be slower
186
than method invocation, but method invocation is known to be slow due to stack
187
frame creation, so avoiding methods in these tight inner loops in unfortunately
188
desirable. We can consider a pyrex version of this with objects in future if
210
from bzrlib.osutils import (
218
class _Bisector(object):
219
"""This just keeps track of information as we are bisecting."""
222
class DirState(object):
223
"""Record directory and metadata state for fast access.
225
A dirstate is a specialised data structure for managing local working
226
tree state information. Its not yet well defined whether it is platform
227
specific, and if it is how we detect/parameterise that.
230
_kind_to_minikind = {'absent':'a', 'file':'f', 'directory':'d', 'relocated':'r', 'symlink':'l'}
231
_minikind_to_kind = {'a':'absent', 'f':'file', 'd':'directory', 'l':'symlink', 'r':'relocated'}
232
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
233
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
236
# TODO: jam 20070221 Make sure we handle when there are duplicated records
237
# (like when we remove + add the same path, or we have a rename)
238
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
239
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
240
# that we are sure a single record will always fit.
241
BISECT_PAGE_SIZE = 4096
244
IN_MEMORY_UNMODIFIED = 1
245
IN_MEMORY_MODIFIED = 2
247
# A pack_stat (the x's) that is just noise and will never match the output
250
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
252
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
253
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
255
def __init__(self, path):
256
"""Create a DirState object.
260
:attr _root_entrie: The root row of the directory/file information,
261
- contains the path to / - '', ''
262
- kind of 'directory',
263
- the file id of the root in utf8
266
- and no sha information.
267
:param path: The path at which the dirstate file on disk should live.
269
# _header_state and _dirblock_state represent the current state
270
# of the dirstate metadata and the per-row data respectiely.
271
# NOT_IN_MEMORY indicates that no data is in memory
272
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
273
# is the same as is on disk
274
# IN_MEMORY_MODIFIED indicates that we have a modified version
275
# of what is on disk.
276
# In future we will add more granularity, for instance _dirblock_state
277
# will probably support partially-in-memory as a separate variable,
278
# allowing for partially-in-memory unmodified and partially-in-memory
280
self._header_state = DirState.NOT_IN_MEMORY
281
self._dirblock_state = DirState.NOT_IN_MEMORY
285
self._state_file = None
286
self._filename = path
287
self._lock_token = None
288
self._id_index = None
289
self._end_of_header = None
290
self._split_path_cache = {}
291
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
293
def add(self, path, file_id, kind, stat, link_or_sha1):
294
"""Add a path to be tracked.
296
:param path: The path within the dirstate - '' is the root, 'foo' is the
297
path foo within the root, 'foo/bar' is the path bar within foo
299
:param file_id: The file id of the path being added.
300
:param kind: The kind of the path.
301
:param stat: The output of os.lstat for the path.
302
:param link_or_sha1: The sha value of the file, or the target of a
303
symlink. '' for directories.
306
# find the block its in.
307
# find the location in the block.
308
# check its not there
310
#------- copied from inventory.make_entry
311
# --- normalized_filename wants a unicode basename only, so get one.
312
dirname, basename = osutils.split(path)
313
# we dont import normalized_filename directly because we want to be
314
# able to change the implementation at runtime for tests.
315
norm_name, can_access = osutils.normalized_filename(basename)
316
if norm_name != basename:
320
raise errors.InvalidNormalization(path)
321
# now that we've normalised, we need the correct utf8 path and
322
# dirname and basename elements. This single encode and split should be
323
# faster than three separate encodes.
324
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
325
dirname, basename = osutils.split(utf8path)
326
assert file_id.__class__ == str, \
327
"must be a utf8 file_id not %s" % (type(file_id))
328
# Make sure the file_id does not exist in this tree
329
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
330
if file_id_entry != (None, None):
331
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
332
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
333
info = '%s:%s' % (kind, path)
334
raise errors.DuplicateFileId(file_id, info)
335
first_key = (dirname, basename, '')
336
block_index, present = self._find_block_index_from_key(first_key)
338
# check the path is not in the tree
339
block = self._dirblocks[block_index][1]
340
entry_index, _ = self._find_entry_index(first_key, block)
341
while (entry_index < len(block) and
342
block[entry_index][0][0:2] == first_key[0:2]):
343
if block[entry_index][1][0][0] not in 'ar':
344
# this path is in the dirstate in the current tree.
345
raise Exception, "adding already added path!"
348
# The block where we want to put the file is not present. But it
349
# might be because the directory was empty, or not loaded yet. Look
350
# for a parent entry, if not found, raise NotVersionedError
351
parent_dir, parent_base = osutils.split(dirname)
352
parent_block_idx, parent_entry_idx, _, parent_present = \
353
self._get_block_entry_index(parent_dir, parent_base, 0)
354
if not parent_present:
355
raise errors.NotVersionedError(path, str(self))
356
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
357
block = self._dirblocks[block_index][1]
358
entry_key = (dirname, basename, file_id)
361
packed_stat = DirState.NULLSTAT
364
packed_stat = pack_stat(stat)
365
parent_info = self._empty_parent_info()
366
minikind = DirState._kind_to_minikind[kind]
368
entry_data = entry_key, [
369
(minikind, link_or_sha1, size, False, packed_stat),
371
elif kind == 'directory':
372
entry_data = entry_key, [
373
(minikind, '', 0, False, packed_stat),
375
elif kind == 'symlink':
376
entry_data = entry_key, [
377
(minikind, link_or_sha1, size, False, packed_stat),
380
raise errors.BzrError('unknown kind %r' % kind)
381
entry_index, present = self._find_entry_index(entry_key, block)
382
assert not present, "basename %r already added" % basename
383
block.insert(entry_index, entry_data)
385
if kind == 'directory':
386
# insert a new dirblock
387
self._ensure_block(block_index, entry_index, utf8path)
388
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
390
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
392
def _bisect(self, dir_name_list):
393
"""Bisect through the disk structure for specific rows.
395
:param dir_name_list: A list of (dir, name) pairs.
396
:return: A dict mapping (dir, name) => entry for found entries. Missing
397
entries will not be in the map.
399
self._requires_lock()
400
# We need the file pointer to be right after the initial header block
401
self._read_header_if_needed()
402
# If _dirblock_state was in memory, we should just return info from
403
# there, this function is only meant to handle when we want to read
405
assert self._dirblock_state == DirState.NOT_IN_MEMORY
407
# The disk representation is generally info + '\0\n\0' at the end. But
408
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
409
# Because it means we can sync on the '\n'
410
state_file = self._state_file
411
file_size = os.fstat(state_file.fileno()).st_size
412
# We end up with 2 extra fields, we should have a trailing '\n' to
413
# ensure that we read the whole record, and we should have a precursur
414
# '' which ensures that we start after the previous '\n'
415
entry_field_count = self._fields_per_entry() + 1
417
low = self._end_of_header
418
high = file_size - 1 # Ignore the final '\0'
419
# Map from (dir, name) => entry
422
# Avoid infinite seeking
423
max_count = 30*len(dir_name_list)
425
# pending is a list of places to look.
426
# each entry is a tuple of low, high, dir_names
427
# low -> the first byte offset to read (inclusive)
428
# high -> the last byte offset (inclusive)
429
# dir_names -> The list of (dir, name) pairs that should be found in
430
# the [low, high] range
431
pending = [(low, high, dir_name_list)]
433
page_size = self._bisect_page_size
435
fields_to_entry = self._get_fields_to_entry()
438
low, high, cur_files = pending.pop()
440
if not cur_files or low >= high:
445
if count > max_count:
446
raise errors.BzrError('Too many seeks, most likely a bug.')
448
mid = max(low, (low+high-page_size)/2)
451
# limit the read size, so we don't end up reading data that we have
453
read_size = min(page_size, (high-mid)+1)
454
block = state_file.read(read_size)
457
entries = block.split('\n')
460
# We didn't find a '\n', so we cannot have found any records.
461
# So put this range back and try again. But we know we have to
462
# increase the page size, because a single read did not contain
463
# a record break (so records must be larger than page_size)
465
pending.append((low, high, cur_files))
468
# Check the first and last entries, in case they are partial, or if
469
# we don't care about the rest of this page
471
first_fields = entries[0].split('\0')
472
if len(first_fields) < entry_field_count:
473
# We didn't get the complete first entry
474
# so move start, and grab the next, which
475
# should be a full entry
476
start += len(entries[0])+1
477
first_fields = entries[1].split('\0')
480
if len(first_fields) <= 2:
481
# We didn't even get a filename here... what do we do?
482
# Try a large page size and repeat this query
484
pending.append((low, high, cur_files))
487
# Find what entries we are looking for, which occur before and
488
# after this first record.
490
first_dir_name = (first_fields[1], first_fields[2])
491
first_loc = bisect.bisect_left(cur_files, first_dir_name)
493
# These exist before the current location
494
pre = cur_files[:first_loc]
495
# These occur after the current location, which may be in the
496
# data we read, or might be after the last entry
497
post = cur_files[first_loc:]
499
if post and len(first_fields) >= entry_field_count:
500
# We have files after the first entry
502
# Parse the last entry
503
last_entry_num = len(entries)-1
504
last_fields = entries[last_entry_num].split('\0')
505
if len(last_fields) < entry_field_count:
506
# The very last hunk was not complete,
507
# read the previous hunk
508
after = mid + len(block) - len(entries[-1])
510
last_fields = entries[last_entry_num].split('\0')
512
after = mid + len(block)
514
last_dir_name = (last_fields[1], last_fields[2])
515
last_loc = bisect.bisect_right(post, last_dir_name)
517
middle_files = post[:last_loc]
518
post = post[last_loc:]
521
# We have files that should occur in this block
522
# (>= first, <= last)
523
# Either we will find them here, or we can mark them as
526
if middle_files[0] == first_dir_name:
527
# We might need to go before this location
528
pre.append(first_dir_name)
529
if middle_files[-1] == last_dir_name:
530
post.insert(0, last_dir_name)
532
# Find out what paths we have
533
paths = {first_dir_name:[first_fields]}
534
# last_dir_name might == first_dir_name so we need to be
535
# careful if we should append rather than overwrite
536
if last_entry_num != first_entry_num:
537
paths.setdefault(last_dir_name, []).append(last_fields)
538
for num in xrange(first_entry_num+1, last_entry_num):
539
# TODO: jam 20070223 We are already splitting here, so
540
# shouldn't we just split the whole thing rather
541
# than doing the split again in add_one_record?
542
fields = entries[num].split('\0')
543
dir_name = (fields[1], fields[2])
544
paths.setdefault(dir_name, []).append(fields)
546
for dir_name in middle_files:
547
for fields in paths.get(dir_name, []):
548
# offset by 1 because of the opening '\0'
549
# consider changing fields_to_entry to avoid the
551
entry = fields_to_entry(fields[1:])
552
found.setdefault(dir_name, []).append(entry)
554
# Now we have split up everything into pre, middle, and post, and
555
# we have handled everything that fell in 'middle'.
556
# We add 'post' first, so that we prefer to seek towards the
557
# beginning, so that we will tend to go as early as we need, and
558
# then only seek forward after that.
560
pending.append((after, high, post))
562
pending.append((low, start-1, pre))
564
# Consider that we may want to return the directory entries in sorted
565
# order. For now, we just return them in whatever order we found them,
566
# and leave it up to the caller if they care if it is ordered or not.
569
def _bisect_dirblocks(self, dir_list):
570
"""Bisect through the disk structure to find entries in given dirs.
572
_bisect_dirblocks is meant to find the contents of directories, which
573
differs from _bisect, which only finds individual entries.
575
:param dir_list: An sorted list of directory names ['', 'dir', 'foo'].
576
:return: A map from dir => entries_for_dir
578
# TODO: jam 20070223 A lot of the bisecting logic could be shared
579
# between this and _bisect. It would require parameterizing the
580
# inner loop with a function, though. We should evaluate the
581
# performance difference.
582
self._requires_lock()
583
# We need the file pointer to be right after the initial header block
584
self._read_header_if_needed()
585
# If _dirblock_state was in memory, we should just return info from
586
# there, this function is only meant to handle when we want to read
588
assert self._dirblock_state == DirState.NOT_IN_MEMORY
590
# The disk representation is generally info + '\0\n\0' at the end. But
591
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
592
# Because it means we can sync on the '\n'
593
state_file = self._state_file
594
file_size = os.fstat(state_file.fileno()).st_size
595
# We end up with 2 extra fields, we should have a trailing '\n' to
596
# ensure that we read the whole record, and we should have a precursur
597
# '' which ensures that we start after the previous '\n'
598
entry_field_count = self._fields_per_entry() + 1
600
low = self._end_of_header
601
high = file_size - 1 # Ignore the final '\0'
602
# Map from dir => entry
605
# Avoid infinite seeking
606
max_count = 30*len(dir_list)
608
# pending is a list of places to look.
609
# each entry is a tuple of low, high, dir_names
610
# low -> the first byte offset to read (inclusive)
611
# high -> the last byte offset (inclusive)
612
# dirs -> The list of directories that should be found in
613
# the [low, high] range
614
pending = [(low, high, dir_list)]
616
page_size = self._bisect_page_size
618
fields_to_entry = self._get_fields_to_entry()
621
low, high, cur_dirs = pending.pop()
623
if not cur_dirs or low >= high:
628
if count > max_count:
629
raise errors.BzrError('Too many seeks, most likely a bug.')
631
mid = max(low, (low+high-page_size)/2)
634
# limit the read size, so we don't end up reading data that we have
636
read_size = min(page_size, (high-mid)+1)
637
block = state_file.read(read_size)
640
entries = block.split('\n')
643
# We didn't find a '\n', so we cannot have found any records.
644
# So put this range back and try again. But we know we have to
645
# increase the page size, because a single read did not contain
646
# a record break (so records must be larger than page_size)
648
pending.append((low, high, cur_dirs))
651
# Check the first and last entries, in case they are partial, or if
652
# we don't care about the rest of this page
654
first_fields = entries[0].split('\0')
655
if len(first_fields) < entry_field_count:
656
# We didn't get the complete first entry
657
# so move start, and grab the next, which
658
# should be a full entry
659
start += len(entries[0])+1
660
first_fields = entries[1].split('\0')
663
if len(first_fields) <= 1:
664
# We didn't even get a dirname here... what do we do?
665
# Try a large page size and repeat this query
667
pending.append((low, high, cur_dirs))
670
# Find what entries we are looking for, which occur before and
671
# after this first record.
673
first_dir = first_fields[1]
674
first_loc = bisect.bisect_left(cur_dirs, first_dir)
676
# These exist before the current location
677
pre = cur_dirs[:first_loc]
678
# These occur after the current location, which may be in the
679
# data we read, or might be after the last entry
680
post = cur_dirs[first_loc:]
682
if post and len(first_fields) >= entry_field_count:
683
# We have records to look at after the first entry
685
# Parse the last entry
686
last_entry_num = len(entries)-1
687
last_fields = entries[last_entry_num].split('\0')
688
if len(last_fields) < entry_field_count:
689
# The very last hunk was not complete,
690
# read the previous hunk
691
after = mid + len(block) - len(entries[-1])
693
last_fields = entries[last_entry_num].split('\0')
695
after = mid + len(block)
697
last_dir = last_fields[1]
698
last_loc = bisect.bisect_right(post, last_dir)
700
middle_files = post[:last_loc]
701
post = post[last_loc:]
704
# We have files that should occur in this block
705
# (>= first, <= last)
706
# Either we will find them here, or we can mark them as
709
if middle_files[0] == first_dir:
710
# We might need to go before this location
711
pre.append(first_dir)
712
if middle_files[-1] == last_dir:
713
post.insert(0, last_dir)
715
# Find out what paths we have
716
paths = {first_dir:[first_fields]}
717
# last_dir might == first_dir so we need to be
718
# careful if we should append rather than overwrite
719
if last_entry_num != first_entry_num:
720
paths.setdefault(last_dir, []).append(last_fields)
721
for num in xrange(first_entry_num+1, last_entry_num):
722
# TODO: jam 20070223 We are already splitting here, so
723
# shouldn't we just split the whole thing rather
724
# than doing the split again in add_one_record?
725
fields = entries[num].split('\0')
726
paths.setdefault(fields[1], []).append(fields)
728
for cur_dir in middle_files:
729
for fields in paths.get(cur_dir, []):
730
# offset by 1 because of the opening '\0'
731
# consider changing fields_to_entry to avoid the
733
entry = fields_to_entry(fields[1:])
734
found.setdefault(cur_dir, []).append(entry)
736
# Now we have split up everything into pre, middle, and post, and
737
# we have handled everything that fell in 'middle'.
738
# We add 'post' first, so that we prefer to seek towards the
739
# beginning, so that we will tend to go as early as we need, and
740
# then only seek forward after that.
742
pending.append((after, high, post))
744
pending.append((low, start-1, pre))
748
def _bisect_recursive(self, dir_name_list):
749
"""Bisect for entries for all paths and their children.
751
This will use bisect to find all records for the supplied paths. It
752
will then continue to bisect for any records which are marked as
753
directories. (and renames?)
755
:param paths: A sorted list of (dir, name) pairs
756
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
757
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
759
# Map from (dir, name, file_id) => [tree_info]
762
found_dir_names = set()
764
# Directories that have been read
765
processed_dirs = set()
766
# Get the ball rolling with the first bisect for all entries.
767
newly_found = self._bisect(dir_name_list)
770
# Directories that need to be read
772
paths_to_search = set()
773
for entry_list in newly_found.itervalues():
774
for dir_name_id, trees_info in entry_list:
775
found[dir_name_id] = trees_info
776
found_dir_names.add(dir_name_id[:2])
778
for tree_info in trees_info:
779
minikind = tree_info[0]
782
# We already processed this one as a directory,
783
# we don't need to do the extra work again.
785
subdir, name, file_id = dir_name_id
786
path = osutils.pathjoin(subdir, name)
788
if path not in processed_dirs:
789
pending_dirs.add(path)
790
elif minikind == 'r':
791
# Rename, we need to directly search the target
792
# which is contained in the fingerprint column
793
dir_name = osutils.split(tree_info[1])
794
if dir_name[0] in pending_dirs:
795
# This entry will be found in the dir search
797
# TODO: We need to check if this entry has
798
# already been found. Otherwise we might be
799
# hitting infinite recursion.
800
if dir_name not in found_dir_names:
801
paths_to_search.add(dir_name)
802
# Now we have a list of paths to look for directly, and
803
# directory blocks that need to be read.
804
# newly_found is mixing the keys between (dir, name) and path
805
# entries, but that is okay, because we only really care about the
807
newly_found = self._bisect(sorted(paths_to_search))
808
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
809
processed_dirs.update(pending_dirs)
812
def _empty_parent_info(self):
813
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
816
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
817
"""Ensure a block for dirname exists.
819
This function exists to let callers which know that there is a
820
directory dirname ensure that the block for it exists. This block can
821
fail to exist because of demand loading, or because a directory had no
822
children. In either case it is not an error. It is however an error to
823
call this if there is no parent entry for the directory, and thus the
824
function requires the coordinates of such an entry to be provided.
826
The root row is special cased and can be indicated with a parent block
829
:param parent_block_index: The index of the block in which dirname's row
831
:param parent_row_index: The index in the parent block where the row
833
:param dirname: The utf8 dirname to ensure there is a block for.
834
:return: The index for the block.
836
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
837
# This is the signature of the root row, and the
838
# contents-of-root row is always index 1
840
# the basename of the directory must be the end of its full name.
841
if not (parent_block_index == -1 and
842
parent_block_index == -1 and dirname == ''):
843
assert dirname.endswith(
844
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
845
block_index, present = self._find_block_index_from_key((dirname, '', ''))
847
## In future, when doing partial parsing, this should load and
848
# populate the entire block.
849
self._dirblocks.insert(block_index, (dirname, []))
852
def _entries_to_current_state(self, new_entries):
853
"""Load new_entries into self.dirblocks.
855
Process new_entries into the current state object, making them the active
858
:param new_entries: A sorted list of entries. This function does not sort
859
to prevent unneeded overhead when callers have a sorted list already.
862
assert new_entries[0][0][0:2] == ('', ''), \
863
"Missing root row %r" % (new_entries[0][0],)
864
# The two blocks here are deliberate: the root block and the
865
# contents-of-root block.
866
self._dirblocks = [('', []), ('', [])]
867
current_block = self._dirblocks[0][1]
870
append_entry = current_block.append
871
for entry in new_entries:
872
if entry[0][0] != current_dirname:
873
# new block - different dirname
875
current_dirname = entry[0][0]
876
self._dirblocks.append((current_dirname, current_block))
877
append_entry = current_block.append
878
# append the entry to the current block
880
self._split_root_dirblock_into_contents()
882
def _split_root_dirblock_into_contents(self):
883
"""Split the root dirblocks into root and contents-of-root.
885
After parsing by path, we end up with root entries and contents-of-root
886
entries in the same block. This loop splits them out again.
888
# The above loop leaves the "root block" entries mixed with the
889
# "contents-of-root block". But we don't want an if check on
890
# all entries, so instead we just fix it up here.
891
assert self._dirblocks[1] == ('', [])
893
contents_of_root_block = []
894
for entry in self._dirblocks[0][1]:
895
if not entry[0][1]: # This is a root entry
896
root_block.append(entry)
898
contents_of_root_block.append(entry)
899
self._dirblocks[0] = ('', root_block)
900
self._dirblocks[1] = ('', contents_of_root_block)
902
def _entry_to_line(self, entry):
903
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
905
:param entry: An entry_tuple as defined in the module docstring.
907
entire_entry = list(entry[0])
908
for tree_number, tree_data in enumerate(entry[1]):
909
# (minikind, fingerprint, size, executable, tree_specific_string)
910
entire_entry.extend(tree_data)
911
# 3 for the key, 5 for the fields per tree.
912
tree_offset = 3 + tree_number * 5
914
entire_entry[tree_offset + 0] = tree_data[0]
916
entire_entry[tree_offset + 2] = str(tree_data[2])
918
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
919
return '\0'.join(entire_entry)
921
def _fields_per_entry(self):
922
"""How many null separated fields should be in each entry row.
924
Each line now has an extra '\n' field which is not used
925
so we just skip over it
928
+ number of fields per tree_data (5) * tree count
931
tree_count = 1 + self._num_present_parents()
932
return 3 + 5 * tree_count + 1
934
def _find_block(self, key, add_if_missing=False):
935
"""Return the block that key should be present in.
937
:param key: A dirstate entry key.
938
:return: The block tuple.
940
block_index, present = self._find_block_index_from_key(key)
942
if not add_if_missing:
943
# check to see if key is versioned itself - we might want to
944
# add it anyway, because dirs with no entries dont get a
945
# dirblock at parse time.
946
# This is an uncommon branch to take: most dirs have children,
947
# and most code works with versioned paths.
948
parent_base, parent_name = osutils.split(key[0])
949
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
950
# some parent path has not been added - its an error to add
952
raise errors.NotVersionedError(key[0:2], str(self))
953
self._dirblocks.insert(block_index, (key[0], []))
954
return self._dirblocks[block_index]
956
def _find_block_index_from_key(self, key):
957
"""Find the dirblock index for a key.
959
:return: The block index, True if the block for the key is present.
961
if key[0:2] == ('', ''):
963
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
964
cache=self._split_path_cache)
965
# _right returns one-past-where-key is so we have to subtract
966
# one to use it. we use _right here because there are two
967
# '' blocks - the root, and the contents of root
968
# we always have a minimum of 2 in self._dirblocks: root and
969
# root-contents, and for '', we get 2 back, so this is
970
# simple and correct:
971
present = (block_index < len(self._dirblocks) and
972
self._dirblocks[block_index][0] == key[0])
973
return block_index, present
975
def _find_entry_index(self, key, block):
976
"""Find the entry index for a key in a block.
978
:return: The entry index, True if the entry for the key is present.
980
entry_index = bisect.bisect_left(block, (key, []))
981
present = (entry_index < len(block) and
982
block[entry_index][0] == key)
983
return entry_index, present
986
def from_tree(tree, dir_state_filename):
987
"""Create a dirstate from a bzr Tree.
989
:param tree: The tree which should provide parent information and
991
:return: a DirState object which is currently locked for writing.
992
(it was locked by DirState.initialize)
994
result = DirState.initialize(dir_state_filename)
998
parent_ids = tree.get_parent_ids()
999
num_parents = len(parent_ids)
1001
for parent_id in parent_ids:
1002
parent_tree = tree.branch.repository.revision_tree(parent_id)
1003
parent_trees.append((parent_id, parent_tree))
1004
parent_tree.lock_read()
1005
result.set_parent_trees(parent_trees, [])
1006
result.set_state_from_inventory(tree.inventory)
1008
for revid, parent_tree in parent_trees:
1009
parent_tree.unlock()
1012
# The caller won't have a chance to unlock this, so make sure we
1018
def get_sha1_for_entry(self, entry, abspath, stat_value=None):
1019
"""Return the sha1 sum for a given file.
1021
:param entry: This is the dirblock entry for the file in question.
1022
:param abspath: The path on disk for this file.
1023
:param stat_value: (optional) if we already have done a stat on the
1025
:return: The sha1 hexdigest of the file (40 bytes)
1027
# This code assumes that the entry passed in is directly held in one of
1028
# the internal _dirblocks. So the dirblock state must have already been
1030
assert self._dirblock_state != DirState.NOT_IN_MEMORY
1031
assert entry[1][0][0] == 'f', \
1032
'can only get sha1 for a file not: %s %s' % (
1033
DirState._minikind_to_kind[entry[1][0][0]], entry[0])
1034
if stat_value is None:
1035
stat_value = os.lstat(abspath)
1036
packed_stat = pack_stat(stat_value)
1037
saved_sha1_digest = entry[1][0][1]
1038
saved_file_size = entry[1][0][2]
1039
saved_packed_stat = entry[1][0][4]
1040
if (packed_stat == saved_packed_stat
1041
and saved_sha1_digest != ''
1042
# size should also be in packed_stat
1043
and saved_file_size == stat_value.st_size):
1044
# The stat hasn't changed since we saved, so we can potentially
1045
# re-use the saved sha hash.
1046
cutoff = self._sha_cutoff_time()
1047
if (stat_value.st_mtime < cutoff
1048
and stat_value.st_ctime < cutoff):
1049
# Return the existing fingerprint
1050
return saved_sha1_digest
1051
# If we have gotten this far, that means that we need to actually
1053
new_sha1_digest = self._sha1_file(abspath)
1054
# TODO: jam 20070301 Is it worth checking to see if the new sha is
1055
# actually different? I'm guessing not, because we wouldn't have
1056
# gotten this far otherwise.
1057
entry[1][0] = ('f', new_sha1_digest, stat_value.st_size,
1058
entry[1][0][3], # Executable?
1061
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1062
return new_sha1_digest
1064
def _sha_cutoff_time(self):
1065
"""Return cutoff time.
1067
Files modified more recently than this time are at risk of being
1068
undetectably modified and so can't be cached.
1070
return int(time.time()) - 3
1072
def _sha1_file(self, abspath):
1073
"""Calculate the SHA1 of a file by reading the full text"""
1074
f = file(abspath, 'rb', buffering=65000)
1076
return osutils.sha_file(f)
1080
def get_ghosts(self):
1081
"""Return a list of the parent tree revision ids that are ghosts."""
1082
self._read_header_if_needed()
1085
def get_lines(self):
1086
"""Serialise the entire dirstate to a sequence of lines."""
1087
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1088
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1089
# read whats on disk.
1090
self._state_file.seek(0)
1091
return self._state_file.readlines()
1093
lines.append(self._get_parents_line(self.get_parent_ids()))
1094
lines.append(self._get_ghosts_line(self._ghosts))
1095
# append the root line which is special cased
1096
lines.extend(map(self._entry_to_line, self._iter_entries()))
1097
return self._get_output_lines(lines)
1099
def _get_ghosts_line(self, ghost_ids):
1100
"""Create a line for the state file for ghost information."""
1101
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1103
def _get_parents_line(self, parent_ids):
1104
"""Create a line for the state file for parents information."""
1105
return '\0'.join([str(len(parent_ids))] + parent_ids)
1107
def _get_fields_to_entry(self):
1108
"""Get a function which converts entry fields into a entry record.
1110
This handles size and executable, as well as parent records.
1112
:return: A function which takes a list of fields, and returns an
1113
appropriate record for storing in memory.
1115
# This is intentionally unrolled for performance
1116
num_present_parents = self._num_present_parents()
1117
if num_present_parents == 0:
1118
def fields_to_entry_0_parents(fields, _int=int):
1119
path_name_file_id_key = (fields[0], fields[1], fields[2])
1120
return (path_name_file_id_key, [
1122
fields[3], # minikind
1123
fields[4], # fingerprint
1124
_int(fields[5]), # size
1125
fields[6] == 'y', # executable
1126
fields[7], # packed_stat or revision_id
1128
return fields_to_entry_0_parents
1129
elif num_present_parents == 1:
1130
def fields_to_entry_1_parent(fields, _int=int):
1131
path_name_file_id_key = (fields[0], fields[1], fields[2])
1132
return (path_name_file_id_key, [
1134
fields[3], # minikind
1135
fields[4], # fingerprint
1136
_int(fields[5]), # size
1137
fields[6] == 'y', # executable
1138
fields[7], # packed_stat or revision_id
1141
fields[8], # minikind
1142
fields[9], # fingerprint
1143
_int(fields[10]), # size
1144
fields[11] == 'y', # executable
1145
fields[12], # packed_stat or revision_id
1148
return fields_to_entry_1_parent
1149
elif num_present_parents == 2:
1150
def fields_to_entry_2_parents(fields, _int=int):
1151
path_name_file_id_key = (fields[0], fields[1], fields[2])
1152
return (path_name_file_id_key, [
1154
fields[3], # minikind
1155
fields[4], # fingerprint
1156
_int(fields[5]), # size
1157
fields[6] == 'y', # executable
1158
fields[7], # packed_stat or revision_id
1161
fields[8], # minikind
1162
fields[9], # fingerprint
1163
_int(fields[10]), # size
1164
fields[11] == 'y', # executable
1165
fields[12], # packed_stat or revision_id
1168
fields[13], # minikind
1169
fields[14], # fingerprint
1170
_int(fields[15]), # size
1171
fields[16] == 'y', # executable
1172
fields[17], # packed_stat or revision_id
1175
return fields_to_entry_2_parents
1177
def fields_to_entry_n_parents(fields, _int=int):
1178
path_name_file_id_key = (fields[0], fields[1], fields[2])
1179
trees = [(fields[cur], # minikind
1180
fields[cur+1], # fingerprint
1181
_int(fields[cur+2]), # size
1182
fields[cur+3] == 'y', # executable
1183
fields[cur+4], # stat or revision_id
1184
) for cur in xrange(3, len(fields)-1, 5)]
1185
return path_name_file_id_key, trees
1186
return fields_to_entry_n_parents
1188
def get_parent_ids(self):
1189
"""Return a list of the parent tree ids for the directory state."""
1190
self._read_header_if_needed()
1191
return list(self._parents)
1193
def _get_block_entry_index(self, dirname, basename, tree_index):
1194
"""Get the coordinates for a path in the state structure.
1196
:param dirname: The utf8 dirname to lookup.
1197
:param basename: The utf8 basename to lookup.
1198
:param tree_index: The index of the tree for which this lookup should
1200
:return: A tuple describing where the path is located, or should be
1201
inserted. The tuple contains four fields: the block index, the row
1202
index, anda two booleans are True when the directory is present, and
1203
when the entire path is present. There is no guarantee that either
1204
coordinate is currently reachable unless the found field for it is
1205
True. For instance, a directory not present in the searched tree
1206
may be returned with a value one greater than the current highest
1207
block offset. The directory present field will always be True when
1208
the path present field is True. The directory present field does
1209
NOT indicate that the directory is present in the searched tree,
1210
rather it indicates that there are at least some files in some
1213
self._read_dirblocks_if_needed()
1214
key = dirname, basename, ''
1215
block_index, present = self._find_block_index_from_key(key)
1217
# no such directory - return the dir index and 0 for the row.
1218
return block_index, 0, False, False
1219
block = self._dirblocks[block_index][1] # access the entries only
1220
entry_index, present = self._find_entry_index(key, block)
1221
# linear search through present entries at this path to find the one
1223
while entry_index < len(block) and block[entry_index][0][1] == basename:
1224
if block[entry_index][1][tree_index][0] not in \
1225
('a', 'r'): # absent, relocated
1226
return block_index, entry_index, True, True
1228
return block_index, entry_index, True, False
1230
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1231
"""Get the dirstate entry for path in tree tree_index
1233
If either file_id or path is supplied, it is used as the key to lookup.
1234
If both are supplied, the fastest lookup is used, and an error is
1235
raised if they do not both point at the same row.
1237
:param tree_index: The index of the tree we wish to locate this path
1238
in. If the path is present in that tree, the entry containing its
1239
details is returned, otherwise (None, None) is returned
1240
:param fileid_utf8: A utf8 file_id to look up.
1241
:param path_utf8: An utf8 path to be looked up.
1242
:return: The dirstate entry tuple for path, or (None, None)
1244
self._read_dirblocks_if_needed()
1245
if path_utf8 is not None:
1246
assert path_utf8.__class__ == str, 'path_utf8 is not a str: %s %s' % (type(path_utf8), path_utf8)
1247
# path lookups are faster
1248
dirname, basename = osutils.split(path_utf8)
1249
block_index, entry_index, dir_present, file_present = \
1250
self._get_block_entry_index(dirname, basename, tree_index)
1251
if not file_present:
1253
entry = self._dirblocks[block_index][1][entry_index]
1254
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1256
if entry[0][2] != fileid_utf8:
1257
raise errors.BzrError('integrity error ? : mismatching'
1258
' tree_index, file_id and path')
1261
assert fileid_utf8 is not None
1262
possible_keys = self._get_id_index().get(fileid_utf8, None)
1263
if not possible_keys:
1265
for key in possible_keys:
1266
block_index, present = \
1267
self._find_block_index_from_key(key)
1268
# strange, probably indicates an out of date
1269
# id index - for now, allow this.
1272
# WARNING: DO not change this code to use _get_block_entry_index
1273
# as that function is not suitable: it does not use the key
1274
# to lookup, and thus the wront coordinates are returned.
1275
block = self._dirblocks[block_index][1]
1276
entry_index, present = self._find_entry_index(key, block)
1278
entry = self._dirblocks[block_index][1][entry_index]
1279
if entry[1][tree_index][0] in 'fdl':
1280
# this is the result we are looking for: the
1281
# real home of this file_id in this tree.
1283
if entry[1][tree_index][0] == 'a':
1284
# there is no home for this entry in this tree
1286
assert entry[1][tree_index][0] == 'r'
1287
real_path = entry[1][tree_index][1]
1288
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1289
path_utf8=real_path)
1293
def initialize(cls, path):
1294
"""Create a new dirstate on path.
1296
The new dirstate will be an empty tree - that is it has no parents,
1297
and only a root node - which has id ROOT_ID.
1299
The object will be write locked when returned to the caller,
1300
unless there was an exception in the writing, in which case it
1303
:param path: The name of the file for the dirstate.
1304
:return: A DirState object.
1306
# This constructs a new DirState object on a path, sets the _state_file
1307
# to a new empty file for that path. It then calls _set_data() with our
1308
# stock empty dirstate information - a root with ROOT_ID, no children,
1309
# and no parents. Finally it calls save() to ensure that this data will
1312
# root dir and root dir contents with no children.
1313
empty_tree_dirblocks = [('', []), ('', [])]
1314
# a new root directory, with a NULLSTAT.
1315
empty_tree_dirblocks[0][1].append(
1316
(('', '', inventory.ROOT_ID), [
1317
('d', '', 0, False, DirState.NULLSTAT),
1321
result._set_data([], empty_tree_dirblocks)
1328
def _inv_entry_to_details(self, inv_entry):
1329
"""Convert an inventory entry (from a revision tree) to state details.
1331
:param inv_entry: An inventory entry whose sha1 and link targets can be
1332
relied upon, and which has a revision set.
1333
:return: A details tuple - the details for a single tree at a path +
1336
kind = inv_entry.kind
1337
minikind = DirState._kind_to_minikind[kind]
1338
tree_data = inv_entry.revision
1339
assert len(tree_data) > 0, 'empty revision for the inv_entry.'
1340
if kind == 'directory':
1344
elif kind == 'symlink':
1345
fingerprint = inv_entry.symlink_target or ''
1348
elif kind == 'file':
1349
fingerprint = inv_entry.text_sha1 or ''
1350
size = inv_entry.text_size or 0
1351
executable = inv_entry.executable
1354
return (minikind, fingerprint, size, executable, tree_data)
1356
def _iter_entries(self):
1357
"""Iterate over all the entries in the dirstate.
1359
Each yelt item is an entry in the standard format described in the
1360
docstring of bzrlib.dirstate.
1362
self._read_dirblocks_if_needed()
1363
for directory in self._dirblocks:
1364
for entry in directory[1]:
1367
def _get_id_index(self):
1368
"""Get an id index of self._dirblocks."""
1369
if self._id_index is None:
1371
for key, tree_details in self._iter_entries():
1372
id_index.setdefault(key[2], set()).add(key)
1373
self._id_index = id_index
1374
return self._id_index
1376
def _get_output_lines(self, lines):
1377
"""format lines for final output.
1379
:param lines: A sequece of lines containing the parents list and the
1382
output_lines = [DirState.HEADER_FORMAT_3]
1383
lines.append('') # a final newline
1384
inventory_text = '\0\n\0'.join(lines)
1385
output_lines.append('adler32: %s\n' % (zlib.adler32(inventory_text),))
1386
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1387
num_entries = len(lines)-3
1388
output_lines.append('num_entries: %s\n' % (num_entries,))
1389
output_lines.append(inventory_text)
1392
def _make_deleted_row(self, fileid_utf8, parents):
1393
"""Return a deleted for for fileid_utf8."""
1394
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1397
def _num_present_parents(self):
1398
"""The number of parent entries in each record row."""
1399
return len(self._parents) - len(self._ghosts)
1403
"""Construct a DirState on the file at path path.
1405
:return: An unlocked DirState object, associated with the given path.
1407
result = DirState(path)
1410
def _read_dirblocks_if_needed(self):
1411
"""Read in all the dirblocks from the file if they are not in memory.
1413
This populates self._dirblocks, and sets self._dirblock_state to
1414
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1417
self._read_header_if_needed()
1418
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1419
# move the _state_file pointer to after the header (in case bisect
1420
# has been called in the mean time)
1421
self._state_file.seek(self._end_of_header)
1422
text = self._state_file.read()
1423
# TODO: check the adler checksums. adler_measured = zlib.adler32(text)
1425
fields = text.split('\0')
1426
# Remove the last blank entry
1427
trailing = fields.pop()
1428
assert trailing == ''
1429
# consider turning fields into a tuple.
1431
# skip the first field which is the trailing null from the header.
1433
# Each line now has an extra '\n' field which is not used
1434
# so we just skip over it
1436
# 3 fields for the key
1437
# + number of fields per tree_data (5) * tree count
1439
num_present_parents = self._num_present_parents()
1440
tree_count = 1 + num_present_parents
1441
entry_size = self._fields_per_entry()
1442
expected_field_count = entry_size * self._num_entries
1443
if len(fields) - cur > expected_field_count:
1444
fields = fields[:expected_field_count + cur]
1445
trace.mutter('Unexpectedly long dirstate field count!')
1446
print "XXX: incorrectly truncated dirstate file bug triggered."
1447
field_count = len(fields)
1448
# this checks our adjustment, and also catches file too short.
1449
assert field_count - cur == expected_field_count, \
1450
'field count incorrect %s != %s, entry_size=%s, '\
1451
'num_entries=%s fields=%r' % (
1452
field_count - cur, expected_field_count, entry_size,
1453
self._num_entries, fields)
1455
if num_present_parents == 1:
1456
# Bind external functions to local names
1458
# We access all fields in order, so we can just iterate over
1459
# them. Grab an straight iterator over the fields. (We use an
1460
# iterator because we don't want to do a lot of additions, nor
1461
# do we want to do a lot of slicing)
1462
next = iter(fields).next
1463
# Move the iterator to the current position
1464
for x in xrange(cur):
1466
# The two blocks here are deliberate: the root block and the
1467
# contents-of-root block.
1468
self._dirblocks = [('', []), ('', [])]
1469
current_block = self._dirblocks[0][1]
1470
current_dirname = ''
1471
append_entry = current_block.append
1472
for count in xrange(self._num_entries):
1476
if dirname != current_dirname:
1477
# new block - different dirname
1479
current_dirname = dirname
1480
self._dirblocks.append((current_dirname, current_block))
1481
append_entry = current_block.append
1482
# we know current_dirname == dirname, so re-use it to avoid
1483
# creating new strings
1484
entry = ((current_dirname, name, file_id),
1487
next(), # fingerprint
1488
_int(next()), # size
1489
next() == 'y', # executable
1490
next(), # packed_stat or revision_id
1494
next(), # fingerprint
1495
_int(next()), # size
1496
next() == 'y', # executable
1497
next(), # packed_stat or revision_id
1501
assert trailing == '\n'
1502
# append the entry to the current block
1504
self._split_root_dirblock_into_contents()
1506
fields_to_entry = self._get_fields_to_entry()
1507
entries = [fields_to_entry(fields[pos:pos+entry_size])
1508
for pos in xrange(cur, field_count, entry_size)]
1509
self._entries_to_current_state(entries)
1510
# To convert from format 2 => format 3
1511
# self._dirblocks = sorted(self._dirblocks,
1512
# key=lambda blk:blk[0].split('/'))
1513
# To convert from format 3 => format 2
1514
# self._dirblocks = sorted(self._dirblocks)
1515
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1517
def _read_header(self):
1518
"""This reads in the metadata header, and the parent ids.
1520
After reading in, the file should be positioned at the null
1521
just before the start of the first record in the file.
1523
:return: (expected adler checksum, number of entries, parent list)
1525
self._read_prelude()
1526
parent_line = self._state_file.readline()
1527
info = parent_line.split('\0')
1528
num_parents = int(info[0])
1529
assert num_parents == len(info)-2, 'incorrect parent info line'
1530
self._parents = info[1:-1]
1532
ghost_line = self._state_file.readline()
1533
info = ghost_line.split('\0')
1534
num_ghosts = int(info[1])
1535
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1536
self._ghosts = info[2:-1]
1537
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1538
self._end_of_header = self._state_file.tell()
1540
def _read_header_if_needed(self):
1541
"""Read the header of the dirstate file if needed."""
1542
# inline this as it will be called a lot
1543
if not self._lock_token:
1544
raise errors.ObjectNotLocked(self)
1545
if self._header_state == DirState.NOT_IN_MEMORY:
1548
def _read_prelude(self):
1549
"""Read in the prelude header of the dirstate file
1551
This only reads in the stuff that is not connected to the adler
1552
checksum. The position will be correct to read in the rest of
1553
the file and check the checksum after this point.
1554
The next entry in the file should be the number of parents,
1555
and their ids. Followed by a newline.
1557
header = self._state_file.readline()
1558
assert header == DirState.HEADER_FORMAT_3, \
1559
'invalid header line: %r' % (header,)
1560
adler_line = self._state_file.readline()
1561
assert adler_line.startswith('adler32: '), 'missing adler32 checksum'
1562
self.adler_expected = int(adler_line[len('adler32: '):-1])
1563
num_entries_line = self._state_file.readline()
1564
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1565
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1568
"""Save any pending changes created during this session.
1570
We reuse the existing file, because that prevents race conditions with
1571
file creation, and we expect to be using oslocks on it in the near
1572
future to prevent concurrent modification and reads - because dirstates
1573
incremental data aggretation is not compatible with reading a modified
1574
file, and replacing a file in use by another process is impossible on
1577
A dirstate in read only mode should be smart enough though to validate
1578
that the file has not changed, and otherwise discard its cache and
1579
start over, to allow for fine grained read lock duration, so 'status'
1580
wont block 'commit' - for example.
1582
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1583
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1584
self._state_file.seek(0)
1585
self._state_file.writelines(self.get_lines())
1586
self._state_file.truncate()
1587
self._state_file.flush()
1588
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1589
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1591
def _set_data(self, parent_ids, dirblocks):
1592
"""Set the full dirstate data in memory.
1594
This is an internal function used to completely replace the objects
1595
in memory state. It puts the dirstate into state 'full-dirty'.
1597
:param parent_ids: A list of parent tree revision ids.
1598
:param dirblocks: A list containing one tuple for each directory in the
1599
tree. Each tuple contains the directory path and a list of entries
1600
found in that directory.
1602
# our memory copy is now authoritative.
1603
self._dirblocks = dirblocks
1604
self._header_state = DirState.IN_MEMORY_MODIFIED
1605
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1606
self._parents = list(parent_ids)
1607
self._id_index = None
1609
def set_path_id(self, path, new_id):
1610
"""Change the id of path to new_id in the current working tree.
1612
:param path: The path inside the tree to set - '' is the root, 'foo'
1613
is the path foo in the root.
1614
:param new_id: The new id to assign to the path. This must be a utf8
1615
file id (not unicode, and not None).
1617
# TODO: start warning here.
1618
assert new_id.__class__ == str
1619
self._read_dirblocks_if_needed()
1621
import pdb;pdb.set_trace()
1623
raise NotImplementedError(self.set_path_id)
1624
# TODO: check new id is unique
1625
entry = self._get_entry(0, path_utf8=path)
1626
if entry[0][2] == new_id:
1627
# Nothing to change.
1629
# mark the old path absent, and insert a new root path
1630
self._make_absent(entry)
1631
self.update_minimal(('', '', new_id), 'd',
1632
path_utf8='', packed_stat=entry[1][0][4])
1633
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1634
if self._id_index is not None:
1635
self._id_index.setdefault(new_id, set()).add(entry[0])
1637
def set_parent_trees(self, trees, ghosts):
1638
"""Set the parent trees for the dirstate.
1640
:param trees: A list of revision_id, tree tuples. tree must be provided
1641
even if the revision_id refers to a ghost: supply an empty tree in
1643
:param ghosts: A list of the revision_ids that are ghosts at the time
1646
# TODO: generate a list of parent indexes to preserve to save
1647
# processing specific parent trees. In the common case one tree will
1648
# be preserved - the left most parent.
1649
# TODO: if the parent tree is a dirstate, we might want to walk them
1650
# all by path in parallel for 'optimal' common-case performance.
1651
# generate new root row.
1652
self._read_dirblocks_if_needed()
1653
# TODO future sketch: Examine the existing parents to generate a change
1654
# map and then walk the new parent trees only, mapping them into the
1655
# dirstate. Walk the dirstate at the same time to remove unreferenced
1658
# sketch: loop over all entries in the dirstate, cherry picking
1659
# entries from the parent trees, if they are not ghost trees.
1660
# after we finish walking the dirstate, all entries not in the dirstate
1661
# are deletes, so we want to append them to the end as per the design
1662
# discussions. So do a set difference on ids with the parents to
1663
# get deletes, and add them to the end.
1664
# During the update process we need to answer the following questions:
1665
# - find other keys containing a fileid in order to create cross-path
1666
# links. We dont't trivially use the inventory from other trees
1667
# because this leads to either double touching, or to accessing
1669
# - find other keys containing a path
1670
# We accumulate each entry via this dictionary, including the root
1673
# we could do parallel iterators, but because file id data may be
1674
# scattered throughout, we dont save on index overhead: we have to look
1675
# at everything anyway. We can probably save cycles by reusing parent
1676
# data and doing an incremental update when adding an additional
1677
# parent, but for now the common cases are adding a new parent (merge),
1678
# and replacing completely (commit), and commit is more common: so
1679
# optimise merge later.
1681
# ---- start generation of full tree mapping data
1682
# what trees should we use?
1683
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
1684
# how many trees do we end up with
1685
parent_count = len(parent_trees)
1687
# one: the current tree
1688
for entry in self._iter_entries():
1689
# skip entries not in the current tree
1690
if entry[1][0][0] in ('a', 'r'): # absent, relocated
1692
by_path[entry[0]] = [entry[1][0]] + \
1693
[DirState.NULL_PARENT_DETAILS] * parent_count
1694
id_index[entry[0][2]] = set([entry[0]])
1696
# now the parent trees:
1697
for tree_index, tree in enumerate(parent_trees):
1698
# the index is off by one, adjust it.
1699
tree_index = tree_index + 1
1700
# when we add new locations for a fileid we need these ranges for
1701
# any fileid in this tree as we set the by_path[id] to:
1702
# already_processed_tree_details + new_details + new_location_suffix
1703
# the suffix is from tree_index+1:parent_count+1.
1704
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
1705
# now stitch in all the entries from this tree
1706
for path, entry in tree.inventory.iter_entries_by_dir():
1707
# here we process each trees details for each item in the tree.
1708
# we first update any existing entries for the id at other paths,
1709
# then we either create or update the entry for the id at the
1710
# right path, and finally we add (if needed) a mapping from
1711
# file_id to this path. We do it in this order to allow us to
1712
# avoid checking all known paths for the id when generating a
1713
# new entry at this path: by adding the id->path mapping last,
1714
# all the mappings are valid and have correct relocation
1715
# records where needed.
1716
file_id = entry.file_id
1717
path_utf8 = path.encode('utf8')
1718
dirname, basename = osutils.split(path_utf8)
1719
new_entry_key = (dirname, basename, file_id)
1720
# tree index consistency: All other paths for this id in this tree
1721
# index must point to the correct path.
1722
for entry_key in id_index.setdefault(file_id, set()):
1723
# TODO:PROFILING: It might be faster to just update
1724
# rather than checking if we need to, and then overwrite
1725
# the one we are located at.
1726
if entry_key != new_entry_key:
1727
# this file id is at a different path in one of the
1728
# other trees, so put absent pointers there
1729
# This is the vertical axis in the matrix, all pointing
1731
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
1732
# by path consistency: Insert into an existing path record (trivial), or
1733
# add a new one with relocation pointers for the other tree indexes.
1734
if new_entry_key in id_index[file_id]:
1735
# there is already an entry where this data belongs, just insert it.
1736
by_path[new_entry_key][tree_index] = \
1737
self._inv_entry_to_details(entry)
1739
# add relocated entries to the horizontal axis - this row
1740
# mapping from path,id. We need to look up the correct path
1741
# for the indexes from 0 to tree_index -1
1743
for lookup_index in xrange(tree_index):
1744
# boundary case: this is the first occurence of file_id
1745
# so there are no id_indexs, possibly take this out of
1747
if not len(id_index[file_id]):
1748
new_details.append(DirState.NULL_PARENT_DETAILS)
1750
# grab any one entry, use it to find the right path.
1751
# TODO: optimise this to reduce memory use in highly
1752
# fragmented situations by reusing the relocation
1754
a_key = iter(id_index[file_id]).next()
1755
if by_path[a_key][lookup_index][0] in ('r', 'a'):
1756
# its a pointer or missing statement, use it as is.
1757
new_details.append(by_path[a_key][lookup_index])
1759
# we have the right key, make a pointer to it.
1760
real_path = ('/'.join(a_key[0:2])).strip('/')
1761
new_details.append(('r', real_path, 0, False, ''))
1762
new_details.append(self._inv_entry_to_details(entry))
1763
new_details.extend(new_location_suffix)
1764
by_path[new_entry_key] = new_details
1765
id_index[file_id].add(new_entry_key)
1766
# --- end generation of full tree mappings
1768
# sort and output all the entries
1769
new_entries = sorted(by_path.items(),
1770
key=lambda entry:(entry[0][0].split('/'), entry[0][1]))
1771
self._entries_to_current_state(new_entries)
1772
self._parents = [rev_id for rev_id, tree in trees]
1773
self._ghosts = list(ghosts)
1774
self._header_state = DirState.IN_MEMORY_MODIFIED
1775
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1776
self._id_index = id_index
1778
def set_state_from_inventory(self, new_inv):
1779
"""Set new_inv as the current state.
1781
This API is called by tree transform, and will usually occur with
1782
existing parent trees.
1784
:param new_inv: The inventory object to set current state from.
1786
self._read_dirblocks_if_needed()
1788
# incremental algorithm:
1789
# two iterators: current data and new data, both in dirblock order.
1790
new_iterator = new_inv.iter_entries_by_dir()
1791
# we will be modifying the dirstate, so we need a stable iterator. In
1792
# future we might write one, for now we just clone the state into a
1793
# list - which is a shallow copy, so each
1794
old_iterator = iter(list(self._iter_entries()))
1795
# both must have roots so this is safe:
1796
current_new = new_iterator.next()
1797
current_old = old_iterator.next()
1798
def advance(iterator):
1800
return iterator.next()
1801
except StopIteration:
1803
while current_new or current_old:
1804
# skip entries in old that are not really there
1805
if current_old and current_old[1][0][0] in ('r', 'a'):
1806
# relocated or absent
1807
current_old = advance(old_iterator)
1810
# convert new into dirblock style
1811
new_path_utf8 = current_new[0].encode('utf8')
1812
new_dirname, new_basename = osutils.split(new_path_utf8)
1813
new_id = current_new[1].file_id
1814
new_entry_key = (new_dirname, new_basename, new_id)
1815
current_new_minikind = \
1816
DirState._kind_to_minikind[current_new[1].kind]
1818
# for safety disable variables
1819
new_path_utf8 = new_dirname = new_basename = new_id = new_entry_key = None
1820
# 5 cases, we dont have a value that is strictly greater than everything, so
1821
# we make both end conditions explicit
1823
# old is finished: insert current_new into the state.
1824
self.update_minimal(new_entry_key, current_new_minikind,
1825
executable=current_new[1].executable,
1826
path_utf8=new_path_utf8)
1827
current_new = advance(new_iterator)
1828
elif not current_new:
1830
self._make_absent(current_old)
1831
current_old = advance(old_iterator)
1832
elif new_entry_key == current_old[0]:
1833
# same - common case
1834
# TODO: update the record if anything significant has changed.
1835
# the minimal required trigger is if the execute bit or cached
1837
if (current_old[1][0][3] != current_new[1].executable or
1838
current_old[1][0][0] != current_new_minikind):
1839
self.update_minimal(current_old[0], current_new_minikind,
1840
executable=current_new[1].executable,
1841
path_utf8=new_path_utf8)
1842
# both sides are dealt with, move on
1843
current_old = advance(old_iterator)
1844
current_new = advance(new_iterator)
1845
elif new_entry_key < current_old[0]:
1847
# add a entry for this and advance new
1848
self.update_minimal(new_entry_key, current_new_minikind,
1849
executable=current_new[1].executable,
1850
path_utf8=new_path_utf8)
1851
current_new = advance(new_iterator)
1854
self._make_absent(current_old)
1855
current_old = advance(old_iterator)
1856
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1857
self._id_index = None
1859
def _make_absent(self, current_old):
1860
"""Mark current_old - an entry - as absent for tree 0.
1862
:return: True if this was the last details entry for they entry key:
1863
that is, if the underlying block has had the entry removed, thus
1864
shrinking in length.
1866
# build up paths that this id will be left at after the change is made,
1867
# so we can update their cross references in tree 0
1868
all_remaining_keys = set()
1869
# Dont check the working tree, because its going.
1870
for details in current_old[1][1:]:
1871
if details[0] not in ('a', 'r'): # absent, relocated
1872
all_remaining_keys.add(current_old[0])
1873
elif details[0] == 'r': # relocated
1874
# record the key for the real path.
1875
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
1876
# absent rows are not present at any path.
1877
last_reference = current_old[0] not in all_remaining_keys
1879
# the current row consists entire of the current item (being marked
1880
# absent), and relocated or absent entries for the other trees:
1881
# Remove it, its meaningless.
1882
block = self._find_block(current_old[0])
1883
entry_index, present = self._find_entry_index(current_old[0], block[1])
1884
assert present, 'could not find entry for %s' % (current_old,)
1885
block[1].pop(entry_index)
1886
# if we have an id_index in use, remove this key from it for this id.
1887
if self._id_index is not None:
1888
self._id_index[current_old[0][2]].remove(current_old[0])
1889
# update all remaining keys for this id to record it as absent. The
1890
# existing details may either be the record we are making as deleted
1891
# (if there were other trees with the id present at this path), or may
1893
for update_key in all_remaining_keys:
1894
update_block_index, present = \
1895
self._find_block_index_from_key(update_key)
1896
assert present, 'could not find block for %s' % (update_key,)
1897
update_entry_index, present = \
1898
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
1899
assert present, 'could not find entry for %s' % (update_key,)
1900
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
1901
# it must not be absent at the moment
1902
assert update_tree_details[0][0] != 'a' # absent
1903
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
1904
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1905
return last_reference
1907
def update_minimal(self, key, minikind, executable=False, fingerprint='',
1908
packed_stat=None, size=0, path_utf8=None):
1909
"""Update an entry to the state in tree 0.
1911
This will either create a new entry at 'key' or update an existing one.
1912
It also makes sure that any other records which might mention this are
1915
:param key: (dir, name, file_id) for the new entry
1916
:param minikind: The type for the entry ('f' == 'file', 'd' ==
1918
:param executable: Should the executable bit be set?
1919
:param fingerprint: Simple fingerprint for new entry.
1920
:param packed_stat: packed stat value for new entry.
1921
:param size: Size information for new entry
1922
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
1925
block = self._find_block(key)[1]
1926
if packed_stat is None:
1927
packed_stat = DirState.NULLSTAT
1928
entry_index, present = self._find_entry_index(key, block)
1929
new_details = (minikind, fingerprint, size, executable, packed_stat)
1930
id_index = self._get_id_index()
1932
# new entry, synthesis cross reference here,
1933
existing_keys = id_index.setdefault(key[2], set())
1934
if not existing_keys:
1935
# not currently in the state, simplest case
1936
new_entry = key, [new_details] + self._empty_parent_info()
1938
# present at one or more existing other paths.
1939
# grab one of them and use it to generate parent
1940
# relocation/absent entries.
1941
new_entry = key, [new_details]
1942
for other_key in existing_keys:
1943
# change the record at other to be a pointer to this new
1944
# record. The loop looks similar to the change to
1945
# relocations when updating an existing record but its not:
1946
# the test for existing kinds is different: this can be
1947
# factored out to a helper though.
1948
other_block_index, present = self._find_block_index_from_key(other_key)
1950
import pdb; pdb.set_trace()
1951
assert present, 'could not find block for %s' % (other_key,)
1952
other_entry_index, present = self._find_entry_index(other_key,
1953
self._dirblocks[other_block_index][1])
1955
import pdb; pdb.set_trace()
1956
assert present, 'could not find entry for %s' % (other_key,)
1957
assert path_utf8 is not None
1958
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
1959
('r', path_utf8, 0, False, '')
1961
num_present_parents = self._num_present_parents()
1962
for lookup_index in xrange(1, num_present_parents + 1):
1963
# grab any one entry, use it to find the right path.
1964
# TODO: optimise this to reduce memory use in highly
1965
# fragmented situations by reusing the relocation
1967
update_block_index, present = \
1968
self._find_block_index_from_key(other_key)
1969
assert present, 'could not find block for %s' % (other_key,)
1970
update_entry_index, present = \
1971
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
1972
assert present, 'could not find entry for %s' % (other_key,)
1973
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
1974
if update_details[0] in ('r', 'a'): # relocated, absent
1975
# its a pointer or absent in lookup_index's tree, use
1977
new_entry[1].append(update_details)
1979
# we have the right key, make a pointer to it.
1980
pointer_path = osutils.pathjoin(*other_key[0:2])
1981
new_entry[1].append(('r', pointer_path, 0, False, ''))
1982
block.insert(entry_index, new_entry)
1983
existing_keys.add(key)
1985
# Does the new state matter?
1986
block[entry_index][1][0] = new_details
1987
# parents cannot be affected by what we do.
1988
# other occurences of this id can be found
1989
# from the id index.
1991
# tree index consistency: All other paths for this id in this tree
1992
# index must point to the correct path. We have to loop here because
1993
# we may have passed entries in the state with this file id already
1994
# that were absent - where parent entries are - and they need to be
1995
# converted to relocated.
1996
assert path_utf8 is not None
1997
for entry_key in id_index.setdefault(key[2], set()):
1998
# TODO:PROFILING: It might be faster to just update
1999
# rather than checking if we need to, and then overwrite
2000
# the one we are located at.
2001
if entry_key != key:
2002
# this file id is at a different path in one of the
2003
# other trees, so put absent pointers there
2004
# This is the vertical axis in the matrix, all pointing
2006
block_index, present = self._find_block_index_from_key(entry_key)
2008
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2010
self._dirblocks[block_index][1][entry_index][1][0] = \
2011
('r', path_utf8, 0, False, '')
2012
# add a containing dirblock if needed.
2013
if new_details[0] == 'd':
2014
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2015
block_index, present = self._find_block_index_from_key(subdir_key)
2017
self._dirblocks.insert(block_index, (subdir_key[0], []))
2019
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2021
def _validate(self):
2022
"""Check that invariants on the dirblock are correct.
2024
This can be useful in debugging; it shouldn't be necessary in
2027
from pprint import pformat
2028
if len(self._dirblocks) > 0:
2029
assert self._dirblocks[0][0] == '', \
2030
"dirblocks don't start with root block:\n" + \
2032
if len(self._dirblocks) > 1:
2033
assert self._dirblocks[1][0] == '', \
2034
"dirblocks missing root directory:\n" + \
2036
assert self._dirblocks[1:] == sorted(self._dirblocks[1:]), \
2037
"dirblocks are not in sorted order:\n" + \
2038
pformat(self._dirblocks)
2039
for dirblock in self._dirblocks:
2040
assert dirblock[1] == sorted(dirblock[1]), \
2041
"dirblock for %r is not sorted:\n%s" % \
2042
(dirblock[0], pformat(dirblock))
2044
def _wipe_state(self):
2045
"""Forget all state information about the dirstate."""
2046
self._header_state = DirState.NOT_IN_MEMORY
2047
self._dirblock_state = DirState.NOT_IN_MEMORY
2050
self._dirblocks = []
2052
def lock_read(self):
2053
"""Acquire a read lock on the dirstate"""
2054
if self._lock_token is not None:
2055
raise errors.LockContention(self._lock_token)
2056
self._lock_token = lock.ReadLock(self._filename)
2057
self._state_file = self._lock_token.f
2060
def lock_write(self):
2061
"""Acquire a write lock on the dirstate"""
2062
if self._lock_token is not None:
2063
raise errors.LockContention(self._lock_token)
2064
self._lock_token = lock.WriteLock(self._filename)
2065
self._state_file = self._lock_token.f
2069
"""Drop any locks held on the dirstate"""
2070
if self._lock_token is None:
2071
raise errors.LockNotHeld(self)
2072
self._state_file = None
2073
self._lock_token.unlock()
2074
self._lock_token = None
2075
self._split_path_cache = {}
2077
def _requires_lock(self):
2078
"""Checks that a lock is currently held by someone on the dirstate"""
2079
if not self._lock_token:
2080
raise errors.ObjectNotLocked(self)
2083
def bisect_dirblock(dirblocks, dirname, lo=0, hi=None, cache={}):
2084
"""Return the index where to insert dirname into the dirblocks.
2086
The return value idx is such that all directories blocks in dirblock[:idx]
2087
have names < dirname, and all blocks in dirblock[idx:] have names >=
2090
Optional args lo (default 0) and hi (default len(dirblocks)) bound the
2091
slice of a to be searched.
2096
dirname_split = cache[dirname]
2098
dirname_split = dirname.split('/')
2099
cache[dirname] = dirname_split
2102
# Grab the dirname for the current dirblock
2103
cur = dirblocks[mid][0]
2105
cur_split = cache[cur]
2107
cur_split = cur.split('/')
2108
cache[cur] = cur_split
2109
if cur_split < dirname_split: lo = mid+1
2114
_base64_encoder = codecs.getencoder('base64')
2117
def pack_stat(st, _encode=_base64_encoder, _pack=struct.pack):
2118
"""Convert stat values into a packed representation."""
2119
# jam 20060614 it isn't really worth removing more entries if we
2120
# are going to leave it in packed form.
2121
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
2122
# With all entries filesize is 5.9M and read time is mabye 280ms
2123
# well within the noise margin
2125
# base64.encode always adds a final newline, so strip it off
2126
return _encode(_pack('>llllll'
2127
, st.st_size, int(st.st_mtime), int(st.st_ctime)
2128
, st.st_dev, st.st_ino, st.st_mode))[0][:-1]
added
removed
bzrlib/dirstate.py
