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" | "t";
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 = "crc32: ", ["-"], 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
89
't' is a reference to a nested subtree; the fingerprint is the referenced
94
The entries on disk and in memory are ordered according to the following keys:
96
directory, as a list of components
100
--- Format 1 had the following different definition: ---
101
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
102
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
104
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
105
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
108
PARENT ROW's are emitted for every parent that is not in the ghosts details
109
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
110
each row will have a PARENT ROW for foo and baz, but not for bar.
113
In any tree, a kind of 'moved' indicates that the fingerprint field
114
(which we treat as opaque data specific to the 'kind' anyway) has the
115
details for the id of this row in that tree.
117
I'm strongly tempted to add a id->path index as well, but I think that
118
where we need id->path mapping; we also usually read the whole file, so
119
I'm going to skip that for the moment, as we have the ability to locate
120
via bisect any path in any tree, and if we lookup things by path, we can
121
accumulate a id->path mapping as we go, which will tend to match what we
124
I plan to implement this asap, so please speak up now to alter/tweak the
125
design - and once we stabilise on this, I'll update the wiki page for
128
The rationale for all this is that we want fast operations for the
129
common case (diff/status/commit/merge on all files) and extremely fast
130
operations for the less common but still occurs a lot status/diff/commit
131
on specific files). Operations on specific files involve a scan for all
132
the children of a path, *in every involved tree*, which the current
133
format did not accommodate.
137
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
138
2) fall back current object model as needed.
139
3) scale usably to the largest trees known today - say 50K entries. (mozilla
140
is an example of this)
144
Eventually reuse dirstate objects across locks IFF the dirstate file has not
145
been modified, but will require that we flush/ignore cached stat-hit data
146
because we wont want to restat all files on disk just because a lock was
147
acquired, yet we cannot trust the data after the previous lock was released.
149
Memory representation:
150
vector of all directories, and vector of the childen ?
152
root_entrie = (direntry for root, [parent_direntries_for_root]),
154
('', ['data for achild', 'data for bchild', 'data for cchild'])
155
('dir', ['achild', 'cchild', 'echild'])
157
- single bisect to find N subtrees from a path spec
158
- in-order for serialisation - this is 'dirblock' grouping.
159
- insertion of a file '/a' affects only the '/' child-vector, that is, to
160
insert 10K elements from scratch does not generates O(N^2) memoves of a
161
single vector, rather each individual, which tends to be limited to a
162
manageable number. Will scale badly on trees with 10K entries in a
163
single directory. compare with Inventory.InventoryDirectory which has
164
a dictionary for the children. No bisect capability, can only probe for
165
exact matches, or grab all elements and sorta.
166
- Whats the risk of error here? Once we have the base format being processed
167
we should have a net win regardless of optimality. So we are going to
168
go with what seems reasonably.
171
maybe we should do a test profile of these core structure - 10K simulated searches/lookups/etc?
173
Objects for each row?
174
The lifetime of Dirstate objects is current per lock, but see above for
175
possible extensions. The lifetime of a row from a dirstate is expected to be
176
very short in the optimistic case: which we are optimising for. For instance,
177
subtree status will determine from analysis of the disk data what rows need to
178
be examined at all, and will be able to determine from a single row whether
179
that file has altered or not, so we are aiming to process tens of thousands of
180
entries each second within the dirstate context, before exposing anything to
181
the larger codebase. This suggests we want the time for a single file
182
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
183
processed, and to scale to 100 thousand we'll another order of magnitude to do
184
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
185
the file on disk, and then immediately discard, the overhead of object creation
186
becomes a significant cost.
188
Figures: Creating a tuple from from 3 elements was profiled at 0.0625
189
microseconds, whereas creating a object which is subclassed from tuple was
190
0.500 microseconds, and creating an object with 3 elements and slots was 3
191
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
192
down to 10 microseconds for the total processing - having 33% of that be object
193
creation is a huge overhead. There is a potential cost in using tuples within
194
each row which is that the conditional code to do comparisons may be slower
195
than method invocation, but method invocation is known to be slow due to stack
196
frame creation, so avoiding methods in these tight inner loops in unfortunately
197
desirable. We can consider a pyrex version of this with objects in future if
206
from stat import S_IEXEC
222
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
223
"""Convert stat values into a packed representation."""
224
# jam 20060614 it isn't really worth removing more entries if we
225
# are going to leave it in packed form.
226
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
227
# With all entries filesize is 5.9M and read time is mabye 280ms
228
# well within the noise margin
230
# base64 encoding always adds a final newline, so strip it off
231
# The current version
232
return _encode(_pack('>LLLLLL'
233
, st.st_size, int(st.st_mtime), int(st.st_ctime)
234
, st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
235
# This is 0.060s / 1.520s faster by not encoding as much information
236
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
237
# This is not strictly faster than _encode(_pack())[:-1]
238
# return '%X.%X.%X.%X.%X.%X' % (
239
# st.st_size, int(st.st_mtime), int(st.st_ctime),
240
# st.st_dev, st.st_ino, st.st_mode)
241
# Similar to the _encode(_pack('>LL'))
242
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
245
class DirState(object):
246
"""Record directory and metadata state for fast access.
248
A dirstate is a specialised data structure for managing local working
249
tree state information. Its not yet well defined whether it is platform
250
specific, and if it is how we detect/parameterise that.
252
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
253
Unlike most bzr disk formats, DirStates must be locked for reading, using
254
lock_read. (This is an os file lock internally.) This is necessary
255
because the file can be rewritten in place.
257
DirStates must be explicitly written with save() to commit changes; just
258
unlocking them does not write the changes to disk.
261
_kind_to_minikind = {
267
'tree-reference': 't',
269
_minikind_to_kind = {
275
't': 'tree-reference',
277
_stat_to_minikind = {
282
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
283
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
286
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
287
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
288
# that we are sure a single record will always fit.
289
BISECT_PAGE_SIZE = 4096
292
IN_MEMORY_UNMODIFIED = 1
293
IN_MEMORY_MODIFIED = 2
295
# A pack_stat (the x's) that is just noise and will never match the output
298
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
300
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
301
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
303
def __init__(self, path):
304
"""Create a DirState object.
308
:attr _root_entrie: The root row of the directory/file information,
309
- contains the path to / - '', ''
310
- kind of 'directory',
311
- the file id of the root in utf8
314
- and no sha information.
315
:param path: The path at which the dirstate file on disk should live.
317
# _header_state and _dirblock_state represent the current state
318
# of the dirstate metadata and the per-row data respectiely.
319
# NOT_IN_MEMORY indicates that no data is in memory
320
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
321
# is the same as is on disk
322
# IN_MEMORY_MODIFIED indicates that we have a modified version
323
# of what is on disk.
324
# In future we will add more granularity, for instance _dirblock_state
325
# will probably support partially-in-memory as a separate variable,
326
# allowing for partially-in-memory unmodified and partially-in-memory
328
self._header_state = DirState.NOT_IN_MEMORY
329
self._dirblock_state = DirState.NOT_IN_MEMORY
333
self._state_file = None
334
self._filename = path
335
self._lock_token = None
336
self._lock_state = None
337
self._id_index = None
338
self._end_of_header = None
339
self._cutoff_time = None
340
self._split_path_cache = {}
341
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
345
(self.__class__.__name__, self._filename)
347
def add(self, path, file_id, kind, stat, fingerprint):
348
"""Add a path to be tracked.
350
:param path: The path within the dirstate - '' is the root, 'foo' is the
351
path foo within the root, 'foo/bar' is the path bar within foo
353
:param file_id: The file id of the path being added.
354
:param kind: The kind of the path, as a string like 'file',
356
:param stat: The output of os.lstat for the path.
357
:param fingerprint: The sha value of the file,
358
or the target of a symlink,
359
or the referenced revision id for tree-references,
360
or '' for directories.
363
# find the block its in.
364
# find the location in the block.
365
# check its not there
367
#------- copied from inventory.make_entry
368
# --- normalized_filename wants a unicode basename only, so get one.
369
dirname, basename = osutils.split(path)
370
# we dont import normalized_filename directly because we want to be
371
# able to change the implementation at runtime for tests.
372
norm_name, can_access = osutils.normalized_filename(basename)
373
if norm_name != basename:
377
raise errors.InvalidNormalization(path)
378
# you should never have files called . or ..; just add the directory
379
# in the parent, or according to the special treatment for the root
380
if basename == '.' or basename == '..':
381
raise errors.InvalidEntryName(path)
382
# now that we've normalised, we need the correct utf8 path and
383
# dirname and basename elements. This single encode and split should be
384
# faster than three separate encodes.
385
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
386
dirname, basename = osutils.split(utf8path)
387
assert file_id.__class__ == str, \
388
"must be a utf8 file_id not %s" % (type(file_id))
389
# Make sure the file_id does not exist in this tree
390
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
391
if file_id_entry != (None, None):
392
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
393
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
394
info = '%s:%s' % (kind, path)
395
raise errors.DuplicateFileId(file_id, info)
396
first_key = (dirname, basename, '')
397
block_index, present = self._find_block_index_from_key(first_key)
399
# check the path is not in the tree
400
block = self._dirblocks[block_index][1]
401
entry_index, _ = self._find_entry_index(first_key, block)
402
while (entry_index < len(block) and
403
block[entry_index][0][0:2] == first_key[0:2]):
404
if block[entry_index][1][0][0] not in 'ar':
405
# this path is in the dirstate in the current tree.
406
raise Exception, "adding already added path!"
409
# The block where we want to put the file is not present. But it
410
# might be because the directory was empty, or not loaded yet. Look
411
# for a parent entry, if not found, raise NotVersionedError
412
parent_dir, parent_base = osutils.split(dirname)
413
parent_block_idx, parent_entry_idx, _, parent_present = \
414
self._get_block_entry_index(parent_dir, parent_base, 0)
415
if not parent_present:
416
raise errors.NotVersionedError(path, str(self))
417
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
418
block = self._dirblocks[block_index][1]
419
entry_key = (dirname, basename, file_id)
422
packed_stat = DirState.NULLSTAT
425
packed_stat = pack_stat(stat)
426
parent_info = self._empty_parent_info()
427
minikind = DirState._kind_to_minikind[kind]
429
entry_data = entry_key, [
430
(minikind, fingerprint, size, False, packed_stat),
432
elif kind == 'directory':
433
entry_data = entry_key, [
434
(minikind, '', 0, False, packed_stat),
436
elif kind == 'symlink':
437
entry_data = entry_key, [
438
(minikind, fingerprint, size, False, packed_stat),
440
elif kind == 'tree-reference':
441
entry_data = entry_key, [
442
(minikind, fingerprint, 0, False, packed_stat),
445
raise errors.BzrError('unknown kind %r' % kind)
446
entry_index, present = self._find_entry_index(entry_key, block)
448
block.insert(entry_index, entry_data)
450
assert block[entry_index][1][0][0] == 'a', " %r(%r) already added" % (basename, file_id)
451
block[entry_index][1][0] = entry_data[1][0]
453
if kind == 'directory':
454
# insert a new dirblock
455
self._ensure_block(block_index, entry_index, utf8path)
456
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
458
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
460
def _bisect(self, paths):
461
"""Bisect through the disk structure for specific rows.
463
:param paths: A list of paths to find
464
:return: A dict mapping path => entries for found entries. Missing
465
entries will not be in the map.
466
The list is not sorted, and entries will be populated
467
based on when they were read.
469
self._requires_lock()
470
# We need the file pointer to be right after the initial header block
471
self._read_header_if_needed()
472
# If _dirblock_state was in memory, we should just return info from
473
# there, this function is only meant to handle when we want to read
475
assert self._dirblock_state == DirState.NOT_IN_MEMORY
477
# The disk representation is generally info + '\0\n\0' at the end. But
478
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
479
# Because it means we can sync on the '\n'
480
state_file = self._state_file
481
file_size = os.fstat(state_file.fileno()).st_size
482
# We end up with 2 extra fields, we should have a trailing '\n' to
483
# ensure that we read the whole record, and we should have a precursur
484
# '' which ensures that we start after the previous '\n'
485
entry_field_count = self._fields_per_entry() + 1
487
low = self._end_of_header
488
high = file_size - 1 # Ignore the final '\0'
489
# Map from (dir, name) => entry
492
# Avoid infinite seeking
493
max_count = 30*len(paths)
495
# pending is a list of places to look.
496
# each entry is a tuple of low, high, dir_names
497
# low -> the first byte offset to read (inclusive)
498
# high -> the last byte offset (inclusive)
499
# dir_names -> The list of (dir, name) pairs that should be found in
500
# the [low, high] range
501
pending = [(low, high, paths)]
503
page_size = self._bisect_page_size
505
fields_to_entry = self._get_fields_to_entry()
508
low, high, cur_files = pending.pop()
510
if not cur_files or low >= high:
515
if count > max_count:
516
raise errors.BzrError('Too many seeks, most likely a bug.')
518
mid = max(low, (low+high-page_size)/2)
521
# limit the read size, so we don't end up reading data that we have
523
read_size = min(page_size, (high-mid)+1)
524
block = state_file.read(read_size)
527
entries = block.split('\n')
530
# We didn't find a '\n', so we cannot have found any records.
531
# So put this range back and try again. But we know we have to
532
# increase the page size, because a single read did not contain
533
# a record break (so records must be larger than page_size)
535
pending.append((low, high, cur_files))
538
# Check the first and last entries, in case they are partial, or if
539
# we don't care about the rest of this page
541
first_fields = entries[0].split('\0')
542
if len(first_fields) < entry_field_count:
543
# We didn't get the complete first entry
544
# so move start, and grab the next, which
545
# should be a full entry
546
start += len(entries[0])+1
547
first_fields = entries[1].split('\0')
550
if len(first_fields) <= 2:
551
# We didn't even get a filename here... what do we do?
552
# Try a large page size and repeat this query
554
pending.append((low, high, cur_files))
557
# Find what entries we are looking for, which occur before and
558
# after this first record.
561
first_path = first_fields[1] + '/' + first_fields[2]
563
first_path = first_fields[2]
564
first_loc = _bisect_path_left(cur_files, first_path)
566
# These exist before the current location
567
pre = cur_files[:first_loc]
568
# These occur after the current location, which may be in the
569
# data we read, or might be after the last entry
570
post = cur_files[first_loc:]
572
if post and len(first_fields) >= entry_field_count:
573
# We have files after the first entry
575
# Parse the last entry
576
last_entry_num = len(entries)-1
577
last_fields = entries[last_entry_num].split('\0')
578
if len(last_fields) < entry_field_count:
579
# The very last hunk was not complete,
580
# read the previous hunk
581
after = mid + len(block) - len(entries[-1])
583
last_fields = entries[last_entry_num].split('\0')
585
after = mid + len(block)
588
last_path = last_fields[1] + '/' + last_fields[2]
590
last_path = last_fields[2]
591
last_loc = _bisect_path_right(post, last_path)
593
middle_files = post[:last_loc]
594
post = post[last_loc:]
597
# We have files that should occur in this block
598
# (>= first, <= last)
599
# Either we will find them here, or we can mark them as
602
if middle_files[0] == first_path:
603
# We might need to go before this location
604
pre.append(first_path)
605
if middle_files[-1] == last_path:
606
post.insert(0, last_path)
608
# Find out what paths we have
609
paths = {first_path:[first_fields]}
610
# last_path might == first_path so we need to be
611
# careful if we should append rather than overwrite
612
if last_entry_num != first_entry_num:
613
paths.setdefault(last_path, []).append(last_fields)
614
for num in xrange(first_entry_num+1, last_entry_num):
615
# TODO: jam 20070223 We are already splitting here, so
616
# shouldn't we just split the whole thing rather
617
# than doing the split again in add_one_record?
618
fields = entries[num].split('\0')
620
path = fields[1] + '/' + fields[2]
623
paths.setdefault(path, []).append(fields)
625
for path in middle_files:
626
for fields in paths.get(path, []):
627
# offset by 1 because of the opening '\0'
628
# consider changing fields_to_entry to avoid the
630
entry = fields_to_entry(fields[1:])
631
found.setdefault(path, []).append(entry)
633
# Now we have split up everything into pre, middle, and post, and
634
# we have handled everything that fell in 'middle'.
635
# We add 'post' first, so that we prefer to seek towards the
636
# beginning, so that we will tend to go as early as we need, and
637
# then only seek forward after that.
639
pending.append((after, high, post))
641
pending.append((low, start-1, pre))
643
# Consider that we may want to return the directory entries in sorted
644
# order. For now, we just return them in whatever order we found them,
645
# and leave it up to the caller if they care if it is ordered or not.
648
def _bisect_dirblocks(self, dir_list):
649
"""Bisect through the disk structure to find entries in given dirs.
651
_bisect_dirblocks is meant to find the contents of directories, which
652
differs from _bisect, which only finds individual entries.
654
:param dir_list: An sorted list of directory names ['', 'dir', 'foo'].
655
:return: A map from dir => entries_for_dir
657
# TODO: jam 20070223 A lot of the bisecting logic could be shared
658
# between this and _bisect. It would require parameterizing the
659
# inner loop with a function, though. We should evaluate the
660
# performance difference.
661
self._requires_lock()
662
# We need the file pointer to be right after the initial header block
663
self._read_header_if_needed()
664
# If _dirblock_state was in memory, we should just return info from
665
# there, this function is only meant to handle when we want to read
667
assert self._dirblock_state == DirState.NOT_IN_MEMORY
669
# The disk representation is generally info + '\0\n\0' at the end. But
670
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
671
# Because it means we can sync on the '\n'
672
state_file = self._state_file
673
file_size = os.fstat(state_file.fileno()).st_size
674
# We end up with 2 extra fields, we should have a trailing '\n' to
675
# ensure that we read the whole record, and we should have a precursur
676
# '' which ensures that we start after the previous '\n'
677
entry_field_count = self._fields_per_entry() + 1
679
low = self._end_of_header
680
high = file_size - 1 # Ignore the final '\0'
681
# Map from dir => entry
684
# Avoid infinite seeking
685
max_count = 30*len(dir_list)
687
# pending is a list of places to look.
688
# each entry is a tuple of low, high, dir_names
689
# low -> the first byte offset to read (inclusive)
690
# high -> the last byte offset (inclusive)
691
# dirs -> The list of directories that should be found in
692
# the [low, high] range
693
pending = [(low, high, dir_list)]
695
page_size = self._bisect_page_size
697
fields_to_entry = self._get_fields_to_entry()
700
low, high, cur_dirs = pending.pop()
702
if not cur_dirs or low >= high:
707
if count > max_count:
708
raise errors.BzrError('Too many seeks, most likely a bug.')
710
mid = max(low, (low+high-page_size)/2)
713
# limit the read size, so we don't end up reading data that we have
715
read_size = min(page_size, (high-mid)+1)
716
block = state_file.read(read_size)
719
entries = block.split('\n')
722
# We didn't find a '\n', so we cannot have found any records.
723
# So put this range back and try again. But we know we have to
724
# increase the page size, because a single read did not contain
725
# a record break (so records must be larger than page_size)
727
pending.append((low, high, cur_dirs))
730
# Check the first and last entries, in case they are partial, or if
731
# we don't care about the rest of this page
733
first_fields = entries[0].split('\0')
734
if len(first_fields) < entry_field_count:
735
# We didn't get the complete first entry
736
# so move start, and grab the next, which
737
# should be a full entry
738
start += len(entries[0])+1
739
first_fields = entries[1].split('\0')
742
if len(first_fields) <= 1:
743
# We didn't even get a dirname here... what do we do?
744
# Try a large page size and repeat this query
746
pending.append((low, high, cur_dirs))
749
# Find what entries we are looking for, which occur before and
750
# after this first record.
752
first_dir = first_fields[1]
753
first_loc = bisect.bisect_left(cur_dirs, first_dir)
755
# These exist before the current location
756
pre = cur_dirs[:first_loc]
757
# These occur after the current location, which may be in the
758
# data we read, or might be after the last entry
759
post = cur_dirs[first_loc:]
761
if post and len(first_fields) >= entry_field_count:
762
# We have records to look at after the first entry
764
# Parse the last entry
765
last_entry_num = len(entries)-1
766
last_fields = entries[last_entry_num].split('\0')
767
if len(last_fields) < entry_field_count:
768
# The very last hunk was not complete,
769
# read the previous hunk
770
after = mid + len(block) - len(entries[-1])
772
last_fields = entries[last_entry_num].split('\0')
774
after = mid + len(block)
776
last_dir = last_fields[1]
777
last_loc = bisect.bisect_right(post, last_dir)
779
middle_files = post[:last_loc]
780
post = post[last_loc:]
783
# We have files that should occur in this block
784
# (>= first, <= last)
785
# Either we will find them here, or we can mark them as
788
if middle_files[0] == first_dir:
789
# We might need to go before this location
790
pre.append(first_dir)
791
if middle_files[-1] == last_dir:
792
post.insert(0, last_dir)
794
# Find out what paths we have
795
paths = {first_dir:[first_fields]}
796
# last_dir might == first_dir so we need to be
797
# careful if we should append rather than overwrite
798
if last_entry_num != first_entry_num:
799
paths.setdefault(last_dir, []).append(last_fields)
800
for num in xrange(first_entry_num+1, last_entry_num):
801
# TODO: jam 20070223 We are already splitting here, so
802
# shouldn't we just split the whole thing rather
803
# than doing the split again in add_one_record?
804
fields = entries[num].split('\0')
805
paths.setdefault(fields[1], []).append(fields)
807
for cur_dir in middle_files:
808
for fields in paths.get(cur_dir, []):
809
# offset by 1 because of the opening '\0'
810
# consider changing fields_to_entry to avoid the
812
entry = fields_to_entry(fields[1:])
813
found.setdefault(cur_dir, []).append(entry)
815
# Now we have split up everything into pre, middle, and post, and
816
# we have handled everything that fell in 'middle'.
817
# We add 'post' first, so that we prefer to seek towards the
818
# beginning, so that we will tend to go as early as we need, and
819
# then only seek forward after that.
821
pending.append((after, high, post))
823
pending.append((low, start-1, pre))
827
def _bisect_recursive(self, paths):
828
"""Bisect for entries for all paths and their children.
830
This will use bisect to find all records for the supplied paths. It
831
will then continue to bisect for any records which are marked as
832
directories. (and renames?)
834
:param paths: A sorted list of (dir, name) pairs
835
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
836
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
838
# Map from (dir, name, file_id) => [tree_info]
841
found_dir_names = set()
843
# Directories that have been read
844
processed_dirs = set()
845
# Get the ball rolling with the first bisect for all entries.
846
newly_found = self._bisect(paths)
849
# Directories that need to be read
851
paths_to_search = set()
852
for entry_list in newly_found.itervalues():
853
for dir_name_id, trees_info in entry_list:
854
found[dir_name_id] = trees_info
855
found_dir_names.add(dir_name_id[:2])
857
for tree_info in trees_info:
858
minikind = tree_info[0]
861
# We already processed this one as a directory,
862
# we don't need to do the extra work again.
864
subdir, name, file_id = dir_name_id
865
path = osutils.pathjoin(subdir, name)
867
if path not in processed_dirs:
868
pending_dirs.add(path)
869
elif minikind == 'r':
870
# Rename, we need to directly search the target
871
# which is contained in the fingerprint column
872
dir_name = osutils.split(tree_info[1])
873
if dir_name[0] in pending_dirs:
874
# This entry will be found in the dir search
876
if dir_name not in found_dir_names:
877
paths_to_search.add(tree_info[1])
878
# Now we have a list of paths to look for directly, and
879
# directory blocks that need to be read.
880
# newly_found is mixing the keys between (dir, name) and path
881
# entries, but that is okay, because we only really care about the
883
newly_found = self._bisect(sorted(paths_to_search))
884
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
885
processed_dirs.update(pending_dirs)
888
def _empty_parent_info(self):
889
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
892
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
893
"""Ensure a block for dirname exists.
895
This function exists to let callers which know that there is a
896
directory dirname ensure that the block for it exists. This block can
897
fail to exist because of demand loading, or because a directory had no
898
children. In either case it is not an error. It is however an error to
899
call this if there is no parent entry for the directory, and thus the
900
function requires the coordinates of such an entry to be provided.
902
The root row is special cased and can be indicated with a parent block
905
:param parent_block_index: The index of the block in which dirname's row
907
:param parent_row_index: The index in the parent block where the row
909
:param dirname: The utf8 dirname to ensure there is a block for.
910
:return: The index for the block.
912
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
913
# This is the signature of the root row, and the
914
# contents-of-root row is always index 1
916
# the basename of the directory must be the end of its full name.
917
if not (parent_block_index == -1 and
918
parent_block_index == -1 and dirname == ''):
919
assert dirname.endswith(
920
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
921
block_index, present = self._find_block_index_from_key((dirname, '', ''))
923
## In future, when doing partial parsing, this should load and
924
# populate the entire block.
925
self._dirblocks.insert(block_index, (dirname, []))
928
def _entries_to_current_state(self, new_entries):
929
"""Load new_entries into self.dirblocks.
931
Process new_entries into the current state object, making them the active
932
state. The entries are grouped together by directory to form dirblocks.
934
:param new_entries: A sorted list of entries. This function does not sort
935
to prevent unneeded overhead when callers have a sorted list already.
938
assert new_entries[0][0][0:2] == ('', ''), \
939
"Missing root row %r" % (new_entries[0][0],)
940
# The two blocks here are deliberate: the root block and the
941
# contents-of-root block.
942
self._dirblocks = [('', []), ('', [])]
943
current_block = self._dirblocks[0][1]
946
append_entry = current_block.append
947
for entry in new_entries:
948
if entry[0][0] != current_dirname:
949
# new block - different dirname
951
current_dirname = entry[0][0]
952
self._dirblocks.append((current_dirname, current_block))
953
append_entry = current_block.append
954
# append the entry to the current block
956
self._split_root_dirblock_into_contents()
958
def _split_root_dirblock_into_contents(self):
959
"""Split the root dirblocks into root and contents-of-root.
961
After parsing by path, we end up with root entries and contents-of-root
962
entries in the same block. This loop splits them out again.
964
# The above loop leaves the "root block" entries mixed with the
965
# "contents-of-root block". But we don't want an if check on
966
# all entries, so instead we just fix it up here.
967
assert self._dirblocks[1] == ('', [])
969
contents_of_root_block = []
970
for entry in self._dirblocks[0][1]:
971
if not entry[0][1]: # This is a root entry
972
root_block.append(entry)
974
contents_of_root_block.append(entry)
975
self._dirblocks[0] = ('', root_block)
976
self._dirblocks[1] = ('', contents_of_root_block)
978
def _entry_to_line(self, entry):
979
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
981
:param entry: An entry_tuple as defined in the module docstring.
983
entire_entry = list(entry[0])
984
for tree_number, tree_data in enumerate(entry[1]):
985
# (minikind, fingerprint, size, executable, tree_specific_string)
986
entire_entry.extend(tree_data)
987
# 3 for the key, 5 for the fields per tree.
988
tree_offset = 3 + tree_number * 5
990
entire_entry[tree_offset + 0] = tree_data[0]
992
entire_entry[tree_offset + 2] = str(tree_data[2])
994
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
995
return '\0'.join(entire_entry)
997
def _fields_per_entry(self):
998
"""How many null separated fields should be in each entry row.
1000
Each line now has an extra '\n' field which is not used
1001
so we just skip over it
1003
3 fields for the key
1004
+ number of fields per tree_data (5) * tree count
1007
tree_count = 1 + self._num_present_parents()
1008
return 3 + 5 * tree_count + 1
1010
def _find_block(self, key, add_if_missing=False):
1011
"""Return the block that key should be present in.
1013
:param key: A dirstate entry key.
1014
:return: The block tuple.
1016
block_index, present = self._find_block_index_from_key(key)
1018
if not add_if_missing:
1019
# check to see if key is versioned itself - we might want to
1020
# add it anyway, because dirs with no entries dont get a
1021
# dirblock at parse time.
1022
# This is an uncommon branch to take: most dirs have children,
1023
# and most code works with versioned paths.
1024
parent_base, parent_name = osutils.split(key[0])
1025
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1026
# some parent path has not been added - its an error to add
1028
raise errors.NotVersionedError(key[0:2], str(self))
1029
self._dirblocks.insert(block_index, (key[0], []))
1030
return self._dirblocks[block_index]
1032
def _find_block_index_from_key(self, key):
1033
"""Find the dirblock index for a key.
1035
:return: The block index, True if the block for the key is present.
1037
if key[0:2] == ('', ''):
1039
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1040
cache=self._split_path_cache)
1041
# _right returns one-past-where-key is so we have to subtract
1042
# one to use it. we use _right here because there are two
1043
# '' blocks - the root, and the contents of root
1044
# we always have a minimum of 2 in self._dirblocks: root and
1045
# root-contents, and for '', we get 2 back, so this is
1046
# simple and correct:
1047
present = (block_index < len(self._dirblocks) and
1048
self._dirblocks[block_index][0] == key[0])
1049
return block_index, present
1051
def _find_entry_index(self, key, block):
1052
"""Find the entry index for a key in a block.
1054
:return: The entry index, True if the entry for the key is present.
1056
entry_index = bisect.bisect_left(block, (key, []))
1057
present = (entry_index < len(block) and
1058
block[entry_index][0] == key)
1059
return entry_index, present
1062
def from_tree(tree, dir_state_filename):
1063
"""Create a dirstate from a bzr Tree.
1065
:param tree: The tree which should provide parent information and
1067
:return: a DirState object which is currently locked for writing.
1068
(it was locked by DirState.initialize)
1070
result = DirState.initialize(dir_state_filename)
1074
parent_ids = tree.get_parent_ids()
1075
num_parents = len(parent_ids)
1077
for parent_id in parent_ids:
1078
parent_tree = tree.branch.repository.revision_tree(parent_id)
1079
parent_trees.append((parent_id, parent_tree))
1080
parent_tree.lock_read()
1081
result.set_parent_trees(parent_trees, [])
1082
result.set_state_from_inventory(tree.inventory)
1084
for revid, parent_tree in parent_trees:
1085
parent_tree.unlock()
1088
# The caller won't have a chance to unlock this, so make sure we
1094
def update_entry(self, entry, abspath, stat_value,
1095
_stat_to_minikind=_stat_to_minikind,
1096
_pack_stat=pack_stat):
1097
"""Update the entry based on what is actually on disk.
1099
:param entry: This is the dirblock entry for the file in question.
1100
:param abspath: The path on disk for this file.
1101
:param stat_value: (optional) if we already have done a stat on the
1103
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1107
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1111
packed_stat = _pack_stat(stat_value)
1112
(saved_minikind, saved_link_or_sha1, saved_file_size,
1113
saved_executable, saved_packed_stat) = entry[1][0]
1115
if (minikind == saved_minikind
1116
and packed_stat == saved_packed_stat):
1117
# The stat hasn't changed since we saved, so we can re-use the
1122
# size should also be in packed_stat
1123
if saved_file_size == stat_value.st_size:
1124
return saved_link_or_sha1
1126
# If we have gotten this far, that means that we need to actually
1127
# process this entry.
1130
link_or_sha1 = self._sha1_file(abspath, entry)
1131
executable = self._is_executable(stat_value.st_mode,
1133
if self._cutoff_time is None:
1134
self._sha_cutoff_time()
1135
if (stat_value.st_mtime < self._cutoff_time
1136
and stat_value.st_ctime < self._cutoff_time):
1137
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1138
executable, packed_stat)
1140
entry[1][0] = ('f', '', stat_value.st_size,
1141
executable, DirState.NULLSTAT)
1142
elif minikind == 'd':
1144
entry[1][0] = ('d', '', 0, False, packed_stat)
1145
if saved_minikind != 'd':
1146
# This changed from something into a directory. Make sure we
1147
# have a directory block for it. This doesn't happen very
1148
# often, so this doesn't have to be super fast.
1149
block_index, entry_index, dir_present, file_present = \
1150
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1151
self._ensure_block(block_index, entry_index,
1152
osutils.pathjoin(entry[0][0], entry[0][1]))
1153
elif minikind == 'l':
1154
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1155
if self._cutoff_time is None:
1156
self._sha_cutoff_time()
1157
if (stat_value.st_mtime < self._cutoff_time
1158
and stat_value.st_ctime < self._cutoff_time):
1159
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1162
entry[1][0] = ('l', '', stat_value.st_size,
1163
False, DirState.NULLSTAT)
1164
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1167
def _sha_cutoff_time(self):
1168
"""Return cutoff time.
1170
Files modified more recently than this time are at risk of being
1171
undetectably modified and so can't be cached.
1173
# Cache the cutoff time as long as we hold a lock.
1174
# time.time() isn't super expensive (approx 3.38us), but
1175
# when you call it 50,000 times it adds up.
1176
# For comparison, os.lstat() costs 7.2us if it is hot.
1177
self._cutoff_time = int(time.time()) - 3
1178
return self._cutoff_time
1180
def _lstat(self, abspath, entry):
1181
"""Return the os.lstat value for this path."""
1182
return os.lstat(abspath)
1184
def _sha1_file(self, abspath, entry):
1185
"""Calculate the SHA1 of a file by reading the full text"""
1186
f = file(abspath, 'rb', buffering=65000)
1188
return osutils.sha_file(f)
1192
def _is_executable(self, mode, old_executable):
1193
"""Is this file executable?"""
1194
return bool(S_IEXEC & mode)
1196
def _is_executable_win32(self, mode, old_executable):
1197
"""On win32 the executable bit is stored in the dirstate."""
1198
return old_executable
1200
if sys.platform == 'win32':
1201
_is_executable = _is_executable_win32
1203
def _read_link(self, abspath, old_link):
1204
"""Read the target of a symlink"""
1205
# TODO: jam 200700301 On Win32, this could just return the value
1206
# already in memory. However, this really needs to be done at a
1207
# higher level, because there either won't be anything on disk,
1208
# or the thing on disk will be a file.
1209
return os.readlink(abspath)
1211
def get_ghosts(self):
1212
"""Return a list of the parent tree revision ids that are ghosts."""
1213
self._read_header_if_needed()
1216
def get_lines(self):
1217
"""Serialise the entire dirstate to a sequence of lines."""
1218
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1219
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1220
# read whats on disk.
1221
self._state_file.seek(0)
1222
return self._state_file.readlines()
1224
lines.append(self._get_parents_line(self.get_parent_ids()))
1225
lines.append(self._get_ghosts_line(self._ghosts))
1226
# append the root line which is special cased
1227
lines.extend(map(self._entry_to_line, self._iter_entries()))
1228
return self._get_output_lines(lines)
1230
def _get_ghosts_line(self, ghost_ids):
1231
"""Create a line for the state file for ghost information."""
1232
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1234
def _get_parents_line(self, parent_ids):
1235
"""Create a line for the state file for parents information."""
1236
return '\0'.join([str(len(parent_ids))] + parent_ids)
1238
def _get_fields_to_entry(self):
1239
"""Get a function which converts entry fields into a entry record.
1241
This handles size and executable, as well as parent records.
1243
:return: A function which takes a list of fields, and returns an
1244
appropriate record for storing in memory.
1246
# This is intentionally unrolled for performance
1247
num_present_parents = self._num_present_parents()
1248
if num_present_parents == 0:
1249
def fields_to_entry_0_parents(fields, _int=int):
1250
path_name_file_id_key = (fields[0], fields[1], fields[2])
1251
return (path_name_file_id_key, [
1253
fields[3], # minikind
1254
fields[4], # fingerprint
1255
_int(fields[5]), # size
1256
fields[6] == 'y', # executable
1257
fields[7], # packed_stat or revision_id
1259
return fields_to_entry_0_parents
1260
elif num_present_parents == 1:
1261
def fields_to_entry_1_parent(fields, _int=int):
1262
path_name_file_id_key = (fields[0], fields[1], fields[2])
1263
return (path_name_file_id_key, [
1265
fields[3], # minikind
1266
fields[4], # fingerprint
1267
_int(fields[5]), # size
1268
fields[6] == 'y', # executable
1269
fields[7], # packed_stat or revision_id
1272
fields[8], # minikind
1273
fields[9], # fingerprint
1274
_int(fields[10]), # size
1275
fields[11] == 'y', # executable
1276
fields[12], # packed_stat or revision_id
1279
return fields_to_entry_1_parent
1280
elif num_present_parents == 2:
1281
def fields_to_entry_2_parents(fields, _int=int):
1282
path_name_file_id_key = (fields[0], fields[1], fields[2])
1283
return (path_name_file_id_key, [
1285
fields[3], # minikind
1286
fields[4], # fingerprint
1287
_int(fields[5]), # size
1288
fields[6] == 'y', # executable
1289
fields[7], # packed_stat or revision_id
1292
fields[8], # minikind
1293
fields[9], # fingerprint
1294
_int(fields[10]), # size
1295
fields[11] == 'y', # executable
1296
fields[12], # packed_stat or revision_id
1299
fields[13], # minikind
1300
fields[14], # fingerprint
1301
_int(fields[15]), # size
1302
fields[16] == 'y', # executable
1303
fields[17], # packed_stat or revision_id
1306
return fields_to_entry_2_parents
1308
def fields_to_entry_n_parents(fields, _int=int):
1309
path_name_file_id_key = (fields[0], fields[1], fields[2])
1310
trees = [(fields[cur], # minikind
1311
fields[cur+1], # fingerprint
1312
_int(fields[cur+2]), # size
1313
fields[cur+3] == 'y', # executable
1314
fields[cur+4], # stat or revision_id
1315
) for cur in xrange(3, len(fields)-1, 5)]
1316
return path_name_file_id_key, trees
1317
return fields_to_entry_n_parents
1319
def get_parent_ids(self):
1320
"""Return a list of the parent tree ids for the directory state."""
1321
self._read_header_if_needed()
1322
return list(self._parents)
1324
def _get_block_entry_index(self, dirname, basename, tree_index):
1325
"""Get the coordinates for a path in the state structure.
1327
:param dirname: The utf8 dirname to lookup.
1328
:param basename: The utf8 basename to lookup.
1329
:param tree_index: The index of the tree for which this lookup should
1331
:return: A tuple describing where the path is located, or should be
1332
inserted. The tuple contains four fields: the block index, the row
1333
index, anda two booleans are True when the directory is present, and
1334
when the entire path is present. There is no guarantee that either
1335
coordinate is currently reachable unless the found field for it is
1336
True. For instance, a directory not present in the searched tree
1337
may be returned with a value one greater than the current highest
1338
block offset. The directory present field will always be True when
1339
the path present field is True. The directory present field does
1340
NOT indicate that the directory is present in the searched tree,
1341
rather it indicates that there are at least some files in some
1344
self._read_dirblocks_if_needed()
1345
key = dirname, basename, ''
1346
block_index, present = self._find_block_index_from_key(key)
1348
# no such directory - return the dir index and 0 for the row.
1349
return block_index, 0, False, False
1350
block = self._dirblocks[block_index][1] # access the entries only
1351
entry_index, present = self._find_entry_index(key, block)
1352
# linear search through present entries at this path to find the one
1354
while entry_index < len(block) and block[entry_index][0][1] == basename:
1355
if block[entry_index][1][tree_index][0] not in \
1356
('a', 'r'): # absent, relocated
1357
return block_index, entry_index, True, True
1359
return block_index, entry_index, True, False
1361
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1362
"""Get the dirstate entry for path in tree tree_index
1364
If either file_id or path is supplied, it is used as the key to lookup.
1365
If both are supplied, the fastest lookup is used, and an error is
1366
raised if they do not both point at the same row.
1368
:param tree_index: The index of the tree we wish to locate this path
1369
in. If the path is present in that tree, the entry containing its
1370
details is returned, otherwise (None, None) is returned
1371
0 is the working tree, higher indexes are successive parent
1373
:param fileid_utf8: A utf8 file_id to look up.
1374
:param path_utf8: An utf8 path to be looked up.
1375
:return: The dirstate entry tuple for path, or (None, None)
1377
self._read_dirblocks_if_needed()
1378
if path_utf8 is not None:
1379
assert path_utf8.__class__ == str, 'path_utf8 is not a str: %s %s' % (type(path_utf8), path_utf8)
1380
# path lookups are faster
1381
dirname, basename = osutils.split(path_utf8)
1382
block_index, entry_index, dir_present, file_present = \
1383
self._get_block_entry_index(dirname, basename, tree_index)
1384
if not file_present:
1386
entry = self._dirblocks[block_index][1][entry_index]
1387
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1389
if entry[0][2] != fileid_utf8:
1390
raise errors.BzrError('integrity error ? : mismatching'
1391
' tree_index, file_id and path')
1394
assert fileid_utf8 is not None
1395
possible_keys = self._get_id_index().get(fileid_utf8, None)
1396
if not possible_keys:
1398
for key in possible_keys:
1399
block_index, present = \
1400
self._find_block_index_from_key(key)
1401
# strange, probably indicates an out of date
1402
# id index - for now, allow this.
1405
# WARNING: DO not change this code to use _get_block_entry_index
1406
# as that function is not suitable: it does not use the key
1407
# to lookup, and thus the wront coordinates are returned.
1408
block = self._dirblocks[block_index][1]
1409
entry_index, present = self._find_entry_index(key, block)
1411
entry = self._dirblocks[block_index][1][entry_index]
1412
if entry[1][tree_index][0] in 'fdlt':
1413
# this is the result we are looking for: the
1414
# real home of this file_id in this tree.
1416
if entry[1][tree_index][0] == 'a':
1417
# there is no home for this entry in this tree
1419
assert entry[1][tree_index][0] == 'r', \
1420
"entry %r has invalid minikind %r for tree %r" \
1422
entry[1][tree_index][0],
1424
real_path = entry[1][tree_index][1]
1425
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1426
path_utf8=real_path)
1430
def initialize(cls, path):
1431
"""Create a new dirstate on path.
1433
The new dirstate will be an empty tree - that is it has no parents,
1434
and only a root node - which has id ROOT_ID.
1436
:param path: The name of the file for the dirstate.
1437
:return: A write-locked DirState object.
1439
# This constructs a new DirState object on a path, sets the _state_file
1440
# to a new empty file for that path. It then calls _set_data() with our
1441
# stock empty dirstate information - a root with ROOT_ID, no children,
1442
# and no parents. Finally it calls save() to ensure that this data will
1445
# root dir and root dir contents with no children.
1446
empty_tree_dirblocks = [('', []), ('', [])]
1447
# a new root directory, with a NULLSTAT.
1448
empty_tree_dirblocks[0][1].append(
1449
(('', '', inventory.ROOT_ID), [
1450
('d', '', 0, False, DirState.NULLSTAT),
1454
result._set_data([], empty_tree_dirblocks)
1461
def _inv_entry_to_details(self, inv_entry):
1462
"""Convert an inventory entry (from a revision tree) to state details.
1464
:param inv_entry: An inventory entry whose sha1 and link targets can be
1465
relied upon, and which has a revision set.
1466
:return: A details tuple - the details for a single tree at a path +
1469
kind = inv_entry.kind
1470
minikind = DirState._kind_to_minikind[kind]
1471
tree_data = inv_entry.revision
1472
assert len(tree_data) > 0, 'empty revision for the inv_entry.'
1473
if kind == 'directory':
1477
elif kind == 'symlink':
1478
fingerprint = inv_entry.symlink_target or ''
1481
elif kind == 'file':
1482
fingerprint = inv_entry.text_sha1 or ''
1483
size = inv_entry.text_size or 0
1484
executable = inv_entry.executable
1485
elif kind == 'tree-reference':
1486
fingerprint = inv_entry.reference_revision or ''
1490
raise Exception("can't pack %s" % inv_entry)
1491
return (minikind, fingerprint, size, executable, tree_data)
1493
def _iter_entries(self):
1494
"""Iterate over all the entries in the dirstate.
1496
Each yelt item is an entry in the standard format described in the
1497
docstring of bzrlib.dirstate.
1499
self._read_dirblocks_if_needed()
1500
for directory in self._dirblocks:
1501
for entry in directory[1]:
1504
def _get_id_index(self):
1505
"""Get an id index of self._dirblocks."""
1506
if self._id_index is None:
1508
for key, tree_details in self._iter_entries():
1509
id_index.setdefault(key[2], set()).add(key)
1510
self._id_index = id_index
1511
return self._id_index
1513
def _get_output_lines(self, lines):
1514
"""format lines for final output.
1516
:param lines: A sequece of lines containing the parents list and the
1519
output_lines = [DirState.HEADER_FORMAT_3]
1520
lines.append('') # a final newline
1521
inventory_text = '\0\n\0'.join(lines)
1522
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1523
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1524
num_entries = len(lines)-3
1525
output_lines.append('num_entries: %s\n' % (num_entries,))
1526
output_lines.append(inventory_text)
1529
def _make_deleted_row(self, fileid_utf8, parents):
1530
"""Return a deleted for for fileid_utf8."""
1531
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1534
def _num_present_parents(self):
1535
"""The number of parent entries in each record row."""
1536
return len(self._parents) - len(self._ghosts)
1540
"""Construct a DirState on the file at path path.
1542
:return: An unlocked DirState object, associated with the given path.
1544
result = DirState(path)
1547
def _read_dirblocks_if_needed(self):
1548
"""Read in all the dirblocks from the file if they are not in memory.
1550
This populates self._dirblocks, and sets self._dirblock_state to
1551
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1554
self._read_header_if_needed()
1555
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1556
_read_dirblocks(self)
1558
def _read_header(self):
1559
"""This reads in the metadata header, and the parent ids.
1561
After reading in, the file should be positioned at the null
1562
just before the start of the first record in the file.
1564
:return: (expected crc checksum, number of entries, parent list)
1566
self._read_prelude()
1567
parent_line = self._state_file.readline()
1568
info = parent_line.split('\0')
1569
num_parents = int(info[0])
1570
assert num_parents == len(info)-2, 'incorrect parent info line'
1571
self._parents = info[1:-1]
1573
ghost_line = self._state_file.readline()
1574
info = ghost_line.split('\0')
1575
num_ghosts = int(info[1])
1576
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1577
self._ghosts = info[2:-1]
1578
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1579
self._end_of_header = self._state_file.tell()
1581
def _read_header_if_needed(self):
1582
"""Read the header of the dirstate file if needed."""
1583
# inline this as it will be called a lot
1584
if not self._lock_token:
1585
raise errors.ObjectNotLocked(self)
1586
if self._header_state == DirState.NOT_IN_MEMORY:
1589
def _read_prelude(self):
1590
"""Read in the prelude header of the dirstate file
1592
This only reads in the stuff that is not connected to the crc
1593
checksum. The position will be correct to read in the rest of
1594
the file and check the checksum after this point.
1595
The next entry in the file should be the number of parents,
1596
and their ids. Followed by a newline.
1598
header = self._state_file.readline()
1599
assert header == DirState.HEADER_FORMAT_3, \
1600
'invalid header line: %r' % (header,)
1601
crc_line = self._state_file.readline()
1602
assert crc_line.startswith('crc32: '), 'missing crc32 checksum'
1603
self.crc_expected = int(crc_line[len('crc32: '):-1])
1604
num_entries_line = self._state_file.readline()
1605
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1606
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1609
"""Save any pending changes created during this session.
1611
We reuse the existing file, because that prevents race conditions with
1612
file creation, and use oslocks on it to prevent concurrent modification
1613
and reads - because dirstates incremental data aggretation is not
1614
compatible with reading a modified file, and replacing a file in use by
1615
another process is impossible on windows.
1617
A dirstate in read only mode should be smart enough though to validate
1618
that the file has not changed, and otherwise discard its cache and
1619
start over, to allow for fine grained read lock duration, so 'status'
1620
wont block 'commit' - for example.
1622
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1623
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1625
grabbed_write_lock = False
1626
if self._lock_state != 'w':
1627
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
1628
# Switch over to the new lock, as the old one may be closed.
1629
# TODO: jam 20070315 We should validate the disk file has
1630
# not changed contents. Since temporary_write_lock may
1631
# not be an atomic operation.
1632
self._lock_token = new_lock
1633
self._state_file = new_lock.f
1634
if not grabbed_write_lock:
1635
# We couldn't grab a write lock, so we switch back to a read one
1638
self._state_file.seek(0)
1639
self._state_file.writelines(self.get_lines())
1640
self._state_file.truncate()
1641
self._state_file.flush()
1642
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1643
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1645
if grabbed_write_lock:
1646
self._lock_token = self._lock_token.restore_read_lock()
1647
self._state_file = self._lock_token.f
1648
# TODO: jam 20070315 We should validate the disk file has
1649
# not changed contents. Since restore_read_lock may
1650
# not be an atomic operation.
1652
def _set_data(self, parent_ids, dirblocks):
1653
"""Set the full dirstate data in memory.
1655
This is an internal function used to completely replace the objects
1656
in memory state. It puts the dirstate into state 'full-dirty'.
1658
:param parent_ids: A list of parent tree revision ids.
1659
:param dirblocks: A list containing one tuple for each directory in the
1660
tree. Each tuple contains the directory path and a list of entries
1661
found in that directory.
1663
# our memory copy is now authoritative.
1664
self._dirblocks = dirblocks
1665
self._header_state = DirState.IN_MEMORY_MODIFIED
1666
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1667
self._parents = list(parent_ids)
1668
self._id_index = None
1670
def set_path_id(self, path, new_id):
1671
"""Change the id of path to new_id in the current working tree.
1673
:param path: The path inside the tree to set - '' is the root, 'foo'
1674
is the path foo in the root.
1675
:param new_id: The new id to assign to the path. This must be a utf8
1676
file id (not unicode, and not None).
1678
assert new_id.__class__ == str, \
1679
"path_id %r is not a plain string" % (new_id,)
1680
self._read_dirblocks_if_needed()
1683
raise NotImplementedError(self.set_path_id)
1684
# TODO: check new id is unique
1685
entry = self._get_entry(0, path_utf8=path)
1686
if entry[0][2] == new_id:
1687
# Nothing to change.
1689
# mark the old path absent, and insert a new root path
1690
self._make_absent(entry)
1691
self.update_minimal(('', '', new_id), 'd',
1692
path_utf8='', packed_stat=entry[1][0][4])
1693
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1694
if self._id_index is not None:
1695
self._id_index.setdefault(new_id, set()).add(entry[0])
1697
def set_parent_trees(self, trees, ghosts):
1698
"""Set the parent trees for the dirstate.
1700
:param trees: A list of revision_id, tree tuples. tree must be provided
1701
even if the revision_id refers to a ghost: supply an empty tree in
1703
:param ghosts: A list of the revision_ids that are ghosts at the time
1706
# TODO: generate a list of parent indexes to preserve to save
1707
# processing specific parent trees. In the common case one tree will
1708
# be preserved - the left most parent.
1709
# TODO: if the parent tree is a dirstate, we might want to walk them
1710
# all by path in parallel for 'optimal' common-case performance.
1711
# generate new root row.
1712
self._read_dirblocks_if_needed()
1713
# TODO future sketch: Examine the existing parents to generate a change
1714
# map and then walk the new parent trees only, mapping them into the
1715
# dirstate. Walk the dirstate at the same time to remove unreferenced
1718
# sketch: loop over all entries in the dirstate, cherry picking
1719
# entries from the parent trees, if they are not ghost trees.
1720
# after we finish walking the dirstate, all entries not in the dirstate
1721
# are deletes, so we want to append them to the end as per the design
1722
# discussions. So do a set difference on ids with the parents to
1723
# get deletes, and add them to the end.
1724
# During the update process we need to answer the following questions:
1725
# - find other keys containing a fileid in order to create cross-path
1726
# links. We dont't trivially use the inventory from other trees
1727
# because this leads to either double touching, or to accessing
1729
# - find other keys containing a path
1730
# We accumulate each entry via this dictionary, including the root
1733
# we could do parallel iterators, but because file id data may be
1734
# scattered throughout, we dont save on index overhead: we have to look
1735
# at everything anyway. We can probably save cycles by reusing parent
1736
# data and doing an incremental update when adding an additional
1737
# parent, but for now the common cases are adding a new parent (merge),
1738
# and replacing completely (commit), and commit is more common: so
1739
# optimise merge later.
1741
# ---- start generation of full tree mapping data
1742
# what trees should we use?
1743
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
1744
# how many trees do we end up with
1745
parent_count = len(parent_trees)
1747
# one: the current tree
1748
for entry in self._iter_entries():
1749
# skip entries not in the current tree
1750
if entry[1][0][0] in ('a', 'r'): # absent, relocated
1752
by_path[entry[0]] = [entry[1][0]] + \
1753
[DirState.NULL_PARENT_DETAILS] * parent_count
1754
id_index[entry[0][2]] = set([entry[0]])
1756
# now the parent trees:
1757
for tree_index, tree in enumerate(parent_trees):
1758
# the index is off by one, adjust it.
1759
tree_index = tree_index + 1
1760
# when we add new locations for a fileid we need these ranges for
1761
# any fileid in this tree as we set the by_path[id] to:
1762
# already_processed_tree_details + new_details + new_location_suffix
1763
# the suffix is from tree_index+1:parent_count+1.
1764
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
1765
# now stitch in all the entries from this tree
1766
for path, entry in tree.inventory.iter_entries_by_dir():
1767
# here we process each trees details for each item in the tree.
1768
# we first update any existing entries for the id at other paths,
1769
# then we either create or update the entry for the id at the
1770
# right path, and finally we add (if needed) a mapping from
1771
# file_id to this path. We do it in this order to allow us to
1772
# avoid checking all known paths for the id when generating a
1773
# new entry at this path: by adding the id->path mapping last,
1774
# all the mappings are valid and have correct relocation
1775
# records where needed.
1776
file_id = entry.file_id
1777
path_utf8 = path.encode('utf8')
1778
dirname, basename = osutils.split(path_utf8)
1779
new_entry_key = (dirname, basename, file_id)
1780
# tree index consistency: All other paths for this id in this tree
1781
# index must point to the correct path.
1782
for entry_key in id_index.setdefault(file_id, set()):
1783
# TODO:PROFILING: It might be faster to just update
1784
# rather than checking if we need to, and then overwrite
1785
# the one we are located at.
1786
if entry_key != new_entry_key:
1787
# this file id is at a different path in one of the
1788
# other trees, so put absent pointers there
1789
# This is the vertical axis in the matrix, all pointing
1791
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
1792
# by path consistency: Insert into an existing path record (trivial), or
1793
# add a new one with relocation pointers for the other tree indexes.
1794
if new_entry_key in id_index[file_id]:
1795
# there is already an entry where this data belongs, just insert it.
1796
by_path[new_entry_key][tree_index] = \
1797
self._inv_entry_to_details(entry)
1799
# add relocated entries to the horizontal axis - this row
1800
# mapping from path,id. We need to look up the correct path
1801
# for the indexes from 0 to tree_index -1
1803
for lookup_index in xrange(tree_index):
1804
# boundary case: this is the first occurence of file_id
1805
# so there are no id_indexs, possibly take this out of
1807
if not len(id_index[file_id]):
1808
new_details.append(DirState.NULL_PARENT_DETAILS)
1810
# grab any one entry, use it to find the right path.
1811
# TODO: optimise this to reduce memory use in highly
1812
# fragmented situations by reusing the relocation
1814
a_key = iter(id_index[file_id]).next()
1815
if by_path[a_key][lookup_index][0] in ('r', 'a'):
1816
# its a pointer or missing statement, use it as is.
1817
new_details.append(by_path[a_key][lookup_index])
1819
# we have the right key, make a pointer to it.
1820
real_path = ('/'.join(a_key[0:2])).strip('/')
1821
new_details.append(('r', real_path, 0, False, ''))
1822
new_details.append(self._inv_entry_to_details(entry))
1823
new_details.extend(new_location_suffix)
1824
by_path[new_entry_key] = new_details
1825
id_index[file_id].add(new_entry_key)
1826
# --- end generation of full tree mappings
1828
# sort and output all the entries
1829
new_entries = self._sort_entries(by_path.items())
1830
self._entries_to_current_state(new_entries)
1831
self._parents = [rev_id for rev_id, tree in trees]
1832
self._ghosts = list(ghosts)
1833
self._header_state = DirState.IN_MEMORY_MODIFIED
1834
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1835
self._id_index = id_index
1837
def _sort_entries(self, entry_list):
1838
"""Given a list of entries, sort them into the right order.
1840
This is done when constructing a new dirstate from trees - normally we
1841
try to keep everything in sorted blocks all the time, but sometimes
1842
it's easier to sort after the fact.
1844
# TODO: Might be faster to do a schwartzian transform?
1846
# sort by: directory parts, file name, file id
1847
return entry[0][0].split('/'), entry[0][1], entry[0][2]
1848
return sorted(entry_list, key=_key)
1850
def set_state_from_inventory(self, new_inv):
1851
"""Set new_inv as the current state.
1853
This API is called by tree transform, and will usually occur with
1854
existing parent trees.
1856
:param new_inv: The inventory object to set current state from.
1858
self._read_dirblocks_if_needed()
1860
# incremental algorithm:
1861
# two iterators: current data and new data, both in dirblock order.
1862
new_iterator = new_inv.iter_entries_by_dir()
1863
# we will be modifying the dirstate, so we need a stable iterator. In
1864
# future we might write one, for now we just clone the state into a
1865
# list - which is a shallow copy, so each
1866
old_iterator = iter(list(self._iter_entries()))
1867
# both must have roots so this is safe:
1868
current_new = new_iterator.next()
1869
current_old = old_iterator.next()
1870
def advance(iterator):
1872
return iterator.next()
1873
except StopIteration:
1875
while current_new or current_old:
1876
# skip entries in old that are not really there
1877
if current_old and current_old[1][0][0] in ('r', 'a'):
1878
# relocated or absent
1879
current_old = advance(old_iterator)
1882
# convert new into dirblock style
1883
new_path_utf8 = current_new[0].encode('utf8')
1884
new_dirname, new_basename = osutils.split(new_path_utf8)
1885
new_id = current_new[1].file_id
1886
new_entry_key = (new_dirname, new_basename, new_id)
1887
current_new_minikind = \
1888
DirState._kind_to_minikind[current_new[1].kind]
1889
if current_new_minikind == 't':
1890
fingerprint = current_new[1].reference_revision
1894
# for safety disable variables
1895
new_path_utf8 = new_dirname = new_basename = new_id = new_entry_key = None
1896
# 5 cases, we dont have a value that is strictly greater than everything, so
1897
# we make both end conditions explicit
1899
# old is finished: insert current_new into the state.
1900
self.update_minimal(new_entry_key, current_new_minikind,
1901
executable=current_new[1].executable,
1902
path_utf8=new_path_utf8, fingerprint=fingerprint)
1903
current_new = advance(new_iterator)
1904
elif not current_new:
1906
self._make_absent(current_old)
1907
current_old = advance(old_iterator)
1908
elif new_entry_key == current_old[0]:
1909
# same - common case
1910
# TODO: update the record if anything significant has changed.
1911
# the minimal required trigger is if the execute bit or cached
1913
if (current_old[1][0][3] != current_new[1].executable or
1914
current_old[1][0][0] != current_new_minikind):
1915
self.update_minimal(current_old[0], current_new_minikind,
1916
executable=current_new[1].executable,
1917
path_utf8=new_path_utf8, fingerprint=fingerprint)
1918
# both sides are dealt with, move on
1919
current_old = advance(old_iterator)
1920
current_new = advance(new_iterator)
1921
elif (new_entry_key[0].split('/') < current_old[0][0].split('/')
1922
and new_entry_key[1:] < current_old[0][1:]):
1924
# add a entry for this and advance new
1925
self.update_minimal(new_entry_key, current_new_minikind,
1926
executable=current_new[1].executable,
1927
path_utf8=new_path_utf8, fingerprint=fingerprint)
1928
current_new = advance(new_iterator)
1931
self._make_absent(current_old)
1932
current_old = advance(old_iterator)
1933
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1934
self._id_index = None
1936
def _make_absent(self, current_old):
1937
"""Mark current_old - an entry - as absent for tree 0.
1939
:return: True if this was the last details entry for they entry key:
1940
that is, if the underlying block has had the entry removed, thus
1941
shrinking in length.
1943
# build up paths that this id will be left at after the change is made,
1944
# so we can update their cross references in tree 0
1945
all_remaining_keys = set()
1946
# Dont check the working tree, because its going.
1947
for details in current_old[1][1:]:
1948
if details[0] not in ('a', 'r'): # absent, relocated
1949
all_remaining_keys.add(current_old[0])
1950
elif details[0] == 'r': # relocated
1951
# record the key for the real path.
1952
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
1953
# absent rows are not present at any path.
1954
last_reference = current_old[0] not in all_remaining_keys
1956
# the current row consists entire of the current item (being marked
1957
# absent), and relocated or absent entries for the other trees:
1958
# Remove it, its meaningless.
1959
block = self._find_block(current_old[0])
1960
entry_index, present = self._find_entry_index(current_old[0], block[1])
1961
assert present, 'could not find entry for %s' % (current_old,)
1962
block[1].pop(entry_index)
1963
# if we have an id_index in use, remove this key from it for this id.
1964
if self._id_index is not None:
1965
self._id_index[current_old[0][2]].remove(current_old[0])
1966
# update all remaining keys for this id to record it as absent. The
1967
# existing details may either be the record we are making as deleted
1968
# (if there were other trees with the id present at this path), or may
1970
for update_key in all_remaining_keys:
1971
update_block_index, present = \
1972
self._find_block_index_from_key(update_key)
1973
assert present, 'could not find block for %s' % (update_key,)
1974
update_entry_index, present = \
1975
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
1976
assert present, 'could not find entry for %s' % (update_key,)
1977
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
1978
# it must not be absent at the moment
1979
assert update_tree_details[0][0] != 'a' # absent
1980
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
1981
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1982
return last_reference
1984
def update_minimal(self, key, minikind, executable=False, fingerprint='',
1985
packed_stat=None, size=0, path_utf8=None):
1986
"""Update an entry to the state in tree 0.
1988
This will either create a new entry at 'key' or update an existing one.
1989
It also makes sure that any other records which might mention this are
1992
:param key: (dir, name, file_id) for the new entry
1993
:param minikind: The type for the entry ('f' == 'file', 'd' ==
1995
:param executable: Should the executable bit be set?
1996
:param fingerprint: Simple fingerprint for new entry.
1997
:param packed_stat: packed stat value for new entry.
1998
:param size: Size information for new entry
1999
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2002
block = self._find_block(key)[1]
2003
if packed_stat is None:
2004
packed_stat = DirState.NULLSTAT
2005
entry_index, present = self._find_entry_index(key, block)
2006
new_details = (minikind, fingerprint, size, executable, packed_stat)
2007
id_index = self._get_id_index()
2009
# new entry, synthesis cross reference here,
2010
existing_keys = id_index.setdefault(key[2], set())
2011
if not existing_keys:
2012
# not currently in the state, simplest case
2013
new_entry = key, [new_details] + self._empty_parent_info()
2015
# present at one or more existing other paths.
2016
# grab one of them and use it to generate parent
2017
# relocation/absent entries.
2018
new_entry = key, [new_details]
2019
for other_key in existing_keys:
2020
# change the record at other to be a pointer to this new
2021
# record. The loop looks similar to the change to
2022
# relocations when updating an existing record but its not:
2023
# the test for existing kinds is different: this can be
2024
# factored out to a helper though.
2025
other_block_index, present = self._find_block_index_from_key(other_key)
2026
assert present, 'could not find block for %s' % (other_key,)
2027
other_entry_index, present = self._find_entry_index(other_key,
2028
self._dirblocks[other_block_index][1])
2029
assert present, 'could not find entry for %s' % (other_key,)
2030
assert path_utf8 is not None
2031
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2032
('r', path_utf8, 0, False, '')
2034
num_present_parents = self._num_present_parents()
2035
for lookup_index in xrange(1, num_present_parents + 1):
2036
# grab any one entry, use it to find the right path.
2037
# TODO: optimise this to reduce memory use in highly
2038
# fragmented situations by reusing the relocation
2040
update_block_index, present = \
2041
self._find_block_index_from_key(other_key)
2042
assert present, 'could not find block for %s' % (other_key,)
2043
update_entry_index, present = \
2044
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2045
assert present, 'could not find entry for %s' % (other_key,)
2046
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2047
if update_details[0] in ('r', 'a'): # relocated, absent
2048
# its a pointer or absent in lookup_index's tree, use
2050
new_entry[1].append(update_details)
2052
# we have the right key, make a pointer to it.
2053
pointer_path = osutils.pathjoin(*other_key[0:2])
2054
new_entry[1].append(('r', pointer_path, 0, False, ''))
2055
block.insert(entry_index, new_entry)
2056
existing_keys.add(key)
2058
# Does the new state matter?
2059
block[entry_index][1][0] = new_details
2060
# parents cannot be affected by what we do.
2061
# other occurences of this id can be found
2062
# from the id index.
2064
# tree index consistency: All other paths for this id in this tree
2065
# index must point to the correct path. We have to loop here because
2066
# we may have passed entries in the state with this file id already
2067
# that were absent - where parent entries are - and they need to be
2068
# converted to relocated.
2069
assert path_utf8 is not None
2070
for entry_key in id_index.setdefault(key[2], set()):
2071
# TODO:PROFILING: It might be faster to just update
2072
# rather than checking if we need to, and then overwrite
2073
# the one we are located at.
2074
if entry_key != key:
2075
# this file id is at a different path in one of the
2076
# other trees, so put absent pointers there
2077
# This is the vertical axis in the matrix, all pointing
2079
block_index, present = self._find_block_index_from_key(entry_key)
2081
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2083
self._dirblocks[block_index][1][entry_index][1][0] = \
2084
('r', path_utf8, 0, False, '')
2085
# add a containing dirblock if needed.
2086
if new_details[0] == 'd':
2087
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2088
block_index, present = self._find_block_index_from_key(subdir_key)
2090
self._dirblocks.insert(block_index, (subdir_key[0], []))
2092
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2094
def _validate(self):
2095
"""Check that invariants on the dirblock are correct.
2097
This can be useful in debugging; it shouldn't be necessary in
2100
This must be called with a lock held.
2102
# NOTE: This must always raise AssertionError not just assert,
2103
# otherwise it may not behave properly under python -O
2105
# TODO: All entries must have some content that's not 'a' or 'r',
2106
# otherwise it could just be removed.
2108
# TODO: All relocations must point directly to a real entry.
2110
# TODO: No repeated keys.
2113
from pprint import pformat
2114
self._read_dirblocks_if_needed()
2115
if len(self._dirblocks) > 0:
2116
if not self._dirblocks[0][0] == '':
2117
raise AssertionError(
2118
"dirblocks don't start with root block:\n" + \
2120
if len(self._dirblocks) > 1:
2121
if not self._dirblocks[1][0] == '':
2122
raise AssertionError(
2123
"dirblocks missing root directory:\n" + \
2125
# the dirblocks are sorted by their path components, name, and dir id
2126
dir_names = [d[0].split('/')
2127
for d in self._dirblocks[1:]]
2128
if dir_names != sorted(dir_names):
2129
raise AssertionError(
2130
"dir names are not in sorted order:\n" + \
2131
pformat(self._dirblocks) + \
2134
for dirblock in self._dirblocks:
2135
# within each dirblock, the entries are sorted by filename and
2137
for entry in dirblock[1]:
2138
if dirblock[0] != entry[0][0]:
2139
raise AssertionError(
2141
"doesn't match directory name in\n%r" %
2142
(entry, pformat(dirblock)))
2143
if dirblock[1] != sorted(dirblock[1]):
2144
raise AssertionError(
2145
"dirblock for %r is not sorted:\n%s" % \
2146
(dirblock[0], pformat(dirblock)))
2149
def check_valid_parent():
2150
"""Check that the current entry has a valid parent.
2152
This makes sure that the parent has a record,
2153
and that the parent isn't marked as "absent" in the
2154
current tree. (It is invalid to have a non-absent file in an absent
2157
if entry[0][0:2] == ('', ''):
2158
# There should be no parent for the root row
2160
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2161
if parent_entry == (None, None):
2162
raise AssertionError(
2163
"no parent entry for: %s in tree %s"
2164
% (this_path, tree_index))
2165
if parent_entry[1][tree_index][0] != 'd':
2166
raise AssertionError(
2167
"Parent entry for %s is not marked as a valid"
2168
" directory. %s" % (this_path, parent_entry,))
2170
# For each file id, for each tree: either
2171
# the file id is not present at all; all rows with that id in the
2172
# key have it marked as 'absent'
2173
# OR the file id is present under exactly one name; any other entries
2174
# that mention that id point to the correct name.
2176
# We check this with a dict per tree pointing either to the present
2177
# name, or None if absent.
2178
tree_count = self._num_present_parents() + 1
2179
id_path_maps = [dict() for i in range(tree_count)]
2180
# Make sure that all renamed entries point to the correct location.
2181
for entry in self._iter_entries():
2182
file_id = entry[0][2]
2183
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2184
if len(entry[1]) != tree_count:
2185
raise AssertionError(
2186
"wrong number of entry details for row\n%s" \
2187
",\nexpected %d" % \
2188
(pformat(entry), tree_count))
2189
for tree_index, tree_state in enumerate(entry[1]):
2190
this_tree_map = id_path_maps[tree_index]
2191
minikind = tree_state[0]
2192
# have we seen this id before in this column?
2193
if file_id in this_tree_map:
2194
previous_path = this_tree_map[file_id]
2195
# any later mention of this file must be consistent with
2196
# what was said before
2198
if previous_path is not None:
2199
raise AssertionError(
2200
"file %s is absent in row %r but also present " \
2202
(file_id, entry, previous_path))
2203
elif minikind == 'r':
2204
target_location = tree_state[1]
2205
if previous_path != target_location:
2206
raise AssertionError(
2207
"file %s relocation in row %r but also at %r" \
2208
% (file_id, entry, previous_path))
2210
# a file, directory, etc - may have been previously
2211
# pointed to by a relocation, which must point here
2212
if previous_path != this_path:
2213
raise AssertionError(
2214
"entry %r inconsistent with previous path %r" % \
2215
(entry, previous_path))
2216
check_valid_parent()
2219
# absent; should not occur anywhere else
2220
this_tree_map[file_id] = None
2221
elif minikind == 'r':
2222
# relocation, must occur at expected location
2223
this_tree_map[file_id] = tree_state[1]
2225
this_tree_map[file_id] = this_path
2226
check_valid_parent()
2228
def _wipe_state(self):
2229
"""Forget all state information about the dirstate."""
2230
self._header_state = DirState.NOT_IN_MEMORY
2231
self._dirblock_state = DirState.NOT_IN_MEMORY
2234
self._dirblocks = []
2235
self._id_index = None
2236
self._end_of_header = None
2237
self._cutoff_time = None
2238
self._split_path_cache = {}
2240
def lock_read(self):
2241
"""Acquire a read lock on the dirstate"""
2242
if self._lock_token is not None:
2243
raise errors.LockContention(self._lock_token)
2244
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2245
# already in memory, we could read just the header and check for
2246
# any modification. If not modified, we can just leave things
2248
self._lock_token = lock.ReadLock(self._filename)
2249
self._lock_state = 'r'
2250
self._state_file = self._lock_token.f
2253
def lock_write(self):
2254
"""Acquire a write lock on the dirstate"""
2255
if self._lock_token is not None:
2256
raise errors.LockContention(self._lock_token)
2257
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2258
# already in memory, we could read just the header and check for
2259
# any modification. If not modified, we can just leave things
2261
self._lock_token = lock.WriteLock(self._filename)
2262
self._lock_state = 'w'
2263
self._state_file = self._lock_token.f
2267
"""Drop any locks held on the dirstate"""
2268
if self._lock_token is None:
2269
raise errors.LockNotHeld(self)
2270
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2271
# already in memory, we could read just the header and check for
2272
# any modification. If not modified, we can just leave things
2274
self._state_file = None
2275
self._lock_state = None
2276
self._lock_token.unlock()
2277
self._lock_token = None
2278
self._split_path_cache = {}
2280
def _requires_lock(self):
2281
"""Checks that a lock is currently held by someone on the dirstate"""
2282
if not self._lock_token:
2283
raise errors.ObjectNotLocked(self)
2286
# Try to load the compiled form if possible
2288
from bzrlib._dirstate_helpers_c import (
2289
_read_dirblocks_c as _read_dirblocks,
2290
bisect_dirblock_c as bisect_dirblock,
2291
_bisect_path_left_c as _bisect_path_left,
2292
_bisect_path_right_c as _bisect_path_right,
2293
cmp_by_dirs_c as cmp_by_dirs,
2296
from bzrlib._dirstate_helpers_py import (
2297
_read_dirblocks_py as _read_dirblocks,
2298
bisect_dirblock_py as bisect_dirblock,
2299
_bisect_path_left_py as _bisect_path_left,
2300
_bisect_path_right_py as _bisect_path_right,
2301
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
