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 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, 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 an 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 won't 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 sort.
166
- What's 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 reasonable.
171
Maybe we should do a test profile of the core structure - 10K simulated
172
searches/lookups/etc?
174
Objects for each row?
175
The lifetime of Dirstate objects is current per lock, but see above for
176
possible extensions. The lifetime of a row from a dirstate is expected to be
177
very short in the optimistic case: which we are optimising for. For instance,
178
subtree status will determine from analysis of the disk data what rows need to
179
be examined at all, and will be able to determine from a single row whether
180
that file has altered or not, so we are aiming to process tens of thousands of
181
entries each second within the dirstate context, before exposing anything to
182
the larger codebase. This suggests we want the time for a single file
183
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
184
processed, and to scale to 100 thousand we'll another order of magnitude to do
185
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
186
the file on disk, and then immediately discard, the overhead of object creation
187
becomes a significant cost.
189
Figures: Creating a tuple from from 3 elements was profiled at 0.0625
190
microseconds, whereas creating a object which is subclassed from tuple was
191
0.500 microseconds, and creating an object with 3 elements and slots was 3
192
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
193
down to 10 microseconds for the total processing - having 33% of that be object
194
creation is a huge overhead. There is a potential cost in using tuples within
195
each row which is that the conditional code to do comparisons may be slower
196
than method invocation, but method invocation is known to be slow due to stack
197
frame creation, so avoiding methods in these tight inner loops in unfortunately
198
desirable. We can consider a pyrex version of this with objects in future if
207
from stat import S_IEXEC
223
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
224
"""Convert stat values into a packed representation."""
225
# jam 20060614 it isn't really worth removing more entries if we
226
# are going to leave it in packed form.
227
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
228
# With all entries, filesize is 5.9M and read time is maybe 280ms
229
# well within the noise margin
231
# base64 encoding always adds a final newline, so strip it off
232
# The current version
233
return _encode(_pack('>LLLLLL'
234
, st.st_size, int(st.st_mtime), int(st.st_ctime)
235
, st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
236
# This is 0.060s / 1.520s faster by not encoding as much information
237
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
238
# This is not strictly faster than _encode(_pack())[:-1]
239
# return '%X.%X.%X.%X.%X.%X' % (
240
# st.st_size, int(st.st_mtime), int(st.st_ctime),
241
# st.st_dev, st.st_ino, st.st_mode)
242
# Similar to the _encode(_pack('>LL'))
243
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
246
class DirState(object):
247
"""Record directory and metadata state for fast access.
249
A dirstate is a specialised data structure for managing local working
250
tree state information. Its not yet well defined whether it is platform
251
specific, and if it is how we detect/parameterise that.
253
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
254
Unlike most bzr disk formats, DirStates must be locked for reading, using
255
lock_read. (This is an os file lock internally.) This is necessary
256
because the file can be rewritten in place.
258
DirStates must be explicitly written with save() to commit changes; just
259
unlocking them does not write the changes to disk.
262
_kind_to_minikind = {
268
'tree-reference': 't',
270
_minikind_to_kind = {
276
't': 'tree-reference',
278
_stat_to_minikind = {
283
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
284
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
287
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
288
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
289
# that we are sure a single record will always fit.
290
BISECT_PAGE_SIZE = 4096
293
IN_MEMORY_UNMODIFIED = 1
294
IN_MEMORY_MODIFIED = 2
296
# A pack_stat (the x's) that is just noise and will never match the output
299
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
301
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
302
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
304
def __init__(self, path):
305
"""Create a DirState object.
309
:attr _root_entrie: The root row of the directory/file information,
310
- contains the path to / - '', ''
311
- kind of 'directory',
312
- the file id of the root in utf8
315
- and no sha information.
316
:param path: The path at which the dirstate file on disk should live.
318
# _header_state and _dirblock_state represent the current state
319
# of the dirstate metadata and the per-row data respectiely.
320
# NOT_IN_MEMORY indicates that no data is in memory
321
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
322
# is the same as is on disk
323
# IN_MEMORY_MODIFIED indicates that we have a modified version
324
# of what is on disk.
325
# In future we will add more granularity, for instance _dirblock_state
326
# will probably support partially-in-memory as a separate variable,
327
# allowing for partially-in-memory unmodified and partially-in-memory
329
self._header_state = DirState.NOT_IN_MEMORY
330
self._dirblock_state = DirState.NOT_IN_MEMORY
334
self._state_file = None
335
self._filename = path
336
self._lock_token = None
337
self._lock_state = None
338
self._id_index = None
339
self._end_of_header = None
340
self._cutoff_time = None
341
self._split_path_cache = {}
342
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
346
(self.__class__.__name__, self._filename)
348
def add(self, path, file_id, kind, stat, fingerprint):
349
"""Add a path to be tracked.
351
:param path: The path within the dirstate - '' is the root, 'foo' is the
352
path foo within the root, 'foo/bar' is the path bar within foo
354
:param file_id: The file id of the path being added.
355
:param kind: The kind of the path, as a string like 'file',
357
:param stat: The output of os.lstat for the path.
358
:param fingerprint: The sha value of the file,
359
or the target of a symlink,
360
or the referenced revision id for tree-references,
361
or '' for directories.
364
# find the block its in.
365
# find the location in the block.
366
# check its not there
368
#------- copied from inventory.ensure_normalized_name - keep synced.
369
# --- normalized_filename wants a unicode basename only, so get one.
370
dirname, basename = osutils.split(path)
371
# we dont import normalized_filename directly because we want to be
372
# able to change the implementation at runtime for tests.
373
norm_name, can_access = osutils.normalized_filename(basename)
374
if norm_name != basename:
378
raise errors.InvalidNormalization(path)
379
# you should never have files called . or ..; just add the directory
380
# in the parent, or according to the special treatment for the root
381
if basename == '.' or basename == '..':
382
raise errors.InvalidEntryName(path)
383
# now that we've normalised, we need the correct utf8 path and
384
# dirname and basename elements. This single encode and split should be
385
# faster than three separate encodes.
386
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
387
dirname, basename = osutils.split(utf8path)
388
assert file_id.__class__ == str, \
389
"must be a utf8 file_id not %s" % (type(file_id))
390
# Make sure the file_id does not exist in this tree
391
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
392
if file_id_entry != (None, None):
393
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
394
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
395
info = '%s:%s' % (kind, path)
396
raise errors.DuplicateFileId(file_id, info)
397
first_key = (dirname, basename, '')
398
block_index, present = self._find_block_index_from_key(first_key)
400
# check the path is not in the tree
401
block = self._dirblocks[block_index][1]
402
entry_index, _ = self._find_entry_index(first_key, block)
403
while (entry_index < len(block) and
404
block[entry_index][0][0:2] == first_key[0:2]):
405
if block[entry_index][1][0][0] not in 'ar':
406
# this path is in the dirstate in the current tree.
407
raise Exception, "adding already added path!"
410
# The block where we want to put the file is not present. But it
411
# might be because the directory was empty, or not loaded yet. Look
412
# for a parent entry, if not found, raise NotVersionedError
413
parent_dir, parent_base = osutils.split(dirname)
414
parent_block_idx, parent_entry_idx, _, parent_present = \
415
self._get_block_entry_index(parent_dir, parent_base, 0)
416
if not parent_present:
417
raise errors.NotVersionedError(path, str(self))
418
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
419
block = self._dirblocks[block_index][1]
420
entry_key = (dirname, basename, file_id)
423
packed_stat = DirState.NULLSTAT
426
packed_stat = pack_stat(stat)
427
parent_info = self._empty_parent_info()
428
minikind = DirState._kind_to_minikind[kind]
430
entry_data = entry_key, [
431
(minikind, fingerprint, size, False, packed_stat),
433
elif kind == 'directory':
434
entry_data = entry_key, [
435
(minikind, '', 0, False, packed_stat),
437
elif kind == 'symlink':
438
entry_data = entry_key, [
439
(minikind, fingerprint, size, False, packed_stat),
441
elif kind == 'tree-reference':
442
entry_data = entry_key, [
443
(minikind, fingerprint, 0, False, packed_stat),
446
raise errors.BzrError('unknown kind %r' % kind)
447
entry_index, present = self._find_entry_index(entry_key, block)
449
block.insert(entry_index, entry_data)
451
assert block[entry_index][1][0][0] == 'a', " %r(%r) already added" % (basename, file_id)
452
block[entry_index][1][0] = entry_data[1][0]
454
if kind == 'directory':
455
# insert a new dirblock
456
self._ensure_block(block_index, entry_index, utf8path)
457
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
459
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
461
def _bisect(self, paths):
462
"""Bisect through the disk structure for specific rows.
464
:param paths: A list of paths to find
465
:return: A dict mapping path => entries for found entries. Missing
466
entries will not be in the map.
467
The list is not sorted, and entries will be populated
468
based on when they were read.
470
self._requires_lock()
471
# We need the file pointer to be right after the initial header block
472
self._read_header_if_needed()
473
# If _dirblock_state was in memory, we should just return info from
474
# there, this function is only meant to handle when we want to read
476
assert self._dirblock_state == DirState.NOT_IN_MEMORY
478
# The disk representation is generally info + '\0\n\0' at the end. But
479
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
480
# Because it means we can sync on the '\n'
481
state_file = self._state_file
482
file_size = os.fstat(state_file.fileno()).st_size
483
# We end up with 2 extra fields, we should have a trailing '\n' to
484
# ensure that we read the whole record, and we should have a precursur
485
# '' which ensures that we start after the previous '\n'
486
entry_field_count = self._fields_per_entry() + 1
488
low = self._end_of_header
489
high = file_size - 1 # Ignore the final '\0'
490
# Map from (dir, name) => entry
493
# Avoid infinite seeking
494
max_count = 30*len(paths)
496
# pending is a list of places to look.
497
# each entry is a tuple of low, high, dir_names
498
# low -> the first byte offset to read (inclusive)
499
# high -> the last byte offset (inclusive)
500
# dir_names -> The list of (dir, name) pairs that should be found in
501
# the [low, high] range
502
pending = [(low, high, paths)]
504
page_size = self._bisect_page_size
506
fields_to_entry = self._get_fields_to_entry()
509
low, high, cur_files = pending.pop()
511
if not cur_files or low >= high:
516
if count > max_count:
517
raise errors.BzrError('Too many seeks, most likely a bug.')
519
mid = max(low, (low+high-page_size)/2)
522
# limit the read size, so we don't end up reading data that we have
524
read_size = min(page_size, (high-mid)+1)
525
block = state_file.read(read_size)
528
entries = block.split('\n')
531
# We didn't find a '\n', so we cannot have found any records.
532
# So put this range back and try again. But we know we have to
533
# increase the page size, because a single read did not contain
534
# a record break (so records must be larger than page_size)
536
pending.append((low, high, cur_files))
539
# Check the first and last entries, in case they are partial, or if
540
# we don't care about the rest of this page
542
first_fields = entries[0].split('\0')
543
if len(first_fields) < entry_field_count:
544
# We didn't get the complete first entry
545
# so move start, and grab the next, which
546
# should be a full entry
547
start += len(entries[0])+1
548
first_fields = entries[1].split('\0')
551
if len(first_fields) <= 2:
552
# We didn't even get a filename here... what do we do?
553
# Try a large page size and repeat this query
555
pending.append((low, high, cur_files))
558
# Find what entries we are looking for, which occur before and
559
# after this first record.
562
first_path = first_fields[1] + '/' + first_fields[2]
564
first_path = first_fields[2]
565
first_loc = _bisect_path_left(cur_files, first_path)
567
# These exist before the current location
568
pre = cur_files[:first_loc]
569
# These occur after the current location, which may be in the
570
# data we read, or might be after the last entry
571
post = cur_files[first_loc:]
573
if post and len(first_fields) >= entry_field_count:
574
# We have files after the first entry
576
# Parse the last entry
577
last_entry_num = len(entries)-1
578
last_fields = entries[last_entry_num].split('\0')
579
if len(last_fields) < entry_field_count:
580
# The very last hunk was not complete,
581
# read the previous hunk
582
after = mid + len(block) - len(entries[-1])
584
last_fields = entries[last_entry_num].split('\0')
586
after = mid + len(block)
589
last_path = last_fields[1] + '/' + last_fields[2]
591
last_path = last_fields[2]
592
last_loc = _bisect_path_right(post, last_path)
594
middle_files = post[:last_loc]
595
post = post[last_loc:]
598
# We have files that should occur in this block
599
# (>= first, <= last)
600
# Either we will find them here, or we can mark them as
603
if middle_files[0] == first_path:
604
# We might need to go before this location
605
pre.append(first_path)
606
if middle_files[-1] == last_path:
607
post.insert(0, last_path)
609
# Find out what paths we have
610
paths = {first_path:[first_fields]}
611
# last_path might == first_path so we need to be
612
# careful if we should append rather than overwrite
613
if last_entry_num != first_entry_num:
614
paths.setdefault(last_path, []).append(last_fields)
615
for num in xrange(first_entry_num+1, last_entry_num):
616
# TODO: jam 20070223 We are already splitting here, so
617
# shouldn't we just split the whole thing rather
618
# than doing the split again in add_one_record?
619
fields = entries[num].split('\0')
621
path = fields[1] + '/' + fields[2]
624
paths.setdefault(path, []).append(fields)
626
for path in middle_files:
627
for fields in paths.get(path, []):
628
# offset by 1 because of the opening '\0'
629
# consider changing fields_to_entry to avoid the
631
entry = fields_to_entry(fields[1:])
632
found.setdefault(path, []).append(entry)
634
# Now we have split up everything into pre, middle, and post, and
635
# we have handled everything that fell in 'middle'.
636
# We add 'post' first, so that we prefer to seek towards the
637
# beginning, so that we will tend to go as early as we need, and
638
# then only seek forward after that.
640
pending.append((after, high, post))
642
pending.append((low, start-1, pre))
644
# Consider that we may want to return the directory entries in sorted
645
# order. For now, we just return them in whatever order we found them,
646
# and leave it up to the caller if they care if it is ordered or not.
649
def _bisect_dirblocks(self, dir_list):
650
"""Bisect through the disk structure to find entries in given dirs.
652
_bisect_dirblocks is meant to find the contents of directories, which
653
differs from _bisect, which only finds individual entries.
655
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
656
:return: A map from dir => entries_for_dir
658
# TODO: jam 20070223 A lot of the bisecting logic could be shared
659
# between this and _bisect. It would require parameterizing the
660
# inner loop with a function, though. We should evaluate the
661
# performance difference.
662
self._requires_lock()
663
# We need the file pointer to be right after the initial header block
664
self._read_header_if_needed()
665
# If _dirblock_state was in memory, we should just return info from
666
# there, this function is only meant to handle when we want to read
668
assert self._dirblock_state == DirState.NOT_IN_MEMORY
670
# The disk representation is generally info + '\0\n\0' at the end. But
671
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
672
# Because it means we can sync on the '\n'
673
state_file = self._state_file
674
file_size = os.fstat(state_file.fileno()).st_size
675
# We end up with 2 extra fields, we should have a trailing '\n' to
676
# ensure that we read the whole record, and we should have a precursur
677
# '' which ensures that we start after the previous '\n'
678
entry_field_count = self._fields_per_entry() + 1
680
low = self._end_of_header
681
high = file_size - 1 # Ignore the final '\0'
682
# Map from dir => entry
685
# Avoid infinite seeking
686
max_count = 30*len(dir_list)
688
# pending is a list of places to look.
689
# each entry is a tuple of low, high, dir_names
690
# low -> the first byte offset to read (inclusive)
691
# high -> the last byte offset (inclusive)
692
# dirs -> The list of directories that should be found in
693
# the [low, high] range
694
pending = [(low, high, dir_list)]
696
page_size = self._bisect_page_size
698
fields_to_entry = self._get_fields_to_entry()
701
low, high, cur_dirs = pending.pop()
703
if not cur_dirs or low >= high:
708
if count > max_count:
709
raise errors.BzrError('Too many seeks, most likely a bug.')
711
mid = max(low, (low+high-page_size)/2)
714
# limit the read size, so we don't end up reading data that we have
716
read_size = min(page_size, (high-mid)+1)
717
block = state_file.read(read_size)
720
entries = block.split('\n')
723
# We didn't find a '\n', so we cannot have found any records.
724
# So put this range back and try again. But we know we have to
725
# increase the page size, because a single read did not contain
726
# a record break (so records must be larger than page_size)
728
pending.append((low, high, cur_dirs))
731
# Check the first and last entries, in case they are partial, or if
732
# we don't care about the rest of this page
734
first_fields = entries[0].split('\0')
735
if len(first_fields) < entry_field_count:
736
# We didn't get the complete first entry
737
# so move start, and grab the next, which
738
# should be a full entry
739
start += len(entries[0])+1
740
first_fields = entries[1].split('\0')
743
if len(first_fields) <= 1:
744
# We didn't even get a dirname here... what do we do?
745
# Try a large page size and repeat this query
747
pending.append((low, high, cur_dirs))
750
# Find what entries we are looking for, which occur before and
751
# after this first record.
753
first_dir = first_fields[1]
754
first_loc = bisect.bisect_left(cur_dirs, first_dir)
756
# These exist before the current location
757
pre = cur_dirs[:first_loc]
758
# These occur after the current location, which may be in the
759
# data we read, or might be after the last entry
760
post = cur_dirs[first_loc:]
762
if post and len(first_fields) >= entry_field_count:
763
# We have records to look at after the first entry
765
# Parse the last entry
766
last_entry_num = len(entries)-1
767
last_fields = entries[last_entry_num].split('\0')
768
if len(last_fields) < entry_field_count:
769
# The very last hunk was not complete,
770
# read the previous hunk
771
after = mid + len(block) - len(entries[-1])
773
last_fields = entries[last_entry_num].split('\0')
775
after = mid + len(block)
777
last_dir = last_fields[1]
778
last_loc = bisect.bisect_right(post, last_dir)
780
middle_files = post[:last_loc]
781
post = post[last_loc:]
784
# We have files that should occur in this block
785
# (>= first, <= last)
786
# Either we will find them here, or we can mark them as
789
if middle_files[0] == first_dir:
790
# We might need to go before this location
791
pre.append(first_dir)
792
if middle_files[-1] == last_dir:
793
post.insert(0, last_dir)
795
# Find out what paths we have
796
paths = {first_dir:[first_fields]}
797
# last_dir might == first_dir so we need to be
798
# careful if we should append rather than overwrite
799
if last_entry_num != first_entry_num:
800
paths.setdefault(last_dir, []).append(last_fields)
801
for num in xrange(first_entry_num+1, last_entry_num):
802
# TODO: jam 20070223 We are already splitting here, so
803
# shouldn't we just split the whole thing rather
804
# than doing the split again in add_one_record?
805
fields = entries[num].split('\0')
806
paths.setdefault(fields[1], []).append(fields)
808
for cur_dir in middle_files:
809
for fields in paths.get(cur_dir, []):
810
# offset by 1 because of the opening '\0'
811
# consider changing fields_to_entry to avoid the
813
entry = fields_to_entry(fields[1:])
814
found.setdefault(cur_dir, []).append(entry)
816
# Now we have split up everything into pre, middle, and post, and
817
# we have handled everything that fell in 'middle'.
818
# We add 'post' first, so that we prefer to seek towards the
819
# beginning, so that we will tend to go as early as we need, and
820
# then only seek forward after that.
822
pending.append((after, high, post))
824
pending.append((low, start-1, pre))
828
def _bisect_recursive(self, paths):
829
"""Bisect for entries for all paths and their children.
831
This will use bisect to find all records for the supplied paths. It
832
will then continue to bisect for any records which are marked as
833
directories. (and renames?)
835
:param paths: A sorted list of (dir, name) pairs
836
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
837
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
839
# Map from (dir, name, file_id) => [tree_info]
842
found_dir_names = set()
844
# Directories that have been read
845
processed_dirs = set()
846
# Get the ball rolling with the first bisect for all entries.
847
newly_found = self._bisect(paths)
850
# Directories that need to be read
852
paths_to_search = set()
853
for entry_list in newly_found.itervalues():
854
for dir_name_id, trees_info in entry_list:
855
found[dir_name_id] = trees_info
856
found_dir_names.add(dir_name_id[:2])
858
for tree_info in trees_info:
859
minikind = tree_info[0]
862
# We already processed this one as a directory,
863
# we don't need to do the extra work again.
865
subdir, name, file_id = dir_name_id
866
path = osutils.pathjoin(subdir, name)
868
if path not in processed_dirs:
869
pending_dirs.add(path)
870
elif minikind == 'r':
871
# Rename, we need to directly search the target
872
# which is contained in the fingerprint column
873
dir_name = osutils.split(tree_info[1])
874
if dir_name[0] in pending_dirs:
875
# This entry will be found in the dir search
877
if dir_name not in found_dir_names:
878
paths_to_search.add(tree_info[1])
879
# Now we have a list of paths to look for directly, and
880
# directory blocks that need to be read.
881
# newly_found is mixing the keys between (dir, name) and path
882
# entries, but that is okay, because we only really care about the
884
newly_found = self._bisect(sorted(paths_to_search))
885
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
886
processed_dirs.update(pending_dirs)
889
def _empty_parent_info(self):
890
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
893
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
894
"""Ensure a block for dirname exists.
896
This function exists to let callers which know that there is a
897
directory dirname ensure that the block for it exists. This block can
898
fail to exist because of demand loading, or because a directory had no
899
children. In either case it is not an error. It is however an error to
900
call this if there is no parent entry for the directory, and thus the
901
function requires the coordinates of such an entry to be provided.
903
The root row is special cased and can be indicated with a parent block
906
:param parent_block_index: The index of the block in which dirname's row
908
:param parent_row_index: The index in the parent block where the row
910
:param dirname: The utf8 dirname to ensure there is a block for.
911
:return: The index for the block.
913
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
914
# This is the signature of the root row, and the
915
# contents-of-root row is always index 1
917
# the basename of the directory must be the end of its full name.
918
if not (parent_block_index == -1 and
919
parent_block_index == -1 and dirname == ''):
920
assert dirname.endswith(
921
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
922
block_index, present = self._find_block_index_from_key((dirname, '', ''))
924
## In future, when doing partial parsing, this should load and
925
# populate the entire block.
926
self._dirblocks.insert(block_index, (dirname, []))
929
def _entries_to_current_state(self, new_entries):
930
"""Load new_entries into self.dirblocks.
932
Process new_entries into the current state object, making them the active
933
state. The entries are grouped together by directory to form dirblocks.
935
:param new_entries: A sorted list of entries. This function does not sort
936
to prevent unneeded overhead when callers have a sorted list already.
939
assert new_entries[0][0][0:2] == ('', ''), \
940
"Missing root row %r" % (new_entries[0][0],)
941
# The two blocks here are deliberate: the root block and the
942
# contents-of-root block.
943
self._dirblocks = [('', []), ('', [])]
944
current_block = self._dirblocks[0][1]
947
append_entry = current_block.append
948
for entry in new_entries:
949
if entry[0][0] != current_dirname:
950
# new block - different dirname
952
current_dirname = entry[0][0]
953
self._dirblocks.append((current_dirname, current_block))
954
append_entry = current_block.append
955
# append the entry to the current block
957
self._split_root_dirblock_into_contents()
959
def _split_root_dirblock_into_contents(self):
960
"""Split the root dirblocks into root and contents-of-root.
962
After parsing by path, we end up with root entries and contents-of-root
963
entries in the same block. This loop splits them out again.
965
# The above loop leaves the "root block" entries mixed with the
966
# "contents-of-root block". But we don't want an if check on
967
# all entries, so instead we just fix it up here.
968
assert self._dirblocks[1] == ('', [])
970
contents_of_root_block = []
971
for entry in self._dirblocks[0][1]:
972
if not entry[0][1]: # This is a root entry
973
root_block.append(entry)
975
contents_of_root_block.append(entry)
976
self._dirblocks[0] = ('', root_block)
977
self._dirblocks[1] = ('', contents_of_root_block)
979
def _entry_to_line(self, entry):
980
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
982
:param entry: An entry_tuple as defined in the module docstring.
984
entire_entry = list(entry[0])
985
for tree_number, tree_data in enumerate(entry[1]):
986
# (minikind, fingerprint, size, executable, tree_specific_string)
987
entire_entry.extend(tree_data)
988
# 3 for the key, 5 for the fields per tree.
989
tree_offset = 3 + tree_number * 5
991
entire_entry[tree_offset + 0] = tree_data[0]
993
entire_entry[tree_offset + 2] = str(tree_data[2])
995
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
996
return '\0'.join(entire_entry)
998
def _fields_per_entry(self):
999
"""How many null separated fields should be in each entry row.
1001
Each line now has an extra '\n' field which is not used
1002
so we just skip over it
1004
3 fields for the key
1005
+ number of fields per tree_data (5) * tree count
1008
tree_count = 1 + self._num_present_parents()
1009
return 3 + 5 * tree_count + 1
1011
def _find_block(self, key, add_if_missing=False):
1012
"""Return the block that key should be present in.
1014
:param key: A dirstate entry key.
1015
:return: The block tuple.
1017
block_index, present = self._find_block_index_from_key(key)
1019
if not add_if_missing:
1020
# check to see if key is versioned itself - we might want to
1021
# add it anyway, because dirs with no entries dont get a
1022
# dirblock at parse time.
1023
# This is an uncommon branch to take: most dirs have children,
1024
# and most code works with versioned paths.
1025
parent_base, parent_name = osutils.split(key[0])
1026
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1027
# some parent path has not been added - its an error to add
1029
raise errors.NotVersionedError(key[0:2], str(self))
1030
self._dirblocks.insert(block_index, (key[0], []))
1031
return self._dirblocks[block_index]
1033
def _find_block_index_from_key(self, key):
1034
"""Find the dirblock index for a key.
1036
:return: The block index, True if the block for the key is present.
1038
if key[0:2] == ('', ''):
1040
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1041
cache=self._split_path_cache)
1042
# _right returns one-past-where-key is so we have to subtract
1043
# one to use it. we use _right here because there are two
1044
# '' blocks - the root, and the contents of root
1045
# we always have a minimum of 2 in self._dirblocks: root and
1046
# root-contents, and for '', we get 2 back, so this is
1047
# simple and correct:
1048
present = (block_index < len(self._dirblocks) and
1049
self._dirblocks[block_index][0] == key[0])
1050
return block_index, present
1052
def _find_entry_index(self, key, block):
1053
"""Find the entry index for a key in a block.
1055
:return: The entry index, True if the entry for the key is present.
1057
entry_index = bisect.bisect_left(block, (key, []))
1058
present = (entry_index < len(block) and
1059
block[entry_index][0] == key)
1060
return entry_index, present
1063
def from_tree(tree, dir_state_filename):
1064
"""Create a dirstate from a bzr Tree.
1066
:param tree: The tree which should provide parent information and
1068
:return: a DirState object which is currently locked for writing.
1069
(it was locked by DirState.initialize)
1071
result = DirState.initialize(dir_state_filename)
1075
parent_ids = tree.get_parent_ids()
1076
num_parents = len(parent_ids)
1078
for parent_id in parent_ids:
1079
parent_tree = tree.branch.repository.revision_tree(parent_id)
1080
parent_trees.append((parent_id, parent_tree))
1081
parent_tree.lock_read()
1082
result.set_parent_trees(parent_trees, [])
1083
result.set_state_from_inventory(tree.inventory)
1085
for revid, parent_tree in parent_trees:
1086
parent_tree.unlock()
1089
# The caller won't have a chance to unlock this, so make sure we
1095
def update_entry(self, entry, abspath, stat_value,
1096
_stat_to_minikind=_stat_to_minikind,
1097
_pack_stat=pack_stat):
1098
"""Update the entry based on what is actually on disk.
1100
:param entry: This is the dirblock entry for the file in question.
1101
:param abspath: The path on disk for this file.
1102
:param stat_value: (optional) if we already have done a stat on the
1104
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1108
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1112
packed_stat = _pack_stat(stat_value)
1113
(saved_minikind, saved_link_or_sha1, saved_file_size,
1114
saved_executable, saved_packed_stat) = entry[1][0]
1116
if (minikind == saved_minikind
1117
and packed_stat == saved_packed_stat):
1118
# The stat hasn't changed since we saved, so we can re-use the
1123
# size should also be in packed_stat
1124
if saved_file_size == stat_value.st_size:
1125
return saved_link_or_sha1
1127
# If we have gotten this far, that means that we need to actually
1128
# process this entry.
1131
link_or_sha1 = self._sha1_file(abspath, entry)
1132
executable = self._is_executable(stat_value.st_mode,
1134
if self._cutoff_time is None:
1135
self._sha_cutoff_time()
1136
if (stat_value.st_mtime < self._cutoff_time
1137
and stat_value.st_ctime < self._cutoff_time):
1138
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1139
executable, packed_stat)
1141
entry[1][0] = ('f', '', stat_value.st_size,
1142
executable, DirState.NULLSTAT)
1143
elif minikind == 'd':
1145
entry[1][0] = ('d', '', 0, False, packed_stat)
1146
if saved_minikind != 'd':
1147
# This changed from something into a directory. Make sure we
1148
# have a directory block for it. This doesn't happen very
1149
# often, so this doesn't have to be super fast.
1150
block_index, entry_index, dir_present, file_present = \
1151
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1152
self._ensure_block(block_index, entry_index,
1153
osutils.pathjoin(entry[0][0], entry[0][1]))
1154
elif minikind == 'l':
1155
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1156
if self._cutoff_time is None:
1157
self._sha_cutoff_time()
1158
if (stat_value.st_mtime < self._cutoff_time
1159
and stat_value.st_ctime < self._cutoff_time):
1160
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1163
entry[1][0] = ('l', '', stat_value.st_size,
1164
False, DirState.NULLSTAT)
1165
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1168
def _sha_cutoff_time(self):
1169
"""Return cutoff time.
1171
Files modified more recently than this time are at risk of being
1172
undetectably modified and so can't be cached.
1174
# Cache the cutoff time as long as we hold a lock.
1175
# time.time() isn't super expensive (approx 3.38us), but
1176
# when you call it 50,000 times it adds up.
1177
# For comparison, os.lstat() costs 7.2us if it is hot.
1178
self._cutoff_time = int(time.time()) - 3
1179
return self._cutoff_time
1181
def _lstat(self, abspath, entry):
1182
"""Return the os.lstat value for this path."""
1183
return os.lstat(abspath)
1185
def _sha1_file(self, abspath, entry):
1186
"""Calculate the SHA1 of a file by reading the full text"""
1187
f = file(abspath, 'rb', buffering=65000)
1189
return osutils.sha_file(f)
1193
def _is_executable(self, mode, old_executable):
1194
"""Is this file executable?"""
1195
return bool(S_IEXEC & mode)
1197
def _is_executable_win32(self, mode, old_executable):
1198
"""On win32 the executable bit is stored in the dirstate."""
1199
return old_executable
1201
if sys.platform == 'win32':
1202
_is_executable = _is_executable_win32
1204
def _read_link(self, abspath, old_link):
1205
"""Read the target of a symlink"""
1206
# TODO: jam 200700301 On Win32, this could just return the value
1207
# already in memory. However, this really needs to be done at a
1208
# higher level, because there either won't be anything on disk,
1209
# or the thing on disk will be a file.
1210
return os.readlink(abspath)
1212
def get_ghosts(self):
1213
"""Return a list of the parent tree revision ids that are ghosts."""
1214
self._read_header_if_needed()
1217
def get_lines(self):
1218
"""Serialise the entire dirstate to a sequence of lines."""
1219
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1220
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1221
# read whats on disk.
1222
self._state_file.seek(0)
1223
return self._state_file.readlines()
1225
lines.append(self._get_parents_line(self.get_parent_ids()))
1226
lines.append(self._get_ghosts_line(self._ghosts))
1227
# append the root line which is special cased
1228
lines.extend(map(self._entry_to_line, self._iter_entries()))
1229
return self._get_output_lines(lines)
1231
def _get_ghosts_line(self, ghost_ids):
1232
"""Create a line for the state file for ghost information."""
1233
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1235
def _get_parents_line(self, parent_ids):
1236
"""Create a line for the state file for parents information."""
1237
return '\0'.join([str(len(parent_ids))] + parent_ids)
1239
def _get_fields_to_entry(self):
1240
"""Get a function which converts entry fields into a entry record.
1242
This handles size and executable, as well as parent records.
1244
:return: A function which takes a list of fields, and returns an
1245
appropriate record for storing in memory.
1247
# This is intentionally unrolled for performance
1248
num_present_parents = self._num_present_parents()
1249
if num_present_parents == 0:
1250
def fields_to_entry_0_parents(fields, _int=int):
1251
path_name_file_id_key = (fields[0], fields[1], fields[2])
1252
return (path_name_file_id_key, [
1254
fields[3], # minikind
1255
fields[4], # fingerprint
1256
_int(fields[5]), # size
1257
fields[6] == 'y', # executable
1258
fields[7], # packed_stat or revision_id
1260
return fields_to_entry_0_parents
1261
elif num_present_parents == 1:
1262
def fields_to_entry_1_parent(fields, _int=int):
1263
path_name_file_id_key = (fields[0], fields[1], fields[2])
1264
return (path_name_file_id_key, [
1266
fields[3], # minikind
1267
fields[4], # fingerprint
1268
_int(fields[5]), # size
1269
fields[6] == 'y', # executable
1270
fields[7], # packed_stat or revision_id
1273
fields[8], # minikind
1274
fields[9], # fingerprint
1275
_int(fields[10]), # size
1276
fields[11] == 'y', # executable
1277
fields[12], # packed_stat or revision_id
1280
return fields_to_entry_1_parent
1281
elif num_present_parents == 2:
1282
def fields_to_entry_2_parents(fields, _int=int):
1283
path_name_file_id_key = (fields[0], fields[1], fields[2])
1284
return (path_name_file_id_key, [
1286
fields[3], # minikind
1287
fields[4], # fingerprint
1288
_int(fields[5]), # size
1289
fields[6] == 'y', # executable
1290
fields[7], # packed_stat or revision_id
1293
fields[8], # minikind
1294
fields[9], # fingerprint
1295
_int(fields[10]), # size
1296
fields[11] == 'y', # executable
1297
fields[12], # packed_stat or revision_id
1300
fields[13], # minikind
1301
fields[14], # fingerprint
1302
_int(fields[15]), # size
1303
fields[16] == 'y', # executable
1304
fields[17], # packed_stat or revision_id
1307
return fields_to_entry_2_parents
1309
def fields_to_entry_n_parents(fields, _int=int):
1310
path_name_file_id_key = (fields[0], fields[1], fields[2])
1311
trees = [(fields[cur], # minikind
1312
fields[cur+1], # fingerprint
1313
_int(fields[cur+2]), # size
1314
fields[cur+3] == 'y', # executable
1315
fields[cur+4], # stat or revision_id
1316
) for cur in xrange(3, len(fields)-1, 5)]
1317
return path_name_file_id_key, trees
1318
return fields_to_entry_n_parents
1320
def get_parent_ids(self):
1321
"""Return a list of the parent tree ids for the directory state."""
1322
self._read_header_if_needed()
1323
return list(self._parents)
1325
def _get_block_entry_index(self, dirname, basename, tree_index):
1326
"""Get the coordinates for a path in the state structure.
1328
:param dirname: The utf8 dirname to lookup.
1329
:param basename: The utf8 basename to lookup.
1330
:param tree_index: The index of the tree for which this lookup should
1332
:return: A tuple describing where the path is located, or should be
1333
inserted. The tuple contains four fields: the block index, the row
1334
index, the directory is present (boolean), the entire path is
1335
present (boolean). There is no guarantee that either
1336
coordinate is currently reachable unless the found field for it is
1337
True. For instance, a directory not present in the searched tree
1338
may be returned with a value one greater than the current highest
1339
block offset. The directory present field will always be True when
1340
the path present field is True. The directory present field does
1341
NOT indicate that the directory is present in the searched tree,
1342
rather it indicates that there are at least some files in some
1345
self._read_dirblocks_if_needed()
1346
key = dirname, basename, ''
1347
block_index, present = self._find_block_index_from_key(key)
1349
# no such directory - return the dir index and 0 for the row.
1350
return block_index, 0, False, False
1351
block = self._dirblocks[block_index][1] # access the entries only
1352
entry_index, present = self._find_entry_index(key, block)
1353
# linear search through present entries at this path to find the one
1355
while entry_index < len(block) and block[entry_index][0][1] == basename:
1356
if block[entry_index][1][tree_index][0] not in \
1357
('a', 'r'): # absent, relocated
1358
return block_index, entry_index, True, True
1360
return block_index, entry_index, True, False
1362
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1363
"""Get the dirstate entry for path in tree tree_index.
1365
If either file_id or path is supplied, it is used as the key to lookup.
1366
If both are supplied, the fastest lookup is used, and an error is
1367
raised if they do not both point at the same row.
1369
:param tree_index: The index of the tree we wish to locate this path
1370
in. If the path is present in that tree, the entry containing its
1371
details is returned, otherwise (None, None) is returned
1372
0 is the working tree, higher indexes are successive parent
1374
:param fileid_utf8: A utf8 file_id to look up.
1375
:param path_utf8: An utf8 path to be looked up.
1376
:return: The dirstate entry tuple for path, or (None, None)
1378
self._read_dirblocks_if_needed()
1379
if path_utf8 is not None:
1380
assert path_utf8.__class__ == str, 'path_utf8 is not a str: %s %s' % (type(path_utf8), path_utf8)
1381
# path lookups are faster
1382
dirname, basename = osutils.split(path_utf8)
1383
block_index, entry_index, dir_present, file_present = \
1384
self._get_block_entry_index(dirname, basename, tree_index)
1385
if not file_present:
1387
entry = self._dirblocks[block_index][1][entry_index]
1388
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1390
if entry[0][2] != fileid_utf8:
1391
raise errors.BzrError('integrity error ? : mismatching'
1392
' tree_index, file_id and path')
1395
assert fileid_utf8 is not None
1396
possible_keys = self._get_id_index().get(fileid_utf8, None)
1397
if not possible_keys:
1399
for key in possible_keys:
1400
block_index, present = \
1401
self._find_block_index_from_key(key)
1402
# strange, probably indicates an out of date
1403
# id index - for now, allow this.
1406
# WARNING: DO not change this code to use _get_block_entry_index
1407
# as that function is not suitable: it does not use the key
1408
# to lookup, and thus the wrong coordinates are returned.
1409
block = self._dirblocks[block_index][1]
1410
entry_index, present = self._find_entry_index(key, block)
1412
entry = self._dirblocks[block_index][1][entry_index]
1413
if entry[1][tree_index][0] in 'fdlt':
1414
# this is the result we are looking for: the
1415
# real home of this file_id in this tree.
1417
if entry[1][tree_index][0] == 'a':
1418
# there is no home for this entry in this tree
1420
assert entry[1][tree_index][0] == 'r', \
1421
"entry %r has invalid minikind %r for tree %r" \
1423
entry[1][tree_index][0],
1425
real_path = entry[1][tree_index][1]
1426
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1427
path_utf8=real_path)
1431
def initialize(cls, path):
1432
"""Create a new dirstate on path.
1434
The new dirstate will be an empty tree - that is it has no parents,
1435
and only a root node - which has id ROOT_ID.
1437
:param path: The name of the file for the dirstate.
1438
:return: A write-locked DirState object.
1440
# This constructs a new DirState object on a path, sets the _state_file
1441
# to a new empty file for that path. It then calls _set_data() with our
1442
# stock empty dirstate information - a root with ROOT_ID, no children,
1443
# and no parents. Finally it calls save() to ensure that this data will
1446
# root dir and root dir contents with no children.
1447
empty_tree_dirblocks = [('', []), ('', [])]
1448
# a new root directory, with a NULLSTAT.
1449
empty_tree_dirblocks[0][1].append(
1450
(('', '', inventory.ROOT_ID), [
1451
('d', '', 0, False, DirState.NULLSTAT),
1455
result._set_data([], empty_tree_dirblocks)
1462
def _inv_entry_to_details(self, inv_entry):
1463
"""Convert an inventory entry (from a revision tree) to state details.
1465
:param inv_entry: An inventory entry whose sha1 and link targets can be
1466
relied upon, and which has a revision set.
1467
:return: A details tuple - the details for a single tree at a path +
1470
kind = inv_entry.kind
1471
minikind = DirState._kind_to_minikind[kind]
1472
tree_data = inv_entry.revision
1473
assert tree_data, 'empty revision for the inv_entry %s.' % \
1475
if kind == 'directory':
1479
elif kind == 'symlink':
1480
fingerprint = inv_entry.symlink_target or ''
1483
elif kind == 'file':
1484
fingerprint = inv_entry.text_sha1 or ''
1485
size = inv_entry.text_size or 0
1486
executable = inv_entry.executable
1487
elif kind == 'tree-reference':
1488
fingerprint = inv_entry.reference_revision or ''
1492
raise Exception("can't pack %s" % inv_entry)
1493
return (minikind, fingerprint, size, executable, tree_data)
1495
def _iter_entries(self):
1496
"""Iterate over all the entries in the dirstate.
1498
Each yelt item is an entry in the standard format described in the
1499
docstring of bzrlib.dirstate.
1501
self._read_dirblocks_if_needed()
1502
for directory in self._dirblocks:
1503
for entry in directory[1]:
1506
def _get_id_index(self):
1507
"""Get an id index of self._dirblocks."""
1508
if self._id_index is None:
1510
for key, tree_details in self._iter_entries():
1511
id_index.setdefault(key[2], set()).add(key)
1512
self._id_index = id_index
1513
return self._id_index
1515
def _get_output_lines(self, lines):
1516
"""Format lines for final output.
1518
:param lines: A sequence of lines containing the parents list and the
1521
output_lines = [DirState.HEADER_FORMAT_3]
1522
lines.append('') # a final newline
1523
inventory_text = '\0\n\0'.join(lines)
1524
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1525
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1526
num_entries = len(lines)-3
1527
output_lines.append('num_entries: %s\n' % (num_entries,))
1528
output_lines.append(inventory_text)
1531
def _make_deleted_row(self, fileid_utf8, parents):
1532
"""Return a deleted row for fileid_utf8."""
1533
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1536
def _num_present_parents(self):
1537
"""The number of parent entries in each record row."""
1538
return len(self._parents) - len(self._ghosts)
1542
"""Construct a DirState on the file at path path.
1544
:return: An unlocked DirState object, associated with the given path.
1546
result = DirState(path)
1549
def _read_dirblocks_if_needed(self):
1550
"""Read in all the dirblocks from the file if they are not in memory.
1552
This populates self._dirblocks, and sets self._dirblock_state to
1553
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1556
self._read_header_if_needed()
1557
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1558
_read_dirblocks(self)
1560
def _read_header(self):
1561
"""This reads in the metadata header, and the parent ids.
1563
After reading in, the file should be positioned at the null
1564
just before the start of the first record in the file.
1566
:return: (expected crc checksum, number of entries, parent list)
1568
self._read_prelude()
1569
parent_line = self._state_file.readline()
1570
info = parent_line.split('\0')
1571
num_parents = int(info[0])
1572
assert num_parents == len(info)-2, 'incorrect parent info line'
1573
self._parents = info[1:-1]
1575
ghost_line = self._state_file.readline()
1576
info = ghost_line.split('\0')
1577
num_ghosts = int(info[1])
1578
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1579
self._ghosts = info[2:-1]
1580
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1581
self._end_of_header = self._state_file.tell()
1583
def _read_header_if_needed(self):
1584
"""Read the header of the dirstate file if needed."""
1585
# inline this as it will be called a lot
1586
if not self._lock_token:
1587
raise errors.ObjectNotLocked(self)
1588
if self._header_state == DirState.NOT_IN_MEMORY:
1591
def _read_prelude(self):
1592
"""Read in the prelude header of the dirstate file.
1594
This only reads in the stuff that is not connected to the crc
1595
checksum. The position will be correct to read in the rest of
1596
the file and check the checksum after this point.
1597
The next entry in the file should be the number of parents,
1598
and their ids. Followed by a newline.
1600
header = self._state_file.readline()
1601
assert header == DirState.HEADER_FORMAT_3, \
1602
'invalid header line: %r' % (header,)
1603
crc_line = self._state_file.readline()
1604
assert crc_line.startswith('crc32: '), 'missing crc32 checksum'
1605
self.crc_expected = int(crc_line[len('crc32: '):-1])
1606
num_entries_line = self._state_file.readline()
1607
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1608
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1611
"""Save any pending changes created during this session.
1613
We reuse the existing file, because that prevents race conditions with
1614
file creation, and use oslocks on it to prevent concurrent modification
1615
and reads - because dirstate's incremental data aggregation is not
1616
compatible with reading a modified file, and replacing a file in use by
1617
another process is impossible on Windows.
1619
A dirstate in read only mode should be smart enough though to validate
1620
that the file has not changed, and otherwise discard its cache and
1621
start over, to allow for fine grained read lock duration, so 'status'
1622
wont block 'commit' - for example.
1624
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1625
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1627
grabbed_write_lock = False
1628
if self._lock_state != 'w':
1629
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
1630
# Switch over to the new lock, as the old one may be closed.
1631
# TODO: jam 20070315 We should validate the disk file has
1632
# not changed contents. Since temporary_write_lock may
1633
# not be an atomic operation.
1634
self._lock_token = new_lock
1635
self._state_file = new_lock.f
1636
if not grabbed_write_lock:
1637
# We couldn't grab a write lock, so we switch back to a read one
1640
self._state_file.seek(0)
1641
self._state_file.writelines(self.get_lines())
1642
self._state_file.truncate()
1643
self._state_file.flush()
1644
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1645
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1647
if grabbed_write_lock:
1648
self._lock_token = self._lock_token.restore_read_lock()
1649
self._state_file = self._lock_token.f
1650
# TODO: jam 20070315 We should validate the disk file has
1651
# not changed contents. Since restore_read_lock may
1652
# not be an atomic operation.
1654
def _set_data(self, parent_ids, dirblocks):
1655
"""Set the full dirstate data in memory.
1657
This is an internal function used to completely replace the objects
1658
in memory state. It puts the dirstate into state 'full-dirty'.
1660
:param parent_ids: A list of parent tree revision ids.
1661
:param dirblocks: A list containing one tuple for each directory in the
1662
tree. Each tuple contains the directory path and a list of entries
1663
found in that directory.
1665
# our memory copy is now authoritative.
1666
self._dirblocks = dirblocks
1667
self._header_state = DirState.IN_MEMORY_MODIFIED
1668
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1669
self._parents = list(parent_ids)
1670
self._id_index = None
1672
def set_path_id(self, path, new_id):
1673
"""Change the id of path to new_id in the current working tree.
1675
:param path: The path inside the tree to set - '' is the root, 'foo'
1676
is the path foo in the root.
1677
:param new_id: The new id to assign to the path. This must be a utf8
1678
file id (not unicode, and not None).
1680
assert new_id.__class__ == str, \
1681
"path_id %r is not a plain string" % (new_id,)
1682
self._read_dirblocks_if_needed()
1684
# TODO: logic not written
1685
raise NotImplementedError(self.set_path_id)
1686
# TODO: check new id is unique
1687
entry = self._get_entry(0, path_utf8=path)
1688
if entry[0][2] == new_id:
1689
# Nothing to change.
1691
# mark the old path absent, and insert a new root path
1692
self._make_absent(entry)
1693
self.update_minimal(('', '', new_id), 'd',
1694
path_utf8='', packed_stat=entry[1][0][4])
1695
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1696
if self._id_index is not None:
1697
self._id_index.setdefault(new_id, set()).add(entry[0])
1699
def set_parent_trees(self, trees, ghosts):
1700
"""Set the parent trees for the dirstate.
1702
:param trees: A list of revision_id, tree tuples. tree must be provided
1703
even if the revision_id refers to a ghost: supply an empty tree in
1705
:param ghosts: A list of the revision_ids that are ghosts at the time
1708
# TODO: generate a list of parent indexes to preserve to save
1709
# processing specific parent trees. In the common case one tree will
1710
# be preserved - the left most parent.
1711
# TODO: if the parent tree is a dirstate, we might want to walk them
1712
# all by path in parallel for 'optimal' common-case performance.
1713
# generate new root row.
1714
self._read_dirblocks_if_needed()
1715
# TODO future sketch: Examine the existing parents to generate a change
1716
# map and then walk the new parent trees only, mapping them into the
1717
# dirstate. Walk the dirstate at the same time to remove unreferenced
1720
# sketch: loop over all entries in the dirstate, cherry picking
1721
# entries from the parent trees, if they are not ghost trees.
1722
# after we finish walking the dirstate, all entries not in the dirstate
1723
# are deletes, so we want to append them to the end as per the design
1724
# discussions. So do a set difference on ids with the parents to
1725
# get deletes, and add them to the end.
1726
# During the update process we need to answer the following questions:
1727
# - find other keys containing a fileid in order to create cross-path
1728
# links. We dont't trivially use the inventory from other trees
1729
# because this leads to either double touching, or to accessing
1731
# - find other keys containing a path
1732
# We accumulate each entry via this dictionary, including the root
1735
# we could do parallel iterators, but because file id data may be
1736
# scattered throughout, we dont save on index overhead: we have to look
1737
# at everything anyway. We can probably save cycles by reusing parent
1738
# data and doing an incremental update when adding an additional
1739
# parent, but for now the common cases are adding a new parent (merge),
1740
# and replacing completely (commit), and commit is more common: so
1741
# optimise merge later.
1743
# ---- start generation of full tree mapping data
1744
# what trees should we use?
1745
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
1746
# how many trees do we end up with
1747
parent_count = len(parent_trees)
1749
# one: the current tree
1750
for entry in self._iter_entries():
1751
# skip entries not in the current tree
1752
if entry[1][0][0] in ('a', 'r'): # absent, relocated
1754
by_path[entry[0]] = [entry[1][0]] + \
1755
[DirState.NULL_PARENT_DETAILS] * parent_count
1756
id_index[entry[0][2]] = set([entry[0]])
1758
# now the parent trees:
1759
for tree_index, tree in enumerate(parent_trees):
1760
# the index is off by one, adjust it.
1761
tree_index = tree_index + 1
1762
# when we add new locations for a fileid we need these ranges for
1763
# any fileid in this tree as we set the by_path[id] to:
1764
# already_processed_tree_details + new_details + new_location_suffix
1765
# the suffix is from tree_index+1:parent_count+1.
1766
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
1767
# now stitch in all the entries from this tree
1768
for path, entry in tree.inventory.iter_entries_by_dir():
1769
# here we process each trees details for each item in the tree.
1770
# we first update any existing entries for the id at other paths,
1771
# then we either create or update the entry for the id at the
1772
# right path, and finally we add (if needed) a mapping from
1773
# file_id to this path. We do it in this order to allow us to
1774
# avoid checking all known paths for the id when generating a
1775
# new entry at this path: by adding the id->path mapping last,
1776
# all the mappings are valid and have correct relocation
1777
# records where needed.
1778
file_id = entry.file_id
1779
path_utf8 = path.encode('utf8')
1780
dirname, basename = osutils.split(path_utf8)
1781
new_entry_key = (dirname, basename, file_id)
1782
# tree index consistency: All other paths for this id in this tree
1783
# index must point to the correct path.
1784
for entry_key in id_index.setdefault(file_id, set()):
1785
# TODO:PROFILING: It might be faster to just update
1786
# rather than checking if we need to, and then overwrite
1787
# the one we are located at.
1788
if entry_key != new_entry_key:
1789
# this file id is at a different path in one of the
1790
# other trees, so put absent pointers there
1791
# This is the vertical axis in the matrix, all pointing
1793
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
1794
# by path consistency: Insert into an existing path record (trivial), or
1795
# add a new one with relocation pointers for the other tree indexes.
1796
if new_entry_key in id_index[file_id]:
1797
# there is already an entry where this data belongs, just insert it.
1798
by_path[new_entry_key][tree_index] = \
1799
self._inv_entry_to_details(entry)
1801
# add relocated entries to the horizontal axis - this row
1802
# mapping from path,id. We need to look up the correct path
1803
# for the indexes from 0 to tree_index -1
1805
for lookup_index in xrange(tree_index):
1806
# boundary case: this is the first occurence of file_id
1807
# so there are no id_indexs, possibly take this out of
1809
if not len(id_index[file_id]):
1810
new_details.append(DirState.NULL_PARENT_DETAILS)
1812
# grab any one entry, use it to find the right path.
1813
# TODO: optimise this to reduce memory use in highly
1814
# fragmented situations by reusing the relocation
1816
a_key = iter(id_index[file_id]).next()
1817
if by_path[a_key][lookup_index][0] in ('r', 'a'):
1818
# its a pointer or missing statement, use it as is.
1819
new_details.append(by_path[a_key][lookup_index])
1821
# we have the right key, make a pointer to it.
1822
real_path = ('/'.join(a_key[0:2])).strip('/')
1823
new_details.append(('r', real_path, 0, False, ''))
1824
new_details.append(self._inv_entry_to_details(entry))
1825
new_details.extend(new_location_suffix)
1826
by_path[new_entry_key] = new_details
1827
id_index[file_id].add(new_entry_key)
1828
# --- end generation of full tree mappings
1830
# sort and output all the entries
1831
new_entries = self._sort_entries(by_path.items())
1832
self._entries_to_current_state(new_entries)
1833
self._parents = [rev_id for rev_id, tree in trees]
1834
self._ghosts = list(ghosts)
1835
self._header_state = DirState.IN_MEMORY_MODIFIED
1836
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1837
self._id_index = id_index
1839
def _sort_entries(self, entry_list):
1840
"""Given a list of entries, sort them into the right order.
1842
This is done when constructing a new dirstate from trees - normally we
1843
try to keep everything in sorted blocks all the time, but sometimes
1844
it's easier to sort after the fact.
1846
# TODO: Might be faster to do a schwartzian transform?
1848
# sort by: directory parts, file name, file id
1849
return entry[0][0].split('/'), entry[0][1], entry[0][2]
1850
return sorted(entry_list, key=_key)
1852
def set_state_from_inventory(self, new_inv):
1853
"""Set new_inv as the current state.
1855
This API is called by tree transform, and will usually occur with
1856
existing parent trees.
1858
:param new_inv: The inventory object to set current state from.
1860
self._read_dirblocks_if_needed()
1862
# incremental algorithm:
1863
# two iterators: current data and new data, both in dirblock order.
1864
new_iterator = new_inv.iter_entries_by_dir()
1865
# we will be modifying the dirstate, so we need a stable iterator. In
1866
# future we might write one, for now we just clone the state into a
1867
# list - which is a shallow copy.
1868
old_iterator = iter(list(self._iter_entries()))
1869
# both must have roots so this is safe:
1870
current_new = new_iterator.next()
1871
current_old = old_iterator.next()
1872
def advance(iterator):
1874
return iterator.next()
1875
except StopIteration:
1877
while current_new or current_old:
1878
# skip entries in old that are not really there
1879
if current_old and current_old[1][0][0] in ('r', 'a'):
1880
# relocated or absent
1881
current_old = advance(old_iterator)
1884
# convert new into dirblock style
1885
new_path_utf8 = current_new[0].encode('utf8')
1886
new_dirname, new_basename = osutils.split(new_path_utf8)
1887
new_id = current_new[1].file_id
1888
new_entry_key = (new_dirname, new_basename, new_id)
1889
current_new_minikind = \
1890
DirState._kind_to_minikind[current_new[1].kind]
1891
if current_new_minikind == 't':
1892
fingerprint = current_new[1].reference_revision or ''
1896
# for safety disable variables
1897
new_path_utf8 = new_dirname = new_basename = new_id = new_entry_key = None
1898
# 5 cases, we dont have a value that is strictly greater than everything, so
1899
# we make both end conditions explicit
1901
# old is finished: insert current_new into the state.
1902
self.update_minimal(new_entry_key, current_new_minikind,
1903
executable=current_new[1].executable,
1904
path_utf8=new_path_utf8, fingerprint=fingerprint)
1905
current_new = advance(new_iterator)
1906
elif not current_new:
1908
self._make_absent(current_old)
1909
current_old = advance(old_iterator)
1910
elif new_entry_key == current_old[0]:
1911
# same - common case
1912
# TODO: update the record if anything significant has changed.
1913
# the minimal required trigger is if the execute bit or cached
1915
if (current_old[1][0][3] != current_new[1].executable or
1916
current_old[1][0][0] != current_new_minikind):
1917
self.update_minimal(current_old[0], current_new_minikind,
1918
executable=current_new[1].executable,
1919
path_utf8=new_path_utf8, fingerprint=fingerprint)
1920
# both sides are dealt with, move on
1921
current_old = advance(old_iterator)
1922
current_new = advance(new_iterator)
1923
elif (new_entry_key[0].split('/') < current_old[0][0].split('/')
1924
and new_entry_key[1:] < current_old[0][1:]):
1926
# add a entry for this and advance new
1927
self.update_minimal(new_entry_key, current_new_minikind,
1928
executable=current_new[1].executable,
1929
path_utf8=new_path_utf8, fingerprint=fingerprint)
1930
current_new = advance(new_iterator)
1933
self._make_absent(current_old)
1934
current_old = advance(old_iterator)
1935
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1936
self._id_index = None
1938
def _make_absent(self, current_old):
1939
"""Mark current_old - an entry - as absent for tree 0.
1941
:return: True if this was the last details entry for the entry key:
1942
that is, if the underlying block has had the entry removed, thus
1943
shrinking in length.
1945
# build up paths that this id will be left at after the change is made,
1946
# so we can update their cross references in tree 0
1947
all_remaining_keys = set()
1948
# Dont check the working tree, because it's going.
1949
for details in current_old[1][1:]:
1950
if details[0] not in ('a', 'r'): # absent, relocated
1951
all_remaining_keys.add(current_old[0])
1952
elif details[0] == 'r': # relocated
1953
# record the key for the real path.
1954
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
1955
# absent rows are not present at any path.
1956
last_reference = current_old[0] not in all_remaining_keys
1958
# the current row consists entire of the current item (being marked
1959
# absent), and relocated or absent entries for the other trees:
1960
# Remove it, its meaningless.
1961
block = self._find_block(current_old[0])
1962
entry_index, present = self._find_entry_index(current_old[0], block[1])
1963
assert present, 'could not find entry for %s' % (current_old,)
1964
block[1].pop(entry_index)
1965
# if we have an id_index in use, remove this key from it for this id.
1966
if self._id_index is not None:
1967
self._id_index[current_old[0][2]].remove(current_old[0])
1968
# update all remaining keys for this id to record it as absent. The
1969
# existing details may either be the record we are making as deleted
1970
# (if there were other trees with the id present at this path), or may
1972
for update_key in all_remaining_keys:
1973
update_block_index, present = \
1974
self._find_block_index_from_key(update_key)
1975
assert present, 'could not find block for %s' % (update_key,)
1976
update_entry_index, present = \
1977
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
1978
assert present, 'could not find entry for %s' % (update_key,)
1979
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
1980
# it must not be absent at the moment
1981
assert update_tree_details[0][0] != 'a' # absent
1982
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
1983
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1984
return last_reference
1986
def update_minimal(self, key, minikind, executable=False, fingerprint='',
1987
packed_stat=None, size=0, path_utf8=None):
1988
"""Update an entry to the state in tree 0.
1990
This will either create a new entry at 'key' or update an existing one.
1991
It also makes sure that any other records which might mention this are
1994
:param key: (dir, name, file_id) for the new entry
1995
:param minikind: The type for the entry ('f' == 'file', 'd' ==
1997
:param executable: Should the executable bit be set?
1998
:param fingerprint: Simple fingerprint for new entry.
1999
:param packed_stat: packed stat value for new entry.
2000
:param size: Size information for new entry
2001
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2004
block = self._find_block(key)[1]
2005
if packed_stat is None:
2006
packed_stat = DirState.NULLSTAT
2007
entry_index, present = self._find_entry_index(key, block)
2008
new_details = (minikind, fingerprint, size, executable, packed_stat)
2009
id_index = self._get_id_index()
2011
# new entry, synthesis cross reference here,
2012
existing_keys = id_index.setdefault(key[2], set())
2013
if not existing_keys:
2014
# not currently in the state, simplest case
2015
new_entry = key, [new_details] + self._empty_parent_info()
2017
# present at one or more existing other paths.
2018
# grab one of them and use it to generate parent
2019
# relocation/absent entries.
2020
new_entry = key, [new_details]
2021
for other_key in existing_keys:
2022
# change the record at other to be a pointer to this new
2023
# record. The loop looks similar to the change to
2024
# relocations when updating an existing record but its not:
2025
# the test for existing kinds is different: this can be
2026
# factored out to a helper though.
2027
other_block_index, present = self._find_block_index_from_key(other_key)
2028
assert present, 'could not find block for %s' % (other_key,)
2029
other_entry_index, present = self._find_entry_index(other_key,
2030
self._dirblocks[other_block_index][1])
2031
assert present, 'could not find entry for %s' % (other_key,)
2032
assert path_utf8 is not None
2033
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2034
('r', path_utf8, 0, False, '')
2036
num_present_parents = self._num_present_parents()
2037
for lookup_index in xrange(1, num_present_parents + 1):
2038
# grab any one entry, use it to find the right path.
2039
# TODO: optimise this to reduce memory use in highly
2040
# fragmented situations by reusing the relocation
2042
update_block_index, present = \
2043
self._find_block_index_from_key(other_key)
2044
assert present, 'could not find block for %s' % (other_key,)
2045
update_entry_index, present = \
2046
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2047
assert present, 'could not find entry for %s' % (other_key,)
2048
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2049
if update_details[0] in ('r', 'a'): # relocated, absent
2050
# its a pointer or absent in lookup_index's tree, use
2052
new_entry[1].append(update_details)
2054
# we have the right key, make a pointer to it.
2055
pointer_path = osutils.pathjoin(*other_key[0:2])
2056
new_entry[1].append(('r', pointer_path, 0, False, ''))
2057
block.insert(entry_index, new_entry)
2058
existing_keys.add(key)
2060
# Does the new state matter?
2061
block[entry_index][1][0] = new_details
2062
# parents cannot be affected by what we do.
2063
# other occurences of this id can be found
2064
# from the id index.
2066
# tree index consistency: All other paths for this id in this tree
2067
# index must point to the correct path. We have to loop here because
2068
# we may have passed entries in the state with this file id already
2069
# that were absent - where parent entries are - and they need to be
2070
# converted to relocated.
2071
assert path_utf8 is not None
2072
for entry_key in id_index.setdefault(key[2], set()):
2073
# TODO:PROFILING: It might be faster to just update
2074
# rather than checking if we need to, and then overwrite
2075
# the one we are located at.
2076
if entry_key != key:
2077
# this file id is at a different path in one of the
2078
# other trees, so put absent pointers there
2079
# This is the vertical axis in the matrix, all pointing
2081
block_index, present = self._find_block_index_from_key(entry_key)
2083
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2085
self._dirblocks[block_index][1][entry_index][1][0] = \
2086
('r', path_utf8, 0, False, '')
2087
# add a containing dirblock if needed.
2088
if new_details[0] == 'd':
2089
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2090
block_index, present = self._find_block_index_from_key(subdir_key)
2092
self._dirblocks.insert(block_index, (subdir_key[0], []))
2094
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2096
def _validate(self):
2097
"""Check that invariants on the dirblock are correct.
2099
This can be useful in debugging; it shouldn't be necessary in
2102
This must be called with a lock held.
2104
# NOTE: This must always raise AssertionError not just assert,
2105
# otherwise it may not behave properly under python -O
2107
# TODO: All entries must have some content that's not 'a' or 'r',
2108
# otherwise it could just be removed.
2110
# TODO: All relocations must point directly to a real entry.
2112
# TODO: No repeated keys.
2115
from pprint import pformat
2116
self._read_dirblocks_if_needed()
2117
if len(self._dirblocks) > 0:
2118
if not self._dirblocks[0][0] == '':
2119
raise AssertionError(
2120
"dirblocks don't start with root block:\n" + \
2122
if len(self._dirblocks) > 1:
2123
if not self._dirblocks[1][0] == '':
2124
raise AssertionError(
2125
"dirblocks missing root directory:\n" + \
2127
# the dirblocks are sorted by their path components, name, and dir id
2128
dir_names = [d[0].split('/')
2129
for d in self._dirblocks[1:]]
2130
if dir_names != sorted(dir_names):
2131
raise AssertionError(
2132
"dir names are not in sorted order:\n" + \
2133
pformat(self._dirblocks) + \
2136
for dirblock in self._dirblocks:
2137
# within each dirblock, the entries are sorted by filename and
2139
for entry in dirblock[1]:
2140
if dirblock[0] != entry[0][0]:
2141
raise AssertionError(
2143
"doesn't match directory name in\n%r" %
2144
(entry, pformat(dirblock)))
2145
if dirblock[1] != sorted(dirblock[1]):
2146
raise AssertionError(
2147
"dirblock for %r is not sorted:\n%s" % \
2148
(dirblock[0], pformat(dirblock)))
2151
def check_valid_parent():
2152
"""Check that the current entry has a valid parent.
2154
This makes sure that the parent has a record,
2155
and that the parent isn't marked as "absent" in the
2156
current tree. (It is invalid to have a non-absent file in an absent
2159
if entry[0][0:2] == ('', ''):
2160
# There should be no parent for the root row
2162
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2163
if parent_entry == (None, None):
2164
raise AssertionError(
2165
"no parent entry for: %s in tree %s"
2166
% (this_path, tree_index))
2167
if parent_entry[1][tree_index][0] != 'd':
2168
raise AssertionError(
2169
"Parent entry for %s is not marked as a valid"
2170
" directory. %s" % (this_path, parent_entry,))
2172
# For each file id, for each tree: either
2173
# the file id is not present at all; all rows with that id in the
2174
# key have it marked as 'absent'
2175
# OR the file id is present under exactly one name; any other entries
2176
# that mention that id point to the correct name.
2178
# We check this with a dict per tree pointing either to the present
2179
# name, or None if absent.
2180
tree_count = self._num_present_parents() + 1
2181
id_path_maps = [dict() for i in range(tree_count)]
2182
# Make sure that all renamed entries point to the correct location.
2183
for entry in self._iter_entries():
2184
file_id = entry[0][2]
2185
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2186
if len(entry[1]) != tree_count:
2187
raise AssertionError(
2188
"wrong number of entry details for row\n%s" \
2189
",\nexpected %d" % \
2190
(pformat(entry), tree_count))
2191
for tree_index, tree_state in enumerate(entry[1]):
2192
this_tree_map = id_path_maps[tree_index]
2193
minikind = tree_state[0]
2194
# have we seen this id before in this column?
2195
if file_id in this_tree_map:
2196
previous_path = this_tree_map[file_id]
2197
# any later mention of this file must be consistent with
2198
# what was said before
2200
if previous_path is not None:
2201
raise AssertionError(
2202
"file %s is absent in row %r but also present " \
2204
(file_id, entry, previous_path))
2205
elif minikind == 'r':
2206
target_location = tree_state[1]
2207
if previous_path != target_location:
2208
raise AssertionError(
2209
"file %s relocation in row %r but also at %r" \
2210
% (file_id, entry, previous_path))
2212
# a file, directory, etc - may have been previously
2213
# pointed to by a relocation, which must point here
2214
if previous_path != this_path:
2215
raise AssertionError(
2216
"entry %r inconsistent with previous path %r" % \
2217
(entry, previous_path))
2218
check_valid_parent()
2221
# absent; should not occur anywhere else
2222
this_tree_map[file_id] = None
2223
elif minikind == 'r':
2224
# relocation, must occur at expected location
2225
this_tree_map[file_id] = tree_state[1]
2227
this_tree_map[file_id] = this_path
2228
check_valid_parent()
2230
def _wipe_state(self):
2231
"""Forget all state information about the dirstate."""
2232
self._header_state = DirState.NOT_IN_MEMORY
2233
self._dirblock_state = DirState.NOT_IN_MEMORY
2236
self._dirblocks = []
2237
self._id_index = None
2238
self._end_of_header = None
2239
self._cutoff_time = None
2240
self._split_path_cache = {}
2242
def lock_read(self):
2243
"""Acquire a read lock on the dirstate."""
2244
if self._lock_token is not None:
2245
raise errors.LockContention(self._lock_token)
2246
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2247
# already in memory, we could read just the header and check for
2248
# any modification. If not modified, we can just leave things
2250
self._lock_token = lock.ReadLock(self._filename)
2251
self._lock_state = 'r'
2252
self._state_file = self._lock_token.f
2255
def lock_write(self):
2256
"""Acquire a write lock on the dirstate."""
2257
if self._lock_token is not None:
2258
raise errors.LockContention(self._lock_token)
2259
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2260
# already in memory, we could read just the header and check for
2261
# any modification. If not modified, we can just leave things
2263
self._lock_token = lock.WriteLock(self._filename)
2264
self._lock_state = 'w'
2265
self._state_file = self._lock_token.f
2269
"""Drop any locks held on the dirstate."""
2270
if self._lock_token is None:
2271
raise errors.LockNotHeld(self)
2272
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2273
# already in memory, we could read just the header and check for
2274
# any modification. If not modified, we can just leave things
2276
self._state_file = None
2277
self._lock_state = None
2278
self._lock_token.unlock()
2279
self._lock_token = None
2280
self._split_path_cache = {}
2282
def _requires_lock(self):
2283
"""Check that a lock is currently held by someone on the dirstate."""
2284
if not self._lock_token:
2285
raise errors.ObjectNotLocked(self)
2288
# Try to load the compiled form if possible
2290
from bzrlib._dirstate_helpers_c import (
2291
_read_dirblocks_c as _read_dirblocks,
2292
bisect_dirblock_c as bisect_dirblock,
2293
_bisect_path_left_c as _bisect_path_left,
2294
_bisect_path_right_c as _bisect_path_right,
2295
cmp_by_dirs_c as cmp_by_dirs,
2298
from bzrlib._dirstate_helpers_py import (
2299
_read_dirblocks_py as _read_dirblocks,
2300
bisect_dirblock_py as bisect_dirblock,
2301
_bisect_path_left_py as _bisect_path_left,
2302
_bisect_path_right_py as _bisect_path_right,
2303
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
