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 = "adler32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", digit, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but its a file. The fingerprint is a
87
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
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
207
from stat import S_IEXEC
222
class _Bisector(object):
223
"""This just keeps track of information as we are bisecting."""
226
class DirState(object):
227
"""Record directory and metadata state for fast access.
229
A dirstate is a specialised data structure for managing local working
230
tree state information. Its not yet well defined whether it is platform
231
specific, and if it is how we detect/parameterise that.
233
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
234
Unlike most bzr disk formats, DirStates must be locked for reading, using
235
lock_read. (This is an os file lock internally.) This is necessary
236
because the file can be rewritten in place.
238
DirStates must be explicitly written with save() to commit changes; just
239
unlocking them does not write the changes to disk.
242
_kind_to_minikind = {
248
'tree-reference': 't',
250
_minikind_to_kind = {
256
't': 'tree-reference',
258
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
259
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
262
# TODO: jam 20070221 Make sure we handle when there are duplicated records
263
# (like when we remove + add the same path, or we have a rename)
264
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
265
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
266
# that we are sure a single record will always fit.
267
BISECT_PAGE_SIZE = 4096
270
IN_MEMORY_UNMODIFIED = 1
271
IN_MEMORY_MODIFIED = 2
273
# A pack_stat (the x's) that is just noise and will never match the output
276
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
278
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
279
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
281
def __init__(self, path):
282
"""Create a DirState object.
286
:attr _root_entrie: The root row of the directory/file information,
287
- contains the path to / - '', ''
288
- kind of 'directory',
289
- the file id of the root in utf8
292
- and no sha information.
293
:param path: The path at which the dirstate file on disk should live.
295
# _header_state and _dirblock_state represent the current state
296
# of the dirstate metadata and the per-row data respectiely.
297
# NOT_IN_MEMORY indicates that no data is in memory
298
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
299
# is the same as is on disk
300
# IN_MEMORY_MODIFIED indicates that we have a modified version
301
# of what is on disk.
302
# In future we will add more granularity, for instance _dirblock_state
303
# will probably support partially-in-memory as a separate variable,
304
# allowing for partially-in-memory unmodified and partially-in-memory
306
self._header_state = DirState.NOT_IN_MEMORY
307
self._dirblock_state = DirState.NOT_IN_MEMORY
311
self._state_file = None
312
self._filename = path
313
self._lock_token = None
314
self._lock_state = None
315
self._id_index = None
316
self._end_of_header = None
317
self._cutoff_time = None
318
self._split_path_cache = {}
319
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
323
(self.__class__.__name__, self._filename)
325
def add(self, path, file_id, kind, stat, fingerprint):
326
"""Add a path to be tracked.
328
:param path: The path within the dirstate - '' is the root, 'foo' is the
329
path foo within the root, 'foo/bar' is the path bar within foo
331
:param file_id: The file id of the path being added.
332
:param kind: The kind of the path, as a string like 'file',
334
:param stat: The output of os.lstat for the path.
335
:param fingerprint: The sha value of the file,
336
or the target of a symlink,
337
or the referenced revision id for tree-references,
338
or '' for directories.
341
# find the block its in.
342
# find the location in the block.
343
# check its not there
345
#------- copied from inventory.make_entry
346
# --- normalized_filename wants a unicode basename only, so get one.
347
dirname, basename = osutils.split(path)
348
# we dont import normalized_filename directly because we want to be
349
# able to change the implementation at runtime for tests.
350
norm_name, can_access = osutils.normalized_filename(basename)
351
if norm_name != basename:
355
raise errors.InvalidNormalization(path)
356
# you should never have files called . or ..; just add the directory
357
# in the parent, or according to the special treatment for the root
358
if basename == '.' or basename == '..':
359
raise errors.InvalidEntryName(path)
360
# now that we've normalised, we need the correct utf8 path and
361
# dirname and basename elements. This single encode and split should be
362
# faster than three separate encodes.
363
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
364
dirname, basename = osutils.split(utf8path)
365
assert file_id.__class__ == str, \
366
"must be a utf8 file_id not %s" % (type(file_id))
367
# Make sure the file_id does not exist in this tree
368
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
369
if file_id_entry != (None, None):
370
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
371
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
372
info = '%s:%s' % (kind, path)
373
raise errors.DuplicateFileId(file_id, info)
374
first_key = (dirname, basename, '')
375
block_index, present = self._find_block_index_from_key(first_key)
377
# check the path is not in the tree
378
block = self._dirblocks[block_index][1]
379
entry_index, _ = self._find_entry_index(first_key, block)
380
while (entry_index < len(block) and
381
block[entry_index][0][0:2] == first_key[0:2]):
382
if block[entry_index][1][0][0] not in 'ar':
383
# this path is in the dirstate in the current tree.
384
raise Exception, "adding already added path!"
387
# The block where we want to put the file is not present. But it
388
# might be because the directory was empty, or not loaded yet. Look
389
# for a parent entry, if not found, raise NotVersionedError
390
parent_dir, parent_base = osutils.split(dirname)
391
parent_block_idx, parent_entry_idx, _, parent_present = \
392
self._get_block_entry_index(parent_dir, parent_base, 0)
393
if not parent_present:
394
raise errors.NotVersionedError(path, str(self))
395
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
396
block = self._dirblocks[block_index][1]
397
entry_key = (dirname, basename, file_id)
400
packed_stat = DirState.NULLSTAT
403
packed_stat = pack_stat(stat)
404
parent_info = self._empty_parent_info()
405
minikind = DirState._kind_to_minikind[kind]
407
entry_data = entry_key, [
408
(minikind, fingerprint, size, False, packed_stat),
410
elif kind == 'directory':
411
entry_data = entry_key, [
412
(minikind, '', 0, False, packed_stat),
414
elif kind == 'symlink':
415
entry_data = entry_key, [
416
(minikind, fingerprint, size, False, packed_stat),
418
elif kind == 'tree-reference':
419
entry_data = entry_key, [
420
(minikind, fingerprint, 0, False, packed_stat),
423
raise errors.BzrError('unknown kind %r' % kind)
424
entry_index, present = self._find_entry_index(entry_key, block)
425
assert not present, "basename %r already added" % basename
426
block.insert(entry_index, entry_data)
428
if kind == 'directory':
429
# insert a new dirblock
430
self._ensure_block(block_index, entry_index, utf8path)
431
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
433
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
435
def _bisect(self, dir_name_list):
436
"""Bisect through the disk structure for specific rows.
438
:param dir_name_list: A list of (dir, name) pairs.
439
:return: A dict mapping (dir, name) => entry for found entries. Missing
440
entries will not be in the map.
442
self._requires_lock()
443
# We need the file pointer to be right after the initial header block
444
self._read_header_if_needed()
445
# If _dirblock_state was in memory, we should just return info from
446
# there, this function is only meant to handle when we want to read
448
assert self._dirblock_state == DirState.NOT_IN_MEMORY
450
# The disk representation is generally info + '\0\n\0' at the end. But
451
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
452
# Because it means we can sync on the '\n'
453
state_file = self._state_file
454
file_size = os.fstat(state_file.fileno()).st_size
455
# We end up with 2 extra fields, we should have a trailing '\n' to
456
# ensure that we read the whole record, and we should have a precursur
457
# '' which ensures that we start after the previous '\n'
458
entry_field_count = self._fields_per_entry() + 1
460
low = self._end_of_header
461
high = file_size - 1 # Ignore the final '\0'
462
# Map from (dir, name) => entry
465
# Avoid infinite seeking
466
max_count = 30*len(dir_name_list)
468
# pending is a list of places to look.
469
# each entry is a tuple of low, high, dir_names
470
# low -> the first byte offset to read (inclusive)
471
# high -> the last byte offset (inclusive)
472
# dir_names -> The list of (dir, name) pairs that should be found in
473
# the [low, high] range
474
pending = [(low, high, dir_name_list)]
476
page_size = self._bisect_page_size
478
fields_to_entry = self._get_fields_to_entry()
481
low, high, cur_files = pending.pop()
483
if not cur_files or low >= high:
488
if count > max_count:
489
raise errors.BzrError('Too many seeks, most likely a bug.')
491
mid = max(low, (low+high-page_size)/2)
494
# limit the read size, so we don't end up reading data that we have
496
read_size = min(page_size, (high-mid)+1)
497
block = state_file.read(read_size)
500
entries = block.split('\n')
503
# We didn't find a '\n', so we cannot have found any records.
504
# So put this range back and try again. But we know we have to
505
# increase the page size, because a single read did not contain
506
# a record break (so records must be larger than page_size)
508
pending.append((low, high, cur_files))
511
# Check the first and last entries, in case they are partial, or if
512
# we don't care about the rest of this page
514
first_fields = entries[0].split('\0')
515
if len(first_fields) < entry_field_count:
516
# We didn't get the complete first entry
517
# so move start, and grab the next, which
518
# should be a full entry
519
start += len(entries[0])+1
520
first_fields = entries[1].split('\0')
523
if len(first_fields) <= 2:
524
# We didn't even get a filename here... what do we do?
525
# Try a large page size and repeat this query
527
pending.append((low, high, cur_files))
530
# Find what entries we are looking for, which occur before and
531
# after this first record.
533
first_dir_name = (first_fields[1], first_fields[2])
534
first_loc = bisect.bisect_left(cur_files, first_dir_name)
536
# These exist before the current location
537
pre = cur_files[:first_loc]
538
# These occur after the current location, which may be in the
539
# data we read, or might be after the last entry
540
post = cur_files[first_loc:]
542
if post and len(first_fields) >= entry_field_count:
543
# We have files after the first entry
545
# Parse the last entry
546
last_entry_num = len(entries)-1
547
last_fields = entries[last_entry_num].split('\0')
548
if len(last_fields) < entry_field_count:
549
# The very last hunk was not complete,
550
# read the previous hunk
551
after = mid + len(block) - len(entries[-1])
553
last_fields = entries[last_entry_num].split('\0')
555
after = mid + len(block)
557
last_dir_name = (last_fields[1], last_fields[2])
558
last_loc = bisect.bisect_right(post, last_dir_name)
560
middle_files = post[:last_loc]
561
post = post[last_loc:]
564
# We have files that should occur in this block
565
# (>= first, <= last)
566
# Either we will find them here, or we can mark them as
569
if middle_files[0] == first_dir_name:
570
# We might need to go before this location
571
pre.append(first_dir_name)
572
if middle_files[-1] == last_dir_name:
573
post.insert(0, last_dir_name)
575
# Find out what paths we have
576
paths = {first_dir_name:[first_fields]}
577
# last_dir_name might == first_dir_name so we need to be
578
# careful if we should append rather than overwrite
579
if last_entry_num != first_entry_num:
580
paths.setdefault(last_dir_name, []).append(last_fields)
581
for num in xrange(first_entry_num+1, last_entry_num):
582
# TODO: jam 20070223 We are already splitting here, so
583
# shouldn't we just split the whole thing rather
584
# than doing the split again in add_one_record?
585
fields = entries[num].split('\0')
586
dir_name = (fields[1], fields[2])
587
paths.setdefault(dir_name, []).append(fields)
589
for dir_name in middle_files:
590
for fields in paths.get(dir_name, []):
591
# offset by 1 because of the opening '\0'
592
# consider changing fields_to_entry to avoid the
594
entry = fields_to_entry(fields[1:])
595
found.setdefault(dir_name, []).append(entry)
597
# Now we have split up everything into pre, middle, and post, and
598
# we have handled everything that fell in 'middle'.
599
# We add 'post' first, so that we prefer to seek towards the
600
# beginning, so that we will tend to go as early as we need, and
601
# then only seek forward after that.
603
pending.append((after, high, post))
605
pending.append((low, start-1, pre))
607
# Consider that we may want to return the directory entries in sorted
608
# order. For now, we just return them in whatever order we found them,
609
# and leave it up to the caller if they care if it is ordered or not.
612
def _bisect_dirblocks(self, dir_list):
613
"""Bisect through the disk structure to find entries in given dirs.
615
_bisect_dirblocks is meant to find the contents of directories, which
616
differs from _bisect, which only finds individual entries.
618
:param dir_list: An sorted list of directory names ['', 'dir', 'foo'].
619
:return: A map from dir => entries_for_dir
621
# TODO: jam 20070223 A lot of the bisecting logic could be shared
622
# between this and _bisect. It would require parameterizing the
623
# inner loop with a function, though. We should evaluate the
624
# performance difference.
625
self._requires_lock()
626
# We need the file pointer to be right after the initial header block
627
self._read_header_if_needed()
628
# If _dirblock_state was in memory, we should just return info from
629
# there, this function is only meant to handle when we want to read
631
assert self._dirblock_state == DirState.NOT_IN_MEMORY
633
# The disk representation is generally info + '\0\n\0' at the end. But
634
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
635
# Because it means we can sync on the '\n'
636
state_file = self._state_file
637
file_size = os.fstat(state_file.fileno()).st_size
638
# We end up with 2 extra fields, we should have a trailing '\n' to
639
# ensure that we read the whole record, and we should have a precursur
640
# '' which ensures that we start after the previous '\n'
641
entry_field_count = self._fields_per_entry() + 1
643
low = self._end_of_header
644
high = file_size - 1 # Ignore the final '\0'
645
# Map from dir => entry
648
# Avoid infinite seeking
649
max_count = 30*len(dir_list)
651
# pending is a list of places to look.
652
# each entry is a tuple of low, high, dir_names
653
# low -> the first byte offset to read (inclusive)
654
# high -> the last byte offset (inclusive)
655
# dirs -> The list of directories that should be found in
656
# the [low, high] range
657
pending = [(low, high, dir_list)]
659
page_size = self._bisect_page_size
661
fields_to_entry = self._get_fields_to_entry()
664
low, high, cur_dirs = pending.pop()
666
if not cur_dirs or low >= high:
671
if count > max_count:
672
raise errors.BzrError('Too many seeks, most likely a bug.')
674
mid = max(low, (low+high-page_size)/2)
677
# limit the read size, so we don't end up reading data that we have
679
read_size = min(page_size, (high-mid)+1)
680
block = state_file.read(read_size)
683
entries = block.split('\n')
686
# We didn't find a '\n', so we cannot have found any records.
687
# So put this range back and try again. But we know we have to
688
# increase the page size, because a single read did not contain
689
# a record break (so records must be larger than page_size)
691
pending.append((low, high, cur_dirs))
694
# Check the first and last entries, in case they are partial, or if
695
# we don't care about the rest of this page
697
first_fields = entries[0].split('\0')
698
if len(first_fields) < entry_field_count:
699
# We didn't get the complete first entry
700
# so move start, and grab the next, which
701
# should be a full entry
702
start += len(entries[0])+1
703
first_fields = entries[1].split('\0')
706
if len(first_fields) <= 1:
707
# We didn't even get a dirname here... what do we do?
708
# Try a large page size and repeat this query
710
pending.append((low, high, cur_dirs))
713
# Find what entries we are looking for, which occur before and
714
# after this first record.
716
first_dir = first_fields[1]
717
first_loc = bisect.bisect_left(cur_dirs, first_dir)
719
# These exist before the current location
720
pre = cur_dirs[:first_loc]
721
# These occur after the current location, which may be in the
722
# data we read, or might be after the last entry
723
post = cur_dirs[first_loc:]
725
if post and len(first_fields) >= entry_field_count:
726
# We have records to look at after the first entry
728
# Parse the last entry
729
last_entry_num = len(entries)-1
730
last_fields = entries[last_entry_num].split('\0')
731
if len(last_fields) < entry_field_count:
732
# The very last hunk was not complete,
733
# read the previous hunk
734
after = mid + len(block) - len(entries[-1])
736
last_fields = entries[last_entry_num].split('\0')
738
after = mid + len(block)
740
last_dir = last_fields[1]
741
last_loc = bisect.bisect_right(post, last_dir)
743
middle_files = post[:last_loc]
744
post = post[last_loc:]
747
# We have files that should occur in this block
748
# (>= first, <= last)
749
# Either we will find them here, or we can mark them as
752
if middle_files[0] == first_dir:
753
# We might need to go before this location
754
pre.append(first_dir)
755
if middle_files[-1] == last_dir:
756
post.insert(0, last_dir)
758
# Find out what paths we have
759
paths = {first_dir:[first_fields]}
760
# last_dir might == first_dir so we need to be
761
# careful if we should append rather than overwrite
762
if last_entry_num != first_entry_num:
763
paths.setdefault(last_dir, []).append(last_fields)
764
for num in xrange(first_entry_num+1, last_entry_num):
765
# TODO: jam 20070223 We are already splitting here, so
766
# shouldn't we just split the whole thing rather
767
# than doing the split again in add_one_record?
768
fields = entries[num].split('\0')
769
paths.setdefault(fields[1], []).append(fields)
771
for cur_dir in middle_files:
772
for fields in paths.get(cur_dir, []):
773
# offset by 1 because of the opening '\0'
774
# consider changing fields_to_entry to avoid the
776
entry = fields_to_entry(fields[1:])
777
found.setdefault(cur_dir, []).append(entry)
779
# Now we have split up everything into pre, middle, and post, and
780
# we have handled everything that fell in 'middle'.
781
# We add 'post' first, so that we prefer to seek towards the
782
# beginning, so that we will tend to go as early as we need, and
783
# then only seek forward after that.
785
pending.append((after, high, post))
787
pending.append((low, start-1, pre))
791
def _bisect_recursive(self, dir_name_list):
792
"""Bisect for entries for all paths and their children.
794
This will use bisect to find all records for the supplied paths. It
795
will then continue to bisect for any records which are marked as
796
directories. (and renames?)
798
:param paths: A sorted list of (dir, name) pairs
799
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
800
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
802
# Map from (dir, name, file_id) => [tree_info]
805
found_dir_names = set()
807
# Directories that have been read
808
processed_dirs = set()
809
# Get the ball rolling with the first bisect for all entries.
810
newly_found = self._bisect(dir_name_list)
813
# Directories that need to be read
815
paths_to_search = set()
816
for entry_list in newly_found.itervalues():
817
for dir_name_id, trees_info in entry_list:
818
found[dir_name_id] = trees_info
819
found_dir_names.add(dir_name_id[:2])
821
for tree_info in trees_info:
822
minikind = tree_info[0]
825
# We already processed this one as a directory,
826
# we don't need to do the extra work again.
828
subdir, name, file_id = dir_name_id
829
path = osutils.pathjoin(subdir, name)
831
if path not in processed_dirs:
832
pending_dirs.add(path)
833
elif minikind == 'r':
834
# Rename, we need to directly search the target
835
# which is contained in the fingerprint column
836
dir_name = osutils.split(tree_info[1])
837
if dir_name[0] in pending_dirs:
838
# This entry will be found in the dir search
840
# TODO: We need to check if this entry has
841
# already been found. Otherwise we might be
842
# hitting infinite recursion.
843
if dir_name not in found_dir_names:
844
paths_to_search.add(dir_name)
845
# Now we have a list of paths to look for directly, and
846
# directory blocks that need to be read.
847
# newly_found is mixing the keys between (dir, name) and path
848
# entries, but that is okay, because we only really care about the
850
newly_found = self._bisect(sorted(paths_to_search))
851
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
852
processed_dirs.update(pending_dirs)
855
def _empty_parent_info(self):
856
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
859
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
860
"""Ensure a block for dirname exists.
862
This function exists to let callers which know that there is a
863
directory dirname ensure that the block for it exists. This block can
864
fail to exist because of demand loading, or because a directory had no
865
children. In either case it is not an error. It is however an error to
866
call this if there is no parent entry for the directory, and thus the
867
function requires the coordinates of such an entry to be provided.
869
The root row is special cased and can be indicated with a parent block
872
:param parent_block_index: The index of the block in which dirname's row
874
:param parent_row_index: The index in the parent block where the row
876
:param dirname: The utf8 dirname to ensure there is a block for.
877
:return: The index for the block.
879
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
880
# This is the signature of the root row, and the
881
# contents-of-root row is always index 1
883
# the basename of the directory must be the end of its full name.
884
if not (parent_block_index == -1 and
885
parent_block_index == -1 and dirname == ''):
886
assert dirname.endswith(
887
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
888
block_index, present = self._find_block_index_from_key((dirname, '', ''))
890
## In future, when doing partial parsing, this should load and
891
# populate the entire block.
892
self._dirblocks.insert(block_index, (dirname, []))
895
def _entries_to_current_state(self, new_entries):
896
"""Load new_entries into self.dirblocks.
898
Process new_entries into the current state object, making them the active
899
state. The entries are grouped together by directory to form dirblocks.
901
:param new_entries: A sorted list of entries. This function does not sort
902
to prevent unneeded overhead when callers have a sorted list already.
905
assert new_entries[0][0][0:2] == ('', ''), \
906
"Missing root row %r" % (new_entries[0][0],)
907
# The two blocks here are deliberate: the root block and the
908
# contents-of-root block.
909
self._dirblocks = [('', []), ('', [])]
910
current_block = self._dirblocks[0][1]
913
append_entry = current_block.append
914
for entry in new_entries:
915
if entry[0][0] != current_dirname:
916
# new block - different dirname
918
current_dirname = entry[0][0]
919
self._dirblocks.append((current_dirname, current_block))
920
append_entry = current_block.append
921
# append the entry to the current block
923
self._split_root_dirblock_into_contents()
925
def _split_root_dirblock_into_contents(self):
926
"""Split the root dirblocks into root and contents-of-root.
928
After parsing by path, we end up with root entries and contents-of-root
929
entries in the same block. This loop splits them out again.
931
# The above loop leaves the "root block" entries mixed with the
932
# "contents-of-root block". But we don't want an if check on
933
# all entries, so instead we just fix it up here.
934
assert self._dirblocks[1] == ('', [])
936
contents_of_root_block = []
937
for entry in self._dirblocks[0][1]:
938
if not entry[0][1]: # This is a root entry
939
root_block.append(entry)
941
contents_of_root_block.append(entry)
942
self._dirblocks[0] = ('', root_block)
943
self._dirblocks[1] = ('', contents_of_root_block)
945
def _entry_to_line(self, entry):
946
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
948
:param entry: An entry_tuple as defined in the module docstring.
950
entire_entry = list(entry[0])
951
for tree_number, tree_data in enumerate(entry[1]):
952
# (minikind, fingerprint, size, executable, tree_specific_string)
953
entire_entry.extend(tree_data)
954
# 3 for the key, 5 for the fields per tree.
955
tree_offset = 3 + tree_number * 5
957
entire_entry[tree_offset + 0] = tree_data[0]
959
entire_entry[tree_offset + 2] = str(tree_data[2])
961
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
962
return '\0'.join(entire_entry)
964
def _fields_per_entry(self):
965
"""How many null separated fields should be in each entry row.
967
Each line now has an extra '\n' field which is not used
968
so we just skip over it
971
+ number of fields per tree_data (5) * tree count
974
tree_count = 1 + self._num_present_parents()
975
return 3 + 5 * tree_count + 1
977
def _find_block(self, key, add_if_missing=False):
978
"""Return the block that key should be present in.
980
:param key: A dirstate entry key.
981
:return: The block tuple.
983
block_index, present = self._find_block_index_from_key(key)
985
if not add_if_missing:
986
# check to see if key is versioned itself - we might want to
987
# add it anyway, because dirs with no entries dont get a
988
# dirblock at parse time.
989
# This is an uncommon branch to take: most dirs have children,
990
# and most code works with versioned paths.
991
parent_base, parent_name = osutils.split(key[0])
992
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
993
# some parent path has not been added - its an error to add
995
raise errors.NotVersionedError(key[0:2], str(self))
996
self._dirblocks.insert(block_index, (key[0], []))
997
return self._dirblocks[block_index]
999
def _find_block_index_from_key(self, key):
1000
"""Find the dirblock index for a key.
1002
:return: The block index, True if the block for the key is present.
1004
if key[0:2] == ('', ''):
1006
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1007
cache=self._split_path_cache)
1008
# _right returns one-past-where-key is so we have to subtract
1009
# one to use it. we use _right here because there are two
1010
# '' blocks - the root, and the contents of root
1011
# we always have a minimum of 2 in self._dirblocks: root and
1012
# root-contents, and for '', we get 2 back, so this is
1013
# simple and correct:
1014
present = (block_index < len(self._dirblocks) and
1015
self._dirblocks[block_index][0] == key[0])
1016
return block_index, present
1018
def _find_entry_index(self, key, block):
1019
"""Find the entry index for a key in a block.
1021
:return: The entry index, True if the entry for the key is present.
1023
entry_index = bisect.bisect_left(block, (key, []))
1024
present = (entry_index < len(block) and
1025
block[entry_index][0] == key)
1026
return entry_index, present
1029
def from_tree(tree, dir_state_filename):
1030
"""Create a dirstate from a bzr Tree.
1032
:param tree: The tree which should provide parent information and
1034
:return: a DirState object which is currently locked for writing.
1035
(it was locked by DirState.initialize)
1037
result = DirState.initialize(dir_state_filename)
1041
parent_ids = tree.get_parent_ids()
1042
num_parents = len(parent_ids)
1044
for parent_id in parent_ids:
1045
parent_tree = tree.branch.repository.revision_tree(parent_id)
1046
parent_trees.append((parent_id, parent_tree))
1047
parent_tree.lock_read()
1048
result.set_parent_trees(parent_trees, [])
1049
result.set_state_from_inventory(tree.inventory)
1051
for revid, parent_tree in parent_trees:
1052
parent_tree.unlock()
1055
# The caller won't have a chance to unlock this, so make sure we
1061
def update_entry(self, entry, abspath, stat_value=None):
1062
"""Update the entry based on what is actually on disk.
1064
:param entry: This is the dirblock entry for the file in question.
1065
:param abspath: The path on disk for this file.
1066
:param stat_value: (optional) if we already have done a stat on the
1068
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1071
# This code assumes that the entry passed in is directly held in one of
1072
# the internal _dirblocks. So the dirblock state must have already been
1074
assert self._dirblock_state != DirState.NOT_IN_MEMORY
1075
if stat_value is None:
1077
# We could inline os.lstat but the common case is that
1078
# stat_value will be passed in, not read here.
1079
stat_value = self._lstat(abspath, entry)
1080
except (OSError, IOError), e:
1081
if e.errno in (errno.ENOENT, errno.EACCES,
1083
# The entry is missing, consider it gone
1087
kind = osutils.file_kind_from_stat_mode(stat_value.st_mode)
1089
minikind = DirState._kind_to_minikind[kind]
1090
except KeyError: # Unknown kind
1092
packed_stat = pack_stat(stat_value)
1093
(saved_minikind, saved_link_or_sha1, saved_file_size,
1094
saved_executable, saved_packed_stat) = entry[1][0]
1096
if (minikind == saved_minikind
1097
and packed_stat == saved_packed_stat
1098
# size should also be in packed_stat
1099
and saved_file_size == stat_value.st_size):
1100
# The stat hasn't changed since we saved, so we can potentially
1101
# re-use the saved sha hash.
1105
if self._cutoff_time is None:
1106
self._sha_cutoff_time()
1108
if (stat_value.st_mtime < self._cutoff_time
1109
and stat_value.st_ctime < self._cutoff_time):
1110
# Return the existing fingerprint
1111
return saved_link_or_sha1
1113
# If we have gotten this far, that means that we need to actually
1114
# process this entry.
1117
link_or_sha1 = self._sha1_file(abspath, entry)
1118
executable = self._is_executable(stat_value.st_mode,
1120
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1121
executable, packed_stat)
1122
elif minikind == 'd':
1124
entry[1][0] = ('d', '', 0, False, packed_stat)
1125
if saved_minikind != 'd':
1126
# This changed from something into a directory. Make sure we
1127
# have a directory block for it. This doesn't happen very
1128
# often, so this doesn't have to be super fast.
1129
block_index, entry_index, dir_present, file_present = \
1130
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1131
self._ensure_block(block_index, entry_index,
1132
osutils.pathjoin(entry[0][0], entry[0][1]))
1133
elif minikind == 'l':
1134
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1135
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1137
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1140
def _sha_cutoff_time(self):
1141
"""Return cutoff time.
1143
Files modified more recently than this time are at risk of being
1144
undetectably modified and so can't be cached.
1146
# Cache the cutoff time as long as we hold a lock.
1147
# time.time() isn't super expensive (approx 3.38us), but
1148
# when you call it 50,000 times it adds up.
1149
# For comparison, os.lstat() costs 7.2us if it is hot.
1150
self._cutoff_time = int(time.time()) - 3
1151
return self._cutoff_time
1153
def _lstat(self, abspath, entry):
1154
"""Return the os.lstat value for this path."""
1155
return os.lstat(abspath)
1157
def _sha1_file(self, abspath, entry):
1158
"""Calculate the SHA1 of a file by reading the full text"""
1159
f = file(abspath, 'rb', buffering=65000)
1161
return osutils.sha_file(f)
1165
def _is_executable(self, mode, old_executable):
1166
"""Is this file executable?"""
1167
return bool(S_IEXEC & mode)
1169
def _is_executable_win32(self, mode, old_executable):
1170
"""On win32 the executable bit is stored in the dirstate."""
1171
return old_executable
1173
if sys.platform == 'win32':
1174
_is_executable = _is_executable_win32
1176
def _read_link(self, abspath, old_link):
1177
"""Read the target of a symlink"""
1178
# TODO: jam 200700301 On Win32, this could just return the value
1179
# already in memory. However, this really needs to be done at a
1180
# higher level, because there either won't be anything on disk,
1181
# or the thing on disk will be a file.
1182
return os.readlink(abspath)
1184
def get_ghosts(self):
1185
"""Return a list of the parent tree revision ids that are ghosts."""
1186
self._read_header_if_needed()
1189
def get_lines(self):
1190
"""Serialise the entire dirstate to a sequence of lines."""
1191
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1192
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1193
# read whats on disk.
1194
self._state_file.seek(0)
1195
return self._state_file.readlines()
1197
lines.append(self._get_parents_line(self.get_parent_ids()))
1198
lines.append(self._get_ghosts_line(self._ghosts))
1199
# append the root line which is special cased
1200
lines.extend(map(self._entry_to_line, self._iter_entries()))
1201
return self._get_output_lines(lines)
1203
def _get_ghosts_line(self, ghost_ids):
1204
"""Create a line for the state file for ghost information."""
1205
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1207
def _get_parents_line(self, parent_ids):
1208
"""Create a line for the state file for parents information."""
1209
return '\0'.join([str(len(parent_ids))] + parent_ids)
1211
def _get_fields_to_entry(self):
1212
"""Get a function which converts entry fields into a entry record.
1214
This handles size and executable, as well as parent records.
1216
:return: A function which takes a list of fields, and returns an
1217
appropriate record for storing in memory.
1219
# This is intentionally unrolled for performance
1220
num_present_parents = self._num_present_parents()
1221
if num_present_parents == 0:
1222
def fields_to_entry_0_parents(fields, _int=int):
1223
path_name_file_id_key = (fields[0], fields[1], fields[2])
1224
return (path_name_file_id_key, [
1226
fields[3], # minikind
1227
fields[4], # fingerprint
1228
_int(fields[5]), # size
1229
fields[6] == 'y', # executable
1230
fields[7], # packed_stat or revision_id
1232
return fields_to_entry_0_parents
1233
elif num_present_parents == 1:
1234
def fields_to_entry_1_parent(fields, _int=int):
1235
path_name_file_id_key = (fields[0], fields[1], fields[2])
1236
return (path_name_file_id_key, [
1238
fields[3], # minikind
1239
fields[4], # fingerprint
1240
_int(fields[5]), # size
1241
fields[6] == 'y', # executable
1242
fields[7], # packed_stat or revision_id
1245
fields[8], # minikind
1246
fields[9], # fingerprint
1247
_int(fields[10]), # size
1248
fields[11] == 'y', # executable
1249
fields[12], # packed_stat or revision_id
1252
return fields_to_entry_1_parent
1253
elif num_present_parents == 2:
1254
def fields_to_entry_2_parents(fields, _int=int):
1255
path_name_file_id_key = (fields[0], fields[1], fields[2])
1256
return (path_name_file_id_key, [
1258
fields[3], # minikind
1259
fields[4], # fingerprint
1260
_int(fields[5]), # size
1261
fields[6] == 'y', # executable
1262
fields[7], # packed_stat or revision_id
1265
fields[8], # minikind
1266
fields[9], # fingerprint
1267
_int(fields[10]), # size
1268
fields[11] == 'y', # executable
1269
fields[12], # packed_stat or revision_id
1272
fields[13], # minikind
1273
fields[14], # fingerprint
1274
_int(fields[15]), # size
1275
fields[16] == 'y', # executable
1276
fields[17], # packed_stat or revision_id
1279
return fields_to_entry_2_parents
1281
def fields_to_entry_n_parents(fields, _int=int):
1282
path_name_file_id_key = (fields[0], fields[1], fields[2])
1283
trees = [(fields[cur], # minikind
1284
fields[cur+1], # fingerprint
1285
_int(fields[cur+2]), # size
1286
fields[cur+3] == 'y', # executable
1287
fields[cur+4], # stat or revision_id
1288
) for cur in xrange(3, len(fields)-1, 5)]
1289
return path_name_file_id_key, trees
1290
return fields_to_entry_n_parents
1292
def get_parent_ids(self):
1293
"""Return a list of the parent tree ids for the directory state."""
1294
self._read_header_if_needed()
1295
return list(self._parents)
1297
def _get_block_entry_index(self, dirname, basename, tree_index):
1298
"""Get the coordinates for a path in the state structure.
1300
:param dirname: The utf8 dirname to lookup.
1301
:param basename: The utf8 basename to lookup.
1302
:param tree_index: The index of the tree for which this lookup should
1304
:return: A tuple describing where the path is located, or should be
1305
inserted. The tuple contains four fields: the block index, the row
1306
index, anda two booleans are True when the directory is present, and
1307
when the entire path is present. There is no guarantee that either
1308
coordinate is currently reachable unless the found field for it is
1309
True. For instance, a directory not present in the searched tree
1310
may be returned with a value one greater than the current highest
1311
block offset. The directory present field will always be True when
1312
the path present field is True. The directory present field does
1313
NOT indicate that the directory is present in the searched tree,
1314
rather it indicates that there are at least some files in some
1317
self._read_dirblocks_if_needed()
1318
key = dirname, basename, ''
1319
block_index, present = self._find_block_index_from_key(key)
1321
# no such directory - return the dir index and 0 for the row.
1322
return block_index, 0, False, False
1323
block = self._dirblocks[block_index][1] # access the entries only
1324
entry_index, present = self._find_entry_index(key, block)
1325
# linear search through present entries at this path to find the one
1327
while entry_index < len(block) and block[entry_index][0][1] == basename:
1328
if block[entry_index][1][tree_index][0] not in \
1329
('a', 'r'): # absent, relocated
1330
return block_index, entry_index, True, True
1332
return block_index, entry_index, True, False
1334
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1335
"""Get the dirstate entry for path in tree tree_index
1337
If either file_id or path is supplied, it is used as the key to lookup.
1338
If both are supplied, the fastest lookup is used, and an error is
1339
raised if they do not both point at the same row.
1341
:param tree_index: The index of the tree we wish to locate this path
1342
in. If the path is present in that tree, the entry containing its
1343
details is returned, otherwise (None, None) is returned
1344
0 is the working tree, higher indexes are successive parent
1346
:param fileid_utf8: A utf8 file_id to look up.
1347
:param path_utf8: An utf8 path to be looked up.
1348
:return: The dirstate entry tuple for path, or (None, None)
1350
self._read_dirblocks_if_needed()
1351
if path_utf8 is not None:
1352
assert path_utf8.__class__ == str, 'path_utf8 is not a str: %s %s' % (type(path_utf8), path_utf8)
1353
# path lookups are faster
1354
dirname, basename = osutils.split(path_utf8)
1355
block_index, entry_index, dir_present, file_present = \
1356
self._get_block_entry_index(dirname, basename, tree_index)
1357
if not file_present:
1359
entry = self._dirblocks[block_index][1][entry_index]
1360
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1362
if entry[0][2] != fileid_utf8:
1363
raise errors.BzrError('integrity error ? : mismatching'
1364
' tree_index, file_id and path')
1367
assert fileid_utf8 is not None
1368
possible_keys = self._get_id_index().get(fileid_utf8, None)
1369
if not possible_keys:
1371
for key in possible_keys:
1372
block_index, present = \
1373
self._find_block_index_from_key(key)
1374
# strange, probably indicates an out of date
1375
# id index - for now, allow this.
1378
# WARNING: DO not change this code to use _get_block_entry_index
1379
# as that function is not suitable: it does not use the key
1380
# to lookup, and thus the wront coordinates are returned.
1381
block = self._dirblocks[block_index][1]
1382
entry_index, present = self._find_entry_index(key, block)
1384
entry = self._dirblocks[block_index][1][entry_index]
1385
if entry[1][tree_index][0] in 'fdlt':
1386
# this is the result we are looking for: the
1387
# real home of this file_id in this tree.
1389
if entry[1][tree_index][0] == 'a':
1390
# there is no home for this entry in this tree
1392
assert entry[1][tree_index][0] == 'r', \
1393
"entry %r has invalid minikind %r for tree %r" \
1395
entry[1][tree_index][0],
1397
real_path = entry[1][tree_index][1]
1398
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1399
path_utf8=real_path)
1403
def initialize(cls, path):
1404
"""Create a new dirstate on path.
1406
The new dirstate will be an empty tree - that is it has no parents,
1407
and only a root node - which has id ROOT_ID.
1409
The object will be write locked when returned to the caller,
1410
unless there was an exception in the writing, in which case it
1413
:param path: The name of the file for the dirstate.
1414
:return: A DirState object.
1416
# This constructs a new DirState object on a path, sets the _state_file
1417
# to a new empty file for that path. It then calls _set_data() with our
1418
# stock empty dirstate information - a root with ROOT_ID, no children,
1419
# and no parents. Finally it calls save() to ensure that this data will
1422
# root dir and root dir contents with no children.
1423
empty_tree_dirblocks = [('', []), ('', [])]
1424
# a new root directory, with a NULLSTAT.
1425
empty_tree_dirblocks[0][1].append(
1426
(('', '', inventory.ROOT_ID), [
1427
('d', '', 0, False, DirState.NULLSTAT),
1431
result._set_data([], empty_tree_dirblocks)
1438
def _inv_entry_to_details(self, inv_entry):
1439
"""Convert an inventory entry (from a revision tree) to state details.
1441
:param inv_entry: An inventory entry whose sha1 and link targets can be
1442
relied upon, and which has a revision set.
1443
:return: A details tuple - the details for a single tree at a path +
1446
kind = inv_entry.kind
1447
minikind = DirState._kind_to_minikind[kind]
1448
tree_data = inv_entry.revision
1449
assert len(tree_data) > 0, 'empty revision for the inv_entry.'
1450
if kind == 'directory':
1454
elif kind == 'symlink':
1455
fingerprint = inv_entry.symlink_target or ''
1458
elif kind == 'file':
1459
fingerprint = inv_entry.text_sha1 or ''
1460
size = inv_entry.text_size or 0
1461
executable = inv_entry.executable
1462
elif kind == 'tree-reference':
1463
fingerprint = inv_entry.reference_revision or ''
1467
raise Exception("can't pack %s" % inv_entry)
1468
return (minikind, fingerprint, size, executable, tree_data)
1470
def _iter_entries(self):
1471
"""Iterate over all the entries in the dirstate.
1473
Each yelt item is an entry in the standard format described in the
1474
docstring of bzrlib.dirstate.
1476
self._read_dirblocks_if_needed()
1477
for directory in self._dirblocks:
1478
for entry in directory[1]:
1481
def _get_id_index(self):
1482
"""Get an id index of self._dirblocks."""
1483
if self._id_index is None:
1485
for key, tree_details in self._iter_entries():
1486
id_index.setdefault(key[2], set()).add(key)
1487
self._id_index = id_index
1488
return self._id_index
1490
def _get_output_lines(self, lines):
1491
"""format lines for final output.
1493
:param lines: A sequece of lines containing the parents list and the
1496
output_lines = [DirState.HEADER_FORMAT_3]
1497
lines.append('') # a final newline
1498
inventory_text = '\0\n\0'.join(lines)
1499
output_lines.append('adler32: %s\n' % (zlib.adler32(inventory_text),))
1500
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1501
num_entries = len(lines)-3
1502
output_lines.append('num_entries: %s\n' % (num_entries,))
1503
output_lines.append(inventory_text)
1506
def _make_deleted_row(self, fileid_utf8, parents):
1507
"""Return a deleted for for fileid_utf8."""
1508
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1511
def _num_present_parents(self):
1512
"""The number of parent entries in each record row."""
1513
return len(self._parents) - len(self._ghosts)
1517
"""Construct a DirState on the file at path path.
1519
:return: An unlocked DirState object, associated with the given path.
1521
result = DirState(path)
1524
def _read_dirblocks_if_needed(self):
1525
"""Read in all the dirblocks from the file if they are not in memory.
1527
This populates self._dirblocks, and sets self._dirblock_state to
1528
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1531
self._read_header_if_needed()
1532
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1533
# move the _state_file pointer to after the header (in case bisect
1534
# has been called in the mean time)
1535
self._state_file.seek(self._end_of_header)
1536
text = self._state_file.read()
1537
# TODO: check the adler checksums. adler_measured = zlib.adler32(text)
1539
fields = text.split('\0')
1540
# Remove the last blank entry
1541
trailing = fields.pop()
1542
assert trailing == ''
1543
# consider turning fields into a tuple.
1545
# skip the first field which is the trailing null from the header.
1547
# Each line now has an extra '\n' field which is not used
1548
# so we just skip over it
1550
# 3 fields for the key
1551
# + number of fields per tree_data (5) * tree count
1553
num_present_parents = self._num_present_parents()
1554
tree_count = 1 + num_present_parents
1555
entry_size = self._fields_per_entry()
1556
expected_field_count = entry_size * self._num_entries
1557
if len(fields) - cur > expected_field_count:
1558
fields = fields[:expected_field_count + cur]
1559
trace.mutter('Unexpectedly long dirstate field count!')
1560
print "XXX: incorrectly truncated dirstate file bug triggered."
1561
field_count = len(fields)
1562
# this checks our adjustment, and also catches file too short.
1563
assert field_count - cur == expected_field_count, \
1564
'field count incorrect %s != %s, entry_size=%s, '\
1565
'num_entries=%s fields=%r' % (
1566
field_count - cur, expected_field_count, entry_size,
1567
self._num_entries, fields)
1569
if num_present_parents == 1:
1570
# Bind external functions to local names
1572
# We access all fields in order, so we can just iterate over
1573
# them. Grab an straight iterator over the fields. (We use an
1574
# iterator because we don't want to do a lot of additions, nor
1575
# do we want to do a lot of slicing)
1576
next = iter(fields).next
1577
# Move the iterator to the current position
1578
for x in xrange(cur):
1580
# The two blocks here are deliberate: the root block and the
1581
# contents-of-root block.
1582
self._dirblocks = [('', []), ('', [])]
1583
current_block = self._dirblocks[0][1]
1584
current_dirname = ''
1585
append_entry = current_block.append
1586
for count in xrange(self._num_entries):
1590
if dirname != current_dirname:
1591
# new block - different dirname
1593
current_dirname = dirname
1594
self._dirblocks.append((current_dirname, current_block))
1595
append_entry = current_block.append
1596
# we know current_dirname == dirname, so re-use it to avoid
1597
# creating new strings
1598
entry = ((current_dirname, name, file_id),
1601
next(), # fingerprint
1602
_int(next()), # size
1603
next() == 'y', # executable
1604
next(), # packed_stat or revision_id
1608
next(), # fingerprint
1609
_int(next()), # size
1610
next() == 'y', # executable
1611
next(), # packed_stat or revision_id
1615
assert trailing == '\n'
1616
# append the entry to the current block
1618
self._split_root_dirblock_into_contents()
1620
fields_to_entry = self._get_fields_to_entry()
1621
entries = [fields_to_entry(fields[pos:pos+entry_size])
1622
for pos in xrange(cur, field_count, entry_size)]
1623
self._entries_to_current_state(entries)
1624
# To convert from format 2 => format 3
1625
# self._dirblocks = sorted(self._dirblocks,
1626
# key=lambda blk:blk[0].split('/'))
1627
# To convert from format 3 => format 2
1628
# self._dirblocks = sorted(self._dirblocks)
1629
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1631
def _read_header(self):
1632
"""This reads in the metadata header, and the parent ids.
1634
After reading in, the file should be positioned at the null
1635
just before the start of the first record in the file.
1637
:return: (expected adler checksum, number of entries, parent list)
1639
self._read_prelude()
1640
parent_line = self._state_file.readline()
1641
info = parent_line.split('\0')
1642
num_parents = int(info[0])
1643
assert num_parents == len(info)-2, 'incorrect parent info line'
1644
self._parents = info[1:-1]
1646
ghost_line = self._state_file.readline()
1647
info = ghost_line.split('\0')
1648
num_ghosts = int(info[1])
1649
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1650
self._ghosts = info[2:-1]
1651
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1652
self._end_of_header = self._state_file.tell()
1654
def _read_header_if_needed(self):
1655
"""Read the header of the dirstate file if needed."""
1656
# inline this as it will be called a lot
1657
if not self._lock_token:
1658
raise errors.ObjectNotLocked(self)
1659
if self._header_state == DirState.NOT_IN_MEMORY:
1662
def _read_prelude(self):
1663
"""Read in the prelude header of the dirstate file
1665
This only reads in the stuff that is not connected to the adler
1666
checksum. The position will be correct to read in the rest of
1667
the file and check the checksum after this point.
1668
The next entry in the file should be the number of parents,
1669
and their ids. Followed by a newline.
1671
header = self._state_file.readline()
1672
assert header == DirState.HEADER_FORMAT_3, \
1673
'invalid header line: %r' % (header,)
1674
adler_line = self._state_file.readline()
1675
assert adler_line.startswith('adler32: '), 'missing adler32 checksum'
1676
self.adler_expected = int(adler_line[len('adler32: '):-1])
1677
num_entries_line = self._state_file.readline()
1678
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1679
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1682
"""Save any pending changes created during this session.
1684
We reuse the existing file, because that prevents race conditions with
1685
file creation, and use oslocks on it to prevent concurrent modification
1686
and reads - because dirstates incremental data aggretation is not
1687
compatible with reading a modified file, and replacing a file in use by
1688
another process is impossible on windows.
1690
A dirstate in read only mode should be smart enough though to validate
1691
that the file has not changed, and otherwise discard its cache and
1692
start over, to allow for fine grained read lock duration, so 'status'
1693
wont block 'commit' - for example.
1695
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1696
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1698
if self._lock_state == 'w':
1699
out_file = self._state_file
1702
# Try to grab a write lock so that we can update the file.
1704
wlock = lock.WriteLock(self._filename)
1705
except (errors.LockError, errors.LockContention), e:
1706
# We couldn't grab the lock, so just leave things dirty in
1710
# This may be a read-only tree, or someone else may have a
1711
# ReadLock. so handle the case when we cannot grab a write
1713
if e.errno in (errno.ENOENT, errno.EPERM, errno.EACCES,
1715
# Ignore these errors and just don't save anything
1721
out_file.writelines(self.get_lines())
1724
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1725
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1727
if wlock is not None:
1730
def _set_data(self, parent_ids, dirblocks):
1731
"""Set the full dirstate data in memory.
1733
This is an internal function used to completely replace the objects
1734
in memory state. It puts the dirstate into state 'full-dirty'.
1736
:param parent_ids: A list of parent tree revision ids.
1737
:param dirblocks: A list containing one tuple for each directory in the
1738
tree. Each tuple contains the directory path and a list of entries
1739
found in that directory.
1741
# our memory copy is now authoritative.
1742
self._dirblocks = dirblocks
1743
self._header_state = DirState.IN_MEMORY_MODIFIED
1744
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1745
self._parents = list(parent_ids)
1746
self._id_index = None
1748
def set_path_id(self, path, new_id):
1749
"""Change the id of path to new_id in the current working tree.
1751
:param path: The path inside the tree to set - '' is the root, 'foo'
1752
is the path foo in the root.
1753
:param new_id: The new id to assign to the path. This must be a utf8
1754
file id (not unicode, and not None).
1756
# TODO: start warning here.
1757
assert new_id.__class__ == str
1758
self._read_dirblocks_if_needed()
1760
import pdb;pdb.set_trace()
1762
raise NotImplementedError(self.set_path_id)
1763
# TODO: check new id is unique
1764
entry = self._get_entry(0, path_utf8=path)
1765
if entry[0][2] == new_id:
1766
# Nothing to change.
1768
# mark the old path absent, and insert a new root path
1769
self._make_absent(entry)
1770
self.update_minimal(('', '', new_id), 'd',
1771
path_utf8='', packed_stat=entry[1][0][4])
1772
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1773
if self._id_index is not None:
1774
self._id_index.setdefault(new_id, set()).add(entry[0])
1776
def set_parent_trees(self, trees, ghosts):
1777
"""Set the parent trees for the dirstate.
1779
:param trees: A list of revision_id, tree tuples. tree must be provided
1780
even if the revision_id refers to a ghost: supply an empty tree in
1782
:param ghosts: A list of the revision_ids that are ghosts at the time
1786
# TODO: generate a list of parent indexes to preserve to save
1787
# processing specific parent trees. In the common case one tree will
1788
# be preserved - the left most parent.
1789
# TODO: if the parent tree is a dirstate, we might want to walk them
1790
# all by path in parallel for 'optimal' common-case performance.
1791
# generate new root row.
1792
self._read_dirblocks_if_needed()
1793
# TODO future sketch: Examine the existing parents to generate a change
1794
# map and then walk the new parent trees only, mapping them into the
1795
# dirstate. Walk the dirstate at the same time to remove unreferenced
1798
# sketch: loop over all entries in the dirstate, cherry picking
1799
# entries from the parent trees, if they are not ghost trees.
1800
# after we finish walking the dirstate, all entries not in the dirstate
1801
# are deletes, so we want to append them to the end as per the design
1802
# discussions. So do a set difference on ids with the parents to
1803
# get deletes, and add them to the end.
1804
# During the update process we need to answer the following questions:
1805
# - find other keys containing a fileid in order to create cross-path
1806
# links. We dont't trivially use the inventory from other trees
1807
# because this leads to either double touching, or to accessing
1809
# - find other keys containing a path
1810
# We accumulate each entry via this dictionary, including the root
1813
# we could do parallel iterators, but because file id data may be
1814
# scattered throughout, we dont save on index overhead: we have to look
1815
# at everything anyway. We can probably save cycles by reusing parent
1816
# data and doing an incremental update when adding an additional
1817
# parent, but for now the common cases are adding a new parent (merge),
1818
# and replacing completely (commit), and commit is more common: so
1819
# optimise merge later.
1821
# ---- start generation of full tree mapping data
1822
# what trees should we use?
1823
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
1824
# how many trees do we end up with
1825
parent_count = len(parent_trees)
1827
# one: the current tree
1828
for entry in self._iter_entries():
1829
# skip entries not in the current tree
1830
if entry[1][0][0] in ('a', 'r'): # absent, relocated
1832
by_path[entry[0]] = [entry[1][0]] + \
1833
[DirState.NULL_PARENT_DETAILS] * parent_count
1834
id_index[entry[0][2]] = set([entry[0]])
1836
# now the parent trees:
1837
for tree_index, tree in enumerate(parent_trees):
1838
# the index is off by one, adjust it.
1839
tree_index = tree_index + 1
1840
# when we add new locations for a fileid we need these ranges for
1841
# any fileid in this tree as we set the by_path[id] to:
1842
# already_processed_tree_details + new_details + new_location_suffix
1843
# the suffix is from tree_index+1:parent_count+1.
1844
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
1845
# now stitch in all the entries from this tree
1846
for path, entry in tree.inventory.iter_entries_by_dir():
1847
# here we process each trees details for each item in the tree.
1848
# we first update any existing entries for the id at other paths,
1849
# then we either create or update the entry for the id at the
1850
# right path, and finally we add (if needed) a mapping from
1851
# file_id to this path. We do it in this order to allow us to
1852
# avoid checking all known paths for the id when generating a
1853
# new entry at this path: by adding the id->path mapping last,
1854
# all the mappings are valid and have correct relocation
1855
# records where needed.
1856
file_id = entry.file_id
1857
path_utf8 = path.encode('utf8')
1858
dirname, basename = osutils.split(path_utf8)
1859
new_entry_key = (dirname, basename, file_id)
1860
# tree index consistency: All other paths for this id in this tree
1861
# index must point to the correct path.
1862
for entry_key in id_index.setdefault(file_id, set()):
1863
# TODO:PROFILING: It might be faster to just update
1864
# rather than checking if we need to, and then overwrite
1865
# the one we are located at.
1866
if entry_key != new_entry_key:
1867
# this file id is at a different path in one of the
1868
# other trees, so put absent pointers there
1869
# This is the vertical axis in the matrix, all pointing
1871
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
1872
# by path consistency: Insert into an existing path record (trivial), or
1873
# add a new one with relocation pointers for the other tree indexes.
1874
if new_entry_key in id_index[file_id]:
1875
# there is already an entry where this data belongs, just insert it.
1876
by_path[new_entry_key][tree_index] = \
1877
self._inv_entry_to_details(entry)
1879
# add relocated entries to the horizontal axis - this row
1880
# mapping from path,id. We need to look up the correct path
1881
# for the indexes from 0 to tree_index -1
1883
for lookup_index in xrange(tree_index):
1884
# boundary case: this is the first occurence of file_id
1885
# so there are no id_indexs, possibly take this out of
1887
if not len(id_index[file_id]):
1888
new_details.append(DirState.NULL_PARENT_DETAILS)
1890
# grab any one entry, use it to find the right path.
1891
# TODO: optimise this to reduce memory use in highly
1892
# fragmented situations by reusing the relocation
1894
a_key = iter(id_index[file_id]).next()
1895
if by_path[a_key][lookup_index][0] in ('r', 'a'):
1896
# its a pointer or missing statement, use it as is.
1897
new_details.append(by_path[a_key][lookup_index])
1899
# we have the right key, make a pointer to it.
1900
real_path = ('/'.join(a_key[0:2])).strip('/')
1901
new_details.append(('r', real_path, 0, False, ''))
1902
new_details.append(self._inv_entry_to_details(entry))
1903
new_details.extend(new_location_suffix)
1904
by_path[new_entry_key] = new_details
1905
id_index[file_id].add(new_entry_key)
1906
# --- end generation of full tree mappings
1908
# sort and output all the entries
1909
new_entries = self._sort_entries(by_path.items())
1910
self._entries_to_current_state(new_entries)
1911
self._parents = [rev_id for rev_id, tree in trees]
1912
self._ghosts = list(ghosts)
1913
self._header_state = DirState.IN_MEMORY_MODIFIED
1914
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1915
self._id_index = id_index
1918
def _sort_entries(self, entry_list):
1919
"""Given a list of entries, sort them into the right order.
1921
This is done when constructing a new dirstate from trees - normally we
1922
try to keep everything in sorted blocks all the time, but sometimes
1923
it's easier to sort after the fact.
1925
# TODO: Might be faster to do a scwartzian transform?
1927
# sort by: directory parts, file name, file id
1928
return entry[0][0].split('/'), entry[0][1], entry[0][2]
1929
return sorted(entry_list, key=_key)
1931
def set_state_from_inventory(self, new_inv):
1932
"""Set new_inv as the current state.
1934
This API is called by tree transform, and will usually occur with
1935
existing parent trees.
1937
:param new_inv: The inventory object to set current state from.
1939
self._read_dirblocks_if_needed()
1941
# incremental algorithm:
1942
# two iterators: current data and new data, both in dirblock order.
1943
new_iterator = new_inv.iter_entries_by_dir()
1944
# we will be modifying the dirstate, so we need a stable iterator. In
1945
# future we might write one, for now we just clone the state into a
1946
# list - which is a shallow copy, so each
1947
old_iterator = iter(list(self._iter_entries()))
1948
# both must have roots so this is safe:
1949
current_new = new_iterator.next()
1950
current_old = old_iterator.next()
1951
def advance(iterator):
1953
return iterator.next()
1954
except StopIteration:
1956
while current_new or current_old:
1957
# skip entries in old that are not really there
1958
if current_old and current_old[1][0][0] in ('r', 'a'):
1959
# relocated or absent
1960
current_old = advance(old_iterator)
1963
# convert new into dirblock style
1964
new_path_utf8 = current_new[0].encode('utf8')
1965
new_dirname, new_basename = osutils.split(new_path_utf8)
1966
new_id = current_new[1].file_id
1967
new_entry_key = (new_dirname, new_basename, new_id)
1968
current_new_minikind = \
1969
DirState._kind_to_minikind[current_new[1].kind]
1970
if current_new_minikind == 't':
1971
fingerprint = current_new[1].reference_revision
1975
# for safety disable variables
1976
new_path_utf8 = new_dirname = new_basename = new_id = new_entry_key = None
1977
# 5 cases, we dont have a value that is strictly greater than everything, so
1978
# we make both end conditions explicit
1980
# old is finished: insert current_new into the state.
1981
self.update_minimal(new_entry_key, current_new_minikind,
1982
executable=current_new[1].executable,
1983
path_utf8=new_path_utf8, fingerprint=fingerprint)
1984
current_new = advance(new_iterator)
1985
elif not current_new:
1987
self._make_absent(current_old)
1988
current_old = advance(old_iterator)
1989
elif new_entry_key == current_old[0]:
1990
# same - common case
1991
# TODO: update the record if anything significant has changed.
1992
# the minimal required trigger is if the execute bit or cached
1994
if (current_old[1][0][3] != current_new[1].executable or
1995
current_old[1][0][0] != current_new_minikind):
1996
self.update_minimal(current_old[0], current_new_minikind,
1997
executable=current_new[1].executable,
1998
path_utf8=new_path_utf8, fingerprint=fingerprint)
1999
# both sides are dealt with, move on
2000
current_old = advance(old_iterator)
2001
current_new = advance(new_iterator)
2002
elif new_entry_key < current_old[0]:
2004
# add a entry for this and advance new
2005
self.update_minimal(new_entry_key, current_new_minikind,
2006
executable=current_new[1].executable,
2007
path_utf8=new_path_utf8, fingerprint=fingerprint)
2008
current_new = advance(new_iterator)
2011
self._make_absent(current_old)
2012
current_old = advance(old_iterator)
2013
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2014
self._id_index = None
2016
def _make_absent(self, current_old):
2017
"""Mark current_old - an entry - as absent for tree 0.
2019
:return: True if this was the last details entry for they entry key:
2020
that is, if the underlying block has had the entry removed, thus
2021
shrinking in length.
2023
# build up paths that this id will be left at after the change is made,
2024
# so we can update their cross references in tree 0
2025
all_remaining_keys = set()
2026
# Dont check the working tree, because its going.
2027
for details in current_old[1][1:]:
2028
if details[0] not in ('a', 'r'): # absent, relocated
2029
all_remaining_keys.add(current_old[0])
2030
elif details[0] == 'r': # relocated
2031
# record the key for the real path.
2032
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2033
# absent rows are not present at any path.
2034
last_reference = current_old[0] not in all_remaining_keys
2036
# the current row consists entire of the current item (being marked
2037
# absent), and relocated or absent entries for the other trees:
2038
# Remove it, its meaningless.
2039
block = self._find_block(current_old[0])
2040
entry_index, present = self._find_entry_index(current_old[0], block[1])
2041
assert present, 'could not find entry for %s' % (current_old,)
2042
block[1].pop(entry_index)
2043
# if we have an id_index in use, remove this key from it for this id.
2044
if self._id_index is not None:
2045
self._id_index[current_old[0][2]].remove(current_old[0])
2046
# update all remaining keys for this id to record it as absent. The
2047
# existing details may either be the record we are making as deleted
2048
# (if there were other trees with the id present at this path), or may
2050
for update_key in all_remaining_keys:
2051
update_block_index, present = \
2052
self._find_block_index_from_key(update_key)
2053
assert present, 'could not find block for %s' % (update_key,)
2054
update_entry_index, present = \
2055
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2056
assert present, 'could not find entry for %s' % (update_key,)
2057
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2058
# it must not be absent at the moment
2059
assert update_tree_details[0][0] != 'a' # absent
2060
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2061
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2062
return last_reference
2064
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2065
packed_stat=None, size=0, path_utf8=None):
2066
"""Update an entry to the state in tree 0.
2068
This will either create a new entry at 'key' or update an existing one.
2069
It also makes sure that any other records which might mention this are
2072
:param key: (dir, name, file_id) for the new entry
2073
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2075
:param executable: Should the executable bit be set?
2076
:param fingerprint: Simple fingerprint for new entry.
2077
:param packed_stat: packed stat value for new entry.
2078
:param size: Size information for new entry
2079
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2082
block = self._find_block(key)[1]
2083
if packed_stat is None:
2084
packed_stat = DirState.NULLSTAT
2085
entry_index, present = self._find_entry_index(key, block)
2086
new_details = (minikind, fingerprint, size, executable, packed_stat)
2087
id_index = self._get_id_index()
2089
# new entry, synthesis cross reference here,
2090
existing_keys = id_index.setdefault(key[2], set())
2091
if not existing_keys:
2092
# not currently in the state, simplest case
2093
new_entry = key, [new_details] + self._empty_parent_info()
2095
# present at one or more existing other paths.
2096
# grab one of them and use it to generate parent
2097
# relocation/absent entries.
2098
new_entry = key, [new_details]
2099
for other_key in existing_keys:
2100
# change the record at other to be a pointer to this new
2101
# record. The loop looks similar to the change to
2102
# relocations when updating an existing record but its not:
2103
# the test for existing kinds is different: this can be
2104
# factored out to a helper though.
2105
other_block_index, present = self._find_block_index_from_key(other_key)
2107
import pdb; pdb.set_trace()
2108
assert present, 'could not find block for %s' % (other_key,)
2109
other_entry_index, present = self._find_entry_index(other_key,
2110
self._dirblocks[other_block_index][1])
2112
import pdb; pdb.set_trace()
2113
assert present, 'could not find entry for %s' % (other_key,)
2114
assert path_utf8 is not None
2115
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2116
('r', path_utf8, 0, False, '')
2118
num_present_parents = self._num_present_parents()
2119
for lookup_index in xrange(1, num_present_parents + 1):
2120
# grab any one entry, use it to find the right path.
2121
# TODO: optimise this to reduce memory use in highly
2122
# fragmented situations by reusing the relocation
2124
update_block_index, present = \
2125
self._find_block_index_from_key(other_key)
2126
assert present, 'could not find block for %s' % (other_key,)
2127
update_entry_index, present = \
2128
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2129
assert present, 'could not find entry for %s' % (other_key,)
2130
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2131
if update_details[0] in ('r', 'a'): # relocated, absent
2132
# its a pointer or absent in lookup_index's tree, use
2134
new_entry[1].append(update_details)
2136
# we have the right key, make a pointer to it.
2137
pointer_path = osutils.pathjoin(*other_key[0:2])
2138
new_entry[1].append(('r', pointer_path, 0, False, ''))
2139
block.insert(entry_index, new_entry)
2140
existing_keys.add(key)
2142
# Does the new state matter?
2143
block[entry_index][1][0] = new_details
2144
# parents cannot be affected by what we do.
2145
# other occurences of this id can be found
2146
# from the id index.
2148
# tree index consistency: All other paths for this id in this tree
2149
# index must point to the correct path. We have to loop here because
2150
# we may have passed entries in the state with this file id already
2151
# that were absent - where parent entries are - and they need to be
2152
# converted to relocated.
2153
assert path_utf8 is not None
2154
for entry_key in id_index.setdefault(key[2], set()):
2155
# TODO:PROFILING: It might be faster to just update
2156
# rather than checking if we need to, and then overwrite
2157
# the one we are located at.
2158
if entry_key != key:
2159
# this file id is at a different path in one of the
2160
# other trees, so put absent pointers there
2161
# This is the vertical axis in the matrix, all pointing
2163
block_index, present = self._find_block_index_from_key(entry_key)
2165
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2167
self._dirblocks[block_index][1][entry_index][1][0] = \
2168
('r', path_utf8, 0, False, '')
2169
# add a containing dirblock if needed.
2170
if new_details[0] == 'd':
2171
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2172
block_index, present = self._find_block_index_from_key(subdir_key)
2174
self._dirblocks.insert(block_index, (subdir_key[0], []))
2176
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2178
def _validate(self):
2179
"""Check that invariants on the dirblock are correct.
2181
This can be useful in debugging; it shouldn't be necessary in
2184
from pprint import pformat
2185
if len(self._dirblocks) > 0:
2186
assert self._dirblocks[0][0] == '', \
2187
"dirblocks don't start with root block:\n" + \
2189
if len(self._dirblocks) > 1:
2190
assert self._dirblocks[1][0] == '', \
2191
"dirblocks missing root directory:\n" + \
2193
# the dirblocks are sorted by their path components, name, and dir id
2194
dir_names = [d[0].split('/')
2195
for d in self._dirblocks[1:]]
2196
if dir_names != sorted(dir_names):
2197
raise AssertionError(
2198
"dir names are not in sorted order:\n" + \
2199
pformat(self._dirblocks) + \
2202
for dirblock in self._dirblocks:
2203
# within each dirblock, the entries are sorted by filename and
2205
assert dirblock[1] == sorted(dirblock[1]), \
2206
"dirblock for %r is not sorted:\n%s" % \
2207
(dirblock[0], pformat(dirblock))
2209
def _wipe_state(self):
2210
"""Forget all state information about the dirstate."""
2211
self._header_state = DirState.NOT_IN_MEMORY
2212
self._dirblock_state = DirState.NOT_IN_MEMORY
2215
self._dirblocks = []
2216
self._id_index = None
2217
self._end_of_header = None
2218
self._cutoff_time = None
2219
self._split_path_cache = {}
2221
def lock_read(self):
2222
"""Acquire a read lock on the dirstate"""
2223
if self._lock_token is not None:
2224
raise errors.LockContention(self._lock_token)
2225
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2226
# already in memory, we could read just the header and check for
2227
# any modification. If not modified, we can just leave things
2229
self._lock_token = lock.ReadLock(self._filename)
2230
self._lock_state = 'r'
2231
self._state_file = self._lock_token.f
2234
def lock_write(self):
2235
"""Acquire a write lock on the dirstate"""
2236
if self._lock_token is not None:
2237
raise errors.LockContention(self._lock_token)
2238
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2239
# already in memory, we could read just the header and check for
2240
# any modification. If not modified, we can just leave things
2242
self._lock_token = lock.WriteLock(self._filename)
2243
self._lock_state = 'w'
2244
self._state_file = self._lock_token.f
2248
"""Drop any locks held on the dirstate"""
2249
if self._lock_token is None:
2250
raise errors.LockNotHeld(self)
2251
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2252
# already in memory, we could read just the header and check for
2253
# any modification. If not modified, we can just leave things
2255
self._state_file = None
2256
self._lock_state = None
2257
self._lock_token.unlock()
2258
self._lock_token = None
2259
self._split_path_cache = {}
2261
def _requires_lock(self):
2262
"""Checks that a lock is currently held by someone on the dirstate"""
2263
if not self._lock_token:
2264
raise errors.ObjectNotLocked(self)
2267
def bisect_dirblock(dirblocks, dirname, lo=0, hi=None, cache={}):
2268
"""Return the index where to insert dirname into the dirblocks.
2270
The return value idx is such that all directories blocks in dirblock[:idx]
2271
have names < dirname, and all blocks in dirblock[idx:] have names >=
2274
Optional args lo (default 0) and hi (default len(dirblocks)) bound the
2275
slice of a to be searched.
2280
dirname_split = cache[dirname]
2282
dirname_split = dirname.split('/')
2283
cache[dirname] = dirname_split
2286
# Grab the dirname for the current dirblock
2287
cur = dirblocks[mid][0]
2289
cur_split = cache[cur]
2291
cur_split = cur.split('/')
2292
cache[cur] = cur_split
2293
if cur_split < dirname_split: lo = mid+1
2299
def pack_stat(st, _encode=base64.encodestring, _pack=struct.pack):
2300
"""Convert stat values into a packed representation."""
2301
# jam 20060614 it isn't really worth removing more entries if we
2302
# are going to leave it in packed form.
2303
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
2304
# With all entries filesize is 5.9M and read time is mabye 280ms
2305
# well within the noise margin
2307
# base64.encode always adds a final newline, so strip it off
2308
return _encode(_pack('>llllll'
2309
, st.st_size, int(st.st_mtime), int(st.st_ctime)
2310
, st.st_dev, st.st_ino, st.st_mode))[:-1]
added
removed
bzrlib/dirstate.py
