1
# Copyright (C) 2006, 2007, 2008 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
225
# This is the Windows equivalent of ENOTDIR
226
# It is defined in pywin32.winerror, but we don't want a strong dependency for
227
# just an error code.
228
ERROR_PATH_NOT_FOUND = 3
229
ERROR_DIRECTORY = 267
232
if not getattr(struct, '_compile', None):
233
# Cannot pre-compile the dirstate pack_stat
234
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
235
"""Convert stat values into a packed representation."""
236
return _encode(_pack('>LLLLLL', st.st_size, int(st.st_mtime),
237
int(st.st_ctime), st.st_dev, st.st_ino & 0xFFFFFFFF,
240
# compile the struct compiler we need, so as to only do it once
241
from _struct import Struct
242
_compiled_pack = Struct('>LLLLLL').pack
243
def pack_stat(st, _encode=binascii.b2a_base64, _pack=_compiled_pack):
244
"""Convert stat values into a packed representation."""
245
# jam 20060614 it isn't really worth removing more entries if we
246
# are going to leave it in packed form.
247
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
248
# With all entries, filesize is 5.9M and read time is maybe 280ms
249
# well within the noise margin
251
# base64 encoding always adds a final newline, so strip it off
252
# The current version
253
return _encode(_pack(st.st_size, int(st.st_mtime), int(st.st_ctime),
254
st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
255
# This is 0.060s / 1.520s faster by not encoding as much information
256
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
257
# This is not strictly faster than _encode(_pack())[:-1]
258
# return '%X.%X.%X.%X.%X.%X' % (
259
# st.st_size, int(st.st_mtime), int(st.st_ctime),
260
# st.st_dev, st.st_ino, st.st_mode)
261
# Similar to the _encode(_pack('>LL'))
262
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
265
class DirState(object):
266
"""Record directory and metadata state for fast access.
268
A dirstate is a specialised data structure for managing local working
269
tree state information. Its not yet well defined whether it is platform
270
specific, and if it is how we detect/parameterize that.
272
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
273
Unlike most bzr disk formats, DirStates must be locked for reading, using
274
lock_read. (This is an os file lock internally.) This is necessary
275
because the file can be rewritten in place.
277
DirStates must be explicitly written with save() to commit changes; just
278
unlocking them does not write the changes to disk.
281
_kind_to_minikind = {
287
'tree-reference': 't',
289
_minikind_to_kind = {
295
't': 'tree-reference',
297
_stat_to_minikind = {
302
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
303
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
306
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
307
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
308
# that we are sure a single record will always fit.
309
BISECT_PAGE_SIZE = 4096
312
IN_MEMORY_UNMODIFIED = 1
313
IN_MEMORY_MODIFIED = 2
315
# A pack_stat (the x's) that is just noise and will never match the output
318
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
320
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
321
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
323
def __init__(self, path):
324
"""Create a DirState object.
326
:param path: The path at which the dirstate file on disk should live.
328
# _header_state and _dirblock_state represent the current state
329
# of the dirstate metadata and the per-row data respectiely.
330
# NOT_IN_MEMORY indicates that no data is in memory
331
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
332
# is the same as is on disk
333
# IN_MEMORY_MODIFIED indicates that we have a modified version
334
# of what is on disk.
335
# In future we will add more granularity, for instance _dirblock_state
336
# will probably support partially-in-memory as a separate variable,
337
# allowing for partially-in-memory unmodified and partially-in-memory
339
self._header_state = DirState.NOT_IN_MEMORY
340
self._dirblock_state = DirState.NOT_IN_MEMORY
341
# If true, an error has been detected while updating the dirstate, and
342
# for safety we're not going to commit to disk.
343
self._changes_aborted = False
347
self._state_file = None
348
self._filename = path
349
self._lock_token = None
350
self._lock_state = None
351
self._id_index = None
352
# a map from packed_stat to sha's.
353
self._packed_stat_index = None
354
self._end_of_header = None
355
self._cutoff_time = None
356
self._split_path_cache = {}
357
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
358
if 'hashcache' in debug.debug_flags:
359
self._sha1_file = self._sha1_file_and_mutter
361
self._sha1_file = osutils.sha_file_by_name
362
# These two attributes provide a simple cache for lookups into the
363
# dirstate in-memory vectors. By probing respectively for the last
364
# block, and for the next entry, we save nearly 2 bisections per path
366
self._last_block_index = None
367
self._last_entry_index = None
371
(self.__class__.__name__, self._filename)
373
def add(self, path, file_id, kind, stat, fingerprint):
374
"""Add a path to be tracked.
376
:param path: The path within the dirstate - '' is the root, 'foo' is the
377
path foo within the root, 'foo/bar' is the path bar within foo
379
:param file_id: The file id of the path being added.
380
:param kind: The kind of the path, as a string like 'file',
382
:param stat: The output of os.lstat for the path.
383
:param fingerprint: The sha value of the file,
384
or the target of a symlink,
385
or the referenced revision id for tree-references,
386
or '' for directories.
389
# find the block its in.
390
# find the location in the block.
391
# check its not there
393
#------- copied from inventory.ensure_normalized_name - keep synced.
394
# --- normalized_filename wants a unicode basename only, so get one.
395
dirname, basename = osutils.split(path)
396
# we dont import normalized_filename directly because we want to be
397
# able to change the implementation at runtime for tests.
398
norm_name, can_access = osutils.normalized_filename(basename)
399
if norm_name != basename:
403
raise errors.InvalidNormalization(path)
404
# you should never have files called . or ..; just add the directory
405
# in the parent, or according to the special treatment for the root
406
if basename == '.' or basename == '..':
407
raise errors.InvalidEntryName(path)
408
# now that we've normalised, we need the correct utf8 path and
409
# dirname and basename elements. This single encode and split should be
410
# faster than three separate encodes.
411
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
412
dirname, basename = osutils.split(utf8path)
413
# uses __class__ for speed; the check is needed for safety
414
if file_id.__class__ is not str:
415
raise AssertionError(
416
"must be a utf8 file_id not %s" % (type(file_id), ))
417
# Make sure the file_id does not exist in this tree
418
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
419
if file_id_entry != (None, None):
420
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
421
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
422
info = '%s:%s' % (kind, path)
423
raise errors.DuplicateFileId(file_id, info)
424
first_key = (dirname, basename, '')
425
block_index, present = self._find_block_index_from_key(first_key)
427
# check the path is not in the tree
428
block = self._dirblocks[block_index][1]
429
entry_index, _ = self._find_entry_index(first_key, block)
430
while (entry_index < len(block) and
431
block[entry_index][0][0:2] == first_key[0:2]):
432
if block[entry_index][1][0][0] not in 'ar':
433
# this path is in the dirstate in the current tree.
434
raise Exception, "adding already added path!"
437
# The block where we want to put the file is not present. But it
438
# might be because the directory was empty, or not loaded yet. Look
439
# for a parent entry, if not found, raise NotVersionedError
440
parent_dir, parent_base = osutils.split(dirname)
441
parent_block_idx, parent_entry_idx, _, parent_present = \
442
self._get_block_entry_index(parent_dir, parent_base, 0)
443
if not parent_present:
444
raise errors.NotVersionedError(path, str(self))
445
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
446
block = self._dirblocks[block_index][1]
447
entry_key = (dirname, basename, file_id)
450
packed_stat = DirState.NULLSTAT
453
packed_stat = pack_stat(stat)
454
parent_info = self._empty_parent_info()
455
minikind = DirState._kind_to_minikind[kind]
457
entry_data = entry_key, [
458
(minikind, fingerprint, size, False, packed_stat),
460
elif kind == 'directory':
461
entry_data = entry_key, [
462
(minikind, '', 0, False, packed_stat),
464
elif kind == 'symlink':
465
entry_data = entry_key, [
466
(minikind, fingerprint, size, False, packed_stat),
468
elif kind == 'tree-reference':
469
entry_data = entry_key, [
470
(minikind, fingerprint, 0, False, packed_stat),
473
raise errors.BzrError('unknown kind %r' % kind)
474
entry_index, present = self._find_entry_index(entry_key, block)
476
block.insert(entry_index, entry_data)
478
if block[entry_index][1][0][0] != 'a':
479
raise AssertionError(" %r(%r) already added" % (basename, file_id))
480
block[entry_index][1][0] = entry_data[1][0]
482
if kind == 'directory':
483
# insert a new dirblock
484
self._ensure_block(block_index, entry_index, utf8path)
485
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
487
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
489
def _bisect(self, paths):
490
"""Bisect through the disk structure for specific rows.
492
:param paths: A list of paths to find
493
:return: A dict mapping path => entries for found entries. Missing
494
entries will not be in the map.
495
The list is not sorted, and entries will be populated
496
based on when they were read.
498
self._requires_lock()
499
# We need the file pointer to be right after the initial header block
500
self._read_header_if_needed()
501
# If _dirblock_state was in memory, we should just return info from
502
# there, this function is only meant to handle when we want to read
504
if self._dirblock_state != DirState.NOT_IN_MEMORY:
505
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
507
# The disk representation is generally info + '\0\n\0' at the end. But
508
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
509
# Because it means we can sync on the '\n'
510
state_file = self._state_file
511
file_size = os.fstat(state_file.fileno()).st_size
512
# We end up with 2 extra fields, we should have a trailing '\n' to
513
# ensure that we read the whole record, and we should have a precursur
514
# '' which ensures that we start after the previous '\n'
515
entry_field_count = self._fields_per_entry() + 1
517
low = self._end_of_header
518
high = file_size - 1 # Ignore the final '\0'
519
# Map from (dir, name) => entry
522
# Avoid infinite seeking
523
max_count = 30*len(paths)
525
# pending is a list of places to look.
526
# each entry is a tuple of low, high, dir_names
527
# low -> the first byte offset to read (inclusive)
528
# high -> the last byte offset (inclusive)
529
# dir_names -> The list of (dir, name) pairs that should be found in
530
# the [low, high] range
531
pending = [(low, high, paths)]
533
page_size = self._bisect_page_size
535
fields_to_entry = self._get_fields_to_entry()
538
low, high, cur_files = pending.pop()
540
if not cur_files or low >= high:
545
if count > max_count:
546
raise errors.BzrError('Too many seeks, most likely a bug.')
548
mid = max(low, (low+high-page_size)/2)
551
# limit the read size, so we don't end up reading data that we have
553
read_size = min(page_size, (high-mid)+1)
554
block = state_file.read(read_size)
557
entries = block.split('\n')
560
# We didn't find a '\n', so we cannot have found any records.
561
# So put this range back and try again. But we know we have to
562
# increase the page size, because a single read did not contain
563
# a record break (so records must be larger than page_size)
565
pending.append((low, high, cur_files))
568
# Check the first and last entries, in case they are partial, or if
569
# we don't care about the rest of this page
571
first_fields = entries[0].split('\0')
572
if len(first_fields) < entry_field_count:
573
# We didn't get the complete first entry
574
# so move start, and grab the next, which
575
# should be a full entry
576
start += len(entries[0])+1
577
first_fields = entries[1].split('\0')
580
if len(first_fields) <= 2:
581
# We didn't even get a filename here... what do we do?
582
# Try a large page size and repeat this query
584
pending.append((low, high, cur_files))
587
# Find what entries we are looking for, which occur before and
588
# after this first record.
591
first_path = first_fields[1] + '/' + first_fields[2]
593
first_path = first_fields[2]
594
first_loc = _bisect_path_left(cur_files, first_path)
596
# These exist before the current location
597
pre = cur_files[:first_loc]
598
# These occur after the current location, which may be in the
599
# data we read, or might be after the last entry
600
post = cur_files[first_loc:]
602
if post and len(first_fields) >= entry_field_count:
603
# We have files after the first entry
605
# Parse the last entry
606
last_entry_num = len(entries)-1
607
last_fields = entries[last_entry_num].split('\0')
608
if len(last_fields) < entry_field_count:
609
# The very last hunk was not complete,
610
# read the previous hunk
611
after = mid + len(block) - len(entries[-1])
613
last_fields = entries[last_entry_num].split('\0')
615
after = mid + len(block)
618
last_path = last_fields[1] + '/' + last_fields[2]
620
last_path = last_fields[2]
621
last_loc = _bisect_path_right(post, last_path)
623
middle_files = post[:last_loc]
624
post = post[last_loc:]
627
# We have files that should occur in this block
628
# (>= first, <= last)
629
# Either we will find them here, or we can mark them as
632
if middle_files[0] == first_path:
633
# We might need to go before this location
634
pre.append(first_path)
635
if middle_files[-1] == last_path:
636
post.insert(0, last_path)
638
# Find out what paths we have
639
paths = {first_path:[first_fields]}
640
# last_path might == first_path so we need to be
641
# careful if we should append rather than overwrite
642
if last_entry_num != first_entry_num:
643
paths.setdefault(last_path, []).append(last_fields)
644
for num in xrange(first_entry_num+1, last_entry_num):
645
# TODO: jam 20070223 We are already splitting here, so
646
# shouldn't we just split the whole thing rather
647
# than doing the split again in add_one_record?
648
fields = entries[num].split('\0')
650
path = fields[1] + '/' + fields[2]
653
paths.setdefault(path, []).append(fields)
655
for path in middle_files:
656
for fields in paths.get(path, []):
657
# offset by 1 because of the opening '\0'
658
# consider changing fields_to_entry to avoid the
660
entry = fields_to_entry(fields[1:])
661
found.setdefault(path, []).append(entry)
663
# Now we have split up everything into pre, middle, and post, and
664
# we have handled everything that fell in 'middle'.
665
# We add 'post' first, so that we prefer to seek towards the
666
# beginning, so that we will tend to go as early as we need, and
667
# then only seek forward after that.
669
pending.append((after, high, post))
671
pending.append((low, start-1, pre))
673
# Consider that we may want to return the directory entries in sorted
674
# order. For now, we just return them in whatever order we found them,
675
# and leave it up to the caller if they care if it is ordered or not.
678
def _bisect_dirblocks(self, dir_list):
679
"""Bisect through the disk structure to find entries in given dirs.
681
_bisect_dirblocks is meant to find the contents of directories, which
682
differs from _bisect, which only finds individual entries.
684
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
685
:return: A map from dir => entries_for_dir
687
# TODO: jam 20070223 A lot of the bisecting logic could be shared
688
# between this and _bisect. It would require parameterizing the
689
# inner loop with a function, though. We should evaluate the
690
# performance difference.
691
self._requires_lock()
692
# We need the file pointer to be right after the initial header block
693
self._read_header_if_needed()
694
# If _dirblock_state was in memory, we should just return info from
695
# there, this function is only meant to handle when we want to read
697
if self._dirblock_state != DirState.NOT_IN_MEMORY:
698
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
699
# The disk representation is generally info + '\0\n\0' at the end. But
700
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
701
# Because it means we can sync on the '\n'
702
state_file = self._state_file
703
file_size = os.fstat(state_file.fileno()).st_size
704
# We end up with 2 extra fields, we should have a trailing '\n' to
705
# ensure that we read the whole record, and we should have a precursur
706
# '' which ensures that we start after the previous '\n'
707
entry_field_count = self._fields_per_entry() + 1
709
low = self._end_of_header
710
high = file_size - 1 # Ignore the final '\0'
711
# Map from dir => entry
714
# Avoid infinite seeking
715
max_count = 30*len(dir_list)
717
# pending is a list of places to look.
718
# each entry is a tuple of low, high, dir_names
719
# low -> the first byte offset to read (inclusive)
720
# high -> the last byte offset (inclusive)
721
# dirs -> The list of directories that should be found in
722
# the [low, high] range
723
pending = [(low, high, dir_list)]
725
page_size = self._bisect_page_size
727
fields_to_entry = self._get_fields_to_entry()
730
low, high, cur_dirs = pending.pop()
732
if not cur_dirs or low >= high:
737
if count > max_count:
738
raise errors.BzrError('Too many seeks, most likely a bug.')
740
mid = max(low, (low+high-page_size)/2)
743
# limit the read size, so we don't end up reading data that we have
745
read_size = min(page_size, (high-mid)+1)
746
block = state_file.read(read_size)
749
entries = block.split('\n')
752
# We didn't find a '\n', so we cannot have found any records.
753
# So put this range back and try again. But we know we have to
754
# increase the page size, because a single read did not contain
755
# a record break (so records must be larger than page_size)
757
pending.append((low, high, cur_dirs))
760
# Check the first and last entries, in case they are partial, or if
761
# we don't care about the rest of this page
763
first_fields = entries[0].split('\0')
764
if len(first_fields) < entry_field_count:
765
# We didn't get the complete first entry
766
# so move start, and grab the next, which
767
# should be a full entry
768
start += len(entries[0])+1
769
first_fields = entries[1].split('\0')
772
if len(first_fields) <= 1:
773
# We didn't even get a dirname here... what do we do?
774
# Try a large page size and repeat this query
776
pending.append((low, high, cur_dirs))
779
# Find what entries we are looking for, which occur before and
780
# after this first record.
782
first_dir = first_fields[1]
783
first_loc = bisect.bisect_left(cur_dirs, first_dir)
785
# These exist before the current location
786
pre = cur_dirs[:first_loc]
787
# These occur after the current location, which may be in the
788
# data we read, or might be after the last entry
789
post = cur_dirs[first_loc:]
791
if post and len(first_fields) >= entry_field_count:
792
# We have records to look at after the first entry
794
# Parse the last entry
795
last_entry_num = len(entries)-1
796
last_fields = entries[last_entry_num].split('\0')
797
if len(last_fields) < entry_field_count:
798
# The very last hunk was not complete,
799
# read the previous hunk
800
after = mid + len(block) - len(entries[-1])
802
last_fields = entries[last_entry_num].split('\0')
804
after = mid + len(block)
806
last_dir = last_fields[1]
807
last_loc = bisect.bisect_right(post, last_dir)
809
middle_files = post[:last_loc]
810
post = post[last_loc:]
813
# We have files that should occur in this block
814
# (>= first, <= last)
815
# Either we will find them here, or we can mark them as
818
if middle_files[0] == first_dir:
819
# We might need to go before this location
820
pre.append(first_dir)
821
if middle_files[-1] == last_dir:
822
post.insert(0, last_dir)
824
# Find out what paths we have
825
paths = {first_dir:[first_fields]}
826
# last_dir might == first_dir so we need to be
827
# careful if we should append rather than overwrite
828
if last_entry_num != first_entry_num:
829
paths.setdefault(last_dir, []).append(last_fields)
830
for num in xrange(first_entry_num+1, last_entry_num):
831
# TODO: jam 20070223 We are already splitting here, so
832
# shouldn't we just split the whole thing rather
833
# than doing the split again in add_one_record?
834
fields = entries[num].split('\0')
835
paths.setdefault(fields[1], []).append(fields)
837
for cur_dir in middle_files:
838
for fields in paths.get(cur_dir, []):
839
# offset by 1 because of the opening '\0'
840
# consider changing fields_to_entry to avoid the
842
entry = fields_to_entry(fields[1:])
843
found.setdefault(cur_dir, []).append(entry)
845
# Now we have split up everything into pre, middle, and post, and
846
# we have handled everything that fell in 'middle'.
847
# We add 'post' first, so that we prefer to seek towards the
848
# beginning, so that we will tend to go as early as we need, and
849
# then only seek forward after that.
851
pending.append((after, high, post))
853
pending.append((low, start-1, pre))
857
def _bisect_recursive(self, paths):
858
"""Bisect for entries for all paths and their children.
860
This will use bisect to find all records for the supplied paths. It
861
will then continue to bisect for any records which are marked as
862
directories. (and renames?)
864
:param paths: A sorted list of (dir, name) pairs
865
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
866
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
868
# Map from (dir, name, file_id) => [tree_info]
871
found_dir_names = set()
873
# Directories that have been read
874
processed_dirs = set()
875
# Get the ball rolling with the first bisect for all entries.
876
newly_found = self._bisect(paths)
879
# Directories that need to be read
881
paths_to_search = set()
882
for entry_list in newly_found.itervalues():
883
for dir_name_id, trees_info in entry_list:
884
found[dir_name_id] = trees_info
885
found_dir_names.add(dir_name_id[:2])
887
for tree_info in trees_info:
888
minikind = tree_info[0]
891
# We already processed this one as a directory,
892
# we don't need to do the extra work again.
894
subdir, name, file_id = dir_name_id
895
path = osutils.pathjoin(subdir, name)
897
if path not in processed_dirs:
898
pending_dirs.add(path)
899
elif minikind == 'r':
900
# Rename, we need to directly search the target
901
# which is contained in the fingerprint column
902
dir_name = osutils.split(tree_info[1])
903
if dir_name[0] in pending_dirs:
904
# This entry will be found in the dir search
906
if dir_name not in found_dir_names:
907
paths_to_search.add(tree_info[1])
908
# Now we have a list of paths to look for directly, and
909
# directory blocks that need to be read.
910
# newly_found is mixing the keys between (dir, name) and path
911
# entries, but that is okay, because we only really care about the
913
newly_found = self._bisect(sorted(paths_to_search))
914
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
915
processed_dirs.update(pending_dirs)
918
def _discard_merge_parents(self):
919
"""Discard any parents trees beyond the first.
921
Note that if this fails the dirstate is corrupted.
923
After this function returns the dirstate contains 2 trees, neither of
926
self._read_header_if_needed()
927
parents = self.get_parent_ids()
930
# only require all dirblocks if we are doing a full-pass removal.
931
self._read_dirblocks_if_needed()
932
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
933
def iter_entries_removable():
934
for block in self._dirblocks:
935
deleted_positions = []
936
for pos, entry in enumerate(block[1]):
938
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
939
deleted_positions.append(pos)
940
if deleted_positions:
941
if len(deleted_positions) == len(block[1]):
944
for pos in reversed(deleted_positions):
946
# if the first parent is a ghost:
947
if parents[0] in self.get_ghosts():
948
empty_parent = [DirState.NULL_PARENT_DETAILS]
949
for entry in iter_entries_removable():
950
entry[1][1:] = empty_parent
952
for entry in iter_entries_removable():
956
self._parents = [parents[0]]
957
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
958
self._header_state = DirState.IN_MEMORY_MODIFIED
960
def _empty_parent_info(self):
961
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
964
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
965
"""Ensure a block for dirname exists.
967
This function exists to let callers which know that there is a
968
directory dirname ensure that the block for it exists. This block can
969
fail to exist because of demand loading, or because a directory had no
970
children. In either case it is not an error. It is however an error to
971
call this if there is no parent entry for the directory, and thus the
972
function requires the coordinates of such an entry to be provided.
974
The root row is special cased and can be indicated with a parent block
977
:param parent_block_index: The index of the block in which dirname's row
979
:param parent_row_index: The index in the parent block where the row
981
:param dirname: The utf8 dirname to ensure there is a block for.
982
:return: The index for the block.
984
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
985
# This is the signature of the root row, and the
986
# contents-of-root row is always index 1
988
# the basename of the directory must be the end of its full name.
989
if not (parent_block_index == -1 and
990
parent_block_index == -1 and dirname == ''):
991
if not dirname.endswith(
992
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
993
raise AssertionError("bad dirname %r" % dirname)
994
block_index, present = self._find_block_index_from_key((dirname, '', ''))
996
## In future, when doing partial parsing, this should load and
997
# populate the entire block.
998
self._dirblocks.insert(block_index, (dirname, []))
1001
def _entries_to_current_state(self, new_entries):
1002
"""Load new_entries into self.dirblocks.
1004
Process new_entries into the current state object, making them the active
1005
state. The entries are grouped together by directory to form dirblocks.
1007
:param new_entries: A sorted list of entries. This function does not sort
1008
to prevent unneeded overhead when callers have a sorted list already.
1011
if new_entries[0][0][0:2] != ('', ''):
1012
raise AssertionError(
1013
"Missing root row %r" % (new_entries[0][0],))
1014
# The two blocks here are deliberate: the root block and the
1015
# contents-of-root block.
1016
self._dirblocks = [('', []), ('', [])]
1017
current_block = self._dirblocks[0][1]
1018
current_dirname = ''
1020
append_entry = current_block.append
1021
for entry in new_entries:
1022
if entry[0][0] != current_dirname:
1023
# new block - different dirname
1025
current_dirname = entry[0][0]
1026
self._dirblocks.append((current_dirname, current_block))
1027
append_entry = current_block.append
1028
# append the entry to the current block
1030
self._split_root_dirblock_into_contents()
1032
def _split_root_dirblock_into_contents(self):
1033
"""Split the root dirblocks into root and contents-of-root.
1035
After parsing by path, we end up with root entries and contents-of-root
1036
entries in the same block. This loop splits them out again.
1038
# The above loop leaves the "root block" entries mixed with the
1039
# "contents-of-root block". But we don't want an if check on
1040
# all entries, so instead we just fix it up here.
1041
if self._dirblocks[1] != ('', []):
1042
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1044
contents_of_root_block = []
1045
for entry in self._dirblocks[0][1]:
1046
if not entry[0][1]: # This is a root entry
1047
root_block.append(entry)
1049
contents_of_root_block.append(entry)
1050
self._dirblocks[0] = ('', root_block)
1051
self._dirblocks[1] = ('', contents_of_root_block)
1053
def _entries_for_path(self, path):
1054
"""Return a list with all the entries that match path for all ids."""
1055
dirname, basename = os.path.split(path)
1056
key = (dirname, basename, '')
1057
block_index, present = self._find_block_index_from_key(key)
1059
# the block which should contain path is absent.
1062
block = self._dirblocks[block_index][1]
1063
entry_index, _ = self._find_entry_index(key, block)
1064
# we may need to look at multiple entries at this path: walk while the specific_files match.
1065
while (entry_index < len(block) and
1066
block[entry_index][0][0:2] == key[0:2]):
1067
result.append(block[entry_index])
1071
def _entry_to_line(self, entry):
1072
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1074
:param entry: An entry_tuple as defined in the module docstring.
1076
entire_entry = list(entry[0])
1077
for tree_number, tree_data in enumerate(entry[1]):
1078
# (minikind, fingerprint, size, executable, tree_specific_string)
1079
entire_entry.extend(tree_data)
1080
# 3 for the key, 5 for the fields per tree.
1081
tree_offset = 3 + tree_number * 5
1083
entire_entry[tree_offset + 0] = tree_data[0]
1085
entire_entry[tree_offset + 2] = str(tree_data[2])
1087
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1088
return '\0'.join(entire_entry)
1090
def _fields_per_entry(self):
1091
"""How many null separated fields should be in each entry row.
1093
Each line now has an extra '\n' field which is not used
1094
so we just skip over it
1096
3 fields for the key
1097
+ number of fields per tree_data (5) * tree count
1100
tree_count = 1 + self._num_present_parents()
1101
return 3 + 5 * tree_count + 1
1103
def _find_block(self, key, add_if_missing=False):
1104
"""Return the block that key should be present in.
1106
:param key: A dirstate entry key.
1107
:return: The block tuple.
1109
block_index, present = self._find_block_index_from_key(key)
1111
if not add_if_missing:
1112
# check to see if key is versioned itself - we might want to
1113
# add it anyway, because dirs with no entries dont get a
1114
# dirblock at parse time.
1115
# This is an uncommon branch to take: most dirs have children,
1116
# and most code works with versioned paths.
1117
parent_base, parent_name = osutils.split(key[0])
1118
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1119
# some parent path has not been added - its an error to add
1121
raise errors.NotVersionedError(key[0:2], str(self))
1122
self._dirblocks.insert(block_index, (key[0], []))
1123
return self._dirblocks[block_index]
1125
def _find_block_index_from_key(self, key):
1126
"""Find the dirblock index for a key.
1128
:return: The block index, True if the block for the key is present.
1130
if key[0:2] == ('', ''):
1133
if (self._last_block_index is not None and
1134
self._dirblocks[self._last_block_index][0] == key[0]):
1135
return self._last_block_index, True
1138
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1139
cache=self._split_path_cache)
1140
# _right returns one-past-where-key is so we have to subtract
1141
# one to use it. we use _right here because there are two
1142
# '' blocks - the root, and the contents of root
1143
# we always have a minimum of 2 in self._dirblocks: root and
1144
# root-contents, and for '', we get 2 back, so this is
1145
# simple and correct:
1146
present = (block_index < len(self._dirblocks) and
1147
self._dirblocks[block_index][0] == key[0])
1148
self._last_block_index = block_index
1149
# Reset the entry index cache to the beginning of the block.
1150
self._last_entry_index = -1
1151
return block_index, present
1153
def _find_entry_index(self, key, block):
1154
"""Find the entry index for a key in a block.
1156
:return: The entry index, True if the entry for the key is present.
1158
len_block = len(block)
1160
if self._last_entry_index is not None:
1162
entry_index = self._last_entry_index + 1
1163
# A hit is when the key is after the last slot, and before or
1164
# equal to the next slot.
1165
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1166
key <= block[entry_index][0]):
1167
self._last_entry_index = entry_index
1168
present = (block[entry_index][0] == key)
1169
return entry_index, present
1172
entry_index = bisect.bisect_left(block, (key, []))
1173
present = (entry_index < len_block and
1174
block[entry_index][0] == key)
1175
self._last_entry_index = entry_index
1176
return entry_index, present
1179
def from_tree(tree, dir_state_filename):
1180
"""Create a dirstate from a bzr Tree.
1182
:param tree: The tree which should provide parent information and
1184
:return: a DirState object which is currently locked for writing.
1185
(it was locked by DirState.initialize)
1187
result = DirState.initialize(dir_state_filename)
1191
parent_ids = tree.get_parent_ids()
1192
num_parents = len(parent_ids)
1194
for parent_id in parent_ids:
1195
parent_tree = tree.branch.repository.revision_tree(parent_id)
1196
parent_trees.append((parent_id, parent_tree))
1197
parent_tree.lock_read()
1198
result.set_parent_trees(parent_trees, [])
1199
result.set_state_from_inventory(tree.inventory)
1201
for revid, parent_tree in parent_trees:
1202
parent_tree.unlock()
1205
# The caller won't have a chance to unlock this, so make sure we
1211
def update_by_delta(self, delta):
1212
"""Apply an inventory delta to the dirstate for tree 0
1214
:param delta: An inventory delta. See Inventory.apply_delta for
1217
self._read_dirblocks_if_needed()
1220
for old_path, new_path, file_id, inv_entry in sorted(delta, reverse=True):
1221
if (file_id in insertions) or (file_id in removals):
1222
raise AssertionError("repeated file id in delta %r" % (file_id,))
1223
if old_path is not None:
1224
old_path = old_path.encode('utf-8')
1225
removals[file_id] = old_path
1226
if new_path is not None:
1227
new_path = new_path.encode('utf-8')
1228
dirname, basename = osutils.split(new_path)
1229
key = (dirname, basename, file_id)
1230
minikind = DirState._kind_to_minikind[inv_entry.kind]
1232
fingerprint = inv_entry.reference_revision
1235
insertions[file_id] = (key, minikind, inv_entry.executable,
1236
fingerprint, new_path)
1237
# Transform moves into delete+add pairs
1238
if None not in (old_path, new_path):
1239
for child in self._iter_child_entries(0, old_path):
1240
if child[0][2] in insertions or child[0][2] in removals:
1242
child_dirname = child[0][0]
1243
child_basename = child[0][1]
1244
minikind = child[1][0][0]
1245
fingerprint = child[1][0][4]
1246
executable = child[1][0][3]
1247
old_child_path = osutils.pathjoin(child[0][0],
1249
removals[child[0][2]] = old_child_path
1250
child_suffix = child_dirname[len(old_path):]
1251
new_child_dirname = (new_path + child_suffix)
1252
key = (new_child_dirname, child_basename, child[0][2])
1253
new_child_path = os.path.join(new_child_dirname,
1255
insertions[child[0][2]] = (key, minikind, executable,
1256
fingerprint, new_child_path)
1257
self._apply_removals(removals.values())
1258
self._apply_insertions(insertions.values())
1260
def _apply_removals(self, removals):
1261
for path in sorted(removals, reverse=True):
1262
dirname, basename = osutils.split(path)
1263
block_i, entry_i, d_present, f_present = \
1264
self._get_block_entry_index(dirname, basename, 0)
1265
entry = self._dirblocks[block_i][1][entry_i]
1266
self._make_absent(entry)
1267
# See if we have a malformed delta: deleting a directory must not
1268
# leave crud behind. This increases the number of bisects needed
1269
# substantially, but deletion or renames of large numbers of paths
1270
# is rare enough it shouldn't be an issue (famous last words?) RBC
1272
block_i, entry_i, d_present, f_present = \
1273
self._get_block_entry_index(path, '', 0)
1275
# The dir block is still present in the dirstate; this could
1276
# be due to it being in a parent tree, or a corrupt delta.
1277
for child_entry in self._dirblocks[block_i][1]:
1278
if child_entry[1][0][0] not in ('r', 'a'):
1279
raise errors.InconsistentDelta(path, entry[0][2],
1280
"The file id was deleted but its children were "
1283
def _apply_insertions(self, adds):
1284
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1285
self.update_minimal(key, minikind, executable, fingerprint,
1286
path_utf8=path_utf8)
1288
def update_basis_by_delta(self, delta, new_revid):
1289
"""Update the parents of this tree after a commit.
1291
This gives the tree one parent, with revision id new_revid. The
1292
inventory delta is applied to the current basis tree to generate the
1293
inventory for the parent new_revid, and all other parent trees are
1296
Note that an exception during the operation of this method will leave
1297
the dirstate in a corrupt state where it should not be saved.
1299
Finally, we expect all changes to be synchronising the basis tree with
1302
:param new_revid: The new revision id for the trees parent.
1303
:param delta: An inventory delta (see apply_inventory_delta) describing
1304
the changes from the current left most parent revision to new_revid.
1306
self._read_dirblocks_if_needed()
1307
self._discard_merge_parents()
1308
if self._ghosts != []:
1309
raise NotImplementedError(self.update_basis_by_delta)
1310
if len(self._parents) == 0:
1311
# setup a blank tree, the most simple way.
1312
empty_parent = DirState.NULL_PARENT_DETAILS
1313
for entry in self._iter_entries():
1314
entry[1].append(empty_parent)
1315
self._parents.append(new_revid)
1317
self._parents[0] = new_revid
1319
delta = sorted(delta, reverse=True)
1323
# The paths this function accepts are unicode and must be encoded as we
1325
encode = cache_utf8.encode
1326
inv_to_entry = self._inv_entry_to_details
1327
# delta is now (deletes, changes), (adds) in reverse lexographical
1329
# deletes in reverse lexographic order are safe to process in situ.
1330
# renames are not, as a rename from any path could go to a path
1331
# lexographically lower, so we transform renames into delete, add pairs,
1332
# expanding them recursively as needed.
1333
# At the same time, to reduce interface friction we convert the input
1334
# inventory entries to dirstate.
1335
root_only = ('', '')
1336
for old_path, new_path, file_id, inv_entry in delta:
1337
if old_path is None:
1338
adds.append((None, encode(new_path), file_id,
1339
inv_to_entry(inv_entry), True))
1340
elif new_path is None:
1341
deletes.append((encode(old_path), None, file_id, None, True))
1342
elif (old_path, new_path) != root_only:
1344
# Because renames must preserve their children we must have
1345
# processed all relocations and removes before hand. The sort
1346
# order ensures we've examined the child paths, but we also
1347
# have to execute the removals, or the split to an add/delete
1348
# pair will result in the deleted item being reinserted, or
1349
# renamed items being reinserted twice - and possibly at the
1350
# wrong place. Splitting into a delete/add pair also simplifies
1351
# the handling of entries with ('f', ...), ('r' ...) because
1352
# the target of the 'r' is old_path here, and we add that to
1353
# deletes, meaning that the add handler does not need to check
1354
# for 'r' items on every pass.
1355
self._update_basis_apply_deletes(deletes)
1357
new_path_utf8 = encode(new_path)
1358
# Split into an add/delete pair recursively.
1359
adds.append((None, new_path_utf8, file_id,
1360
inv_to_entry(inv_entry), False))
1361
# Expunge deletes that we've seen so that deleted/renamed
1362
# children of a rename directory are handled correctly.
1363
new_deletes = reversed(list(self._iter_child_entries(1,
1365
# Remove the current contents of the tree at orig_path, and
1366
# reinsert at the correct new path.
1367
for entry in new_deletes:
1369
source_path = entry[0][0] + '/' + entry[0][1]
1371
source_path = entry[0][1]
1373
target_path = new_path_utf8 + source_path[len(old_path):]
1376
raise AssertionError("cannot rename directory to"
1378
target_path = source_path[len(old_path) + 1:]
1379
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1381
(source_path, target_path, entry[0][2], None, False))
1383
(encode(old_path), new_path, file_id, None, False))
1385
# changes to just the root should not require remove/insertion
1387
changes.append((encode(old_path), encode(new_path), file_id,
1388
inv_to_entry(inv_entry)))
1390
# Finish expunging deletes/first half of renames.
1391
self._update_basis_apply_deletes(deletes)
1392
# Reinstate second half of renames and new paths.
1393
self._update_basis_apply_adds(adds)
1394
# Apply in-situ changes.
1395
self._update_basis_apply_changes(changes)
1397
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1398
self._header_state = DirState.IN_MEMORY_MODIFIED
1399
self._id_index = None
1402
def _update_basis_apply_adds(self, adds):
1403
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1405
They may be adds, or renames that have been split into add/delete
1408
:param adds: A sequence of adds. Each add is a tuple:
1409
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1410
is False when the add is the second half of a remove-and-reinsert
1411
pair created to handle renames and deletes.
1413
# Adds are accumulated partly from renames, so can be in any input
1416
# adds is now in lexographic order, which places all parents before
1417
# their children, so we can process it linearly.
1419
for old_path, new_path, file_id, new_details, real_add in adds:
1420
# the entry for this file_id must be in tree 0.
1421
entry = self._get_entry(0, file_id, new_path)
1422
if entry[0] is None or entry[0][2] != file_id:
1423
self._changes_aborted = True
1424
raise errors.InconsistentDelta(new_path, file_id,
1425
'working tree does not contain new entry')
1426
if real_add and entry[1][1][0] not in absent:
1427
self._changes_aborted = True
1428
raise errors.InconsistentDelta(new_path, file_id,
1429
'The entry was considered to be a genuinely new record,'
1430
' but there was already an old record for it.')
1431
# We don't need to update the target of an 'r' because the handling
1432
# of renames turns all 'r' situations into a delete at the original
1434
entry[1][1] = new_details
1436
def _update_basis_apply_changes(self, changes):
1437
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1439
:param adds: A sequence of changes. Each change is a tuple:
1440
(path_utf8, path_utf8, file_id, (entry_details))
1443
for old_path, new_path, file_id, new_details in changes:
1444
# the entry for this file_id must be in tree 0.
1445
entry = self._get_entry(0, file_id, new_path)
1446
if entry[0] is None or entry[0][2] != file_id:
1447
self._changes_aborted = True
1448
raise errors.InconsistentDelta(new_path, file_id,
1449
'working tree does not contain new entry')
1450
if (entry[1][0][0] in absent or
1451
entry[1][1][0] in absent):
1452
self._changes_aborted = True
1453
raise errors.InconsistentDelta(new_path, file_id,
1454
'changed considered absent')
1455
entry[1][1] = new_details
1457
def _update_basis_apply_deletes(self, deletes):
1458
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1460
They may be deletes, or renames that have been split into add/delete
1463
:param deletes: A sequence of deletes. Each delete is a tuple:
1464
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1465
real_delete is True when the desired outcome is an actual deletion
1466
rather than the rename handling logic temporarily deleting a path
1467
during the replacement of a parent.
1469
null = DirState.NULL_PARENT_DETAILS
1470
for old_path, new_path, file_id, _, real_delete in deletes:
1471
if real_delete != (new_path is None):
1472
raise AssertionError("bad delete delta")
1473
# the entry for this file_id must be in tree 1.
1474
dirname, basename = osutils.split(old_path)
1475
block_index, entry_index, dir_present, file_present = \
1476
self._get_block_entry_index(dirname, basename, 1)
1477
if not file_present:
1478
self._changes_aborted = True
1479
raise errors.InconsistentDelta(old_path, file_id,
1480
'basis tree does not contain removed entry')
1481
entry = self._dirblocks[block_index][1][entry_index]
1482
if entry[0][2] != file_id:
1483
self._changes_aborted = True
1484
raise errors.InconsistentDelta(old_path, file_id,
1485
'mismatched file_id in tree 1')
1487
if entry[1][0][0] != 'a':
1488
self._changes_aborted = True
1489
raise errors.InconsistentDelta(old_path, file_id,
1490
'This was marked as a real delete, but the WT state'
1491
' claims that it still exists and is versioned.')
1492
del self._dirblocks[block_index][1][entry_index]
1494
if entry[1][0][0] == 'a':
1495
self._changes_aborted = True
1496
raise errors.InconsistentDelta(old_path, file_id,
1497
'The entry was considered a rename, but the source path'
1498
' is marked as absent.')
1499
# For whatever reason, we were asked to rename an entry
1500
# that was originally marked as deleted. This could be
1501
# because we are renaming the parent directory, and the WT
1502
# current state has the file marked as deleted.
1503
elif entry[1][0][0] == 'r':
1504
# implement the rename
1505
del self._dirblocks[block_index][1][entry_index]
1507
# it is being resurrected here, so blank it out temporarily.
1508
self._dirblocks[block_index][1][entry_index][1][1] = null
1510
def _observed_sha1(self, entry, sha1, stat_value,
1511
_stat_to_minikind=_stat_to_minikind, _pack_stat=pack_stat):
1512
"""Note the sha1 of a file.
1514
:param entry: The entry the sha1 is for.
1515
:param sha1: The observed sha1.
1516
:param stat_value: The os.lstat for the file.
1519
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1523
packed_stat = _pack_stat(stat_value)
1525
if self._cutoff_time is None:
1526
self._sha_cutoff_time()
1527
if (stat_value.st_mtime < self._cutoff_time
1528
and stat_value.st_ctime < self._cutoff_time):
1529
entry[1][0] = ('f', sha1, entry[1][0][2], entry[1][0][3],
1531
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1533
def _sha_cutoff_time(self):
1534
"""Return cutoff time.
1536
Files modified more recently than this time are at risk of being
1537
undetectably modified and so can't be cached.
1539
# Cache the cutoff time as long as we hold a lock.
1540
# time.time() isn't super expensive (approx 3.38us), but
1541
# when you call it 50,000 times it adds up.
1542
# For comparison, os.lstat() costs 7.2us if it is hot.
1543
self._cutoff_time = int(time.time()) - 3
1544
return self._cutoff_time
1546
def _lstat(self, abspath, entry):
1547
"""Return the os.lstat value for this path."""
1548
return os.lstat(abspath)
1550
def _sha1_file_and_mutter(self, abspath):
1551
# when -Dhashcache is turned on, this is monkey-patched in to log
1553
trace.mutter("dirstate sha1 " + abspath)
1554
return osutils.sha_file_by_name(abspath)
1556
def _is_executable(self, mode, old_executable):
1557
"""Is this file executable?"""
1558
return bool(S_IEXEC & mode)
1560
def _is_executable_win32(self, mode, old_executable):
1561
"""On win32 the executable bit is stored in the dirstate."""
1562
return old_executable
1564
if sys.platform == 'win32':
1565
_is_executable = _is_executable_win32
1567
def _read_link(self, abspath, old_link):
1568
"""Read the target of a symlink"""
1569
# TODO: jam 200700301 On Win32, this could just return the value
1570
# already in memory. However, this really needs to be done at a
1571
# higher level, because there either won't be anything on disk,
1572
# or the thing on disk will be a file.
1573
return os.readlink(abspath)
1575
def get_ghosts(self):
1576
"""Return a list of the parent tree revision ids that are ghosts."""
1577
self._read_header_if_needed()
1580
def get_lines(self):
1581
"""Serialise the entire dirstate to a sequence of lines."""
1582
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1583
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1584
# read whats on disk.
1585
self._state_file.seek(0)
1586
return self._state_file.readlines()
1588
lines.append(self._get_parents_line(self.get_parent_ids()))
1589
lines.append(self._get_ghosts_line(self._ghosts))
1590
# append the root line which is special cased
1591
lines.extend(map(self._entry_to_line, self._iter_entries()))
1592
return self._get_output_lines(lines)
1594
def _get_ghosts_line(self, ghost_ids):
1595
"""Create a line for the state file for ghost information."""
1596
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1598
def _get_parents_line(self, parent_ids):
1599
"""Create a line for the state file for parents information."""
1600
return '\0'.join([str(len(parent_ids))] + parent_ids)
1602
def _get_fields_to_entry(self):
1603
"""Get a function which converts entry fields into a entry record.
1605
This handles size and executable, as well as parent records.
1607
:return: A function which takes a list of fields, and returns an
1608
appropriate record for storing in memory.
1610
# This is intentionally unrolled for performance
1611
num_present_parents = self._num_present_parents()
1612
if num_present_parents == 0:
1613
def fields_to_entry_0_parents(fields, _int=int):
1614
path_name_file_id_key = (fields[0], fields[1], fields[2])
1615
return (path_name_file_id_key, [
1617
fields[3], # minikind
1618
fields[4], # fingerprint
1619
_int(fields[5]), # size
1620
fields[6] == 'y', # executable
1621
fields[7], # packed_stat or revision_id
1623
return fields_to_entry_0_parents
1624
elif num_present_parents == 1:
1625
def fields_to_entry_1_parent(fields, _int=int):
1626
path_name_file_id_key = (fields[0], fields[1], fields[2])
1627
return (path_name_file_id_key, [
1629
fields[3], # minikind
1630
fields[4], # fingerprint
1631
_int(fields[5]), # size
1632
fields[6] == 'y', # executable
1633
fields[7], # packed_stat or revision_id
1636
fields[8], # minikind
1637
fields[9], # fingerprint
1638
_int(fields[10]), # size
1639
fields[11] == 'y', # executable
1640
fields[12], # packed_stat or revision_id
1643
return fields_to_entry_1_parent
1644
elif num_present_parents == 2:
1645
def fields_to_entry_2_parents(fields, _int=int):
1646
path_name_file_id_key = (fields[0], fields[1], fields[2])
1647
return (path_name_file_id_key, [
1649
fields[3], # minikind
1650
fields[4], # fingerprint
1651
_int(fields[5]), # size
1652
fields[6] == 'y', # executable
1653
fields[7], # packed_stat or revision_id
1656
fields[8], # minikind
1657
fields[9], # fingerprint
1658
_int(fields[10]), # size
1659
fields[11] == 'y', # executable
1660
fields[12], # packed_stat or revision_id
1663
fields[13], # minikind
1664
fields[14], # fingerprint
1665
_int(fields[15]), # size
1666
fields[16] == 'y', # executable
1667
fields[17], # packed_stat or revision_id
1670
return fields_to_entry_2_parents
1672
def fields_to_entry_n_parents(fields, _int=int):
1673
path_name_file_id_key = (fields[0], fields[1], fields[2])
1674
trees = [(fields[cur], # minikind
1675
fields[cur+1], # fingerprint
1676
_int(fields[cur+2]), # size
1677
fields[cur+3] == 'y', # executable
1678
fields[cur+4], # stat or revision_id
1679
) for cur in xrange(3, len(fields)-1, 5)]
1680
return path_name_file_id_key, trees
1681
return fields_to_entry_n_parents
1683
def get_parent_ids(self):
1684
"""Return a list of the parent tree ids for the directory state."""
1685
self._read_header_if_needed()
1686
return list(self._parents)
1688
def _get_block_entry_index(self, dirname, basename, tree_index):
1689
"""Get the coordinates for a path in the state structure.
1691
:param dirname: The utf8 dirname to lookup.
1692
:param basename: The utf8 basename to lookup.
1693
:param tree_index: The index of the tree for which this lookup should
1695
:return: A tuple describing where the path is located, or should be
1696
inserted. The tuple contains four fields: the block index, the row
1697
index, the directory is present (boolean), the entire path is
1698
present (boolean). There is no guarantee that either
1699
coordinate is currently reachable unless the found field for it is
1700
True. For instance, a directory not present in the searched tree
1701
may be returned with a value one greater than the current highest
1702
block offset. The directory present field will always be True when
1703
the path present field is True. The directory present field does
1704
NOT indicate that the directory is present in the searched tree,
1705
rather it indicates that there are at least some files in some
1708
self._read_dirblocks_if_needed()
1709
key = dirname, basename, ''
1710
block_index, present = self._find_block_index_from_key(key)
1712
# no such directory - return the dir index and 0 for the row.
1713
return block_index, 0, False, False
1714
block = self._dirblocks[block_index][1] # access the entries only
1715
entry_index, present = self._find_entry_index(key, block)
1716
# linear search through entries at this path to find the one
1718
while entry_index < len(block) and block[entry_index][0][1] == basename:
1719
if block[entry_index][1][tree_index][0] not in 'ar':
1720
# neither absent or relocated
1721
return block_index, entry_index, True, True
1723
return block_index, entry_index, True, False
1725
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1726
"""Get the dirstate entry for path in tree tree_index.
1728
If either file_id or path is supplied, it is used as the key to lookup.
1729
If both are supplied, the fastest lookup is used, and an error is
1730
raised if they do not both point at the same row.
1732
:param tree_index: The index of the tree we wish to locate this path
1733
in. If the path is present in that tree, the entry containing its
1734
details is returned, otherwise (None, None) is returned
1735
0 is the working tree, higher indexes are successive parent
1737
:param fileid_utf8: A utf8 file_id to look up.
1738
:param path_utf8: An utf8 path to be looked up.
1739
:return: The dirstate entry tuple for path, or (None, None)
1741
self._read_dirblocks_if_needed()
1742
if path_utf8 is not None:
1743
if type(path_utf8) is not str:
1744
raise AssertionError('path_utf8 is not a str: %s %s'
1745
% (type(path_utf8), path_utf8))
1746
# path lookups are faster
1747
dirname, basename = osutils.split(path_utf8)
1748
block_index, entry_index, dir_present, file_present = \
1749
self._get_block_entry_index(dirname, basename, tree_index)
1750
if not file_present:
1752
entry = self._dirblocks[block_index][1][entry_index]
1753
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
1754
raise AssertionError('unversioned entry?')
1756
if entry[0][2] != fileid_utf8:
1757
self._changes_aborted = True
1758
raise errors.BzrError('integrity error ? : mismatching'
1759
' tree_index, file_id and path')
1762
possible_keys = self._get_id_index().get(fileid_utf8, None)
1763
if not possible_keys:
1765
for key in possible_keys:
1766
block_index, present = \
1767
self._find_block_index_from_key(key)
1768
# strange, probably indicates an out of date
1769
# id index - for now, allow this.
1772
# WARNING: DO not change this code to use _get_block_entry_index
1773
# as that function is not suitable: it does not use the key
1774
# to lookup, and thus the wrong coordinates are returned.
1775
block = self._dirblocks[block_index][1]
1776
entry_index, present = self._find_entry_index(key, block)
1778
entry = self._dirblocks[block_index][1][entry_index]
1779
if entry[1][tree_index][0] in 'fdlt':
1780
# this is the result we are looking for: the
1781
# real home of this file_id in this tree.
1783
if entry[1][tree_index][0] == 'a':
1784
# there is no home for this entry in this tree
1786
if entry[1][tree_index][0] != 'r':
1787
raise AssertionError(
1788
"entry %r has invalid minikind %r for tree %r" \
1790
entry[1][tree_index][0],
1792
real_path = entry[1][tree_index][1]
1793
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1794
path_utf8=real_path)
1798
def initialize(cls, path):
1799
"""Create a new dirstate on path.
1801
The new dirstate will be an empty tree - that is it has no parents,
1802
and only a root node - which has id ROOT_ID.
1804
:param path: The name of the file for the dirstate.
1805
:return: A write-locked DirState object.
1807
# This constructs a new DirState object on a path, sets the _state_file
1808
# to a new empty file for that path. It then calls _set_data() with our
1809
# stock empty dirstate information - a root with ROOT_ID, no children,
1810
# and no parents. Finally it calls save() to ensure that this data will
1813
# root dir and root dir contents with no children.
1814
empty_tree_dirblocks = [('', []), ('', [])]
1815
# a new root directory, with a NULLSTAT.
1816
empty_tree_dirblocks[0][1].append(
1817
(('', '', inventory.ROOT_ID), [
1818
('d', '', 0, False, DirState.NULLSTAT),
1822
result._set_data([], empty_tree_dirblocks)
1830
def _inv_entry_to_details(inv_entry):
1831
"""Convert an inventory entry (from a revision tree) to state details.
1833
:param inv_entry: An inventory entry whose sha1 and link targets can be
1834
relied upon, and which has a revision set.
1835
:return: A details tuple - the details for a single tree at a path +
1838
kind = inv_entry.kind
1839
minikind = DirState._kind_to_minikind[kind]
1840
tree_data = inv_entry.revision
1841
if kind == 'directory':
1845
elif kind == 'symlink':
1846
# We don't support non-ascii targets for symlinks yet.
1847
fingerprint = str(inv_entry.symlink_target or '')
1850
elif kind == 'file':
1851
fingerprint = inv_entry.text_sha1 or ''
1852
size = inv_entry.text_size or 0
1853
executable = inv_entry.executable
1854
elif kind == 'tree-reference':
1855
fingerprint = inv_entry.reference_revision or ''
1859
raise Exception("can't pack %s" % inv_entry)
1860
return (minikind, fingerprint, size, executable, tree_data)
1862
def _iter_child_entries(self, tree_index, path_utf8):
1863
"""Iterate over all the entries that are children of path_utf.
1865
This only returns entries that are present (not in 'a', 'r') in
1866
tree_index. tree_index data is not refreshed, so if tree 0 is used,
1867
results may differ from that obtained if paths were statted to
1868
determine what ones were directories.
1870
Asking for the children of a non-directory will return an empty
1874
next_pending_dirs = [path_utf8]
1876
while next_pending_dirs:
1877
pending_dirs = next_pending_dirs
1878
next_pending_dirs = []
1879
for path in pending_dirs:
1880
block_index, present = self._find_block_index_from_key(
1882
if block_index == 0:
1884
if len(self._dirblocks) == 1:
1885
# asked for the children of the root with no other
1889
# children of a non-directory asked for.
1891
block = self._dirblocks[block_index]
1892
for entry in block[1]:
1893
kind = entry[1][tree_index][0]
1894
if kind not in absent:
1898
path = entry[0][0] + '/' + entry[0][1]
1901
next_pending_dirs.append(path)
1903
def _iter_entries(self):
1904
"""Iterate over all the entries in the dirstate.
1906
Each yelt item is an entry in the standard format described in the
1907
docstring of bzrlib.dirstate.
1909
self._read_dirblocks_if_needed()
1910
for directory in self._dirblocks:
1911
for entry in directory[1]:
1914
def _get_id_index(self):
1915
"""Get an id index of self._dirblocks."""
1916
if self._id_index is None:
1918
for key, tree_details in self._iter_entries():
1919
id_index.setdefault(key[2], set()).add(key)
1920
self._id_index = id_index
1921
return self._id_index
1923
def _get_output_lines(self, lines):
1924
"""Format lines for final output.
1926
:param lines: A sequence of lines containing the parents list and the
1929
output_lines = [DirState.HEADER_FORMAT_3]
1930
lines.append('') # a final newline
1931
inventory_text = '\0\n\0'.join(lines)
1932
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1933
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1934
num_entries = len(lines)-3
1935
output_lines.append('num_entries: %s\n' % (num_entries,))
1936
output_lines.append(inventory_text)
1939
def _make_deleted_row(self, fileid_utf8, parents):
1940
"""Return a deleted row for fileid_utf8."""
1941
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1944
def _num_present_parents(self):
1945
"""The number of parent entries in each record row."""
1946
return len(self._parents) - len(self._ghosts)
1950
"""Construct a DirState on the file at path path.
1952
:return: An unlocked DirState object, associated with the given path.
1954
result = DirState(path)
1957
def _read_dirblocks_if_needed(self):
1958
"""Read in all the dirblocks from the file if they are not in memory.
1960
This populates self._dirblocks, and sets self._dirblock_state to
1961
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1964
self._read_header_if_needed()
1965
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1966
_read_dirblocks(self)
1968
def _read_header(self):
1969
"""This reads in the metadata header, and the parent ids.
1971
After reading in, the file should be positioned at the null
1972
just before the start of the first record in the file.
1974
:return: (expected crc checksum, number of entries, parent list)
1976
self._read_prelude()
1977
parent_line = self._state_file.readline()
1978
info = parent_line.split('\0')
1979
num_parents = int(info[0])
1980
self._parents = info[1:-1]
1981
ghost_line = self._state_file.readline()
1982
info = ghost_line.split('\0')
1983
num_ghosts = int(info[1])
1984
self._ghosts = info[2:-1]
1985
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1986
self._end_of_header = self._state_file.tell()
1988
def _read_header_if_needed(self):
1989
"""Read the header of the dirstate file if needed."""
1990
# inline this as it will be called a lot
1991
if not self._lock_token:
1992
raise errors.ObjectNotLocked(self)
1993
if self._header_state == DirState.NOT_IN_MEMORY:
1996
def _read_prelude(self):
1997
"""Read in the prelude header of the dirstate file.
1999
This only reads in the stuff that is not connected to the crc
2000
checksum. The position will be correct to read in the rest of
2001
the file and check the checksum after this point.
2002
The next entry in the file should be the number of parents,
2003
and their ids. Followed by a newline.
2005
header = self._state_file.readline()
2006
if header != DirState.HEADER_FORMAT_3:
2007
raise errors.BzrError(
2008
'invalid header line: %r' % (header,))
2009
crc_line = self._state_file.readline()
2010
if not crc_line.startswith('crc32: '):
2011
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2012
self.crc_expected = int(crc_line[len('crc32: '):-1])
2013
num_entries_line = self._state_file.readline()
2014
if not num_entries_line.startswith('num_entries: '):
2015
raise errors.BzrError('missing num_entries line')
2016
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2018
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
2019
"""Find a sha1 given a stat lookup."""
2020
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
2022
def _get_packed_stat_index(self):
2023
"""Get a packed_stat index of self._dirblocks."""
2024
if self._packed_stat_index is None:
2026
for key, tree_details in self._iter_entries():
2027
if tree_details[0][0] == 'f':
2028
index[tree_details[0][4]] = tree_details[0][1]
2029
self._packed_stat_index = index
2030
return self._packed_stat_index
2033
"""Save any pending changes created during this session.
2035
We reuse the existing file, because that prevents race conditions with
2036
file creation, and use oslocks on it to prevent concurrent modification
2037
and reads - because dirstate's incremental data aggregation is not
2038
compatible with reading a modified file, and replacing a file in use by
2039
another process is impossible on Windows.
2041
A dirstate in read only mode should be smart enough though to validate
2042
that the file has not changed, and otherwise discard its cache and
2043
start over, to allow for fine grained read lock duration, so 'status'
2044
wont block 'commit' - for example.
2046
if self._changes_aborted:
2047
# Should this be a warning? For now, I'm expecting that places that
2048
# mark it inconsistent will warn, making a warning here redundant.
2049
trace.mutter('Not saving DirState because '
2050
'_changes_aborted is set.')
2052
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
2053
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2055
grabbed_write_lock = False
2056
if self._lock_state != 'w':
2057
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2058
# Switch over to the new lock, as the old one may be closed.
2059
# TODO: jam 20070315 We should validate the disk file has
2060
# not changed contents. Since temporary_write_lock may
2061
# not be an atomic operation.
2062
self._lock_token = new_lock
2063
self._state_file = new_lock.f
2064
if not grabbed_write_lock:
2065
# We couldn't grab a write lock, so we switch back to a read one
2068
self._state_file.seek(0)
2069
self._state_file.writelines(self.get_lines())
2070
self._state_file.truncate()
2071
self._state_file.flush()
2072
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2073
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
2075
if grabbed_write_lock:
2076
self._lock_token = self._lock_token.restore_read_lock()
2077
self._state_file = self._lock_token.f
2078
# TODO: jam 20070315 We should validate the disk file has
2079
# not changed contents. Since restore_read_lock may
2080
# not be an atomic operation.
2082
def _set_data(self, parent_ids, dirblocks):
2083
"""Set the full dirstate data in memory.
2085
This is an internal function used to completely replace the objects
2086
in memory state. It puts the dirstate into state 'full-dirty'.
2088
:param parent_ids: A list of parent tree revision ids.
2089
:param dirblocks: A list containing one tuple for each directory in the
2090
tree. Each tuple contains the directory path and a list of entries
2091
found in that directory.
2093
# our memory copy is now authoritative.
2094
self._dirblocks = dirblocks
2095
self._header_state = DirState.IN_MEMORY_MODIFIED
2096
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2097
self._parents = list(parent_ids)
2098
self._id_index = None
2099
self._packed_stat_index = None
2101
def set_path_id(self, path, new_id):
2102
"""Change the id of path to new_id in the current working tree.
2104
:param path: The path inside the tree to set - '' is the root, 'foo'
2105
is the path foo in the root.
2106
:param new_id: The new id to assign to the path. This must be a utf8
2107
file id (not unicode, and not None).
2109
self._read_dirblocks_if_needed()
2111
# TODO: logic not written
2112
raise NotImplementedError(self.set_path_id)
2113
# TODO: check new id is unique
2114
entry = self._get_entry(0, path_utf8=path)
2115
if entry[0][2] == new_id:
2116
# Nothing to change.
2118
# mark the old path absent, and insert a new root path
2119
self._make_absent(entry)
2120
self.update_minimal(('', '', new_id), 'd',
2121
path_utf8='', packed_stat=entry[1][0][4])
2122
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2123
if self._id_index is not None:
2124
self._id_index.setdefault(new_id, set()).add(entry[0])
2126
def set_parent_trees(self, trees, ghosts):
2127
"""Set the parent trees for the dirstate.
2129
:param trees: A list of revision_id, tree tuples. tree must be provided
2130
even if the revision_id refers to a ghost: supply an empty tree in
2132
:param ghosts: A list of the revision_ids that are ghosts at the time
2135
# TODO: generate a list of parent indexes to preserve to save
2136
# processing specific parent trees. In the common case one tree will
2137
# be preserved - the left most parent.
2138
# TODO: if the parent tree is a dirstate, we might want to walk them
2139
# all by path in parallel for 'optimal' common-case performance.
2140
# generate new root row.
2141
self._read_dirblocks_if_needed()
2142
# TODO future sketch: Examine the existing parents to generate a change
2143
# map and then walk the new parent trees only, mapping them into the
2144
# dirstate. Walk the dirstate at the same time to remove unreferenced
2147
# sketch: loop over all entries in the dirstate, cherry picking
2148
# entries from the parent trees, if they are not ghost trees.
2149
# after we finish walking the dirstate, all entries not in the dirstate
2150
# are deletes, so we want to append them to the end as per the design
2151
# discussions. So do a set difference on ids with the parents to
2152
# get deletes, and add them to the end.
2153
# During the update process we need to answer the following questions:
2154
# - find other keys containing a fileid in order to create cross-path
2155
# links. We dont't trivially use the inventory from other trees
2156
# because this leads to either double touching, or to accessing
2158
# - find other keys containing a path
2159
# We accumulate each entry via this dictionary, including the root
2162
# we could do parallel iterators, but because file id data may be
2163
# scattered throughout, we dont save on index overhead: we have to look
2164
# at everything anyway. We can probably save cycles by reusing parent
2165
# data and doing an incremental update when adding an additional
2166
# parent, but for now the common cases are adding a new parent (merge),
2167
# and replacing completely (commit), and commit is more common: so
2168
# optimise merge later.
2170
# ---- start generation of full tree mapping data
2171
# what trees should we use?
2172
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2173
# how many trees do we end up with
2174
parent_count = len(parent_trees)
2176
# one: the current tree
2177
for entry in self._iter_entries():
2178
# skip entries not in the current tree
2179
if entry[1][0][0] in 'ar': # absent, relocated
2181
by_path[entry[0]] = [entry[1][0]] + \
2182
[DirState.NULL_PARENT_DETAILS] * parent_count
2183
id_index[entry[0][2]] = set([entry[0]])
2185
# now the parent trees:
2186
for tree_index, tree in enumerate(parent_trees):
2187
# the index is off by one, adjust it.
2188
tree_index = tree_index + 1
2189
# when we add new locations for a fileid we need these ranges for
2190
# any fileid in this tree as we set the by_path[id] to:
2191
# already_processed_tree_details + new_details + new_location_suffix
2192
# the suffix is from tree_index+1:parent_count+1.
2193
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2194
# now stitch in all the entries from this tree
2195
for path, entry in tree.inventory.iter_entries_by_dir():
2196
# here we process each trees details for each item in the tree.
2197
# we first update any existing entries for the id at other paths,
2198
# then we either create or update the entry for the id at the
2199
# right path, and finally we add (if needed) a mapping from
2200
# file_id to this path. We do it in this order to allow us to
2201
# avoid checking all known paths for the id when generating a
2202
# new entry at this path: by adding the id->path mapping last,
2203
# all the mappings are valid and have correct relocation
2204
# records where needed.
2205
file_id = entry.file_id
2206
path_utf8 = path.encode('utf8')
2207
dirname, basename = osutils.split(path_utf8)
2208
new_entry_key = (dirname, basename, file_id)
2209
# tree index consistency: All other paths for this id in this tree
2210
# index must point to the correct path.
2211
for entry_key in id_index.setdefault(file_id, set()):
2212
# TODO:PROFILING: It might be faster to just update
2213
# rather than checking if we need to, and then overwrite
2214
# the one we are located at.
2215
if entry_key != new_entry_key:
2216
# this file id is at a different path in one of the
2217
# other trees, so put absent pointers there
2218
# This is the vertical axis in the matrix, all pointing
2220
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2221
# by path consistency: Insert into an existing path record (trivial), or
2222
# add a new one with relocation pointers for the other tree indexes.
2223
if new_entry_key in id_index[file_id]:
2224
# there is already an entry where this data belongs, just insert it.
2225
by_path[new_entry_key][tree_index] = \
2226
self._inv_entry_to_details(entry)
2228
# add relocated entries to the horizontal axis - this row
2229
# mapping from path,id. We need to look up the correct path
2230
# for the indexes from 0 to tree_index -1
2232
for lookup_index in xrange(tree_index):
2233
# boundary case: this is the first occurence of file_id
2234
# so there are no id_indexs, possibly take this out of
2236
if not len(id_index[file_id]):
2237
new_details.append(DirState.NULL_PARENT_DETAILS)
2239
# grab any one entry, use it to find the right path.
2240
# TODO: optimise this to reduce memory use in highly
2241
# fragmented situations by reusing the relocation
2243
a_key = iter(id_index[file_id]).next()
2244
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2245
# its a pointer or missing statement, use it as is.
2246
new_details.append(by_path[a_key][lookup_index])
2248
# we have the right key, make a pointer to it.
2249
real_path = ('/'.join(a_key[0:2])).strip('/')
2250
new_details.append(('r', real_path, 0, False, ''))
2251
new_details.append(self._inv_entry_to_details(entry))
2252
new_details.extend(new_location_suffix)
2253
by_path[new_entry_key] = new_details
2254
id_index[file_id].add(new_entry_key)
2255
# --- end generation of full tree mappings
2257
# sort and output all the entries
2258
new_entries = self._sort_entries(by_path.items())
2259
self._entries_to_current_state(new_entries)
2260
self._parents = [rev_id for rev_id, tree in trees]
2261
self._ghosts = list(ghosts)
2262
self._header_state = DirState.IN_MEMORY_MODIFIED
2263
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2264
self._id_index = id_index
2266
def _sort_entries(self, entry_list):
2267
"""Given a list of entries, sort them into the right order.
2269
This is done when constructing a new dirstate from trees - normally we
2270
try to keep everything in sorted blocks all the time, but sometimes
2271
it's easier to sort after the fact.
2274
# sort by: directory parts, file name, file id
2275
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2276
return sorted(entry_list, key=_key)
2278
def set_state_from_inventory(self, new_inv):
2279
"""Set new_inv as the current state.
2281
This API is called by tree transform, and will usually occur with
2282
existing parent trees.
2284
:param new_inv: The inventory object to set current state from.
2286
if 'evil' in debug.debug_flags:
2287
trace.mutter_callsite(1,
2288
"set_state_from_inventory called; please mutate the tree instead")
2289
self._read_dirblocks_if_needed()
2291
# Two iterators: current data and new data, both in dirblock order.
2292
# We zip them together, which tells about entries that are new in the
2293
# inventory, or removed in the inventory, or present in both and
2296
# You might think we could just synthesize a new dirstate directly
2297
# since we're processing it in the right order. However, we need to
2298
# also consider there may be any number of parent trees and relocation
2299
# pointers, and we don't want to duplicate that here.
2300
new_iterator = new_inv.iter_entries_by_dir()
2301
# we will be modifying the dirstate, so we need a stable iterator. In
2302
# future we might write one, for now we just clone the state into a
2303
# list - which is a shallow copy.
2304
old_iterator = iter(list(self._iter_entries()))
2305
# both must have roots so this is safe:
2306
current_new = new_iterator.next()
2307
current_old = old_iterator.next()
2308
def advance(iterator):
2310
return iterator.next()
2311
except StopIteration:
2313
while current_new or current_old:
2314
# skip entries in old that are not really there
2315
if current_old and current_old[1][0][0] in 'ar':
2316
# relocated or absent
2317
current_old = advance(old_iterator)
2320
# convert new into dirblock style
2321
new_path_utf8 = current_new[0].encode('utf8')
2322
new_dirname, new_basename = osutils.split(new_path_utf8)
2323
new_id = current_new[1].file_id
2324
new_entry_key = (new_dirname, new_basename, new_id)
2325
current_new_minikind = \
2326
DirState._kind_to_minikind[current_new[1].kind]
2327
if current_new_minikind == 't':
2328
fingerprint = current_new[1].reference_revision or ''
2330
# We normally only insert or remove records, or update
2331
# them when it has significantly changed. Then we want to
2332
# erase its fingerprint. Unaffected records should
2333
# normally not be updated at all.
2336
# for safety disable variables
2337
new_path_utf8 = new_dirname = new_basename = new_id = \
2338
new_entry_key = None
2339
# 5 cases, we dont have a value that is strictly greater than everything, so
2340
# we make both end conditions explicit
2342
# old is finished: insert current_new into the state.
2343
self.update_minimal(new_entry_key, current_new_minikind,
2344
executable=current_new[1].executable,
2345
path_utf8=new_path_utf8, fingerprint=fingerprint)
2346
current_new = advance(new_iterator)
2347
elif not current_new:
2349
self._make_absent(current_old)
2350
current_old = advance(old_iterator)
2351
elif new_entry_key == current_old[0]:
2352
# same - common case
2353
# We're looking at the same path and id in both the dirstate
2354
# and inventory, so just need to update the fields in the
2355
# dirstate from the one in the inventory.
2356
# TODO: update the record if anything significant has changed.
2357
# the minimal required trigger is if the execute bit or cached
2359
if (current_old[1][0][3] != current_new[1].executable or
2360
current_old[1][0][0] != current_new_minikind):
2361
self.update_minimal(current_old[0], current_new_minikind,
2362
executable=current_new[1].executable,
2363
path_utf8=new_path_utf8, fingerprint=fingerprint)
2364
# both sides are dealt with, move on
2365
current_old = advance(old_iterator)
2366
current_new = advance(new_iterator)
2367
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2368
or (new_dirname == current_old[0][0]
2369
and new_entry_key[1:] < current_old[0][1:])):
2371
# add a entry for this and advance new
2372
self.update_minimal(new_entry_key, current_new_minikind,
2373
executable=current_new[1].executable,
2374
path_utf8=new_path_utf8, fingerprint=fingerprint)
2375
current_new = advance(new_iterator)
2377
# we've advanced past the place where the old key would be,
2378
# without seeing it in the new list. so it must be gone.
2379
self._make_absent(current_old)
2380
current_old = advance(old_iterator)
2381
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2382
self._id_index = None
2383
self._packed_stat_index = None
2385
def _make_absent(self, current_old):
2386
"""Mark current_old - an entry - as absent for tree 0.
2388
:return: True if this was the last details entry for the entry key:
2389
that is, if the underlying block has had the entry removed, thus
2390
shrinking in length.
2392
# build up paths that this id will be left at after the change is made,
2393
# so we can update their cross references in tree 0
2394
all_remaining_keys = set()
2395
# Dont check the working tree, because it's going.
2396
for details in current_old[1][1:]:
2397
if details[0] not in 'ar': # absent, relocated
2398
all_remaining_keys.add(current_old[0])
2399
elif details[0] == 'r': # relocated
2400
# record the key for the real path.
2401
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2402
# absent rows are not present at any path.
2403
last_reference = current_old[0] not in all_remaining_keys
2405
# the current row consists entire of the current item (being marked
2406
# absent), and relocated or absent entries for the other trees:
2407
# Remove it, its meaningless.
2408
block = self._find_block(current_old[0])
2409
entry_index, present = self._find_entry_index(current_old[0], block[1])
2411
raise AssertionError('could not find entry for %s' % (current_old,))
2412
block[1].pop(entry_index)
2413
# if we have an id_index in use, remove this key from it for this id.
2414
if self._id_index is not None:
2415
self._id_index[current_old[0][2]].remove(current_old[0])
2416
# update all remaining keys for this id to record it as absent. The
2417
# existing details may either be the record we are marking as deleted
2418
# (if there were other trees with the id present at this path), or may
2420
for update_key in all_remaining_keys:
2421
update_block_index, present = \
2422
self._find_block_index_from_key(update_key)
2424
raise AssertionError('could not find block for %s' % (update_key,))
2425
update_entry_index, present = \
2426
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2428
raise AssertionError('could not find entry for %s' % (update_key,))
2429
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2430
# it must not be absent at the moment
2431
if update_tree_details[0][0] == 'a': # absent
2432
raise AssertionError('bad row %r' % (update_tree_details,))
2433
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2434
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2435
return last_reference
2437
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2438
packed_stat=None, size=0, path_utf8=None):
2439
"""Update an entry to the state in tree 0.
2441
This will either create a new entry at 'key' or update an existing one.
2442
It also makes sure that any other records which might mention this are
2445
:param key: (dir, name, file_id) for the new entry
2446
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2448
:param executable: Should the executable bit be set?
2449
:param fingerprint: Simple fingerprint for new entry: sha1 for files,
2450
referenced revision id for subtrees, etc.
2451
:param packed_stat: Packed stat value for new entry.
2452
:param size: Size information for new entry
2453
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2456
If packed_stat and fingerprint are not given, they're invalidated in
2459
block = self._find_block(key)[1]
2460
if packed_stat is None:
2461
packed_stat = DirState.NULLSTAT
2462
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2463
# sometimes present in the dirstate - this seems oddly inconsistent.
2465
entry_index, present = self._find_entry_index(key, block)
2466
new_details = (minikind, fingerprint, size, executable, packed_stat)
2467
id_index = self._get_id_index()
2469
# new entry, synthesis cross reference here,
2470
existing_keys = id_index.setdefault(key[2], set())
2471
if not existing_keys:
2472
# not currently in the state, simplest case
2473
new_entry = key, [new_details] + self._empty_parent_info()
2475
# present at one or more existing other paths.
2476
# grab one of them and use it to generate parent
2477
# relocation/absent entries.
2478
new_entry = key, [new_details]
2479
for other_key in existing_keys:
2480
# change the record at other to be a pointer to this new
2481
# record. The loop looks similar to the change to
2482
# relocations when updating an existing record but its not:
2483
# the test for existing kinds is different: this can be
2484
# factored out to a helper though.
2485
other_block_index, present = self._find_block_index_from_key(other_key)
2487
raise AssertionError('could not find block for %s' % (other_key,))
2488
other_entry_index, present = self._find_entry_index(other_key,
2489
self._dirblocks[other_block_index][1])
2491
raise AssertionError('could not find entry for %s' % (other_key,))
2492
if path_utf8 is None:
2493
raise AssertionError('no path')
2494
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2495
('r', path_utf8, 0, False, '')
2497
num_present_parents = self._num_present_parents()
2498
for lookup_index in xrange(1, num_present_parents + 1):
2499
# grab any one entry, use it to find the right path.
2500
# TODO: optimise this to reduce memory use in highly
2501
# fragmented situations by reusing the relocation
2503
update_block_index, present = \
2504
self._find_block_index_from_key(other_key)
2506
raise AssertionError('could not find block for %s' % (other_key,))
2507
update_entry_index, present = \
2508
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2510
raise AssertionError('could not find entry for %s' % (other_key,))
2511
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2512
if update_details[0] in 'ar': # relocated, absent
2513
# its a pointer or absent in lookup_index's tree, use
2515
new_entry[1].append(update_details)
2517
# we have the right key, make a pointer to it.
2518
pointer_path = osutils.pathjoin(*other_key[0:2])
2519
new_entry[1].append(('r', pointer_path, 0, False, ''))
2520
block.insert(entry_index, new_entry)
2521
existing_keys.add(key)
2523
# Does the new state matter?
2524
block[entry_index][1][0] = new_details
2525
# parents cannot be affected by what we do.
2526
# other occurences of this id can be found
2527
# from the id index.
2529
# tree index consistency: All other paths for this id in this tree
2530
# index must point to the correct path. We have to loop here because
2531
# we may have passed entries in the state with this file id already
2532
# that were absent - where parent entries are - and they need to be
2533
# converted to relocated.
2534
if path_utf8 is None:
2535
raise AssertionError('no path')
2536
for entry_key in id_index.setdefault(key[2], set()):
2537
# TODO:PROFILING: It might be faster to just update
2538
# rather than checking if we need to, and then overwrite
2539
# the one we are located at.
2540
if entry_key != key:
2541
# this file id is at a different path in one of the
2542
# other trees, so put absent pointers there
2543
# This is the vertical axis in the matrix, all pointing
2545
block_index, present = self._find_block_index_from_key(entry_key)
2547
raise AssertionError('not present: %r', entry_key)
2548
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2550
raise AssertionError('not present: %r', entry_key)
2551
self._dirblocks[block_index][1][entry_index][1][0] = \
2552
('r', path_utf8, 0, False, '')
2553
# add a containing dirblock if needed.
2554
if new_details[0] == 'd':
2555
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2556
block_index, present = self._find_block_index_from_key(subdir_key)
2558
self._dirblocks.insert(block_index, (subdir_key[0], []))
2560
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2562
def _validate(self):
2563
"""Check that invariants on the dirblock are correct.
2565
This can be useful in debugging; it shouldn't be necessary in
2568
This must be called with a lock held.
2570
# NOTE: This must always raise AssertionError not just assert,
2571
# otherwise it may not behave properly under python -O
2573
# TODO: All entries must have some content that's not 'a' or 'r',
2574
# otherwise it could just be removed.
2576
# TODO: All relocations must point directly to a real entry.
2578
# TODO: No repeated keys.
2581
from pprint import pformat
2582
self._read_dirblocks_if_needed()
2583
if len(self._dirblocks) > 0:
2584
if not self._dirblocks[0][0] == '':
2585
raise AssertionError(
2586
"dirblocks don't start with root block:\n" + \
2587
pformat(self._dirblocks))
2588
if len(self._dirblocks) > 1:
2589
if not self._dirblocks[1][0] == '':
2590
raise AssertionError(
2591
"dirblocks missing root directory:\n" + \
2592
pformat(self._dirblocks))
2593
# the dirblocks are sorted by their path components, name, and dir id
2594
dir_names = [d[0].split('/')
2595
for d in self._dirblocks[1:]]
2596
if dir_names != sorted(dir_names):
2597
raise AssertionError(
2598
"dir names are not in sorted order:\n" + \
2599
pformat(self._dirblocks) + \
2602
for dirblock in self._dirblocks:
2603
# within each dirblock, the entries are sorted by filename and
2605
for entry in dirblock[1]:
2606
if dirblock[0] != entry[0][0]:
2607
raise AssertionError(
2609
"doesn't match directory name in\n%r" %
2610
(entry, pformat(dirblock)))
2611
if dirblock[1] != sorted(dirblock[1]):
2612
raise AssertionError(
2613
"dirblock for %r is not sorted:\n%s" % \
2614
(dirblock[0], pformat(dirblock)))
2616
def check_valid_parent():
2617
"""Check that the current entry has a valid parent.
2619
This makes sure that the parent has a record,
2620
and that the parent isn't marked as "absent" in the
2621
current tree. (It is invalid to have a non-absent file in an absent
2624
if entry[0][0:2] == ('', ''):
2625
# There should be no parent for the root row
2627
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2628
if parent_entry == (None, None):
2629
raise AssertionError(
2630
"no parent entry for: %s in tree %s"
2631
% (this_path, tree_index))
2632
if parent_entry[1][tree_index][0] != 'd':
2633
raise AssertionError(
2634
"Parent entry for %s is not marked as a valid"
2635
" directory. %s" % (this_path, parent_entry,))
2637
# For each file id, for each tree: either
2638
# the file id is not present at all; all rows with that id in the
2639
# key have it marked as 'absent'
2640
# OR the file id is present under exactly one name; any other entries
2641
# that mention that id point to the correct name.
2643
# We check this with a dict per tree pointing either to the present
2644
# name, or None if absent.
2645
tree_count = self._num_present_parents() + 1
2646
id_path_maps = [dict() for i in range(tree_count)]
2647
# Make sure that all renamed entries point to the correct location.
2648
for entry in self._iter_entries():
2649
file_id = entry[0][2]
2650
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2651
if len(entry[1]) != tree_count:
2652
raise AssertionError(
2653
"wrong number of entry details for row\n%s" \
2654
",\nexpected %d" % \
2655
(pformat(entry), tree_count))
2656
absent_positions = 0
2657
for tree_index, tree_state in enumerate(entry[1]):
2658
this_tree_map = id_path_maps[tree_index]
2659
minikind = tree_state[0]
2660
if minikind in 'ar':
2661
absent_positions += 1
2662
# have we seen this id before in this column?
2663
if file_id in this_tree_map:
2664
previous_path, previous_loc = this_tree_map[file_id]
2665
# any later mention of this file must be consistent with
2666
# what was said before
2668
if previous_path is not None:
2669
raise AssertionError(
2670
"file %s is absent in row %r but also present " \
2672
(file_id, entry, previous_path))
2673
elif minikind == 'r':
2674
target_location = tree_state[1]
2675
if previous_path != target_location:
2676
raise AssertionError(
2677
"file %s relocation in row %r but also at %r" \
2678
% (file_id, entry, previous_path))
2680
# a file, directory, etc - may have been previously
2681
# pointed to by a relocation, which must point here
2682
if previous_path != this_path:
2683
raise AssertionError(
2684
"entry %r inconsistent with previous path %r "
2686
(entry, previous_path, previous_loc))
2687
check_valid_parent()
2690
# absent; should not occur anywhere else
2691
this_tree_map[file_id] = None, this_path
2692
elif minikind == 'r':
2693
# relocation, must occur at expected location
2694
this_tree_map[file_id] = tree_state[1], this_path
2696
this_tree_map[file_id] = this_path, this_path
2697
check_valid_parent()
2698
if absent_positions == tree_count:
2699
raise AssertionError(
2700
"entry %r has no data for any tree." % (entry,))
2702
def _wipe_state(self):
2703
"""Forget all state information about the dirstate."""
2704
self._header_state = DirState.NOT_IN_MEMORY
2705
self._dirblock_state = DirState.NOT_IN_MEMORY
2706
self._changes_aborted = False
2709
self._dirblocks = []
2710
self._id_index = None
2711
self._packed_stat_index = None
2712
self._end_of_header = None
2713
self._cutoff_time = None
2714
self._split_path_cache = {}
2716
def lock_read(self):
2717
"""Acquire a read lock on the dirstate."""
2718
if self._lock_token is not None:
2719
raise errors.LockContention(self._lock_token)
2720
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2721
# already in memory, we could read just the header and check for
2722
# any modification. If not modified, we can just leave things
2724
self._lock_token = lock.ReadLock(self._filename)
2725
self._lock_state = 'r'
2726
self._state_file = self._lock_token.f
2729
def lock_write(self):
2730
"""Acquire a write lock on the dirstate."""
2731
if self._lock_token is not None:
2732
raise errors.LockContention(self._lock_token)
2733
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2734
# already in memory, we could read just the header and check for
2735
# any modification. If not modified, we can just leave things
2737
self._lock_token = lock.WriteLock(self._filename)
2738
self._lock_state = 'w'
2739
self._state_file = self._lock_token.f
2743
"""Drop any locks held on the dirstate."""
2744
if self._lock_token is None:
2745
raise errors.LockNotHeld(self)
2746
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2747
# already in memory, we could read just the header and check for
2748
# any modification. If not modified, we can just leave things
2750
self._state_file = None
2751
self._lock_state = None
2752
self._lock_token.unlock()
2753
self._lock_token = None
2754
self._split_path_cache = {}
2756
def _requires_lock(self):
2757
"""Check that a lock is currently held by someone on the dirstate."""
2758
if not self._lock_token:
2759
raise errors.ObjectNotLocked(self)
2762
def py_update_entry(state, entry, abspath, stat_value,
2763
_stat_to_minikind=DirState._stat_to_minikind,
2764
_pack_stat=pack_stat):
2765
"""Update the entry based on what is actually on disk.
2767
This function only calculates the sha if it needs to - if the entry is
2768
uncachable, or clearly different to the first parent's entry, no sha
2769
is calculated, and None is returned.
2771
:param state: The dirstate this entry is in.
2772
:param entry: This is the dirblock entry for the file in question.
2773
:param abspath: The path on disk for this file.
2774
:param stat_value: The stat value done on the path.
2775
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
2776
target of a symlink.
2779
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
2783
packed_stat = _pack_stat(stat_value)
2784
(saved_minikind, saved_link_or_sha1, saved_file_size,
2785
saved_executable, saved_packed_stat) = entry[1][0]
2787
if (minikind == saved_minikind
2788
and packed_stat == saved_packed_stat):
2789
# The stat hasn't changed since we saved, so we can re-use the
2794
# size should also be in packed_stat
2795
if saved_file_size == stat_value.st_size:
2796
return saved_link_or_sha1
2798
# If we have gotten this far, that means that we need to actually
2799
# process this entry.
2802
executable = state._is_executable(stat_value.st_mode,
2804
if state._cutoff_time is None:
2805
state._sha_cutoff_time()
2806
if (stat_value.st_mtime < state._cutoff_time
2807
and stat_value.st_ctime < state._cutoff_time
2808
and len(entry[1]) > 1
2809
and entry[1][1][0] != 'a'):
2810
# Could check for size changes for further optimised
2811
# avoidance of sha1's. However the most prominent case of
2812
# over-shaing is during initial add, which this catches.
2813
link_or_sha1 = state._sha1_file(abspath)
2814
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
2815
executable, packed_stat)
2817
entry[1][0] = ('f', '', stat_value.st_size,
2818
executable, DirState.NULLSTAT)
2819
elif minikind == 'd':
2821
entry[1][0] = ('d', '', 0, False, packed_stat)
2822
if saved_minikind != 'd':
2823
# This changed from something into a directory. Make sure we
2824
# have a directory block for it. This doesn't happen very
2825
# often, so this doesn't have to be super fast.
2826
block_index, entry_index, dir_present, file_present = \
2827
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
2828
state._ensure_block(block_index, entry_index,
2829
osutils.pathjoin(entry[0][0], entry[0][1]))
2830
elif minikind == 'l':
2831
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
2832
if state._cutoff_time is None:
2833
state._sha_cutoff_time()
2834
if (stat_value.st_mtime < state._cutoff_time
2835
and stat_value.st_ctime < state._cutoff_time):
2836
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
2839
entry[1][0] = ('l', '', stat_value.st_size,
2840
False, DirState.NULLSTAT)
2841
state._dirblock_state = DirState.IN_MEMORY_MODIFIED
2843
update_entry = py_update_entry
2846
class ProcessEntryPython(object):
2848
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id", "uninteresting",
2849
"last_source_parent", "last_target_parent", "include_unchanged",
2850
"use_filesystem_for_exec", "utf8_decode", "searched_specific_files",
2851
"search_specific_files", "state", "source_index", "target_index",
2852
"want_unversioned", "tree"]
2854
def __init__(self, include_unchanged, use_filesystem_for_exec,
2855
search_specific_files, state, source_index, target_index,
2856
want_unversioned, tree):
2857
self.old_dirname_to_file_id = {}
2858
self.new_dirname_to_file_id = {}
2859
# Just a sentry, so that _process_entry can say that this
2860
# record is handled, but isn't interesting to process (unchanged)
2861
self.uninteresting = object()
2862
# Using a list so that we can access the values and change them in
2863
# nested scope. Each one is [path, file_id, entry]
2864
self.last_source_parent = [None, None]
2865
self.last_target_parent = [None, None]
2866
self.include_unchanged = include_unchanged
2867
self.use_filesystem_for_exec = use_filesystem_for_exec
2868
self.utf8_decode = cache_utf8._utf8_decode
2869
# for all search_indexs in each path at or under each element of
2870
# search_specific_files, if the detail is relocated: add the id, and add the
2871
# relocated path as one to search if its not searched already. If the
2872
# detail is not relocated, add the id.
2873
self.searched_specific_files = set()
2874
self.search_specific_files = search_specific_files
2876
self.source_index = source_index
2877
self.target_index = target_index
2878
self.want_unversioned = want_unversioned
2881
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
2882
"""Compare an entry and real disk to generate delta information.
2884
:param path_info: top_relpath, basename, kind, lstat, abspath for
2885
the path of entry. If None, then the path is considered absent.
2886
(Perhaps we should pass in a concrete entry for this ?)
2887
Basename is returned as a utf8 string because we expect this
2888
tuple will be ignored, and don't want to take the time to
2890
:return: None if these don't match
2891
A tuple of information about the change, or
2892
the object 'uninteresting' if these match, but are
2893
basically identical.
2895
if self.source_index is None:
2896
source_details = DirState.NULL_PARENT_DETAILS
2898
source_details = entry[1][self.source_index]
2899
target_details = entry[1][self.target_index]
2900
target_minikind = target_details[0]
2901
if path_info is not None and target_minikind in 'fdlt':
2902
if not (self.target_index == 0):
2903
raise AssertionError()
2904
link_or_sha1 = update_entry(self.state, entry,
2905
abspath=path_info[4], stat_value=path_info[3])
2906
# The entry may have been modified by update_entry
2907
target_details = entry[1][self.target_index]
2908
target_minikind = target_details[0]
2911
file_id = entry[0][2]
2912
source_minikind = source_details[0]
2913
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
2914
# claimed content in both: diff
2915
# r | fdlt | | add source to search, add id path move and perform
2916
# | | | diff check on source-target
2917
# r | fdlt | a | dangling file that was present in the basis.
2919
if source_minikind in 'r':
2920
# add the source to the search path to find any children it
2921
# has. TODO ? : only add if it is a container ?
2922
if not osutils.is_inside_any(self.searched_specific_files,
2924
self.search_specific_files.add(source_details[1])
2925
# generate the old path; this is needed for stating later
2927
old_path = source_details[1]
2928
old_dirname, old_basename = os.path.split(old_path)
2929
path = pathjoin(entry[0][0], entry[0][1])
2930
old_entry = self.state._get_entry(self.source_index,
2932
# update the source details variable to be the real
2934
if old_entry == (None, None):
2935
raise errors.CorruptDirstate(self.state._filename,
2936
"entry '%s/%s' is considered renamed from %r"
2937
" but source does not exist\n"
2938
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
2939
source_details = old_entry[1][self.source_index]
2940
source_minikind = source_details[0]
2942
old_dirname = entry[0][0]
2943
old_basename = entry[0][1]
2944
old_path = path = None
2945
if path_info is None:
2946
# the file is missing on disk, show as removed.
2947
content_change = True
2951
# source and target are both versioned and disk file is present.
2952
target_kind = path_info[2]
2953
if target_kind == 'directory':
2955
old_path = path = pathjoin(old_dirname, old_basename)
2956
self.new_dirname_to_file_id[path] = file_id
2957
if source_minikind != 'd':
2958
content_change = True
2960
# directories have no fingerprint
2961
content_change = False
2963
elif target_kind == 'file':
2964
if source_minikind != 'f':
2965
content_change = True
2967
# If the size is the same, check the sha:
2968
if target_details[2] == source_details[2]:
2969
if link_or_sha1 is None:
2971
file_obj = file(path_info[4], 'rb')
2973
statvalue = os.fstat(file_obj.fileno())
2974
link_or_sha1 = osutils.sha_file(file_obj)
2977
self.state._observed_sha1(entry, link_or_sha1,
2979
content_change = (link_or_sha1 != source_details[1])
2981
# Size changed, so must be different
2982
content_change = True
2983
# Target details is updated at update_entry time
2984
if self.use_filesystem_for_exec:
2985
# We don't need S_ISREG here, because we are sure
2986
# we are dealing with a file.
2987
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
2989
target_exec = target_details[3]
2990
elif target_kind == 'symlink':
2991
if source_minikind != 'l':
2992
content_change = True
2994
content_change = (link_or_sha1 != source_details[1])
2996
elif target_kind == 'tree-reference':
2997
if source_minikind != 't':
2998
content_change = True
3000
content_change = False
3003
raise Exception, "unknown kind %s" % path_info[2]
3004
if source_minikind == 'd':
3006
old_path = path = pathjoin(old_dirname, old_basename)
3007
self.old_dirname_to_file_id[old_path] = file_id
3008
# parent id is the entry for the path in the target tree
3009
if old_dirname == self.last_source_parent[0]:
3010
source_parent_id = self.last_source_parent[1]
3013
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3015
source_parent_entry = self.state._get_entry(self.source_index,
3016
path_utf8=old_dirname)
3017
source_parent_id = source_parent_entry[0][2]
3018
if source_parent_id == entry[0][2]:
3019
# This is the root, so the parent is None
3020
source_parent_id = None
3022
self.last_source_parent[0] = old_dirname
3023
self.last_source_parent[1] = source_parent_id
3024
new_dirname = entry[0][0]
3025
if new_dirname == self.last_target_parent[0]:
3026
target_parent_id = self.last_target_parent[1]
3029
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3031
# TODO: We don't always need to do the lookup, because the
3032
# parent entry will be the same as the source entry.
3033
target_parent_entry = self.state._get_entry(self.target_index,
3034
path_utf8=new_dirname)
3035
if target_parent_entry == (None, None):
3036
raise AssertionError(
3037
"Could not find target parent in wt: %s\nparent of: %s"
3038
% (new_dirname, entry))
3039
target_parent_id = target_parent_entry[0][2]
3040
if target_parent_id == entry[0][2]:
3041
# This is the root, so the parent is None
3042
target_parent_id = None
3044
self.last_target_parent[0] = new_dirname
3045
self.last_target_parent[1] = target_parent_id
3047
source_exec = source_details[3]
3048
if (self.include_unchanged
3050
or source_parent_id != target_parent_id
3051
or old_basename != entry[0][1]
3052
or source_exec != target_exec
3054
if old_path is None:
3055
old_path = path = pathjoin(old_dirname, old_basename)
3056
old_path_u = self.utf8_decode(old_path)[0]
3059
old_path_u = self.utf8_decode(old_path)[0]
3060
if old_path == path:
3063
path_u = self.utf8_decode(path)[0]
3064
source_kind = DirState._minikind_to_kind[source_minikind]
3065
return (entry[0][2],
3066
(old_path_u, path_u),
3069
(source_parent_id, target_parent_id),
3070
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3071
(source_kind, target_kind),
3072
(source_exec, target_exec))
3074
return self.uninteresting
3075
elif source_minikind in 'a' and target_minikind in 'fdlt':
3076
# looks like a new file
3077
path = pathjoin(entry[0][0], entry[0][1])
3078
# parent id is the entry for the path in the target tree
3079
# TODO: these are the same for an entire directory: cache em.
3080
parent_id = self.state._get_entry(self.target_index,
3081
path_utf8=entry[0][0])[0][2]
3082
if parent_id == entry[0][2]:
3084
if path_info is not None:
3086
if self.use_filesystem_for_exec:
3087
# We need S_ISREG here, because we aren't sure if this
3090
stat.S_ISREG(path_info[3].st_mode)
3091
and stat.S_IEXEC & path_info[3].st_mode)
3093
target_exec = target_details[3]
3094
return (entry[0][2],
3095
(None, self.utf8_decode(path)[0]),
3099
(None, self.utf8_decode(entry[0][1])[0]),
3100
(None, path_info[2]),
3101
(None, target_exec))
3103
# Its a missing file, report it as such.
3104
return (entry[0][2],
3105
(None, self.utf8_decode(path)[0]),
3109
(None, self.utf8_decode(entry[0][1])[0]),
3112
elif source_minikind in 'fdlt' and target_minikind in 'a':
3113
# unversioned, possibly, or possibly not deleted: we dont care.
3114
# if its still on disk, *and* theres no other entry at this
3115
# path [we dont know this in this routine at the moment -
3116
# perhaps we should change this - then it would be an unknown.
3117
old_path = pathjoin(entry[0][0], entry[0][1])
3118
# parent id is the entry for the path in the target tree
3119
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3120
if parent_id == entry[0][2]:
3122
return (entry[0][2],
3123
(self.utf8_decode(old_path)[0], None),
3127
(self.utf8_decode(entry[0][1])[0], None),
3128
(DirState._minikind_to_kind[source_minikind], None),
3129
(source_details[3], None))
3130
elif source_minikind in 'fdlt' and target_minikind in 'r':
3131
# a rename; could be a true rename, or a rename inherited from
3132
# a renamed parent. TODO: handle this efficiently. Its not
3133
# common case to rename dirs though, so a correct but slow
3134
# implementation will do.
3135
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3136
self.search_specific_files.add(target_details[1])
3137
elif source_minikind in 'ra' and target_minikind in 'ra':
3138
# neither of the selected trees contain this file,
3139
# so skip over it. This is not currently directly tested, but
3140
# is indirectly via test_too_much.TestCommands.test_conflicts.
3143
raise AssertionError("don't know how to compare "
3144
"source_minikind=%r, target_minikind=%r"
3145
% (source_minikind, target_minikind))
3146
## import pdb;pdb.set_trace()
3152
def iter_changes(self):
3153
"""Iterate over the changes."""
3154
utf8_decode = cache_utf8._utf8_decode
3155
_cmp_by_dirs = cmp_by_dirs
3156
_process_entry = self._process_entry
3157
uninteresting = self.uninteresting
3158
search_specific_files = self.search_specific_files
3159
searched_specific_files = self.searched_specific_files
3160
splitpath = osutils.splitpath
3162
# compare source_index and target_index at or under each element of search_specific_files.
3163
# follow the following comparison table. Note that we only want to do diff operations when
3164
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3168
# Source | Target | disk | action
3169
# r | fdlt | | add source to search, add id path move and perform
3170
# | | | diff check on source-target
3171
# r | fdlt | a | dangling file that was present in the basis.
3173
# r | a | | add source to search
3175
# r | r | | this path is present in a non-examined tree, skip.
3176
# r | r | a | this path is present in a non-examined tree, skip.
3177
# a | fdlt | | add new id
3178
# a | fdlt | a | dangling locally added file, skip
3179
# a | a | | not present in either tree, skip
3180
# a | a | a | not present in any tree, skip
3181
# a | r | | not present in either tree at this path, skip as it
3182
# | | | may not be selected by the users list of paths.
3183
# a | r | a | not present in either tree at this path, skip as it
3184
# | | | may not be selected by the users list of paths.
3185
# fdlt | fdlt | | content in both: diff them
3186
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3187
# fdlt | a | | unversioned: output deleted id for now
3188
# fdlt | a | a | unversioned and deleted: output deleted id
3189
# fdlt | r | | relocated in this tree, so add target to search.
3190
# | | | Dont diff, we will see an r,fd; pair when we reach
3191
# | | | this id at the other path.
3192
# fdlt | r | a | relocated in this tree, so add target to search.
3193
# | | | Dont diff, we will see an r,fd; pair when we reach
3194
# | | | this id at the other path.
3196
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3197
# keeping a cache of directories that we have seen.
3199
while search_specific_files:
3200
# TODO: the pending list should be lexically sorted? the
3201
# interface doesn't require it.
3202
current_root = search_specific_files.pop()
3203
current_root_unicode = current_root.decode('utf8')
3204
searched_specific_files.add(current_root)
3205
# process the entries for this containing directory: the rest will be
3206
# found by their parents recursively.
3207
root_entries = self.state._entries_for_path(current_root)
3208
root_abspath = self.tree.abspath(current_root_unicode)
3210
root_stat = os.lstat(root_abspath)
3212
if e.errno == errno.ENOENT:
3213
# the path does not exist: let _process_entry know that.
3214
root_dir_info = None
3216
# some other random error: hand it up.
3219
root_dir_info = ('', current_root,
3220
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3222
if root_dir_info[2] == 'directory':
3223
if self.tree._directory_is_tree_reference(
3224
current_root.decode('utf8')):
3225
root_dir_info = root_dir_info[:2] + \
3226
('tree-reference',) + root_dir_info[3:]
3228
if not root_entries and not root_dir_info:
3229
# this specified path is not present at all, skip it.
3231
path_handled = False
3232
for entry in root_entries:
3233
result = _process_entry(entry, root_dir_info)
3234
if result is not None:
3236
if result is not uninteresting:
3238
if self.want_unversioned and not path_handled and root_dir_info:
3239
new_executable = bool(
3240
stat.S_ISREG(root_dir_info[3].st_mode)
3241
and stat.S_IEXEC & root_dir_info[3].st_mode)
3243
(None, current_root_unicode),
3247
(None, splitpath(current_root_unicode)[-1]),
3248
(None, root_dir_info[2]),
3249
(None, new_executable)
3251
initial_key = (current_root, '', '')
3252
block_index, _ = self.state._find_block_index_from_key(initial_key)
3253
if block_index == 0:
3254
# we have processed the total root already, but because the
3255
# initial key matched it we should skip it here.
3257
if root_dir_info and root_dir_info[2] == 'tree-reference':
3258
current_dir_info = None
3260
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3262
current_dir_info = dir_iterator.next()
3264
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3265
# python 2.5 has e.errno == EINVAL,
3266
# and e.winerror == ERROR_DIRECTORY
3267
e_winerror = getattr(e, 'winerror', None)
3268
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3269
# there may be directories in the inventory even though
3270
# this path is not a file on disk: so mark it as end of
3272
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3273
current_dir_info = None
3274
elif (sys.platform == 'win32'
3275
and (e.errno in win_errors
3276
or e_winerror in win_errors)):
3277
current_dir_info = None
3281
if current_dir_info[0][0] == '':
3282
# remove .bzr from iteration
3283
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3284
if current_dir_info[1][bzr_index][0] != '.bzr':
3285
raise AssertionError()
3286
del current_dir_info[1][bzr_index]
3287
# walk until both the directory listing and the versioned metadata
3289
if (block_index < len(self.state._dirblocks) and
3290
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3291
current_block = self.state._dirblocks[block_index]
3293
current_block = None
3294
while (current_dir_info is not None or
3295
current_block is not None):
3296
if (current_dir_info and current_block
3297
and current_dir_info[0][0] != current_block[0]):
3298
if _cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
3299
# filesystem data refers to paths not covered by the dirblock.
3300
# this has two possibilities:
3301
# A) it is versioned but empty, so there is no block for it
3302
# B) it is not versioned.
3304
# if (A) then we need to recurse into it to check for
3305
# new unknown files or directories.
3306
# if (B) then we should ignore it, because we don't
3307
# recurse into unknown directories.
3309
while path_index < len(current_dir_info[1]):
3310
current_path_info = current_dir_info[1][path_index]
3311
if self.want_unversioned:
3312
if current_path_info[2] == 'directory':
3313
if self.tree._directory_is_tree_reference(
3314
current_path_info[0].decode('utf8')):
3315
current_path_info = current_path_info[:2] + \
3316
('tree-reference',) + current_path_info[3:]
3317
new_executable = bool(
3318
stat.S_ISREG(current_path_info[3].st_mode)
3319
and stat.S_IEXEC & current_path_info[3].st_mode)
3321
(None, utf8_decode(current_path_info[0])[0]),
3325
(None, utf8_decode(current_path_info[1])[0]),
3326
(None, current_path_info[2]),
3327
(None, new_executable))
3328
# dont descend into this unversioned path if it is
3330
if current_path_info[2] in ('directory',
3332
del current_dir_info[1][path_index]
3336
# This dir info has been handled, go to the next
3338
current_dir_info = dir_iterator.next()
3339
except StopIteration:
3340
current_dir_info = None
3342
# We have a dirblock entry for this location, but there
3343
# is no filesystem path for this. This is most likely
3344
# because a directory was removed from the disk.
3345
# We don't have to report the missing directory,
3346
# because that should have already been handled, but we
3347
# need to handle all of the files that are contained
3349
for current_entry in current_block[1]:
3350
# entry referring to file not present on disk.
3351
# advance the entry only, after processing.
3352
result = _process_entry(current_entry, None)
3353
if result is not None:
3354
if result is not uninteresting:
3357
if (block_index < len(self.state._dirblocks) and
3358
osutils.is_inside(current_root,
3359
self.state._dirblocks[block_index][0])):
3360
current_block = self.state._dirblocks[block_index]
3362
current_block = None
3365
if current_block and entry_index < len(current_block[1]):
3366
current_entry = current_block[1][entry_index]
3368
current_entry = None
3369
advance_entry = True
3371
if current_dir_info and path_index < len(current_dir_info[1]):
3372
current_path_info = current_dir_info[1][path_index]
3373
if current_path_info[2] == 'directory':
3374
if self.tree._directory_is_tree_reference(
3375
current_path_info[0].decode('utf8')):
3376
current_path_info = current_path_info[:2] + \
3377
('tree-reference',) + current_path_info[3:]
3379
current_path_info = None
3381
path_handled = False
3382
while (current_entry is not None or
3383
current_path_info is not None):
3384
if current_entry is None:
3385
# the check for path_handled when the path is adnvaced
3386
# will yield this path if needed.
3388
elif current_path_info is None:
3389
# no path is fine: the per entry code will handle it.
3390
result = _process_entry(current_entry, current_path_info)
3391
if result is not None:
3392
if result is not uninteresting:
3394
elif (current_entry[0][1] != current_path_info[1]
3395
or current_entry[1][self.target_index][0] in 'ar'):
3396
# The current path on disk doesn't match the dirblock
3397
# record. Either the dirblock is marked as absent, or
3398
# the file on disk is not present at all in the
3399
# dirblock. Either way, report about the dirblock
3400
# entry, and let other code handle the filesystem one.
3402
# Compare the basename for these files to determine
3404
if current_path_info[1] < current_entry[0][1]:
3405
# extra file on disk: pass for now, but only
3406
# increment the path, not the entry
3407
advance_entry = False
3409
# entry referring to file not present on disk.
3410
# advance the entry only, after processing.
3411
result = _process_entry(current_entry, None)
3412
if result is not None:
3413
if result is not uninteresting:
3415
advance_path = False
3417
result = _process_entry(current_entry, current_path_info)
3418
if result is not None:
3420
if result is not uninteresting:
3422
if advance_entry and current_entry is not None:
3424
if entry_index < len(current_block[1]):
3425
current_entry = current_block[1][entry_index]
3427
current_entry = None
3429
advance_entry = True # reset the advance flaga
3430
if advance_path and current_path_info is not None:
3431
if not path_handled:
3432
# unversioned in all regards
3433
if self.want_unversioned:
3434
new_executable = bool(
3435
stat.S_ISREG(current_path_info[3].st_mode)
3436
and stat.S_IEXEC & current_path_info[3].st_mode)
3438
relpath_unicode = utf8_decode(current_path_info[0])[0]
3439
except UnicodeDecodeError:
3440
raise errors.BadFilenameEncoding(
3441
current_path_info[0], osutils._fs_enc)
3443
(None, relpath_unicode),
3447
(None, utf8_decode(current_path_info[1])[0]),
3448
(None, current_path_info[2]),
3449
(None, new_executable))
3450
# dont descend into this unversioned path if it is
3452
if current_path_info[2] in ('directory'):
3453
del current_dir_info[1][path_index]
3455
# dont descend the disk iterator into any tree
3457
if current_path_info[2] == 'tree-reference':
3458
del current_dir_info[1][path_index]
3461
if path_index < len(current_dir_info[1]):
3462
current_path_info = current_dir_info[1][path_index]
3463
if current_path_info[2] == 'directory':
3464
if self.tree._directory_is_tree_reference(
3465
current_path_info[0].decode('utf8')):
3466
current_path_info = current_path_info[:2] + \
3467
('tree-reference',) + current_path_info[3:]
3469
current_path_info = None
3470
path_handled = False
3472
advance_path = True # reset the advance flagg.
3473
if current_block is not None:
3475
if (block_index < len(self.state._dirblocks) and
3476
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3477
current_block = self.state._dirblocks[block_index]
3479
current_block = None
3480
if current_dir_info is not None:
3482
current_dir_info = dir_iterator.next()
3483
except StopIteration:
3484
current_dir_info = None
3485
_process_entry = ProcessEntryPython
3488
# Try to load the compiled form if possible
3490
from bzrlib._dirstate_helpers_c import (
3491
_read_dirblocks_c as _read_dirblocks,
3492
bisect_dirblock_c as bisect_dirblock,
3493
_bisect_path_left_c as _bisect_path_left,
3494
_bisect_path_right_c as _bisect_path_right,
3495
cmp_by_dirs_c as cmp_by_dirs,
3496
ProcessEntryC as _process_entry,
3497
update_entry as update_entry,
3500
from bzrlib._dirstate_helpers_py import (
3501
_read_dirblocks_py as _read_dirblocks,
3502
bisect_dirblock_py as bisect_dirblock,
3503
_bisect_path_left_py as _bisect_path_left,
3504
_bisect_path_right_py as _bisect_path_right,
3505
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
