1
# Copyright (C) 2006-2011 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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 it's a file. The fingerprint is the
86
sha1 value of the file's canonical form, i.e. after any read filters have
87
been applied to the convenience form stored in the working tree.
88
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
90
't' is a reference to a nested subtree; the fingerprint is the referenced
95
The entries on disk and in memory are ordered according to the following keys:
97
directory, as a list of components
101
--- Format 1 had the following different definition: ---
102
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
103
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
105
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
106
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
109
PARENT ROW's are emitted for every parent that is not in the ghosts details
110
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
111
each row will have a PARENT ROW for foo and baz, but not for bar.
114
In any tree, a kind of 'moved' indicates that the fingerprint field
115
(which we treat as opaque data specific to the 'kind' anyway) has the
116
details for the id of this row in that tree.
118
I'm strongly tempted to add a id->path index as well, but I think that
119
where we need id->path mapping; we also usually read the whole file, so
120
I'm going to skip that for the moment, as we have the ability to locate
121
via bisect any path in any tree, and if we lookup things by path, we can
122
accumulate an id->path mapping as we go, which will tend to match what we
125
I plan to implement this asap, so please speak up now to alter/tweak the
126
design - and once we stabilise on this, I'll update the wiki page for
129
The rationale for all this is that we want fast operations for the
130
common case (diff/status/commit/merge on all files) and extremely fast
131
operations for the less common but still occurs a lot status/diff/commit
132
on specific files). Operations on specific files involve a scan for all
133
the children of a path, *in every involved tree*, which the current
134
format did not accommodate.
138
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
139
2) fall back current object model as needed.
140
3) scale usably to the largest trees known today - say 50K entries. (mozilla
141
is an example of this)
145
Eventually reuse dirstate objects across locks IFF the dirstate file has not
146
been modified, but will require that we flush/ignore cached stat-hit data
147
because we won't want to restat all files on disk just because a lock was
148
acquired, yet we cannot trust the data after the previous lock was released.
150
Memory representation:
151
vector of all directories, and vector of the childen ?
153
root_entrie = (direntry for root, [parent_direntries_for_root]),
155
('', ['data for achild', 'data for bchild', 'data for cchild'])
156
('dir', ['achild', 'cchild', 'echild'])
158
- single bisect to find N subtrees from a path spec
159
- in-order for serialisation - this is 'dirblock' grouping.
160
- insertion of a file '/a' affects only the '/' child-vector, that is, to
161
insert 10K elements from scratch does not generates O(N^2) memoves of a
162
single vector, rather each individual, which tends to be limited to a
163
manageable number. Will scale badly on trees with 10K entries in a
164
single directory. compare with Inventory.InventoryDirectory which has
165
a dictionary for the children. No bisect capability, can only probe for
166
exact matches, or grab all elements and sort.
167
- What's the risk of error here? Once we have the base format being processed
168
we should have a net win regardless of optimality. So we are going to
169
go with what seems reasonable.
172
Maybe we should do a test profile of the core structure - 10K simulated
173
searches/lookups/etc?
175
Objects for each row?
176
The lifetime of Dirstate objects is current per lock, but see above for
177
possible extensions. The lifetime of a row from a dirstate is expected to be
178
very short in the optimistic case: which we are optimising for. For instance,
179
subtree status will determine from analysis of the disk data what rows need to
180
be examined at all, and will be able to determine from a single row whether
181
that file has altered or not, so we are aiming to process tens of thousands of
182
entries each second within the dirstate context, before exposing anything to
183
the larger codebase. This suggests we want the time for a single file
184
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
185
processed, and to scale to 100 thousand we'll another order of magnitude to do
186
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
187
the file on disk, and then immediately discard, the overhead of object creation
188
becomes a significant cost.
190
Figures: Creating a tuple from 3 elements was profiled at 0.0625
191
microseconds, whereas creating a object which is subclassed from tuple was
192
0.500 microseconds, and creating an object with 3 elements and slots was 3
193
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
194
down to 10 microseconds for the total processing - having 33% of that be object
195
creation is a huge overhead. There is a potential cost in using tuples within
196
each row which is that the conditional code to do comparisons may be slower
197
than method invocation, but method invocation is known to be slow due to stack
198
frame creation, so avoiding methods in these tight inner loops in unfortunately
199
desirable. We can consider a pyrex version of this with objects in future if
209
from stat import S_IEXEC
228
# This is the Windows equivalent of ENOTDIR
229
# It is defined in pywin32.winerror, but we don't want a strong dependency for
230
# just an error code.
231
ERROR_PATH_NOT_FOUND = 3
232
ERROR_DIRECTORY = 267
235
if not getattr(struct, '_compile', None):
236
# Cannot pre-compile the dirstate pack_stat
237
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
238
"""Convert stat values into a packed representation."""
239
return _encode(_pack('>LLLLLL', st.st_size, int(st.st_mtime),
240
int(st.st_ctime), st.st_dev, st.st_ino & 0xFFFFFFFF,
243
# compile the struct compiler we need, so as to only do it once
244
from _struct import Struct
245
_compiled_pack = Struct('>LLLLLL').pack
246
def pack_stat(st, _encode=binascii.b2a_base64, _pack=_compiled_pack):
247
"""Convert stat values into a packed representation."""
248
# jam 20060614 it isn't really worth removing more entries if we
249
# are going to leave it in packed form.
250
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
251
# With all entries, filesize is 5.9M and read time is maybe 280ms
252
# well within the noise margin
254
# base64 encoding always adds a final newline, so strip it off
255
# The current version
256
return _encode(_pack(st.st_size, int(st.st_mtime), int(st.st_ctime),
257
st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
258
# This is 0.060s / 1.520s faster by not encoding as much information
259
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
260
# This is not strictly faster than _encode(_pack())[:-1]
261
# return '%X.%X.%X.%X.%X.%X' % (
262
# st.st_size, int(st.st_mtime), int(st.st_ctime),
263
# st.st_dev, st.st_ino, st.st_mode)
264
# Similar to the _encode(_pack('>LL'))
265
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
268
def _unpack_stat(packed_stat):
269
"""Turn a packed_stat back into the stat fields.
271
This is meant as a debugging tool, should not be used in real code.
273
(st_size, st_mtime, st_ctime, st_dev, st_ino,
274
st_mode) = struct.unpack('>LLLLLL', binascii.a2b_base64(packed_stat))
275
return dict(st_size=st_size, st_mtime=st_mtime, st_ctime=st_ctime,
276
st_dev=st_dev, st_ino=st_ino, st_mode=st_mode)
279
class SHA1Provider(object):
280
"""An interface for getting sha1s of a file."""
282
def sha1(self, abspath):
283
"""Return the sha1 of a file given its absolute path.
285
:param abspath: May be a filesystem encoded absolute path
288
raise NotImplementedError(self.sha1)
290
def stat_and_sha1(self, abspath):
291
"""Return the stat and sha1 of a file given its absolute path.
293
:param abspath: May be a filesystem encoded absolute path
296
Note: the stat should be the stat of the physical file
297
while the sha may be the sha of its canonical content.
299
raise NotImplementedError(self.stat_and_sha1)
302
class DefaultSHA1Provider(SHA1Provider):
303
"""A SHA1Provider that reads directly from the filesystem."""
305
def sha1(self, abspath):
306
"""Return the sha1 of a file given its absolute path."""
307
return osutils.sha_file_by_name(abspath)
309
def stat_and_sha1(self, abspath):
310
"""Return the stat and sha1 of a file given its absolute path."""
311
file_obj = file(abspath, 'rb')
313
statvalue = os.fstat(file_obj.fileno())
314
sha1 = osutils.sha_file(file_obj)
317
return statvalue, sha1
320
class DirState(object):
321
"""Record directory and metadata state for fast access.
323
A dirstate is a specialised data structure for managing local working
324
tree state information. Its not yet well defined whether it is platform
325
specific, and if it is how we detect/parameterize that.
327
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
328
Unlike most bzr disk formats, DirStates must be locked for reading, using
329
lock_read. (This is an os file lock internally.) This is necessary
330
because the file can be rewritten in place.
332
DirStates must be explicitly written with save() to commit changes; just
333
unlocking them does not write the changes to disk.
336
_kind_to_minikind = {
342
'tree-reference': 't',
344
_minikind_to_kind = {
350
't': 'tree-reference',
352
_stat_to_minikind = {
357
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
358
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
361
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
362
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
363
# that we are sure a single record will always fit.
364
BISECT_PAGE_SIZE = 4096
367
IN_MEMORY_UNMODIFIED = 1
368
IN_MEMORY_MODIFIED = 2
370
# A pack_stat (the x's) that is just noise and will never match the output
373
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
375
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
376
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
378
def __init__(self, path, sha1_provider):
379
"""Create a DirState object.
381
:param path: The path at which the dirstate file on disk should live.
382
:param sha1_provider: an object meeting the SHA1Provider interface.
384
# _header_state and _dirblock_state represent the current state
385
# of the dirstate metadata and the per-row data respectiely.
386
# NOT_IN_MEMORY indicates that no data is in memory
387
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
388
# is the same as is on disk
389
# IN_MEMORY_MODIFIED indicates that we have a modified version
390
# of what is on disk.
391
# In future we will add more granularity, for instance _dirblock_state
392
# will probably support partially-in-memory as a separate variable,
393
# allowing for partially-in-memory unmodified and partially-in-memory
395
self._header_state = DirState.NOT_IN_MEMORY
396
self._dirblock_state = DirState.NOT_IN_MEMORY
397
# If true, an error has been detected while updating the dirstate, and
398
# for safety we're not going to commit to disk.
399
self._changes_aborted = False
403
self._state_file = None
404
self._filename = path
405
self._lock_token = None
406
self._lock_state = None
407
self._id_index = None
408
# a map from packed_stat to sha's.
409
self._packed_stat_index = None
410
self._end_of_header = None
411
self._cutoff_time = None
412
self._split_path_cache = {}
413
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
414
self._sha1_provider = sha1_provider
415
if 'hashcache' in debug.debug_flags:
416
self._sha1_file = self._sha1_file_and_mutter
418
self._sha1_file = self._sha1_provider.sha1
419
# These two attributes provide a simple cache for lookups into the
420
# dirstate in-memory vectors. By probing respectively for the last
421
# block, and for the next entry, we save nearly 2 bisections per path
423
self._last_block_index = None
424
self._last_entry_index = None
428
(self.__class__.__name__, self._filename)
430
def add(self, path, file_id, kind, stat, fingerprint):
431
"""Add a path to be tracked.
433
:param path: The path within the dirstate - '' is the root, 'foo' is the
434
path foo within the root, 'foo/bar' is the path bar within foo
436
:param file_id: The file id of the path being added.
437
:param kind: The kind of the path, as a string like 'file',
439
:param stat: The output of os.lstat for the path.
440
:param fingerprint: The sha value of the file's canonical form (i.e.
441
after any read filters have been applied),
442
or the target of a symlink,
443
or the referenced revision id for tree-references,
444
or '' for directories.
447
# find the block its in.
448
# find the location in the block.
449
# check its not there
451
#------- copied from inventory.ensure_normalized_name - keep synced.
452
# --- normalized_filename wants a unicode basename only, so get one.
453
dirname, basename = osutils.split(path)
454
# we dont import normalized_filename directly because we want to be
455
# able to change the implementation at runtime for tests.
456
norm_name, can_access = osutils.normalized_filename(basename)
457
if norm_name != basename:
461
raise errors.InvalidNormalization(path)
462
# you should never have files called . or ..; just add the directory
463
# in the parent, or according to the special treatment for the root
464
if basename == '.' or basename == '..':
465
raise errors.InvalidEntryName(path)
466
# now that we've normalised, we need the correct utf8 path and
467
# dirname and basename elements. This single encode and split should be
468
# faster than three separate encodes.
469
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
470
dirname, basename = osutils.split(utf8path)
471
# uses __class__ for speed; the check is needed for safety
472
if file_id.__class__ is not str:
473
raise AssertionError(
474
"must be a utf8 file_id not %s" % (type(file_id), ))
475
# Make sure the file_id does not exist in this tree
477
file_id_entry = self._get_entry(0, fileid_utf8=file_id, include_deleted=True)
478
if file_id_entry != (None, None):
479
if file_id_entry[1][0][0] == 'a':
480
if file_id_entry[0] != (dirname, basename, file_id):
481
# set the old name's current operation to rename
482
self.update_minimal(file_id_entry[0],
488
rename_from = file_id_entry[0][0:2]
490
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
491
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
492
info = '%s:%s' % (kind, path)
493
raise errors.DuplicateFileId(file_id, info)
494
first_key = (dirname, basename, '')
495
block_index, present = self._find_block_index_from_key(first_key)
497
# check the path is not in the tree
498
block = self._dirblocks[block_index][1]
499
entry_index, _ = self._find_entry_index(first_key, block)
500
while (entry_index < len(block) and
501
block[entry_index][0][0:2] == first_key[0:2]):
502
if block[entry_index][1][0][0] not in 'ar':
503
# this path is in the dirstate in the current tree.
504
raise Exception, "adding already added path!"
507
# The block where we want to put the file is not present. But it
508
# might be because the directory was empty, or not loaded yet. Look
509
# for a parent entry, if not found, raise NotVersionedError
510
parent_dir, parent_base = osutils.split(dirname)
511
parent_block_idx, parent_entry_idx, _, parent_present = \
512
self._get_block_entry_index(parent_dir, parent_base, 0)
513
if not parent_present:
514
raise errors.NotVersionedError(path, str(self))
515
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
516
block = self._dirblocks[block_index][1]
517
entry_key = (dirname, basename, file_id)
520
packed_stat = DirState.NULLSTAT
523
packed_stat = pack_stat(stat)
524
parent_info = self._empty_parent_info()
525
minikind = DirState._kind_to_minikind[kind]
526
if rename_from is not None:
528
old_path_utf8 = '%s/%s' % rename_from
530
old_path_utf8 = rename_from[1]
531
parent_info[0] = ('r', old_path_utf8, 0, False, '')
533
entry_data = entry_key, [
534
(minikind, fingerprint, size, False, packed_stat),
536
elif kind == 'directory':
537
entry_data = entry_key, [
538
(minikind, '', 0, False, packed_stat),
540
elif kind == 'symlink':
541
entry_data = entry_key, [
542
(minikind, fingerprint, size, False, packed_stat),
544
elif kind == 'tree-reference':
545
entry_data = entry_key, [
546
(minikind, fingerprint, 0, False, packed_stat),
549
raise errors.BzrError('unknown kind %r' % kind)
550
entry_index, present = self._find_entry_index(entry_key, block)
552
block.insert(entry_index, entry_data)
554
if block[entry_index][1][0][0] != 'a':
555
raise AssertionError(" %r(%r) already added" % (basename, file_id))
556
block[entry_index][1][0] = entry_data[1][0]
558
if kind == 'directory':
559
# insert a new dirblock
560
self._ensure_block(block_index, entry_index, utf8path)
561
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
563
self._add_to_id_index(self._id_index, entry_key)
565
def _bisect(self, paths):
566
"""Bisect through the disk structure for specific rows.
568
:param paths: A list of paths to find
569
:return: A dict mapping path => entries for found entries. Missing
570
entries will not be in the map.
571
The list is not sorted, and entries will be populated
572
based on when they were read.
574
self._requires_lock()
575
# We need the file pointer to be right after the initial header block
576
self._read_header_if_needed()
577
# If _dirblock_state was in memory, we should just return info from
578
# there, this function is only meant to handle when we want to read
580
if self._dirblock_state != DirState.NOT_IN_MEMORY:
581
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
583
# The disk representation is generally info + '\0\n\0' at the end. But
584
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
585
# Because it means we can sync on the '\n'
586
state_file = self._state_file
587
file_size = os.fstat(state_file.fileno()).st_size
588
# We end up with 2 extra fields, we should have a trailing '\n' to
589
# ensure that we read the whole record, and we should have a precursur
590
# '' which ensures that we start after the previous '\n'
591
entry_field_count = self._fields_per_entry() + 1
593
low = self._end_of_header
594
high = file_size - 1 # Ignore the final '\0'
595
# Map from (dir, name) => entry
598
# Avoid infinite seeking
599
max_count = 30*len(paths)
601
# pending is a list of places to look.
602
# each entry is a tuple of low, high, dir_names
603
# low -> the first byte offset to read (inclusive)
604
# high -> the last byte offset (inclusive)
605
# dir_names -> The list of (dir, name) pairs that should be found in
606
# the [low, high] range
607
pending = [(low, high, paths)]
609
page_size = self._bisect_page_size
611
fields_to_entry = self._get_fields_to_entry()
614
low, high, cur_files = pending.pop()
616
if not cur_files or low >= high:
621
if count > max_count:
622
raise errors.BzrError('Too many seeks, most likely a bug.')
624
mid = max(low, (low+high-page_size)/2)
627
# limit the read size, so we don't end up reading data that we have
629
read_size = min(page_size, (high-mid)+1)
630
block = state_file.read(read_size)
633
entries = block.split('\n')
636
# We didn't find a '\n', so we cannot have found any records.
637
# So put this range back and try again. But we know we have to
638
# increase the page size, because a single read did not contain
639
# a record break (so records must be larger than page_size)
641
pending.append((low, high, cur_files))
644
# Check the first and last entries, in case they are partial, or if
645
# we don't care about the rest of this page
647
first_fields = entries[0].split('\0')
648
if len(first_fields) < entry_field_count:
649
# We didn't get the complete first entry
650
# so move start, and grab the next, which
651
# should be a full entry
652
start += len(entries[0])+1
653
first_fields = entries[1].split('\0')
656
if len(first_fields) <= 2:
657
# We didn't even get a filename here... what do we do?
658
# Try a large page size and repeat this query
660
pending.append((low, high, cur_files))
663
# Find what entries we are looking for, which occur before and
664
# after this first record.
667
first_path = first_fields[1] + '/' + first_fields[2]
669
first_path = first_fields[2]
670
first_loc = _bisect_path_left(cur_files, first_path)
672
# These exist before the current location
673
pre = cur_files[:first_loc]
674
# These occur after the current location, which may be in the
675
# data we read, or might be after the last entry
676
post = cur_files[first_loc:]
678
if post and len(first_fields) >= entry_field_count:
679
# We have files after the first entry
681
# Parse the last entry
682
last_entry_num = len(entries)-1
683
last_fields = entries[last_entry_num].split('\0')
684
if len(last_fields) < entry_field_count:
685
# The very last hunk was not complete,
686
# read the previous hunk
687
after = mid + len(block) - len(entries[-1])
689
last_fields = entries[last_entry_num].split('\0')
691
after = mid + len(block)
694
last_path = last_fields[1] + '/' + last_fields[2]
696
last_path = last_fields[2]
697
last_loc = _bisect_path_right(post, last_path)
699
middle_files = post[:last_loc]
700
post = post[last_loc:]
703
# We have files that should occur in this block
704
# (>= first, <= last)
705
# Either we will find them here, or we can mark them as
708
if middle_files[0] == first_path:
709
# We might need to go before this location
710
pre.append(first_path)
711
if middle_files[-1] == last_path:
712
post.insert(0, last_path)
714
# Find out what paths we have
715
paths = {first_path:[first_fields]}
716
# last_path might == first_path so we need to be
717
# careful if we should append rather than overwrite
718
if last_entry_num != first_entry_num:
719
paths.setdefault(last_path, []).append(last_fields)
720
for num in xrange(first_entry_num+1, last_entry_num):
721
# TODO: jam 20070223 We are already splitting here, so
722
# shouldn't we just split the whole thing rather
723
# than doing the split again in add_one_record?
724
fields = entries[num].split('\0')
726
path = fields[1] + '/' + fields[2]
729
paths.setdefault(path, []).append(fields)
731
for path in middle_files:
732
for fields in paths.get(path, []):
733
# offset by 1 because of the opening '\0'
734
# consider changing fields_to_entry to avoid the
736
entry = fields_to_entry(fields[1:])
737
found.setdefault(path, []).append(entry)
739
# Now we have split up everything into pre, middle, and post, and
740
# we have handled everything that fell in 'middle'.
741
# We add 'post' first, so that we prefer to seek towards the
742
# beginning, so that we will tend to go as early as we need, and
743
# then only seek forward after that.
745
pending.append((after, high, post))
747
pending.append((low, start-1, pre))
749
# Consider that we may want to return the directory entries in sorted
750
# order. For now, we just return them in whatever order we found them,
751
# and leave it up to the caller if they care if it is ordered or not.
754
def _bisect_dirblocks(self, dir_list):
755
"""Bisect through the disk structure to find entries in given dirs.
757
_bisect_dirblocks is meant to find the contents of directories, which
758
differs from _bisect, which only finds individual entries.
760
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
761
:return: A map from dir => entries_for_dir
763
# TODO: jam 20070223 A lot of the bisecting logic could be shared
764
# between this and _bisect. It would require parameterizing the
765
# inner loop with a function, though. We should evaluate the
766
# performance difference.
767
self._requires_lock()
768
# We need the file pointer to be right after the initial header block
769
self._read_header_if_needed()
770
# If _dirblock_state was in memory, we should just return info from
771
# there, this function is only meant to handle when we want to read
773
if self._dirblock_state != DirState.NOT_IN_MEMORY:
774
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
775
# The disk representation is generally info + '\0\n\0' at the end. But
776
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
777
# Because it means we can sync on the '\n'
778
state_file = self._state_file
779
file_size = os.fstat(state_file.fileno()).st_size
780
# We end up with 2 extra fields, we should have a trailing '\n' to
781
# ensure that we read the whole record, and we should have a precursur
782
# '' which ensures that we start after the previous '\n'
783
entry_field_count = self._fields_per_entry() + 1
785
low = self._end_of_header
786
high = file_size - 1 # Ignore the final '\0'
787
# Map from dir => entry
790
# Avoid infinite seeking
791
max_count = 30*len(dir_list)
793
# pending is a list of places to look.
794
# each entry is a tuple of low, high, dir_names
795
# low -> the first byte offset to read (inclusive)
796
# high -> the last byte offset (inclusive)
797
# dirs -> The list of directories that should be found in
798
# the [low, high] range
799
pending = [(low, high, dir_list)]
801
page_size = self._bisect_page_size
803
fields_to_entry = self._get_fields_to_entry()
806
low, high, cur_dirs = pending.pop()
808
if not cur_dirs or low >= high:
813
if count > max_count:
814
raise errors.BzrError('Too many seeks, most likely a bug.')
816
mid = max(low, (low+high-page_size)/2)
819
# limit the read size, so we don't end up reading data that we have
821
read_size = min(page_size, (high-mid)+1)
822
block = state_file.read(read_size)
825
entries = block.split('\n')
828
# We didn't find a '\n', so we cannot have found any records.
829
# So put this range back and try again. But we know we have to
830
# increase the page size, because a single read did not contain
831
# a record break (so records must be larger than page_size)
833
pending.append((low, high, cur_dirs))
836
# Check the first and last entries, in case they are partial, or if
837
# we don't care about the rest of this page
839
first_fields = entries[0].split('\0')
840
if len(first_fields) < entry_field_count:
841
# We didn't get the complete first entry
842
# so move start, and grab the next, which
843
# should be a full entry
844
start += len(entries[0])+1
845
first_fields = entries[1].split('\0')
848
if len(first_fields) <= 1:
849
# We didn't even get a dirname here... what do we do?
850
# Try a large page size and repeat this query
852
pending.append((low, high, cur_dirs))
855
# Find what entries we are looking for, which occur before and
856
# after this first record.
858
first_dir = first_fields[1]
859
first_loc = bisect.bisect_left(cur_dirs, first_dir)
861
# These exist before the current location
862
pre = cur_dirs[:first_loc]
863
# These occur after the current location, which may be in the
864
# data we read, or might be after the last entry
865
post = cur_dirs[first_loc:]
867
if post and len(first_fields) >= entry_field_count:
868
# We have records to look at after the first entry
870
# Parse the last entry
871
last_entry_num = len(entries)-1
872
last_fields = entries[last_entry_num].split('\0')
873
if len(last_fields) < entry_field_count:
874
# The very last hunk was not complete,
875
# read the previous hunk
876
after = mid + len(block) - len(entries[-1])
878
last_fields = entries[last_entry_num].split('\0')
880
after = mid + len(block)
882
last_dir = last_fields[1]
883
last_loc = bisect.bisect_right(post, last_dir)
885
middle_files = post[:last_loc]
886
post = post[last_loc:]
889
# We have files that should occur in this block
890
# (>= first, <= last)
891
# Either we will find them here, or we can mark them as
894
if middle_files[0] == first_dir:
895
# We might need to go before this location
896
pre.append(first_dir)
897
if middle_files[-1] == last_dir:
898
post.insert(0, last_dir)
900
# Find out what paths we have
901
paths = {first_dir:[first_fields]}
902
# last_dir might == first_dir so we need to be
903
# careful if we should append rather than overwrite
904
if last_entry_num != first_entry_num:
905
paths.setdefault(last_dir, []).append(last_fields)
906
for num in xrange(first_entry_num+1, last_entry_num):
907
# TODO: jam 20070223 We are already splitting here, so
908
# shouldn't we just split the whole thing rather
909
# than doing the split again in add_one_record?
910
fields = entries[num].split('\0')
911
paths.setdefault(fields[1], []).append(fields)
913
for cur_dir in middle_files:
914
for fields in paths.get(cur_dir, []):
915
# offset by 1 because of the opening '\0'
916
# consider changing fields_to_entry to avoid the
918
entry = fields_to_entry(fields[1:])
919
found.setdefault(cur_dir, []).append(entry)
921
# Now we have split up everything into pre, middle, and post, and
922
# we have handled everything that fell in 'middle'.
923
# We add 'post' first, so that we prefer to seek towards the
924
# beginning, so that we will tend to go as early as we need, and
925
# then only seek forward after that.
927
pending.append((after, high, post))
929
pending.append((low, start-1, pre))
933
def _bisect_recursive(self, paths):
934
"""Bisect for entries for all paths and their children.
936
This will use bisect to find all records for the supplied paths. It
937
will then continue to bisect for any records which are marked as
938
directories. (and renames?)
940
:param paths: A sorted list of (dir, name) pairs
941
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
942
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
944
# Map from (dir, name, file_id) => [tree_info]
947
found_dir_names = set()
949
# Directories that have been read
950
processed_dirs = set()
951
# Get the ball rolling with the first bisect for all entries.
952
newly_found = self._bisect(paths)
955
# Directories that need to be read
957
paths_to_search = set()
958
for entry_list in newly_found.itervalues():
959
for dir_name_id, trees_info in entry_list:
960
found[dir_name_id] = trees_info
961
found_dir_names.add(dir_name_id[:2])
963
for tree_info in trees_info:
964
minikind = tree_info[0]
967
# We already processed this one as a directory,
968
# we don't need to do the extra work again.
970
subdir, name, file_id = dir_name_id
971
path = osutils.pathjoin(subdir, name)
973
if path not in processed_dirs:
974
pending_dirs.add(path)
975
elif minikind == 'r':
976
# Rename, we need to directly search the target
977
# which is contained in the fingerprint column
978
dir_name = osutils.split(tree_info[1])
979
if dir_name[0] in pending_dirs:
980
# This entry will be found in the dir search
982
if dir_name not in found_dir_names:
983
paths_to_search.add(tree_info[1])
984
# Now we have a list of paths to look for directly, and
985
# directory blocks that need to be read.
986
# newly_found is mixing the keys between (dir, name) and path
987
# entries, but that is okay, because we only really care about the
989
newly_found = self._bisect(sorted(paths_to_search))
990
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
991
processed_dirs.update(pending_dirs)
994
def _discard_merge_parents(self):
995
"""Discard any parents trees beyond the first.
997
Note that if this fails the dirstate is corrupted.
999
After this function returns the dirstate contains 2 trees, neither of
1002
self._read_header_if_needed()
1003
parents = self.get_parent_ids()
1004
if len(parents) < 1:
1006
# only require all dirblocks if we are doing a full-pass removal.
1007
self._read_dirblocks_if_needed()
1008
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
1009
def iter_entries_removable():
1010
for block in self._dirblocks:
1011
deleted_positions = []
1012
for pos, entry in enumerate(block[1]):
1014
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
1015
deleted_positions.append(pos)
1016
if deleted_positions:
1017
if len(deleted_positions) == len(block[1]):
1020
for pos in reversed(deleted_positions):
1022
# if the first parent is a ghost:
1023
if parents[0] in self.get_ghosts():
1024
empty_parent = [DirState.NULL_PARENT_DETAILS]
1025
for entry in iter_entries_removable():
1026
entry[1][1:] = empty_parent
1028
for entry in iter_entries_removable():
1032
self._parents = [parents[0]]
1033
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1034
self._header_state = DirState.IN_MEMORY_MODIFIED
1036
def _empty_parent_info(self):
1037
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
1040
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
1041
"""Ensure a block for dirname exists.
1043
This function exists to let callers which know that there is a
1044
directory dirname ensure that the block for it exists. This block can
1045
fail to exist because of demand loading, or because a directory had no
1046
children. In either case it is not an error. It is however an error to
1047
call this if there is no parent entry for the directory, and thus the
1048
function requires the coordinates of such an entry to be provided.
1050
The root row is special cased and can be indicated with a parent block
1053
:param parent_block_index: The index of the block in which dirname's row
1055
:param parent_row_index: The index in the parent block where the row
1057
:param dirname: The utf8 dirname to ensure there is a block for.
1058
:return: The index for the block.
1060
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
1061
# This is the signature of the root row, and the
1062
# contents-of-root row is always index 1
1064
# the basename of the directory must be the end of its full name.
1065
if not (parent_block_index == -1 and
1066
parent_block_index == -1 and dirname == ''):
1067
if not dirname.endswith(
1068
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
1069
raise AssertionError("bad dirname %r" % dirname)
1070
block_index, present = self._find_block_index_from_key((dirname, '', ''))
1072
## In future, when doing partial parsing, this should load and
1073
# populate the entire block.
1074
self._dirblocks.insert(block_index, (dirname, []))
1077
def _entries_to_current_state(self, new_entries):
1078
"""Load new_entries into self.dirblocks.
1080
Process new_entries into the current state object, making them the active
1081
state. The entries are grouped together by directory to form dirblocks.
1083
:param new_entries: A sorted list of entries. This function does not sort
1084
to prevent unneeded overhead when callers have a sorted list already.
1087
if new_entries[0][0][0:2] != ('', ''):
1088
raise AssertionError(
1089
"Missing root row %r" % (new_entries[0][0],))
1090
# The two blocks here are deliberate: the root block and the
1091
# contents-of-root block.
1092
self._dirblocks = [('', []), ('', [])]
1093
current_block = self._dirblocks[0][1]
1094
current_dirname = ''
1096
append_entry = current_block.append
1097
for entry in new_entries:
1098
if entry[0][0] != current_dirname:
1099
# new block - different dirname
1101
current_dirname = entry[0][0]
1102
self._dirblocks.append((current_dirname, current_block))
1103
append_entry = current_block.append
1104
# append the entry to the current block
1106
self._split_root_dirblock_into_contents()
1108
def _split_root_dirblock_into_contents(self):
1109
"""Split the root dirblocks into root and contents-of-root.
1111
After parsing by path, we end up with root entries and contents-of-root
1112
entries in the same block. This loop splits them out again.
1114
# The above loop leaves the "root block" entries mixed with the
1115
# "contents-of-root block". But we don't want an if check on
1116
# all entries, so instead we just fix it up here.
1117
if self._dirblocks[1] != ('', []):
1118
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1120
contents_of_root_block = []
1121
for entry in self._dirblocks[0][1]:
1122
if not entry[0][1]: # This is a root entry
1123
root_block.append(entry)
1125
contents_of_root_block.append(entry)
1126
self._dirblocks[0] = ('', root_block)
1127
self._dirblocks[1] = ('', contents_of_root_block)
1129
def _entries_for_path(self, path):
1130
"""Return a list with all the entries that match path for all ids."""
1131
dirname, basename = os.path.split(path)
1132
key = (dirname, basename, '')
1133
block_index, present = self._find_block_index_from_key(key)
1135
# the block which should contain path is absent.
1138
block = self._dirblocks[block_index][1]
1139
entry_index, _ = self._find_entry_index(key, block)
1140
# we may need to look at multiple entries at this path: walk while the specific_files match.
1141
while (entry_index < len(block) and
1142
block[entry_index][0][0:2] == key[0:2]):
1143
result.append(block[entry_index])
1147
def _entry_to_line(self, entry):
1148
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1150
:param entry: An entry_tuple as defined in the module docstring.
1152
entire_entry = list(entry[0])
1153
for tree_number, tree_data in enumerate(entry[1]):
1154
# (minikind, fingerprint, size, executable, tree_specific_string)
1155
entire_entry.extend(tree_data)
1156
# 3 for the key, 5 for the fields per tree.
1157
tree_offset = 3 + tree_number * 5
1159
entire_entry[tree_offset + 0] = tree_data[0]
1161
entire_entry[tree_offset + 2] = str(tree_data[2])
1163
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1164
return '\0'.join(entire_entry)
1166
def _fields_per_entry(self):
1167
"""How many null separated fields should be in each entry row.
1169
Each line now has an extra '\n' field which is not used
1170
so we just skip over it
1172
3 fields for the key
1173
+ number of fields per tree_data (5) * tree count
1176
tree_count = 1 + self._num_present_parents()
1177
return 3 + 5 * tree_count + 1
1179
def _find_block(self, key, add_if_missing=False):
1180
"""Return the block that key should be present in.
1182
:param key: A dirstate entry key.
1183
:return: The block tuple.
1185
block_index, present = self._find_block_index_from_key(key)
1187
if not add_if_missing:
1188
# check to see if key is versioned itself - we might want to
1189
# add it anyway, because dirs with no entries dont get a
1190
# dirblock at parse time.
1191
# This is an uncommon branch to take: most dirs have children,
1192
# and most code works with versioned paths.
1193
parent_base, parent_name = osutils.split(key[0])
1194
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1195
# some parent path has not been added - its an error to add
1197
raise errors.NotVersionedError(key[0:2], str(self))
1198
self._dirblocks.insert(block_index, (key[0], []))
1199
return self._dirblocks[block_index]
1201
def _find_block_index_from_key(self, key):
1202
"""Find the dirblock index for a key.
1204
:return: The block index, True if the block for the key is present.
1206
if key[0:2] == ('', ''):
1209
if (self._last_block_index is not None and
1210
self._dirblocks[self._last_block_index][0] == key[0]):
1211
return self._last_block_index, True
1214
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1215
cache=self._split_path_cache)
1216
# _right returns one-past-where-key is so we have to subtract
1217
# one to use it. we use _right here because there are two
1218
# '' blocks - the root, and the contents of root
1219
# we always have a minimum of 2 in self._dirblocks: root and
1220
# root-contents, and for '', we get 2 back, so this is
1221
# simple and correct:
1222
present = (block_index < len(self._dirblocks) and
1223
self._dirblocks[block_index][0] == key[0])
1224
self._last_block_index = block_index
1225
# Reset the entry index cache to the beginning of the block.
1226
self._last_entry_index = -1
1227
return block_index, present
1229
def _find_entry_index(self, key, block):
1230
"""Find the entry index for a key in a block.
1232
:return: The entry index, True if the entry for the key is present.
1234
len_block = len(block)
1236
if self._last_entry_index is not None:
1238
entry_index = self._last_entry_index + 1
1239
# A hit is when the key is after the last slot, and before or
1240
# equal to the next slot.
1241
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1242
key <= block[entry_index][0]):
1243
self._last_entry_index = entry_index
1244
present = (block[entry_index][0] == key)
1245
return entry_index, present
1248
entry_index = bisect.bisect_left(block, (key, []))
1249
present = (entry_index < len_block and
1250
block[entry_index][0] == key)
1251
self._last_entry_index = entry_index
1252
return entry_index, present
1255
def from_tree(tree, dir_state_filename, sha1_provider=None):
1256
"""Create a dirstate from a bzr Tree.
1258
:param tree: The tree which should provide parent information and
1260
:param sha1_provider: an object meeting the SHA1Provider interface.
1261
If None, a DefaultSHA1Provider is used.
1262
:return: a DirState object which is currently locked for writing.
1263
(it was locked by DirState.initialize)
1265
result = DirState.initialize(dir_state_filename,
1266
sha1_provider=sha1_provider)
1270
parent_ids = tree.get_parent_ids()
1271
num_parents = len(parent_ids)
1273
for parent_id in parent_ids:
1274
parent_tree = tree.branch.repository.revision_tree(parent_id)
1275
parent_trees.append((parent_id, parent_tree))
1276
parent_tree.lock_read()
1277
result.set_parent_trees(parent_trees, [])
1278
result.set_state_from_inventory(tree.inventory)
1280
for revid, parent_tree in parent_trees:
1281
parent_tree.unlock()
1284
# The caller won't have a chance to unlock this, so make sure we
1290
def update_by_delta(self, delta):
1291
"""Apply an inventory delta to the dirstate for tree 0
1293
This is the workhorse for apply_inventory_delta in dirstate based
1296
:param delta: An inventory delta. See Inventory.apply_delta for
1299
self._read_dirblocks_if_needed()
1300
encode = cache_utf8.encode
1303
# Accumulate parent references (path_utf8, id), to check for parentless
1304
# items or items placed under files/links/tree-references. We get
1305
# references from every item in the delta that is not a deletion and
1306
# is not itself the root.
1308
# Added ids must not be in the dirstate already. This set holds those
1311
# This loop transforms the delta to single atomic operations that can
1312
# be executed and validated.
1313
for old_path, new_path, file_id, inv_entry in sorted(
1314
inventory._check_delta_unique_old_paths(
1315
inventory._check_delta_unique_new_paths(
1316
inventory._check_delta_ids_match_entry(
1317
inventory._check_delta_ids_are_valid(
1318
inventory._check_delta_new_path_entry_both_or_None(delta))))),
1320
if (file_id in insertions) or (file_id in removals):
1321
raise errors.InconsistentDelta(old_path or new_path, file_id,
1323
if old_path is not None:
1324
old_path = old_path.encode('utf-8')
1325
removals[file_id] = old_path
1327
new_ids.add(file_id)
1328
if new_path is not None:
1329
if inv_entry is None:
1330
raise errors.InconsistentDelta(new_path, file_id,
1331
"new_path with no entry")
1332
new_path = new_path.encode('utf-8')
1333
dirname_utf8, basename = osutils.split(new_path)
1335
parents.add((dirname_utf8, inv_entry.parent_id))
1336
key = (dirname_utf8, basename, file_id)
1337
minikind = DirState._kind_to_minikind[inv_entry.kind]
1339
fingerprint = inv_entry.reference_revision or ''
1342
insertions[file_id] = (key, minikind, inv_entry.executable,
1343
fingerprint, new_path)
1344
# Transform moves into delete+add pairs
1345
if None not in (old_path, new_path):
1346
for child in self._iter_child_entries(0, old_path):
1347
if child[0][2] in insertions or child[0][2] in removals:
1349
child_dirname = child[0][0]
1350
child_basename = child[0][1]
1351
minikind = child[1][0][0]
1352
fingerprint = child[1][0][4]
1353
executable = child[1][0][3]
1354
old_child_path = osutils.pathjoin(child_dirname,
1356
removals[child[0][2]] = old_child_path
1357
child_suffix = child_dirname[len(old_path):]
1358
new_child_dirname = (new_path + child_suffix)
1359
key = (new_child_dirname, child_basename, child[0][2])
1360
new_child_path = osutils.pathjoin(new_child_dirname,
1362
insertions[child[0][2]] = (key, minikind, executable,
1363
fingerprint, new_child_path)
1364
self._check_delta_ids_absent(new_ids, delta, 0)
1366
self._apply_removals(removals.iteritems())
1367
self._apply_insertions(insertions.values())
1369
self._after_delta_check_parents(parents, 0)
1370
except errors.BzrError, e:
1371
self._changes_aborted = True
1372
if 'integrity error' not in str(e):
1374
# _get_entry raises BzrError when a request is inconsistent; we
1375
# want such errors to be shown as InconsistentDelta - and that
1376
# fits the behaviour we trigger.
1377
raise errors.InconsistentDeltaDelta(delta, "error from _get_entry.")
1379
def _apply_removals(self, removals):
1380
for file_id, path in sorted(removals, reverse=True,
1381
key=operator.itemgetter(1)):
1382
dirname, basename = osutils.split(path)
1383
block_i, entry_i, d_present, f_present = \
1384
self._get_block_entry_index(dirname, basename, 0)
1386
entry = self._dirblocks[block_i][1][entry_i]
1388
self._changes_aborted = True
1389
raise errors.InconsistentDelta(path, file_id,
1390
"Wrong path for old path.")
1391
if not f_present or entry[1][0][0] in 'ar':
1392
self._changes_aborted = True
1393
raise errors.InconsistentDelta(path, file_id,
1394
"Wrong path for old path.")
1395
if file_id != entry[0][2]:
1396
self._changes_aborted = True
1397
raise errors.InconsistentDelta(path, file_id,
1398
"Attempt to remove path has wrong id - found %r."
1400
self._make_absent(entry)
1401
# See if we have a malformed delta: deleting a directory must not
1402
# leave crud behind. This increases the number of bisects needed
1403
# substantially, but deletion or renames of large numbers of paths
1404
# is rare enough it shouldn't be an issue (famous last words?) RBC
1406
block_i, entry_i, d_present, f_present = \
1407
self._get_block_entry_index(path, '', 0)
1409
# The dir block is still present in the dirstate; this could
1410
# be due to it being in a parent tree, or a corrupt delta.
1411
for child_entry in self._dirblocks[block_i][1]:
1412
if child_entry[1][0][0] not in ('r', 'a'):
1413
self._changes_aborted = True
1414
raise errors.InconsistentDelta(path, entry[0][2],
1415
"The file id was deleted but its children were "
1418
def _apply_insertions(self, adds):
1420
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1421
self.update_minimal(key, minikind, executable, fingerprint,
1422
path_utf8=path_utf8)
1423
except errors.NotVersionedError:
1424
self._changes_aborted = True
1425
raise errors.InconsistentDelta(path_utf8.decode('utf8'), key[2],
1428
def update_basis_by_delta(self, delta, new_revid):
1429
"""Update the parents of this tree after a commit.
1431
This gives the tree one parent, with revision id new_revid. The
1432
inventory delta is applied to the current basis tree to generate the
1433
inventory for the parent new_revid, and all other parent trees are
1436
Note that an exception during the operation of this method will leave
1437
the dirstate in a corrupt state where it should not be saved.
1439
Finally, we expect all changes to be synchronising the basis tree with
1442
:param new_revid: The new revision id for the trees parent.
1443
:param delta: An inventory delta (see apply_inventory_delta) describing
1444
the changes from the current left most parent revision to new_revid.
1446
self._read_dirblocks_if_needed()
1447
self._discard_merge_parents()
1448
if self._ghosts != []:
1449
raise NotImplementedError(self.update_basis_by_delta)
1450
if len(self._parents) == 0:
1451
# setup a blank tree, the most simple way.
1452
empty_parent = DirState.NULL_PARENT_DETAILS
1453
for entry in self._iter_entries():
1454
entry[1].append(empty_parent)
1455
self._parents.append(new_revid)
1457
self._parents[0] = new_revid
1459
delta = sorted(delta, reverse=True)
1463
# The paths this function accepts are unicode and must be encoded as we
1465
encode = cache_utf8.encode
1466
inv_to_entry = self._inv_entry_to_details
1467
# delta is now (deletes, changes), (adds) in reverse lexographical
1469
# deletes in reverse lexographic order are safe to process in situ.
1470
# renames are not, as a rename from any path could go to a path
1471
# lexographically lower, so we transform renames into delete, add pairs,
1472
# expanding them recursively as needed.
1473
# At the same time, to reduce interface friction we convert the input
1474
# inventory entries to dirstate.
1475
root_only = ('', '')
1476
# Accumulate parent references (path_utf8, id), to check for parentless
1477
# items or items placed under files/links/tree-references. We get
1478
# references from every item in the delta that is not a deletion and
1479
# is not itself the root.
1481
# Added ids must not be in the dirstate already. This set holds those
1484
for old_path, new_path, file_id, inv_entry in delta:
1485
if inv_entry is not None and file_id != inv_entry.file_id:
1486
raise errors.InconsistentDelta(new_path, file_id,
1487
"mismatched entry file_id %r" % inv_entry)
1488
if new_path is not None:
1489
if inv_entry is None:
1490
raise errors.InconsistentDelta(new_path, file_id,
1491
"new_path with no entry")
1492
new_path_utf8 = encode(new_path)
1493
# note the parent for validation
1494
dirname_utf8, basename_utf8 = osutils.split(new_path_utf8)
1496
parents.add((dirname_utf8, inv_entry.parent_id))
1497
if old_path is None:
1498
adds.append((None, encode(new_path), file_id,
1499
inv_to_entry(inv_entry), True))
1500
new_ids.add(file_id)
1501
elif new_path is None:
1502
deletes.append((encode(old_path), None, file_id, None, True))
1503
elif (old_path, new_path) != root_only:
1505
# Because renames must preserve their children we must have
1506
# processed all relocations and removes before hand. The sort
1507
# order ensures we've examined the child paths, but we also
1508
# have to execute the removals, or the split to an add/delete
1509
# pair will result in the deleted item being reinserted, or
1510
# renamed items being reinserted twice - and possibly at the
1511
# wrong place. Splitting into a delete/add pair also simplifies
1512
# the handling of entries with ('f', ...), ('r' ...) because
1513
# the target of the 'r' is old_path here, and we add that to
1514
# deletes, meaning that the add handler does not need to check
1515
# for 'r' items on every pass.
1516
self._update_basis_apply_deletes(deletes)
1518
# Split into an add/delete pair recursively.
1519
adds.append((None, new_path_utf8, file_id,
1520
inv_to_entry(inv_entry), False))
1521
# Expunge deletes that we've seen so that deleted/renamed
1522
# children of a rename directory are handled correctly.
1523
new_deletes = reversed(list(self._iter_child_entries(1,
1525
# Remove the current contents of the tree at orig_path, and
1526
# reinsert at the correct new path.
1527
for entry in new_deletes:
1529
source_path = entry[0][0] + '/' + entry[0][1]
1531
source_path = entry[0][1]
1533
target_path = new_path_utf8 + source_path[len(old_path):]
1536
raise AssertionError("cannot rename directory to"
1538
target_path = source_path[len(old_path) + 1:]
1539
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1541
(source_path, target_path, entry[0][2], None, False))
1543
(encode(old_path), new_path, file_id, None, False))
1545
# changes to just the root should not require remove/insertion
1547
changes.append((encode(old_path), encode(new_path), file_id,
1548
inv_to_entry(inv_entry)))
1549
self._check_delta_ids_absent(new_ids, delta, 1)
1551
# Finish expunging deletes/first half of renames.
1552
self._update_basis_apply_deletes(deletes)
1553
# Reinstate second half of renames and new paths.
1554
self._update_basis_apply_adds(adds)
1555
# Apply in-situ changes.
1556
self._update_basis_apply_changes(changes)
1558
self._after_delta_check_parents(parents, 1)
1559
except errors.BzrError, e:
1560
self._changes_aborted = True
1561
if 'integrity error' not in str(e):
1563
# _get_entry raises BzrError when a request is inconsistent; we
1564
# want such errors to be shown as InconsistentDelta - and that
1565
# fits the behaviour we trigger. Partof this is driven by dirstate
1566
# only supporting deltas that turn the basis into a closer fit to
1568
raise errors.InconsistentDeltaDelta(delta, "error from _get_entry.")
1570
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1571
self._header_state = DirState.IN_MEMORY_MODIFIED
1572
self._id_index = None
1575
def _check_delta_ids_absent(self, new_ids, delta, tree_index):
1576
"""Check that none of the file_ids in new_ids are present in a tree."""
1579
id_index = self._get_id_index()
1580
for file_id in new_ids:
1581
for key in id_index.get(file_id, ()):
1582
block_i, entry_i, d_present, f_present = \
1583
self._get_block_entry_index(key[0], key[1], tree_index)
1585
# In a different tree
1587
entry = self._dirblocks[block_i][1][entry_i]
1588
if entry[0][2] != file_id:
1589
# Different file_id, so not what we want.
1591
# NB: No changes made before this helper is called, so no need
1592
# to set the _changes_aborted flag.
1593
raise errors.InconsistentDelta(
1594
("%s/%s" % key[0:2]).decode('utf8'), file_id,
1595
"This file_id is new in the delta but already present in "
1598
def _update_basis_apply_adds(self, adds):
1599
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1601
They may be adds, or renames that have been split into add/delete
1604
:param adds: A sequence of adds. Each add is a tuple:
1605
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1606
is False when the add is the second half of a remove-and-reinsert
1607
pair created to handle renames and deletes.
1609
# Adds are accumulated partly from renames, so can be in any input
1612
# adds is now in lexographic order, which places all parents before
1613
# their children, so we can process it linearly.
1615
for old_path, new_path, file_id, new_details, real_add in adds:
1616
# the entry for this file_id must be in tree 0.
1617
entry = self._get_entry(0, file_id, new_path)
1618
if entry[0] is None or entry[0][2] != file_id:
1619
self._changes_aborted = True
1620
raise errors.InconsistentDelta(new_path, file_id,
1621
'working tree does not contain new entry')
1622
if real_add and entry[1][1][0] not in absent:
1623
self._changes_aborted = True
1624
raise errors.InconsistentDelta(new_path, file_id,
1625
'The entry was considered to be a genuinely new record,'
1626
' but there was already an old record for it.')
1627
# We don't need to update the target of an 'r' because the handling
1628
# of renames turns all 'r' situations into a delete at the original
1630
entry[1][1] = new_details
1632
def _update_basis_apply_changes(self, changes):
1633
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1635
:param adds: A sequence of changes. Each change is a tuple:
1636
(path_utf8, path_utf8, file_id, (entry_details))
1639
for old_path, new_path, file_id, new_details in changes:
1640
# the entry for this file_id must be in tree 0.
1641
entry = self._get_entry(0, file_id, new_path)
1642
if entry[0] is None or entry[0][2] != file_id:
1643
self._changes_aborted = True
1644
raise errors.InconsistentDelta(new_path, file_id,
1645
'working tree does not contain new entry')
1646
if (entry[1][0][0] in absent or
1647
entry[1][1][0] in absent):
1648
self._changes_aborted = True
1649
raise errors.InconsistentDelta(new_path, file_id,
1650
'changed considered absent')
1651
entry[1][1] = new_details
1653
def _update_basis_apply_deletes(self, deletes):
1654
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1656
They may be deletes, or renames that have been split into add/delete
1659
:param deletes: A sequence of deletes. Each delete is a tuple:
1660
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1661
real_delete is True when the desired outcome is an actual deletion
1662
rather than the rename handling logic temporarily deleting a path
1663
during the replacement of a parent.
1665
null = DirState.NULL_PARENT_DETAILS
1666
for old_path, new_path, file_id, _, real_delete in deletes:
1667
if real_delete != (new_path is None):
1668
self._changes_aborted = True
1669
raise AssertionError("bad delete delta")
1670
# the entry for this file_id must be in tree 1.
1671
dirname, basename = osutils.split(old_path)
1672
block_index, entry_index, dir_present, file_present = \
1673
self._get_block_entry_index(dirname, basename, 1)
1674
if not file_present:
1675
self._changes_aborted = True
1676
raise errors.InconsistentDelta(old_path, file_id,
1677
'basis tree does not contain removed entry')
1678
entry = self._dirblocks[block_index][1][entry_index]
1679
if entry[0][2] != file_id:
1680
self._changes_aborted = True
1681
raise errors.InconsistentDelta(old_path, file_id,
1682
'mismatched file_id in tree 1')
1684
if entry[1][0][0] != 'a':
1685
self._changes_aborted = True
1686
raise errors.InconsistentDelta(old_path, file_id,
1687
'This was marked as a real delete, but the WT state'
1688
' claims that it still exists and is versioned.')
1689
del self._dirblocks[block_index][1][entry_index]
1691
if entry[1][0][0] == 'a':
1692
self._changes_aborted = True
1693
raise errors.InconsistentDelta(old_path, file_id,
1694
'The entry was considered a rename, but the source path'
1695
' is marked as absent.')
1696
# For whatever reason, we were asked to rename an entry
1697
# that was originally marked as deleted. This could be
1698
# because we are renaming the parent directory, and the WT
1699
# current state has the file marked as deleted.
1700
elif entry[1][0][0] == 'r':
1701
# implement the rename
1702
del self._dirblocks[block_index][1][entry_index]
1704
# it is being resurrected here, so blank it out temporarily.
1705
self._dirblocks[block_index][1][entry_index][1][1] = null
1707
def _after_delta_check_parents(self, parents, index):
1708
"""Check that parents required by the delta are all intact.
1710
:param parents: An iterable of (path_utf8, file_id) tuples which are
1711
required to be present in tree 'index' at path_utf8 with id file_id
1713
:param index: The column in the dirstate to check for parents in.
1715
for dirname_utf8, file_id in parents:
1716
# Get the entry - the ensures that file_id, dirname_utf8 exists and
1717
# has the right file id.
1718
entry = self._get_entry(index, file_id, dirname_utf8)
1719
if entry[1] is None:
1720
self._changes_aborted = True
1721
raise errors.InconsistentDelta(dirname_utf8.decode('utf8'),
1722
file_id, "This parent is not present.")
1723
# Parents of things must be directories
1724
if entry[1][index][0] != 'd':
1725
self._changes_aborted = True
1726
raise errors.InconsistentDelta(dirname_utf8.decode('utf8'),
1727
file_id, "This parent is not a directory.")
1729
def _observed_sha1(self, entry, sha1, stat_value,
1730
_stat_to_minikind=_stat_to_minikind, _pack_stat=pack_stat):
1731
"""Note the sha1 of a file.
1733
:param entry: The entry the sha1 is for.
1734
:param sha1: The observed sha1.
1735
:param stat_value: The os.lstat for the file.
1738
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1742
packed_stat = _pack_stat(stat_value)
1744
if self._cutoff_time is None:
1745
self._sha_cutoff_time()
1746
if (stat_value.st_mtime < self._cutoff_time
1747
and stat_value.st_ctime < self._cutoff_time):
1748
entry[1][0] = ('f', sha1, stat_value.st_size, entry[1][0][3],
1750
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1752
def _sha_cutoff_time(self):
1753
"""Return cutoff time.
1755
Files modified more recently than this time are at risk of being
1756
undetectably modified and so can't be cached.
1758
# Cache the cutoff time as long as we hold a lock.
1759
# time.time() isn't super expensive (approx 3.38us), but
1760
# when you call it 50,000 times it adds up.
1761
# For comparison, os.lstat() costs 7.2us if it is hot.
1762
self._cutoff_time = int(time.time()) - 3
1763
return self._cutoff_time
1765
def _lstat(self, abspath, entry):
1766
"""Return the os.lstat value for this path."""
1767
return os.lstat(abspath)
1769
def _sha1_file_and_mutter(self, abspath):
1770
# when -Dhashcache is turned on, this is monkey-patched in to log
1772
trace.mutter("dirstate sha1 " + abspath)
1773
return self._sha1_provider.sha1(abspath)
1775
def _is_executable(self, mode, old_executable):
1776
"""Is this file executable?"""
1777
return bool(S_IEXEC & mode)
1779
def _is_executable_win32(self, mode, old_executable):
1780
"""On win32 the executable bit is stored in the dirstate."""
1781
return old_executable
1783
if sys.platform == 'win32':
1784
_is_executable = _is_executable_win32
1786
def _read_link(self, abspath, old_link):
1787
"""Read the target of a symlink"""
1788
# TODO: jam 200700301 On Win32, this could just return the value
1789
# already in memory. However, this really needs to be done at a
1790
# higher level, because there either won't be anything on disk,
1791
# or the thing on disk will be a file.
1792
fs_encoding = osutils._fs_enc
1793
if isinstance(abspath, unicode):
1794
# abspath is defined as the path to pass to lstat. readlink is
1795
# buggy in python < 2.6 (it doesn't encode unicode path into FS
1796
# encoding), so we need to encode ourselves knowing that unicode
1797
# paths are produced by UnicodeDirReader on purpose.
1798
abspath = abspath.encode(fs_encoding)
1799
target = os.readlink(abspath)
1800
if fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1801
# Change encoding if needed
1802
target = target.decode(fs_encoding).encode('UTF-8')
1805
def get_ghosts(self):
1806
"""Return a list of the parent tree revision ids that are ghosts."""
1807
self._read_header_if_needed()
1810
def get_lines(self):
1811
"""Serialise the entire dirstate to a sequence of lines."""
1812
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1813
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1814
# read whats on disk.
1815
self._state_file.seek(0)
1816
return self._state_file.readlines()
1818
lines.append(self._get_parents_line(self.get_parent_ids()))
1819
lines.append(self._get_ghosts_line(self._ghosts))
1820
# append the root line which is special cased
1821
lines.extend(map(self._entry_to_line, self._iter_entries()))
1822
return self._get_output_lines(lines)
1824
def _get_ghosts_line(self, ghost_ids):
1825
"""Create a line for the state file for ghost information."""
1826
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1828
def _get_parents_line(self, parent_ids):
1829
"""Create a line for the state file for parents information."""
1830
return '\0'.join([str(len(parent_ids))] + parent_ids)
1832
def _get_fields_to_entry(self):
1833
"""Get a function which converts entry fields into a entry record.
1835
This handles size and executable, as well as parent records.
1837
:return: A function which takes a list of fields, and returns an
1838
appropriate record for storing in memory.
1840
# This is intentionally unrolled for performance
1841
num_present_parents = self._num_present_parents()
1842
if num_present_parents == 0:
1843
def fields_to_entry_0_parents(fields, _int=int):
1844
path_name_file_id_key = (fields[0], fields[1], fields[2])
1845
return (path_name_file_id_key, [
1847
fields[3], # minikind
1848
fields[4], # fingerprint
1849
_int(fields[5]), # size
1850
fields[6] == 'y', # executable
1851
fields[7], # packed_stat or revision_id
1853
return fields_to_entry_0_parents
1854
elif num_present_parents == 1:
1855
def fields_to_entry_1_parent(fields, _int=int):
1856
path_name_file_id_key = (fields[0], fields[1], fields[2])
1857
return (path_name_file_id_key, [
1859
fields[3], # minikind
1860
fields[4], # fingerprint
1861
_int(fields[5]), # size
1862
fields[6] == 'y', # executable
1863
fields[7], # packed_stat or revision_id
1866
fields[8], # minikind
1867
fields[9], # fingerprint
1868
_int(fields[10]), # size
1869
fields[11] == 'y', # executable
1870
fields[12], # packed_stat or revision_id
1873
return fields_to_entry_1_parent
1874
elif num_present_parents == 2:
1875
def fields_to_entry_2_parents(fields, _int=int):
1876
path_name_file_id_key = (fields[0], fields[1], fields[2])
1877
return (path_name_file_id_key, [
1879
fields[3], # minikind
1880
fields[4], # fingerprint
1881
_int(fields[5]), # size
1882
fields[6] == 'y', # executable
1883
fields[7], # packed_stat or revision_id
1886
fields[8], # minikind
1887
fields[9], # fingerprint
1888
_int(fields[10]), # size
1889
fields[11] == 'y', # executable
1890
fields[12], # packed_stat or revision_id
1893
fields[13], # minikind
1894
fields[14], # fingerprint
1895
_int(fields[15]), # size
1896
fields[16] == 'y', # executable
1897
fields[17], # packed_stat or revision_id
1900
return fields_to_entry_2_parents
1902
def fields_to_entry_n_parents(fields, _int=int):
1903
path_name_file_id_key = (fields[0], fields[1], fields[2])
1904
trees = [(fields[cur], # minikind
1905
fields[cur+1], # fingerprint
1906
_int(fields[cur+2]), # size
1907
fields[cur+3] == 'y', # executable
1908
fields[cur+4], # stat or revision_id
1909
) for cur in xrange(3, len(fields)-1, 5)]
1910
return path_name_file_id_key, trees
1911
return fields_to_entry_n_parents
1913
def get_parent_ids(self):
1914
"""Return a list of the parent tree ids for the directory state."""
1915
self._read_header_if_needed()
1916
return list(self._parents)
1918
def _get_block_entry_index(self, dirname, basename, tree_index):
1919
"""Get the coordinates for a path in the state structure.
1921
:param dirname: The utf8 dirname to lookup.
1922
:param basename: The utf8 basename to lookup.
1923
:param tree_index: The index of the tree for which this lookup should
1925
:return: A tuple describing where the path is located, or should be
1926
inserted. The tuple contains four fields: the block index, the row
1927
index, the directory is present (boolean), the entire path is
1928
present (boolean). There is no guarantee that either
1929
coordinate is currently reachable unless the found field for it is
1930
True. For instance, a directory not present in the searched tree
1931
may be returned with a value one greater than the current highest
1932
block offset. The directory present field will always be True when
1933
the path present field is True. The directory present field does
1934
NOT indicate that the directory is present in the searched tree,
1935
rather it indicates that there are at least some files in some
1938
self._read_dirblocks_if_needed()
1939
key = dirname, basename, ''
1940
block_index, present = self._find_block_index_from_key(key)
1942
# no such directory - return the dir index and 0 for the row.
1943
return block_index, 0, False, False
1944
block = self._dirblocks[block_index][1] # access the entries only
1945
entry_index, present = self._find_entry_index(key, block)
1946
# linear search through entries at this path to find the one
1948
while entry_index < len(block) and block[entry_index][0][1] == basename:
1949
if block[entry_index][1][tree_index][0] not in 'ar':
1950
# neither absent or relocated
1951
return block_index, entry_index, True, True
1953
return block_index, entry_index, True, False
1955
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None, include_deleted=False):
1956
"""Get the dirstate entry for path in tree tree_index.
1958
If either file_id or path is supplied, it is used as the key to lookup.
1959
If both are supplied, the fastest lookup is used, and an error is
1960
raised if they do not both point at the same row.
1962
:param tree_index: The index of the tree we wish to locate this path
1963
in. If the path is present in that tree, the entry containing its
1964
details is returned, otherwise (None, None) is returned
1965
0 is the working tree, higher indexes are successive parent
1967
:param fileid_utf8: A utf8 file_id to look up.
1968
:param path_utf8: An utf8 path to be looked up.
1969
:param include_deleted: If True, and performing a lookup via
1970
fileid_utf8 rather than path_utf8, return an entry for deleted
1972
:return: The dirstate entry tuple for path, or (None, None)
1974
self._read_dirblocks_if_needed()
1975
if path_utf8 is not None:
1976
if type(path_utf8) is not str:
1977
raise errors.BzrError('path_utf8 is not a str: %s %r'
1978
% (type(path_utf8), path_utf8))
1979
# path lookups are faster
1980
dirname, basename = osutils.split(path_utf8)
1981
block_index, entry_index, dir_present, file_present = \
1982
self._get_block_entry_index(dirname, basename, tree_index)
1983
if not file_present:
1985
entry = self._dirblocks[block_index][1][entry_index]
1986
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
1987
raise AssertionError('unversioned entry?')
1989
if entry[0][2] != fileid_utf8:
1990
self._changes_aborted = True
1991
raise errors.BzrError('integrity error ? : mismatching'
1992
' tree_index, file_id and path')
1995
possible_keys = self._get_id_index().get(fileid_utf8, ())
1996
if not possible_keys:
1998
for key in possible_keys:
1999
block_index, present = \
2000
self._find_block_index_from_key(key)
2001
# strange, probably indicates an out of date
2002
# id index - for now, allow this.
2005
# WARNING: DO not change this code to use _get_block_entry_index
2006
# as that function is not suitable: it does not use the key
2007
# to lookup, and thus the wrong coordinates are returned.
2008
block = self._dirblocks[block_index][1]
2009
entry_index, present = self._find_entry_index(key, block)
2011
entry = self._dirblocks[block_index][1][entry_index]
2012
# TODO: We might want to assert that entry[0][2] ==
2014
if entry[1][tree_index][0] in 'fdlt':
2015
# this is the result we are looking for: the
2016
# real home of this file_id in this tree.
2018
if entry[1][tree_index][0] == 'a':
2019
# there is no home for this entry in this tree
2023
if entry[1][tree_index][0] != 'r':
2024
raise AssertionError(
2025
"entry %r has invalid minikind %r for tree %r" \
2027
entry[1][tree_index][0],
2029
real_path = entry[1][tree_index][1]
2030
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
2031
path_utf8=real_path)
2035
def initialize(cls, path, sha1_provider=None):
2036
"""Create a new dirstate on path.
2038
The new dirstate will be an empty tree - that is it has no parents,
2039
and only a root node - which has id ROOT_ID.
2041
:param path: The name of the file for the dirstate.
2042
:param sha1_provider: an object meeting the SHA1Provider interface.
2043
If None, a DefaultSHA1Provider is used.
2044
:return: A write-locked DirState object.
2046
# This constructs a new DirState object on a path, sets the _state_file
2047
# to a new empty file for that path. It then calls _set_data() with our
2048
# stock empty dirstate information - a root with ROOT_ID, no children,
2049
# and no parents. Finally it calls save() to ensure that this data will
2051
if sha1_provider is None:
2052
sha1_provider = DefaultSHA1Provider()
2053
result = cls(path, sha1_provider)
2054
# root dir and root dir contents with no children.
2055
empty_tree_dirblocks = [('', []), ('', [])]
2056
# a new root directory, with a NULLSTAT.
2057
empty_tree_dirblocks[0][1].append(
2058
(('', '', inventory.ROOT_ID), [
2059
('d', '', 0, False, DirState.NULLSTAT),
2063
result._set_data([], empty_tree_dirblocks)
2071
def _inv_entry_to_details(inv_entry):
2072
"""Convert an inventory entry (from a revision tree) to state details.
2074
:param inv_entry: An inventory entry whose sha1 and link targets can be
2075
relied upon, and which has a revision set.
2076
:return: A details tuple - the details for a single tree at a path +
2079
kind = inv_entry.kind
2080
minikind = DirState._kind_to_minikind[kind]
2081
tree_data = inv_entry.revision
2082
if kind == 'directory':
2086
elif kind == 'symlink':
2087
if inv_entry.symlink_target is None:
2090
fingerprint = inv_entry.symlink_target.encode('utf8')
2093
elif kind == 'file':
2094
fingerprint = inv_entry.text_sha1 or ''
2095
size = inv_entry.text_size or 0
2096
executable = inv_entry.executable
2097
elif kind == 'tree-reference':
2098
fingerprint = inv_entry.reference_revision or ''
2102
raise Exception("can't pack %s" % inv_entry)
2103
return (minikind, fingerprint, size, executable, tree_data)
2105
def _iter_child_entries(self, tree_index, path_utf8):
2106
"""Iterate over all the entries that are children of path_utf.
2108
This only returns entries that are present (not in 'a', 'r') in
2109
tree_index. tree_index data is not refreshed, so if tree 0 is used,
2110
results may differ from that obtained if paths were statted to
2111
determine what ones were directories.
2113
Asking for the children of a non-directory will return an empty
2117
next_pending_dirs = [path_utf8]
2119
while next_pending_dirs:
2120
pending_dirs = next_pending_dirs
2121
next_pending_dirs = []
2122
for path in pending_dirs:
2123
block_index, present = self._find_block_index_from_key(
2125
if block_index == 0:
2127
if len(self._dirblocks) == 1:
2128
# asked for the children of the root with no other
2132
# children of a non-directory asked for.
2134
block = self._dirblocks[block_index]
2135
for entry in block[1]:
2136
kind = entry[1][tree_index][0]
2137
if kind not in absent:
2141
path = entry[0][0] + '/' + entry[0][1]
2144
next_pending_dirs.append(path)
2146
def _iter_entries(self):
2147
"""Iterate over all the entries in the dirstate.
2149
Each yelt item is an entry in the standard format described in the
2150
docstring of bzrlib.dirstate.
2152
self._read_dirblocks_if_needed()
2153
for directory in self._dirblocks:
2154
for entry in directory[1]:
2157
def _get_id_index(self):
2158
"""Get an id index of self._dirblocks.
2160
This maps from file_id => [(directory, name, file_id)] entries where
2161
that file_id appears in one of the trees.
2163
if self._id_index is None:
2165
for key, tree_details in self._iter_entries():
2166
self._add_to_id_index(id_index, key)
2167
self._id_index = id_index
2168
return self._id_index
2170
def _add_to_id_index(self, id_index, entry_key):
2171
"""Add this entry to the _id_index mapping."""
2172
# This code used to use a set for every entry in the id_index. However,
2173
# it is *rare* to have more than one entry. So a set is a large
2174
# overkill. And even when we do, we won't ever have more than the
2175
# number of parent trees. Which is still a small number (rarely >2). As
2176
# such, we use a simple tuple, and do our own uniqueness checks. While
2177
# the 'in' check is O(N) since N is nicely bounded it shouldn't ever
2178
# cause quadratic failure.
2179
# TODO: This should use StaticTuple
2180
file_id = entry_key[2]
2181
entry_key = static_tuple.StaticTuple.from_sequence(entry_key)
2182
if file_id not in id_index:
2183
id_index[file_id] = static_tuple.StaticTuple(entry_key,)
2185
entry_keys = id_index[file_id]
2186
if entry_key not in entry_keys:
2187
id_index[file_id] = entry_keys + (entry_key,)
2189
def _remove_from_id_index(self, id_index, entry_key):
2190
"""Remove this entry from the _id_index mapping.
2192
It is an programming error to call this when the entry_key is not
2195
file_id = entry_key[2]
2196
entry_keys = list(id_index[file_id])
2197
entry_keys.remove(entry_key)
2198
id_index[file_id] = static_tuple.StaticTuple.from_sequence(entry_keys)
2200
def _get_output_lines(self, lines):
2201
"""Format lines for final output.
2203
:param lines: A sequence of lines containing the parents list and the
2206
output_lines = [DirState.HEADER_FORMAT_3]
2207
lines.append('') # a final newline
2208
inventory_text = '\0\n\0'.join(lines)
2209
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
2210
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
2211
num_entries = len(lines)-3
2212
output_lines.append('num_entries: %s\n' % (num_entries,))
2213
output_lines.append(inventory_text)
2216
def _make_deleted_row(self, fileid_utf8, parents):
2217
"""Return a deleted row for fileid_utf8."""
2218
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
2221
def _num_present_parents(self):
2222
"""The number of parent entries in each record row."""
2223
return len(self._parents) - len(self._ghosts)
2226
def on_file(path, sha1_provider=None):
2227
"""Construct a DirState on the file at path "path".
2229
:param path: The path at which the dirstate file on disk should live.
2230
:param sha1_provider: an object meeting the SHA1Provider interface.
2231
If None, a DefaultSHA1Provider is used.
2232
:return: An unlocked DirState object, associated with the given path.
2234
if sha1_provider is None:
2235
sha1_provider = DefaultSHA1Provider()
2236
result = DirState(path, sha1_provider)
2239
def _read_dirblocks_if_needed(self):
2240
"""Read in all the dirblocks from the file if they are not in memory.
2242
This populates self._dirblocks, and sets self._dirblock_state to
2243
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
2246
self._read_header_if_needed()
2247
if self._dirblock_state == DirState.NOT_IN_MEMORY:
2248
_read_dirblocks(self)
2250
def _read_header(self):
2251
"""This reads in the metadata header, and the parent ids.
2253
After reading in, the file should be positioned at the null
2254
just before the start of the first record in the file.
2256
:return: (expected crc checksum, number of entries, parent list)
2258
self._read_prelude()
2259
parent_line = self._state_file.readline()
2260
info = parent_line.split('\0')
2261
num_parents = int(info[0])
2262
self._parents = info[1:-1]
2263
ghost_line = self._state_file.readline()
2264
info = ghost_line.split('\0')
2265
num_ghosts = int(info[1])
2266
self._ghosts = info[2:-1]
2267
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2268
self._end_of_header = self._state_file.tell()
2270
def _read_header_if_needed(self):
2271
"""Read the header of the dirstate file if needed."""
2272
# inline this as it will be called a lot
2273
if not self._lock_token:
2274
raise errors.ObjectNotLocked(self)
2275
if self._header_state == DirState.NOT_IN_MEMORY:
2278
def _read_prelude(self):
2279
"""Read in the prelude header of the dirstate file.
2281
This only reads in the stuff that is not connected to the crc
2282
checksum. The position will be correct to read in the rest of
2283
the file and check the checksum after this point.
2284
The next entry in the file should be the number of parents,
2285
and their ids. Followed by a newline.
2287
header = self._state_file.readline()
2288
if header != DirState.HEADER_FORMAT_3:
2289
raise errors.BzrError(
2290
'invalid header line: %r' % (header,))
2291
crc_line = self._state_file.readline()
2292
if not crc_line.startswith('crc32: '):
2293
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2294
self.crc_expected = int(crc_line[len('crc32: '):-1])
2295
num_entries_line = self._state_file.readline()
2296
if not num_entries_line.startswith('num_entries: '):
2297
raise errors.BzrError('missing num_entries line')
2298
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2300
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
2301
"""Find a sha1 given a stat lookup."""
2302
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
2304
def _get_packed_stat_index(self):
2305
"""Get a packed_stat index of self._dirblocks."""
2306
if self._packed_stat_index is None:
2308
for key, tree_details in self._iter_entries():
2309
if tree_details[0][0] == 'f':
2310
index[tree_details[0][4]] = tree_details[0][1]
2311
self._packed_stat_index = index
2312
return self._packed_stat_index
2315
"""Save any pending changes created during this session.
2317
We reuse the existing file, because that prevents race conditions with
2318
file creation, and use oslocks on it to prevent concurrent modification
2319
and reads - because dirstate's incremental data aggregation is not
2320
compatible with reading a modified file, and replacing a file in use by
2321
another process is impossible on Windows.
2323
A dirstate in read only mode should be smart enough though to validate
2324
that the file has not changed, and otherwise discard its cache and
2325
start over, to allow for fine grained read lock duration, so 'status'
2326
wont block 'commit' - for example.
2328
if self._changes_aborted:
2329
# Should this be a warning? For now, I'm expecting that places that
2330
# mark it inconsistent will warn, making a warning here redundant.
2331
trace.mutter('Not saving DirState because '
2332
'_changes_aborted is set.')
2334
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
2335
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2337
grabbed_write_lock = False
2338
if self._lock_state != 'w':
2339
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2340
# Switch over to the new lock, as the old one may be closed.
2341
# TODO: jam 20070315 We should validate the disk file has
2342
# not changed contents. Since temporary_write_lock may
2343
# not be an atomic operation.
2344
self._lock_token = new_lock
2345
self._state_file = new_lock.f
2346
if not grabbed_write_lock:
2347
# We couldn't grab a write lock, so we switch back to a read one
2350
self._state_file.seek(0)
2351
self._state_file.writelines(self.get_lines())
2352
self._state_file.truncate()
2353
self._state_file.flush()
2354
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2355
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
2357
if grabbed_write_lock:
2358
self._lock_token = self._lock_token.restore_read_lock()
2359
self._state_file = self._lock_token.f
2360
# TODO: jam 20070315 We should validate the disk file has
2361
# not changed contents. Since restore_read_lock may
2362
# not be an atomic operation.
2364
def _set_data(self, parent_ids, dirblocks):
2365
"""Set the full dirstate data in memory.
2367
This is an internal function used to completely replace the objects
2368
in memory state. It puts the dirstate into state 'full-dirty'.
2370
:param parent_ids: A list of parent tree revision ids.
2371
:param dirblocks: A list containing one tuple for each directory in the
2372
tree. Each tuple contains the directory path and a list of entries
2373
found in that directory.
2375
# our memory copy is now authoritative.
2376
self._dirblocks = dirblocks
2377
self._header_state = DirState.IN_MEMORY_MODIFIED
2378
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2379
self._parents = list(parent_ids)
2380
self._id_index = None
2381
self._packed_stat_index = None
2383
def set_path_id(self, path, new_id):
2384
"""Change the id of path to new_id in the current working tree.
2386
:param path: The path inside the tree to set - '' is the root, 'foo'
2387
is the path foo in the root.
2388
:param new_id: The new id to assign to the path. This must be a utf8
2389
file id (not unicode, and not None).
2391
self._read_dirblocks_if_needed()
2393
# TODO: logic not written
2394
raise NotImplementedError(self.set_path_id)
2395
# TODO: check new id is unique
2396
entry = self._get_entry(0, path_utf8=path)
2397
if entry[0][2] == new_id:
2398
# Nothing to change.
2400
# mark the old path absent, and insert a new root path
2401
self._make_absent(entry)
2402
self.update_minimal(('', '', new_id), 'd',
2403
path_utf8='', packed_stat=entry[1][0][4])
2404
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2406
def set_parent_trees(self, trees, ghosts):
2407
"""Set the parent trees for the dirstate.
2409
:param trees: A list of revision_id, tree tuples. tree must be provided
2410
even if the revision_id refers to a ghost: supply an empty tree in
2412
:param ghosts: A list of the revision_ids that are ghosts at the time
2415
# TODO: generate a list of parent indexes to preserve to save
2416
# processing specific parent trees. In the common case one tree will
2417
# be preserved - the left most parent.
2418
# TODO: if the parent tree is a dirstate, we might want to walk them
2419
# all by path in parallel for 'optimal' common-case performance.
2420
# generate new root row.
2421
self._read_dirblocks_if_needed()
2422
# TODO future sketch: Examine the existing parents to generate a change
2423
# map and then walk the new parent trees only, mapping them into the
2424
# dirstate. Walk the dirstate at the same time to remove unreferenced
2427
# sketch: loop over all entries in the dirstate, cherry picking
2428
# entries from the parent trees, if they are not ghost trees.
2429
# after we finish walking the dirstate, all entries not in the dirstate
2430
# are deletes, so we want to append them to the end as per the design
2431
# discussions. So do a set difference on ids with the parents to
2432
# get deletes, and add them to the end.
2433
# During the update process we need to answer the following questions:
2434
# - find other keys containing a fileid in order to create cross-path
2435
# links. We dont't trivially use the inventory from other trees
2436
# because this leads to either double touching, or to accessing
2438
# - find other keys containing a path
2439
# We accumulate each entry via this dictionary, including the root
2442
# we could do parallel iterators, but because file id data may be
2443
# scattered throughout, we dont save on index overhead: we have to look
2444
# at everything anyway. We can probably save cycles by reusing parent
2445
# data and doing an incremental update when adding an additional
2446
# parent, but for now the common cases are adding a new parent (merge),
2447
# and replacing completely (commit), and commit is more common: so
2448
# optimise merge later.
2450
# ---- start generation of full tree mapping data
2451
# what trees should we use?
2452
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2453
# how many trees do we end up with
2454
parent_count = len(parent_trees)
2456
# one: the current tree
2457
for entry in self._iter_entries():
2458
# skip entries not in the current tree
2459
if entry[1][0][0] in 'ar': # absent, relocated
2461
by_path[entry[0]] = [entry[1][0]] + \
2462
[DirState.NULL_PARENT_DETAILS] * parent_count
2463
# TODO: Possibly inline this, since we know it isn't present yet
2464
# id_index[entry[0][2]] = (entry[0],)
2465
self._add_to_id_index(id_index, entry[0])
2467
# now the parent trees:
2468
for tree_index, tree in enumerate(parent_trees):
2469
# the index is off by one, adjust it.
2470
tree_index = tree_index + 1
2471
# when we add new locations for a fileid we need these ranges for
2472
# any fileid in this tree as we set the by_path[id] to:
2473
# already_processed_tree_details + new_details + new_location_suffix
2474
# the suffix is from tree_index+1:parent_count+1.
2475
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2476
# now stitch in all the entries from this tree
2477
for path, entry in tree.inventory.iter_entries_by_dir():
2478
# here we process each trees details for each item in the tree.
2479
# we first update any existing entries for the id at other paths,
2480
# then we either create or update the entry for the id at the
2481
# right path, and finally we add (if needed) a mapping from
2482
# file_id to this path. We do it in this order to allow us to
2483
# avoid checking all known paths for the id when generating a
2484
# new entry at this path: by adding the id->path mapping last,
2485
# all the mappings are valid and have correct relocation
2486
# records where needed.
2487
file_id = entry.file_id
2488
path_utf8 = path.encode('utf8')
2489
dirname, basename = osutils.split(path_utf8)
2490
new_entry_key = (dirname, basename, file_id)
2491
# tree index consistency: All other paths for this id in this tree
2492
# index must point to the correct path.
2493
for entry_key in id_index.get(file_id, ()):
2494
# TODO:PROFILING: It might be faster to just update
2495
# rather than checking if we need to, and then overwrite
2496
# the one we are located at.
2497
if entry_key != new_entry_key:
2498
# this file id is at a different path in one of the
2499
# other trees, so put absent pointers there
2500
# This is the vertical axis in the matrix, all pointing
2502
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2503
# by path consistency: Insert into an existing path record (trivial), or
2504
# add a new one with relocation pointers for the other tree indexes.
2505
entry_keys = id_index.get(file_id, ())
2506
if new_entry_key in entry_keys:
2507
# there is already an entry where this data belongs, just insert it.
2508
by_path[new_entry_key][tree_index] = \
2509
self._inv_entry_to_details(entry)
2511
# add relocated entries to the horizontal axis - this row
2512
# mapping from path,id. We need to look up the correct path
2513
# for the indexes from 0 to tree_index -1
2515
for lookup_index in xrange(tree_index):
2516
# boundary case: this is the first occurence of file_id
2517
# so there are no id_indexes, possibly take this out of
2519
if not len(entry_keys):
2520
new_details.append(DirState.NULL_PARENT_DETAILS)
2522
# grab any one entry, use it to find the right path.
2523
# TODO: optimise this to reduce memory use in highly
2524
# fragmented situations by reusing the relocation
2526
a_key = iter(entry_keys).next()
2527
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2528
# its a pointer or missing statement, use it as is.
2529
new_details.append(by_path[a_key][lookup_index])
2531
# we have the right key, make a pointer to it.
2532
real_path = ('/'.join(a_key[0:2])).strip('/')
2533
new_details.append(('r', real_path, 0, False, ''))
2534
new_details.append(self._inv_entry_to_details(entry))
2535
new_details.extend(new_location_suffix)
2536
by_path[new_entry_key] = new_details
2537
self._add_to_id_index(id_index, new_entry_key)
2538
# --- end generation of full tree mappings
2540
# sort and output all the entries
2541
new_entries = self._sort_entries(by_path.items())
2542
self._entries_to_current_state(new_entries)
2543
self._parents = [rev_id for rev_id, tree in trees]
2544
self._ghosts = list(ghosts)
2545
self._header_state = DirState.IN_MEMORY_MODIFIED
2546
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2547
self._id_index = id_index
2549
def _sort_entries(self, entry_list):
2550
"""Given a list of entries, sort them into the right order.
2552
This is done when constructing a new dirstate from trees - normally we
2553
try to keep everything in sorted blocks all the time, but sometimes
2554
it's easier to sort after the fact.
2557
# sort by: directory parts, file name, file id
2558
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2559
return sorted(entry_list, key=_key)
2561
def set_state_from_inventory(self, new_inv):
2562
"""Set new_inv as the current state.
2564
This API is called by tree transform, and will usually occur with
2565
existing parent trees.
2567
:param new_inv: The inventory object to set current state from.
2569
if 'evil' in debug.debug_flags:
2570
trace.mutter_callsite(1,
2571
"set_state_from_inventory called; please mutate the tree instead")
2572
tracing = 'dirstate' in debug.debug_flags
2574
trace.mutter("set_state_from_inventory trace:")
2575
self._read_dirblocks_if_needed()
2577
# Two iterators: current data and new data, both in dirblock order.
2578
# We zip them together, which tells about entries that are new in the
2579
# inventory, or removed in the inventory, or present in both and
2582
# You might think we could just synthesize a new dirstate directly
2583
# since we're processing it in the right order. However, we need to
2584
# also consider there may be any number of parent trees and relocation
2585
# pointers, and we don't want to duplicate that here.
2586
new_iterator = new_inv.iter_entries_by_dir()
2587
# we will be modifying the dirstate, so we need a stable iterator. In
2588
# future we might write one, for now we just clone the state into a
2589
# list using a copy so that we see every original item and don't have
2590
# to adjust the position when items are inserted or deleted in the
2591
# underlying dirstate.
2592
old_iterator = iter(list(self._iter_entries()))
2593
# both must have roots so this is safe:
2594
current_new = new_iterator.next()
2595
current_old = old_iterator.next()
2596
def advance(iterator):
2598
return iterator.next()
2599
except StopIteration:
2601
while current_new or current_old:
2602
# skip entries in old that are not really there
2603
if current_old and current_old[1][0][0] in 'ar':
2604
# relocated or absent
2605
current_old = advance(old_iterator)
2608
# convert new into dirblock style
2609
new_path_utf8 = current_new[0].encode('utf8')
2610
new_dirname, new_basename = osutils.split(new_path_utf8)
2611
new_id = current_new[1].file_id
2612
new_entry_key = (new_dirname, new_basename, new_id)
2613
current_new_minikind = \
2614
DirState._kind_to_minikind[current_new[1].kind]
2615
if current_new_minikind == 't':
2616
fingerprint = current_new[1].reference_revision or ''
2618
# We normally only insert or remove records, or update
2619
# them when it has significantly changed. Then we want to
2620
# erase its fingerprint. Unaffected records should
2621
# normally not be updated at all.
2624
# for safety disable variables
2625
new_path_utf8 = new_dirname = new_basename = new_id = \
2626
new_entry_key = None
2627
# 5 cases, we dont have a value that is strictly greater than everything, so
2628
# we make both end conditions explicit
2630
# old is finished: insert current_new into the state.
2632
trace.mutter("Appending from new '%s'.",
2633
new_path_utf8.decode('utf8'))
2634
self.update_minimal(new_entry_key, current_new_minikind,
2635
executable=current_new[1].executable,
2636
path_utf8=new_path_utf8, fingerprint=fingerprint,
2638
current_new = advance(new_iterator)
2639
elif not current_new:
2642
trace.mutter("Truncating from old '%s/%s'.",
2643
current_old[0][0].decode('utf8'),
2644
current_old[0][1].decode('utf8'))
2645
self._make_absent(current_old)
2646
current_old = advance(old_iterator)
2647
elif new_entry_key == current_old[0]:
2648
# same - common case
2649
# We're looking at the same path and id in both the dirstate
2650
# and inventory, so just need to update the fields in the
2651
# dirstate from the one in the inventory.
2652
# TODO: update the record if anything significant has changed.
2653
# the minimal required trigger is if the execute bit or cached
2655
if (current_old[1][0][3] != current_new[1].executable or
2656
current_old[1][0][0] != current_new_minikind):
2658
trace.mutter("Updating in-place change '%s'.",
2659
new_path_utf8.decode('utf8'))
2660
self.update_minimal(current_old[0], current_new_minikind,
2661
executable=current_new[1].executable,
2662
path_utf8=new_path_utf8, fingerprint=fingerprint,
2664
# both sides are dealt with, move on
2665
current_old = advance(old_iterator)
2666
current_new = advance(new_iterator)
2667
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2668
or (new_dirname == current_old[0][0]
2669
and new_entry_key[1:] < current_old[0][1:])):
2671
# add a entry for this and advance new
2673
trace.mutter("Inserting from new '%s'.",
2674
new_path_utf8.decode('utf8'))
2675
self.update_minimal(new_entry_key, current_new_minikind,
2676
executable=current_new[1].executable,
2677
path_utf8=new_path_utf8, fingerprint=fingerprint,
2679
current_new = advance(new_iterator)
2681
# we've advanced past the place where the old key would be,
2682
# without seeing it in the new list. so it must be gone.
2684
trace.mutter("Deleting from old '%s/%s'.",
2685
current_old[0][0].decode('utf8'),
2686
current_old[0][1].decode('utf8'))
2687
self._make_absent(current_old)
2688
current_old = advance(old_iterator)
2689
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2690
self._id_index = None
2691
self._packed_stat_index = None
2693
trace.mutter("set_state_from_inventory complete.")
2695
def set_state_from_scratch(self, working_inv, parent_trees, parent_ghosts):
2696
"""Wipe the currently stored state and set it to something new.
2698
This is a hard-reset for the data we are working with.
2700
# Technically, we really want a write lock, but until we write, we
2701
# don't really need it.
2702
self._requires_lock()
2703
# root dir and root dir contents with no children. We have to have a
2704
# root for set_state_from_inventory to work correctly.
2705
empty_root = (('', '', inventory.ROOT_ID),
2706
[('d', '', 0, False, DirState.NULLSTAT)])
2707
empty_tree_dirblocks = [('', [empty_root]), ('', [])]
2708
self._set_data([], empty_tree_dirblocks)
2709
self.set_state_from_inventory(working_inv)
2710
self.set_parent_trees(parent_trees, parent_ghosts)
2712
def _make_absent(self, current_old):
2713
"""Mark current_old - an entry - as absent for tree 0.
2715
:return: True if this was the last details entry for the entry key:
2716
that is, if the underlying block has had the entry removed, thus
2717
shrinking in length.
2719
# build up paths that this id will be left at after the change is made,
2720
# so we can update their cross references in tree 0
2721
all_remaining_keys = set()
2722
# Dont check the working tree, because it's going.
2723
for details in current_old[1][1:]:
2724
if details[0] not in 'ar': # absent, relocated
2725
all_remaining_keys.add(current_old[0])
2726
elif details[0] == 'r': # relocated
2727
# record the key for the real path.
2728
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2729
# absent rows are not present at any path.
2730
last_reference = current_old[0] not in all_remaining_keys
2732
# the current row consists entire of the current item (being marked
2733
# absent), and relocated or absent entries for the other trees:
2734
# Remove it, its meaningless.
2735
block = self._find_block(current_old[0])
2736
entry_index, present = self._find_entry_index(current_old[0], block[1])
2738
raise AssertionError('could not find entry for %s' % (current_old,))
2739
block[1].pop(entry_index)
2740
# if we have an id_index in use, remove this key from it for this id.
2741
if self._id_index is not None:
2742
self._remove_from_id_index(self._id_index, current_old[0])
2743
# update all remaining keys for this id to record it as absent. The
2744
# existing details may either be the record we are marking as deleted
2745
# (if there were other trees with the id present at this path), or may
2747
for update_key in all_remaining_keys:
2748
update_block_index, present = \
2749
self._find_block_index_from_key(update_key)
2751
raise AssertionError('could not find block for %s' % (update_key,))
2752
update_entry_index, present = \
2753
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2755
raise AssertionError('could not find entry for %s' % (update_key,))
2756
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2757
# it must not be absent at the moment
2758
if update_tree_details[0][0] == 'a': # absent
2759
raise AssertionError('bad row %r' % (update_tree_details,))
2760
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2761
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2762
return last_reference
2764
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2765
packed_stat=None, size=0, path_utf8=None, fullscan=False):
2766
"""Update an entry to the state in tree 0.
2768
This will either create a new entry at 'key' or update an existing one.
2769
It also makes sure that any other records which might mention this are
2772
:param key: (dir, name, file_id) for the new entry
2773
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2775
:param executable: Should the executable bit be set?
2776
:param fingerprint: Simple fingerprint for new entry: canonical-form
2777
sha1 for files, referenced revision id for subtrees, etc.
2778
:param packed_stat: Packed stat value for new entry.
2779
:param size: Size information for new entry
2780
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2782
:param fullscan: If True then a complete scan of the dirstate is being
2783
done and checking for duplicate rows should not be done. This
2784
should only be set by set_state_from_inventory and similar methods.
2786
If packed_stat and fingerprint are not given, they're invalidated in
2789
block = self._find_block(key)[1]
2790
if packed_stat is None:
2791
packed_stat = DirState.NULLSTAT
2792
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2793
# sometimes present in the dirstate - this seems oddly inconsistent.
2795
entry_index, present = self._find_entry_index(key, block)
2796
new_details = (minikind, fingerprint, size, executable, packed_stat)
2797
id_index = self._get_id_index()
2799
# New record. Check there isn't a entry at this path already.
2801
low_index, _ = self._find_entry_index(key[0:2] + ('',), block)
2802
while low_index < len(block):
2803
entry = block[low_index]
2804
if entry[0][0:2] == key[0:2]:
2805
if entry[1][0][0] not in 'ar':
2806
# This entry has the same path (but a different id) as
2807
# the new entry we're adding, and is present in ths
2809
raise errors.InconsistentDelta(
2810
("%s/%s" % key[0:2]).decode('utf8'), key[2],
2811
"Attempt to add item at path already occupied by "
2812
"id %r" % entry[0][2])
2816
# new entry, synthesis cross reference here,
2817
existing_keys = id_index.get(key[2], ())
2818
if not existing_keys:
2819
# not currently in the state, simplest case
2820
new_entry = key, [new_details] + self._empty_parent_info()
2822
# present at one or more existing other paths.
2823
# grab one of them and use it to generate parent
2824
# relocation/absent entries.
2825
new_entry = key, [new_details]
2826
# existing_keys can be changed as we iterate.
2827
for other_key in tuple(existing_keys):
2828
# change the record at other to be a pointer to this new
2829
# record. The loop looks similar to the change to
2830
# relocations when updating an existing record but its not:
2831
# the test for existing kinds is different: this can be
2832
# factored out to a helper though.
2833
other_block_index, present = self._find_block_index_from_key(
2836
raise AssertionError('could not find block for %s' % (
2838
other_block = self._dirblocks[other_block_index][1]
2839
other_entry_index, present = self._find_entry_index(
2840
other_key, other_block)
2842
raise AssertionError(
2843
'update_minimal: could not find other entry for %s'
2845
if path_utf8 is None:
2846
raise AssertionError('no path')
2847
# Turn this other location into a reference to the new
2848
# location. This also updates the aliased iterator
2849
# (current_old in set_state_from_inventory) so that the old
2850
# entry, if not already examined, is skipped over by that
2852
other_entry = other_block[other_entry_index]
2853
other_entry[1][0] = ('r', path_utf8, 0, False, '')
2854
if self._maybe_remove_row(other_block, other_entry_index,
2856
# If the row holding this was removed, we need to
2857
# recompute where this entry goes
2858
entry_index, _ = self._find_entry_index(key, block)
2861
# adds a tuple to the new details for each column
2862
# - either by copying an existing relocation pointer inside that column
2863
# - or by creating a new pointer to the right row inside that column
2864
num_present_parents = self._num_present_parents()
2865
if num_present_parents:
2866
# TODO: This re-evaluates the existing_keys set, do we need
2867
# to do that ourselves?
2868
other_key = list(existing_keys)[0]
2869
for lookup_index in xrange(1, num_present_parents + 1):
2870
# grab any one entry, use it to find the right path.
2871
# TODO: optimise this to reduce memory use in highly
2872
# fragmented situations by reusing the relocation
2874
update_block_index, present = \
2875
self._find_block_index_from_key(other_key)
2877
raise AssertionError('could not find block for %s' % (other_key,))
2878
update_entry_index, present = \
2879
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2881
raise AssertionError('update_minimal: could not find entry for %s' % (other_key,))
2882
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2883
if update_details[0] in 'ar': # relocated, absent
2884
# its a pointer or absent in lookup_index's tree, use
2886
new_entry[1].append(update_details)
2888
# we have the right key, make a pointer to it.
2889
pointer_path = osutils.pathjoin(*other_key[0:2])
2890
new_entry[1].append(('r', pointer_path, 0, False, ''))
2891
block.insert(entry_index, new_entry)
2892
self._add_to_id_index(id_index, key)
2894
# Does the new state matter?
2895
block[entry_index][1][0] = new_details
2896
# parents cannot be affected by what we do.
2897
# other occurences of this id can be found
2898
# from the id index.
2900
# tree index consistency: All other paths for this id in this tree
2901
# index must point to the correct path. We have to loop here because
2902
# we may have passed entries in the state with this file id already
2903
# that were absent - where parent entries are - and they need to be
2904
# converted to relocated.
2905
if path_utf8 is None:
2906
raise AssertionError('no path')
2907
existing_keys = id_index.get(key[2], ())
2908
if key not in existing_keys:
2909
raise AssertionError('We found the entry in the blocks, but'
2910
' the key is not in the id_index.'
2911
' key: %s, existing_keys: %s' % (key, existing_keys))
2912
for entry_key in existing_keys:
2913
# TODO:PROFILING: It might be faster to just update
2914
# rather than checking if we need to, and then overwrite
2915
# the one we are located at.
2916
if entry_key != key:
2917
# this file id is at a different path in one of the
2918
# other trees, so put absent pointers there
2919
# This is the vertical axis in the matrix, all pointing
2921
block_index, present = self._find_block_index_from_key(entry_key)
2923
raise AssertionError('not present: %r', entry_key)
2924
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2926
raise AssertionError('not present: %r', entry_key)
2927
self._dirblocks[block_index][1][entry_index][1][0] = \
2928
('r', path_utf8, 0, False, '')
2929
# add a containing dirblock if needed.
2930
if new_details[0] == 'd':
2931
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2932
block_index, present = self._find_block_index_from_key(subdir_key)
2934
self._dirblocks.insert(block_index, (subdir_key[0], []))
2936
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2938
def _maybe_remove_row(self, block, index, id_index):
2939
"""Remove index if it is absent or relocated across the row.
2941
id_index is updated accordingly.
2942
:return: True if we removed the row, False otherwise
2944
present_in_row = False
2945
entry = block[index]
2946
for column in entry[1]:
2947
if column[0] not in 'ar':
2948
present_in_row = True
2950
if not present_in_row:
2952
self._remove_from_id_index(id_index, entry[0])
2956
def _validate(self):
2957
"""Check that invariants on the dirblock are correct.
2959
This can be useful in debugging; it shouldn't be necessary in
2962
This must be called with a lock held.
2964
# NOTE: This must always raise AssertionError not just assert,
2965
# otherwise it may not behave properly under python -O
2967
# TODO: All entries must have some content that's not 'a' or 'r',
2968
# otherwise it could just be removed.
2970
# TODO: All relocations must point directly to a real entry.
2972
# TODO: No repeated keys.
2975
from pprint import pformat
2976
self._read_dirblocks_if_needed()
2977
if len(self._dirblocks) > 0:
2978
if not self._dirblocks[0][0] == '':
2979
raise AssertionError(
2980
"dirblocks don't start with root block:\n" + \
2981
pformat(self._dirblocks))
2982
if len(self._dirblocks) > 1:
2983
if not self._dirblocks[1][0] == '':
2984
raise AssertionError(
2985
"dirblocks missing root directory:\n" + \
2986
pformat(self._dirblocks))
2987
# the dirblocks are sorted by their path components, name, and dir id
2988
dir_names = [d[0].split('/')
2989
for d in self._dirblocks[1:]]
2990
if dir_names != sorted(dir_names):
2991
raise AssertionError(
2992
"dir names are not in sorted order:\n" + \
2993
pformat(self._dirblocks) + \
2996
for dirblock in self._dirblocks:
2997
# within each dirblock, the entries are sorted by filename and
2999
for entry in dirblock[1]:
3000
if dirblock[0] != entry[0][0]:
3001
raise AssertionError(
3003
"doesn't match directory name in\n%r" %
3004
(entry, pformat(dirblock)))
3005
if dirblock[1] != sorted(dirblock[1]):
3006
raise AssertionError(
3007
"dirblock for %r is not sorted:\n%s" % \
3008
(dirblock[0], pformat(dirblock)))
3010
def check_valid_parent():
3011
"""Check that the current entry has a valid parent.
3013
This makes sure that the parent has a record,
3014
and that the parent isn't marked as "absent" in the
3015
current tree. (It is invalid to have a non-absent file in an absent
3018
if entry[0][0:2] == ('', ''):
3019
# There should be no parent for the root row
3021
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
3022
if parent_entry == (None, None):
3023
raise AssertionError(
3024
"no parent entry for: %s in tree %s"
3025
% (this_path, tree_index))
3026
if parent_entry[1][tree_index][0] != 'd':
3027
raise AssertionError(
3028
"Parent entry for %s is not marked as a valid"
3029
" directory. %s" % (this_path, parent_entry,))
3031
# For each file id, for each tree: either
3032
# the file id is not present at all; all rows with that id in the
3033
# key have it marked as 'absent'
3034
# OR the file id is present under exactly one name; any other entries
3035
# that mention that id point to the correct name.
3037
# We check this with a dict per tree pointing either to the present
3038
# name, or None if absent.
3039
tree_count = self._num_present_parents() + 1
3040
id_path_maps = [dict() for i in range(tree_count)]
3041
# Make sure that all renamed entries point to the correct location.
3042
for entry in self._iter_entries():
3043
file_id = entry[0][2]
3044
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
3045
if len(entry[1]) != tree_count:
3046
raise AssertionError(
3047
"wrong number of entry details for row\n%s" \
3048
",\nexpected %d" % \
3049
(pformat(entry), tree_count))
3050
absent_positions = 0
3051
for tree_index, tree_state in enumerate(entry[1]):
3052
this_tree_map = id_path_maps[tree_index]
3053
minikind = tree_state[0]
3054
if minikind in 'ar':
3055
absent_positions += 1
3056
# have we seen this id before in this column?
3057
if file_id in this_tree_map:
3058
previous_path, previous_loc = this_tree_map[file_id]
3059
# any later mention of this file must be consistent with
3060
# what was said before
3062
if previous_path is not None:
3063
raise AssertionError(
3064
"file %s is absent in row %r but also present " \
3066
(file_id, entry, previous_path))
3067
elif minikind == 'r':
3068
target_location = tree_state[1]
3069
if previous_path != target_location:
3070
raise AssertionError(
3071
"file %s relocation in row %r but also at %r" \
3072
% (file_id, entry, previous_path))
3074
# a file, directory, etc - may have been previously
3075
# pointed to by a relocation, which must point here
3076
if previous_path != this_path:
3077
raise AssertionError(
3078
"entry %r inconsistent with previous path %r "
3080
(entry, previous_path, previous_loc))
3081
check_valid_parent()
3084
# absent; should not occur anywhere else
3085
this_tree_map[file_id] = None, this_path
3086
elif minikind == 'r':
3087
# relocation, must occur at expected location
3088
this_tree_map[file_id] = tree_state[1], this_path
3090
this_tree_map[file_id] = this_path, this_path
3091
check_valid_parent()
3092
if absent_positions == tree_count:
3093
raise AssertionError(
3094
"entry %r has no data for any tree." % (entry,))
3095
if self._id_index is not None:
3096
for file_id, entry_keys in self._id_index.iteritems():
3097
for entry_key in entry_keys:
3098
if entry_key[2] != file_id:
3099
raise AssertionError(
3100
'file_id %r did not match entry key %s'
3101
% (file_id, entry_key))
3102
if len(entry_keys) != len(set(entry_keys)):
3103
raise AssertionError(
3104
'id_index contained non-unique data for %s'
3107
def _wipe_state(self):
3108
"""Forget all state information about the dirstate."""
3109
self._header_state = DirState.NOT_IN_MEMORY
3110
self._dirblock_state = DirState.NOT_IN_MEMORY
3111
self._changes_aborted = False
3114
self._dirblocks = []
3115
self._id_index = None
3116
self._packed_stat_index = None
3117
self._end_of_header = None
3118
self._cutoff_time = None
3119
self._split_path_cache = {}
3121
def lock_read(self):
3122
"""Acquire a read lock on the dirstate."""
3123
if self._lock_token is not None:
3124
raise errors.LockContention(self._lock_token)
3125
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3126
# already in memory, we could read just the header and check for
3127
# any modification. If not modified, we can just leave things
3129
self._lock_token = lock.ReadLock(self._filename)
3130
self._lock_state = 'r'
3131
self._state_file = self._lock_token.f
3134
def lock_write(self):
3135
"""Acquire a write lock on the dirstate."""
3136
if self._lock_token is not None:
3137
raise errors.LockContention(self._lock_token)
3138
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3139
# already in memory, we could read just the header and check for
3140
# any modification. If not modified, we can just leave things
3142
self._lock_token = lock.WriteLock(self._filename)
3143
self._lock_state = 'w'
3144
self._state_file = self._lock_token.f
3148
"""Drop any locks held on the dirstate."""
3149
if self._lock_token is None:
3150
raise errors.LockNotHeld(self)
3151
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3152
# already in memory, we could read just the header and check for
3153
# any modification. If not modified, we can just leave things
3155
self._state_file = None
3156
self._lock_state = None
3157
self._lock_token.unlock()
3158
self._lock_token = None
3159
self._split_path_cache = {}
3161
def _requires_lock(self):
3162
"""Check that a lock is currently held by someone on the dirstate."""
3163
if not self._lock_token:
3164
raise errors.ObjectNotLocked(self)
3167
def py_update_entry(state, entry, abspath, stat_value,
3168
_stat_to_minikind=DirState._stat_to_minikind,
3169
_pack_stat=pack_stat):
3170
"""Update the entry based on what is actually on disk.
3172
This function only calculates the sha if it needs to - if the entry is
3173
uncachable, or clearly different to the first parent's entry, no sha
3174
is calculated, and None is returned.
3176
:param state: The dirstate this entry is in.
3177
:param entry: This is the dirblock entry for the file in question.
3178
:param abspath: The path on disk for this file.
3179
:param stat_value: The stat value done on the path.
3180
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
3181
target of a symlink.
3184
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
3188
packed_stat = _pack_stat(stat_value)
3189
(saved_minikind, saved_link_or_sha1, saved_file_size,
3190
saved_executable, saved_packed_stat) = entry[1][0]
3192
if minikind == 'd' and saved_minikind == 't':
3194
if (minikind == saved_minikind
3195
and packed_stat == saved_packed_stat):
3196
# The stat hasn't changed since we saved, so we can re-use the
3201
# size should also be in packed_stat
3202
if saved_file_size == stat_value.st_size:
3203
return saved_link_or_sha1
3205
# If we have gotten this far, that means that we need to actually
3206
# process this entry.
3210
executable = state._is_executable(stat_value.st_mode,
3212
if state._cutoff_time is None:
3213
state._sha_cutoff_time()
3214
if (stat_value.st_mtime < state._cutoff_time
3215
and stat_value.st_ctime < state._cutoff_time
3216
and len(entry[1]) > 1
3217
and entry[1][1][0] != 'a'):
3218
# Could check for size changes for further optimised
3219
# avoidance of sha1's. However the most prominent case of
3220
# over-shaing is during initial add, which this catches.
3221
# Besides, if content filtering happens, size and sha
3222
# are calculated at the same time, so checking just the size
3223
# gains nothing w.r.t. performance.
3224
link_or_sha1 = state._sha1_file(abspath)
3225
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
3226
executable, packed_stat)
3228
entry[1][0] = ('f', '', stat_value.st_size,
3229
executable, DirState.NULLSTAT)
3230
worth_saving = False
3231
elif minikind == 'd':
3233
entry[1][0] = ('d', '', 0, False, packed_stat)
3234
if saved_minikind != 'd':
3235
# This changed from something into a directory. Make sure we
3236
# have a directory block for it. This doesn't happen very
3237
# often, so this doesn't have to be super fast.
3238
block_index, entry_index, dir_present, file_present = \
3239
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
3240
state._ensure_block(block_index, entry_index,
3241
osutils.pathjoin(entry[0][0], entry[0][1]))
3243
worth_saving = False
3244
elif minikind == 'l':
3245
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
3246
if state._cutoff_time is None:
3247
state._sha_cutoff_time()
3248
if (stat_value.st_mtime < state._cutoff_time
3249
and stat_value.st_ctime < state._cutoff_time):
3250
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
3253
entry[1][0] = ('l', '', stat_value.st_size,
3254
False, DirState.NULLSTAT)
3256
state._dirblock_state = DirState.IN_MEMORY_MODIFIED
3260
class ProcessEntryPython(object):
3262
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id",
3263
"last_source_parent", "last_target_parent", "include_unchanged",
3264
"partial", "use_filesystem_for_exec", "utf8_decode",
3265
"searched_specific_files", "search_specific_files",
3266
"searched_exact_paths", "search_specific_file_parents", "seen_ids",
3267
"state", "source_index", "target_index", "want_unversioned", "tree"]
3269
def __init__(self, include_unchanged, use_filesystem_for_exec,
3270
search_specific_files, state, source_index, target_index,
3271
want_unversioned, tree):
3272
self.old_dirname_to_file_id = {}
3273
self.new_dirname_to_file_id = {}
3274
# Are we doing a partial iter_changes?
3275
self.partial = search_specific_files != set([''])
3276
# Using a list so that we can access the values and change them in
3277
# nested scope. Each one is [path, file_id, entry]
3278
self.last_source_parent = [None, None]
3279
self.last_target_parent = [None, None]
3280
self.include_unchanged = include_unchanged
3281
self.use_filesystem_for_exec = use_filesystem_for_exec
3282
self.utf8_decode = cache_utf8._utf8_decode
3283
# for all search_indexs in each path at or under each element of
3284
# search_specific_files, if the detail is relocated: add the id, and
3285
# add the relocated path as one to search if its not searched already.
3286
# If the detail is not relocated, add the id.
3287
self.searched_specific_files = set()
3288
# When we search exact paths without expanding downwards, we record
3290
self.searched_exact_paths = set()
3291
self.search_specific_files = search_specific_files
3292
# The parents up to the root of the paths we are searching.
3293
# After all normal paths are returned, these specific items are returned.
3294
self.search_specific_file_parents = set()
3295
# The ids we've sent out in the delta.
3296
self.seen_ids = set()
3298
self.source_index = source_index
3299
self.target_index = target_index
3300
if target_index != 0:
3301
# A lot of code in here depends on target_index == 0
3302
raise errors.BzrError('unsupported target index')
3303
self.want_unversioned = want_unversioned
3306
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
3307
"""Compare an entry and real disk to generate delta information.
3309
:param path_info: top_relpath, basename, kind, lstat, abspath for
3310
the path of entry. If None, then the path is considered absent in
3311
the target (Perhaps we should pass in a concrete entry for this ?)
3312
Basename is returned as a utf8 string because we expect this
3313
tuple will be ignored, and don't want to take the time to
3315
:return: (iter_changes_result, changed). If the entry has not been
3316
handled then changed is None. Otherwise it is False if no content
3317
or metadata changes have occurred, and True if any content or
3318
metadata change has occurred. If self.include_unchanged is True then
3319
if changed is not None, iter_changes_result will always be a result
3320
tuple. Otherwise, iter_changes_result is None unless changed is
3323
if self.source_index is None:
3324
source_details = DirState.NULL_PARENT_DETAILS
3326
source_details = entry[1][self.source_index]
3327
target_details = entry[1][self.target_index]
3328
target_minikind = target_details[0]
3329
if path_info is not None and target_minikind in 'fdlt':
3330
if not (self.target_index == 0):
3331
raise AssertionError()
3332
link_or_sha1 = update_entry(self.state, entry,
3333
abspath=path_info[4], stat_value=path_info[3])
3334
# The entry may have been modified by update_entry
3335
target_details = entry[1][self.target_index]
3336
target_minikind = target_details[0]
3339
file_id = entry[0][2]
3340
source_minikind = source_details[0]
3341
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
3342
# claimed content in both: diff
3343
# r | fdlt | | add source to search, add id path move and perform
3344
# | | | diff check on source-target
3345
# r | fdlt | a | dangling file that was present in the basis.
3347
if source_minikind in 'r':
3348
# add the source to the search path to find any children it
3349
# has. TODO ? : only add if it is a container ?
3350
if not osutils.is_inside_any(self.searched_specific_files,
3352
self.search_specific_files.add(source_details[1])
3353
# generate the old path; this is needed for stating later
3355
old_path = source_details[1]
3356
old_dirname, old_basename = os.path.split(old_path)
3357
path = pathjoin(entry[0][0], entry[0][1])
3358
old_entry = self.state._get_entry(self.source_index,
3360
# update the source details variable to be the real
3362
if old_entry == (None, None):
3363
raise errors.CorruptDirstate(self.state._filename,
3364
"entry '%s/%s' is considered renamed from %r"
3365
" but source does not exist\n"
3366
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
3367
source_details = old_entry[1][self.source_index]
3368
source_minikind = source_details[0]
3370
old_dirname = entry[0][0]
3371
old_basename = entry[0][1]
3372
old_path = path = None
3373
if path_info is None:
3374
# the file is missing on disk, show as removed.
3375
content_change = True
3379
# source and target are both versioned and disk file is present.
3380
target_kind = path_info[2]
3381
if target_kind == 'directory':
3383
old_path = path = pathjoin(old_dirname, old_basename)
3384
self.new_dirname_to_file_id[path] = file_id
3385
if source_minikind != 'd':
3386
content_change = True
3388
# directories have no fingerprint
3389
content_change = False
3391
elif target_kind == 'file':
3392
if source_minikind != 'f':
3393
content_change = True
3395
# Check the sha. We can't just rely on the size as
3396
# content filtering may mean differ sizes actually
3397
# map to the same content
3398
if link_or_sha1 is None:
3400
statvalue, link_or_sha1 = \
3401
self.state._sha1_provider.stat_and_sha1(
3403
self.state._observed_sha1(entry, link_or_sha1,
3405
content_change = (link_or_sha1 != source_details[1])
3406
# Target details is updated at update_entry time
3407
if self.use_filesystem_for_exec:
3408
# We don't need S_ISREG here, because we are sure
3409
# we are dealing with a file.
3410
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
3412
target_exec = target_details[3]
3413
elif target_kind == 'symlink':
3414
if source_minikind != 'l':
3415
content_change = True
3417
content_change = (link_or_sha1 != source_details[1])
3419
elif target_kind == 'tree-reference':
3420
if source_minikind != 't':
3421
content_change = True
3423
content_change = False
3427
path = pathjoin(old_dirname, old_basename)
3428
raise errors.BadFileKindError(path, path_info[2])
3429
if source_minikind == 'd':
3431
old_path = path = pathjoin(old_dirname, old_basename)
3432
self.old_dirname_to_file_id[old_path] = file_id
3433
# parent id is the entry for the path in the target tree
3434
if old_basename and old_dirname == self.last_source_parent[0]:
3435
source_parent_id = self.last_source_parent[1]
3438
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3440
source_parent_entry = self.state._get_entry(self.source_index,
3441
path_utf8=old_dirname)
3442
source_parent_id = source_parent_entry[0][2]
3443
if source_parent_id == entry[0][2]:
3444
# This is the root, so the parent is None
3445
source_parent_id = None
3447
self.last_source_parent[0] = old_dirname
3448
self.last_source_parent[1] = source_parent_id
3449
new_dirname = entry[0][0]
3450
if entry[0][1] and new_dirname == self.last_target_parent[0]:
3451
target_parent_id = self.last_target_parent[1]
3454
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3456
# TODO: We don't always need to do the lookup, because the
3457
# parent entry will be the same as the source entry.
3458
target_parent_entry = self.state._get_entry(self.target_index,
3459
path_utf8=new_dirname)
3460
if target_parent_entry == (None, None):
3461
raise AssertionError(
3462
"Could not find target parent in wt: %s\nparent of: %s"
3463
% (new_dirname, entry))
3464
target_parent_id = target_parent_entry[0][2]
3465
if target_parent_id == entry[0][2]:
3466
# This is the root, so the parent is None
3467
target_parent_id = None
3469
self.last_target_parent[0] = new_dirname
3470
self.last_target_parent[1] = target_parent_id
3472
source_exec = source_details[3]
3473
changed = (content_change
3474
or source_parent_id != target_parent_id
3475
or old_basename != entry[0][1]
3476
or source_exec != target_exec
3478
if not changed and not self.include_unchanged:
3481
if old_path is None:
3482
old_path = path = pathjoin(old_dirname, old_basename)
3483
old_path_u = self.utf8_decode(old_path)[0]
3486
old_path_u = self.utf8_decode(old_path)[0]
3487
if old_path == path:
3490
path_u = self.utf8_decode(path)[0]
3491
source_kind = DirState._minikind_to_kind[source_minikind]
3492
return (entry[0][2],
3493
(old_path_u, path_u),
3496
(source_parent_id, target_parent_id),
3497
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3498
(source_kind, target_kind),
3499
(source_exec, target_exec)), changed
3500
elif source_minikind in 'a' and target_minikind in 'fdlt':
3501
# looks like a new file
3502
path = pathjoin(entry[0][0], entry[0][1])
3503
# parent id is the entry for the path in the target tree
3504
# TODO: these are the same for an entire directory: cache em.
3505
parent_id = self.state._get_entry(self.target_index,
3506
path_utf8=entry[0][0])[0][2]
3507
if parent_id == entry[0][2]:
3509
if path_info is not None:
3511
if self.use_filesystem_for_exec:
3512
# We need S_ISREG here, because we aren't sure if this
3515
stat.S_ISREG(path_info[3].st_mode)
3516
and stat.S_IEXEC & path_info[3].st_mode)
3518
target_exec = target_details[3]
3519
return (entry[0][2],
3520
(None, self.utf8_decode(path)[0]),
3524
(None, self.utf8_decode(entry[0][1])[0]),
3525
(None, path_info[2]),
3526
(None, target_exec)), True
3528
# Its a missing file, report it as such.
3529
return (entry[0][2],
3530
(None, self.utf8_decode(path)[0]),
3534
(None, self.utf8_decode(entry[0][1])[0]),
3536
(None, False)), True
3537
elif source_minikind in 'fdlt' and target_minikind in 'a':
3538
# unversioned, possibly, or possibly not deleted: we dont care.
3539
# if its still on disk, *and* theres no other entry at this
3540
# path [we dont know this in this routine at the moment -
3541
# perhaps we should change this - then it would be an unknown.
3542
old_path = pathjoin(entry[0][0], entry[0][1])
3543
# parent id is the entry for the path in the target tree
3544
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3545
if parent_id == entry[0][2]:
3547
return (entry[0][2],
3548
(self.utf8_decode(old_path)[0], None),
3552
(self.utf8_decode(entry[0][1])[0], None),
3553
(DirState._minikind_to_kind[source_minikind], None),
3554
(source_details[3], None)), True
3555
elif source_minikind in 'fdlt' and target_minikind in 'r':
3556
# a rename; could be a true rename, or a rename inherited from
3557
# a renamed parent. TODO: handle this efficiently. Its not
3558
# common case to rename dirs though, so a correct but slow
3559
# implementation will do.
3560
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3561
self.search_specific_files.add(target_details[1])
3562
elif source_minikind in 'ra' and target_minikind in 'ra':
3563
# neither of the selected trees contain this file,
3564
# so skip over it. This is not currently directly tested, but
3565
# is indirectly via test_too_much.TestCommands.test_conflicts.
3568
raise AssertionError("don't know how to compare "
3569
"source_minikind=%r, target_minikind=%r"
3570
% (source_minikind, target_minikind))
3571
## import pdb;pdb.set_trace()
3577
def _gather_result_for_consistency(self, result):
3578
"""Check a result we will yield to make sure we are consistent later.
3580
This gathers result's parents into a set to output later.
3582
:param result: A result tuple.
3584
if not self.partial or not result[0]:
3586
self.seen_ids.add(result[0])
3587
new_path = result[1][1]
3589
# Not the root and not a delete: queue up the parents of the path.
3590
self.search_specific_file_parents.update(
3591
osutils.parent_directories(new_path.encode('utf8')))
3592
# Add the root directory which parent_directories does not
3594
self.search_specific_file_parents.add('')
3596
def iter_changes(self):
3597
"""Iterate over the changes."""
3598
utf8_decode = cache_utf8._utf8_decode
3599
_cmp_by_dirs = cmp_by_dirs
3600
_process_entry = self._process_entry
3601
search_specific_files = self.search_specific_files
3602
searched_specific_files = self.searched_specific_files
3603
splitpath = osutils.splitpath
3605
# compare source_index and target_index at or under each element of search_specific_files.
3606
# follow the following comparison table. Note that we only want to do diff operations when
3607
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3611
# Source | Target | disk | action
3612
# r | fdlt | | add source to search, add id path move and perform
3613
# | | | diff check on source-target
3614
# r | fdlt | a | dangling file that was present in the basis.
3616
# r | a | | add source to search
3618
# r | r | | this path is present in a non-examined tree, skip.
3619
# r | r | a | this path is present in a non-examined tree, skip.
3620
# a | fdlt | | add new id
3621
# a | fdlt | a | dangling locally added file, skip
3622
# a | a | | not present in either tree, skip
3623
# a | a | a | not present in any tree, skip
3624
# a | r | | not present in either tree at this path, skip as it
3625
# | | | may not be selected by the users list of paths.
3626
# a | r | a | not present in either tree at this path, skip as it
3627
# | | | may not be selected by the users list of paths.
3628
# fdlt | fdlt | | content in both: diff them
3629
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3630
# fdlt | a | | unversioned: output deleted id for now
3631
# fdlt | a | a | unversioned and deleted: output deleted id
3632
# fdlt | r | | relocated in this tree, so add target to search.
3633
# | | | Dont diff, we will see an r,fd; pair when we reach
3634
# | | | this id at the other path.
3635
# fdlt | r | a | relocated in this tree, so add target to search.
3636
# | | | Dont diff, we will see an r,fd; pair when we reach
3637
# | | | this id at the other path.
3639
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3640
# keeping a cache of directories that we have seen.
3642
while search_specific_files:
3643
# TODO: the pending list should be lexically sorted? the
3644
# interface doesn't require it.
3645
current_root = search_specific_files.pop()
3646
current_root_unicode = current_root.decode('utf8')
3647
searched_specific_files.add(current_root)
3648
# process the entries for this containing directory: the rest will be
3649
# found by their parents recursively.
3650
root_entries = self.state._entries_for_path(current_root)
3651
root_abspath = self.tree.abspath(current_root_unicode)
3653
root_stat = os.lstat(root_abspath)
3655
if e.errno == errno.ENOENT:
3656
# the path does not exist: let _process_entry know that.
3657
root_dir_info = None
3659
# some other random error: hand it up.
3662
root_dir_info = ('', current_root,
3663
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3665
if root_dir_info[2] == 'directory':
3666
if self.tree._directory_is_tree_reference(
3667
current_root.decode('utf8')):
3668
root_dir_info = root_dir_info[:2] + \
3669
('tree-reference',) + root_dir_info[3:]
3671
if not root_entries and not root_dir_info:
3672
# this specified path is not present at all, skip it.
3674
path_handled = False
3675
for entry in root_entries:
3676
result, changed = _process_entry(entry, root_dir_info)
3677
if changed is not None:
3680
self._gather_result_for_consistency(result)
3681
if changed or self.include_unchanged:
3683
if self.want_unversioned and not path_handled and root_dir_info:
3684
new_executable = bool(
3685
stat.S_ISREG(root_dir_info[3].st_mode)
3686
and stat.S_IEXEC & root_dir_info[3].st_mode)
3688
(None, current_root_unicode),
3692
(None, splitpath(current_root_unicode)[-1]),
3693
(None, root_dir_info[2]),
3694
(None, new_executable)
3696
initial_key = (current_root, '', '')
3697
block_index, _ = self.state._find_block_index_from_key(initial_key)
3698
if block_index == 0:
3699
# we have processed the total root already, but because the
3700
# initial key matched it we should skip it here.
3702
if root_dir_info and root_dir_info[2] == 'tree-reference':
3703
current_dir_info = None
3705
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3707
current_dir_info = dir_iterator.next()
3709
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3710
# python 2.5 has e.errno == EINVAL,
3711
# and e.winerror == ERROR_DIRECTORY
3712
e_winerror = getattr(e, 'winerror', None)
3713
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3714
# there may be directories in the inventory even though
3715
# this path is not a file on disk: so mark it as end of
3717
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3718
current_dir_info = None
3719
elif (sys.platform == 'win32'
3720
and (e.errno in win_errors
3721
or e_winerror in win_errors)):
3722
current_dir_info = None
3726
if current_dir_info[0][0] == '':
3727
# remove .bzr from iteration
3728
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3729
if current_dir_info[1][bzr_index][0] != '.bzr':
3730
raise AssertionError()
3731
del current_dir_info[1][bzr_index]
3732
# walk until both the directory listing and the versioned metadata
3734
if (block_index < len(self.state._dirblocks) and
3735
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3736
current_block = self.state._dirblocks[block_index]
3738
current_block = None
3739
while (current_dir_info is not None or
3740
current_block is not None):
3741
if (current_dir_info and current_block
3742
and current_dir_info[0][0] != current_block[0]):
3743
if _cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
3744
# filesystem data refers to paths not covered by the dirblock.
3745
# this has two possibilities:
3746
# A) it is versioned but empty, so there is no block for it
3747
# B) it is not versioned.
3749
# if (A) then we need to recurse into it to check for
3750
# new unknown files or directories.
3751
# if (B) then we should ignore it, because we don't
3752
# recurse into unknown directories.
3754
while path_index < len(current_dir_info[1]):
3755
current_path_info = current_dir_info[1][path_index]
3756
if self.want_unversioned:
3757
if current_path_info[2] == 'directory':
3758
if self.tree._directory_is_tree_reference(
3759
current_path_info[0].decode('utf8')):
3760
current_path_info = current_path_info[:2] + \
3761
('tree-reference',) + current_path_info[3:]
3762
new_executable = bool(
3763
stat.S_ISREG(current_path_info[3].st_mode)
3764
and stat.S_IEXEC & current_path_info[3].st_mode)
3766
(None, utf8_decode(current_path_info[0])[0]),
3770
(None, utf8_decode(current_path_info[1])[0]),
3771
(None, current_path_info[2]),
3772
(None, new_executable))
3773
# dont descend into this unversioned path if it is
3775
if current_path_info[2] in ('directory',
3777
del current_dir_info[1][path_index]
3781
# This dir info has been handled, go to the next
3783
current_dir_info = dir_iterator.next()
3784
except StopIteration:
3785
current_dir_info = None
3787
# We have a dirblock entry for this location, but there
3788
# is no filesystem path for this. This is most likely
3789
# because a directory was removed from the disk.
3790
# We don't have to report the missing directory,
3791
# because that should have already been handled, but we
3792
# need to handle all of the files that are contained
3794
for current_entry in current_block[1]:
3795
# entry referring to file not present on disk.
3796
# advance the entry only, after processing.
3797
result, changed = _process_entry(current_entry, None)
3798
if changed is not None:
3800
self._gather_result_for_consistency(result)
3801
if changed or self.include_unchanged:
3804
if (block_index < len(self.state._dirblocks) and
3805
osutils.is_inside(current_root,
3806
self.state._dirblocks[block_index][0])):
3807
current_block = self.state._dirblocks[block_index]
3809
current_block = None
3812
if current_block and entry_index < len(current_block[1]):
3813
current_entry = current_block[1][entry_index]
3815
current_entry = None
3816
advance_entry = True
3818
if current_dir_info and path_index < len(current_dir_info[1]):
3819
current_path_info = current_dir_info[1][path_index]
3820
if current_path_info[2] == 'directory':
3821
if self.tree._directory_is_tree_reference(
3822
current_path_info[0].decode('utf8')):
3823
current_path_info = current_path_info[:2] + \
3824
('tree-reference',) + current_path_info[3:]
3826
current_path_info = None
3828
path_handled = False
3829
while (current_entry is not None or
3830
current_path_info is not None):
3831
if current_entry is None:
3832
# the check for path_handled when the path is advanced
3833
# will yield this path if needed.
3835
elif current_path_info is None:
3836
# no path is fine: the per entry code will handle it.
3837
result, changed = _process_entry(current_entry, current_path_info)
3838
if changed is not None:
3840
self._gather_result_for_consistency(result)
3841
if changed or self.include_unchanged:
3843
elif (current_entry[0][1] != current_path_info[1]
3844
or current_entry[1][self.target_index][0] in 'ar'):
3845
# The current path on disk doesn't match the dirblock
3846
# record. Either the dirblock is marked as absent, or
3847
# the file on disk is not present at all in the
3848
# dirblock. Either way, report about the dirblock
3849
# entry, and let other code handle the filesystem one.
3851
# Compare the basename for these files to determine
3853
if current_path_info[1] < current_entry[0][1]:
3854
# extra file on disk: pass for now, but only
3855
# increment the path, not the entry
3856
advance_entry = False
3858
# entry referring to file not present on disk.
3859
# advance the entry only, after processing.
3860
result, changed = _process_entry(current_entry, None)
3861
if changed is not None:
3863
self._gather_result_for_consistency(result)
3864
if changed or self.include_unchanged:
3866
advance_path = False
3868
result, changed = _process_entry(current_entry, current_path_info)
3869
if changed is not None:
3872
self._gather_result_for_consistency(result)
3873
if changed or self.include_unchanged:
3875
if advance_entry and current_entry is not None:
3877
if entry_index < len(current_block[1]):
3878
current_entry = current_block[1][entry_index]
3880
current_entry = None
3882
advance_entry = True # reset the advance flaga
3883
if advance_path and current_path_info is not None:
3884
if not path_handled:
3885
# unversioned in all regards
3886
if self.want_unversioned:
3887
new_executable = bool(
3888
stat.S_ISREG(current_path_info[3].st_mode)
3889
and stat.S_IEXEC & current_path_info[3].st_mode)
3891
relpath_unicode = utf8_decode(current_path_info[0])[0]
3892
except UnicodeDecodeError:
3893
raise errors.BadFilenameEncoding(
3894
current_path_info[0], osutils._fs_enc)
3896
(None, relpath_unicode),
3900
(None, utf8_decode(current_path_info[1])[0]),
3901
(None, current_path_info[2]),
3902
(None, new_executable))
3903
# dont descend into this unversioned path if it is
3905
if current_path_info[2] in ('directory'):
3906
del current_dir_info[1][path_index]
3908
# dont descend the disk iterator into any tree
3910
if current_path_info[2] == 'tree-reference':
3911
del current_dir_info[1][path_index]
3914
if path_index < len(current_dir_info[1]):
3915
current_path_info = current_dir_info[1][path_index]
3916
if current_path_info[2] == 'directory':
3917
if self.tree._directory_is_tree_reference(
3918
current_path_info[0].decode('utf8')):
3919
current_path_info = current_path_info[:2] + \
3920
('tree-reference',) + current_path_info[3:]
3922
current_path_info = None
3923
path_handled = False
3925
advance_path = True # reset the advance flagg.
3926
if current_block is not None:
3928
if (block_index < len(self.state._dirblocks) and
3929
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3930
current_block = self.state._dirblocks[block_index]
3932
current_block = None
3933
if current_dir_info is not None:
3935
current_dir_info = dir_iterator.next()
3936
except StopIteration:
3937
current_dir_info = None
3938
for result in self._iter_specific_file_parents():
3941
def _iter_specific_file_parents(self):
3942
"""Iter over the specific file parents."""
3943
while self.search_specific_file_parents:
3944
# Process the parent directories for the paths we were iterating.
3945
# Even in extremely large trees this should be modest, so currently
3946
# no attempt is made to optimise.
3947
path_utf8 = self.search_specific_file_parents.pop()
3948
if osutils.is_inside_any(self.searched_specific_files, path_utf8):
3949
# We've examined this path.
3951
if path_utf8 in self.searched_exact_paths:
3952
# We've examined this path.
3954
path_entries = self.state._entries_for_path(path_utf8)
3955
# We need either one or two entries. If the path in
3956
# self.target_index has moved (so the entry in source_index is in
3957
# 'ar') then we need to also look for the entry for this path in
3958
# self.source_index, to output the appropriate delete-or-rename.
3959
selected_entries = []
3961
for candidate_entry in path_entries:
3962
# Find entries present in target at this path:
3963
if candidate_entry[1][self.target_index][0] not in 'ar':
3965
selected_entries.append(candidate_entry)
3966
# Find entries present in source at this path:
3967
elif (self.source_index is not None and
3968
candidate_entry[1][self.source_index][0] not in 'ar'):
3970
if candidate_entry[1][self.target_index][0] == 'a':
3971
# Deleted, emit it here.
3972
selected_entries.append(candidate_entry)
3974
# renamed, emit it when we process the directory it
3976
self.search_specific_file_parents.add(
3977
candidate_entry[1][self.target_index][1])
3979
raise AssertionError(
3980
"Missing entry for specific path parent %r, %r" % (
3981
path_utf8, path_entries))
3982
path_info = self._path_info(path_utf8, path_utf8.decode('utf8'))
3983
for entry in selected_entries:
3984
if entry[0][2] in self.seen_ids:
3986
result, changed = self._process_entry(entry, path_info)
3988
raise AssertionError(
3989
"Got entry<->path mismatch for specific path "
3990
"%r entry %r path_info %r " % (
3991
path_utf8, entry, path_info))
3992
# Only include changes - we're outside the users requested
3995
self._gather_result_for_consistency(result)
3996
if (result[6][0] == 'directory' and
3997
result[6][1] != 'directory'):
3998
# This stopped being a directory, the old children have
4000
if entry[1][self.source_index][0] == 'r':
4001
# renamed, take the source path
4002
entry_path_utf8 = entry[1][self.source_index][1]
4004
entry_path_utf8 = path_utf8
4005
initial_key = (entry_path_utf8, '', '')
4006
block_index, _ = self.state._find_block_index_from_key(
4008
if block_index == 0:
4009
# The children of the root are in block index 1.
4011
current_block = None
4012
if block_index < len(self.state._dirblocks):
4013
current_block = self.state._dirblocks[block_index]
4014
if not osutils.is_inside(
4015
entry_path_utf8, current_block[0]):
4016
# No entries for this directory at all.
4017
current_block = None
4018
if current_block is not None:
4019
for entry in current_block[1]:
4020
if entry[1][self.source_index][0] in 'ar':
4021
# Not in the source tree, so doesn't have to be
4024
# Path of the entry itself.
4026
self.search_specific_file_parents.add(
4027
osutils.pathjoin(*entry[0][:2]))
4028
if changed or self.include_unchanged:
4030
self.searched_exact_paths.add(path_utf8)
4032
def _path_info(self, utf8_path, unicode_path):
4033
"""Generate path_info for unicode_path.
4035
:return: None if unicode_path does not exist, or a path_info tuple.
4037
abspath = self.tree.abspath(unicode_path)
4039
stat = os.lstat(abspath)
4041
if e.errno == errno.ENOENT:
4042
# the path does not exist.
4046
utf8_basename = utf8_path.rsplit('/', 1)[-1]
4047
dir_info = (utf8_path, utf8_basename,
4048
osutils.file_kind_from_stat_mode(stat.st_mode), stat,
4050
if dir_info[2] == 'directory':
4051
if self.tree._directory_is_tree_reference(
4053
self.root_dir_info = self.root_dir_info[:2] + \
4054
('tree-reference',) + self.root_dir_info[3:]
4058
# Try to load the compiled form if possible
4060
from bzrlib._dirstate_helpers_pyx import (
4066
ProcessEntryC as _process_entry,
4067
update_entry as update_entry,
4069
except ImportError, e:
4070
osutils.failed_to_load_extension(e)
4071
from bzrlib._dirstate_helpers_py import (
4078
# FIXME: It would be nice to be able to track moved lines so that the
4079
# corresponding python code can be moved to the _dirstate_helpers_py
4080
# module. I don't want to break the history for this important piece of
4081
# code so I left the code here -- vila 20090622
4082
update_entry = py_update_entry
4083
_process_entry = ProcessEntryPython
added
removed
bzrlib/dirstate.py
