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.
25
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
28
WHOLE_NUMBER = {digit}, digit;
30
REVISION_ID = a non-empty utf8 string;
32
dirstate format = header line, full checksum, row count, parent details,
33
ghost_details, entries;
34
header line = "#bazaar dirstate flat format 3", NL;
35
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
36
row count = "num_entries: ", WHOLE_NUMBER, NL;
37
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
38
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
40
entry = entry_key, current_entry_details, {parent_entry_details};
41
entry_key = dirname, basename, fileid;
42
current_entry_details = common_entry_details, working_entry_details;
43
parent_entry_details = common_entry_details, history_entry_details;
44
common_entry_details = MINIKIND, fingerprint, size, executable
45
working_entry_details = packed_stat
46
history_entry_details = REVISION_ID;
49
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
51
Given this definition, the following is useful to know::
53
entry (aka row) - all the data for a given key.
54
entry[0]: The key (dirname, basename, fileid)
58
entry[1]: The tree(s) data for this path and id combination.
59
entry[1][0]: The current tree
60
entry[1][1]: The second tree
62
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate::
64
entry[1][0][0]: minikind
65
entry[1][0][1]: fingerprint
67
entry[1][0][3]: executable
68
entry[1][0][4]: packed_stat
72
entry[1][1][4]: revision_id
74
There may be multiple rows at the root, one per id present in the root, so the
75
in memory root row is now::
77
self._dirblocks[0] -> ('', [entry ...]),
79
and the entries in there are::
83
entries[0][2]: file_id
84
entries[1][0]: The tree data for the current tree for this fileid at /
89
'r' is a relocated entry: This path is not present in this tree with this
90
id, but the id can be found at another location. The fingerprint is
91
used to point to the target location.
92
'a' is an absent entry: In that tree the id is not present at this path.
93
'd' is a directory entry: This path in this tree is a directory with the
94
current file id. There is no fingerprint for directories.
95
'f' is a file entry: As for directory, but it's a file. The fingerprint is
96
the sha1 value of the file's canonical form, i.e. after any read
97
filters have been applied to the convenience form stored in the working
99
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is
101
't' is a reference to a nested subtree; the fingerprint is the referenced
106
The entries on disk and in memory are ordered according to the following keys::
108
directory, as a list of components
112
--- Format 1 had the following different definition: ---
116
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
117
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
119
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
120
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
123
PARENT ROW's are emitted for every parent that is not in the ghosts details
124
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
125
each row will have a PARENT ROW for foo and baz, but not for bar.
128
In any tree, a kind of 'moved' indicates that the fingerprint field
129
(which we treat as opaque data specific to the 'kind' anyway) has the
130
details for the id of this row in that tree.
132
I'm strongly tempted to add a id->path index as well, but I think that
133
where we need id->path mapping; we also usually read the whole file, so
134
I'm going to skip that for the moment, as we have the ability to locate
135
via bisect any path in any tree, and if we lookup things by path, we can
136
accumulate an id->path mapping as we go, which will tend to match what we
139
I plan to implement this asap, so please speak up now to alter/tweak the
140
design - and once we stabilise on this, I'll update the wiki page for
143
The rationale for all this is that we want fast operations for the
144
common case (diff/status/commit/merge on all files) and extremely fast
145
operations for the less common but still occurs a lot status/diff/commit
146
on specific files). Operations on specific files involve a scan for all
147
the children of a path, *in every involved tree*, which the current
148
format did not accommodate.
152
1. Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
153
2. fall back current object model as needed.
154
3. scale usably to the largest trees known today - say 50K entries. (mozilla
155
is an example of this)
160
Eventually reuse dirstate objects across locks IFF the dirstate file has not
161
been modified, but will require that we flush/ignore cached stat-hit data
162
because we won't want to restat all files on disk just because a lock was
163
acquired, yet we cannot trust the data after the previous lock was released.
165
Memory representation::
167
vector of all directories, and vector of the childen ?
169
root_entrie = (direntry for root, [parent_direntries_for_root]),
171
('', ['data for achild', 'data for bchild', 'data for cchild'])
172
('dir', ['achild', 'cchild', 'echild'])
174
- single bisect to find N subtrees from a path spec
175
- in-order for serialisation - this is 'dirblock' grouping.
176
- insertion of a file '/a' affects only the '/' child-vector, that is, to
177
insert 10K elements from scratch does not generates O(N^2) memoves of a
178
single vector, rather each individual, which tends to be limited to a
179
manageable number. Will scale badly on trees with 10K entries in a
180
single directory. compare with Inventory.InventoryDirectory which has
181
a dictionary for the children. No bisect capability, can only probe for
182
exact matches, or grab all elements and sort.
183
- What's the risk of error here? Once we have the base format being processed
184
we should have a net win regardless of optimality. So we are going to
185
go with what seems reasonable.
189
Maybe we should do a test profile of the core structure - 10K simulated
190
searches/lookups/etc?
192
Objects for each row?
193
The lifetime of Dirstate objects is current per lock, but see above for
194
possible extensions. The lifetime of a row from a dirstate is expected to be
195
very short in the optimistic case: which we are optimising for. For instance,
196
subtree status will determine from analysis of the disk data what rows need to
197
be examined at all, and will be able to determine from a single row whether
198
that file has altered or not, so we are aiming to process tens of thousands of
199
entries each second within the dirstate context, before exposing anything to
200
the larger codebase. This suggests we want the time for a single file
201
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
202
processed, and to scale to 100 thousand we'll another order of magnitude to do
203
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
204
the file on disk, and then immediately discard, the overhead of object creation
205
becomes a significant cost.
207
Figures: Creating a tuple from 3 elements was profiled at 0.0625
208
microseconds, whereas creating a object which is subclassed from tuple was
209
0.500 microseconds, and creating an object with 3 elements and slots was 3
210
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
211
down to 10 microseconds for the total processing - having 33% of that be object
212
creation is a huge overhead. There is a potential cost in using tuples within
213
each row which is that the conditional code to do comparisons may be slower
214
than method invocation, but method invocation is known to be slow due to stack
215
frame creation, so avoiding methods in these tight inner loops in unfortunately
216
desirable. We can consider a pyrex version of this with objects in future if
221
from __future__ import absolute_import
227
from stat import S_IEXEC
245
from .sixish import (
252
# This is the Windows equivalent of ENOTDIR
253
# It is defined in pywin32.winerror, but we don't want a strong dependency for
254
# just an error code.
255
ERROR_PATH_NOT_FOUND = 3
256
ERROR_DIRECTORY = 267
259
class SHA1Provider(object):
260
"""An interface for getting sha1s of a file."""
262
def sha1(self, abspath):
263
"""Return the sha1 of a file given its absolute path.
265
:param abspath: May be a filesystem encoded absolute path
268
raise NotImplementedError(self.sha1)
270
def stat_and_sha1(self, abspath):
271
"""Return the stat and sha1 of a file given its absolute path.
273
:param abspath: May be a filesystem encoded absolute path
276
Note: the stat should be the stat of the physical file
277
while the sha may be the sha of its canonical content.
279
raise NotImplementedError(self.stat_and_sha1)
282
class DefaultSHA1Provider(SHA1Provider):
283
"""A SHA1Provider that reads directly from the filesystem."""
285
def sha1(self, abspath):
286
"""Return the sha1 of a file given its absolute path."""
287
return osutils.sha_file_by_name(abspath)
289
def stat_and_sha1(self, abspath):
290
"""Return the stat and sha1 of a file given its absolute path."""
291
file_obj = file(abspath, 'rb')
293
statvalue = os.fstat(file_obj.fileno())
294
sha1 = osutils.sha_file(file_obj)
297
return statvalue, sha1
300
class DirState(object):
301
"""Record directory and metadata state for fast access.
303
A dirstate is a specialised data structure for managing local working
304
tree state information. Its not yet well defined whether it is platform
305
specific, and if it is how we detect/parameterize that.
307
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
308
Unlike most bzr disk formats, DirStates must be locked for reading, using
309
lock_read. (This is an os file lock internally.) This is necessary
310
because the file can be rewritten in place.
312
DirStates must be explicitly written with save() to commit changes; just
313
unlocking them does not write the changes to disk.
316
_kind_to_minikind = {
322
'tree-reference': 't',
324
_minikind_to_kind = {
330
't': 'tree-reference',
332
_stat_to_minikind = {
337
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
338
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
341
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
342
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
343
# that we are sure a single record will always fit.
344
BISECT_PAGE_SIZE = 4096
347
IN_MEMORY_UNMODIFIED = 1
348
IN_MEMORY_MODIFIED = 2
349
IN_MEMORY_HASH_MODIFIED = 3 # Only hash-cache updates
351
# A pack_stat (the x's) that is just noise and will never match the output
354
NULL_PARENT_DETAILS = static_tuple.StaticTuple('a', '', 0, False, '')
356
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
357
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
359
def __init__(self, path, sha1_provider, worth_saving_limit=0):
360
"""Create a DirState object.
362
:param path: The path at which the dirstate file on disk should live.
363
:param sha1_provider: an object meeting the SHA1Provider interface.
364
:param worth_saving_limit: when the exact number of hash changed
365
entries is known, only bother saving the dirstate if more than
366
this count of entries have changed.
367
-1 means never save hash changes, 0 means always save hash changes.
369
# _header_state and _dirblock_state represent the current state
370
# of the dirstate metadata and the per-row data respectiely.
371
# NOT_IN_MEMORY indicates that no data is in memory
372
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
373
# is the same as is on disk
374
# IN_MEMORY_MODIFIED indicates that we have a modified version
375
# of what is on disk.
376
# In future we will add more granularity, for instance _dirblock_state
377
# will probably support partially-in-memory as a separate variable,
378
# allowing for partially-in-memory unmodified and partially-in-memory
380
self._header_state = DirState.NOT_IN_MEMORY
381
self._dirblock_state = DirState.NOT_IN_MEMORY
382
# If true, an error has been detected while updating the dirstate, and
383
# for safety we're not going to commit to disk.
384
self._changes_aborted = False
388
self._state_file = None
389
self._filename = path
390
self._lock_token = None
391
self._lock_state = None
392
self._id_index = None
393
# a map from packed_stat to sha's.
394
self._packed_stat_index = None
395
self._end_of_header = None
396
self._cutoff_time = None
397
self._split_path_cache = {}
398
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
399
self._sha1_provider = sha1_provider
400
if 'hashcache' in debug.debug_flags:
401
self._sha1_file = self._sha1_file_and_mutter
403
self._sha1_file = self._sha1_provider.sha1
404
# These two attributes provide a simple cache for lookups into the
405
# dirstate in-memory vectors. By probing respectively for the last
406
# block, and for the next entry, we save nearly 2 bisections per path
408
self._last_block_index = None
409
self._last_entry_index = None
410
# The set of known hash changes
411
self._known_hash_changes = set()
412
# How many hash changed entries can we have without saving
413
self._worth_saving_limit = worth_saving_limit
414
self._config_stack = config.LocationStack(urlutils.local_path_to_url(
419
(self.__class__.__name__, self._filename)
421
def _mark_modified(self, hash_changed_entries=None, header_modified=False):
422
"""Mark this dirstate as modified.
424
:param hash_changed_entries: if non-None, mark just these entries as
425
having their hash modified.
426
:param header_modified: mark the header modified as well, not just the
429
#trace.mutter_callsite(3, "modified hash entries: %s", hash_changed_entries)
430
if hash_changed_entries:
431
self._known_hash_changes.update([e[0] for e in hash_changed_entries])
432
if self._dirblock_state in (DirState.NOT_IN_MEMORY,
433
DirState.IN_MEMORY_UNMODIFIED):
434
# If the dirstate is already marked a IN_MEMORY_MODIFIED, then
435
# that takes precedence.
436
self._dirblock_state = DirState.IN_MEMORY_HASH_MODIFIED
438
# TODO: Since we now have a IN_MEMORY_HASH_MODIFIED state, we
439
# should fail noisily if someone tries to set
440
# IN_MEMORY_MODIFIED but we don't have a write-lock!
441
# We don't know exactly what changed so disable smart saving
442
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
444
self._header_state = DirState.IN_MEMORY_MODIFIED
446
def _mark_unmodified(self):
447
"""Mark this dirstate as unmodified."""
448
self._header_state = DirState.IN_MEMORY_UNMODIFIED
449
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
450
self._known_hash_changes = set()
452
def add(self, path, file_id, kind, stat, fingerprint):
453
"""Add a path to be tracked.
455
:param path: The path within the dirstate - '' is the root, 'foo' is the
456
path foo within the root, 'foo/bar' is the path bar within foo
458
:param file_id: The file id of the path being added.
459
:param kind: The kind of the path, as a string like 'file',
461
:param stat: The output of os.lstat for the path.
462
:param fingerprint: The sha value of the file's canonical form (i.e.
463
after any read filters have been applied),
464
or the target of a symlink,
465
or the referenced revision id for tree-references,
466
or '' for directories.
469
# find the block its in.
470
# find the location in the block.
471
# check its not there
473
#------- copied from inventory.ensure_normalized_name - keep synced.
474
# --- normalized_filename wants a unicode basename only, so get one.
475
dirname, basename = osutils.split(path)
476
# we dont import normalized_filename directly because we want to be
477
# able to change the implementation at runtime for tests.
478
norm_name, can_access = osutils.normalized_filename(basename)
479
if norm_name != basename:
483
raise errors.InvalidNormalization(path)
484
# you should never have files called . or ..; just add the directory
485
# in the parent, or according to the special treatment for the root
486
if basename == '.' or basename == '..':
487
raise errors.InvalidEntryName(path)
488
# now that we've normalised, we need the correct utf8 path and
489
# dirname and basename elements. This single encode and split should be
490
# faster than three separate encodes.
491
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
492
dirname, basename = osutils.split(utf8path)
493
# uses __class__ for speed; the check is needed for safety
494
if file_id.__class__ is not str:
495
raise AssertionError(
496
"must be a utf8 file_id not %s" % (type(file_id), ))
497
# Make sure the file_id does not exist in this tree
499
file_id_entry = self._get_entry(0, fileid_utf8=file_id, include_deleted=True)
500
if file_id_entry != (None, None):
501
if file_id_entry[1][0][0] == 'a':
502
if file_id_entry[0] != (dirname, basename, file_id):
503
# set the old name's current operation to rename
504
self.update_minimal(file_id_entry[0],
510
rename_from = file_id_entry[0][0:2]
512
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
513
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
514
info = '%s:%s' % (kind, path)
515
raise errors.DuplicateFileId(file_id, info)
516
first_key = (dirname, basename, '')
517
block_index, present = self._find_block_index_from_key(first_key)
519
# check the path is not in the tree
520
block = self._dirblocks[block_index][1]
521
entry_index, _ = self._find_entry_index(first_key, block)
522
while (entry_index < len(block) and
523
block[entry_index][0][0:2] == first_key[0:2]):
524
if block[entry_index][1][0][0] not in 'ar':
525
# this path is in the dirstate in the current tree.
526
raise Exception("adding already added path!")
529
# The block where we want to put the file is not present. But it
530
# might be because the directory was empty, or not loaded yet. Look
531
# for a parent entry, if not found, raise NotVersionedError
532
parent_dir, parent_base = osutils.split(dirname)
533
parent_block_idx, parent_entry_idx, _, parent_present = \
534
self._get_block_entry_index(parent_dir, parent_base, 0)
535
if not parent_present:
536
raise errors.NotVersionedError(path, str(self))
537
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
538
block = self._dirblocks[block_index][1]
539
entry_key = (dirname, basename, file_id)
542
packed_stat = DirState.NULLSTAT
545
packed_stat = pack_stat(stat)
546
parent_info = self._empty_parent_info()
547
minikind = DirState._kind_to_minikind[kind]
548
if rename_from is not None:
550
old_path_utf8 = '%s/%s' % rename_from
552
old_path_utf8 = rename_from[1]
553
parent_info[0] = ('r', old_path_utf8, 0, False, '')
555
entry_data = entry_key, [
556
(minikind, fingerprint, size, False, packed_stat),
558
elif kind == 'directory':
559
entry_data = entry_key, [
560
(minikind, '', 0, False, packed_stat),
562
elif kind == 'symlink':
563
entry_data = entry_key, [
564
(minikind, fingerprint, size, False, packed_stat),
566
elif kind == 'tree-reference':
567
entry_data = entry_key, [
568
(minikind, fingerprint, 0, False, packed_stat),
571
raise errors.BzrError('unknown kind %r' % kind)
572
entry_index, present = self._find_entry_index(entry_key, block)
574
block.insert(entry_index, entry_data)
576
if block[entry_index][1][0][0] != 'a':
577
raise AssertionError(" %r(%r) already added" % (basename, file_id))
578
block[entry_index][1][0] = entry_data[1][0]
580
if kind == 'directory':
581
# insert a new dirblock
582
self._ensure_block(block_index, entry_index, utf8path)
583
self._mark_modified()
585
self._add_to_id_index(self._id_index, entry_key)
587
def _bisect(self, paths):
588
"""Bisect through the disk structure for specific rows.
590
:param paths: A list of paths to find
591
:return: A dict mapping path => entries for found entries. Missing
592
entries will not be in the map.
593
The list is not sorted, and entries will be populated
594
based on when they were read.
596
self._requires_lock()
597
# We need the file pointer to be right after the initial header block
598
self._read_header_if_needed()
599
# If _dirblock_state was in memory, we should just return info from
600
# there, this function is only meant to handle when we want to read
602
if self._dirblock_state != DirState.NOT_IN_MEMORY:
603
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
605
# The disk representation is generally info + '\0\n\0' at the end. But
606
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
607
# Because it means we can sync on the '\n'
608
state_file = self._state_file
609
file_size = os.fstat(state_file.fileno()).st_size
610
# We end up with 2 extra fields, we should have a trailing '\n' to
611
# ensure that we read the whole record, and we should have a precursur
612
# '' which ensures that we start after the previous '\n'
613
entry_field_count = self._fields_per_entry() + 1
615
low = self._end_of_header
616
high = file_size - 1 # Ignore the final '\0'
617
# Map from (dir, name) => entry
620
# Avoid infinite seeking
621
max_count = 30*len(paths)
623
# pending is a list of places to look.
624
# each entry is a tuple of low, high, dir_names
625
# low -> the first byte offset to read (inclusive)
626
# high -> the last byte offset (inclusive)
627
# dir_names -> The list of (dir, name) pairs that should be found in
628
# the [low, high] range
629
pending = [(low, high, paths)]
631
page_size = self._bisect_page_size
633
fields_to_entry = self._get_fields_to_entry()
636
low, high, cur_files = pending.pop()
638
if not cur_files or low >= high:
643
if count > max_count:
644
raise errors.BzrError('Too many seeks, most likely a bug.')
646
mid = max(low, (low+high-page_size)/2)
649
# limit the read size, so we don't end up reading data that we have
651
read_size = min(page_size, (high-mid)+1)
652
block = state_file.read(read_size)
655
entries = block.split('\n')
658
# We didn't find a '\n', so we cannot have found any records.
659
# So put this range back and try again. But we know we have to
660
# increase the page size, because a single read did not contain
661
# a record break (so records must be larger than page_size)
663
pending.append((low, high, cur_files))
666
# Check the first and last entries, in case they are partial, or if
667
# we don't care about the rest of this page
669
first_fields = entries[0].split('\0')
670
if len(first_fields) < entry_field_count:
671
# We didn't get the complete first entry
672
# so move start, and grab the next, which
673
# should be a full entry
674
start += len(entries[0])+1
675
first_fields = entries[1].split('\0')
678
if len(first_fields) <= 2:
679
# We didn't even get a filename here... what do we do?
680
# Try a large page size and repeat this query
682
pending.append((low, high, cur_files))
685
# Find what entries we are looking for, which occur before and
686
# after this first record.
689
first_path = first_fields[1] + '/' + first_fields[2]
691
first_path = first_fields[2]
692
first_loc = _bisect_path_left(cur_files, first_path)
694
# These exist before the current location
695
pre = cur_files[:first_loc]
696
# These occur after the current location, which may be in the
697
# data we read, or might be after the last entry
698
post = cur_files[first_loc:]
700
if post and len(first_fields) >= entry_field_count:
701
# We have files after the first entry
703
# Parse the last entry
704
last_entry_num = len(entries)-1
705
last_fields = entries[last_entry_num].split('\0')
706
if len(last_fields) < entry_field_count:
707
# The very last hunk was not complete,
708
# read the previous hunk
709
after = mid + len(block) - len(entries[-1])
711
last_fields = entries[last_entry_num].split('\0')
713
after = mid + len(block)
716
last_path = last_fields[1] + '/' + last_fields[2]
718
last_path = last_fields[2]
719
last_loc = _bisect_path_right(post, last_path)
721
middle_files = post[:last_loc]
722
post = post[last_loc:]
725
# We have files that should occur in this block
726
# (>= first, <= last)
727
# Either we will find them here, or we can mark them as
730
if middle_files[0] == first_path:
731
# We might need to go before this location
732
pre.append(first_path)
733
if middle_files[-1] == last_path:
734
post.insert(0, last_path)
736
# Find out what paths we have
737
paths = {first_path:[first_fields]}
738
# last_path might == first_path so we need to be
739
# careful if we should append rather than overwrite
740
if last_entry_num != first_entry_num:
741
paths.setdefault(last_path, []).append(last_fields)
742
for num in range(first_entry_num+1, last_entry_num):
743
# TODO: jam 20070223 We are already splitting here, so
744
# shouldn't we just split the whole thing rather
745
# than doing the split again in add_one_record?
746
fields = entries[num].split('\0')
748
path = fields[1] + '/' + fields[2]
751
paths.setdefault(path, []).append(fields)
753
for path in middle_files:
754
for fields in paths.get(path, []):
755
# offset by 1 because of the opening '\0'
756
# consider changing fields_to_entry to avoid the
758
entry = fields_to_entry(fields[1:])
759
found.setdefault(path, []).append(entry)
761
# Now we have split up everything into pre, middle, and post, and
762
# we have handled everything that fell in 'middle'.
763
# We add 'post' first, so that we prefer to seek towards the
764
# beginning, so that we will tend to go as early as we need, and
765
# then only seek forward after that.
767
pending.append((after, high, post))
769
pending.append((low, start-1, pre))
771
# Consider that we may want to return the directory entries in sorted
772
# order. For now, we just return them in whatever order we found them,
773
# and leave it up to the caller if they care if it is ordered or not.
776
def _bisect_dirblocks(self, dir_list):
777
"""Bisect through the disk structure to find entries in given dirs.
779
_bisect_dirblocks is meant to find the contents of directories, which
780
differs from _bisect, which only finds individual entries.
782
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
783
:return: A map from dir => entries_for_dir
785
# TODO: jam 20070223 A lot of the bisecting logic could be shared
786
# between this and _bisect. It would require parameterizing the
787
# inner loop with a function, though. We should evaluate the
788
# performance difference.
789
self._requires_lock()
790
# We need the file pointer to be right after the initial header block
791
self._read_header_if_needed()
792
# If _dirblock_state was in memory, we should just return info from
793
# there, this function is only meant to handle when we want to read
795
if self._dirblock_state != DirState.NOT_IN_MEMORY:
796
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
797
# The disk representation is generally info + '\0\n\0' at the end. But
798
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
799
# Because it means we can sync on the '\n'
800
state_file = self._state_file
801
file_size = os.fstat(state_file.fileno()).st_size
802
# We end up with 2 extra fields, we should have a trailing '\n' to
803
# ensure that we read the whole record, and we should have a precursur
804
# '' which ensures that we start after the previous '\n'
805
entry_field_count = self._fields_per_entry() + 1
807
low = self._end_of_header
808
high = file_size - 1 # Ignore the final '\0'
809
# Map from dir => entry
812
# Avoid infinite seeking
813
max_count = 30*len(dir_list)
815
# pending is a list of places to look.
816
# each entry is a tuple of low, high, dir_names
817
# low -> the first byte offset to read (inclusive)
818
# high -> the last byte offset (inclusive)
819
# dirs -> The list of directories that should be found in
820
# the [low, high] range
821
pending = [(low, high, dir_list)]
823
page_size = self._bisect_page_size
825
fields_to_entry = self._get_fields_to_entry()
828
low, high, cur_dirs = pending.pop()
830
if not cur_dirs or low >= high:
835
if count > max_count:
836
raise errors.BzrError('Too many seeks, most likely a bug.')
838
mid = max(low, (low+high-page_size)/2)
841
# limit the read size, so we don't end up reading data that we have
843
read_size = min(page_size, (high-mid)+1)
844
block = state_file.read(read_size)
847
entries = block.split('\n')
850
# We didn't find a '\n', so we cannot have found any records.
851
# So put this range back and try again. But we know we have to
852
# increase the page size, because a single read did not contain
853
# a record break (so records must be larger than page_size)
855
pending.append((low, high, cur_dirs))
858
# Check the first and last entries, in case they are partial, or if
859
# we don't care about the rest of this page
861
first_fields = entries[0].split('\0')
862
if len(first_fields) < entry_field_count:
863
# We didn't get the complete first entry
864
# so move start, and grab the next, which
865
# should be a full entry
866
start += len(entries[0])+1
867
first_fields = entries[1].split('\0')
870
if len(first_fields) <= 1:
871
# We didn't even get a dirname here... what do we do?
872
# Try a large page size and repeat this query
874
pending.append((low, high, cur_dirs))
877
# Find what entries we are looking for, which occur before and
878
# after this first record.
880
first_dir = first_fields[1]
881
first_loc = bisect.bisect_left(cur_dirs, first_dir)
883
# These exist before the current location
884
pre = cur_dirs[:first_loc]
885
# These occur after the current location, which may be in the
886
# data we read, or might be after the last entry
887
post = cur_dirs[first_loc:]
889
if post and len(first_fields) >= entry_field_count:
890
# We have records to look at after the first entry
892
# Parse the last entry
893
last_entry_num = len(entries)-1
894
last_fields = entries[last_entry_num].split('\0')
895
if len(last_fields) < entry_field_count:
896
# The very last hunk was not complete,
897
# read the previous hunk
898
after = mid + len(block) - len(entries[-1])
900
last_fields = entries[last_entry_num].split('\0')
902
after = mid + len(block)
904
last_dir = last_fields[1]
905
last_loc = bisect.bisect_right(post, last_dir)
907
middle_files = post[:last_loc]
908
post = post[last_loc:]
911
# We have files that should occur in this block
912
# (>= first, <= last)
913
# Either we will find them here, or we can mark them as
916
if middle_files[0] == first_dir:
917
# We might need to go before this location
918
pre.append(first_dir)
919
if middle_files[-1] == last_dir:
920
post.insert(0, last_dir)
922
# Find out what paths we have
923
paths = {first_dir:[first_fields]}
924
# last_dir might == first_dir so we need to be
925
# careful if we should append rather than overwrite
926
if last_entry_num != first_entry_num:
927
paths.setdefault(last_dir, []).append(last_fields)
928
for num in range(first_entry_num+1, last_entry_num):
929
# TODO: jam 20070223 We are already splitting here, so
930
# shouldn't we just split the whole thing rather
931
# than doing the split again in add_one_record?
932
fields = entries[num].split('\0')
933
paths.setdefault(fields[1], []).append(fields)
935
for cur_dir in middle_files:
936
for fields in paths.get(cur_dir, []):
937
# offset by 1 because of the opening '\0'
938
# consider changing fields_to_entry to avoid the
940
entry = fields_to_entry(fields[1:])
941
found.setdefault(cur_dir, []).append(entry)
943
# Now we have split up everything into pre, middle, and post, and
944
# we have handled everything that fell in 'middle'.
945
# We add 'post' first, so that we prefer to seek towards the
946
# beginning, so that we will tend to go as early as we need, and
947
# then only seek forward after that.
949
pending.append((after, high, post))
951
pending.append((low, start-1, pre))
955
def _bisect_recursive(self, paths):
956
"""Bisect for entries for all paths and their children.
958
This will use bisect to find all records for the supplied paths. It
959
will then continue to bisect for any records which are marked as
960
directories. (and renames?)
962
:param paths: A sorted list of (dir, name) pairs
963
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
964
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
966
# Map from (dir, name, file_id) => [tree_info]
969
found_dir_names = set()
971
# Directories that have been read
972
processed_dirs = set()
973
# Get the ball rolling with the first bisect for all entries.
974
newly_found = self._bisect(paths)
977
# Directories that need to be read
979
paths_to_search = set()
980
for entry_list in viewvalues(newly_found):
981
for dir_name_id, trees_info in entry_list:
982
found[dir_name_id] = trees_info
983
found_dir_names.add(dir_name_id[:2])
985
for tree_info in trees_info:
986
minikind = tree_info[0]
989
# We already processed this one as a directory,
990
# we don't need to do the extra work again.
992
subdir, name, file_id = dir_name_id
993
path = osutils.pathjoin(subdir, name)
995
if path not in processed_dirs:
996
pending_dirs.add(path)
997
elif minikind == 'r':
998
# Rename, we need to directly search the target
999
# which is contained in the fingerprint column
1000
dir_name = osutils.split(tree_info[1])
1001
if dir_name[0] in pending_dirs:
1002
# This entry will be found in the dir search
1004
if dir_name not in found_dir_names:
1005
paths_to_search.add(tree_info[1])
1006
# Now we have a list of paths to look for directly, and
1007
# directory blocks that need to be read.
1008
# newly_found is mixing the keys between (dir, name) and path
1009
# entries, but that is okay, because we only really care about the
1011
newly_found = self._bisect(sorted(paths_to_search))
1012
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
1013
processed_dirs.update(pending_dirs)
1016
def _discard_merge_parents(self):
1017
"""Discard any parents trees beyond the first.
1019
Note that if this fails the dirstate is corrupted.
1021
After this function returns the dirstate contains 2 trees, neither of
1024
self._read_header_if_needed()
1025
parents = self.get_parent_ids()
1026
if len(parents) < 1:
1028
# only require all dirblocks if we are doing a full-pass removal.
1029
self._read_dirblocks_if_needed()
1030
dead_patterns = {('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')}
1031
def iter_entries_removable():
1032
for block in self._dirblocks:
1033
deleted_positions = []
1034
for pos, entry in enumerate(block[1]):
1036
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
1037
deleted_positions.append(pos)
1038
if deleted_positions:
1039
if len(deleted_positions) == len(block[1]):
1042
for pos in reversed(deleted_positions):
1044
# if the first parent is a ghost:
1045
if parents[0] in self.get_ghosts():
1046
empty_parent = [DirState.NULL_PARENT_DETAILS]
1047
for entry in iter_entries_removable():
1048
entry[1][1:] = empty_parent
1050
for entry in iter_entries_removable():
1054
self._parents = [parents[0]]
1055
self._mark_modified(header_modified=True)
1057
def _empty_parent_info(self):
1058
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
1061
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
1062
"""Ensure a block for dirname exists.
1064
This function exists to let callers which know that there is a
1065
directory dirname ensure that the block for it exists. This block can
1066
fail to exist because of demand loading, or because a directory had no
1067
children. In either case it is not an error. It is however an error to
1068
call this if there is no parent entry for the directory, and thus the
1069
function requires the coordinates of such an entry to be provided.
1071
The root row is special cased and can be indicated with a parent block
1074
:param parent_block_index: The index of the block in which dirname's row
1076
:param parent_row_index: The index in the parent block where the row
1078
:param dirname: The utf8 dirname to ensure there is a block for.
1079
:return: The index for the block.
1081
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
1082
# This is the signature of the root row, and the
1083
# contents-of-root row is always index 1
1085
# the basename of the directory must be the end of its full name.
1086
if not (parent_block_index == -1 and
1087
parent_block_index == -1 and dirname == ''):
1088
if not dirname.endswith(
1089
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
1090
raise AssertionError("bad dirname %r" % dirname)
1091
block_index, present = self._find_block_index_from_key((dirname, '', ''))
1093
## In future, when doing partial parsing, this should load and
1094
# populate the entire block.
1095
self._dirblocks.insert(block_index, (dirname, []))
1098
def _entries_to_current_state(self, new_entries):
1099
"""Load new_entries into self.dirblocks.
1101
Process new_entries into the current state object, making them the active
1102
state. The entries are grouped together by directory to form dirblocks.
1104
:param new_entries: A sorted list of entries. This function does not sort
1105
to prevent unneeded overhead when callers have a sorted list already.
1108
if new_entries[0][0][0:2] != ('', ''):
1109
raise AssertionError(
1110
"Missing root row %r" % (new_entries[0][0],))
1111
# The two blocks here are deliberate: the root block and the
1112
# contents-of-root block.
1113
self._dirblocks = [('', []), ('', [])]
1114
current_block = self._dirblocks[0][1]
1115
current_dirname = ''
1117
append_entry = current_block.append
1118
for entry in new_entries:
1119
if entry[0][0] != current_dirname:
1120
# new block - different dirname
1122
current_dirname = entry[0][0]
1123
self._dirblocks.append((current_dirname, current_block))
1124
append_entry = current_block.append
1125
# append the entry to the current block
1127
self._split_root_dirblock_into_contents()
1129
def _split_root_dirblock_into_contents(self):
1130
"""Split the root dirblocks into root and contents-of-root.
1132
After parsing by path, we end up with root entries and contents-of-root
1133
entries in the same block. This loop splits them out again.
1135
# The above loop leaves the "root block" entries mixed with the
1136
# "contents-of-root block". But we don't want an if check on
1137
# all entries, so instead we just fix it up here.
1138
if self._dirblocks[1] != ('', []):
1139
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1141
contents_of_root_block = []
1142
for entry in self._dirblocks[0][1]:
1143
if not entry[0][1]: # This is a root entry
1144
root_block.append(entry)
1146
contents_of_root_block.append(entry)
1147
self._dirblocks[0] = ('', root_block)
1148
self._dirblocks[1] = ('', contents_of_root_block)
1150
def _entries_for_path(self, path):
1151
"""Return a list with all the entries that match path for all ids."""
1152
dirname, basename = os.path.split(path)
1153
key = (dirname, basename, '')
1154
block_index, present = self._find_block_index_from_key(key)
1156
# the block which should contain path is absent.
1159
block = self._dirblocks[block_index][1]
1160
entry_index, _ = self._find_entry_index(key, block)
1161
# we may need to look at multiple entries at this path: walk while the specific_files match.
1162
while (entry_index < len(block) and
1163
block[entry_index][0][0:2] == key[0:2]):
1164
result.append(block[entry_index])
1168
def _entry_to_line(self, entry):
1169
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1171
:param entry: An entry_tuple as defined in the module docstring.
1173
entire_entry = list(entry[0])
1174
for tree_number, tree_data in enumerate(entry[1]):
1175
# (minikind, fingerprint, size, executable, tree_specific_string)
1176
entire_entry.extend(tree_data)
1177
# 3 for the key, 5 for the fields per tree.
1178
tree_offset = 3 + tree_number * 5
1180
entire_entry[tree_offset + 0] = tree_data[0]
1182
entire_entry[tree_offset + 2] = str(tree_data[2])
1184
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1185
return '\0'.join(entire_entry)
1187
def _fields_per_entry(self):
1188
"""How many null separated fields should be in each entry row.
1190
Each line now has an extra '\\n' field which is not used
1191
so we just skip over it
1194
3 fields for the key
1195
+ number of fields per tree_data (5) * tree count
1198
tree_count = 1 + self._num_present_parents()
1199
return 3 + 5 * tree_count + 1
1201
def _find_block(self, key, add_if_missing=False):
1202
"""Return the block that key should be present in.
1204
:param key: A dirstate entry key.
1205
:return: The block tuple.
1207
block_index, present = self._find_block_index_from_key(key)
1209
if not add_if_missing:
1210
# check to see if key is versioned itself - we might want to
1211
# add it anyway, because dirs with no entries dont get a
1212
# dirblock at parse time.
1213
# This is an uncommon branch to take: most dirs have children,
1214
# and most code works with versioned paths.
1215
parent_base, parent_name = osutils.split(key[0])
1216
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1217
# some parent path has not been added - its an error to add
1219
raise errors.NotVersionedError(key[0:2], str(self))
1220
self._dirblocks.insert(block_index, (key[0], []))
1221
return self._dirblocks[block_index]
1223
def _find_block_index_from_key(self, key):
1224
"""Find the dirblock index for a key.
1226
:return: The block index, True if the block for the key is present.
1228
if key[0:2] == ('', ''):
1231
if (self._last_block_index is not None and
1232
self._dirblocks[self._last_block_index][0] == key[0]):
1233
return self._last_block_index, True
1236
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1237
cache=self._split_path_cache)
1238
# _right returns one-past-where-key is so we have to subtract
1239
# one to use it. we use _right here because there are two
1240
# '' blocks - the root, and the contents of root
1241
# we always have a minimum of 2 in self._dirblocks: root and
1242
# root-contents, and for '', we get 2 back, so this is
1243
# simple and correct:
1244
present = (block_index < len(self._dirblocks) and
1245
self._dirblocks[block_index][0] == key[0])
1246
self._last_block_index = block_index
1247
# Reset the entry index cache to the beginning of the block.
1248
self._last_entry_index = -1
1249
return block_index, present
1251
def _find_entry_index(self, key, block):
1252
"""Find the entry index for a key in a block.
1254
:return: The entry index, True if the entry for the key is present.
1256
len_block = len(block)
1258
if self._last_entry_index is not None:
1260
entry_index = self._last_entry_index + 1
1261
# A hit is when the key is after the last slot, and before or
1262
# equal to the next slot.
1263
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1264
key <= block[entry_index][0]):
1265
self._last_entry_index = entry_index
1266
present = (block[entry_index][0] == key)
1267
return entry_index, present
1270
entry_index = bisect.bisect_left(block, (key, []))
1271
present = (entry_index < len_block and
1272
block[entry_index][0] == key)
1273
self._last_entry_index = entry_index
1274
return entry_index, present
1277
def from_tree(tree, dir_state_filename, sha1_provider=None):
1278
"""Create a dirstate from a bzr Tree.
1280
:param tree: The tree which should provide parent information and
1282
:param sha1_provider: an object meeting the SHA1Provider interface.
1283
If None, a DefaultSHA1Provider is used.
1284
:return: a DirState object which is currently locked for writing.
1285
(it was locked by DirState.initialize)
1287
result = DirState.initialize(dir_state_filename,
1288
sha1_provider=sha1_provider)
1292
parent_ids = tree.get_parent_ids()
1293
num_parents = len(parent_ids)
1295
for parent_id in parent_ids:
1296
parent_tree = tree.branch.repository.revision_tree(parent_id)
1297
parent_trees.append((parent_id, parent_tree))
1298
parent_tree.lock_read()
1299
result.set_parent_trees(parent_trees, [])
1300
result.set_state_from_inventory(tree.root_inventory)
1302
for revid, parent_tree in parent_trees:
1303
parent_tree.unlock()
1306
# The caller won't have a chance to unlock this, so make sure we
1312
def _check_delta_is_valid(self, delta):
1313
return list(inventory._check_delta_unique_ids(
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
def update_by_delta(self, delta):
1321
"""Apply an inventory delta to the dirstate for tree 0
1323
This is the workhorse for apply_inventory_delta in dirstate based
1326
:param delta: An inventory delta. See Inventory.apply_delta for
1329
self._read_dirblocks_if_needed()
1330
encode = cache_utf8.encode
1333
# Accumulate parent references (path_utf8, id), to check for parentless
1334
# items or items placed under files/links/tree-references. We get
1335
# references from every item in the delta that is not a deletion and
1336
# is not itself the root.
1338
# Added ids must not be in the dirstate already. This set holds those
1341
# This loop transforms the delta to single atomic operations that can
1342
# be executed and validated.
1343
delta = sorted(self._check_delta_is_valid(delta), reverse=True)
1344
for old_path, new_path, file_id, inv_entry in delta:
1345
if (file_id in insertions) or (file_id in removals):
1346
self._raise_invalid(old_path or new_path, file_id,
1348
if old_path is not None:
1349
old_path = old_path.encode('utf-8')
1350
removals[file_id] = old_path
1352
new_ids.add(file_id)
1353
if new_path is not None:
1354
if inv_entry is None:
1355
self._raise_invalid(new_path, file_id,
1356
"new_path with no entry")
1357
new_path = new_path.encode('utf-8')
1358
dirname_utf8, basename = osutils.split(new_path)
1360
parents.add((dirname_utf8, inv_entry.parent_id))
1361
key = (dirname_utf8, basename, file_id)
1362
minikind = DirState._kind_to_minikind[inv_entry.kind]
1364
fingerprint = inv_entry.reference_revision or ''
1367
insertions[file_id] = (key, minikind, inv_entry.executable,
1368
fingerprint, new_path)
1369
# Transform moves into delete+add pairs
1370
if None not in (old_path, new_path):
1371
for child in self._iter_child_entries(0, old_path):
1372
if child[0][2] in insertions or child[0][2] in removals:
1374
child_dirname = child[0][0]
1375
child_basename = child[0][1]
1376
minikind = child[1][0][0]
1377
fingerprint = child[1][0][4]
1378
executable = child[1][0][3]
1379
old_child_path = osutils.pathjoin(child_dirname,
1381
removals[child[0][2]] = old_child_path
1382
child_suffix = child_dirname[len(old_path):]
1383
new_child_dirname = (new_path + child_suffix)
1384
key = (new_child_dirname, child_basename, child[0][2])
1385
new_child_path = osutils.pathjoin(new_child_dirname,
1387
insertions[child[0][2]] = (key, minikind, executable,
1388
fingerprint, new_child_path)
1389
self._check_delta_ids_absent(new_ids, delta, 0)
1391
self._apply_removals(viewitems(removals))
1392
self._apply_insertions(viewvalues(insertions))
1394
self._after_delta_check_parents(parents, 0)
1395
except errors.BzrError as e:
1396
self._changes_aborted = True
1397
if 'integrity error' not in str(e):
1399
# _get_entry raises BzrError when a request is inconsistent; we
1400
# want such errors to be shown as InconsistentDelta - and that
1401
# fits the behaviour we trigger.
1402
raise errors.InconsistentDeltaDelta(delta,
1403
"error from _get_entry. %s" % (e,))
1405
def _apply_removals(self, removals):
1406
for file_id, path in sorted(removals, reverse=True,
1407
key=operator.itemgetter(1)):
1408
dirname, basename = osutils.split(path)
1409
block_i, entry_i, d_present, f_present = \
1410
self._get_block_entry_index(dirname, basename, 0)
1412
entry = self._dirblocks[block_i][1][entry_i]
1414
self._raise_invalid(path, file_id,
1415
"Wrong path for old path.")
1416
if not f_present or entry[1][0][0] in 'ar':
1417
self._raise_invalid(path, file_id,
1418
"Wrong path for old path.")
1419
if file_id != entry[0][2]:
1420
self._raise_invalid(path, file_id,
1421
"Attempt to remove path has wrong id - found %r."
1423
self._make_absent(entry)
1424
# See if we have a malformed delta: deleting a directory must not
1425
# leave crud behind. This increases the number of bisects needed
1426
# substantially, but deletion or renames of large numbers of paths
1427
# is rare enough it shouldn't be an issue (famous last words?) RBC
1429
block_i, entry_i, d_present, f_present = \
1430
self._get_block_entry_index(path, '', 0)
1432
# The dir block is still present in the dirstate; this could
1433
# be due to it being in a parent tree, or a corrupt delta.
1434
for child_entry in self._dirblocks[block_i][1]:
1435
if child_entry[1][0][0] not in ('r', 'a'):
1436
self._raise_invalid(path, entry[0][2],
1437
"The file id was deleted but its children were "
1440
def _apply_insertions(self, adds):
1442
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1443
self.update_minimal(key, minikind, executable, fingerprint,
1444
path_utf8=path_utf8)
1445
except errors.NotVersionedError:
1446
self._raise_invalid(path_utf8.decode('utf8'), key[2],
1449
def update_basis_by_delta(self, delta, new_revid):
1450
"""Update the parents of this tree after a commit.
1452
This gives the tree one parent, with revision id new_revid. The
1453
inventory delta is applied to the current basis tree to generate the
1454
inventory for the parent new_revid, and all other parent trees are
1457
Note that an exception during the operation of this method will leave
1458
the dirstate in a corrupt state where it should not be saved.
1460
:param new_revid: The new revision id for the trees parent.
1461
:param delta: An inventory delta (see apply_inventory_delta) describing
1462
the changes from the current left most parent revision to new_revid.
1464
self._read_dirblocks_if_needed()
1465
self._discard_merge_parents()
1466
if self._ghosts != []:
1467
raise NotImplementedError(self.update_basis_by_delta)
1468
if len(self._parents) == 0:
1469
# setup a blank tree, the most simple way.
1470
empty_parent = DirState.NULL_PARENT_DETAILS
1471
for entry in self._iter_entries():
1472
entry[1].append(empty_parent)
1473
self._parents.append(new_revid)
1475
self._parents[0] = new_revid
1477
delta = sorted(self._check_delta_is_valid(delta), reverse=True)
1481
# The paths this function accepts are unicode and must be encoded as we
1483
encode = cache_utf8.encode
1484
inv_to_entry = self._inv_entry_to_details
1485
# delta is now (deletes, changes), (adds) in reverse lexographical
1487
# deletes in reverse lexographic order are safe to process in situ.
1488
# renames are not, as a rename from any path could go to a path
1489
# lexographically lower, so we transform renames into delete, add pairs,
1490
# expanding them recursively as needed.
1491
# At the same time, to reduce interface friction we convert the input
1492
# inventory entries to dirstate.
1493
root_only = ('', '')
1494
# Accumulate parent references (path_utf8, id), to check for parentless
1495
# items or items placed under files/links/tree-references. We get
1496
# references from every item in the delta that is not a deletion and
1497
# is not itself the root.
1499
# Added ids must not be in the dirstate already. This set holds those
1502
for old_path, new_path, file_id, inv_entry in delta:
1503
if inv_entry is not None and file_id != inv_entry.file_id:
1504
self._raise_invalid(new_path, file_id,
1505
"mismatched entry file_id %r" % inv_entry)
1506
if new_path is None:
1507
new_path_utf8 = None
1509
if inv_entry is None:
1510
self._raise_invalid(new_path, file_id,
1511
"new_path with no entry")
1512
new_path_utf8 = encode(new_path)
1513
# note the parent for validation
1514
dirname_utf8, basename_utf8 = osutils.split(new_path_utf8)
1516
parents.add((dirname_utf8, inv_entry.parent_id))
1517
if old_path is None:
1518
old_path_utf8 = None
1520
old_path_utf8 = encode(old_path)
1521
if old_path is None:
1522
adds.append((None, new_path_utf8, file_id,
1523
inv_to_entry(inv_entry), True))
1524
new_ids.add(file_id)
1525
elif new_path is None:
1526
deletes.append((old_path_utf8, None, file_id, None, True))
1527
elif (old_path, new_path) == root_only:
1528
# change things in-place
1529
# Note: the case of a parent directory changing its file_id
1530
# tends to break optimizations here, because officially
1531
# the file has actually been moved, it just happens to
1532
# end up at the same path. If we can figure out how to
1533
# handle that case, we can avoid a lot of add+delete
1534
# pairs for objects that stay put.
1535
# elif old_path == new_path:
1536
changes.append((old_path_utf8, new_path_utf8, file_id,
1537
inv_to_entry(inv_entry)))
1540
# Because renames must preserve their children we must have
1541
# processed all relocations and removes before hand. The sort
1542
# order ensures we've examined the child paths, but we also
1543
# have to execute the removals, or the split to an add/delete
1544
# pair will result in the deleted item being reinserted, or
1545
# renamed items being reinserted twice - and possibly at the
1546
# wrong place. Splitting into a delete/add pair also simplifies
1547
# the handling of entries with ('f', ...), ('r' ...) because
1548
# the target of the 'r' is old_path here, and we add that to
1549
# deletes, meaning that the add handler does not need to check
1550
# for 'r' items on every pass.
1551
self._update_basis_apply_deletes(deletes)
1553
# Split into an add/delete pair recursively.
1554
adds.append((old_path_utf8, new_path_utf8, file_id,
1555
inv_to_entry(inv_entry), False))
1556
# Expunge deletes that we've seen so that deleted/renamed
1557
# children of a rename directory are handled correctly.
1558
new_deletes = reversed(list(
1559
self._iter_child_entries(1, old_path_utf8)))
1560
# Remove the current contents of the tree at orig_path, and
1561
# reinsert at the correct new path.
1562
for entry in new_deletes:
1563
child_dirname, child_basename, child_file_id = entry[0]
1565
source_path = child_dirname + '/' + child_basename
1567
source_path = child_basename
1570
new_path_utf8 + source_path[len(old_path_utf8):]
1572
if old_path_utf8 == '':
1573
raise AssertionError("cannot rename directory to"
1575
target_path = source_path[len(old_path_utf8) + 1:]
1576
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1578
(source_path, target_path, entry[0][2], None, False))
1580
(old_path_utf8, new_path_utf8, file_id, None, False))
1582
self._check_delta_ids_absent(new_ids, delta, 1)
1584
# Finish expunging deletes/first half of renames.
1585
self._update_basis_apply_deletes(deletes)
1586
# Reinstate second half of renames and new paths.
1587
self._update_basis_apply_adds(adds)
1588
# Apply in-situ changes.
1589
self._update_basis_apply_changes(changes)
1591
self._after_delta_check_parents(parents, 1)
1592
except errors.BzrError as e:
1593
self._changes_aborted = True
1594
if 'integrity error' not in str(e):
1596
# _get_entry raises BzrError when a request is inconsistent; we
1597
# want such errors to be shown as InconsistentDelta - and that
1598
# fits the behaviour we trigger.
1599
raise errors.InconsistentDeltaDelta(delta,
1600
"error from _get_entry. %s" % (e,))
1602
self._mark_modified(header_modified=True)
1603
self._id_index = None
1606
def _check_delta_ids_absent(self, new_ids, delta, tree_index):
1607
"""Check that none of the file_ids in new_ids are present in a tree."""
1610
id_index = self._get_id_index()
1611
for file_id in new_ids:
1612
for key in id_index.get(file_id, ()):
1613
block_i, entry_i, d_present, f_present = \
1614
self._get_block_entry_index(key[0], key[1], tree_index)
1616
# In a different tree
1618
entry = self._dirblocks[block_i][1][entry_i]
1619
if entry[0][2] != file_id:
1620
# Different file_id, so not what we want.
1622
self._raise_invalid(("%s/%s" % key[0:2]).decode('utf8'), file_id,
1623
"This file_id is new in the delta but already present in "
1626
def _raise_invalid(self, path, file_id, reason):
1627
self._changes_aborted = True
1628
raise errors.InconsistentDelta(path, file_id, reason)
1630
def _update_basis_apply_adds(self, adds):
1631
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1633
They may be adds, or renames that have been split into add/delete
1636
:param adds: A sequence of adds. Each add is a tuple:
1637
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1638
is False when the add is the second half of a remove-and-reinsert
1639
pair created to handle renames and deletes.
1641
# Adds are accumulated partly from renames, so can be in any input
1643
# TODO: we may want to sort in dirblocks order. That way each entry
1644
# will end up in the same directory, allowing the _get_entry
1645
# fast-path for looking up 2 items in the same dir work.
1646
adds.sort(key=lambda x: x[1])
1647
# adds is now in lexographic order, which places all parents before
1648
# their children, so we can process it linearly.
1650
st = static_tuple.StaticTuple
1651
for old_path, new_path, file_id, new_details, real_add in adds:
1652
dirname, basename = osutils.split(new_path)
1653
entry_key = st(dirname, basename, file_id)
1654
block_index, present = self._find_block_index_from_key(entry_key)
1656
# The block where we want to put the file is not present.
1657
# However, it might have just been an empty directory. Look for
1658
# the parent in the basis-so-far before throwing an error.
1659
parent_dir, parent_base = osutils.split(dirname)
1660
parent_block_idx, parent_entry_idx, _, parent_present = \
1661
self._get_block_entry_index(parent_dir, parent_base, 1)
1662
if not parent_present:
1663
self._raise_invalid(new_path, file_id,
1664
"Unable to find block for this record."
1665
" Was the parent added?")
1666
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
1668
block = self._dirblocks[block_index][1]
1669
entry_index, present = self._find_entry_index(entry_key, block)
1671
if old_path is not None:
1672
self._raise_invalid(new_path, file_id,
1673
'considered a real add but still had old_path at %s'
1676
entry = block[entry_index]
1677
basis_kind = entry[1][1][0]
1678
if basis_kind == 'a':
1679
entry[1][1] = new_details
1680
elif basis_kind == 'r':
1681
raise NotImplementedError()
1683
self._raise_invalid(new_path, file_id,
1684
"An entry was marked as a new add"
1685
" but the basis target already existed")
1687
# The exact key was not found in the block. However, we need to
1688
# check if there is a key next to us that would have matched.
1689
# We only need to check 2 locations, because there are only 2
1691
for maybe_index in range(entry_index-1, entry_index+1):
1692
if maybe_index < 0 or maybe_index >= len(block):
1694
maybe_entry = block[maybe_index]
1695
if maybe_entry[0][:2] != (dirname, basename):
1696
# Just a random neighbor
1698
if maybe_entry[0][2] == file_id:
1699
raise AssertionError(
1700
'_find_entry_index didnt find a key match'
1701
' but walking the data did, for %s'
1703
basis_kind = maybe_entry[1][1][0]
1704
if basis_kind not in 'ar':
1705
self._raise_invalid(new_path, file_id,
1706
"we have an add record for path, but the path"
1707
" is already present with another file_id %s"
1708
% (maybe_entry[0][2],))
1710
entry = (entry_key, [DirState.NULL_PARENT_DETAILS,
1712
block.insert(entry_index, entry)
1714
active_kind = entry[1][0][0]
1715
if active_kind == 'a':
1716
# The active record shows up as absent, this could be genuine,
1717
# or it could be present at some other location. We need to
1719
id_index = self._get_id_index()
1720
# The id_index may not be perfectly accurate for tree1, because
1721
# we haven't been keeping it updated. However, it should be
1722
# fine for tree0, and that gives us enough info for what we
1724
keys = id_index.get(file_id, ())
1726
block_i, entry_i, d_present, f_present = \
1727
self._get_block_entry_index(key[0], key[1], 0)
1730
active_entry = self._dirblocks[block_i][1][entry_i]
1731
if (active_entry[0][2] != file_id):
1732
# Some other file is at this path, we don't need to
1735
real_active_kind = active_entry[1][0][0]
1736
if real_active_kind in 'ar':
1737
# We found a record, which was not *this* record,
1738
# which matches the file_id, but is not actually
1739
# present. Something seems *really* wrong.
1740
self._raise_invalid(new_path, file_id,
1741
"We found a tree0 entry that doesnt make sense")
1742
# Now, we've found a tree0 entry which matches the file_id
1743
# but is at a different location. So update them to be
1745
active_dir, active_name = active_entry[0][:2]
1747
active_path = active_dir + '/' + active_name
1749
active_path = active_name
1750
active_entry[1][1] = st('r', new_path, 0, False, '')
1751
entry[1][0] = st('r', active_path, 0, False, '')
1752
elif active_kind == 'r':
1753
raise NotImplementedError()
1755
new_kind = new_details[0]
1757
self._ensure_block(block_index, entry_index, new_path)
1759
def _update_basis_apply_changes(self, changes):
1760
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1762
:param adds: A sequence of changes. Each change is a tuple:
1763
(path_utf8, path_utf8, file_id, (entry_details))
1766
for old_path, new_path, file_id, new_details in changes:
1767
# the entry for this file_id must be in tree 0.
1768
entry = self._get_entry(1, file_id, new_path)
1769
if entry[0] is None or entry[1][1][0] in 'ar':
1770
self._raise_invalid(new_path, file_id,
1771
'changed entry considered not present')
1772
entry[1][1] = new_details
1774
def _update_basis_apply_deletes(self, deletes):
1775
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1777
They may be deletes, or renames that have been split into add/delete
1780
:param deletes: A sequence of deletes. Each delete is a tuple:
1781
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1782
real_delete is True when the desired outcome is an actual deletion
1783
rather than the rename handling logic temporarily deleting a path
1784
during the replacement of a parent.
1786
null = DirState.NULL_PARENT_DETAILS
1787
for old_path, new_path, file_id, _, real_delete in deletes:
1788
if real_delete != (new_path is None):
1789
self._raise_invalid(old_path, file_id, "bad delete delta")
1790
# the entry for this file_id must be in tree 1.
1791
dirname, basename = osutils.split(old_path)
1792
block_index, entry_index, dir_present, file_present = \
1793
self._get_block_entry_index(dirname, basename, 1)
1794
if not file_present:
1795
self._raise_invalid(old_path, file_id,
1796
'basis tree does not contain removed entry')
1797
entry = self._dirblocks[block_index][1][entry_index]
1798
# The state of the entry in the 'active' WT
1799
active_kind = entry[1][0][0]
1800
if entry[0][2] != file_id:
1801
self._raise_invalid(old_path, file_id,
1802
'mismatched file_id in tree 1')
1804
old_kind = entry[1][1][0]
1805
if active_kind in 'ar':
1806
# The active tree doesn't have this file_id.
1807
# The basis tree is changing this record. If this is a
1808
# rename, then we don't want the record here at all
1809
# anymore. If it is just an in-place change, we want the
1810
# record here, but we'll add it if we need to. So we just
1812
if active_kind == 'r':
1813
active_path = entry[1][0][1]
1814
active_entry = self._get_entry(0, file_id, active_path)
1815
if active_entry[1][1][0] != 'r':
1816
self._raise_invalid(old_path, file_id,
1817
"Dirstate did not have matching rename entries")
1818
elif active_entry[1][0][0] in 'ar':
1819
self._raise_invalid(old_path, file_id,
1820
"Dirstate had a rename pointing at an inactive"
1822
active_entry[1][1] = null
1823
del self._dirblocks[block_index][1][entry_index]
1825
# This was a directory, and the active tree says it
1826
# doesn't exist, and now the basis tree says it doesn't
1827
# exist. Remove its dirblock if present
1829
present) = self._find_block_index_from_key(
1832
dir_block = self._dirblocks[dir_block_index][1]
1834
# This entry is empty, go ahead and just remove it
1835
del self._dirblocks[dir_block_index]
1837
# There is still an active record, so just mark this
1840
block_i, entry_i, d_present, f_present = \
1841
self._get_block_entry_index(old_path, '', 1)
1843
dir_block = self._dirblocks[block_i][1]
1844
for child_entry in dir_block:
1845
child_basis_kind = child_entry[1][1][0]
1846
if child_basis_kind not in 'ar':
1847
self._raise_invalid(old_path, file_id,
1848
"The file id was deleted but its children were "
1851
def _after_delta_check_parents(self, parents, index):
1852
"""Check that parents required by the delta are all intact.
1854
:param parents: An iterable of (path_utf8, file_id) tuples which are
1855
required to be present in tree 'index' at path_utf8 with id file_id
1857
:param index: The column in the dirstate to check for parents in.
1859
for dirname_utf8, file_id in parents:
1860
# Get the entry - the ensures that file_id, dirname_utf8 exists and
1861
# has the right file id.
1862
entry = self._get_entry(index, file_id, dirname_utf8)
1863
if entry[1] is None:
1864
self._raise_invalid(dirname_utf8.decode('utf8'),
1865
file_id, "This parent is not present.")
1866
# Parents of things must be directories
1867
if entry[1][index][0] != 'd':
1868
self._raise_invalid(dirname_utf8.decode('utf8'),
1869
file_id, "This parent is not a directory.")
1871
def _observed_sha1(self, entry, sha1, stat_value,
1872
_stat_to_minikind=_stat_to_minikind):
1873
"""Note the sha1 of a file.
1875
:param entry: The entry the sha1 is for.
1876
:param sha1: The observed sha1.
1877
:param stat_value: The os.lstat for the file.
1880
minikind = _stat_to_minikind[stat_value.st_mode & 0o170000]
1885
if self._cutoff_time is None:
1886
self._sha_cutoff_time()
1887
if (stat_value.st_mtime < self._cutoff_time
1888
and stat_value.st_ctime < self._cutoff_time):
1889
entry[1][0] = ('f', sha1, stat_value.st_size, entry[1][0][3],
1890
pack_stat(stat_value))
1891
self._mark_modified([entry])
1893
def _sha_cutoff_time(self):
1894
"""Return cutoff time.
1896
Files modified more recently than this time are at risk of being
1897
undetectably modified and so can't be cached.
1899
# Cache the cutoff time as long as we hold a lock.
1900
# time.time() isn't super expensive (approx 3.38us), but
1901
# when you call it 50,000 times it adds up.
1902
# For comparison, os.lstat() costs 7.2us if it is hot.
1903
self._cutoff_time = int(time.time()) - 3
1904
return self._cutoff_time
1906
def _lstat(self, abspath, entry):
1907
"""Return the os.lstat value for this path."""
1908
return os.lstat(abspath)
1910
def _sha1_file_and_mutter(self, abspath):
1911
# when -Dhashcache is turned on, this is monkey-patched in to log
1913
trace.mutter("dirstate sha1 " + abspath)
1914
return self._sha1_provider.sha1(abspath)
1916
def _is_executable(self, mode, old_executable):
1917
"""Is this file executable?"""
1918
return bool(S_IEXEC & mode)
1920
def _is_executable_win32(self, mode, old_executable):
1921
"""On win32 the executable bit is stored in the dirstate."""
1922
return old_executable
1924
if sys.platform == 'win32':
1925
_is_executable = _is_executable_win32
1927
def _read_link(self, abspath, old_link):
1928
"""Read the target of a symlink"""
1929
# TODO: jam 200700301 On Win32, this could just return the value
1930
# already in memory. However, this really needs to be done at a
1931
# higher level, because there either won't be anything on disk,
1932
# or the thing on disk will be a file.
1933
fs_encoding = osutils._fs_enc
1934
if isinstance(abspath, unicode):
1935
# abspath is defined as the path to pass to lstat. readlink is
1936
# buggy in python < 2.6 (it doesn't encode unicode path into FS
1937
# encoding), so we need to encode ourselves knowing that unicode
1938
# paths are produced by UnicodeDirReader on purpose.
1939
abspath = abspath.encode(fs_encoding)
1940
target = os.readlink(abspath)
1941
if fs_encoding not in ('utf-8', 'ascii'):
1942
# Change encoding if needed
1943
target = target.decode(fs_encoding).encode('UTF-8')
1946
def get_ghosts(self):
1947
"""Return a list of the parent tree revision ids that are ghosts."""
1948
self._read_header_if_needed()
1951
def get_lines(self):
1952
"""Serialise the entire dirstate to a sequence of lines."""
1953
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1954
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1955
# read what's on disk.
1956
self._state_file.seek(0)
1957
return self._state_file.readlines()
1959
lines.append(self._get_parents_line(self.get_parent_ids()))
1960
lines.append(self._get_ghosts_line(self._ghosts))
1961
lines.extend(self._iter_entry_lines())
1962
return self._get_output_lines(lines)
1964
def _get_ghosts_line(self, ghost_ids):
1965
"""Create a line for the state file for ghost information."""
1966
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1968
def _get_parents_line(self, parent_ids):
1969
"""Create a line for the state file for parents information."""
1970
return '\0'.join([str(len(parent_ids))] + parent_ids)
1972
def _iter_entry_lines(self):
1973
"""Create lines for entries."""
1974
return map(self._entry_to_line, self._iter_entries())
1976
def _get_fields_to_entry(self):
1977
"""Get a function which converts entry fields into a entry record.
1979
This handles size and executable, as well as parent records.
1981
:return: A function which takes a list of fields, and returns an
1982
appropriate record for storing in memory.
1984
# This is intentionally unrolled for performance
1985
num_present_parents = self._num_present_parents()
1986
if num_present_parents == 0:
1987
def fields_to_entry_0_parents(fields, _int=int):
1988
path_name_file_id_key = (fields[0], fields[1], fields[2])
1989
return (path_name_file_id_key, [
1991
fields[3], # minikind
1992
fields[4], # fingerprint
1993
_int(fields[5]), # size
1994
fields[6] == 'y', # executable
1995
fields[7], # packed_stat or revision_id
1997
return fields_to_entry_0_parents
1998
elif num_present_parents == 1:
1999
def fields_to_entry_1_parent(fields, _int=int):
2000
path_name_file_id_key = (fields[0], fields[1], fields[2])
2001
return (path_name_file_id_key, [
2003
fields[3], # minikind
2004
fields[4], # fingerprint
2005
_int(fields[5]), # size
2006
fields[6] == 'y', # executable
2007
fields[7], # packed_stat or revision_id
2010
fields[8], # minikind
2011
fields[9], # fingerprint
2012
_int(fields[10]), # size
2013
fields[11] == 'y', # executable
2014
fields[12], # packed_stat or revision_id
2017
return fields_to_entry_1_parent
2018
elif num_present_parents == 2:
2019
def fields_to_entry_2_parents(fields, _int=int):
2020
path_name_file_id_key = (fields[0], fields[1], fields[2])
2021
return (path_name_file_id_key, [
2023
fields[3], # minikind
2024
fields[4], # fingerprint
2025
_int(fields[5]), # size
2026
fields[6] == 'y', # executable
2027
fields[7], # packed_stat or revision_id
2030
fields[8], # minikind
2031
fields[9], # fingerprint
2032
_int(fields[10]), # size
2033
fields[11] == 'y', # executable
2034
fields[12], # packed_stat or revision_id
2037
fields[13], # minikind
2038
fields[14], # fingerprint
2039
_int(fields[15]), # size
2040
fields[16] == 'y', # executable
2041
fields[17], # packed_stat or revision_id
2044
return fields_to_entry_2_parents
2046
def fields_to_entry_n_parents(fields, _int=int):
2047
path_name_file_id_key = (fields[0], fields[1], fields[2])
2048
trees = [(fields[cur], # minikind
2049
fields[cur+1], # fingerprint
2050
_int(fields[cur+2]), # size
2051
fields[cur+3] == 'y', # executable
2052
fields[cur+4], # stat or revision_id
2053
) for cur in range(3, len(fields)-1, 5)]
2054
return path_name_file_id_key, trees
2055
return fields_to_entry_n_parents
2057
def get_parent_ids(self):
2058
"""Return a list of the parent tree ids for the directory state."""
2059
self._read_header_if_needed()
2060
return list(self._parents)
2062
def _get_block_entry_index(self, dirname, basename, tree_index):
2063
"""Get the coordinates for a path in the state structure.
2065
:param dirname: The utf8 dirname to lookup.
2066
:param basename: The utf8 basename to lookup.
2067
:param tree_index: The index of the tree for which this lookup should
2069
:return: A tuple describing where the path is located, or should be
2070
inserted. The tuple contains four fields: the block index, the row
2071
index, the directory is present (boolean), the entire path is
2072
present (boolean). There is no guarantee that either
2073
coordinate is currently reachable unless the found field for it is
2074
True. For instance, a directory not present in the searched tree
2075
may be returned with a value one greater than the current highest
2076
block offset. The directory present field will always be True when
2077
the path present field is True. The directory present field does
2078
NOT indicate that the directory is present in the searched tree,
2079
rather it indicates that there are at least some files in some
2082
self._read_dirblocks_if_needed()
2083
key = dirname, basename, ''
2084
block_index, present = self._find_block_index_from_key(key)
2086
# no such directory - return the dir index and 0 for the row.
2087
return block_index, 0, False, False
2088
block = self._dirblocks[block_index][1] # access the entries only
2089
entry_index, present = self._find_entry_index(key, block)
2090
# linear search through entries at this path to find the one
2092
while entry_index < len(block) and block[entry_index][0][1] == basename:
2093
if block[entry_index][1][tree_index][0] not in 'ar':
2094
# neither absent or relocated
2095
return block_index, entry_index, True, True
2097
return block_index, entry_index, True, False
2099
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None,
2100
include_deleted=False):
2101
"""Get the dirstate entry for path in tree tree_index.
2103
If either file_id or path is supplied, it is used as the key to lookup.
2104
If both are supplied, the fastest lookup is used, and an error is
2105
raised if they do not both point at the same row.
2107
:param tree_index: The index of the tree we wish to locate this path
2108
in. If the path is present in that tree, the entry containing its
2109
details is returned, otherwise (None, None) is returned
2110
0 is the working tree, higher indexes are successive parent
2112
:param fileid_utf8: A utf8 file_id to look up.
2113
:param path_utf8: An utf8 path to be looked up.
2114
:param include_deleted: If True, and performing a lookup via
2115
fileid_utf8 rather than path_utf8, return an entry for deleted
2117
:return: The dirstate entry tuple for path, or (None, None)
2119
self._read_dirblocks_if_needed()
2120
if path_utf8 is not None:
2121
if not isinstance(path_utf8, str):
2122
raise errors.BzrError('path_utf8 is not a str: %s %r'
2123
% (type(path_utf8), path_utf8))
2124
# path lookups are faster
2125
dirname, basename = osutils.split(path_utf8)
2126
block_index, entry_index, dir_present, file_present = \
2127
self._get_block_entry_index(dirname, basename, tree_index)
2128
if not file_present:
2130
entry = self._dirblocks[block_index][1][entry_index]
2131
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
2132
raise AssertionError('unversioned entry?')
2134
if entry[0][2] != fileid_utf8:
2135
self._changes_aborted = True
2136
raise errors.BzrError('integrity error ? : mismatching'
2137
' tree_index, file_id and path')
2140
possible_keys = self._get_id_index().get(fileid_utf8, ())
2141
if not possible_keys:
2143
for key in possible_keys:
2144
block_index, present = \
2145
self._find_block_index_from_key(key)
2146
# strange, probably indicates an out of date
2147
# id index - for now, allow this.
2150
# WARNING: DO not change this code to use _get_block_entry_index
2151
# as that function is not suitable: it does not use the key
2152
# to lookup, and thus the wrong coordinates are returned.
2153
block = self._dirblocks[block_index][1]
2154
entry_index, present = self._find_entry_index(key, block)
2156
entry = self._dirblocks[block_index][1][entry_index]
2157
# TODO: We might want to assert that entry[0][2] ==
2159
if entry[1][tree_index][0] in 'fdlt':
2160
# this is the result we are looking for: the
2161
# real home of this file_id in this tree.
2163
if entry[1][tree_index][0] == 'a':
2164
# there is no home for this entry in this tree
2168
if entry[1][tree_index][0] != 'r':
2169
raise AssertionError(
2170
"entry %r has invalid minikind %r for tree %r" \
2172
entry[1][tree_index][0],
2174
real_path = entry[1][tree_index][1]
2175
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
2176
path_utf8=real_path)
2180
def initialize(cls, path, sha1_provider=None):
2181
"""Create a new dirstate on path.
2183
The new dirstate will be an empty tree - that is it has no parents,
2184
and only a root node - which has id ROOT_ID.
2186
:param path: The name of the file for the dirstate.
2187
:param sha1_provider: an object meeting the SHA1Provider interface.
2188
If None, a DefaultSHA1Provider is used.
2189
:return: A write-locked DirState object.
2191
# This constructs a new DirState object on a path, sets the _state_file
2192
# to a new empty file for that path. It then calls _set_data() with our
2193
# stock empty dirstate information - a root with ROOT_ID, no children,
2194
# and no parents. Finally it calls save() to ensure that this data will
2196
if sha1_provider is None:
2197
sha1_provider = DefaultSHA1Provider()
2198
result = cls(path, sha1_provider)
2199
# root dir and root dir contents with no children.
2200
empty_tree_dirblocks = [('', []), ('', [])]
2201
# a new root directory, with a NULLSTAT.
2202
empty_tree_dirblocks[0][1].append(
2203
(('', '', inventory.ROOT_ID), [
2204
('d', '', 0, False, DirState.NULLSTAT),
2208
result._set_data([], empty_tree_dirblocks)
2216
def _inv_entry_to_details(inv_entry):
2217
"""Convert an inventory entry (from a revision tree) to state details.
2219
:param inv_entry: An inventory entry whose sha1 and link targets can be
2220
relied upon, and which has a revision set.
2221
:return: A details tuple - the details for a single tree at a path +
2224
kind = inv_entry.kind
2225
minikind = DirState._kind_to_minikind[kind]
2226
tree_data = inv_entry.revision
2227
if kind == 'directory':
2231
elif kind == 'symlink':
2232
if inv_entry.symlink_target is None:
2235
fingerprint = inv_entry.symlink_target.encode('utf8')
2238
elif kind == 'file':
2239
fingerprint = inv_entry.text_sha1 or ''
2240
size = inv_entry.text_size or 0
2241
executable = inv_entry.executable
2242
elif kind == 'tree-reference':
2243
fingerprint = inv_entry.reference_revision or ''
2247
raise Exception("can't pack %s" % inv_entry)
2248
return static_tuple.StaticTuple(minikind, fingerprint, size,
2249
executable, tree_data)
2251
def _iter_child_entries(self, tree_index, path_utf8):
2252
"""Iterate over all the entries that are children of path_utf.
2254
This only returns entries that are present (not in 'a', 'r') in
2255
tree_index. tree_index data is not refreshed, so if tree 0 is used,
2256
results may differ from that obtained if paths were statted to
2257
determine what ones were directories.
2259
Asking for the children of a non-directory will return an empty
2263
next_pending_dirs = [path_utf8]
2265
while next_pending_dirs:
2266
pending_dirs = next_pending_dirs
2267
next_pending_dirs = []
2268
for path in pending_dirs:
2269
block_index, present = self._find_block_index_from_key(
2271
if block_index == 0:
2273
if len(self._dirblocks) == 1:
2274
# asked for the children of the root with no other
2278
# children of a non-directory asked for.
2280
block = self._dirblocks[block_index]
2281
for entry in block[1]:
2282
kind = entry[1][tree_index][0]
2283
if kind not in absent:
2287
path = entry[0][0] + '/' + entry[0][1]
2290
next_pending_dirs.append(path)
2292
def _iter_entries(self):
2293
"""Iterate over all the entries in the dirstate.
2295
Each yelt item is an entry in the standard format described in the
2296
docstring of breezy.dirstate.
2298
self._read_dirblocks_if_needed()
2299
for directory in self._dirblocks:
2300
for entry in directory[1]:
2303
def _get_id_index(self):
2304
"""Get an id index of self._dirblocks.
2306
This maps from file_id => [(directory, name, file_id)] entries where
2307
that file_id appears in one of the trees.
2309
if self._id_index is None:
2311
for key, tree_details in self._iter_entries():
2312
self._add_to_id_index(id_index, key)
2313
self._id_index = id_index
2314
return self._id_index
2316
def _add_to_id_index(self, id_index, entry_key):
2317
"""Add this entry to the _id_index mapping."""
2318
# This code used to use a set for every entry in the id_index. However,
2319
# it is *rare* to have more than one entry. So a set is a large
2320
# overkill. And even when we do, we won't ever have more than the
2321
# number of parent trees. Which is still a small number (rarely >2). As
2322
# such, we use a simple tuple, and do our own uniqueness checks. While
2323
# the 'in' check is O(N) since N is nicely bounded it shouldn't ever
2324
# cause quadratic failure.
2325
file_id = entry_key[2]
2326
entry_key = static_tuple.StaticTuple.from_sequence(entry_key)
2327
if file_id not in id_index:
2328
id_index[file_id] = static_tuple.StaticTuple(entry_key,)
2330
entry_keys = id_index[file_id]
2331
if entry_key not in entry_keys:
2332
id_index[file_id] = entry_keys + (entry_key,)
2334
def _remove_from_id_index(self, id_index, entry_key):
2335
"""Remove this entry from the _id_index mapping.
2337
It is an programming error to call this when the entry_key is not
2340
file_id = entry_key[2]
2341
entry_keys = list(id_index[file_id])
2342
entry_keys.remove(entry_key)
2343
id_index[file_id] = static_tuple.StaticTuple.from_sequence(entry_keys)
2345
def _get_output_lines(self, lines):
2346
"""Format lines for final output.
2348
:param lines: A sequence of lines containing the parents list and the
2351
output_lines = [DirState.HEADER_FORMAT_3]
2352
lines.append('') # a final newline
2353
inventory_text = '\0\n\0'.join(lines)
2354
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
2355
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
2356
num_entries = len(lines)-3
2357
output_lines.append('num_entries: %s\n' % (num_entries,))
2358
output_lines.append(inventory_text)
2361
def _make_deleted_row(self, fileid_utf8, parents):
2362
"""Return a deleted row for fileid_utf8."""
2363
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
2366
def _num_present_parents(self):
2367
"""The number of parent entries in each record row."""
2368
return len(self._parents) - len(self._ghosts)
2371
def on_file(cls, path, sha1_provider=None, worth_saving_limit=0):
2372
"""Construct a DirState on the file at path "path".
2374
:param path: The path at which the dirstate file on disk should live.
2375
:param sha1_provider: an object meeting the SHA1Provider interface.
2376
If None, a DefaultSHA1Provider is used.
2377
:param worth_saving_limit: when the exact number of hash changed
2378
entries is known, only bother saving the dirstate if more than
2379
this count of entries have changed. -1 means never save.
2380
:return: An unlocked DirState object, associated with the given path.
2382
if sha1_provider is None:
2383
sha1_provider = DefaultSHA1Provider()
2384
result = cls(path, sha1_provider,
2385
worth_saving_limit=worth_saving_limit)
2388
def _read_dirblocks_if_needed(self):
2389
"""Read in all the dirblocks from the file if they are not in memory.
2391
This populates self._dirblocks, and sets self._dirblock_state to
2392
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
2395
self._read_header_if_needed()
2396
if self._dirblock_state == DirState.NOT_IN_MEMORY:
2397
_read_dirblocks(self)
2399
def _read_header(self):
2400
"""This reads in the metadata header, and the parent ids.
2402
After reading in, the file should be positioned at the null
2403
just before the start of the first record in the file.
2405
:return: (expected crc checksum, number of entries, parent list)
2407
self._read_prelude()
2408
parent_line = self._state_file.readline()
2409
info = parent_line.split('\0')
2410
num_parents = int(info[0])
2411
self._parents = info[1:-1]
2412
ghost_line = self._state_file.readline()
2413
info = ghost_line.split('\0')
2414
num_ghosts = int(info[1])
2415
self._ghosts = info[2:-1]
2416
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2417
self._end_of_header = self._state_file.tell()
2419
def _read_header_if_needed(self):
2420
"""Read the header of the dirstate file if needed."""
2421
# inline this as it will be called a lot
2422
if not self._lock_token:
2423
raise errors.ObjectNotLocked(self)
2424
if self._header_state == DirState.NOT_IN_MEMORY:
2427
def _read_prelude(self):
2428
"""Read in the prelude header of the dirstate file.
2430
This only reads in the stuff that is not connected to the crc
2431
checksum. The position will be correct to read in the rest of
2432
the file and check the checksum after this point.
2433
The next entry in the file should be the number of parents,
2434
and their ids. Followed by a newline.
2436
header = self._state_file.readline()
2437
if header != DirState.HEADER_FORMAT_3:
2438
raise errors.BzrError(
2439
'invalid header line: %r' % (header,))
2440
crc_line = self._state_file.readline()
2441
if not crc_line.startswith('crc32: '):
2442
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2443
self.crc_expected = int(crc_line[len('crc32: '):-1])
2444
num_entries_line = self._state_file.readline()
2445
if not num_entries_line.startswith('num_entries: '):
2446
raise errors.BzrError('missing num_entries line')
2447
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2449
def sha1_from_stat(self, path, stat_result):
2450
"""Find a sha1 given a stat lookup."""
2451
return self._get_packed_stat_index().get(pack_stat(stat_result), None)
2453
def _get_packed_stat_index(self):
2454
"""Get a packed_stat index of self._dirblocks."""
2455
if self._packed_stat_index is None:
2457
for key, tree_details in self._iter_entries():
2458
if tree_details[0][0] == 'f':
2459
index[tree_details[0][4]] = tree_details[0][1]
2460
self._packed_stat_index = index
2461
return self._packed_stat_index
2464
"""Save any pending changes created during this session.
2466
We reuse the existing file, because that prevents race conditions with
2467
file creation, and use oslocks on it to prevent concurrent modification
2468
and reads - because dirstate's incremental data aggregation is not
2469
compatible with reading a modified file, and replacing a file in use by
2470
another process is impossible on Windows.
2472
A dirstate in read only mode should be smart enough though to validate
2473
that the file has not changed, and otherwise discard its cache and
2474
start over, to allow for fine grained read lock duration, so 'status'
2475
wont block 'commit' - for example.
2477
if self._changes_aborted:
2478
# Should this be a warning? For now, I'm expecting that places that
2479
# mark it inconsistent will warn, making a warning here redundant.
2480
trace.mutter('Not saving DirState because '
2481
'_changes_aborted is set.')
2483
# TODO: Since we now distinguish IN_MEMORY_MODIFIED from
2484
# IN_MEMORY_HASH_MODIFIED, we should only fail quietly if we fail
2485
# to save an IN_MEMORY_HASH_MODIFIED, and fail *noisily* if we
2486
# fail to save IN_MEMORY_MODIFIED
2487
if not self._worth_saving():
2490
grabbed_write_lock = False
2491
if self._lock_state != 'w':
2492
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2493
# Switch over to the new lock, as the old one may be closed.
2494
# TODO: jam 20070315 We should validate the disk file has
2495
# not changed contents, since temporary_write_lock may
2496
# not be an atomic operation.
2497
self._lock_token = new_lock
2498
self._state_file = new_lock.f
2499
if not grabbed_write_lock:
2500
# We couldn't grab a write lock, so we switch back to a read one
2503
lines = self.get_lines()
2504
self._state_file.seek(0)
2505
self._state_file.writelines(lines)
2506
self._state_file.truncate()
2507
self._state_file.flush()
2508
self._maybe_fdatasync()
2509
self._mark_unmodified()
2511
if grabbed_write_lock:
2512
self._lock_token = self._lock_token.restore_read_lock()
2513
self._state_file = self._lock_token.f
2514
# TODO: jam 20070315 We should validate the disk file has
2515
# not changed contents. Since restore_read_lock may
2516
# not be an atomic operation.
2518
def _maybe_fdatasync(self):
2519
"""Flush to disk if possible and if not configured off."""
2520
if self._config_stack.get('dirstate.fdatasync'):
2521
osutils.fdatasync(self._state_file.fileno())
2523
def _worth_saving(self):
2524
"""Is it worth saving the dirstate or not?"""
2525
if (self._header_state == DirState.IN_MEMORY_MODIFIED
2526
or self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2528
if self._dirblock_state == DirState.IN_MEMORY_HASH_MODIFIED:
2529
if self._worth_saving_limit == -1:
2530
# We never save hash changes when the limit is -1
2532
# If we're using smart saving and only a small number of
2533
# entries have changed their hash, don't bother saving. John has
2534
# suggested using a heuristic here based on the size of the
2535
# changed files and/or tree. For now, we go with a configurable
2536
# number of changes, keeping the calculation time
2537
# as low overhead as possible. (This also keeps all existing
2538
# tests passing as the default is 0, i.e. always save.)
2539
if len(self._known_hash_changes) >= self._worth_saving_limit:
2543
def _set_data(self, parent_ids, dirblocks):
2544
"""Set the full dirstate data in memory.
2546
This is an internal function used to completely replace the objects
2547
in memory state. It puts the dirstate into state 'full-dirty'.
2549
:param parent_ids: A list of parent tree revision ids.
2550
:param dirblocks: A list containing one tuple for each directory in the
2551
tree. Each tuple contains the directory path and a list of entries
2552
found in that directory.
2554
# our memory copy is now authoritative.
2555
self._dirblocks = dirblocks
2556
self._mark_modified(header_modified=True)
2557
self._parents = list(parent_ids)
2558
self._id_index = None
2559
self._packed_stat_index = None
2561
def set_path_id(self, path, new_id):
2562
"""Change the id of path to new_id in the current working tree.
2564
:param path: The path inside the tree to set - '' is the root, 'foo'
2565
is the path foo in the root.
2566
:param new_id: The new id to assign to the path. This must be a utf8
2567
file id (not unicode, and not None).
2569
self._read_dirblocks_if_needed()
2571
# TODO: logic not written
2572
raise NotImplementedError(self.set_path_id)
2573
# TODO: check new id is unique
2574
entry = self._get_entry(0, path_utf8=path)
2575
if entry[0][2] == new_id:
2576
# Nothing to change.
2578
# mark the old path absent, and insert a new root path
2579
self._make_absent(entry)
2580
self.update_minimal(('', '', new_id), 'd',
2581
path_utf8='', packed_stat=entry[1][0][4])
2582
self._mark_modified()
2584
def set_parent_trees(self, trees, ghosts):
2585
"""Set the parent trees for the dirstate.
2587
:param trees: A list of revision_id, tree tuples. tree must be provided
2588
even if the revision_id refers to a ghost: supply an empty tree in
2590
:param ghosts: A list of the revision_ids that are ghosts at the time
2593
# TODO: generate a list of parent indexes to preserve to save
2594
# processing specific parent trees. In the common case one tree will
2595
# be preserved - the left most parent.
2596
# TODO: if the parent tree is a dirstate, we might want to walk them
2597
# all by path in parallel for 'optimal' common-case performance.
2598
# generate new root row.
2599
self._read_dirblocks_if_needed()
2600
# TODO future sketch: Examine the existing parents to generate a change
2601
# map and then walk the new parent trees only, mapping them into the
2602
# dirstate. Walk the dirstate at the same time to remove unreferenced
2605
# sketch: loop over all entries in the dirstate, cherry picking
2606
# entries from the parent trees, if they are not ghost trees.
2607
# after we finish walking the dirstate, all entries not in the dirstate
2608
# are deletes, so we want to append them to the end as per the design
2609
# discussions. So do a set difference on ids with the parents to
2610
# get deletes, and add them to the end.
2611
# During the update process we need to answer the following questions:
2612
# - find other keys containing a fileid in order to create cross-path
2613
# links. We dont't trivially use the inventory from other trees
2614
# because this leads to either double touching, or to accessing
2616
# - find other keys containing a path
2617
# We accumulate each entry via this dictionary, including the root
2620
# we could do parallel iterators, but because file id data may be
2621
# scattered throughout, we dont save on index overhead: we have to look
2622
# at everything anyway. We can probably save cycles by reusing parent
2623
# data and doing an incremental update when adding an additional
2624
# parent, but for now the common cases are adding a new parent (merge),
2625
# and replacing completely (commit), and commit is more common: so
2626
# optimise merge later.
2628
# ---- start generation of full tree mapping data
2629
# what trees should we use?
2630
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2631
# how many trees do we end up with
2632
parent_count = len(parent_trees)
2633
st = static_tuple.StaticTuple
2635
# one: the current tree
2636
for entry in self._iter_entries():
2637
# skip entries not in the current tree
2638
if entry[1][0][0] in 'ar': # absent, relocated
2640
by_path[entry[0]] = [entry[1][0]] + \
2641
[DirState.NULL_PARENT_DETAILS] * parent_count
2642
# TODO: Possibly inline this, since we know it isn't present yet
2643
# id_index[entry[0][2]] = (entry[0],)
2644
self._add_to_id_index(id_index, entry[0])
2646
# now the parent trees:
2647
for tree_index, tree in enumerate(parent_trees):
2648
# the index is off by one, adjust it.
2649
tree_index = tree_index + 1
2650
# when we add new locations for a fileid we need these ranges for
2651
# any fileid in this tree as we set the by_path[id] to:
2652
# already_processed_tree_details + new_details + new_location_suffix
2653
# the suffix is from tree_index+1:parent_count+1.
2654
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2655
# now stitch in all the entries from this tree
2657
for path, entry in tree.iter_entries_by_dir():
2658
# here we process each trees details for each item in the tree.
2659
# we first update any existing entries for the id at other paths,
2660
# then we either create or update the entry for the id at the
2661
# right path, and finally we add (if needed) a mapping from
2662
# file_id to this path. We do it in this order to allow us to
2663
# avoid checking all known paths for the id when generating a
2664
# new entry at this path: by adding the id->path mapping last,
2665
# all the mappings are valid and have correct relocation
2666
# records where needed.
2667
file_id = entry.file_id
2668
path_utf8 = path.encode('utf8')
2669
dirname, basename = osutils.split(path_utf8)
2670
if dirname == last_dirname:
2671
# Try to re-use objects as much as possible
2672
dirname = last_dirname
2674
last_dirname = dirname
2675
new_entry_key = st(dirname, basename, file_id)
2676
# tree index consistency: All other paths for this id in this tree
2677
# index must point to the correct path.
2678
entry_keys = id_index.get(file_id, ())
2679
for entry_key in entry_keys:
2680
# TODO:PROFILING: It might be faster to just update
2681
# rather than checking if we need to, and then overwrite
2682
# the one we are located at.
2683
if entry_key != new_entry_key:
2684
# this file id is at a different path in one of the
2685
# other trees, so put absent pointers there
2686
# This is the vertical axis in the matrix, all pointing
2688
by_path[entry_key][tree_index] = st('r', path_utf8, 0,
2690
# by path consistency: Insert into an existing path record
2691
# (trivial), or add a new one with relocation pointers for the
2692
# other tree indexes.
2693
if new_entry_key in entry_keys:
2694
# there is already an entry where this data belongs, just
2696
by_path[new_entry_key][tree_index] = \
2697
self._inv_entry_to_details(entry)
2699
# add relocated entries to the horizontal axis - this row
2700
# mapping from path,id. We need to look up the correct path
2701
# for the indexes from 0 to tree_index -1
2703
for lookup_index in range(tree_index):
2704
# boundary case: this is the first occurence of file_id
2705
# so there are no id_indexes, possibly take this out of
2707
if not len(entry_keys):
2708
new_details.append(DirState.NULL_PARENT_DETAILS)
2710
# grab any one entry, use it to find the right path.
2711
a_key = next(iter(entry_keys))
2712
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2713
# its a pointer or missing statement, use it as
2715
new_details.append(by_path[a_key][lookup_index])
2717
# we have the right key, make a pointer to it.
2718
real_path = ('/'.join(a_key[0:2])).strip('/')
2719
new_details.append(st('r', real_path, 0, False,
2721
new_details.append(self._inv_entry_to_details(entry))
2722
new_details.extend(new_location_suffix)
2723
by_path[new_entry_key] = new_details
2724
self._add_to_id_index(id_index, new_entry_key)
2725
# --- end generation of full tree mappings
2727
# sort and output all the entries
2728
new_entries = self._sort_entries(viewitems(by_path))
2729
self._entries_to_current_state(new_entries)
2730
self._parents = [rev_id for rev_id, tree in trees]
2731
self._ghosts = list(ghosts)
2732
self._mark_modified(header_modified=True)
2733
self._id_index = id_index
2735
def _sort_entries(self, entry_list):
2736
"""Given a list of entries, sort them into the right order.
2738
This is done when constructing a new dirstate from trees - normally we
2739
try to keep everything in sorted blocks all the time, but sometimes
2740
it's easier to sort after the fact.
2742
# When sorting, we usually have 10x more entries than directories. (69k
2743
# total entries, 4k directories). So cache the results of splitting.
2744
# Saving time and objects. Also, use StaticTuple to avoid putting all
2745
# of these object into python's garbage collector.
2747
def _key(entry, _split_dirs=split_dirs, _st=static_tuple.StaticTuple):
2748
# sort by: directory parts, file name, file id
2749
dirpath, fname, file_id = entry[0]
2751
split = _split_dirs[dirpath]
2753
split = _st.from_sequence(dirpath.split('/'))
2754
_split_dirs[dirpath] = split
2755
return _st(split, fname, file_id)
2756
return sorted(entry_list, key=_key)
2758
def set_state_from_inventory(self, new_inv):
2759
"""Set new_inv as the current state.
2761
This API is called by tree transform, and will usually occur with
2762
existing parent trees.
2764
:param new_inv: The inventory object to set current state from.
2766
if 'evil' in debug.debug_flags:
2767
trace.mutter_callsite(1,
2768
"set_state_from_inventory called; please mutate the tree instead")
2769
tracing = 'dirstate' in debug.debug_flags
2771
trace.mutter("set_state_from_inventory trace:")
2772
self._read_dirblocks_if_needed()
2774
# Two iterators: current data and new data, both in dirblock order.
2775
# We zip them together, which tells about entries that are new in the
2776
# inventory, or removed in the inventory, or present in both and
2779
# You might think we could just synthesize a new dirstate directly
2780
# since we're processing it in the right order. However, we need to
2781
# also consider there may be any number of parent trees and relocation
2782
# pointers, and we don't want to duplicate that here.
2783
new_iterator = new_inv.iter_entries_by_dir()
2784
# we will be modifying the dirstate, so we need a stable iterator. In
2785
# future we might write one, for now we just clone the state into a
2786
# list using a copy so that we see every original item and don't have
2787
# to adjust the position when items are inserted or deleted in the
2788
# underlying dirstate.
2789
old_iterator = iter(list(self._iter_entries()))
2790
# both must have roots so this is safe:
2791
current_new = next(new_iterator)
2792
current_old = next(old_iterator)
2793
def advance(iterator):
2795
return next(iterator)
2796
except StopIteration:
2798
while current_new or current_old:
2799
# skip entries in old that are not really there
2800
if current_old and current_old[1][0][0] in 'ar':
2801
# relocated or absent
2802
current_old = advance(old_iterator)
2805
# convert new into dirblock style
2806
new_path_utf8 = current_new[0].encode('utf8')
2807
new_dirname, new_basename = osutils.split(new_path_utf8)
2808
new_id = current_new[1].file_id
2809
new_entry_key = (new_dirname, new_basename, new_id)
2810
current_new_minikind = \
2811
DirState._kind_to_minikind[current_new[1].kind]
2812
if current_new_minikind == 't':
2813
fingerprint = current_new[1].reference_revision or ''
2815
# We normally only insert or remove records, or update
2816
# them when it has significantly changed. Then we want to
2817
# erase its fingerprint. Unaffected records should
2818
# normally not be updated at all.
2821
# for safety disable variables
2822
new_path_utf8 = new_dirname = new_basename = new_id = \
2823
new_entry_key = None
2824
# 5 cases, we dont have a value that is strictly greater than everything, so
2825
# we make both end conditions explicit
2827
# old is finished: insert current_new into the state.
2829
trace.mutter("Appending from new '%s'.",
2830
new_path_utf8.decode('utf8'))
2831
self.update_minimal(new_entry_key, current_new_minikind,
2832
executable=current_new[1].executable,
2833
path_utf8=new_path_utf8, fingerprint=fingerprint,
2835
current_new = advance(new_iterator)
2836
elif not current_new:
2839
trace.mutter("Truncating from old '%s/%s'.",
2840
current_old[0][0].decode('utf8'),
2841
current_old[0][1].decode('utf8'))
2842
self._make_absent(current_old)
2843
current_old = advance(old_iterator)
2844
elif new_entry_key == current_old[0]:
2845
# same - common case
2846
# We're looking at the same path and id in both the dirstate
2847
# and inventory, so just need to update the fields in the
2848
# dirstate from the one in the inventory.
2849
# TODO: update the record if anything significant has changed.
2850
# the minimal required trigger is if the execute bit or cached
2852
if (current_old[1][0][3] != current_new[1].executable or
2853
current_old[1][0][0] != current_new_minikind):
2855
trace.mutter("Updating in-place change '%s'.",
2856
new_path_utf8.decode('utf8'))
2857
self.update_minimal(current_old[0], current_new_minikind,
2858
executable=current_new[1].executable,
2859
path_utf8=new_path_utf8, fingerprint=fingerprint,
2861
# both sides are dealt with, move on
2862
current_old = advance(old_iterator)
2863
current_new = advance(new_iterator)
2864
elif (lt_by_dirs(new_dirname, current_old[0][0])
2865
or (new_dirname == current_old[0][0]
2866
and new_entry_key[1:] < current_old[0][1:])):
2868
# add a entry for this and advance new
2870
trace.mutter("Inserting from new '%s'.",
2871
new_path_utf8.decode('utf8'))
2872
self.update_minimal(new_entry_key, current_new_minikind,
2873
executable=current_new[1].executable,
2874
path_utf8=new_path_utf8, fingerprint=fingerprint,
2876
current_new = advance(new_iterator)
2878
# we've advanced past the place where the old key would be,
2879
# without seeing it in the new list. so it must be gone.
2881
trace.mutter("Deleting from old '%s/%s'.",
2882
current_old[0][0].decode('utf8'),
2883
current_old[0][1].decode('utf8'))
2884
self._make_absent(current_old)
2885
current_old = advance(old_iterator)
2886
self._mark_modified()
2887
self._id_index = None
2888
self._packed_stat_index = None
2890
trace.mutter("set_state_from_inventory complete.")
2892
def set_state_from_scratch(self, working_inv, parent_trees, parent_ghosts):
2893
"""Wipe the currently stored state and set it to something new.
2895
This is a hard-reset for the data we are working with.
2897
# Technically, we really want a write lock, but until we write, we
2898
# don't really need it.
2899
self._requires_lock()
2900
# root dir and root dir contents with no children. We have to have a
2901
# root for set_state_from_inventory to work correctly.
2902
empty_root = (('', '', inventory.ROOT_ID),
2903
[('d', '', 0, False, DirState.NULLSTAT)])
2904
empty_tree_dirblocks = [('', [empty_root]), ('', [])]
2905
self._set_data([], empty_tree_dirblocks)
2906
self.set_state_from_inventory(working_inv)
2907
self.set_parent_trees(parent_trees, parent_ghosts)
2909
def _make_absent(self, current_old):
2910
"""Mark current_old - an entry - as absent for tree 0.
2912
:return: True if this was the last details entry for the entry key:
2913
that is, if the underlying block has had the entry removed, thus
2914
shrinking in length.
2916
# build up paths that this id will be left at after the change is made,
2917
# so we can update their cross references in tree 0
2918
all_remaining_keys = set()
2919
# Dont check the working tree, because it's going.
2920
for details in current_old[1][1:]:
2921
if details[0] not in 'ar': # absent, relocated
2922
all_remaining_keys.add(current_old[0])
2923
elif details[0] == 'r': # relocated
2924
# record the key for the real path.
2925
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2926
# absent rows are not present at any path.
2927
last_reference = current_old[0] not in all_remaining_keys
2929
# the current row consists entire of the current item (being marked
2930
# absent), and relocated or absent entries for the other trees:
2931
# Remove it, its meaningless.
2932
block = self._find_block(current_old[0])
2933
entry_index, present = self._find_entry_index(current_old[0], block[1])
2935
raise AssertionError('could not find entry for %s' % (current_old,))
2936
block[1].pop(entry_index)
2937
# if we have an id_index in use, remove this key from it for this id.
2938
if self._id_index is not None:
2939
self._remove_from_id_index(self._id_index, current_old[0])
2940
# update all remaining keys for this id to record it as absent. The
2941
# existing details may either be the record we are marking as deleted
2942
# (if there were other trees with the id present at this path), or may
2944
for update_key in all_remaining_keys:
2945
update_block_index, present = \
2946
self._find_block_index_from_key(update_key)
2948
raise AssertionError('could not find block for %s' % (update_key,))
2949
update_entry_index, present = \
2950
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2952
raise AssertionError('could not find entry for %s' % (update_key,))
2953
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2954
# it must not be absent at the moment
2955
if update_tree_details[0][0] == 'a': # absent
2956
raise AssertionError('bad row %r' % (update_tree_details,))
2957
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2958
self._mark_modified()
2959
return last_reference
2961
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2962
packed_stat=None, size=0, path_utf8=None, fullscan=False):
2963
"""Update an entry to the state in tree 0.
2965
This will either create a new entry at 'key' or update an existing one.
2966
It also makes sure that any other records which might mention this are
2969
:param key: (dir, name, file_id) for the new entry
2970
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2972
:param executable: Should the executable bit be set?
2973
:param fingerprint: Simple fingerprint for new entry: canonical-form
2974
sha1 for files, referenced revision id for subtrees, etc.
2975
:param packed_stat: Packed stat value for new entry.
2976
:param size: Size information for new entry
2977
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2979
:param fullscan: If True then a complete scan of the dirstate is being
2980
done and checking for duplicate rows should not be done. This
2981
should only be set by set_state_from_inventory and similar methods.
2983
If packed_stat and fingerprint are not given, they're invalidated in
2986
block = self._find_block(key)[1]
2987
if packed_stat is None:
2988
packed_stat = DirState.NULLSTAT
2989
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2990
# sometimes present in the dirstate - this seems oddly inconsistent.
2992
entry_index, present = self._find_entry_index(key, block)
2993
new_details = (minikind, fingerprint, size, executable, packed_stat)
2994
id_index = self._get_id_index()
2996
# New record. Check there isn't a entry at this path already.
2998
low_index, _ = self._find_entry_index(key[0:2] + ('',), block)
2999
while low_index < len(block):
3000
entry = block[low_index]
3001
if entry[0][0:2] == key[0:2]:
3002
if entry[1][0][0] not in 'ar':
3003
# This entry has the same path (but a different id) as
3004
# the new entry we're adding, and is present in ths
3006
self._raise_invalid(
3007
("%s/%s" % key[0:2]).decode('utf8'), key[2],
3008
"Attempt to add item at path already occupied by "
3009
"id %r" % entry[0][2])
3013
# new entry, synthesis cross reference here,
3014
existing_keys = id_index.get(key[2], ())
3015
if not existing_keys:
3016
# not currently in the state, simplest case
3017
new_entry = key, [new_details] + self._empty_parent_info()
3019
# present at one or more existing other paths.
3020
# grab one of them and use it to generate parent
3021
# relocation/absent entries.
3022
new_entry = key, [new_details]
3023
# existing_keys can be changed as we iterate.
3024
for other_key in tuple(existing_keys):
3025
# change the record at other to be a pointer to this new
3026
# record. The loop looks similar to the change to
3027
# relocations when updating an existing record but its not:
3028
# the test for existing kinds is different: this can be
3029
# factored out to a helper though.
3030
other_block_index, present = self._find_block_index_from_key(
3033
raise AssertionError('could not find block for %s' % (
3035
other_block = self._dirblocks[other_block_index][1]
3036
other_entry_index, present = self._find_entry_index(
3037
other_key, other_block)
3039
raise AssertionError(
3040
'update_minimal: could not find other entry for %s'
3042
if path_utf8 is None:
3043
raise AssertionError('no path')
3044
# Turn this other location into a reference to the new
3045
# location. This also updates the aliased iterator
3046
# (current_old in set_state_from_inventory) so that the old
3047
# entry, if not already examined, is skipped over by that
3049
other_entry = other_block[other_entry_index]
3050
other_entry[1][0] = ('r', path_utf8, 0, False, '')
3051
if self._maybe_remove_row(other_block, other_entry_index,
3053
# If the row holding this was removed, we need to
3054
# recompute where this entry goes
3055
entry_index, _ = self._find_entry_index(key, block)
3058
# adds a tuple to the new details for each column
3059
# - either by copying an existing relocation pointer inside that column
3060
# - or by creating a new pointer to the right row inside that column
3061
num_present_parents = self._num_present_parents()
3062
if num_present_parents:
3063
# TODO: This re-evaluates the existing_keys set, do we need
3064
# to do that ourselves?
3065
other_key = list(existing_keys)[0]
3066
for lookup_index in range(1, num_present_parents + 1):
3067
# grab any one entry, use it to find the right path.
3068
# TODO: optimise this to reduce memory use in highly
3069
# fragmented situations by reusing the relocation
3071
update_block_index, present = \
3072
self._find_block_index_from_key(other_key)
3074
raise AssertionError('could not find block for %s' % (other_key,))
3075
update_entry_index, present = \
3076
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
3078
raise AssertionError('update_minimal: could not find entry for %s' % (other_key,))
3079
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
3080
if update_details[0] in 'ar': # relocated, absent
3081
# its a pointer or absent in lookup_index's tree, use
3083
new_entry[1].append(update_details)
3085
# we have the right key, make a pointer to it.
3086
pointer_path = osutils.pathjoin(*other_key[0:2])
3087
new_entry[1].append(('r', pointer_path, 0, False, ''))
3088
block.insert(entry_index, new_entry)
3089
self._add_to_id_index(id_index, key)
3091
# Does the new state matter?
3092
block[entry_index][1][0] = new_details
3093
# parents cannot be affected by what we do.
3094
# other occurences of this id can be found
3095
# from the id index.
3097
# tree index consistency: All other paths for this id in this tree
3098
# index must point to the correct path. We have to loop here because
3099
# we may have passed entries in the state with this file id already
3100
# that were absent - where parent entries are - and they need to be
3101
# converted to relocated.
3102
if path_utf8 is None:
3103
raise AssertionError('no path')
3104
existing_keys = id_index.get(key[2], ())
3105
if key not in existing_keys:
3106
raise AssertionError('We found the entry in the blocks, but'
3107
' the key is not in the id_index.'
3108
' key: %s, existing_keys: %s' % (key, existing_keys))
3109
for entry_key in existing_keys:
3110
# TODO:PROFILING: It might be faster to just update
3111
# rather than checking if we need to, and then overwrite
3112
# the one we are located at.
3113
if entry_key != key:
3114
# this file id is at a different path in one of the
3115
# other trees, so put absent pointers there
3116
# This is the vertical axis in the matrix, all pointing
3118
block_index, present = self._find_block_index_from_key(entry_key)
3120
raise AssertionError('not present: %r', entry_key)
3121
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
3123
raise AssertionError('not present: %r', entry_key)
3124
self._dirblocks[block_index][1][entry_index][1][0] = \
3125
('r', path_utf8, 0, False, '')
3126
# add a containing dirblock if needed.
3127
if new_details[0] == 'd':
3128
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
3129
block_index, present = self._find_block_index_from_key(subdir_key)
3131
self._dirblocks.insert(block_index, (subdir_key[0], []))
3133
self._mark_modified()
3135
def _maybe_remove_row(self, block, index, id_index):
3136
"""Remove index if it is absent or relocated across the row.
3138
id_index is updated accordingly.
3139
:return: True if we removed the row, False otherwise
3141
present_in_row = False
3142
entry = block[index]
3143
for column in entry[1]:
3144
if column[0] not in 'ar':
3145
present_in_row = True
3147
if not present_in_row:
3149
self._remove_from_id_index(id_index, entry[0])
3153
def _validate(self):
3154
"""Check that invariants on the dirblock are correct.
3156
This can be useful in debugging; it shouldn't be necessary in
3159
This must be called with a lock held.
3161
# NOTE: This must always raise AssertionError not just assert,
3162
# otherwise it may not behave properly under python -O
3164
# TODO: All entries must have some content that's not 'a' or 'r',
3165
# otherwise it could just be removed.
3167
# TODO: All relocations must point directly to a real entry.
3169
# TODO: No repeated keys.
3172
from pprint import pformat
3173
self._read_dirblocks_if_needed()
3174
if len(self._dirblocks) > 0:
3175
if not self._dirblocks[0][0] == '':
3176
raise AssertionError(
3177
"dirblocks don't start with root block:\n" + \
3178
pformat(self._dirblocks))
3179
if len(self._dirblocks) > 1:
3180
if not self._dirblocks[1][0] == '':
3181
raise AssertionError(
3182
"dirblocks missing root directory:\n" + \
3183
pformat(self._dirblocks))
3184
# the dirblocks are sorted by their path components, name, and dir id
3185
dir_names = [d[0].split('/')
3186
for d in self._dirblocks[1:]]
3187
if dir_names != sorted(dir_names):
3188
raise AssertionError(
3189
"dir names are not in sorted order:\n" + \
3190
pformat(self._dirblocks) + \
3193
for dirblock in self._dirblocks:
3194
# within each dirblock, the entries are sorted by filename and
3196
for entry in dirblock[1]:
3197
if dirblock[0] != entry[0][0]:
3198
raise AssertionError(
3200
"doesn't match directory name in\n%r" %
3201
(entry, pformat(dirblock)))
3202
if dirblock[1] != sorted(dirblock[1]):
3203
raise AssertionError(
3204
"dirblock for %r is not sorted:\n%s" % \
3205
(dirblock[0], pformat(dirblock)))
3207
def check_valid_parent():
3208
"""Check that the current entry has a valid parent.
3210
This makes sure that the parent has a record,
3211
and that the parent isn't marked as "absent" in the
3212
current tree. (It is invalid to have a non-absent file in an absent
3215
if entry[0][0:2] == ('', ''):
3216
# There should be no parent for the root row
3218
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
3219
if parent_entry == (None, None):
3220
raise AssertionError(
3221
"no parent entry for: %s in tree %s"
3222
% (this_path, tree_index))
3223
if parent_entry[1][tree_index][0] != 'd':
3224
raise AssertionError(
3225
"Parent entry for %s is not marked as a valid"
3226
" directory. %s" % (this_path, parent_entry,))
3228
# For each file id, for each tree: either
3229
# the file id is not present at all; all rows with that id in the
3230
# key have it marked as 'absent'
3231
# OR the file id is present under exactly one name; any other entries
3232
# that mention that id point to the correct name.
3234
# We check this with a dict per tree pointing either to the present
3235
# name, or None if absent.
3236
tree_count = self._num_present_parents() + 1
3237
id_path_maps = [{} for _ in range(tree_count)]
3238
# Make sure that all renamed entries point to the correct location.
3239
for entry in self._iter_entries():
3240
file_id = entry[0][2]
3241
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
3242
if len(entry[1]) != tree_count:
3243
raise AssertionError(
3244
"wrong number of entry details for row\n%s" \
3245
",\nexpected %d" % \
3246
(pformat(entry), tree_count))
3247
absent_positions = 0
3248
for tree_index, tree_state in enumerate(entry[1]):
3249
this_tree_map = id_path_maps[tree_index]
3250
minikind = tree_state[0]
3251
if minikind in 'ar':
3252
absent_positions += 1
3253
# have we seen this id before in this column?
3254
if file_id in this_tree_map:
3255
previous_path, previous_loc = this_tree_map[file_id]
3256
# any later mention of this file must be consistent with
3257
# what was said before
3259
if previous_path is not None:
3260
raise AssertionError(
3261
"file %s is absent in row %r but also present " \
3263
(file_id, entry, previous_path))
3264
elif minikind == 'r':
3265
target_location = tree_state[1]
3266
if previous_path != target_location:
3267
raise AssertionError(
3268
"file %s relocation in row %r but also at %r" \
3269
% (file_id, entry, previous_path))
3271
# a file, directory, etc - may have been previously
3272
# pointed to by a relocation, which must point here
3273
if previous_path != this_path:
3274
raise AssertionError(
3275
"entry %r inconsistent with previous path %r "
3277
(entry, previous_path, previous_loc))
3278
check_valid_parent()
3281
# absent; should not occur anywhere else
3282
this_tree_map[file_id] = None, this_path
3283
elif minikind == 'r':
3284
# relocation, must occur at expected location
3285
this_tree_map[file_id] = tree_state[1], this_path
3287
this_tree_map[file_id] = this_path, this_path
3288
check_valid_parent()
3289
if absent_positions == tree_count:
3290
raise AssertionError(
3291
"entry %r has no data for any tree." % (entry,))
3292
if self._id_index is not None:
3293
for file_id, entry_keys in viewitems(self._id_index):
3294
for entry_key in entry_keys:
3295
# Check that the entry in the map is pointing to the same
3297
if entry_key[2] != file_id:
3298
raise AssertionError(
3299
'file_id %r did not match entry key %s'
3300
% (file_id, entry_key))
3301
# And that from this entry key, we can look up the original
3303
block_index, present = self._find_block_index_from_key(entry_key)
3305
raise AssertionError('missing block for entry key: %r', entry_key)
3306
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
3308
raise AssertionError('missing entry for key: %r', entry_key)
3309
if len(entry_keys) != len(set(entry_keys)):
3310
raise AssertionError(
3311
'id_index contained non-unique data for %s'
3314
def _wipe_state(self):
3315
"""Forget all state information about the dirstate."""
3316
self._header_state = DirState.NOT_IN_MEMORY
3317
self._dirblock_state = DirState.NOT_IN_MEMORY
3318
self._changes_aborted = False
3321
self._dirblocks = []
3322
self._id_index = None
3323
self._packed_stat_index = None
3324
self._end_of_header = None
3325
self._cutoff_time = None
3326
self._split_path_cache = {}
3328
def lock_read(self):
3329
"""Acquire a read lock on the dirstate."""
3330
if self._lock_token is not None:
3331
raise errors.LockContention(self._lock_token)
3332
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3333
# already in memory, we could read just the header and check for
3334
# any modification. If not modified, we can just leave things
3336
self._lock_token = lock.ReadLock(self._filename)
3337
self._lock_state = 'r'
3338
self._state_file = self._lock_token.f
3341
def lock_write(self):
3342
"""Acquire a write lock on the dirstate."""
3343
if self._lock_token is not None:
3344
raise errors.LockContention(self._lock_token)
3345
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3346
# already in memory, we could read just the header and check for
3347
# any modification. If not modified, we can just leave things
3349
self._lock_token = lock.WriteLock(self._filename)
3350
self._lock_state = 'w'
3351
self._state_file = self._lock_token.f
3355
"""Drop any locks held on the dirstate."""
3356
if self._lock_token is None:
3357
raise errors.LockNotHeld(self)
3358
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3359
# already in memory, we could read just the header and check for
3360
# any modification. If not modified, we can just leave things
3362
self._state_file = None
3363
self._lock_state = None
3364
self._lock_token.unlock()
3365
self._lock_token = None
3366
self._split_path_cache = {}
3368
def _requires_lock(self):
3369
"""Check that a lock is currently held by someone on the dirstate."""
3370
if not self._lock_token:
3371
raise errors.ObjectNotLocked(self)
3374
def py_update_entry(state, entry, abspath, stat_value,
3375
_stat_to_minikind=DirState._stat_to_minikind):
3376
"""Update the entry based on what is actually on disk.
3378
This function only calculates the sha if it needs to - if the entry is
3379
uncachable, or clearly different to the first parent's entry, no sha
3380
is calculated, and None is returned.
3382
:param state: The dirstate this entry is in.
3383
:param entry: This is the dirblock entry for the file in question.
3384
:param abspath: The path on disk for this file.
3385
:param stat_value: The stat value done on the path.
3386
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
3387
target of a symlink.
3390
minikind = _stat_to_minikind[stat_value.st_mode & 0o170000]
3394
packed_stat = pack_stat(stat_value)
3395
(saved_minikind, saved_link_or_sha1, saved_file_size,
3396
saved_executable, saved_packed_stat) = entry[1][0]
3398
if minikind == 'd' and saved_minikind == 't':
3400
if (minikind == saved_minikind
3401
and packed_stat == saved_packed_stat):
3402
# The stat hasn't changed since we saved, so we can re-use the
3407
# size should also be in packed_stat
3408
if saved_file_size == stat_value.st_size:
3409
return saved_link_or_sha1
3411
# If we have gotten this far, that means that we need to actually
3412
# process this entry.
3416
executable = state._is_executable(stat_value.st_mode,
3418
if state._cutoff_time is None:
3419
state._sha_cutoff_time()
3420
if (stat_value.st_mtime < state._cutoff_time
3421
and stat_value.st_ctime < state._cutoff_time
3422
and len(entry[1]) > 1
3423
and entry[1][1][0] != 'a'):
3424
# Could check for size changes for further optimised
3425
# avoidance of sha1's. However the most prominent case of
3426
# over-shaing is during initial add, which this catches.
3427
# Besides, if content filtering happens, size and sha
3428
# are calculated at the same time, so checking just the size
3429
# gains nothing w.r.t. performance.
3430
link_or_sha1 = state._sha1_file(abspath)
3431
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
3432
executable, packed_stat)
3434
entry[1][0] = ('f', '', stat_value.st_size,
3435
executable, DirState.NULLSTAT)
3436
worth_saving = False
3437
elif minikind == 'd':
3439
entry[1][0] = ('d', '', 0, False, packed_stat)
3440
if saved_minikind != 'd':
3441
# This changed from something into a directory. Make sure we
3442
# have a directory block for it. This doesn't happen very
3443
# often, so this doesn't have to be super fast.
3444
block_index, entry_index, dir_present, file_present = \
3445
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
3446
state._ensure_block(block_index, entry_index,
3447
osutils.pathjoin(entry[0][0], entry[0][1]))
3449
worth_saving = False
3450
elif minikind == 'l':
3451
if saved_minikind == 'l':
3452
worth_saving = False
3453
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
3454
if state._cutoff_time is None:
3455
state._sha_cutoff_time()
3456
if (stat_value.st_mtime < state._cutoff_time
3457
and stat_value.st_ctime < state._cutoff_time):
3458
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
3461
entry[1][0] = ('l', '', stat_value.st_size,
3462
False, DirState.NULLSTAT)
3464
state._mark_modified([entry])
3468
class ProcessEntryPython(object):
3470
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id",
3471
"last_source_parent", "last_target_parent", "include_unchanged",
3472
"partial", "use_filesystem_for_exec", "utf8_decode",
3473
"searched_specific_files", "search_specific_files",
3474
"searched_exact_paths", "search_specific_file_parents", "seen_ids",
3475
"state", "source_index", "target_index", "want_unversioned", "tree"]
3477
def __init__(self, include_unchanged, use_filesystem_for_exec,
3478
search_specific_files, state, source_index, target_index,
3479
want_unversioned, tree):
3480
self.old_dirname_to_file_id = {}
3481
self.new_dirname_to_file_id = {}
3482
# Are we doing a partial iter_changes?
3483
self.partial = search_specific_files != {''}
3484
# Using a list so that we can access the values and change them in
3485
# nested scope. Each one is [path, file_id, entry]
3486
self.last_source_parent = [None, None]
3487
self.last_target_parent = [None, None]
3488
self.include_unchanged = include_unchanged
3489
self.use_filesystem_for_exec = use_filesystem_for_exec
3490
self.utf8_decode = cache_utf8._utf8_decode
3491
# for all search_indexs in each path at or under each element of
3492
# search_specific_files, if the detail is relocated: add the id, and
3493
# add the relocated path as one to search if its not searched already.
3494
# If the detail is not relocated, add the id.
3495
self.searched_specific_files = set()
3496
# When we search exact paths without expanding downwards, we record
3498
self.searched_exact_paths = set()
3499
self.search_specific_files = search_specific_files
3500
# The parents up to the root of the paths we are searching.
3501
# After all normal paths are returned, these specific items are returned.
3502
self.search_specific_file_parents = set()
3503
# The ids we've sent out in the delta.
3504
self.seen_ids = set()
3506
self.source_index = source_index
3507
self.target_index = target_index
3508
if target_index != 0:
3509
# A lot of code in here depends on target_index == 0
3510
raise errors.BzrError('unsupported target index')
3511
self.want_unversioned = want_unversioned
3514
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
3515
"""Compare an entry and real disk to generate delta information.
3517
:param path_info: top_relpath, basename, kind, lstat, abspath for
3518
the path of entry. If None, then the path is considered absent in
3519
the target (Perhaps we should pass in a concrete entry for this ?)
3520
Basename is returned as a utf8 string because we expect this
3521
tuple will be ignored, and don't want to take the time to
3523
:return: (iter_changes_result, changed). If the entry has not been
3524
handled then changed is None. Otherwise it is False if no content
3525
or metadata changes have occurred, and True if any content or
3526
metadata change has occurred. If self.include_unchanged is True then
3527
if changed is not None, iter_changes_result will always be a result
3528
tuple. Otherwise, iter_changes_result is None unless changed is
3531
if self.source_index is None:
3532
source_details = DirState.NULL_PARENT_DETAILS
3534
source_details = entry[1][self.source_index]
3535
target_details = entry[1][self.target_index]
3536
target_minikind = target_details[0]
3537
if path_info is not None and target_minikind in 'fdlt':
3538
if not (self.target_index == 0):
3539
raise AssertionError()
3540
link_or_sha1 = update_entry(self.state, entry,
3541
abspath=path_info[4], stat_value=path_info[3])
3542
# The entry may have been modified by update_entry
3543
target_details = entry[1][self.target_index]
3544
target_minikind = target_details[0]
3547
file_id = entry[0][2]
3548
source_minikind = source_details[0]
3549
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
3550
# claimed content in both: diff
3551
# r | fdlt | | add source to search, add id path move and perform
3552
# | | | diff check on source-target
3553
# r | fdlt | a | dangling file that was present in the basis.
3555
if source_minikind in 'r':
3556
# add the source to the search path to find any children it
3557
# has. TODO ? : only add if it is a container ?
3558
if not osutils.is_inside_any(self.searched_specific_files,
3560
self.search_specific_files.add(source_details[1])
3561
# generate the old path; this is needed for stating later
3563
old_path = source_details[1]
3564
old_dirname, old_basename = os.path.split(old_path)
3565
path = pathjoin(entry[0][0], entry[0][1])
3566
old_entry = self.state._get_entry(self.source_index,
3568
# update the source details variable to be the real
3570
if old_entry == (None, None):
3571
raise errors.CorruptDirstate(self.state._filename,
3572
"entry '%s/%s' is considered renamed from %r"
3573
" but source does not exist\n"
3574
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
3575
source_details = old_entry[1][self.source_index]
3576
source_minikind = source_details[0]
3578
old_dirname = entry[0][0]
3579
old_basename = entry[0][1]
3580
old_path = path = None
3581
if path_info is None:
3582
# the file is missing on disk, show as removed.
3583
content_change = True
3587
# source and target are both versioned and disk file is present.
3588
target_kind = path_info[2]
3589
if target_kind == 'directory':
3591
old_path = path = pathjoin(old_dirname, old_basename)
3592
self.new_dirname_to_file_id[path] = file_id
3593
if source_minikind != 'd':
3594
content_change = True
3596
# directories have no fingerprint
3597
content_change = False
3599
elif target_kind == 'file':
3600
if source_minikind != 'f':
3601
content_change = True
3603
# Check the sha. We can't just rely on the size as
3604
# content filtering may mean differ sizes actually
3605
# map to the same content
3606
if link_or_sha1 is None:
3608
statvalue, link_or_sha1 = \
3609
self.state._sha1_provider.stat_and_sha1(
3611
self.state._observed_sha1(entry, link_or_sha1,
3613
content_change = (link_or_sha1 != source_details[1])
3614
# Target details is updated at update_entry time
3615
if self.use_filesystem_for_exec:
3616
# We don't need S_ISREG here, because we are sure
3617
# we are dealing with a file.
3618
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
3620
target_exec = target_details[3]
3621
elif target_kind == 'symlink':
3622
if source_minikind != 'l':
3623
content_change = True
3625
content_change = (link_or_sha1 != source_details[1])
3627
elif target_kind == 'tree-reference':
3628
if source_minikind != 't':
3629
content_change = True
3631
content_change = False
3635
path = pathjoin(old_dirname, old_basename)
3636
raise errors.BadFileKindError(path, path_info[2])
3637
if source_minikind == 'd':
3639
old_path = path = pathjoin(old_dirname, old_basename)
3640
self.old_dirname_to_file_id[old_path] = file_id
3641
# parent id is the entry for the path in the target tree
3642
if old_basename and old_dirname == self.last_source_parent[0]:
3643
source_parent_id = self.last_source_parent[1]
3646
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3648
source_parent_entry = self.state._get_entry(self.source_index,
3649
path_utf8=old_dirname)
3650
source_parent_id = source_parent_entry[0][2]
3651
if source_parent_id == entry[0][2]:
3652
# This is the root, so the parent is None
3653
source_parent_id = None
3655
self.last_source_parent[0] = old_dirname
3656
self.last_source_parent[1] = source_parent_id
3657
new_dirname = entry[0][0]
3658
if entry[0][1] and new_dirname == self.last_target_parent[0]:
3659
target_parent_id = self.last_target_parent[1]
3662
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3664
# TODO: We don't always need to do the lookup, because the
3665
# parent entry will be the same as the source entry.
3666
target_parent_entry = self.state._get_entry(self.target_index,
3667
path_utf8=new_dirname)
3668
if target_parent_entry == (None, None):
3669
raise AssertionError(
3670
"Could not find target parent in wt: %s\nparent of: %s"
3671
% (new_dirname, entry))
3672
target_parent_id = target_parent_entry[0][2]
3673
if target_parent_id == entry[0][2]:
3674
# This is the root, so the parent is None
3675
target_parent_id = None
3677
self.last_target_parent[0] = new_dirname
3678
self.last_target_parent[1] = target_parent_id
3680
source_exec = source_details[3]
3681
changed = (content_change
3682
or source_parent_id != target_parent_id
3683
or old_basename != entry[0][1]
3684
or source_exec != target_exec
3686
if not changed and not self.include_unchanged:
3689
if old_path is None:
3690
old_path = path = pathjoin(old_dirname, old_basename)
3691
old_path_u = self.utf8_decode(old_path)[0]
3694
old_path_u = self.utf8_decode(old_path)[0]
3695
if old_path == path:
3698
path_u = self.utf8_decode(path)[0]
3699
source_kind = DirState._minikind_to_kind[source_minikind]
3700
return (entry[0][2],
3701
(old_path_u, path_u),
3704
(source_parent_id, target_parent_id),
3705
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3706
(source_kind, target_kind),
3707
(source_exec, target_exec)), changed
3708
elif source_minikind in 'a' and target_minikind in 'fdlt':
3709
# looks like a new file
3710
path = pathjoin(entry[0][0], entry[0][1])
3711
# parent id is the entry for the path in the target tree
3712
# TODO: these are the same for an entire directory: cache em.
3713
parent_id = self.state._get_entry(self.target_index,
3714
path_utf8=entry[0][0])[0][2]
3715
if parent_id == entry[0][2]:
3717
if path_info is not None:
3719
if self.use_filesystem_for_exec:
3720
# We need S_ISREG here, because we aren't sure if this
3723
stat.S_ISREG(path_info[3].st_mode)
3724
and stat.S_IEXEC & path_info[3].st_mode)
3726
target_exec = target_details[3]
3727
return (entry[0][2],
3728
(None, self.utf8_decode(path)[0]),
3732
(None, self.utf8_decode(entry[0][1])[0]),
3733
(None, path_info[2]),
3734
(None, target_exec)), True
3736
# Its a missing file, report it as such.
3737
return (entry[0][2],
3738
(None, self.utf8_decode(path)[0]),
3742
(None, self.utf8_decode(entry[0][1])[0]),
3744
(None, False)), True
3745
elif source_minikind in 'fdlt' and target_minikind in 'a':
3746
# unversioned, possibly, or possibly not deleted: we dont care.
3747
# if its still on disk, *and* theres no other entry at this
3748
# path [we dont know this in this routine at the moment -
3749
# perhaps we should change this - then it would be an unknown.
3750
old_path = pathjoin(entry[0][0], entry[0][1])
3751
# parent id is the entry for the path in the target tree
3752
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3753
if parent_id == entry[0][2]:
3755
return (entry[0][2],
3756
(self.utf8_decode(old_path)[0], None),
3760
(self.utf8_decode(entry[0][1])[0], None),
3761
(DirState._minikind_to_kind[source_minikind], None),
3762
(source_details[3], None)), True
3763
elif source_minikind in 'fdlt' and target_minikind in 'r':
3764
# a rename; could be a true rename, or a rename inherited from
3765
# a renamed parent. TODO: handle this efficiently. Its not
3766
# common case to rename dirs though, so a correct but slow
3767
# implementation will do.
3768
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3769
self.search_specific_files.add(target_details[1])
3770
elif source_minikind in 'ra' and target_minikind in 'ra':
3771
# neither of the selected trees contain this file,
3772
# so skip over it. This is not currently directly tested, but
3773
# is indirectly via test_too_much.TestCommands.test_conflicts.
3776
raise AssertionError("don't know how to compare "
3777
"source_minikind=%r, target_minikind=%r"
3778
% (source_minikind, target_minikind))
3784
def _gather_result_for_consistency(self, result):
3785
"""Check a result we will yield to make sure we are consistent later.
3787
This gathers result's parents into a set to output later.
3789
:param result: A result tuple.
3791
if not self.partial or not result[0]:
3793
self.seen_ids.add(result[0])
3794
new_path = result[1][1]
3796
# Not the root and not a delete: queue up the parents of the path.
3797
self.search_specific_file_parents.update(
3798
osutils.parent_directories(new_path.encode('utf8')))
3799
# Add the root directory which parent_directories does not
3801
self.search_specific_file_parents.add('')
3803
def iter_changes(self):
3804
"""Iterate over the changes."""
3805
utf8_decode = cache_utf8._utf8_decode
3806
_lt_by_dirs = lt_by_dirs
3807
_process_entry = self._process_entry
3808
search_specific_files = self.search_specific_files
3809
searched_specific_files = self.searched_specific_files
3810
splitpath = osutils.splitpath
3812
# compare source_index and target_index at or under each element of search_specific_files.
3813
# follow the following comparison table. Note that we only want to do diff operations when
3814
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3818
# Source | Target | disk | action
3819
# r | fdlt | | add source to search, add id path move and perform
3820
# | | | diff check on source-target
3821
# r | fdlt | a | dangling file that was present in the basis.
3823
# r | a | | add source to search
3825
# r | r | | this path is present in a non-examined tree, skip.
3826
# r | r | a | this path is present in a non-examined tree, skip.
3827
# a | fdlt | | add new id
3828
# a | fdlt | a | dangling locally added file, skip
3829
# a | a | | not present in either tree, skip
3830
# a | a | a | not present in any tree, skip
3831
# a | r | | not present in either tree at this path, skip as it
3832
# | | | may not be selected by the users list of paths.
3833
# a | r | a | not present in either tree at this path, skip as it
3834
# | | | may not be selected by the users list of paths.
3835
# fdlt | fdlt | | content in both: diff them
3836
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3837
# fdlt | a | | unversioned: output deleted id for now
3838
# fdlt | a | a | unversioned and deleted: output deleted id
3839
# fdlt | r | | relocated in this tree, so add target to search.
3840
# | | | Dont diff, we will see an r,fd; pair when we reach
3841
# | | | this id at the other path.
3842
# fdlt | r | a | relocated in this tree, so add target to search.
3843
# | | | Dont diff, we will see an r,fd; pair when we reach
3844
# | | | this id at the other path.
3846
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3847
# keeping a cache of directories that we have seen.
3849
while search_specific_files:
3850
# TODO: the pending list should be lexically sorted? the
3851
# interface doesn't require it.
3852
current_root = search_specific_files.pop()
3853
current_root_unicode = current_root.decode('utf8')
3854
searched_specific_files.add(current_root)
3855
# process the entries for this containing directory: the rest will be
3856
# found by their parents recursively.
3857
root_entries = self.state._entries_for_path(current_root)
3858
root_abspath = self.tree.abspath(current_root_unicode)
3860
root_stat = os.lstat(root_abspath)
3861
except OSError as e:
3862
if e.errno == errno.ENOENT:
3863
# the path does not exist: let _process_entry know that.
3864
root_dir_info = None
3866
# some other random error: hand it up.
3869
root_dir_info = ('', current_root,
3870
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3872
if root_dir_info[2] == 'directory':
3873
if self.tree._directory_is_tree_reference(
3874
current_root.decode('utf8')):
3875
root_dir_info = root_dir_info[:2] + \
3876
('tree-reference',) + root_dir_info[3:]
3878
if not root_entries and not root_dir_info:
3879
# this specified path is not present at all, skip it.
3881
path_handled = False
3882
for entry in root_entries:
3883
result, changed = _process_entry(entry, root_dir_info)
3884
if changed is not None:
3887
self._gather_result_for_consistency(result)
3888
if changed or self.include_unchanged:
3890
if self.want_unversioned and not path_handled and root_dir_info:
3891
new_executable = bool(
3892
stat.S_ISREG(root_dir_info[3].st_mode)
3893
and stat.S_IEXEC & root_dir_info[3].st_mode)
3895
(None, current_root_unicode),
3899
(None, splitpath(current_root_unicode)[-1]),
3900
(None, root_dir_info[2]),
3901
(None, new_executable)
3903
initial_key = (current_root, '', '')
3904
block_index, _ = self.state._find_block_index_from_key(initial_key)
3905
if block_index == 0:
3906
# we have processed the total root already, but because the
3907
# initial key matched it we should skip it here.
3909
if root_dir_info and root_dir_info[2] == 'tree-reference':
3910
current_dir_info = None
3912
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3914
current_dir_info = next(dir_iterator)
3915
except OSError as e:
3916
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3917
# python 2.5 has e.errno == EINVAL,
3918
# and e.winerror == ERROR_DIRECTORY
3919
e_winerror = getattr(e, 'winerror', None)
3920
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3921
# there may be directories in the inventory even though
3922
# this path is not a file on disk: so mark it as end of
3924
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3925
current_dir_info = None
3926
elif (sys.platform == 'win32'
3927
and (e.errno in win_errors
3928
or e_winerror in win_errors)):
3929
current_dir_info = None
3933
if current_dir_info[0][0] == '':
3934
# remove .bzr from iteration
3935
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3936
if current_dir_info[1][bzr_index][0] != '.bzr':
3937
raise AssertionError()
3938
del current_dir_info[1][bzr_index]
3939
# walk until both the directory listing and the versioned metadata
3941
if (block_index < len(self.state._dirblocks) and
3942
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3943
current_block = self.state._dirblocks[block_index]
3945
current_block = None
3946
while (current_dir_info is not None or
3947
current_block is not None):
3948
if (current_dir_info and current_block
3949
and current_dir_info[0][0] != current_block[0]):
3950
if _lt_by_dirs(current_dir_info[0][0], current_block[0]):
3951
# filesystem data refers to paths not covered by the dirblock.
3952
# this has two possibilities:
3953
# A) it is versioned but empty, so there is no block for it
3954
# B) it is not versioned.
3956
# if (A) then we need to recurse into it to check for
3957
# new unknown files or directories.
3958
# if (B) then we should ignore it, because we don't
3959
# recurse into unknown directories.
3961
while path_index < len(current_dir_info[1]):
3962
current_path_info = current_dir_info[1][path_index]
3963
if self.want_unversioned:
3964
if current_path_info[2] == 'directory':
3965
if self.tree._directory_is_tree_reference(
3966
current_path_info[0].decode('utf8')):
3967
current_path_info = current_path_info[:2] + \
3968
('tree-reference',) + current_path_info[3:]
3969
new_executable = bool(
3970
stat.S_ISREG(current_path_info[3].st_mode)
3971
and stat.S_IEXEC & current_path_info[3].st_mode)
3973
(None, utf8_decode(current_path_info[0])[0]),
3977
(None, utf8_decode(current_path_info[1])[0]),
3978
(None, current_path_info[2]),
3979
(None, new_executable))
3980
# dont descend into this unversioned path if it is
3982
if current_path_info[2] in ('directory',
3984
del current_dir_info[1][path_index]
3988
# This dir info has been handled, go to the next
3990
current_dir_info = next(dir_iterator)
3991
except StopIteration:
3992
current_dir_info = None
3994
# We have a dirblock entry for this location, but there
3995
# is no filesystem path for this. This is most likely
3996
# because a directory was removed from the disk.
3997
# We don't have to report the missing directory,
3998
# because that should have already been handled, but we
3999
# need to handle all of the files that are contained
4001
for current_entry in current_block[1]:
4002
# entry referring to file not present on disk.
4003
# advance the entry only, after processing.
4004
result, changed = _process_entry(current_entry, None)
4005
if changed is not None:
4007
self._gather_result_for_consistency(result)
4008
if changed or self.include_unchanged:
4011
if (block_index < len(self.state._dirblocks) and
4012
osutils.is_inside(current_root,
4013
self.state._dirblocks[block_index][0])):
4014
current_block = self.state._dirblocks[block_index]
4016
current_block = None
4019
if current_block and entry_index < len(current_block[1]):
4020
current_entry = current_block[1][entry_index]
4022
current_entry = None
4023
advance_entry = True
4025
if current_dir_info and path_index < len(current_dir_info[1]):
4026
current_path_info = current_dir_info[1][path_index]
4027
if current_path_info[2] == 'directory':
4028
if self.tree._directory_is_tree_reference(
4029
current_path_info[0].decode('utf8')):
4030
current_path_info = current_path_info[:2] + \
4031
('tree-reference',) + current_path_info[3:]
4033
current_path_info = None
4035
path_handled = False
4036
while (current_entry is not None or
4037
current_path_info is not None):
4038
if current_entry is None:
4039
# the check for path_handled when the path is advanced
4040
# will yield this path if needed.
4042
elif current_path_info is None:
4043
# no path is fine: the per entry code will handle it.
4044
result, changed = _process_entry(current_entry, current_path_info)
4045
if changed is not None:
4047
self._gather_result_for_consistency(result)
4048
if changed or self.include_unchanged:
4050
elif (current_entry[0][1] != current_path_info[1]
4051
or current_entry[1][self.target_index][0] in 'ar'):
4052
# The current path on disk doesn't match the dirblock
4053
# record. Either the dirblock is marked as absent, or
4054
# the file on disk is not present at all in the
4055
# dirblock. Either way, report about the dirblock
4056
# entry, and let other code handle the filesystem one.
4058
# Compare the basename for these files to determine
4060
if current_path_info[1] < current_entry[0][1]:
4061
# extra file on disk: pass for now, but only
4062
# increment the path, not the entry
4063
advance_entry = False
4065
# entry referring to file not present on disk.
4066
# advance the entry only, after processing.
4067
result, changed = _process_entry(current_entry, None)
4068
if changed is not None:
4070
self._gather_result_for_consistency(result)
4071
if changed or self.include_unchanged:
4073
advance_path = False
4075
result, changed = _process_entry(current_entry, current_path_info)
4076
if changed is not None:
4079
self._gather_result_for_consistency(result)
4080
if changed or self.include_unchanged:
4082
if advance_entry and current_entry is not None:
4084
if entry_index < len(current_block[1]):
4085
current_entry = current_block[1][entry_index]
4087
current_entry = None
4089
advance_entry = True # reset the advance flaga
4090
if advance_path and current_path_info is not None:
4091
if not path_handled:
4092
# unversioned in all regards
4093
if self.want_unversioned:
4094
new_executable = bool(
4095
stat.S_ISREG(current_path_info[3].st_mode)
4096
and stat.S_IEXEC & current_path_info[3].st_mode)
4098
relpath_unicode = utf8_decode(current_path_info[0])[0]
4099
except UnicodeDecodeError:
4100
raise errors.BadFilenameEncoding(
4101
current_path_info[0], osutils._fs_enc)
4103
(None, relpath_unicode),
4107
(None, utf8_decode(current_path_info[1])[0]),
4108
(None, current_path_info[2]),
4109
(None, new_executable))
4110
# dont descend into this unversioned path if it is
4112
if current_path_info[2] in ('directory'):
4113
del current_dir_info[1][path_index]
4115
# dont descend the disk iterator into any tree
4117
if current_path_info[2] == 'tree-reference':
4118
del current_dir_info[1][path_index]
4121
if path_index < len(current_dir_info[1]):
4122
current_path_info = current_dir_info[1][path_index]
4123
if current_path_info[2] == 'directory':
4124
if self.tree._directory_is_tree_reference(
4125
current_path_info[0].decode('utf8')):
4126
current_path_info = current_path_info[:2] + \
4127
('tree-reference',) + current_path_info[3:]
4129
current_path_info = None
4130
path_handled = False
4132
advance_path = True # reset the advance flagg.
4133
if current_block is not None:
4135
if (block_index < len(self.state._dirblocks) and
4136
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
4137
current_block = self.state._dirblocks[block_index]
4139
current_block = None
4140
if current_dir_info is not None:
4142
current_dir_info = next(dir_iterator)
4143
except StopIteration:
4144
current_dir_info = None
4145
for result in self._iter_specific_file_parents():
4148
def _iter_specific_file_parents(self):
4149
"""Iter over the specific file parents."""
4150
while self.search_specific_file_parents:
4151
# Process the parent directories for the paths we were iterating.
4152
# Even in extremely large trees this should be modest, so currently
4153
# no attempt is made to optimise.
4154
path_utf8 = self.search_specific_file_parents.pop()
4155
if osutils.is_inside_any(self.searched_specific_files, path_utf8):
4156
# We've examined this path.
4158
if path_utf8 in self.searched_exact_paths:
4159
# We've examined this path.
4161
path_entries = self.state._entries_for_path(path_utf8)
4162
# We need either one or two entries. If the path in
4163
# self.target_index has moved (so the entry in source_index is in
4164
# 'ar') then we need to also look for the entry for this path in
4165
# self.source_index, to output the appropriate delete-or-rename.
4166
selected_entries = []
4168
for candidate_entry in path_entries:
4169
# Find entries present in target at this path:
4170
if candidate_entry[1][self.target_index][0] not in 'ar':
4172
selected_entries.append(candidate_entry)
4173
# Find entries present in source at this path:
4174
elif (self.source_index is not None and
4175
candidate_entry[1][self.source_index][0] not in 'ar'):
4177
if candidate_entry[1][self.target_index][0] == 'a':
4178
# Deleted, emit it here.
4179
selected_entries.append(candidate_entry)
4181
# renamed, emit it when we process the directory it
4183
self.search_specific_file_parents.add(
4184
candidate_entry[1][self.target_index][1])
4186
raise AssertionError(
4187
"Missing entry for specific path parent %r, %r" % (
4188
path_utf8, path_entries))
4189
path_info = self._path_info(path_utf8, path_utf8.decode('utf8'))
4190
for entry in selected_entries:
4191
if entry[0][2] in self.seen_ids:
4193
result, changed = self._process_entry(entry, path_info)
4195
raise AssertionError(
4196
"Got entry<->path mismatch for specific path "
4197
"%r entry %r path_info %r " % (
4198
path_utf8, entry, path_info))
4199
# Only include changes - we're outside the users requested
4202
self._gather_result_for_consistency(result)
4203
if (result[6][0] == 'directory' and
4204
result[6][1] != 'directory'):
4205
# This stopped being a directory, the old children have
4207
if entry[1][self.source_index][0] == 'r':
4208
# renamed, take the source path
4209
entry_path_utf8 = entry[1][self.source_index][1]
4211
entry_path_utf8 = path_utf8
4212
initial_key = (entry_path_utf8, '', '')
4213
block_index, _ = self.state._find_block_index_from_key(
4215
if block_index == 0:
4216
# The children of the root are in block index 1.
4218
current_block = None
4219
if block_index < len(self.state._dirblocks):
4220
current_block = self.state._dirblocks[block_index]
4221
if not osutils.is_inside(
4222
entry_path_utf8, current_block[0]):
4223
# No entries for this directory at all.
4224
current_block = None
4225
if current_block is not None:
4226
for entry in current_block[1]:
4227
if entry[1][self.source_index][0] in 'ar':
4228
# Not in the source tree, so doesn't have to be
4231
# Path of the entry itself.
4233
self.search_specific_file_parents.add(
4234
osutils.pathjoin(*entry[0][:2]))
4235
if changed or self.include_unchanged:
4237
self.searched_exact_paths.add(path_utf8)
4239
def _path_info(self, utf8_path, unicode_path):
4240
"""Generate path_info for unicode_path.
4242
:return: None if unicode_path does not exist, or a path_info tuple.
4244
abspath = self.tree.abspath(unicode_path)
4246
stat = os.lstat(abspath)
4247
except OSError as e:
4248
if e.errno == errno.ENOENT:
4249
# the path does not exist.
4253
utf8_basename = utf8_path.rsplit('/', 1)[-1]
4254
dir_info = (utf8_path, utf8_basename,
4255
osutils.file_kind_from_stat_mode(stat.st_mode), stat,
4257
if dir_info[2] == 'directory':
4258
if self.tree._directory_is_tree_reference(
4260
self.root_dir_info = self.root_dir_info[:2] + \
4261
('tree-reference',) + self.root_dir_info[3:]
4265
# Try to load the compiled form if possible
4267
from breezy._dirstate_helpers_pyx import (
4274
ProcessEntryC as _process_entry,
4275
update_entry as update_entry,
4277
except ImportError as e:
4278
osutils.failed_to_load_extension(e)
4279
from breezy._dirstate_helpers_py import (
4287
# FIXME: It would be nice to be able to track moved lines so that the
4288
# corresponding python code can be moved to the _dirstate_helpers_py
4289
# module. I don't want to break the history for this important piece of
4290
# code so I left the code here -- vila 20090622
4291
update_entry = py_update_entry
4292
_process_entry = ProcessEntryPython
added
removed
breezy/dirstate.py
