1
# Copyright (C) 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but it's a file. The fingerprint is the
86
sha1 value of the file's canonical form, i.e. after any read filters have
87
been applied to the convenience form stored in the working tree.
88
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
90
't' is a reference to a nested subtree; the fingerprint is the referenced
95
The entries on disk and in memory are ordered according to the following keys:
97
directory, as a list of components
101
--- Format 1 had the following different definition: ---
102
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
103
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
105
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
106
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
109
PARENT ROW's are emitted for every parent that is not in the ghosts details
110
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
111
each row will have a PARENT ROW for foo and baz, but not for bar.
114
In any tree, a kind of 'moved' indicates that the fingerprint field
115
(which we treat as opaque data specific to the 'kind' anyway) has the
116
details for the id of this row in that tree.
118
I'm strongly tempted to add a id->path index as well, but I think that
119
where we need id->path mapping; we also usually read the whole file, so
120
I'm going to skip that for the moment, as we have the ability to locate
121
via bisect any path in any tree, and if we lookup things by path, we can
122
accumulate an id->path mapping as we go, which will tend to match what we
125
I plan to implement this asap, so please speak up now to alter/tweak the
126
design - and once we stabilise on this, I'll update the wiki page for
129
The rationale for all this is that we want fast operations for the
130
common case (diff/status/commit/merge on all files) and extremely fast
131
operations for the less common but still occurs a lot status/diff/commit
132
on specific files). Operations on specific files involve a scan for all
133
the children of a path, *in every involved tree*, which the current
134
format did not accommodate.
138
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
139
2) fall back current object model as needed.
140
3) scale usably to the largest trees known today - say 50K entries. (mozilla
141
is an example of this)
145
Eventually reuse dirstate objects across locks IFF the dirstate file has not
146
been modified, but will require that we flush/ignore cached stat-hit data
147
because we won't want to restat all files on disk just because a lock was
148
acquired, yet we cannot trust the data after the previous lock was released.
150
Memory representation:
151
vector of all directories, and vector of the childen ?
153
root_entrie = (direntry for root, [parent_direntries_for_root]),
155
('', ['data for achild', 'data for bchild', 'data for cchild'])
156
('dir', ['achild', 'cchild', 'echild'])
158
- single bisect to find N subtrees from a path spec
159
- in-order for serialisation - this is 'dirblock' grouping.
160
- insertion of a file '/a' affects only the '/' child-vector, that is, to
161
insert 10K elements from scratch does not generates O(N^2) memoves of a
162
single vector, rather each individual, which tends to be limited to a
163
manageable number. Will scale badly on trees with 10K entries in a
164
single directory. compare with Inventory.InventoryDirectory which has
165
a dictionary for the children. No bisect capability, can only probe for
166
exact matches, or grab all elements and sort.
167
- What's the risk of error here? Once we have the base format being processed
168
we should have a net win regardless of optimality. So we are going to
169
go with what seems reasonable.
172
Maybe we should do a test profile of the core structure - 10K simulated
173
searches/lookups/etc?
175
Objects for each row?
176
The lifetime of Dirstate objects is current per lock, but see above for
177
possible extensions. The lifetime of a row from a dirstate is expected to be
178
very short in the optimistic case: which we are optimising for. For instance,
179
subtree status will determine from analysis of the disk data what rows need to
180
be examined at all, and will be able to determine from a single row whether
181
that file has altered or not, so we are aiming to process tens of thousands of
182
entries each second within the dirstate context, before exposing anything to
183
the larger codebase. This suggests we want the time for a single file
184
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
185
processed, and to scale to 100 thousand we'll another order of magnitude to do
186
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
187
the file on disk, and then immediately discard, the overhead of object creation
188
becomes a significant cost.
190
Figures: Creating a tuple from 3 elements was profiled at 0.0625
191
microseconds, whereas creating a object which is subclassed from tuple was
192
0.500 microseconds, and creating an object with 3 elements and slots was 3
193
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
194
down to 10 microseconds for the total processing - having 33% of that be object
195
creation is a huge overhead. There is a potential cost in using tuples within
196
each row which is that the conditional code to do comparisons may be slower
197
than method invocation, but method invocation is known to be slow due to stack
198
frame creation, so avoiding methods in these tight inner loops in unfortunately
199
desirable. We can consider a pyrex version of this with objects in future if
208
from stat import S_IEXEC
226
# This is the Windows equivalent of ENOTDIR
227
# It is defined in pywin32.winerror, but we don't want a strong dependency for
228
# just an error code.
229
ERROR_PATH_NOT_FOUND = 3
230
ERROR_DIRECTORY = 267
233
if not getattr(struct, '_compile', None):
234
# Cannot pre-compile the dirstate pack_stat
235
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
236
"""Convert stat values into a packed representation."""
237
return _encode(_pack('>LLLLLL', st.st_size, int(st.st_mtime),
238
int(st.st_ctime), st.st_dev, st.st_ino & 0xFFFFFFFF,
241
# compile the struct compiler we need, so as to only do it once
242
from _struct import Struct
243
_compiled_pack = Struct('>LLLLLL').pack
244
def pack_stat(st, _encode=binascii.b2a_base64, _pack=_compiled_pack):
245
"""Convert stat values into a packed representation."""
246
# jam 20060614 it isn't really worth removing more entries if we
247
# are going to leave it in packed form.
248
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
249
# With all entries, filesize is 5.9M and read time is maybe 280ms
250
# well within the noise margin
252
# base64 encoding always adds a final newline, so strip it off
253
# The current version
254
return _encode(_pack(st.st_size, int(st.st_mtime), int(st.st_ctime),
255
st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
256
# This is 0.060s / 1.520s faster by not encoding as much information
257
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
258
# This is not strictly faster than _encode(_pack())[:-1]
259
# return '%X.%X.%X.%X.%X.%X' % (
260
# st.st_size, int(st.st_mtime), int(st.st_ctime),
261
# st.st_dev, st.st_ino, st.st_mode)
262
# Similar to the _encode(_pack('>LL'))
263
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
266
class SHA1Provider(object):
267
"""An interface for getting sha1s of a file."""
269
def sha1(self, abspath):
270
"""Return the sha1 of a file given its absolute path.
272
:param abspath: May be a filesystem encoded absolute path
275
raise NotImplementedError(self.sha1)
277
def stat_and_sha1(self, abspath):
278
"""Return the stat and sha1 of a file given its absolute path.
280
:param abspath: May be a filesystem encoded absolute path
283
Note: the stat should be the stat of the physical file
284
while the sha may be the sha of its canonical content.
286
raise NotImplementedError(self.stat_and_sha1)
289
class DefaultSHA1Provider(SHA1Provider):
290
"""A SHA1Provider that reads directly from the filesystem."""
292
def sha1(self, abspath):
293
"""Return the sha1 of a file given its absolute path."""
294
return osutils.sha_file_by_name(abspath)
296
def stat_and_sha1(self, abspath):
297
"""Return the stat and sha1 of a file given its absolute path."""
298
file_obj = file(abspath, 'rb')
300
statvalue = os.fstat(file_obj.fileno())
301
sha1 = osutils.sha_file(file_obj)
304
return statvalue, sha1
307
class DirState(object):
308
"""Record directory and metadata state for fast access.
310
A dirstate is a specialised data structure for managing local working
311
tree state information. Its not yet well defined whether it is platform
312
specific, and if it is how we detect/parameterize that.
314
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
315
Unlike most bzr disk formats, DirStates must be locked for reading, using
316
lock_read. (This is an os file lock internally.) This is necessary
317
because the file can be rewritten in place.
319
DirStates must be explicitly written with save() to commit changes; just
320
unlocking them does not write the changes to disk.
323
_kind_to_minikind = {
329
'tree-reference': 't',
331
_minikind_to_kind = {
337
't': 'tree-reference',
339
_stat_to_minikind = {
344
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
345
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
348
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
349
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
350
# that we are sure a single record will always fit.
351
BISECT_PAGE_SIZE = 4096
354
IN_MEMORY_UNMODIFIED = 1
355
IN_MEMORY_MODIFIED = 2
357
# A pack_stat (the x's) that is just noise and will never match the output
360
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
362
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
363
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
365
def __init__(self, path, sha1_provider):
366
"""Create a DirState object.
368
:param path: The path at which the dirstate file on disk should live.
369
:param sha1_provider: an object meeting the SHA1Provider interface.
371
# _header_state and _dirblock_state represent the current state
372
# of the dirstate metadata and the per-row data respectiely.
373
# NOT_IN_MEMORY indicates that no data is in memory
374
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
375
# is the same as is on disk
376
# IN_MEMORY_MODIFIED indicates that we have a modified version
377
# of what is on disk.
378
# In future we will add more granularity, for instance _dirblock_state
379
# will probably support partially-in-memory as a separate variable,
380
# allowing for partially-in-memory unmodified and partially-in-memory
382
self._header_state = DirState.NOT_IN_MEMORY
383
self._dirblock_state = DirState.NOT_IN_MEMORY
384
# If true, an error has been detected while updating the dirstate, and
385
# for safety we're not going to commit to disk.
386
self._changes_aborted = False
390
self._state_file = None
391
self._filename = path
392
self._lock_token = None
393
self._lock_state = None
394
self._id_index = None
395
# a map from packed_stat to sha's.
396
self._packed_stat_index = None
397
self._end_of_header = None
398
self._cutoff_time = None
399
self._split_path_cache = {}
400
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
401
self._sha1_provider = sha1_provider
402
if 'hashcache' in debug.debug_flags:
403
self._sha1_file = self._sha1_file_and_mutter
405
self._sha1_file = self._sha1_provider.sha1
406
# These two attributes provide a simple cache for lookups into the
407
# dirstate in-memory vectors. By probing respectively for the last
408
# block, and for the next entry, we save nearly 2 bisections per path
410
self._last_block_index = None
411
self._last_entry_index = None
415
(self.__class__.__name__, self._filename)
417
def add(self, path, file_id, kind, stat, fingerprint):
418
"""Add a path to be tracked.
420
:param path: The path within the dirstate - '' is the root, 'foo' is the
421
path foo within the root, 'foo/bar' is the path bar within foo
423
:param file_id: The file id of the path being added.
424
:param kind: The kind of the path, as a string like 'file',
426
:param stat: The output of os.lstat for the path.
427
:param fingerprint: The sha value of the file's canonical form (i.e.
428
after any read filters have been applied),
429
or the target of a symlink,
430
or the referenced revision id for tree-references,
431
or '' for directories.
434
# find the block its in.
435
# find the location in the block.
436
# check its not there
438
#------- copied from inventory.ensure_normalized_name - keep synced.
439
# --- normalized_filename wants a unicode basename only, so get one.
440
dirname, basename = osutils.split(path)
441
# we dont import normalized_filename directly because we want to be
442
# able to change the implementation at runtime for tests.
443
norm_name, can_access = osutils.normalized_filename(basename)
444
if norm_name != basename:
448
raise errors.InvalidNormalization(path)
449
# you should never have files called . or ..; just add the directory
450
# in the parent, or according to the special treatment for the root
451
if basename == '.' or basename == '..':
452
raise errors.InvalidEntryName(path)
453
# now that we've normalised, we need the correct utf8 path and
454
# dirname and basename elements. This single encode and split should be
455
# faster than three separate encodes.
456
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
457
dirname, basename = osutils.split(utf8path)
458
# uses __class__ for speed; the check is needed for safety
459
if file_id.__class__ is not str:
460
raise AssertionError(
461
"must be a utf8 file_id not %s" % (type(file_id), ))
462
# Make sure the file_id does not exist in this tree
464
file_id_entry = self._get_entry(0, fileid_utf8=file_id, include_deleted=True)
465
if file_id_entry != (None, None):
466
if file_id_entry[1][0][0] == 'a':
467
if file_id_entry[0] != (dirname, basename, file_id):
468
# set the old name's current operation to rename
469
self.update_minimal(file_id_entry[0],
475
rename_from = file_id_entry[0][0:2]
477
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
478
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
479
info = '%s:%s' % (kind, path)
480
raise errors.DuplicateFileId(file_id, info)
481
first_key = (dirname, basename, '')
482
block_index, present = self._find_block_index_from_key(first_key)
484
# check the path is not in the tree
485
block = self._dirblocks[block_index][1]
486
entry_index, _ = self._find_entry_index(first_key, block)
487
while (entry_index < len(block) and
488
block[entry_index][0][0:2] == first_key[0:2]):
489
if block[entry_index][1][0][0] not in 'ar':
490
# this path is in the dirstate in the current tree.
491
raise Exception, "adding already added path!"
494
# The block where we want to put the file is not present. But it
495
# might be because the directory was empty, or not loaded yet. Look
496
# for a parent entry, if not found, raise NotVersionedError
497
parent_dir, parent_base = osutils.split(dirname)
498
parent_block_idx, parent_entry_idx, _, parent_present = \
499
self._get_block_entry_index(parent_dir, parent_base, 0)
500
if not parent_present:
501
raise errors.NotVersionedError(path, str(self))
502
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
503
block = self._dirblocks[block_index][1]
504
entry_key = (dirname, basename, file_id)
507
packed_stat = DirState.NULLSTAT
510
packed_stat = pack_stat(stat)
511
parent_info = self._empty_parent_info()
512
minikind = DirState._kind_to_minikind[kind]
513
if rename_from is not None:
515
old_path_utf8 = '%s/%s' % rename_from
517
old_path_utf8 = rename_from[1]
518
parent_info[0] = ('r', old_path_utf8, 0, False, '')
520
entry_data = entry_key, [
521
(minikind, fingerprint, size, False, packed_stat),
523
elif kind == 'directory':
524
entry_data = entry_key, [
525
(minikind, '', 0, False, packed_stat),
527
elif kind == 'symlink':
528
entry_data = entry_key, [
529
(minikind, fingerprint, size, False, packed_stat),
531
elif kind == 'tree-reference':
532
entry_data = entry_key, [
533
(minikind, fingerprint, 0, False, packed_stat),
536
raise errors.BzrError('unknown kind %r' % kind)
537
entry_index, present = self._find_entry_index(entry_key, block)
539
block.insert(entry_index, entry_data)
541
if block[entry_index][1][0][0] != 'a':
542
raise AssertionError(" %r(%r) already added" % (basename, file_id))
543
block[entry_index][1][0] = entry_data[1][0]
545
if kind == 'directory':
546
# insert a new dirblock
547
self._ensure_block(block_index, entry_index, utf8path)
548
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
550
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
552
def _bisect(self, paths):
553
"""Bisect through the disk structure for specific rows.
555
:param paths: A list of paths to find
556
:return: A dict mapping path => entries for found entries. Missing
557
entries will not be in the map.
558
The list is not sorted, and entries will be populated
559
based on when they were read.
561
self._requires_lock()
562
# We need the file pointer to be right after the initial header block
563
self._read_header_if_needed()
564
# If _dirblock_state was in memory, we should just return info from
565
# there, this function is only meant to handle when we want to read
567
if self._dirblock_state != DirState.NOT_IN_MEMORY:
568
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
570
# The disk representation is generally info + '\0\n\0' at the end. But
571
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
572
# Because it means we can sync on the '\n'
573
state_file = self._state_file
574
file_size = os.fstat(state_file.fileno()).st_size
575
# We end up with 2 extra fields, we should have a trailing '\n' to
576
# ensure that we read the whole record, and we should have a precursur
577
# '' which ensures that we start after the previous '\n'
578
entry_field_count = self._fields_per_entry() + 1
580
low = self._end_of_header
581
high = file_size - 1 # Ignore the final '\0'
582
# Map from (dir, name) => entry
585
# Avoid infinite seeking
586
max_count = 30*len(paths)
588
# pending is a list of places to look.
589
# each entry is a tuple of low, high, dir_names
590
# low -> the first byte offset to read (inclusive)
591
# high -> the last byte offset (inclusive)
592
# dir_names -> The list of (dir, name) pairs that should be found in
593
# the [low, high] range
594
pending = [(low, high, paths)]
596
page_size = self._bisect_page_size
598
fields_to_entry = self._get_fields_to_entry()
601
low, high, cur_files = pending.pop()
603
if not cur_files or low >= high:
608
if count > max_count:
609
raise errors.BzrError('Too many seeks, most likely a bug.')
611
mid = max(low, (low+high-page_size)/2)
614
# limit the read size, so we don't end up reading data that we have
616
read_size = min(page_size, (high-mid)+1)
617
block = state_file.read(read_size)
620
entries = block.split('\n')
623
# We didn't find a '\n', so we cannot have found any records.
624
# So put this range back and try again. But we know we have to
625
# increase the page size, because a single read did not contain
626
# a record break (so records must be larger than page_size)
628
pending.append((low, high, cur_files))
631
# Check the first and last entries, in case they are partial, or if
632
# we don't care about the rest of this page
634
first_fields = entries[0].split('\0')
635
if len(first_fields) < entry_field_count:
636
# We didn't get the complete first entry
637
# so move start, and grab the next, which
638
# should be a full entry
639
start += len(entries[0])+1
640
first_fields = entries[1].split('\0')
643
if len(first_fields) <= 2:
644
# We didn't even get a filename here... what do we do?
645
# Try a large page size and repeat this query
647
pending.append((low, high, cur_files))
650
# Find what entries we are looking for, which occur before and
651
# after this first record.
654
first_path = first_fields[1] + '/' + first_fields[2]
656
first_path = first_fields[2]
657
first_loc = _bisect_path_left(cur_files, first_path)
659
# These exist before the current location
660
pre = cur_files[:first_loc]
661
# These occur after the current location, which may be in the
662
# data we read, or might be after the last entry
663
post = cur_files[first_loc:]
665
if post and len(first_fields) >= entry_field_count:
666
# We have files after the first entry
668
# Parse the last entry
669
last_entry_num = len(entries)-1
670
last_fields = entries[last_entry_num].split('\0')
671
if len(last_fields) < entry_field_count:
672
# The very last hunk was not complete,
673
# read the previous hunk
674
after = mid + len(block) - len(entries[-1])
676
last_fields = entries[last_entry_num].split('\0')
678
after = mid + len(block)
681
last_path = last_fields[1] + '/' + last_fields[2]
683
last_path = last_fields[2]
684
last_loc = _bisect_path_right(post, last_path)
686
middle_files = post[:last_loc]
687
post = post[last_loc:]
690
# We have files that should occur in this block
691
# (>= first, <= last)
692
# Either we will find them here, or we can mark them as
695
if middle_files[0] == first_path:
696
# We might need to go before this location
697
pre.append(first_path)
698
if middle_files[-1] == last_path:
699
post.insert(0, last_path)
701
# Find out what paths we have
702
paths = {first_path:[first_fields]}
703
# last_path might == first_path so we need to be
704
# careful if we should append rather than overwrite
705
if last_entry_num != first_entry_num:
706
paths.setdefault(last_path, []).append(last_fields)
707
for num in xrange(first_entry_num+1, last_entry_num):
708
# TODO: jam 20070223 We are already splitting here, so
709
# shouldn't we just split the whole thing rather
710
# than doing the split again in add_one_record?
711
fields = entries[num].split('\0')
713
path = fields[1] + '/' + fields[2]
716
paths.setdefault(path, []).append(fields)
718
for path in middle_files:
719
for fields in paths.get(path, []):
720
# offset by 1 because of the opening '\0'
721
# consider changing fields_to_entry to avoid the
723
entry = fields_to_entry(fields[1:])
724
found.setdefault(path, []).append(entry)
726
# Now we have split up everything into pre, middle, and post, and
727
# we have handled everything that fell in 'middle'.
728
# We add 'post' first, so that we prefer to seek towards the
729
# beginning, so that we will tend to go as early as we need, and
730
# then only seek forward after that.
732
pending.append((after, high, post))
734
pending.append((low, start-1, pre))
736
# Consider that we may want to return the directory entries in sorted
737
# order. For now, we just return them in whatever order we found them,
738
# and leave it up to the caller if they care if it is ordered or not.
741
def _bisect_dirblocks(self, dir_list):
742
"""Bisect through the disk structure to find entries in given dirs.
744
_bisect_dirblocks is meant to find the contents of directories, which
745
differs from _bisect, which only finds individual entries.
747
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
748
:return: A map from dir => entries_for_dir
750
# TODO: jam 20070223 A lot of the bisecting logic could be shared
751
# between this and _bisect. It would require parameterizing the
752
# inner loop with a function, though. We should evaluate the
753
# performance difference.
754
self._requires_lock()
755
# We need the file pointer to be right after the initial header block
756
self._read_header_if_needed()
757
# If _dirblock_state was in memory, we should just return info from
758
# there, this function is only meant to handle when we want to read
760
if self._dirblock_state != DirState.NOT_IN_MEMORY:
761
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
762
# The disk representation is generally info + '\0\n\0' at the end. But
763
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
764
# Because it means we can sync on the '\n'
765
state_file = self._state_file
766
file_size = os.fstat(state_file.fileno()).st_size
767
# We end up with 2 extra fields, we should have a trailing '\n' to
768
# ensure that we read the whole record, and we should have a precursur
769
# '' which ensures that we start after the previous '\n'
770
entry_field_count = self._fields_per_entry() + 1
772
low = self._end_of_header
773
high = file_size - 1 # Ignore the final '\0'
774
# Map from dir => entry
777
# Avoid infinite seeking
778
max_count = 30*len(dir_list)
780
# pending is a list of places to look.
781
# each entry is a tuple of low, high, dir_names
782
# low -> the first byte offset to read (inclusive)
783
# high -> the last byte offset (inclusive)
784
# dirs -> The list of directories that should be found in
785
# the [low, high] range
786
pending = [(low, high, dir_list)]
788
page_size = self._bisect_page_size
790
fields_to_entry = self._get_fields_to_entry()
793
low, high, cur_dirs = pending.pop()
795
if not cur_dirs or low >= high:
800
if count > max_count:
801
raise errors.BzrError('Too many seeks, most likely a bug.')
803
mid = max(low, (low+high-page_size)/2)
806
# limit the read size, so we don't end up reading data that we have
808
read_size = min(page_size, (high-mid)+1)
809
block = state_file.read(read_size)
812
entries = block.split('\n')
815
# We didn't find a '\n', so we cannot have found any records.
816
# So put this range back and try again. But we know we have to
817
# increase the page size, because a single read did not contain
818
# a record break (so records must be larger than page_size)
820
pending.append((low, high, cur_dirs))
823
# Check the first and last entries, in case they are partial, or if
824
# we don't care about the rest of this page
826
first_fields = entries[0].split('\0')
827
if len(first_fields) < entry_field_count:
828
# We didn't get the complete first entry
829
# so move start, and grab the next, which
830
# should be a full entry
831
start += len(entries[0])+1
832
first_fields = entries[1].split('\0')
835
if len(first_fields) <= 1:
836
# We didn't even get a dirname here... what do we do?
837
# Try a large page size and repeat this query
839
pending.append((low, high, cur_dirs))
842
# Find what entries we are looking for, which occur before and
843
# after this first record.
845
first_dir = first_fields[1]
846
first_loc = bisect.bisect_left(cur_dirs, first_dir)
848
# These exist before the current location
849
pre = cur_dirs[:first_loc]
850
# These occur after the current location, which may be in the
851
# data we read, or might be after the last entry
852
post = cur_dirs[first_loc:]
854
if post and len(first_fields) >= entry_field_count:
855
# We have records to look at after the first entry
857
# Parse the last entry
858
last_entry_num = len(entries)-1
859
last_fields = entries[last_entry_num].split('\0')
860
if len(last_fields) < entry_field_count:
861
# The very last hunk was not complete,
862
# read the previous hunk
863
after = mid + len(block) - len(entries[-1])
865
last_fields = entries[last_entry_num].split('\0')
867
after = mid + len(block)
869
last_dir = last_fields[1]
870
last_loc = bisect.bisect_right(post, last_dir)
872
middle_files = post[:last_loc]
873
post = post[last_loc:]
876
# We have files that should occur in this block
877
# (>= first, <= last)
878
# Either we will find them here, or we can mark them as
881
if middle_files[0] == first_dir:
882
# We might need to go before this location
883
pre.append(first_dir)
884
if middle_files[-1] == last_dir:
885
post.insert(0, last_dir)
887
# Find out what paths we have
888
paths = {first_dir:[first_fields]}
889
# last_dir might == first_dir so we need to be
890
# careful if we should append rather than overwrite
891
if last_entry_num != first_entry_num:
892
paths.setdefault(last_dir, []).append(last_fields)
893
for num in xrange(first_entry_num+1, last_entry_num):
894
# TODO: jam 20070223 We are already splitting here, so
895
# shouldn't we just split the whole thing rather
896
# than doing the split again in add_one_record?
897
fields = entries[num].split('\0')
898
paths.setdefault(fields[1], []).append(fields)
900
for cur_dir in middle_files:
901
for fields in paths.get(cur_dir, []):
902
# offset by 1 because of the opening '\0'
903
# consider changing fields_to_entry to avoid the
905
entry = fields_to_entry(fields[1:])
906
found.setdefault(cur_dir, []).append(entry)
908
# Now we have split up everything into pre, middle, and post, and
909
# we have handled everything that fell in 'middle'.
910
# We add 'post' first, so that we prefer to seek towards the
911
# beginning, so that we will tend to go as early as we need, and
912
# then only seek forward after that.
914
pending.append((after, high, post))
916
pending.append((low, start-1, pre))
920
def _bisect_recursive(self, paths):
921
"""Bisect for entries for all paths and their children.
923
This will use bisect to find all records for the supplied paths. It
924
will then continue to bisect for any records which are marked as
925
directories. (and renames?)
927
:param paths: A sorted list of (dir, name) pairs
928
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
929
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
931
# Map from (dir, name, file_id) => [tree_info]
934
found_dir_names = set()
936
# Directories that have been read
937
processed_dirs = set()
938
# Get the ball rolling with the first bisect for all entries.
939
newly_found = self._bisect(paths)
942
# Directories that need to be read
944
paths_to_search = set()
945
for entry_list in newly_found.itervalues():
946
for dir_name_id, trees_info in entry_list:
947
found[dir_name_id] = trees_info
948
found_dir_names.add(dir_name_id[:2])
950
for tree_info in trees_info:
951
minikind = tree_info[0]
954
# We already processed this one as a directory,
955
# we don't need to do the extra work again.
957
subdir, name, file_id = dir_name_id
958
path = osutils.pathjoin(subdir, name)
960
if path not in processed_dirs:
961
pending_dirs.add(path)
962
elif minikind == 'r':
963
# Rename, we need to directly search the target
964
# which is contained in the fingerprint column
965
dir_name = osutils.split(tree_info[1])
966
if dir_name[0] in pending_dirs:
967
# This entry will be found in the dir search
969
if dir_name not in found_dir_names:
970
paths_to_search.add(tree_info[1])
971
# Now we have a list of paths to look for directly, and
972
# directory blocks that need to be read.
973
# newly_found is mixing the keys between (dir, name) and path
974
# entries, but that is okay, because we only really care about the
976
newly_found = self._bisect(sorted(paths_to_search))
977
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
978
processed_dirs.update(pending_dirs)
981
def _discard_merge_parents(self):
982
"""Discard any parents trees beyond the first.
984
Note that if this fails the dirstate is corrupted.
986
After this function returns the dirstate contains 2 trees, neither of
989
self._read_header_if_needed()
990
parents = self.get_parent_ids()
993
# only require all dirblocks if we are doing a full-pass removal.
994
self._read_dirblocks_if_needed()
995
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
996
def iter_entries_removable():
997
for block in self._dirblocks:
998
deleted_positions = []
999
for pos, entry in enumerate(block[1]):
1001
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
1002
deleted_positions.append(pos)
1003
if deleted_positions:
1004
if len(deleted_positions) == len(block[1]):
1007
for pos in reversed(deleted_positions):
1009
# if the first parent is a ghost:
1010
if parents[0] in self.get_ghosts():
1011
empty_parent = [DirState.NULL_PARENT_DETAILS]
1012
for entry in iter_entries_removable():
1013
entry[1][1:] = empty_parent
1015
for entry in iter_entries_removable():
1019
self._parents = [parents[0]]
1020
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1021
self._header_state = DirState.IN_MEMORY_MODIFIED
1023
def _empty_parent_info(self):
1024
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
1027
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
1028
"""Ensure a block for dirname exists.
1030
This function exists to let callers which know that there is a
1031
directory dirname ensure that the block for it exists. This block can
1032
fail to exist because of demand loading, or because a directory had no
1033
children. In either case it is not an error. It is however an error to
1034
call this if there is no parent entry for the directory, and thus the
1035
function requires the coordinates of such an entry to be provided.
1037
The root row is special cased and can be indicated with a parent block
1040
:param parent_block_index: The index of the block in which dirname's row
1042
:param parent_row_index: The index in the parent block where the row
1044
:param dirname: The utf8 dirname to ensure there is a block for.
1045
:return: The index for the block.
1047
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
1048
# This is the signature of the root row, and the
1049
# contents-of-root row is always index 1
1051
# the basename of the directory must be the end of its full name.
1052
if not (parent_block_index == -1 and
1053
parent_block_index == -1 and dirname == ''):
1054
if not dirname.endswith(
1055
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
1056
raise AssertionError("bad dirname %r" % dirname)
1057
block_index, present = self._find_block_index_from_key((dirname, '', ''))
1059
## In future, when doing partial parsing, this should load and
1060
# populate the entire block.
1061
self._dirblocks.insert(block_index, (dirname, []))
1064
def _entries_to_current_state(self, new_entries):
1065
"""Load new_entries into self.dirblocks.
1067
Process new_entries into the current state object, making them the active
1068
state. The entries are grouped together by directory to form dirblocks.
1070
:param new_entries: A sorted list of entries. This function does not sort
1071
to prevent unneeded overhead when callers have a sorted list already.
1074
if new_entries[0][0][0:2] != ('', ''):
1075
raise AssertionError(
1076
"Missing root row %r" % (new_entries[0][0],))
1077
# The two blocks here are deliberate: the root block and the
1078
# contents-of-root block.
1079
self._dirblocks = [('', []), ('', [])]
1080
current_block = self._dirblocks[0][1]
1081
current_dirname = ''
1083
append_entry = current_block.append
1084
for entry in new_entries:
1085
if entry[0][0] != current_dirname:
1086
# new block - different dirname
1088
current_dirname = entry[0][0]
1089
self._dirblocks.append((current_dirname, current_block))
1090
append_entry = current_block.append
1091
# append the entry to the current block
1093
self._split_root_dirblock_into_contents()
1095
def _split_root_dirblock_into_contents(self):
1096
"""Split the root dirblocks into root and contents-of-root.
1098
After parsing by path, we end up with root entries and contents-of-root
1099
entries in the same block. This loop splits them out again.
1101
# The above loop leaves the "root block" entries mixed with the
1102
# "contents-of-root block". But we don't want an if check on
1103
# all entries, so instead we just fix it up here.
1104
if self._dirblocks[1] != ('', []):
1105
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1107
contents_of_root_block = []
1108
for entry in self._dirblocks[0][1]:
1109
if not entry[0][1]: # This is a root entry
1110
root_block.append(entry)
1112
contents_of_root_block.append(entry)
1113
self._dirblocks[0] = ('', root_block)
1114
self._dirblocks[1] = ('', contents_of_root_block)
1116
def _entries_for_path(self, path):
1117
"""Return a list with all the entries that match path for all ids."""
1118
dirname, basename = os.path.split(path)
1119
key = (dirname, basename, '')
1120
block_index, present = self._find_block_index_from_key(key)
1122
# the block which should contain path is absent.
1125
block = self._dirblocks[block_index][1]
1126
entry_index, _ = self._find_entry_index(key, block)
1127
# we may need to look at multiple entries at this path: walk while the specific_files match.
1128
while (entry_index < len(block) and
1129
block[entry_index][0][0:2] == key[0:2]):
1130
result.append(block[entry_index])
1134
def _entry_to_line(self, entry):
1135
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1137
:param entry: An entry_tuple as defined in the module docstring.
1139
entire_entry = list(entry[0])
1140
for tree_number, tree_data in enumerate(entry[1]):
1141
# (minikind, fingerprint, size, executable, tree_specific_string)
1142
entire_entry.extend(tree_data)
1143
# 3 for the key, 5 for the fields per tree.
1144
tree_offset = 3 + tree_number * 5
1146
entire_entry[tree_offset + 0] = tree_data[0]
1148
entire_entry[tree_offset + 2] = str(tree_data[2])
1150
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1151
return '\0'.join(entire_entry)
1153
def _fields_per_entry(self):
1154
"""How many null separated fields should be in each entry row.
1156
Each line now has an extra '\n' field which is not used
1157
so we just skip over it
1159
3 fields for the key
1160
+ number of fields per tree_data (5) * tree count
1163
tree_count = 1 + self._num_present_parents()
1164
return 3 + 5 * tree_count + 1
1166
def _find_block(self, key, add_if_missing=False):
1167
"""Return the block that key should be present in.
1169
:param key: A dirstate entry key.
1170
:return: The block tuple.
1172
block_index, present = self._find_block_index_from_key(key)
1174
if not add_if_missing:
1175
# check to see if key is versioned itself - we might want to
1176
# add it anyway, because dirs with no entries dont get a
1177
# dirblock at parse time.
1178
# This is an uncommon branch to take: most dirs have children,
1179
# and most code works with versioned paths.
1180
parent_base, parent_name = osutils.split(key[0])
1181
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1182
# some parent path has not been added - its an error to add
1184
raise errors.NotVersionedError(key[0:2], str(self))
1185
self._dirblocks.insert(block_index, (key[0], []))
1186
return self._dirblocks[block_index]
1188
def _find_block_index_from_key(self, key):
1189
"""Find the dirblock index for a key.
1191
:return: The block index, True if the block for the key is present.
1193
if key[0:2] == ('', ''):
1196
if (self._last_block_index is not None and
1197
self._dirblocks[self._last_block_index][0] == key[0]):
1198
return self._last_block_index, True
1201
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1202
cache=self._split_path_cache)
1203
# _right returns one-past-where-key is so we have to subtract
1204
# one to use it. we use _right here because there are two
1205
# '' blocks - the root, and the contents of root
1206
# we always have a minimum of 2 in self._dirblocks: root and
1207
# root-contents, and for '', we get 2 back, so this is
1208
# simple and correct:
1209
present = (block_index < len(self._dirblocks) and
1210
self._dirblocks[block_index][0] == key[0])
1211
self._last_block_index = block_index
1212
# Reset the entry index cache to the beginning of the block.
1213
self._last_entry_index = -1
1214
return block_index, present
1216
def _find_entry_index(self, key, block):
1217
"""Find the entry index for a key in a block.
1219
:return: The entry index, True if the entry for the key is present.
1221
len_block = len(block)
1223
if self._last_entry_index is not None:
1225
entry_index = self._last_entry_index + 1
1226
# A hit is when the key is after the last slot, and before or
1227
# equal to the next slot.
1228
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1229
key <= block[entry_index][0]):
1230
self._last_entry_index = entry_index
1231
present = (block[entry_index][0] == key)
1232
return entry_index, present
1235
entry_index = bisect.bisect_left(block, (key, []))
1236
present = (entry_index < len_block and
1237
block[entry_index][0] == key)
1238
self._last_entry_index = entry_index
1239
return entry_index, present
1242
def from_tree(tree, dir_state_filename, sha1_provider=None):
1243
"""Create a dirstate from a bzr Tree.
1245
:param tree: The tree which should provide parent information and
1247
:param sha1_provider: an object meeting the SHA1Provider interface.
1248
If None, a DefaultSHA1Provider is used.
1249
:return: a DirState object which is currently locked for writing.
1250
(it was locked by DirState.initialize)
1252
result = DirState.initialize(dir_state_filename,
1253
sha1_provider=sha1_provider)
1257
parent_ids = tree.get_parent_ids()
1258
num_parents = len(parent_ids)
1260
for parent_id in parent_ids:
1261
parent_tree = tree.branch.repository.revision_tree(parent_id)
1262
parent_trees.append((parent_id, parent_tree))
1263
parent_tree.lock_read()
1264
result.set_parent_trees(parent_trees, [])
1265
result.set_state_from_inventory(tree.inventory)
1267
for revid, parent_tree in parent_trees:
1268
parent_tree.unlock()
1271
# The caller won't have a chance to unlock this, so make sure we
1277
def update_by_delta(self, delta):
1278
"""Apply an inventory delta to the dirstate for tree 0
1280
:param delta: An inventory delta. See Inventory.apply_delta for
1283
self._read_dirblocks_if_needed()
1286
for old_path, new_path, file_id, inv_entry in sorted(delta, reverse=True):
1287
if (file_id in insertions) or (file_id in removals):
1288
raise errors.InconsistentDelta(old_path or new_path, file_id,
1290
if old_path is not None:
1291
old_path = old_path.encode('utf-8')
1292
removals[file_id] = old_path
1293
if new_path is not None:
1294
new_path = new_path.encode('utf-8')
1295
dirname, basename = osutils.split(new_path)
1296
key = (dirname, basename, file_id)
1297
minikind = DirState._kind_to_minikind[inv_entry.kind]
1299
fingerprint = inv_entry.reference_revision
1302
insertions[file_id] = (key, minikind, inv_entry.executable,
1303
fingerprint, new_path)
1304
# Transform moves into delete+add pairs
1305
if None not in (old_path, new_path):
1306
for child in self._iter_child_entries(0, old_path):
1307
if child[0][2] in insertions or child[0][2] in removals:
1309
child_dirname = child[0][0]
1310
child_basename = child[0][1]
1311
minikind = child[1][0][0]
1312
fingerprint = child[1][0][4]
1313
executable = child[1][0][3]
1314
old_child_path = osutils.pathjoin(child[0][0],
1316
removals[child[0][2]] = old_child_path
1317
child_suffix = child_dirname[len(old_path):]
1318
new_child_dirname = (new_path + child_suffix)
1319
key = (new_child_dirname, child_basename, child[0][2])
1320
new_child_path = os.path.join(new_child_dirname,
1322
insertions[child[0][2]] = (key, minikind, executable,
1323
fingerprint, new_child_path)
1324
self._apply_removals(removals.values())
1325
self._apply_insertions(insertions.values())
1327
def _apply_removals(self, removals):
1328
for path in sorted(removals, reverse=True):
1329
dirname, basename = osutils.split(path)
1330
block_i, entry_i, d_present, f_present = \
1331
self._get_block_entry_index(dirname, basename, 0)
1332
entry = self._dirblocks[block_i][1][entry_i]
1333
self._make_absent(entry)
1334
# See if we have a malformed delta: deleting a directory must not
1335
# leave crud behind. This increases the number of bisects needed
1336
# substantially, but deletion or renames of large numbers of paths
1337
# is rare enough it shouldn't be an issue (famous last words?) RBC
1339
block_i, entry_i, d_present, f_present = \
1340
self._get_block_entry_index(path, '', 0)
1342
# The dir block is still present in the dirstate; this could
1343
# be due to it being in a parent tree, or a corrupt delta.
1344
for child_entry in self._dirblocks[block_i][1]:
1345
if child_entry[1][0][0] not in ('r', 'a'):
1346
raise errors.InconsistentDelta(path, entry[0][2],
1347
"The file id was deleted but its children were "
1350
def _apply_insertions(self, adds):
1351
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1352
self.update_minimal(key, minikind, executable, fingerprint,
1353
path_utf8=path_utf8)
1355
def update_basis_by_delta(self, delta, new_revid):
1356
"""Update the parents of this tree after a commit.
1358
This gives the tree one parent, with revision id new_revid. The
1359
inventory delta is applied to the current basis tree to generate the
1360
inventory for the parent new_revid, and all other parent trees are
1363
Note that an exception during the operation of this method will leave
1364
the dirstate in a corrupt state where it should not be saved.
1366
Finally, we expect all changes to be synchronising the basis tree with
1369
:param new_revid: The new revision id for the trees parent.
1370
:param delta: An inventory delta (see apply_inventory_delta) describing
1371
the changes from the current left most parent revision to new_revid.
1373
self._read_dirblocks_if_needed()
1374
self._discard_merge_parents()
1375
if self._ghosts != []:
1376
raise NotImplementedError(self.update_basis_by_delta)
1377
if len(self._parents) == 0:
1378
# setup a blank tree, the most simple way.
1379
empty_parent = DirState.NULL_PARENT_DETAILS
1380
for entry in self._iter_entries():
1381
entry[1].append(empty_parent)
1382
self._parents.append(new_revid)
1384
self._parents[0] = new_revid
1386
delta = sorted(delta, reverse=True)
1390
# The paths this function accepts are unicode and must be encoded as we
1392
encode = cache_utf8.encode
1393
inv_to_entry = self._inv_entry_to_details
1394
# delta is now (deletes, changes), (adds) in reverse lexographical
1396
# deletes in reverse lexographic order are safe to process in situ.
1397
# renames are not, as a rename from any path could go to a path
1398
# lexographically lower, so we transform renames into delete, add pairs,
1399
# expanding them recursively as needed.
1400
# At the same time, to reduce interface friction we convert the input
1401
# inventory entries to dirstate.
1402
root_only = ('', '')
1403
# Accumulate parent references (path and id), to check for parentless
1404
# items or items placed under files/links/tree-references.
1406
for old_path, new_path, file_id, inv_entry in delta:
1407
if inv_entry is not None and file_id != inv_entry.file_id:
1408
raise errors.InconsistentDelta(new_path, file_id,
1409
"mismatched entry file_id %r" % inv_entry)
1410
if new_path is not None:
1411
new_path_utf8 = encode(new_path)
1412
# note the parent for validation
1413
dirname, basename = osutils.split(new_path_utf8)
1415
parents.add((dirname, inv_entry.parent_id))
1416
if old_path is None:
1417
adds.append((None, encode(new_path), file_id,
1418
inv_to_entry(inv_entry), True))
1419
elif new_path is None:
1420
deletes.append((encode(old_path), None, file_id, None, True))
1421
elif (old_path, new_path) != root_only:
1423
# Because renames must preserve their children we must have
1424
# processed all relocations and removes before hand. The sort
1425
# order ensures we've examined the child paths, but we also
1426
# have to execute the removals, or the split to an add/delete
1427
# pair will result in the deleted item being reinserted, or
1428
# renamed items being reinserted twice - and possibly at the
1429
# wrong place. Splitting into a delete/add pair also simplifies
1430
# the handling of entries with ('f', ...), ('r' ...) because
1431
# the target of the 'r' is old_path here, and we add that to
1432
# deletes, meaning that the add handler does not need to check
1433
# for 'r' items on every pass.
1434
self._update_basis_apply_deletes(deletes)
1436
# Split into an add/delete pair recursively.
1437
adds.append((None, new_path_utf8, file_id,
1438
inv_to_entry(inv_entry), False))
1439
# Expunge deletes that we've seen so that deleted/renamed
1440
# children of a rename directory are handled correctly.
1441
new_deletes = reversed(list(self._iter_child_entries(1,
1443
# Remove the current contents of the tree at orig_path, and
1444
# reinsert at the correct new path.
1445
for entry in new_deletes:
1447
source_path = entry[0][0] + '/' + entry[0][1]
1449
source_path = entry[0][1]
1451
target_path = new_path_utf8 + source_path[len(old_path):]
1454
raise AssertionError("cannot rename directory to"
1456
target_path = source_path[len(old_path) + 1:]
1457
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1459
(source_path, target_path, entry[0][2], None, False))
1461
(encode(old_path), new_path, file_id, None, False))
1462
# note the parent for validation
1463
dirname, basename = osutils.split(new_path)
1464
parents.add((dirname, inv_entry.parent_id))
1466
# changes to just the root should not require remove/insertion
1468
changes.append((encode(old_path), encode(new_path), file_id,
1469
inv_to_entry(inv_entry)))
1472
# Finish expunging deletes/first half of renames.
1473
self._update_basis_apply_deletes(deletes)
1474
# Reinstate second half of renames and new paths.
1475
self._update_basis_apply_adds(adds)
1476
# Apply in-situ changes.
1477
self._update_basis_apply_changes(changes)
1479
self._update_basis_check_parents(parents)
1480
except errors.BzrError, e:
1481
if 'integrity error' not in str(e):
1483
# _get_entry raises BzrError when a request is inconsistent; we
1484
# want such errors to be shown as InconsistentDelta - and that
1485
# fits the behaviour we trigger. Partof this is driven by dirstate
1486
# only supporting deltas that turn the basis into a closer fit to
1488
self._changes_aborted = True
1489
raise errors.InconsistentDeltaDelta(delta, "error from _get_entry.")
1491
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1492
self._header_state = DirState.IN_MEMORY_MODIFIED
1493
self._id_index = None
1496
def _update_basis_apply_adds(self, adds):
1497
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1499
They may be adds, or renames that have been split into add/delete
1502
:param adds: A sequence of adds. Each add is a tuple:
1503
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1504
is False when the add is the second half of a remove-and-reinsert
1505
pair created to handle renames and deletes.
1507
# Adds are accumulated partly from renames, so can be in any input
1510
# adds is now in lexographic order, which places all parents before
1511
# their children, so we can process it linearly.
1513
for old_path, new_path, file_id, new_details, real_add in adds:
1514
# the entry for this file_id must be in tree 0.
1515
entry = self._get_entry(0, file_id, new_path)
1516
if entry[0] is None or entry[0][2] != file_id:
1517
self._changes_aborted = True
1518
raise errors.InconsistentDelta(new_path, file_id,
1519
'working tree does not contain new entry')
1520
if real_add and entry[1][1][0] not in absent:
1521
self._changes_aborted = True
1522
raise errors.InconsistentDelta(new_path, file_id,
1523
'The entry was considered to be a genuinely new record,'
1524
' but there was already an old record for it.')
1525
# We don't need to update the target of an 'r' because the handling
1526
# of renames turns all 'r' situations into a delete at the original
1528
entry[1][1] = new_details
1530
def _update_basis_apply_changes(self, changes):
1531
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1533
:param adds: A sequence of changes. Each change is a tuple:
1534
(path_utf8, path_utf8, file_id, (entry_details))
1537
for old_path, new_path, file_id, new_details in changes:
1538
# the entry for this file_id must be in tree 0.
1539
entry = self._get_entry(0, file_id, new_path)
1540
if entry[0] is None or entry[0][2] != file_id:
1541
self._changes_aborted = True
1542
raise errors.InconsistentDelta(new_path, file_id,
1543
'working tree does not contain new entry')
1544
if (entry[1][0][0] in absent or
1545
entry[1][1][0] in absent):
1546
self._changes_aborted = True
1547
raise errors.InconsistentDelta(new_path, file_id,
1548
'changed considered absent')
1549
entry[1][1] = new_details
1551
def _update_basis_apply_deletes(self, deletes):
1552
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1554
They may be deletes, or renames that have been split into add/delete
1557
:param deletes: A sequence of deletes. Each delete is a tuple:
1558
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1559
real_delete is True when the desired outcome is an actual deletion
1560
rather than the rename handling logic temporarily deleting a path
1561
during the replacement of a parent.
1563
null = DirState.NULL_PARENT_DETAILS
1564
for old_path, new_path, file_id, _, real_delete in deletes:
1565
if real_delete != (new_path is None):
1566
raise AssertionError("bad delete delta")
1567
# the entry for this file_id must be in tree 1.
1568
dirname, basename = osutils.split(old_path)
1569
block_index, entry_index, dir_present, file_present = \
1570
self._get_block_entry_index(dirname, basename, 1)
1571
if not file_present:
1572
self._changes_aborted = True
1573
raise errors.InconsistentDelta(old_path, file_id,
1574
'basis tree does not contain removed entry')
1575
entry = self._dirblocks[block_index][1][entry_index]
1576
if entry[0][2] != file_id:
1577
self._changes_aborted = True
1578
raise errors.InconsistentDelta(old_path, file_id,
1579
'mismatched file_id in tree 1')
1581
if entry[1][0][0] != 'a':
1582
self._changes_aborted = True
1583
raise errors.InconsistentDelta(old_path, file_id,
1584
'This was marked as a real delete, but the WT state'
1585
' claims that it still exists and is versioned.')
1586
del self._dirblocks[block_index][1][entry_index]
1588
if entry[1][0][0] == 'a':
1589
self._changes_aborted = True
1590
raise errors.InconsistentDelta(old_path, file_id,
1591
'The entry was considered a rename, but the source path'
1592
' is marked as absent.')
1593
# For whatever reason, we were asked to rename an entry
1594
# that was originally marked as deleted. This could be
1595
# because we are renaming the parent directory, and the WT
1596
# current state has the file marked as deleted.
1597
elif entry[1][0][0] == 'r':
1598
# implement the rename
1599
del self._dirblocks[block_index][1][entry_index]
1601
# it is being resurrected here, so blank it out temporarily.
1602
self._dirblocks[block_index][1][entry_index][1][1] = null
1604
def _update_basis_check_parents(self, parents):
1605
"""Check that parents required by the delta are all intact."""
1606
for dirname, file_id in parents:
1607
# Get the entry - the ensures that file_id, dirname exists and has
1608
# the right file id.
1609
entry = self._get_entry(1, file_id, dirname)
1610
if entry[1] is None:
1611
self._changes_aborted = True
1612
raise errors.InconsistentDelta(dirname, file_id,
1613
"This parent is not present.")
1614
# Parents of things must be directories
1615
if entry[1][1][0] != 'd':
1616
self._changes_aborted = True
1617
raise errors.InconsistentDelta(dirname, file_id,
1618
"This parent is not a directory.")
1620
def _observed_sha1(self, entry, sha1, stat_value,
1621
_stat_to_minikind=_stat_to_minikind, _pack_stat=pack_stat):
1622
"""Note the sha1 of a file.
1624
:param entry: The entry the sha1 is for.
1625
:param sha1: The observed sha1.
1626
:param stat_value: The os.lstat for the file.
1629
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1633
packed_stat = _pack_stat(stat_value)
1635
if self._cutoff_time is None:
1636
self._sha_cutoff_time()
1637
if (stat_value.st_mtime < self._cutoff_time
1638
and stat_value.st_ctime < self._cutoff_time):
1639
entry[1][0] = ('f', sha1, entry[1][0][2], entry[1][0][3],
1641
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1643
def _sha_cutoff_time(self):
1644
"""Return cutoff time.
1646
Files modified more recently than this time are at risk of being
1647
undetectably modified and so can't be cached.
1649
# Cache the cutoff time as long as we hold a lock.
1650
# time.time() isn't super expensive (approx 3.38us), but
1651
# when you call it 50,000 times it adds up.
1652
# For comparison, os.lstat() costs 7.2us if it is hot.
1653
self._cutoff_time = int(time.time()) - 3
1654
return self._cutoff_time
1656
def _lstat(self, abspath, entry):
1657
"""Return the os.lstat value for this path."""
1658
return os.lstat(abspath)
1660
def _sha1_file_and_mutter(self, abspath):
1661
# when -Dhashcache is turned on, this is monkey-patched in to log
1663
trace.mutter("dirstate sha1 " + abspath)
1664
return self._sha1_provider.sha1(abspath)
1666
def _is_executable(self, mode, old_executable):
1667
"""Is this file executable?"""
1668
return bool(S_IEXEC & mode)
1670
def _is_executable_win32(self, mode, old_executable):
1671
"""On win32 the executable bit is stored in the dirstate."""
1672
return old_executable
1674
if sys.platform == 'win32':
1675
_is_executable = _is_executable_win32
1677
def _read_link(self, abspath, old_link):
1678
"""Read the target of a symlink"""
1679
# TODO: jam 200700301 On Win32, this could just return the value
1680
# already in memory. However, this really needs to be done at a
1681
# higher level, because there either won't be anything on disk,
1682
# or the thing on disk will be a file.
1683
fs_encoding = osutils._fs_enc
1684
if isinstance(abspath, unicode):
1685
# abspath is defined as the path to pass to lstat. readlink is
1686
# buggy in python < 2.6 (it doesn't encode unicode path into FS
1687
# encoding), so we need to encode ourselves knowing that unicode
1688
# paths are produced by UnicodeDirReader on purpose.
1689
abspath = abspath.encode(fs_encoding)
1690
target = os.readlink(abspath)
1691
if fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1692
# Change encoding if needed
1693
target = target.decode(fs_encoding).encode('UTF-8')
1696
def get_ghosts(self):
1697
"""Return a list of the parent tree revision ids that are ghosts."""
1698
self._read_header_if_needed()
1701
def get_lines(self):
1702
"""Serialise the entire dirstate to a sequence of lines."""
1703
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1704
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1705
# read whats on disk.
1706
self._state_file.seek(0)
1707
return self._state_file.readlines()
1709
lines.append(self._get_parents_line(self.get_parent_ids()))
1710
lines.append(self._get_ghosts_line(self._ghosts))
1711
# append the root line which is special cased
1712
lines.extend(map(self._entry_to_line, self._iter_entries()))
1713
return self._get_output_lines(lines)
1715
def _get_ghosts_line(self, ghost_ids):
1716
"""Create a line for the state file for ghost information."""
1717
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1719
def _get_parents_line(self, parent_ids):
1720
"""Create a line for the state file for parents information."""
1721
return '\0'.join([str(len(parent_ids))] + parent_ids)
1723
def _get_fields_to_entry(self):
1724
"""Get a function which converts entry fields into a entry record.
1726
This handles size and executable, as well as parent records.
1728
:return: A function which takes a list of fields, and returns an
1729
appropriate record for storing in memory.
1731
# This is intentionally unrolled for performance
1732
num_present_parents = self._num_present_parents()
1733
if num_present_parents == 0:
1734
def fields_to_entry_0_parents(fields, _int=int):
1735
path_name_file_id_key = (fields[0], fields[1], fields[2])
1736
return (path_name_file_id_key, [
1738
fields[3], # minikind
1739
fields[4], # fingerprint
1740
_int(fields[5]), # size
1741
fields[6] == 'y', # executable
1742
fields[7], # packed_stat or revision_id
1744
return fields_to_entry_0_parents
1745
elif num_present_parents == 1:
1746
def fields_to_entry_1_parent(fields, _int=int):
1747
path_name_file_id_key = (fields[0], fields[1], fields[2])
1748
return (path_name_file_id_key, [
1750
fields[3], # minikind
1751
fields[4], # fingerprint
1752
_int(fields[5]), # size
1753
fields[6] == 'y', # executable
1754
fields[7], # packed_stat or revision_id
1757
fields[8], # minikind
1758
fields[9], # fingerprint
1759
_int(fields[10]), # size
1760
fields[11] == 'y', # executable
1761
fields[12], # packed_stat or revision_id
1764
return fields_to_entry_1_parent
1765
elif num_present_parents == 2:
1766
def fields_to_entry_2_parents(fields, _int=int):
1767
path_name_file_id_key = (fields[0], fields[1], fields[2])
1768
return (path_name_file_id_key, [
1770
fields[3], # minikind
1771
fields[4], # fingerprint
1772
_int(fields[5]), # size
1773
fields[6] == 'y', # executable
1774
fields[7], # packed_stat or revision_id
1777
fields[8], # minikind
1778
fields[9], # fingerprint
1779
_int(fields[10]), # size
1780
fields[11] == 'y', # executable
1781
fields[12], # packed_stat or revision_id
1784
fields[13], # minikind
1785
fields[14], # fingerprint
1786
_int(fields[15]), # size
1787
fields[16] == 'y', # executable
1788
fields[17], # packed_stat or revision_id
1791
return fields_to_entry_2_parents
1793
def fields_to_entry_n_parents(fields, _int=int):
1794
path_name_file_id_key = (fields[0], fields[1], fields[2])
1795
trees = [(fields[cur], # minikind
1796
fields[cur+1], # fingerprint
1797
_int(fields[cur+2]), # size
1798
fields[cur+3] == 'y', # executable
1799
fields[cur+4], # stat or revision_id
1800
) for cur in xrange(3, len(fields)-1, 5)]
1801
return path_name_file_id_key, trees
1802
return fields_to_entry_n_parents
1804
def get_parent_ids(self):
1805
"""Return a list of the parent tree ids for the directory state."""
1806
self._read_header_if_needed()
1807
return list(self._parents)
1809
def _get_block_entry_index(self, dirname, basename, tree_index):
1810
"""Get the coordinates for a path in the state structure.
1812
:param dirname: The utf8 dirname to lookup.
1813
:param basename: The utf8 basename to lookup.
1814
:param tree_index: The index of the tree for which this lookup should
1816
:return: A tuple describing where the path is located, or should be
1817
inserted. The tuple contains four fields: the block index, the row
1818
index, the directory is present (boolean), the entire path is
1819
present (boolean). There is no guarantee that either
1820
coordinate is currently reachable unless the found field for it is
1821
True. For instance, a directory not present in the searched tree
1822
may be returned with a value one greater than the current highest
1823
block offset. The directory present field will always be True when
1824
the path present field is True. The directory present field does
1825
NOT indicate that the directory is present in the searched tree,
1826
rather it indicates that there are at least some files in some
1829
self._read_dirblocks_if_needed()
1830
key = dirname, basename, ''
1831
block_index, present = self._find_block_index_from_key(key)
1833
# no such directory - return the dir index and 0 for the row.
1834
return block_index, 0, False, False
1835
block = self._dirblocks[block_index][1] # access the entries only
1836
entry_index, present = self._find_entry_index(key, block)
1837
# linear search through entries at this path to find the one
1839
while entry_index < len(block) and block[entry_index][0][1] == basename:
1840
if block[entry_index][1][tree_index][0] not in 'ar':
1841
# neither absent or relocated
1842
return block_index, entry_index, True, True
1844
return block_index, entry_index, True, False
1846
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None, include_deleted=False):
1847
"""Get the dirstate entry for path in tree tree_index.
1849
If either file_id or path is supplied, it is used as the key to lookup.
1850
If both are supplied, the fastest lookup is used, and an error is
1851
raised if they do not both point at the same row.
1853
:param tree_index: The index of the tree we wish to locate this path
1854
in. If the path is present in that tree, the entry containing its
1855
details is returned, otherwise (None, None) is returned
1856
0 is the working tree, higher indexes are successive parent
1858
:param fileid_utf8: A utf8 file_id to look up.
1859
:param path_utf8: An utf8 path to be looked up.
1860
:param include_deleted: If True, and performing a lookup via
1861
fileid_utf8 rather than path_utf8, return an entry for deleted
1863
:return: The dirstate entry tuple for path, or (None, None)
1865
self._read_dirblocks_if_needed()
1866
if path_utf8 is not None:
1867
if type(path_utf8) is not str:
1868
raise AssertionError('path_utf8 is not a str: %s %s'
1869
% (type(path_utf8), path_utf8))
1870
# path lookups are faster
1871
dirname, basename = osutils.split(path_utf8)
1872
block_index, entry_index, dir_present, file_present = \
1873
self._get_block_entry_index(dirname, basename, tree_index)
1874
if not file_present:
1876
entry = self._dirblocks[block_index][1][entry_index]
1877
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
1878
raise AssertionError('unversioned entry?')
1880
if entry[0][2] != fileid_utf8:
1881
self._changes_aborted = True
1882
raise errors.BzrError('integrity error ? : mismatching'
1883
' tree_index, file_id and path')
1886
possible_keys = self._get_id_index().get(fileid_utf8, None)
1887
if not possible_keys:
1889
for key in possible_keys:
1890
block_index, present = \
1891
self._find_block_index_from_key(key)
1892
# strange, probably indicates an out of date
1893
# id index - for now, allow this.
1896
# WARNING: DO not change this code to use _get_block_entry_index
1897
# as that function is not suitable: it does not use the key
1898
# to lookup, and thus the wrong coordinates are returned.
1899
block = self._dirblocks[block_index][1]
1900
entry_index, present = self._find_entry_index(key, block)
1902
entry = self._dirblocks[block_index][1][entry_index]
1903
if entry[1][tree_index][0] in 'fdlt':
1904
# this is the result we are looking for: the
1905
# real home of this file_id in this tree.
1907
if entry[1][tree_index][0] == 'a':
1908
# there is no home for this entry in this tree
1912
if entry[1][tree_index][0] != 'r':
1913
raise AssertionError(
1914
"entry %r has invalid minikind %r for tree %r" \
1916
entry[1][tree_index][0],
1918
real_path = entry[1][tree_index][1]
1919
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1920
path_utf8=real_path)
1924
def initialize(cls, path, sha1_provider=None):
1925
"""Create a new dirstate on path.
1927
The new dirstate will be an empty tree - that is it has no parents,
1928
and only a root node - which has id ROOT_ID.
1930
:param path: The name of the file for the dirstate.
1931
:param sha1_provider: an object meeting the SHA1Provider interface.
1932
If None, a DefaultSHA1Provider is used.
1933
:return: A write-locked DirState object.
1935
# This constructs a new DirState object on a path, sets the _state_file
1936
# to a new empty file for that path. It then calls _set_data() with our
1937
# stock empty dirstate information - a root with ROOT_ID, no children,
1938
# and no parents. Finally it calls save() to ensure that this data will
1940
if sha1_provider is None:
1941
sha1_provider = DefaultSHA1Provider()
1942
result = cls(path, sha1_provider)
1943
# root dir and root dir contents with no children.
1944
empty_tree_dirblocks = [('', []), ('', [])]
1945
# a new root directory, with a NULLSTAT.
1946
empty_tree_dirblocks[0][1].append(
1947
(('', '', inventory.ROOT_ID), [
1948
('d', '', 0, False, DirState.NULLSTAT),
1952
result._set_data([], empty_tree_dirblocks)
1960
def _inv_entry_to_details(inv_entry):
1961
"""Convert an inventory entry (from a revision tree) to state details.
1963
:param inv_entry: An inventory entry whose sha1 and link targets can be
1964
relied upon, and which has a revision set.
1965
:return: A details tuple - the details for a single tree at a path +
1968
kind = inv_entry.kind
1969
minikind = DirState._kind_to_minikind[kind]
1970
tree_data = inv_entry.revision
1971
if kind == 'directory':
1975
elif kind == 'symlink':
1976
if inv_entry.symlink_target is None:
1979
fingerprint = inv_entry.symlink_target.encode('utf8')
1982
elif kind == 'file':
1983
fingerprint = inv_entry.text_sha1 or ''
1984
size = inv_entry.text_size or 0
1985
executable = inv_entry.executable
1986
elif kind == 'tree-reference':
1987
fingerprint = inv_entry.reference_revision or ''
1991
raise Exception("can't pack %s" % inv_entry)
1992
return (minikind, fingerprint, size, executable, tree_data)
1994
def _iter_child_entries(self, tree_index, path_utf8):
1995
"""Iterate over all the entries that are children of path_utf.
1997
This only returns entries that are present (not in 'a', 'r') in
1998
tree_index. tree_index data is not refreshed, so if tree 0 is used,
1999
results may differ from that obtained if paths were statted to
2000
determine what ones were directories.
2002
Asking for the children of a non-directory will return an empty
2006
next_pending_dirs = [path_utf8]
2008
while next_pending_dirs:
2009
pending_dirs = next_pending_dirs
2010
next_pending_dirs = []
2011
for path in pending_dirs:
2012
block_index, present = self._find_block_index_from_key(
2014
if block_index == 0:
2016
if len(self._dirblocks) == 1:
2017
# asked for the children of the root with no other
2021
# children of a non-directory asked for.
2023
block = self._dirblocks[block_index]
2024
for entry in block[1]:
2025
kind = entry[1][tree_index][0]
2026
if kind not in absent:
2030
path = entry[0][0] + '/' + entry[0][1]
2033
next_pending_dirs.append(path)
2035
def _iter_entries(self):
2036
"""Iterate over all the entries in the dirstate.
2038
Each yelt item is an entry in the standard format described in the
2039
docstring of bzrlib.dirstate.
2041
self._read_dirblocks_if_needed()
2042
for directory in self._dirblocks:
2043
for entry in directory[1]:
2046
def _get_id_index(self):
2047
"""Get an id index of self._dirblocks."""
2048
if self._id_index is None:
2050
for key, tree_details in self._iter_entries():
2051
id_index.setdefault(key[2], set()).add(key)
2052
self._id_index = id_index
2053
return self._id_index
2055
def _get_output_lines(self, lines):
2056
"""Format lines for final output.
2058
:param lines: A sequence of lines containing the parents list and the
2061
output_lines = [DirState.HEADER_FORMAT_3]
2062
lines.append('') # a final newline
2063
inventory_text = '\0\n\0'.join(lines)
2064
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
2065
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
2066
num_entries = len(lines)-3
2067
output_lines.append('num_entries: %s\n' % (num_entries,))
2068
output_lines.append(inventory_text)
2071
def _make_deleted_row(self, fileid_utf8, parents):
2072
"""Return a deleted row for fileid_utf8."""
2073
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
2076
def _num_present_parents(self):
2077
"""The number of parent entries in each record row."""
2078
return len(self._parents) - len(self._ghosts)
2081
def on_file(path, sha1_provider=None):
2082
"""Construct a DirState on the file at path "path".
2084
:param path: The path at which the dirstate file on disk should live.
2085
:param sha1_provider: an object meeting the SHA1Provider interface.
2086
If None, a DefaultSHA1Provider is used.
2087
:return: An unlocked DirState object, associated with the given path.
2089
if sha1_provider is None:
2090
sha1_provider = DefaultSHA1Provider()
2091
result = DirState(path, sha1_provider)
2094
def _read_dirblocks_if_needed(self):
2095
"""Read in all the dirblocks from the file if they are not in memory.
2097
This populates self._dirblocks, and sets self._dirblock_state to
2098
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
2101
self._read_header_if_needed()
2102
if self._dirblock_state == DirState.NOT_IN_MEMORY:
2103
_read_dirblocks(self)
2105
def _read_header(self):
2106
"""This reads in the metadata header, and the parent ids.
2108
After reading in, the file should be positioned at the null
2109
just before the start of the first record in the file.
2111
:return: (expected crc checksum, number of entries, parent list)
2113
self._read_prelude()
2114
parent_line = self._state_file.readline()
2115
info = parent_line.split('\0')
2116
num_parents = int(info[0])
2117
self._parents = info[1:-1]
2118
ghost_line = self._state_file.readline()
2119
info = ghost_line.split('\0')
2120
num_ghosts = int(info[1])
2121
self._ghosts = info[2:-1]
2122
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2123
self._end_of_header = self._state_file.tell()
2125
def _read_header_if_needed(self):
2126
"""Read the header of the dirstate file if needed."""
2127
# inline this as it will be called a lot
2128
if not self._lock_token:
2129
raise errors.ObjectNotLocked(self)
2130
if self._header_state == DirState.NOT_IN_MEMORY:
2133
def _read_prelude(self):
2134
"""Read in the prelude header of the dirstate file.
2136
This only reads in the stuff that is not connected to the crc
2137
checksum. The position will be correct to read in the rest of
2138
the file and check the checksum after this point.
2139
The next entry in the file should be the number of parents,
2140
and their ids. Followed by a newline.
2142
header = self._state_file.readline()
2143
if header != DirState.HEADER_FORMAT_3:
2144
raise errors.BzrError(
2145
'invalid header line: %r' % (header,))
2146
crc_line = self._state_file.readline()
2147
if not crc_line.startswith('crc32: '):
2148
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2149
self.crc_expected = int(crc_line[len('crc32: '):-1])
2150
num_entries_line = self._state_file.readline()
2151
if not num_entries_line.startswith('num_entries: '):
2152
raise errors.BzrError('missing num_entries line')
2153
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2155
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
2156
"""Find a sha1 given a stat lookup."""
2157
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
2159
def _get_packed_stat_index(self):
2160
"""Get a packed_stat index of self._dirblocks."""
2161
if self._packed_stat_index is None:
2163
for key, tree_details in self._iter_entries():
2164
if tree_details[0][0] == 'f':
2165
index[tree_details[0][4]] = tree_details[0][1]
2166
self._packed_stat_index = index
2167
return self._packed_stat_index
2170
"""Save any pending changes created during this session.
2172
We reuse the existing file, because that prevents race conditions with
2173
file creation, and use oslocks on it to prevent concurrent modification
2174
and reads - because dirstate's incremental data aggregation is not
2175
compatible with reading a modified file, and replacing a file in use by
2176
another process is impossible on Windows.
2178
A dirstate in read only mode should be smart enough though to validate
2179
that the file has not changed, and otherwise discard its cache and
2180
start over, to allow for fine grained read lock duration, so 'status'
2181
wont block 'commit' - for example.
2183
if self._changes_aborted:
2184
# Should this be a warning? For now, I'm expecting that places that
2185
# mark it inconsistent will warn, making a warning here redundant.
2186
trace.mutter('Not saving DirState because '
2187
'_changes_aborted is set.')
2189
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
2190
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2192
grabbed_write_lock = False
2193
if self._lock_state != 'w':
2194
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2195
# Switch over to the new lock, as the old one may be closed.
2196
# TODO: jam 20070315 We should validate the disk file has
2197
# not changed contents. Since temporary_write_lock may
2198
# not be an atomic operation.
2199
self._lock_token = new_lock
2200
self._state_file = new_lock.f
2201
if not grabbed_write_lock:
2202
# We couldn't grab a write lock, so we switch back to a read one
2205
self._state_file.seek(0)
2206
self._state_file.writelines(self.get_lines())
2207
self._state_file.truncate()
2208
self._state_file.flush()
2209
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2210
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
2212
if grabbed_write_lock:
2213
self._lock_token = self._lock_token.restore_read_lock()
2214
self._state_file = self._lock_token.f
2215
# TODO: jam 20070315 We should validate the disk file has
2216
# not changed contents. Since restore_read_lock may
2217
# not be an atomic operation.
2219
def _set_data(self, parent_ids, dirblocks):
2220
"""Set the full dirstate data in memory.
2222
This is an internal function used to completely replace the objects
2223
in memory state. It puts the dirstate into state 'full-dirty'.
2225
:param parent_ids: A list of parent tree revision ids.
2226
:param dirblocks: A list containing one tuple for each directory in the
2227
tree. Each tuple contains the directory path and a list of entries
2228
found in that directory.
2230
# our memory copy is now authoritative.
2231
self._dirblocks = dirblocks
2232
self._header_state = DirState.IN_MEMORY_MODIFIED
2233
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2234
self._parents = list(parent_ids)
2235
self._id_index = None
2236
self._packed_stat_index = None
2238
def set_path_id(self, path, new_id):
2239
"""Change the id of path to new_id in the current working tree.
2241
:param path: The path inside the tree to set - '' is the root, 'foo'
2242
is the path foo in the root.
2243
:param new_id: The new id to assign to the path. This must be a utf8
2244
file id (not unicode, and not None).
2246
self._read_dirblocks_if_needed()
2248
# TODO: logic not written
2249
raise NotImplementedError(self.set_path_id)
2250
# TODO: check new id is unique
2251
entry = self._get_entry(0, path_utf8=path)
2252
if entry[0][2] == new_id:
2253
# Nothing to change.
2255
# mark the old path absent, and insert a new root path
2256
self._make_absent(entry)
2257
self.update_minimal(('', '', new_id), 'd',
2258
path_utf8='', packed_stat=entry[1][0][4])
2259
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2260
if self._id_index is not None:
2261
self._id_index.setdefault(new_id, set()).add(entry[0])
2263
def set_parent_trees(self, trees, ghosts):
2264
"""Set the parent trees for the dirstate.
2266
:param trees: A list of revision_id, tree tuples. tree must be provided
2267
even if the revision_id refers to a ghost: supply an empty tree in
2269
:param ghosts: A list of the revision_ids that are ghosts at the time
2272
# TODO: generate a list of parent indexes to preserve to save
2273
# processing specific parent trees. In the common case one tree will
2274
# be preserved - the left most parent.
2275
# TODO: if the parent tree is a dirstate, we might want to walk them
2276
# all by path in parallel for 'optimal' common-case performance.
2277
# generate new root row.
2278
self._read_dirblocks_if_needed()
2279
# TODO future sketch: Examine the existing parents to generate a change
2280
# map and then walk the new parent trees only, mapping them into the
2281
# dirstate. Walk the dirstate at the same time to remove unreferenced
2284
# sketch: loop over all entries in the dirstate, cherry picking
2285
# entries from the parent trees, if they are not ghost trees.
2286
# after we finish walking the dirstate, all entries not in the dirstate
2287
# are deletes, so we want to append them to the end as per the design
2288
# discussions. So do a set difference on ids with the parents to
2289
# get deletes, and add them to the end.
2290
# During the update process we need to answer the following questions:
2291
# - find other keys containing a fileid in order to create cross-path
2292
# links. We dont't trivially use the inventory from other trees
2293
# because this leads to either double touching, or to accessing
2295
# - find other keys containing a path
2296
# We accumulate each entry via this dictionary, including the root
2299
# we could do parallel iterators, but because file id data may be
2300
# scattered throughout, we dont save on index overhead: we have to look
2301
# at everything anyway. We can probably save cycles by reusing parent
2302
# data and doing an incremental update when adding an additional
2303
# parent, but for now the common cases are adding a new parent (merge),
2304
# and replacing completely (commit), and commit is more common: so
2305
# optimise merge later.
2307
# ---- start generation of full tree mapping data
2308
# what trees should we use?
2309
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2310
# how many trees do we end up with
2311
parent_count = len(parent_trees)
2313
# one: the current tree
2314
for entry in self._iter_entries():
2315
# skip entries not in the current tree
2316
if entry[1][0][0] in 'ar': # absent, relocated
2318
by_path[entry[0]] = [entry[1][0]] + \
2319
[DirState.NULL_PARENT_DETAILS] * parent_count
2320
id_index[entry[0][2]] = set([entry[0]])
2322
# now the parent trees:
2323
for tree_index, tree in enumerate(parent_trees):
2324
# the index is off by one, adjust it.
2325
tree_index = tree_index + 1
2326
# when we add new locations for a fileid we need these ranges for
2327
# any fileid in this tree as we set the by_path[id] to:
2328
# already_processed_tree_details + new_details + new_location_suffix
2329
# the suffix is from tree_index+1:parent_count+1.
2330
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2331
# now stitch in all the entries from this tree
2332
for path, entry in tree.inventory.iter_entries_by_dir():
2333
# here we process each trees details for each item in the tree.
2334
# we first update any existing entries for the id at other paths,
2335
# then we either create or update the entry for the id at the
2336
# right path, and finally we add (if needed) a mapping from
2337
# file_id to this path. We do it in this order to allow us to
2338
# avoid checking all known paths for the id when generating a
2339
# new entry at this path: by adding the id->path mapping last,
2340
# all the mappings are valid and have correct relocation
2341
# records where needed.
2342
file_id = entry.file_id
2343
path_utf8 = path.encode('utf8')
2344
dirname, basename = osutils.split(path_utf8)
2345
new_entry_key = (dirname, basename, file_id)
2346
# tree index consistency: All other paths for this id in this tree
2347
# index must point to the correct path.
2348
for entry_key in id_index.setdefault(file_id, set()):
2349
# TODO:PROFILING: It might be faster to just update
2350
# rather than checking if we need to, and then overwrite
2351
# the one we are located at.
2352
if entry_key != new_entry_key:
2353
# this file id is at a different path in one of the
2354
# other trees, so put absent pointers there
2355
# This is the vertical axis in the matrix, all pointing
2357
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2358
# by path consistency: Insert into an existing path record (trivial), or
2359
# add a new one with relocation pointers for the other tree indexes.
2360
if new_entry_key in id_index[file_id]:
2361
# there is already an entry where this data belongs, just insert it.
2362
by_path[new_entry_key][tree_index] = \
2363
self._inv_entry_to_details(entry)
2365
# add relocated entries to the horizontal axis - this row
2366
# mapping from path,id. We need to look up the correct path
2367
# for the indexes from 0 to tree_index -1
2369
for lookup_index in xrange(tree_index):
2370
# boundary case: this is the first occurence of file_id
2371
# so there are no id_indexs, possibly take this out of
2373
if not len(id_index[file_id]):
2374
new_details.append(DirState.NULL_PARENT_DETAILS)
2376
# grab any one entry, use it to find the right path.
2377
# TODO: optimise this to reduce memory use in highly
2378
# fragmented situations by reusing the relocation
2380
a_key = iter(id_index[file_id]).next()
2381
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2382
# its a pointer or missing statement, use it as is.
2383
new_details.append(by_path[a_key][lookup_index])
2385
# we have the right key, make a pointer to it.
2386
real_path = ('/'.join(a_key[0:2])).strip('/')
2387
new_details.append(('r', real_path, 0, False, ''))
2388
new_details.append(self._inv_entry_to_details(entry))
2389
new_details.extend(new_location_suffix)
2390
by_path[new_entry_key] = new_details
2391
id_index[file_id].add(new_entry_key)
2392
# --- end generation of full tree mappings
2394
# sort and output all the entries
2395
new_entries = self._sort_entries(by_path.items())
2396
self._entries_to_current_state(new_entries)
2397
self._parents = [rev_id for rev_id, tree in trees]
2398
self._ghosts = list(ghosts)
2399
self._header_state = DirState.IN_MEMORY_MODIFIED
2400
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2401
self._id_index = id_index
2403
def _sort_entries(self, entry_list):
2404
"""Given a list of entries, sort them into the right order.
2406
This is done when constructing a new dirstate from trees - normally we
2407
try to keep everything in sorted blocks all the time, but sometimes
2408
it's easier to sort after the fact.
2411
# sort by: directory parts, file name, file id
2412
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2413
return sorted(entry_list, key=_key)
2415
def set_state_from_inventory(self, new_inv):
2416
"""Set new_inv as the current state.
2418
This API is called by tree transform, and will usually occur with
2419
existing parent trees.
2421
:param new_inv: The inventory object to set current state from.
2423
if 'evil' in debug.debug_flags:
2424
trace.mutter_callsite(1,
2425
"set_state_from_inventory called; please mutate the tree instead")
2426
self._read_dirblocks_if_needed()
2428
# Two iterators: current data and new data, both in dirblock order.
2429
# We zip them together, which tells about entries that are new in the
2430
# inventory, or removed in the inventory, or present in both and
2433
# You might think we could just synthesize a new dirstate directly
2434
# since we're processing it in the right order. However, we need to
2435
# also consider there may be any number of parent trees and relocation
2436
# pointers, and we don't want to duplicate that here.
2437
new_iterator = new_inv.iter_entries_by_dir()
2438
# we will be modifying the dirstate, so we need a stable iterator. In
2439
# future we might write one, for now we just clone the state into a
2440
# list - which is a shallow copy.
2441
old_iterator = iter(list(self._iter_entries()))
2442
# both must have roots so this is safe:
2443
current_new = new_iterator.next()
2444
current_old = old_iterator.next()
2445
def advance(iterator):
2447
return iterator.next()
2448
except StopIteration:
2450
while current_new or current_old:
2451
# skip entries in old that are not really there
2452
if current_old and current_old[1][0][0] in 'ar':
2453
# relocated or absent
2454
current_old = advance(old_iterator)
2457
# convert new into dirblock style
2458
new_path_utf8 = current_new[0].encode('utf8')
2459
new_dirname, new_basename = osutils.split(new_path_utf8)
2460
new_id = current_new[1].file_id
2461
new_entry_key = (new_dirname, new_basename, new_id)
2462
current_new_minikind = \
2463
DirState._kind_to_minikind[current_new[1].kind]
2464
if current_new_minikind == 't':
2465
fingerprint = current_new[1].reference_revision or ''
2467
# We normally only insert or remove records, or update
2468
# them when it has significantly changed. Then we want to
2469
# erase its fingerprint. Unaffected records should
2470
# normally not be updated at all.
2473
# for safety disable variables
2474
new_path_utf8 = new_dirname = new_basename = new_id = \
2475
new_entry_key = None
2476
# 5 cases, we dont have a value that is strictly greater than everything, so
2477
# we make both end conditions explicit
2479
# old is finished: insert current_new into the state.
2480
self.update_minimal(new_entry_key, current_new_minikind,
2481
executable=current_new[1].executable,
2482
path_utf8=new_path_utf8, fingerprint=fingerprint)
2483
current_new = advance(new_iterator)
2484
elif not current_new:
2486
self._make_absent(current_old)
2487
current_old = advance(old_iterator)
2488
elif new_entry_key == current_old[0]:
2489
# same - common case
2490
# We're looking at the same path and id in both the dirstate
2491
# and inventory, so just need to update the fields in the
2492
# dirstate from the one in the inventory.
2493
# TODO: update the record if anything significant has changed.
2494
# the minimal required trigger is if the execute bit or cached
2496
if (current_old[1][0][3] != current_new[1].executable or
2497
current_old[1][0][0] != current_new_minikind):
2498
self.update_minimal(current_old[0], current_new_minikind,
2499
executable=current_new[1].executable,
2500
path_utf8=new_path_utf8, fingerprint=fingerprint)
2501
# both sides are dealt with, move on
2502
current_old = advance(old_iterator)
2503
current_new = advance(new_iterator)
2504
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2505
or (new_dirname == current_old[0][0]
2506
and new_entry_key[1:] < current_old[0][1:])):
2508
# add a entry for this and advance new
2509
self.update_minimal(new_entry_key, current_new_minikind,
2510
executable=current_new[1].executable,
2511
path_utf8=new_path_utf8, fingerprint=fingerprint)
2512
current_new = advance(new_iterator)
2514
# we've advanced past the place where the old key would be,
2515
# without seeing it in the new list. so it must be gone.
2516
self._make_absent(current_old)
2517
current_old = advance(old_iterator)
2518
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2519
self._id_index = None
2520
self._packed_stat_index = None
2522
def _make_absent(self, current_old):
2523
"""Mark current_old - an entry - as absent for tree 0.
2525
:return: True if this was the last details entry for the entry key:
2526
that is, if the underlying block has had the entry removed, thus
2527
shrinking in length.
2529
# build up paths that this id will be left at after the change is made,
2530
# so we can update their cross references in tree 0
2531
all_remaining_keys = set()
2532
# Dont check the working tree, because it's going.
2533
for details in current_old[1][1:]:
2534
if details[0] not in 'ar': # absent, relocated
2535
all_remaining_keys.add(current_old[0])
2536
elif details[0] == 'r': # relocated
2537
# record the key for the real path.
2538
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2539
# absent rows are not present at any path.
2540
last_reference = current_old[0] not in all_remaining_keys
2542
# the current row consists entire of the current item (being marked
2543
# absent), and relocated or absent entries for the other trees:
2544
# Remove it, its meaningless.
2545
block = self._find_block(current_old[0])
2546
entry_index, present = self._find_entry_index(current_old[0], block[1])
2548
raise AssertionError('could not find entry for %s' % (current_old,))
2549
block[1].pop(entry_index)
2550
# if we have an id_index in use, remove this key from it for this id.
2551
if self._id_index is not None:
2552
self._id_index[current_old[0][2]].remove(current_old[0])
2553
# update all remaining keys for this id to record it as absent. The
2554
# existing details may either be the record we are marking as deleted
2555
# (if there were other trees with the id present at this path), or may
2557
for update_key in all_remaining_keys:
2558
update_block_index, present = \
2559
self._find_block_index_from_key(update_key)
2561
raise AssertionError('could not find block for %s' % (update_key,))
2562
update_entry_index, present = \
2563
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2565
raise AssertionError('could not find entry for %s' % (update_key,))
2566
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2567
# it must not be absent at the moment
2568
if update_tree_details[0][0] == 'a': # absent
2569
raise AssertionError('bad row %r' % (update_tree_details,))
2570
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2571
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2572
return last_reference
2574
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2575
packed_stat=None, size=0, path_utf8=None):
2576
"""Update an entry to the state in tree 0.
2578
This will either create a new entry at 'key' or update an existing one.
2579
It also makes sure that any other records which might mention this are
2582
:param key: (dir, name, file_id) for the new entry
2583
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2585
:param executable: Should the executable bit be set?
2586
:param fingerprint: Simple fingerprint for new entry: canonical-form
2587
sha1 for files, referenced revision id for subtrees, etc.
2588
:param packed_stat: Packed stat value for new entry.
2589
:param size: Size information for new entry
2590
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2593
If packed_stat and fingerprint are not given, they're invalidated in
2596
block = self._find_block(key)[1]
2597
if packed_stat is None:
2598
packed_stat = DirState.NULLSTAT
2599
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2600
# sometimes present in the dirstate - this seems oddly inconsistent.
2602
entry_index, present = self._find_entry_index(key, block)
2603
new_details = (minikind, fingerprint, size, executable, packed_stat)
2604
id_index = self._get_id_index()
2606
# new entry, synthesis cross reference here,
2607
existing_keys = id_index.setdefault(key[2], set())
2608
if not existing_keys:
2609
# not currently in the state, simplest case
2610
new_entry = key, [new_details] + self._empty_parent_info()
2612
# present at one or more existing other paths.
2613
# grab one of them and use it to generate parent
2614
# relocation/absent entries.
2615
new_entry = key, [new_details]
2616
for other_key in existing_keys:
2617
# change the record at other to be a pointer to this new
2618
# record. The loop looks similar to the change to
2619
# relocations when updating an existing record but its not:
2620
# the test for existing kinds is different: this can be
2621
# factored out to a helper though.
2622
other_block_index, present = self._find_block_index_from_key(other_key)
2624
raise AssertionError('could not find block for %s' % (other_key,))
2625
other_entry_index, present = self._find_entry_index(other_key,
2626
self._dirblocks[other_block_index][1])
2628
raise AssertionError('could not find entry for %s' % (other_key,))
2629
if path_utf8 is None:
2630
raise AssertionError('no path')
2631
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2632
('r', path_utf8, 0, False, '')
2634
num_present_parents = self._num_present_parents()
2635
for lookup_index in xrange(1, num_present_parents + 1):
2636
# grab any one entry, use it to find the right path.
2637
# TODO: optimise this to reduce memory use in highly
2638
# fragmented situations by reusing the relocation
2640
update_block_index, present = \
2641
self._find_block_index_from_key(other_key)
2643
raise AssertionError('could not find block for %s' % (other_key,))
2644
update_entry_index, present = \
2645
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2647
raise AssertionError('could not find entry for %s' % (other_key,))
2648
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2649
if update_details[0] in 'ar': # relocated, absent
2650
# its a pointer or absent in lookup_index's tree, use
2652
new_entry[1].append(update_details)
2654
# we have the right key, make a pointer to it.
2655
pointer_path = osutils.pathjoin(*other_key[0:2])
2656
new_entry[1].append(('r', pointer_path, 0, False, ''))
2657
block.insert(entry_index, new_entry)
2658
existing_keys.add(key)
2660
# Does the new state matter?
2661
block[entry_index][1][0] = new_details
2662
# parents cannot be affected by what we do.
2663
# other occurences of this id can be found
2664
# from the id index.
2666
# tree index consistency: All other paths for this id in this tree
2667
# index must point to the correct path. We have to loop here because
2668
# we may have passed entries in the state with this file id already
2669
# that were absent - where parent entries are - and they need to be
2670
# converted to relocated.
2671
if path_utf8 is None:
2672
raise AssertionError('no path')
2673
for entry_key in id_index.setdefault(key[2], set()):
2674
# TODO:PROFILING: It might be faster to just update
2675
# rather than checking if we need to, and then overwrite
2676
# the one we are located at.
2677
if entry_key != key:
2678
# this file id is at a different path in one of the
2679
# other trees, so put absent pointers there
2680
# This is the vertical axis in the matrix, all pointing
2682
block_index, present = self._find_block_index_from_key(entry_key)
2684
raise AssertionError('not present: %r', entry_key)
2685
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2687
raise AssertionError('not present: %r', entry_key)
2688
self._dirblocks[block_index][1][entry_index][1][0] = \
2689
('r', path_utf8, 0, False, '')
2690
# add a containing dirblock if needed.
2691
if new_details[0] == 'd':
2692
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2693
block_index, present = self._find_block_index_from_key(subdir_key)
2695
self._dirblocks.insert(block_index, (subdir_key[0], []))
2697
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2699
def _validate(self):
2700
"""Check that invariants on the dirblock are correct.
2702
This can be useful in debugging; it shouldn't be necessary in
2705
This must be called with a lock held.
2707
# NOTE: This must always raise AssertionError not just assert,
2708
# otherwise it may not behave properly under python -O
2710
# TODO: All entries must have some content that's not 'a' or 'r',
2711
# otherwise it could just be removed.
2713
# TODO: All relocations must point directly to a real entry.
2715
# TODO: No repeated keys.
2718
from pprint import pformat
2719
self._read_dirblocks_if_needed()
2720
if len(self._dirblocks) > 0:
2721
if not self._dirblocks[0][0] == '':
2722
raise AssertionError(
2723
"dirblocks don't start with root block:\n" + \
2724
pformat(self._dirblocks))
2725
if len(self._dirblocks) > 1:
2726
if not self._dirblocks[1][0] == '':
2727
raise AssertionError(
2728
"dirblocks missing root directory:\n" + \
2729
pformat(self._dirblocks))
2730
# the dirblocks are sorted by their path components, name, and dir id
2731
dir_names = [d[0].split('/')
2732
for d in self._dirblocks[1:]]
2733
if dir_names != sorted(dir_names):
2734
raise AssertionError(
2735
"dir names are not in sorted order:\n" + \
2736
pformat(self._dirblocks) + \
2739
for dirblock in self._dirblocks:
2740
# within each dirblock, the entries are sorted by filename and
2742
for entry in dirblock[1]:
2743
if dirblock[0] != entry[0][0]:
2744
raise AssertionError(
2746
"doesn't match directory name in\n%r" %
2747
(entry, pformat(dirblock)))
2748
if dirblock[1] != sorted(dirblock[1]):
2749
raise AssertionError(
2750
"dirblock for %r is not sorted:\n%s" % \
2751
(dirblock[0], pformat(dirblock)))
2753
def check_valid_parent():
2754
"""Check that the current entry has a valid parent.
2756
This makes sure that the parent has a record,
2757
and that the parent isn't marked as "absent" in the
2758
current tree. (It is invalid to have a non-absent file in an absent
2761
if entry[0][0:2] == ('', ''):
2762
# There should be no parent for the root row
2764
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2765
if parent_entry == (None, None):
2766
raise AssertionError(
2767
"no parent entry for: %s in tree %s"
2768
% (this_path, tree_index))
2769
if parent_entry[1][tree_index][0] != 'd':
2770
raise AssertionError(
2771
"Parent entry for %s is not marked as a valid"
2772
" directory. %s" % (this_path, parent_entry,))
2774
# For each file id, for each tree: either
2775
# the file id is not present at all; all rows with that id in the
2776
# key have it marked as 'absent'
2777
# OR the file id is present under exactly one name; any other entries
2778
# that mention that id point to the correct name.
2780
# We check this with a dict per tree pointing either to the present
2781
# name, or None if absent.
2782
tree_count = self._num_present_parents() + 1
2783
id_path_maps = [dict() for i in range(tree_count)]
2784
# Make sure that all renamed entries point to the correct location.
2785
for entry in self._iter_entries():
2786
file_id = entry[0][2]
2787
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2788
if len(entry[1]) != tree_count:
2789
raise AssertionError(
2790
"wrong number of entry details for row\n%s" \
2791
",\nexpected %d" % \
2792
(pformat(entry), tree_count))
2793
absent_positions = 0
2794
for tree_index, tree_state in enumerate(entry[1]):
2795
this_tree_map = id_path_maps[tree_index]
2796
minikind = tree_state[0]
2797
if minikind in 'ar':
2798
absent_positions += 1
2799
# have we seen this id before in this column?
2800
if file_id in this_tree_map:
2801
previous_path, previous_loc = this_tree_map[file_id]
2802
# any later mention of this file must be consistent with
2803
# what was said before
2805
if previous_path is not None:
2806
raise AssertionError(
2807
"file %s is absent in row %r but also present " \
2809
(file_id, entry, previous_path))
2810
elif minikind == 'r':
2811
target_location = tree_state[1]
2812
if previous_path != target_location:
2813
raise AssertionError(
2814
"file %s relocation in row %r but also at %r" \
2815
% (file_id, entry, previous_path))
2817
# a file, directory, etc - may have been previously
2818
# pointed to by a relocation, which must point here
2819
if previous_path != this_path:
2820
raise AssertionError(
2821
"entry %r inconsistent with previous path %r "
2823
(entry, previous_path, previous_loc))
2824
check_valid_parent()
2827
# absent; should not occur anywhere else
2828
this_tree_map[file_id] = None, this_path
2829
elif minikind == 'r':
2830
# relocation, must occur at expected location
2831
this_tree_map[file_id] = tree_state[1], this_path
2833
this_tree_map[file_id] = this_path, this_path
2834
check_valid_parent()
2835
if absent_positions == tree_count:
2836
raise AssertionError(
2837
"entry %r has no data for any tree." % (entry,))
2839
def _wipe_state(self):
2840
"""Forget all state information about the dirstate."""
2841
self._header_state = DirState.NOT_IN_MEMORY
2842
self._dirblock_state = DirState.NOT_IN_MEMORY
2843
self._changes_aborted = False
2846
self._dirblocks = []
2847
self._id_index = None
2848
self._packed_stat_index = None
2849
self._end_of_header = None
2850
self._cutoff_time = None
2851
self._split_path_cache = {}
2853
def lock_read(self):
2854
"""Acquire a read lock on the dirstate."""
2855
if self._lock_token is not None:
2856
raise errors.LockContention(self._lock_token)
2857
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2858
# already in memory, we could read just the header and check for
2859
# any modification. If not modified, we can just leave things
2861
self._lock_token = lock.ReadLock(self._filename)
2862
self._lock_state = 'r'
2863
self._state_file = self._lock_token.f
2866
def lock_write(self):
2867
"""Acquire a write lock on the dirstate."""
2868
if self._lock_token is not None:
2869
raise errors.LockContention(self._lock_token)
2870
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2871
# already in memory, we could read just the header and check for
2872
# any modification. If not modified, we can just leave things
2874
self._lock_token = lock.WriteLock(self._filename)
2875
self._lock_state = 'w'
2876
self._state_file = self._lock_token.f
2880
"""Drop any locks held on the dirstate."""
2881
if self._lock_token is None:
2882
raise errors.LockNotHeld(self)
2883
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2884
# already in memory, we could read just the header and check for
2885
# any modification. If not modified, we can just leave things
2887
self._state_file = None
2888
self._lock_state = None
2889
self._lock_token.unlock()
2890
self._lock_token = None
2891
self._split_path_cache = {}
2893
def _requires_lock(self):
2894
"""Check that a lock is currently held by someone on the dirstate."""
2895
if not self._lock_token:
2896
raise errors.ObjectNotLocked(self)
2899
def py_update_entry(state, entry, abspath, stat_value,
2900
_stat_to_minikind=DirState._stat_to_minikind,
2901
_pack_stat=pack_stat):
2902
"""Update the entry based on what is actually on disk.
2904
This function only calculates the sha if it needs to - if the entry is
2905
uncachable, or clearly different to the first parent's entry, no sha
2906
is calculated, and None is returned.
2908
:param state: The dirstate this entry is in.
2909
:param entry: This is the dirblock entry for the file in question.
2910
:param abspath: The path on disk for this file.
2911
:param stat_value: The stat value done on the path.
2912
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
2913
target of a symlink.
2916
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
2920
packed_stat = _pack_stat(stat_value)
2921
(saved_minikind, saved_link_or_sha1, saved_file_size,
2922
saved_executable, saved_packed_stat) = entry[1][0]
2924
if minikind == 'd' and saved_minikind == 't':
2926
if (minikind == saved_minikind
2927
and packed_stat == saved_packed_stat):
2928
# The stat hasn't changed since we saved, so we can re-use the
2933
# size should also be in packed_stat
2934
if saved_file_size == stat_value.st_size:
2935
return saved_link_or_sha1
2937
# If we have gotten this far, that means that we need to actually
2938
# process this entry.
2941
executable = state._is_executable(stat_value.st_mode,
2943
if state._cutoff_time is None:
2944
state._sha_cutoff_time()
2945
if (stat_value.st_mtime < state._cutoff_time
2946
and stat_value.st_ctime < state._cutoff_time
2947
and len(entry[1]) > 1
2948
and entry[1][1][0] != 'a'):
2949
# Could check for size changes for further optimised
2950
# avoidance of sha1's. However the most prominent case of
2951
# over-shaing is during initial add, which this catches.
2952
# Besides, if content filtering happens, size and sha
2953
# are calculated at the same time, so checking just the size
2954
# gains nothing w.r.t. performance.
2955
link_or_sha1 = state._sha1_file(abspath)
2956
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
2957
executable, packed_stat)
2959
entry[1][0] = ('f', '', stat_value.st_size,
2960
executable, DirState.NULLSTAT)
2961
elif minikind == 'd':
2963
entry[1][0] = ('d', '', 0, False, packed_stat)
2964
if saved_minikind != 'd':
2965
# This changed from something into a directory. Make sure we
2966
# have a directory block for it. This doesn't happen very
2967
# often, so this doesn't have to be super fast.
2968
block_index, entry_index, dir_present, file_present = \
2969
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
2970
state._ensure_block(block_index, entry_index,
2971
osutils.pathjoin(entry[0][0], entry[0][1]))
2972
elif minikind == 'l':
2973
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
2974
if state._cutoff_time is None:
2975
state._sha_cutoff_time()
2976
if (stat_value.st_mtime < state._cutoff_time
2977
and stat_value.st_ctime < state._cutoff_time):
2978
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
2981
entry[1][0] = ('l', '', stat_value.st_size,
2982
False, DirState.NULLSTAT)
2983
state._dirblock_state = DirState.IN_MEMORY_MODIFIED
2987
class ProcessEntryPython(object):
2989
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id", "uninteresting",
2990
"last_source_parent", "last_target_parent", "include_unchanged",
2991
"use_filesystem_for_exec", "utf8_decode", "searched_specific_files",
2992
"search_specific_files", "state", "source_index", "target_index",
2993
"want_unversioned", "tree"]
2995
def __init__(self, include_unchanged, use_filesystem_for_exec,
2996
search_specific_files, state, source_index, target_index,
2997
want_unversioned, tree):
2998
self.old_dirname_to_file_id = {}
2999
self.new_dirname_to_file_id = {}
3000
# Just a sentry, so that _process_entry can say that this
3001
# record is handled, but isn't interesting to process (unchanged)
3002
self.uninteresting = object()
3003
# Using a list so that we can access the values and change them in
3004
# nested scope. Each one is [path, file_id, entry]
3005
self.last_source_parent = [None, None]
3006
self.last_target_parent = [None, None]
3007
self.include_unchanged = include_unchanged
3008
self.use_filesystem_for_exec = use_filesystem_for_exec
3009
self.utf8_decode = cache_utf8._utf8_decode
3010
# for all search_indexs in each path at or under each element of
3011
# search_specific_files, if the detail is relocated: add the id, and add the
3012
# relocated path as one to search if its not searched already. If the
3013
# detail is not relocated, add the id.
3014
self.searched_specific_files = set()
3015
self.search_specific_files = search_specific_files
3017
self.source_index = source_index
3018
self.target_index = target_index
3019
self.want_unversioned = want_unversioned
3022
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
3023
"""Compare an entry and real disk to generate delta information.
3025
:param path_info: top_relpath, basename, kind, lstat, abspath for
3026
the path of entry. If None, then the path is considered absent.
3027
(Perhaps we should pass in a concrete entry for this ?)
3028
Basename is returned as a utf8 string because we expect this
3029
tuple will be ignored, and don't want to take the time to
3031
:return: None if these don't match
3032
A tuple of information about the change, or
3033
the object 'uninteresting' if these match, but are
3034
basically identical.
3036
if self.source_index is None:
3037
source_details = DirState.NULL_PARENT_DETAILS
3039
source_details = entry[1][self.source_index]
3040
target_details = entry[1][self.target_index]
3041
target_minikind = target_details[0]
3042
if path_info is not None and target_minikind in 'fdlt':
3043
if not (self.target_index == 0):
3044
raise AssertionError()
3045
link_or_sha1 = update_entry(self.state, entry,
3046
abspath=path_info[4], stat_value=path_info[3])
3047
# The entry may have been modified by update_entry
3048
target_details = entry[1][self.target_index]
3049
target_minikind = target_details[0]
3052
file_id = entry[0][2]
3053
source_minikind = source_details[0]
3054
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
3055
# claimed content in both: diff
3056
# r | fdlt | | add source to search, add id path move and perform
3057
# | | | diff check on source-target
3058
# r | fdlt | a | dangling file that was present in the basis.
3060
if source_minikind in 'r':
3061
# add the source to the search path to find any children it
3062
# has. TODO ? : only add if it is a container ?
3063
if not osutils.is_inside_any(self.searched_specific_files,
3065
self.search_specific_files.add(source_details[1])
3066
# generate the old path; this is needed for stating later
3068
old_path = source_details[1]
3069
old_dirname, old_basename = os.path.split(old_path)
3070
path = pathjoin(entry[0][0], entry[0][1])
3071
old_entry = self.state._get_entry(self.source_index,
3073
# update the source details variable to be the real
3075
if old_entry == (None, None):
3076
raise errors.CorruptDirstate(self.state._filename,
3077
"entry '%s/%s' is considered renamed from %r"
3078
" but source does not exist\n"
3079
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
3080
source_details = old_entry[1][self.source_index]
3081
source_minikind = source_details[0]
3083
old_dirname = entry[0][0]
3084
old_basename = entry[0][1]
3085
old_path = path = None
3086
if path_info is None:
3087
# the file is missing on disk, show as removed.
3088
content_change = True
3092
# source and target are both versioned and disk file is present.
3093
target_kind = path_info[2]
3094
if target_kind == 'directory':
3096
old_path = path = pathjoin(old_dirname, old_basename)
3097
self.new_dirname_to_file_id[path] = file_id
3098
if source_minikind != 'd':
3099
content_change = True
3101
# directories have no fingerprint
3102
content_change = False
3104
elif target_kind == 'file':
3105
if source_minikind != 'f':
3106
content_change = True
3108
# Check the sha. We can't just rely on the size as
3109
# content filtering may mean differ sizes actually
3110
# map to the same content
3111
if link_or_sha1 is None:
3113
statvalue, link_or_sha1 = \
3114
self.state._sha1_provider.stat_and_sha1(
3116
self.state._observed_sha1(entry, link_or_sha1,
3118
content_change = (link_or_sha1 != source_details[1])
3119
# Target details is updated at update_entry time
3120
if self.use_filesystem_for_exec:
3121
# We don't need S_ISREG here, because we are sure
3122
# we are dealing with a file.
3123
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
3125
target_exec = target_details[3]
3126
elif target_kind == 'symlink':
3127
if source_minikind != 'l':
3128
content_change = True
3130
content_change = (link_or_sha1 != source_details[1])
3132
elif target_kind == 'tree-reference':
3133
if source_minikind != 't':
3134
content_change = True
3136
content_change = False
3139
raise Exception, "unknown kind %s" % path_info[2]
3140
if source_minikind == 'd':
3142
old_path = path = pathjoin(old_dirname, old_basename)
3143
self.old_dirname_to_file_id[old_path] = file_id
3144
# parent id is the entry for the path in the target tree
3145
if old_dirname == self.last_source_parent[0]:
3146
source_parent_id = self.last_source_parent[1]
3149
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3151
source_parent_entry = self.state._get_entry(self.source_index,
3152
path_utf8=old_dirname)
3153
source_parent_id = source_parent_entry[0][2]
3154
if source_parent_id == entry[0][2]:
3155
# This is the root, so the parent is None
3156
source_parent_id = None
3158
self.last_source_parent[0] = old_dirname
3159
self.last_source_parent[1] = source_parent_id
3160
new_dirname = entry[0][0]
3161
if new_dirname == self.last_target_parent[0]:
3162
target_parent_id = self.last_target_parent[1]
3165
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3167
# TODO: We don't always need to do the lookup, because the
3168
# parent entry will be the same as the source entry.
3169
target_parent_entry = self.state._get_entry(self.target_index,
3170
path_utf8=new_dirname)
3171
if target_parent_entry == (None, None):
3172
raise AssertionError(
3173
"Could not find target parent in wt: %s\nparent of: %s"
3174
% (new_dirname, entry))
3175
target_parent_id = target_parent_entry[0][2]
3176
if target_parent_id == entry[0][2]:
3177
# This is the root, so the parent is None
3178
target_parent_id = None
3180
self.last_target_parent[0] = new_dirname
3181
self.last_target_parent[1] = target_parent_id
3183
source_exec = source_details[3]
3184
if (self.include_unchanged
3186
or source_parent_id != target_parent_id
3187
or old_basename != entry[0][1]
3188
or source_exec != target_exec
3190
if old_path is None:
3191
old_path = path = pathjoin(old_dirname, old_basename)
3192
old_path_u = self.utf8_decode(old_path)[0]
3195
old_path_u = self.utf8_decode(old_path)[0]
3196
if old_path == path:
3199
path_u = self.utf8_decode(path)[0]
3200
source_kind = DirState._minikind_to_kind[source_minikind]
3201
return (entry[0][2],
3202
(old_path_u, path_u),
3205
(source_parent_id, target_parent_id),
3206
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3207
(source_kind, target_kind),
3208
(source_exec, target_exec))
3210
return self.uninteresting
3211
elif source_minikind in 'a' and target_minikind in 'fdlt':
3212
# looks like a new file
3213
path = pathjoin(entry[0][0], entry[0][1])
3214
# parent id is the entry for the path in the target tree
3215
# TODO: these are the same for an entire directory: cache em.
3216
parent_id = self.state._get_entry(self.target_index,
3217
path_utf8=entry[0][0])[0][2]
3218
if parent_id == entry[0][2]:
3220
if path_info is not None:
3222
if self.use_filesystem_for_exec:
3223
# We need S_ISREG here, because we aren't sure if this
3226
stat.S_ISREG(path_info[3].st_mode)
3227
and stat.S_IEXEC & path_info[3].st_mode)
3229
target_exec = target_details[3]
3230
return (entry[0][2],
3231
(None, self.utf8_decode(path)[0]),
3235
(None, self.utf8_decode(entry[0][1])[0]),
3236
(None, path_info[2]),
3237
(None, target_exec))
3239
# Its a missing file, report it as such.
3240
return (entry[0][2],
3241
(None, self.utf8_decode(path)[0]),
3245
(None, self.utf8_decode(entry[0][1])[0]),
3248
elif source_minikind in 'fdlt' and target_minikind in 'a':
3249
# unversioned, possibly, or possibly not deleted: we dont care.
3250
# if its still on disk, *and* theres no other entry at this
3251
# path [we dont know this in this routine at the moment -
3252
# perhaps we should change this - then it would be an unknown.
3253
old_path = pathjoin(entry[0][0], entry[0][1])
3254
# parent id is the entry for the path in the target tree
3255
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3256
if parent_id == entry[0][2]:
3258
return (entry[0][2],
3259
(self.utf8_decode(old_path)[0], None),
3263
(self.utf8_decode(entry[0][1])[0], None),
3264
(DirState._minikind_to_kind[source_minikind], None),
3265
(source_details[3], None))
3266
elif source_minikind in 'fdlt' and target_minikind in 'r':
3267
# a rename; could be a true rename, or a rename inherited from
3268
# a renamed parent. TODO: handle this efficiently. Its not
3269
# common case to rename dirs though, so a correct but slow
3270
# implementation will do.
3271
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3272
self.search_specific_files.add(target_details[1])
3273
elif source_minikind in 'ra' and target_minikind in 'ra':
3274
# neither of the selected trees contain this file,
3275
# so skip over it. This is not currently directly tested, but
3276
# is indirectly via test_too_much.TestCommands.test_conflicts.
3279
raise AssertionError("don't know how to compare "
3280
"source_minikind=%r, target_minikind=%r"
3281
% (source_minikind, target_minikind))
3282
## import pdb;pdb.set_trace()
3288
def iter_changes(self):
3289
"""Iterate over the changes."""
3290
utf8_decode = cache_utf8._utf8_decode
3291
_cmp_by_dirs = cmp_by_dirs
3292
_process_entry = self._process_entry
3293
uninteresting = self.uninteresting
3294
search_specific_files = self.search_specific_files
3295
searched_specific_files = self.searched_specific_files
3296
splitpath = osutils.splitpath
3298
# compare source_index and target_index at or under each element of search_specific_files.
3299
# follow the following comparison table. Note that we only want to do diff operations when
3300
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3304
# Source | Target | disk | action
3305
# r | fdlt | | add source to search, add id path move and perform
3306
# | | | diff check on source-target
3307
# r | fdlt | a | dangling file that was present in the basis.
3309
# r | a | | add source to search
3311
# r | r | | this path is present in a non-examined tree, skip.
3312
# r | r | a | this path is present in a non-examined tree, skip.
3313
# a | fdlt | | add new id
3314
# a | fdlt | a | dangling locally added file, skip
3315
# a | a | | not present in either tree, skip
3316
# a | a | a | not present in any tree, skip
3317
# a | r | | not present in either tree at this path, skip as it
3318
# | | | may not be selected by the users list of paths.
3319
# a | r | a | not present in either tree at this path, skip as it
3320
# | | | may not be selected by the users list of paths.
3321
# fdlt | fdlt | | content in both: diff them
3322
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3323
# fdlt | a | | unversioned: output deleted id for now
3324
# fdlt | a | a | unversioned and deleted: output deleted id
3325
# fdlt | r | | relocated in this tree, so add target to search.
3326
# | | | Dont diff, we will see an r,fd; pair when we reach
3327
# | | | this id at the other path.
3328
# fdlt | r | a | relocated in this tree, so add target to search.
3329
# | | | Dont diff, we will see an r,fd; pair when we reach
3330
# | | | this id at the other path.
3332
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3333
# keeping a cache of directories that we have seen.
3335
while search_specific_files:
3336
# TODO: the pending list should be lexically sorted? the
3337
# interface doesn't require it.
3338
current_root = search_specific_files.pop()
3339
current_root_unicode = current_root.decode('utf8')
3340
searched_specific_files.add(current_root)
3341
# process the entries for this containing directory: the rest will be
3342
# found by their parents recursively.
3343
root_entries = self.state._entries_for_path(current_root)
3344
root_abspath = self.tree.abspath(current_root_unicode)
3346
root_stat = os.lstat(root_abspath)
3348
if e.errno == errno.ENOENT:
3349
# the path does not exist: let _process_entry know that.
3350
root_dir_info = None
3352
# some other random error: hand it up.
3355
root_dir_info = ('', current_root,
3356
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3358
if root_dir_info[2] == 'directory':
3359
if self.tree._directory_is_tree_reference(
3360
current_root.decode('utf8')):
3361
root_dir_info = root_dir_info[:2] + \
3362
('tree-reference',) + root_dir_info[3:]
3364
if not root_entries and not root_dir_info:
3365
# this specified path is not present at all, skip it.
3367
path_handled = False
3368
for entry in root_entries:
3369
result = _process_entry(entry, root_dir_info)
3370
if result is not None:
3372
if result is not uninteresting:
3374
if self.want_unversioned and not path_handled and root_dir_info:
3375
new_executable = bool(
3376
stat.S_ISREG(root_dir_info[3].st_mode)
3377
and stat.S_IEXEC & root_dir_info[3].st_mode)
3379
(None, current_root_unicode),
3383
(None, splitpath(current_root_unicode)[-1]),
3384
(None, root_dir_info[2]),
3385
(None, new_executable)
3387
initial_key = (current_root, '', '')
3388
block_index, _ = self.state._find_block_index_from_key(initial_key)
3389
if block_index == 0:
3390
# we have processed the total root already, but because the
3391
# initial key matched it we should skip it here.
3393
if root_dir_info and root_dir_info[2] == 'tree-reference':
3394
current_dir_info = None
3396
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3398
current_dir_info = dir_iterator.next()
3400
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3401
# python 2.5 has e.errno == EINVAL,
3402
# and e.winerror == ERROR_DIRECTORY
3403
e_winerror = getattr(e, 'winerror', None)
3404
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3405
# there may be directories in the inventory even though
3406
# this path is not a file on disk: so mark it as end of
3408
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3409
current_dir_info = None
3410
elif (sys.platform == 'win32'
3411
and (e.errno in win_errors
3412
or e_winerror in win_errors)):
3413
current_dir_info = None
3417
if current_dir_info[0][0] == '':
3418
# remove .bzr from iteration
3419
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3420
if current_dir_info[1][bzr_index][0] != '.bzr':
3421
raise AssertionError()
3422
del current_dir_info[1][bzr_index]
3423
# walk until both the directory listing and the versioned metadata
3425
if (block_index < len(self.state._dirblocks) and
3426
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3427
current_block = self.state._dirblocks[block_index]
3429
current_block = None
3430
while (current_dir_info is not None or
3431
current_block is not None):
3432
if (current_dir_info and current_block
3433
and current_dir_info[0][0] != current_block[0]):
3434
if _cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
3435
# filesystem data refers to paths not covered by the dirblock.
3436
# this has two possibilities:
3437
# A) it is versioned but empty, so there is no block for it
3438
# B) it is not versioned.
3440
# if (A) then we need to recurse into it to check for
3441
# new unknown files or directories.
3442
# if (B) then we should ignore it, because we don't
3443
# recurse into unknown directories.
3445
while path_index < len(current_dir_info[1]):
3446
current_path_info = current_dir_info[1][path_index]
3447
if self.want_unversioned:
3448
if current_path_info[2] == 'directory':
3449
if self.tree._directory_is_tree_reference(
3450
current_path_info[0].decode('utf8')):
3451
current_path_info = current_path_info[:2] + \
3452
('tree-reference',) + current_path_info[3:]
3453
new_executable = bool(
3454
stat.S_ISREG(current_path_info[3].st_mode)
3455
and stat.S_IEXEC & current_path_info[3].st_mode)
3457
(None, utf8_decode(current_path_info[0])[0]),
3461
(None, utf8_decode(current_path_info[1])[0]),
3462
(None, current_path_info[2]),
3463
(None, new_executable))
3464
# dont descend into this unversioned path if it is
3466
if current_path_info[2] in ('directory',
3468
del current_dir_info[1][path_index]
3472
# This dir info has been handled, go to the next
3474
current_dir_info = dir_iterator.next()
3475
except StopIteration:
3476
current_dir_info = None
3478
# We have a dirblock entry for this location, but there
3479
# is no filesystem path for this. This is most likely
3480
# because a directory was removed from the disk.
3481
# We don't have to report the missing directory,
3482
# because that should have already been handled, but we
3483
# need to handle all of the files that are contained
3485
for current_entry in current_block[1]:
3486
# entry referring to file not present on disk.
3487
# advance the entry only, after processing.
3488
result = _process_entry(current_entry, None)
3489
if result is not None:
3490
if result is not uninteresting:
3493
if (block_index < len(self.state._dirblocks) and
3494
osutils.is_inside(current_root,
3495
self.state._dirblocks[block_index][0])):
3496
current_block = self.state._dirblocks[block_index]
3498
current_block = None
3501
if current_block and entry_index < len(current_block[1]):
3502
current_entry = current_block[1][entry_index]
3504
current_entry = None
3505
advance_entry = True
3507
if current_dir_info and path_index < len(current_dir_info[1]):
3508
current_path_info = current_dir_info[1][path_index]
3509
if current_path_info[2] == 'directory':
3510
if self.tree._directory_is_tree_reference(
3511
current_path_info[0].decode('utf8')):
3512
current_path_info = current_path_info[:2] + \
3513
('tree-reference',) + current_path_info[3:]
3515
current_path_info = None
3517
path_handled = False
3518
while (current_entry is not None or
3519
current_path_info is not None):
3520
if current_entry is None:
3521
# the check for path_handled when the path is advanced
3522
# will yield this path if needed.
3524
elif current_path_info is None:
3525
# no path is fine: the per entry code will handle it.
3526
result = _process_entry(current_entry, current_path_info)
3527
if result is not None:
3528
if result is not uninteresting:
3530
elif (current_entry[0][1] != current_path_info[1]
3531
or current_entry[1][self.target_index][0] in 'ar'):
3532
# The current path on disk doesn't match the dirblock
3533
# record. Either the dirblock is marked as absent, or
3534
# the file on disk is not present at all in the
3535
# dirblock. Either way, report about the dirblock
3536
# entry, and let other code handle the filesystem one.
3538
# Compare the basename for these files to determine
3540
if current_path_info[1] < current_entry[0][1]:
3541
# extra file on disk: pass for now, but only
3542
# increment the path, not the entry
3543
advance_entry = False
3545
# entry referring to file not present on disk.
3546
# advance the entry only, after processing.
3547
result = _process_entry(current_entry, None)
3548
if result is not None:
3549
if result is not uninteresting:
3551
advance_path = False
3553
result = _process_entry(current_entry, current_path_info)
3554
if result is not None:
3556
if result is not uninteresting:
3558
if advance_entry and current_entry is not None:
3560
if entry_index < len(current_block[1]):
3561
current_entry = current_block[1][entry_index]
3563
current_entry = None
3565
advance_entry = True # reset the advance flaga
3566
if advance_path and current_path_info is not None:
3567
if not path_handled:
3568
# unversioned in all regards
3569
if self.want_unversioned:
3570
new_executable = bool(
3571
stat.S_ISREG(current_path_info[3].st_mode)
3572
and stat.S_IEXEC & current_path_info[3].st_mode)
3574
relpath_unicode = utf8_decode(current_path_info[0])[0]
3575
except UnicodeDecodeError:
3576
raise errors.BadFilenameEncoding(
3577
current_path_info[0], osutils._fs_enc)
3579
(None, relpath_unicode),
3583
(None, utf8_decode(current_path_info[1])[0]),
3584
(None, current_path_info[2]),
3585
(None, new_executable))
3586
# dont descend into this unversioned path if it is
3588
if current_path_info[2] in ('directory'):
3589
del current_dir_info[1][path_index]
3591
# dont descend the disk iterator into any tree
3593
if current_path_info[2] == 'tree-reference':
3594
del current_dir_info[1][path_index]
3597
if path_index < len(current_dir_info[1]):
3598
current_path_info = current_dir_info[1][path_index]
3599
if current_path_info[2] == 'directory':
3600
if self.tree._directory_is_tree_reference(
3601
current_path_info[0].decode('utf8')):
3602
current_path_info = current_path_info[:2] + \
3603
('tree-reference',) + current_path_info[3:]
3605
current_path_info = None
3606
path_handled = False
3608
advance_path = True # reset the advance flagg.
3609
if current_block is not None:
3611
if (block_index < len(self.state._dirblocks) and
3612
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3613
current_block = self.state._dirblocks[block_index]
3615
current_block = None
3616
if current_dir_info is not None:
3618
current_dir_info = dir_iterator.next()
3619
except StopIteration:
3620
current_dir_info = None
3623
# Try to load the compiled form if possible
3625
from bzrlib._dirstate_helpers_pyx import (
3631
ProcessEntryC as _process_entry,
3632
update_entry as update_entry,
3635
from bzrlib._dirstate_helpers_py import (
3642
# FIXME: It would be nice to be able to track moved lines so that the
3643
# corresponding python code can be moved to the _dirstate_helpers_py
3644
# module. I don't want to break the history for this important piece of
3645
# code so I left the code here -- vila 20090622
3646
update_entry = py_update_entry
3647
_process_entry = ProcessEntryPython
added
removed
bzrlib/dirstate.py
