1
# Copyright (C) 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but 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 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."""
271
raise NotImplementedError(self.sha1)
273
def stat_and_sha1(self, abspath):
274
"""Return the stat and sha1 of a file given its absolute path."""
275
raise NotImplementedError(self.stat_and_sha1)
278
class DefaultSha1Provider(Sha1Provider):
279
"""A Sha1Provider that reads directly from the filesystem."""
281
def sha1(self, abspath):
282
"""Return the sha1 of a file given it's absolute path."""
283
return osutils.sha_file_by_name(abspath)
285
def stat_and_sha1(self, abspath):
286
"""Return the stat and sha1 of a file given it's absolute path."""
287
file_obj = file(abspath, 'rb')
289
statvalue = os.fstat(file_obj.fileno())
290
sha1 = osutils.sha_file(file_obj)
293
return statvalue, sha1
296
class DirState(object):
297
"""Record directory and metadata state for fast access.
299
A dirstate is a specialised data structure for managing local working
300
tree state information. Its not yet well defined whether it is platform
301
specific, and if it is how we detect/parameterize that.
303
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
304
Unlike most bzr disk formats, DirStates must be locked for reading, using
305
lock_read. (This is an os file lock internally.) This is necessary
306
because the file can be rewritten in place.
308
DirStates must be explicitly written with save() to commit changes; just
309
unlocking them does not write the changes to disk.
312
_kind_to_minikind = {
318
'tree-reference': 't',
320
_minikind_to_kind = {
326
't': 'tree-reference',
328
_stat_to_minikind = {
333
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
334
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
337
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
338
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
339
# that we are sure a single record will always fit.
340
BISECT_PAGE_SIZE = 4096
343
IN_MEMORY_UNMODIFIED = 1
344
IN_MEMORY_MODIFIED = 2
346
# A pack_stat (the x's) that is just noise and will never match the output
349
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
351
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
352
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
354
def __init__(self, path, sha1_provider):
355
"""Create a DirState object.
357
:param path: The path at which the dirstate file on disk should live.
358
:param sha1_provider: an object meeting the Sha1Provider interface.
360
# _header_state and _dirblock_state represent the current state
361
# of the dirstate metadata and the per-row data respectiely.
362
# NOT_IN_MEMORY indicates that no data is in memory
363
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
364
# is the same as is on disk
365
# IN_MEMORY_MODIFIED indicates that we have a modified version
366
# of what is on disk.
367
# In future we will add more granularity, for instance _dirblock_state
368
# will probably support partially-in-memory as a separate variable,
369
# allowing for partially-in-memory unmodified and partially-in-memory
371
self._header_state = DirState.NOT_IN_MEMORY
372
self._dirblock_state = DirState.NOT_IN_MEMORY
373
# If true, an error has been detected while updating the dirstate, and
374
# for safety we're not going to commit to disk.
375
self._changes_aborted = False
379
self._state_file = None
380
self._filename = path
381
self._lock_token = None
382
self._lock_state = None
383
self._id_index = None
384
# a map from packed_stat to sha's.
385
self._packed_stat_index = None
386
self._end_of_header = None
387
self._cutoff_time = None
388
self._split_path_cache = {}
389
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
390
self._sha1_provider = sha1_provider
391
if 'hashcache' in debug.debug_flags:
392
self._sha1_file = self._sha1_file_and_mutter
394
self._sha1_file = self._sha1_provider.sha1
395
# These two attributes provide a simple cache for lookups into the
396
# dirstate in-memory vectors. By probing respectively for the last
397
# block, and for the next entry, we save nearly 2 bisections per path
399
self._last_block_index = None
400
self._last_entry_index = None
404
(self.__class__.__name__, self._filename)
406
def add(self, path, file_id, kind, stat, fingerprint):
407
"""Add a path to be tracked.
409
:param path: The path within the dirstate - '' is the root, 'foo' is the
410
path foo within the root, 'foo/bar' is the path bar within foo
412
:param file_id: The file id of the path being added.
413
:param kind: The kind of the path, as a string like 'file',
415
:param stat: The output of os.lstat for the path.
416
:param fingerprint: The sha value of the file's canonical form (i.e.
417
after any read filters have been applied),
418
or the target of a symlink,
419
or the referenced revision id for tree-references,
420
or '' for directories.
423
# find the block its in.
424
# find the location in the block.
425
# check its not there
427
#------- copied from inventory.ensure_normalized_name - keep synced.
428
# --- normalized_filename wants a unicode basename only, so get one.
429
dirname, basename = osutils.split(path)
430
# we dont import normalized_filename directly because we want to be
431
# able to change the implementation at runtime for tests.
432
norm_name, can_access = osutils.normalized_filename(basename)
433
if norm_name != basename:
437
raise errors.InvalidNormalization(path)
438
# you should never have files called . or ..; just add the directory
439
# in the parent, or according to the special treatment for the root
440
if basename == '.' or basename == '..':
441
raise errors.InvalidEntryName(path)
442
# now that we've normalised, we need the correct utf8 path and
443
# dirname and basename elements. This single encode and split should be
444
# faster than three separate encodes.
445
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
446
dirname, basename = osutils.split(utf8path)
447
# uses __class__ for speed; the check is needed for safety
448
if file_id.__class__ is not str:
449
raise AssertionError(
450
"must be a utf8 file_id not %s" % (type(file_id), ))
451
# Make sure the file_id does not exist in this tree
453
file_id_entry = self._get_entry(0, fileid_utf8=file_id, include_deleted=True)
454
if file_id_entry != (None, None):
455
if file_id_entry[1][0][0] == 'a':
456
if file_id_entry[0] != (dirname, basename, file_id):
457
# set the old name's current operation to rename
458
self.update_minimal(file_id_entry[0],
464
rename_from = file_id_entry[0][0:2]
466
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
467
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
468
info = '%s:%s' % (kind, path)
469
raise errors.DuplicateFileId(file_id, info)
470
first_key = (dirname, basename, '')
471
block_index, present = self._find_block_index_from_key(first_key)
473
# check the path is not in the tree
474
block = self._dirblocks[block_index][1]
475
entry_index, _ = self._find_entry_index(first_key, block)
476
while (entry_index < len(block) and
477
block[entry_index][0][0:2] == first_key[0:2]):
478
if block[entry_index][1][0][0] not in 'ar':
479
# this path is in the dirstate in the current tree.
480
raise Exception, "adding already added path!"
483
# The block where we want to put the file is not present. But it
484
# might be because the directory was empty, or not loaded yet. Look
485
# for a parent entry, if not found, raise NotVersionedError
486
parent_dir, parent_base = osutils.split(dirname)
487
parent_block_idx, parent_entry_idx, _, parent_present = \
488
self._get_block_entry_index(parent_dir, parent_base, 0)
489
if not parent_present:
490
raise errors.NotVersionedError(path, str(self))
491
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
492
block = self._dirblocks[block_index][1]
493
entry_key = (dirname, basename, file_id)
496
packed_stat = DirState.NULLSTAT
499
packed_stat = pack_stat(stat)
500
parent_info = self._empty_parent_info()
501
minikind = DirState._kind_to_minikind[kind]
502
if rename_from is not None:
504
old_path_utf8 = '%s/%s' % rename_from
506
old_path_utf8 = rename_from[1]
507
parent_info[0] = ('r', old_path_utf8, 0, False, '')
509
entry_data = entry_key, [
510
(minikind, fingerprint, size, False, packed_stat),
512
elif kind == 'directory':
513
entry_data = entry_key, [
514
(minikind, '', 0, False, packed_stat),
516
elif kind == 'symlink':
517
entry_data = entry_key, [
518
(minikind, fingerprint, size, False, packed_stat),
520
elif kind == 'tree-reference':
521
entry_data = entry_key, [
522
(minikind, fingerprint, 0, False, packed_stat),
525
raise errors.BzrError('unknown kind %r' % kind)
526
entry_index, present = self._find_entry_index(entry_key, block)
528
block.insert(entry_index, entry_data)
530
if block[entry_index][1][0][0] != 'a':
531
raise AssertionError(" %r(%r) already added" % (basename, file_id))
532
block[entry_index][1][0] = entry_data[1][0]
534
if kind == 'directory':
535
# insert a new dirblock
536
self._ensure_block(block_index, entry_index, utf8path)
537
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
539
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
541
def _bisect(self, paths):
542
"""Bisect through the disk structure for specific rows.
544
:param paths: A list of paths to find
545
:return: A dict mapping path => entries for found entries. Missing
546
entries will not be in the map.
547
The list is not sorted, and entries will be populated
548
based on when they were read.
550
self._requires_lock()
551
# We need the file pointer to be right after the initial header block
552
self._read_header_if_needed()
553
# If _dirblock_state was in memory, we should just return info from
554
# there, this function is only meant to handle when we want to read
556
if self._dirblock_state != DirState.NOT_IN_MEMORY:
557
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
559
# The disk representation is generally info + '\0\n\0' at the end. But
560
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
561
# Because it means we can sync on the '\n'
562
state_file = self._state_file
563
file_size = os.fstat(state_file.fileno()).st_size
564
# We end up with 2 extra fields, we should have a trailing '\n' to
565
# ensure that we read the whole record, and we should have a precursur
566
# '' which ensures that we start after the previous '\n'
567
entry_field_count = self._fields_per_entry() + 1
569
low = self._end_of_header
570
high = file_size - 1 # Ignore the final '\0'
571
# Map from (dir, name) => entry
574
# Avoid infinite seeking
575
max_count = 30*len(paths)
577
# pending is a list of places to look.
578
# each entry is a tuple of low, high, dir_names
579
# low -> the first byte offset to read (inclusive)
580
# high -> the last byte offset (inclusive)
581
# dir_names -> The list of (dir, name) pairs that should be found in
582
# the [low, high] range
583
pending = [(low, high, paths)]
585
page_size = self._bisect_page_size
587
fields_to_entry = self._get_fields_to_entry()
590
low, high, cur_files = pending.pop()
592
if not cur_files or low >= high:
597
if count > max_count:
598
raise errors.BzrError('Too many seeks, most likely a bug.')
600
mid = max(low, (low+high-page_size)/2)
603
# limit the read size, so we don't end up reading data that we have
605
read_size = min(page_size, (high-mid)+1)
606
block = state_file.read(read_size)
609
entries = block.split('\n')
612
# We didn't find a '\n', so we cannot have found any records.
613
# So put this range back and try again. But we know we have to
614
# increase the page size, because a single read did not contain
615
# a record break (so records must be larger than page_size)
617
pending.append((low, high, cur_files))
620
# Check the first and last entries, in case they are partial, or if
621
# we don't care about the rest of this page
623
first_fields = entries[0].split('\0')
624
if len(first_fields) < entry_field_count:
625
# We didn't get the complete first entry
626
# so move start, and grab the next, which
627
# should be a full entry
628
start += len(entries[0])+1
629
first_fields = entries[1].split('\0')
632
if len(first_fields) <= 2:
633
# We didn't even get a filename here... what do we do?
634
# Try a large page size and repeat this query
636
pending.append((low, high, cur_files))
639
# Find what entries we are looking for, which occur before and
640
# after this first record.
643
first_path = first_fields[1] + '/' + first_fields[2]
645
first_path = first_fields[2]
646
first_loc = _bisect_path_left(cur_files, first_path)
648
# These exist before the current location
649
pre = cur_files[:first_loc]
650
# These occur after the current location, which may be in the
651
# data we read, or might be after the last entry
652
post = cur_files[first_loc:]
654
if post and len(first_fields) >= entry_field_count:
655
# We have files after the first entry
657
# Parse the last entry
658
last_entry_num = len(entries)-1
659
last_fields = entries[last_entry_num].split('\0')
660
if len(last_fields) < entry_field_count:
661
# The very last hunk was not complete,
662
# read the previous hunk
663
after = mid + len(block) - len(entries[-1])
665
last_fields = entries[last_entry_num].split('\0')
667
after = mid + len(block)
670
last_path = last_fields[1] + '/' + last_fields[2]
672
last_path = last_fields[2]
673
last_loc = _bisect_path_right(post, last_path)
675
middle_files = post[:last_loc]
676
post = post[last_loc:]
679
# We have files that should occur in this block
680
# (>= first, <= last)
681
# Either we will find them here, or we can mark them as
684
if middle_files[0] == first_path:
685
# We might need to go before this location
686
pre.append(first_path)
687
if middle_files[-1] == last_path:
688
post.insert(0, last_path)
690
# Find out what paths we have
691
paths = {first_path:[first_fields]}
692
# last_path might == first_path so we need to be
693
# careful if we should append rather than overwrite
694
if last_entry_num != first_entry_num:
695
paths.setdefault(last_path, []).append(last_fields)
696
for num in xrange(first_entry_num+1, last_entry_num):
697
# TODO: jam 20070223 We are already splitting here, so
698
# shouldn't we just split the whole thing rather
699
# than doing the split again in add_one_record?
700
fields = entries[num].split('\0')
702
path = fields[1] + '/' + fields[2]
705
paths.setdefault(path, []).append(fields)
707
for path in middle_files:
708
for fields in paths.get(path, []):
709
# offset by 1 because of the opening '\0'
710
# consider changing fields_to_entry to avoid the
712
entry = fields_to_entry(fields[1:])
713
found.setdefault(path, []).append(entry)
715
# Now we have split up everything into pre, middle, and post, and
716
# we have handled everything that fell in 'middle'.
717
# We add 'post' first, so that we prefer to seek towards the
718
# beginning, so that we will tend to go as early as we need, and
719
# then only seek forward after that.
721
pending.append((after, high, post))
723
pending.append((low, start-1, pre))
725
# Consider that we may want to return the directory entries in sorted
726
# order. For now, we just return them in whatever order we found them,
727
# and leave it up to the caller if they care if it is ordered or not.
730
def _bisect_dirblocks(self, dir_list):
731
"""Bisect through the disk structure to find entries in given dirs.
733
_bisect_dirblocks is meant to find the contents of directories, which
734
differs from _bisect, which only finds individual entries.
736
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
737
:return: A map from dir => entries_for_dir
739
# TODO: jam 20070223 A lot of the bisecting logic could be shared
740
# between this and _bisect. It would require parameterizing the
741
# inner loop with a function, though. We should evaluate the
742
# performance difference.
743
self._requires_lock()
744
# We need the file pointer to be right after the initial header block
745
self._read_header_if_needed()
746
# If _dirblock_state was in memory, we should just return info from
747
# there, this function is only meant to handle when we want to read
749
if self._dirblock_state != DirState.NOT_IN_MEMORY:
750
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
751
# The disk representation is generally info + '\0\n\0' at the end. But
752
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
753
# Because it means we can sync on the '\n'
754
state_file = self._state_file
755
file_size = os.fstat(state_file.fileno()).st_size
756
# We end up with 2 extra fields, we should have a trailing '\n' to
757
# ensure that we read the whole record, and we should have a precursur
758
# '' which ensures that we start after the previous '\n'
759
entry_field_count = self._fields_per_entry() + 1
761
low = self._end_of_header
762
high = file_size - 1 # Ignore the final '\0'
763
# Map from dir => entry
766
# Avoid infinite seeking
767
max_count = 30*len(dir_list)
769
# pending is a list of places to look.
770
# each entry is a tuple of low, high, dir_names
771
# low -> the first byte offset to read (inclusive)
772
# high -> the last byte offset (inclusive)
773
# dirs -> The list of directories that should be found in
774
# the [low, high] range
775
pending = [(low, high, dir_list)]
777
page_size = self._bisect_page_size
779
fields_to_entry = self._get_fields_to_entry()
782
low, high, cur_dirs = pending.pop()
784
if not cur_dirs or low >= high:
789
if count > max_count:
790
raise errors.BzrError('Too many seeks, most likely a bug.')
792
mid = max(low, (low+high-page_size)/2)
795
# limit the read size, so we don't end up reading data that we have
797
read_size = min(page_size, (high-mid)+1)
798
block = state_file.read(read_size)
801
entries = block.split('\n')
804
# We didn't find a '\n', so we cannot have found any records.
805
# So put this range back and try again. But we know we have to
806
# increase the page size, because a single read did not contain
807
# a record break (so records must be larger than page_size)
809
pending.append((low, high, cur_dirs))
812
# Check the first and last entries, in case they are partial, or if
813
# we don't care about the rest of this page
815
first_fields = entries[0].split('\0')
816
if len(first_fields) < entry_field_count:
817
# We didn't get the complete first entry
818
# so move start, and grab the next, which
819
# should be a full entry
820
start += len(entries[0])+1
821
first_fields = entries[1].split('\0')
824
if len(first_fields) <= 1:
825
# We didn't even get a dirname here... what do we do?
826
# Try a large page size and repeat this query
828
pending.append((low, high, cur_dirs))
831
# Find what entries we are looking for, which occur before and
832
# after this first record.
834
first_dir = first_fields[1]
835
first_loc = bisect.bisect_left(cur_dirs, first_dir)
837
# These exist before the current location
838
pre = cur_dirs[:first_loc]
839
# These occur after the current location, which may be in the
840
# data we read, or might be after the last entry
841
post = cur_dirs[first_loc:]
843
if post and len(first_fields) >= entry_field_count:
844
# We have records to look at after the first entry
846
# Parse the last entry
847
last_entry_num = len(entries)-1
848
last_fields = entries[last_entry_num].split('\0')
849
if len(last_fields) < entry_field_count:
850
# The very last hunk was not complete,
851
# read the previous hunk
852
after = mid + len(block) - len(entries[-1])
854
last_fields = entries[last_entry_num].split('\0')
856
after = mid + len(block)
858
last_dir = last_fields[1]
859
last_loc = bisect.bisect_right(post, last_dir)
861
middle_files = post[:last_loc]
862
post = post[last_loc:]
865
# We have files that should occur in this block
866
# (>= first, <= last)
867
# Either we will find them here, or we can mark them as
870
if middle_files[0] == first_dir:
871
# We might need to go before this location
872
pre.append(first_dir)
873
if middle_files[-1] == last_dir:
874
post.insert(0, last_dir)
876
# Find out what paths we have
877
paths = {first_dir:[first_fields]}
878
# last_dir might == first_dir so we need to be
879
# careful if we should append rather than overwrite
880
if last_entry_num != first_entry_num:
881
paths.setdefault(last_dir, []).append(last_fields)
882
for num in xrange(first_entry_num+1, last_entry_num):
883
# TODO: jam 20070223 We are already splitting here, so
884
# shouldn't we just split the whole thing rather
885
# than doing the split again in add_one_record?
886
fields = entries[num].split('\0')
887
paths.setdefault(fields[1], []).append(fields)
889
for cur_dir in middle_files:
890
for fields in paths.get(cur_dir, []):
891
# offset by 1 because of the opening '\0'
892
# consider changing fields_to_entry to avoid the
894
entry = fields_to_entry(fields[1:])
895
found.setdefault(cur_dir, []).append(entry)
897
# Now we have split up everything into pre, middle, and post, and
898
# we have handled everything that fell in 'middle'.
899
# We add 'post' first, so that we prefer to seek towards the
900
# beginning, so that we will tend to go as early as we need, and
901
# then only seek forward after that.
903
pending.append((after, high, post))
905
pending.append((low, start-1, pre))
909
def _bisect_recursive(self, paths):
910
"""Bisect for entries for all paths and their children.
912
This will use bisect to find all records for the supplied paths. It
913
will then continue to bisect for any records which are marked as
914
directories. (and renames?)
916
:param paths: A sorted list of (dir, name) pairs
917
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
918
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
920
# Map from (dir, name, file_id) => [tree_info]
923
found_dir_names = set()
925
# Directories that have been read
926
processed_dirs = set()
927
# Get the ball rolling with the first bisect for all entries.
928
newly_found = self._bisect(paths)
931
# Directories that need to be read
933
paths_to_search = set()
934
for entry_list in newly_found.itervalues():
935
for dir_name_id, trees_info in entry_list:
936
found[dir_name_id] = trees_info
937
found_dir_names.add(dir_name_id[:2])
939
for tree_info in trees_info:
940
minikind = tree_info[0]
943
# We already processed this one as a directory,
944
# we don't need to do the extra work again.
946
subdir, name, file_id = dir_name_id
947
path = osutils.pathjoin(subdir, name)
949
if path not in processed_dirs:
950
pending_dirs.add(path)
951
elif minikind == 'r':
952
# Rename, we need to directly search the target
953
# which is contained in the fingerprint column
954
dir_name = osutils.split(tree_info[1])
955
if dir_name[0] in pending_dirs:
956
# This entry will be found in the dir search
958
if dir_name not in found_dir_names:
959
paths_to_search.add(tree_info[1])
960
# Now we have a list of paths to look for directly, and
961
# directory blocks that need to be read.
962
# newly_found is mixing the keys between (dir, name) and path
963
# entries, but that is okay, because we only really care about the
965
newly_found = self._bisect(sorted(paths_to_search))
966
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
967
processed_dirs.update(pending_dirs)
970
def _discard_merge_parents(self):
971
"""Discard any parents trees beyond the first.
973
Note that if this fails the dirstate is corrupted.
975
After this function returns the dirstate contains 2 trees, neither of
978
self._read_header_if_needed()
979
parents = self.get_parent_ids()
982
# only require all dirblocks if we are doing a full-pass removal.
983
self._read_dirblocks_if_needed()
984
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
985
def iter_entries_removable():
986
for block in self._dirblocks:
987
deleted_positions = []
988
for pos, entry in enumerate(block[1]):
990
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
991
deleted_positions.append(pos)
992
if deleted_positions:
993
if len(deleted_positions) == len(block[1]):
996
for pos in reversed(deleted_positions):
998
# if the first parent is a ghost:
999
if parents[0] in self.get_ghosts():
1000
empty_parent = [DirState.NULL_PARENT_DETAILS]
1001
for entry in iter_entries_removable():
1002
entry[1][1:] = empty_parent
1004
for entry in iter_entries_removable():
1008
self._parents = [parents[0]]
1009
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1010
self._header_state = DirState.IN_MEMORY_MODIFIED
1012
def _empty_parent_info(self):
1013
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
1016
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
1017
"""Ensure a block for dirname exists.
1019
This function exists to let callers which know that there is a
1020
directory dirname ensure that the block for it exists. This block can
1021
fail to exist because of demand loading, or because a directory had no
1022
children. In either case it is not an error. It is however an error to
1023
call this if there is no parent entry for the directory, and thus the
1024
function requires the coordinates of such an entry to be provided.
1026
The root row is special cased and can be indicated with a parent block
1029
:param parent_block_index: The index of the block in which dirname's row
1031
:param parent_row_index: The index in the parent block where the row
1033
:param dirname: The utf8 dirname to ensure there is a block for.
1034
:return: The index for the block.
1036
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
1037
# This is the signature of the root row, and the
1038
# contents-of-root row is always index 1
1040
# the basename of the directory must be the end of its full name.
1041
if not (parent_block_index == -1 and
1042
parent_block_index == -1 and dirname == ''):
1043
if not dirname.endswith(
1044
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
1045
raise AssertionError("bad dirname %r" % dirname)
1046
block_index, present = self._find_block_index_from_key((dirname, '', ''))
1048
## In future, when doing partial parsing, this should load and
1049
# populate the entire block.
1050
self._dirblocks.insert(block_index, (dirname, []))
1053
def _entries_to_current_state(self, new_entries):
1054
"""Load new_entries into self.dirblocks.
1056
Process new_entries into the current state object, making them the active
1057
state. The entries are grouped together by directory to form dirblocks.
1059
:param new_entries: A sorted list of entries. This function does not sort
1060
to prevent unneeded overhead when callers have a sorted list already.
1063
if new_entries[0][0][0:2] != ('', ''):
1064
raise AssertionError(
1065
"Missing root row %r" % (new_entries[0][0],))
1066
# The two blocks here are deliberate: the root block and the
1067
# contents-of-root block.
1068
self._dirblocks = [('', []), ('', [])]
1069
current_block = self._dirblocks[0][1]
1070
current_dirname = ''
1072
append_entry = current_block.append
1073
for entry in new_entries:
1074
if entry[0][0] != current_dirname:
1075
# new block - different dirname
1077
current_dirname = entry[0][0]
1078
self._dirblocks.append((current_dirname, current_block))
1079
append_entry = current_block.append
1080
# append the entry to the current block
1082
self._split_root_dirblock_into_contents()
1084
def _split_root_dirblock_into_contents(self):
1085
"""Split the root dirblocks into root and contents-of-root.
1087
After parsing by path, we end up with root entries and contents-of-root
1088
entries in the same block. This loop splits them out again.
1090
# The above loop leaves the "root block" entries mixed with the
1091
# "contents-of-root block". But we don't want an if check on
1092
# all entries, so instead we just fix it up here.
1093
if self._dirblocks[1] != ('', []):
1094
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1096
contents_of_root_block = []
1097
for entry in self._dirblocks[0][1]:
1098
if not entry[0][1]: # This is a root entry
1099
root_block.append(entry)
1101
contents_of_root_block.append(entry)
1102
self._dirblocks[0] = ('', root_block)
1103
self._dirblocks[1] = ('', contents_of_root_block)
1105
def _entries_for_path(self, path):
1106
"""Return a list with all the entries that match path for all ids."""
1107
dirname, basename = os.path.split(path)
1108
key = (dirname, basename, '')
1109
block_index, present = self._find_block_index_from_key(key)
1111
# the block which should contain path is absent.
1114
block = self._dirblocks[block_index][1]
1115
entry_index, _ = self._find_entry_index(key, block)
1116
# we may need to look at multiple entries at this path: walk while the specific_files match.
1117
while (entry_index < len(block) and
1118
block[entry_index][0][0:2] == key[0:2]):
1119
result.append(block[entry_index])
1123
def _entry_to_line(self, entry):
1124
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1126
:param entry: An entry_tuple as defined in the module docstring.
1128
entire_entry = list(entry[0])
1129
for tree_number, tree_data in enumerate(entry[1]):
1130
# (minikind, fingerprint, size, executable, tree_specific_string)
1131
entire_entry.extend(tree_data)
1132
# 3 for the key, 5 for the fields per tree.
1133
tree_offset = 3 + tree_number * 5
1135
entire_entry[tree_offset + 0] = tree_data[0]
1137
entire_entry[tree_offset + 2] = str(tree_data[2])
1139
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1140
return '\0'.join(entire_entry)
1142
def _fields_per_entry(self):
1143
"""How many null separated fields should be in each entry row.
1145
Each line now has an extra '\n' field which is not used
1146
so we just skip over it
1148
3 fields for the key
1149
+ number of fields per tree_data (5) * tree count
1152
tree_count = 1 + self._num_present_parents()
1153
return 3 + 5 * tree_count + 1
1155
def _find_block(self, key, add_if_missing=False):
1156
"""Return the block that key should be present in.
1158
:param key: A dirstate entry key.
1159
:return: The block tuple.
1161
block_index, present = self._find_block_index_from_key(key)
1163
if not add_if_missing:
1164
# check to see if key is versioned itself - we might want to
1165
# add it anyway, because dirs with no entries dont get a
1166
# dirblock at parse time.
1167
# This is an uncommon branch to take: most dirs have children,
1168
# and most code works with versioned paths.
1169
parent_base, parent_name = osutils.split(key[0])
1170
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1171
# some parent path has not been added - its an error to add
1173
raise errors.NotVersionedError(key[0:2], str(self))
1174
self._dirblocks.insert(block_index, (key[0], []))
1175
return self._dirblocks[block_index]
1177
def _find_block_index_from_key(self, key):
1178
"""Find the dirblock index for a key.
1180
:return: The block index, True if the block for the key is present.
1182
if key[0:2] == ('', ''):
1185
if (self._last_block_index is not None and
1186
self._dirblocks[self._last_block_index][0] == key[0]):
1187
return self._last_block_index, True
1190
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1191
cache=self._split_path_cache)
1192
# _right returns one-past-where-key is so we have to subtract
1193
# one to use it. we use _right here because there are two
1194
# '' blocks - the root, and the contents of root
1195
# we always have a minimum of 2 in self._dirblocks: root and
1196
# root-contents, and for '', we get 2 back, so this is
1197
# simple and correct:
1198
present = (block_index < len(self._dirblocks) and
1199
self._dirblocks[block_index][0] == key[0])
1200
self._last_block_index = block_index
1201
# Reset the entry index cache to the beginning of the block.
1202
self._last_entry_index = -1
1203
return block_index, present
1205
def _find_entry_index(self, key, block):
1206
"""Find the entry index for a key in a block.
1208
:return: The entry index, True if the entry for the key is present.
1210
len_block = len(block)
1212
if self._last_entry_index is not None:
1214
entry_index = self._last_entry_index + 1
1215
# A hit is when the key is after the last slot, and before or
1216
# equal to the next slot.
1217
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1218
key <= block[entry_index][0]):
1219
self._last_entry_index = entry_index
1220
present = (block[entry_index][0] == key)
1221
return entry_index, present
1224
entry_index = bisect.bisect_left(block, (key, []))
1225
present = (entry_index < len_block and
1226
block[entry_index][0] == key)
1227
self._last_entry_index = entry_index
1228
return entry_index, present
1231
def from_tree(tree, dir_state_filename, sha1_provider=None):
1232
"""Create a dirstate from a bzr Tree.
1234
:param tree: The tree which should provide parent information and
1236
:param sha1_provider: an object meeting the Sha1Provider interface.
1237
If None, a DefaultSha1Provider is used.
1238
:return: a DirState object which is currently locked for writing.
1239
(it was locked by DirState.initialize)
1241
result = DirState.initialize(dir_state_filename,
1242
sha1_provider=sha1_provider)
1246
parent_ids = tree.get_parent_ids()
1247
num_parents = len(parent_ids)
1249
for parent_id in parent_ids:
1250
parent_tree = tree.branch.repository.revision_tree(parent_id)
1251
parent_trees.append((parent_id, parent_tree))
1252
parent_tree.lock_read()
1253
result.set_parent_trees(parent_trees, [])
1254
result.set_state_from_inventory(tree.inventory)
1256
for revid, parent_tree in parent_trees:
1257
parent_tree.unlock()
1260
# The caller won't have a chance to unlock this, so make sure we
1266
def update_by_delta(self, delta):
1267
"""Apply an inventory delta to the dirstate for tree 0
1269
:param delta: An inventory delta. See Inventory.apply_delta for
1272
self._read_dirblocks_if_needed()
1275
for old_path, new_path, file_id, inv_entry in sorted(delta, reverse=True):
1276
if (file_id in insertions) or (file_id in removals):
1277
raise AssertionError("repeated file id in delta %r" % (file_id,))
1278
if old_path is not None:
1279
old_path = old_path.encode('utf-8')
1280
removals[file_id] = old_path
1281
if new_path is not None:
1282
new_path = new_path.encode('utf-8')
1283
dirname, basename = osutils.split(new_path)
1284
key = (dirname, basename, file_id)
1285
minikind = DirState._kind_to_minikind[inv_entry.kind]
1287
fingerprint = inv_entry.reference_revision
1290
insertions[file_id] = (key, minikind, inv_entry.executable,
1291
fingerprint, new_path)
1292
# Transform moves into delete+add pairs
1293
if None not in (old_path, new_path):
1294
for child in self._iter_child_entries(0, old_path):
1295
if child[0][2] in insertions or child[0][2] in removals:
1297
child_dirname = child[0][0]
1298
child_basename = child[0][1]
1299
minikind = child[1][0][0]
1300
fingerprint = child[1][0][4]
1301
executable = child[1][0][3]
1302
old_child_path = osutils.pathjoin(child[0][0],
1304
removals[child[0][2]] = old_child_path
1305
child_suffix = child_dirname[len(old_path):]
1306
new_child_dirname = (new_path + child_suffix)
1307
key = (new_child_dirname, child_basename, child[0][2])
1308
new_child_path = os.path.join(new_child_dirname,
1310
insertions[child[0][2]] = (key, minikind, executable,
1311
fingerprint, new_child_path)
1312
self._apply_removals(removals.values())
1313
self._apply_insertions(insertions.values())
1315
def _apply_removals(self, removals):
1316
for path in sorted(removals, reverse=True):
1317
dirname, basename = osutils.split(path)
1318
block_i, entry_i, d_present, f_present = \
1319
self._get_block_entry_index(dirname, basename, 0)
1320
entry = self._dirblocks[block_i][1][entry_i]
1321
self._make_absent(entry)
1322
# See if we have a malformed delta: deleting a directory must not
1323
# leave crud behind. This increases the number of bisects needed
1324
# substantially, but deletion or renames of large numbers of paths
1325
# is rare enough it shouldn't be an issue (famous last words?) RBC
1327
block_i, entry_i, d_present, f_present = \
1328
self._get_block_entry_index(path, '', 0)
1330
# The dir block is still present in the dirstate; this could
1331
# be due to it being in a parent tree, or a corrupt delta.
1332
for child_entry in self._dirblocks[block_i][1]:
1333
if child_entry[1][0][0] not in ('r', 'a'):
1334
raise errors.InconsistentDelta(path, entry[0][2],
1335
"The file id was deleted but its children were "
1338
def _apply_insertions(self, adds):
1339
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1340
self.update_minimal(key, minikind, executable, fingerprint,
1341
path_utf8=path_utf8)
1343
def update_basis_by_delta(self, delta, new_revid):
1344
"""Update the parents of this tree after a commit.
1346
This gives the tree one parent, with revision id new_revid. The
1347
inventory delta is applied to the current basis tree to generate the
1348
inventory for the parent new_revid, and all other parent trees are
1351
Note that an exception during the operation of this method will leave
1352
the dirstate in a corrupt state where it should not be saved.
1354
Finally, we expect all changes to be synchronising the basis tree with
1357
:param new_revid: The new revision id for the trees parent.
1358
:param delta: An inventory delta (see apply_inventory_delta) describing
1359
the changes from the current left most parent revision to new_revid.
1361
self._read_dirblocks_if_needed()
1362
self._discard_merge_parents()
1363
if self._ghosts != []:
1364
raise NotImplementedError(self.update_basis_by_delta)
1365
if len(self._parents) == 0:
1366
# setup a blank tree, the most simple way.
1367
empty_parent = DirState.NULL_PARENT_DETAILS
1368
for entry in self._iter_entries():
1369
entry[1].append(empty_parent)
1370
self._parents.append(new_revid)
1372
self._parents[0] = new_revid
1374
delta = sorted(delta, reverse=True)
1378
# The paths this function accepts are unicode and must be encoded as we
1380
encode = cache_utf8.encode
1381
inv_to_entry = self._inv_entry_to_details
1382
# delta is now (deletes, changes), (adds) in reverse lexographical
1384
# deletes in reverse lexographic order are safe to process in situ.
1385
# renames are not, as a rename from any path could go to a path
1386
# lexographically lower, so we transform renames into delete, add pairs,
1387
# expanding them recursively as needed.
1388
# At the same time, to reduce interface friction we convert the input
1389
# inventory entries to dirstate.
1390
root_only = ('', '')
1391
for old_path, new_path, file_id, inv_entry in delta:
1392
if old_path is None:
1393
adds.append((None, encode(new_path), file_id,
1394
inv_to_entry(inv_entry), True))
1395
elif new_path is None:
1396
deletes.append((encode(old_path), None, file_id, None, True))
1397
elif (old_path, new_path) != root_only:
1399
# Because renames must preserve their children we must have
1400
# processed all relocations and removes before hand. The sort
1401
# order ensures we've examined the child paths, but we also
1402
# have to execute the removals, or the split to an add/delete
1403
# pair will result in the deleted item being reinserted, or
1404
# renamed items being reinserted twice - and possibly at the
1405
# wrong place. Splitting into a delete/add pair also simplifies
1406
# the handling of entries with ('f', ...), ('r' ...) because
1407
# the target of the 'r' is old_path here, and we add that to
1408
# deletes, meaning that the add handler does not need to check
1409
# for 'r' items on every pass.
1410
self._update_basis_apply_deletes(deletes)
1412
new_path_utf8 = encode(new_path)
1413
# Split into an add/delete pair recursively.
1414
adds.append((None, new_path_utf8, file_id,
1415
inv_to_entry(inv_entry), False))
1416
# Expunge deletes that we've seen so that deleted/renamed
1417
# children of a rename directory are handled correctly.
1418
new_deletes = reversed(list(self._iter_child_entries(1,
1420
# Remove the current contents of the tree at orig_path, and
1421
# reinsert at the correct new path.
1422
for entry in new_deletes:
1424
source_path = entry[0][0] + '/' + entry[0][1]
1426
source_path = entry[0][1]
1428
target_path = new_path_utf8 + source_path[len(old_path):]
1431
raise AssertionError("cannot rename directory to"
1433
target_path = source_path[len(old_path) + 1:]
1434
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1436
(source_path, target_path, entry[0][2], None, False))
1438
(encode(old_path), new_path, file_id, None, False))
1440
# changes to just the root should not require remove/insertion
1442
changes.append((encode(old_path), encode(new_path), file_id,
1443
inv_to_entry(inv_entry)))
1445
# Finish expunging deletes/first half of renames.
1446
self._update_basis_apply_deletes(deletes)
1447
# Reinstate second half of renames and new paths.
1448
self._update_basis_apply_adds(adds)
1449
# Apply in-situ changes.
1450
self._update_basis_apply_changes(changes)
1452
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1453
self._header_state = DirState.IN_MEMORY_MODIFIED
1454
self._id_index = None
1457
def _update_basis_apply_adds(self, adds):
1458
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1460
They may be adds, or renames that have been split into add/delete
1463
:param adds: A sequence of adds. Each add is a tuple:
1464
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1465
is False when the add is the second half of a remove-and-reinsert
1466
pair created to handle renames and deletes.
1468
# Adds are accumulated partly from renames, so can be in any input
1471
# adds is now in lexographic order, which places all parents before
1472
# their children, so we can process it linearly.
1474
for old_path, new_path, file_id, new_details, real_add in adds:
1475
# the entry for this file_id must be in tree 0.
1476
entry = self._get_entry(0, file_id, new_path)
1477
if entry[0] is None or entry[0][2] != file_id:
1478
self._changes_aborted = True
1479
raise errors.InconsistentDelta(new_path, file_id,
1480
'working tree does not contain new entry')
1481
if real_add and entry[1][1][0] not in absent:
1482
self._changes_aborted = True
1483
raise errors.InconsistentDelta(new_path, file_id,
1484
'The entry was considered to be a genuinely new record,'
1485
' but there was already an old record for it.')
1486
# We don't need to update the target of an 'r' because the handling
1487
# of renames turns all 'r' situations into a delete at the original
1489
entry[1][1] = new_details
1491
def _update_basis_apply_changes(self, changes):
1492
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1494
:param adds: A sequence of changes. Each change is a tuple:
1495
(path_utf8, path_utf8, file_id, (entry_details))
1498
for old_path, new_path, file_id, new_details in changes:
1499
# the entry for this file_id must be in tree 0.
1500
entry = self._get_entry(0, file_id, new_path)
1501
if entry[0] is None or entry[0][2] != file_id:
1502
self._changes_aborted = True
1503
raise errors.InconsistentDelta(new_path, file_id,
1504
'working tree does not contain new entry')
1505
if (entry[1][0][0] in absent or
1506
entry[1][1][0] in absent):
1507
self._changes_aborted = True
1508
raise errors.InconsistentDelta(new_path, file_id,
1509
'changed considered absent')
1510
entry[1][1] = new_details
1512
def _update_basis_apply_deletes(self, deletes):
1513
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1515
They may be deletes, or renames that have been split into add/delete
1518
:param deletes: A sequence of deletes. Each delete is a tuple:
1519
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1520
real_delete is True when the desired outcome is an actual deletion
1521
rather than the rename handling logic temporarily deleting a path
1522
during the replacement of a parent.
1524
null = DirState.NULL_PARENT_DETAILS
1525
for old_path, new_path, file_id, _, real_delete in deletes:
1526
if real_delete != (new_path is None):
1527
raise AssertionError("bad delete delta")
1528
# the entry for this file_id must be in tree 1.
1529
dirname, basename = osutils.split(old_path)
1530
block_index, entry_index, dir_present, file_present = \
1531
self._get_block_entry_index(dirname, basename, 1)
1532
if not file_present:
1533
self._changes_aborted = True
1534
raise errors.InconsistentDelta(old_path, file_id,
1535
'basis tree does not contain removed entry')
1536
entry = self._dirblocks[block_index][1][entry_index]
1537
if entry[0][2] != file_id:
1538
self._changes_aborted = True
1539
raise errors.InconsistentDelta(old_path, file_id,
1540
'mismatched file_id in tree 1')
1542
if entry[1][0][0] != 'a':
1543
self._changes_aborted = True
1544
raise errors.InconsistentDelta(old_path, file_id,
1545
'This was marked as a real delete, but the WT state'
1546
' claims that it still exists and is versioned.')
1547
del self._dirblocks[block_index][1][entry_index]
1549
if entry[1][0][0] == 'a':
1550
self._changes_aborted = True
1551
raise errors.InconsistentDelta(old_path, file_id,
1552
'The entry was considered a rename, but the source path'
1553
' is marked as absent.')
1554
# For whatever reason, we were asked to rename an entry
1555
# that was originally marked as deleted. This could be
1556
# because we are renaming the parent directory, and the WT
1557
# current state has the file marked as deleted.
1558
elif entry[1][0][0] == 'r':
1559
# implement the rename
1560
del self._dirblocks[block_index][1][entry_index]
1562
# it is being resurrected here, so blank it out temporarily.
1563
self._dirblocks[block_index][1][entry_index][1][1] = null
1565
def _observed_sha1(self, entry, sha1, stat_value,
1566
_stat_to_minikind=_stat_to_minikind, _pack_stat=pack_stat):
1567
"""Note the sha1 of a file.
1569
:param entry: The entry the sha1 is for.
1570
:param sha1: The observed sha1.
1571
:param stat_value: The os.lstat for the file.
1574
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1578
packed_stat = _pack_stat(stat_value)
1580
if self._cutoff_time is None:
1581
self._sha_cutoff_time()
1582
if (stat_value.st_mtime < self._cutoff_time
1583
and stat_value.st_ctime < self._cutoff_time):
1584
entry[1][0] = ('f', sha1, entry[1][0][2], entry[1][0][3],
1586
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1588
def _sha_cutoff_time(self):
1589
"""Return cutoff time.
1591
Files modified more recently than this time are at risk of being
1592
undetectably modified and so can't be cached.
1594
# Cache the cutoff time as long as we hold a lock.
1595
# time.time() isn't super expensive (approx 3.38us), but
1596
# when you call it 50,000 times it adds up.
1597
# For comparison, os.lstat() costs 7.2us if it is hot.
1598
self._cutoff_time = int(time.time()) - 3
1599
return self._cutoff_time
1601
def _lstat(self, abspath, entry):
1602
"""Return the os.lstat value for this path."""
1603
return os.lstat(abspath)
1605
def _sha1_file_and_mutter(self, abspath):
1606
# when -Dhashcache is turned on, this is monkey-patched in to log
1608
trace.mutter("dirstate sha1 " + abspath)
1609
return self._sha1_provider.sha1(abspath)
1611
def _is_executable(self, mode, old_executable):
1612
"""Is this file executable?"""
1613
return bool(S_IEXEC & mode)
1615
def _is_executable_win32(self, mode, old_executable):
1616
"""On win32 the executable bit is stored in the dirstate."""
1617
return old_executable
1619
if sys.platform == 'win32':
1620
_is_executable = _is_executable_win32
1622
def _read_link(self, abspath, old_link):
1623
"""Read the target of a symlink"""
1624
# TODO: jam 200700301 On Win32, this could just return the value
1625
# already in memory. However, this really needs to be done at a
1626
# higher level, because there either won't be anything on disk,
1627
# or the thing on disk will be a file.
1628
fs_encoding = osutils._fs_enc
1629
if isinstance(abspath, unicode):
1630
# abspath is defined as the path to pass to lstat. readlink is
1631
# buggy in python < 2.6 (it doesn't encode unicode path into FS
1632
# encoding), so we need to encode ourselves knowing that unicode
1633
# paths are produced by UnicodeDirReader on purpose.
1634
abspath = abspath.encode(fs_encoding)
1635
target = os.readlink(abspath)
1636
if fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1637
# Change encoding if needed
1638
target = target.decode(fs_encoding).encode('UTF-8')
1641
def get_ghosts(self):
1642
"""Return a list of the parent tree revision ids that are ghosts."""
1643
self._read_header_if_needed()
1646
def get_lines(self):
1647
"""Serialise the entire dirstate to a sequence of lines."""
1648
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1649
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1650
# read whats on disk.
1651
self._state_file.seek(0)
1652
return self._state_file.readlines()
1654
lines.append(self._get_parents_line(self.get_parent_ids()))
1655
lines.append(self._get_ghosts_line(self._ghosts))
1656
# append the root line which is special cased
1657
lines.extend(map(self._entry_to_line, self._iter_entries()))
1658
return self._get_output_lines(lines)
1660
def _get_ghosts_line(self, ghost_ids):
1661
"""Create a line for the state file for ghost information."""
1662
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1664
def _get_parents_line(self, parent_ids):
1665
"""Create a line for the state file for parents information."""
1666
return '\0'.join([str(len(parent_ids))] + parent_ids)
1668
def _get_fields_to_entry(self):
1669
"""Get a function which converts entry fields into a entry record.
1671
This handles size and executable, as well as parent records.
1673
:return: A function which takes a list of fields, and returns an
1674
appropriate record for storing in memory.
1676
# This is intentionally unrolled for performance
1677
num_present_parents = self._num_present_parents()
1678
if num_present_parents == 0:
1679
def fields_to_entry_0_parents(fields, _int=int):
1680
path_name_file_id_key = (fields[0], fields[1], fields[2])
1681
return (path_name_file_id_key, [
1683
fields[3], # minikind
1684
fields[4], # fingerprint
1685
_int(fields[5]), # size
1686
fields[6] == 'y', # executable
1687
fields[7], # packed_stat or revision_id
1689
return fields_to_entry_0_parents
1690
elif num_present_parents == 1:
1691
def fields_to_entry_1_parent(fields, _int=int):
1692
path_name_file_id_key = (fields[0], fields[1], fields[2])
1693
return (path_name_file_id_key, [
1695
fields[3], # minikind
1696
fields[4], # fingerprint
1697
_int(fields[5]), # size
1698
fields[6] == 'y', # executable
1699
fields[7], # packed_stat or revision_id
1702
fields[8], # minikind
1703
fields[9], # fingerprint
1704
_int(fields[10]), # size
1705
fields[11] == 'y', # executable
1706
fields[12], # packed_stat or revision_id
1709
return fields_to_entry_1_parent
1710
elif num_present_parents == 2:
1711
def fields_to_entry_2_parents(fields, _int=int):
1712
path_name_file_id_key = (fields[0], fields[1], fields[2])
1713
return (path_name_file_id_key, [
1715
fields[3], # minikind
1716
fields[4], # fingerprint
1717
_int(fields[5]), # size
1718
fields[6] == 'y', # executable
1719
fields[7], # packed_stat or revision_id
1722
fields[8], # minikind
1723
fields[9], # fingerprint
1724
_int(fields[10]), # size
1725
fields[11] == 'y', # executable
1726
fields[12], # packed_stat or revision_id
1729
fields[13], # minikind
1730
fields[14], # fingerprint
1731
_int(fields[15]), # size
1732
fields[16] == 'y', # executable
1733
fields[17], # packed_stat or revision_id
1736
return fields_to_entry_2_parents
1738
def fields_to_entry_n_parents(fields, _int=int):
1739
path_name_file_id_key = (fields[0], fields[1], fields[2])
1740
trees = [(fields[cur], # minikind
1741
fields[cur+1], # fingerprint
1742
_int(fields[cur+2]), # size
1743
fields[cur+3] == 'y', # executable
1744
fields[cur+4], # stat or revision_id
1745
) for cur in xrange(3, len(fields)-1, 5)]
1746
return path_name_file_id_key, trees
1747
return fields_to_entry_n_parents
1749
def get_parent_ids(self):
1750
"""Return a list of the parent tree ids for the directory state."""
1751
self._read_header_if_needed()
1752
return list(self._parents)
1754
def _get_block_entry_index(self, dirname, basename, tree_index):
1755
"""Get the coordinates for a path in the state structure.
1757
:param dirname: The utf8 dirname to lookup.
1758
:param basename: The utf8 basename to lookup.
1759
:param tree_index: The index of the tree for which this lookup should
1761
:return: A tuple describing where the path is located, or should be
1762
inserted. The tuple contains four fields: the block index, the row
1763
index, the directory is present (boolean), the entire path is
1764
present (boolean). There is no guarantee that either
1765
coordinate is currently reachable unless the found field for it is
1766
True. For instance, a directory not present in the searched tree
1767
may be returned with a value one greater than the current highest
1768
block offset. The directory present field will always be True when
1769
the path present field is True. The directory present field does
1770
NOT indicate that the directory is present in the searched tree,
1771
rather it indicates that there are at least some files in some
1774
self._read_dirblocks_if_needed()
1775
key = dirname, basename, ''
1776
block_index, present = self._find_block_index_from_key(key)
1778
# no such directory - return the dir index and 0 for the row.
1779
return block_index, 0, False, False
1780
block = self._dirblocks[block_index][1] # access the entries only
1781
entry_index, present = self._find_entry_index(key, block)
1782
# linear search through entries at this path to find the one
1784
while entry_index < len(block) and block[entry_index][0][1] == basename:
1785
if block[entry_index][1][tree_index][0] not in 'ar':
1786
# neither absent or relocated
1787
return block_index, entry_index, True, True
1789
return block_index, entry_index, True, False
1791
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None, include_deleted=False):
1792
"""Get the dirstate entry for path in tree tree_index.
1794
If either file_id or path is supplied, it is used as the key to lookup.
1795
If both are supplied, the fastest lookup is used, and an error is
1796
raised if they do not both point at the same row.
1798
:param tree_index: The index of the tree we wish to locate this path
1799
in. If the path is present in that tree, the entry containing its
1800
details is returned, otherwise (None, None) is returned
1801
0 is the working tree, higher indexes are successive parent
1803
:param fileid_utf8: A utf8 file_id to look up.
1804
:param path_utf8: An utf8 path to be looked up.
1805
:param include_deleted: If True, and performing a lookup via
1806
fileid_utf8 rather than path_utf8, return an entry for deleted
1808
:return: The dirstate entry tuple for path, or (None, None)
1810
self._read_dirblocks_if_needed()
1811
if path_utf8 is not None:
1812
if type(path_utf8) is not str:
1813
raise AssertionError('path_utf8 is not a str: %s %s'
1814
% (type(path_utf8), path_utf8))
1815
# path lookups are faster
1816
dirname, basename = osutils.split(path_utf8)
1817
block_index, entry_index, dir_present, file_present = \
1818
self._get_block_entry_index(dirname, basename, tree_index)
1819
if not file_present:
1821
entry = self._dirblocks[block_index][1][entry_index]
1822
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
1823
raise AssertionError('unversioned entry?')
1825
if entry[0][2] != fileid_utf8:
1826
self._changes_aborted = True
1827
raise errors.BzrError('integrity error ? : mismatching'
1828
' tree_index, file_id and path')
1831
possible_keys = self._get_id_index().get(fileid_utf8, None)
1832
if not possible_keys:
1834
for key in possible_keys:
1835
block_index, present = \
1836
self._find_block_index_from_key(key)
1837
# strange, probably indicates an out of date
1838
# id index - for now, allow this.
1841
# WARNING: DO not change this code to use _get_block_entry_index
1842
# as that function is not suitable: it does not use the key
1843
# to lookup, and thus the wrong coordinates are returned.
1844
block = self._dirblocks[block_index][1]
1845
entry_index, present = self._find_entry_index(key, block)
1847
entry = self._dirblocks[block_index][1][entry_index]
1848
if entry[1][tree_index][0] in 'fdlt':
1849
# this is the result we are looking for: the
1850
# real home of this file_id in this tree.
1852
if entry[1][tree_index][0] == 'a':
1853
# there is no home for this entry in this tree
1857
if entry[1][tree_index][0] != 'r':
1858
raise AssertionError(
1859
"entry %r has invalid minikind %r for tree %r" \
1861
entry[1][tree_index][0],
1863
real_path = entry[1][tree_index][1]
1864
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1865
path_utf8=real_path)
1869
def initialize(cls, path, sha1_provider=None):
1870
"""Create a new dirstate on path.
1872
The new dirstate will be an empty tree - that is it has no parents,
1873
and only a root node - which has id ROOT_ID.
1875
:param path: The name of the file for the dirstate.
1876
:param sha1_provider: an object meeting the Sha1Provider interface.
1877
If None, a DefaultSha1Provider is used.
1878
:return: A write-locked DirState object.
1880
# This constructs a new DirState object on a path, sets the _state_file
1881
# to a new empty file for that path. It then calls _set_data() with our
1882
# stock empty dirstate information - a root with ROOT_ID, no children,
1883
# and no parents. Finally it calls save() to ensure that this data will
1885
if sha1_provider is None:
1886
sha1_provider = DefaultSha1Provider()
1887
result = cls(path, sha1_provider)
1888
# root dir and root dir contents with no children.
1889
empty_tree_dirblocks = [('', []), ('', [])]
1890
# a new root directory, with a NULLSTAT.
1891
empty_tree_dirblocks[0][1].append(
1892
(('', '', inventory.ROOT_ID), [
1893
('d', '', 0, False, DirState.NULLSTAT),
1897
result._set_data([], empty_tree_dirblocks)
1905
def _inv_entry_to_details(inv_entry):
1906
"""Convert an inventory entry (from a revision tree) to state details.
1908
:param inv_entry: An inventory entry whose sha1 and link targets can be
1909
relied upon, and which has a revision set.
1910
:return: A details tuple - the details for a single tree at a path +
1913
kind = inv_entry.kind
1914
minikind = DirState._kind_to_minikind[kind]
1915
tree_data = inv_entry.revision
1916
if kind == 'directory':
1920
elif kind == 'symlink':
1921
if inv_entry.symlink_target is None:
1924
fingerprint = inv_entry.symlink_target.encode('utf8')
1927
elif kind == 'file':
1928
fingerprint = inv_entry.text_sha1 or ''
1929
size = inv_entry.text_size or 0
1930
executable = inv_entry.executable
1931
elif kind == 'tree-reference':
1932
fingerprint = inv_entry.reference_revision or ''
1936
raise Exception("can't pack %s" % inv_entry)
1937
return (minikind, fingerprint, size, executable, tree_data)
1939
def _iter_child_entries(self, tree_index, path_utf8):
1940
"""Iterate over all the entries that are children of path_utf.
1942
This only returns entries that are present (not in 'a', 'r') in
1943
tree_index. tree_index data is not refreshed, so if tree 0 is used,
1944
results may differ from that obtained if paths were statted to
1945
determine what ones were directories.
1947
Asking for the children of a non-directory will return an empty
1951
next_pending_dirs = [path_utf8]
1953
while next_pending_dirs:
1954
pending_dirs = next_pending_dirs
1955
next_pending_dirs = []
1956
for path in pending_dirs:
1957
block_index, present = self._find_block_index_from_key(
1959
if block_index == 0:
1961
if len(self._dirblocks) == 1:
1962
# asked for the children of the root with no other
1966
# children of a non-directory asked for.
1968
block = self._dirblocks[block_index]
1969
for entry in block[1]:
1970
kind = entry[1][tree_index][0]
1971
if kind not in absent:
1975
path = entry[0][0] + '/' + entry[0][1]
1978
next_pending_dirs.append(path)
1980
def _iter_entries(self):
1981
"""Iterate over all the entries in the dirstate.
1983
Each yelt item is an entry in the standard format described in the
1984
docstring of bzrlib.dirstate.
1986
self._read_dirblocks_if_needed()
1987
for directory in self._dirblocks:
1988
for entry in directory[1]:
1991
def _get_id_index(self):
1992
"""Get an id index of self._dirblocks."""
1993
if self._id_index is None:
1995
for key, tree_details in self._iter_entries():
1996
id_index.setdefault(key[2], set()).add(key)
1997
self._id_index = id_index
1998
return self._id_index
2000
def _get_output_lines(self, lines):
2001
"""Format lines for final output.
2003
:param lines: A sequence of lines containing the parents list and the
2006
output_lines = [DirState.HEADER_FORMAT_3]
2007
lines.append('') # a final newline
2008
inventory_text = '\0\n\0'.join(lines)
2009
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
2010
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
2011
num_entries = len(lines)-3
2012
output_lines.append('num_entries: %s\n' % (num_entries,))
2013
output_lines.append(inventory_text)
2016
def _make_deleted_row(self, fileid_utf8, parents):
2017
"""Return a deleted row for fileid_utf8."""
2018
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
2021
def _num_present_parents(self):
2022
"""The number of parent entries in each record row."""
2023
return len(self._parents) - len(self._ghosts)
2026
def on_file(path, sha1_provider=None):
2027
"""Construct a DirState on the file at path path.
2029
:param path: The path at which the dirstate file on disk should live.
2030
:param sha1_provider: an object meeting the Sha1Provider interface.
2031
If None, a DefaultSha1Provider is used.
2032
:return: An unlocked DirState object, associated with the given path.
2034
if sha1_provider is None:
2035
sha1_provider = DefaultSha1Provider()
2036
result = DirState(path, sha1_provider)
2039
def _read_dirblocks_if_needed(self):
2040
"""Read in all the dirblocks from the file if they are not in memory.
2042
This populates self._dirblocks, and sets self._dirblock_state to
2043
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
2046
self._read_header_if_needed()
2047
if self._dirblock_state == DirState.NOT_IN_MEMORY:
2048
_read_dirblocks(self)
2050
def _read_header(self):
2051
"""This reads in the metadata header, and the parent ids.
2053
After reading in, the file should be positioned at the null
2054
just before the start of the first record in the file.
2056
:return: (expected crc checksum, number of entries, parent list)
2058
self._read_prelude()
2059
parent_line = self._state_file.readline()
2060
info = parent_line.split('\0')
2061
num_parents = int(info[0])
2062
self._parents = info[1:-1]
2063
ghost_line = self._state_file.readline()
2064
info = ghost_line.split('\0')
2065
num_ghosts = int(info[1])
2066
self._ghosts = info[2:-1]
2067
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2068
self._end_of_header = self._state_file.tell()
2070
def _read_header_if_needed(self):
2071
"""Read the header of the dirstate file if needed."""
2072
# inline this as it will be called a lot
2073
if not self._lock_token:
2074
raise errors.ObjectNotLocked(self)
2075
if self._header_state == DirState.NOT_IN_MEMORY:
2078
def _read_prelude(self):
2079
"""Read in the prelude header of the dirstate file.
2081
This only reads in the stuff that is not connected to the crc
2082
checksum. The position will be correct to read in the rest of
2083
the file and check the checksum after this point.
2084
The next entry in the file should be the number of parents,
2085
and their ids. Followed by a newline.
2087
header = self._state_file.readline()
2088
if header != DirState.HEADER_FORMAT_3:
2089
raise errors.BzrError(
2090
'invalid header line: %r' % (header,))
2091
crc_line = self._state_file.readline()
2092
if not crc_line.startswith('crc32: '):
2093
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2094
self.crc_expected = int(crc_line[len('crc32: '):-1])
2095
num_entries_line = self._state_file.readline()
2096
if not num_entries_line.startswith('num_entries: '):
2097
raise errors.BzrError('missing num_entries line')
2098
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2100
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
2101
"""Find a sha1 given a stat lookup."""
2102
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
2104
def _get_packed_stat_index(self):
2105
"""Get a packed_stat index of self._dirblocks."""
2106
if self._packed_stat_index is None:
2108
for key, tree_details in self._iter_entries():
2109
if tree_details[0][0] == 'f':
2110
index[tree_details[0][4]] = tree_details[0][1]
2111
self._packed_stat_index = index
2112
return self._packed_stat_index
2115
"""Save any pending changes created during this session.
2117
We reuse the existing file, because that prevents race conditions with
2118
file creation, and use oslocks on it to prevent concurrent modification
2119
and reads - because dirstate's incremental data aggregation is not
2120
compatible with reading a modified file, and replacing a file in use by
2121
another process is impossible on Windows.
2123
A dirstate in read only mode should be smart enough though to validate
2124
that the file has not changed, and otherwise discard its cache and
2125
start over, to allow for fine grained read lock duration, so 'status'
2126
wont block 'commit' - for example.
2128
if self._changes_aborted:
2129
# Should this be a warning? For now, I'm expecting that places that
2130
# mark it inconsistent will warn, making a warning here redundant.
2131
trace.mutter('Not saving DirState because '
2132
'_changes_aborted is set.')
2134
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
2135
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2137
grabbed_write_lock = False
2138
if self._lock_state != 'w':
2139
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2140
# Switch over to the new lock, as the old one may be closed.
2141
# TODO: jam 20070315 We should validate the disk file has
2142
# not changed contents. Since temporary_write_lock may
2143
# not be an atomic operation.
2144
self._lock_token = new_lock
2145
self._state_file = new_lock.f
2146
if not grabbed_write_lock:
2147
# We couldn't grab a write lock, so we switch back to a read one
2150
self._state_file.seek(0)
2151
self._state_file.writelines(self.get_lines())
2152
self._state_file.truncate()
2153
self._state_file.flush()
2154
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2155
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
2157
if grabbed_write_lock:
2158
self._lock_token = self._lock_token.restore_read_lock()
2159
self._state_file = self._lock_token.f
2160
# TODO: jam 20070315 We should validate the disk file has
2161
# not changed contents. Since restore_read_lock may
2162
# not be an atomic operation.
2164
def _set_data(self, parent_ids, dirblocks):
2165
"""Set the full dirstate data in memory.
2167
This is an internal function used to completely replace the objects
2168
in memory state. It puts the dirstate into state 'full-dirty'.
2170
:param parent_ids: A list of parent tree revision ids.
2171
:param dirblocks: A list containing one tuple for each directory in the
2172
tree. Each tuple contains the directory path and a list of entries
2173
found in that directory.
2175
# our memory copy is now authoritative.
2176
self._dirblocks = dirblocks
2177
self._header_state = DirState.IN_MEMORY_MODIFIED
2178
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2179
self._parents = list(parent_ids)
2180
self._id_index = None
2181
self._packed_stat_index = None
2183
def set_path_id(self, path, new_id):
2184
"""Change the id of path to new_id in the current working tree.
2186
:param path: The path inside the tree to set - '' is the root, 'foo'
2187
is the path foo in the root.
2188
:param new_id: The new id to assign to the path. This must be a utf8
2189
file id (not unicode, and not None).
2191
self._read_dirblocks_if_needed()
2193
# TODO: logic not written
2194
raise NotImplementedError(self.set_path_id)
2195
# TODO: check new id is unique
2196
entry = self._get_entry(0, path_utf8=path)
2197
if entry[0][2] == new_id:
2198
# Nothing to change.
2200
# mark the old path absent, and insert a new root path
2201
self._make_absent(entry)
2202
self.update_minimal(('', '', new_id), 'd',
2203
path_utf8='', packed_stat=entry[1][0][4])
2204
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2205
if self._id_index is not None:
2206
self._id_index.setdefault(new_id, set()).add(entry[0])
2208
def set_parent_trees(self, trees, ghosts):
2209
"""Set the parent trees for the dirstate.
2211
:param trees: A list of revision_id, tree tuples. tree must be provided
2212
even if the revision_id refers to a ghost: supply an empty tree in
2214
:param ghosts: A list of the revision_ids that are ghosts at the time
2217
# TODO: generate a list of parent indexes to preserve to save
2218
# processing specific parent trees. In the common case one tree will
2219
# be preserved - the left most parent.
2220
# TODO: if the parent tree is a dirstate, we might want to walk them
2221
# all by path in parallel for 'optimal' common-case performance.
2222
# generate new root row.
2223
self._read_dirblocks_if_needed()
2224
# TODO future sketch: Examine the existing parents to generate a change
2225
# map and then walk the new parent trees only, mapping them into the
2226
# dirstate. Walk the dirstate at the same time to remove unreferenced
2229
# sketch: loop over all entries in the dirstate, cherry picking
2230
# entries from the parent trees, if they are not ghost trees.
2231
# after we finish walking the dirstate, all entries not in the dirstate
2232
# are deletes, so we want to append them to the end as per the design
2233
# discussions. So do a set difference on ids with the parents to
2234
# get deletes, and add them to the end.
2235
# During the update process we need to answer the following questions:
2236
# - find other keys containing a fileid in order to create cross-path
2237
# links. We dont't trivially use the inventory from other trees
2238
# because this leads to either double touching, or to accessing
2240
# - find other keys containing a path
2241
# We accumulate each entry via this dictionary, including the root
2244
# we could do parallel iterators, but because file id data may be
2245
# scattered throughout, we dont save on index overhead: we have to look
2246
# at everything anyway. We can probably save cycles by reusing parent
2247
# data and doing an incremental update when adding an additional
2248
# parent, but for now the common cases are adding a new parent (merge),
2249
# and replacing completely (commit), and commit is more common: so
2250
# optimise merge later.
2252
# ---- start generation of full tree mapping data
2253
# what trees should we use?
2254
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2255
# how many trees do we end up with
2256
parent_count = len(parent_trees)
2258
# one: the current tree
2259
for entry in self._iter_entries():
2260
# skip entries not in the current tree
2261
if entry[1][0][0] in 'ar': # absent, relocated
2263
by_path[entry[0]] = [entry[1][0]] + \
2264
[DirState.NULL_PARENT_DETAILS] * parent_count
2265
id_index[entry[0][2]] = set([entry[0]])
2267
# now the parent trees:
2268
for tree_index, tree in enumerate(parent_trees):
2269
# the index is off by one, adjust it.
2270
tree_index = tree_index + 1
2271
# when we add new locations for a fileid we need these ranges for
2272
# any fileid in this tree as we set the by_path[id] to:
2273
# already_processed_tree_details + new_details + new_location_suffix
2274
# the suffix is from tree_index+1:parent_count+1.
2275
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2276
# now stitch in all the entries from this tree
2277
for path, entry in tree.inventory.iter_entries_by_dir():
2278
# here we process each trees details for each item in the tree.
2279
# we first update any existing entries for the id at other paths,
2280
# then we either create or update the entry for the id at the
2281
# right path, and finally we add (if needed) a mapping from
2282
# file_id to this path. We do it in this order to allow us to
2283
# avoid checking all known paths for the id when generating a
2284
# new entry at this path: by adding the id->path mapping last,
2285
# all the mappings are valid and have correct relocation
2286
# records where needed.
2287
file_id = entry.file_id
2288
path_utf8 = path.encode('utf8')
2289
dirname, basename = osutils.split(path_utf8)
2290
new_entry_key = (dirname, basename, file_id)
2291
# tree index consistency: All other paths for this id in this tree
2292
# index must point to the correct path.
2293
for entry_key in id_index.setdefault(file_id, set()):
2294
# TODO:PROFILING: It might be faster to just update
2295
# rather than checking if we need to, and then overwrite
2296
# the one we are located at.
2297
if entry_key != new_entry_key:
2298
# this file id is at a different path in one of the
2299
# other trees, so put absent pointers there
2300
# This is the vertical axis in the matrix, all pointing
2302
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2303
# by path consistency: Insert into an existing path record (trivial), or
2304
# add a new one with relocation pointers for the other tree indexes.
2305
if new_entry_key in id_index[file_id]:
2306
# there is already an entry where this data belongs, just insert it.
2307
by_path[new_entry_key][tree_index] = \
2308
self._inv_entry_to_details(entry)
2310
# add relocated entries to the horizontal axis - this row
2311
# mapping from path,id. We need to look up the correct path
2312
# for the indexes from 0 to tree_index -1
2314
for lookup_index in xrange(tree_index):
2315
# boundary case: this is the first occurence of file_id
2316
# so there are no id_indexs, possibly take this out of
2318
if not len(id_index[file_id]):
2319
new_details.append(DirState.NULL_PARENT_DETAILS)
2321
# grab any one entry, use it to find the right path.
2322
# TODO: optimise this to reduce memory use in highly
2323
# fragmented situations by reusing the relocation
2325
a_key = iter(id_index[file_id]).next()
2326
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2327
# its a pointer or missing statement, use it as is.
2328
new_details.append(by_path[a_key][lookup_index])
2330
# we have the right key, make a pointer to it.
2331
real_path = ('/'.join(a_key[0:2])).strip('/')
2332
new_details.append(('r', real_path, 0, False, ''))
2333
new_details.append(self._inv_entry_to_details(entry))
2334
new_details.extend(new_location_suffix)
2335
by_path[new_entry_key] = new_details
2336
id_index[file_id].add(new_entry_key)
2337
# --- end generation of full tree mappings
2339
# sort and output all the entries
2340
new_entries = self._sort_entries(by_path.items())
2341
self._entries_to_current_state(new_entries)
2342
self._parents = [rev_id for rev_id, tree in trees]
2343
self._ghosts = list(ghosts)
2344
self._header_state = DirState.IN_MEMORY_MODIFIED
2345
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2346
self._id_index = id_index
2348
def _sort_entries(self, entry_list):
2349
"""Given a list of entries, sort them into the right order.
2351
This is done when constructing a new dirstate from trees - normally we
2352
try to keep everything in sorted blocks all the time, but sometimes
2353
it's easier to sort after the fact.
2356
# sort by: directory parts, file name, file id
2357
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2358
return sorted(entry_list, key=_key)
2360
def set_state_from_inventory(self, new_inv):
2361
"""Set new_inv as the current state.
2363
This API is called by tree transform, and will usually occur with
2364
existing parent trees.
2366
:param new_inv: The inventory object to set current state from.
2368
if 'evil' in debug.debug_flags:
2369
trace.mutter_callsite(1,
2370
"set_state_from_inventory called; please mutate the tree instead")
2371
self._read_dirblocks_if_needed()
2373
# Two iterators: current data and new data, both in dirblock order.
2374
# We zip them together, which tells about entries that are new in the
2375
# inventory, or removed in the inventory, or present in both and
2378
# You might think we could just synthesize a new dirstate directly
2379
# since we're processing it in the right order. However, we need to
2380
# also consider there may be any number of parent trees and relocation
2381
# pointers, and we don't want to duplicate that here.
2382
new_iterator = new_inv.iter_entries_by_dir()
2383
# we will be modifying the dirstate, so we need a stable iterator. In
2384
# future we might write one, for now we just clone the state into a
2385
# list - which is a shallow copy.
2386
old_iterator = iter(list(self._iter_entries()))
2387
# both must have roots so this is safe:
2388
current_new = new_iterator.next()
2389
current_old = old_iterator.next()
2390
def advance(iterator):
2392
return iterator.next()
2393
except StopIteration:
2395
while current_new or current_old:
2396
# skip entries in old that are not really there
2397
if current_old and current_old[1][0][0] in 'ar':
2398
# relocated or absent
2399
current_old = advance(old_iterator)
2402
# convert new into dirblock style
2403
new_path_utf8 = current_new[0].encode('utf8')
2404
new_dirname, new_basename = osutils.split(new_path_utf8)
2405
new_id = current_new[1].file_id
2406
new_entry_key = (new_dirname, new_basename, new_id)
2407
current_new_minikind = \
2408
DirState._kind_to_minikind[current_new[1].kind]
2409
if current_new_minikind == 't':
2410
fingerprint = current_new[1].reference_revision or ''
2412
# We normally only insert or remove records, or update
2413
# them when it has significantly changed. Then we want to
2414
# erase its fingerprint. Unaffected records should
2415
# normally not be updated at all.
2418
# for safety disable variables
2419
new_path_utf8 = new_dirname = new_basename = new_id = \
2420
new_entry_key = None
2421
# 5 cases, we dont have a value that is strictly greater than everything, so
2422
# we make both end conditions explicit
2424
# old is finished: insert current_new into the state.
2425
self.update_minimal(new_entry_key, current_new_minikind,
2426
executable=current_new[1].executable,
2427
path_utf8=new_path_utf8, fingerprint=fingerprint)
2428
current_new = advance(new_iterator)
2429
elif not current_new:
2431
self._make_absent(current_old)
2432
current_old = advance(old_iterator)
2433
elif new_entry_key == current_old[0]:
2434
# same - common case
2435
# We're looking at the same path and id in both the dirstate
2436
# and inventory, so just need to update the fields in the
2437
# dirstate from the one in the inventory.
2438
# TODO: update the record if anything significant has changed.
2439
# the minimal required trigger is if the execute bit or cached
2441
if (current_old[1][0][3] != current_new[1].executable or
2442
current_old[1][0][0] != current_new_minikind):
2443
self.update_minimal(current_old[0], current_new_minikind,
2444
executable=current_new[1].executable,
2445
path_utf8=new_path_utf8, fingerprint=fingerprint)
2446
# both sides are dealt with, move on
2447
current_old = advance(old_iterator)
2448
current_new = advance(new_iterator)
2449
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2450
or (new_dirname == current_old[0][0]
2451
and new_entry_key[1:] < current_old[0][1:])):
2453
# add a entry for this and advance new
2454
self.update_minimal(new_entry_key, current_new_minikind,
2455
executable=current_new[1].executable,
2456
path_utf8=new_path_utf8, fingerprint=fingerprint)
2457
current_new = advance(new_iterator)
2459
# we've advanced past the place where the old key would be,
2460
# without seeing it in the new list. so it must be gone.
2461
self._make_absent(current_old)
2462
current_old = advance(old_iterator)
2463
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2464
self._id_index = None
2465
self._packed_stat_index = None
2467
def _make_absent(self, current_old):
2468
"""Mark current_old - an entry - as absent for tree 0.
2470
:return: True if this was the last details entry for the entry key:
2471
that is, if the underlying block has had the entry removed, thus
2472
shrinking in length.
2474
# build up paths that this id will be left at after the change is made,
2475
# so we can update their cross references in tree 0
2476
all_remaining_keys = set()
2477
# Dont check the working tree, because it's going.
2478
for details in current_old[1][1:]:
2479
if details[0] not in 'ar': # absent, relocated
2480
all_remaining_keys.add(current_old[0])
2481
elif details[0] == 'r': # relocated
2482
# record the key for the real path.
2483
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2484
# absent rows are not present at any path.
2485
last_reference = current_old[0] not in all_remaining_keys
2487
# the current row consists entire of the current item (being marked
2488
# absent), and relocated or absent entries for the other trees:
2489
# Remove it, its meaningless.
2490
block = self._find_block(current_old[0])
2491
entry_index, present = self._find_entry_index(current_old[0], block[1])
2493
raise AssertionError('could not find entry for %s' % (current_old,))
2494
block[1].pop(entry_index)
2495
# if we have an id_index in use, remove this key from it for this id.
2496
if self._id_index is not None:
2497
self._id_index[current_old[0][2]].remove(current_old[0])
2498
# update all remaining keys for this id to record it as absent. The
2499
# existing details may either be the record we are marking as deleted
2500
# (if there were other trees with the id present at this path), or may
2502
for update_key in all_remaining_keys:
2503
update_block_index, present = \
2504
self._find_block_index_from_key(update_key)
2506
raise AssertionError('could not find block for %s' % (update_key,))
2507
update_entry_index, present = \
2508
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2510
raise AssertionError('could not find entry for %s' % (update_key,))
2511
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2512
# it must not be absent at the moment
2513
if update_tree_details[0][0] == 'a': # absent
2514
raise AssertionError('bad row %r' % (update_tree_details,))
2515
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2516
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2517
return last_reference
2519
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2520
packed_stat=None, size=0, path_utf8=None):
2521
"""Update an entry to the state in tree 0.
2523
This will either create a new entry at 'key' or update an existing one.
2524
It also makes sure that any other records which might mention this are
2527
:param key: (dir, name, file_id) for the new entry
2528
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2530
:param executable: Should the executable bit be set?
2531
:param fingerprint: Simple fingerprint for new entry: canonical-form
2532
sha1 for files, referenced revision id for subtrees, etc.
2533
:param packed_stat: Packed stat value for new entry.
2534
:param size: Size information for new entry
2535
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2538
If packed_stat and fingerprint are not given, they're invalidated in
2541
block = self._find_block(key)[1]
2542
if packed_stat is None:
2543
packed_stat = DirState.NULLSTAT
2544
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2545
# sometimes present in the dirstate - this seems oddly inconsistent.
2547
entry_index, present = self._find_entry_index(key, block)
2548
new_details = (minikind, fingerprint, size, executable, packed_stat)
2549
id_index = self._get_id_index()
2551
# new entry, synthesis cross reference here,
2552
existing_keys = id_index.setdefault(key[2], set())
2553
if not existing_keys:
2554
# not currently in the state, simplest case
2555
new_entry = key, [new_details] + self._empty_parent_info()
2557
# present at one or more existing other paths.
2558
# grab one of them and use it to generate parent
2559
# relocation/absent entries.
2560
new_entry = key, [new_details]
2561
for other_key in existing_keys:
2562
# change the record at other to be a pointer to this new
2563
# record. The loop looks similar to the change to
2564
# relocations when updating an existing record but its not:
2565
# the test for existing kinds is different: this can be
2566
# factored out to a helper though.
2567
other_block_index, present = self._find_block_index_from_key(other_key)
2569
raise AssertionError('could not find block for %s' % (other_key,))
2570
other_entry_index, present = self._find_entry_index(other_key,
2571
self._dirblocks[other_block_index][1])
2573
raise AssertionError('could not find entry for %s' % (other_key,))
2574
if path_utf8 is None:
2575
raise AssertionError('no path')
2576
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2577
('r', path_utf8, 0, False, '')
2579
num_present_parents = self._num_present_parents()
2580
for lookup_index in xrange(1, num_present_parents + 1):
2581
# grab any one entry, use it to find the right path.
2582
# TODO: optimise this to reduce memory use in highly
2583
# fragmented situations by reusing the relocation
2585
update_block_index, present = \
2586
self._find_block_index_from_key(other_key)
2588
raise AssertionError('could not find block for %s' % (other_key,))
2589
update_entry_index, present = \
2590
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2592
raise AssertionError('could not find entry for %s' % (other_key,))
2593
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2594
if update_details[0] in 'ar': # relocated, absent
2595
# its a pointer or absent in lookup_index's tree, use
2597
new_entry[1].append(update_details)
2599
# we have the right key, make a pointer to it.
2600
pointer_path = osutils.pathjoin(*other_key[0:2])
2601
new_entry[1].append(('r', pointer_path, 0, False, ''))
2602
block.insert(entry_index, new_entry)
2603
existing_keys.add(key)
2605
# Does the new state matter?
2606
block[entry_index][1][0] = new_details
2607
# parents cannot be affected by what we do.
2608
# other occurences of this id can be found
2609
# from the id index.
2611
# tree index consistency: All other paths for this id in this tree
2612
# index must point to the correct path. We have to loop here because
2613
# we may have passed entries in the state with this file id already
2614
# that were absent - where parent entries are - and they need to be
2615
# converted to relocated.
2616
if path_utf8 is None:
2617
raise AssertionError('no path')
2618
for entry_key in id_index.setdefault(key[2], set()):
2619
# TODO:PROFILING: It might be faster to just update
2620
# rather than checking if we need to, and then overwrite
2621
# the one we are located at.
2622
if entry_key != key:
2623
# this file id is at a different path in one of the
2624
# other trees, so put absent pointers there
2625
# This is the vertical axis in the matrix, all pointing
2627
block_index, present = self._find_block_index_from_key(entry_key)
2629
raise AssertionError('not present: %r', entry_key)
2630
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2632
raise AssertionError('not present: %r', entry_key)
2633
self._dirblocks[block_index][1][entry_index][1][0] = \
2634
('r', path_utf8, 0, False, '')
2635
# add a containing dirblock if needed.
2636
if new_details[0] == 'd':
2637
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2638
block_index, present = self._find_block_index_from_key(subdir_key)
2640
self._dirblocks.insert(block_index, (subdir_key[0], []))
2642
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2644
def _validate(self):
2645
"""Check that invariants on the dirblock are correct.
2647
This can be useful in debugging; it shouldn't be necessary in
2650
This must be called with a lock held.
2652
# NOTE: This must always raise AssertionError not just assert,
2653
# otherwise it may not behave properly under python -O
2655
# TODO: All entries must have some content that's not 'a' or 'r',
2656
# otherwise it could just be removed.
2658
# TODO: All relocations must point directly to a real entry.
2660
# TODO: No repeated keys.
2663
from pprint import pformat
2664
self._read_dirblocks_if_needed()
2665
if len(self._dirblocks) > 0:
2666
if not self._dirblocks[0][0] == '':
2667
raise AssertionError(
2668
"dirblocks don't start with root block:\n" + \
2669
pformat(self._dirblocks))
2670
if len(self._dirblocks) > 1:
2671
if not self._dirblocks[1][0] == '':
2672
raise AssertionError(
2673
"dirblocks missing root directory:\n" + \
2674
pformat(self._dirblocks))
2675
# the dirblocks are sorted by their path components, name, and dir id
2676
dir_names = [d[0].split('/')
2677
for d in self._dirblocks[1:]]
2678
if dir_names != sorted(dir_names):
2679
raise AssertionError(
2680
"dir names are not in sorted order:\n" + \
2681
pformat(self._dirblocks) + \
2684
for dirblock in self._dirblocks:
2685
# within each dirblock, the entries are sorted by filename and
2687
for entry in dirblock[1]:
2688
if dirblock[0] != entry[0][0]:
2689
raise AssertionError(
2691
"doesn't match directory name in\n%r" %
2692
(entry, pformat(dirblock)))
2693
if dirblock[1] != sorted(dirblock[1]):
2694
raise AssertionError(
2695
"dirblock for %r is not sorted:\n%s" % \
2696
(dirblock[0], pformat(dirblock)))
2698
def check_valid_parent():
2699
"""Check that the current entry has a valid parent.
2701
This makes sure that the parent has a record,
2702
and that the parent isn't marked as "absent" in the
2703
current tree. (It is invalid to have a non-absent file in an absent
2706
if entry[0][0:2] == ('', ''):
2707
# There should be no parent for the root row
2709
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2710
if parent_entry == (None, None):
2711
raise AssertionError(
2712
"no parent entry for: %s in tree %s"
2713
% (this_path, tree_index))
2714
if parent_entry[1][tree_index][0] != 'd':
2715
raise AssertionError(
2716
"Parent entry for %s is not marked as a valid"
2717
" directory. %s" % (this_path, parent_entry,))
2719
# For each file id, for each tree: either
2720
# the file id is not present at all; all rows with that id in the
2721
# key have it marked as 'absent'
2722
# OR the file id is present under exactly one name; any other entries
2723
# that mention that id point to the correct name.
2725
# We check this with a dict per tree pointing either to the present
2726
# name, or None if absent.
2727
tree_count = self._num_present_parents() + 1
2728
id_path_maps = [dict() for i in range(tree_count)]
2729
# Make sure that all renamed entries point to the correct location.
2730
for entry in self._iter_entries():
2731
file_id = entry[0][2]
2732
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2733
if len(entry[1]) != tree_count:
2734
raise AssertionError(
2735
"wrong number of entry details for row\n%s" \
2736
",\nexpected %d" % \
2737
(pformat(entry), tree_count))
2738
absent_positions = 0
2739
for tree_index, tree_state in enumerate(entry[1]):
2740
this_tree_map = id_path_maps[tree_index]
2741
minikind = tree_state[0]
2742
if minikind in 'ar':
2743
absent_positions += 1
2744
# have we seen this id before in this column?
2745
if file_id in this_tree_map:
2746
previous_path, previous_loc = this_tree_map[file_id]
2747
# any later mention of this file must be consistent with
2748
# what was said before
2750
if previous_path is not None:
2751
raise AssertionError(
2752
"file %s is absent in row %r but also present " \
2754
(file_id, entry, previous_path))
2755
elif minikind == 'r':
2756
target_location = tree_state[1]
2757
if previous_path != target_location:
2758
raise AssertionError(
2759
"file %s relocation in row %r but also at %r" \
2760
% (file_id, entry, previous_path))
2762
# a file, directory, etc - may have been previously
2763
# pointed to by a relocation, which must point here
2764
if previous_path != this_path:
2765
raise AssertionError(
2766
"entry %r inconsistent with previous path %r "
2768
(entry, previous_path, previous_loc))
2769
check_valid_parent()
2772
# absent; should not occur anywhere else
2773
this_tree_map[file_id] = None, this_path
2774
elif minikind == 'r':
2775
# relocation, must occur at expected location
2776
this_tree_map[file_id] = tree_state[1], this_path
2778
this_tree_map[file_id] = this_path, this_path
2779
check_valid_parent()
2780
if absent_positions == tree_count:
2781
raise AssertionError(
2782
"entry %r has no data for any tree." % (entry,))
2784
def _wipe_state(self):
2785
"""Forget all state information about the dirstate."""
2786
self._header_state = DirState.NOT_IN_MEMORY
2787
self._dirblock_state = DirState.NOT_IN_MEMORY
2788
self._changes_aborted = False
2791
self._dirblocks = []
2792
self._id_index = None
2793
self._packed_stat_index = None
2794
self._end_of_header = None
2795
self._cutoff_time = None
2796
self._split_path_cache = {}
2798
def lock_read(self):
2799
"""Acquire a read lock on the dirstate."""
2800
if self._lock_token is not None:
2801
raise errors.LockContention(self._lock_token)
2802
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2803
# already in memory, we could read just the header and check for
2804
# any modification. If not modified, we can just leave things
2806
self._lock_token = lock.ReadLock(self._filename)
2807
self._lock_state = 'r'
2808
self._state_file = self._lock_token.f
2811
def lock_write(self):
2812
"""Acquire a write lock on the dirstate."""
2813
if self._lock_token is not None:
2814
raise errors.LockContention(self._lock_token)
2815
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2816
# already in memory, we could read just the header and check for
2817
# any modification. If not modified, we can just leave things
2819
self._lock_token = lock.WriteLock(self._filename)
2820
self._lock_state = 'w'
2821
self._state_file = self._lock_token.f
2825
"""Drop any locks held on the dirstate."""
2826
if self._lock_token is None:
2827
raise errors.LockNotHeld(self)
2828
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2829
# already in memory, we could read just the header and check for
2830
# any modification. If not modified, we can just leave things
2832
self._state_file = None
2833
self._lock_state = None
2834
self._lock_token.unlock()
2835
self._lock_token = None
2836
self._split_path_cache = {}
2838
def _requires_lock(self):
2839
"""Check that a lock is currently held by someone on the dirstate."""
2840
if not self._lock_token:
2841
raise errors.ObjectNotLocked(self)
2844
def py_update_entry(state, entry, abspath, stat_value,
2845
_stat_to_minikind=DirState._stat_to_minikind,
2846
_pack_stat=pack_stat):
2847
"""Update the entry based on what is actually on disk.
2849
This function only calculates the sha if it needs to - if the entry is
2850
uncachable, or clearly different to the first parent's entry, no sha
2851
is calculated, and None is returned.
2853
:param state: The dirstate this entry is in.
2854
:param entry: This is the dirblock entry for the file in question.
2855
:param abspath: The path on disk for this file.
2856
:param stat_value: The stat value done on the path.
2857
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
2858
target of a symlink.
2861
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
2865
packed_stat = _pack_stat(stat_value)
2866
(saved_minikind, saved_link_or_sha1, saved_file_size,
2867
saved_executable, saved_packed_stat) = entry[1][0]
2869
if minikind == 'd' and saved_minikind == 't':
2871
if (minikind == saved_minikind
2872
and packed_stat == saved_packed_stat):
2873
# The stat hasn't changed since we saved, so we can re-use the
2878
# size should also be in packed_stat
2879
if saved_file_size == stat_value.st_size:
2880
return saved_link_or_sha1
2882
# If we have gotten this far, that means that we need to actually
2883
# process this entry.
2886
executable = state._is_executable(stat_value.st_mode,
2888
if state._cutoff_time is None:
2889
state._sha_cutoff_time()
2890
if (stat_value.st_mtime < state._cutoff_time
2891
and stat_value.st_ctime < state._cutoff_time
2892
and len(entry[1]) > 1
2893
and entry[1][1][0] != 'a'):
2894
# Could check for size changes for further optimised
2895
# avoidance of sha1's. However the most prominent case of
2896
# over-shaing is during initial add, which this catches.
2897
# Besides, if content filtering happens, size and sha
2898
# are calculated at the same time, so checking just the size
2899
# gains nothing w.r.t. performance.
2900
link_or_sha1 = state._sha1_file(abspath)
2901
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
2902
executable, packed_stat)
2904
entry[1][0] = ('f', '', stat_value.st_size,
2905
executable, DirState.NULLSTAT)
2906
elif minikind == 'd':
2908
entry[1][0] = ('d', '', 0, False, packed_stat)
2909
if saved_minikind != 'd':
2910
# This changed from something into a directory. Make sure we
2911
# have a directory block for it. This doesn't happen very
2912
# often, so this doesn't have to be super fast.
2913
block_index, entry_index, dir_present, file_present = \
2914
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
2915
state._ensure_block(block_index, entry_index,
2916
osutils.pathjoin(entry[0][0], entry[0][1]))
2917
elif minikind == 'l':
2918
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
2919
if state._cutoff_time is None:
2920
state._sha_cutoff_time()
2921
if (stat_value.st_mtime < state._cutoff_time
2922
and stat_value.st_ctime < state._cutoff_time):
2923
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
2926
entry[1][0] = ('l', '', stat_value.st_size,
2927
False, DirState.NULLSTAT)
2928
state._dirblock_state = DirState.IN_MEMORY_MODIFIED
2930
update_entry = py_update_entry
2933
class ProcessEntryPython(object):
2935
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id", "uninteresting",
2936
"last_source_parent", "last_target_parent", "include_unchanged",
2937
"use_filesystem_for_exec", "utf8_decode", "searched_specific_files",
2938
"search_specific_files", "state", "source_index", "target_index",
2939
"want_unversioned", "tree"]
2941
def __init__(self, include_unchanged, use_filesystem_for_exec,
2942
search_specific_files, state, source_index, target_index,
2943
want_unversioned, tree):
2944
self.old_dirname_to_file_id = {}
2945
self.new_dirname_to_file_id = {}
2946
# Just a sentry, so that _process_entry can say that this
2947
# record is handled, but isn't interesting to process (unchanged)
2948
self.uninteresting = object()
2949
# Using a list so that we can access the values and change them in
2950
# nested scope. Each one is [path, file_id, entry]
2951
self.last_source_parent = [None, None]
2952
self.last_target_parent = [None, None]
2953
self.include_unchanged = include_unchanged
2954
self.use_filesystem_for_exec = use_filesystem_for_exec
2955
self.utf8_decode = cache_utf8._utf8_decode
2956
# for all search_indexs in each path at or under each element of
2957
# search_specific_files, if the detail is relocated: add the id, and add the
2958
# relocated path as one to search if its not searched already. If the
2959
# detail is not relocated, add the id.
2960
self.searched_specific_files = set()
2961
self.search_specific_files = search_specific_files
2963
self.source_index = source_index
2964
self.target_index = target_index
2965
self.want_unversioned = want_unversioned
2968
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
2969
"""Compare an entry and real disk to generate delta information.
2971
:param path_info: top_relpath, basename, kind, lstat, abspath for
2972
the path of entry. If None, then the path is considered absent.
2973
(Perhaps we should pass in a concrete entry for this ?)
2974
Basename is returned as a utf8 string because we expect this
2975
tuple will be ignored, and don't want to take the time to
2977
:return: None if these don't match
2978
A tuple of information about the change, or
2979
the object 'uninteresting' if these match, but are
2980
basically identical.
2982
if self.source_index is None:
2983
source_details = DirState.NULL_PARENT_DETAILS
2985
source_details = entry[1][self.source_index]
2986
target_details = entry[1][self.target_index]
2987
target_minikind = target_details[0]
2988
if path_info is not None and target_minikind in 'fdlt':
2989
if not (self.target_index == 0):
2990
raise AssertionError()
2991
link_or_sha1 = update_entry(self.state, entry,
2992
abspath=path_info[4], stat_value=path_info[3])
2993
# The entry may have been modified by update_entry
2994
target_details = entry[1][self.target_index]
2995
target_minikind = target_details[0]
2998
file_id = entry[0][2]
2999
source_minikind = source_details[0]
3000
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
3001
# claimed content in both: diff
3002
# r | fdlt | | add source to search, add id path move and perform
3003
# | | | diff check on source-target
3004
# r | fdlt | a | dangling file that was present in the basis.
3006
if source_minikind in 'r':
3007
# add the source to the search path to find any children it
3008
# has. TODO ? : only add if it is a container ?
3009
if not osutils.is_inside_any(self.searched_specific_files,
3011
self.search_specific_files.add(source_details[1])
3012
# generate the old path; this is needed for stating later
3014
old_path = source_details[1]
3015
old_dirname, old_basename = os.path.split(old_path)
3016
path = pathjoin(entry[0][0], entry[0][1])
3017
old_entry = self.state._get_entry(self.source_index,
3019
# update the source details variable to be the real
3021
if old_entry == (None, None):
3022
raise errors.CorruptDirstate(self.state._filename,
3023
"entry '%s/%s' is considered renamed from %r"
3024
" but source does not exist\n"
3025
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
3026
source_details = old_entry[1][self.source_index]
3027
source_minikind = source_details[0]
3029
old_dirname = entry[0][0]
3030
old_basename = entry[0][1]
3031
old_path = path = None
3032
if path_info is None:
3033
# the file is missing on disk, show as removed.
3034
content_change = True
3038
# source and target are both versioned and disk file is present.
3039
target_kind = path_info[2]
3040
if target_kind == 'directory':
3042
old_path = path = pathjoin(old_dirname, old_basename)
3043
self.new_dirname_to_file_id[path] = file_id
3044
if source_minikind != 'd':
3045
content_change = True
3047
# directories have no fingerprint
3048
content_change = False
3050
elif target_kind == 'file':
3051
if source_minikind != 'f':
3052
content_change = True
3054
# If the size is the same, check the sha:
3055
if target_details[2] == source_details[2]:
3056
if link_or_sha1 is None:
3058
statvalue, link_or_sha1 = \
3059
self.sha1_provider.stat_and_sha1(
3061
self.state._observed_sha1(entry, link_or_sha1,
3063
content_change = (link_or_sha1 != source_details[1])
3065
# Size changed, so must be different
3066
content_change = True
3067
# Target details is updated at update_entry time
3068
if self.use_filesystem_for_exec:
3069
# We don't need S_ISREG here, because we are sure
3070
# we are dealing with a file.
3071
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
3073
target_exec = target_details[3]
3074
elif target_kind == 'symlink':
3075
if source_minikind != 'l':
3076
content_change = True
3078
content_change = (link_or_sha1 != source_details[1])
3080
elif target_kind == 'tree-reference':
3081
if source_minikind != 't':
3082
content_change = True
3084
content_change = False
3087
raise Exception, "unknown kind %s" % path_info[2]
3088
if source_minikind == 'd':
3090
old_path = path = pathjoin(old_dirname, old_basename)
3091
self.old_dirname_to_file_id[old_path] = file_id
3092
# parent id is the entry for the path in the target tree
3093
if old_dirname == self.last_source_parent[0]:
3094
source_parent_id = self.last_source_parent[1]
3097
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3099
source_parent_entry = self.state._get_entry(self.source_index,
3100
path_utf8=old_dirname)
3101
source_parent_id = source_parent_entry[0][2]
3102
if source_parent_id == entry[0][2]:
3103
# This is the root, so the parent is None
3104
source_parent_id = None
3106
self.last_source_parent[0] = old_dirname
3107
self.last_source_parent[1] = source_parent_id
3108
new_dirname = entry[0][0]
3109
if new_dirname == self.last_target_parent[0]:
3110
target_parent_id = self.last_target_parent[1]
3113
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3115
# TODO: We don't always need to do the lookup, because the
3116
# parent entry will be the same as the source entry.
3117
target_parent_entry = self.state._get_entry(self.target_index,
3118
path_utf8=new_dirname)
3119
if target_parent_entry == (None, None):
3120
raise AssertionError(
3121
"Could not find target parent in wt: %s\nparent of: %s"
3122
% (new_dirname, entry))
3123
target_parent_id = target_parent_entry[0][2]
3124
if target_parent_id == entry[0][2]:
3125
# This is the root, so the parent is None
3126
target_parent_id = None
3128
self.last_target_parent[0] = new_dirname
3129
self.last_target_parent[1] = target_parent_id
3131
source_exec = source_details[3]
3132
if (self.include_unchanged
3134
or source_parent_id != target_parent_id
3135
or old_basename != entry[0][1]
3136
or source_exec != target_exec
3138
if old_path is None:
3139
old_path = path = pathjoin(old_dirname, old_basename)
3140
old_path_u = self.utf8_decode(old_path)[0]
3143
old_path_u = self.utf8_decode(old_path)[0]
3144
if old_path == path:
3147
path_u = self.utf8_decode(path)[0]
3148
source_kind = DirState._minikind_to_kind[source_minikind]
3149
return (entry[0][2],
3150
(old_path_u, path_u),
3153
(source_parent_id, target_parent_id),
3154
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3155
(source_kind, target_kind),
3156
(source_exec, target_exec))
3158
return self.uninteresting
3159
elif source_minikind in 'a' and target_minikind in 'fdlt':
3160
# looks like a new file
3161
path = pathjoin(entry[0][0], entry[0][1])
3162
# parent id is the entry for the path in the target tree
3163
# TODO: these are the same for an entire directory: cache em.
3164
parent_id = self.state._get_entry(self.target_index,
3165
path_utf8=entry[0][0])[0][2]
3166
if parent_id == entry[0][2]:
3168
if path_info is not None:
3170
if self.use_filesystem_for_exec:
3171
# We need S_ISREG here, because we aren't sure if this
3174
stat.S_ISREG(path_info[3].st_mode)
3175
and stat.S_IEXEC & path_info[3].st_mode)
3177
target_exec = target_details[3]
3178
return (entry[0][2],
3179
(None, self.utf8_decode(path)[0]),
3183
(None, self.utf8_decode(entry[0][1])[0]),
3184
(None, path_info[2]),
3185
(None, target_exec))
3187
# Its a missing file, report it as such.
3188
return (entry[0][2],
3189
(None, self.utf8_decode(path)[0]),
3193
(None, self.utf8_decode(entry[0][1])[0]),
3196
elif source_minikind in 'fdlt' and target_minikind in 'a':
3197
# unversioned, possibly, or possibly not deleted: we dont care.
3198
# if its still on disk, *and* theres no other entry at this
3199
# path [we dont know this in this routine at the moment -
3200
# perhaps we should change this - then it would be an unknown.
3201
old_path = pathjoin(entry[0][0], entry[0][1])
3202
# parent id is the entry for the path in the target tree
3203
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3204
if parent_id == entry[0][2]:
3206
return (entry[0][2],
3207
(self.utf8_decode(old_path)[0], None),
3211
(self.utf8_decode(entry[0][1])[0], None),
3212
(DirState._minikind_to_kind[source_minikind], None),
3213
(source_details[3], None))
3214
elif source_minikind in 'fdlt' and target_minikind in 'r':
3215
# a rename; could be a true rename, or a rename inherited from
3216
# a renamed parent. TODO: handle this efficiently. Its not
3217
# common case to rename dirs though, so a correct but slow
3218
# implementation will do.
3219
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3220
self.search_specific_files.add(target_details[1])
3221
elif source_minikind in 'ra' and target_minikind in 'ra':
3222
# neither of the selected trees contain this file,
3223
# so skip over it. This is not currently directly tested, but
3224
# is indirectly via test_too_much.TestCommands.test_conflicts.
3227
raise AssertionError("don't know how to compare "
3228
"source_minikind=%r, target_minikind=%r"
3229
% (source_minikind, target_minikind))
3230
## import pdb;pdb.set_trace()
3236
def iter_changes(self):
3237
"""Iterate over the changes."""
3238
utf8_decode = cache_utf8._utf8_decode
3239
_cmp_by_dirs = cmp_by_dirs
3240
_process_entry = self._process_entry
3241
uninteresting = self.uninteresting
3242
search_specific_files = self.search_specific_files
3243
searched_specific_files = self.searched_specific_files
3244
splitpath = osutils.splitpath
3246
# compare source_index and target_index at or under each element of search_specific_files.
3247
# follow the following comparison table. Note that we only want to do diff operations when
3248
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3252
# Source | Target | disk | action
3253
# r | fdlt | | add source to search, add id path move and perform
3254
# | | | diff check on source-target
3255
# r | fdlt | a | dangling file that was present in the basis.
3257
# r | a | | add source to search
3259
# r | r | | this path is present in a non-examined tree, skip.
3260
# r | r | a | this path is present in a non-examined tree, skip.
3261
# a | fdlt | | add new id
3262
# a | fdlt | a | dangling locally added file, skip
3263
# a | a | | not present in either tree, skip
3264
# a | a | a | not present in any tree, skip
3265
# a | r | | not present in either tree at this path, skip as it
3266
# | | | may not be selected by the users list of paths.
3267
# a | r | a | not present in either tree at this path, skip as it
3268
# | | | may not be selected by the users list of paths.
3269
# fdlt | fdlt | | content in both: diff them
3270
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3271
# fdlt | a | | unversioned: output deleted id for now
3272
# fdlt | a | a | unversioned and deleted: output deleted id
3273
# fdlt | r | | relocated in this tree, so add target to search.
3274
# | | | Dont diff, we will see an r,fd; pair when we reach
3275
# | | | this id at the other path.
3276
# fdlt | r | a | relocated in this tree, so add target to search.
3277
# | | | Dont diff, we will see an r,fd; pair when we reach
3278
# | | | this id at the other path.
3280
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3281
# keeping a cache of directories that we have seen.
3283
while search_specific_files:
3284
# TODO: the pending list should be lexically sorted? the
3285
# interface doesn't require it.
3286
current_root = search_specific_files.pop()
3287
current_root_unicode = current_root.decode('utf8')
3288
searched_specific_files.add(current_root)
3289
# process the entries for this containing directory: the rest will be
3290
# found by their parents recursively.
3291
root_entries = self.state._entries_for_path(current_root)
3292
root_abspath = self.tree.abspath(current_root_unicode)
3294
root_stat = os.lstat(root_abspath)
3296
if e.errno == errno.ENOENT:
3297
# the path does not exist: let _process_entry know that.
3298
root_dir_info = None
3300
# some other random error: hand it up.
3303
root_dir_info = ('', current_root,
3304
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3306
if root_dir_info[2] == 'directory':
3307
if self.tree._directory_is_tree_reference(
3308
current_root.decode('utf8')):
3309
root_dir_info = root_dir_info[:2] + \
3310
('tree-reference',) + root_dir_info[3:]
3312
if not root_entries and not root_dir_info:
3313
# this specified path is not present at all, skip it.
3315
path_handled = False
3316
for entry in root_entries:
3317
result = _process_entry(entry, root_dir_info)
3318
if result is not None:
3320
if result is not uninteresting:
3322
if self.want_unversioned and not path_handled and root_dir_info:
3323
new_executable = bool(
3324
stat.S_ISREG(root_dir_info[3].st_mode)
3325
and stat.S_IEXEC & root_dir_info[3].st_mode)
3327
(None, current_root_unicode),
3331
(None, splitpath(current_root_unicode)[-1]),
3332
(None, root_dir_info[2]),
3333
(None, new_executable)
3335
initial_key = (current_root, '', '')
3336
block_index, _ = self.state._find_block_index_from_key(initial_key)
3337
if block_index == 0:
3338
# we have processed the total root already, but because the
3339
# initial key matched it we should skip it here.
3341
if root_dir_info and root_dir_info[2] == 'tree-reference':
3342
current_dir_info = None
3344
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3346
current_dir_info = dir_iterator.next()
3348
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3349
# python 2.5 has e.errno == EINVAL,
3350
# and e.winerror == ERROR_DIRECTORY
3351
e_winerror = getattr(e, 'winerror', None)
3352
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3353
# there may be directories in the inventory even though
3354
# this path is not a file on disk: so mark it as end of
3356
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3357
current_dir_info = None
3358
elif (sys.platform == 'win32'
3359
and (e.errno in win_errors
3360
or e_winerror in win_errors)):
3361
current_dir_info = None
3365
if current_dir_info[0][0] == '':
3366
# remove .bzr from iteration
3367
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3368
if current_dir_info[1][bzr_index][0] != '.bzr':
3369
raise AssertionError()
3370
del current_dir_info[1][bzr_index]
3371
# walk until both the directory listing and the versioned metadata
3373
if (block_index < len(self.state._dirblocks) and
3374
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3375
current_block = self.state._dirblocks[block_index]
3377
current_block = None
3378
while (current_dir_info is not None or
3379
current_block is not None):
3380
if (current_dir_info and current_block
3381
and current_dir_info[0][0] != current_block[0]):
3382
if _cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
3383
# filesystem data refers to paths not covered by the dirblock.
3384
# this has two possibilities:
3385
# A) it is versioned but empty, so there is no block for it
3386
# B) it is not versioned.
3388
# if (A) then we need to recurse into it to check for
3389
# new unknown files or directories.
3390
# if (B) then we should ignore it, because we don't
3391
# recurse into unknown directories.
3393
while path_index < len(current_dir_info[1]):
3394
current_path_info = current_dir_info[1][path_index]
3395
if self.want_unversioned:
3396
if current_path_info[2] == 'directory':
3397
if self.tree._directory_is_tree_reference(
3398
current_path_info[0].decode('utf8')):
3399
current_path_info = current_path_info[:2] + \
3400
('tree-reference',) + current_path_info[3:]
3401
new_executable = bool(
3402
stat.S_ISREG(current_path_info[3].st_mode)
3403
and stat.S_IEXEC & current_path_info[3].st_mode)
3405
(None, utf8_decode(current_path_info[0])[0]),
3409
(None, utf8_decode(current_path_info[1])[0]),
3410
(None, current_path_info[2]),
3411
(None, new_executable))
3412
# dont descend into this unversioned path if it is
3414
if current_path_info[2] in ('directory',
3416
del current_dir_info[1][path_index]
3420
# This dir info has been handled, go to the next
3422
current_dir_info = dir_iterator.next()
3423
except StopIteration:
3424
current_dir_info = None
3426
# We have a dirblock entry for this location, but there
3427
# is no filesystem path for this. This is most likely
3428
# because a directory was removed from the disk.
3429
# We don't have to report the missing directory,
3430
# because that should have already been handled, but we
3431
# need to handle all of the files that are contained
3433
for current_entry in current_block[1]:
3434
# entry referring to file not present on disk.
3435
# advance the entry only, after processing.
3436
result = _process_entry(current_entry, None)
3437
if result is not None:
3438
if result is not uninteresting:
3441
if (block_index < len(self.state._dirblocks) and
3442
osutils.is_inside(current_root,
3443
self.state._dirblocks[block_index][0])):
3444
current_block = self.state._dirblocks[block_index]
3446
current_block = None
3449
if current_block and entry_index < len(current_block[1]):
3450
current_entry = current_block[1][entry_index]
3452
current_entry = None
3453
advance_entry = True
3455
if current_dir_info and path_index < len(current_dir_info[1]):
3456
current_path_info = current_dir_info[1][path_index]
3457
if current_path_info[2] == 'directory':
3458
if self.tree._directory_is_tree_reference(
3459
current_path_info[0].decode('utf8')):
3460
current_path_info = current_path_info[:2] + \
3461
('tree-reference',) + current_path_info[3:]
3463
current_path_info = None
3465
path_handled = False
3466
while (current_entry is not None or
3467
current_path_info is not None):
3468
if current_entry is None:
3469
# the check for path_handled when the path is advanced
3470
# will yield this path if needed.
3472
elif current_path_info is None:
3473
# no path is fine: the per entry code will handle it.
3474
result = _process_entry(current_entry, current_path_info)
3475
if result is not None:
3476
if result is not uninteresting:
3478
elif (current_entry[0][1] != current_path_info[1]
3479
or current_entry[1][self.target_index][0] in 'ar'):
3480
# The current path on disk doesn't match the dirblock
3481
# record. Either the dirblock is marked as absent, or
3482
# the file on disk is not present at all in the
3483
# dirblock. Either way, report about the dirblock
3484
# entry, and let other code handle the filesystem one.
3486
# Compare the basename for these files to determine
3488
if current_path_info[1] < current_entry[0][1]:
3489
# extra file on disk: pass for now, but only
3490
# increment the path, not the entry
3491
advance_entry = False
3493
# entry referring to file not present on disk.
3494
# advance the entry only, after processing.
3495
result = _process_entry(current_entry, None)
3496
if result is not None:
3497
if result is not uninteresting:
3499
advance_path = False
3501
result = _process_entry(current_entry, current_path_info)
3502
if result is not None:
3504
if result is not uninteresting:
3506
if advance_entry and current_entry is not None:
3508
if entry_index < len(current_block[1]):
3509
current_entry = current_block[1][entry_index]
3511
current_entry = None
3513
advance_entry = True # reset the advance flaga
3514
if advance_path and current_path_info is not None:
3515
if not path_handled:
3516
# unversioned in all regards
3517
if self.want_unversioned:
3518
new_executable = bool(
3519
stat.S_ISREG(current_path_info[3].st_mode)
3520
and stat.S_IEXEC & current_path_info[3].st_mode)
3522
relpath_unicode = utf8_decode(current_path_info[0])[0]
3523
except UnicodeDecodeError:
3524
raise errors.BadFilenameEncoding(
3525
current_path_info[0], osutils._fs_enc)
3527
(None, relpath_unicode),
3531
(None, utf8_decode(current_path_info[1])[0]),
3532
(None, current_path_info[2]),
3533
(None, new_executable))
3534
# dont descend into this unversioned path if it is
3536
if current_path_info[2] in ('directory'):
3537
del current_dir_info[1][path_index]
3539
# dont descend the disk iterator into any tree
3541
if current_path_info[2] == 'tree-reference':
3542
del current_dir_info[1][path_index]
3545
if path_index < len(current_dir_info[1]):
3546
current_path_info = current_dir_info[1][path_index]
3547
if current_path_info[2] == 'directory':
3548
if self.tree._directory_is_tree_reference(
3549
current_path_info[0].decode('utf8')):
3550
current_path_info = current_path_info[:2] + \
3551
('tree-reference',) + current_path_info[3:]
3553
current_path_info = None
3554
path_handled = False
3556
advance_path = True # reset the advance flagg.
3557
if current_block is not None:
3559
if (block_index < len(self.state._dirblocks) and
3560
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3561
current_block = self.state._dirblocks[block_index]
3563
current_block = None
3564
if current_dir_info is not None:
3566
current_dir_info = dir_iterator.next()
3567
except StopIteration:
3568
current_dir_info = None
3569
_process_entry = ProcessEntryPython
3572
# Try to load the compiled form if possible
3574
from bzrlib._dirstate_helpers_c import (
3575
_read_dirblocks_c as _read_dirblocks,
3576
bisect_dirblock_c as bisect_dirblock,
3577
_bisect_path_left_c as _bisect_path_left,
3578
_bisect_path_right_c as _bisect_path_right,
3579
cmp_by_dirs_c as cmp_by_dirs,
3580
ProcessEntryC as _process_entry,
3581
update_entry as update_entry,
3584
from bzrlib._dirstate_helpers_py import (
3585
_read_dirblocks_py as _read_dirblocks,
3586
bisect_dirblock_py as bisect_dirblock,
3587
_bisect_path_left_py as _bisect_path_left,
3588
_bisect_path_right_py as _bisect_path_right,
3589
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
