1
# Copyright (C) 2006, 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but it's a file. The fingerprint is the
86
sha1 value of the file's canonical form, i.e. after any read filters have
87
been applied to the convenience form stored in the working tree.
88
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
90
't' is a reference to a nested subtree; the fingerprint is the referenced
95
The entries on disk and in memory are ordered according to the following keys:
97
directory, as a list of components
101
--- Format 1 had the following different definition: ---
102
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
103
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
105
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
106
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
109
PARENT ROW's are emitted for every parent that is not in the ghosts details
110
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
111
each row will have a PARENT ROW for foo and baz, but not for bar.
114
In any tree, a kind of 'moved' indicates that the fingerprint field
115
(which we treat as opaque data specific to the 'kind' anyway) has the
116
details for the id of this row in that tree.
118
I'm strongly tempted to add a id->path index as well, but I think that
119
where we need id->path mapping; we also usually read the whole file, so
120
I'm going to skip that for the moment, as we have the ability to locate
121
via bisect any path in any tree, and if we lookup things by path, we can
122
accumulate an id->path mapping as we go, which will tend to match what we
125
I plan to implement this asap, so please speak up now to alter/tweak the
126
design - and once we stabilise on this, I'll update the wiki page for
129
The rationale for all this is that we want fast operations for the
130
common case (diff/status/commit/merge on all files) and extremely fast
131
operations for the less common but still occurs a lot status/diff/commit
132
on specific files). Operations on specific files involve a scan for all
133
the children of a path, *in every involved tree*, which the current
134
format did not accommodate.
138
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
139
2) fall back current object model as needed.
140
3) scale usably to the largest trees known today - say 50K entries. (mozilla
141
is an example of this)
145
Eventually reuse dirstate objects across locks IFF the dirstate file has not
146
been modified, but will require that we flush/ignore cached stat-hit data
147
because we won't want to restat all files on disk just because a lock was
148
acquired, yet we cannot trust the data after the previous lock was released.
150
Memory representation:
151
vector of all directories, and vector of the childen ?
153
root_entrie = (direntry for root, [parent_direntries_for_root]),
155
('', ['data for achild', 'data for bchild', 'data for cchild'])
156
('dir', ['achild', 'cchild', 'echild'])
158
- single bisect to find N subtrees from a path spec
159
- in-order for serialisation - this is 'dirblock' grouping.
160
- insertion of a file '/a' affects only the '/' child-vector, that is, to
161
insert 10K elements from scratch does not generates O(N^2) memoves of a
162
single vector, rather each individual, which tends to be limited to a
163
manageable number. Will scale badly on trees with 10K entries in a
164
single directory. compare with Inventory.InventoryDirectory which has
165
a dictionary for the children. No bisect capability, can only probe for
166
exact matches, or grab all elements and sort.
167
- What's the risk of error here? Once we have the base format being processed
168
we should have a net win regardless of optimality. So we are going to
169
go with what seems reasonable.
172
Maybe we should do a test profile of the core structure - 10K simulated
173
searches/lookups/etc?
175
Objects for each row?
176
The lifetime of Dirstate objects is current per lock, but see above for
177
possible extensions. The lifetime of a row from a dirstate is expected to be
178
very short in the optimistic case: which we are optimising for. For instance,
179
subtree status will determine from analysis of the disk data what rows need to
180
be examined at all, and will be able to determine from a single row whether
181
that file has altered or not, so we are aiming to process tens of thousands of
182
entries each second within the dirstate context, before exposing anything to
183
the larger codebase. This suggests we want the time for a single file
184
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
185
processed, and to scale to 100 thousand we'll another order of magnitude to do
186
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
187
the file on disk, and then immediately discard, the overhead of object creation
188
becomes a significant cost.
190
Figures: Creating a tuple from 3 elements was profiled at 0.0625
191
microseconds, whereas creating a object which is subclassed from tuple was
192
0.500 microseconds, and creating an object with 3 elements and slots was 3
193
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
194
down to 10 microseconds for the total processing - having 33% of that be object
195
creation is a huge overhead. There is a potential cost in using tuples within
196
each row which is that the conditional code to do comparisons may be slower
197
than method invocation, but method invocation is known to be slow due to stack
198
frame creation, so avoiding methods in these tight inner loops in unfortunately
199
desirable. We can consider a pyrex version of this with objects in future if
208
from stat import S_IEXEC
226
# This is the Windows equivalent of ENOTDIR
227
# It is defined in pywin32.winerror, but we don't want a strong dependency for
228
# just an error code.
229
ERROR_PATH_NOT_FOUND = 3
230
ERROR_DIRECTORY = 267
233
if not getattr(struct, '_compile', None):
234
# Cannot pre-compile the dirstate pack_stat
235
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
236
"""Convert stat values into a packed representation."""
237
return _encode(_pack('>LLLLLL', st.st_size, int(st.st_mtime),
238
int(st.st_ctime), st.st_dev, st.st_ino & 0xFFFFFFFF,
241
# compile the struct compiler we need, so as to only do it once
242
from _struct import Struct
243
_compiled_pack = Struct('>LLLLLL').pack
244
def pack_stat(st, _encode=binascii.b2a_base64, _pack=_compiled_pack):
245
"""Convert stat values into a packed representation."""
246
# jam 20060614 it isn't really worth removing more entries if we
247
# are going to leave it in packed form.
248
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
249
# With all entries, filesize is 5.9M and read time is maybe 280ms
250
# well within the noise margin
252
# base64 encoding always adds a final newline, so strip it off
253
# The current version
254
return _encode(_pack(st.st_size, int(st.st_mtime), int(st.st_ctime),
255
st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
256
# This is 0.060s / 1.520s faster by not encoding as much information
257
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
258
# This is not strictly faster than _encode(_pack())[:-1]
259
# return '%X.%X.%X.%X.%X.%X' % (
260
# st.st_size, int(st.st_mtime), int(st.st_ctime),
261
# st.st_dev, st.st_ino, st.st_mode)
262
# Similar to the _encode(_pack('>LL'))
263
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
266
class SHA1Provider(object):
267
"""An interface for getting sha1s of a file."""
269
def sha1(self, abspath):
270
"""Return the sha1 of a file given its absolute path."""
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.
276
Note: the stat should be the stat of the physical file
277
while the sha may be the sha of its canonical content.
279
raise NotImplementedError(self.stat_and_sha1)
282
class DefaultSHA1Provider(SHA1Provider):
283
"""A SHA1Provider that reads directly from the filesystem."""
285
def sha1(self, abspath):
286
"""Return the sha1 of a file given its absolute path."""
287
return osutils.sha_file_by_name(abspath)
289
def stat_and_sha1(self, abspath):
290
"""Return the stat and sha1 of a file given its absolute path."""
291
file_obj = file(abspath, 'rb')
293
statvalue = os.fstat(file_obj.fileno())
294
sha1 = osutils.sha_file(file_obj)
297
return statvalue, sha1
300
class DirState(object):
301
"""Record directory and metadata state for fast access.
303
A dirstate is a specialised data structure for managing local working
304
tree state information. Its not yet well defined whether it is platform
305
specific, and if it is how we detect/parameterize that.
307
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
308
Unlike most bzr disk formats, DirStates must be locked for reading, using
309
lock_read. (This is an os file lock internally.) This is necessary
310
because the file can be rewritten in place.
312
DirStates must be explicitly written with save() to commit changes; just
313
unlocking them does not write the changes to disk.
316
_kind_to_minikind = {
322
'tree-reference': 't',
324
_minikind_to_kind = {
330
't': 'tree-reference',
332
_stat_to_minikind = {
337
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
338
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
341
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
342
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
343
# that we are sure a single record will always fit.
344
BISECT_PAGE_SIZE = 4096
347
IN_MEMORY_UNMODIFIED = 1
348
IN_MEMORY_MODIFIED = 2
350
# A pack_stat (the x's) that is just noise and will never match the output
353
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
355
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
356
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
358
def __init__(self, path, sha1_provider):
359
"""Create a DirState object.
361
:param path: The path at which the dirstate file on disk should live.
362
:param sha1_provider: an object meeting the SHA1Provider interface.
364
# _header_state and _dirblock_state represent the current state
365
# of the dirstate metadata and the per-row data respectiely.
366
# NOT_IN_MEMORY indicates that no data is in memory
367
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
368
# is the same as is on disk
369
# IN_MEMORY_MODIFIED indicates that we have a modified version
370
# of what is on disk.
371
# In future we will add more granularity, for instance _dirblock_state
372
# will probably support partially-in-memory as a separate variable,
373
# allowing for partially-in-memory unmodified and partially-in-memory
375
self._header_state = DirState.NOT_IN_MEMORY
376
self._dirblock_state = DirState.NOT_IN_MEMORY
377
# If true, an error has been detected while updating the dirstate, and
378
# for safety we're not going to commit to disk.
379
self._changes_aborted = False
383
self._state_file = None
384
self._filename = path
385
self._lock_token = None
386
self._lock_state = None
387
self._id_index = None
388
# a map from packed_stat to sha's.
389
self._packed_stat_index = None
390
self._end_of_header = None
391
self._cutoff_time = None
392
self._split_path_cache = {}
393
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
394
self._sha1_provider = sha1_provider
395
if 'hashcache' in debug.debug_flags:
396
self._sha1_file = self._sha1_file_and_mutter
398
self._sha1_file = self._sha1_provider.sha1
399
# These two attributes provide a simple cache for lookups into the
400
# dirstate in-memory vectors. By probing respectively for the last
401
# block, and for the next entry, we save nearly 2 bisections per path
403
self._last_block_index = None
404
self._last_entry_index = None
408
(self.__class__.__name__, self._filename)
410
def add(self, path, file_id, kind, stat, fingerprint):
411
"""Add a path to be tracked.
413
:param path: The path within the dirstate - '' is the root, 'foo' is the
414
path foo within the root, 'foo/bar' is the path bar within foo
416
:param file_id: The file id of the path being added.
417
:param kind: The kind of the path, as a string like 'file',
419
:param stat: The output of os.lstat for the path.
420
:param fingerprint: The sha value of the file's canonical form (i.e.
421
after any read filters have been applied),
422
or the target of a symlink,
423
or the referenced revision id for tree-references,
424
or '' for directories.
427
# find the block its in.
428
# find the location in the block.
429
# check its not there
431
#------- copied from inventory.ensure_normalized_name - keep synced.
432
# --- normalized_filename wants a unicode basename only, so get one.
433
dirname, basename = osutils.split(path)
434
# we dont import normalized_filename directly because we want to be
435
# able to change the implementation at runtime for tests.
436
norm_name, can_access = osutils.normalized_filename(basename)
437
if norm_name != basename:
441
raise errors.InvalidNormalization(path)
442
# you should never have files called . or ..; just add the directory
443
# in the parent, or according to the special treatment for the root
444
if basename == '.' or basename == '..':
445
raise errors.InvalidEntryName(path)
446
# now that we've normalised, we need the correct utf8 path and
447
# dirname and basename elements. This single encode and split should be
448
# faster than three separate encodes.
449
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
450
dirname, basename = osutils.split(utf8path)
451
# uses __class__ for speed; the check is needed for safety
452
if file_id.__class__ is not str:
453
raise AssertionError(
454
"must be a utf8 file_id not %s" % (type(file_id), ))
455
# Make sure the file_id does not exist in this tree
457
file_id_entry = self._get_entry(0, fileid_utf8=file_id, include_deleted=True)
458
if file_id_entry != (None, None):
459
if file_id_entry[1][0][0] == 'a':
460
if file_id_entry[0] != (dirname, basename, file_id):
461
# set the old name's current operation to rename
462
self.update_minimal(file_id_entry[0],
468
rename_from = file_id_entry[0][0:2]
470
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
471
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
472
info = '%s:%s' % (kind, path)
473
raise errors.DuplicateFileId(file_id, info)
474
first_key = (dirname, basename, '')
475
block_index, present = self._find_block_index_from_key(first_key)
477
# check the path is not in the tree
478
block = self._dirblocks[block_index][1]
479
entry_index, _ = self._find_entry_index(first_key, block)
480
while (entry_index < len(block) and
481
block[entry_index][0][0:2] == first_key[0:2]):
482
if block[entry_index][1][0][0] not in 'ar':
483
# this path is in the dirstate in the current tree.
484
raise Exception, "adding already added path!"
487
# The block where we want to put the file is not present. But it
488
# might be because the directory was empty, or not loaded yet. Look
489
# for a parent entry, if not found, raise NotVersionedError
490
parent_dir, parent_base = osutils.split(dirname)
491
parent_block_idx, parent_entry_idx, _, parent_present = \
492
self._get_block_entry_index(parent_dir, parent_base, 0)
493
if not parent_present:
494
raise errors.NotVersionedError(path, str(self))
495
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
496
block = self._dirblocks[block_index][1]
497
entry_key = (dirname, basename, file_id)
500
packed_stat = DirState.NULLSTAT
503
packed_stat = pack_stat(stat)
504
parent_info = self._empty_parent_info()
505
minikind = DirState._kind_to_minikind[kind]
506
if rename_from is not None:
508
old_path_utf8 = '%s/%s' % rename_from
510
old_path_utf8 = rename_from[1]
511
parent_info[0] = ('r', old_path_utf8, 0, False, '')
513
entry_data = entry_key, [
514
(minikind, fingerprint, size, False, packed_stat),
516
elif kind == 'directory':
517
entry_data = entry_key, [
518
(minikind, '', 0, False, packed_stat),
520
elif kind == 'symlink':
521
entry_data = entry_key, [
522
(minikind, fingerprint, size, False, packed_stat),
524
elif kind == 'tree-reference':
525
entry_data = entry_key, [
526
(minikind, fingerprint, 0, False, packed_stat),
529
raise errors.BzrError('unknown kind %r' % kind)
530
entry_index, present = self._find_entry_index(entry_key, block)
532
block.insert(entry_index, entry_data)
534
if block[entry_index][1][0][0] != 'a':
535
raise AssertionError(" %r(%r) already added" % (basename, file_id))
536
block[entry_index][1][0] = entry_data[1][0]
538
if kind == 'directory':
539
# insert a new dirblock
540
self._ensure_block(block_index, entry_index, utf8path)
541
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
543
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
545
def _bisect(self, paths):
546
"""Bisect through the disk structure for specific rows.
548
:param paths: A list of paths to find
549
:return: A dict mapping path => entries for found entries. Missing
550
entries will not be in the map.
551
The list is not sorted, and entries will be populated
552
based on when they were read.
554
self._requires_lock()
555
# We need the file pointer to be right after the initial header block
556
self._read_header_if_needed()
557
# If _dirblock_state was in memory, we should just return info from
558
# there, this function is only meant to handle when we want to read
560
if self._dirblock_state != DirState.NOT_IN_MEMORY:
561
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
563
# The disk representation is generally info + '\0\n\0' at the end. But
564
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
565
# Because it means we can sync on the '\n'
566
state_file = self._state_file
567
file_size = os.fstat(state_file.fileno()).st_size
568
# We end up with 2 extra fields, we should have a trailing '\n' to
569
# ensure that we read the whole record, and we should have a precursur
570
# '' which ensures that we start after the previous '\n'
571
entry_field_count = self._fields_per_entry() + 1
573
low = self._end_of_header
574
high = file_size - 1 # Ignore the final '\0'
575
# Map from (dir, name) => entry
578
# Avoid infinite seeking
579
max_count = 30*len(paths)
581
# pending is a list of places to look.
582
# each entry is a tuple of low, high, dir_names
583
# low -> the first byte offset to read (inclusive)
584
# high -> the last byte offset (inclusive)
585
# dir_names -> The list of (dir, name) pairs that should be found in
586
# the [low, high] range
587
pending = [(low, high, paths)]
589
page_size = self._bisect_page_size
591
fields_to_entry = self._get_fields_to_entry()
594
low, high, cur_files = pending.pop()
596
if not cur_files or low >= high:
601
if count > max_count:
602
raise errors.BzrError('Too many seeks, most likely a bug.')
604
mid = max(low, (low+high-page_size)/2)
607
# limit the read size, so we don't end up reading data that we have
609
read_size = min(page_size, (high-mid)+1)
610
block = state_file.read(read_size)
613
entries = block.split('\n')
616
# We didn't find a '\n', so we cannot have found any records.
617
# So put this range back and try again. But we know we have to
618
# increase the page size, because a single read did not contain
619
# a record break (so records must be larger than page_size)
621
pending.append((low, high, cur_files))
624
# Check the first and last entries, in case they are partial, or if
625
# we don't care about the rest of this page
627
first_fields = entries[0].split('\0')
628
if len(first_fields) < entry_field_count:
629
# We didn't get the complete first entry
630
# so move start, and grab the next, which
631
# should be a full entry
632
start += len(entries[0])+1
633
first_fields = entries[1].split('\0')
636
if len(first_fields) <= 2:
637
# We didn't even get a filename here... what do we do?
638
# Try a large page size and repeat this query
640
pending.append((low, high, cur_files))
643
# Find what entries we are looking for, which occur before and
644
# after this first record.
647
first_path = first_fields[1] + '/' + first_fields[2]
649
first_path = first_fields[2]
650
first_loc = _bisect_path_left(cur_files, first_path)
652
# These exist before the current location
653
pre = cur_files[:first_loc]
654
# These occur after the current location, which may be in the
655
# data we read, or might be after the last entry
656
post = cur_files[first_loc:]
658
if post and len(first_fields) >= entry_field_count:
659
# We have files after the first entry
661
# Parse the last entry
662
last_entry_num = len(entries)-1
663
last_fields = entries[last_entry_num].split('\0')
664
if len(last_fields) < entry_field_count:
665
# The very last hunk was not complete,
666
# read the previous hunk
667
after = mid + len(block) - len(entries[-1])
669
last_fields = entries[last_entry_num].split('\0')
671
after = mid + len(block)
674
last_path = last_fields[1] + '/' + last_fields[2]
676
last_path = last_fields[2]
677
last_loc = _bisect_path_right(post, last_path)
679
middle_files = post[:last_loc]
680
post = post[last_loc:]
683
# We have files that should occur in this block
684
# (>= first, <= last)
685
# Either we will find them here, or we can mark them as
688
if middle_files[0] == first_path:
689
# We might need to go before this location
690
pre.append(first_path)
691
if middle_files[-1] == last_path:
692
post.insert(0, last_path)
694
# Find out what paths we have
695
paths = {first_path:[first_fields]}
696
# last_path might == first_path so we need to be
697
# careful if we should append rather than overwrite
698
if last_entry_num != first_entry_num:
699
paths.setdefault(last_path, []).append(last_fields)
700
for num in xrange(first_entry_num+1, last_entry_num):
701
# TODO: jam 20070223 We are already splitting here, so
702
# shouldn't we just split the whole thing rather
703
# than doing the split again in add_one_record?
704
fields = entries[num].split('\0')
706
path = fields[1] + '/' + fields[2]
709
paths.setdefault(path, []).append(fields)
711
for path in middle_files:
712
for fields in paths.get(path, []):
713
# offset by 1 because of the opening '\0'
714
# consider changing fields_to_entry to avoid the
716
entry = fields_to_entry(fields[1:])
717
found.setdefault(path, []).append(entry)
719
# Now we have split up everything into pre, middle, and post, and
720
# we have handled everything that fell in 'middle'.
721
# We add 'post' first, so that we prefer to seek towards the
722
# beginning, so that we will tend to go as early as we need, and
723
# then only seek forward after that.
725
pending.append((after, high, post))
727
pending.append((low, start-1, pre))
729
# Consider that we may want to return the directory entries in sorted
730
# order. For now, we just return them in whatever order we found them,
731
# and leave it up to the caller if they care if it is ordered or not.
734
def _bisect_dirblocks(self, dir_list):
735
"""Bisect through the disk structure to find entries in given dirs.
737
_bisect_dirblocks is meant to find the contents of directories, which
738
differs from _bisect, which only finds individual entries.
740
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
741
:return: A map from dir => entries_for_dir
743
# TODO: jam 20070223 A lot of the bisecting logic could be shared
744
# between this and _bisect. It would require parameterizing the
745
# inner loop with a function, though. We should evaluate the
746
# performance difference.
747
self._requires_lock()
748
# We need the file pointer to be right after the initial header block
749
self._read_header_if_needed()
750
# If _dirblock_state was in memory, we should just return info from
751
# there, this function is only meant to handle when we want to read
753
if self._dirblock_state != DirState.NOT_IN_MEMORY:
754
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
755
# The disk representation is generally info + '\0\n\0' at the end. But
756
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
757
# Because it means we can sync on the '\n'
758
state_file = self._state_file
759
file_size = os.fstat(state_file.fileno()).st_size
760
# We end up with 2 extra fields, we should have a trailing '\n' to
761
# ensure that we read the whole record, and we should have a precursur
762
# '' which ensures that we start after the previous '\n'
763
entry_field_count = self._fields_per_entry() + 1
765
low = self._end_of_header
766
high = file_size - 1 # Ignore the final '\0'
767
# Map from dir => entry
770
# Avoid infinite seeking
771
max_count = 30*len(dir_list)
773
# pending is a list of places to look.
774
# each entry is a tuple of low, high, dir_names
775
# low -> the first byte offset to read (inclusive)
776
# high -> the last byte offset (inclusive)
777
# dirs -> The list of directories that should be found in
778
# the [low, high] range
779
pending = [(low, high, dir_list)]
781
page_size = self._bisect_page_size
783
fields_to_entry = self._get_fields_to_entry()
786
low, high, cur_dirs = pending.pop()
788
if not cur_dirs or low >= high:
793
if count > max_count:
794
raise errors.BzrError('Too many seeks, most likely a bug.')
796
mid = max(low, (low+high-page_size)/2)
799
# limit the read size, so we don't end up reading data that we have
801
read_size = min(page_size, (high-mid)+1)
802
block = state_file.read(read_size)
805
entries = block.split('\n')
808
# We didn't find a '\n', so we cannot have found any records.
809
# So put this range back and try again. But we know we have to
810
# increase the page size, because a single read did not contain
811
# a record break (so records must be larger than page_size)
813
pending.append((low, high, cur_dirs))
816
# Check the first and last entries, in case they are partial, or if
817
# we don't care about the rest of this page
819
first_fields = entries[0].split('\0')
820
if len(first_fields) < entry_field_count:
821
# We didn't get the complete first entry
822
# so move start, and grab the next, which
823
# should be a full entry
824
start += len(entries[0])+1
825
first_fields = entries[1].split('\0')
828
if len(first_fields) <= 1:
829
# We didn't even get a dirname here... what do we do?
830
# Try a large page size and repeat this query
832
pending.append((low, high, cur_dirs))
835
# Find what entries we are looking for, which occur before and
836
# after this first record.
838
first_dir = first_fields[1]
839
first_loc = bisect.bisect_left(cur_dirs, first_dir)
841
# These exist before the current location
842
pre = cur_dirs[:first_loc]
843
# These occur after the current location, which may be in the
844
# data we read, or might be after the last entry
845
post = cur_dirs[first_loc:]
847
if post and len(first_fields) >= entry_field_count:
848
# We have records to look at after the first entry
850
# Parse the last entry
851
last_entry_num = len(entries)-1
852
last_fields = entries[last_entry_num].split('\0')
853
if len(last_fields) < entry_field_count:
854
# The very last hunk was not complete,
855
# read the previous hunk
856
after = mid + len(block) - len(entries[-1])
858
last_fields = entries[last_entry_num].split('\0')
860
after = mid + len(block)
862
last_dir = last_fields[1]
863
last_loc = bisect.bisect_right(post, last_dir)
865
middle_files = post[:last_loc]
866
post = post[last_loc:]
869
# We have files that should occur in this block
870
# (>= first, <= last)
871
# Either we will find them here, or we can mark them as
874
if middle_files[0] == first_dir:
875
# We might need to go before this location
876
pre.append(first_dir)
877
if middle_files[-1] == last_dir:
878
post.insert(0, last_dir)
880
# Find out what paths we have
881
paths = {first_dir:[first_fields]}
882
# last_dir might == first_dir so we need to be
883
# careful if we should append rather than overwrite
884
if last_entry_num != first_entry_num:
885
paths.setdefault(last_dir, []).append(last_fields)
886
for num in xrange(first_entry_num+1, last_entry_num):
887
# TODO: jam 20070223 We are already splitting here, so
888
# shouldn't we just split the whole thing rather
889
# than doing the split again in add_one_record?
890
fields = entries[num].split('\0')
891
paths.setdefault(fields[1], []).append(fields)
893
for cur_dir in middle_files:
894
for fields in paths.get(cur_dir, []):
895
# offset by 1 because of the opening '\0'
896
# consider changing fields_to_entry to avoid the
898
entry = fields_to_entry(fields[1:])
899
found.setdefault(cur_dir, []).append(entry)
901
# Now we have split up everything into pre, middle, and post, and
902
# we have handled everything that fell in 'middle'.
903
# We add 'post' first, so that we prefer to seek towards the
904
# beginning, so that we will tend to go as early as we need, and
905
# then only seek forward after that.
907
pending.append((after, high, post))
909
pending.append((low, start-1, pre))
913
def _bisect_recursive(self, paths):
914
"""Bisect for entries for all paths and their children.
916
This will use bisect to find all records for the supplied paths. It
917
will then continue to bisect for any records which are marked as
918
directories. (and renames?)
920
:param paths: A sorted list of (dir, name) pairs
921
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
922
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
924
# Map from (dir, name, file_id) => [tree_info]
927
found_dir_names = set()
929
# Directories that have been read
930
processed_dirs = set()
931
# Get the ball rolling with the first bisect for all entries.
932
newly_found = self._bisect(paths)
935
# Directories that need to be read
937
paths_to_search = set()
938
for entry_list in newly_found.itervalues():
939
for dir_name_id, trees_info in entry_list:
940
found[dir_name_id] = trees_info
941
found_dir_names.add(dir_name_id[:2])
943
for tree_info in trees_info:
944
minikind = tree_info[0]
947
# We already processed this one as a directory,
948
# we don't need to do the extra work again.
950
subdir, name, file_id = dir_name_id
951
path = osutils.pathjoin(subdir, name)
953
if path not in processed_dirs:
954
pending_dirs.add(path)
955
elif minikind == 'r':
956
# Rename, we need to directly search the target
957
# which is contained in the fingerprint column
958
dir_name = osutils.split(tree_info[1])
959
if dir_name[0] in pending_dirs:
960
# This entry will be found in the dir search
962
if dir_name not in found_dir_names:
963
paths_to_search.add(tree_info[1])
964
# Now we have a list of paths to look for directly, and
965
# directory blocks that need to be read.
966
# newly_found is mixing the keys between (dir, name) and path
967
# entries, but that is okay, because we only really care about the
969
newly_found = self._bisect(sorted(paths_to_search))
970
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
971
processed_dirs.update(pending_dirs)
974
def _discard_merge_parents(self):
975
"""Discard any parents trees beyond the first.
977
Note that if this fails the dirstate is corrupted.
979
After this function returns the dirstate contains 2 trees, neither of
982
self._read_header_if_needed()
983
parents = self.get_parent_ids()
986
# only require all dirblocks if we are doing a full-pass removal.
987
self._read_dirblocks_if_needed()
988
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
989
def iter_entries_removable():
990
for block in self._dirblocks:
991
deleted_positions = []
992
for pos, entry in enumerate(block[1]):
994
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
995
deleted_positions.append(pos)
996
if deleted_positions:
997
if len(deleted_positions) == len(block[1]):
1000
for pos in reversed(deleted_positions):
1002
# if the first parent is a ghost:
1003
if parents[0] in self.get_ghosts():
1004
empty_parent = [DirState.NULL_PARENT_DETAILS]
1005
for entry in iter_entries_removable():
1006
entry[1][1:] = empty_parent
1008
for entry in iter_entries_removable():
1012
self._parents = [parents[0]]
1013
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1014
self._header_state = DirState.IN_MEMORY_MODIFIED
1016
def _empty_parent_info(self):
1017
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
1020
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
1021
"""Ensure a block for dirname exists.
1023
This function exists to let callers which know that there is a
1024
directory dirname ensure that the block for it exists. This block can
1025
fail to exist because of demand loading, or because a directory had no
1026
children. In either case it is not an error. It is however an error to
1027
call this if there is no parent entry for the directory, and thus the
1028
function requires the coordinates of such an entry to be provided.
1030
The root row is special cased and can be indicated with a parent block
1033
:param parent_block_index: The index of the block in which dirname's row
1035
:param parent_row_index: The index in the parent block where the row
1037
:param dirname: The utf8 dirname to ensure there is a block for.
1038
:return: The index for the block.
1040
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
1041
# This is the signature of the root row, and the
1042
# contents-of-root row is always index 1
1044
# the basename of the directory must be the end of its full name.
1045
if not (parent_block_index == -1 and
1046
parent_block_index == -1 and dirname == ''):
1047
if not dirname.endswith(
1048
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
1049
raise AssertionError("bad dirname %r" % dirname)
1050
block_index, present = self._find_block_index_from_key((dirname, '', ''))
1052
## In future, when doing partial parsing, this should load and
1053
# populate the entire block.
1054
self._dirblocks.insert(block_index, (dirname, []))
1057
def _entries_to_current_state(self, new_entries):
1058
"""Load new_entries into self.dirblocks.
1060
Process new_entries into the current state object, making them the active
1061
state. The entries are grouped together by directory to form dirblocks.
1063
:param new_entries: A sorted list of entries. This function does not sort
1064
to prevent unneeded overhead when callers have a sorted list already.
1067
if new_entries[0][0][0:2] != ('', ''):
1068
raise AssertionError(
1069
"Missing root row %r" % (new_entries[0][0],))
1070
# The two blocks here are deliberate: the root block and the
1071
# contents-of-root block.
1072
self._dirblocks = [('', []), ('', [])]
1073
current_block = self._dirblocks[0][1]
1074
current_dirname = ''
1076
append_entry = current_block.append
1077
for entry in new_entries:
1078
if entry[0][0] != current_dirname:
1079
# new block - different dirname
1081
current_dirname = entry[0][0]
1082
self._dirblocks.append((current_dirname, current_block))
1083
append_entry = current_block.append
1084
# append the entry to the current block
1086
self._split_root_dirblock_into_contents()
1088
def _split_root_dirblock_into_contents(self):
1089
"""Split the root dirblocks into root and contents-of-root.
1091
After parsing by path, we end up with root entries and contents-of-root
1092
entries in the same block. This loop splits them out again.
1094
# The above loop leaves the "root block" entries mixed with the
1095
# "contents-of-root block". But we don't want an if check on
1096
# all entries, so instead we just fix it up here.
1097
if self._dirblocks[1] != ('', []):
1098
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1100
contents_of_root_block = []
1101
for entry in self._dirblocks[0][1]:
1102
if not entry[0][1]: # This is a root entry
1103
root_block.append(entry)
1105
contents_of_root_block.append(entry)
1106
self._dirblocks[0] = ('', root_block)
1107
self._dirblocks[1] = ('', contents_of_root_block)
1109
def _entries_for_path(self, path):
1110
"""Return a list with all the entries that match path for all ids."""
1111
dirname, basename = os.path.split(path)
1112
key = (dirname, basename, '')
1113
block_index, present = self._find_block_index_from_key(key)
1115
# the block which should contain path is absent.
1118
block = self._dirblocks[block_index][1]
1119
entry_index, _ = self._find_entry_index(key, block)
1120
# we may need to look at multiple entries at this path: walk while the specific_files match.
1121
while (entry_index < len(block) and
1122
block[entry_index][0][0:2] == key[0:2]):
1123
result.append(block[entry_index])
1127
def _entry_to_line(self, entry):
1128
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1130
:param entry: An entry_tuple as defined in the module docstring.
1132
entire_entry = list(entry[0])
1133
for tree_number, tree_data in enumerate(entry[1]):
1134
# (minikind, fingerprint, size, executable, tree_specific_string)
1135
entire_entry.extend(tree_data)
1136
# 3 for the key, 5 for the fields per tree.
1137
tree_offset = 3 + tree_number * 5
1139
entire_entry[tree_offset + 0] = tree_data[0]
1141
entire_entry[tree_offset + 2] = str(tree_data[2])
1143
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1144
return '\0'.join(entire_entry)
1146
def _fields_per_entry(self):
1147
"""How many null separated fields should be in each entry row.
1149
Each line now has an extra '\n' field which is not used
1150
so we just skip over it
1152
3 fields for the key
1153
+ number of fields per tree_data (5) * tree count
1156
tree_count = 1 + self._num_present_parents()
1157
return 3 + 5 * tree_count + 1
1159
def _find_block(self, key, add_if_missing=False):
1160
"""Return the block that key should be present in.
1162
:param key: A dirstate entry key.
1163
:return: The block tuple.
1165
block_index, present = self._find_block_index_from_key(key)
1167
if not add_if_missing:
1168
# check to see if key is versioned itself - we might want to
1169
# add it anyway, because dirs with no entries dont get a
1170
# dirblock at parse time.
1171
# This is an uncommon branch to take: most dirs have children,
1172
# and most code works with versioned paths.
1173
parent_base, parent_name = osutils.split(key[0])
1174
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1175
# some parent path has not been added - its an error to add
1177
raise errors.NotVersionedError(key[0:2], str(self))
1178
self._dirblocks.insert(block_index, (key[0], []))
1179
return self._dirblocks[block_index]
1181
def _find_block_index_from_key(self, key):
1182
"""Find the dirblock index for a key.
1184
:return: The block index, True if the block for the key is present.
1186
if key[0:2] == ('', ''):
1189
if (self._last_block_index is not None and
1190
self._dirblocks[self._last_block_index][0] == key[0]):
1191
return self._last_block_index, True
1194
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1195
cache=self._split_path_cache)
1196
# _right returns one-past-where-key is so we have to subtract
1197
# one to use it. we use _right here because there are two
1198
# '' blocks - the root, and the contents of root
1199
# we always have a minimum of 2 in self._dirblocks: root and
1200
# root-contents, and for '', we get 2 back, so this is
1201
# simple and correct:
1202
present = (block_index < len(self._dirblocks) and
1203
self._dirblocks[block_index][0] == key[0])
1204
self._last_block_index = block_index
1205
# Reset the entry index cache to the beginning of the block.
1206
self._last_entry_index = -1
1207
return block_index, present
1209
def _find_entry_index(self, key, block):
1210
"""Find the entry index for a key in a block.
1212
:return: The entry index, True if the entry for the key is present.
1214
len_block = len(block)
1216
if self._last_entry_index is not None:
1218
entry_index = self._last_entry_index + 1
1219
# A hit is when the key is after the last slot, and before or
1220
# equal to the next slot.
1221
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1222
key <= block[entry_index][0]):
1223
self._last_entry_index = entry_index
1224
present = (block[entry_index][0] == key)
1225
return entry_index, present
1228
entry_index = bisect.bisect_left(block, (key, []))
1229
present = (entry_index < len_block and
1230
block[entry_index][0] == key)
1231
self._last_entry_index = entry_index
1232
return entry_index, present
1235
def from_tree(tree, dir_state_filename, sha1_provider=None):
1236
"""Create a dirstate from a bzr Tree.
1238
:param tree: The tree which should provide parent information and
1240
:param sha1_provider: an object meeting the SHA1Provider interface.
1241
If None, a DefaultSHA1Provider is used.
1242
:return: a DirState object which is currently locked for writing.
1243
(it was locked by DirState.initialize)
1245
result = DirState.initialize(dir_state_filename,
1246
sha1_provider=sha1_provider)
1250
parent_ids = tree.get_parent_ids()
1251
num_parents = len(parent_ids)
1253
for parent_id in parent_ids:
1254
parent_tree = tree.branch.repository.revision_tree(parent_id)
1255
parent_trees.append((parent_id, parent_tree))
1256
parent_tree.lock_read()
1257
result.set_parent_trees(parent_trees, [])
1258
result.set_state_from_inventory(tree.inventory)
1260
for revid, parent_tree in parent_trees:
1261
parent_tree.unlock()
1264
# The caller won't have a chance to unlock this, so make sure we
1270
def update_by_delta(self, delta):
1271
"""Apply an inventory delta to the dirstate for tree 0
1273
:param delta: An inventory delta. See Inventory.apply_delta for
1276
self._read_dirblocks_if_needed()
1279
for old_path, new_path, file_id, inv_entry in sorted(delta, reverse=True):
1280
if (file_id in insertions) or (file_id in removals):
1281
raise AssertionError("repeated file id in delta %r" % (file_id,))
1282
if old_path is not None:
1283
old_path = old_path.encode('utf-8')
1284
removals[file_id] = old_path
1285
if new_path is not None:
1286
new_path = new_path.encode('utf-8')
1287
dirname, basename = osutils.split(new_path)
1288
key = (dirname, basename, file_id)
1289
minikind = DirState._kind_to_minikind[inv_entry.kind]
1291
fingerprint = inv_entry.reference_revision
1294
insertions[file_id] = (key, minikind, inv_entry.executable,
1295
fingerprint, new_path)
1296
# Transform moves into delete+add pairs
1297
if None not in (old_path, new_path):
1298
for child in self._iter_child_entries(0, old_path):
1299
if child[0][2] in insertions or child[0][2] in removals:
1301
child_dirname = child[0][0]
1302
child_basename = child[0][1]
1303
minikind = child[1][0][0]
1304
fingerprint = child[1][0][4]
1305
executable = child[1][0][3]
1306
old_child_path = osutils.pathjoin(child[0][0],
1308
removals[child[0][2]] = old_child_path
1309
child_suffix = child_dirname[len(old_path):]
1310
new_child_dirname = (new_path + child_suffix)
1311
key = (new_child_dirname, child_basename, child[0][2])
1312
new_child_path = os.path.join(new_child_dirname,
1314
insertions[child[0][2]] = (key, minikind, executable,
1315
fingerprint, new_child_path)
1316
self._apply_removals(removals.values())
1317
self._apply_insertions(insertions.values())
1319
def _apply_removals(self, removals):
1320
for path in sorted(removals, reverse=True):
1321
dirname, basename = osutils.split(path)
1322
block_i, entry_i, d_present, f_present = \
1323
self._get_block_entry_index(dirname, basename, 0)
1324
entry = self._dirblocks[block_i][1][entry_i]
1325
self._make_absent(entry)
1326
# See if we have a malformed delta: deleting a directory must not
1327
# leave crud behind. This increases the number of bisects needed
1328
# substantially, but deletion or renames of large numbers of paths
1329
# is rare enough it shouldn't be an issue (famous last words?) RBC
1331
block_i, entry_i, d_present, f_present = \
1332
self._get_block_entry_index(path, '', 0)
1334
# The dir block is still present in the dirstate; this could
1335
# be due to it being in a parent tree, or a corrupt delta.
1336
for child_entry in self._dirblocks[block_i][1]:
1337
if child_entry[1][0][0] not in ('r', 'a'):
1338
raise errors.InconsistentDelta(path, entry[0][2],
1339
"The file id was deleted but its children were "
1342
def _apply_insertions(self, adds):
1343
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1344
self.update_minimal(key, minikind, executable, fingerprint,
1345
path_utf8=path_utf8)
1347
def update_basis_by_delta(self, delta, new_revid):
1348
"""Update the parents of this tree after a commit.
1350
This gives the tree one parent, with revision id new_revid. The
1351
inventory delta is applied to the current basis tree to generate the
1352
inventory for the parent new_revid, and all other parent trees are
1355
Note that an exception during the operation of this method will leave
1356
the dirstate in a corrupt state where it should not be saved.
1358
Finally, we expect all changes to be synchronising the basis tree with
1361
:param new_revid: The new revision id for the trees parent.
1362
:param delta: An inventory delta (see apply_inventory_delta) describing
1363
the changes from the current left most parent revision to new_revid.
1365
self._read_dirblocks_if_needed()
1366
self._discard_merge_parents()
1367
if self._ghosts != []:
1368
raise NotImplementedError(self.update_basis_by_delta)
1369
if len(self._parents) == 0:
1370
# setup a blank tree, the most simple way.
1371
empty_parent = DirState.NULL_PARENT_DETAILS
1372
for entry in self._iter_entries():
1373
entry[1].append(empty_parent)
1374
self._parents.append(new_revid)
1376
self._parents[0] = new_revid
1378
delta = sorted(delta, reverse=True)
1382
# The paths this function accepts are unicode and must be encoded as we
1384
encode = cache_utf8.encode
1385
inv_to_entry = self._inv_entry_to_details
1386
# delta is now (deletes, changes), (adds) in reverse lexographical
1388
# deletes in reverse lexographic order are safe to process in situ.
1389
# renames are not, as a rename from any path could go to a path
1390
# lexographically lower, so we transform renames into delete, add pairs,
1391
# expanding them recursively as needed.
1392
# At the same time, to reduce interface friction we convert the input
1393
# inventory entries to dirstate.
1394
root_only = ('', '')
1395
for old_path, new_path, file_id, inv_entry in delta:
1396
if old_path is None:
1397
adds.append((None, encode(new_path), file_id,
1398
inv_to_entry(inv_entry), True))
1399
elif new_path is None:
1400
deletes.append((encode(old_path), None, file_id, None, True))
1401
elif (old_path, new_path) != root_only:
1403
# Because renames must preserve their children we must have
1404
# processed all relocations and removes before hand. The sort
1405
# order ensures we've examined the child paths, but we also
1406
# have to execute the removals, or the split to an add/delete
1407
# pair will result in the deleted item being reinserted, or
1408
# renamed items being reinserted twice - and possibly at the
1409
# wrong place. Splitting into a delete/add pair also simplifies
1410
# the handling of entries with ('f', ...), ('r' ...) because
1411
# the target of the 'r' is old_path here, and we add that to
1412
# deletes, meaning that the add handler does not need to check
1413
# for 'r' items on every pass.
1414
self._update_basis_apply_deletes(deletes)
1416
new_path_utf8 = encode(new_path)
1417
# Split into an add/delete pair recursively.
1418
adds.append((None, new_path_utf8, file_id,
1419
inv_to_entry(inv_entry), False))
1420
# Expunge deletes that we've seen so that deleted/renamed
1421
# children of a rename directory are handled correctly.
1422
new_deletes = reversed(list(self._iter_child_entries(1,
1424
# Remove the current contents of the tree at orig_path, and
1425
# reinsert at the correct new path.
1426
for entry in new_deletes:
1428
source_path = entry[0][0] + '/' + entry[0][1]
1430
source_path = entry[0][1]
1432
target_path = new_path_utf8 + source_path[len(old_path):]
1435
raise AssertionError("cannot rename directory to"
1437
target_path = source_path[len(old_path) + 1:]
1438
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1440
(source_path, target_path, entry[0][2], None, False))
1442
(encode(old_path), new_path, file_id, None, False))
1444
# changes to just the root should not require remove/insertion
1446
changes.append((encode(old_path), encode(new_path), file_id,
1447
inv_to_entry(inv_entry)))
1449
# Finish expunging deletes/first half of renames.
1450
self._update_basis_apply_deletes(deletes)
1451
# Reinstate second half of renames and new paths.
1452
self._update_basis_apply_adds(adds)
1453
# Apply in-situ changes.
1454
self._update_basis_apply_changes(changes)
1456
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1457
self._header_state = DirState.IN_MEMORY_MODIFIED
1458
self._id_index = None
1461
def _update_basis_apply_adds(self, adds):
1462
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1464
They may be adds, or renames that have been split into add/delete
1467
:param adds: A sequence of adds. Each add is a tuple:
1468
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1469
is False when the add is the second half of a remove-and-reinsert
1470
pair created to handle renames and deletes.
1472
# Adds are accumulated partly from renames, so can be in any input
1475
# adds is now in lexographic order, which places all parents before
1476
# their children, so we can process it linearly.
1478
for old_path, new_path, file_id, new_details, real_add in adds:
1479
# the entry for this file_id must be in tree 0.
1480
entry = self._get_entry(0, file_id, new_path)
1481
if entry[0] is None or entry[0][2] != file_id:
1482
self._changes_aborted = True
1483
raise errors.InconsistentDelta(new_path, file_id,
1484
'working tree does not contain new entry')
1485
if real_add and entry[1][1][0] not in absent:
1486
self._changes_aborted = True
1487
raise errors.InconsistentDelta(new_path, file_id,
1488
'The entry was considered to be a genuinely new record,'
1489
' but there was already an old record for it.')
1490
# We don't need to update the target of an 'r' because the handling
1491
# of renames turns all 'r' situations into a delete at the original
1493
entry[1][1] = new_details
1495
def _update_basis_apply_changes(self, changes):
1496
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1498
:param adds: A sequence of changes. Each change is a tuple:
1499
(path_utf8, path_utf8, file_id, (entry_details))
1502
for old_path, new_path, file_id, new_details in changes:
1503
# the entry for this file_id must be in tree 0.
1504
entry = self._get_entry(0, file_id, new_path)
1505
if entry[0] is None or entry[0][2] != file_id:
1506
self._changes_aborted = True
1507
raise errors.InconsistentDelta(new_path, file_id,
1508
'working tree does not contain new entry')
1509
if (entry[1][0][0] in absent or
1510
entry[1][1][0] in absent):
1511
self._changes_aborted = True
1512
raise errors.InconsistentDelta(new_path, file_id,
1513
'changed considered absent')
1514
entry[1][1] = new_details
1516
def _update_basis_apply_deletes(self, deletes):
1517
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1519
They may be deletes, or renames that have been split into add/delete
1522
:param deletes: A sequence of deletes. Each delete is a tuple:
1523
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1524
real_delete is True when the desired outcome is an actual deletion
1525
rather than the rename handling logic temporarily deleting a path
1526
during the replacement of a parent.
1528
null = DirState.NULL_PARENT_DETAILS
1529
for old_path, new_path, file_id, _, real_delete in deletes:
1530
if real_delete != (new_path is None):
1531
raise AssertionError("bad delete delta")
1532
# the entry for this file_id must be in tree 1.
1533
dirname, basename = osutils.split(old_path)
1534
block_index, entry_index, dir_present, file_present = \
1535
self._get_block_entry_index(dirname, basename, 1)
1536
if not file_present:
1537
self._changes_aborted = True
1538
raise errors.InconsistentDelta(old_path, file_id,
1539
'basis tree does not contain removed entry')
1540
entry = self._dirblocks[block_index][1][entry_index]
1541
if entry[0][2] != file_id:
1542
self._changes_aborted = True
1543
raise errors.InconsistentDelta(old_path, file_id,
1544
'mismatched file_id in tree 1')
1546
if entry[1][0][0] != 'a':
1547
self._changes_aborted = True
1548
raise errors.InconsistentDelta(old_path, file_id,
1549
'This was marked as a real delete, but the WT state'
1550
' claims that it still exists and is versioned.')
1551
del self._dirblocks[block_index][1][entry_index]
1553
if entry[1][0][0] == 'a':
1554
self._changes_aborted = True
1555
raise errors.InconsistentDelta(old_path, file_id,
1556
'The entry was considered a rename, but the source path'
1557
' is marked as absent.')
1558
# For whatever reason, we were asked to rename an entry
1559
# that was originally marked as deleted. This could be
1560
# because we are renaming the parent directory, and the WT
1561
# current state has the file marked as deleted.
1562
elif entry[1][0][0] == 'r':
1563
# implement the rename
1564
del self._dirblocks[block_index][1][entry_index]
1566
# it is being resurrected here, so blank it out temporarily.
1567
self._dirblocks[block_index][1][entry_index][1][1] = null
1569
def _observed_sha1(self, entry, sha1, stat_value,
1570
_stat_to_minikind=_stat_to_minikind, _pack_stat=pack_stat):
1571
"""Note the sha1 of a file.
1573
:param entry: The entry the sha1 is for.
1574
:param sha1: The observed sha1.
1575
:param stat_value: The os.lstat for the file.
1578
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1582
packed_stat = _pack_stat(stat_value)
1584
if self._cutoff_time is None:
1585
self._sha_cutoff_time()
1586
if (stat_value.st_mtime < self._cutoff_time
1587
and stat_value.st_ctime < self._cutoff_time):
1588
entry[1][0] = ('f', sha1, entry[1][0][2], entry[1][0][3],
1590
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1592
def _sha_cutoff_time(self):
1593
"""Return cutoff time.
1595
Files modified more recently than this time are at risk of being
1596
undetectably modified and so can't be cached.
1598
# Cache the cutoff time as long as we hold a lock.
1599
# time.time() isn't super expensive (approx 3.38us), but
1600
# when you call it 50,000 times it adds up.
1601
# For comparison, os.lstat() costs 7.2us if it is hot.
1602
self._cutoff_time = int(time.time()) - 3
1603
return self._cutoff_time
1605
def _lstat(self, abspath, entry):
1606
"""Return the os.lstat value for this path."""
1607
return os.lstat(abspath)
1609
def _sha1_file_and_mutter(self, abspath):
1610
# when -Dhashcache is turned on, this is monkey-patched in to log
1612
trace.mutter("dirstate sha1 " + abspath)
1613
return self._sha1_provider.sha1(abspath)
1615
def _is_executable(self, mode, old_executable):
1616
"""Is this file executable?"""
1617
return bool(S_IEXEC & mode)
1619
def _is_executable_win32(self, mode, old_executable):
1620
"""On win32 the executable bit is stored in the dirstate."""
1621
return old_executable
1623
if sys.platform == 'win32':
1624
_is_executable = _is_executable_win32
1626
def _read_link(self, abspath, old_link):
1627
"""Read the target of a symlink"""
1628
# TODO: jam 200700301 On Win32, this could just return the value
1629
# already in memory. However, this really needs to be done at a
1630
# higher level, because there either won't be anything on disk,
1631
# or the thing on disk will be a file.
1632
fs_encoding = osutils._fs_enc
1633
if isinstance(abspath, unicode):
1634
# abspath is defined as the path to pass to lstat. readlink is
1635
# buggy in python < 2.6 (it doesn't encode unicode path into FS
1636
# encoding), so we need to encode ourselves knowing that unicode
1637
# paths are produced by UnicodeDirReader on purpose.
1638
abspath = abspath.encode(fs_encoding)
1639
target = os.readlink(abspath)
1640
if fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1641
# Change encoding if needed
1642
target = target.decode(fs_encoding).encode('UTF-8')
1645
def get_ghosts(self):
1646
"""Return a list of the parent tree revision ids that are ghosts."""
1647
self._read_header_if_needed()
1650
def get_lines(self):
1651
"""Serialise the entire dirstate to a sequence of lines."""
1652
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1653
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1654
# read whats on disk.
1655
self._state_file.seek(0)
1656
return self._state_file.readlines()
1658
lines.append(self._get_parents_line(self.get_parent_ids()))
1659
lines.append(self._get_ghosts_line(self._ghosts))
1660
# append the root line which is special cased
1661
lines.extend(map(self._entry_to_line, self._iter_entries()))
1662
return self._get_output_lines(lines)
1664
def _get_ghosts_line(self, ghost_ids):
1665
"""Create a line for the state file for ghost information."""
1666
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1668
def _get_parents_line(self, parent_ids):
1669
"""Create a line for the state file for parents information."""
1670
return '\0'.join([str(len(parent_ids))] + parent_ids)
1672
def _get_fields_to_entry(self):
1673
"""Get a function which converts entry fields into a entry record.
1675
This handles size and executable, as well as parent records.
1677
:return: A function which takes a list of fields, and returns an
1678
appropriate record for storing in memory.
1680
# This is intentionally unrolled for performance
1681
num_present_parents = self._num_present_parents()
1682
if num_present_parents == 0:
1683
def fields_to_entry_0_parents(fields, _int=int):
1684
path_name_file_id_key = (fields[0], fields[1], fields[2])
1685
return (path_name_file_id_key, [
1687
fields[3], # minikind
1688
fields[4], # fingerprint
1689
_int(fields[5]), # size
1690
fields[6] == 'y', # executable
1691
fields[7], # packed_stat or revision_id
1693
return fields_to_entry_0_parents
1694
elif num_present_parents == 1:
1695
def fields_to_entry_1_parent(fields, _int=int):
1696
path_name_file_id_key = (fields[0], fields[1], fields[2])
1697
return (path_name_file_id_key, [
1699
fields[3], # minikind
1700
fields[4], # fingerprint
1701
_int(fields[5]), # size
1702
fields[6] == 'y', # executable
1703
fields[7], # packed_stat or revision_id
1706
fields[8], # minikind
1707
fields[9], # fingerprint
1708
_int(fields[10]), # size
1709
fields[11] == 'y', # executable
1710
fields[12], # packed_stat or revision_id
1713
return fields_to_entry_1_parent
1714
elif num_present_parents == 2:
1715
def fields_to_entry_2_parents(fields, _int=int):
1716
path_name_file_id_key = (fields[0], fields[1], fields[2])
1717
return (path_name_file_id_key, [
1719
fields[3], # minikind
1720
fields[4], # fingerprint
1721
_int(fields[5]), # size
1722
fields[6] == 'y', # executable
1723
fields[7], # packed_stat or revision_id
1726
fields[8], # minikind
1727
fields[9], # fingerprint
1728
_int(fields[10]), # size
1729
fields[11] == 'y', # executable
1730
fields[12], # packed_stat or revision_id
1733
fields[13], # minikind
1734
fields[14], # fingerprint
1735
_int(fields[15]), # size
1736
fields[16] == 'y', # executable
1737
fields[17], # packed_stat or revision_id
1740
return fields_to_entry_2_parents
1742
def fields_to_entry_n_parents(fields, _int=int):
1743
path_name_file_id_key = (fields[0], fields[1], fields[2])
1744
trees = [(fields[cur], # minikind
1745
fields[cur+1], # fingerprint
1746
_int(fields[cur+2]), # size
1747
fields[cur+3] == 'y', # executable
1748
fields[cur+4], # stat or revision_id
1749
) for cur in xrange(3, len(fields)-1, 5)]
1750
return path_name_file_id_key, trees
1751
return fields_to_entry_n_parents
1753
def get_parent_ids(self):
1754
"""Return a list of the parent tree ids for the directory state."""
1755
self._read_header_if_needed()
1756
return list(self._parents)
1758
def _get_block_entry_index(self, dirname, basename, tree_index):
1759
"""Get the coordinates for a path in the state structure.
1761
:param dirname: The utf8 dirname to lookup.
1762
:param basename: The utf8 basename to lookup.
1763
:param tree_index: The index of the tree for which this lookup should
1765
:return: A tuple describing where the path is located, or should be
1766
inserted. The tuple contains four fields: the block index, the row
1767
index, the directory is present (boolean), the entire path is
1768
present (boolean). There is no guarantee that either
1769
coordinate is currently reachable unless the found field for it is
1770
True. For instance, a directory not present in the searched tree
1771
may be returned with a value one greater than the current highest
1772
block offset. The directory present field will always be True when
1773
the path present field is True. The directory present field does
1774
NOT indicate that the directory is present in the searched tree,
1775
rather it indicates that there are at least some files in some
1778
self._read_dirblocks_if_needed()
1779
key = dirname, basename, ''
1780
block_index, present = self._find_block_index_from_key(key)
1782
# no such directory - return the dir index and 0 for the row.
1783
return block_index, 0, False, False
1784
block = self._dirblocks[block_index][1] # access the entries only
1785
entry_index, present = self._find_entry_index(key, block)
1786
# linear search through entries at this path to find the one
1788
while entry_index < len(block) and block[entry_index][0][1] == basename:
1789
if block[entry_index][1][tree_index][0] not in 'ar':
1790
# neither absent or relocated
1791
return block_index, entry_index, True, True
1793
return block_index, entry_index, True, False
1795
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None, include_deleted=False):
1796
"""Get the dirstate entry for path in tree tree_index.
1798
If either file_id or path is supplied, it is used as the key to lookup.
1799
If both are supplied, the fastest lookup is used, and an error is
1800
raised if they do not both point at the same row.
1802
:param tree_index: The index of the tree we wish to locate this path
1803
in. If the path is present in that tree, the entry containing its
1804
details is returned, otherwise (None, None) is returned
1805
0 is the working tree, higher indexes are successive parent
1807
:param fileid_utf8: A utf8 file_id to look up.
1808
:param path_utf8: An utf8 path to be looked up.
1809
:param include_deleted: If True, and performing a lookup via
1810
fileid_utf8 rather than path_utf8, return an entry for deleted
1812
:return: The dirstate entry tuple for path, or (None, None)
1814
self._read_dirblocks_if_needed()
1815
if path_utf8 is not None:
1816
if type(path_utf8) is not str:
1817
raise AssertionError('path_utf8 is not a str: %s %s'
1818
% (type(path_utf8), path_utf8))
1819
# path lookups are faster
1820
dirname, basename = osutils.split(path_utf8)
1821
block_index, entry_index, dir_present, file_present = \
1822
self._get_block_entry_index(dirname, basename, tree_index)
1823
if not file_present:
1825
entry = self._dirblocks[block_index][1][entry_index]
1826
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
1827
raise AssertionError('unversioned entry?')
1829
if entry[0][2] != fileid_utf8:
1830
self._changes_aborted = True
1831
raise errors.BzrError('integrity error ? : mismatching'
1832
' tree_index, file_id and path')
1835
possible_keys = self._get_id_index().get(fileid_utf8, None)
1836
if not possible_keys:
1838
for key in possible_keys:
1839
block_index, present = \
1840
self._find_block_index_from_key(key)
1841
# strange, probably indicates an out of date
1842
# id index - for now, allow this.
1845
# WARNING: DO not change this code to use _get_block_entry_index
1846
# as that function is not suitable: it does not use the key
1847
# to lookup, and thus the wrong coordinates are returned.
1848
block = self._dirblocks[block_index][1]
1849
entry_index, present = self._find_entry_index(key, block)
1851
entry = self._dirblocks[block_index][1][entry_index]
1852
if entry[1][tree_index][0] in 'fdlt':
1853
# this is the result we are looking for: the
1854
# real home of this file_id in this tree.
1856
if entry[1][tree_index][0] == 'a':
1857
# there is no home for this entry in this tree
1861
if entry[1][tree_index][0] != 'r':
1862
raise AssertionError(
1863
"entry %r has invalid minikind %r for tree %r" \
1865
entry[1][tree_index][0],
1867
real_path = entry[1][tree_index][1]
1868
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1869
path_utf8=real_path)
1873
def initialize(cls, path, sha1_provider=None):
1874
"""Create a new dirstate on path.
1876
The new dirstate will be an empty tree - that is it has no parents,
1877
and only a root node - which has id ROOT_ID.
1879
:param path: The name of the file for the dirstate.
1880
:param sha1_provider: an object meeting the SHA1Provider interface.
1881
If None, a DefaultSHA1Provider is used.
1882
:return: A write-locked DirState object.
1884
# This constructs a new DirState object on a path, sets the _state_file
1885
# to a new empty file for that path. It then calls _set_data() with our
1886
# stock empty dirstate information - a root with ROOT_ID, no children,
1887
# and no parents. Finally it calls save() to ensure that this data will
1889
if sha1_provider is None:
1890
sha1_provider = DefaultSHA1Provider()
1891
result = cls(path, sha1_provider)
1892
# root dir and root dir contents with no children.
1893
empty_tree_dirblocks = [('', []), ('', [])]
1894
# a new root directory, with a NULLSTAT.
1895
empty_tree_dirblocks[0][1].append(
1896
(('', '', inventory.ROOT_ID), [
1897
('d', '', 0, False, DirState.NULLSTAT),
1901
result._set_data([], empty_tree_dirblocks)
1909
def _inv_entry_to_details(inv_entry):
1910
"""Convert an inventory entry (from a revision tree) to state details.
1912
:param inv_entry: An inventory entry whose sha1 and link targets can be
1913
relied upon, and which has a revision set.
1914
:return: A details tuple - the details for a single tree at a path +
1917
kind = inv_entry.kind
1918
minikind = DirState._kind_to_minikind[kind]
1919
tree_data = inv_entry.revision
1920
if kind == 'directory':
1924
elif kind == 'symlink':
1925
if inv_entry.symlink_target is None:
1928
fingerprint = inv_entry.symlink_target.encode('utf8')
1931
elif kind == 'file':
1932
fingerprint = inv_entry.text_sha1 or ''
1933
size = inv_entry.text_size or 0
1934
executable = inv_entry.executable
1935
elif kind == 'tree-reference':
1936
fingerprint = inv_entry.reference_revision or ''
1940
raise Exception("can't pack %s" % inv_entry)
1941
return (minikind, fingerprint, size, executable, tree_data)
1943
def _iter_child_entries(self, tree_index, path_utf8):
1944
"""Iterate over all the entries that are children of path_utf.
1946
This only returns entries that are present (not in 'a', 'r') in
1947
tree_index. tree_index data is not refreshed, so if tree 0 is used,
1948
results may differ from that obtained if paths were statted to
1949
determine what ones were directories.
1951
Asking for the children of a non-directory will return an empty
1955
next_pending_dirs = [path_utf8]
1957
while next_pending_dirs:
1958
pending_dirs = next_pending_dirs
1959
next_pending_dirs = []
1960
for path in pending_dirs:
1961
block_index, present = self._find_block_index_from_key(
1963
if block_index == 0:
1965
if len(self._dirblocks) == 1:
1966
# asked for the children of the root with no other
1970
# children of a non-directory asked for.
1972
block = self._dirblocks[block_index]
1973
for entry in block[1]:
1974
kind = entry[1][tree_index][0]
1975
if kind not in absent:
1979
path = entry[0][0] + '/' + entry[0][1]
1982
next_pending_dirs.append(path)
1984
def _iter_entries(self):
1985
"""Iterate over all the entries in the dirstate.
1987
Each yelt item is an entry in the standard format described in the
1988
docstring of bzrlib.dirstate.
1990
self._read_dirblocks_if_needed()
1991
for directory in self._dirblocks:
1992
for entry in directory[1]:
1995
def _get_id_index(self):
1996
"""Get an id index of self._dirblocks."""
1997
if self._id_index is None:
1999
for key, tree_details in self._iter_entries():
2000
id_index.setdefault(key[2], set()).add(key)
2001
self._id_index = id_index
2002
return self._id_index
2004
def _get_output_lines(self, lines):
2005
"""Format lines for final output.
2007
:param lines: A sequence of lines containing the parents list and the
2010
output_lines = [DirState.HEADER_FORMAT_3]
2011
lines.append('') # a final newline
2012
inventory_text = '\0\n\0'.join(lines)
2013
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
2014
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
2015
num_entries = len(lines)-3
2016
output_lines.append('num_entries: %s\n' % (num_entries,))
2017
output_lines.append(inventory_text)
2020
def _make_deleted_row(self, fileid_utf8, parents):
2021
"""Return a deleted row for fileid_utf8."""
2022
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
2025
def _num_present_parents(self):
2026
"""The number of parent entries in each record row."""
2027
return len(self._parents) - len(self._ghosts)
2030
def on_file(path, sha1_provider=None):
2031
"""Construct a DirState on the file at path "path".
2033
:param path: The path at which the dirstate file on disk should live.
2034
:param sha1_provider: an object meeting the SHA1Provider interface.
2035
If None, a DefaultSHA1Provider is used.
2036
:return: An unlocked DirState object, associated with the given path.
2038
if sha1_provider is None:
2039
sha1_provider = DefaultSHA1Provider()
2040
result = DirState(path, sha1_provider)
2043
def _read_dirblocks_if_needed(self):
2044
"""Read in all the dirblocks from the file if they are not in memory.
2046
This populates self._dirblocks, and sets self._dirblock_state to
2047
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
2050
self._read_header_if_needed()
2051
if self._dirblock_state == DirState.NOT_IN_MEMORY:
2052
_read_dirblocks(self)
2054
def _read_header(self):
2055
"""This reads in the metadata header, and the parent ids.
2057
After reading in, the file should be positioned at the null
2058
just before the start of the first record in the file.
2060
:return: (expected crc checksum, number of entries, parent list)
2062
self._read_prelude()
2063
parent_line = self._state_file.readline()
2064
info = parent_line.split('\0')
2065
num_parents = int(info[0])
2066
self._parents = info[1:-1]
2067
ghost_line = self._state_file.readline()
2068
info = ghost_line.split('\0')
2069
num_ghosts = int(info[1])
2070
self._ghosts = info[2:-1]
2071
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2072
self._end_of_header = self._state_file.tell()
2074
def _read_header_if_needed(self):
2075
"""Read the header of the dirstate file if needed."""
2076
# inline this as it will be called a lot
2077
if not self._lock_token:
2078
raise errors.ObjectNotLocked(self)
2079
if self._header_state == DirState.NOT_IN_MEMORY:
2082
def _read_prelude(self):
2083
"""Read in the prelude header of the dirstate file.
2085
This only reads in the stuff that is not connected to the crc
2086
checksum. The position will be correct to read in the rest of
2087
the file and check the checksum after this point.
2088
The next entry in the file should be the number of parents,
2089
and their ids. Followed by a newline.
2091
header = self._state_file.readline()
2092
if header != DirState.HEADER_FORMAT_3:
2093
raise errors.BzrError(
2094
'invalid header line: %r' % (header,))
2095
crc_line = self._state_file.readline()
2096
if not crc_line.startswith('crc32: '):
2097
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2098
self.crc_expected = int(crc_line[len('crc32: '):-1])
2099
num_entries_line = self._state_file.readline()
2100
if not num_entries_line.startswith('num_entries: '):
2101
raise errors.BzrError('missing num_entries line')
2102
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2104
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
2105
"""Find a sha1 given a stat lookup."""
2106
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
2108
def _get_packed_stat_index(self):
2109
"""Get a packed_stat index of self._dirblocks."""
2110
if self._packed_stat_index is None:
2112
for key, tree_details in self._iter_entries():
2113
if tree_details[0][0] == 'f':
2114
index[tree_details[0][4]] = tree_details[0][1]
2115
self._packed_stat_index = index
2116
return self._packed_stat_index
2119
"""Save any pending changes created during this session.
2121
We reuse the existing file, because that prevents race conditions with
2122
file creation, and use oslocks on it to prevent concurrent modification
2123
and reads - because dirstate's incremental data aggregation is not
2124
compatible with reading a modified file, and replacing a file in use by
2125
another process is impossible on Windows.
2127
A dirstate in read only mode should be smart enough though to validate
2128
that the file has not changed, and otherwise discard its cache and
2129
start over, to allow for fine grained read lock duration, so 'status'
2130
wont block 'commit' - for example.
2132
if self._changes_aborted:
2133
# Should this be a warning? For now, I'm expecting that places that
2134
# mark it inconsistent will warn, making a warning here redundant.
2135
trace.mutter('Not saving DirState because '
2136
'_changes_aborted is set.')
2138
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
2139
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2141
grabbed_write_lock = False
2142
if self._lock_state != 'w':
2143
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2144
# Switch over to the new lock, as the old one may be closed.
2145
# TODO: jam 20070315 We should validate the disk file has
2146
# not changed contents. Since temporary_write_lock may
2147
# not be an atomic operation.
2148
self._lock_token = new_lock
2149
self._state_file = new_lock.f
2150
if not grabbed_write_lock:
2151
# We couldn't grab a write lock, so we switch back to a read one
2154
self._state_file.seek(0)
2155
self._state_file.writelines(self.get_lines())
2156
self._state_file.truncate()
2157
self._state_file.flush()
2158
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2159
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
2161
if grabbed_write_lock:
2162
self._lock_token = self._lock_token.restore_read_lock()
2163
self._state_file = self._lock_token.f
2164
# TODO: jam 20070315 We should validate the disk file has
2165
# not changed contents. Since restore_read_lock may
2166
# not be an atomic operation.
2168
def _set_data(self, parent_ids, dirblocks):
2169
"""Set the full dirstate data in memory.
2171
This is an internal function used to completely replace the objects
2172
in memory state. It puts the dirstate into state 'full-dirty'.
2174
:param parent_ids: A list of parent tree revision ids.
2175
:param dirblocks: A list containing one tuple for each directory in the
2176
tree. Each tuple contains the directory path and a list of entries
2177
found in that directory.
2179
# our memory copy is now authoritative.
2180
self._dirblocks = dirblocks
2181
self._header_state = DirState.IN_MEMORY_MODIFIED
2182
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2183
self._parents = list(parent_ids)
2184
self._id_index = None
2185
self._packed_stat_index = None
2187
def set_path_id(self, path, new_id):
2188
"""Change the id of path to new_id in the current working tree.
2190
:param path: The path inside the tree to set - '' is the root, 'foo'
2191
is the path foo in the root.
2192
:param new_id: The new id to assign to the path. This must be a utf8
2193
file id (not unicode, and not None).
2195
self._read_dirblocks_if_needed()
2197
# TODO: logic not written
2198
raise NotImplementedError(self.set_path_id)
2199
# TODO: check new id is unique
2200
entry = self._get_entry(0, path_utf8=path)
2201
if entry[0][2] == new_id:
2202
# Nothing to change.
2204
# mark the old path absent, and insert a new root path
2205
self._make_absent(entry)
2206
self.update_minimal(('', '', new_id), 'd',
2207
path_utf8='', packed_stat=entry[1][0][4])
2208
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2209
if self._id_index is not None:
2210
self._id_index.setdefault(new_id, set()).add(entry[0])
2212
def set_parent_trees(self, trees, ghosts):
2213
"""Set the parent trees for the dirstate.
2215
:param trees: A list of revision_id, tree tuples. tree must be provided
2216
even if the revision_id refers to a ghost: supply an empty tree in
2218
:param ghosts: A list of the revision_ids that are ghosts at the time
2221
# TODO: generate a list of parent indexes to preserve to save
2222
# processing specific parent trees. In the common case one tree will
2223
# be preserved - the left most parent.
2224
# TODO: if the parent tree is a dirstate, we might want to walk them
2225
# all by path in parallel for 'optimal' common-case performance.
2226
# generate new root row.
2227
self._read_dirblocks_if_needed()
2228
# TODO future sketch: Examine the existing parents to generate a change
2229
# map and then walk the new parent trees only, mapping them into the
2230
# dirstate. Walk the dirstate at the same time to remove unreferenced
2233
# sketch: loop over all entries in the dirstate, cherry picking
2234
# entries from the parent trees, if they are not ghost trees.
2235
# after we finish walking the dirstate, all entries not in the dirstate
2236
# are deletes, so we want to append them to the end as per the design
2237
# discussions. So do a set difference on ids with the parents to
2238
# get deletes, and add them to the end.
2239
# During the update process we need to answer the following questions:
2240
# - find other keys containing a fileid in order to create cross-path
2241
# links. We dont't trivially use the inventory from other trees
2242
# because this leads to either double touching, or to accessing
2244
# - find other keys containing a path
2245
# We accumulate each entry via this dictionary, including the root
2248
# we could do parallel iterators, but because file id data may be
2249
# scattered throughout, we dont save on index overhead: we have to look
2250
# at everything anyway. We can probably save cycles by reusing parent
2251
# data and doing an incremental update when adding an additional
2252
# parent, but for now the common cases are adding a new parent (merge),
2253
# and replacing completely (commit), and commit is more common: so
2254
# optimise merge later.
2256
# ---- start generation of full tree mapping data
2257
# what trees should we use?
2258
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2259
# how many trees do we end up with
2260
parent_count = len(parent_trees)
2262
# one: the current tree
2263
for entry in self._iter_entries():
2264
# skip entries not in the current tree
2265
if entry[1][0][0] in 'ar': # absent, relocated
2267
by_path[entry[0]] = [entry[1][0]] + \
2268
[DirState.NULL_PARENT_DETAILS] * parent_count
2269
id_index[entry[0][2]] = set([entry[0]])
2271
# now the parent trees:
2272
for tree_index, tree in enumerate(parent_trees):
2273
# the index is off by one, adjust it.
2274
tree_index = tree_index + 1
2275
# when we add new locations for a fileid we need these ranges for
2276
# any fileid in this tree as we set the by_path[id] to:
2277
# already_processed_tree_details + new_details + new_location_suffix
2278
# the suffix is from tree_index+1:parent_count+1.
2279
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2280
# now stitch in all the entries from this tree
2281
for path, entry in tree.inventory.iter_entries_by_dir():
2282
# here we process each trees details for each item in the tree.
2283
# we first update any existing entries for the id at other paths,
2284
# then we either create or update the entry for the id at the
2285
# right path, and finally we add (if needed) a mapping from
2286
# file_id to this path. We do it in this order to allow us to
2287
# avoid checking all known paths for the id when generating a
2288
# new entry at this path: by adding the id->path mapping last,
2289
# all the mappings are valid and have correct relocation
2290
# records where needed.
2291
file_id = entry.file_id
2292
path_utf8 = path.encode('utf8')
2293
dirname, basename = osutils.split(path_utf8)
2294
new_entry_key = (dirname, basename, file_id)
2295
# tree index consistency: All other paths for this id in this tree
2296
# index must point to the correct path.
2297
for entry_key in id_index.setdefault(file_id, set()):
2298
# TODO:PROFILING: It might be faster to just update
2299
# rather than checking if we need to, and then overwrite
2300
# the one we are located at.
2301
if entry_key != new_entry_key:
2302
# this file id is at a different path in one of the
2303
# other trees, so put absent pointers there
2304
# This is the vertical axis in the matrix, all pointing
2306
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2307
# by path consistency: Insert into an existing path record (trivial), or
2308
# add a new one with relocation pointers for the other tree indexes.
2309
if new_entry_key in id_index[file_id]:
2310
# there is already an entry where this data belongs, just insert it.
2311
by_path[new_entry_key][tree_index] = \
2312
self._inv_entry_to_details(entry)
2314
# add relocated entries to the horizontal axis - this row
2315
# mapping from path,id. We need to look up the correct path
2316
# for the indexes from 0 to tree_index -1
2318
for lookup_index in xrange(tree_index):
2319
# boundary case: this is the first occurence of file_id
2320
# so there are no id_indexs, possibly take this out of
2322
if not len(id_index[file_id]):
2323
new_details.append(DirState.NULL_PARENT_DETAILS)
2325
# grab any one entry, use it to find the right path.
2326
# TODO: optimise this to reduce memory use in highly
2327
# fragmented situations by reusing the relocation
2329
a_key = iter(id_index[file_id]).next()
2330
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2331
# its a pointer or missing statement, use it as is.
2332
new_details.append(by_path[a_key][lookup_index])
2334
# we have the right key, make a pointer to it.
2335
real_path = ('/'.join(a_key[0:2])).strip('/')
2336
new_details.append(('r', real_path, 0, False, ''))
2337
new_details.append(self._inv_entry_to_details(entry))
2338
new_details.extend(new_location_suffix)
2339
by_path[new_entry_key] = new_details
2340
id_index[file_id].add(new_entry_key)
2341
# --- end generation of full tree mappings
2343
# sort and output all the entries
2344
new_entries = self._sort_entries(by_path.items())
2345
self._entries_to_current_state(new_entries)
2346
self._parents = [rev_id for rev_id, tree in trees]
2347
self._ghosts = list(ghosts)
2348
self._header_state = DirState.IN_MEMORY_MODIFIED
2349
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2350
self._id_index = id_index
2352
def _sort_entries(self, entry_list):
2353
"""Given a list of entries, sort them into the right order.
2355
This is done when constructing a new dirstate from trees - normally we
2356
try to keep everything in sorted blocks all the time, but sometimes
2357
it's easier to sort after the fact.
2360
# sort by: directory parts, file name, file id
2361
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2362
return sorted(entry_list, key=_key)
2364
def set_state_from_inventory(self, new_inv):
2365
"""Set new_inv as the current state.
2367
This API is called by tree transform, and will usually occur with
2368
existing parent trees.
2370
:param new_inv: The inventory object to set current state from.
2372
if 'evil' in debug.debug_flags:
2373
trace.mutter_callsite(1,
2374
"set_state_from_inventory called; please mutate the tree instead")
2375
self._read_dirblocks_if_needed()
2377
# Two iterators: current data and new data, both in dirblock order.
2378
# We zip them together, which tells about entries that are new in the
2379
# inventory, or removed in the inventory, or present in both and
2382
# You might think we could just synthesize a new dirstate directly
2383
# since we're processing it in the right order. However, we need to
2384
# also consider there may be any number of parent trees and relocation
2385
# pointers, and we don't want to duplicate that here.
2386
new_iterator = new_inv.iter_entries_by_dir()
2387
# we will be modifying the dirstate, so we need a stable iterator. In
2388
# future we might write one, for now we just clone the state into a
2389
# list - which is a shallow copy.
2390
old_iterator = iter(list(self._iter_entries()))
2391
# both must have roots so this is safe:
2392
current_new = new_iterator.next()
2393
current_old = old_iterator.next()
2394
def advance(iterator):
2396
return iterator.next()
2397
except StopIteration:
2399
while current_new or current_old:
2400
# skip entries in old that are not really there
2401
if current_old and current_old[1][0][0] in 'ar':
2402
# relocated or absent
2403
current_old = advance(old_iterator)
2406
# convert new into dirblock style
2407
new_path_utf8 = current_new[0].encode('utf8')
2408
new_dirname, new_basename = osutils.split(new_path_utf8)
2409
new_id = current_new[1].file_id
2410
new_entry_key = (new_dirname, new_basename, new_id)
2411
current_new_minikind = \
2412
DirState._kind_to_minikind[current_new[1].kind]
2413
if current_new_minikind == 't':
2414
fingerprint = current_new[1].reference_revision or ''
2416
# We normally only insert or remove records, or update
2417
# them when it has significantly changed. Then we want to
2418
# erase its fingerprint. Unaffected records should
2419
# normally not be updated at all.
2422
# for safety disable variables
2423
new_path_utf8 = new_dirname = new_basename = new_id = \
2424
new_entry_key = None
2425
# 5 cases, we dont have a value that is strictly greater than everything, so
2426
# we make both end conditions explicit
2428
# old is finished: insert current_new into the state.
2429
self.update_minimal(new_entry_key, current_new_minikind,
2430
executable=current_new[1].executable,
2431
path_utf8=new_path_utf8, fingerprint=fingerprint)
2432
current_new = advance(new_iterator)
2433
elif not current_new:
2435
self._make_absent(current_old)
2436
current_old = advance(old_iterator)
2437
elif new_entry_key == current_old[0]:
2438
# same - common case
2439
# We're looking at the same path and id in both the dirstate
2440
# and inventory, so just need to update the fields in the
2441
# dirstate from the one in the inventory.
2442
# TODO: update the record if anything significant has changed.
2443
# the minimal required trigger is if the execute bit or cached
2445
if (current_old[1][0][3] != current_new[1].executable or
2446
current_old[1][0][0] != current_new_minikind):
2447
self.update_minimal(current_old[0], current_new_minikind,
2448
executable=current_new[1].executable,
2449
path_utf8=new_path_utf8, fingerprint=fingerprint)
2450
# both sides are dealt with, move on
2451
current_old = advance(old_iterator)
2452
current_new = advance(new_iterator)
2453
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2454
or (new_dirname == current_old[0][0]
2455
and new_entry_key[1:] < current_old[0][1:])):
2457
# add a entry for this and advance new
2458
self.update_minimal(new_entry_key, current_new_minikind,
2459
executable=current_new[1].executable,
2460
path_utf8=new_path_utf8, fingerprint=fingerprint)
2461
current_new = advance(new_iterator)
2463
# we've advanced past the place where the old key would be,
2464
# without seeing it in the new list. so it must be gone.
2465
self._make_absent(current_old)
2466
current_old = advance(old_iterator)
2467
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2468
self._id_index = None
2469
self._packed_stat_index = None
2471
def _make_absent(self, current_old):
2472
"""Mark current_old - an entry - as absent for tree 0.
2474
:return: True if this was the last details entry for the entry key:
2475
that is, if the underlying block has had the entry removed, thus
2476
shrinking in length.
2478
# build up paths that this id will be left at after the change is made,
2479
# so we can update their cross references in tree 0
2480
all_remaining_keys = set()
2481
# Dont check the working tree, because it's going.
2482
for details in current_old[1][1:]:
2483
if details[0] not in 'ar': # absent, relocated
2484
all_remaining_keys.add(current_old[0])
2485
elif details[0] == 'r': # relocated
2486
# record the key for the real path.
2487
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2488
# absent rows are not present at any path.
2489
last_reference = current_old[0] not in all_remaining_keys
2491
# the current row consists entire of the current item (being marked
2492
# absent), and relocated or absent entries for the other trees:
2493
# Remove it, its meaningless.
2494
block = self._find_block(current_old[0])
2495
entry_index, present = self._find_entry_index(current_old[0], block[1])
2497
raise AssertionError('could not find entry for %s' % (current_old,))
2498
block[1].pop(entry_index)
2499
# if we have an id_index in use, remove this key from it for this id.
2500
if self._id_index is not None:
2501
self._id_index[current_old[0][2]].remove(current_old[0])
2502
# update all remaining keys for this id to record it as absent. The
2503
# existing details may either be the record we are marking as deleted
2504
# (if there were other trees with the id present at this path), or may
2506
for update_key in all_remaining_keys:
2507
update_block_index, present = \
2508
self._find_block_index_from_key(update_key)
2510
raise AssertionError('could not find block for %s' % (update_key,))
2511
update_entry_index, present = \
2512
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2514
raise AssertionError('could not find entry for %s' % (update_key,))
2515
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2516
# it must not be absent at the moment
2517
if update_tree_details[0][0] == 'a': # absent
2518
raise AssertionError('bad row %r' % (update_tree_details,))
2519
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2520
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2521
return last_reference
2523
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2524
packed_stat=None, size=0, path_utf8=None):
2525
"""Update an entry to the state in tree 0.
2527
This will either create a new entry at 'key' or update an existing one.
2528
It also makes sure that any other records which might mention this are
2531
:param key: (dir, name, file_id) for the new entry
2532
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2534
:param executable: Should the executable bit be set?
2535
:param fingerprint: Simple fingerprint for new entry: canonical-form
2536
sha1 for files, referenced revision id for subtrees, etc.
2537
:param packed_stat: Packed stat value for new entry.
2538
:param size: Size information for new entry
2539
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2542
If packed_stat and fingerprint are not given, they're invalidated in
2545
block = self._find_block(key)[1]
2546
if packed_stat is None:
2547
packed_stat = DirState.NULLSTAT
2548
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2549
# sometimes present in the dirstate - this seems oddly inconsistent.
2551
entry_index, present = self._find_entry_index(key, block)
2552
new_details = (minikind, fingerprint, size, executable, packed_stat)
2553
id_index = self._get_id_index()
2555
# new entry, synthesis cross reference here,
2556
existing_keys = id_index.setdefault(key[2], set())
2557
if not existing_keys:
2558
# not currently in the state, simplest case
2559
new_entry = key, [new_details] + self._empty_parent_info()
2561
# present at one or more existing other paths.
2562
# grab one of them and use it to generate parent
2563
# relocation/absent entries.
2564
new_entry = key, [new_details]
2565
for other_key in existing_keys:
2566
# change the record at other to be a pointer to this new
2567
# record. The loop looks similar to the change to
2568
# relocations when updating an existing record but its not:
2569
# the test for existing kinds is different: this can be
2570
# factored out to a helper though.
2571
other_block_index, present = self._find_block_index_from_key(other_key)
2573
raise AssertionError('could not find block for %s' % (other_key,))
2574
other_entry_index, present = self._find_entry_index(other_key,
2575
self._dirblocks[other_block_index][1])
2577
raise AssertionError('could not find entry for %s' % (other_key,))
2578
if path_utf8 is None:
2579
raise AssertionError('no path')
2580
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2581
('r', path_utf8, 0, False, '')
2583
num_present_parents = self._num_present_parents()
2584
for lookup_index in xrange(1, num_present_parents + 1):
2585
# grab any one entry, use it to find the right path.
2586
# TODO: optimise this to reduce memory use in highly
2587
# fragmented situations by reusing the relocation
2589
update_block_index, present = \
2590
self._find_block_index_from_key(other_key)
2592
raise AssertionError('could not find block for %s' % (other_key,))
2593
update_entry_index, present = \
2594
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2596
raise AssertionError('could not find entry for %s' % (other_key,))
2597
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2598
if update_details[0] in 'ar': # relocated, absent
2599
# its a pointer or absent in lookup_index's tree, use
2601
new_entry[1].append(update_details)
2603
# we have the right key, make a pointer to it.
2604
pointer_path = osutils.pathjoin(*other_key[0:2])
2605
new_entry[1].append(('r', pointer_path, 0, False, ''))
2606
block.insert(entry_index, new_entry)
2607
existing_keys.add(key)
2609
# Does the new state matter?
2610
block[entry_index][1][0] = new_details
2611
# parents cannot be affected by what we do.
2612
# other occurences of this id can be found
2613
# from the id index.
2615
# tree index consistency: All other paths for this id in this tree
2616
# index must point to the correct path. We have to loop here because
2617
# we may have passed entries in the state with this file id already
2618
# that were absent - where parent entries are - and they need to be
2619
# converted to relocated.
2620
if path_utf8 is None:
2621
raise AssertionError('no path')
2622
for entry_key in id_index.setdefault(key[2], set()):
2623
# TODO:PROFILING: It might be faster to just update
2624
# rather than checking if we need to, and then overwrite
2625
# the one we are located at.
2626
if entry_key != key:
2627
# this file id is at a different path in one of the
2628
# other trees, so put absent pointers there
2629
# This is the vertical axis in the matrix, all pointing
2631
block_index, present = self._find_block_index_from_key(entry_key)
2633
raise AssertionError('not present: %r', entry_key)
2634
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2636
raise AssertionError('not present: %r', entry_key)
2637
self._dirblocks[block_index][1][entry_index][1][0] = \
2638
('r', path_utf8, 0, False, '')
2639
# add a containing dirblock if needed.
2640
if new_details[0] == 'd':
2641
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2642
block_index, present = self._find_block_index_from_key(subdir_key)
2644
self._dirblocks.insert(block_index, (subdir_key[0], []))
2646
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2648
def _validate(self):
2649
"""Check that invariants on the dirblock are correct.
2651
This can be useful in debugging; it shouldn't be necessary in
2654
This must be called with a lock held.
2656
# NOTE: This must always raise AssertionError not just assert,
2657
# otherwise it may not behave properly under python -O
2659
# TODO: All entries must have some content that's not 'a' or 'r',
2660
# otherwise it could just be removed.
2662
# TODO: All relocations must point directly to a real entry.
2664
# TODO: No repeated keys.
2667
from pprint import pformat
2668
self._read_dirblocks_if_needed()
2669
if len(self._dirblocks) > 0:
2670
if not self._dirblocks[0][0] == '':
2671
raise AssertionError(
2672
"dirblocks don't start with root block:\n" + \
2673
pformat(self._dirblocks))
2674
if len(self._dirblocks) > 1:
2675
if not self._dirblocks[1][0] == '':
2676
raise AssertionError(
2677
"dirblocks missing root directory:\n" + \
2678
pformat(self._dirblocks))
2679
# the dirblocks are sorted by their path components, name, and dir id
2680
dir_names = [d[0].split('/')
2681
for d in self._dirblocks[1:]]
2682
if dir_names != sorted(dir_names):
2683
raise AssertionError(
2684
"dir names are not in sorted order:\n" + \
2685
pformat(self._dirblocks) + \
2688
for dirblock in self._dirblocks:
2689
# within each dirblock, the entries are sorted by filename and
2691
for entry in dirblock[1]:
2692
if dirblock[0] != entry[0][0]:
2693
raise AssertionError(
2695
"doesn't match directory name in\n%r" %
2696
(entry, pformat(dirblock)))
2697
if dirblock[1] != sorted(dirblock[1]):
2698
raise AssertionError(
2699
"dirblock for %r is not sorted:\n%s" % \
2700
(dirblock[0], pformat(dirblock)))
2702
def check_valid_parent():
2703
"""Check that the current entry has a valid parent.
2705
This makes sure that the parent has a record,
2706
and that the parent isn't marked as "absent" in the
2707
current tree. (It is invalid to have a non-absent file in an absent
2710
if entry[0][0:2] == ('', ''):
2711
# There should be no parent for the root row
2713
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2714
if parent_entry == (None, None):
2715
raise AssertionError(
2716
"no parent entry for: %s in tree %s"
2717
% (this_path, tree_index))
2718
if parent_entry[1][tree_index][0] != 'd':
2719
raise AssertionError(
2720
"Parent entry for %s is not marked as a valid"
2721
" directory. %s" % (this_path, parent_entry,))
2723
# For each file id, for each tree: either
2724
# the file id is not present at all; all rows with that id in the
2725
# key have it marked as 'absent'
2726
# OR the file id is present under exactly one name; any other entries
2727
# that mention that id point to the correct name.
2729
# We check this with a dict per tree pointing either to the present
2730
# name, or None if absent.
2731
tree_count = self._num_present_parents() + 1
2732
id_path_maps = [dict() for i in range(tree_count)]
2733
# Make sure that all renamed entries point to the correct location.
2734
for entry in self._iter_entries():
2735
file_id = entry[0][2]
2736
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2737
if len(entry[1]) != tree_count:
2738
raise AssertionError(
2739
"wrong number of entry details for row\n%s" \
2740
",\nexpected %d" % \
2741
(pformat(entry), tree_count))
2742
absent_positions = 0
2743
for tree_index, tree_state in enumerate(entry[1]):
2744
this_tree_map = id_path_maps[tree_index]
2745
minikind = tree_state[0]
2746
if minikind in 'ar':
2747
absent_positions += 1
2748
# have we seen this id before in this column?
2749
if file_id in this_tree_map:
2750
previous_path, previous_loc = this_tree_map[file_id]
2751
# any later mention of this file must be consistent with
2752
# what was said before
2754
if previous_path is not None:
2755
raise AssertionError(
2756
"file %s is absent in row %r but also present " \
2758
(file_id, entry, previous_path))
2759
elif minikind == 'r':
2760
target_location = tree_state[1]
2761
if previous_path != target_location:
2762
raise AssertionError(
2763
"file %s relocation in row %r but also at %r" \
2764
% (file_id, entry, previous_path))
2766
# a file, directory, etc - may have been previously
2767
# pointed to by a relocation, which must point here
2768
if previous_path != this_path:
2769
raise AssertionError(
2770
"entry %r inconsistent with previous path %r "
2772
(entry, previous_path, previous_loc))
2773
check_valid_parent()
2776
# absent; should not occur anywhere else
2777
this_tree_map[file_id] = None, this_path
2778
elif minikind == 'r':
2779
# relocation, must occur at expected location
2780
this_tree_map[file_id] = tree_state[1], this_path
2782
this_tree_map[file_id] = this_path, this_path
2783
check_valid_parent()
2784
if absent_positions == tree_count:
2785
raise AssertionError(
2786
"entry %r has no data for any tree." % (entry,))
2788
def _wipe_state(self):
2789
"""Forget all state information about the dirstate."""
2790
self._header_state = DirState.NOT_IN_MEMORY
2791
self._dirblock_state = DirState.NOT_IN_MEMORY
2792
self._changes_aborted = False
2795
self._dirblocks = []
2796
self._id_index = None
2797
self._packed_stat_index = None
2798
self._end_of_header = None
2799
self._cutoff_time = None
2800
self._split_path_cache = {}
2802
def lock_read(self):
2803
"""Acquire a read lock on the dirstate."""
2804
if self._lock_token is not None:
2805
raise errors.LockContention(self._lock_token)
2806
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2807
# already in memory, we could read just the header and check for
2808
# any modification. If not modified, we can just leave things
2810
self._lock_token = lock.ReadLock(self._filename)
2811
self._lock_state = 'r'
2812
self._state_file = self._lock_token.f
2815
def lock_write(self):
2816
"""Acquire a write lock on the dirstate."""
2817
if self._lock_token is not None:
2818
raise errors.LockContention(self._lock_token)
2819
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2820
# already in memory, we could read just the header and check for
2821
# any modification. If not modified, we can just leave things
2823
self._lock_token = lock.WriteLock(self._filename)
2824
self._lock_state = 'w'
2825
self._state_file = self._lock_token.f
2829
"""Drop any locks held on the dirstate."""
2830
if self._lock_token is None:
2831
raise errors.LockNotHeld(self)
2832
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2833
# already in memory, we could read just the header and check for
2834
# any modification. If not modified, we can just leave things
2836
self._state_file = None
2837
self._lock_state = None
2838
self._lock_token.unlock()
2839
self._lock_token = None
2840
self._split_path_cache = {}
2842
def _requires_lock(self):
2843
"""Check that a lock is currently held by someone on the dirstate."""
2844
if not self._lock_token:
2845
raise errors.ObjectNotLocked(self)
2848
def py_update_entry(state, entry, abspath, stat_value,
2849
_stat_to_minikind=DirState._stat_to_minikind,
2850
_pack_stat=pack_stat):
2851
"""Update the entry based on what is actually on disk.
2853
This function only calculates the sha if it needs to - if the entry is
2854
uncachable, or clearly different to the first parent's entry, no sha
2855
is calculated, and None is returned.
2857
:param state: The dirstate this entry is in.
2858
:param entry: This is the dirblock entry for the file in question.
2859
:param abspath: The path on disk for this file.
2860
:param stat_value: The stat value done on the path.
2861
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
2862
target of a symlink.
2865
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
2869
packed_stat = _pack_stat(stat_value)
2870
(saved_minikind, saved_link_or_sha1, saved_file_size,
2871
saved_executable, saved_packed_stat) = entry[1][0]
2873
if minikind == 'd' and saved_minikind == 't':
2875
if (minikind == saved_minikind
2876
and packed_stat == saved_packed_stat):
2877
# The stat hasn't changed since we saved, so we can re-use the
2882
# size should also be in packed_stat
2883
if saved_file_size == stat_value.st_size:
2884
return saved_link_or_sha1
2886
# If we have gotten this far, that means that we need to actually
2887
# process this entry.
2890
executable = state._is_executable(stat_value.st_mode,
2892
if state._cutoff_time is None:
2893
state._sha_cutoff_time()
2894
if (stat_value.st_mtime < state._cutoff_time
2895
and stat_value.st_ctime < state._cutoff_time
2896
and len(entry[1]) > 1
2897
and entry[1][1][0] != 'a'):
2898
# Could check for size changes for further optimised
2899
# avoidance of sha1's. However the most prominent case of
2900
# over-shaing is during initial add, which this catches.
2901
# Besides, if content filtering happens, size and sha
2902
# are calculated at the same time, so checking just the size
2903
# gains nothing w.r.t. performance.
2904
link_or_sha1 = state._sha1_file(abspath)
2905
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
2906
executable, packed_stat)
2908
entry[1][0] = ('f', '', stat_value.st_size,
2909
executable, DirState.NULLSTAT)
2910
elif minikind == 'd':
2912
entry[1][0] = ('d', '', 0, False, packed_stat)
2913
if saved_minikind != 'd':
2914
# This changed from something into a directory. Make sure we
2915
# have a directory block for it. This doesn't happen very
2916
# often, so this doesn't have to be super fast.
2917
block_index, entry_index, dir_present, file_present = \
2918
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
2919
state._ensure_block(block_index, entry_index,
2920
osutils.pathjoin(entry[0][0], entry[0][1]))
2921
elif minikind == 'l':
2922
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
2923
if state._cutoff_time is None:
2924
state._sha_cutoff_time()
2925
if (stat_value.st_mtime < state._cutoff_time
2926
and stat_value.st_ctime < state._cutoff_time):
2927
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
2930
entry[1][0] = ('l', '', stat_value.st_size,
2931
False, DirState.NULLSTAT)
2932
state._dirblock_state = DirState.IN_MEMORY_MODIFIED
2934
update_entry = py_update_entry
2937
class ProcessEntryPython(object):
2939
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id", "uninteresting",
2940
"last_source_parent", "last_target_parent", "include_unchanged",
2941
"use_filesystem_for_exec", "utf8_decode", "searched_specific_files",
2942
"search_specific_files", "state", "source_index", "target_index",
2943
"want_unversioned", "tree"]
2945
def __init__(self, include_unchanged, use_filesystem_for_exec,
2946
search_specific_files, state, source_index, target_index,
2947
want_unversioned, tree):
2948
self.old_dirname_to_file_id = {}
2949
self.new_dirname_to_file_id = {}
2950
# Just a sentry, so that _process_entry can say that this
2951
# record is handled, but isn't interesting to process (unchanged)
2952
self.uninteresting = object()
2953
# Using a list so that we can access the values and change them in
2954
# nested scope. Each one is [path, file_id, entry]
2955
self.last_source_parent = [None, None]
2956
self.last_target_parent = [None, None]
2957
self.include_unchanged = include_unchanged
2958
self.use_filesystem_for_exec = use_filesystem_for_exec
2959
self.utf8_decode = cache_utf8._utf8_decode
2960
# for all search_indexs in each path at or under each element of
2961
# search_specific_files, if the detail is relocated: add the id, and add the
2962
# relocated path as one to search if its not searched already. If the
2963
# detail is not relocated, add the id.
2964
self.searched_specific_files = set()
2965
self.search_specific_files = search_specific_files
2967
self.source_index = source_index
2968
self.target_index = target_index
2969
self.want_unversioned = want_unversioned
2972
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
2973
"""Compare an entry and real disk to generate delta information.
2975
:param path_info: top_relpath, basename, kind, lstat, abspath for
2976
the path of entry. If None, then the path is considered absent.
2977
(Perhaps we should pass in a concrete entry for this ?)
2978
Basename is returned as a utf8 string because we expect this
2979
tuple will be ignored, and don't want to take the time to
2981
:return: None if these don't match
2982
A tuple of information about the change, or
2983
the object 'uninteresting' if these match, but are
2984
basically identical.
2986
if self.source_index is None:
2987
source_details = DirState.NULL_PARENT_DETAILS
2989
source_details = entry[1][self.source_index]
2990
target_details = entry[1][self.target_index]
2991
target_minikind = target_details[0]
2992
if path_info is not None and target_minikind in 'fdlt':
2993
if not (self.target_index == 0):
2994
raise AssertionError()
2995
link_or_sha1 = update_entry(self.state, entry,
2996
abspath=path_info[4], stat_value=path_info[3])
2997
# The entry may have been modified by update_entry
2998
target_details = entry[1][self.target_index]
2999
target_minikind = target_details[0]
3002
file_id = entry[0][2]
3003
source_minikind = source_details[0]
3004
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
3005
# claimed content in both: diff
3006
# r | fdlt | | add source to search, add id path move and perform
3007
# | | | diff check on source-target
3008
# r | fdlt | a | dangling file that was present in the basis.
3010
if source_minikind in 'r':
3011
# add the source to the search path to find any children it
3012
# has. TODO ? : only add if it is a container ?
3013
if not osutils.is_inside_any(self.searched_specific_files,
3015
self.search_specific_files.add(source_details[1])
3016
# generate the old path; this is needed for stating later
3018
old_path = source_details[1]
3019
old_dirname, old_basename = os.path.split(old_path)
3020
path = pathjoin(entry[0][0], entry[0][1])
3021
old_entry = self.state._get_entry(self.source_index,
3023
# update the source details variable to be the real
3025
if old_entry == (None, None):
3026
raise errors.CorruptDirstate(self.state._filename,
3027
"entry '%s/%s' is considered renamed from %r"
3028
" but source does not exist\n"
3029
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
3030
source_details = old_entry[1][self.source_index]
3031
source_minikind = source_details[0]
3033
old_dirname = entry[0][0]
3034
old_basename = entry[0][1]
3035
old_path = path = None
3036
if path_info is None:
3037
# the file is missing on disk, show as removed.
3038
content_change = True
3042
# source and target are both versioned and disk file is present.
3043
target_kind = path_info[2]
3044
if target_kind == 'directory':
3046
old_path = path = pathjoin(old_dirname, old_basename)
3047
self.new_dirname_to_file_id[path] = file_id
3048
if source_minikind != 'd':
3049
content_change = True
3051
# directories have no fingerprint
3052
content_change = False
3054
elif target_kind == 'file':
3055
if source_minikind != 'f':
3056
content_change = True
3058
# If the size is the same, check the sha:
3059
if target_details[2] == source_details[2]:
3060
if link_or_sha1 is None:
3062
statvalue, link_or_sha1 = \
3063
self.state._sha1_provider.stat_and_sha1(
3065
self.state._observed_sha1(entry, link_or_sha1,
3067
content_change = (link_or_sha1 != source_details[1])
3069
# Size changed, so must be different
3070
content_change = True
3071
# Target details is updated at update_entry time
3072
if self.use_filesystem_for_exec:
3073
# We don't need S_ISREG here, because we are sure
3074
# we are dealing with a file.
3075
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
3077
target_exec = target_details[3]
3078
elif target_kind == 'symlink':
3079
if source_minikind != 'l':
3080
content_change = True
3082
content_change = (link_or_sha1 != source_details[1])
3084
elif target_kind == 'tree-reference':
3085
if source_minikind != 't':
3086
content_change = True
3088
content_change = False
3091
raise Exception, "unknown kind %s" % path_info[2]
3092
if source_minikind == 'd':
3094
old_path = path = pathjoin(old_dirname, old_basename)
3095
self.old_dirname_to_file_id[old_path] = file_id
3096
# parent id is the entry for the path in the target tree
3097
if old_dirname == self.last_source_parent[0]:
3098
source_parent_id = self.last_source_parent[1]
3101
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3103
source_parent_entry = self.state._get_entry(self.source_index,
3104
path_utf8=old_dirname)
3105
source_parent_id = source_parent_entry[0][2]
3106
if source_parent_id == entry[0][2]:
3107
# This is the root, so the parent is None
3108
source_parent_id = None
3110
self.last_source_parent[0] = old_dirname
3111
self.last_source_parent[1] = source_parent_id
3112
new_dirname = entry[0][0]
3113
if new_dirname == self.last_target_parent[0]:
3114
target_parent_id = self.last_target_parent[1]
3117
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3119
# TODO: We don't always need to do the lookup, because the
3120
# parent entry will be the same as the source entry.
3121
target_parent_entry = self.state._get_entry(self.target_index,
3122
path_utf8=new_dirname)
3123
if target_parent_entry == (None, None):
3124
raise AssertionError(
3125
"Could not find target parent in wt: %s\nparent of: %s"
3126
% (new_dirname, entry))
3127
target_parent_id = target_parent_entry[0][2]
3128
if target_parent_id == entry[0][2]:
3129
# This is the root, so the parent is None
3130
target_parent_id = None
3132
self.last_target_parent[0] = new_dirname
3133
self.last_target_parent[1] = target_parent_id
3135
source_exec = source_details[3]
3136
if (self.include_unchanged
3138
or source_parent_id != target_parent_id
3139
or old_basename != entry[0][1]
3140
or source_exec != target_exec
3142
if old_path is None:
3143
old_path = path = pathjoin(old_dirname, old_basename)
3144
old_path_u = self.utf8_decode(old_path)[0]
3147
old_path_u = self.utf8_decode(old_path)[0]
3148
if old_path == path:
3151
path_u = self.utf8_decode(path)[0]
3152
source_kind = DirState._minikind_to_kind[source_minikind]
3153
return (entry[0][2],
3154
(old_path_u, path_u),
3157
(source_parent_id, target_parent_id),
3158
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3159
(source_kind, target_kind),
3160
(source_exec, target_exec))
3162
return self.uninteresting
3163
elif source_minikind in 'a' and target_minikind in 'fdlt':
3164
# looks like a new file
3165
path = pathjoin(entry[0][0], entry[0][1])
3166
# parent id is the entry for the path in the target tree
3167
# TODO: these are the same for an entire directory: cache em.
3168
parent_id = self.state._get_entry(self.target_index,
3169
path_utf8=entry[0][0])[0][2]
3170
if parent_id == entry[0][2]:
3172
if path_info is not None:
3174
if self.use_filesystem_for_exec:
3175
# We need S_ISREG here, because we aren't sure if this
3178
stat.S_ISREG(path_info[3].st_mode)
3179
and stat.S_IEXEC & path_info[3].st_mode)
3181
target_exec = target_details[3]
3182
return (entry[0][2],
3183
(None, self.utf8_decode(path)[0]),
3187
(None, self.utf8_decode(entry[0][1])[0]),
3188
(None, path_info[2]),
3189
(None, target_exec))
3191
# Its a missing file, report it as such.
3192
return (entry[0][2],
3193
(None, self.utf8_decode(path)[0]),
3197
(None, self.utf8_decode(entry[0][1])[0]),
3200
elif source_minikind in 'fdlt' and target_minikind in 'a':
3201
# unversioned, possibly, or possibly not deleted: we dont care.
3202
# if its still on disk, *and* theres no other entry at this
3203
# path [we dont know this in this routine at the moment -
3204
# perhaps we should change this - then it would be an unknown.
3205
old_path = pathjoin(entry[0][0], entry[0][1])
3206
# parent id is the entry for the path in the target tree
3207
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3208
if parent_id == entry[0][2]:
3210
return (entry[0][2],
3211
(self.utf8_decode(old_path)[0], None),
3215
(self.utf8_decode(entry[0][1])[0], None),
3216
(DirState._minikind_to_kind[source_minikind], None),
3217
(source_details[3], None))
3218
elif source_minikind in 'fdlt' and target_minikind in 'r':
3219
# a rename; could be a true rename, or a rename inherited from
3220
# a renamed parent. TODO: handle this efficiently. Its not
3221
# common case to rename dirs though, so a correct but slow
3222
# implementation will do.
3223
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3224
self.search_specific_files.add(target_details[1])
3225
elif source_minikind in 'ra' and target_minikind in 'ra':
3226
# neither of the selected trees contain this file,
3227
# so skip over it. This is not currently directly tested, but
3228
# is indirectly via test_too_much.TestCommands.test_conflicts.
3231
raise AssertionError("don't know how to compare "
3232
"source_minikind=%r, target_minikind=%r"
3233
% (source_minikind, target_minikind))
3234
## import pdb;pdb.set_trace()
3240
def iter_changes(self):
3241
"""Iterate over the changes."""
3242
utf8_decode = cache_utf8._utf8_decode
3243
_cmp_by_dirs = cmp_by_dirs
3244
_process_entry = self._process_entry
3245
uninteresting = self.uninteresting
3246
search_specific_files = self.search_specific_files
3247
searched_specific_files = self.searched_specific_files
3248
splitpath = osutils.splitpath
3250
# compare source_index and target_index at or under each element of search_specific_files.
3251
# follow the following comparison table. Note that we only want to do diff operations when
3252
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3256
# Source | Target | disk | action
3257
# r | fdlt | | add source to search, add id path move and perform
3258
# | | | diff check on source-target
3259
# r | fdlt | a | dangling file that was present in the basis.
3261
# r | a | | add source to search
3263
# r | r | | this path is present in a non-examined tree, skip.
3264
# r | r | a | this path is present in a non-examined tree, skip.
3265
# a | fdlt | | add new id
3266
# a | fdlt | a | dangling locally added file, skip
3267
# a | a | | not present in either tree, skip
3268
# a | a | a | not present in any tree, skip
3269
# a | r | | not present in either tree at this path, skip as it
3270
# | | | may not be selected by the users list of paths.
3271
# a | r | a | not present in either tree at this path, skip as it
3272
# | | | may not be selected by the users list of paths.
3273
# fdlt | fdlt | | content in both: diff them
3274
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3275
# fdlt | a | | unversioned: output deleted id for now
3276
# fdlt | a | a | unversioned and deleted: output deleted id
3277
# fdlt | r | | relocated in this tree, so add target to search.
3278
# | | | Dont diff, we will see an r,fd; pair when we reach
3279
# | | | this id at the other path.
3280
# fdlt | r | a | relocated in this tree, so add target to search.
3281
# | | | Dont diff, we will see an r,fd; pair when we reach
3282
# | | | this id at the other path.
3284
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3285
# keeping a cache of directories that we have seen.
3287
while search_specific_files:
3288
# TODO: the pending list should be lexically sorted? the
3289
# interface doesn't require it.
3290
current_root = search_specific_files.pop()
3291
current_root_unicode = current_root.decode('utf8')
3292
searched_specific_files.add(current_root)
3293
# process the entries for this containing directory: the rest will be
3294
# found by their parents recursively.
3295
root_entries = self.state._entries_for_path(current_root)
3296
root_abspath = self.tree.abspath(current_root_unicode)
3298
root_stat = os.lstat(root_abspath)
3300
if e.errno == errno.ENOENT:
3301
# the path does not exist: let _process_entry know that.
3302
root_dir_info = None
3304
# some other random error: hand it up.
3307
root_dir_info = ('', current_root,
3308
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3310
if root_dir_info[2] == 'directory':
3311
if self.tree._directory_is_tree_reference(
3312
current_root.decode('utf8')):
3313
root_dir_info = root_dir_info[:2] + \
3314
('tree-reference',) + root_dir_info[3:]
3316
if not root_entries and not root_dir_info:
3317
# this specified path is not present at all, skip it.
3319
path_handled = False
3320
for entry in root_entries:
3321
result = _process_entry(entry, root_dir_info)
3322
if result is not None:
3324
if result is not uninteresting:
3326
if self.want_unversioned and not path_handled and root_dir_info:
3327
new_executable = bool(
3328
stat.S_ISREG(root_dir_info[3].st_mode)
3329
and stat.S_IEXEC & root_dir_info[3].st_mode)
3331
(None, current_root_unicode),
3335
(None, splitpath(current_root_unicode)[-1]),
3336
(None, root_dir_info[2]),
3337
(None, new_executable)
3339
initial_key = (current_root, '', '')
3340
block_index, _ = self.state._find_block_index_from_key(initial_key)
3341
if block_index == 0:
3342
# we have processed the total root already, but because the
3343
# initial key matched it we should skip it here.
3345
if root_dir_info and root_dir_info[2] == 'tree-reference':
3346
current_dir_info = None
3348
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3350
current_dir_info = dir_iterator.next()
3352
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3353
# python 2.5 has e.errno == EINVAL,
3354
# and e.winerror == ERROR_DIRECTORY
3355
e_winerror = getattr(e, 'winerror', None)
3356
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3357
# there may be directories in the inventory even though
3358
# this path is not a file on disk: so mark it as end of
3360
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3361
current_dir_info = None
3362
elif (sys.platform == 'win32'
3363
and (e.errno in win_errors
3364
or e_winerror in win_errors)):
3365
current_dir_info = None
3369
if current_dir_info[0][0] == '':
3370
# remove .bzr from iteration
3371
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3372
if current_dir_info[1][bzr_index][0] != '.bzr':
3373
raise AssertionError()
3374
del current_dir_info[1][bzr_index]
3375
# walk until both the directory listing and the versioned metadata
3377
if (block_index < len(self.state._dirblocks) and
3378
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3379
current_block = self.state._dirblocks[block_index]
3381
current_block = None
3382
while (current_dir_info is not None or
3383
current_block is not None):
3384
if (current_dir_info and current_block
3385
and current_dir_info[0][0] != current_block[0]):
3386
if _cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
3387
# filesystem data refers to paths not covered by the dirblock.
3388
# this has two possibilities:
3389
# A) it is versioned but empty, so there is no block for it
3390
# B) it is not versioned.
3392
# if (A) then we need to recurse into it to check for
3393
# new unknown files or directories.
3394
# if (B) then we should ignore it, because we don't
3395
# recurse into unknown directories.
3397
while path_index < len(current_dir_info[1]):
3398
current_path_info = current_dir_info[1][path_index]
3399
if self.want_unversioned:
3400
if current_path_info[2] == 'directory':
3401
if self.tree._directory_is_tree_reference(
3402
current_path_info[0].decode('utf8')):
3403
current_path_info = current_path_info[:2] + \
3404
('tree-reference',) + current_path_info[3:]
3405
new_executable = bool(
3406
stat.S_ISREG(current_path_info[3].st_mode)
3407
and stat.S_IEXEC & current_path_info[3].st_mode)
3409
(None, utf8_decode(current_path_info[0])[0]),
3413
(None, utf8_decode(current_path_info[1])[0]),
3414
(None, current_path_info[2]),
3415
(None, new_executable))
3416
# dont descend into this unversioned path if it is
3418
if current_path_info[2] in ('directory',
3420
del current_dir_info[1][path_index]
3424
# This dir info has been handled, go to the next
3426
current_dir_info = dir_iterator.next()
3427
except StopIteration:
3428
current_dir_info = None
3430
# We have a dirblock entry for this location, but there
3431
# is no filesystem path for this. This is most likely
3432
# because a directory was removed from the disk.
3433
# We don't have to report the missing directory,
3434
# because that should have already been handled, but we
3435
# need to handle all of the files that are contained
3437
for current_entry in current_block[1]:
3438
# entry referring to file not present on disk.
3439
# advance the entry only, after processing.
3440
result = _process_entry(current_entry, None)
3441
if result is not None:
3442
if result is not uninteresting:
3445
if (block_index < len(self.state._dirblocks) and
3446
osutils.is_inside(current_root,
3447
self.state._dirblocks[block_index][0])):
3448
current_block = self.state._dirblocks[block_index]
3450
current_block = None
3453
if current_block and entry_index < len(current_block[1]):
3454
current_entry = current_block[1][entry_index]
3456
current_entry = None
3457
advance_entry = True
3459
if current_dir_info and path_index < len(current_dir_info[1]):
3460
current_path_info = current_dir_info[1][path_index]
3461
if current_path_info[2] == 'directory':
3462
if self.tree._directory_is_tree_reference(
3463
current_path_info[0].decode('utf8')):
3464
current_path_info = current_path_info[:2] + \
3465
('tree-reference',) + current_path_info[3:]
3467
current_path_info = None
3469
path_handled = False
3470
while (current_entry is not None or
3471
current_path_info is not None):
3472
if current_entry is None:
3473
# the check for path_handled when the path is advanced
3474
# will yield this path if needed.
3476
elif current_path_info is None:
3477
# no path is fine: the per entry code will handle it.
3478
result = _process_entry(current_entry, current_path_info)
3479
if result is not None:
3480
if result is not uninteresting:
3482
elif (current_entry[0][1] != current_path_info[1]
3483
or current_entry[1][self.target_index][0] in 'ar'):
3484
# The current path on disk doesn't match the dirblock
3485
# record. Either the dirblock is marked as absent, or
3486
# the file on disk is not present at all in the
3487
# dirblock. Either way, report about the dirblock
3488
# entry, and let other code handle the filesystem one.
3490
# Compare the basename for these files to determine
3492
if current_path_info[1] < current_entry[0][1]:
3493
# extra file on disk: pass for now, but only
3494
# increment the path, not the entry
3495
advance_entry = False
3497
# entry referring to file not present on disk.
3498
# advance the entry only, after processing.
3499
result = _process_entry(current_entry, None)
3500
if result is not None:
3501
if result is not uninteresting:
3503
advance_path = False
3505
result = _process_entry(current_entry, current_path_info)
3506
if result is not None:
3508
if result is not uninteresting:
3510
if advance_entry and current_entry is not None:
3512
if entry_index < len(current_block[1]):
3513
current_entry = current_block[1][entry_index]
3515
current_entry = None
3517
advance_entry = True # reset the advance flaga
3518
if advance_path and current_path_info is not None:
3519
if not path_handled:
3520
# unversioned in all regards
3521
if self.want_unversioned:
3522
new_executable = bool(
3523
stat.S_ISREG(current_path_info[3].st_mode)
3524
and stat.S_IEXEC & current_path_info[3].st_mode)
3526
relpath_unicode = utf8_decode(current_path_info[0])[0]
3527
except UnicodeDecodeError:
3528
raise errors.BadFilenameEncoding(
3529
current_path_info[0], osutils._fs_enc)
3531
(None, relpath_unicode),
3535
(None, utf8_decode(current_path_info[1])[0]),
3536
(None, current_path_info[2]),
3537
(None, new_executable))
3538
# dont descend into this unversioned path if it is
3540
if current_path_info[2] in ('directory'):
3541
del current_dir_info[1][path_index]
3543
# dont descend the disk iterator into any tree
3545
if current_path_info[2] == 'tree-reference':
3546
del current_dir_info[1][path_index]
3549
if path_index < len(current_dir_info[1]):
3550
current_path_info = current_dir_info[1][path_index]
3551
if current_path_info[2] == 'directory':
3552
if self.tree._directory_is_tree_reference(
3553
current_path_info[0].decode('utf8')):
3554
current_path_info = current_path_info[:2] + \
3555
('tree-reference',) + current_path_info[3:]
3557
current_path_info = None
3558
path_handled = False
3560
advance_path = True # reset the advance flagg.
3561
if current_block is not None:
3563
if (block_index < len(self.state._dirblocks) and
3564
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3565
current_block = self.state._dirblocks[block_index]
3567
current_block = None
3568
if current_dir_info is not None:
3570
current_dir_info = dir_iterator.next()
3571
except StopIteration:
3572
current_dir_info = None
3573
_process_entry = ProcessEntryPython
3576
# Try to load the compiled form if possible
3578
from bzrlib._dirstate_helpers_c import (
3579
_read_dirblocks_c as _read_dirblocks,
3580
bisect_dirblock_c as bisect_dirblock,
3581
_bisect_path_left_c as _bisect_path_left,
3582
_bisect_path_right_c as _bisect_path_right,
3583
cmp_by_dirs_c as cmp_by_dirs,
3584
ProcessEntryC as _process_entry,
3585
update_entry as update_entry,
3588
from bzrlib._dirstate_helpers_py import (
3589
_read_dirblocks_py as _read_dirblocks,
3590
bisect_dirblock_py as bisect_dirblock,
3591
_bisect_path_left_py as _bisect_path_left,
3592
_bisect_path_right_py as _bisect_path_right,
3593
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
