1
# Copyright (C) 2006-2011 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but it's a file. The fingerprint is the
86
sha1 value of the file's canonical form, i.e. after any read filters have
87
been applied to the convenience form stored in the working tree.
88
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
90
't' is a reference to a nested subtree; the fingerprint is the referenced
95
The entries on disk and in memory are ordered according to the following keys:
97
directory, as a list of components
101
--- Format 1 had the following different definition: ---
102
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
103
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
105
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
106
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
109
PARENT ROW's are emitted for every parent that is not in the ghosts details
110
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
111
each row will have a PARENT ROW for foo and baz, but not for bar.
114
In any tree, a kind of 'moved' indicates that the fingerprint field
115
(which we treat as opaque data specific to the 'kind' anyway) has the
116
details for the id of this row in that tree.
118
I'm strongly tempted to add a id->path index as well, but I think that
119
where we need id->path mapping; we also usually read the whole file, so
120
I'm going to skip that for the moment, as we have the ability to locate
121
via bisect any path in any tree, and if we lookup things by path, we can
122
accumulate an id->path mapping as we go, which will tend to match what we
125
I plan to implement this asap, so please speak up now to alter/tweak the
126
design - and once we stabilise on this, I'll update the wiki page for
129
The rationale for all this is that we want fast operations for the
130
common case (diff/status/commit/merge on all files) and extremely fast
131
operations for the less common but still occurs a lot status/diff/commit
132
on specific files). Operations on specific files involve a scan for all
133
the children of a path, *in every involved tree*, which the current
134
format did not accommodate.
138
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
139
2) fall back current object model as needed.
140
3) scale usably to the largest trees known today - say 50K entries. (mozilla
141
is an example of this)
145
Eventually reuse dirstate objects across locks IFF the dirstate file has not
146
been modified, but will require that we flush/ignore cached stat-hit data
147
because we won't want to restat all files on disk just because a lock was
148
acquired, yet we cannot trust the data after the previous lock was released.
150
Memory representation:
151
vector of all directories, and vector of the childen ?
153
root_entrie = (direntry for root, [parent_direntries_for_root]),
155
('', ['data for achild', 'data for bchild', 'data for cchild'])
156
('dir', ['achild', 'cchild', 'echild'])
158
- single bisect to find N subtrees from a path spec
159
- in-order for serialisation - this is 'dirblock' grouping.
160
- insertion of a file '/a' affects only the '/' child-vector, that is, to
161
insert 10K elements from scratch does not generates O(N^2) memoves of a
162
single vector, rather each individual, which tends to be limited to a
163
manageable number. Will scale badly on trees with 10K entries in a
164
single directory. compare with Inventory.InventoryDirectory which has
165
a dictionary for the children. No bisect capability, can only probe for
166
exact matches, or grab all elements and sort.
167
- What's the risk of error here? Once we have the base format being processed
168
we should have a net win regardless of optimality. So we are going to
169
go with what seems reasonable.
172
Maybe we should do a test profile of the core structure - 10K simulated
173
searches/lookups/etc?
175
Objects for each row?
176
The lifetime of Dirstate objects is current per lock, but see above for
177
possible extensions. The lifetime of a row from a dirstate is expected to be
178
very short in the optimistic case: which we are optimising for. For instance,
179
subtree status will determine from analysis of the disk data what rows need to
180
be examined at all, and will be able to determine from a single row whether
181
that file has altered or not, so we are aiming to process tens of thousands of
182
entries each second within the dirstate context, before exposing anything to
183
the larger codebase. This suggests we want the time for a single file
184
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
185
processed, and to scale to 100 thousand we'll another order of magnitude to do
186
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
187
the file on disk, and then immediately discard, the overhead of object creation
188
becomes a significant cost.
190
Figures: Creating a tuple from 3 elements was profiled at 0.0625
191
microseconds, whereas creating a object which is subclassed from tuple was
192
0.500 microseconds, and creating an object with 3 elements and slots was 3
193
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
194
down to 10 microseconds for the total processing - having 33% of that be object
195
creation is a huge overhead. There is a potential cost in using tuples within
196
each row which is that the conditional code to do comparisons may be slower
197
than method invocation, but method invocation is known to be slow due to stack
198
frame creation, so avoiding methods in these tight inner loops in unfortunately
199
desirable. We can consider a pyrex version of this with objects in future if
209
from stat import S_IEXEC
228
# This is the Windows equivalent of ENOTDIR
229
# It is defined in pywin32.winerror, but we don't want a strong dependency for
230
# just an error code.
231
ERROR_PATH_NOT_FOUND = 3
232
ERROR_DIRECTORY = 267
235
if not getattr(struct, '_compile', None):
236
# Cannot pre-compile the dirstate pack_stat
237
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
238
"""Convert stat values into a packed representation."""
239
return _encode(_pack('>LLLLLL', st.st_size, int(st.st_mtime),
240
int(st.st_ctime), st.st_dev, st.st_ino & 0xFFFFFFFF,
243
# compile the struct compiler we need, so as to only do it once
244
from _struct import Struct
245
_compiled_pack = Struct('>LLLLLL').pack
246
def pack_stat(st, _encode=binascii.b2a_base64, _pack=_compiled_pack):
247
"""Convert stat values into a packed representation."""
248
# jam 20060614 it isn't really worth removing more entries if we
249
# are going to leave it in packed form.
250
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
251
# With all entries, filesize is 5.9M and read time is maybe 280ms
252
# well within the noise margin
254
# base64 encoding always adds a final newline, so strip it off
255
# The current version
256
return _encode(_pack(st.st_size, int(st.st_mtime), int(st.st_ctime),
257
st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
258
# This is 0.060s / 1.520s faster by not encoding as much information
259
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
260
# This is not strictly faster than _encode(_pack())[:-1]
261
# return '%X.%X.%X.%X.%X.%X' % (
262
# st.st_size, int(st.st_mtime), int(st.st_ctime),
263
# st.st_dev, st.st_ino, st.st_mode)
264
# Similar to the _encode(_pack('>LL'))
265
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
268
def _unpack_stat(packed_stat):
269
"""Turn a packed_stat back into the stat fields.
271
This is meant as a debugging tool, should not be used in real code.
273
(st_size, st_mtime, st_ctime, st_dev, st_ino,
274
st_mode) = struct.unpack('>LLLLLL', binascii.a2b_base64(packed_stat))
275
return dict(st_size=st_size, st_mtime=st_mtime, st_ctime=st_ctime,
276
st_dev=st_dev, st_ino=st_ino, st_mode=st_mode)
279
class SHA1Provider(object):
280
"""An interface for getting sha1s of a file."""
282
def sha1(self, abspath):
283
"""Return the sha1 of a file given its absolute path.
285
:param abspath: May be a filesystem encoded absolute path
288
raise NotImplementedError(self.sha1)
290
def stat_and_sha1(self, abspath):
291
"""Return the stat and sha1 of a file given its absolute path.
293
:param abspath: May be a filesystem encoded absolute path
296
Note: the stat should be the stat of the physical file
297
while the sha may be the sha of its canonical content.
299
raise NotImplementedError(self.stat_and_sha1)
302
class DefaultSHA1Provider(SHA1Provider):
303
"""A SHA1Provider that reads directly from the filesystem."""
305
def sha1(self, abspath):
306
"""Return the sha1 of a file given its absolute path."""
307
return osutils.sha_file_by_name(abspath)
309
def stat_and_sha1(self, abspath):
310
"""Return the stat and sha1 of a file given its absolute path."""
311
file_obj = file(abspath, 'rb')
313
statvalue = os.fstat(file_obj.fileno())
314
sha1 = osutils.sha_file(file_obj)
317
return statvalue, sha1
320
class DirState(object):
321
"""Record directory and metadata state for fast access.
323
A dirstate is a specialised data structure for managing local working
324
tree state information. Its not yet well defined whether it is platform
325
specific, and if it is how we detect/parameterize that.
327
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
328
Unlike most bzr disk formats, DirStates must be locked for reading, using
329
lock_read. (This is an os file lock internally.) This is necessary
330
because the file can be rewritten in place.
332
DirStates must be explicitly written with save() to commit changes; just
333
unlocking them does not write the changes to disk.
336
_kind_to_minikind = {
342
'tree-reference': 't',
344
_minikind_to_kind = {
350
't': 'tree-reference',
352
_stat_to_minikind = {
357
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
358
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
361
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
362
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
363
# that we are sure a single record will always fit.
364
BISECT_PAGE_SIZE = 4096
367
IN_MEMORY_UNMODIFIED = 1
368
IN_MEMORY_MODIFIED = 2
369
IN_MEMORY_HASH_MODIFIED = 3 # Only hash-cache updates
371
# A pack_stat (the x's) that is just noise and will never match the output
374
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
376
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
377
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
379
def __init__(self, path, sha1_provider, worth_saving_limit=0):
380
"""Create a DirState object.
382
:param path: The path at which the dirstate file on disk should live.
383
:param sha1_provider: an object meeting the SHA1Provider interface.
384
:param worth_saving_limit: when the exact number of hash changed
385
entries is known, only bother saving the dirstate if more than
386
this count of entries have changed.
387
-1 means never save hash changes, 0 means always save hash changes.
389
# _header_state and _dirblock_state represent the current state
390
# of the dirstate metadata and the per-row data respectiely.
391
# NOT_IN_MEMORY indicates that no data is in memory
392
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
393
# is the same as is on disk
394
# IN_MEMORY_MODIFIED indicates that we have a modified version
395
# of what is on disk.
396
# In future we will add more granularity, for instance _dirblock_state
397
# will probably support partially-in-memory as a separate variable,
398
# allowing for partially-in-memory unmodified and partially-in-memory
400
self._header_state = DirState.NOT_IN_MEMORY
401
self._dirblock_state = DirState.NOT_IN_MEMORY
402
# If true, an error has been detected while updating the dirstate, and
403
# for safety we're not going to commit to disk.
404
self._changes_aborted = False
408
self._state_file = None
409
self._filename = path
410
self._lock_token = None
411
self._lock_state = None
412
self._id_index = None
413
# a map from packed_stat to sha's.
414
self._packed_stat_index = None
415
self._end_of_header = None
416
self._cutoff_time = None
417
self._split_path_cache = {}
418
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
419
self._sha1_provider = sha1_provider
420
if 'hashcache' in debug.debug_flags:
421
self._sha1_file = self._sha1_file_and_mutter
423
self._sha1_file = self._sha1_provider.sha1
424
# These two attributes provide a simple cache for lookups into the
425
# dirstate in-memory vectors. By probing respectively for the last
426
# block, and for the next entry, we save nearly 2 bisections per path
428
self._last_block_index = None
429
self._last_entry_index = None
430
# The set of known hash changes
431
self._known_hash_changes = set()
432
# How many hash changed entries can we have without saving
433
self._worth_saving_limit = worth_saving_limit
437
(self.__class__.__name__, self._filename)
439
def _mark_modified(self, hash_changed_entries=None, header_modified=False):
440
"""Mark this dirstate as modified.
442
:param hash_changed_entries: if non-None, mark just these entries as
443
having their hash modified.
444
:param header_modified: mark the header modified as well, not just the
447
#trace.mutter_callsite(3, "modified hash entries: %s", hash_changed_entries)
448
if hash_changed_entries:
449
self._known_hash_changes.update([e[0] for e in hash_changed_entries])
450
if self._dirblock_state in (DirState.NOT_IN_MEMORY,
451
DirState.IN_MEMORY_UNMODIFIED):
452
# If the dirstate is already marked a IN_MEMORY_MODIFIED, then
453
# that takes precedence.
454
self._dirblock_state = DirState.IN_MEMORY_HASH_MODIFIED
456
# TODO: Since we now have a IN_MEMORY_HASH_MODIFIED state, we
457
# should fail noisily if someone tries to set
458
# IN_MEMORY_MODIFIED but we don't have a write-lock!
459
# We don't know exactly what changed so disable smart saving
460
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
462
self._header_state = DirState.IN_MEMORY_MODIFIED
464
def _mark_unmodified(self):
465
"""Mark this dirstate as unmodified."""
466
self._header_state = DirState.IN_MEMORY_UNMODIFIED
467
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
468
self._known_hash_changes = set()
470
def add(self, path, file_id, kind, stat, fingerprint):
471
"""Add a path to be tracked.
473
:param path: The path within the dirstate - '' is the root, 'foo' is the
474
path foo within the root, 'foo/bar' is the path bar within foo
476
:param file_id: The file id of the path being added.
477
:param kind: The kind of the path, as a string like 'file',
479
:param stat: The output of os.lstat for the path.
480
:param fingerprint: The sha value of the file's canonical form (i.e.
481
after any read filters have been applied),
482
or the target of a symlink,
483
or the referenced revision id for tree-references,
484
or '' for directories.
487
# find the block its in.
488
# find the location in the block.
489
# check its not there
491
#------- copied from inventory.ensure_normalized_name - keep synced.
492
# --- normalized_filename wants a unicode basename only, so get one.
493
dirname, basename = osutils.split(path)
494
# we dont import normalized_filename directly because we want to be
495
# able to change the implementation at runtime for tests.
496
norm_name, can_access = osutils.normalized_filename(basename)
497
if norm_name != basename:
501
raise errors.InvalidNormalization(path)
502
# you should never have files called . or ..; just add the directory
503
# in the parent, or according to the special treatment for the root
504
if basename == '.' or basename == '..':
505
raise errors.InvalidEntryName(path)
506
# now that we've normalised, we need the correct utf8 path and
507
# dirname and basename elements. This single encode and split should be
508
# faster than three separate encodes.
509
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
510
dirname, basename = osutils.split(utf8path)
511
# uses __class__ for speed; the check is needed for safety
512
if file_id.__class__ is not str:
513
raise AssertionError(
514
"must be a utf8 file_id not %s" % (type(file_id), ))
515
# Make sure the file_id does not exist in this tree
517
file_id_entry = self._get_entry(0, fileid_utf8=file_id, include_deleted=True)
518
if file_id_entry != (None, None):
519
if file_id_entry[1][0][0] == 'a':
520
if file_id_entry[0] != (dirname, basename, file_id):
521
# set the old name's current operation to rename
522
self.update_minimal(file_id_entry[0],
528
rename_from = file_id_entry[0][0:2]
530
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
531
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
532
info = '%s:%s' % (kind, path)
533
raise errors.DuplicateFileId(file_id, info)
534
first_key = (dirname, basename, '')
535
block_index, present = self._find_block_index_from_key(first_key)
537
# check the path is not in the tree
538
block = self._dirblocks[block_index][1]
539
entry_index, _ = self._find_entry_index(first_key, block)
540
while (entry_index < len(block) and
541
block[entry_index][0][0:2] == first_key[0:2]):
542
if block[entry_index][1][0][0] not in 'ar':
543
# this path is in the dirstate in the current tree.
544
raise Exception, "adding already added path!"
547
# The block where we want to put the file is not present. But it
548
# might be because the directory was empty, or not loaded yet. Look
549
# for a parent entry, if not found, raise NotVersionedError
550
parent_dir, parent_base = osutils.split(dirname)
551
parent_block_idx, parent_entry_idx, _, parent_present = \
552
self._get_block_entry_index(parent_dir, parent_base, 0)
553
if not parent_present:
554
raise errors.NotVersionedError(path, str(self))
555
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
556
block = self._dirblocks[block_index][1]
557
entry_key = (dirname, basename, file_id)
560
packed_stat = DirState.NULLSTAT
563
packed_stat = pack_stat(stat)
564
parent_info = self._empty_parent_info()
565
minikind = DirState._kind_to_minikind[kind]
566
if rename_from is not None:
568
old_path_utf8 = '%s/%s' % rename_from
570
old_path_utf8 = rename_from[1]
571
parent_info[0] = ('r', old_path_utf8, 0, False, '')
573
entry_data = entry_key, [
574
(minikind, fingerprint, size, False, packed_stat),
576
elif kind == 'directory':
577
entry_data = entry_key, [
578
(minikind, '', 0, False, packed_stat),
580
elif kind == 'symlink':
581
entry_data = entry_key, [
582
(minikind, fingerprint, size, False, packed_stat),
584
elif kind == 'tree-reference':
585
entry_data = entry_key, [
586
(minikind, fingerprint, 0, False, packed_stat),
589
raise errors.BzrError('unknown kind %r' % kind)
590
entry_index, present = self._find_entry_index(entry_key, block)
592
block.insert(entry_index, entry_data)
594
if block[entry_index][1][0][0] != 'a':
595
raise AssertionError(" %r(%r) already added" % (basename, file_id))
596
block[entry_index][1][0] = entry_data[1][0]
598
if kind == 'directory':
599
# insert a new dirblock
600
self._ensure_block(block_index, entry_index, utf8path)
601
self._mark_modified()
603
self._add_to_id_index(self._id_index, entry_key)
605
def _bisect(self, paths):
606
"""Bisect through the disk structure for specific rows.
608
:param paths: A list of paths to find
609
:return: A dict mapping path => entries for found entries. Missing
610
entries will not be in the map.
611
The list is not sorted, and entries will be populated
612
based on when they were read.
614
self._requires_lock()
615
# We need the file pointer to be right after the initial header block
616
self._read_header_if_needed()
617
# If _dirblock_state was in memory, we should just return info from
618
# there, this function is only meant to handle when we want to read
620
if self._dirblock_state != DirState.NOT_IN_MEMORY:
621
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
623
# The disk representation is generally info + '\0\n\0' at the end. But
624
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
625
# Because it means we can sync on the '\n'
626
state_file = self._state_file
627
file_size = os.fstat(state_file.fileno()).st_size
628
# We end up with 2 extra fields, we should have a trailing '\n' to
629
# ensure that we read the whole record, and we should have a precursur
630
# '' which ensures that we start after the previous '\n'
631
entry_field_count = self._fields_per_entry() + 1
633
low = self._end_of_header
634
high = file_size - 1 # Ignore the final '\0'
635
# Map from (dir, name) => entry
638
# Avoid infinite seeking
639
max_count = 30*len(paths)
641
# pending is a list of places to look.
642
# each entry is a tuple of low, high, dir_names
643
# low -> the first byte offset to read (inclusive)
644
# high -> the last byte offset (inclusive)
645
# dir_names -> The list of (dir, name) pairs that should be found in
646
# the [low, high] range
647
pending = [(low, high, paths)]
649
page_size = self._bisect_page_size
651
fields_to_entry = self._get_fields_to_entry()
654
low, high, cur_files = pending.pop()
656
if not cur_files or low >= high:
661
if count > max_count:
662
raise errors.BzrError('Too many seeks, most likely a bug.')
664
mid = max(low, (low+high-page_size)/2)
667
# limit the read size, so we don't end up reading data that we have
669
read_size = min(page_size, (high-mid)+1)
670
block = state_file.read(read_size)
673
entries = block.split('\n')
676
# We didn't find a '\n', so we cannot have found any records.
677
# So put this range back and try again. But we know we have to
678
# increase the page size, because a single read did not contain
679
# a record break (so records must be larger than page_size)
681
pending.append((low, high, cur_files))
684
# Check the first and last entries, in case they are partial, or if
685
# we don't care about the rest of this page
687
first_fields = entries[0].split('\0')
688
if len(first_fields) < entry_field_count:
689
# We didn't get the complete first entry
690
# so move start, and grab the next, which
691
# should be a full entry
692
start += len(entries[0])+1
693
first_fields = entries[1].split('\0')
696
if len(first_fields) <= 2:
697
# We didn't even get a filename here... what do we do?
698
# Try a large page size and repeat this query
700
pending.append((low, high, cur_files))
703
# Find what entries we are looking for, which occur before and
704
# after this first record.
707
first_path = first_fields[1] + '/' + first_fields[2]
709
first_path = first_fields[2]
710
first_loc = _bisect_path_left(cur_files, first_path)
712
# These exist before the current location
713
pre = cur_files[:first_loc]
714
# These occur after the current location, which may be in the
715
# data we read, or might be after the last entry
716
post = cur_files[first_loc:]
718
if post and len(first_fields) >= entry_field_count:
719
# We have files after the first entry
721
# Parse the last entry
722
last_entry_num = len(entries)-1
723
last_fields = entries[last_entry_num].split('\0')
724
if len(last_fields) < entry_field_count:
725
# The very last hunk was not complete,
726
# read the previous hunk
727
after = mid + len(block) - len(entries[-1])
729
last_fields = entries[last_entry_num].split('\0')
731
after = mid + len(block)
734
last_path = last_fields[1] + '/' + last_fields[2]
736
last_path = last_fields[2]
737
last_loc = _bisect_path_right(post, last_path)
739
middle_files = post[:last_loc]
740
post = post[last_loc:]
743
# We have files that should occur in this block
744
# (>= first, <= last)
745
# Either we will find them here, or we can mark them as
748
if middle_files[0] == first_path:
749
# We might need to go before this location
750
pre.append(first_path)
751
if middle_files[-1] == last_path:
752
post.insert(0, last_path)
754
# Find out what paths we have
755
paths = {first_path:[first_fields]}
756
# last_path might == first_path so we need to be
757
# careful if we should append rather than overwrite
758
if last_entry_num != first_entry_num:
759
paths.setdefault(last_path, []).append(last_fields)
760
for num in xrange(first_entry_num+1, last_entry_num):
761
# TODO: jam 20070223 We are already splitting here, so
762
# shouldn't we just split the whole thing rather
763
# than doing the split again in add_one_record?
764
fields = entries[num].split('\0')
766
path = fields[1] + '/' + fields[2]
769
paths.setdefault(path, []).append(fields)
771
for path in middle_files:
772
for fields in paths.get(path, []):
773
# offset by 1 because of the opening '\0'
774
# consider changing fields_to_entry to avoid the
776
entry = fields_to_entry(fields[1:])
777
found.setdefault(path, []).append(entry)
779
# Now we have split up everything into pre, middle, and post, and
780
# we have handled everything that fell in 'middle'.
781
# We add 'post' first, so that we prefer to seek towards the
782
# beginning, so that we will tend to go as early as we need, and
783
# then only seek forward after that.
785
pending.append((after, high, post))
787
pending.append((low, start-1, pre))
789
# Consider that we may want to return the directory entries in sorted
790
# order. For now, we just return them in whatever order we found them,
791
# and leave it up to the caller if they care if it is ordered or not.
794
def _bisect_dirblocks(self, dir_list):
795
"""Bisect through the disk structure to find entries in given dirs.
797
_bisect_dirblocks is meant to find the contents of directories, which
798
differs from _bisect, which only finds individual entries.
800
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
801
:return: A map from dir => entries_for_dir
803
# TODO: jam 20070223 A lot of the bisecting logic could be shared
804
# between this and _bisect. It would require parameterizing the
805
# inner loop with a function, though. We should evaluate the
806
# performance difference.
807
self._requires_lock()
808
# We need the file pointer to be right after the initial header block
809
self._read_header_if_needed()
810
# If _dirblock_state was in memory, we should just return info from
811
# there, this function is only meant to handle when we want to read
813
if self._dirblock_state != DirState.NOT_IN_MEMORY:
814
raise AssertionError("bad dirblock state %r" % self._dirblock_state)
815
# The disk representation is generally info + '\0\n\0' at the end. But
816
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
817
# Because it means we can sync on the '\n'
818
state_file = self._state_file
819
file_size = os.fstat(state_file.fileno()).st_size
820
# We end up with 2 extra fields, we should have a trailing '\n' to
821
# ensure that we read the whole record, and we should have a precursur
822
# '' which ensures that we start after the previous '\n'
823
entry_field_count = self._fields_per_entry() + 1
825
low = self._end_of_header
826
high = file_size - 1 # Ignore the final '\0'
827
# Map from dir => entry
830
# Avoid infinite seeking
831
max_count = 30*len(dir_list)
833
# pending is a list of places to look.
834
# each entry is a tuple of low, high, dir_names
835
# low -> the first byte offset to read (inclusive)
836
# high -> the last byte offset (inclusive)
837
# dirs -> The list of directories that should be found in
838
# the [low, high] range
839
pending = [(low, high, dir_list)]
841
page_size = self._bisect_page_size
843
fields_to_entry = self._get_fields_to_entry()
846
low, high, cur_dirs = pending.pop()
848
if not cur_dirs or low >= high:
853
if count > max_count:
854
raise errors.BzrError('Too many seeks, most likely a bug.')
856
mid = max(low, (low+high-page_size)/2)
859
# limit the read size, so we don't end up reading data that we have
861
read_size = min(page_size, (high-mid)+1)
862
block = state_file.read(read_size)
865
entries = block.split('\n')
868
# We didn't find a '\n', so we cannot have found any records.
869
# So put this range back and try again. But we know we have to
870
# increase the page size, because a single read did not contain
871
# a record break (so records must be larger than page_size)
873
pending.append((low, high, cur_dirs))
876
# Check the first and last entries, in case they are partial, or if
877
# we don't care about the rest of this page
879
first_fields = entries[0].split('\0')
880
if len(first_fields) < entry_field_count:
881
# We didn't get the complete first entry
882
# so move start, and grab the next, which
883
# should be a full entry
884
start += len(entries[0])+1
885
first_fields = entries[1].split('\0')
888
if len(first_fields) <= 1:
889
# We didn't even get a dirname here... what do we do?
890
# Try a large page size and repeat this query
892
pending.append((low, high, cur_dirs))
895
# Find what entries we are looking for, which occur before and
896
# after this first record.
898
first_dir = first_fields[1]
899
first_loc = bisect.bisect_left(cur_dirs, first_dir)
901
# These exist before the current location
902
pre = cur_dirs[:first_loc]
903
# These occur after the current location, which may be in the
904
# data we read, or might be after the last entry
905
post = cur_dirs[first_loc:]
907
if post and len(first_fields) >= entry_field_count:
908
# We have records to look at after the first entry
910
# Parse the last entry
911
last_entry_num = len(entries)-1
912
last_fields = entries[last_entry_num].split('\0')
913
if len(last_fields) < entry_field_count:
914
# The very last hunk was not complete,
915
# read the previous hunk
916
after = mid + len(block) - len(entries[-1])
918
last_fields = entries[last_entry_num].split('\0')
920
after = mid + len(block)
922
last_dir = last_fields[1]
923
last_loc = bisect.bisect_right(post, last_dir)
925
middle_files = post[:last_loc]
926
post = post[last_loc:]
929
# We have files that should occur in this block
930
# (>= first, <= last)
931
# Either we will find them here, or we can mark them as
934
if middle_files[0] == first_dir:
935
# We might need to go before this location
936
pre.append(first_dir)
937
if middle_files[-1] == last_dir:
938
post.insert(0, last_dir)
940
# Find out what paths we have
941
paths = {first_dir:[first_fields]}
942
# last_dir might == first_dir so we need to be
943
# careful if we should append rather than overwrite
944
if last_entry_num != first_entry_num:
945
paths.setdefault(last_dir, []).append(last_fields)
946
for num in xrange(first_entry_num+1, last_entry_num):
947
# TODO: jam 20070223 We are already splitting here, so
948
# shouldn't we just split the whole thing rather
949
# than doing the split again in add_one_record?
950
fields = entries[num].split('\0')
951
paths.setdefault(fields[1], []).append(fields)
953
for cur_dir in middle_files:
954
for fields in paths.get(cur_dir, []):
955
# offset by 1 because of the opening '\0'
956
# consider changing fields_to_entry to avoid the
958
entry = fields_to_entry(fields[1:])
959
found.setdefault(cur_dir, []).append(entry)
961
# Now we have split up everything into pre, middle, and post, and
962
# we have handled everything that fell in 'middle'.
963
# We add 'post' first, so that we prefer to seek towards the
964
# beginning, so that we will tend to go as early as we need, and
965
# then only seek forward after that.
967
pending.append((after, high, post))
969
pending.append((low, start-1, pre))
973
def _bisect_recursive(self, paths):
974
"""Bisect for entries for all paths and their children.
976
This will use bisect to find all records for the supplied paths. It
977
will then continue to bisect for any records which are marked as
978
directories. (and renames?)
980
:param paths: A sorted list of (dir, name) pairs
981
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
982
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
984
# Map from (dir, name, file_id) => [tree_info]
987
found_dir_names = set()
989
# Directories that have been read
990
processed_dirs = set()
991
# Get the ball rolling with the first bisect for all entries.
992
newly_found = self._bisect(paths)
995
# Directories that need to be read
997
paths_to_search = set()
998
for entry_list in newly_found.itervalues():
999
for dir_name_id, trees_info in entry_list:
1000
found[dir_name_id] = trees_info
1001
found_dir_names.add(dir_name_id[:2])
1003
for tree_info in trees_info:
1004
minikind = tree_info[0]
1007
# We already processed this one as a directory,
1008
# we don't need to do the extra work again.
1010
subdir, name, file_id = dir_name_id
1011
path = osutils.pathjoin(subdir, name)
1013
if path not in processed_dirs:
1014
pending_dirs.add(path)
1015
elif minikind == 'r':
1016
# Rename, we need to directly search the target
1017
# which is contained in the fingerprint column
1018
dir_name = osutils.split(tree_info[1])
1019
if dir_name[0] in pending_dirs:
1020
# This entry will be found in the dir search
1022
if dir_name not in found_dir_names:
1023
paths_to_search.add(tree_info[1])
1024
# Now we have a list of paths to look for directly, and
1025
# directory blocks that need to be read.
1026
# newly_found is mixing the keys between (dir, name) and path
1027
# entries, but that is okay, because we only really care about the
1029
newly_found = self._bisect(sorted(paths_to_search))
1030
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
1031
processed_dirs.update(pending_dirs)
1034
def _discard_merge_parents(self):
1035
"""Discard any parents trees beyond the first.
1037
Note that if this fails the dirstate is corrupted.
1039
After this function returns the dirstate contains 2 trees, neither of
1042
self._read_header_if_needed()
1043
parents = self.get_parent_ids()
1044
if len(parents) < 1:
1046
# only require all dirblocks if we are doing a full-pass removal.
1047
self._read_dirblocks_if_needed()
1048
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
1049
def iter_entries_removable():
1050
for block in self._dirblocks:
1051
deleted_positions = []
1052
for pos, entry in enumerate(block[1]):
1054
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
1055
deleted_positions.append(pos)
1056
if deleted_positions:
1057
if len(deleted_positions) == len(block[1]):
1060
for pos in reversed(deleted_positions):
1062
# if the first parent is a ghost:
1063
if parents[0] in self.get_ghosts():
1064
empty_parent = [DirState.NULL_PARENT_DETAILS]
1065
for entry in iter_entries_removable():
1066
entry[1][1:] = empty_parent
1068
for entry in iter_entries_removable():
1072
self._parents = [parents[0]]
1073
self._mark_modified(header_modified=True)
1075
def _empty_parent_info(self):
1076
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
1079
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
1080
"""Ensure a block for dirname exists.
1082
This function exists to let callers which know that there is a
1083
directory dirname ensure that the block for it exists. This block can
1084
fail to exist because of demand loading, or because a directory had no
1085
children. In either case it is not an error. It is however an error to
1086
call this if there is no parent entry for the directory, and thus the
1087
function requires the coordinates of such an entry to be provided.
1089
The root row is special cased and can be indicated with a parent block
1092
:param parent_block_index: The index of the block in which dirname's row
1094
:param parent_row_index: The index in the parent block where the row
1096
:param dirname: The utf8 dirname to ensure there is a block for.
1097
:return: The index for the block.
1099
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
1100
# This is the signature of the root row, and the
1101
# contents-of-root row is always index 1
1103
# the basename of the directory must be the end of its full name.
1104
if not (parent_block_index == -1 and
1105
parent_block_index == -1 and dirname == ''):
1106
if not dirname.endswith(
1107
self._dirblocks[parent_block_index][1][parent_row_index][0][1]):
1108
raise AssertionError("bad dirname %r" % dirname)
1109
block_index, present = self._find_block_index_from_key((dirname, '', ''))
1111
## In future, when doing partial parsing, this should load and
1112
# populate the entire block.
1113
self._dirblocks.insert(block_index, (dirname, []))
1116
def _entries_to_current_state(self, new_entries):
1117
"""Load new_entries into self.dirblocks.
1119
Process new_entries into the current state object, making them the active
1120
state. The entries are grouped together by directory to form dirblocks.
1122
:param new_entries: A sorted list of entries. This function does not sort
1123
to prevent unneeded overhead when callers have a sorted list already.
1126
if new_entries[0][0][0:2] != ('', ''):
1127
raise AssertionError(
1128
"Missing root row %r" % (new_entries[0][0],))
1129
# The two blocks here are deliberate: the root block and the
1130
# contents-of-root block.
1131
self._dirblocks = [('', []), ('', [])]
1132
current_block = self._dirblocks[0][1]
1133
current_dirname = ''
1135
append_entry = current_block.append
1136
for entry in new_entries:
1137
if entry[0][0] != current_dirname:
1138
# new block - different dirname
1140
current_dirname = entry[0][0]
1141
self._dirblocks.append((current_dirname, current_block))
1142
append_entry = current_block.append
1143
# append the entry to the current block
1145
self._split_root_dirblock_into_contents()
1147
def _split_root_dirblock_into_contents(self):
1148
"""Split the root dirblocks into root and contents-of-root.
1150
After parsing by path, we end up with root entries and contents-of-root
1151
entries in the same block. This loop splits them out again.
1153
# The above loop leaves the "root block" entries mixed with the
1154
# "contents-of-root block". But we don't want an if check on
1155
# all entries, so instead we just fix it up here.
1156
if self._dirblocks[1] != ('', []):
1157
raise ValueError("bad dirblock start %r" % (self._dirblocks[1],))
1159
contents_of_root_block = []
1160
for entry in self._dirblocks[0][1]:
1161
if not entry[0][1]: # This is a root entry
1162
root_block.append(entry)
1164
contents_of_root_block.append(entry)
1165
self._dirblocks[0] = ('', root_block)
1166
self._dirblocks[1] = ('', contents_of_root_block)
1168
def _entries_for_path(self, path):
1169
"""Return a list with all the entries that match path for all ids."""
1170
dirname, basename = os.path.split(path)
1171
key = (dirname, basename, '')
1172
block_index, present = self._find_block_index_from_key(key)
1174
# the block which should contain path is absent.
1177
block = self._dirblocks[block_index][1]
1178
entry_index, _ = self._find_entry_index(key, block)
1179
# we may need to look at multiple entries at this path: walk while the specific_files match.
1180
while (entry_index < len(block) and
1181
block[entry_index][0][0:2] == key[0:2]):
1182
result.append(block[entry_index])
1186
def _entry_to_line(self, entry):
1187
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1189
:param entry: An entry_tuple as defined in the module docstring.
1191
entire_entry = list(entry[0])
1192
for tree_number, tree_data in enumerate(entry[1]):
1193
# (minikind, fingerprint, size, executable, tree_specific_string)
1194
entire_entry.extend(tree_data)
1195
# 3 for the key, 5 for the fields per tree.
1196
tree_offset = 3 + tree_number * 5
1198
entire_entry[tree_offset + 0] = tree_data[0]
1200
entire_entry[tree_offset + 2] = str(tree_data[2])
1202
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1203
return '\0'.join(entire_entry)
1205
def _fields_per_entry(self):
1206
"""How many null separated fields should be in each entry row.
1208
Each line now has an extra '\n' field which is not used
1209
so we just skip over it
1211
3 fields for the key
1212
+ number of fields per tree_data (5) * tree count
1215
tree_count = 1 + self._num_present_parents()
1216
return 3 + 5 * tree_count + 1
1218
def _find_block(self, key, add_if_missing=False):
1219
"""Return the block that key should be present in.
1221
:param key: A dirstate entry key.
1222
:return: The block tuple.
1224
block_index, present = self._find_block_index_from_key(key)
1226
if not add_if_missing:
1227
# check to see if key is versioned itself - we might want to
1228
# add it anyway, because dirs with no entries dont get a
1229
# dirblock at parse time.
1230
# This is an uncommon branch to take: most dirs have children,
1231
# and most code works with versioned paths.
1232
parent_base, parent_name = osutils.split(key[0])
1233
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1234
# some parent path has not been added - its an error to add
1236
raise errors.NotVersionedError(key[0:2], str(self))
1237
self._dirblocks.insert(block_index, (key[0], []))
1238
return self._dirblocks[block_index]
1240
def _find_block_index_from_key(self, key):
1241
"""Find the dirblock index for a key.
1243
:return: The block index, True if the block for the key is present.
1245
if key[0:2] == ('', ''):
1248
if (self._last_block_index is not None and
1249
self._dirblocks[self._last_block_index][0] == key[0]):
1250
return self._last_block_index, True
1253
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1254
cache=self._split_path_cache)
1255
# _right returns one-past-where-key is so we have to subtract
1256
# one to use it. we use _right here because there are two
1257
# '' blocks - the root, and the contents of root
1258
# we always have a minimum of 2 in self._dirblocks: root and
1259
# root-contents, and for '', we get 2 back, so this is
1260
# simple and correct:
1261
present = (block_index < len(self._dirblocks) and
1262
self._dirblocks[block_index][0] == key[0])
1263
self._last_block_index = block_index
1264
# Reset the entry index cache to the beginning of the block.
1265
self._last_entry_index = -1
1266
return block_index, present
1268
def _find_entry_index(self, key, block):
1269
"""Find the entry index for a key in a block.
1271
:return: The entry index, True if the entry for the key is present.
1273
len_block = len(block)
1275
if self._last_entry_index is not None:
1277
entry_index = self._last_entry_index + 1
1278
# A hit is when the key is after the last slot, and before or
1279
# equal to the next slot.
1280
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1281
key <= block[entry_index][0]):
1282
self._last_entry_index = entry_index
1283
present = (block[entry_index][0] == key)
1284
return entry_index, present
1287
entry_index = bisect.bisect_left(block, (key, []))
1288
present = (entry_index < len_block and
1289
block[entry_index][0] == key)
1290
self._last_entry_index = entry_index
1291
return entry_index, present
1294
def from_tree(tree, dir_state_filename, sha1_provider=None):
1295
"""Create a dirstate from a bzr Tree.
1297
:param tree: The tree which should provide parent information and
1299
:param sha1_provider: an object meeting the SHA1Provider interface.
1300
If None, a DefaultSHA1Provider is used.
1301
:return: a DirState object which is currently locked for writing.
1302
(it was locked by DirState.initialize)
1304
result = DirState.initialize(dir_state_filename,
1305
sha1_provider=sha1_provider)
1309
parent_ids = tree.get_parent_ids()
1310
num_parents = len(parent_ids)
1312
for parent_id in parent_ids:
1313
parent_tree = tree.branch.repository.revision_tree(parent_id)
1314
parent_trees.append((parent_id, parent_tree))
1315
parent_tree.lock_read()
1316
result.set_parent_trees(parent_trees, [])
1317
result.set_state_from_inventory(tree.inventory)
1319
for revid, parent_tree in parent_trees:
1320
parent_tree.unlock()
1323
# The caller won't have a chance to unlock this, so make sure we
1329
def update_by_delta(self, delta):
1330
"""Apply an inventory delta to the dirstate for tree 0
1332
This is the workhorse for apply_inventory_delta in dirstate based
1335
:param delta: An inventory delta. See Inventory.apply_delta for
1338
self._read_dirblocks_if_needed()
1339
encode = cache_utf8.encode
1342
# Accumulate parent references (path_utf8, id), to check for parentless
1343
# items or items placed under files/links/tree-references. We get
1344
# references from every item in the delta that is not a deletion and
1345
# is not itself the root.
1347
# Added ids must not be in the dirstate already. This set holds those
1350
# This loop transforms the delta to single atomic operations that can
1351
# be executed and validated.
1352
for old_path, new_path, file_id, inv_entry in sorted(
1353
inventory._check_delta_unique_old_paths(
1354
inventory._check_delta_unique_new_paths(
1355
inventory._check_delta_ids_match_entry(
1356
inventory._check_delta_ids_are_valid(
1357
inventory._check_delta_new_path_entry_both_or_None(delta))))),
1359
if (file_id in insertions) or (file_id in removals):
1360
raise errors.InconsistentDelta(old_path or new_path, file_id,
1362
if old_path is not None:
1363
old_path = old_path.encode('utf-8')
1364
removals[file_id] = old_path
1366
new_ids.add(file_id)
1367
if new_path is not None:
1368
if inv_entry is None:
1369
raise errors.InconsistentDelta(new_path, file_id,
1370
"new_path with no entry")
1371
new_path = new_path.encode('utf-8')
1372
dirname_utf8, basename = osutils.split(new_path)
1374
parents.add((dirname_utf8, inv_entry.parent_id))
1375
key = (dirname_utf8, basename, file_id)
1376
minikind = DirState._kind_to_minikind[inv_entry.kind]
1378
fingerprint = inv_entry.reference_revision or ''
1381
insertions[file_id] = (key, minikind, inv_entry.executable,
1382
fingerprint, new_path)
1383
# Transform moves into delete+add pairs
1384
if None not in (old_path, new_path):
1385
for child in self._iter_child_entries(0, old_path):
1386
if child[0][2] in insertions or child[0][2] in removals:
1388
child_dirname = child[0][0]
1389
child_basename = child[0][1]
1390
minikind = child[1][0][0]
1391
fingerprint = child[1][0][4]
1392
executable = child[1][0][3]
1393
old_child_path = osutils.pathjoin(child_dirname,
1395
removals[child[0][2]] = old_child_path
1396
child_suffix = child_dirname[len(old_path):]
1397
new_child_dirname = (new_path + child_suffix)
1398
key = (new_child_dirname, child_basename, child[0][2])
1399
new_child_path = osutils.pathjoin(new_child_dirname,
1401
insertions[child[0][2]] = (key, minikind, executable,
1402
fingerprint, new_child_path)
1403
self._check_delta_ids_absent(new_ids, delta, 0)
1405
self._apply_removals(removals.iteritems())
1406
self._apply_insertions(insertions.values())
1408
self._after_delta_check_parents(parents, 0)
1409
except errors.BzrError, e:
1410
self._changes_aborted = True
1411
if 'integrity error' not in str(e):
1413
# _get_entry raises BzrError when a request is inconsistent; we
1414
# want such errors to be shown as InconsistentDelta - and that
1415
# fits the behaviour we trigger.
1416
raise errors.InconsistentDeltaDelta(delta, "error from _get_entry.")
1418
def _apply_removals(self, removals):
1419
for file_id, path in sorted(removals, reverse=True,
1420
key=operator.itemgetter(1)):
1421
dirname, basename = osutils.split(path)
1422
block_i, entry_i, d_present, f_present = \
1423
self._get_block_entry_index(dirname, basename, 0)
1425
entry = self._dirblocks[block_i][1][entry_i]
1427
self._changes_aborted = True
1428
raise errors.InconsistentDelta(path, file_id,
1429
"Wrong path for old path.")
1430
if not f_present or entry[1][0][0] in 'ar':
1431
self._changes_aborted = True
1432
raise errors.InconsistentDelta(path, file_id,
1433
"Wrong path for old path.")
1434
if file_id != entry[0][2]:
1435
self._changes_aborted = True
1436
raise errors.InconsistentDelta(path, file_id,
1437
"Attempt to remove path has wrong id - found %r."
1439
self._make_absent(entry)
1440
# See if we have a malformed delta: deleting a directory must not
1441
# leave crud behind. This increases the number of bisects needed
1442
# substantially, but deletion or renames of large numbers of paths
1443
# is rare enough it shouldn't be an issue (famous last words?) RBC
1445
block_i, entry_i, d_present, f_present = \
1446
self._get_block_entry_index(path, '', 0)
1448
# The dir block is still present in the dirstate; this could
1449
# be due to it being in a parent tree, or a corrupt delta.
1450
for child_entry in self._dirblocks[block_i][1]:
1451
if child_entry[1][0][0] not in ('r', 'a'):
1452
self._changes_aborted = True
1453
raise errors.InconsistentDelta(path, entry[0][2],
1454
"The file id was deleted but its children were "
1457
def _apply_insertions(self, adds):
1459
for key, minikind, executable, fingerprint, path_utf8 in sorted(adds):
1460
self.update_minimal(key, minikind, executable, fingerprint,
1461
path_utf8=path_utf8)
1462
except errors.NotVersionedError:
1463
self._changes_aborted = True
1464
raise errors.InconsistentDelta(path_utf8.decode('utf8'), key[2],
1467
def update_basis_by_delta(self, delta, new_revid):
1468
"""Update the parents of this tree after a commit.
1470
This gives the tree one parent, with revision id new_revid. The
1471
inventory delta is applied to the current basis tree to generate the
1472
inventory for the parent new_revid, and all other parent trees are
1475
Note that an exception during the operation of this method will leave
1476
the dirstate in a corrupt state where it should not be saved.
1478
Finally, we expect all changes to be synchronising the basis tree with
1481
:param new_revid: The new revision id for the trees parent.
1482
:param delta: An inventory delta (see apply_inventory_delta) describing
1483
the changes from the current left most parent revision to new_revid.
1485
self._read_dirblocks_if_needed()
1486
self._discard_merge_parents()
1487
if self._ghosts != []:
1488
raise NotImplementedError(self.update_basis_by_delta)
1489
if len(self._parents) == 0:
1490
# setup a blank tree, the most simple way.
1491
empty_parent = DirState.NULL_PARENT_DETAILS
1492
for entry in self._iter_entries():
1493
entry[1].append(empty_parent)
1494
self._parents.append(new_revid)
1496
self._parents[0] = new_revid
1498
delta = sorted(delta, reverse=True)
1502
# The paths this function accepts are unicode and must be encoded as we
1504
encode = cache_utf8.encode
1505
inv_to_entry = self._inv_entry_to_details
1506
# delta is now (deletes, changes), (adds) in reverse lexographical
1508
# deletes in reverse lexographic order are safe to process in situ.
1509
# renames are not, as a rename from any path could go to a path
1510
# lexographically lower, so we transform renames into delete, add pairs,
1511
# expanding them recursively as needed.
1512
# At the same time, to reduce interface friction we convert the input
1513
# inventory entries to dirstate.
1514
root_only = ('', '')
1515
# Accumulate parent references (path_utf8, id), to check for parentless
1516
# items or items placed under files/links/tree-references. We get
1517
# references from every item in the delta that is not a deletion and
1518
# is not itself the root.
1520
# Added ids must not be in the dirstate already. This set holds those
1523
for old_path, new_path, file_id, inv_entry in delta:
1524
if inv_entry is not None and file_id != inv_entry.file_id:
1525
raise errors.InconsistentDelta(new_path, file_id,
1526
"mismatched entry file_id %r" % inv_entry)
1527
if new_path is not None:
1528
if inv_entry is None:
1529
raise errors.InconsistentDelta(new_path, file_id,
1530
"new_path with no entry")
1531
new_path_utf8 = encode(new_path)
1532
# note the parent for validation
1533
dirname_utf8, basename_utf8 = osutils.split(new_path_utf8)
1535
parents.add((dirname_utf8, inv_entry.parent_id))
1536
if old_path is None:
1537
adds.append((None, encode(new_path), file_id,
1538
inv_to_entry(inv_entry), True))
1539
new_ids.add(file_id)
1540
elif new_path is None:
1541
deletes.append((encode(old_path), None, file_id, None, True))
1542
elif (old_path, new_path) != root_only:
1544
# Because renames must preserve their children we must have
1545
# processed all relocations and removes before hand. The sort
1546
# order ensures we've examined the child paths, but we also
1547
# have to execute the removals, or the split to an add/delete
1548
# pair will result in the deleted item being reinserted, or
1549
# renamed items being reinserted twice - and possibly at the
1550
# wrong place. Splitting into a delete/add pair also simplifies
1551
# the handling of entries with ('f', ...), ('r' ...) because
1552
# the target of the 'r' is old_path here, and we add that to
1553
# deletes, meaning that the add handler does not need to check
1554
# for 'r' items on every pass.
1555
self._update_basis_apply_deletes(deletes)
1557
# Split into an add/delete pair recursively.
1558
adds.append((None, new_path_utf8, file_id,
1559
inv_to_entry(inv_entry), False))
1560
# Expunge deletes that we've seen so that deleted/renamed
1561
# children of a rename directory are handled correctly.
1562
new_deletes = reversed(list(self._iter_child_entries(1,
1564
# Remove the current contents of the tree at orig_path, and
1565
# reinsert at the correct new path.
1566
for entry in new_deletes:
1568
source_path = entry[0][0] + '/' + entry[0][1]
1570
source_path = entry[0][1]
1572
target_path = new_path_utf8 + source_path[len(old_path):]
1575
raise AssertionError("cannot rename directory to"
1577
target_path = source_path[len(old_path) + 1:]
1578
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1580
(source_path, target_path, entry[0][2], None, False))
1582
(encode(old_path), new_path, file_id, None, False))
1584
# changes to just the root should not require remove/insertion
1586
changes.append((encode(old_path), encode(new_path), file_id,
1587
inv_to_entry(inv_entry)))
1588
self._check_delta_ids_absent(new_ids, delta, 1)
1590
# Finish expunging deletes/first half of renames.
1591
self._update_basis_apply_deletes(deletes)
1592
# Reinstate second half of renames and new paths.
1593
self._update_basis_apply_adds(adds)
1594
# Apply in-situ changes.
1595
self._update_basis_apply_changes(changes)
1597
self._after_delta_check_parents(parents, 1)
1598
except errors.BzrError, e:
1599
self._changes_aborted = True
1600
if 'integrity error' not in str(e):
1602
# _get_entry raises BzrError when a request is inconsistent; we
1603
# want such errors to be shown as InconsistentDelta - and that
1604
# fits the behaviour we trigger. Partof this is driven by dirstate
1605
# only supporting deltas that turn the basis into a closer fit to
1607
raise errors.InconsistentDeltaDelta(delta, "error from _get_entry.")
1609
self._mark_modified(header_modified=True)
1610
self._id_index = None
1613
def _check_delta_ids_absent(self, new_ids, delta, tree_index):
1614
"""Check that none of the file_ids in new_ids are present in a tree."""
1617
id_index = self._get_id_index()
1618
for file_id in new_ids:
1619
for key in id_index.get(file_id, ()):
1620
block_i, entry_i, d_present, f_present = \
1621
self._get_block_entry_index(key[0], key[1], tree_index)
1623
# In a different tree
1625
entry = self._dirblocks[block_i][1][entry_i]
1626
if entry[0][2] != file_id:
1627
# Different file_id, so not what we want.
1629
# NB: No changes made before this helper is called, so no need
1630
# to set the _changes_aborted flag.
1631
raise errors.InconsistentDelta(
1632
("%s/%s" % key[0:2]).decode('utf8'), file_id,
1633
"This file_id is new in the delta but already present in "
1636
def _update_basis_apply_adds(self, adds):
1637
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1639
They may be adds, or renames that have been split into add/delete
1642
:param adds: A sequence of adds. Each add is a tuple:
1643
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1644
is False when the add is the second half of a remove-and-reinsert
1645
pair created to handle renames and deletes.
1647
# Adds are accumulated partly from renames, so can be in any input
1650
# adds is now in lexographic order, which places all parents before
1651
# their children, so we can process it linearly.
1653
for old_path, new_path, file_id, new_details, real_add in adds:
1654
# the entry for this file_id must be in tree 0.
1655
entry = self._get_entry(0, file_id, new_path)
1656
if entry[0] is None or entry[0][2] != file_id:
1657
self._changes_aborted = True
1658
raise errors.InconsistentDelta(new_path, file_id,
1659
'working tree does not contain new entry')
1660
if real_add and entry[1][1][0] not in absent:
1661
self._changes_aborted = True
1662
raise errors.InconsistentDelta(new_path, file_id,
1663
'The entry was considered to be a genuinely new record,'
1664
' but there was already an old record for it.')
1665
# We don't need to update the target of an 'r' because the handling
1666
# of renames turns all 'r' situations into a delete at the original
1668
entry[1][1] = new_details
1670
def _update_basis_apply_changes(self, changes):
1671
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1673
:param adds: A sequence of changes. Each change is a tuple:
1674
(path_utf8, path_utf8, file_id, (entry_details))
1677
for old_path, new_path, file_id, new_details in changes:
1678
# the entry for this file_id must be in tree 0.
1679
entry = self._get_entry(0, file_id, new_path)
1680
if entry[0] is None or entry[0][2] != file_id:
1681
self._changes_aborted = True
1682
raise errors.InconsistentDelta(new_path, file_id,
1683
'working tree does not contain new entry')
1684
if (entry[1][0][0] in absent or
1685
entry[1][1][0] in absent):
1686
self._changes_aborted = True
1687
raise errors.InconsistentDelta(new_path, file_id,
1688
'changed considered absent')
1689
entry[1][1] = new_details
1691
def _update_basis_apply_deletes(self, deletes):
1692
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1694
They may be deletes, or renames that have been split into add/delete
1697
:param deletes: A sequence of deletes. Each delete is a tuple:
1698
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1699
real_delete is True when the desired outcome is an actual deletion
1700
rather than the rename handling logic temporarily deleting a path
1701
during the replacement of a parent.
1703
null = DirState.NULL_PARENT_DETAILS
1704
for old_path, new_path, file_id, _, real_delete in deletes:
1705
if real_delete != (new_path is None):
1706
self._changes_aborted = True
1707
raise AssertionError("bad delete delta")
1708
# the entry for this file_id must be in tree 1.
1709
dirname, basename = osutils.split(old_path)
1710
block_index, entry_index, dir_present, file_present = \
1711
self._get_block_entry_index(dirname, basename, 1)
1712
if not file_present:
1713
self._changes_aborted = True
1714
raise errors.InconsistentDelta(old_path, file_id,
1715
'basis tree does not contain removed entry')
1716
entry = self._dirblocks[block_index][1][entry_index]
1717
if entry[0][2] != file_id:
1718
self._changes_aborted = True
1719
raise errors.InconsistentDelta(old_path, file_id,
1720
'mismatched file_id in tree 1')
1722
if entry[1][0][0] != 'a':
1723
self._changes_aborted = True
1724
raise errors.InconsistentDelta(old_path, file_id,
1725
'This was marked as a real delete, but the WT state'
1726
' claims that it still exists and is versioned.')
1727
del self._dirblocks[block_index][1][entry_index]
1729
if entry[1][0][0] == 'a':
1730
self._changes_aborted = True
1731
raise errors.InconsistentDelta(old_path, file_id,
1732
'The entry was considered a rename, but the source path'
1733
' is marked as absent.')
1734
# For whatever reason, we were asked to rename an entry
1735
# that was originally marked as deleted. This could be
1736
# because we are renaming the parent directory, and the WT
1737
# current state has the file marked as deleted.
1738
elif entry[1][0][0] == 'r':
1739
# implement the rename
1740
del self._dirblocks[block_index][1][entry_index]
1742
# it is being resurrected here, so blank it out temporarily.
1743
self._dirblocks[block_index][1][entry_index][1][1] = null
1745
def _after_delta_check_parents(self, parents, index):
1746
"""Check that parents required by the delta are all intact.
1748
:param parents: An iterable of (path_utf8, file_id) tuples which are
1749
required to be present in tree 'index' at path_utf8 with id file_id
1751
:param index: The column in the dirstate to check for parents in.
1753
for dirname_utf8, file_id in parents:
1754
# Get the entry - the ensures that file_id, dirname_utf8 exists and
1755
# has the right file id.
1756
entry = self._get_entry(index, file_id, dirname_utf8)
1757
if entry[1] is None:
1758
self._changes_aborted = True
1759
raise errors.InconsistentDelta(dirname_utf8.decode('utf8'),
1760
file_id, "This parent is not present.")
1761
# Parents of things must be directories
1762
if entry[1][index][0] != 'd':
1763
self._changes_aborted = True
1764
raise errors.InconsistentDelta(dirname_utf8.decode('utf8'),
1765
file_id, "This parent is not a directory.")
1767
def _observed_sha1(self, entry, sha1, stat_value,
1768
_stat_to_minikind=_stat_to_minikind, _pack_stat=pack_stat):
1769
"""Note the sha1 of a file.
1771
:param entry: The entry the sha1 is for.
1772
:param sha1: The observed sha1.
1773
:param stat_value: The os.lstat for the file.
1776
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1780
packed_stat = _pack_stat(stat_value)
1782
if self._cutoff_time is None:
1783
self._sha_cutoff_time()
1784
if (stat_value.st_mtime < self._cutoff_time
1785
and stat_value.st_ctime < self._cutoff_time):
1786
entry[1][0] = ('f', sha1, stat_value.st_size, entry[1][0][3],
1788
self._mark_modified([entry])
1790
def _sha_cutoff_time(self):
1791
"""Return cutoff time.
1793
Files modified more recently than this time are at risk of being
1794
undetectably modified and so can't be cached.
1796
# Cache the cutoff time as long as we hold a lock.
1797
# time.time() isn't super expensive (approx 3.38us), but
1798
# when you call it 50,000 times it adds up.
1799
# For comparison, os.lstat() costs 7.2us if it is hot.
1800
self._cutoff_time = int(time.time()) - 3
1801
return self._cutoff_time
1803
def _lstat(self, abspath, entry):
1804
"""Return the os.lstat value for this path."""
1805
return os.lstat(abspath)
1807
def _sha1_file_and_mutter(self, abspath):
1808
# when -Dhashcache is turned on, this is monkey-patched in to log
1810
trace.mutter("dirstate sha1 " + abspath)
1811
return self._sha1_provider.sha1(abspath)
1813
def _is_executable(self, mode, old_executable):
1814
"""Is this file executable?"""
1815
return bool(S_IEXEC & mode)
1817
def _is_executable_win32(self, mode, old_executable):
1818
"""On win32 the executable bit is stored in the dirstate."""
1819
return old_executable
1821
if sys.platform == 'win32':
1822
_is_executable = _is_executable_win32
1824
def _read_link(self, abspath, old_link):
1825
"""Read the target of a symlink"""
1826
# TODO: jam 200700301 On Win32, this could just return the value
1827
# already in memory. However, this really needs to be done at a
1828
# higher level, because there either won't be anything on disk,
1829
# or the thing on disk will be a file.
1830
fs_encoding = osutils._fs_enc
1831
if isinstance(abspath, unicode):
1832
# abspath is defined as the path to pass to lstat. readlink is
1833
# buggy in python < 2.6 (it doesn't encode unicode path into FS
1834
# encoding), so we need to encode ourselves knowing that unicode
1835
# paths are produced by UnicodeDirReader on purpose.
1836
abspath = abspath.encode(fs_encoding)
1837
target = os.readlink(abspath)
1838
if fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1839
# Change encoding if needed
1840
target = target.decode(fs_encoding).encode('UTF-8')
1843
def get_ghosts(self):
1844
"""Return a list of the parent tree revision ids that are ghosts."""
1845
self._read_header_if_needed()
1848
def get_lines(self):
1849
"""Serialise the entire dirstate to a sequence of lines."""
1850
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1851
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1852
# read what's on disk.
1853
self._state_file.seek(0)
1854
return self._state_file.readlines()
1856
lines.append(self._get_parents_line(self.get_parent_ids()))
1857
lines.append(self._get_ghosts_line(self._ghosts))
1858
lines.extend(self._get_entry_lines())
1859
return self._get_output_lines(lines)
1861
def _get_ghosts_line(self, ghost_ids):
1862
"""Create a line for the state file for ghost information."""
1863
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1865
def _get_parents_line(self, parent_ids):
1866
"""Create a line for the state file for parents information."""
1867
return '\0'.join([str(len(parent_ids))] + parent_ids)
1869
def _get_entry_lines(self):
1870
"""Create lines for entries."""
1871
return map(self._entry_to_line, self._iter_entries())
1873
def _get_fields_to_entry(self):
1874
"""Get a function which converts entry fields into a entry record.
1876
This handles size and executable, as well as parent records.
1878
:return: A function which takes a list of fields, and returns an
1879
appropriate record for storing in memory.
1881
# This is intentionally unrolled for performance
1882
num_present_parents = self._num_present_parents()
1883
if num_present_parents == 0:
1884
def fields_to_entry_0_parents(fields, _int=int):
1885
path_name_file_id_key = (fields[0], fields[1], fields[2])
1886
return (path_name_file_id_key, [
1888
fields[3], # minikind
1889
fields[4], # fingerprint
1890
_int(fields[5]), # size
1891
fields[6] == 'y', # executable
1892
fields[7], # packed_stat or revision_id
1894
return fields_to_entry_0_parents
1895
elif num_present_parents == 1:
1896
def fields_to_entry_1_parent(fields, _int=int):
1897
path_name_file_id_key = (fields[0], fields[1], fields[2])
1898
return (path_name_file_id_key, [
1900
fields[3], # minikind
1901
fields[4], # fingerprint
1902
_int(fields[5]), # size
1903
fields[6] == 'y', # executable
1904
fields[7], # packed_stat or revision_id
1907
fields[8], # minikind
1908
fields[9], # fingerprint
1909
_int(fields[10]), # size
1910
fields[11] == 'y', # executable
1911
fields[12], # packed_stat or revision_id
1914
return fields_to_entry_1_parent
1915
elif num_present_parents == 2:
1916
def fields_to_entry_2_parents(fields, _int=int):
1917
path_name_file_id_key = (fields[0], fields[1], fields[2])
1918
return (path_name_file_id_key, [
1920
fields[3], # minikind
1921
fields[4], # fingerprint
1922
_int(fields[5]), # size
1923
fields[6] == 'y', # executable
1924
fields[7], # packed_stat or revision_id
1927
fields[8], # minikind
1928
fields[9], # fingerprint
1929
_int(fields[10]), # size
1930
fields[11] == 'y', # executable
1931
fields[12], # packed_stat or revision_id
1934
fields[13], # minikind
1935
fields[14], # fingerprint
1936
_int(fields[15]), # size
1937
fields[16] == 'y', # executable
1938
fields[17], # packed_stat or revision_id
1941
return fields_to_entry_2_parents
1943
def fields_to_entry_n_parents(fields, _int=int):
1944
path_name_file_id_key = (fields[0], fields[1], fields[2])
1945
trees = [(fields[cur], # minikind
1946
fields[cur+1], # fingerprint
1947
_int(fields[cur+2]), # size
1948
fields[cur+3] == 'y', # executable
1949
fields[cur+4], # stat or revision_id
1950
) for cur in xrange(3, len(fields)-1, 5)]
1951
return path_name_file_id_key, trees
1952
return fields_to_entry_n_parents
1954
def get_parent_ids(self):
1955
"""Return a list of the parent tree ids for the directory state."""
1956
self._read_header_if_needed()
1957
return list(self._parents)
1959
def _get_block_entry_index(self, dirname, basename, tree_index):
1960
"""Get the coordinates for a path in the state structure.
1962
:param dirname: The utf8 dirname to lookup.
1963
:param basename: The utf8 basename to lookup.
1964
:param tree_index: The index of the tree for which this lookup should
1966
:return: A tuple describing where the path is located, or should be
1967
inserted. The tuple contains four fields: the block index, the row
1968
index, the directory is present (boolean), the entire path is
1969
present (boolean). There is no guarantee that either
1970
coordinate is currently reachable unless the found field for it is
1971
True. For instance, a directory not present in the searched tree
1972
may be returned with a value one greater than the current highest
1973
block offset. The directory present field will always be True when
1974
the path present field is True. The directory present field does
1975
NOT indicate that the directory is present in the searched tree,
1976
rather it indicates that there are at least some files in some
1979
self._read_dirblocks_if_needed()
1980
key = dirname, basename, ''
1981
block_index, present = self._find_block_index_from_key(key)
1983
# no such directory - return the dir index and 0 for the row.
1984
return block_index, 0, False, False
1985
block = self._dirblocks[block_index][1] # access the entries only
1986
entry_index, present = self._find_entry_index(key, block)
1987
# linear search through entries at this path to find the one
1989
while entry_index < len(block) and block[entry_index][0][1] == basename:
1990
if block[entry_index][1][tree_index][0] not in 'ar':
1991
# neither absent or relocated
1992
return block_index, entry_index, True, True
1994
return block_index, entry_index, True, False
1996
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None, include_deleted=False):
1997
"""Get the dirstate entry for path in tree tree_index.
1999
If either file_id or path is supplied, it is used as the key to lookup.
2000
If both are supplied, the fastest lookup is used, and an error is
2001
raised if they do not both point at the same row.
2003
:param tree_index: The index of the tree we wish to locate this path
2004
in. If the path is present in that tree, the entry containing its
2005
details is returned, otherwise (None, None) is returned
2006
0 is the working tree, higher indexes are successive parent
2008
:param fileid_utf8: A utf8 file_id to look up.
2009
:param path_utf8: An utf8 path to be looked up.
2010
:param include_deleted: If True, and performing a lookup via
2011
fileid_utf8 rather than path_utf8, return an entry for deleted
2013
:return: The dirstate entry tuple for path, or (None, None)
2015
self._read_dirblocks_if_needed()
2016
if path_utf8 is not None:
2017
if type(path_utf8) is not str:
2018
raise errors.BzrError('path_utf8 is not a str: %s %r'
2019
% (type(path_utf8), path_utf8))
2020
# path lookups are faster
2021
dirname, basename = osutils.split(path_utf8)
2022
block_index, entry_index, dir_present, file_present = \
2023
self._get_block_entry_index(dirname, basename, tree_index)
2024
if not file_present:
2026
entry = self._dirblocks[block_index][1][entry_index]
2027
if not (entry[0][2] and entry[1][tree_index][0] not in ('a', 'r')):
2028
raise AssertionError('unversioned entry?')
2030
if entry[0][2] != fileid_utf8:
2031
self._changes_aborted = True
2032
raise errors.BzrError('integrity error ? : mismatching'
2033
' tree_index, file_id and path')
2036
possible_keys = self._get_id_index().get(fileid_utf8, ())
2037
if not possible_keys:
2039
for key in possible_keys:
2040
block_index, present = \
2041
self._find_block_index_from_key(key)
2042
# strange, probably indicates an out of date
2043
# id index - for now, allow this.
2046
# WARNING: DO not change this code to use _get_block_entry_index
2047
# as that function is not suitable: it does not use the key
2048
# to lookup, and thus the wrong coordinates are returned.
2049
block = self._dirblocks[block_index][1]
2050
entry_index, present = self._find_entry_index(key, block)
2052
entry = self._dirblocks[block_index][1][entry_index]
2053
# TODO: We might want to assert that entry[0][2] ==
2055
if entry[1][tree_index][0] in 'fdlt':
2056
# this is the result we are looking for: the
2057
# real home of this file_id in this tree.
2059
if entry[1][tree_index][0] == 'a':
2060
# there is no home for this entry in this tree
2064
if entry[1][tree_index][0] != 'r':
2065
raise AssertionError(
2066
"entry %r has invalid minikind %r for tree %r" \
2068
entry[1][tree_index][0],
2070
real_path = entry[1][tree_index][1]
2071
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
2072
path_utf8=real_path)
2076
def initialize(cls, path, sha1_provider=None):
2077
"""Create a new dirstate on path.
2079
The new dirstate will be an empty tree - that is it has no parents,
2080
and only a root node - which has id ROOT_ID.
2082
:param path: The name of the file for the dirstate.
2083
:param sha1_provider: an object meeting the SHA1Provider interface.
2084
If None, a DefaultSHA1Provider is used.
2085
:return: A write-locked DirState object.
2087
# This constructs a new DirState object on a path, sets the _state_file
2088
# to a new empty file for that path. It then calls _set_data() with our
2089
# stock empty dirstate information - a root with ROOT_ID, no children,
2090
# and no parents. Finally it calls save() to ensure that this data will
2092
if sha1_provider is None:
2093
sha1_provider = DefaultSHA1Provider()
2094
result = cls(path, sha1_provider)
2095
# root dir and root dir contents with no children.
2096
empty_tree_dirblocks = [('', []), ('', [])]
2097
# a new root directory, with a NULLSTAT.
2098
empty_tree_dirblocks[0][1].append(
2099
(('', '', inventory.ROOT_ID), [
2100
('d', '', 0, False, DirState.NULLSTAT),
2104
result._set_data([], empty_tree_dirblocks)
2112
def _inv_entry_to_details(inv_entry):
2113
"""Convert an inventory entry (from a revision tree) to state details.
2115
:param inv_entry: An inventory entry whose sha1 and link targets can be
2116
relied upon, and which has a revision set.
2117
:return: A details tuple - the details for a single tree at a path +
2120
kind = inv_entry.kind
2121
minikind = DirState._kind_to_minikind[kind]
2122
tree_data = inv_entry.revision
2123
if kind == 'directory':
2127
elif kind == 'symlink':
2128
if inv_entry.symlink_target is None:
2131
fingerprint = inv_entry.symlink_target.encode('utf8')
2134
elif kind == 'file':
2135
fingerprint = inv_entry.text_sha1 or ''
2136
size = inv_entry.text_size or 0
2137
executable = inv_entry.executable
2138
elif kind == 'tree-reference':
2139
fingerprint = inv_entry.reference_revision or ''
2143
raise Exception("can't pack %s" % inv_entry)
2144
return (minikind, fingerprint, size, executable, tree_data)
2146
def _iter_child_entries(self, tree_index, path_utf8):
2147
"""Iterate over all the entries that are children of path_utf.
2149
This only returns entries that are present (not in 'a', 'r') in
2150
tree_index. tree_index data is not refreshed, so if tree 0 is used,
2151
results may differ from that obtained if paths were statted to
2152
determine what ones were directories.
2154
Asking for the children of a non-directory will return an empty
2158
next_pending_dirs = [path_utf8]
2160
while next_pending_dirs:
2161
pending_dirs = next_pending_dirs
2162
next_pending_dirs = []
2163
for path in pending_dirs:
2164
block_index, present = self._find_block_index_from_key(
2166
if block_index == 0:
2168
if len(self._dirblocks) == 1:
2169
# asked for the children of the root with no other
2173
# children of a non-directory asked for.
2175
block = self._dirblocks[block_index]
2176
for entry in block[1]:
2177
kind = entry[1][tree_index][0]
2178
if kind not in absent:
2182
path = entry[0][0] + '/' + entry[0][1]
2185
next_pending_dirs.append(path)
2187
def _iter_entries(self):
2188
"""Iterate over all the entries in the dirstate.
2190
Each yelt item is an entry in the standard format described in the
2191
docstring of bzrlib.dirstate.
2193
self._read_dirblocks_if_needed()
2194
for directory in self._dirblocks:
2195
for entry in directory[1]:
2198
def _get_id_index(self):
2199
"""Get an id index of self._dirblocks.
2201
This maps from file_id => [(directory, name, file_id)] entries where
2202
that file_id appears in one of the trees.
2204
if self._id_index is None:
2206
for key, tree_details in self._iter_entries():
2207
self._add_to_id_index(id_index, key)
2208
self._id_index = id_index
2209
return self._id_index
2211
def _add_to_id_index(self, id_index, entry_key):
2212
"""Add this entry to the _id_index mapping."""
2213
# This code used to use a set for every entry in the id_index. However,
2214
# it is *rare* to have more than one entry. So a set is a large
2215
# overkill. And even when we do, we won't ever have more than the
2216
# number of parent trees. Which is still a small number (rarely >2). As
2217
# such, we use a simple tuple, and do our own uniqueness checks. While
2218
# the 'in' check is O(N) since N is nicely bounded it shouldn't ever
2219
# cause quadratic failure.
2220
# TODO: This should use StaticTuple
2221
file_id = entry_key[2]
2222
entry_key = static_tuple.StaticTuple.from_sequence(entry_key)
2223
if file_id not in id_index:
2224
id_index[file_id] = static_tuple.StaticTuple(entry_key,)
2226
entry_keys = id_index[file_id]
2227
if entry_key not in entry_keys:
2228
id_index[file_id] = entry_keys + (entry_key,)
2230
def _remove_from_id_index(self, id_index, entry_key):
2231
"""Remove this entry from the _id_index mapping.
2233
It is an programming error to call this when the entry_key is not
2236
file_id = entry_key[2]
2237
entry_keys = list(id_index[file_id])
2238
entry_keys.remove(entry_key)
2239
id_index[file_id] = static_tuple.StaticTuple.from_sequence(entry_keys)
2241
def _get_output_lines(self, lines):
2242
"""Format lines for final output.
2244
:param lines: A sequence of lines containing the parents list and the
2247
output_lines = [DirState.HEADER_FORMAT_3]
2248
lines.append('') # a final newline
2249
inventory_text = '\0\n\0'.join(lines)
2250
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
2251
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
2252
num_entries = len(lines)-3
2253
output_lines.append('num_entries: %s\n' % (num_entries,))
2254
output_lines.append(inventory_text)
2257
def _make_deleted_row(self, fileid_utf8, parents):
2258
"""Return a deleted row for fileid_utf8."""
2259
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
2262
def _num_present_parents(self):
2263
"""The number of parent entries in each record row."""
2264
return len(self._parents) - len(self._ghosts)
2267
def on_file(cls, path, sha1_provider=None, worth_saving_limit=0):
2268
"""Construct a DirState on the file at path "path".
2270
:param path: The path at which the dirstate file on disk should live.
2271
:param sha1_provider: an object meeting the SHA1Provider interface.
2272
If None, a DefaultSHA1Provider is used.
2273
:param worth_saving_limit: when the exact number of hash changed
2274
entries is known, only bother saving the dirstate if more than
2275
this count of entries have changed. -1 means never save.
2276
:return: An unlocked DirState object, associated with the given path.
2278
if sha1_provider is None:
2279
sha1_provider = DefaultSHA1Provider()
2280
result = cls(path, sha1_provider,
2281
worth_saving_limit=worth_saving_limit)
2284
def _read_dirblocks_if_needed(self):
2285
"""Read in all the dirblocks from the file if they are not in memory.
2287
This populates self._dirblocks, and sets self._dirblock_state to
2288
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
2291
self._read_header_if_needed()
2292
if self._dirblock_state == DirState.NOT_IN_MEMORY:
2293
_read_dirblocks(self)
2295
def _read_header(self):
2296
"""This reads in the metadata header, and the parent ids.
2298
After reading in, the file should be positioned at the null
2299
just before the start of the first record in the file.
2301
:return: (expected crc checksum, number of entries, parent list)
2303
self._read_prelude()
2304
parent_line = self._state_file.readline()
2305
info = parent_line.split('\0')
2306
num_parents = int(info[0])
2307
self._parents = info[1:-1]
2308
ghost_line = self._state_file.readline()
2309
info = ghost_line.split('\0')
2310
num_ghosts = int(info[1])
2311
self._ghosts = info[2:-1]
2312
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2313
self._end_of_header = self._state_file.tell()
2315
def _read_header_if_needed(self):
2316
"""Read the header of the dirstate file if needed."""
2317
# inline this as it will be called a lot
2318
if not self._lock_token:
2319
raise errors.ObjectNotLocked(self)
2320
if self._header_state == DirState.NOT_IN_MEMORY:
2323
def _read_prelude(self):
2324
"""Read in the prelude header of the dirstate file.
2326
This only reads in the stuff that is not connected to the crc
2327
checksum. The position will be correct to read in the rest of
2328
the file and check the checksum after this point.
2329
The next entry in the file should be the number of parents,
2330
and their ids. Followed by a newline.
2332
header = self._state_file.readline()
2333
if header != DirState.HEADER_FORMAT_3:
2334
raise errors.BzrError(
2335
'invalid header line: %r' % (header,))
2336
crc_line = self._state_file.readline()
2337
if not crc_line.startswith('crc32: '):
2338
raise errors.BzrError('missing crc32 checksum: %r' % crc_line)
2339
self.crc_expected = int(crc_line[len('crc32: '):-1])
2340
num_entries_line = self._state_file.readline()
2341
if not num_entries_line.startswith('num_entries: '):
2342
raise errors.BzrError('missing num_entries line')
2343
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
2345
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
2346
"""Find a sha1 given a stat lookup."""
2347
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
2349
def _get_packed_stat_index(self):
2350
"""Get a packed_stat index of self._dirblocks."""
2351
if self._packed_stat_index is None:
2353
for key, tree_details in self._iter_entries():
2354
if tree_details[0][0] == 'f':
2355
index[tree_details[0][4]] = tree_details[0][1]
2356
self._packed_stat_index = index
2357
return self._packed_stat_index
2360
"""Save any pending changes created during this session.
2362
We reuse the existing file, because that prevents race conditions with
2363
file creation, and use oslocks on it to prevent concurrent modification
2364
and reads - because dirstate's incremental data aggregation is not
2365
compatible with reading a modified file, and replacing a file in use by
2366
another process is impossible on Windows.
2368
A dirstate in read only mode should be smart enough though to validate
2369
that the file has not changed, and otherwise discard its cache and
2370
start over, to allow for fine grained read lock duration, so 'status'
2371
wont block 'commit' - for example.
2373
if self._changes_aborted:
2374
# Should this be a warning? For now, I'm expecting that places that
2375
# mark it inconsistent will warn, making a warning here redundant.
2376
trace.mutter('Not saving DirState because '
2377
'_changes_aborted is set.')
2379
# TODO: Since we now distinguish IN_MEMORY_MODIFIED from
2380
# IN_MEMORY_HASH_MODIFIED, we should only fail quietly if we fail
2381
# to save an IN_MEMORY_HASH_MODIFIED, and fail *noisily* if we
2382
# fail to save IN_MEMORY_MODIFIED
2383
if self._worth_saving():
2384
grabbed_write_lock = False
2385
if self._lock_state != 'w':
2386
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
2387
# Switch over to the new lock, as the old one may be closed.
2388
# TODO: jam 20070315 We should validate the disk file has
2389
# not changed contents. Since temporary_write_lock may
2390
# not be an atomic operation.
2391
self._lock_token = new_lock
2392
self._state_file = new_lock.f
2393
if not grabbed_write_lock:
2394
# We couldn't grab a write lock, so we switch back to a read one
2397
lines = self.get_lines()
2398
self._state_file.seek(0)
2399
self._state_file.writelines(lines)
2400
self._state_file.truncate()
2401
self._state_file.flush()
2402
self._mark_unmodified()
2404
if grabbed_write_lock:
2405
self._lock_token = self._lock_token.restore_read_lock()
2406
self._state_file = self._lock_token.f
2407
# TODO: jam 20070315 We should validate the disk file has
2408
# not changed contents. Since restore_read_lock may
2409
# not be an atomic operation.
2411
def _worth_saving(self):
2412
"""Is it worth saving the dirstate or not?"""
2413
if (self._header_state == DirState.IN_MEMORY_MODIFIED
2414
or self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
2416
if self._dirblock_state == DirState.IN_MEMORY_HASH_MODIFIED:
2417
if self._worth_saving_limit == -1:
2418
# We never save hash changes when the limit is -1
2420
# If we're using smart saving and only a small number of
2421
# entries have changed their hash, don't bother saving. John has
2422
# suggested using a heuristic here based on the size of the
2423
# changed files and/or tree. For now, we go with a configurable
2424
# number of changes, keeping the calculation time
2425
# as low overhead as possible. (This also keeps all existing
2426
# tests passing as the default is 0, i.e. always save.)
2427
if len(self._known_hash_changes) >= self._worth_saving_limit:
2431
def _set_data(self, parent_ids, dirblocks):
2432
"""Set the full dirstate data in memory.
2434
This is an internal function used to completely replace the objects
2435
in memory state. It puts the dirstate into state 'full-dirty'.
2437
:param parent_ids: A list of parent tree revision ids.
2438
:param dirblocks: A list containing one tuple for each directory in the
2439
tree. Each tuple contains the directory path and a list of entries
2440
found in that directory.
2442
# our memory copy is now authoritative.
2443
self._dirblocks = dirblocks
2444
self._mark_modified(header_modified=True)
2445
self._parents = list(parent_ids)
2446
self._id_index = None
2447
self._packed_stat_index = None
2449
def set_path_id(self, path, new_id):
2450
"""Change the id of path to new_id in the current working tree.
2452
:param path: The path inside the tree to set - '' is the root, 'foo'
2453
is the path foo in the root.
2454
:param new_id: The new id to assign to the path. This must be a utf8
2455
file id (not unicode, and not None).
2457
self._read_dirblocks_if_needed()
2459
# TODO: logic not written
2460
raise NotImplementedError(self.set_path_id)
2461
# TODO: check new id is unique
2462
entry = self._get_entry(0, path_utf8=path)
2463
if entry[0][2] == new_id:
2464
# Nothing to change.
2466
# mark the old path absent, and insert a new root path
2467
self._make_absent(entry)
2468
self.update_minimal(('', '', new_id), 'd',
2469
path_utf8='', packed_stat=entry[1][0][4])
2470
self._mark_modified()
2471
# XXX: This was added by Ian, we need to make sure there
2472
# are tests for it, because it isn't in bzr.dev TRUNK
2473
# It looks like the only place it is called is in setting the root
2474
# id of the tree. So probably we never had an _id_index when we
2475
# don't even have a root yet.
2476
if self._id_index is not None:
2477
self._add_to_id_index(self._id_index, entry[0])
2479
def set_parent_trees(self, trees, ghosts):
2480
"""Set the parent trees for the dirstate.
2482
:param trees: A list of revision_id, tree tuples. tree must be provided
2483
even if the revision_id refers to a ghost: supply an empty tree in
2485
:param ghosts: A list of the revision_ids that are ghosts at the time
2488
# TODO: generate a list of parent indexes to preserve to save
2489
# processing specific parent trees. In the common case one tree will
2490
# be preserved - the left most parent.
2491
# TODO: if the parent tree is a dirstate, we might want to walk them
2492
# all by path in parallel for 'optimal' common-case performance.
2493
# generate new root row.
2494
self._read_dirblocks_if_needed()
2495
# TODO future sketch: Examine the existing parents to generate a change
2496
# map and then walk the new parent trees only, mapping them into the
2497
# dirstate. Walk the dirstate at the same time to remove unreferenced
2500
# sketch: loop over all entries in the dirstate, cherry picking
2501
# entries from the parent trees, if they are not ghost trees.
2502
# after we finish walking the dirstate, all entries not in the dirstate
2503
# are deletes, so we want to append them to the end as per the design
2504
# discussions. So do a set difference on ids with the parents to
2505
# get deletes, and add them to the end.
2506
# During the update process we need to answer the following questions:
2507
# - find other keys containing a fileid in order to create cross-path
2508
# links. We dont't trivially use the inventory from other trees
2509
# because this leads to either double touching, or to accessing
2511
# - find other keys containing a path
2512
# We accumulate each entry via this dictionary, including the root
2515
# we could do parallel iterators, but because file id data may be
2516
# scattered throughout, we dont save on index overhead: we have to look
2517
# at everything anyway. We can probably save cycles by reusing parent
2518
# data and doing an incremental update when adding an additional
2519
# parent, but for now the common cases are adding a new parent (merge),
2520
# and replacing completely (commit), and commit is more common: so
2521
# optimise merge later.
2523
# ---- start generation of full tree mapping data
2524
# what trees should we use?
2525
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2526
# how many trees do we end up with
2527
parent_count = len(parent_trees)
2529
# one: the current tree
2530
for entry in self._iter_entries():
2531
# skip entries not in the current tree
2532
if entry[1][0][0] in 'ar': # absent, relocated
2534
by_path[entry[0]] = [entry[1][0]] + \
2535
[DirState.NULL_PARENT_DETAILS] * parent_count
2536
# TODO: Possibly inline this, since we know it isn't present yet
2537
# id_index[entry[0][2]] = (entry[0],)
2538
self._add_to_id_index(id_index, entry[0])
2540
# now the parent trees:
2541
for tree_index, tree in enumerate(parent_trees):
2542
# the index is off by one, adjust it.
2543
tree_index = tree_index + 1
2544
# when we add new locations for a fileid we need these ranges for
2545
# any fileid in this tree as we set the by_path[id] to:
2546
# already_processed_tree_details + new_details + new_location_suffix
2547
# the suffix is from tree_index+1:parent_count+1.
2548
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2549
# now stitch in all the entries from this tree
2550
for path, entry in tree.inventory.iter_entries_by_dir():
2551
# here we process each trees details for each item in the tree.
2552
# we first update any existing entries for the id at other paths,
2553
# then we either create or update the entry for the id at the
2554
# right path, and finally we add (if needed) a mapping from
2555
# file_id to this path. We do it in this order to allow us to
2556
# avoid checking all known paths for the id when generating a
2557
# new entry at this path: by adding the id->path mapping last,
2558
# all the mappings are valid and have correct relocation
2559
# records where needed.
2560
file_id = entry.file_id
2561
path_utf8 = path.encode('utf8')
2562
dirname, basename = osutils.split(path_utf8)
2563
new_entry_key = (dirname, basename, file_id)
2564
# tree index consistency: All other paths for this id in this tree
2565
# index must point to the correct path.
2566
for entry_key in id_index.get(file_id, ()):
2567
# TODO:PROFILING: It might be faster to just update
2568
# rather than checking if we need to, and then overwrite
2569
# the one we are located at.
2570
if entry_key != new_entry_key:
2571
# this file id is at a different path in one of the
2572
# other trees, so put absent pointers there
2573
# This is the vertical axis in the matrix, all pointing
2575
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2576
# by path consistency: Insert into an existing path record (trivial), or
2577
# add a new one with relocation pointers for the other tree indexes.
2578
entry_keys = id_index.get(file_id, ())
2579
if new_entry_key in entry_keys:
2580
# there is already an entry where this data belongs, just insert it.
2581
by_path[new_entry_key][tree_index] = \
2582
self._inv_entry_to_details(entry)
2584
# add relocated entries to the horizontal axis - this row
2585
# mapping from path,id. We need to look up the correct path
2586
# for the indexes from 0 to tree_index -1
2588
for lookup_index in xrange(tree_index):
2589
# boundary case: this is the first occurence of file_id
2590
# so there are no id_indexes, possibly take this out of
2592
if not len(entry_keys):
2593
new_details.append(DirState.NULL_PARENT_DETAILS)
2595
# grab any one entry, use it to find the right path.
2596
# TODO: optimise this to reduce memory use in highly
2597
# fragmented situations by reusing the relocation
2599
a_key = iter(entry_keys).next()
2600
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2601
# its a pointer or missing statement, use it as is.
2602
new_details.append(by_path[a_key][lookup_index])
2604
# we have the right key, make a pointer to it.
2605
real_path = ('/'.join(a_key[0:2])).strip('/')
2606
new_details.append(('r', real_path, 0, False, ''))
2607
new_details.append(self._inv_entry_to_details(entry))
2608
new_details.extend(new_location_suffix)
2609
by_path[new_entry_key] = new_details
2610
self._add_to_id_index(id_index, new_entry_key)
2611
# --- end generation of full tree mappings
2613
# sort and output all the entries
2614
new_entries = self._sort_entries(by_path.items())
2615
self._entries_to_current_state(new_entries)
2616
self._parents = [rev_id for rev_id, tree in trees]
2617
self._ghosts = list(ghosts)
2618
self._mark_modified(header_modified=True)
2619
self._id_index = id_index
2621
def _sort_entries(self, entry_list):
2622
"""Given a list of entries, sort them into the right order.
2624
This is done when constructing a new dirstate from trees - normally we
2625
try to keep everything in sorted blocks all the time, but sometimes
2626
it's easier to sort after the fact.
2629
# sort by: directory parts, file name, file id
2630
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2631
return sorted(entry_list, key=_key)
2633
def set_state_from_inventory(self, new_inv):
2634
"""Set new_inv as the current state.
2636
This API is called by tree transform, and will usually occur with
2637
existing parent trees.
2639
:param new_inv: The inventory object to set current state from.
2641
if 'evil' in debug.debug_flags:
2642
trace.mutter_callsite(1,
2643
"set_state_from_inventory called; please mutate the tree instead")
2644
tracing = 'dirstate' in debug.debug_flags
2646
trace.mutter("set_state_from_inventory trace:")
2647
self._read_dirblocks_if_needed()
2649
# Two iterators: current data and new data, both in dirblock order.
2650
# We zip them together, which tells about entries that are new in the
2651
# inventory, or removed in the inventory, or present in both and
2654
# You might think we could just synthesize a new dirstate directly
2655
# since we're processing it in the right order. However, we need to
2656
# also consider there may be any number of parent trees and relocation
2657
# pointers, and we don't want to duplicate that here.
2658
new_iterator = new_inv.iter_entries_by_dir()
2659
# we will be modifying the dirstate, so we need a stable iterator. In
2660
# future we might write one, for now we just clone the state into a
2661
# list using a copy so that we see every original item and don't have
2662
# to adjust the position when items are inserted or deleted in the
2663
# underlying dirstate.
2664
old_iterator = iter(list(self._iter_entries()))
2665
# both must have roots so this is safe:
2666
current_new = new_iterator.next()
2667
current_old = old_iterator.next()
2668
def advance(iterator):
2670
return iterator.next()
2671
except StopIteration:
2673
while current_new or current_old:
2674
# skip entries in old that are not really there
2675
if current_old and current_old[1][0][0] in 'ar':
2676
# relocated or absent
2677
current_old = advance(old_iterator)
2680
# convert new into dirblock style
2681
new_path_utf8 = current_new[0].encode('utf8')
2682
new_dirname, new_basename = osutils.split(new_path_utf8)
2683
new_id = current_new[1].file_id
2684
new_entry_key = (new_dirname, new_basename, new_id)
2685
current_new_minikind = \
2686
DirState._kind_to_minikind[current_new[1].kind]
2687
if current_new_minikind == 't':
2688
fingerprint = current_new[1].reference_revision or ''
2690
# We normally only insert or remove records, or update
2691
# them when it has significantly changed. Then we want to
2692
# erase its fingerprint. Unaffected records should
2693
# normally not be updated at all.
2696
# for safety disable variables
2697
new_path_utf8 = new_dirname = new_basename = new_id = \
2698
new_entry_key = None
2699
# 5 cases, we dont have a value that is strictly greater than everything, so
2700
# we make both end conditions explicit
2702
# old is finished: insert current_new into the state.
2704
trace.mutter("Appending from new '%s'.",
2705
new_path_utf8.decode('utf8'))
2706
self.update_minimal(new_entry_key, current_new_minikind,
2707
executable=current_new[1].executable,
2708
path_utf8=new_path_utf8, fingerprint=fingerprint,
2710
current_new = advance(new_iterator)
2711
elif not current_new:
2714
trace.mutter("Truncating from old '%s/%s'.",
2715
current_old[0][0].decode('utf8'),
2716
current_old[0][1].decode('utf8'))
2717
self._make_absent(current_old)
2718
current_old = advance(old_iterator)
2719
elif new_entry_key == current_old[0]:
2720
# same - common case
2721
# We're looking at the same path and id in both the dirstate
2722
# and inventory, so just need to update the fields in the
2723
# dirstate from the one in the inventory.
2724
# TODO: update the record if anything significant has changed.
2725
# the minimal required trigger is if the execute bit or cached
2727
if (current_old[1][0][3] != current_new[1].executable or
2728
current_old[1][0][0] != current_new_minikind):
2730
trace.mutter("Updating in-place change '%s'.",
2731
new_path_utf8.decode('utf8'))
2732
self.update_minimal(current_old[0], current_new_minikind,
2733
executable=current_new[1].executable,
2734
path_utf8=new_path_utf8, fingerprint=fingerprint,
2736
# both sides are dealt with, move on
2737
current_old = advance(old_iterator)
2738
current_new = advance(new_iterator)
2739
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2740
or (new_dirname == current_old[0][0]
2741
and new_entry_key[1:] < current_old[0][1:])):
2743
# add a entry for this and advance new
2745
trace.mutter("Inserting from new '%s'.",
2746
new_path_utf8.decode('utf8'))
2747
self.update_minimal(new_entry_key, current_new_minikind,
2748
executable=current_new[1].executable,
2749
path_utf8=new_path_utf8, fingerprint=fingerprint,
2751
current_new = advance(new_iterator)
2753
# we've advanced past the place where the old key would be,
2754
# without seeing it in the new list. so it must be gone.
2756
trace.mutter("Deleting from old '%s/%s'.",
2757
current_old[0][0].decode('utf8'),
2758
current_old[0][1].decode('utf8'))
2759
self._make_absent(current_old)
2760
current_old = advance(old_iterator)
2761
self._mark_modified()
2762
self._id_index = None
2763
self._packed_stat_index = None
2765
trace.mutter("set_state_from_inventory complete.")
2767
def set_state_from_scratch(self, working_inv, parent_trees, parent_ghosts):
2768
"""Wipe the currently stored state and set it to something new.
2770
This is a hard-reset for the data we are working with.
2772
# Technically, we really want a write lock, but until we write, we
2773
# don't really need it.
2774
self._requires_lock()
2775
# root dir and root dir contents with no children. We have to have a
2776
# root for set_state_from_inventory to work correctly.
2777
empty_root = (('', '', inventory.ROOT_ID),
2778
[('d', '', 0, False, DirState.NULLSTAT)])
2779
empty_tree_dirblocks = [('', [empty_root]), ('', [])]
2780
self._set_data([], empty_tree_dirblocks)
2781
self.set_state_from_inventory(working_inv)
2782
self.set_parent_trees(parent_trees, parent_ghosts)
2784
def _make_absent(self, current_old):
2785
"""Mark current_old - an entry - as absent for tree 0.
2787
:return: True if this was the last details entry for the entry key:
2788
that is, if the underlying block has had the entry removed, thus
2789
shrinking in length.
2791
# build up paths that this id will be left at after the change is made,
2792
# so we can update their cross references in tree 0
2793
all_remaining_keys = set()
2794
# Dont check the working tree, because it's going.
2795
for details in current_old[1][1:]:
2796
if details[0] not in 'ar': # absent, relocated
2797
all_remaining_keys.add(current_old[0])
2798
elif details[0] == 'r': # relocated
2799
# record the key for the real path.
2800
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2801
# absent rows are not present at any path.
2802
last_reference = current_old[0] not in all_remaining_keys
2804
# the current row consists entire of the current item (being marked
2805
# absent), and relocated or absent entries for the other trees:
2806
# Remove it, its meaningless.
2807
block = self._find_block(current_old[0])
2808
entry_index, present = self._find_entry_index(current_old[0], block[1])
2810
raise AssertionError('could not find entry for %s' % (current_old,))
2811
block[1].pop(entry_index)
2812
# if we have an id_index in use, remove this key from it for this id.
2813
if self._id_index is not None:
2814
self._remove_from_id_index(self._id_index, current_old[0])
2815
# update all remaining keys for this id to record it as absent. The
2816
# existing details may either be the record we are marking as deleted
2817
# (if there were other trees with the id present at this path), or may
2819
for update_key in all_remaining_keys:
2820
update_block_index, present = \
2821
self._find_block_index_from_key(update_key)
2823
raise AssertionError('could not find block for %s' % (update_key,))
2824
update_entry_index, present = \
2825
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2827
raise AssertionError('could not find entry for %s' % (update_key,))
2828
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2829
# it must not be absent at the moment
2830
if update_tree_details[0][0] == 'a': # absent
2831
raise AssertionError('bad row %r' % (update_tree_details,))
2832
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2833
self._mark_modified()
2834
return last_reference
2836
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2837
packed_stat=None, size=0, path_utf8=None, fullscan=False):
2838
"""Update an entry to the state in tree 0.
2840
This will either create a new entry at 'key' or update an existing one.
2841
It also makes sure that any other records which might mention this are
2844
:param key: (dir, name, file_id) for the new entry
2845
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2847
:param executable: Should the executable bit be set?
2848
:param fingerprint: Simple fingerprint for new entry: canonical-form
2849
sha1 for files, referenced revision id for subtrees, etc.
2850
:param packed_stat: Packed stat value for new entry.
2851
:param size: Size information for new entry
2852
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2854
:param fullscan: If True then a complete scan of the dirstate is being
2855
done and checking for duplicate rows should not be done. This
2856
should only be set by set_state_from_inventory and similar methods.
2858
If packed_stat and fingerprint are not given, they're invalidated in
2861
block = self._find_block(key)[1]
2862
if packed_stat is None:
2863
packed_stat = DirState.NULLSTAT
2864
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2865
# sometimes present in the dirstate - this seems oddly inconsistent.
2867
entry_index, present = self._find_entry_index(key, block)
2868
new_details = (minikind, fingerprint, size, executable, packed_stat)
2869
id_index = self._get_id_index()
2871
# New record. Check there isn't a entry at this path already.
2873
low_index, _ = self._find_entry_index(key[0:2] + ('',), block)
2874
while low_index < len(block):
2875
entry = block[low_index]
2876
if entry[0][0:2] == key[0:2]:
2877
if entry[1][0][0] not in 'ar':
2878
# This entry has the same path (but a different id) as
2879
# the new entry we're adding, and is present in ths
2881
raise errors.InconsistentDelta(
2882
("%s/%s" % key[0:2]).decode('utf8'), key[2],
2883
"Attempt to add item at path already occupied by "
2884
"id %r" % entry[0][2])
2888
# new entry, synthesis cross reference here,
2889
existing_keys = id_index.get(key[2], ())
2890
if not existing_keys:
2891
# not currently in the state, simplest case
2892
new_entry = key, [new_details] + self._empty_parent_info()
2894
# present at one or more existing other paths.
2895
# grab one of them and use it to generate parent
2896
# relocation/absent entries.
2897
new_entry = key, [new_details]
2898
# existing_keys can be changed as we iterate.
2899
for other_key in tuple(existing_keys):
2900
# change the record at other to be a pointer to this new
2901
# record. The loop looks similar to the change to
2902
# relocations when updating an existing record but its not:
2903
# the test for existing kinds is different: this can be
2904
# factored out to a helper though.
2905
other_block_index, present = self._find_block_index_from_key(
2908
raise AssertionError('could not find block for %s' % (
2910
other_block = self._dirblocks[other_block_index][1]
2911
other_entry_index, present = self._find_entry_index(
2912
other_key, other_block)
2914
raise AssertionError(
2915
'update_minimal: could not find other entry for %s'
2917
if path_utf8 is None:
2918
raise AssertionError('no path')
2919
# Turn this other location into a reference to the new
2920
# location. This also updates the aliased iterator
2921
# (current_old in set_state_from_inventory) so that the old
2922
# entry, if not already examined, is skipped over by that
2924
other_entry = other_block[other_entry_index]
2925
other_entry[1][0] = ('r', path_utf8, 0, False, '')
2926
if self._maybe_remove_row(other_block, other_entry_index,
2928
# If the row holding this was removed, we need to
2929
# recompute where this entry goes
2930
entry_index, _ = self._find_entry_index(key, block)
2933
# adds a tuple to the new details for each column
2934
# - either by copying an existing relocation pointer inside that column
2935
# - or by creating a new pointer to the right row inside that column
2936
num_present_parents = self._num_present_parents()
2937
if num_present_parents:
2938
# TODO: This re-evaluates the existing_keys set, do we need
2939
# to do that ourselves?
2940
other_key = list(existing_keys)[0]
2941
for lookup_index in xrange(1, num_present_parents + 1):
2942
# grab any one entry, use it to find the right path.
2943
# TODO: optimise this to reduce memory use in highly
2944
# fragmented situations by reusing the relocation
2946
update_block_index, present = \
2947
self._find_block_index_from_key(other_key)
2949
raise AssertionError('could not find block for %s' % (other_key,))
2950
update_entry_index, present = \
2951
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2953
raise AssertionError('update_minimal: could not find entry for %s' % (other_key,))
2954
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2955
if update_details[0] in 'ar': # relocated, absent
2956
# its a pointer or absent in lookup_index's tree, use
2958
new_entry[1].append(update_details)
2960
# we have the right key, make a pointer to it.
2961
pointer_path = osutils.pathjoin(*other_key[0:2])
2962
new_entry[1].append(('r', pointer_path, 0, False, ''))
2963
block.insert(entry_index, new_entry)
2964
self._add_to_id_index(id_index, key)
2966
# Does the new state matter?
2967
block[entry_index][1][0] = new_details
2968
# parents cannot be affected by what we do.
2969
# other occurences of this id can be found
2970
# from the id index.
2972
# tree index consistency: All other paths for this id in this tree
2973
# index must point to the correct path. We have to loop here because
2974
# we may have passed entries in the state with this file id already
2975
# that were absent - where parent entries are - and they need to be
2976
# converted to relocated.
2977
if path_utf8 is None:
2978
raise AssertionError('no path')
2979
existing_keys = id_index.get(key[2], ())
2980
if key not in existing_keys:
2981
raise AssertionError('We found the entry in the blocks, but'
2982
' the key is not in the id_index.'
2983
' key: %s, existing_keys: %s' % (key, existing_keys))
2984
for entry_key in existing_keys:
2985
# TODO:PROFILING: It might be faster to just update
2986
# rather than checking if we need to, and then overwrite
2987
# the one we are located at.
2988
if entry_key != key:
2989
# this file id is at a different path in one of the
2990
# other trees, so put absent pointers there
2991
# This is the vertical axis in the matrix, all pointing
2993
block_index, present = self._find_block_index_from_key(entry_key)
2995
raise AssertionError('not present: %r', entry_key)
2996
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2998
raise AssertionError('not present: %r', entry_key)
2999
self._dirblocks[block_index][1][entry_index][1][0] = \
3000
('r', path_utf8, 0, False, '')
3001
# add a containing dirblock if needed.
3002
if new_details[0] == 'd':
3003
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
3004
block_index, present = self._find_block_index_from_key(subdir_key)
3006
self._dirblocks.insert(block_index, (subdir_key[0], []))
3008
self._mark_modified()
3010
def _maybe_remove_row(self, block, index, id_index):
3011
"""Remove index if it is absent or relocated across the row.
3013
id_index is updated accordingly.
3014
:return: True if we removed the row, False otherwise
3016
present_in_row = False
3017
entry = block[index]
3018
for column in entry[1]:
3019
if column[0] not in 'ar':
3020
present_in_row = True
3022
if not present_in_row:
3024
self._remove_from_id_index(id_index, entry[0])
3028
def _validate(self):
3029
"""Check that invariants on the dirblock are correct.
3031
This can be useful in debugging; it shouldn't be necessary in
3034
This must be called with a lock held.
3036
# NOTE: This must always raise AssertionError not just assert,
3037
# otherwise it may not behave properly under python -O
3039
# TODO: All entries must have some content that's not 'a' or 'r',
3040
# otherwise it could just be removed.
3042
# TODO: All relocations must point directly to a real entry.
3044
# TODO: No repeated keys.
3047
from pprint import pformat
3048
self._read_dirblocks_if_needed()
3049
if len(self._dirblocks) > 0:
3050
if not self._dirblocks[0][0] == '':
3051
raise AssertionError(
3052
"dirblocks don't start with root block:\n" + \
3053
pformat(self._dirblocks))
3054
if len(self._dirblocks) > 1:
3055
if not self._dirblocks[1][0] == '':
3056
raise AssertionError(
3057
"dirblocks missing root directory:\n" + \
3058
pformat(self._dirblocks))
3059
# the dirblocks are sorted by their path components, name, and dir id
3060
dir_names = [d[0].split('/')
3061
for d in self._dirblocks[1:]]
3062
if dir_names != sorted(dir_names):
3063
raise AssertionError(
3064
"dir names are not in sorted order:\n" + \
3065
pformat(self._dirblocks) + \
3068
for dirblock in self._dirblocks:
3069
# within each dirblock, the entries are sorted by filename and
3071
for entry in dirblock[1]:
3072
if dirblock[0] != entry[0][0]:
3073
raise AssertionError(
3075
"doesn't match directory name in\n%r" %
3076
(entry, pformat(dirblock)))
3077
if dirblock[1] != sorted(dirblock[1]):
3078
raise AssertionError(
3079
"dirblock for %r is not sorted:\n%s" % \
3080
(dirblock[0], pformat(dirblock)))
3082
def check_valid_parent():
3083
"""Check that the current entry has a valid parent.
3085
This makes sure that the parent has a record,
3086
and that the parent isn't marked as "absent" in the
3087
current tree. (It is invalid to have a non-absent file in an absent
3090
if entry[0][0:2] == ('', ''):
3091
# There should be no parent for the root row
3093
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
3094
if parent_entry == (None, None):
3095
raise AssertionError(
3096
"no parent entry for: %s in tree %s"
3097
% (this_path, tree_index))
3098
if parent_entry[1][tree_index][0] != 'd':
3099
raise AssertionError(
3100
"Parent entry for %s is not marked as a valid"
3101
" directory. %s" % (this_path, parent_entry,))
3103
# For each file id, for each tree: either
3104
# the file id is not present at all; all rows with that id in the
3105
# key have it marked as 'absent'
3106
# OR the file id is present under exactly one name; any other entries
3107
# that mention that id point to the correct name.
3109
# We check this with a dict per tree pointing either to the present
3110
# name, or None if absent.
3111
tree_count = self._num_present_parents() + 1
3112
id_path_maps = [dict() for i in range(tree_count)]
3113
# Make sure that all renamed entries point to the correct location.
3114
for entry in self._iter_entries():
3115
file_id = entry[0][2]
3116
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
3117
if len(entry[1]) != tree_count:
3118
raise AssertionError(
3119
"wrong number of entry details for row\n%s" \
3120
",\nexpected %d" % \
3121
(pformat(entry), tree_count))
3122
absent_positions = 0
3123
for tree_index, tree_state in enumerate(entry[1]):
3124
this_tree_map = id_path_maps[tree_index]
3125
minikind = tree_state[0]
3126
if minikind in 'ar':
3127
absent_positions += 1
3128
# have we seen this id before in this column?
3129
if file_id in this_tree_map:
3130
previous_path, previous_loc = this_tree_map[file_id]
3131
# any later mention of this file must be consistent with
3132
# what was said before
3134
if previous_path is not None:
3135
raise AssertionError(
3136
"file %s is absent in row %r but also present " \
3138
(file_id, entry, previous_path))
3139
elif minikind == 'r':
3140
target_location = tree_state[1]
3141
if previous_path != target_location:
3142
raise AssertionError(
3143
"file %s relocation in row %r but also at %r" \
3144
% (file_id, entry, previous_path))
3146
# a file, directory, etc - may have been previously
3147
# pointed to by a relocation, which must point here
3148
if previous_path != this_path:
3149
raise AssertionError(
3150
"entry %r inconsistent with previous path %r "
3152
(entry, previous_path, previous_loc))
3153
check_valid_parent()
3156
# absent; should not occur anywhere else
3157
this_tree_map[file_id] = None, this_path
3158
elif minikind == 'r':
3159
# relocation, must occur at expected location
3160
this_tree_map[file_id] = tree_state[1], this_path
3162
this_tree_map[file_id] = this_path, this_path
3163
check_valid_parent()
3164
if absent_positions == tree_count:
3165
raise AssertionError(
3166
"entry %r has no data for any tree." % (entry,))
3167
if self._id_index is not None:
3168
for file_id, entry_keys in self._id_index.iteritems():
3169
for entry_key in entry_keys:
3170
if entry_key[2] != file_id:
3171
raise AssertionError(
3172
'file_id %r did not match entry key %s'
3173
% (file_id, entry_key))
3174
if len(entry_keys) != len(set(entry_keys)):
3175
raise AssertionError(
3176
'id_index contained non-unique data for %s'
3179
def _wipe_state(self):
3180
"""Forget all state information about the dirstate."""
3181
self._header_state = DirState.NOT_IN_MEMORY
3182
self._dirblock_state = DirState.NOT_IN_MEMORY
3183
self._changes_aborted = False
3186
self._dirblocks = []
3187
self._id_index = None
3188
self._packed_stat_index = None
3189
self._end_of_header = None
3190
self._cutoff_time = None
3191
self._split_path_cache = {}
3193
def lock_read(self):
3194
"""Acquire a read lock on the dirstate."""
3195
if self._lock_token is not None:
3196
raise errors.LockContention(self._lock_token)
3197
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3198
# already in memory, we could read just the header and check for
3199
# any modification. If not modified, we can just leave things
3201
self._lock_token = lock.ReadLock(self._filename)
3202
self._lock_state = 'r'
3203
self._state_file = self._lock_token.f
3206
def lock_write(self):
3207
"""Acquire a write lock on the dirstate."""
3208
if self._lock_token is not None:
3209
raise errors.LockContention(self._lock_token)
3210
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3211
# already in memory, we could read just the header and check for
3212
# any modification. If not modified, we can just leave things
3214
self._lock_token = lock.WriteLock(self._filename)
3215
self._lock_state = 'w'
3216
self._state_file = self._lock_token.f
3220
"""Drop any locks held on the dirstate."""
3221
if self._lock_token is None:
3222
raise errors.LockNotHeld(self)
3223
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
3224
# already in memory, we could read just the header and check for
3225
# any modification. If not modified, we can just leave things
3227
self._state_file = None
3228
self._lock_state = None
3229
self._lock_token.unlock()
3230
self._lock_token = None
3231
self._split_path_cache = {}
3233
def _requires_lock(self):
3234
"""Check that a lock is currently held by someone on the dirstate."""
3235
if not self._lock_token:
3236
raise errors.ObjectNotLocked(self)
3239
def py_update_entry(state, entry, abspath, stat_value,
3240
_stat_to_minikind=DirState._stat_to_minikind,
3241
_pack_stat=pack_stat):
3242
"""Update the entry based on what is actually on disk.
3244
This function only calculates the sha if it needs to - if the entry is
3245
uncachable, or clearly different to the first parent's entry, no sha
3246
is calculated, and None is returned.
3248
:param state: The dirstate this entry is in.
3249
:param entry: This is the dirblock entry for the file in question.
3250
:param abspath: The path on disk for this file.
3251
:param stat_value: The stat value done on the path.
3252
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
3253
target of a symlink.
3256
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
3260
packed_stat = _pack_stat(stat_value)
3261
(saved_minikind, saved_link_or_sha1, saved_file_size,
3262
saved_executable, saved_packed_stat) = entry[1][0]
3264
if minikind == 'd' and saved_minikind == 't':
3266
if (minikind == saved_minikind
3267
and packed_stat == saved_packed_stat):
3268
# The stat hasn't changed since we saved, so we can re-use the
3273
# size should also be in packed_stat
3274
if saved_file_size == stat_value.st_size:
3275
return saved_link_or_sha1
3277
# If we have gotten this far, that means that we need to actually
3278
# process this entry.
3282
executable = state._is_executable(stat_value.st_mode,
3284
if state._cutoff_time is None:
3285
state._sha_cutoff_time()
3286
if (stat_value.st_mtime < state._cutoff_time
3287
and stat_value.st_ctime < state._cutoff_time
3288
and len(entry[1]) > 1
3289
and entry[1][1][0] != 'a'):
3290
# Could check for size changes for further optimised
3291
# avoidance of sha1's. However the most prominent case of
3292
# over-shaing is during initial add, which this catches.
3293
# Besides, if content filtering happens, size and sha
3294
# are calculated at the same time, so checking just the size
3295
# gains nothing w.r.t. performance.
3296
link_or_sha1 = state._sha1_file(abspath)
3297
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
3298
executable, packed_stat)
3300
entry[1][0] = ('f', '', stat_value.st_size,
3301
executable, DirState.NULLSTAT)
3302
worth_saving = False
3303
elif minikind == 'd':
3305
entry[1][0] = ('d', '', 0, False, packed_stat)
3306
if saved_minikind != 'd':
3307
# This changed from something into a directory. Make sure we
3308
# have a directory block for it. This doesn't happen very
3309
# often, so this doesn't have to be super fast.
3310
block_index, entry_index, dir_present, file_present = \
3311
state._get_block_entry_index(entry[0][0], entry[0][1], 0)
3312
state._ensure_block(block_index, entry_index,
3313
osutils.pathjoin(entry[0][0], entry[0][1]))
3315
worth_saving = False
3316
elif minikind == 'l':
3317
if saved_minikind == 'l':
3318
worth_saving = False
3319
link_or_sha1 = state._read_link(abspath, saved_link_or_sha1)
3320
if state._cutoff_time is None:
3321
state._sha_cutoff_time()
3322
if (stat_value.st_mtime < state._cutoff_time
3323
and stat_value.st_ctime < state._cutoff_time):
3324
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
3327
entry[1][0] = ('l', '', stat_value.st_size,
3328
False, DirState.NULLSTAT)
3330
state._mark_modified([entry])
3334
class ProcessEntryPython(object):
3336
__slots__ = ["old_dirname_to_file_id", "new_dirname_to_file_id",
3337
"last_source_parent", "last_target_parent", "include_unchanged",
3338
"partial", "use_filesystem_for_exec", "utf8_decode",
3339
"searched_specific_files", "search_specific_files",
3340
"searched_exact_paths", "search_specific_file_parents", "seen_ids",
3341
"state", "source_index", "target_index", "want_unversioned", "tree"]
3343
def __init__(self, include_unchanged, use_filesystem_for_exec,
3344
search_specific_files, state, source_index, target_index,
3345
want_unversioned, tree):
3346
self.old_dirname_to_file_id = {}
3347
self.new_dirname_to_file_id = {}
3348
# Are we doing a partial iter_changes?
3349
self.partial = search_specific_files != set([''])
3350
# Using a list so that we can access the values and change them in
3351
# nested scope. Each one is [path, file_id, entry]
3352
self.last_source_parent = [None, None]
3353
self.last_target_parent = [None, None]
3354
self.include_unchanged = include_unchanged
3355
self.use_filesystem_for_exec = use_filesystem_for_exec
3356
self.utf8_decode = cache_utf8._utf8_decode
3357
# for all search_indexs in each path at or under each element of
3358
# search_specific_files, if the detail is relocated: add the id, and
3359
# add the relocated path as one to search if its not searched already.
3360
# If the detail is not relocated, add the id.
3361
self.searched_specific_files = set()
3362
# When we search exact paths without expanding downwards, we record
3364
self.searched_exact_paths = set()
3365
self.search_specific_files = search_specific_files
3366
# The parents up to the root of the paths we are searching.
3367
# After all normal paths are returned, these specific items are returned.
3368
self.search_specific_file_parents = set()
3369
# The ids we've sent out in the delta.
3370
self.seen_ids = set()
3372
self.source_index = source_index
3373
self.target_index = target_index
3374
if target_index != 0:
3375
# A lot of code in here depends on target_index == 0
3376
raise errors.BzrError('unsupported target index')
3377
self.want_unversioned = want_unversioned
3380
def _process_entry(self, entry, path_info, pathjoin=osutils.pathjoin):
3381
"""Compare an entry and real disk to generate delta information.
3383
:param path_info: top_relpath, basename, kind, lstat, abspath for
3384
the path of entry. If None, then the path is considered absent in
3385
the target (Perhaps we should pass in a concrete entry for this ?)
3386
Basename is returned as a utf8 string because we expect this
3387
tuple will be ignored, and don't want to take the time to
3389
:return: (iter_changes_result, changed). If the entry has not been
3390
handled then changed is None. Otherwise it is False if no content
3391
or metadata changes have occurred, and True if any content or
3392
metadata change has occurred. If self.include_unchanged is True then
3393
if changed is not None, iter_changes_result will always be a result
3394
tuple. Otherwise, iter_changes_result is None unless changed is
3397
if self.source_index is None:
3398
source_details = DirState.NULL_PARENT_DETAILS
3400
source_details = entry[1][self.source_index]
3401
target_details = entry[1][self.target_index]
3402
target_minikind = target_details[0]
3403
if path_info is not None and target_minikind in 'fdlt':
3404
if not (self.target_index == 0):
3405
raise AssertionError()
3406
link_or_sha1 = update_entry(self.state, entry,
3407
abspath=path_info[4], stat_value=path_info[3])
3408
# The entry may have been modified by update_entry
3409
target_details = entry[1][self.target_index]
3410
target_minikind = target_details[0]
3413
file_id = entry[0][2]
3414
source_minikind = source_details[0]
3415
if source_minikind in 'fdltr' and target_minikind in 'fdlt':
3416
# claimed content in both: diff
3417
# r | fdlt | | add source to search, add id path move and perform
3418
# | | | diff check on source-target
3419
# r | fdlt | a | dangling file that was present in the basis.
3421
if source_minikind in 'r':
3422
# add the source to the search path to find any children it
3423
# has. TODO ? : only add if it is a container ?
3424
if not osutils.is_inside_any(self.searched_specific_files,
3426
self.search_specific_files.add(source_details[1])
3427
# generate the old path; this is needed for stating later
3429
old_path = source_details[1]
3430
old_dirname, old_basename = os.path.split(old_path)
3431
path = pathjoin(entry[0][0], entry[0][1])
3432
old_entry = self.state._get_entry(self.source_index,
3434
# update the source details variable to be the real
3436
if old_entry == (None, None):
3437
raise errors.CorruptDirstate(self.state._filename,
3438
"entry '%s/%s' is considered renamed from %r"
3439
" but source does not exist\n"
3440
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
3441
source_details = old_entry[1][self.source_index]
3442
source_minikind = source_details[0]
3444
old_dirname = entry[0][0]
3445
old_basename = entry[0][1]
3446
old_path = path = None
3447
if path_info is None:
3448
# the file is missing on disk, show as removed.
3449
content_change = True
3453
# source and target are both versioned and disk file is present.
3454
target_kind = path_info[2]
3455
if target_kind == 'directory':
3457
old_path = path = pathjoin(old_dirname, old_basename)
3458
self.new_dirname_to_file_id[path] = file_id
3459
if source_minikind != 'd':
3460
content_change = True
3462
# directories have no fingerprint
3463
content_change = False
3465
elif target_kind == 'file':
3466
if source_minikind != 'f':
3467
content_change = True
3469
# Check the sha. We can't just rely on the size as
3470
# content filtering may mean differ sizes actually
3471
# map to the same content
3472
if link_or_sha1 is None:
3474
statvalue, link_or_sha1 = \
3475
self.state._sha1_provider.stat_and_sha1(
3477
self.state._observed_sha1(entry, link_or_sha1,
3479
content_change = (link_or_sha1 != source_details[1])
3480
# Target details is updated at update_entry time
3481
if self.use_filesystem_for_exec:
3482
# We don't need S_ISREG here, because we are sure
3483
# we are dealing with a file.
3484
target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
3486
target_exec = target_details[3]
3487
elif target_kind == 'symlink':
3488
if source_minikind != 'l':
3489
content_change = True
3491
content_change = (link_or_sha1 != source_details[1])
3493
elif target_kind == 'tree-reference':
3494
if source_minikind != 't':
3495
content_change = True
3497
content_change = False
3501
path = pathjoin(old_dirname, old_basename)
3502
raise errors.BadFileKindError(path, path_info[2])
3503
if source_minikind == 'd':
3505
old_path = path = pathjoin(old_dirname, old_basename)
3506
self.old_dirname_to_file_id[old_path] = file_id
3507
# parent id is the entry for the path in the target tree
3508
if old_basename and old_dirname == self.last_source_parent[0]:
3509
source_parent_id = self.last_source_parent[1]
3512
source_parent_id = self.old_dirname_to_file_id[old_dirname]
3514
source_parent_entry = self.state._get_entry(self.source_index,
3515
path_utf8=old_dirname)
3516
source_parent_id = source_parent_entry[0][2]
3517
if source_parent_id == entry[0][2]:
3518
# This is the root, so the parent is None
3519
source_parent_id = None
3521
self.last_source_parent[0] = old_dirname
3522
self.last_source_parent[1] = source_parent_id
3523
new_dirname = entry[0][0]
3524
if entry[0][1] and new_dirname == self.last_target_parent[0]:
3525
target_parent_id = self.last_target_parent[1]
3528
target_parent_id = self.new_dirname_to_file_id[new_dirname]
3530
# TODO: We don't always need to do the lookup, because the
3531
# parent entry will be the same as the source entry.
3532
target_parent_entry = self.state._get_entry(self.target_index,
3533
path_utf8=new_dirname)
3534
if target_parent_entry == (None, None):
3535
raise AssertionError(
3536
"Could not find target parent in wt: %s\nparent of: %s"
3537
% (new_dirname, entry))
3538
target_parent_id = target_parent_entry[0][2]
3539
if target_parent_id == entry[0][2]:
3540
# This is the root, so the parent is None
3541
target_parent_id = None
3543
self.last_target_parent[0] = new_dirname
3544
self.last_target_parent[1] = target_parent_id
3546
source_exec = source_details[3]
3547
changed = (content_change
3548
or source_parent_id != target_parent_id
3549
or old_basename != entry[0][1]
3550
or source_exec != target_exec
3552
if not changed and not self.include_unchanged:
3555
if old_path is None:
3556
old_path = path = pathjoin(old_dirname, old_basename)
3557
old_path_u = self.utf8_decode(old_path)[0]
3560
old_path_u = self.utf8_decode(old_path)[0]
3561
if old_path == path:
3564
path_u = self.utf8_decode(path)[0]
3565
source_kind = DirState._minikind_to_kind[source_minikind]
3566
return (entry[0][2],
3567
(old_path_u, path_u),
3570
(source_parent_id, target_parent_id),
3571
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
3572
(source_kind, target_kind),
3573
(source_exec, target_exec)), changed
3574
elif source_minikind in 'a' and target_minikind in 'fdlt':
3575
# looks like a new file
3576
path = pathjoin(entry[0][0], entry[0][1])
3577
# parent id is the entry for the path in the target tree
3578
# TODO: these are the same for an entire directory: cache em.
3579
parent_id = self.state._get_entry(self.target_index,
3580
path_utf8=entry[0][0])[0][2]
3581
if parent_id == entry[0][2]:
3583
if path_info is not None:
3585
if self.use_filesystem_for_exec:
3586
# We need S_ISREG here, because we aren't sure if this
3589
stat.S_ISREG(path_info[3].st_mode)
3590
and stat.S_IEXEC & path_info[3].st_mode)
3592
target_exec = target_details[3]
3593
return (entry[0][2],
3594
(None, self.utf8_decode(path)[0]),
3598
(None, self.utf8_decode(entry[0][1])[0]),
3599
(None, path_info[2]),
3600
(None, target_exec)), True
3602
# Its a missing file, report it as such.
3603
return (entry[0][2],
3604
(None, self.utf8_decode(path)[0]),
3608
(None, self.utf8_decode(entry[0][1])[0]),
3610
(None, False)), True
3611
elif source_minikind in 'fdlt' and target_minikind in 'a':
3612
# unversioned, possibly, or possibly not deleted: we dont care.
3613
# if its still on disk, *and* theres no other entry at this
3614
# path [we dont know this in this routine at the moment -
3615
# perhaps we should change this - then it would be an unknown.
3616
old_path = pathjoin(entry[0][0], entry[0][1])
3617
# parent id is the entry for the path in the target tree
3618
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
3619
if parent_id == entry[0][2]:
3621
return (entry[0][2],
3622
(self.utf8_decode(old_path)[0], None),
3626
(self.utf8_decode(entry[0][1])[0], None),
3627
(DirState._minikind_to_kind[source_minikind], None),
3628
(source_details[3], None)), True
3629
elif source_minikind in 'fdlt' and target_minikind in 'r':
3630
# a rename; could be a true rename, or a rename inherited from
3631
# a renamed parent. TODO: handle this efficiently. Its not
3632
# common case to rename dirs though, so a correct but slow
3633
# implementation will do.
3634
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
3635
self.search_specific_files.add(target_details[1])
3636
elif source_minikind in 'ra' and target_minikind in 'ra':
3637
# neither of the selected trees contain this file,
3638
# so skip over it. This is not currently directly tested, but
3639
# is indirectly via test_too_much.TestCommands.test_conflicts.
3642
raise AssertionError("don't know how to compare "
3643
"source_minikind=%r, target_minikind=%r"
3644
% (source_minikind, target_minikind))
3645
## import pdb;pdb.set_trace()
3651
def _gather_result_for_consistency(self, result):
3652
"""Check a result we will yield to make sure we are consistent later.
3654
This gathers result's parents into a set to output later.
3656
:param result: A result tuple.
3658
if not self.partial or not result[0]:
3660
self.seen_ids.add(result[0])
3661
new_path = result[1][1]
3663
# Not the root and not a delete: queue up the parents of the path.
3664
self.search_specific_file_parents.update(
3665
osutils.parent_directories(new_path.encode('utf8')))
3666
# Add the root directory which parent_directories does not
3668
self.search_specific_file_parents.add('')
3670
def iter_changes(self):
3671
"""Iterate over the changes."""
3672
utf8_decode = cache_utf8._utf8_decode
3673
_cmp_by_dirs = cmp_by_dirs
3674
_process_entry = self._process_entry
3675
search_specific_files = self.search_specific_files
3676
searched_specific_files = self.searched_specific_files
3677
splitpath = osutils.splitpath
3679
# compare source_index and target_index at or under each element of search_specific_files.
3680
# follow the following comparison table. Note that we only want to do diff operations when
3681
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
3685
# Source | Target | disk | action
3686
# r | fdlt | | add source to search, add id path move and perform
3687
# | | | diff check on source-target
3688
# r | fdlt | a | dangling file that was present in the basis.
3690
# r | a | | add source to search
3692
# r | r | | this path is present in a non-examined tree, skip.
3693
# r | r | a | this path is present in a non-examined tree, skip.
3694
# a | fdlt | | add new id
3695
# a | fdlt | a | dangling locally added file, skip
3696
# a | a | | not present in either tree, skip
3697
# a | a | a | not present in any tree, skip
3698
# a | r | | not present in either tree at this path, skip as it
3699
# | | | may not be selected by the users list of paths.
3700
# a | r | a | not present in either tree at this path, skip as it
3701
# | | | may not be selected by the users list of paths.
3702
# fdlt | fdlt | | content in both: diff them
3703
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
3704
# fdlt | a | | unversioned: output deleted id for now
3705
# fdlt | a | a | unversioned and deleted: output deleted id
3706
# fdlt | r | | relocated in this tree, so add target to search.
3707
# | | | Dont diff, we will see an r,fd; pair when we reach
3708
# | | | this id at the other path.
3709
# fdlt | r | a | relocated in this tree, so add target to search.
3710
# | | | Dont diff, we will see an r,fd; pair when we reach
3711
# | | | this id at the other path.
3713
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
3714
# keeping a cache of directories that we have seen.
3716
while search_specific_files:
3717
# TODO: the pending list should be lexically sorted? the
3718
# interface doesn't require it.
3719
current_root = search_specific_files.pop()
3720
current_root_unicode = current_root.decode('utf8')
3721
searched_specific_files.add(current_root)
3722
# process the entries for this containing directory: the rest will be
3723
# found by their parents recursively.
3724
root_entries = self.state._entries_for_path(current_root)
3725
root_abspath = self.tree.abspath(current_root_unicode)
3727
root_stat = os.lstat(root_abspath)
3729
if e.errno == errno.ENOENT:
3730
# the path does not exist: let _process_entry know that.
3731
root_dir_info = None
3733
# some other random error: hand it up.
3736
root_dir_info = ('', current_root,
3737
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
3739
if root_dir_info[2] == 'directory':
3740
if self.tree._directory_is_tree_reference(
3741
current_root.decode('utf8')):
3742
root_dir_info = root_dir_info[:2] + \
3743
('tree-reference',) + root_dir_info[3:]
3745
if not root_entries and not root_dir_info:
3746
# this specified path is not present at all, skip it.
3748
path_handled = False
3749
for entry in root_entries:
3750
result, changed = _process_entry(entry, root_dir_info)
3751
if changed is not None:
3754
self._gather_result_for_consistency(result)
3755
if changed or self.include_unchanged:
3757
if self.want_unversioned and not path_handled and root_dir_info:
3758
new_executable = bool(
3759
stat.S_ISREG(root_dir_info[3].st_mode)
3760
and stat.S_IEXEC & root_dir_info[3].st_mode)
3762
(None, current_root_unicode),
3766
(None, splitpath(current_root_unicode)[-1]),
3767
(None, root_dir_info[2]),
3768
(None, new_executable)
3770
initial_key = (current_root, '', '')
3771
block_index, _ = self.state._find_block_index_from_key(initial_key)
3772
if block_index == 0:
3773
# we have processed the total root already, but because the
3774
# initial key matched it we should skip it here.
3776
if root_dir_info and root_dir_info[2] == 'tree-reference':
3777
current_dir_info = None
3779
dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
3781
current_dir_info = dir_iterator.next()
3783
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
3784
# python 2.5 has e.errno == EINVAL,
3785
# and e.winerror == ERROR_DIRECTORY
3786
e_winerror = getattr(e, 'winerror', None)
3787
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
3788
# there may be directories in the inventory even though
3789
# this path is not a file on disk: so mark it as end of
3791
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
3792
current_dir_info = None
3793
elif (sys.platform == 'win32'
3794
and (e.errno in win_errors
3795
or e_winerror in win_errors)):
3796
current_dir_info = None
3800
if current_dir_info[0][0] == '':
3801
# remove .bzr from iteration
3802
bzr_index = bisect.bisect_left(current_dir_info[1], ('.bzr',))
3803
if current_dir_info[1][bzr_index][0] != '.bzr':
3804
raise AssertionError()
3805
del current_dir_info[1][bzr_index]
3806
# walk until both the directory listing and the versioned metadata
3808
if (block_index < len(self.state._dirblocks) and
3809
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
3810
current_block = self.state._dirblocks[block_index]
3812
current_block = None
3813
while (current_dir_info is not None or
3814
current_block is not None):
3815
if (current_dir_info and current_block
3816
and current_dir_info[0][0] != current_block[0]):
3817
if _cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
3818
# filesystem data refers to paths not covered by the dirblock.
3819
# this has two possibilities:
3820
# A) it is versioned but empty, so there is no block for it
3821
# B) it is not versioned.
3823
# if (A) then we need to recurse into it to check for
3824
# new unknown files or directories.
3825
# if (B) then we should ignore it, because we don't
3826
# recurse into unknown directories.
3828
while path_index < len(current_dir_info[1]):
3829
current_path_info = current_dir_info[1][path_index]
3830
if self.want_unversioned:
3831
if current_path_info[2] == 'directory':
3832
if self.tree._directory_is_tree_reference(
3833
current_path_info[0].decode('utf8')):
3834
current_path_info = current_path_info[:2] + \
3835
('tree-reference',) + current_path_info[3:]
3836
new_executable = bool(
3837
stat.S_ISREG(current_path_info[3].st_mode)
3838
and stat.S_IEXEC & current_path_info[3].st_mode)
3840
(None, utf8_decode(current_path_info[0])[0]),
3844
(None, utf8_decode(current_path_info[1])[0]),
3845
(None, current_path_info[2]),
3846
(None, new_executable))
3847
# dont descend into this unversioned path if it is
3849
if current_path_info[2] in ('directory',
3851
del current_dir_info[1][path_index]
3855
# This dir info has been handled, go to the next
3857
current_dir_info = dir_iterator.next()
3858
except StopIteration:
3859
current_dir_info = None
3861
# We have a dirblock entry for this location, but there
3862
# is no filesystem path for this. This is most likely
3863
# because a directory was removed from the disk.
3864
# We don't have to report the missing directory,
3865
# because that should have already been handled, but we
3866
# need to handle all of the files that are contained
3868
for current_entry in current_block[1]:
3869
# entry referring to file not present on disk.
3870
# advance the entry only, after processing.
3871
result, changed = _process_entry(current_entry, None)
3872
if changed is not None:
3874
self._gather_result_for_consistency(result)
3875
if changed or self.include_unchanged:
3878
if (block_index < len(self.state._dirblocks) and
3879
osutils.is_inside(current_root,
3880
self.state._dirblocks[block_index][0])):
3881
current_block = self.state._dirblocks[block_index]
3883
current_block = None
3886
if current_block and entry_index < len(current_block[1]):
3887
current_entry = current_block[1][entry_index]
3889
current_entry = None
3890
advance_entry = True
3892
if current_dir_info and path_index < len(current_dir_info[1]):
3893
current_path_info = current_dir_info[1][path_index]
3894
if current_path_info[2] == 'directory':
3895
if self.tree._directory_is_tree_reference(
3896
current_path_info[0].decode('utf8')):
3897
current_path_info = current_path_info[:2] + \
3898
('tree-reference',) + current_path_info[3:]
3900
current_path_info = None
3902
path_handled = False
3903
while (current_entry is not None or
3904
current_path_info is not None):
3905
if current_entry is None:
3906
# the check for path_handled when the path is advanced
3907
# will yield this path if needed.
3909
elif current_path_info is None:
3910
# no path is fine: the per entry code will handle it.
3911
result, changed = _process_entry(current_entry, current_path_info)
3912
if changed is not None:
3914
self._gather_result_for_consistency(result)
3915
if changed or self.include_unchanged:
3917
elif (current_entry[0][1] != current_path_info[1]
3918
or current_entry[1][self.target_index][0] in 'ar'):
3919
# The current path on disk doesn't match the dirblock
3920
# record. Either the dirblock is marked as absent, or
3921
# the file on disk is not present at all in the
3922
# dirblock. Either way, report about the dirblock
3923
# entry, and let other code handle the filesystem one.
3925
# Compare the basename for these files to determine
3927
if current_path_info[1] < current_entry[0][1]:
3928
# extra file on disk: pass for now, but only
3929
# increment the path, not the entry
3930
advance_entry = False
3932
# entry referring to file not present on disk.
3933
# advance the entry only, after processing.
3934
result, changed = _process_entry(current_entry, None)
3935
if changed is not None:
3937
self._gather_result_for_consistency(result)
3938
if changed or self.include_unchanged:
3940
advance_path = False
3942
result, changed = _process_entry(current_entry, current_path_info)
3943
if changed is not None:
3946
self._gather_result_for_consistency(result)
3947
if changed or self.include_unchanged:
3949
if advance_entry and current_entry is not None:
3951
if entry_index < len(current_block[1]):
3952
current_entry = current_block[1][entry_index]
3954
current_entry = None
3956
advance_entry = True # reset the advance flaga
3957
if advance_path and current_path_info is not None:
3958
if not path_handled:
3959
# unversioned in all regards
3960
if self.want_unversioned:
3961
new_executable = bool(
3962
stat.S_ISREG(current_path_info[3].st_mode)
3963
and stat.S_IEXEC & current_path_info[3].st_mode)
3965
relpath_unicode = utf8_decode(current_path_info[0])[0]
3966
except UnicodeDecodeError:
3967
raise errors.BadFilenameEncoding(
3968
current_path_info[0], osutils._fs_enc)
3970
(None, relpath_unicode),
3974
(None, utf8_decode(current_path_info[1])[0]),
3975
(None, current_path_info[2]),
3976
(None, new_executable))
3977
# dont descend into this unversioned path if it is
3979
if current_path_info[2] in ('directory'):
3980
del current_dir_info[1][path_index]
3982
# dont descend the disk iterator into any tree
3984
if current_path_info[2] == 'tree-reference':
3985
del current_dir_info[1][path_index]
3988
if path_index < len(current_dir_info[1]):
3989
current_path_info = current_dir_info[1][path_index]
3990
if current_path_info[2] == 'directory':
3991
if self.tree._directory_is_tree_reference(
3992
current_path_info[0].decode('utf8')):
3993
current_path_info = current_path_info[:2] + \
3994
('tree-reference',) + current_path_info[3:]
3996
current_path_info = None
3997
path_handled = False
3999
advance_path = True # reset the advance flagg.
4000
if current_block is not None:
4002
if (block_index < len(self.state._dirblocks) and
4003
osutils.is_inside(current_root, self.state._dirblocks[block_index][0])):
4004
current_block = self.state._dirblocks[block_index]
4006
current_block = None
4007
if current_dir_info is not None:
4009
current_dir_info = dir_iterator.next()
4010
except StopIteration:
4011
current_dir_info = None
4012
for result in self._iter_specific_file_parents():
4015
def _iter_specific_file_parents(self):
4016
"""Iter over the specific file parents."""
4017
while self.search_specific_file_parents:
4018
# Process the parent directories for the paths we were iterating.
4019
# Even in extremely large trees this should be modest, so currently
4020
# no attempt is made to optimise.
4021
path_utf8 = self.search_specific_file_parents.pop()
4022
if osutils.is_inside_any(self.searched_specific_files, path_utf8):
4023
# We've examined this path.
4025
if path_utf8 in self.searched_exact_paths:
4026
# We've examined this path.
4028
path_entries = self.state._entries_for_path(path_utf8)
4029
# We need either one or two entries. If the path in
4030
# self.target_index has moved (so the entry in source_index is in
4031
# 'ar') then we need to also look for the entry for this path in
4032
# self.source_index, to output the appropriate delete-or-rename.
4033
selected_entries = []
4035
for candidate_entry in path_entries:
4036
# Find entries present in target at this path:
4037
if candidate_entry[1][self.target_index][0] not in 'ar':
4039
selected_entries.append(candidate_entry)
4040
# Find entries present in source at this path:
4041
elif (self.source_index is not None and
4042
candidate_entry[1][self.source_index][0] not in 'ar'):
4044
if candidate_entry[1][self.target_index][0] == 'a':
4045
# Deleted, emit it here.
4046
selected_entries.append(candidate_entry)
4048
# renamed, emit it when we process the directory it
4050
self.search_specific_file_parents.add(
4051
candidate_entry[1][self.target_index][1])
4053
raise AssertionError(
4054
"Missing entry for specific path parent %r, %r" % (
4055
path_utf8, path_entries))
4056
path_info = self._path_info(path_utf8, path_utf8.decode('utf8'))
4057
for entry in selected_entries:
4058
if entry[0][2] in self.seen_ids:
4060
result, changed = self._process_entry(entry, path_info)
4062
raise AssertionError(
4063
"Got entry<->path mismatch for specific path "
4064
"%r entry %r path_info %r " % (
4065
path_utf8, entry, path_info))
4066
# Only include changes - we're outside the users requested
4069
self._gather_result_for_consistency(result)
4070
if (result[6][0] == 'directory' and
4071
result[6][1] != 'directory'):
4072
# This stopped being a directory, the old children have
4074
if entry[1][self.source_index][0] == 'r':
4075
# renamed, take the source path
4076
entry_path_utf8 = entry[1][self.source_index][1]
4078
entry_path_utf8 = path_utf8
4079
initial_key = (entry_path_utf8, '', '')
4080
block_index, _ = self.state._find_block_index_from_key(
4082
if block_index == 0:
4083
# The children of the root are in block index 1.
4085
current_block = None
4086
if block_index < len(self.state._dirblocks):
4087
current_block = self.state._dirblocks[block_index]
4088
if not osutils.is_inside(
4089
entry_path_utf8, current_block[0]):
4090
# No entries for this directory at all.
4091
current_block = None
4092
if current_block is not None:
4093
for entry in current_block[1]:
4094
if entry[1][self.source_index][0] in 'ar':
4095
# Not in the source tree, so doesn't have to be
4098
# Path of the entry itself.
4100
self.search_specific_file_parents.add(
4101
osutils.pathjoin(*entry[0][:2]))
4102
if changed or self.include_unchanged:
4104
self.searched_exact_paths.add(path_utf8)
4106
def _path_info(self, utf8_path, unicode_path):
4107
"""Generate path_info for unicode_path.
4109
:return: None if unicode_path does not exist, or a path_info tuple.
4111
abspath = self.tree.abspath(unicode_path)
4113
stat = os.lstat(abspath)
4115
if e.errno == errno.ENOENT:
4116
# the path does not exist.
4120
utf8_basename = utf8_path.rsplit('/', 1)[-1]
4121
dir_info = (utf8_path, utf8_basename,
4122
osutils.file_kind_from_stat_mode(stat.st_mode), stat,
4124
if dir_info[2] == 'directory':
4125
if self.tree._directory_is_tree_reference(
4127
self.root_dir_info = self.root_dir_info[:2] + \
4128
('tree-reference',) + self.root_dir_info[3:]
4132
# Try to load the compiled form if possible
4134
from bzrlib._dirstate_helpers_pyx import (
4140
ProcessEntryC as _process_entry,
4141
update_entry as update_entry,
4143
except ImportError, e:
4144
osutils.failed_to_load_extension(e)
4145
from bzrlib._dirstate_helpers_py import (
4152
# FIXME: It would be nice to be able to track moved lines so that the
4153
# corresponding python code can be moved to the _dirstate_helpers_py
4154
# module. I don't want to break the history for this important piece of
4155
# code so I left the code here -- vila 20090622
4156
update_entry = py_update_entry
4157
_process_entry = ProcessEntryPython
added
removed
bzrlib/dirstate.py
