1
# Copyright (C) 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but its a file. The fingerprint is a
87
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
89
't' is a reference to a nested subtree; the fingerprint is the referenced
94
The entries on disk and in memory are ordered according to the following keys:
96
directory, as a list of components
100
--- Format 1 had the following different definition: ---
101
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
102
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
104
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
105
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
108
PARENT ROW's are emitted for every parent that is not in the ghosts details
109
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
110
each row will have a PARENT ROW for foo and baz, but not for bar.
113
In any tree, a kind of 'moved' indicates that the fingerprint field
114
(which we treat as opaque data specific to the 'kind' anyway) has the
115
details for the id of this row in that tree.
117
I'm strongly tempted to add a id->path index as well, but I think that
118
where we need id->path mapping; we also usually read the whole file, so
119
I'm going to skip that for the moment, as we have the ability to locate
120
via bisect any path in any tree, and if we lookup things by path, we can
121
accumulate an id->path mapping as we go, which will tend to match what we
124
I plan to implement this asap, so please speak up now to alter/tweak the
125
design - and once we stabilise on this, I'll update the wiki page for
128
The rationale for all this is that we want fast operations for the
129
common case (diff/status/commit/merge on all files) and extremely fast
130
operations for the less common but still occurs a lot status/diff/commit
131
on specific files). Operations on specific files involve a scan for all
132
the children of a path, *in every involved tree*, which the current
133
format did not accommodate.
137
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
138
2) fall back current object model as needed.
139
3) scale usably to the largest trees known today - say 50K entries. (mozilla
140
is an example of this)
144
Eventually reuse dirstate objects across locks IFF the dirstate file has not
145
been modified, but will require that we flush/ignore cached stat-hit data
146
because we won't want to restat all files on disk just because a lock was
147
acquired, yet we cannot trust the data after the previous lock was released.
149
Memory representation:
150
vector of all directories, and vector of the childen ?
152
root_entrie = (direntry for root, [parent_direntries_for_root]),
154
('', ['data for achild', 'data for bchild', 'data for cchild'])
155
('dir', ['achild', 'cchild', 'echild'])
157
- single bisect to find N subtrees from a path spec
158
- in-order for serialisation - this is 'dirblock' grouping.
159
- insertion of a file '/a' affects only the '/' child-vector, that is, to
160
insert 10K elements from scratch does not generates O(N^2) memoves of a
161
single vector, rather each individual, which tends to be limited to a
162
manageable number. Will scale badly on trees with 10K entries in a
163
single directory. compare with Inventory.InventoryDirectory which has
164
a dictionary for the children. No bisect capability, can only probe for
165
exact matches, or grab all elements and sort.
166
- What's the risk of error here? Once we have the base format being processed
167
we should have a net win regardless of optimality. So we are going to
168
go with what seems reasonable.
171
Maybe we should do a test profile of the core structure - 10K simulated
172
searches/lookups/etc?
174
Objects for each row?
175
The lifetime of Dirstate objects is current per lock, but see above for
176
possible extensions. The lifetime of a row from a dirstate is expected to be
177
very short in the optimistic case: which we are optimising for. For instance,
178
subtree status will determine from analysis of the disk data what rows need to
179
be examined at all, and will be able to determine from a single row whether
180
that file has altered or not, so we are aiming to process tens of thousands of
181
entries each second within the dirstate context, before exposing anything to
182
the larger codebase. This suggests we want the time for a single file
183
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
184
processed, and to scale to 100 thousand we'll another order of magnitude to do
185
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
186
the file on disk, and then immediately discard, the overhead of object creation
187
becomes a significant cost.
189
Figures: Creating a tuple from from 3 elements was profiled at 0.0625
190
microseconds, whereas creating a object which is subclassed from tuple was
191
0.500 microseconds, and creating an object with 3 elements and slots was 3
192
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
193
down to 10 microseconds for the total processing - having 33% of that be object
194
creation is a huge overhead. There is a potential cost in using tuples within
195
each row which is that the conditional code to do comparisons may be slower
196
than method invocation, but method invocation is known to be slow due to stack
197
frame creation, so avoiding methods in these tight inner loops in unfortunately
198
desirable. We can consider a pyrex version of this with objects in future if
207
from stat import S_IEXEC
225
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
226
"""Convert stat values into a packed representation."""
227
# jam 20060614 it isn't really worth removing more entries if we
228
# are going to leave it in packed form.
229
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
230
# With all entries, filesize is 5.9M and read time is maybe 280ms
231
# well within the noise margin
233
# base64 encoding always adds a final newline, so strip it off
234
# The current version
235
return _encode(_pack('>LLLLLL'
236
, st.st_size, int(st.st_mtime), int(st.st_ctime)
237
, st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
238
# This is 0.060s / 1.520s faster by not encoding as much information
239
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
240
# This is not strictly faster than _encode(_pack())[:-1]
241
# return '%X.%X.%X.%X.%X.%X' % (
242
# st.st_size, int(st.st_mtime), int(st.st_ctime),
243
# st.st_dev, st.st_ino, st.st_mode)
244
# Similar to the _encode(_pack('>LL'))
245
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
248
class DirState(object):
249
"""Record directory and metadata state for fast access.
251
A dirstate is a specialised data structure for managing local working
252
tree state information. Its not yet well defined whether it is platform
253
specific, and if it is how we detect/parameterise that.
255
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
256
Unlike most bzr disk formats, DirStates must be locked for reading, using
257
lock_read. (This is an os file lock internally.) This is necessary
258
because the file can be rewritten in place.
260
DirStates must be explicitly written with save() to commit changes; just
261
unlocking them does not write the changes to disk.
264
_kind_to_minikind = {
270
'tree-reference': 't',
272
_minikind_to_kind = {
278
't': 'tree-reference',
280
_stat_to_minikind = {
285
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
286
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
289
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
290
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
291
# that we are sure a single record will always fit.
292
BISECT_PAGE_SIZE = 4096
295
IN_MEMORY_UNMODIFIED = 1
296
IN_MEMORY_MODIFIED = 2
298
# A pack_stat (the x's) that is just noise and will never match the output
301
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
303
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
304
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
306
def __init__(self, path):
307
"""Create a DirState object.
309
:param path: The path at which the dirstate file on disk should live.
311
# _header_state and _dirblock_state represent the current state
312
# of the dirstate metadata and the per-row data respectiely.
313
# NOT_IN_MEMORY indicates that no data is in memory
314
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
315
# is the same as is on disk
316
# IN_MEMORY_MODIFIED indicates that we have a modified version
317
# of what is on disk.
318
# In future we will add more granularity, for instance _dirblock_state
319
# will probably support partially-in-memory as a separate variable,
320
# allowing for partially-in-memory unmodified and partially-in-memory
322
self._header_state = DirState.NOT_IN_MEMORY
323
self._dirblock_state = DirState.NOT_IN_MEMORY
327
self._state_file = None
328
self._filename = path
329
self._lock_token = None
330
self._lock_state = None
331
self._id_index = None
332
# a map from packed_stat to sha's.
333
self._packed_stat_index = None
334
self._end_of_header = None
335
self._cutoff_time = None
336
self._split_path_cache = {}
337
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
338
if 'hashcache' in debug.debug_flags:
339
self._sha1_file = self._sha1_file_and_mutter
341
self._sha1_file = osutils.sha_file_by_name
342
# These two attributes provide a simple cache for lookups into the
343
# dirstate in-memory vectors. By probing respectively for the last
344
# block, and for the next entry, we save nearly 2 bisections per path
346
self._last_block_index = None
347
self._last_entry_index = None
351
(self.__class__.__name__, self._filename)
353
def add(self, path, file_id, kind, stat, fingerprint):
354
"""Add a path to be tracked.
356
:param path: The path within the dirstate - '' is the root, 'foo' is the
357
path foo within the root, 'foo/bar' is the path bar within foo
359
:param file_id: The file id of the path being added.
360
:param kind: The kind of the path, as a string like 'file',
362
:param stat: The output of os.lstat for the path.
363
:param fingerprint: The sha value of the file,
364
or the target of a symlink,
365
or the referenced revision id for tree-references,
366
or '' for directories.
369
# find the block its in.
370
# find the location in the block.
371
# check its not there
373
#------- copied from inventory.ensure_normalized_name - keep synced.
374
# --- normalized_filename wants a unicode basename only, so get one.
375
dirname, basename = osutils.split(path)
376
# we dont import normalized_filename directly because we want to be
377
# able to change the implementation at runtime for tests.
378
norm_name, can_access = osutils.normalized_filename(basename)
379
if norm_name != basename:
383
raise errors.InvalidNormalization(path)
384
# you should never have files called . or ..; just add the directory
385
# in the parent, or according to the special treatment for the root
386
if basename == '.' or basename == '..':
387
raise errors.InvalidEntryName(path)
388
# now that we've normalised, we need the correct utf8 path and
389
# dirname and basename elements. This single encode and split should be
390
# faster than three separate encodes.
391
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
392
dirname, basename = osutils.split(utf8path)
393
assert file_id.__class__ == str, \
394
"must be a utf8 file_id not %s" % (type(file_id))
395
# Make sure the file_id does not exist in this tree
396
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
397
if file_id_entry != (None, None):
398
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
399
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
400
info = '%s:%s' % (kind, path)
401
raise errors.DuplicateFileId(file_id, info)
402
first_key = (dirname, basename, '')
403
block_index, present = self._find_block_index_from_key(first_key)
405
# check the path is not in the tree
406
block = self._dirblocks[block_index][1]
407
entry_index, _ = self._find_entry_index(first_key, block)
408
while (entry_index < len(block) and
409
block[entry_index][0][0:2] == first_key[0:2]):
410
if block[entry_index][1][0][0] not in 'ar':
411
# this path is in the dirstate in the current tree.
412
raise Exception, "adding already added path!"
415
# The block where we want to put the file is not present. But it
416
# might be because the directory was empty, or not loaded yet. Look
417
# for a parent entry, if not found, raise NotVersionedError
418
parent_dir, parent_base = osutils.split(dirname)
419
parent_block_idx, parent_entry_idx, _, parent_present = \
420
self._get_block_entry_index(parent_dir, parent_base, 0)
421
if not parent_present:
422
raise errors.NotVersionedError(path, str(self))
423
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
424
block = self._dirblocks[block_index][1]
425
entry_key = (dirname, basename, file_id)
428
packed_stat = DirState.NULLSTAT
431
packed_stat = pack_stat(stat)
432
parent_info = self._empty_parent_info()
433
minikind = DirState._kind_to_minikind[kind]
435
entry_data = entry_key, [
436
(minikind, fingerprint, size, False, packed_stat),
438
elif kind == 'directory':
439
entry_data = entry_key, [
440
(minikind, '', 0, False, packed_stat),
442
elif kind == 'symlink':
443
entry_data = entry_key, [
444
(minikind, fingerprint, size, False, packed_stat),
446
elif kind == 'tree-reference':
447
entry_data = entry_key, [
448
(minikind, fingerprint, 0, False, packed_stat),
451
raise errors.BzrError('unknown kind %r' % kind)
452
entry_index, present = self._find_entry_index(entry_key, block)
454
block.insert(entry_index, entry_data)
456
assert block[entry_index][1][0][0] == 'a', " %r(%r) already added" % (basename, file_id)
457
block[entry_index][1][0] = entry_data[1][0]
459
if kind == 'directory':
460
# insert a new dirblock
461
self._ensure_block(block_index, entry_index, utf8path)
462
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
464
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
466
def _bisect(self, paths):
467
"""Bisect through the disk structure for specific rows.
469
:param paths: A list of paths to find
470
:return: A dict mapping path => entries for found entries. Missing
471
entries will not be in the map.
472
The list is not sorted, and entries will be populated
473
based on when they were read.
475
self._requires_lock()
476
# We need the file pointer to be right after the initial header block
477
self._read_header_if_needed()
478
# If _dirblock_state was in memory, we should just return info from
479
# there, this function is only meant to handle when we want to read
481
assert self._dirblock_state == DirState.NOT_IN_MEMORY
483
# The disk representation is generally info + '\0\n\0' at the end. But
484
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
485
# Because it means we can sync on the '\n'
486
state_file = self._state_file
487
file_size = os.fstat(state_file.fileno()).st_size
488
# We end up with 2 extra fields, we should have a trailing '\n' to
489
# ensure that we read the whole record, and we should have a precursur
490
# '' which ensures that we start after the previous '\n'
491
entry_field_count = self._fields_per_entry() + 1
493
low = self._end_of_header
494
high = file_size - 1 # Ignore the final '\0'
495
# Map from (dir, name) => entry
498
# Avoid infinite seeking
499
max_count = 30*len(paths)
501
# pending is a list of places to look.
502
# each entry is a tuple of low, high, dir_names
503
# low -> the first byte offset to read (inclusive)
504
# high -> the last byte offset (inclusive)
505
# dir_names -> The list of (dir, name) pairs that should be found in
506
# the [low, high] range
507
pending = [(low, high, paths)]
509
page_size = self._bisect_page_size
511
fields_to_entry = self._get_fields_to_entry()
514
low, high, cur_files = pending.pop()
516
if not cur_files or low >= high:
521
if count > max_count:
522
raise errors.BzrError('Too many seeks, most likely a bug.')
524
mid = max(low, (low+high-page_size)/2)
527
# limit the read size, so we don't end up reading data that we have
529
read_size = min(page_size, (high-mid)+1)
530
block = state_file.read(read_size)
533
entries = block.split('\n')
536
# We didn't find a '\n', so we cannot have found any records.
537
# So put this range back and try again. But we know we have to
538
# increase the page size, because a single read did not contain
539
# a record break (so records must be larger than page_size)
541
pending.append((low, high, cur_files))
544
# Check the first and last entries, in case they are partial, or if
545
# we don't care about the rest of this page
547
first_fields = entries[0].split('\0')
548
if len(first_fields) < entry_field_count:
549
# We didn't get the complete first entry
550
# so move start, and grab the next, which
551
# should be a full entry
552
start += len(entries[0])+1
553
first_fields = entries[1].split('\0')
556
if len(first_fields) <= 2:
557
# We didn't even get a filename here... what do we do?
558
# Try a large page size and repeat this query
560
pending.append((low, high, cur_files))
563
# Find what entries we are looking for, which occur before and
564
# after this first record.
567
first_path = first_fields[1] + '/' + first_fields[2]
569
first_path = first_fields[2]
570
first_loc = _bisect_path_left(cur_files, first_path)
572
# These exist before the current location
573
pre = cur_files[:first_loc]
574
# These occur after the current location, which may be in the
575
# data we read, or might be after the last entry
576
post = cur_files[first_loc:]
578
if post and len(first_fields) >= entry_field_count:
579
# We have files after the first entry
581
# Parse the last entry
582
last_entry_num = len(entries)-1
583
last_fields = entries[last_entry_num].split('\0')
584
if len(last_fields) < entry_field_count:
585
# The very last hunk was not complete,
586
# read the previous hunk
587
after = mid + len(block) - len(entries[-1])
589
last_fields = entries[last_entry_num].split('\0')
591
after = mid + len(block)
594
last_path = last_fields[1] + '/' + last_fields[2]
596
last_path = last_fields[2]
597
last_loc = _bisect_path_right(post, last_path)
599
middle_files = post[:last_loc]
600
post = post[last_loc:]
603
# We have files that should occur in this block
604
# (>= first, <= last)
605
# Either we will find them here, or we can mark them as
608
if middle_files[0] == first_path:
609
# We might need to go before this location
610
pre.append(first_path)
611
if middle_files[-1] == last_path:
612
post.insert(0, last_path)
614
# Find out what paths we have
615
paths = {first_path:[first_fields]}
616
# last_path might == first_path so we need to be
617
# careful if we should append rather than overwrite
618
if last_entry_num != first_entry_num:
619
paths.setdefault(last_path, []).append(last_fields)
620
for num in xrange(first_entry_num+1, last_entry_num):
621
# TODO: jam 20070223 We are already splitting here, so
622
# shouldn't we just split the whole thing rather
623
# than doing the split again in add_one_record?
624
fields = entries[num].split('\0')
626
path = fields[1] + '/' + fields[2]
629
paths.setdefault(path, []).append(fields)
631
for path in middle_files:
632
for fields in paths.get(path, []):
633
# offset by 1 because of the opening '\0'
634
# consider changing fields_to_entry to avoid the
636
entry = fields_to_entry(fields[1:])
637
found.setdefault(path, []).append(entry)
639
# Now we have split up everything into pre, middle, and post, and
640
# we have handled everything that fell in 'middle'.
641
# We add 'post' first, so that we prefer to seek towards the
642
# beginning, so that we will tend to go as early as we need, and
643
# then only seek forward after that.
645
pending.append((after, high, post))
647
pending.append((low, start-1, pre))
649
# Consider that we may want to return the directory entries in sorted
650
# order. For now, we just return them in whatever order we found them,
651
# and leave it up to the caller if they care if it is ordered or not.
654
def _bisect_dirblocks(self, dir_list):
655
"""Bisect through the disk structure to find entries in given dirs.
657
_bisect_dirblocks is meant to find the contents of directories, which
658
differs from _bisect, which only finds individual entries.
660
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
661
:return: A map from dir => entries_for_dir
663
# TODO: jam 20070223 A lot of the bisecting logic could be shared
664
# between this and _bisect. It would require parameterizing the
665
# inner loop with a function, though. We should evaluate the
666
# performance difference.
667
self._requires_lock()
668
# We need the file pointer to be right after the initial header block
669
self._read_header_if_needed()
670
# If _dirblock_state was in memory, we should just return info from
671
# there, this function is only meant to handle when we want to read
673
assert self._dirblock_state == DirState.NOT_IN_MEMORY
675
# The disk representation is generally info + '\0\n\0' at the end. But
676
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
677
# Because it means we can sync on the '\n'
678
state_file = self._state_file
679
file_size = os.fstat(state_file.fileno()).st_size
680
# We end up with 2 extra fields, we should have a trailing '\n' to
681
# ensure that we read the whole record, and we should have a precursur
682
# '' which ensures that we start after the previous '\n'
683
entry_field_count = self._fields_per_entry() + 1
685
low = self._end_of_header
686
high = file_size - 1 # Ignore the final '\0'
687
# Map from dir => entry
690
# Avoid infinite seeking
691
max_count = 30*len(dir_list)
693
# pending is a list of places to look.
694
# each entry is a tuple of low, high, dir_names
695
# low -> the first byte offset to read (inclusive)
696
# high -> the last byte offset (inclusive)
697
# dirs -> The list of directories that should be found in
698
# the [low, high] range
699
pending = [(low, high, dir_list)]
701
page_size = self._bisect_page_size
703
fields_to_entry = self._get_fields_to_entry()
706
low, high, cur_dirs = pending.pop()
708
if not cur_dirs or low >= high:
713
if count > max_count:
714
raise errors.BzrError('Too many seeks, most likely a bug.')
716
mid = max(low, (low+high-page_size)/2)
719
# limit the read size, so we don't end up reading data that we have
721
read_size = min(page_size, (high-mid)+1)
722
block = state_file.read(read_size)
725
entries = block.split('\n')
728
# We didn't find a '\n', so we cannot have found any records.
729
# So put this range back and try again. But we know we have to
730
# increase the page size, because a single read did not contain
731
# a record break (so records must be larger than page_size)
733
pending.append((low, high, cur_dirs))
736
# Check the first and last entries, in case they are partial, or if
737
# we don't care about the rest of this page
739
first_fields = entries[0].split('\0')
740
if len(first_fields) < entry_field_count:
741
# We didn't get the complete first entry
742
# so move start, and grab the next, which
743
# should be a full entry
744
start += len(entries[0])+1
745
first_fields = entries[1].split('\0')
748
if len(first_fields) <= 1:
749
# We didn't even get a dirname here... what do we do?
750
# Try a large page size and repeat this query
752
pending.append((low, high, cur_dirs))
755
# Find what entries we are looking for, which occur before and
756
# after this first record.
758
first_dir = first_fields[1]
759
first_loc = bisect.bisect_left(cur_dirs, first_dir)
761
# These exist before the current location
762
pre = cur_dirs[:first_loc]
763
# These occur after the current location, which may be in the
764
# data we read, or might be after the last entry
765
post = cur_dirs[first_loc:]
767
if post and len(first_fields) >= entry_field_count:
768
# We have records to look at after the first entry
770
# Parse the last entry
771
last_entry_num = len(entries)-1
772
last_fields = entries[last_entry_num].split('\0')
773
if len(last_fields) < entry_field_count:
774
# The very last hunk was not complete,
775
# read the previous hunk
776
after = mid + len(block) - len(entries[-1])
778
last_fields = entries[last_entry_num].split('\0')
780
after = mid + len(block)
782
last_dir = last_fields[1]
783
last_loc = bisect.bisect_right(post, last_dir)
785
middle_files = post[:last_loc]
786
post = post[last_loc:]
789
# We have files that should occur in this block
790
# (>= first, <= last)
791
# Either we will find them here, or we can mark them as
794
if middle_files[0] == first_dir:
795
# We might need to go before this location
796
pre.append(first_dir)
797
if middle_files[-1] == last_dir:
798
post.insert(0, last_dir)
800
# Find out what paths we have
801
paths = {first_dir:[first_fields]}
802
# last_dir might == first_dir so we need to be
803
# careful if we should append rather than overwrite
804
if last_entry_num != first_entry_num:
805
paths.setdefault(last_dir, []).append(last_fields)
806
for num in xrange(first_entry_num+1, last_entry_num):
807
# TODO: jam 20070223 We are already splitting here, so
808
# shouldn't we just split the whole thing rather
809
# than doing the split again in add_one_record?
810
fields = entries[num].split('\0')
811
paths.setdefault(fields[1], []).append(fields)
813
for cur_dir in middle_files:
814
for fields in paths.get(cur_dir, []):
815
# offset by 1 because of the opening '\0'
816
# consider changing fields_to_entry to avoid the
818
entry = fields_to_entry(fields[1:])
819
found.setdefault(cur_dir, []).append(entry)
821
# Now we have split up everything into pre, middle, and post, and
822
# we have handled everything that fell in 'middle'.
823
# We add 'post' first, so that we prefer to seek towards the
824
# beginning, so that we will tend to go as early as we need, and
825
# then only seek forward after that.
827
pending.append((after, high, post))
829
pending.append((low, start-1, pre))
833
def _bisect_recursive(self, paths):
834
"""Bisect for entries for all paths and their children.
836
This will use bisect to find all records for the supplied paths. It
837
will then continue to bisect for any records which are marked as
838
directories. (and renames?)
840
:param paths: A sorted list of (dir, name) pairs
841
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
842
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
844
# Map from (dir, name, file_id) => [tree_info]
847
found_dir_names = set()
849
# Directories that have been read
850
processed_dirs = set()
851
# Get the ball rolling with the first bisect for all entries.
852
newly_found = self._bisect(paths)
855
# Directories that need to be read
857
paths_to_search = set()
858
for entry_list in newly_found.itervalues():
859
for dir_name_id, trees_info in entry_list:
860
found[dir_name_id] = trees_info
861
found_dir_names.add(dir_name_id[:2])
863
for tree_info in trees_info:
864
minikind = tree_info[0]
867
# We already processed this one as a directory,
868
# we don't need to do the extra work again.
870
subdir, name, file_id = dir_name_id
871
path = osutils.pathjoin(subdir, name)
873
if path not in processed_dirs:
874
pending_dirs.add(path)
875
elif minikind == 'r':
876
# Rename, we need to directly search the target
877
# which is contained in the fingerprint column
878
dir_name = osutils.split(tree_info[1])
879
if dir_name[0] in pending_dirs:
880
# This entry will be found in the dir search
882
if dir_name not in found_dir_names:
883
paths_to_search.add(tree_info[1])
884
# Now we have a list of paths to look for directly, and
885
# directory blocks that need to be read.
886
# newly_found is mixing the keys between (dir, name) and path
887
# entries, but that is okay, because we only really care about the
889
newly_found = self._bisect(sorted(paths_to_search))
890
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
891
processed_dirs.update(pending_dirs)
894
def _discard_merge_parents(self):
895
"""Discard any parents trees beyond the first.
897
Note that if this fails the dirstate is corrupted.
899
After this function returns the dirstate contains 2 trees, neither of
902
self._read_header_if_needed()
903
parents = self.get_parent_ids()
906
# only require all dirblocks if we are doing a full-pass removal.
907
self._read_dirblocks_if_needed()
908
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
909
def iter_entries_removable():
910
for block in self._dirblocks:
911
deleted_positions = []
912
for pos, entry in enumerate(block[1]):
914
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
915
deleted_positions.append(pos)
916
if deleted_positions:
917
if len(deleted_positions) == len(block):
920
for pos in reversed(deleted_positions):
922
# if the first parent is a ghost:
923
if parents[0] in self.get_ghosts():
924
empty_parent = [DirState.NULL_PARENT_DETAILS]
925
for entry in iter_entries_removable():
926
entry[1][1:] = empty_parent
928
for entry in iter_entries_removable():
932
self._parents = [parents[0]]
933
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
934
self._header_state = DirState.IN_MEMORY_MODIFIED
936
def _empty_parent_info(self):
937
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
940
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
941
"""Ensure a block for dirname exists.
943
This function exists to let callers which know that there is a
944
directory dirname ensure that the block for it exists. This block can
945
fail to exist because of demand loading, or because a directory had no
946
children. In either case it is not an error. It is however an error to
947
call this if there is no parent entry for the directory, and thus the
948
function requires the coordinates of such an entry to be provided.
950
The root row is special cased and can be indicated with a parent block
953
:param parent_block_index: The index of the block in which dirname's row
955
:param parent_row_index: The index in the parent block where the row
957
:param dirname: The utf8 dirname to ensure there is a block for.
958
:return: The index for the block.
960
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
961
# This is the signature of the root row, and the
962
# contents-of-root row is always index 1
964
# the basename of the directory must be the end of its full name.
965
if not (parent_block_index == -1 and
966
parent_block_index == -1 and dirname == ''):
967
assert dirname.endswith(
968
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
969
block_index, present = self._find_block_index_from_key((dirname, '', ''))
971
## In future, when doing partial parsing, this should load and
972
# populate the entire block.
973
self._dirblocks.insert(block_index, (dirname, []))
976
def _entries_to_current_state(self, new_entries):
977
"""Load new_entries into self.dirblocks.
979
Process new_entries into the current state object, making them the active
980
state. The entries are grouped together by directory to form dirblocks.
982
:param new_entries: A sorted list of entries. This function does not sort
983
to prevent unneeded overhead when callers have a sorted list already.
986
assert new_entries[0][0][0:2] == ('', ''), \
987
"Missing root row %r" % (new_entries[0][0],)
988
# The two blocks here are deliberate: the root block and the
989
# contents-of-root block.
990
self._dirblocks = [('', []), ('', [])]
991
current_block = self._dirblocks[0][1]
994
append_entry = current_block.append
995
for entry in new_entries:
996
if entry[0][0] != current_dirname:
997
# new block - different dirname
999
current_dirname = entry[0][0]
1000
self._dirblocks.append((current_dirname, current_block))
1001
append_entry = current_block.append
1002
# append the entry to the current block
1004
self._split_root_dirblock_into_contents()
1006
def _split_root_dirblock_into_contents(self):
1007
"""Split the root dirblocks into root and contents-of-root.
1009
After parsing by path, we end up with root entries and contents-of-root
1010
entries in the same block. This loop splits them out again.
1012
# The above loop leaves the "root block" entries mixed with the
1013
# "contents-of-root block". But we don't want an if check on
1014
# all entries, so instead we just fix it up here.
1015
assert self._dirblocks[1] == ('', [])
1017
contents_of_root_block = []
1018
for entry in self._dirblocks[0][1]:
1019
if not entry[0][1]: # This is a root entry
1020
root_block.append(entry)
1022
contents_of_root_block.append(entry)
1023
self._dirblocks[0] = ('', root_block)
1024
self._dirblocks[1] = ('', contents_of_root_block)
1026
def _entry_to_line(self, entry):
1027
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1029
:param entry: An entry_tuple as defined in the module docstring.
1031
entire_entry = list(entry[0])
1032
for tree_number, tree_data in enumerate(entry[1]):
1033
# (minikind, fingerprint, size, executable, tree_specific_string)
1034
entire_entry.extend(tree_data)
1035
# 3 for the key, 5 for the fields per tree.
1036
tree_offset = 3 + tree_number * 5
1038
entire_entry[tree_offset + 0] = tree_data[0]
1040
entire_entry[tree_offset + 2] = str(tree_data[2])
1042
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1043
return '\0'.join(entire_entry)
1045
def _fields_per_entry(self):
1046
"""How many null separated fields should be in each entry row.
1048
Each line now has an extra '\n' field which is not used
1049
so we just skip over it
1051
3 fields for the key
1052
+ number of fields per tree_data (5) * tree count
1055
tree_count = 1 + self._num_present_parents()
1056
return 3 + 5 * tree_count + 1
1058
def _find_block(self, key, add_if_missing=False):
1059
"""Return the block that key should be present in.
1061
:param key: A dirstate entry key.
1062
:return: The block tuple.
1064
block_index, present = self._find_block_index_from_key(key)
1066
if not add_if_missing:
1067
# check to see if key is versioned itself - we might want to
1068
# add it anyway, because dirs with no entries dont get a
1069
# dirblock at parse time.
1070
# This is an uncommon branch to take: most dirs have children,
1071
# and most code works with versioned paths.
1072
parent_base, parent_name = osutils.split(key[0])
1073
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1074
# some parent path has not been added - its an error to add
1076
raise errors.NotVersionedError(key[0:2], str(self))
1077
self._dirblocks.insert(block_index, (key[0], []))
1078
return self._dirblocks[block_index]
1080
def _find_block_index_from_key(self, key):
1081
"""Find the dirblock index for a key.
1083
:return: The block index, True if the block for the key is present.
1085
if key[0:2] == ('', ''):
1088
if (self._last_block_index is not None and
1089
self._dirblocks[self._last_block_index][0] == key[0]):
1090
return self._last_block_index, True
1093
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1094
cache=self._split_path_cache)
1095
# _right returns one-past-where-key is so we have to subtract
1096
# one to use it. we use _right here because there are two
1097
# '' blocks - the root, and the contents of root
1098
# we always have a minimum of 2 in self._dirblocks: root and
1099
# root-contents, and for '', we get 2 back, so this is
1100
# simple and correct:
1101
present = (block_index < len(self._dirblocks) and
1102
self._dirblocks[block_index][0] == key[0])
1103
self._last_block_index = block_index
1104
# Reset the entry index cache to the beginning of the block.
1105
self._last_entry_index = -1
1106
return block_index, present
1108
def _find_entry_index(self, key, block):
1109
"""Find the entry index for a key in a block.
1111
:return: The entry index, True if the entry for the key is present.
1113
len_block = len(block)
1115
if self._last_entry_index is not None:
1117
entry_index = self._last_entry_index + 1
1118
# A hit is when the key is after the last slot, and before or
1119
# equal to the next slot.
1120
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1121
key <= block[entry_index][0]):
1122
self._last_entry_index = entry_index
1123
present = (block[entry_index][0] == key)
1124
return entry_index, present
1127
entry_index = bisect.bisect_left(block, (key, []))
1128
present = (entry_index < len_block and
1129
block[entry_index][0] == key)
1130
self._last_entry_index = entry_index
1131
return entry_index, present
1134
def from_tree(tree, dir_state_filename):
1135
"""Create a dirstate from a bzr Tree.
1137
:param tree: The tree which should provide parent information and
1139
:return: a DirState object which is currently locked for writing.
1140
(it was locked by DirState.initialize)
1142
result = DirState.initialize(dir_state_filename)
1146
parent_ids = tree.get_parent_ids()
1147
num_parents = len(parent_ids)
1149
for parent_id in parent_ids:
1150
parent_tree = tree.branch.repository.revision_tree(parent_id)
1151
parent_trees.append((parent_id, parent_tree))
1152
parent_tree.lock_read()
1153
result.set_parent_trees(parent_trees, [])
1154
result.set_state_from_inventory(tree.inventory)
1156
for revid, parent_tree in parent_trees:
1157
parent_tree.unlock()
1160
# The caller won't have a chance to unlock this, so make sure we
1166
def update_basis_by_delta(self, delta, new_revid):
1167
"""Update the parents of this tree after a commit.
1169
This gives the tree one parent, with revision id new_revid. The
1170
inventory delta is applied to the current basis tree to generate the
1171
inventory for the parent new_revid, and all other parent trees are
1174
Note that an exception during the operation of this method will leave
1175
the dirstate in a corrupt state where it should not be saved.
1177
Finally, we expect all changes to be synchronising the basis tree with
1180
:param new_revid: The new revision id for the trees parent.
1181
:param delta: An inventory delta (see apply_inventory_delta) describing
1182
the changes from the current left most parent revision to new_revid.
1184
self._read_dirblocks_if_needed()
1185
self._discard_merge_parents()
1186
if self._ghosts != []:
1187
raise NotImplementedError(self.update_basis_by_delta)
1188
if len(self._parents) == 0:
1189
# setup a blank tree, the most simple way.
1190
empty_parent = DirState.NULL_PARENT_DETAILS
1191
for entry in self._iter_entries():
1192
entry[1].append(empty_parent)
1193
self._parents.append(new_revid)
1195
self._parents[0] = new_revid
1197
delta = sorted(delta, reverse=True)
1201
# The paths this function accepts are unicode and must be encoded as we
1203
encode = cache_utf8.encode
1204
inv_to_entry = self._inv_entry_to_details
1205
# delta is now (deletes, changes), (adds) in reverse lexographical
1207
# deletes in reverse lexographic order are safe to process in situ.
1208
# renames are not, as a rename from any path could go to a path
1209
# lexographically lower, so we transform renames into delete, add pairs,
1210
# expanding them recursively as needed.
1211
# At the same time, to reduce interface friction we convert the input
1212
# inventory entries to dirstate.
1213
root_only = ('', '')
1214
for old_path, new_path, file_id, inv_entry in delta:
1215
if old_path is None:
1216
adds.append((None, encode(new_path), file_id,
1217
inv_to_entry(inv_entry), True))
1218
elif new_path is None:
1219
deletes.append((encode(old_path), None, file_id, None, True))
1220
elif (old_path, new_path) != root_only:
1222
# Because renames must preserve their children we must have
1223
# processed all relocations and removes before hand. The sort
1224
# order ensures we've examined the child paths, but we also
1225
# have to execute the removals, or the split to an add/delete
1226
# pair will result in the deleted item being reinserted, or
1227
# renamed items being reinserted twice - and possibly at the
1228
# wrong place. Splitting into a delete/add pair also simplifies
1229
# the handling of entries with ('f', ...), ('r' ...) because
1230
# the target of the 'r' is old_path here, and we add that to
1231
# deletes, meaning that the add handler does not need to check
1232
# for 'r' items on every pass.
1233
self._update_basis_apply_deletes(deletes)
1235
new_path_utf8 = encode(new_path)
1236
# Split into an add/delete pair recursively.
1237
adds.append((None, new_path_utf8, file_id,
1238
inv_to_entry(inv_entry), False))
1239
# Expunge deletes that we've seen so that deleted/renamed
1240
# children of a rename directory are handled correctly.
1241
new_deletes = reversed(list(self._iter_child_entries(1,
1243
# Remove the current contents of the tree at orig_path, and
1244
# reinsert at the correct new path.
1245
for entry in new_deletes:
1247
source_path = entry[0][0] + '/' + entry[0][1]
1249
source_path = entry[0][1]
1250
target_path = new_path_utf8 + source_path[len(old_path):]
1251
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1253
(source_path, target_path, entry[0][2], None, False))
1255
(encode(old_path), new_path, file_id, None, False))
1257
# changes to just the root should not require remove/insertion
1259
changes.append((encode(old_path), encode(new_path), file_id,
1260
inv_to_entry(inv_entry)))
1262
# Finish expunging deletes/first half of renames.
1263
self._update_basis_apply_deletes(deletes)
1264
# Reinstate second half of renames and new paths.
1265
self._update_basis_apply_adds(adds)
1266
# Apply in-situ changes.
1267
self._update_basis_apply_changes(changes)
1269
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1270
self._header_state = DirState.IN_MEMORY_MODIFIED
1271
self._id_index = None
1274
def _update_basis_apply_adds(self, adds):
1275
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1277
They may be adds, or renames that have been split into add/delete
1280
:param adds: A sequence of adds. Each add is a tuple:
1281
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1282
is False when the add is the second half of a remove-and-reinsert
1283
pair created to handle renames and deletes.
1285
# Adds are accumulated partly from renames, so can be in any input
1288
# adds is now in lexographic order, which places all parents before
1289
# their children, so we can process it linearly.
1291
for old_path, new_path, file_id, new_details, real_add in adds:
1292
assert old_path is None
1293
# the entry for this file_id must be in tree 0.
1294
entry = self._get_entry(0, file_id, new_path)
1295
if entry[0][2] != file_id:
1296
raise errors.BzrError('dirstate: cannot apply delta, working'
1297
' tree does not contain new entry %r %r' %
1298
(new_path, file_id))
1299
if real_add and entry[1][1][0] not in absent:
1300
raise errors.BzrError('dirstate: inconsistent delta, with '
1301
'tree 0. %r %r' % (new_path, file_id))
1302
# We don't need to update the target of an 'r' because the handling
1303
# of renames turns all 'r' situations into a delete at the original
1305
entry[1][1] = new_details
1307
def _update_basis_apply_changes(self, changes):
1308
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1310
:param adds: A sequence of changes. Each change is a tuple:
1311
(path_utf8, path_utf8, file_id, (entry_details))
1314
for old_path, new_path, file_id, new_details in changes:
1315
assert old_path == new_path
1316
# the entry for this file_id must be in tree 0.
1317
entry = self._get_entry(0, file_id, new_path)
1318
if entry[0][2] != file_id:
1319
raise errors.BzrError('dirstate: cannot apply delta, working'
1320
' tree does not contain new entry %r %r' %
1321
(new_path, file_id))
1322
if (entry[1][0][0] in absent or
1323
entry[1][1][0] in absent):
1324
raise errors.BzrError('dirstate: inconsistent delta, with '
1325
'tree 0. %r %r' % (new_path, file_id))
1326
entry[1][1] = new_details
1328
def _update_basis_apply_deletes(self, deletes):
1329
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1331
They may be deletes, or renames that have been split into add/delete
1334
:param deletes: A sequence of deletes. Each delete is a tuple:
1335
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1336
real_delete is True when the desired outcome is an actual deletion
1337
rather than the rename handling logic temporarily deleting a path
1338
during the replacement of a parent.
1340
null = DirState.NULL_PARENT_DETAILS
1341
for old_path, new_path, file_id, _, real_delete in deletes:
1343
assert new_path is None
1345
assert new_path is not None
1346
# the entry for this file_id must be in tree 1.
1347
dirname, basename = osutils.split(old_path)
1348
block_index, entry_index, dir_present, file_present = \
1349
self._get_block_entry_index(dirname, basename, 1)
1350
if not file_present:
1351
raise errors.BzrError('dirstate: cannot apply delta, basis'
1352
' tree does not contain new entry %r %r' %
1353
(old_path, file_id))
1354
entry = self._dirblocks[block_index][1][entry_index]
1355
if entry[0][2] != file_id:
1356
raise errors.BzrError('mismatched file_id in tree 1 %r %r' %
1357
(old_path, file_id))
1359
if entry[1][0][0] != 'a':
1360
raise errors.BzrError('dirstate: inconsistent delta, with '
1361
'tree 0. %r %r' % (old_path, file_id))
1362
del self._dirblocks[block_index][1][entry_index]
1364
if entry[1][0][0] == 'a':
1365
raise errors.BzrError('dirstate: inconsistent delta, with '
1366
'tree 0. %r %r' % (old_path, file_id))
1367
elif entry[1][0][0] == 'r':
1368
# implement the rename
1369
del self._dirblocks[block_index][1][entry_index]
1371
# it is being resurrected here, so blank it out temporarily.
1372
self._dirblocks[block_index][1][entry_index][1][1] = null
1374
def update_entry(self, entry, abspath, stat_value,
1375
_stat_to_minikind=_stat_to_minikind,
1376
_pack_stat=pack_stat):
1377
"""Update the entry based on what is actually on disk.
1379
:param entry: This is the dirblock entry for the file in question.
1380
:param abspath: The path on disk for this file.
1381
:param stat_value: (optional) if we already have done a stat on the
1383
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1387
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1391
packed_stat = _pack_stat(stat_value)
1392
(saved_minikind, saved_link_or_sha1, saved_file_size,
1393
saved_executable, saved_packed_stat) = entry[1][0]
1395
if (minikind == saved_minikind
1396
and packed_stat == saved_packed_stat):
1397
# The stat hasn't changed since we saved, so we can re-use the
1402
# size should also be in packed_stat
1403
if saved_file_size == stat_value.st_size:
1404
return saved_link_or_sha1
1406
# If we have gotten this far, that means that we need to actually
1407
# process this entry.
1410
link_or_sha1 = self._sha1_file(abspath)
1411
executable = self._is_executable(stat_value.st_mode,
1413
if self._cutoff_time is None:
1414
self._sha_cutoff_time()
1415
if (stat_value.st_mtime < self._cutoff_time
1416
and stat_value.st_ctime < self._cutoff_time):
1417
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1418
executable, packed_stat)
1420
entry[1][0] = ('f', '', stat_value.st_size,
1421
executable, DirState.NULLSTAT)
1422
elif minikind == 'd':
1424
entry[1][0] = ('d', '', 0, False, packed_stat)
1425
if saved_minikind != 'd':
1426
# This changed from something into a directory. Make sure we
1427
# have a directory block for it. This doesn't happen very
1428
# often, so this doesn't have to be super fast.
1429
block_index, entry_index, dir_present, file_present = \
1430
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1431
self._ensure_block(block_index, entry_index,
1432
osutils.pathjoin(entry[0][0], entry[0][1]))
1433
elif minikind == 'l':
1434
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1435
if self._cutoff_time is None:
1436
self._sha_cutoff_time()
1437
if (stat_value.st_mtime < self._cutoff_time
1438
and stat_value.st_ctime < self._cutoff_time):
1439
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1442
entry[1][0] = ('l', '', stat_value.st_size,
1443
False, DirState.NULLSTAT)
1444
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1447
def _sha_cutoff_time(self):
1448
"""Return cutoff time.
1450
Files modified more recently than this time are at risk of being
1451
undetectably modified and so can't be cached.
1453
# Cache the cutoff time as long as we hold a lock.
1454
# time.time() isn't super expensive (approx 3.38us), but
1455
# when you call it 50,000 times it adds up.
1456
# For comparison, os.lstat() costs 7.2us if it is hot.
1457
self._cutoff_time = int(time.time()) - 3
1458
return self._cutoff_time
1460
def _lstat(self, abspath, entry):
1461
"""Return the os.lstat value for this path."""
1462
return os.lstat(abspath)
1464
def _sha1_file_and_mutter(self, abspath):
1465
# when -Dhashcache is turned on, this is monkey-patched in to log
1467
trace.mutter("dirstate sha1 " + abspath)
1468
return osutils.sha_file_by_name(abspath)
1470
def _is_executable(self, mode, old_executable):
1471
"""Is this file executable?"""
1472
return bool(S_IEXEC & mode)
1474
def _is_executable_win32(self, mode, old_executable):
1475
"""On win32 the executable bit is stored in the dirstate."""
1476
return old_executable
1478
if sys.platform == 'win32':
1479
_is_executable = _is_executable_win32
1481
def _read_link(self, abspath, old_link):
1482
"""Read the target of a symlink"""
1483
# TODO: jam 200700301 On Win32, this could just return the value
1484
# already in memory. However, this really needs to be done at a
1485
# higher level, because there either won't be anything on disk,
1486
# or the thing on disk will be a file.
1487
return os.readlink(abspath)
1489
def get_ghosts(self):
1490
"""Return a list of the parent tree revision ids that are ghosts."""
1491
self._read_header_if_needed()
1494
def get_lines(self):
1495
"""Serialise the entire dirstate to a sequence of lines."""
1496
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1497
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1498
# read whats on disk.
1499
self._state_file.seek(0)
1500
return self._state_file.readlines()
1502
lines.append(self._get_parents_line(self.get_parent_ids()))
1503
lines.append(self._get_ghosts_line(self._ghosts))
1504
# append the root line which is special cased
1505
lines.extend(map(self._entry_to_line, self._iter_entries()))
1506
return self._get_output_lines(lines)
1508
def _get_ghosts_line(self, ghost_ids):
1509
"""Create a line for the state file for ghost information."""
1510
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1512
def _get_parents_line(self, parent_ids):
1513
"""Create a line for the state file for parents information."""
1514
return '\0'.join([str(len(parent_ids))] + parent_ids)
1516
def _get_fields_to_entry(self):
1517
"""Get a function which converts entry fields into a entry record.
1519
This handles size and executable, as well as parent records.
1521
:return: A function which takes a list of fields, and returns an
1522
appropriate record for storing in memory.
1524
# This is intentionally unrolled for performance
1525
num_present_parents = self._num_present_parents()
1526
if num_present_parents == 0:
1527
def fields_to_entry_0_parents(fields, _int=int):
1528
path_name_file_id_key = (fields[0], fields[1], fields[2])
1529
return (path_name_file_id_key, [
1531
fields[3], # minikind
1532
fields[4], # fingerprint
1533
_int(fields[5]), # size
1534
fields[6] == 'y', # executable
1535
fields[7], # packed_stat or revision_id
1537
return fields_to_entry_0_parents
1538
elif num_present_parents == 1:
1539
def fields_to_entry_1_parent(fields, _int=int):
1540
path_name_file_id_key = (fields[0], fields[1], fields[2])
1541
return (path_name_file_id_key, [
1543
fields[3], # minikind
1544
fields[4], # fingerprint
1545
_int(fields[5]), # size
1546
fields[6] == 'y', # executable
1547
fields[7], # packed_stat or revision_id
1550
fields[8], # minikind
1551
fields[9], # fingerprint
1552
_int(fields[10]), # size
1553
fields[11] == 'y', # executable
1554
fields[12], # packed_stat or revision_id
1557
return fields_to_entry_1_parent
1558
elif num_present_parents == 2:
1559
def fields_to_entry_2_parents(fields, _int=int):
1560
path_name_file_id_key = (fields[0], fields[1], fields[2])
1561
return (path_name_file_id_key, [
1563
fields[3], # minikind
1564
fields[4], # fingerprint
1565
_int(fields[5]), # size
1566
fields[6] == 'y', # executable
1567
fields[7], # packed_stat or revision_id
1570
fields[8], # minikind
1571
fields[9], # fingerprint
1572
_int(fields[10]), # size
1573
fields[11] == 'y', # executable
1574
fields[12], # packed_stat or revision_id
1577
fields[13], # minikind
1578
fields[14], # fingerprint
1579
_int(fields[15]), # size
1580
fields[16] == 'y', # executable
1581
fields[17], # packed_stat or revision_id
1584
return fields_to_entry_2_parents
1586
def fields_to_entry_n_parents(fields, _int=int):
1587
path_name_file_id_key = (fields[0], fields[1], fields[2])
1588
trees = [(fields[cur], # minikind
1589
fields[cur+1], # fingerprint
1590
_int(fields[cur+2]), # size
1591
fields[cur+3] == 'y', # executable
1592
fields[cur+4], # stat or revision_id
1593
) for cur in xrange(3, len(fields)-1, 5)]
1594
return path_name_file_id_key, trees
1595
return fields_to_entry_n_parents
1597
def get_parent_ids(self):
1598
"""Return a list of the parent tree ids for the directory state."""
1599
self._read_header_if_needed()
1600
return list(self._parents)
1602
def _get_block_entry_index(self, dirname, basename, tree_index):
1603
"""Get the coordinates for a path in the state structure.
1605
:param dirname: The utf8 dirname to lookup.
1606
:param basename: The utf8 basename to lookup.
1607
:param tree_index: The index of the tree for which this lookup should
1609
:return: A tuple describing where the path is located, or should be
1610
inserted. The tuple contains four fields: the block index, the row
1611
index, the directory is present (boolean), the entire path is
1612
present (boolean). There is no guarantee that either
1613
coordinate is currently reachable unless the found field for it is
1614
True. For instance, a directory not present in the searched tree
1615
may be returned with a value one greater than the current highest
1616
block offset. The directory present field will always be True when
1617
the path present field is True. The directory present field does
1618
NOT indicate that the directory is present in the searched tree,
1619
rather it indicates that there are at least some files in some
1622
self._read_dirblocks_if_needed()
1623
key = dirname, basename, ''
1624
block_index, present = self._find_block_index_from_key(key)
1626
# no such directory - return the dir index and 0 for the row.
1627
return block_index, 0, False, False
1628
block = self._dirblocks[block_index][1] # access the entries only
1629
entry_index, present = self._find_entry_index(key, block)
1630
# linear search through entries at this path to find the one
1632
while entry_index < len(block) and block[entry_index][0][1] == basename:
1633
if block[entry_index][1][tree_index][0] not in 'ar':
1634
# neither absent or relocated
1635
return block_index, entry_index, True, True
1637
return block_index, entry_index, True, False
1639
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1640
"""Get the dirstate entry for path in tree tree_index.
1642
If either file_id or path is supplied, it is used as the key to lookup.
1643
If both are supplied, the fastest lookup is used, and an error is
1644
raised if they do not both point at the same row.
1646
:param tree_index: The index of the tree we wish to locate this path
1647
in. If the path is present in that tree, the entry containing its
1648
details is returned, otherwise (None, None) is returned
1649
0 is the working tree, higher indexes are successive parent
1651
:param fileid_utf8: A utf8 file_id to look up.
1652
:param path_utf8: An utf8 path to be looked up.
1653
:return: The dirstate entry tuple for path, or (None, None)
1655
self._read_dirblocks_if_needed()
1656
if path_utf8 is not None:
1657
assert path_utf8.__class__ == str, ('path_utf8 is not a str: %s %s'
1658
% (type(path_utf8), path_utf8))
1659
# path lookups are faster
1660
dirname, basename = osutils.split(path_utf8)
1661
block_index, entry_index, dir_present, file_present = \
1662
self._get_block_entry_index(dirname, basename, tree_index)
1663
if not file_present:
1665
entry = self._dirblocks[block_index][1][entry_index]
1666
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1668
if entry[0][2] != fileid_utf8:
1669
raise errors.BzrError('integrity error ? : mismatching'
1670
' tree_index, file_id and path')
1673
assert fileid_utf8 is not None
1674
possible_keys = self._get_id_index().get(fileid_utf8, None)
1675
if not possible_keys:
1677
for key in possible_keys:
1678
block_index, present = \
1679
self._find_block_index_from_key(key)
1680
# strange, probably indicates an out of date
1681
# id index - for now, allow this.
1684
# WARNING: DO not change this code to use _get_block_entry_index
1685
# as that function is not suitable: it does not use the key
1686
# to lookup, and thus the wrong coordinates are returned.
1687
block = self._dirblocks[block_index][1]
1688
entry_index, present = self._find_entry_index(key, block)
1690
entry = self._dirblocks[block_index][1][entry_index]
1691
if entry[1][tree_index][0] in 'fdlt':
1692
# this is the result we are looking for: the
1693
# real home of this file_id in this tree.
1695
if entry[1][tree_index][0] == 'a':
1696
# there is no home for this entry in this tree
1698
assert entry[1][tree_index][0] == 'r', \
1699
"entry %r has invalid minikind %r for tree %r" \
1701
entry[1][tree_index][0],
1703
real_path = entry[1][tree_index][1]
1704
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1705
path_utf8=real_path)
1709
def initialize(cls, path):
1710
"""Create a new dirstate on path.
1712
The new dirstate will be an empty tree - that is it has no parents,
1713
and only a root node - which has id ROOT_ID.
1715
:param path: The name of the file for the dirstate.
1716
:return: A write-locked DirState object.
1718
# This constructs a new DirState object on a path, sets the _state_file
1719
# to a new empty file for that path. It then calls _set_data() with our
1720
# stock empty dirstate information - a root with ROOT_ID, no children,
1721
# and no parents. Finally it calls save() to ensure that this data will
1724
# root dir and root dir contents with no children.
1725
empty_tree_dirblocks = [('', []), ('', [])]
1726
# a new root directory, with a NULLSTAT.
1727
empty_tree_dirblocks[0][1].append(
1728
(('', '', inventory.ROOT_ID), [
1729
('d', '', 0, False, DirState.NULLSTAT),
1733
result._set_data([], empty_tree_dirblocks)
1740
def _inv_entry_to_details(self, inv_entry):
1741
"""Convert an inventory entry (from a revision tree) to state details.
1743
:param inv_entry: An inventory entry whose sha1 and link targets can be
1744
relied upon, and which has a revision set.
1745
:return: A details tuple - the details for a single tree at a path +
1748
kind = inv_entry.kind
1749
minikind = DirState._kind_to_minikind[kind]
1750
tree_data = inv_entry.revision
1751
assert tree_data, 'empty revision for the inv_entry %s.' % \
1753
if kind == 'directory':
1757
elif kind == 'symlink':
1758
fingerprint = inv_entry.symlink_target or ''
1761
elif kind == 'file':
1762
fingerprint = inv_entry.text_sha1 or ''
1763
size = inv_entry.text_size or 0
1764
executable = inv_entry.executable
1765
elif kind == 'tree-reference':
1766
fingerprint = inv_entry.reference_revision or ''
1770
raise Exception("can't pack %s" % inv_entry)
1771
return (minikind, fingerprint, size, executable, tree_data)
1773
def _iter_child_entries(self, tree_index, path_utf8):
1774
"""Iterate over all the entries that are children of path_utf.
1776
This only returns entries that are present (not in 'a', 'r') in
1777
tree_index. tree_index data is not refreshed, so if tree 0 is used,
1778
results may differ from that obtained if paths were statted to
1779
determine what ones were directories.
1781
Asking for the children of a non-directory will return an empty
1785
next_pending_dirs = [path_utf8]
1787
while next_pending_dirs:
1788
pending_dirs = next_pending_dirs
1789
next_pending_dirs = []
1790
for path in pending_dirs:
1791
block_index, present = self._find_block_index_from_key(
1793
if block_index == 0:
1795
if len(self._dirblocks) == 1:
1796
# asked for the children of the root with no other
1800
# children of a non-directory asked for.
1802
block = self._dirblocks[block_index]
1803
for entry in block[1]:
1804
kind = entry[1][tree_index][0]
1805
if kind not in absent:
1809
path = entry[0][0] + '/' + entry[0][1]
1812
next_pending_dirs.append(path)
1814
def _iter_entries(self):
1815
"""Iterate over all the entries in the dirstate.
1817
Each yelt item is an entry in the standard format described in the
1818
docstring of bzrlib.dirstate.
1820
self._read_dirblocks_if_needed()
1821
for directory in self._dirblocks:
1822
for entry in directory[1]:
1825
def _get_id_index(self):
1826
"""Get an id index of self._dirblocks."""
1827
if self._id_index is None:
1829
for key, tree_details in self._iter_entries():
1830
id_index.setdefault(key[2], set()).add(key)
1831
self._id_index = id_index
1832
return self._id_index
1834
def _get_output_lines(self, lines):
1835
"""Format lines for final output.
1837
:param lines: A sequence of lines containing the parents list and the
1840
output_lines = [DirState.HEADER_FORMAT_3]
1841
lines.append('') # a final newline
1842
inventory_text = '\0\n\0'.join(lines)
1843
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1844
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1845
num_entries = len(lines)-3
1846
output_lines.append('num_entries: %s\n' % (num_entries,))
1847
output_lines.append(inventory_text)
1850
def _make_deleted_row(self, fileid_utf8, parents):
1851
"""Return a deleted row for fileid_utf8."""
1852
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1855
def _num_present_parents(self):
1856
"""The number of parent entries in each record row."""
1857
return len(self._parents) - len(self._ghosts)
1861
"""Construct a DirState on the file at path path.
1863
:return: An unlocked DirState object, associated with the given path.
1865
result = DirState(path)
1868
def _read_dirblocks_if_needed(self):
1869
"""Read in all the dirblocks from the file if they are not in memory.
1871
This populates self._dirblocks, and sets self._dirblock_state to
1872
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1875
self._read_header_if_needed()
1876
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1877
_read_dirblocks(self)
1879
def _read_header(self):
1880
"""This reads in the metadata header, and the parent ids.
1882
After reading in, the file should be positioned at the null
1883
just before the start of the first record in the file.
1885
:return: (expected crc checksum, number of entries, parent list)
1887
self._read_prelude()
1888
parent_line = self._state_file.readline()
1889
info = parent_line.split('\0')
1890
num_parents = int(info[0])
1891
assert num_parents == len(info)-2, 'incorrect parent info line'
1892
self._parents = info[1:-1]
1894
ghost_line = self._state_file.readline()
1895
info = ghost_line.split('\0')
1896
num_ghosts = int(info[1])
1897
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1898
self._ghosts = info[2:-1]
1899
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1900
self._end_of_header = self._state_file.tell()
1902
def _read_header_if_needed(self):
1903
"""Read the header of the dirstate file if needed."""
1904
# inline this as it will be called a lot
1905
if not self._lock_token:
1906
raise errors.ObjectNotLocked(self)
1907
if self._header_state == DirState.NOT_IN_MEMORY:
1910
def _read_prelude(self):
1911
"""Read in the prelude header of the dirstate file.
1913
This only reads in the stuff that is not connected to the crc
1914
checksum. The position will be correct to read in the rest of
1915
the file and check the checksum after this point.
1916
The next entry in the file should be the number of parents,
1917
and their ids. Followed by a newline.
1919
header = self._state_file.readline()
1920
assert header == DirState.HEADER_FORMAT_3, \
1921
'invalid header line: %r' % (header,)
1922
crc_line = self._state_file.readline()
1923
assert crc_line.startswith('crc32: '), 'missing crc32 checksum'
1924
self.crc_expected = int(crc_line[len('crc32: '):-1])
1925
num_entries_line = self._state_file.readline()
1926
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1927
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1929
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
1930
"""Find a sha1 given a stat lookup."""
1931
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
1933
def _get_packed_stat_index(self):
1934
"""Get a packed_stat index of self._dirblocks."""
1935
if self._packed_stat_index is None:
1937
for key, tree_details in self._iter_entries():
1938
if tree_details[0][0] == 'f':
1939
index[tree_details[0][4]] = tree_details[0][1]
1940
self._packed_stat_index = index
1941
return self._packed_stat_index
1944
"""Save any pending changes created during this session.
1946
We reuse the existing file, because that prevents race conditions with
1947
file creation, and use oslocks on it to prevent concurrent modification
1948
and reads - because dirstate's incremental data aggregation is not
1949
compatible with reading a modified file, and replacing a file in use by
1950
another process is impossible on Windows.
1952
A dirstate in read only mode should be smart enough though to validate
1953
that the file has not changed, and otherwise discard its cache and
1954
start over, to allow for fine grained read lock duration, so 'status'
1955
wont block 'commit' - for example.
1957
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1958
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1960
grabbed_write_lock = False
1961
if self._lock_state != 'w':
1962
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
1963
# Switch over to the new lock, as the old one may be closed.
1964
# TODO: jam 20070315 We should validate the disk file has
1965
# not changed contents. Since temporary_write_lock may
1966
# not be an atomic operation.
1967
self._lock_token = new_lock
1968
self._state_file = new_lock.f
1969
if not grabbed_write_lock:
1970
# We couldn't grab a write lock, so we switch back to a read one
1973
self._state_file.seek(0)
1974
self._state_file.writelines(self.get_lines())
1975
self._state_file.truncate()
1976
self._state_file.flush()
1977
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1978
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1980
if grabbed_write_lock:
1981
self._lock_token = self._lock_token.restore_read_lock()
1982
self._state_file = self._lock_token.f
1983
# TODO: jam 20070315 We should validate the disk file has
1984
# not changed contents. Since restore_read_lock may
1985
# not be an atomic operation.
1987
def _set_data(self, parent_ids, dirblocks):
1988
"""Set the full dirstate data in memory.
1990
This is an internal function used to completely replace the objects
1991
in memory state. It puts the dirstate into state 'full-dirty'.
1993
:param parent_ids: A list of parent tree revision ids.
1994
:param dirblocks: A list containing one tuple for each directory in the
1995
tree. Each tuple contains the directory path and a list of entries
1996
found in that directory.
1998
# our memory copy is now authoritative.
1999
self._dirblocks = dirblocks
2000
self._header_state = DirState.IN_MEMORY_MODIFIED
2001
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2002
self._parents = list(parent_ids)
2003
self._id_index = None
2004
self._packed_stat_index = None
2006
def set_path_id(self, path, new_id):
2007
"""Change the id of path to new_id in the current working tree.
2009
:param path: The path inside the tree to set - '' is the root, 'foo'
2010
is the path foo in the root.
2011
:param new_id: The new id to assign to the path. This must be a utf8
2012
file id (not unicode, and not None).
2014
assert new_id.__class__ == str, \
2015
"path_id %r is not a plain string" % (new_id,)
2016
self._read_dirblocks_if_needed()
2018
# TODO: logic not written
2019
raise NotImplementedError(self.set_path_id)
2020
# TODO: check new id is unique
2021
entry = self._get_entry(0, path_utf8=path)
2022
if entry[0][2] == new_id:
2023
# Nothing to change.
2025
# mark the old path absent, and insert a new root path
2026
self._make_absent(entry)
2027
self.update_minimal(('', '', new_id), 'd',
2028
path_utf8='', packed_stat=entry[1][0][4])
2029
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2030
if self._id_index is not None:
2031
self._id_index.setdefault(new_id, set()).add(entry[0])
2033
def set_parent_trees(self, trees, ghosts):
2034
"""Set the parent trees for the dirstate.
2036
:param trees: A list of revision_id, tree tuples. tree must be provided
2037
even if the revision_id refers to a ghost: supply an empty tree in
2039
:param ghosts: A list of the revision_ids that are ghosts at the time
2042
# TODO: generate a list of parent indexes to preserve to save
2043
# processing specific parent trees. In the common case one tree will
2044
# be preserved - the left most parent.
2045
# TODO: if the parent tree is a dirstate, we might want to walk them
2046
# all by path in parallel for 'optimal' common-case performance.
2047
# generate new root row.
2048
self._read_dirblocks_if_needed()
2049
# TODO future sketch: Examine the existing parents to generate a change
2050
# map and then walk the new parent trees only, mapping them into the
2051
# dirstate. Walk the dirstate at the same time to remove unreferenced
2054
# sketch: loop over all entries in the dirstate, cherry picking
2055
# entries from the parent trees, if they are not ghost trees.
2056
# after we finish walking the dirstate, all entries not in the dirstate
2057
# are deletes, so we want to append them to the end as per the design
2058
# discussions. So do a set difference on ids with the parents to
2059
# get deletes, and add them to the end.
2060
# During the update process we need to answer the following questions:
2061
# - find other keys containing a fileid in order to create cross-path
2062
# links. We dont't trivially use the inventory from other trees
2063
# because this leads to either double touching, or to accessing
2065
# - find other keys containing a path
2066
# We accumulate each entry via this dictionary, including the root
2069
# we could do parallel iterators, but because file id data may be
2070
# scattered throughout, we dont save on index overhead: we have to look
2071
# at everything anyway. We can probably save cycles by reusing parent
2072
# data and doing an incremental update when adding an additional
2073
# parent, but for now the common cases are adding a new parent (merge),
2074
# and replacing completely (commit), and commit is more common: so
2075
# optimise merge later.
2077
# ---- start generation of full tree mapping data
2078
# what trees should we use?
2079
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2080
# how many trees do we end up with
2081
parent_count = len(parent_trees)
2083
# one: the current tree
2084
for entry in self._iter_entries():
2085
# skip entries not in the current tree
2086
if entry[1][0][0] in 'ar': # absent, relocated
2088
by_path[entry[0]] = [entry[1][0]] + \
2089
[DirState.NULL_PARENT_DETAILS] * parent_count
2090
id_index[entry[0][2]] = set([entry[0]])
2092
# now the parent trees:
2093
for tree_index, tree in enumerate(parent_trees):
2094
# the index is off by one, adjust it.
2095
tree_index = tree_index + 1
2096
# when we add new locations for a fileid we need these ranges for
2097
# any fileid in this tree as we set the by_path[id] to:
2098
# already_processed_tree_details + new_details + new_location_suffix
2099
# the suffix is from tree_index+1:parent_count+1.
2100
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2101
# now stitch in all the entries from this tree
2102
for path, entry in tree.inventory.iter_entries_by_dir():
2103
# here we process each trees details for each item in the tree.
2104
# we first update any existing entries for the id at other paths,
2105
# then we either create or update the entry for the id at the
2106
# right path, and finally we add (if needed) a mapping from
2107
# file_id to this path. We do it in this order to allow us to
2108
# avoid checking all known paths for the id when generating a
2109
# new entry at this path: by adding the id->path mapping last,
2110
# all the mappings are valid and have correct relocation
2111
# records where needed.
2112
file_id = entry.file_id
2113
path_utf8 = path.encode('utf8')
2114
dirname, basename = osutils.split(path_utf8)
2115
new_entry_key = (dirname, basename, file_id)
2116
# tree index consistency: All other paths for this id in this tree
2117
# index must point to the correct path.
2118
for entry_key in id_index.setdefault(file_id, set()):
2119
# TODO:PROFILING: It might be faster to just update
2120
# rather than checking if we need to, and then overwrite
2121
# the one we are located at.
2122
if entry_key != new_entry_key:
2123
# this file id is at a different path in one of the
2124
# other trees, so put absent pointers there
2125
# This is the vertical axis in the matrix, all pointing
2127
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2128
# by path consistency: Insert into an existing path record (trivial), or
2129
# add a new one with relocation pointers for the other tree indexes.
2130
if new_entry_key in id_index[file_id]:
2131
# there is already an entry where this data belongs, just insert it.
2132
by_path[new_entry_key][tree_index] = \
2133
self._inv_entry_to_details(entry)
2135
# add relocated entries to the horizontal axis - this row
2136
# mapping from path,id. We need to look up the correct path
2137
# for the indexes from 0 to tree_index -1
2139
for lookup_index in xrange(tree_index):
2140
# boundary case: this is the first occurence of file_id
2141
# so there are no id_indexs, possibly take this out of
2143
if not len(id_index[file_id]):
2144
new_details.append(DirState.NULL_PARENT_DETAILS)
2146
# grab any one entry, use it to find the right path.
2147
# TODO: optimise this to reduce memory use in highly
2148
# fragmented situations by reusing the relocation
2150
a_key = iter(id_index[file_id]).next()
2151
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2152
# its a pointer or missing statement, use it as is.
2153
new_details.append(by_path[a_key][lookup_index])
2155
# we have the right key, make a pointer to it.
2156
real_path = ('/'.join(a_key[0:2])).strip('/')
2157
new_details.append(('r', real_path, 0, False, ''))
2158
new_details.append(self._inv_entry_to_details(entry))
2159
new_details.extend(new_location_suffix)
2160
by_path[new_entry_key] = new_details
2161
id_index[file_id].add(new_entry_key)
2162
# --- end generation of full tree mappings
2164
# sort and output all the entries
2165
new_entries = self._sort_entries(by_path.items())
2166
self._entries_to_current_state(new_entries)
2167
self._parents = [rev_id for rev_id, tree in trees]
2168
self._ghosts = list(ghosts)
2169
self._header_state = DirState.IN_MEMORY_MODIFIED
2170
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2171
self._id_index = id_index
2173
def _sort_entries(self, entry_list):
2174
"""Given a list of entries, sort them into the right order.
2176
This is done when constructing a new dirstate from trees - normally we
2177
try to keep everything in sorted blocks all the time, but sometimes
2178
it's easier to sort after the fact.
2181
# sort by: directory parts, file name, file id
2182
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2183
return sorted(entry_list, key=_key)
2185
def set_state_from_inventory(self, new_inv):
2186
"""Set new_inv as the current state.
2188
This API is called by tree transform, and will usually occur with
2189
existing parent trees.
2191
:param new_inv: The inventory object to set current state from.
2193
if 'evil' in debug.debug_flags:
2194
trace.mutter_callsite(1,
2195
"set_state_from_inventory called; please mutate the tree instead")
2196
self._read_dirblocks_if_needed()
2198
# Two iterators: current data and new data, both in dirblock order.
2199
# We zip them together, which tells about entries that are new in the
2200
# inventory, or removed in the inventory, or present in both and
2203
# You might think we could just synthesize a new dirstate directly
2204
# since we're processing it in the right order. However, we need to
2205
# also consider there may be any number of parent trees and relocation
2206
# pointers, and we don't want to duplicate that here.
2207
new_iterator = new_inv.iter_entries_by_dir()
2208
# we will be modifying the dirstate, so we need a stable iterator. In
2209
# future we might write one, for now we just clone the state into a
2210
# list - which is a shallow copy.
2211
old_iterator = iter(list(self._iter_entries()))
2212
# both must have roots so this is safe:
2213
current_new = new_iterator.next()
2214
current_old = old_iterator.next()
2215
def advance(iterator):
2217
return iterator.next()
2218
except StopIteration:
2220
while current_new or current_old:
2221
# skip entries in old that are not really there
2222
if current_old and current_old[1][0][0] in 'ar':
2223
# relocated or absent
2224
current_old = advance(old_iterator)
2227
# convert new into dirblock style
2228
new_path_utf8 = current_new[0].encode('utf8')
2229
new_dirname, new_basename = osutils.split(new_path_utf8)
2230
new_id = current_new[1].file_id
2231
new_entry_key = (new_dirname, new_basename, new_id)
2232
current_new_minikind = \
2233
DirState._kind_to_minikind[current_new[1].kind]
2234
if current_new_minikind == 't':
2235
fingerprint = current_new[1].reference_revision or ''
2237
# We normally only insert or remove records, or update
2238
# them when it has significantly changed. Then we want to
2239
# erase its fingerprint. Unaffected records should
2240
# normally not be updated at all.
2243
# for safety disable variables
2244
new_path_utf8 = new_dirname = new_basename = new_id = \
2245
new_entry_key = None
2246
# 5 cases, we dont have a value that is strictly greater than everything, so
2247
# we make both end conditions explicit
2249
# old is finished: insert current_new into the state.
2250
self.update_minimal(new_entry_key, current_new_minikind,
2251
executable=current_new[1].executable,
2252
path_utf8=new_path_utf8, fingerprint=fingerprint)
2253
current_new = advance(new_iterator)
2254
elif not current_new:
2256
self._make_absent(current_old)
2257
current_old = advance(old_iterator)
2258
elif new_entry_key == current_old[0]:
2259
# same - common case
2260
# We're looking at the same path and id in both the dirstate
2261
# and inventory, so just need to update the fields in the
2262
# dirstate from the one in the inventory.
2263
# TODO: update the record if anything significant has changed.
2264
# the minimal required trigger is if the execute bit or cached
2266
if (current_old[1][0][3] != current_new[1].executable or
2267
current_old[1][0][0] != current_new_minikind):
2268
self.update_minimal(current_old[0], current_new_minikind,
2269
executable=current_new[1].executable,
2270
path_utf8=new_path_utf8, fingerprint=fingerprint)
2271
# both sides are dealt with, move on
2272
current_old = advance(old_iterator)
2273
current_new = advance(new_iterator)
2274
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2275
or (new_dirname == current_old[0][0]
2276
and new_entry_key[1:] < current_old[0][1:])):
2278
# add a entry for this and advance new
2279
self.update_minimal(new_entry_key, current_new_minikind,
2280
executable=current_new[1].executable,
2281
path_utf8=new_path_utf8, fingerprint=fingerprint)
2282
current_new = advance(new_iterator)
2284
# we've advanced past the place where the old key would be,
2285
# without seeing it in the new list. so it must be gone.
2286
self._make_absent(current_old)
2287
current_old = advance(old_iterator)
2288
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2289
self._id_index = None
2290
self._packed_stat_index = None
2292
def _make_absent(self, current_old):
2293
"""Mark current_old - an entry - as absent for tree 0.
2295
:return: True if this was the last details entry for the entry key:
2296
that is, if the underlying block has had the entry removed, thus
2297
shrinking in length.
2299
# build up paths that this id will be left at after the change is made,
2300
# so we can update their cross references in tree 0
2301
all_remaining_keys = set()
2302
# Dont check the working tree, because it's going.
2303
for details in current_old[1][1:]:
2304
if details[0] not in 'ar': # absent, relocated
2305
all_remaining_keys.add(current_old[0])
2306
elif details[0] == 'r': # relocated
2307
# record the key for the real path.
2308
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2309
# absent rows are not present at any path.
2310
last_reference = current_old[0] not in all_remaining_keys
2312
# the current row consists entire of the current item (being marked
2313
# absent), and relocated or absent entries for the other trees:
2314
# Remove it, its meaningless.
2315
block = self._find_block(current_old[0])
2316
entry_index, present = self._find_entry_index(current_old[0], block[1])
2317
assert present, 'could not find entry for %s' % (current_old,)
2318
block[1].pop(entry_index)
2319
# if we have an id_index in use, remove this key from it for this id.
2320
if self._id_index is not None:
2321
self._id_index[current_old[0][2]].remove(current_old[0])
2322
# update all remaining keys for this id to record it as absent. The
2323
# existing details may either be the record we are marking as deleted
2324
# (if there were other trees with the id present at this path), or may
2326
for update_key in all_remaining_keys:
2327
update_block_index, present = \
2328
self._find_block_index_from_key(update_key)
2329
assert present, 'could not find block for %s' % (update_key,)
2330
update_entry_index, present = \
2331
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2332
assert present, 'could not find entry for %s' % (update_key,)
2333
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2334
# it must not be absent at the moment
2335
assert update_tree_details[0][0] != 'a' # absent
2336
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2337
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2338
return last_reference
2340
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2341
packed_stat=None, size=0, path_utf8=None):
2342
"""Update an entry to the state in tree 0.
2344
This will either create a new entry at 'key' or update an existing one.
2345
It also makes sure that any other records which might mention this are
2348
:param key: (dir, name, file_id) for the new entry
2349
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2351
:param executable: Should the executable bit be set?
2352
:param fingerprint: Simple fingerprint for new entry: sha1 for files,
2353
referenced revision id for subtrees, etc.
2354
:param packed_stat: Packed stat value for new entry.
2355
:param size: Size information for new entry
2356
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2359
If packed_stat and fingerprint are not given, they're invalidated in
2362
block = self._find_block(key)[1]
2363
if packed_stat is None:
2364
packed_stat = DirState.NULLSTAT
2365
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2366
# sometimes present in the dirstate - this seems oddly inconsistent.
2368
entry_index, present = self._find_entry_index(key, block)
2369
new_details = (minikind, fingerprint, size, executable, packed_stat)
2370
id_index = self._get_id_index()
2372
# new entry, synthesis cross reference here,
2373
existing_keys = id_index.setdefault(key[2], set())
2374
if not existing_keys:
2375
# not currently in the state, simplest case
2376
new_entry = key, [new_details] + self._empty_parent_info()
2378
# present at one or more existing other paths.
2379
# grab one of them and use it to generate parent
2380
# relocation/absent entries.
2381
new_entry = key, [new_details]
2382
for other_key in existing_keys:
2383
# change the record at other to be a pointer to this new
2384
# record. The loop looks similar to the change to
2385
# relocations when updating an existing record but its not:
2386
# the test for existing kinds is different: this can be
2387
# factored out to a helper though.
2388
other_block_index, present = self._find_block_index_from_key(other_key)
2389
assert present, 'could not find block for %s' % (other_key,)
2390
other_entry_index, present = self._find_entry_index(other_key,
2391
self._dirblocks[other_block_index][1])
2392
assert present, 'could not find entry for %s' % (other_key,)
2393
assert path_utf8 is not None
2394
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2395
('r', path_utf8, 0, False, '')
2397
num_present_parents = self._num_present_parents()
2398
for lookup_index in xrange(1, num_present_parents + 1):
2399
# grab any one entry, use it to find the right path.
2400
# TODO: optimise this to reduce memory use in highly
2401
# fragmented situations by reusing the relocation
2403
update_block_index, present = \
2404
self._find_block_index_from_key(other_key)
2405
assert present, 'could not find block for %s' % (other_key,)
2406
update_entry_index, present = \
2407
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2408
assert present, 'could not find entry for %s' % (other_key,)
2409
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2410
if update_details[0] in 'ar': # relocated, absent
2411
# its a pointer or absent in lookup_index's tree, use
2413
new_entry[1].append(update_details)
2415
# we have the right key, make a pointer to it.
2416
pointer_path = osutils.pathjoin(*other_key[0:2])
2417
new_entry[1].append(('r', pointer_path, 0, False, ''))
2418
block.insert(entry_index, new_entry)
2419
existing_keys.add(key)
2421
# Does the new state matter?
2422
block[entry_index][1][0] = new_details
2423
# parents cannot be affected by what we do.
2424
# other occurences of this id can be found
2425
# from the id index.
2427
# tree index consistency: All other paths for this id in this tree
2428
# index must point to the correct path. We have to loop here because
2429
# we may have passed entries in the state with this file id already
2430
# that were absent - where parent entries are - and they need to be
2431
# converted to relocated.
2432
assert path_utf8 is not None
2433
for entry_key in id_index.setdefault(key[2], set()):
2434
# TODO:PROFILING: It might be faster to just update
2435
# rather than checking if we need to, and then overwrite
2436
# the one we are located at.
2437
if entry_key != key:
2438
# this file id is at a different path in one of the
2439
# other trees, so put absent pointers there
2440
# This is the vertical axis in the matrix, all pointing
2442
block_index, present = self._find_block_index_from_key(entry_key)
2444
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2446
self._dirblocks[block_index][1][entry_index][1][0] = \
2447
('r', path_utf8, 0, False, '')
2448
# add a containing dirblock if needed.
2449
if new_details[0] == 'd':
2450
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2451
block_index, present = self._find_block_index_from_key(subdir_key)
2453
self._dirblocks.insert(block_index, (subdir_key[0], []))
2455
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2457
def _validate(self):
2458
"""Check that invariants on the dirblock are correct.
2460
This can be useful in debugging; it shouldn't be necessary in
2463
This must be called with a lock held.
2465
# NOTE: This must always raise AssertionError not just assert,
2466
# otherwise it may not behave properly under python -O
2468
# TODO: All entries must have some content that's not 'a' or 'r',
2469
# otherwise it could just be removed.
2471
# TODO: All relocations must point directly to a real entry.
2473
# TODO: No repeated keys.
2476
from pprint import pformat
2477
self._read_dirblocks_if_needed()
2478
if len(self._dirblocks) > 0:
2479
if not self._dirblocks[0][0] == '':
2480
raise AssertionError(
2481
"dirblocks don't start with root block:\n" + \
2483
if len(self._dirblocks) > 1:
2484
if not self._dirblocks[1][0] == '':
2485
raise AssertionError(
2486
"dirblocks missing root directory:\n" + \
2488
# the dirblocks are sorted by their path components, name, and dir id
2489
dir_names = [d[0].split('/')
2490
for d in self._dirblocks[1:]]
2491
if dir_names != sorted(dir_names):
2492
raise AssertionError(
2493
"dir names are not in sorted order:\n" + \
2494
pformat(self._dirblocks) + \
2497
for dirblock in self._dirblocks:
2498
# within each dirblock, the entries are sorted by filename and
2500
for entry in dirblock[1]:
2501
if dirblock[0] != entry[0][0]:
2502
raise AssertionError(
2504
"doesn't match directory name in\n%r" %
2505
(entry, pformat(dirblock)))
2506
if dirblock[1] != sorted(dirblock[1]):
2507
raise AssertionError(
2508
"dirblock for %r is not sorted:\n%s" % \
2509
(dirblock[0], pformat(dirblock)))
2511
def check_valid_parent():
2512
"""Check that the current entry has a valid parent.
2514
This makes sure that the parent has a record,
2515
and that the parent isn't marked as "absent" in the
2516
current tree. (It is invalid to have a non-absent file in an absent
2519
if entry[0][0:2] == ('', ''):
2520
# There should be no parent for the root row
2522
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2523
if parent_entry == (None, None):
2524
raise AssertionError(
2525
"no parent entry for: %s in tree %s"
2526
% (this_path, tree_index))
2527
if parent_entry[1][tree_index][0] != 'd':
2528
raise AssertionError(
2529
"Parent entry for %s is not marked as a valid"
2530
" directory. %s" % (this_path, parent_entry,))
2532
# For each file id, for each tree: either
2533
# the file id is not present at all; all rows with that id in the
2534
# key have it marked as 'absent'
2535
# OR the file id is present under exactly one name; any other entries
2536
# that mention that id point to the correct name.
2538
# We check this with a dict per tree pointing either to the present
2539
# name, or None if absent.
2540
tree_count = self._num_present_parents() + 1
2541
id_path_maps = [dict() for i in range(tree_count)]
2542
# Make sure that all renamed entries point to the correct location.
2543
for entry in self._iter_entries():
2544
file_id = entry[0][2]
2545
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2546
if len(entry[1]) != tree_count:
2547
raise AssertionError(
2548
"wrong number of entry details for row\n%s" \
2549
",\nexpected %d" % \
2550
(pformat(entry), tree_count))
2551
absent_positions = 0
2552
for tree_index, tree_state in enumerate(entry[1]):
2553
this_tree_map = id_path_maps[tree_index]
2554
minikind = tree_state[0]
2555
if minikind in 'ar':
2556
absent_positions += 1
2557
# have we seen this id before in this column?
2558
if file_id in this_tree_map:
2559
previous_path, previous_loc = this_tree_map[file_id]
2560
# any later mention of this file must be consistent with
2561
# what was said before
2563
if previous_path is not None:
2564
raise AssertionError(
2565
"file %s is absent in row %r but also present " \
2567
(file_id, entry, previous_path))
2568
elif minikind == 'r':
2569
target_location = tree_state[1]
2570
if previous_path != target_location:
2571
raise AssertionError(
2572
"file %s relocation in row %r but also at %r" \
2573
% (file_id, entry, previous_path))
2575
# a file, directory, etc - may have been previously
2576
# pointed to by a relocation, which must point here
2577
if previous_path != this_path:
2578
raise AssertionError(
2579
"entry %r inconsistent with previous path %r "
2581
(entry, previous_path, previous_loc))
2582
check_valid_parent()
2585
# absent; should not occur anywhere else
2586
this_tree_map[file_id] = None, this_path
2587
elif minikind == 'r':
2588
# relocation, must occur at expected location
2589
this_tree_map[file_id] = tree_state[1], this_path
2591
this_tree_map[file_id] = this_path, this_path
2592
check_valid_parent()
2593
if absent_positions == tree_count:
2594
raise AssertionError(
2595
"entry %r has no data for any tree." % (entry,))
2597
def _wipe_state(self):
2598
"""Forget all state information about the dirstate."""
2599
self._header_state = DirState.NOT_IN_MEMORY
2600
self._dirblock_state = DirState.NOT_IN_MEMORY
2603
self._dirblocks = []
2604
self._id_index = None
2605
self._packed_stat_index = None
2606
self._end_of_header = None
2607
self._cutoff_time = None
2608
self._split_path_cache = {}
2610
def lock_read(self):
2611
"""Acquire a read lock on the dirstate."""
2612
if self._lock_token is not None:
2613
raise errors.LockContention(self._lock_token)
2614
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2615
# already in memory, we could read just the header and check for
2616
# any modification. If not modified, we can just leave things
2618
self._lock_token = lock.ReadLock(self._filename)
2619
self._lock_state = 'r'
2620
self._state_file = self._lock_token.f
2623
def lock_write(self):
2624
"""Acquire a write lock on the dirstate."""
2625
if self._lock_token is not None:
2626
raise errors.LockContention(self._lock_token)
2627
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2628
# already in memory, we could read just the header and check for
2629
# any modification. If not modified, we can just leave things
2631
self._lock_token = lock.WriteLock(self._filename)
2632
self._lock_state = 'w'
2633
self._state_file = self._lock_token.f
2637
"""Drop any locks held on the dirstate."""
2638
if self._lock_token is None:
2639
raise errors.LockNotHeld(self)
2640
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2641
# already in memory, we could read just the header and check for
2642
# any modification. If not modified, we can just leave things
2644
self._state_file = None
2645
self._lock_state = None
2646
self._lock_token.unlock()
2647
self._lock_token = None
2648
self._split_path_cache = {}
2650
def _requires_lock(self):
2651
"""Check that a lock is currently held by someone on the dirstate."""
2652
if not self._lock_token:
2653
raise errors.ObjectNotLocked(self)
2656
# Try to load the compiled form if possible
2658
from bzrlib._dirstate_helpers_c import (
2659
_read_dirblocks_c as _read_dirblocks,
2660
bisect_dirblock_c as bisect_dirblock,
2661
_bisect_path_left_c as _bisect_path_left,
2662
_bisect_path_right_c as _bisect_path_right,
2663
cmp_by_dirs_c as cmp_by_dirs,
2666
from bzrlib._dirstate_helpers_py import (
2667
_read_dirblocks_py as _read_dirblocks,
2668
bisect_dirblock_py as bisect_dirblock,
2669
_bisect_path_left_py as _bisect_path_left,
2670
_bisect_path_right_py as _bisect_path_right,
2671
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
