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
224
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
225
"""Convert stat values into a packed representation."""
226
# jam 20060614 it isn't really worth removing more entries if we
227
# are going to leave it in packed form.
228
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
229
# With all entries, filesize is 5.9M and read time is maybe 280ms
230
# well within the noise margin
232
# base64 encoding always adds a final newline, so strip it off
233
# The current version
234
return _encode(_pack('>LLLLLL'
235
, st.st_size, int(st.st_mtime), int(st.st_ctime)
236
, st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
237
# This is 0.060s / 1.520s faster by not encoding as much information
238
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
239
# This is not strictly faster than _encode(_pack())[:-1]
240
# return '%X.%X.%X.%X.%X.%X' % (
241
# st.st_size, int(st.st_mtime), int(st.st_ctime),
242
# st.st_dev, st.st_ino, st.st_mode)
243
# Similar to the _encode(_pack('>LL'))
244
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
247
class DirState(object):
248
"""Record directory and metadata state for fast access.
250
A dirstate is a specialised data structure for managing local working
251
tree state information. Its not yet well defined whether it is platform
252
specific, and if it is how we detect/parameterise that.
254
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
255
Unlike most bzr disk formats, DirStates must be locked for reading, using
256
lock_read. (This is an os file lock internally.) This is necessary
257
because the file can be rewritten in place.
259
DirStates must be explicitly written with save() to commit changes; just
260
unlocking them does not write the changes to disk.
263
_kind_to_minikind = {
269
'tree-reference': 't',
271
_minikind_to_kind = {
277
't': 'tree-reference',
279
_stat_to_minikind = {
284
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
285
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
288
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
289
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
290
# that we are sure a single record will always fit.
291
BISECT_PAGE_SIZE = 4096
294
IN_MEMORY_UNMODIFIED = 1
295
IN_MEMORY_MODIFIED = 2
297
# A pack_stat (the x's) that is just noise and will never match the output
300
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
302
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
303
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
305
def __init__(self, path):
306
"""Create a DirState object.
310
:attr _root_entrie: The root row of the directory/file information,
311
- contains the path to / - '', ''
312
- kind of 'directory',
313
- the file id of the root in utf8
316
- and no sha information.
317
:param path: The path at which the dirstate file on disk should live.
319
# _header_state and _dirblock_state represent the current state
320
# of the dirstate metadata and the per-row data respectiely.
321
# NOT_IN_MEMORY indicates that no data is in memory
322
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
323
# is the same as is on disk
324
# IN_MEMORY_MODIFIED indicates that we have a modified version
325
# of what is on disk.
326
# In future we will add more granularity, for instance _dirblock_state
327
# will probably support partially-in-memory as a separate variable,
328
# allowing for partially-in-memory unmodified and partially-in-memory
330
self._header_state = DirState.NOT_IN_MEMORY
331
self._dirblock_state = DirState.NOT_IN_MEMORY
335
self._state_file = None
336
self._filename = path
337
self._lock_token = None
338
self._lock_state = None
339
self._id_index = None
340
self._end_of_header = None
341
self._cutoff_time = None
342
self._split_path_cache = {}
343
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
344
if 'hashcache' in debug.debug_flags:
345
self._sha1_file = self._sha1_file_and_mutter
347
self._sha1_file = osutils.sha_file_by_name
348
# These two attributes provide a simple cache for lookups into the
349
# dirstate in-memory vectors. By probing respectively for the last
350
# block, and for the next entry, we save nearly 2 bisections per path
352
self._last_block_index = None
353
self._last_entry_index = None
357
(self.__class__.__name__, self._filename)
359
def add(self, path, file_id, kind, stat, fingerprint):
360
"""Add a path to be tracked.
362
:param path: The path within the dirstate - '' is the root, 'foo' is the
363
path foo within the root, 'foo/bar' is the path bar within foo
365
:param file_id: The file id of the path being added.
366
:param kind: The kind of the path, as a string like 'file',
368
:param stat: The output of os.lstat for the path.
369
:param fingerprint: The sha value of the file,
370
or the target of a symlink,
371
or the referenced revision id for tree-references,
372
or '' for directories.
375
# find the block its in.
376
# find the location in the block.
377
# check its not there
379
#------- copied from inventory.ensure_normalized_name - keep synced.
380
# --- normalized_filename wants a unicode basename only, so get one.
381
dirname, basename = osutils.split(path)
382
# we dont import normalized_filename directly because we want to be
383
# able to change the implementation at runtime for tests.
384
norm_name, can_access = osutils.normalized_filename(basename)
385
if norm_name != basename:
389
raise errors.InvalidNormalization(path)
390
# you should never have files called . or ..; just add the directory
391
# in the parent, or according to the special treatment for the root
392
if basename == '.' or basename == '..':
393
raise errors.InvalidEntryName(path)
394
# now that we've normalised, we need the correct utf8 path and
395
# dirname and basename elements. This single encode and split should be
396
# faster than three separate encodes.
397
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
398
dirname, basename = osutils.split(utf8path)
399
assert file_id.__class__ == str, \
400
"must be a utf8 file_id not %s" % (type(file_id))
401
# Make sure the file_id does not exist in this tree
402
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
403
if file_id_entry != (None, None):
404
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
405
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
406
info = '%s:%s' % (kind, path)
407
raise errors.DuplicateFileId(file_id, info)
408
first_key = (dirname, basename, '')
409
block_index, present = self._find_block_index_from_key(first_key)
411
# check the path is not in the tree
412
block = self._dirblocks[block_index][1]
413
entry_index, _ = self._find_entry_index(first_key, block)
414
while (entry_index < len(block) and
415
block[entry_index][0][0:2] == first_key[0:2]):
416
if block[entry_index][1][0][0] not in 'ar':
417
# this path is in the dirstate in the current tree.
418
raise Exception, "adding already added path!"
421
# The block where we want to put the file is not present. But it
422
# might be because the directory was empty, or not loaded yet. Look
423
# for a parent entry, if not found, raise NotVersionedError
424
parent_dir, parent_base = osutils.split(dirname)
425
parent_block_idx, parent_entry_idx, _, parent_present = \
426
self._get_block_entry_index(parent_dir, parent_base, 0)
427
if not parent_present:
428
raise errors.NotVersionedError(path, str(self))
429
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
430
block = self._dirblocks[block_index][1]
431
entry_key = (dirname, basename, file_id)
434
packed_stat = DirState.NULLSTAT
437
packed_stat = pack_stat(stat)
438
parent_info = self._empty_parent_info()
439
minikind = DirState._kind_to_minikind[kind]
441
entry_data = entry_key, [
442
(minikind, fingerprint, size, False, packed_stat),
444
elif kind == 'directory':
445
entry_data = entry_key, [
446
(minikind, '', 0, False, packed_stat),
448
elif kind == 'symlink':
449
entry_data = entry_key, [
450
(minikind, fingerprint, size, False, packed_stat),
452
elif kind == 'tree-reference':
453
entry_data = entry_key, [
454
(minikind, fingerprint, 0, False, packed_stat),
457
raise errors.BzrError('unknown kind %r' % kind)
458
entry_index, present = self._find_entry_index(entry_key, block)
460
block.insert(entry_index, entry_data)
462
assert block[entry_index][1][0][0] == 'a', " %r(%r) already added" % (basename, file_id)
463
block[entry_index][1][0] = entry_data[1][0]
465
if kind == 'directory':
466
# insert a new dirblock
467
self._ensure_block(block_index, entry_index, utf8path)
468
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
470
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
472
def _bisect(self, paths):
473
"""Bisect through the disk structure for specific rows.
475
:param paths: A list of paths to find
476
:return: A dict mapping path => entries for found entries. Missing
477
entries will not be in the map.
478
The list is not sorted, and entries will be populated
479
based on when they were read.
481
self._requires_lock()
482
# We need the file pointer to be right after the initial header block
483
self._read_header_if_needed()
484
# If _dirblock_state was in memory, we should just return info from
485
# there, this function is only meant to handle when we want to read
487
assert self._dirblock_state == DirState.NOT_IN_MEMORY
489
# The disk representation is generally info + '\0\n\0' at the end. But
490
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
491
# Because it means we can sync on the '\n'
492
state_file = self._state_file
493
file_size = os.fstat(state_file.fileno()).st_size
494
# We end up with 2 extra fields, we should have a trailing '\n' to
495
# ensure that we read the whole record, and we should have a precursur
496
# '' which ensures that we start after the previous '\n'
497
entry_field_count = self._fields_per_entry() + 1
499
low = self._end_of_header
500
high = file_size - 1 # Ignore the final '\0'
501
# Map from (dir, name) => entry
504
# Avoid infinite seeking
505
max_count = 30*len(paths)
507
# pending is a list of places to look.
508
# each entry is a tuple of low, high, dir_names
509
# low -> the first byte offset to read (inclusive)
510
# high -> the last byte offset (inclusive)
511
# dir_names -> The list of (dir, name) pairs that should be found in
512
# the [low, high] range
513
pending = [(low, high, paths)]
515
page_size = self._bisect_page_size
517
fields_to_entry = self._get_fields_to_entry()
520
low, high, cur_files = pending.pop()
522
if not cur_files or low >= high:
527
if count > max_count:
528
raise errors.BzrError('Too many seeks, most likely a bug.')
530
mid = max(low, (low+high-page_size)/2)
533
# limit the read size, so we don't end up reading data that we have
535
read_size = min(page_size, (high-mid)+1)
536
block = state_file.read(read_size)
539
entries = block.split('\n')
542
# We didn't find a '\n', so we cannot have found any records.
543
# So put this range back and try again. But we know we have to
544
# increase the page size, because a single read did not contain
545
# a record break (so records must be larger than page_size)
547
pending.append((low, high, cur_files))
550
# Check the first and last entries, in case they are partial, or if
551
# we don't care about the rest of this page
553
first_fields = entries[0].split('\0')
554
if len(first_fields) < entry_field_count:
555
# We didn't get the complete first entry
556
# so move start, and grab the next, which
557
# should be a full entry
558
start += len(entries[0])+1
559
first_fields = entries[1].split('\0')
562
if len(first_fields) <= 2:
563
# We didn't even get a filename here... what do we do?
564
# Try a large page size and repeat this query
566
pending.append((low, high, cur_files))
569
# Find what entries we are looking for, which occur before and
570
# after this first record.
573
first_path = first_fields[1] + '/' + first_fields[2]
575
first_path = first_fields[2]
576
first_loc = _bisect_path_left(cur_files, first_path)
578
# These exist before the current location
579
pre = cur_files[:first_loc]
580
# These occur after the current location, which may be in the
581
# data we read, or might be after the last entry
582
post = cur_files[first_loc:]
584
if post and len(first_fields) >= entry_field_count:
585
# We have files after the first entry
587
# Parse the last entry
588
last_entry_num = len(entries)-1
589
last_fields = entries[last_entry_num].split('\0')
590
if len(last_fields) < entry_field_count:
591
# The very last hunk was not complete,
592
# read the previous hunk
593
after = mid + len(block) - len(entries[-1])
595
last_fields = entries[last_entry_num].split('\0')
597
after = mid + len(block)
600
last_path = last_fields[1] + '/' + last_fields[2]
602
last_path = last_fields[2]
603
last_loc = _bisect_path_right(post, last_path)
605
middle_files = post[:last_loc]
606
post = post[last_loc:]
609
# We have files that should occur in this block
610
# (>= first, <= last)
611
# Either we will find them here, or we can mark them as
614
if middle_files[0] == first_path:
615
# We might need to go before this location
616
pre.append(first_path)
617
if middle_files[-1] == last_path:
618
post.insert(0, last_path)
620
# Find out what paths we have
621
paths = {first_path:[first_fields]}
622
# last_path might == first_path so we need to be
623
# careful if we should append rather than overwrite
624
if last_entry_num != first_entry_num:
625
paths.setdefault(last_path, []).append(last_fields)
626
for num in xrange(first_entry_num+1, last_entry_num):
627
# TODO: jam 20070223 We are already splitting here, so
628
# shouldn't we just split the whole thing rather
629
# than doing the split again in add_one_record?
630
fields = entries[num].split('\0')
632
path = fields[1] + '/' + fields[2]
635
paths.setdefault(path, []).append(fields)
637
for path in middle_files:
638
for fields in paths.get(path, []):
639
# offset by 1 because of the opening '\0'
640
# consider changing fields_to_entry to avoid the
642
entry = fields_to_entry(fields[1:])
643
found.setdefault(path, []).append(entry)
645
# Now we have split up everything into pre, middle, and post, and
646
# we have handled everything that fell in 'middle'.
647
# We add 'post' first, so that we prefer to seek towards the
648
# beginning, so that we will tend to go as early as we need, and
649
# then only seek forward after that.
651
pending.append((after, high, post))
653
pending.append((low, start-1, pre))
655
# Consider that we may want to return the directory entries in sorted
656
# order. For now, we just return them in whatever order we found them,
657
# and leave it up to the caller if they care if it is ordered or not.
660
def _bisect_dirblocks(self, dir_list):
661
"""Bisect through the disk structure to find entries in given dirs.
663
_bisect_dirblocks is meant to find the contents of directories, which
664
differs from _bisect, which only finds individual entries.
666
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
667
:return: A map from dir => entries_for_dir
669
# TODO: jam 20070223 A lot of the bisecting logic could be shared
670
# between this and _bisect. It would require parameterizing the
671
# inner loop with a function, though. We should evaluate the
672
# performance difference.
673
self._requires_lock()
674
# We need the file pointer to be right after the initial header block
675
self._read_header_if_needed()
676
# If _dirblock_state was in memory, we should just return info from
677
# there, this function is only meant to handle when we want to read
679
assert self._dirblock_state == DirState.NOT_IN_MEMORY
681
# The disk representation is generally info + '\0\n\0' at the end. But
682
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
683
# Because it means we can sync on the '\n'
684
state_file = self._state_file
685
file_size = os.fstat(state_file.fileno()).st_size
686
# We end up with 2 extra fields, we should have a trailing '\n' to
687
# ensure that we read the whole record, and we should have a precursur
688
# '' which ensures that we start after the previous '\n'
689
entry_field_count = self._fields_per_entry() + 1
691
low = self._end_of_header
692
high = file_size - 1 # Ignore the final '\0'
693
# Map from dir => entry
696
# Avoid infinite seeking
697
max_count = 30*len(dir_list)
699
# pending is a list of places to look.
700
# each entry is a tuple of low, high, dir_names
701
# low -> the first byte offset to read (inclusive)
702
# high -> the last byte offset (inclusive)
703
# dirs -> The list of directories that should be found in
704
# the [low, high] range
705
pending = [(low, high, dir_list)]
707
page_size = self._bisect_page_size
709
fields_to_entry = self._get_fields_to_entry()
712
low, high, cur_dirs = pending.pop()
714
if not cur_dirs or low >= high:
719
if count > max_count:
720
raise errors.BzrError('Too many seeks, most likely a bug.')
722
mid = max(low, (low+high-page_size)/2)
725
# limit the read size, so we don't end up reading data that we have
727
read_size = min(page_size, (high-mid)+1)
728
block = state_file.read(read_size)
731
entries = block.split('\n')
734
# We didn't find a '\n', so we cannot have found any records.
735
# So put this range back and try again. But we know we have to
736
# increase the page size, because a single read did not contain
737
# a record break (so records must be larger than page_size)
739
pending.append((low, high, cur_dirs))
742
# Check the first and last entries, in case they are partial, or if
743
# we don't care about the rest of this page
745
first_fields = entries[0].split('\0')
746
if len(first_fields) < entry_field_count:
747
# We didn't get the complete first entry
748
# so move start, and grab the next, which
749
# should be a full entry
750
start += len(entries[0])+1
751
first_fields = entries[1].split('\0')
754
if len(first_fields) <= 1:
755
# We didn't even get a dirname here... what do we do?
756
# Try a large page size and repeat this query
758
pending.append((low, high, cur_dirs))
761
# Find what entries we are looking for, which occur before and
762
# after this first record.
764
first_dir = first_fields[1]
765
first_loc = bisect.bisect_left(cur_dirs, first_dir)
767
# These exist before the current location
768
pre = cur_dirs[:first_loc]
769
# These occur after the current location, which may be in the
770
# data we read, or might be after the last entry
771
post = cur_dirs[first_loc:]
773
if post and len(first_fields) >= entry_field_count:
774
# We have records to look at after the first entry
776
# Parse the last entry
777
last_entry_num = len(entries)-1
778
last_fields = entries[last_entry_num].split('\0')
779
if len(last_fields) < entry_field_count:
780
# The very last hunk was not complete,
781
# read the previous hunk
782
after = mid + len(block) - len(entries[-1])
784
last_fields = entries[last_entry_num].split('\0')
786
after = mid + len(block)
788
last_dir = last_fields[1]
789
last_loc = bisect.bisect_right(post, last_dir)
791
middle_files = post[:last_loc]
792
post = post[last_loc:]
795
# We have files that should occur in this block
796
# (>= first, <= last)
797
# Either we will find them here, or we can mark them as
800
if middle_files[0] == first_dir:
801
# We might need to go before this location
802
pre.append(first_dir)
803
if middle_files[-1] == last_dir:
804
post.insert(0, last_dir)
806
# Find out what paths we have
807
paths = {first_dir:[first_fields]}
808
# last_dir might == first_dir so we need to be
809
# careful if we should append rather than overwrite
810
if last_entry_num != first_entry_num:
811
paths.setdefault(last_dir, []).append(last_fields)
812
for num in xrange(first_entry_num+1, last_entry_num):
813
# TODO: jam 20070223 We are already splitting here, so
814
# shouldn't we just split the whole thing rather
815
# than doing the split again in add_one_record?
816
fields = entries[num].split('\0')
817
paths.setdefault(fields[1], []).append(fields)
819
for cur_dir in middle_files:
820
for fields in paths.get(cur_dir, []):
821
# offset by 1 because of the opening '\0'
822
# consider changing fields_to_entry to avoid the
824
entry = fields_to_entry(fields[1:])
825
found.setdefault(cur_dir, []).append(entry)
827
# Now we have split up everything into pre, middle, and post, and
828
# we have handled everything that fell in 'middle'.
829
# We add 'post' first, so that we prefer to seek towards the
830
# beginning, so that we will tend to go as early as we need, and
831
# then only seek forward after that.
833
pending.append((after, high, post))
835
pending.append((low, start-1, pre))
839
def _bisect_recursive(self, paths):
840
"""Bisect for entries for all paths and their children.
842
This will use bisect to find all records for the supplied paths. It
843
will then continue to bisect for any records which are marked as
844
directories. (and renames?)
846
:param paths: A sorted list of (dir, name) pairs
847
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
848
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
850
# Map from (dir, name, file_id) => [tree_info]
853
found_dir_names = set()
855
# Directories that have been read
856
processed_dirs = set()
857
# Get the ball rolling with the first bisect for all entries.
858
newly_found = self._bisect(paths)
861
# Directories that need to be read
863
paths_to_search = set()
864
for entry_list in newly_found.itervalues():
865
for dir_name_id, trees_info in entry_list:
866
found[dir_name_id] = trees_info
867
found_dir_names.add(dir_name_id[:2])
869
for tree_info in trees_info:
870
minikind = tree_info[0]
873
# We already processed this one as a directory,
874
# we don't need to do the extra work again.
876
subdir, name, file_id = dir_name_id
877
path = osutils.pathjoin(subdir, name)
879
if path not in processed_dirs:
880
pending_dirs.add(path)
881
elif minikind == 'r':
882
# Rename, we need to directly search the target
883
# which is contained in the fingerprint column
884
dir_name = osutils.split(tree_info[1])
885
if dir_name[0] in pending_dirs:
886
# This entry will be found in the dir search
888
if dir_name not in found_dir_names:
889
paths_to_search.add(tree_info[1])
890
# Now we have a list of paths to look for directly, and
891
# directory blocks that need to be read.
892
# newly_found is mixing the keys between (dir, name) and path
893
# entries, but that is okay, because we only really care about the
895
newly_found = self._bisect(sorted(paths_to_search))
896
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
897
processed_dirs.update(pending_dirs)
900
def _empty_parent_info(self):
901
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
904
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
905
"""Ensure a block for dirname exists.
907
This function exists to let callers which know that there is a
908
directory dirname ensure that the block for it exists. This block can
909
fail to exist because of demand loading, or because a directory had no
910
children. In either case it is not an error. It is however an error to
911
call this if there is no parent entry for the directory, and thus the
912
function requires the coordinates of such an entry to be provided.
914
The root row is special cased and can be indicated with a parent block
917
:param parent_block_index: The index of the block in which dirname's row
919
:param parent_row_index: The index in the parent block where the row
921
:param dirname: The utf8 dirname to ensure there is a block for.
922
:return: The index for the block.
924
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
925
# This is the signature of the root row, and the
926
# contents-of-root row is always index 1
928
# the basename of the directory must be the end of its full name.
929
if not (parent_block_index == -1 and
930
parent_block_index == -1 and dirname == ''):
931
assert dirname.endswith(
932
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
933
block_index, present = self._find_block_index_from_key((dirname, '', ''))
935
## In future, when doing partial parsing, this should load and
936
# populate the entire block.
937
self._dirblocks.insert(block_index, (dirname, []))
940
def _entries_to_current_state(self, new_entries):
941
"""Load new_entries into self.dirblocks.
943
Process new_entries into the current state object, making them the active
944
state. The entries are grouped together by directory to form dirblocks.
946
:param new_entries: A sorted list of entries. This function does not sort
947
to prevent unneeded overhead when callers have a sorted list already.
950
assert new_entries[0][0][0:2] == ('', ''), \
951
"Missing root row %r" % (new_entries[0][0],)
952
# The two blocks here are deliberate: the root block and the
953
# contents-of-root block.
954
self._dirblocks = [('', []), ('', [])]
955
current_block = self._dirblocks[0][1]
958
append_entry = current_block.append
959
for entry in new_entries:
960
if entry[0][0] != current_dirname:
961
# new block - different dirname
963
current_dirname = entry[0][0]
964
self._dirblocks.append((current_dirname, current_block))
965
append_entry = current_block.append
966
# append the entry to the current block
968
self._split_root_dirblock_into_contents()
970
def _split_root_dirblock_into_contents(self):
971
"""Split the root dirblocks into root and contents-of-root.
973
After parsing by path, we end up with root entries and contents-of-root
974
entries in the same block. This loop splits them out again.
976
# The above loop leaves the "root block" entries mixed with the
977
# "contents-of-root block". But we don't want an if check on
978
# all entries, so instead we just fix it up here.
979
assert self._dirblocks[1] == ('', [])
981
contents_of_root_block = []
982
for entry in self._dirblocks[0][1]:
983
if not entry[0][1]: # This is a root entry
984
root_block.append(entry)
986
contents_of_root_block.append(entry)
987
self._dirblocks[0] = ('', root_block)
988
self._dirblocks[1] = ('', contents_of_root_block)
990
def _entry_to_line(self, entry):
991
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
993
:param entry: An entry_tuple as defined in the module docstring.
995
entire_entry = list(entry[0])
996
for tree_number, tree_data in enumerate(entry[1]):
997
# (minikind, fingerprint, size, executable, tree_specific_string)
998
entire_entry.extend(tree_data)
999
# 3 for the key, 5 for the fields per tree.
1000
tree_offset = 3 + tree_number * 5
1002
entire_entry[tree_offset + 0] = tree_data[0]
1004
entire_entry[tree_offset + 2] = str(tree_data[2])
1006
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1007
return '\0'.join(entire_entry)
1009
def _fields_per_entry(self):
1010
"""How many null separated fields should be in each entry row.
1012
Each line now has an extra '\n' field which is not used
1013
so we just skip over it
1015
3 fields for the key
1016
+ number of fields per tree_data (5) * tree count
1019
tree_count = 1 + self._num_present_parents()
1020
return 3 + 5 * tree_count + 1
1022
def _find_block(self, key, add_if_missing=False):
1023
"""Return the block that key should be present in.
1025
:param key: A dirstate entry key.
1026
:return: The block tuple.
1028
block_index, present = self._find_block_index_from_key(key)
1030
if not add_if_missing:
1031
# check to see if key is versioned itself - we might want to
1032
# add it anyway, because dirs with no entries dont get a
1033
# dirblock at parse time.
1034
# This is an uncommon branch to take: most dirs have children,
1035
# and most code works with versioned paths.
1036
parent_base, parent_name = osutils.split(key[0])
1037
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1038
# some parent path has not been added - its an error to add
1040
raise errors.NotVersionedError(key[0:2], str(self))
1041
self._dirblocks.insert(block_index, (key[0], []))
1042
return self._dirblocks[block_index]
1044
def _find_block_index_from_key(self, key):
1045
"""Find the dirblock index for a key.
1047
:return: The block index, True if the block for the key is present.
1049
if key[0:2] == ('', ''):
1052
if (self._last_block_index is not None and
1053
self._dirblocks[self._last_block_index][0] == key[0]):
1054
return self._last_block_index, True
1057
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1058
cache=self._split_path_cache)
1059
# _right returns one-past-where-key is so we have to subtract
1060
# one to use it. we use _right here because there are two
1061
# '' blocks - the root, and the contents of root
1062
# we always have a minimum of 2 in self._dirblocks: root and
1063
# root-contents, and for '', we get 2 back, so this is
1064
# simple and correct:
1065
present = (block_index < len(self._dirblocks) and
1066
self._dirblocks[block_index][0] == key[0])
1067
self._last_block_index = block_index
1068
# Reset the entry index cache to the beginning of the block.
1069
self._last_entry_index = -1
1070
return block_index, present
1072
def _find_entry_index(self, key, block):
1073
"""Find the entry index for a key in a block.
1075
:return: The entry index, True if the entry for the key is present.
1077
len_block = len(block)
1079
if self._last_entry_index is not None:
1081
entry_index = self._last_entry_index + 1
1082
# A hit is when the key is after the last slot, and before or
1083
# equal to the next slot.
1084
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1085
key <= block[entry_index][0]):
1086
self._last_entry_index = entry_index
1087
present = (block[entry_index][0] == key)
1088
return entry_index, present
1091
entry_index = bisect.bisect_left(block, (key, []))
1092
present = (entry_index < len_block and
1093
block[entry_index][0] == key)
1094
self._last_entry_index = entry_index
1095
return entry_index, present
1098
def from_tree(tree, dir_state_filename):
1099
"""Create a dirstate from a bzr Tree.
1101
:param tree: The tree which should provide parent information and
1103
:return: a DirState object which is currently locked for writing.
1104
(it was locked by DirState.initialize)
1106
result = DirState.initialize(dir_state_filename)
1110
parent_ids = tree.get_parent_ids()
1111
num_parents = len(parent_ids)
1113
for parent_id in parent_ids:
1114
parent_tree = tree.branch.repository.revision_tree(parent_id)
1115
parent_trees.append((parent_id, parent_tree))
1116
parent_tree.lock_read()
1117
result.set_parent_trees(parent_trees, [])
1118
result.set_state_from_inventory(tree.inventory)
1120
for revid, parent_tree in parent_trees:
1121
parent_tree.unlock()
1124
# The caller won't have a chance to unlock this, so make sure we
1130
def update_entry(self, entry, abspath, stat_value,
1131
_stat_to_minikind=_stat_to_minikind,
1132
_pack_stat=pack_stat):
1133
"""Update the entry based on what is actually on disk.
1135
:param entry: This is the dirblock entry for the file in question.
1136
:param abspath: The path on disk for this file.
1137
:param stat_value: (optional) if we already have done a stat on the
1139
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1143
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1147
packed_stat = _pack_stat(stat_value)
1148
(saved_minikind, saved_link_or_sha1, saved_file_size,
1149
saved_executable, saved_packed_stat) = entry[1][0]
1151
if (minikind == saved_minikind
1152
and packed_stat == saved_packed_stat):
1153
# The stat hasn't changed since we saved, so we can re-use the
1158
# size should also be in packed_stat
1159
if saved_file_size == stat_value.st_size:
1160
return saved_link_or_sha1
1162
# If we have gotten this far, that means that we need to actually
1163
# process this entry.
1166
link_or_sha1 = self._sha1_file(abspath)
1167
executable = self._is_executable(stat_value.st_mode,
1169
if self._cutoff_time is None:
1170
self._sha_cutoff_time()
1171
if (stat_value.st_mtime < self._cutoff_time
1172
and stat_value.st_ctime < self._cutoff_time):
1173
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1174
executable, packed_stat)
1176
entry[1][0] = ('f', '', stat_value.st_size,
1177
executable, DirState.NULLSTAT)
1178
elif minikind == 'd':
1180
entry[1][0] = ('d', '', 0, False, packed_stat)
1181
if saved_minikind != 'd':
1182
# This changed from something into a directory. Make sure we
1183
# have a directory block for it. This doesn't happen very
1184
# often, so this doesn't have to be super fast.
1185
block_index, entry_index, dir_present, file_present = \
1186
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1187
self._ensure_block(block_index, entry_index,
1188
osutils.pathjoin(entry[0][0], entry[0][1]))
1189
elif minikind == 'l':
1190
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1191
if self._cutoff_time is None:
1192
self._sha_cutoff_time()
1193
if (stat_value.st_mtime < self._cutoff_time
1194
and stat_value.st_ctime < self._cutoff_time):
1195
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1198
entry[1][0] = ('l', '', stat_value.st_size,
1199
False, DirState.NULLSTAT)
1200
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1203
def _sha_cutoff_time(self):
1204
"""Return cutoff time.
1206
Files modified more recently than this time are at risk of being
1207
undetectably modified and so can't be cached.
1209
# Cache the cutoff time as long as we hold a lock.
1210
# time.time() isn't super expensive (approx 3.38us), but
1211
# when you call it 50,000 times it adds up.
1212
# For comparison, os.lstat() costs 7.2us if it is hot.
1213
self._cutoff_time = int(time.time()) - 3
1214
return self._cutoff_time
1216
def _lstat(self, abspath, entry):
1217
"""Return the os.lstat value for this path."""
1218
return os.lstat(abspath)
1220
def _sha1_file_and_mutter(self, abspath):
1221
# when -Dhashcache is turned on, this is monkey-patched in to log
1223
trace.mutter("dirstate sha1 " + abspath)
1224
return osutils.sha_file_by_name(abspath)
1226
def _is_executable(self, mode, old_executable):
1227
"""Is this file executable?"""
1228
return bool(S_IEXEC & mode)
1230
def _is_executable_win32(self, mode, old_executable):
1231
"""On win32 the executable bit is stored in the dirstate."""
1232
return old_executable
1234
if sys.platform == 'win32':
1235
_is_executable = _is_executable_win32
1237
def _read_link(self, abspath, old_link):
1238
"""Read the target of a symlink"""
1239
# TODO: jam 200700301 On Win32, this could just return the value
1240
# already in memory. However, this really needs to be done at a
1241
# higher level, because there either won't be anything on disk,
1242
# or the thing on disk will be a file.
1243
return os.readlink(abspath)
1245
def get_ghosts(self):
1246
"""Return a list of the parent tree revision ids that are ghosts."""
1247
self._read_header_if_needed()
1250
def get_lines(self):
1251
"""Serialise the entire dirstate to a sequence of lines."""
1252
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1253
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1254
# read whats on disk.
1255
self._state_file.seek(0)
1256
return self._state_file.readlines()
1258
lines.append(self._get_parents_line(self.get_parent_ids()))
1259
lines.append(self._get_ghosts_line(self._ghosts))
1260
# append the root line which is special cased
1261
lines.extend(map(self._entry_to_line, self._iter_entries()))
1262
return self._get_output_lines(lines)
1264
def _get_ghosts_line(self, ghost_ids):
1265
"""Create a line for the state file for ghost information."""
1266
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1268
def _get_parents_line(self, parent_ids):
1269
"""Create a line for the state file for parents information."""
1270
return '\0'.join([str(len(parent_ids))] + parent_ids)
1272
def _get_fields_to_entry(self):
1273
"""Get a function which converts entry fields into a entry record.
1275
This handles size and executable, as well as parent records.
1277
:return: A function which takes a list of fields, and returns an
1278
appropriate record for storing in memory.
1280
# This is intentionally unrolled for performance
1281
num_present_parents = self._num_present_parents()
1282
if num_present_parents == 0:
1283
def fields_to_entry_0_parents(fields, _int=int):
1284
path_name_file_id_key = (fields[0], fields[1], fields[2])
1285
return (path_name_file_id_key, [
1287
fields[3], # minikind
1288
fields[4], # fingerprint
1289
_int(fields[5]), # size
1290
fields[6] == 'y', # executable
1291
fields[7], # packed_stat or revision_id
1293
return fields_to_entry_0_parents
1294
elif num_present_parents == 1:
1295
def fields_to_entry_1_parent(fields, _int=int):
1296
path_name_file_id_key = (fields[0], fields[1], fields[2])
1297
return (path_name_file_id_key, [
1299
fields[3], # minikind
1300
fields[4], # fingerprint
1301
_int(fields[5]), # size
1302
fields[6] == 'y', # executable
1303
fields[7], # packed_stat or revision_id
1306
fields[8], # minikind
1307
fields[9], # fingerprint
1308
_int(fields[10]), # size
1309
fields[11] == 'y', # executable
1310
fields[12], # packed_stat or revision_id
1313
return fields_to_entry_1_parent
1314
elif num_present_parents == 2:
1315
def fields_to_entry_2_parents(fields, _int=int):
1316
path_name_file_id_key = (fields[0], fields[1], fields[2])
1317
return (path_name_file_id_key, [
1319
fields[3], # minikind
1320
fields[4], # fingerprint
1321
_int(fields[5]), # size
1322
fields[6] == 'y', # executable
1323
fields[7], # packed_stat or revision_id
1326
fields[8], # minikind
1327
fields[9], # fingerprint
1328
_int(fields[10]), # size
1329
fields[11] == 'y', # executable
1330
fields[12], # packed_stat or revision_id
1333
fields[13], # minikind
1334
fields[14], # fingerprint
1335
_int(fields[15]), # size
1336
fields[16] == 'y', # executable
1337
fields[17], # packed_stat or revision_id
1340
return fields_to_entry_2_parents
1342
def fields_to_entry_n_parents(fields, _int=int):
1343
path_name_file_id_key = (fields[0], fields[1], fields[2])
1344
trees = [(fields[cur], # minikind
1345
fields[cur+1], # fingerprint
1346
_int(fields[cur+2]), # size
1347
fields[cur+3] == 'y', # executable
1348
fields[cur+4], # stat or revision_id
1349
) for cur in xrange(3, len(fields)-1, 5)]
1350
return path_name_file_id_key, trees
1351
return fields_to_entry_n_parents
1353
def get_parent_ids(self):
1354
"""Return a list of the parent tree ids for the directory state."""
1355
self._read_header_if_needed()
1356
return list(self._parents)
1358
def _get_block_entry_index(self, dirname, basename, tree_index):
1359
"""Get the coordinates for a path in the state structure.
1361
:param dirname: The utf8 dirname to lookup.
1362
:param basename: The utf8 basename to lookup.
1363
:param tree_index: The index of the tree for which this lookup should
1365
:return: A tuple describing where the path is located, or should be
1366
inserted. The tuple contains four fields: the block index, the row
1367
index, the directory is present (boolean), the entire path is
1368
present (boolean). There is no guarantee that either
1369
coordinate is currently reachable unless the found field for it is
1370
True. For instance, a directory not present in the searched tree
1371
may be returned with a value one greater than the current highest
1372
block offset. The directory present field will always be True when
1373
the path present field is True. The directory present field does
1374
NOT indicate that the directory is present in the searched tree,
1375
rather it indicates that there are at least some files in some
1378
self._read_dirblocks_if_needed()
1379
key = dirname, basename, ''
1380
block_index, present = self._find_block_index_from_key(key)
1382
# no such directory - return the dir index and 0 for the row.
1383
return block_index, 0, False, False
1384
block = self._dirblocks[block_index][1] # access the entries only
1385
entry_index, present = self._find_entry_index(key, block)
1386
# linear search through entries at this path to find the one
1388
while entry_index < len(block) and block[entry_index][0][1] == basename:
1389
if block[entry_index][1][tree_index][0] not in \
1390
('a', 'r'): # absent, relocated
1391
return block_index, entry_index, True, True
1393
return block_index, entry_index, True, False
1395
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1396
"""Get the dirstate entry for path in tree tree_index.
1398
If either file_id or path is supplied, it is used as the key to lookup.
1399
If both are supplied, the fastest lookup is used, and an error is
1400
raised if they do not both point at the same row.
1402
:param tree_index: The index of the tree we wish to locate this path
1403
in. If the path is present in that tree, the entry containing its
1404
details is returned, otherwise (None, None) is returned
1405
0 is the working tree, higher indexes are successive parent
1407
:param fileid_utf8: A utf8 file_id to look up.
1408
:param path_utf8: An utf8 path to be looked up.
1409
:return: The dirstate entry tuple for path, or (None, None)
1411
self._read_dirblocks_if_needed()
1412
if path_utf8 is not None:
1413
assert path_utf8.__class__ == str, ('path_utf8 is not a str: %s %s'
1414
% (type(path_utf8), path_utf8))
1415
# path lookups are faster
1416
dirname, basename = osutils.split(path_utf8)
1417
block_index, entry_index, dir_present, file_present = \
1418
self._get_block_entry_index(dirname, basename, tree_index)
1419
if not file_present:
1421
entry = self._dirblocks[block_index][1][entry_index]
1422
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1424
if entry[0][2] != fileid_utf8:
1425
raise errors.BzrError('integrity error ? : mismatching'
1426
' tree_index, file_id and path')
1429
assert fileid_utf8 is not None
1430
possible_keys = self._get_id_index().get(fileid_utf8, None)
1431
if not possible_keys:
1433
for key in possible_keys:
1434
block_index, present = \
1435
self._find_block_index_from_key(key)
1436
# strange, probably indicates an out of date
1437
# id index - for now, allow this.
1440
# WARNING: DO not change this code to use _get_block_entry_index
1441
# as that function is not suitable: it does not use the key
1442
# to lookup, and thus the wrong coordinates are returned.
1443
block = self._dirblocks[block_index][1]
1444
entry_index, present = self._find_entry_index(key, block)
1446
entry = self._dirblocks[block_index][1][entry_index]
1447
if entry[1][tree_index][0] in 'fdlt':
1448
# this is the result we are looking for: the
1449
# real home of this file_id in this tree.
1451
if entry[1][tree_index][0] == 'a':
1452
# there is no home for this entry in this tree
1454
assert entry[1][tree_index][0] == 'r', \
1455
"entry %r has invalid minikind %r for tree %r" \
1457
entry[1][tree_index][0],
1459
real_path = entry[1][tree_index][1]
1460
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1461
path_utf8=real_path)
1465
def initialize(cls, path):
1466
"""Create a new dirstate on path.
1468
The new dirstate will be an empty tree - that is it has no parents,
1469
and only a root node - which has id ROOT_ID.
1471
:param path: The name of the file for the dirstate.
1472
:return: A write-locked DirState object.
1474
# This constructs a new DirState object on a path, sets the _state_file
1475
# to a new empty file for that path. It then calls _set_data() with our
1476
# stock empty dirstate information - a root with ROOT_ID, no children,
1477
# and no parents. Finally it calls save() to ensure that this data will
1480
# root dir and root dir contents with no children.
1481
empty_tree_dirblocks = [('', []), ('', [])]
1482
# a new root directory, with a NULLSTAT.
1483
empty_tree_dirblocks[0][1].append(
1484
(('', '', inventory.ROOT_ID), [
1485
('d', '', 0, False, DirState.NULLSTAT),
1489
result._set_data([], empty_tree_dirblocks)
1496
def _inv_entry_to_details(self, inv_entry):
1497
"""Convert an inventory entry (from a revision tree) to state details.
1499
:param inv_entry: An inventory entry whose sha1 and link targets can be
1500
relied upon, and which has a revision set.
1501
:return: A details tuple - the details for a single tree at a path +
1504
kind = inv_entry.kind
1505
minikind = DirState._kind_to_minikind[kind]
1506
tree_data = inv_entry.revision
1507
assert tree_data, 'empty revision for the inv_entry %s.' % \
1509
if kind == 'directory':
1513
elif kind == 'symlink':
1514
fingerprint = inv_entry.symlink_target or ''
1517
elif kind == 'file':
1518
fingerprint = inv_entry.text_sha1 or ''
1519
size = inv_entry.text_size or 0
1520
executable = inv_entry.executable
1521
elif kind == 'tree-reference':
1522
fingerprint = inv_entry.reference_revision or ''
1526
raise Exception("can't pack %s" % inv_entry)
1527
return (minikind, fingerprint, size, executable, tree_data)
1529
def _iter_entries(self):
1530
"""Iterate over all the entries in the dirstate.
1532
Each yelt item is an entry in the standard format described in the
1533
docstring of bzrlib.dirstate.
1535
self._read_dirblocks_if_needed()
1536
for directory in self._dirblocks:
1537
for entry in directory[1]:
1540
def _get_id_index(self):
1541
"""Get an id index of self._dirblocks."""
1542
if self._id_index is None:
1544
for key, tree_details in self._iter_entries():
1545
id_index.setdefault(key[2], set()).add(key)
1546
self._id_index = id_index
1547
return self._id_index
1549
def _get_output_lines(self, lines):
1550
"""Format lines for final output.
1552
:param lines: A sequence of lines containing the parents list and the
1555
output_lines = [DirState.HEADER_FORMAT_3]
1556
lines.append('') # a final newline
1557
inventory_text = '\0\n\0'.join(lines)
1558
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1559
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1560
num_entries = len(lines)-3
1561
output_lines.append('num_entries: %s\n' % (num_entries,))
1562
output_lines.append(inventory_text)
1565
def _make_deleted_row(self, fileid_utf8, parents):
1566
"""Return a deleted row for fileid_utf8."""
1567
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1570
def _num_present_parents(self):
1571
"""The number of parent entries in each record row."""
1572
return len(self._parents) - len(self._ghosts)
1576
"""Construct a DirState on the file at path path.
1578
:return: An unlocked DirState object, associated with the given path.
1580
result = DirState(path)
1583
def _read_dirblocks_if_needed(self):
1584
"""Read in all the dirblocks from the file if they are not in memory.
1586
This populates self._dirblocks, and sets self._dirblock_state to
1587
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1590
self._read_header_if_needed()
1591
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1592
_read_dirblocks(self)
1594
def _read_header(self):
1595
"""This reads in the metadata header, and the parent ids.
1597
After reading in, the file should be positioned at the null
1598
just before the start of the first record in the file.
1600
:return: (expected crc checksum, number of entries, parent list)
1602
self._read_prelude()
1603
parent_line = self._state_file.readline()
1604
info = parent_line.split('\0')
1605
num_parents = int(info[0])
1606
assert num_parents == len(info)-2, 'incorrect parent info line'
1607
self._parents = info[1:-1]
1609
ghost_line = self._state_file.readline()
1610
info = ghost_line.split('\0')
1611
num_ghosts = int(info[1])
1612
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1613
self._ghosts = info[2:-1]
1614
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1615
self._end_of_header = self._state_file.tell()
1617
def _read_header_if_needed(self):
1618
"""Read the header of the dirstate file if needed."""
1619
# inline this as it will be called a lot
1620
if not self._lock_token:
1621
raise errors.ObjectNotLocked(self)
1622
if self._header_state == DirState.NOT_IN_MEMORY:
1625
def _read_prelude(self):
1626
"""Read in the prelude header of the dirstate file.
1628
This only reads in the stuff that is not connected to the crc
1629
checksum. The position will be correct to read in the rest of
1630
the file and check the checksum after this point.
1631
The next entry in the file should be the number of parents,
1632
and their ids. Followed by a newline.
1634
header = self._state_file.readline()
1635
assert header == DirState.HEADER_FORMAT_3, \
1636
'invalid header line: %r' % (header,)
1637
crc_line = self._state_file.readline()
1638
assert crc_line.startswith('crc32: '), 'missing crc32 checksum'
1639
self.crc_expected = int(crc_line[len('crc32: '):-1])
1640
num_entries_line = self._state_file.readline()
1641
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1642
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1645
"""Save any pending changes created during this session.
1647
We reuse the existing file, because that prevents race conditions with
1648
file creation, and use oslocks on it to prevent concurrent modification
1649
and reads - because dirstate's incremental data aggregation is not
1650
compatible with reading a modified file, and replacing a file in use by
1651
another process is impossible on Windows.
1653
A dirstate in read only mode should be smart enough though to validate
1654
that the file has not changed, and otherwise discard its cache and
1655
start over, to allow for fine grained read lock duration, so 'status'
1656
wont block 'commit' - for example.
1658
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1659
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1661
grabbed_write_lock = False
1662
if self._lock_state != 'w':
1663
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
1664
# Switch over to the new lock, as the old one may be closed.
1665
# TODO: jam 20070315 We should validate the disk file has
1666
# not changed contents. Since temporary_write_lock may
1667
# not be an atomic operation.
1668
self._lock_token = new_lock
1669
self._state_file = new_lock.f
1670
if not grabbed_write_lock:
1671
# We couldn't grab a write lock, so we switch back to a read one
1674
self._state_file.seek(0)
1675
self._state_file.writelines(self.get_lines())
1676
self._state_file.truncate()
1677
self._state_file.flush()
1678
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1679
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1681
if grabbed_write_lock:
1682
self._lock_token = self._lock_token.restore_read_lock()
1683
self._state_file = self._lock_token.f
1684
# TODO: jam 20070315 We should validate the disk file has
1685
# not changed contents. Since restore_read_lock may
1686
# not be an atomic operation.
1688
def _set_data(self, parent_ids, dirblocks):
1689
"""Set the full dirstate data in memory.
1691
This is an internal function used to completely replace the objects
1692
in memory state. It puts the dirstate into state 'full-dirty'.
1694
:param parent_ids: A list of parent tree revision ids.
1695
:param dirblocks: A list containing one tuple for each directory in the
1696
tree. Each tuple contains the directory path and a list of entries
1697
found in that directory.
1699
# our memory copy is now authoritative.
1700
self._dirblocks = dirblocks
1701
self._header_state = DirState.IN_MEMORY_MODIFIED
1702
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1703
self._parents = list(parent_ids)
1704
self._id_index = None
1706
def set_path_id(self, path, new_id):
1707
"""Change the id of path to new_id in the current working tree.
1709
:param path: The path inside the tree to set - '' is the root, 'foo'
1710
is the path foo in the root.
1711
:param new_id: The new id to assign to the path. This must be a utf8
1712
file id (not unicode, and not None).
1714
assert new_id.__class__ == str, \
1715
"path_id %r is not a plain string" % (new_id,)
1716
self._read_dirblocks_if_needed()
1718
# TODO: logic not written
1719
raise NotImplementedError(self.set_path_id)
1720
# TODO: check new id is unique
1721
entry = self._get_entry(0, path_utf8=path)
1722
if entry[0][2] == new_id:
1723
# Nothing to change.
1725
# mark the old path absent, and insert a new root path
1726
self._make_absent(entry)
1727
self.update_minimal(('', '', new_id), 'd',
1728
path_utf8='', packed_stat=entry[1][0][4])
1729
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1730
if self._id_index is not None:
1731
self._id_index.setdefault(new_id, set()).add(entry[0])
1733
def set_parent_trees(self, trees, ghosts):
1734
"""Set the parent trees for the dirstate.
1736
:param trees: A list of revision_id, tree tuples. tree must be provided
1737
even if the revision_id refers to a ghost: supply an empty tree in
1739
:param ghosts: A list of the revision_ids that are ghosts at the time
1742
# TODO: generate a list of parent indexes to preserve to save
1743
# processing specific parent trees. In the common case one tree will
1744
# be preserved - the left most parent.
1745
# TODO: if the parent tree is a dirstate, we might want to walk them
1746
# all by path in parallel for 'optimal' common-case performance.
1747
# generate new root row.
1748
self._read_dirblocks_if_needed()
1749
# TODO future sketch: Examine the existing parents to generate a change
1750
# map and then walk the new parent trees only, mapping them into the
1751
# dirstate. Walk the dirstate at the same time to remove unreferenced
1754
# sketch: loop over all entries in the dirstate, cherry picking
1755
# entries from the parent trees, if they are not ghost trees.
1756
# after we finish walking the dirstate, all entries not in the dirstate
1757
# are deletes, so we want to append them to the end as per the design
1758
# discussions. So do a set difference on ids with the parents to
1759
# get deletes, and add them to the end.
1760
# During the update process we need to answer the following questions:
1761
# - find other keys containing a fileid in order to create cross-path
1762
# links. We dont't trivially use the inventory from other trees
1763
# because this leads to either double touching, or to accessing
1765
# - find other keys containing a path
1766
# We accumulate each entry via this dictionary, including the root
1769
# we could do parallel iterators, but because file id data may be
1770
# scattered throughout, we dont save on index overhead: we have to look
1771
# at everything anyway. We can probably save cycles by reusing parent
1772
# data and doing an incremental update when adding an additional
1773
# parent, but for now the common cases are adding a new parent (merge),
1774
# and replacing completely (commit), and commit is more common: so
1775
# optimise merge later.
1777
# ---- start generation of full tree mapping data
1778
# what trees should we use?
1779
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
1780
# how many trees do we end up with
1781
parent_count = len(parent_trees)
1783
# one: the current tree
1784
for entry in self._iter_entries():
1785
# skip entries not in the current tree
1786
if entry[1][0][0] in ('a', 'r'): # absent, relocated
1788
by_path[entry[0]] = [entry[1][0]] + \
1789
[DirState.NULL_PARENT_DETAILS] * parent_count
1790
id_index[entry[0][2]] = set([entry[0]])
1792
# now the parent trees:
1793
for tree_index, tree in enumerate(parent_trees):
1794
# the index is off by one, adjust it.
1795
tree_index = tree_index + 1
1796
# when we add new locations for a fileid we need these ranges for
1797
# any fileid in this tree as we set the by_path[id] to:
1798
# already_processed_tree_details + new_details + new_location_suffix
1799
# the suffix is from tree_index+1:parent_count+1.
1800
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
1801
# now stitch in all the entries from this tree
1802
for path, entry in tree.inventory.iter_entries_by_dir():
1803
# here we process each trees details for each item in the tree.
1804
# we first update any existing entries for the id at other paths,
1805
# then we either create or update the entry for the id at the
1806
# right path, and finally we add (if needed) a mapping from
1807
# file_id to this path. We do it in this order to allow us to
1808
# avoid checking all known paths for the id when generating a
1809
# new entry at this path: by adding the id->path mapping last,
1810
# all the mappings are valid and have correct relocation
1811
# records where needed.
1812
file_id = entry.file_id
1813
path_utf8 = path.encode('utf8')
1814
dirname, basename = osutils.split(path_utf8)
1815
new_entry_key = (dirname, basename, file_id)
1816
# tree index consistency: All other paths for this id in this tree
1817
# index must point to the correct path.
1818
for entry_key in id_index.setdefault(file_id, set()):
1819
# TODO:PROFILING: It might be faster to just update
1820
# rather than checking if we need to, and then overwrite
1821
# the one we are located at.
1822
if entry_key != new_entry_key:
1823
# this file id is at a different path in one of the
1824
# other trees, so put absent pointers there
1825
# This is the vertical axis in the matrix, all pointing
1827
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
1828
# by path consistency: Insert into an existing path record (trivial), or
1829
# add a new one with relocation pointers for the other tree indexes.
1830
if new_entry_key in id_index[file_id]:
1831
# there is already an entry where this data belongs, just insert it.
1832
by_path[new_entry_key][tree_index] = \
1833
self._inv_entry_to_details(entry)
1835
# add relocated entries to the horizontal axis - this row
1836
# mapping from path,id. We need to look up the correct path
1837
# for the indexes from 0 to tree_index -1
1839
for lookup_index in xrange(tree_index):
1840
# boundary case: this is the first occurence of file_id
1841
# so there are no id_indexs, possibly take this out of
1843
if not len(id_index[file_id]):
1844
new_details.append(DirState.NULL_PARENT_DETAILS)
1846
# grab any one entry, use it to find the right path.
1847
# TODO: optimise this to reduce memory use in highly
1848
# fragmented situations by reusing the relocation
1850
a_key = iter(id_index[file_id]).next()
1851
if by_path[a_key][lookup_index][0] in ('r', 'a'):
1852
# its a pointer or missing statement, use it as is.
1853
new_details.append(by_path[a_key][lookup_index])
1855
# we have the right key, make a pointer to it.
1856
real_path = ('/'.join(a_key[0:2])).strip('/')
1857
new_details.append(('r', real_path, 0, False, ''))
1858
new_details.append(self._inv_entry_to_details(entry))
1859
new_details.extend(new_location_suffix)
1860
by_path[new_entry_key] = new_details
1861
id_index[file_id].add(new_entry_key)
1862
# --- end generation of full tree mappings
1864
# sort and output all the entries
1865
new_entries = self._sort_entries(by_path.items())
1866
self._entries_to_current_state(new_entries)
1867
self._parents = [rev_id for rev_id, tree in trees]
1868
self._ghosts = list(ghosts)
1869
self._header_state = DirState.IN_MEMORY_MODIFIED
1870
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1871
self._id_index = id_index
1873
def _sort_entries(self, entry_list):
1874
"""Given a list of entries, sort them into the right order.
1876
This is done when constructing a new dirstate from trees - normally we
1877
try to keep everything in sorted blocks all the time, but sometimes
1878
it's easier to sort after the fact.
1881
# sort by: directory parts, file name, file id
1882
return entry[0][0].split('/'), entry[0][1], entry[0][2]
1883
return sorted(entry_list, key=_key)
1885
def set_state_from_inventory(self, new_inv):
1886
"""Set new_inv as the current state.
1888
This API is called by tree transform, and will usually occur with
1889
existing parent trees.
1891
:param new_inv: The inventory object to set current state from.
1893
if 'evil' in debug.debug_flags:
1894
trace.mutter_callsite(1,
1895
"set_state_from_inventory called; please mutate the tree instead")
1896
self._read_dirblocks_if_needed()
1898
# Two iterators: current data and new data, both in dirblock order.
1899
# We zip them together, which tells about entries that are new in the
1900
# inventory, or removed in the inventory, or present in both and
1903
# You might think we could just synthesize a new dirstate directly
1904
# since we're processing it in the right order. However, we need to
1905
# also consider there may be any number of parent trees and relocation
1906
# pointers, and we don't want to duplicate that here.
1907
new_iterator = new_inv.iter_entries_by_dir()
1908
# we will be modifying the dirstate, so we need a stable iterator. In
1909
# future we might write one, for now we just clone the state into a
1910
# list - which is a shallow copy.
1911
old_iterator = iter(list(self._iter_entries()))
1912
# both must have roots so this is safe:
1913
current_new = new_iterator.next()
1914
current_old = old_iterator.next()
1915
def advance(iterator):
1917
return iterator.next()
1918
except StopIteration:
1920
while current_new or current_old:
1921
# skip entries in old that are not really there
1922
if current_old and current_old[1][0][0] in ('r', 'a'):
1923
# relocated or absent
1924
current_old = advance(old_iterator)
1927
# convert new into dirblock style
1928
new_path_utf8 = current_new[0].encode('utf8')
1929
new_dirname, new_basename = osutils.split(new_path_utf8)
1930
new_id = current_new[1].file_id
1931
new_entry_key = (new_dirname, new_basename, new_id)
1932
current_new_minikind = \
1933
DirState._kind_to_minikind[current_new[1].kind]
1934
if current_new_minikind == 't':
1935
fingerprint = current_new[1].reference_revision or ''
1937
# We normally only insert or remove records, or update
1938
# them when it has significantly changed. Then we want to
1939
# erase its fingerprint. Unaffected records should
1940
# normally not be updated at all.
1943
# for safety disable variables
1944
new_path_utf8 = new_dirname = new_basename = new_id = \
1945
new_entry_key = None
1946
# 5 cases, we dont have a value that is strictly greater than everything, so
1947
# we make both end conditions explicit
1949
# old is finished: insert current_new into the state.
1950
self.update_minimal(new_entry_key, current_new_minikind,
1951
executable=current_new[1].executable,
1952
path_utf8=new_path_utf8, fingerprint=fingerprint)
1953
current_new = advance(new_iterator)
1954
elif not current_new:
1956
self._make_absent(current_old)
1957
current_old = advance(old_iterator)
1958
elif new_entry_key == current_old[0]:
1959
# same - common case
1960
# We're looking at the same path and id in both the dirstate
1961
# and inventory, so just need to update the fields in the
1962
# dirstate from the one in the inventory.
1963
# TODO: update the record if anything significant has changed.
1964
# the minimal required trigger is if the execute bit or cached
1966
if (current_old[1][0][3] != current_new[1].executable or
1967
current_old[1][0][0] != current_new_minikind):
1968
self.update_minimal(current_old[0], current_new_minikind,
1969
executable=current_new[1].executable,
1970
path_utf8=new_path_utf8, fingerprint=fingerprint)
1971
# both sides are dealt with, move on
1972
current_old = advance(old_iterator)
1973
current_new = advance(new_iterator)
1974
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
1975
or (new_dirname == current_old[0][0]
1976
and new_entry_key[1:] < current_old[0][1:])):
1978
# add a entry for this and advance new
1979
self.update_minimal(new_entry_key, current_new_minikind,
1980
executable=current_new[1].executable,
1981
path_utf8=new_path_utf8, fingerprint=fingerprint)
1982
current_new = advance(new_iterator)
1984
# we've advanced past the place where the old key would be,
1985
# without seeing it in the new list. so it must be gone.
1986
self._make_absent(current_old)
1987
current_old = advance(old_iterator)
1988
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1989
self._id_index = None
1991
def _make_absent(self, current_old):
1992
"""Mark current_old - an entry - as absent for tree 0.
1994
:return: True if this was the last details entry for the entry key:
1995
that is, if the underlying block has had the entry removed, thus
1996
shrinking in length.
1998
# build up paths that this id will be left at after the change is made,
1999
# so we can update their cross references in tree 0
2000
all_remaining_keys = set()
2001
# Dont check the working tree, because it's going.
2002
for details in current_old[1][1:]:
2003
if details[0] not in ('a', 'r'): # absent, relocated
2004
all_remaining_keys.add(current_old[0])
2005
elif details[0] == 'r': # relocated
2006
# record the key for the real path.
2007
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2008
# absent rows are not present at any path.
2009
last_reference = current_old[0] not in all_remaining_keys
2011
# the current row consists entire of the current item (being marked
2012
# absent), and relocated or absent entries for the other trees:
2013
# Remove it, its meaningless.
2014
block = self._find_block(current_old[0])
2015
entry_index, present = self._find_entry_index(current_old[0], block[1])
2016
assert present, 'could not find entry for %s' % (current_old,)
2017
block[1].pop(entry_index)
2018
# if we have an id_index in use, remove this key from it for this id.
2019
if self._id_index is not None:
2020
self._id_index[current_old[0][2]].remove(current_old[0])
2021
# update all remaining keys for this id to record it as absent. The
2022
# existing details may either be the record we are marking as deleted
2023
# (if there were other trees with the id present at this path), or may
2025
for update_key in all_remaining_keys:
2026
update_block_index, present = \
2027
self._find_block_index_from_key(update_key)
2028
assert present, 'could not find block for %s' % (update_key,)
2029
update_entry_index, present = \
2030
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2031
assert present, 'could not find entry for %s' % (update_key,)
2032
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2033
# it must not be absent at the moment
2034
assert update_tree_details[0][0] != 'a' # absent
2035
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2036
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2037
return last_reference
2039
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2040
packed_stat=None, size=0, path_utf8=None):
2041
"""Update an entry to the state in tree 0.
2043
This will either create a new entry at 'key' or update an existing one.
2044
It also makes sure that any other records which might mention this are
2047
:param key: (dir, name, file_id) for the new entry
2048
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2050
:param executable: Should the executable bit be set?
2051
:param fingerprint: Simple fingerprint for new entry: sha1 for files,
2052
referenced revision id for subtrees, etc.
2053
:param packed_stat: Packed stat value for new entry.
2054
:param size: Size information for new entry
2055
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2058
If packed_stat and fingerprint are not given, they're invalidated in
2061
block = self._find_block(key)[1]
2062
if packed_stat is None:
2063
packed_stat = DirState.NULLSTAT
2064
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2065
# sometimes present in the dirstate - this seems oddly inconsistent.
2067
entry_index, present = self._find_entry_index(key, block)
2068
new_details = (minikind, fingerprint, size, executable, packed_stat)
2069
id_index = self._get_id_index()
2071
# new entry, synthesis cross reference here,
2072
existing_keys = id_index.setdefault(key[2], set())
2073
if not existing_keys:
2074
# not currently in the state, simplest case
2075
new_entry = key, [new_details] + self._empty_parent_info()
2077
# present at one or more existing other paths.
2078
# grab one of them and use it to generate parent
2079
# relocation/absent entries.
2080
new_entry = key, [new_details]
2081
for other_key in existing_keys:
2082
# change the record at other to be a pointer to this new
2083
# record. The loop looks similar to the change to
2084
# relocations when updating an existing record but its not:
2085
# the test for existing kinds is different: this can be
2086
# factored out to a helper though.
2087
other_block_index, present = self._find_block_index_from_key(other_key)
2088
assert present, 'could not find block for %s' % (other_key,)
2089
other_entry_index, present = self._find_entry_index(other_key,
2090
self._dirblocks[other_block_index][1])
2091
assert present, 'could not find entry for %s' % (other_key,)
2092
assert path_utf8 is not None
2093
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2094
('r', path_utf8, 0, False, '')
2096
num_present_parents = self._num_present_parents()
2097
for lookup_index in xrange(1, num_present_parents + 1):
2098
# grab any one entry, use it to find the right path.
2099
# TODO: optimise this to reduce memory use in highly
2100
# fragmented situations by reusing the relocation
2102
update_block_index, present = \
2103
self._find_block_index_from_key(other_key)
2104
assert present, 'could not find block for %s' % (other_key,)
2105
update_entry_index, present = \
2106
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2107
assert present, 'could not find entry for %s' % (other_key,)
2108
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2109
if update_details[0] in ('r', 'a'): # relocated, absent
2110
# its a pointer or absent in lookup_index's tree, use
2112
new_entry[1].append(update_details)
2114
# we have the right key, make a pointer to it.
2115
pointer_path = osutils.pathjoin(*other_key[0:2])
2116
new_entry[1].append(('r', pointer_path, 0, False, ''))
2117
block.insert(entry_index, new_entry)
2118
existing_keys.add(key)
2120
# Does the new state matter?
2121
block[entry_index][1][0] = new_details
2122
# parents cannot be affected by what we do.
2123
# other occurences of this id can be found
2124
# from the id index.
2126
# tree index consistency: All other paths for this id in this tree
2127
# index must point to the correct path. We have to loop here because
2128
# we may have passed entries in the state with this file id already
2129
# that were absent - where parent entries are - and they need to be
2130
# converted to relocated.
2131
assert path_utf8 is not None
2132
for entry_key in id_index.setdefault(key[2], set()):
2133
# TODO:PROFILING: It might be faster to just update
2134
# rather than checking if we need to, and then overwrite
2135
# the one we are located at.
2136
if entry_key != key:
2137
# this file id is at a different path in one of the
2138
# other trees, so put absent pointers there
2139
# This is the vertical axis in the matrix, all pointing
2141
block_index, present = self._find_block_index_from_key(entry_key)
2143
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2145
self._dirblocks[block_index][1][entry_index][1][0] = \
2146
('r', path_utf8, 0, False, '')
2147
# add a containing dirblock if needed.
2148
if new_details[0] == 'd':
2149
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2150
block_index, present = self._find_block_index_from_key(subdir_key)
2152
self._dirblocks.insert(block_index, (subdir_key[0], []))
2154
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2156
def _validate(self):
2157
"""Check that invariants on the dirblock are correct.
2159
This can be useful in debugging; it shouldn't be necessary in
2162
This must be called with a lock held.
2164
# NOTE: This must always raise AssertionError not just assert,
2165
# otherwise it may not behave properly under python -O
2167
# TODO: All entries must have some content that's not 'a' or 'r',
2168
# otherwise it could just be removed.
2170
# TODO: All relocations must point directly to a real entry.
2172
# TODO: No repeated keys.
2175
from pprint import pformat
2176
self._read_dirblocks_if_needed()
2177
if len(self._dirblocks) > 0:
2178
if not self._dirblocks[0][0] == '':
2179
raise AssertionError(
2180
"dirblocks don't start with root block:\n" + \
2182
if len(self._dirblocks) > 1:
2183
if not self._dirblocks[1][0] == '':
2184
raise AssertionError(
2185
"dirblocks missing root directory:\n" + \
2187
# the dirblocks are sorted by their path components, name, and dir id
2188
dir_names = [d[0].split('/')
2189
for d in self._dirblocks[1:]]
2190
if dir_names != sorted(dir_names):
2191
raise AssertionError(
2192
"dir names are not in sorted order:\n" + \
2193
pformat(self._dirblocks) + \
2196
for dirblock in self._dirblocks:
2197
# within each dirblock, the entries are sorted by filename and
2199
for entry in dirblock[1]:
2200
if dirblock[0] != entry[0][0]:
2201
raise AssertionError(
2203
"doesn't match directory name in\n%r" %
2204
(entry, pformat(dirblock)))
2205
if dirblock[1] != sorted(dirblock[1]):
2206
raise AssertionError(
2207
"dirblock for %r is not sorted:\n%s" % \
2208
(dirblock[0], pformat(dirblock)))
2211
def check_valid_parent():
2212
"""Check that the current entry has a valid parent.
2214
This makes sure that the parent has a record,
2215
and that the parent isn't marked as "absent" in the
2216
current tree. (It is invalid to have a non-absent file in an absent
2219
if entry[0][0:2] == ('', ''):
2220
# There should be no parent for the root row
2222
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2223
if parent_entry == (None, None):
2224
raise AssertionError(
2225
"no parent entry for: %s in tree %s"
2226
% (this_path, tree_index))
2227
if parent_entry[1][tree_index][0] != 'd':
2228
raise AssertionError(
2229
"Parent entry for %s is not marked as a valid"
2230
" directory. %s" % (this_path, parent_entry,))
2232
# For each file id, for each tree: either
2233
# the file id is not present at all; all rows with that id in the
2234
# key have it marked as 'absent'
2235
# OR the file id is present under exactly one name; any other entries
2236
# that mention that id point to the correct name.
2238
# We check this with a dict per tree pointing either to the present
2239
# name, or None if absent.
2240
tree_count = self._num_present_parents() + 1
2241
id_path_maps = [dict() for i in range(tree_count)]
2242
# Make sure that all renamed entries point to the correct location.
2243
for entry in self._iter_entries():
2244
file_id = entry[0][2]
2245
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2246
if len(entry[1]) != tree_count:
2247
raise AssertionError(
2248
"wrong number of entry details for row\n%s" \
2249
",\nexpected %d" % \
2250
(pformat(entry), tree_count))
2251
for tree_index, tree_state in enumerate(entry[1]):
2252
this_tree_map = id_path_maps[tree_index]
2253
minikind = tree_state[0]
2254
# have we seen this id before in this column?
2255
if file_id in this_tree_map:
2256
previous_path = this_tree_map[file_id]
2257
# any later mention of this file must be consistent with
2258
# what was said before
2260
if previous_path is not None:
2261
raise AssertionError(
2262
"file %s is absent in row %r but also present " \
2264
(file_id, entry, previous_path))
2265
elif minikind == 'r':
2266
target_location = tree_state[1]
2267
if previous_path != target_location:
2268
raise AssertionError(
2269
"file %s relocation in row %r but also at %r" \
2270
% (file_id, entry, previous_path))
2272
# a file, directory, etc - may have been previously
2273
# pointed to by a relocation, which must point here
2274
if previous_path != this_path:
2275
raise AssertionError(
2276
"entry %r inconsistent with previous path %r" % \
2277
(entry, previous_path))
2278
check_valid_parent()
2281
# absent; should not occur anywhere else
2282
this_tree_map[file_id] = None
2283
elif minikind == 'r':
2284
# relocation, must occur at expected location
2285
this_tree_map[file_id] = tree_state[1]
2287
this_tree_map[file_id] = this_path
2288
check_valid_parent()
2290
def _wipe_state(self):
2291
"""Forget all state information about the dirstate."""
2292
self._header_state = DirState.NOT_IN_MEMORY
2293
self._dirblock_state = DirState.NOT_IN_MEMORY
2296
self._dirblocks = []
2297
self._id_index = None
2298
self._end_of_header = None
2299
self._cutoff_time = None
2300
self._split_path_cache = {}
2302
def lock_read(self):
2303
"""Acquire a read lock on the dirstate."""
2304
if self._lock_token is not None:
2305
raise errors.LockContention(self._lock_token)
2306
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2307
# already in memory, we could read just the header and check for
2308
# any modification. If not modified, we can just leave things
2310
self._lock_token = lock.ReadLock(self._filename)
2311
self._lock_state = 'r'
2312
self._state_file = self._lock_token.f
2315
def lock_write(self):
2316
"""Acquire a write lock on the dirstate."""
2317
if self._lock_token is not None:
2318
raise errors.LockContention(self._lock_token)
2319
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2320
# already in memory, we could read just the header and check for
2321
# any modification. If not modified, we can just leave things
2323
self._lock_token = lock.WriteLock(self._filename)
2324
self._lock_state = 'w'
2325
self._state_file = self._lock_token.f
2329
"""Drop any locks held on the dirstate."""
2330
if self._lock_token is None:
2331
raise errors.LockNotHeld(self)
2332
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2333
# already in memory, we could read just the header and check for
2334
# any modification. If not modified, we can just leave things
2336
self._state_file = None
2337
self._lock_state = None
2338
self._lock_token.unlock()
2339
self._lock_token = None
2340
self._split_path_cache = {}
2342
def _requires_lock(self):
2343
"""Check that a lock is currently held by someone on the dirstate."""
2344
if not self._lock_token:
2345
raise errors.ObjectNotLocked(self)
2348
# Try to load the compiled form if possible
2350
from bzrlib._dirstate_helpers_c import (
2351
_read_dirblocks_c as _read_dirblocks,
2352
bisect_dirblock_c as bisect_dirblock,
2353
_bisect_path_left_c as _bisect_path_left,
2354
_bisect_path_right_c as _bisect_path_right,
2355
cmp_by_dirs_c as cmp_by_dirs,
2358
from bzrlib._dirstate_helpers_py import (
2359
_read_dirblocks_py as _read_dirblocks,
2360
bisect_dirblock_py as bisect_dirblock,
2361
_bisect_path_left_py as _bisect_path_left,
2362
_bisect_path_right_py as _bisect_path_right,
2363
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
