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/parameterize 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
301
# A pack_stat (the x's) that is just noise and will never match the output
304
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
306
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
307
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
309
def __init__(self, path):
310
"""Create a DirState object.
312
:param path: The path at which the dirstate file on disk should live.
314
# _header_state and _dirblock_state represent the current state
315
# of the dirstate metadata and the per-row data respectiely.
316
# NOT_IN_MEMORY indicates that no data is in memory
317
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
318
# is the same as is on disk
319
# IN_MEMORY_MODIFIED indicates that we have a modified version
320
# of what is on disk.
321
# In future we will add more granularity, for instance _dirblock_state
322
# will probably support partially-in-memory as a separate variable,
323
# allowing for partially-in-memory unmodified and partially-in-memory
325
self._header_state = DirState.NOT_IN_MEMORY
326
self._dirblock_state = DirState.NOT_IN_MEMORY
327
self._consistency = DirState.CONSISTENT
331
self._state_file = None
332
self._filename = path
333
self._lock_token = None
334
self._lock_state = None
335
self._id_index = None
336
# a map from packed_stat to sha's.
337
self._packed_stat_index = None
338
self._end_of_header = None
339
self._cutoff_time = None
340
self._split_path_cache = {}
341
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
342
if 'hashcache' in debug.debug_flags:
343
self._sha1_file = self._sha1_file_and_mutter
345
self._sha1_file = osutils.sha_file_by_name
346
# These two attributes provide a simple cache for lookups into the
347
# dirstate in-memory vectors. By probing respectively for the last
348
# block, and for the next entry, we save nearly 2 bisections per path
350
self._last_block_index = None
351
self._last_entry_index = None
355
(self.__class__.__name__, self._filename)
357
def add(self, path, file_id, kind, stat, fingerprint):
358
"""Add a path to be tracked.
360
:param path: The path within the dirstate - '' is the root, 'foo' is the
361
path foo within the root, 'foo/bar' is the path bar within foo
363
:param file_id: The file id of the path being added.
364
:param kind: The kind of the path, as a string like 'file',
366
:param stat: The output of os.lstat for the path.
367
:param fingerprint: The sha value of the file,
368
or the target of a symlink,
369
or the referenced revision id for tree-references,
370
or '' for directories.
373
# find the block its in.
374
# find the location in the block.
375
# check its not there
377
#------- copied from inventory.ensure_normalized_name - keep synced.
378
# --- normalized_filename wants a unicode basename only, so get one.
379
dirname, basename = osutils.split(path)
380
# we dont import normalized_filename directly because we want to be
381
# able to change the implementation at runtime for tests.
382
norm_name, can_access = osutils.normalized_filename(basename)
383
if norm_name != basename:
387
raise errors.InvalidNormalization(path)
388
# you should never have files called . or ..; just add the directory
389
# in the parent, or according to the special treatment for the root
390
if basename == '.' or basename == '..':
391
raise errors.InvalidEntryName(path)
392
# now that we've normalised, we need the correct utf8 path and
393
# dirname and basename elements. This single encode and split should be
394
# faster than three separate encodes.
395
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
396
dirname, basename = osutils.split(utf8path)
397
assert file_id.__class__ == str, \
398
"must be a utf8 file_id not %s" % (type(file_id))
399
# Make sure the file_id does not exist in this tree
400
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
401
if file_id_entry != (None, None):
402
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
403
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
404
info = '%s:%s' % (kind, path)
405
raise errors.DuplicateFileId(file_id, info)
406
first_key = (dirname, basename, '')
407
block_index, present = self._find_block_index_from_key(first_key)
409
# check the path is not in the tree
410
block = self._dirblocks[block_index][1]
411
entry_index, _ = self._find_entry_index(first_key, block)
412
while (entry_index < len(block) and
413
block[entry_index][0][0:2] == first_key[0:2]):
414
if block[entry_index][1][0][0] not in 'ar':
415
# this path is in the dirstate in the current tree.
416
raise Exception, "adding already added path!"
419
# The block where we want to put the file is not present. But it
420
# might be because the directory was empty, or not loaded yet. Look
421
# for a parent entry, if not found, raise NotVersionedError
422
parent_dir, parent_base = osutils.split(dirname)
423
parent_block_idx, parent_entry_idx, _, parent_present = \
424
self._get_block_entry_index(parent_dir, parent_base, 0)
425
if not parent_present:
426
raise errors.NotVersionedError(path, str(self))
427
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
428
block = self._dirblocks[block_index][1]
429
entry_key = (dirname, basename, file_id)
432
packed_stat = DirState.NULLSTAT
435
packed_stat = pack_stat(stat)
436
parent_info = self._empty_parent_info()
437
minikind = DirState._kind_to_minikind[kind]
439
entry_data = entry_key, [
440
(minikind, fingerprint, size, False, packed_stat),
442
elif kind == 'directory':
443
entry_data = entry_key, [
444
(minikind, '', 0, False, packed_stat),
446
elif kind == 'symlink':
447
entry_data = entry_key, [
448
(minikind, fingerprint, size, False, packed_stat),
450
elif kind == 'tree-reference':
451
entry_data = entry_key, [
452
(minikind, fingerprint, 0, False, packed_stat),
455
raise errors.BzrError('unknown kind %r' % kind)
456
entry_index, present = self._find_entry_index(entry_key, block)
458
block.insert(entry_index, entry_data)
460
assert block[entry_index][1][0][0] == 'a', " %r(%r) already added" % (basename, file_id)
461
block[entry_index][1][0] = entry_data[1][0]
463
if kind == 'directory':
464
# insert a new dirblock
465
self._ensure_block(block_index, entry_index, utf8path)
466
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
468
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
470
def _bisect(self, paths):
471
"""Bisect through the disk structure for specific rows.
473
:param paths: A list of paths to find
474
:return: A dict mapping path => entries for found entries. Missing
475
entries will not be in the map.
476
The list is not sorted, and entries will be populated
477
based on when they were read.
479
self._requires_lock()
480
# We need the file pointer to be right after the initial header block
481
self._read_header_if_needed()
482
# If _dirblock_state was in memory, we should just return info from
483
# there, this function is only meant to handle when we want to read
485
assert self._dirblock_state == DirState.NOT_IN_MEMORY
487
# The disk representation is generally info + '\0\n\0' at the end. But
488
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
489
# Because it means we can sync on the '\n'
490
state_file = self._state_file
491
file_size = os.fstat(state_file.fileno()).st_size
492
# We end up with 2 extra fields, we should have a trailing '\n' to
493
# ensure that we read the whole record, and we should have a precursur
494
# '' which ensures that we start after the previous '\n'
495
entry_field_count = self._fields_per_entry() + 1
497
low = self._end_of_header
498
high = file_size - 1 # Ignore the final '\0'
499
# Map from (dir, name) => entry
502
# Avoid infinite seeking
503
max_count = 30*len(paths)
505
# pending is a list of places to look.
506
# each entry is a tuple of low, high, dir_names
507
# low -> the first byte offset to read (inclusive)
508
# high -> the last byte offset (inclusive)
509
# dir_names -> The list of (dir, name) pairs that should be found in
510
# the [low, high] range
511
pending = [(low, high, paths)]
513
page_size = self._bisect_page_size
515
fields_to_entry = self._get_fields_to_entry()
518
low, high, cur_files = pending.pop()
520
if not cur_files or low >= high:
525
if count > max_count:
526
raise errors.BzrError('Too many seeks, most likely a bug.')
528
mid = max(low, (low+high-page_size)/2)
531
# limit the read size, so we don't end up reading data that we have
533
read_size = min(page_size, (high-mid)+1)
534
block = state_file.read(read_size)
537
entries = block.split('\n')
540
# We didn't find a '\n', so we cannot have found any records.
541
# So put this range back and try again. But we know we have to
542
# increase the page size, because a single read did not contain
543
# a record break (so records must be larger than page_size)
545
pending.append((low, high, cur_files))
548
# Check the first and last entries, in case they are partial, or if
549
# we don't care about the rest of this page
551
first_fields = entries[0].split('\0')
552
if len(first_fields) < entry_field_count:
553
# We didn't get the complete first entry
554
# so move start, and grab the next, which
555
# should be a full entry
556
start += len(entries[0])+1
557
first_fields = entries[1].split('\0')
560
if len(first_fields) <= 2:
561
# We didn't even get a filename here... what do we do?
562
# Try a large page size and repeat this query
564
pending.append((low, high, cur_files))
567
# Find what entries we are looking for, which occur before and
568
# after this first record.
571
first_path = first_fields[1] + '/' + first_fields[2]
573
first_path = first_fields[2]
574
first_loc = _bisect_path_left(cur_files, first_path)
576
# These exist before the current location
577
pre = cur_files[:first_loc]
578
# These occur after the current location, which may be in the
579
# data we read, or might be after the last entry
580
post = cur_files[first_loc:]
582
if post and len(first_fields) >= entry_field_count:
583
# We have files after the first entry
585
# Parse the last entry
586
last_entry_num = len(entries)-1
587
last_fields = entries[last_entry_num].split('\0')
588
if len(last_fields) < entry_field_count:
589
# The very last hunk was not complete,
590
# read the previous hunk
591
after = mid + len(block) - len(entries[-1])
593
last_fields = entries[last_entry_num].split('\0')
595
after = mid + len(block)
598
last_path = last_fields[1] + '/' + last_fields[2]
600
last_path = last_fields[2]
601
last_loc = _bisect_path_right(post, last_path)
603
middle_files = post[:last_loc]
604
post = post[last_loc:]
607
# We have files that should occur in this block
608
# (>= first, <= last)
609
# Either we will find them here, or we can mark them as
612
if middle_files[0] == first_path:
613
# We might need to go before this location
614
pre.append(first_path)
615
if middle_files[-1] == last_path:
616
post.insert(0, last_path)
618
# Find out what paths we have
619
paths = {first_path:[first_fields]}
620
# last_path might == first_path so we need to be
621
# careful if we should append rather than overwrite
622
if last_entry_num != first_entry_num:
623
paths.setdefault(last_path, []).append(last_fields)
624
for num in xrange(first_entry_num+1, last_entry_num):
625
# TODO: jam 20070223 We are already splitting here, so
626
# shouldn't we just split the whole thing rather
627
# than doing the split again in add_one_record?
628
fields = entries[num].split('\0')
630
path = fields[1] + '/' + fields[2]
633
paths.setdefault(path, []).append(fields)
635
for path in middle_files:
636
for fields in paths.get(path, []):
637
# offset by 1 because of the opening '\0'
638
# consider changing fields_to_entry to avoid the
640
entry = fields_to_entry(fields[1:])
641
found.setdefault(path, []).append(entry)
643
# Now we have split up everything into pre, middle, and post, and
644
# we have handled everything that fell in 'middle'.
645
# We add 'post' first, so that we prefer to seek towards the
646
# beginning, so that we will tend to go as early as we need, and
647
# then only seek forward after that.
649
pending.append((after, high, post))
651
pending.append((low, start-1, pre))
653
# Consider that we may want to return the directory entries in sorted
654
# order. For now, we just return them in whatever order we found them,
655
# and leave it up to the caller if they care if it is ordered or not.
658
def _bisect_dirblocks(self, dir_list):
659
"""Bisect through the disk structure to find entries in given dirs.
661
_bisect_dirblocks is meant to find the contents of directories, which
662
differs from _bisect, which only finds individual entries.
664
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
665
:return: A map from dir => entries_for_dir
667
# TODO: jam 20070223 A lot of the bisecting logic could be shared
668
# between this and _bisect. It would require parameterizing the
669
# inner loop with a function, though. We should evaluate the
670
# performance difference.
671
self._requires_lock()
672
# We need the file pointer to be right after the initial header block
673
self._read_header_if_needed()
674
# If _dirblock_state was in memory, we should just return info from
675
# there, this function is only meant to handle when we want to read
677
assert self._dirblock_state == DirState.NOT_IN_MEMORY
679
# The disk representation is generally info + '\0\n\0' at the end. But
680
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
681
# Because it means we can sync on the '\n'
682
state_file = self._state_file
683
file_size = os.fstat(state_file.fileno()).st_size
684
# We end up with 2 extra fields, we should have a trailing '\n' to
685
# ensure that we read the whole record, and we should have a precursur
686
# '' which ensures that we start after the previous '\n'
687
entry_field_count = self._fields_per_entry() + 1
689
low = self._end_of_header
690
high = file_size - 1 # Ignore the final '\0'
691
# Map from dir => entry
694
# Avoid infinite seeking
695
max_count = 30*len(dir_list)
697
# pending is a list of places to look.
698
# each entry is a tuple of low, high, dir_names
699
# low -> the first byte offset to read (inclusive)
700
# high -> the last byte offset (inclusive)
701
# dirs -> The list of directories that should be found in
702
# the [low, high] range
703
pending = [(low, high, dir_list)]
705
page_size = self._bisect_page_size
707
fields_to_entry = self._get_fields_to_entry()
710
low, high, cur_dirs = pending.pop()
712
if not cur_dirs or low >= high:
717
if count > max_count:
718
raise errors.BzrError('Too many seeks, most likely a bug.')
720
mid = max(low, (low+high-page_size)/2)
723
# limit the read size, so we don't end up reading data that we have
725
read_size = min(page_size, (high-mid)+1)
726
block = state_file.read(read_size)
729
entries = block.split('\n')
732
# We didn't find a '\n', so we cannot have found any records.
733
# So put this range back and try again. But we know we have to
734
# increase the page size, because a single read did not contain
735
# a record break (so records must be larger than page_size)
737
pending.append((low, high, cur_dirs))
740
# Check the first and last entries, in case they are partial, or if
741
# we don't care about the rest of this page
743
first_fields = entries[0].split('\0')
744
if len(first_fields) < entry_field_count:
745
# We didn't get the complete first entry
746
# so move start, and grab the next, which
747
# should be a full entry
748
start += len(entries[0])+1
749
first_fields = entries[1].split('\0')
752
if len(first_fields) <= 1:
753
# We didn't even get a dirname here... what do we do?
754
# Try a large page size and repeat this query
756
pending.append((low, high, cur_dirs))
759
# Find what entries we are looking for, which occur before and
760
# after this first record.
762
first_dir = first_fields[1]
763
first_loc = bisect.bisect_left(cur_dirs, first_dir)
765
# These exist before the current location
766
pre = cur_dirs[:first_loc]
767
# These occur after the current location, which may be in the
768
# data we read, or might be after the last entry
769
post = cur_dirs[first_loc:]
771
if post and len(first_fields) >= entry_field_count:
772
# We have records to look at after the first entry
774
# Parse the last entry
775
last_entry_num = len(entries)-1
776
last_fields = entries[last_entry_num].split('\0')
777
if len(last_fields) < entry_field_count:
778
# The very last hunk was not complete,
779
# read the previous hunk
780
after = mid + len(block) - len(entries[-1])
782
last_fields = entries[last_entry_num].split('\0')
784
after = mid + len(block)
786
last_dir = last_fields[1]
787
last_loc = bisect.bisect_right(post, last_dir)
789
middle_files = post[:last_loc]
790
post = post[last_loc:]
793
# We have files that should occur in this block
794
# (>= first, <= last)
795
# Either we will find them here, or we can mark them as
798
if middle_files[0] == first_dir:
799
# We might need to go before this location
800
pre.append(first_dir)
801
if middle_files[-1] == last_dir:
802
post.insert(0, last_dir)
804
# Find out what paths we have
805
paths = {first_dir:[first_fields]}
806
# last_dir might == first_dir so we need to be
807
# careful if we should append rather than overwrite
808
if last_entry_num != first_entry_num:
809
paths.setdefault(last_dir, []).append(last_fields)
810
for num in xrange(first_entry_num+1, last_entry_num):
811
# TODO: jam 20070223 We are already splitting here, so
812
# shouldn't we just split the whole thing rather
813
# than doing the split again in add_one_record?
814
fields = entries[num].split('\0')
815
paths.setdefault(fields[1], []).append(fields)
817
for cur_dir in middle_files:
818
for fields in paths.get(cur_dir, []):
819
# offset by 1 because of the opening '\0'
820
# consider changing fields_to_entry to avoid the
822
entry = fields_to_entry(fields[1:])
823
found.setdefault(cur_dir, []).append(entry)
825
# Now we have split up everything into pre, middle, and post, and
826
# we have handled everything that fell in 'middle'.
827
# We add 'post' first, so that we prefer to seek towards the
828
# beginning, so that we will tend to go as early as we need, and
829
# then only seek forward after that.
831
pending.append((after, high, post))
833
pending.append((low, start-1, pre))
837
def _bisect_recursive(self, paths):
838
"""Bisect for entries for all paths and their children.
840
This will use bisect to find all records for the supplied paths. It
841
will then continue to bisect for any records which are marked as
842
directories. (and renames?)
844
:param paths: A sorted list of (dir, name) pairs
845
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
846
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
848
# Map from (dir, name, file_id) => [tree_info]
851
found_dir_names = set()
853
# Directories that have been read
854
processed_dirs = set()
855
# Get the ball rolling with the first bisect for all entries.
856
newly_found = self._bisect(paths)
859
# Directories that need to be read
861
paths_to_search = set()
862
for entry_list in newly_found.itervalues():
863
for dir_name_id, trees_info in entry_list:
864
found[dir_name_id] = trees_info
865
found_dir_names.add(dir_name_id[:2])
867
for tree_info in trees_info:
868
minikind = tree_info[0]
871
# We already processed this one as a directory,
872
# we don't need to do the extra work again.
874
subdir, name, file_id = dir_name_id
875
path = osutils.pathjoin(subdir, name)
877
if path not in processed_dirs:
878
pending_dirs.add(path)
879
elif minikind == 'r':
880
# Rename, we need to directly search the target
881
# which is contained in the fingerprint column
882
dir_name = osutils.split(tree_info[1])
883
if dir_name[0] in pending_dirs:
884
# This entry will be found in the dir search
886
if dir_name not in found_dir_names:
887
paths_to_search.add(tree_info[1])
888
# Now we have a list of paths to look for directly, and
889
# directory blocks that need to be read.
890
# newly_found is mixing the keys between (dir, name) and path
891
# entries, but that is okay, because we only really care about the
893
newly_found = self._bisect(sorted(paths_to_search))
894
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
895
processed_dirs.update(pending_dirs)
898
def _discard_merge_parents(self):
899
"""Discard any parents trees beyond the first.
901
Note that if this fails the dirstate is corrupted.
903
After this function returns the dirstate contains 2 trees, neither of
906
self._read_header_if_needed()
907
parents = self.get_parent_ids()
910
# only require all dirblocks if we are doing a full-pass removal.
911
self._read_dirblocks_if_needed()
912
dead_patterns = set([('a', 'r'), ('a', 'a'), ('r', 'r'), ('r', 'a')])
913
def iter_entries_removable():
914
for block in self._dirblocks:
915
deleted_positions = []
916
for pos, entry in enumerate(block[1]):
918
if (entry[1][0][0], entry[1][1][0]) in dead_patterns:
919
deleted_positions.append(pos)
920
if deleted_positions:
921
if len(deleted_positions) == len(block[1]):
924
for pos in reversed(deleted_positions):
926
# if the first parent is a ghost:
927
if parents[0] in self.get_ghosts():
928
empty_parent = [DirState.NULL_PARENT_DETAILS]
929
for entry in iter_entries_removable():
930
entry[1][1:] = empty_parent
932
for entry in iter_entries_removable():
936
self._parents = [parents[0]]
937
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
938
self._header_state = DirState.IN_MEMORY_MODIFIED
940
def _empty_parent_info(self):
941
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
944
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
945
"""Ensure a block for dirname exists.
947
This function exists to let callers which know that there is a
948
directory dirname ensure that the block for it exists. This block can
949
fail to exist because of demand loading, or because a directory had no
950
children. In either case it is not an error. It is however an error to
951
call this if there is no parent entry for the directory, and thus the
952
function requires the coordinates of such an entry to be provided.
954
The root row is special cased and can be indicated with a parent block
957
:param parent_block_index: The index of the block in which dirname's row
959
:param parent_row_index: The index in the parent block where the row
961
:param dirname: The utf8 dirname to ensure there is a block for.
962
:return: The index for the block.
964
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
965
# This is the signature of the root row, and the
966
# contents-of-root row is always index 1
968
# the basename of the directory must be the end of its full name.
969
if not (parent_block_index == -1 and
970
parent_block_index == -1 and dirname == ''):
971
assert dirname.endswith(
972
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
973
block_index, present = self._find_block_index_from_key((dirname, '', ''))
975
## In future, when doing partial parsing, this should load and
976
# populate the entire block.
977
self._dirblocks.insert(block_index, (dirname, []))
980
def _entries_to_current_state(self, new_entries):
981
"""Load new_entries into self.dirblocks.
983
Process new_entries into the current state object, making them the active
984
state. The entries are grouped together by directory to form dirblocks.
986
:param new_entries: A sorted list of entries. This function does not sort
987
to prevent unneeded overhead when callers have a sorted list already.
990
assert new_entries[0][0][0:2] == ('', ''), \
991
"Missing root row %r" % (new_entries[0][0],)
992
# The two blocks here are deliberate: the root block and the
993
# contents-of-root block.
994
self._dirblocks = [('', []), ('', [])]
995
current_block = self._dirblocks[0][1]
998
append_entry = current_block.append
999
for entry in new_entries:
1000
if entry[0][0] != current_dirname:
1001
# new block - different dirname
1003
current_dirname = entry[0][0]
1004
self._dirblocks.append((current_dirname, current_block))
1005
append_entry = current_block.append
1006
# append the entry to the current block
1008
self._split_root_dirblock_into_contents()
1010
def _split_root_dirblock_into_contents(self):
1011
"""Split the root dirblocks into root and contents-of-root.
1013
After parsing by path, we end up with root entries and contents-of-root
1014
entries in the same block. This loop splits them out again.
1016
# The above loop leaves the "root block" entries mixed with the
1017
# "contents-of-root block". But we don't want an if check on
1018
# all entries, so instead we just fix it up here.
1019
assert self._dirblocks[1] == ('', [])
1021
contents_of_root_block = []
1022
for entry in self._dirblocks[0][1]:
1023
if not entry[0][1]: # This is a root entry
1024
root_block.append(entry)
1026
contents_of_root_block.append(entry)
1027
self._dirblocks[0] = ('', root_block)
1028
self._dirblocks[1] = ('', contents_of_root_block)
1030
def _entry_to_line(self, entry):
1031
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
1033
:param entry: An entry_tuple as defined in the module docstring.
1035
entire_entry = list(entry[0])
1036
for tree_number, tree_data in enumerate(entry[1]):
1037
# (minikind, fingerprint, size, executable, tree_specific_string)
1038
entire_entry.extend(tree_data)
1039
# 3 for the key, 5 for the fields per tree.
1040
tree_offset = 3 + tree_number * 5
1042
entire_entry[tree_offset + 0] = tree_data[0]
1044
entire_entry[tree_offset + 2] = str(tree_data[2])
1046
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1047
return '\0'.join(entire_entry)
1049
def _fields_per_entry(self):
1050
"""How many null separated fields should be in each entry row.
1052
Each line now has an extra '\n' field which is not used
1053
so we just skip over it
1055
3 fields for the key
1056
+ number of fields per tree_data (5) * tree count
1059
tree_count = 1 + self._num_present_parents()
1060
return 3 + 5 * tree_count + 1
1062
def _find_block(self, key, add_if_missing=False):
1063
"""Return the block that key should be present in.
1065
:param key: A dirstate entry key.
1066
:return: The block tuple.
1068
block_index, present = self._find_block_index_from_key(key)
1070
if not add_if_missing:
1071
# check to see if key is versioned itself - we might want to
1072
# add it anyway, because dirs with no entries dont get a
1073
# dirblock at parse time.
1074
# This is an uncommon branch to take: most dirs have children,
1075
# and most code works with versioned paths.
1076
parent_base, parent_name = osutils.split(key[0])
1077
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1078
# some parent path has not been added - its an error to add
1080
raise errors.NotVersionedError(key[0:2], str(self))
1081
self._dirblocks.insert(block_index, (key[0], []))
1082
return self._dirblocks[block_index]
1084
def _find_block_index_from_key(self, key):
1085
"""Find the dirblock index for a key.
1087
:return: The block index, True if the block for the key is present.
1089
if key[0:2] == ('', ''):
1092
if (self._last_block_index is not None and
1093
self._dirblocks[self._last_block_index][0] == key[0]):
1094
return self._last_block_index, True
1097
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1098
cache=self._split_path_cache)
1099
# _right returns one-past-where-key is so we have to subtract
1100
# one to use it. we use _right here because there are two
1101
# '' blocks - the root, and the contents of root
1102
# we always have a minimum of 2 in self._dirblocks: root and
1103
# root-contents, and for '', we get 2 back, so this is
1104
# simple and correct:
1105
present = (block_index < len(self._dirblocks) and
1106
self._dirblocks[block_index][0] == key[0])
1107
self._last_block_index = block_index
1108
# Reset the entry index cache to the beginning of the block.
1109
self._last_entry_index = -1
1110
return block_index, present
1112
def _find_entry_index(self, key, block):
1113
"""Find the entry index for a key in a block.
1115
:return: The entry index, True if the entry for the key is present.
1117
len_block = len(block)
1119
if self._last_entry_index is not None:
1121
entry_index = self._last_entry_index + 1
1122
# A hit is when the key is after the last slot, and before or
1123
# equal to the next slot.
1124
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1125
key <= block[entry_index][0]):
1126
self._last_entry_index = entry_index
1127
present = (block[entry_index][0] == key)
1128
return entry_index, present
1131
entry_index = bisect.bisect_left(block, (key, []))
1132
present = (entry_index < len_block and
1133
block[entry_index][0] == key)
1134
self._last_entry_index = entry_index
1135
return entry_index, present
1138
def from_tree(tree, dir_state_filename):
1139
"""Create a dirstate from a bzr Tree.
1141
:param tree: The tree which should provide parent information and
1143
:return: a DirState object which is currently locked for writing.
1144
(it was locked by DirState.initialize)
1146
result = DirState.initialize(dir_state_filename)
1150
parent_ids = tree.get_parent_ids()
1151
num_parents = len(parent_ids)
1153
for parent_id in parent_ids:
1154
parent_tree = tree.branch.repository.revision_tree(parent_id)
1155
parent_trees.append((parent_id, parent_tree))
1156
parent_tree.lock_read()
1157
result.set_parent_trees(parent_trees, [])
1158
result.set_state_from_inventory(tree.inventory)
1160
for revid, parent_tree in parent_trees:
1161
parent_tree.unlock()
1164
# The caller won't have a chance to unlock this, so make sure we
1170
def update_basis_by_delta(self, delta, new_revid):
1171
"""Update the parents of this tree after a commit.
1173
This gives the tree one parent, with revision id new_revid. The
1174
inventory delta is applied to the current basis tree to generate the
1175
inventory for the parent new_revid, and all other parent trees are
1178
Note that an exception during the operation of this method will leave
1179
the dirstate in a corrupt state where it should not be saved.
1181
Finally, we expect all changes to be synchronising the basis tree with
1184
:param new_revid: The new revision id for the trees parent.
1185
:param delta: An inventory delta (see apply_inventory_delta) describing
1186
the changes from the current left most parent revision to new_revid.
1188
self._read_dirblocks_if_needed()
1189
self._discard_merge_parents()
1190
if self._ghosts != []:
1191
raise NotImplementedError(self.update_basis_by_delta)
1192
if len(self._parents) == 0:
1193
# setup a blank tree, the most simple way.
1194
empty_parent = DirState.NULL_PARENT_DETAILS
1195
for entry in self._iter_entries():
1196
entry[1].append(empty_parent)
1197
self._parents.append(new_revid)
1199
self._parents[0] = new_revid
1201
delta = sorted(delta, reverse=True)
1205
# The paths this function accepts are unicode and must be encoded as we
1207
encode = cache_utf8.encode
1208
inv_to_entry = self._inv_entry_to_details
1209
# delta is now (deletes, changes), (adds) in reverse lexographical
1211
# deletes in reverse lexographic order are safe to process in situ.
1212
# renames are not, as a rename from any path could go to a path
1213
# lexographically lower, so we transform renames into delete, add pairs,
1214
# expanding them recursively as needed.
1215
# At the same time, to reduce interface friction we convert the input
1216
# inventory entries to dirstate.
1217
root_only = ('', '')
1218
for old_path, new_path, file_id, inv_entry in delta:
1219
if old_path is None:
1220
adds.append((None, encode(new_path), file_id,
1221
inv_to_entry(inv_entry), True))
1222
elif new_path is None:
1223
deletes.append((encode(old_path), None, file_id, None, True))
1224
elif (old_path, new_path) != root_only:
1226
# Because renames must preserve their children we must have
1227
# processed all relocations and removes before hand. The sort
1228
# order ensures we've examined the child paths, but we also
1229
# have to execute the removals, or the split to an add/delete
1230
# pair will result in the deleted item being reinserted, or
1231
# renamed items being reinserted twice - and possibly at the
1232
# wrong place. Splitting into a delete/add pair also simplifies
1233
# the handling of entries with ('f', ...), ('r' ...) because
1234
# the target of the 'r' is old_path here, and we add that to
1235
# deletes, meaning that the add handler does not need to check
1236
# for 'r' items on every pass.
1237
self._update_basis_apply_deletes(deletes)
1239
new_path_utf8 = encode(new_path)
1240
# Split into an add/delete pair recursively.
1241
adds.append((None, new_path_utf8, file_id,
1242
inv_to_entry(inv_entry), False))
1243
# Expunge deletes that we've seen so that deleted/renamed
1244
# children of a rename directory are handled correctly.
1245
new_deletes = reversed(list(self._iter_child_entries(1,
1247
# Remove the current contents of the tree at orig_path, and
1248
# reinsert at the correct new path.
1249
for entry in new_deletes:
1251
source_path = entry[0][0] + '/' + entry[0][1]
1253
source_path = entry[0][1]
1254
target_path = new_path_utf8 + source_path[len(old_path):]
1255
adds.append((None, target_path, entry[0][2], entry[1][1], False))
1257
(source_path, target_path, entry[0][2], None, False))
1259
(encode(old_path), new_path, file_id, None, False))
1261
# changes to just the root should not require remove/insertion
1263
changes.append((encode(old_path), encode(new_path), file_id,
1264
inv_to_entry(inv_entry)))
1266
# Finish expunging deletes/first half of renames.
1267
self._update_basis_apply_deletes(deletes)
1268
# Reinstate second half of renames and new paths.
1269
self._update_basis_apply_adds(adds)
1270
# Apply in-situ changes.
1271
self._update_basis_apply_changes(changes)
1273
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1274
self._header_state = DirState.IN_MEMORY_MODIFIED
1275
self._id_index = None
1278
def _update_basis_apply_adds(self, adds):
1279
"""Apply a sequence of adds to tree 1 during update_basis_by_delta.
1281
They may be adds, or renames that have been split into add/delete
1284
:param adds: A sequence of adds. Each add is a tuple:
1285
(None, new_path_utf8, file_id, (entry_details), real_add). real_add
1286
is False when the add is the second half of a remove-and-reinsert
1287
pair created to handle renames and deletes.
1289
# Adds are accumulated partly from renames, so can be in any input
1292
# adds is now in lexographic order, which places all parents before
1293
# their children, so we can process it linearly.
1295
for old_path, new_path, file_id, new_details, real_add in adds:
1296
assert old_path is None
1297
# the entry for this file_id must be in tree 0.
1298
entry = self._get_entry(0, file_id, new_path)
1299
if entry[0] is None or entry[0][2] != file_id:
1300
self._consistency = DirState.INCONSISTENT
1301
raise errors.InconsistentDelta(new_path, file_id,
1302
'working tree does not contain new entry')
1303
if real_add and entry[1][1][0] not in absent:
1304
self._consistency = DirState.INCONSISTENT
1305
raise errors.BzrError(new_path, file_id,
1306
'The entry was considered to be a genuinely new record,'
1307
' but there was already an old record for it.')
1308
# We don't need to update the target of an 'r' because the handling
1309
# of renames turns all 'r' situations into a delete at the original
1311
entry[1][1] = new_details
1313
def _update_basis_apply_changes(self, changes):
1314
"""Apply a sequence of changes to tree 1 during update_basis_by_delta.
1316
:param adds: A sequence of changes. Each change is a tuple:
1317
(path_utf8, path_utf8, file_id, (entry_details))
1320
for old_path, new_path, file_id, new_details in changes:
1321
assert old_path == new_path
1322
# the entry for this file_id must be in tree 0.
1323
entry = self._get_entry(0, file_id, new_path)
1324
if entry[0] is None or entry[0][2] != file_id:
1325
self._consistency = DirState.INCONSISTENT
1326
raise errors.InconsistentDelta(new_path, file_id,
1327
'working tree does not contain new entry')
1328
if (entry[1][0][0] in absent or
1329
entry[1][1][0] in absent):
1330
self._consistency = DirState.INCONSISTENT
1331
raise errors.InconsistentDelta(new_path, file_id,
1332
'changed considered absent')
1333
entry[1][1] = new_details
1335
def _update_basis_apply_deletes(self, deletes):
1336
"""Apply a sequence of deletes to tree 1 during update_basis_by_delta.
1338
They may be deletes, or renames that have been split into add/delete
1341
:param deletes: A sequence of deletes. Each delete is a tuple:
1342
(old_path_utf8, new_path_utf8, file_id, None, real_delete).
1343
real_delete is True when the desired outcome is an actual deletion
1344
rather than the rename handling logic temporarily deleting a path
1345
during the replacement of a parent.
1347
null = DirState.NULL_PARENT_DETAILS
1348
for old_path, new_path, file_id, _, real_delete in deletes:
1350
assert new_path is None
1352
assert new_path is not None
1353
# the entry for this file_id must be in tree 1.
1354
dirname, basename = osutils.split(old_path)
1355
block_index, entry_index, dir_present, file_present = \
1356
self._get_block_entry_index(dirname, basename, 1)
1357
if not file_present:
1358
self._consistency = DirState.INCONSISTENT
1359
raise errors.InconsistentDelta(old_path, file_id,
1360
'basis tree does not contain removed entry')
1361
entry = self._dirblocks[block_index][1][entry_index]
1362
if entry[0][2] != file_id:
1363
self._consistency = DirState.INCONSISTENT
1364
raise errors.InconsistentDelta(old_path, file_id,
1365
'mismatched file_id in tree 1')
1367
if entry[1][0][0] != 'a':
1368
self._consistency = DirState.INCONSISTENT
1369
raise errors.InconsistentDelta(old_path, file_id,
1370
'This was marked as a real delete, but the WT state'
1371
' claims that it still exists and is versioned.')
1372
del self._dirblocks[block_index][1][entry_index]
1374
if entry[1][0][0] == 'a':
1375
self._consistency = DirState.INCONSISTENT
1376
raise errors.InconsistentDelta(old_path, file_id,
1377
'The entry was considered a rename, but the source path'
1378
' is marked as absent.')
1379
# For whatever reason, we were asked to rename an entry
1380
# that was originally marked as deleted. This could be
1381
# because we are renaming the parent directory, and the WT
1382
# current state has the file marked as deleted.
1383
elif entry[1][0][0] == 'r':
1384
# implement the rename
1385
del self._dirblocks[block_index][1][entry_index]
1387
# it is being resurrected here, so blank it out temporarily.
1388
self._dirblocks[block_index][1][entry_index][1][1] = null
1390
def update_entry(self, entry, abspath, stat_value,
1391
_stat_to_minikind=_stat_to_minikind,
1392
_pack_stat=pack_stat):
1393
"""Update the entry based on what is actually on disk.
1395
:param entry: This is the dirblock entry for the file in question.
1396
:param abspath: The path on disk for this file.
1397
:param stat_value: (optional) if we already have done a stat on the
1399
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1403
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1407
packed_stat = _pack_stat(stat_value)
1408
(saved_minikind, saved_link_or_sha1, saved_file_size,
1409
saved_executable, saved_packed_stat) = entry[1][0]
1411
if (minikind == saved_minikind
1412
and packed_stat == saved_packed_stat):
1413
# The stat hasn't changed since we saved, so we can re-use the
1418
# size should also be in packed_stat
1419
if saved_file_size == stat_value.st_size:
1420
return saved_link_or_sha1
1422
# If we have gotten this far, that means that we need to actually
1423
# process this entry.
1426
link_or_sha1 = self._sha1_file(abspath)
1427
executable = self._is_executable(stat_value.st_mode,
1429
if self._cutoff_time is None:
1430
self._sha_cutoff_time()
1431
if (stat_value.st_mtime < self._cutoff_time
1432
and stat_value.st_ctime < self._cutoff_time):
1433
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1434
executable, packed_stat)
1436
entry[1][0] = ('f', '', stat_value.st_size,
1437
executable, DirState.NULLSTAT)
1438
elif minikind == 'd':
1440
entry[1][0] = ('d', '', 0, False, packed_stat)
1441
if saved_minikind != 'd':
1442
# This changed from something into a directory. Make sure we
1443
# have a directory block for it. This doesn't happen very
1444
# often, so this doesn't have to be super fast.
1445
block_index, entry_index, dir_present, file_present = \
1446
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1447
self._ensure_block(block_index, entry_index,
1448
osutils.pathjoin(entry[0][0], entry[0][1]))
1449
elif minikind == 'l':
1450
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1451
if self._cutoff_time is None:
1452
self._sha_cutoff_time()
1453
if (stat_value.st_mtime < self._cutoff_time
1454
and stat_value.st_ctime < self._cutoff_time):
1455
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1458
entry[1][0] = ('l', '', stat_value.st_size,
1459
False, DirState.NULLSTAT)
1460
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1463
def _sha_cutoff_time(self):
1464
"""Return cutoff time.
1466
Files modified more recently than this time are at risk of being
1467
undetectably modified and so can't be cached.
1469
# Cache the cutoff time as long as we hold a lock.
1470
# time.time() isn't super expensive (approx 3.38us), but
1471
# when you call it 50,000 times it adds up.
1472
# For comparison, os.lstat() costs 7.2us if it is hot.
1473
self._cutoff_time = int(time.time()) - 3
1474
return self._cutoff_time
1476
def _lstat(self, abspath, entry):
1477
"""Return the os.lstat value for this path."""
1478
return os.lstat(abspath)
1480
def _sha1_file_and_mutter(self, abspath):
1481
# when -Dhashcache is turned on, this is monkey-patched in to log
1483
trace.mutter("dirstate sha1 " + abspath)
1484
return osutils.sha_file_by_name(abspath)
1486
def _is_executable(self, mode, old_executable):
1487
"""Is this file executable?"""
1488
return bool(S_IEXEC & mode)
1490
def _is_executable_win32(self, mode, old_executable):
1491
"""On win32 the executable bit is stored in the dirstate."""
1492
return old_executable
1494
if sys.platform == 'win32':
1495
_is_executable = _is_executable_win32
1497
def _read_link(self, abspath, old_link):
1498
"""Read the target of a symlink"""
1499
# TODO: jam 200700301 On Win32, this could just return the value
1500
# already in memory. However, this really needs to be done at a
1501
# higher level, because there either won't be anything on disk,
1502
# or the thing on disk will be a file.
1503
return os.readlink(abspath)
1505
def get_ghosts(self):
1506
"""Return a list of the parent tree revision ids that are ghosts."""
1507
self._read_header_if_needed()
1510
def get_lines(self):
1511
"""Serialise the entire dirstate to a sequence of lines."""
1512
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1513
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1514
# read whats on disk.
1515
self._state_file.seek(0)
1516
return self._state_file.readlines()
1518
lines.append(self._get_parents_line(self.get_parent_ids()))
1519
lines.append(self._get_ghosts_line(self._ghosts))
1520
# append the root line which is special cased
1521
lines.extend(map(self._entry_to_line, self._iter_entries()))
1522
return self._get_output_lines(lines)
1524
def _get_ghosts_line(self, ghost_ids):
1525
"""Create a line for the state file for ghost information."""
1526
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1528
def _get_parents_line(self, parent_ids):
1529
"""Create a line for the state file for parents information."""
1530
return '\0'.join([str(len(parent_ids))] + parent_ids)
1532
def _get_fields_to_entry(self):
1533
"""Get a function which converts entry fields into a entry record.
1535
This handles size and executable, as well as parent records.
1537
:return: A function which takes a list of fields, and returns an
1538
appropriate record for storing in memory.
1540
# This is intentionally unrolled for performance
1541
num_present_parents = self._num_present_parents()
1542
if num_present_parents == 0:
1543
def fields_to_entry_0_parents(fields, _int=int):
1544
path_name_file_id_key = (fields[0], fields[1], fields[2])
1545
return (path_name_file_id_key, [
1547
fields[3], # minikind
1548
fields[4], # fingerprint
1549
_int(fields[5]), # size
1550
fields[6] == 'y', # executable
1551
fields[7], # packed_stat or revision_id
1553
return fields_to_entry_0_parents
1554
elif num_present_parents == 1:
1555
def fields_to_entry_1_parent(fields, _int=int):
1556
path_name_file_id_key = (fields[0], fields[1], fields[2])
1557
return (path_name_file_id_key, [
1559
fields[3], # minikind
1560
fields[4], # fingerprint
1561
_int(fields[5]), # size
1562
fields[6] == 'y', # executable
1563
fields[7], # packed_stat or revision_id
1566
fields[8], # minikind
1567
fields[9], # fingerprint
1568
_int(fields[10]), # size
1569
fields[11] == 'y', # executable
1570
fields[12], # packed_stat or revision_id
1573
return fields_to_entry_1_parent
1574
elif num_present_parents == 2:
1575
def fields_to_entry_2_parents(fields, _int=int):
1576
path_name_file_id_key = (fields[0], fields[1], fields[2])
1577
return (path_name_file_id_key, [
1579
fields[3], # minikind
1580
fields[4], # fingerprint
1581
_int(fields[5]), # size
1582
fields[6] == 'y', # executable
1583
fields[7], # packed_stat or revision_id
1586
fields[8], # minikind
1587
fields[9], # fingerprint
1588
_int(fields[10]), # size
1589
fields[11] == 'y', # executable
1590
fields[12], # packed_stat or revision_id
1593
fields[13], # minikind
1594
fields[14], # fingerprint
1595
_int(fields[15]), # size
1596
fields[16] == 'y', # executable
1597
fields[17], # packed_stat or revision_id
1600
return fields_to_entry_2_parents
1602
def fields_to_entry_n_parents(fields, _int=int):
1603
path_name_file_id_key = (fields[0], fields[1], fields[2])
1604
trees = [(fields[cur], # minikind
1605
fields[cur+1], # fingerprint
1606
_int(fields[cur+2]), # size
1607
fields[cur+3] == 'y', # executable
1608
fields[cur+4], # stat or revision_id
1609
) for cur in xrange(3, len(fields)-1, 5)]
1610
return path_name_file_id_key, trees
1611
return fields_to_entry_n_parents
1613
def get_parent_ids(self):
1614
"""Return a list of the parent tree ids for the directory state."""
1615
self._read_header_if_needed()
1616
return list(self._parents)
1618
def _get_block_entry_index(self, dirname, basename, tree_index):
1619
"""Get the coordinates for a path in the state structure.
1621
:param dirname: The utf8 dirname to lookup.
1622
:param basename: The utf8 basename to lookup.
1623
:param tree_index: The index of the tree for which this lookup should
1625
:return: A tuple describing where the path is located, or should be
1626
inserted. The tuple contains four fields: the block index, the row
1627
index, the directory is present (boolean), the entire path is
1628
present (boolean). There is no guarantee that either
1629
coordinate is currently reachable unless the found field for it is
1630
True. For instance, a directory not present in the searched tree
1631
may be returned with a value one greater than the current highest
1632
block offset. The directory present field will always be True when
1633
the path present field is True. The directory present field does
1634
NOT indicate that the directory is present in the searched tree,
1635
rather it indicates that there are at least some files in some
1638
self._read_dirblocks_if_needed()
1639
key = dirname, basename, ''
1640
block_index, present = self._find_block_index_from_key(key)
1642
# no such directory - return the dir index and 0 for the row.
1643
return block_index, 0, False, False
1644
block = self._dirblocks[block_index][1] # access the entries only
1645
entry_index, present = self._find_entry_index(key, block)
1646
# linear search through entries at this path to find the one
1648
while entry_index < len(block) and block[entry_index][0][1] == basename:
1649
if block[entry_index][1][tree_index][0] not in 'ar':
1650
# neither absent or relocated
1651
return block_index, entry_index, True, True
1653
return block_index, entry_index, True, False
1655
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1656
"""Get the dirstate entry for path in tree tree_index.
1658
If either file_id or path is supplied, it is used as the key to lookup.
1659
If both are supplied, the fastest lookup is used, and an error is
1660
raised if they do not both point at the same row.
1662
:param tree_index: The index of the tree we wish to locate this path
1663
in. If the path is present in that tree, the entry containing its
1664
details is returned, otherwise (None, None) is returned
1665
0 is the working tree, higher indexes are successive parent
1667
:param fileid_utf8: A utf8 file_id to look up.
1668
:param path_utf8: An utf8 path to be looked up.
1669
:return: The dirstate entry tuple for path, or (None, None)
1671
self._read_dirblocks_if_needed()
1672
if path_utf8 is not None:
1673
assert path_utf8.__class__ == str, ('path_utf8 is not a str: %s %s'
1674
% (type(path_utf8), path_utf8))
1675
# path lookups are faster
1676
dirname, basename = osutils.split(path_utf8)
1677
block_index, entry_index, dir_present, file_present = \
1678
self._get_block_entry_index(dirname, basename, tree_index)
1679
if not file_present:
1681
entry = self._dirblocks[block_index][1][entry_index]
1682
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1684
if entry[0][2] != fileid_utf8:
1685
self._consistency = DirState.INCONSISTENT
1686
raise errors.BzrError('integrity error ? : mismatching'
1687
' tree_index, file_id and path')
1690
assert fileid_utf8 is not None
1691
possible_keys = self._get_id_index().get(fileid_utf8, None)
1692
if not possible_keys:
1694
for key in possible_keys:
1695
block_index, present = \
1696
self._find_block_index_from_key(key)
1697
# strange, probably indicates an out of date
1698
# id index - for now, allow this.
1701
# WARNING: DO not change this code to use _get_block_entry_index
1702
# as that function is not suitable: it does not use the key
1703
# to lookup, and thus the wrong coordinates are returned.
1704
block = self._dirblocks[block_index][1]
1705
entry_index, present = self._find_entry_index(key, block)
1707
entry = self._dirblocks[block_index][1][entry_index]
1708
if entry[1][tree_index][0] in 'fdlt':
1709
# this is the result we are looking for: the
1710
# real home of this file_id in this tree.
1712
if entry[1][tree_index][0] == 'a':
1713
# there is no home for this entry in this tree
1715
assert entry[1][tree_index][0] == 'r', \
1716
"entry %r has invalid minikind %r for tree %r" \
1718
entry[1][tree_index][0],
1720
real_path = entry[1][tree_index][1]
1721
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1722
path_utf8=real_path)
1726
def initialize(cls, path):
1727
"""Create a new dirstate on path.
1729
The new dirstate will be an empty tree - that is it has no parents,
1730
and only a root node - which has id ROOT_ID.
1732
:param path: The name of the file for the dirstate.
1733
:return: A write-locked DirState object.
1735
# This constructs a new DirState object on a path, sets the _state_file
1736
# to a new empty file for that path. It then calls _set_data() with our
1737
# stock empty dirstate information - a root with ROOT_ID, no children,
1738
# and no parents. Finally it calls save() to ensure that this data will
1741
# root dir and root dir contents with no children.
1742
empty_tree_dirblocks = [('', []), ('', [])]
1743
# a new root directory, with a NULLSTAT.
1744
empty_tree_dirblocks[0][1].append(
1745
(('', '', inventory.ROOT_ID), [
1746
('d', '', 0, False, DirState.NULLSTAT),
1750
result._set_data([], empty_tree_dirblocks)
1757
def _inv_entry_to_details(self, inv_entry):
1758
"""Convert an inventory entry (from a revision tree) to state details.
1760
:param inv_entry: An inventory entry whose sha1 and link targets can be
1761
relied upon, and which has a revision set.
1762
:return: A details tuple - the details for a single tree at a path +
1765
kind = inv_entry.kind
1766
minikind = DirState._kind_to_minikind[kind]
1767
tree_data = inv_entry.revision
1768
assert tree_data, 'empty revision for the inv_entry %s.' % \
1770
if kind == 'directory':
1774
elif kind == 'symlink':
1775
fingerprint = inv_entry.symlink_target or ''
1778
elif kind == 'file':
1779
fingerprint = inv_entry.text_sha1 or ''
1780
size = inv_entry.text_size or 0
1781
executable = inv_entry.executable
1782
elif kind == 'tree-reference':
1783
fingerprint = inv_entry.reference_revision or ''
1787
raise Exception("can't pack %s" % inv_entry)
1788
return (minikind, fingerprint, size, executable, tree_data)
1790
def _iter_child_entries(self, tree_index, path_utf8):
1791
"""Iterate over all the entries that are children of path_utf.
1793
This only returns entries that are present (not in 'a', 'r') in
1794
tree_index. tree_index data is not refreshed, so if tree 0 is used,
1795
results may differ from that obtained if paths were statted to
1796
determine what ones were directories.
1798
Asking for the children of a non-directory will return an empty
1802
next_pending_dirs = [path_utf8]
1804
while next_pending_dirs:
1805
pending_dirs = next_pending_dirs
1806
next_pending_dirs = []
1807
for path in pending_dirs:
1808
block_index, present = self._find_block_index_from_key(
1810
if block_index == 0:
1812
if len(self._dirblocks) == 1:
1813
# asked for the children of the root with no other
1817
# children of a non-directory asked for.
1819
block = self._dirblocks[block_index]
1820
for entry in block[1]:
1821
kind = entry[1][tree_index][0]
1822
if kind not in absent:
1826
path = entry[0][0] + '/' + entry[0][1]
1829
next_pending_dirs.append(path)
1831
def _iter_entries(self):
1832
"""Iterate over all the entries in the dirstate.
1834
Each yelt item is an entry in the standard format described in the
1835
docstring of bzrlib.dirstate.
1837
self._read_dirblocks_if_needed()
1838
for directory in self._dirblocks:
1839
for entry in directory[1]:
1842
def _get_id_index(self):
1843
"""Get an id index of self._dirblocks."""
1844
if self._id_index is None:
1846
for key, tree_details in self._iter_entries():
1847
id_index.setdefault(key[2], set()).add(key)
1848
self._id_index = id_index
1849
return self._id_index
1851
def _get_output_lines(self, lines):
1852
"""Format lines for final output.
1854
:param lines: A sequence of lines containing the parents list and the
1857
output_lines = [DirState.HEADER_FORMAT_3]
1858
lines.append('') # a final newline
1859
inventory_text = '\0\n\0'.join(lines)
1860
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1861
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1862
num_entries = len(lines)-3
1863
output_lines.append('num_entries: %s\n' % (num_entries,))
1864
output_lines.append(inventory_text)
1867
def _make_deleted_row(self, fileid_utf8, parents):
1868
"""Return a deleted row for fileid_utf8."""
1869
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1872
def _num_present_parents(self):
1873
"""The number of parent entries in each record row."""
1874
return len(self._parents) - len(self._ghosts)
1878
"""Construct a DirState on the file at path path.
1880
:return: An unlocked DirState object, associated with the given path.
1882
result = DirState(path)
1885
def _read_dirblocks_if_needed(self):
1886
"""Read in all the dirblocks from the file if they are not in memory.
1888
This populates self._dirblocks, and sets self._dirblock_state to
1889
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1892
self._read_header_if_needed()
1893
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1894
_read_dirblocks(self)
1896
def _read_header(self):
1897
"""This reads in the metadata header, and the parent ids.
1899
After reading in, the file should be positioned at the null
1900
just before the start of the first record in the file.
1902
:return: (expected crc checksum, number of entries, parent list)
1904
self._read_prelude()
1905
parent_line = self._state_file.readline()
1906
info = parent_line.split('\0')
1907
num_parents = int(info[0])
1908
assert num_parents == len(info)-2, 'incorrect parent info line'
1909
self._parents = info[1:-1]
1911
ghost_line = self._state_file.readline()
1912
info = ghost_line.split('\0')
1913
num_ghosts = int(info[1])
1914
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1915
self._ghosts = info[2:-1]
1916
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1917
self._end_of_header = self._state_file.tell()
1919
def _read_header_if_needed(self):
1920
"""Read the header of the dirstate file if needed."""
1921
# inline this as it will be called a lot
1922
if not self._lock_token:
1923
raise errors.ObjectNotLocked(self)
1924
if self._header_state == DirState.NOT_IN_MEMORY:
1927
def _read_prelude(self):
1928
"""Read in the prelude header of the dirstate file.
1930
This only reads in the stuff that is not connected to the crc
1931
checksum. The position will be correct to read in the rest of
1932
the file and check the checksum after this point.
1933
The next entry in the file should be the number of parents,
1934
and their ids. Followed by a newline.
1936
header = self._state_file.readline()
1937
assert header == DirState.HEADER_FORMAT_3, \
1938
'invalid header line: %r' % (header,)
1939
crc_line = self._state_file.readline()
1940
assert crc_line.startswith('crc32: '), 'missing crc32 checksum'
1941
self.crc_expected = int(crc_line[len('crc32: '):-1])
1942
num_entries_line = self._state_file.readline()
1943
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1944
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1946
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
1947
"""Find a sha1 given a stat lookup."""
1948
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
1950
def _get_packed_stat_index(self):
1951
"""Get a packed_stat index of self._dirblocks."""
1952
if self._packed_stat_index is None:
1954
for key, tree_details in self._iter_entries():
1955
if tree_details[0][0] == 'f':
1956
index[tree_details[0][4]] = tree_details[0][1]
1957
self._packed_stat_index = index
1958
return self._packed_stat_index
1961
"""Save any pending changes created during this session.
1963
We reuse the existing file, because that prevents race conditions with
1964
file creation, and use oslocks on it to prevent concurrent modification
1965
and reads - because dirstate's incremental data aggregation is not
1966
compatible with reading a modified file, and replacing a file in use by
1967
another process is impossible on Windows.
1969
A dirstate in read only mode should be smart enough though to validate
1970
that the file has not changed, and otherwise discard its cache and
1971
start over, to allow for fine grained read lock duration, so 'status'
1972
wont block 'commit' - for example.
1974
if self._consistency != DirState.CONSISTENT:
1975
# Should this be a warning? For now, I'm expecting that places that
1976
# mark it inconsistent will warn, making a warning here redundant.
1977
trace.mutter('Not saving DirState because it is in an'
1978
' INCONSISTENT state.')
1980
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1981
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1983
grabbed_write_lock = False
1984
if self._lock_state != 'w':
1985
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
1986
# Switch over to the new lock, as the old one may be closed.
1987
# TODO: jam 20070315 We should validate the disk file has
1988
# not changed contents. Since temporary_write_lock may
1989
# not be an atomic operation.
1990
self._lock_token = new_lock
1991
self._state_file = new_lock.f
1992
if not grabbed_write_lock:
1993
# We couldn't grab a write lock, so we switch back to a read one
1996
self._state_file.seek(0)
1997
self._state_file.writelines(self.get_lines())
1998
self._state_file.truncate()
1999
self._state_file.flush()
2000
self._header_state = DirState.IN_MEMORY_UNMODIFIED
2001
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
2003
if grabbed_write_lock:
2004
self._lock_token = self._lock_token.restore_read_lock()
2005
self._state_file = self._lock_token.f
2006
# TODO: jam 20070315 We should validate the disk file has
2007
# not changed contents. Since restore_read_lock may
2008
# not be an atomic operation.
2010
def _set_data(self, parent_ids, dirblocks):
2011
"""Set the full dirstate data in memory.
2013
This is an internal function used to completely replace the objects
2014
in memory state. It puts the dirstate into state 'full-dirty'.
2016
:param parent_ids: A list of parent tree revision ids.
2017
:param dirblocks: A list containing one tuple for each directory in the
2018
tree. Each tuple contains the directory path and a list of entries
2019
found in that directory.
2021
# our memory copy is now authoritative.
2022
self._dirblocks = dirblocks
2023
self._header_state = DirState.IN_MEMORY_MODIFIED
2024
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2025
self._parents = list(parent_ids)
2026
self._id_index = None
2027
self._packed_stat_index = None
2029
def set_path_id(self, path, new_id):
2030
"""Change the id of path to new_id in the current working tree.
2032
:param path: The path inside the tree to set - '' is the root, 'foo'
2033
is the path foo in the root.
2034
:param new_id: The new id to assign to the path. This must be a utf8
2035
file id (not unicode, and not None).
2037
assert new_id.__class__ == str, \
2038
"path_id %r is not a plain string" % (new_id,)
2039
self._read_dirblocks_if_needed()
2041
# TODO: logic not written
2042
raise NotImplementedError(self.set_path_id)
2043
# TODO: check new id is unique
2044
entry = self._get_entry(0, path_utf8=path)
2045
if entry[0][2] == new_id:
2046
# Nothing to change.
2048
# mark the old path absent, and insert a new root path
2049
self._make_absent(entry)
2050
self.update_minimal(('', '', new_id), 'd',
2051
path_utf8='', packed_stat=entry[1][0][4])
2052
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2053
if self._id_index is not None:
2054
self._id_index.setdefault(new_id, set()).add(entry[0])
2056
def set_parent_trees(self, trees, ghosts):
2057
"""Set the parent trees for the dirstate.
2059
:param trees: A list of revision_id, tree tuples. tree must be provided
2060
even if the revision_id refers to a ghost: supply an empty tree in
2062
:param ghosts: A list of the revision_ids that are ghosts at the time
2065
# TODO: generate a list of parent indexes to preserve to save
2066
# processing specific parent trees. In the common case one tree will
2067
# be preserved - the left most parent.
2068
# TODO: if the parent tree is a dirstate, we might want to walk them
2069
# all by path in parallel for 'optimal' common-case performance.
2070
# generate new root row.
2071
self._read_dirblocks_if_needed()
2072
# TODO future sketch: Examine the existing parents to generate a change
2073
# map and then walk the new parent trees only, mapping them into the
2074
# dirstate. Walk the dirstate at the same time to remove unreferenced
2077
# sketch: loop over all entries in the dirstate, cherry picking
2078
# entries from the parent trees, if they are not ghost trees.
2079
# after we finish walking the dirstate, all entries not in the dirstate
2080
# are deletes, so we want to append them to the end as per the design
2081
# discussions. So do a set difference on ids with the parents to
2082
# get deletes, and add them to the end.
2083
# During the update process we need to answer the following questions:
2084
# - find other keys containing a fileid in order to create cross-path
2085
# links. We dont't trivially use the inventory from other trees
2086
# because this leads to either double touching, or to accessing
2088
# - find other keys containing a path
2089
# We accumulate each entry via this dictionary, including the root
2092
# we could do parallel iterators, but because file id data may be
2093
# scattered throughout, we dont save on index overhead: we have to look
2094
# at everything anyway. We can probably save cycles by reusing parent
2095
# data and doing an incremental update when adding an additional
2096
# parent, but for now the common cases are adding a new parent (merge),
2097
# and replacing completely (commit), and commit is more common: so
2098
# optimise merge later.
2100
# ---- start generation of full tree mapping data
2101
# what trees should we use?
2102
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
2103
# how many trees do we end up with
2104
parent_count = len(parent_trees)
2106
# one: the current tree
2107
for entry in self._iter_entries():
2108
# skip entries not in the current tree
2109
if entry[1][0][0] in 'ar': # absent, relocated
2111
by_path[entry[0]] = [entry[1][0]] + \
2112
[DirState.NULL_PARENT_DETAILS] * parent_count
2113
id_index[entry[0][2]] = set([entry[0]])
2115
# now the parent trees:
2116
for tree_index, tree in enumerate(parent_trees):
2117
# the index is off by one, adjust it.
2118
tree_index = tree_index + 1
2119
# when we add new locations for a fileid we need these ranges for
2120
# any fileid in this tree as we set the by_path[id] to:
2121
# already_processed_tree_details + new_details + new_location_suffix
2122
# the suffix is from tree_index+1:parent_count+1.
2123
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
2124
# now stitch in all the entries from this tree
2125
for path, entry in tree.inventory.iter_entries_by_dir():
2126
# here we process each trees details for each item in the tree.
2127
# we first update any existing entries for the id at other paths,
2128
# then we either create or update the entry for the id at the
2129
# right path, and finally we add (if needed) a mapping from
2130
# file_id to this path. We do it in this order to allow us to
2131
# avoid checking all known paths for the id when generating a
2132
# new entry at this path: by adding the id->path mapping last,
2133
# all the mappings are valid and have correct relocation
2134
# records where needed.
2135
file_id = entry.file_id
2136
path_utf8 = path.encode('utf8')
2137
dirname, basename = osutils.split(path_utf8)
2138
new_entry_key = (dirname, basename, file_id)
2139
# tree index consistency: All other paths for this id in this tree
2140
# index must point to the correct path.
2141
for entry_key in id_index.setdefault(file_id, set()):
2142
# TODO:PROFILING: It might be faster to just update
2143
# rather than checking if we need to, and then overwrite
2144
# the one we are located at.
2145
if entry_key != new_entry_key:
2146
# this file id is at a different path in one of the
2147
# other trees, so put absent pointers there
2148
# This is the vertical axis in the matrix, all pointing
2150
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
2151
# by path consistency: Insert into an existing path record (trivial), or
2152
# add a new one with relocation pointers for the other tree indexes.
2153
if new_entry_key in id_index[file_id]:
2154
# there is already an entry where this data belongs, just insert it.
2155
by_path[new_entry_key][tree_index] = \
2156
self._inv_entry_to_details(entry)
2158
# add relocated entries to the horizontal axis - this row
2159
# mapping from path,id. We need to look up the correct path
2160
# for the indexes from 0 to tree_index -1
2162
for lookup_index in xrange(tree_index):
2163
# boundary case: this is the first occurence of file_id
2164
# so there are no id_indexs, possibly take this out of
2166
if not len(id_index[file_id]):
2167
new_details.append(DirState.NULL_PARENT_DETAILS)
2169
# grab any one entry, use it to find the right path.
2170
# TODO: optimise this to reduce memory use in highly
2171
# fragmented situations by reusing the relocation
2173
a_key = iter(id_index[file_id]).next()
2174
if by_path[a_key][lookup_index][0] in ('r', 'a'):
2175
# its a pointer or missing statement, use it as is.
2176
new_details.append(by_path[a_key][lookup_index])
2178
# we have the right key, make a pointer to it.
2179
real_path = ('/'.join(a_key[0:2])).strip('/')
2180
new_details.append(('r', real_path, 0, False, ''))
2181
new_details.append(self._inv_entry_to_details(entry))
2182
new_details.extend(new_location_suffix)
2183
by_path[new_entry_key] = new_details
2184
id_index[file_id].add(new_entry_key)
2185
# --- end generation of full tree mappings
2187
# sort and output all the entries
2188
new_entries = self._sort_entries(by_path.items())
2189
self._entries_to_current_state(new_entries)
2190
self._parents = [rev_id for rev_id, tree in trees]
2191
self._ghosts = list(ghosts)
2192
self._header_state = DirState.IN_MEMORY_MODIFIED
2193
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2194
self._id_index = id_index
2196
def _sort_entries(self, entry_list):
2197
"""Given a list of entries, sort them into the right order.
2199
This is done when constructing a new dirstate from trees - normally we
2200
try to keep everything in sorted blocks all the time, but sometimes
2201
it's easier to sort after the fact.
2204
# sort by: directory parts, file name, file id
2205
return entry[0][0].split('/'), entry[0][1], entry[0][2]
2206
return sorted(entry_list, key=_key)
2208
def set_state_from_inventory(self, new_inv):
2209
"""Set new_inv as the current state.
2211
This API is called by tree transform, and will usually occur with
2212
existing parent trees.
2214
:param new_inv: The inventory object to set current state from.
2216
if 'evil' in debug.debug_flags:
2217
trace.mutter_callsite(1,
2218
"set_state_from_inventory called; please mutate the tree instead")
2219
self._read_dirblocks_if_needed()
2221
# Two iterators: current data and new data, both in dirblock order.
2222
# We zip them together, which tells about entries that are new in the
2223
# inventory, or removed in the inventory, or present in both and
2226
# You might think we could just synthesize a new dirstate directly
2227
# since we're processing it in the right order. However, we need to
2228
# also consider there may be any number of parent trees and relocation
2229
# pointers, and we don't want to duplicate that here.
2230
new_iterator = new_inv.iter_entries_by_dir()
2231
# we will be modifying the dirstate, so we need a stable iterator. In
2232
# future we might write one, for now we just clone the state into a
2233
# list - which is a shallow copy.
2234
old_iterator = iter(list(self._iter_entries()))
2235
# both must have roots so this is safe:
2236
current_new = new_iterator.next()
2237
current_old = old_iterator.next()
2238
def advance(iterator):
2240
return iterator.next()
2241
except StopIteration:
2243
while current_new or current_old:
2244
# skip entries in old that are not really there
2245
if current_old and current_old[1][0][0] in 'ar':
2246
# relocated or absent
2247
current_old = advance(old_iterator)
2250
# convert new into dirblock style
2251
new_path_utf8 = current_new[0].encode('utf8')
2252
new_dirname, new_basename = osutils.split(new_path_utf8)
2253
new_id = current_new[1].file_id
2254
new_entry_key = (new_dirname, new_basename, new_id)
2255
current_new_minikind = \
2256
DirState._kind_to_minikind[current_new[1].kind]
2257
if current_new_minikind == 't':
2258
fingerprint = current_new[1].reference_revision or ''
2260
# We normally only insert or remove records, or update
2261
# them when it has significantly changed. Then we want to
2262
# erase its fingerprint. Unaffected records should
2263
# normally not be updated at all.
2266
# for safety disable variables
2267
new_path_utf8 = new_dirname = new_basename = new_id = \
2268
new_entry_key = None
2269
# 5 cases, we dont have a value that is strictly greater than everything, so
2270
# we make both end conditions explicit
2272
# old is finished: insert current_new into the state.
2273
self.update_minimal(new_entry_key, current_new_minikind,
2274
executable=current_new[1].executable,
2275
path_utf8=new_path_utf8, fingerprint=fingerprint)
2276
current_new = advance(new_iterator)
2277
elif not current_new:
2279
self._make_absent(current_old)
2280
current_old = advance(old_iterator)
2281
elif new_entry_key == current_old[0]:
2282
# same - common case
2283
# We're looking at the same path and id in both the dirstate
2284
# and inventory, so just need to update the fields in the
2285
# dirstate from the one in the inventory.
2286
# TODO: update the record if anything significant has changed.
2287
# the minimal required trigger is if the execute bit or cached
2289
if (current_old[1][0][3] != current_new[1].executable or
2290
current_old[1][0][0] != current_new_minikind):
2291
self.update_minimal(current_old[0], current_new_minikind,
2292
executable=current_new[1].executable,
2293
path_utf8=new_path_utf8, fingerprint=fingerprint)
2294
# both sides are dealt with, move on
2295
current_old = advance(old_iterator)
2296
current_new = advance(new_iterator)
2297
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
2298
or (new_dirname == current_old[0][0]
2299
and new_entry_key[1:] < current_old[0][1:])):
2301
# add a entry for this and advance new
2302
self.update_minimal(new_entry_key, current_new_minikind,
2303
executable=current_new[1].executable,
2304
path_utf8=new_path_utf8, fingerprint=fingerprint)
2305
current_new = advance(new_iterator)
2307
# we've advanced past the place where the old key would be,
2308
# without seeing it in the new list. so it must be gone.
2309
self._make_absent(current_old)
2310
current_old = advance(old_iterator)
2311
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2312
self._id_index = None
2313
self._packed_stat_index = None
2315
def _make_absent(self, current_old):
2316
"""Mark current_old - an entry - as absent for tree 0.
2318
:return: True if this was the last details entry for the entry key:
2319
that is, if the underlying block has had the entry removed, thus
2320
shrinking in length.
2322
# build up paths that this id will be left at after the change is made,
2323
# so we can update their cross references in tree 0
2324
all_remaining_keys = set()
2325
# Dont check the working tree, because it's going.
2326
for details in current_old[1][1:]:
2327
if details[0] not in 'ar': # absent, relocated
2328
all_remaining_keys.add(current_old[0])
2329
elif details[0] == 'r': # relocated
2330
# record the key for the real path.
2331
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2332
# absent rows are not present at any path.
2333
last_reference = current_old[0] not in all_remaining_keys
2335
# the current row consists entire of the current item (being marked
2336
# absent), and relocated or absent entries for the other trees:
2337
# Remove it, its meaningless.
2338
block = self._find_block(current_old[0])
2339
entry_index, present = self._find_entry_index(current_old[0], block[1])
2340
assert present, 'could not find entry for %s' % (current_old,)
2341
block[1].pop(entry_index)
2342
# if we have an id_index in use, remove this key from it for this id.
2343
if self._id_index is not None:
2344
self._id_index[current_old[0][2]].remove(current_old[0])
2345
# update all remaining keys for this id to record it as absent. The
2346
# existing details may either be the record we are marking as deleted
2347
# (if there were other trees with the id present at this path), or may
2349
for update_key in all_remaining_keys:
2350
update_block_index, present = \
2351
self._find_block_index_from_key(update_key)
2352
assert present, 'could not find block for %s' % (update_key,)
2353
update_entry_index, present = \
2354
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2355
assert present, 'could not find entry for %s' % (update_key,)
2356
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2357
# it must not be absent at the moment
2358
assert update_tree_details[0][0] != 'a' # absent
2359
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2360
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2361
return last_reference
2363
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2364
packed_stat=None, size=0, path_utf8=None):
2365
"""Update an entry to the state in tree 0.
2367
This will either create a new entry at 'key' or update an existing one.
2368
It also makes sure that any other records which might mention this are
2371
:param key: (dir, name, file_id) for the new entry
2372
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2374
:param executable: Should the executable bit be set?
2375
:param fingerprint: Simple fingerprint for new entry: sha1 for files,
2376
referenced revision id for subtrees, etc.
2377
:param packed_stat: Packed stat value for new entry.
2378
:param size: Size information for new entry
2379
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2382
If packed_stat and fingerprint are not given, they're invalidated in
2385
block = self._find_block(key)[1]
2386
if packed_stat is None:
2387
packed_stat = DirState.NULLSTAT
2388
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2389
# sometimes present in the dirstate - this seems oddly inconsistent.
2391
entry_index, present = self._find_entry_index(key, block)
2392
new_details = (minikind, fingerprint, size, executable, packed_stat)
2393
id_index = self._get_id_index()
2395
# new entry, synthesis cross reference here,
2396
existing_keys = id_index.setdefault(key[2], set())
2397
if not existing_keys:
2398
# not currently in the state, simplest case
2399
new_entry = key, [new_details] + self._empty_parent_info()
2401
# present at one or more existing other paths.
2402
# grab one of them and use it to generate parent
2403
# relocation/absent entries.
2404
new_entry = key, [new_details]
2405
for other_key in existing_keys:
2406
# change the record at other to be a pointer to this new
2407
# record. The loop looks similar to the change to
2408
# relocations when updating an existing record but its not:
2409
# the test for existing kinds is different: this can be
2410
# factored out to a helper though.
2411
other_block_index, present = self._find_block_index_from_key(other_key)
2412
assert present, 'could not find block for %s' % (other_key,)
2413
other_entry_index, present = self._find_entry_index(other_key,
2414
self._dirblocks[other_block_index][1])
2415
assert present, 'could not find entry for %s' % (other_key,)
2416
assert path_utf8 is not None
2417
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2418
('r', path_utf8, 0, False, '')
2420
num_present_parents = self._num_present_parents()
2421
for lookup_index in xrange(1, num_present_parents + 1):
2422
# grab any one entry, use it to find the right path.
2423
# TODO: optimise this to reduce memory use in highly
2424
# fragmented situations by reusing the relocation
2426
update_block_index, present = \
2427
self._find_block_index_from_key(other_key)
2428
assert present, 'could not find block for %s' % (other_key,)
2429
update_entry_index, present = \
2430
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2431
assert present, 'could not find entry for %s' % (other_key,)
2432
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2433
if update_details[0] in 'ar': # relocated, absent
2434
# its a pointer or absent in lookup_index's tree, use
2436
new_entry[1].append(update_details)
2438
# we have the right key, make a pointer to it.
2439
pointer_path = osutils.pathjoin(*other_key[0:2])
2440
new_entry[1].append(('r', pointer_path, 0, False, ''))
2441
block.insert(entry_index, new_entry)
2442
existing_keys.add(key)
2444
# Does the new state matter?
2445
block[entry_index][1][0] = new_details
2446
# parents cannot be affected by what we do.
2447
# other occurences of this id can be found
2448
# from the id index.
2450
# tree index consistency: All other paths for this id in this tree
2451
# index must point to the correct path. We have to loop here because
2452
# we may have passed entries in the state with this file id already
2453
# that were absent - where parent entries are - and they need to be
2454
# converted to relocated.
2455
assert path_utf8 is not None
2456
for entry_key in id_index.setdefault(key[2], set()):
2457
# TODO:PROFILING: It might be faster to just update
2458
# rather than checking if we need to, and then overwrite
2459
# the one we are located at.
2460
if entry_key != key:
2461
# this file id is at a different path in one of the
2462
# other trees, so put absent pointers there
2463
# This is the vertical axis in the matrix, all pointing
2465
block_index, present = self._find_block_index_from_key(entry_key)
2467
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2469
self._dirblocks[block_index][1][entry_index][1][0] = \
2470
('r', path_utf8, 0, False, '')
2471
# add a containing dirblock if needed.
2472
if new_details[0] == 'd':
2473
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2474
block_index, present = self._find_block_index_from_key(subdir_key)
2476
self._dirblocks.insert(block_index, (subdir_key[0], []))
2478
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2480
def _validate(self):
2481
"""Check that invariants on the dirblock are correct.
2483
This can be useful in debugging; it shouldn't be necessary in
2486
This must be called with a lock held.
2488
# NOTE: This must always raise AssertionError not just assert,
2489
# otherwise it may not behave properly under python -O
2491
# TODO: All entries must have some content that's not 'a' or 'r',
2492
# otherwise it could just be removed.
2494
# TODO: All relocations must point directly to a real entry.
2496
# TODO: No repeated keys.
2499
from pprint import pformat
2500
self._read_dirblocks_if_needed()
2501
if len(self._dirblocks) > 0:
2502
if not self._dirblocks[0][0] == '':
2503
raise AssertionError(
2504
"dirblocks don't start with root block:\n" + \
2506
if len(self._dirblocks) > 1:
2507
if not self._dirblocks[1][0] == '':
2508
raise AssertionError(
2509
"dirblocks missing root directory:\n" + \
2511
# the dirblocks are sorted by their path components, name, and dir id
2512
dir_names = [d[0].split('/')
2513
for d in self._dirblocks[1:]]
2514
if dir_names != sorted(dir_names):
2515
raise AssertionError(
2516
"dir names are not in sorted order:\n" + \
2517
pformat(self._dirblocks) + \
2520
for dirblock in self._dirblocks:
2521
# within each dirblock, the entries are sorted by filename and
2523
for entry in dirblock[1]:
2524
if dirblock[0] != entry[0][0]:
2525
raise AssertionError(
2527
"doesn't match directory name in\n%r" %
2528
(entry, pformat(dirblock)))
2529
if dirblock[1] != sorted(dirblock[1]):
2530
raise AssertionError(
2531
"dirblock for %r is not sorted:\n%s" % \
2532
(dirblock[0], pformat(dirblock)))
2534
def check_valid_parent():
2535
"""Check that the current entry has a valid parent.
2537
This makes sure that the parent has a record,
2538
and that the parent isn't marked as "absent" in the
2539
current tree. (It is invalid to have a non-absent file in an absent
2542
if entry[0][0:2] == ('', ''):
2543
# There should be no parent for the root row
2545
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2546
if parent_entry == (None, None):
2547
raise AssertionError(
2548
"no parent entry for: %s in tree %s"
2549
% (this_path, tree_index))
2550
if parent_entry[1][tree_index][0] != 'd':
2551
raise AssertionError(
2552
"Parent entry for %s is not marked as a valid"
2553
" directory. %s" % (this_path, parent_entry,))
2555
# For each file id, for each tree: either
2556
# the file id is not present at all; all rows with that id in the
2557
# key have it marked as 'absent'
2558
# OR the file id is present under exactly one name; any other entries
2559
# that mention that id point to the correct name.
2561
# We check this with a dict per tree pointing either to the present
2562
# name, or None if absent.
2563
tree_count = self._num_present_parents() + 1
2564
id_path_maps = [dict() for i in range(tree_count)]
2565
# Make sure that all renamed entries point to the correct location.
2566
for entry in self._iter_entries():
2567
file_id = entry[0][2]
2568
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2569
if len(entry[1]) != tree_count:
2570
raise AssertionError(
2571
"wrong number of entry details for row\n%s" \
2572
",\nexpected %d" % \
2573
(pformat(entry), tree_count))
2574
absent_positions = 0
2575
for tree_index, tree_state in enumerate(entry[1]):
2576
this_tree_map = id_path_maps[tree_index]
2577
minikind = tree_state[0]
2578
if minikind in 'ar':
2579
absent_positions += 1
2580
# have we seen this id before in this column?
2581
if file_id in this_tree_map:
2582
previous_path, previous_loc = this_tree_map[file_id]
2583
# any later mention of this file must be consistent with
2584
# what was said before
2586
if previous_path is not None:
2587
raise AssertionError(
2588
"file %s is absent in row %r but also present " \
2590
(file_id, entry, previous_path))
2591
elif minikind == 'r':
2592
target_location = tree_state[1]
2593
if previous_path != target_location:
2594
raise AssertionError(
2595
"file %s relocation in row %r but also at %r" \
2596
% (file_id, entry, previous_path))
2598
# a file, directory, etc - may have been previously
2599
# pointed to by a relocation, which must point here
2600
if previous_path != this_path:
2601
raise AssertionError(
2602
"entry %r inconsistent with previous path %r "
2604
(entry, previous_path, previous_loc))
2605
check_valid_parent()
2608
# absent; should not occur anywhere else
2609
this_tree_map[file_id] = None, this_path
2610
elif minikind == 'r':
2611
# relocation, must occur at expected location
2612
this_tree_map[file_id] = tree_state[1], this_path
2614
this_tree_map[file_id] = this_path, this_path
2615
check_valid_parent()
2616
if absent_positions == tree_count:
2617
raise AssertionError(
2618
"entry %r has no data for any tree." % (entry,))
2620
def _wipe_state(self):
2621
"""Forget all state information about the dirstate."""
2622
self._header_state = DirState.NOT_IN_MEMORY
2623
self._dirblock_state = DirState.NOT_IN_MEMORY
2624
self._consistency = DirState.CONSISTENT
2627
self._dirblocks = []
2628
self._id_index = None
2629
self._packed_stat_index = None
2630
self._end_of_header = None
2631
self._cutoff_time = None
2632
self._split_path_cache = {}
2634
def lock_read(self):
2635
"""Acquire a read lock on the dirstate."""
2636
if self._lock_token is not None:
2637
raise errors.LockContention(self._lock_token)
2638
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2639
# already in memory, we could read just the header and check for
2640
# any modification. If not modified, we can just leave things
2642
self._lock_token = lock.ReadLock(self._filename)
2643
self._lock_state = 'r'
2644
self._state_file = self._lock_token.f
2647
def lock_write(self):
2648
"""Acquire a write lock on the dirstate."""
2649
if self._lock_token is not None:
2650
raise errors.LockContention(self._lock_token)
2651
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2652
# already in memory, we could read just the header and check for
2653
# any modification. If not modified, we can just leave things
2655
self._lock_token = lock.WriteLock(self._filename)
2656
self._lock_state = 'w'
2657
self._state_file = self._lock_token.f
2661
"""Drop any locks held on the dirstate."""
2662
if self._lock_token is None:
2663
raise errors.LockNotHeld(self)
2664
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2665
# already in memory, we could read just the header and check for
2666
# any modification. If not modified, we can just leave things
2668
self._state_file = None
2669
self._lock_state = None
2670
self._lock_token.unlock()
2671
self._lock_token = None
2672
self._split_path_cache = {}
2674
def _requires_lock(self):
2675
"""Check that a lock is currently held by someone on the dirstate."""
2676
if not self._lock_token:
2677
raise errors.ObjectNotLocked(self)
2680
# Try to load the compiled form if possible
2682
from bzrlib._dirstate_helpers_c import (
2683
_read_dirblocks_c as _read_dirblocks,
2684
bisect_dirblock_c as bisect_dirblock,
2685
_bisect_path_left_c as _bisect_path_left,
2686
_bisect_path_right_c as _bisect_path_right,
2687
cmp_by_dirs_c as cmp_by_dirs,
2690
from bzrlib._dirstate_helpers_py import (
2691
_read_dirblocks_py as _read_dirblocks,
2692
bisect_dirblock_py as bisect_dirblock,
2693
_bisect_path_left_py as _bisect_path_left,
2694
_bisect_path_right_py as _bisect_path_right,
2695
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
