1
# Copyright (C) 2006, 2007 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""DirState objects record the state of a directory and its bzr metadata.
19
Pseudo EBNF grammar for the state file. Fields are separated by NULLs, and
20
lines by NL. The field delimiters are ommitted in the grammar, line delimiters
21
are not - this is done for clarity of reading. All string data is in utf8.
23
MINIKIND = "f" | "d" | "l" | "a" | "r" | "t";
26
WHOLE_NUMBER = {digit}, digit;
28
REVISION_ID = a non-empty utf8 string;
30
dirstate format = header line, full checksum, row count, parent details,
31
ghost_details, entries;
32
header line = "#bazaar dirstate flat format 3", NL;
33
full checksum = "crc32: ", ["-"], WHOLE_NUMBER, NL;
34
row count = "num_entries: ", WHOLE_NUMBER, NL;
35
parent_details = WHOLE NUMBER, {REVISION_ID}* NL;
36
ghost_details = WHOLE NUMBER, {REVISION_ID}*, NL;
38
entry = entry_key, current_entry_details, {parent_entry_details};
39
entry_key = dirname, basename, fileid;
40
current_entry_details = common_entry_details, working_entry_details;
41
parent_entry_details = common_entry_details, history_entry_details;
42
common_entry_details = MINIKIND, fingerprint, size, executable
43
working_entry_details = packed_stat
44
history_entry_details = REVISION_ID;
47
fingerprint = a nonempty utf8 sequence with meaning defined by minikind.
49
Given this definition, the following is useful to know:
50
entry (aka row) - all the data for a given key.
51
entry[0]: The key (dirname, basename, fileid)
55
entry[1]: The tree(s) data for this path and id combination.
56
entry[1][0]: The current tree
57
entry[1][1]: The second tree
59
For an entry for a tree, we have (using tree 0 - current tree) to demonstrate:
60
entry[1][0][0]: minikind
61
entry[1][0][1]: fingerprint
63
entry[1][0][3]: executable
64
entry[1][0][4]: packed_stat
66
entry[1][1][4]: revision_id
68
There may be multiple rows at the root, one per id present in the root, so the
69
in memory root row is now:
70
self._dirblocks[0] -> ('', [entry ...]),
71
and the entries in there are
74
entries[0][2]: file_id
75
entries[1][0]: The tree data for the current tree for this fileid at /
79
'r' is a relocated entry: This path is not present in this tree with this id,
80
but the id can be found at another location. The fingerprint is used to
81
point to the target location.
82
'a' is an absent entry: In that tree the id is not present at this path.
83
'd' is a directory entry: This path in this tree is a directory with the
84
current file id. There is no fingerprint for directories.
85
'f' is a file entry: As for directory, but its a file. The fingerprint is a
87
'l' is a symlink entry: As for directory, but a symlink. The fingerprint is the
89
't' is a reference to a nested subtree; the fingerprint is the referenced
94
The entries on disk and in memory are ordered according to the following keys:
96
directory, as a list of components
100
--- Format 1 had the following different definition: ---
101
rows = dirname, NULL, basename, NULL, MINIKIND, NULL, fileid_utf8, NULL,
102
WHOLE NUMBER (* size *), NULL, packed stat, NULL, sha1|symlink target,
104
PARENT ROW = NULL, revision_utf8, NULL, MINIKIND, NULL, dirname, NULL,
105
basename, NULL, WHOLE NUMBER (* size *), NULL, "y" | "n", NULL,
108
PARENT ROW's are emitted for every parent that is not in the ghosts details
109
line. That is, if the parents are foo, bar, baz, and the ghosts are bar, then
110
each row will have a PARENT ROW for foo and baz, but not for bar.
113
In any tree, a kind of 'moved' indicates that the fingerprint field
114
(which we treat as opaque data specific to the 'kind' anyway) has the
115
details for the id of this row in that tree.
117
I'm strongly tempted to add a id->path index as well, but I think that
118
where we need id->path mapping; we also usually read the whole file, so
119
I'm going to skip that for the moment, as we have the ability to locate
120
via bisect any path in any tree, and if we lookup things by path, we can
121
accumulate an id->path mapping as we go, which will tend to match what we
124
I plan to implement this asap, so please speak up now to alter/tweak the
125
design - and once we stabilise on this, I'll update the wiki page for
128
The rationale for all this is that we want fast operations for the
129
common case (diff/status/commit/merge on all files) and extremely fast
130
operations for the less common but still occurs a lot status/diff/commit
131
on specific files). Operations on specific files involve a scan for all
132
the children of a path, *in every involved tree*, which the current
133
format did not accommodate.
137
1) Fast end to end use for bzr's top 5 uses cases. (commmit/diff/status/merge/???)
138
2) fall back current object model as needed.
139
3) scale usably to the largest trees known today - say 50K entries. (mozilla
140
is an example of this)
144
Eventually reuse dirstate objects across locks IFF the dirstate file has not
145
been modified, but will require that we flush/ignore cached stat-hit data
146
because we won't want to restat all files on disk just because a lock was
147
acquired, yet we cannot trust the data after the previous lock was released.
149
Memory representation:
150
vector of all directories, and vector of the childen ?
152
root_entrie = (direntry for root, [parent_direntries_for_root]),
154
('', ['data for achild', 'data for bchild', 'data for cchild'])
155
('dir', ['achild', 'cchild', 'echild'])
157
- single bisect to find N subtrees from a path spec
158
- in-order for serialisation - this is 'dirblock' grouping.
159
- insertion of a file '/a' affects only the '/' child-vector, that is, to
160
insert 10K elements from scratch does not generates O(N^2) memoves of a
161
single vector, rather each individual, which tends to be limited to a
162
manageable number. Will scale badly on trees with 10K entries in a
163
single directory. compare with Inventory.InventoryDirectory which has
164
a dictionary for the children. No bisect capability, can only probe for
165
exact matches, or grab all elements and sort.
166
- What's the risk of error here? Once we have the base format being processed
167
we should have a net win regardless of optimality. So we are going to
168
go with what seems reasonable.
171
Maybe we should do a test profile of the core structure - 10K simulated
172
searches/lookups/etc?
174
Objects for each row?
175
The lifetime of Dirstate objects is current per lock, but see above for
176
possible extensions. The lifetime of a row from a dirstate is expected to be
177
very short in the optimistic case: which we are optimising for. For instance,
178
subtree status will determine from analysis of the disk data what rows need to
179
be examined at all, and will be able to determine from a single row whether
180
that file has altered or not, so we are aiming to process tens of thousands of
181
entries each second within the dirstate context, before exposing anything to
182
the larger codebase. This suggests we want the time for a single file
183
comparison to be < 0.1 milliseconds. That would give us 10000 paths per second
184
processed, and to scale to 100 thousand we'll another order of magnitude to do
185
that. Now, as the lifetime for all unchanged entries is the time to parse, stat
186
the file on disk, and then immediately discard, the overhead of object creation
187
becomes a significant cost.
189
Figures: Creating a tuple from from 3 elements was profiled at 0.0625
190
microseconds, whereas creating a object which is subclassed from tuple was
191
0.500 microseconds, and creating an object with 3 elements and slots was 3
192
microseconds long. 0.1 milliseconds is 100 microseconds, and ideally we'll get
193
down to 10 microseconds for the total processing - having 33% of that be object
194
creation is a huge overhead. There is a potential cost in using tuples within
195
each row which is that the conditional code to do comparisons may be slower
196
than method invocation, but method invocation is known to be slow due to stack
197
frame creation, so avoiding methods in these tight inner loops in unfortunately
198
desirable. We can consider a pyrex version of this with objects in future if
207
from stat import S_IEXEC
224
def pack_stat(st, _encode=binascii.b2a_base64, _pack=struct.pack):
225
"""Convert stat values into a packed representation."""
226
# jam 20060614 it isn't really worth removing more entries if we
227
# are going to leave it in packed form.
228
# With only st_mtime and st_mode filesize is 5.5M and read time is 275ms
229
# With all entries, filesize is 5.9M and read time is maybe 280ms
230
# well within the noise margin
232
# base64 encoding always adds a final newline, so strip it off
233
# The current version
234
return _encode(_pack('>LLLLLL'
235
, st.st_size, int(st.st_mtime), int(st.st_ctime)
236
, st.st_dev, st.st_ino & 0xFFFFFFFF, st.st_mode))[:-1]
237
# This is 0.060s / 1.520s faster by not encoding as much information
238
# return _encode(_pack('>LL', int(st.st_mtime), st.st_mode))[:-1]
239
# This is not strictly faster than _encode(_pack())[:-1]
240
# return '%X.%X.%X.%X.%X.%X' % (
241
# st.st_size, int(st.st_mtime), int(st.st_ctime),
242
# st.st_dev, st.st_ino, st.st_mode)
243
# Similar to the _encode(_pack('>LL'))
244
# return '%X.%X' % (int(st.st_mtime), st.st_mode)
247
class DirState(object):
248
"""Record directory and metadata state for fast access.
250
A dirstate is a specialised data structure for managing local working
251
tree state information. Its not yet well defined whether it is platform
252
specific, and if it is how we detect/parameterise that.
254
Dirstates use the usual lock_write, lock_read and unlock mechanisms.
255
Unlike most bzr disk formats, DirStates must be locked for reading, using
256
lock_read. (This is an os file lock internally.) This is necessary
257
because the file can be rewritten in place.
259
DirStates must be explicitly written with save() to commit changes; just
260
unlocking them does not write the changes to disk.
263
_kind_to_minikind = {
269
'tree-reference': 't',
271
_minikind_to_kind = {
277
't': 'tree-reference',
279
_stat_to_minikind = {
284
_to_yesno = {True:'y', False: 'n'} # TODO profile the performance gain
285
# of using int conversion rather than a dict here. AND BLAME ANDREW IF
288
# TODO: jam 20070221 Figure out what to do if we have a record that exceeds
289
# the BISECT_PAGE_SIZE. For now, we just have to make it large enough
290
# that we are sure a single record will always fit.
291
BISECT_PAGE_SIZE = 4096
294
IN_MEMORY_UNMODIFIED = 1
295
IN_MEMORY_MODIFIED = 2
297
# A pack_stat (the x's) that is just noise and will never match the output
300
NULL_PARENT_DETAILS = ('a', '', 0, False, '')
302
HEADER_FORMAT_2 = '#bazaar dirstate flat format 2\n'
303
HEADER_FORMAT_3 = '#bazaar dirstate flat format 3\n'
305
def __init__(self, path):
306
"""Create a DirState object.
310
:attr _root_entrie: The root row of the directory/file information,
311
- contains the path to / - '', ''
312
- kind of 'directory',
313
- the file id of the root in utf8
316
- and no sha information.
317
:param path: The path at which the dirstate file on disk should live.
319
# _header_state and _dirblock_state represent the current state
320
# of the dirstate metadata and the per-row data respectiely.
321
# NOT_IN_MEMORY indicates that no data is in memory
322
# IN_MEMORY_UNMODIFIED indicates that what we have in memory
323
# is the same as is on disk
324
# IN_MEMORY_MODIFIED indicates that we have a modified version
325
# of what is on disk.
326
# In future we will add more granularity, for instance _dirblock_state
327
# will probably support partially-in-memory as a separate variable,
328
# allowing for partially-in-memory unmodified and partially-in-memory
330
self._header_state = DirState.NOT_IN_MEMORY
331
self._dirblock_state = DirState.NOT_IN_MEMORY
335
self._state_file = None
336
self._filename = path
337
self._lock_token = None
338
self._lock_state = None
339
self._id_index = None
340
# a map from packed_stat to sha's.
341
self._packed_stat_index = None
342
self._end_of_header = None
343
self._cutoff_time = None
344
self._split_path_cache = {}
345
self._bisect_page_size = DirState.BISECT_PAGE_SIZE
346
if 'hashcache' in debug.debug_flags:
347
self._sha1_file = self._sha1_file_and_mutter
349
self._sha1_file = osutils.sha_file_by_name
350
# These two attributes provide a simple cache for lookups into the
351
# dirstate in-memory vectors. By probing respectively for the last
352
# block, and for the next entry, we save nearly 2 bisections per path
354
self._last_block_index = None
355
self._last_entry_index = None
359
(self.__class__.__name__, self._filename)
361
def add(self, path, file_id, kind, stat, fingerprint):
362
"""Add a path to be tracked.
364
:param path: The path within the dirstate - '' is the root, 'foo' is the
365
path foo within the root, 'foo/bar' is the path bar within foo
367
:param file_id: The file id of the path being added.
368
:param kind: The kind of the path, as a string like 'file',
370
:param stat: The output of os.lstat for the path.
371
:param fingerprint: The sha value of the file,
372
or the target of a symlink,
373
or the referenced revision id for tree-references,
374
or '' for directories.
377
# find the block its in.
378
# find the location in the block.
379
# check its not there
381
#------- copied from inventory.ensure_normalized_name - keep synced.
382
# --- normalized_filename wants a unicode basename only, so get one.
383
dirname, basename = osutils.split(path)
384
# we dont import normalized_filename directly because we want to be
385
# able to change the implementation at runtime for tests.
386
norm_name, can_access = osutils.normalized_filename(basename)
387
if norm_name != basename:
391
raise errors.InvalidNormalization(path)
392
# you should never have files called . or ..; just add the directory
393
# in the parent, or according to the special treatment for the root
394
if basename == '.' or basename == '..':
395
raise errors.InvalidEntryName(path)
396
# now that we've normalised, we need the correct utf8 path and
397
# dirname and basename elements. This single encode and split should be
398
# faster than three separate encodes.
399
utf8path = (dirname + '/' + basename).strip('/').encode('utf8')
400
dirname, basename = osutils.split(utf8path)
401
assert file_id.__class__ == str, \
402
"must be a utf8 file_id not %s" % (type(file_id))
403
# Make sure the file_id does not exist in this tree
404
file_id_entry = self._get_entry(0, fileid_utf8=file_id)
405
if file_id_entry != (None, None):
406
path = osutils.pathjoin(file_id_entry[0][0], file_id_entry[0][1])
407
kind = DirState._minikind_to_kind[file_id_entry[1][0][0]]
408
info = '%s:%s' % (kind, path)
409
raise errors.DuplicateFileId(file_id, info)
410
first_key = (dirname, basename, '')
411
block_index, present = self._find_block_index_from_key(first_key)
413
# check the path is not in the tree
414
block = self._dirblocks[block_index][1]
415
entry_index, _ = self._find_entry_index(first_key, block)
416
while (entry_index < len(block) and
417
block[entry_index][0][0:2] == first_key[0:2]):
418
if block[entry_index][1][0][0] not in 'ar':
419
# this path is in the dirstate in the current tree.
420
raise Exception, "adding already added path!"
423
# The block where we want to put the file is not present. But it
424
# might be because the directory was empty, or not loaded yet. Look
425
# for a parent entry, if not found, raise NotVersionedError
426
parent_dir, parent_base = osutils.split(dirname)
427
parent_block_idx, parent_entry_idx, _, parent_present = \
428
self._get_block_entry_index(parent_dir, parent_base, 0)
429
if not parent_present:
430
raise errors.NotVersionedError(path, str(self))
431
self._ensure_block(parent_block_idx, parent_entry_idx, dirname)
432
block = self._dirblocks[block_index][1]
433
entry_key = (dirname, basename, file_id)
436
packed_stat = DirState.NULLSTAT
439
packed_stat = pack_stat(stat)
440
parent_info = self._empty_parent_info()
441
minikind = DirState._kind_to_minikind[kind]
443
entry_data = entry_key, [
444
(minikind, fingerprint, size, False, packed_stat),
446
elif kind == 'directory':
447
entry_data = entry_key, [
448
(minikind, '', 0, False, packed_stat),
450
elif kind == 'symlink':
451
entry_data = entry_key, [
452
(minikind, fingerprint, size, False, packed_stat),
454
elif kind == 'tree-reference':
455
entry_data = entry_key, [
456
(minikind, fingerprint, 0, False, packed_stat),
459
raise errors.BzrError('unknown kind %r' % kind)
460
entry_index, present = self._find_entry_index(entry_key, block)
462
block.insert(entry_index, entry_data)
464
assert block[entry_index][1][0][0] == 'a', " %r(%r) already added" % (basename, file_id)
465
block[entry_index][1][0] = entry_data[1][0]
467
if kind == 'directory':
468
# insert a new dirblock
469
self._ensure_block(block_index, entry_index, utf8path)
470
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
472
self._id_index.setdefault(entry_key[2], set()).add(entry_key)
474
def _bisect(self, paths):
475
"""Bisect through the disk structure for specific rows.
477
:param paths: A list of paths to find
478
:return: A dict mapping path => entries for found entries. Missing
479
entries will not be in the map.
480
The list is not sorted, and entries will be populated
481
based on when they were read.
483
self._requires_lock()
484
# We need the file pointer to be right after the initial header block
485
self._read_header_if_needed()
486
# If _dirblock_state was in memory, we should just return info from
487
# there, this function is only meant to handle when we want to read
489
assert self._dirblock_state == DirState.NOT_IN_MEMORY
491
# The disk representation is generally info + '\0\n\0' at the end. But
492
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
493
# Because it means we can sync on the '\n'
494
state_file = self._state_file
495
file_size = os.fstat(state_file.fileno()).st_size
496
# We end up with 2 extra fields, we should have a trailing '\n' to
497
# ensure that we read the whole record, and we should have a precursur
498
# '' which ensures that we start after the previous '\n'
499
entry_field_count = self._fields_per_entry() + 1
501
low = self._end_of_header
502
high = file_size - 1 # Ignore the final '\0'
503
# Map from (dir, name) => entry
506
# Avoid infinite seeking
507
max_count = 30*len(paths)
509
# pending is a list of places to look.
510
# each entry is a tuple of low, high, dir_names
511
# low -> the first byte offset to read (inclusive)
512
# high -> the last byte offset (inclusive)
513
# dir_names -> The list of (dir, name) pairs that should be found in
514
# the [low, high] range
515
pending = [(low, high, paths)]
517
page_size = self._bisect_page_size
519
fields_to_entry = self._get_fields_to_entry()
522
low, high, cur_files = pending.pop()
524
if not cur_files or low >= high:
529
if count > max_count:
530
raise errors.BzrError('Too many seeks, most likely a bug.')
532
mid = max(low, (low+high-page_size)/2)
535
# limit the read size, so we don't end up reading data that we have
537
read_size = min(page_size, (high-mid)+1)
538
block = state_file.read(read_size)
541
entries = block.split('\n')
544
# We didn't find a '\n', so we cannot have found any records.
545
# So put this range back and try again. But we know we have to
546
# increase the page size, because a single read did not contain
547
# a record break (so records must be larger than page_size)
549
pending.append((low, high, cur_files))
552
# Check the first and last entries, in case they are partial, or if
553
# we don't care about the rest of this page
555
first_fields = entries[0].split('\0')
556
if len(first_fields) < entry_field_count:
557
# We didn't get the complete first entry
558
# so move start, and grab the next, which
559
# should be a full entry
560
start += len(entries[0])+1
561
first_fields = entries[1].split('\0')
564
if len(first_fields) <= 2:
565
# We didn't even get a filename here... what do we do?
566
# Try a large page size and repeat this query
568
pending.append((low, high, cur_files))
571
# Find what entries we are looking for, which occur before and
572
# after this first record.
575
first_path = first_fields[1] + '/' + first_fields[2]
577
first_path = first_fields[2]
578
first_loc = _bisect_path_left(cur_files, first_path)
580
# These exist before the current location
581
pre = cur_files[:first_loc]
582
# These occur after the current location, which may be in the
583
# data we read, or might be after the last entry
584
post = cur_files[first_loc:]
586
if post and len(first_fields) >= entry_field_count:
587
# We have files after the first entry
589
# Parse the last entry
590
last_entry_num = len(entries)-1
591
last_fields = entries[last_entry_num].split('\0')
592
if len(last_fields) < entry_field_count:
593
# The very last hunk was not complete,
594
# read the previous hunk
595
after = mid + len(block) - len(entries[-1])
597
last_fields = entries[last_entry_num].split('\0')
599
after = mid + len(block)
602
last_path = last_fields[1] + '/' + last_fields[2]
604
last_path = last_fields[2]
605
last_loc = _bisect_path_right(post, last_path)
607
middle_files = post[:last_loc]
608
post = post[last_loc:]
611
# We have files that should occur in this block
612
# (>= first, <= last)
613
# Either we will find them here, or we can mark them as
616
if middle_files[0] == first_path:
617
# We might need to go before this location
618
pre.append(first_path)
619
if middle_files[-1] == last_path:
620
post.insert(0, last_path)
622
# Find out what paths we have
623
paths = {first_path:[first_fields]}
624
# last_path might == first_path so we need to be
625
# careful if we should append rather than overwrite
626
if last_entry_num != first_entry_num:
627
paths.setdefault(last_path, []).append(last_fields)
628
for num in xrange(first_entry_num+1, last_entry_num):
629
# TODO: jam 20070223 We are already splitting here, so
630
# shouldn't we just split the whole thing rather
631
# than doing the split again in add_one_record?
632
fields = entries[num].split('\0')
634
path = fields[1] + '/' + fields[2]
637
paths.setdefault(path, []).append(fields)
639
for path in middle_files:
640
for fields in paths.get(path, []):
641
# offset by 1 because of the opening '\0'
642
# consider changing fields_to_entry to avoid the
644
entry = fields_to_entry(fields[1:])
645
found.setdefault(path, []).append(entry)
647
# Now we have split up everything into pre, middle, and post, and
648
# we have handled everything that fell in 'middle'.
649
# We add 'post' first, so that we prefer to seek towards the
650
# beginning, so that we will tend to go as early as we need, and
651
# then only seek forward after that.
653
pending.append((after, high, post))
655
pending.append((low, start-1, pre))
657
# Consider that we may want to return the directory entries in sorted
658
# order. For now, we just return them in whatever order we found them,
659
# and leave it up to the caller if they care if it is ordered or not.
662
def _bisect_dirblocks(self, dir_list):
663
"""Bisect through the disk structure to find entries in given dirs.
665
_bisect_dirblocks is meant to find the contents of directories, which
666
differs from _bisect, which only finds individual entries.
668
:param dir_list: A sorted list of directory names ['', 'dir', 'foo'].
669
:return: A map from dir => entries_for_dir
671
# TODO: jam 20070223 A lot of the bisecting logic could be shared
672
# between this and _bisect. It would require parameterizing the
673
# inner loop with a function, though. We should evaluate the
674
# performance difference.
675
self._requires_lock()
676
# We need the file pointer to be right after the initial header block
677
self._read_header_if_needed()
678
# If _dirblock_state was in memory, we should just return info from
679
# there, this function is only meant to handle when we want to read
681
assert self._dirblock_state == DirState.NOT_IN_MEMORY
683
# The disk representation is generally info + '\0\n\0' at the end. But
684
# for bisecting, it is easier to treat this as '\0' + info + '\0\n'
685
# Because it means we can sync on the '\n'
686
state_file = self._state_file
687
file_size = os.fstat(state_file.fileno()).st_size
688
# We end up with 2 extra fields, we should have a trailing '\n' to
689
# ensure that we read the whole record, and we should have a precursur
690
# '' which ensures that we start after the previous '\n'
691
entry_field_count = self._fields_per_entry() + 1
693
low = self._end_of_header
694
high = file_size - 1 # Ignore the final '\0'
695
# Map from dir => entry
698
# Avoid infinite seeking
699
max_count = 30*len(dir_list)
701
# pending is a list of places to look.
702
# each entry is a tuple of low, high, dir_names
703
# low -> the first byte offset to read (inclusive)
704
# high -> the last byte offset (inclusive)
705
# dirs -> The list of directories that should be found in
706
# the [low, high] range
707
pending = [(low, high, dir_list)]
709
page_size = self._bisect_page_size
711
fields_to_entry = self._get_fields_to_entry()
714
low, high, cur_dirs = pending.pop()
716
if not cur_dirs or low >= high:
721
if count > max_count:
722
raise errors.BzrError('Too many seeks, most likely a bug.')
724
mid = max(low, (low+high-page_size)/2)
727
# limit the read size, so we don't end up reading data that we have
729
read_size = min(page_size, (high-mid)+1)
730
block = state_file.read(read_size)
733
entries = block.split('\n')
736
# We didn't find a '\n', so we cannot have found any records.
737
# So put this range back and try again. But we know we have to
738
# increase the page size, because a single read did not contain
739
# a record break (so records must be larger than page_size)
741
pending.append((low, high, cur_dirs))
744
# Check the first and last entries, in case they are partial, or if
745
# we don't care about the rest of this page
747
first_fields = entries[0].split('\0')
748
if len(first_fields) < entry_field_count:
749
# We didn't get the complete first entry
750
# so move start, and grab the next, which
751
# should be a full entry
752
start += len(entries[0])+1
753
first_fields = entries[1].split('\0')
756
if len(first_fields) <= 1:
757
# We didn't even get a dirname here... what do we do?
758
# Try a large page size and repeat this query
760
pending.append((low, high, cur_dirs))
763
# Find what entries we are looking for, which occur before and
764
# after this first record.
766
first_dir = first_fields[1]
767
first_loc = bisect.bisect_left(cur_dirs, first_dir)
769
# These exist before the current location
770
pre = cur_dirs[:first_loc]
771
# These occur after the current location, which may be in the
772
# data we read, or might be after the last entry
773
post = cur_dirs[first_loc:]
775
if post and len(first_fields) >= entry_field_count:
776
# We have records to look at after the first entry
778
# Parse the last entry
779
last_entry_num = len(entries)-1
780
last_fields = entries[last_entry_num].split('\0')
781
if len(last_fields) < entry_field_count:
782
# The very last hunk was not complete,
783
# read the previous hunk
784
after = mid + len(block) - len(entries[-1])
786
last_fields = entries[last_entry_num].split('\0')
788
after = mid + len(block)
790
last_dir = last_fields[1]
791
last_loc = bisect.bisect_right(post, last_dir)
793
middle_files = post[:last_loc]
794
post = post[last_loc:]
797
# We have files that should occur in this block
798
# (>= first, <= last)
799
# Either we will find them here, or we can mark them as
802
if middle_files[0] == first_dir:
803
# We might need to go before this location
804
pre.append(first_dir)
805
if middle_files[-1] == last_dir:
806
post.insert(0, last_dir)
808
# Find out what paths we have
809
paths = {first_dir:[first_fields]}
810
# last_dir might == first_dir so we need to be
811
# careful if we should append rather than overwrite
812
if last_entry_num != first_entry_num:
813
paths.setdefault(last_dir, []).append(last_fields)
814
for num in xrange(first_entry_num+1, last_entry_num):
815
# TODO: jam 20070223 We are already splitting here, so
816
# shouldn't we just split the whole thing rather
817
# than doing the split again in add_one_record?
818
fields = entries[num].split('\0')
819
paths.setdefault(fields[1], []).append(fields)
821
for cur_dir in middle_files:
822
for fields in paths.get(cur_dir, []):
823
# offset by 1 because of the opening '\0'
824
# consider changing fields_to_entry to avoid the
826
entry = fields_to_entry(fields[1:])
827
found.setdefault(cur_dir, []).append(entry)
829
# Now we have split up everything into pre, middle, and post, and
830
# we have handled everything that fell in 'middle'.
831
# We add 'post' first, so that we prefer to seek towards the
832
# beginning, so that we will tend to go as early as we need, and
833
# then only seek forward after that.
835
pending.append((after, high, post))
837
pending.append((low, start-1, pre))
841
def _bisect_recursive(self, paths):
842
"""Bisect for entries for all paths and their children.
844
This will use bisect to find all records for the supplied paths. It
845
will then continue to bisect for any records which are marked as
846
directories. (and renames?)
848
:param paths: A sorted list of (dir, name) pairs
849
eg: [('', 'a'), ('', 'f'), ('a/b', 'c')]
850
:return: A dictionary mapping (dir, name, file_id) => [tree_info]
852
# Map from (dir, name, file_id) => [tree_info]
855
found_dir_names = set()
857
# Directories that have been read
858
processed_dirs = set()
859
# Get the ball rolling with the first bisect for all entries.
860
newly_found = self._bisect(paths)
863
# Directories that need to be read
865
paths_to_search = set()
866
for entry_list in newly_found.itervalues():
867
for dir_name_id, trees_info in entry_list:
868
found[dir_name_id] = trees_info
869
found_dir_names.add(dir_name_id[:2])
871
for tree_info in trees_info:
872
minikind = tree_info[0]
875
# We already processed this one as a directory,
876
# we don't need to do the extra work again.
878
subdir, name, file_id = dir_name_id
879
path = osutils.pathjoin(subdir, name)
881
if path not in processed_dirs:
882
pending_dirs.add(path)
883
elif minikind == 'r':
884
# Rename, we need to directly search the target
885
# which is contained in the fingerprint column
886
dir_name = osutils.split(tree_info[1])
887
if dir_name[0] in pending_dirs:
888
# This entry will be found in the dir search
890
if dir_name not in found_dir_names:
891
paths_to_search.add(tree_info[1])
892
# Now we have a list of paths to look for directly, and
893
# directory blocks that need to be read.
894
# newly_found is mixing the keys between (dir, name) and path
895
# entries, but that is okay, because we only really care about the
897
newly_found = self._bisect(sorted(paths_to_search))
898
newly_found.update(self._bisect_dirblocks(sorted(pending_dirs)))
899
processed_dirs.update(pending_dirs)
902
def _empty_parent_info(self):
903
return [DirState.NULL_PARENT_DETAILS] * (len(self._parents) -
906
def _ensure_block(self, parent_block_index, parent_row_index, dirname):
907
"""Ensure a block for dirname exists.
909
This function exists to let callers which know that there is a
910
directory dirname ensure that the block for it exists. This block can
911
fail to exist because of demand loading, or because a directory had no
912
children. In either case it is not an error. It is however an error to
913
call this if there is no parent entry for the directory, and thus the
914
function requires the coordinates of such an entry to be provided.
916
The root row is special cased and can be indicated with a parent block
919
:param parent_block_index: The index of the block in which dirname's row
921
:param parent_row_index: The index in the parent block where the row
923
:param dirname: The utf8 dirname to ensure there is a block for.
924
:return: The index for the block.
926
if dirname == '' and parent_row_index == 0 and parent_block_index == 0:
927
# This is the signature of the root row, and the
928
# contents-of-root row is always index 1
930
# the basename of the directory must be the end of its full name.
931
if not (parent_block_index == -1 and
932
parent_block_index == -1 and dirname == ''):
933
assert dirname.endswith(
934
self._dirblocks[parent_block_index][1][parent_row_index][0][1])
935
block_index, present = self._find_block_index_from_key((dirname, '', ''))
937
## In future, when doing partial parsing, this should load and
938
# populate the entire block.
939
self._dirblocks.insert(block_index, (dirname, []))
942
def _entries_to_current_state(self, new_entries):
943
"""Load new_entries into self.dirblocks.
945
Process new_entries into the current state object, making them the active
946
state. The entries are grouped together by directory to form dirblocks.
948
:param new_entries: A sorted list of entries. This function does not sort
949
to prevent unneeded overhead when callers have a sorted list already.
952
assert new_entries[0][0][0:2] == ('', ''), \
953
"Missing root row %r" % (new_entries[0][0],)
954
# The two blocks here are deliberate: the root block and the
955
# contents-of-root block.
956
self._dirblocks = [('', []), ('', [])]
957
current_block = self._dirblocks[0][1]
960
append_entry = current_block.append
961
for entry in new_entries:
962
if entry[0][0] != current_dirname:
963
# new block - different dirname
965
current_dirname = entry[0][0]
966
self._dirblocks.append((current_dirname, current_block))
967
append_entry = current_block.append
968
# append the entry to the current block
970
self._split_root_dirblock_into_contents()
972
def _split_root_dirblock_into_contents(self):
973
"""Split the root dirblocks into root and contents-of-root.
975
After parsing by path, we end up with root entries and contents-of-root
976
entries in the same block. This loop splits them out again.
978
# The above loop leaves the "root block" entries mixed with the
979
# "contents-of-root block". But we don't want an if check on
980
# all entries, so instead we just fix it up here.
981
assert self._dirblocks[1] == ('', [])
983
contents_of_root_block = []
984
for entry in self._dirblocks[0][1]:
985
if not entry[0][1]: # This is a root entry
986
root_block.append(entry)
988
contents_of_root_block.append(entry)
989
self._dirblocks[0] = ('', root_block)
990
self._dirblocks[1] = ('', contents_of_root_block)
992
def _entry_to_line(self, entry):
993
"""Serialize entry to a NULL delimited line ready for _get_output_lines.
995
:param entry: An entry_tuple as defined in the module docstring.
997
entire_entry = list(entry[0])
998
for tree_number, tree_data in enumerate(entry[1]):
999
# (minikind, fingerprint, size, executable, tree_specific_string)
1000
entire_entry.extend(tree_data)
1001
# 3 for the key, 5 for the fields per tree.
1002
tree_offset = 3 + tree_number * 5
1004
entire_entry[tree_offset + 0] = tree_data[0]
1006
entire_entry[tree_offset + 2] = str(tree_data[2])
1008
entire_entry[tree_offset + 3] = DirState._to_yesno[tree_data[3]]
1009
return '\0'.join(entire_entry)
1011
def _fields_per_entry(self):
1012
"""How many null separated fields should be in each entry row.
1014
Each line now has an extra '\n' field which is not used
1015
so we just skip over it
1017
3 fields for the key
1018
+ number of fields per tree_data (5) * tree count
1021
tree_count = 1 + self._num_present_parents()
1022
return 3 + 5 * tree_count + 1
1024
def _find_block(self, key, add_if_missing=False):
1025
"""Return the block that key should be present in.
1027
:param key: A dirstate entry key.
1028
:return: The block tuple.
1030
block_index, present = self._find_block_index_from_key(key)
1032
if not add_if_missing:
1033
# check to see if key is versioned itself - we might want to
1034
# add it anyway, because dirs with no entries dont get a
1035
# dirblock at parse time.
1036
# This is an uncommon branch to take: most dirs have children,
1037
# and most code works with versioned paths.
1038
parent_base, parent_name = osutils.split(key[0])
1039
if not self._get_block_entry_index(parent_base, parent_name, 0)[3]:
1040
# some parent path has not been added - its an error to add
1042
raise errors.NotVersionedError(key[0:2], str(self))
1043
self._dirblocks.insert(block_index, (key[0], []))
1044
return self._dirblocks[block_index]
1046
def _find_block_index_from_key(self, key):
1047
"""Find the dirblock index for a key.
1049
:return: The block index, True if the block for the key is present.
1051
if key[0:2] == ('', ''):
1054
if (self._last_block_index is not None and
1055
self._dirblocks[self._last_block_index][0] == key[0]):
1056
return self._last_block_index, True
1059
block_index = bisect_dirblock(self._dirblocks, key[0], 1,
1060
cache=self._split_path_cache)
1061
# _right returns one-past-where-key is so we have to subtract
1062
# one to use it. we use _right here because there are two
1063
# '' blocks - the root, and the contents of root
1064
# we always have a minimum of 2 in self._dirblocks: root and
1065
# root-contents, and for '', we get 2 back, so this is
1066
# simple and correct:
1067
present = (block_index < len(self._dirblocks) and
1068
self._dirblocks[block_index][0] == key[0])
1069
self._last_block_index = block_index
1070
# Reset the entry index cache to the beginning of the block.
1071
self._last_entry_index = -1
1072
return block_index, present
1074
def _find_entry_index(self, key, block):
1075
"""Find the entry index for a key in a block.
1077
:return: The entry index, True if the entry for the key is present.
1079
len_block = len(block)
1081
if self._last_entry_index is not None:
1083
entry_index = self._last_entry_index + 1
1084
# A hit is when the key is after the last slot, and before or
1085
# equal to the next slot.
1086
if ((entry_index > 0 and block[entry_index - 1][0] < key) and
1087
key <= block[entry_index][0]):
1088
self._last_entry_index = entry_index
1089
present = (block[entry_index][0] == key)
1090
return entry_index, present
1093
entry_index = bisect.bisect_left(block, (key, []))
1094
present = (entry_index < len_block and
1095
block[entry_index][0] == key)
1096
self._last_entry_index = entry_index
1097
return entry_index, present
1100
def from_tree(tree, dir_state_filename):
1101
"""Create a dirstate from a bzr Tree.
1103
:param tree: The tree which should provide parent information and
1105
:return: a DirState object which is currently locked for writing.
1106
(it was locked by DirState.initialize)
1108
result = DirState.initialize(dir_state_filename)
1112
parent_ids = tree.get_parent_ids()
1113
num_parents = len(parent_ids)
1115
for parent_id in parent_ids:
1116
parent_tree = tree.branch.repository.revision_tree(parent_id)
1117
parent_trees.append((parent_id, parent_tree))
1118
parent_tree.lock_read()
1119
result.set_parent_trees(parent_trees, [])
1120
result.set_state_from_inventory(tree.inventory)
1122
for revid, parent_tree in parent_trees:
1123
parent_tree.unlock()
1126
# The caller won't have a chance to unlock this, so make sure we
1132
def update_entry(self, entry, abspath, stat_value,
1133
_stat_to_minikind=_stat_to_minikind,
1134
_pack_stat=pack_stat):
1135
"""Update the entry based on what is actually on disk.
1137
:param entry: This is the dirblock entry for the file in question.
1138
:param abspath: The path on disk for this file.
1139
:param stat_value: (optional) if we already have done a stat on the
1141
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
1145
minikind = _stat_to_minikind[stat_value.st_mode & 0170000]
1149
packed_stat = _pack_stat(stat_value)
1150
(saved_minikind, saved_link_or_sha1, saved_file_size,
1151
saved_executable, saved_packed_stat) = entry[1][0]
1153
if (minikind == saved_minikind
1154
and packed_stat == saved_packed_stat):
1155
# The stat hasn't changed since we saved, so we can re-use the
1160
# size should also be in packed_stat
1161
if saved_file_size == stat_value.st_size:
1162
return saved_link_or_sha1
1164
# If we have gotten this far, that means that we need to actually
1165
# process this entry.
1168
link_or_sha1 = self._sha1_file(abspath)
1169
executable = self._is_executable(stat_value.st_mode,
1171
if self._cutoff_time is None:
1172
self._sha_cutoff_time()
1173
if (stat_value.st_mtime < self._cutoff_time
1174
and stat_value.st_ctime < self._cutoff_time):
1175
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
1176
executable, packed_stat)
1178
entry[1][0] = ('f', '', stat_value.st_size,
1179
executable, DirState.NULLSTAT)
1180
elif minikind == 'd':
1182
entry[1][0] = ('d', '', 0, False, packed_stat)
1183
if saved_minikind != 'd':
1184
# This changed from something into a directory. Make sure we
1185
# have a directory block for it. This doesn't happen very
1186
# often, so this doesn't have to be super fast.
1187
block_index, entry_index, dir_present, file_present = \
1188
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
1189
self._ensure_block(block_index, entry_index,
1190
osutils.pathjoin(entry[0][0], entry[0][1]))
1191
elif minikind == 'l':
1192
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
1193
if self._cutoff_time is None:
1194
self._sha_cutoff_time()
1195
if (stat_value.st_mtime < self._cutoff_time
1196
and stat_value.st_ctime < self._cutoff_time):
1197
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
1200
entry[1][0] = ('l', '', stat_value.st_size,
1201
False, DirState.NULLSTAT)
1202
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1205
def _sha_cutoff_time(self):
1206
"""Return cutoff time.
1208
Files modified more recently than this time are at risk of being
1209
undetectably modified and so can't be cached.
1211
# Cache the cutoff time as long as we hold a lock.
1212
# time.time() isn't super expensive (approx 3.38us), but
1213
# when you call it 50,000 times it adds up.
1214
# For comparison, os.lstat() costs 7.2us if it is hot.
1215
self._cutoff_time = int(time.time()) - 3
1216
return self._cutoff_time
1218
def _lstat(self, abspath, entry):
1219
"""Return the os.lstat value for this path."""
1220
return os.lstat(abspath)
1222
def _sha1_file_and_mutter(self, abspath):
1223
# when -Dhashcache is turned on, this is monkey-patched in to log
1225
trace.mutter("dirstate sha1 " + abspath)
1226
return osutils.sha_file_by_name(abspath)
1228
def _is_executable(self, mode, old_executable):
1229
"""Is this file executable?"""
1230
return bool(S_IEXEC & mode)
1232
def _is_executable_win32(self, mode, old_executable):
1233
"""On win32 the executable bit is stored in the dirstate."""
1234
return old_executable
1236
if sys.platform == 'win32':
1237
_is_executable = _is_executable_win32
1239
def _read_link(self, abspath, old_link):
1240
"""Read the target of a symlink"""
1241
# TODO: jam 200700301 On Win32, this could just return the value
1242
# already in memory. However, this really needs to be done at a
1243
# higher level, because there either won't be anything on disk,
1244
# or the thing on disk will be a file.
1245
return os.readlink(abspath)
1247
def get_ghosts(self):
1248
"""Return a list of the parent tree revision ids that are ghosts."""
1249
self._read_header_if_needed()
1252
def get_lines(self):
1253
"""Serialise the entire dirstate to a sequence of lines."""
1254
if (self._header_state == DirState.IN_MEMORY_UNMODIFIED and
1255
self._dirblock_state == DirState.IN_MEMORY_UNMODIFIED):
1256
# read whats on disk.
1257
self._state_file.seek(0)
1258
return self._state_file.readlines()
1260
lines.append(self._get_parents_line(self.get_parent_ids()))
1261
lines.append(self._get_ghosts_line(self._ghosts))
1262
# append the root line which is special cased
1263
lines.extend(map(self._entry_to_line, self._iter_entries()))
1264
return self._get_output_lines(lines)
1266
def _get_ghosts_line(self, ghost_ids):
1267
"""Create a line for the state file for ghost information."""
1268
return '\0'.join([str(len(ghost_ids))] + ghost_ids)
1270
def _get_parents_line(self, parent_ids):
1271
"""Create a line for the state file for parents information."""
1272
return '\0'.join([str(len(parent_ids))] + parent_ids)
1274
def _get_fields_to_entry(self):
1275
"""Get a function which converts entry fields into a entry record.
1277
This handles size and executable, as well as parent records.
1279
:return: A function which takes a list of fields, and returns an
1280
appropriate record for storing in memory.
1282
# This is intentionally unrolled for performance
1283
num_present_parents = self._num_present_parents()
1284
if num_present_parents == 0:
1285
def fields_to_entry_0_parents(fields, _int=int):
1286
path_name_file_id_key = (fields[0], fields[1], fields[2])
1287
return (path_name_file_id_key, [
1289
fields[3], # minikind
1290
fields[4], # fingerprint
1291
_int(fields[5]), # size
1292
fields[6] == 'y', # executable
1293
fields[7], # packed_stat or revision_id
1295
return fields_to_entry_0_parents
1296
elif num_present_parents == 1:
1297
def fields_to_entry_1_parent(fields, _int=int):
1298
path_name_file_id_key = (fields[0], fields[1], fields[2])
1299
return (path_name_file_id_key, [
1301
fields[3], # minikind
1302
fields[4], # fingerprint
1303
_int(fields[5]), # size
1304
fields[6] == 'y', # executable
1305
fields[7], # packed_stat or revision_id
1308
fields[8], # minikind
1309
fields[9], # fingerprint
1310
_int(fields[10]), # size
1311
fields[11] == 'y', # executable
1312
fields[12], # packed_stat or revision_id
1315
return fields_to_entry_1_parent
1316
elif num_present_parents == 2:
1317
def fields_to_entry_2_parents(fields, _int=int):
1318
path_name_file_id_key = (fields[0], fields[1], fields[2])
1319
return (path_name_file_id_key, [
1321
fields[3], # minikind
1322
fields[4], # fingerprint
1323
_int(fields[5]), # size
1324
fields[6] == 'y', # executable
1325
fields[7], # packed_stat or revision_id
1328
fields[8], # minikind
1329
fields[9], # fingerprint
1330
_int(fields[10]), # size
1331
fields[11] == 'y', # executable
1332
fields[12], # packed_stat or revision_id
1335
fields[13], # minikind
1336
fields[14], # fingerprint
1337
_int(fields[15]), # size
1338
fields[16] == 'y', # executable
1339
fields[17], # packed_stat or revision_id
1342
return fields_to_entry_2_parents
1344
def fields_to_entry_n_parents(fields, _int=int):
1345
path_name_file_id_key = (fields[0], fields[1], fields[2])
1346
trees = [(fields[cur], # minikind
1347
fields[cur+1], # fingerprint
1348
_int(fields[cur+2]), # size
1349
fields[cur+3] == 'y', # executable
1350
fields[cur+4], # stat or revision_id
1351
) for cur in xrange(3, len(fields)-1, 5)]
1352
return path_name_file_id_key, trees
1353
return fields_to_entry_n_parents
1355
def get_parent_ids(self):
1356
"""Return a list of the parent tree ids for the directory state."""
1357
self._read_header_if_needed()
1358
return list(self._parents)
1360
def _get_block_entry_index(self, dirname, basename, tree_index):
1361
"""Get the coordinates for a path in the state structure.
1363
:param dirname: The utf8 dirname to lookup.
1364
:param basename: The utf8 basename to lookup.
1365
:param tree_index: The index of the tree for which this lookup should
1367
:return: A tuple describing where the path is located, or should be
1368
inserted. The tuple contains four fields: the block index, the row
1369
index, the directory is present (boolean), the entire path is
1370
present (boolean). There is no guarantee that either
1371
coordinate is currently reachable unless the found field for it is
1372
True. For instance, a directory not present in the searched tree
1373
may be returned with a value one greater than the current highest
1374
block offset. The directory present field will always be True when
1375
the path present field is True. The directory present field does
1376
NOT indicate that the directory is present in the searched tree,
1377
rather it indicates that there are at least some files in some
1380
self._read_dirblocks_if_needed()
1381
key = dirname, basename, ''
1382
block_index, present = self._find_block_index_from_key(key)
1384
# no such directory - return the dir index and 0 for the row.
1385
return block_index, 0, False, False
1386
block = self._dirblocks[block_index][1] # access the entries only
1387
entry_index, present = self._find_entry_index(key, block)
1388
# linear search through entries at this path to find the one
1390
while entry_index < len(block) and block[entry_index][0][1] == basename:
1391
if block[entry_index][1][tree_index][0] not in \
1392
('a', 'r'): # absent, relocated
1393
return block_index, entry_index, True, True
1395
return block_index, entry_index, True, False
1397
def _get_entry(self, tree_index, fileid_utf8=None, path_utf8=None):
1398
"""Get the dirstate entry for path in tree tree_index.
1400
If either file_id or path is supplied, it is used as the key to lookup.
1401
If both are supplied, the fastest lookup is used, and an error is
1402
raised if they do not both point at the same row.
1404
:param tree_index: The index of the tree we wish to locate this path
1405
in. If the path is present in that tree, the entry containing its
1406
details is returned, otherwise (None, None) is returned
1407
0 is the working tree, higher indexes are successive parent
1409
:param fileid_utf8: A utf8 file_id to look up.
1410
:param path_utf8: An utf8 path to be looked up.
1411
:return: The dirstate entry tuple for path, or (None, None)
1413
self._read_dirblocks_if_needed()
1414
if path_utf8 is not None:
1415
assert path_utf8.__class__ == str, ('path_utf8 is not a str: %s %s'
1416
% (type(path_utf8), path_utf8))
1417
# path lookups are faster
1418
dirname, basename = osutils.split(path_utf8)
1419
block_index, entry_index, dir_present, file_present = \
1420
self._get_block_entry_index(dirname, basename, tree_index)
1421
if not file_present:
1423
entry = self._dirblocks[block_index][1][entry_index]
1424
assert entry[0][2] and entry[1][tree_index][0] not in ('a', 'r'), 'unversioned entry?!?!'
1426
if entry[0][2] != fileid_utf8:
1427
raise errors.BzrError('integrity error ? : mismatching'
1428
' tree_index, file_id and path')
1431
assert fileid_utf8 is not None
1432
possible_keys = self._get_id_index().get(fileid_utf8, None)
1433
if not possible_keys:
1435
for key in possible_keys:
1436
block_index, present = \
1437
self._find_block_index_from_key(key)
1438
# strange, probably indicates an out of date
1439
# id index - for now, allow this.
1442
# WARNING: DO not change this code to use _get_block_entry_index
1443
# as that function is not suitable: it does not use the key
1444
# to lookup, and thus the wrong coordinates are returned.
1445
block = self._dirblocks[block_index][1]
1446
entry_index, present = self._find_entry_index(key, block)
1448
entry = self._dirblocks[block_index][1][entry_index]
1449
if entry[1][tree_index][0] in 'fdlt':
1450
# this is the result we are looking for: the
1451
# real home of this file_id in this tree.
1453
if entry[1][tree_index][0] == 'a':
1454
# there is no home for this entry in this tree
1456
assert entry[1][tree_index][0] == 'r', \
1457
"entry %r has invalid minikind %r for tree %r" \
1459
entry[1][tree_index][0],
1461
real_path = entry[1][tree_index][1]
1462
return self._get_entry(tree_index, fileid_utf8=fileid_utf8,
1463
path_utf8=real_path)
1467
def initialize(cls, path):
1468
"""Create a new dirstate on path.
1470
The new dirstate will be an empty tree - that is it has no parents,
1471
and only a root node - which has id ROOT_ID.
1473
:param path: The name of the file for the dirstate.
1474
:return: A write-locked DirState object.
1476
# This constructs a new DirState object on a path, sets the _state_file
1477
# to a new empty file for that path. It then calls _set_data() with our
1478
# stock empty dirstate information - a root with ROOT_ID, no children,
1479
# and no parents. Finally it calls save() to ensure that this data will
1482
# root dir and root dir contents with no children.
1483
empty_tree_dirblocks = [('', []), ('', [])]
1484
# a new root directory, with a NULLSTAT.
1485
empty_tree_dirblocks[0][1].append(
1486
(('', '', inventory.ROOT_ID), [
1487
('d', '', 0, False, DirState.NULLSTAT),
1491
result._set_data([], empty_tree_dirblocks)
1498
def _inv_entry_to_details(self, inv_entry):
1499
"""Convert an inventory entry (from a revision tree) to state details.
1501
:param inv_entry: An inventory entry whose sha1 and link targets can be
1502
relied upon, and which has a revision set.
1503
:return: A details tuple - the details for a single tree at a path +
1506
kind = inv_entry.kind
1507
minikind = DirState._kind_to_minikind[kind]
1508
tree_data = inv_entry.revision
1509
assert tree_data, 'empty revision for the inv_entry %s.' % \
1511
if kind == 'directory':
1515
elif kind == 'symlink':
1516
fingerprint = inv_entry.symlink_target or ''
1519
elif kind == 'file':
1520
fingerprint = inv_entry.text_sha1 or ''
1521
size = inv_entry.text_size or 0
1522
executable = inv_entry.executable
1523
elif kind == 'tree-reference':
1524
fingerprint = inv_entry.reference_revision or ''
1528
raise Exception("can't pack %s" % inv_entry)
1529
return (minikind, fingerprint, size, executable, tree_data)
1531
def _iter_entries(self):
1532
"""Iterate over all the entries in the dirstate.
1534
Each yelt item is an entry in the standard format described in the
1535
docstring of bzrlib.dirstate.
1537
self._read_dirblocks_if_needed()
1538
for directory in self._dirblocks:
1539
for entry in directory[1]:
1542
def _get_id_index(self):
1543
"""Get an id index of self._dirblocks."""
1544
if self._id_index is None:
1546
for key, tree_details in self._iter_entries():
1547
id_index.setdefault(key[2], set()).add(key)
1548
self._id_index = id_index
1549
return self._id_index
1551
def _get_output_lines(self, lines):
1552
"""Format lines for final output.
1554
:param lines: A sequence of lines containing the parents list and the
1557
output_lines = [DirState.HEADER_FORMAT_3]
1558
lines.append('') # a final newline
1559
inventory_text = '\0\n\0'.join(lines)
1560
output_lines.append('crc32: %s\n' % (zlib.crc32(inventory_text),))
1561
# -3, 1 for num parents, 1 for ghosts, 1 for final newline
1562
num_entries = len(lines)-3
1563
output_lines.append('num_entries: %s\n' % (num_entries,))
1564
output_lines.append(inventory_text)
1567
def _make_deleted_row(self, fileid_utf8, parents):
1568
"""Return a deleted row for fileid_utf8."""
1569
return ('/', 'RECYCLED.BIN', 'file', fileid_utf8, 0, DirState.NULLSTAT,
1572
def _num_present_parents(self):
1573
"""The number of parent entries in each record row."""
1574
return len(self._parents) - len(self._ghosts)
1578
"""Construct a DirState on the file at path path.
1580
:return: An unlocked DirState object, associated with the given path.
1582
result = DirState(path)
1585
def _read_dirblocks_if_needed(self):
1586
"""Read in all the dirblocks from the file if they are not in memory.
1588
This populates self._dirblocks, and sets self._dirblock_state to
1589
IN_MEMORY_UNMODIFIED. It is not currently ready for incremental block
1592
self._read_header_if_needed()
1593
if self._dirblock_state == DirState.NOT_IN_MEMORY:
1594
_read_dirblocks(self)
1596
def _read_header(self):
1597
"""This reads in the metadata header, and the parent ids.
1599
After reading in, the file should be positioned at the null
1600
just before the start of the first record in the file.
1602
:return: (expected crc checksum, number of entries, parent list)
1604
self._read_prelude()
1605
parent_line = self._state_file.readline()
1606
info = parent_line.split('\0')
1607
num_parents = int(info[0])
1608
assert num_parents == len(info)-2, 'incorrect parent info line'
1609
self._parents = info[1:-1]
1611
ghost_line = self._state_file.readline()
1612
info = ghost_line.split('\0')
1613
num_ghosts = int(info[1])
1614
assert num_ghosts == len(info)-3, 'incorrect ghost info line'
1615
self._ghosts = info[2:-1]
1616
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1617
self._end_of_header = self._state_file.tell()
1619
def _read_header_if_needed(self):
1620
"""Read the header of the dirstate file if needed."""
1621
# inline this as it will be called a lot
1622
if not self._lock_token:
1623
raise errors.ObjectNotLocked(self)
1624
if self._header_state == DirState.NOT_IN_MEMORY:
1627
def _read_prelude(self):
1628
"""Read in the prelude header of the dirstate file.
1630
This only reads in the stuff that is not connected to the crc
1631
checksum. The position will be correct to read in the rest of
1632
the file and check the checksum after this point.
1633
The next entry in the file should be the number of parents,
1634
and their ids. Followed by a newline.
1636
header = self._state_file.readline()
1637
assert header == DirState.HEADER_FORMAT_3, \
1638
'invalid header line: %r' % (header,)
1639
crc_line = self._state_file.readline()
1640
assert crc_line.startswith('crc32: '), 'missing crc32 checksum'
1641
self.crc_expected = int(crc_line[len('crc32: '):-1])
1642
num_entries_line = self._state_file.readline()
1643
assert num_entries_line.startswith('num_entries: '), 'missing num_entries line'
1644
self._num_entries = int(num_entries_line[len('num_entries: '):-1])
1646
def sha1_from_stat(self, path, stat_result, _pack_stat=pack_stat):
1647
"""Find a sha1 given a stat lookup."""
1648
return self._get_packed_stat_index().get(_pack_stat(stat_result), None)
1650
def _get_packed_stat_index(self):
1651
"""Get a packed_stat index of self._dirblocks."""
1652
if self._packed_stat_index is None:
1654
for key, tree_details in self._iter_entries():
1655
if tree_details[0][0] == 'f':
1656
index[tree_details[0][4]] = tree_details[0][1]
1657
self._packed_stat_index = index
1658
return self._packed_stat_index
1661
"""Save any pending changes created during this session.
1663
We reuse the existing file, because that prevents race conditions with
1664
file creation, and use oslocks on it to prevent concurrent modification
1665
and reads - because dirstate's incremental data aggregation is not
1666
compatible with reading a modified file, and replacing a file in use by
1667
another process is impossible on Windows.
1669
A dirstate in read only mode should be smart enough though to validate
1670
that the file has not changed, and otherwise discard its cache and
1671
start over, to allow for fine grained read lock duration, so 'status'
1672
wont block 'commit' - for example.
1674
if (self._header_state == DirState.IN_MEMORY_MODIFIED or
1675
self._dirblock_state == DirState.IN_MEMORY_MODIFIED):
1677
grabbed_write_lock = False
1678
if self._lock_state != 'w':
1679
grabbed_write_lock, new_lock = self._lock_token.temporary_write_lock()
1680
# Switch over to the new lock, as the old one may be closed.
1681
# TODO: jam 20070315 We should validate the disk file has
1682
# not changed contents. Since temporary_write_lock may
1683
# not be an atomic operation.
1684
self._lock_token = new_lock
1685
self._state_file = new_lock.f
1686
if not grabbed_write_lock:
1687
# We couldn't grab a write lock, so we switch back to a read one
1690
self._state_file.seek(0)
1691
self._state_file.writelines(self.get_lines())
1692
self._state_file.truncate()
1693
self._state_file.flush()
1694
self._header_state = DirState.IN_MEMORY_UNMODIFIED
1695
self._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
1697
if grabbed_write_lock:
1698
self._lock_token = self._lock_token.restore_read_lock()
1699
self._state_file = self._lock_token.f
1700
# TODO: jam 20070315 We should validate the disk file has
1701
# not changed contents. Since restore_read_lock may
1702
# not be an atomic operation.
1704
def _set_data(self, parent_ids, dirblocks):
1705
"""Set the full dirstate data in memory.
1707
This is an internal function used to completely replace the objects
1708
in memory state. It puts the dirstate into state 'full-dirty'.
1710
:param parent_ids: A list of parent tree revision ids.
1711
:param dirblocks: A list containing one tuple for each directory in the
1712
tree. Each tuple contains the directory path and a list of entries
1713
found in that directory.
1715
# our memory copy is now authoritative.
1716
self._dirblocks = dirblocks
1717
self._header_state = DirState.IN_MEMORY_MODIFIED
1718
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1719
self._parents = list(parent_ids)
1720
self._id_index = None
1721
self._packed_stat_index = None
1723
def set_path_id(self, path, new_id):
1724
"""Change the id of path to new_id in the current working tree.
1726
:param path: The path inside the tree to set - '' is the root, 'foo'
1727
is the path foo in the root.
1728
:param new_id: The new id to assign to the path. This must be a utf8
1729
file id (not unicode, and not None).
1731
assert new_id.__class__ == str, \
1732
"path_id %r is not a plain string" % (new_id,)
1733
self._read_dirblocks_if_needed()
1735
# TODO: logic not written
1736
raise NotImplementedError(self.set_path_id)
1737
# TODO: check new id is unique
1738
entry = self._get_entry(0, path_utf8=path)
1739
if entry[0][2] == new_id:
1740
# Nothing to change.
1742
# mark the old path absent, and insert a new root path
1743
self._make_absent(entry)
1744
self.update_minimal(('', '', new_id), 'd',
1745
path_utf8='', packed_stat=entry[1][0][4])
1746
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1747
if self._id_index is not None:
1748
self._id_index.setdefault(new_id, set()).add(entry[0])
1750
def set_parent_trees(self, trees, ghosts):
1751
"""Set the parent trees for the dirstate.
1753
:param trees: A list of revision_id, tree tuples. tree must be provided
1754
even if the revision_id refers to a ghost: supply an empty tree in
1756
:param ghosts: A list of the revision_ids that are ghosts at the time
1759
# TODO: generate a list of parent indexes to preserve to save
1760
# processing specific parent trees. In the common case one tree will
1761
# be preserved - the left most parent.
1762
# TODO: if the parent tree is a dirstate, we might want to walk them
1763
# all by path in parallel for 'optimal' common-case performance.
1764
# generate new root row.
1765
self._read_dirblocks_if_needed()
1766
# TODO future sketch: Examine the existing parents to generate a change
1767
# map and then walk the new parent trees only, mapping them into the
1768
# dirstate. Walk the dirstate at the same time to remove unreferenced
1771
# sketch: loop over all entries in the dirstate, cherry picking
1772
# entries from the parent trees, if they are not ghost trees.
1773
# after we finish walking the dirstate, all entries not in the dirstate
1774
# are deletes, so we want to append them to the end as per the design
1775
# discussions. So do a set difference on ids with the parents to
1776
# get deletes, and add them to the end.
1777
# During the update process we need to answer the following questions:
1778
# - find other keys containing a fileid in order to create cross-path
1779
# links. We dont't trivially use the inventory from other trees
1780
# because this leads to either double touching, or to accessing
1782
# - find other keys containing a path
1783
# We accumulate each entry via this dictionary, including the root
1786
# we could do parallel iterators, but because file id data may be
1787
# scattered throughout, we dont save on index overhead: we have to look
1788
# at everything anyway. We can probably save cycles by reusing parent
1789
# data and doing an incremental update when adding an additional
1790
# parent, but for now the common cases are adding a new parent (merge),
1791
# and replacing completely (commit), and commit is more common: so
1792
# optimise merge later.
1794
# ---- start generation of full tree mapping data
1795
# what trees should we use?
1796
parent_trees = [tree for rev_id, tree in trees if rev_id not in ghosts]
1797
# how many trees do we end up with
1798
parent_count = len(parent_trees)
1800
# one: the current tree
1801
for entry in self._iter_entries():
1802
# skip entries not in the current tree
1803
if entry[1][0][0] in ('a', 'r'): # absent, relocated
1805
by_path[entry[0]] = [entry[1][0]] + \
1806
[DirState.NULL_PARENT_DETAILS] * parent_count
1807
id_index[entry[0][2]] = set([entry[0]])
1809
# now the parent trees:
1810
for tree_index, tree in enumerate(parent_trees):
1811
# the index is off by one, adjust it.
1812
tree_index = tree_index + 1
1813
# when we add new locations for a fileid we need these ranges for
1814
# any fileid in this tree as we set the by_path[id] to:
1815
# already_processed_tree_details + new_details + new_location_suffix
1816
# the suffix is from tree_index+1:parent_count+1.
1817
new_location_suffix = [DirState.NULL_PARENT_DETAILS] * (parent_count - tree_index)
1818
# now stitch in all the entries from this tree
1819
for path, entry in tree.inventory.iter_entries_by_dir():
1820
# here we process each trees details for each item in the tree.
1821
# we first update any existing entries for the id at other paths,
1822
# then we either create or update the entry for the id at the
1823
# right path, and finally we add (if needed) a mapping from
1824
# file_id to this path. We do it in this order to allow us to
1825
# avoid checking all known paths for the id when generating a
1826
# new entry at this path: by adding the id->path mapping last,
1827
# all the mappings are valid and have correct relocation
1828
# records where needed.
1829
file_id = entry.file_id
1830
path_utf8 = path.encode('utf8')
1831
dirname, basename = osutils.split(path_utf8)
1832
new_entry_key = (dirname, basename, file_id)
1833
# tree index consistency: All other paths for this id in this tree
1834
# index must point to the correct path.
1835
for entry_key in id_index.setdefault(file_id, set()):
1836
# TODO:PROFILING: It might be faster to just update
1837
# rather than checking if we need to, and then overwrite
1838
# the one we are located at.
1839
if entry_key != new_entry_key:
1840
# this file id is at a different path in one of the
1841
# other trees, so put absent pointers there
1842
# This is the vertical axis in the matrix, all pointing
1844
by_path[entry_key][tree_index] = ('r', path_utf8, 0, False, '')
1845
# by path consistency: Insert into an existing path record (trivial), or
1846
# add a new one with relocation pointers for the other tree indexes.
1847
if new_entry_key in id_index[file_id]:
1848
# there is already an entry where this data belongs, just insert it.
1849
by_path[new_entry_key][tree_index] = \
1850
self._inv_entry_to_details(entry)
1852
# add relocated entries to the horizontal axis - this row
1853
# mapping from path,id. We need to look up the correct path
1854
# for the indexes from 0 to tree_index -1
1856
for lookup_index in xrange(tree_index):
1857
# boundary case: this is the first occurence of file_id
1858
# so there are no id_indexs, possibly take this out of
1860
if not len(id_index[file_id]):
1861
new_details.append(DirState.NULL_PARENT_DETAILS)
1863
# grab any one entry, use it to find the right path.
1864
# TODO: optimise this to reduce memory use in highly
1865
# fragmented situations by reusing the relocation
1867
a_key = iter(id_index[file_id]).next()
1868
if by_path[a_key][lookup_index][0] in ('r', 'a'):
1869
# its a pointer or missing statement, use it as is.
1870
new_details.append(by_path[a_key][lookup_index])
1872
# we have the right key, make a pointer to it.
1873
real_path = ('/'.join(a_key[0:2])).strip('/')
1874
new_details.append(('r', real_path, 0, False, ''))
1875
new_details.append(self._inv_entry_to_details(entry))
1876
new_details.extend(new_location_suffix)
1877
by_path[new_entry_key] = new_details
1878
id_index[file_id].add(new_entry_key)
1879
# --- end generation of full tree mappings
1881
# sort and output all the entries
1882
new_entries = self._sort_entries(by_path.items())
1883
self._entries_to_current_state(new_entries)
1884
self._parents = [rev_id for rev_id, tree in trees]
1885
self._ghosts = list(ghosts)
1886
self._header_state = DirState.IN_MEMORY_MODIFIED
1887
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
1888
self._id_index = id_index
1890
def _sort_entries(self, entry_list):
1891
"""Given a list of entries, sort them into the right order.
1893
This is done when constructing a new dirstate from trees - normally we
1894
try to keep everything in sorted blocks all the time, but sometimes
1895
it's easier to sort after the fact.
1898
# sort by: directory parts, file name, file id
1899
return entry[0][0].split('/'), entry[0][1], entry[0][2]
1900
return sorted(entry_list, key=_key)
1902
def set_state_from_inventory(self, new_inv):
1903
"""Set new_inv as the current state.
1905
This API is called by tree transform, and will usually occur with
1906
existing parent trees.
1908
:param new_inv: The inventory object to set current state from.
1910
if 'evil' in debug.debug_flags:
1911
trace.mutter_callsite(1,
1912
"set_state_from_inventory called; please mutate the tree instead")
1913
self._read_dirblocks_if_needed()
1915
# Two iterators: current data and new data, both in dirblock order.
1916
# We zip them together, which tells about entries that are new in the
1917
# inventory, or removed in the inventory, or present in both and
1920
# You might think we could just synthesize a new dirstate directly
1921
# since we're processing it in the right order. However, we need to
1922
# also consider there may be any number of parent trees and relocation
1923
# pointers, and we don't want to duplicate that here.
1924
new_iterator = new_inv.iter_entries_by_dir()
1925
# we will be modifying the dirstate, so we need a stable iterator. In
1926
# future we might write one, for now we just clone the state into a
1927
# list - which is a shallow copy.
1928
old_iterator = iter(list(self._iter_entries()))
1929
# both must have roots so this is safe:
1930
current_new = new_iterator.next()
1931
current_old = old_iterator.next()
1932
def advance(iterator):
1934
return iterator.next()
1935
except StopIteration:
1937
while current_new or current_old:
1938
# skip entries in old that are not really there
1939
if current_old and current_old[1][0][0] in ('r', 'a'):
1940
# relocated or absent
1941
current_old = advance(old_iterator)
1944
# convert new into dirblock style
1945
new_path_utf8 = current_new[0].encode('utf8')
1946
new_dirname, new_basename = osutils.split(new_path_utf8)
1947
new_id = current_new[1].file_id
1948
new_entry_key = (new_dirname, new_basename, new_id)
1949
current_new_minikind = \
1950
DirState._kind_to_minikind[current_new[1].kind]
1951
if current_new_minikind == 't':
1952
fingerprint = current_new[1].reference_revision or ''
1954
# We normally only insert or remove records, or update
1955
# them when it has significantly changed. Then we want to
1956
# erase its fingerprint. Unaffected records should
1957
# normally not be updated at all.
1960
# for safety disable variables
1961
new_path_utf8 = new_dirname = new_basename = new_id = \
1962
new_entry_key = None
1963
# 5 cases, we dont have a value that is strictly greater than everything, so
1964
# we make both end conditions explicit
1966
# old is finished: insert current_new into the state.
1967
self.update_minimal(new_entry_key, current_new_minikind,
1968
executable=current_new[1].executable,
1969
path_utf8=new_path_utf8, fingerprint=fingerprint)
1970
current_new = advance(new_iterator)
1971
elif not current_new:
1973
self._make_absent(current_old)
1974
current_old = advance(old_iterator)
1975
elif new_entry_key == current_old[0]:
1976
# same - common case
1977
# We're looking at the same path and id in both the dirstate
1978
# and inventory, so just need to update the fields in the
1979
# dirstate from the one in the inventory.
1980
# TODO: update the record if anything significant has changed.
1981
# the minimal required trigger is if the execute bit or cached
1983
if (current_old[1][0][3] != current_new[1].executable or
1984
current_old[1][0][0] != current_new_minikind):
1985
self.update_minimal(current_old[0], current_new_minikind,
1986
executable=current_new[1].executable,
1987
path_utf8=new_path_utf8, fingerprint=fingerprint)
1988
# both sides are dealt with, move on
1989
current_old = advance(old_iterator)
1990
current_new = advance(new_iterator)
1991
elif (cmp_by_dirs(new_dirname, current_old[0][0]) < 0
1992
or (new_dirname == current_old[0][0]
1993
and new_entry_key[1:] < current_old[0][1:])):
1995
# add a entry for this and advance new
1996
self.update_minimal(new_entry_key, current_new_minikind,
1997
executable=current_new[1].executable,
1998
path_utf8=new_path_utf8, fingerprint=fingerprint)
1999
current_new = advance(new_iterator)
2001
# we've advanced past the place where the old key would be,
2002
# without seeing it in the new list. so it must be gone.
2003
self._make_absent(current_old)
2004
current_old = advance(old_iterator)
2005
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2006
self._id_index = None
2007
self._packed_stat_index = None
2009
def _make_absent(self, current_old):
2010
"""Mark current_old - an entry - as absent for tree 0.
2012
:return: True if this was the last details entry for the entry key:
2013
that is, if the underlying block has had the entry removed, thus
2014
shrinking in length.
2016
# build up paths that this id will be left at after the change is made,
2017
# so we can update their cross references in tree 0
2018
all_remaining_keys = set()
2019
# Dont check the working tree, because it's going.
2020
for details in current_old[1][1:]:
2021
if details[0] not in ('a', 'r'): # absent, relocated
2022
all_remaining_keys.add(current_old[0])
2023
elif details[0] == 'r': # relocated
2024
# record the key for the real path.
2025
all_remaining_keys.add(tuple(osutils.split(details[1])) + (current_old[0][2],))
2026
# absent rows are not present at any path.
2027
last_reference = current_old[0] not in all_remaining_keys
2029
# the current row consists entire of the current item (being marked
2030
# absent), and relocated or absent entries for the other trees:
2031
# Remove it, its meaningless.
2032
block = self._find_block(current_old[0])
2033
entry_index, present = self._find_entry_index(current_old[0], block[1])
2034
assert present, 'could not find entry for %s' % (current_old,)
2035
block[1].pop(entry_index)
2036
# if we have an id_index in use, remove this key from it for this id.
2037
if self._id_index is not None:
2038
self._id_index[current_old[0][2]].remove(current_old[0])
2039
# update all remaining keys for this id to record it as absent. The
2040
# existing details may either be the record we are marking as deleted
2041
# (if there were other trees with the id present at this path), or may
2043
for update_key in all_remaining_keys:
2044
update_block_index, present = \
2045
self._find_block_index_from_key(update_key)
2046
assert present, 'could not find block for %s' % (update_key,)
2047
update_entry_index, present = \
2048
self._find_entry_index(update_key, self._dirblocks[update_block_index][1])
2049
assert present, 'could not find entry for %s' % (update_key,)
2050
update_tree_details = self._dirblocks[update_block_index][1][update_entry_index][1]
2051
# it must not be absent at the moment
2052
assert update_tree_details[0][0] != 'a' # absent
2053
update_tree_details[0] = DirState.NULL_PARENT_DETAILS
2054
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2055
return last_reference
2057
def update_minimal(self, key, minikind, executable=False, fingerprint='',
2058
packed_stat=None, size=0, path_utf8=None):
2059
"""Update an entry to the state in tree 0.
2061
This will either create a new entry at 'key' or update an existing one.
2062
It also makes sure that any other records which might mention this are
2065
:param key: (dir, name, file_id) for the new entry
2066
:param minikind: The type for the entry ('f' == 'file', 'd' ==
2068
:param executable: Should the executable bit be set?
2069
:param fingerprint: Simple fingerprint for new entry: sha1 for files,
2070
referenced revision id for subtrees, etc.
2071
:param packed_stat: Packed stat value for new entry.
2072
:param size: Size information for new entry
2073
:param path_utf8: key[0] + '/' + key[1], just passed in to avoid doing
2076
If packed_stat and fingerprint are not given, they're invalidated in
2079
block = self._find_block(key)[1]
2080
if packed_stat is None:
2081
packed_stat = DirState.NULLSTAT
2082
# XXX: Some callers pass '' as the packed_stat, and it seems to be
2083
# sometimes present in the dirstate - this seems oddly inconsistent.
2085
entry_index, present = self._find_entry_index(key, block)
2086
new_details = (minikind, fingerprint, size, executable, packed_stat)
2087
id_index = self._get_id_index()
2089
# new entry, synthesis cross reference here,
2090
existing_keys = id_index.setdefault(key[2], set())
2091
if not existing_keys:
2092
# not currently in the state, simplest case
2093
new_entry = key, [new_details] + self._empty_parent_info()
2095
# present at one or more existing other paths.
2096
# grab one of them and use it to generate parent
2097
# relocation/absent entries.
2098
new_entry = key, [new_details]
2099
for other_key in existing_keys:
2100
# change the record at other to be a pointer to this new
2101
# record. The loop looks similar to the change to
2102
# relocations when updating an existing record but its not:
2103
# the test for existing kinds is different: this can be
2104
# factored out to a helper though.
2105
other_block_index, present = self._find_block_index_from_key(other_key)
2106
assert present, 'could not find block for %s' % (other_key,)
2107
other_entry_index, present = self._find_entry_index(other_key,
2108
self._dirblocks[other_block_index][1])
2109
assert present, 'could not find entry for %s' % (other_key,)
2110
assert path_utf8 is not None
2111
self._dirblocks[other_block_index][1][other_entry_index][1][0] = \
2112
('r', path_utf8, 0, False, '')
2114
num_present_parents = self._num_present_parents()
2115
for lookup_index in xrange(1, num_present_parents + 1):
2116
# grab any one entry, use it to find the right path.
2117
# TODO: optimise this to reduce memory use in highly
2118
# fragmented situations by reusing the relocation
2120
update_block_index, present = \
2121
self._find_block_index_from_key(other_key)
2122
assert present, 'could not find block for %s' % (other_key,)
2123
update_entry_index, present = \
2124
self._find_entry_index(other_key, self._dirblocks[update_block_index][1])
2125
assert present, 'could not find entry for %s' % (other_key,)
2126
update_details = self._dirblocks[update_block_index][1][update_entry_index][1][lookup_index]
2127
if update_details[0] in ('r', 'a'): # relocated, absent
2128
# its a pointer or absent in lookup_index's tree, use
2130
new_entry[1].append(update_details)
2132
# we have the right key, make a pointer to it.
2133
pointer_path = osutils.pathjoin(*other_key[0:2])
2134
new_entry[1].append(('r', pointer_path, 0, False, ''))
2135
block.insert(entry_index, new_entry)
2136
existing_keys.add(key)
2138
# Does the new state matter?
2139
block[entry_index][1][0] = new_details
2140
# parents cannot be affected by what we do.
2141
# other occurences of this id can be found
2142
# from the id index.
2144
# tree index consistency: All other paths for this id in this tree
2145
# index must point to the correct path. We have to loop here because
2146
# we may have passed entries in the state with this file id already
2147
# that were absent - where parent entries are - and they need to be
2148
# converted to relocated.
2149
assert path_utf8 is not None
2150
for entry_key in id_index.setdefault(key[2], set()):
2151
# TODO:PROFILING: It might be faster to just update
2152
# rather than checking if we need to, and then overwrite
2153
# the one we are located at.
2154
if entry_key != key:
2155
# this file id is at a different path in one of the
2156
# other trees, so put absent pointers there
2157
# This is the vertical axis in the matrix, all pointing
2159
block_index, present = self._find_block_index_from_key(entry_key)
2161
entry_index, present = self._find_entry_index(entry_key, self._dirblocks[block_index][1])
2163
self._dirblocks[block_index][1][entry_index][1][0] = \
2164
('r', path_utf8, 0, False, '')
2165
# add a containing dirblock if needed.
2166
if new_details[0] == 'd':
2167
subdir_key = (osutils.pathjoin(*key[0:2]), '', '')
2168
block_index, present = self._find_block_index_from_key(subdir_key)
2170
self._dirblocks.insert(block_index, (subdir_key[0], []))
2172
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
2174
def _validate(self):
2175
"""Check that invariants on the dirblock are correct.
2177
This can be useful in debugging; it shouldn't be necessary in
2180
This must be called with a lock held.
2182
# NOTE: This must always raise AssertionError not just assert,
2183
# otherwise it may not behave properly under python -O
2185
# TODO: All entries must have some content that's not 'a' or 'r',
2186
# otherwise it could just be removed.
2188
# TODO: All relocations must point directly to a real entry.
2190
# TODO: No repeated keys.
2193
from pprint import pformat
2194
self._read_dirblocks_if_needed()
2195
if len(self._dirblocks) > 0:
2196
if not self._dirblocks[0][0] == '':
2197
raise AssertionError(
2198
"dirblocks don't start with root block:\n" + \
2200
if len(self._dirblocks) > 1:
2201
if not self._dirblocks[1][0] == '':
2202
raise AssertionError(
2203
"dirblocks missing root directory:\n" + \
2205
# the dirblocks are sorted by their path components, name, and dir id
2206
dir_names = [d[0].split('/')
2207
for d in self._dirblocks[1:]]
2208
if dir_names != sorted(dir_names):
2209
raise AssertionError(
2210
"dir names are not in sorted order:\n" + \
2211
pformat(self._dirblocks) + \
2214
for dirblock in self._dirblocks:
2215
# within each dirblock, the entries are sorted by filename and
2217
for entry in dirblock[1]:
2218
if dirblock[0] != entry[0][0]:
2219
raise AssertionError(
2221
"doesn't match directory name in\n%r" %
2222
(entry, pformat(dirblock)))
2223
if dirblock[1] != sorted(dirblock[1]):
2224
raise AssertionError(
2225
"dirblock for %r is not sorted:\n%s" % \
2226
(dirblock[0], pformat(dirblock)))
2229
def check_valid_parent():
2230
"""Check that the current entry has a valid parent.
2232
This makes sure that the parent has a record,
2233
and that the parent isn't marked as "absent" in the
2234
current tree. (It is invalid to have a non-absent file in an absent
2237
if entry[0][0:2] == ('', ''):
2238
# There should be no parent for the root row
2240
parent_entry = self._get_entry(tree_index, path_utf8=entry[0][0])
2241
if parent_entry == (None, None):
2242
raise AssertionError(
2243
"no parent entry for: %s in tree %s"
2244
% (this_path, tree_index))
2245
if parent_entry[1][tree_index][0] != 'd':
2246
raise AssertionError(
2247
"Parent entry for %s is not marked as a valid"
2248
" directory. %s" % (this_path, parent_entry,))
2250
# For each file id, for each tree: either
2251
# the file id is not present at all; all rows with that id in the
2252
# key have it marked as 'absent'
2253
# OR the file id is present under exactly one name; any other entries
2254
# that mention that id point to the correct name.
2256
# We check this with a dict per tree pointing either to the present
2257
# name, or None if absent.
2258
tree_count = self._num_present_parents() + 1
2259
id_path_maps = [dict() for i in range(tree_count)]
2260
# Make sure that all renamed entries point to the correct location.
2261
for entry in self._iter_entries():
2262
file_id = entry[0][2]
2263
this_path = osutils.pathjoin(entry[0][0], entry[0][1])
2264
if len(entry[1]) != tree_count:
2265
raise AssertionError(
2266
"wrong number of entry details for row\n%s" \
2267
",\nexpected %d" % \
2268
(pformat(entry), tree_count))
2269
for tree_index, tree_state in enumerate(entry[1]):
2270
this_tree_map = id_path_maps[tree_index]
2271
minikind = tree_state[0]
2272
# have we seen this id before in this column?
2273
if file_id in this_tree_map:
2274
previous_path = this_tree_map[file_id]
2275
# any later mention of this file must be consistent with
2276
# what was said before
2278
if previous_path is not None:
2279
raise AssertionError(
2280
"file %s is absent in row %r but also present " \
2282
(file_id, entry, previous_path))
2283
elif minikind == 'r':
2284
target_location = tree_state[1]
2285
if previous_path != target_location:
2286
raise AssertionError(
2287
"file %s relocation in row %r but also at %r" \
2288
% (file_id, entry, previous_path))
2290
# a file, directory, etc - may have been previously
2291
# pointed to by a relocation, which must point here
2292
if previous_path != this_path:
2293
raise AssertionError(
2294
"entry %r inconsistent with previous path %r" % \
2295
(entry, previous_path))
2296
check_valid_parent()
2299
# absent; should not occur anywhere else
2300
this_tree_map[file_id] = None
2301
elif minikind == 'r':
2302
# relocation, must occur at expected location
2303
this_tree_map[file_id] = tree_state[1]
2305
this_tree_map[file_id] = this_path
2306
check_valid_parent()
2308
def _wipe_state(self):
2309
"""Forget all state information about the dirstate."""
2310
self._header_state = DirState.NOT_IN_MEMORY
2311
self._dirblock_state = DirState.NOT_IN_MEMORY
2314
self._dirblocks = []
2315
self._id_index = None
2316
self._packed_stat_index = None
2317
self._end_of_header = None
2318
self._cutoff_time = None
2319
self._split_path_cache = {}
2321
def lock_read(self):
2322
"""Acquire a read lock on the dirstate."""
2323
if self._lock_token is not None:
2324
raise errors.LockContention(self._lock_token)
2325
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2326
# already in memory, we could read just the header and check for
2327
# any modification. If not modified, we can just leave things
2329
self._lock_token = lock.ReadLock(self._filename)
2330
self._lock_state = 'r'
2331
self._state_file = self._lock_token.f
2334
def lock_write(self):
2335
"""Acquire a write lock on the dirstate."""
2336
if self._lock_token is not None:
2337
raise errors.LockContention(self._lock_token)
2338
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2339
# already in memory, we could read just the header and check for
2340
# any modification. If not modified, we can just leave things
2342
self._lock_token = lock.WriteLock(self._filename)
2343
self._lock_state = 'w'
2344
self._state_file = self._lock_token.f
2348
"""Drop any locks held on the dirstate."""
2349
if self._lock_token is None:
2350
raise errors.LockNotHeld(self)
2351
# TODO: jam 20070301 Rather than wiping completely, if the blocks are
2352
# already in memory, we could read just the header and check for
2353
# any modification. If not modified, we can just leave things
2355
self._state_file = None
2356
self._lock_state = None
2357
self._lock_token.unlock()
2358
self._lock_token = None
2359
self._split_path_cache = {}
2361
def _requires_lock(self):
2362
"""Check that a lock is currently held by someone on the dirstate."""
2363
if not self._lock_token:
2364
raise errors.ObjectNotLocked(self)
2367
# Try to load the compiled form if possible
2369
from bzrlib._dirstate_helpers_c import (
2370
_read_dirblocks_c as _read_dirblocks,
2371
bisect_dirblock_c as bisect_dirblock,
2372
_bisect_path_left_c as _bisect_path_left,
2373
_bisect_path_right_c as _bisect_path_right,
2374
cmp_by_dirs_c as cmp_by_dirs,
2377
from bzrlib._dirstate_helpers_py import (
2378
_read_dirblocks_py as _read_dirblocks,
2379
bisect_dirblock_py as bisect_dirblock,
2380
_bisect_path_left_py as _bisect_path_left,
2381
_bisect_path_right_py as _bisect_path_right,
2382
cmp_by_dirs_py as cmp_by_dirs,
added
removed
bzrlib/dirstate.py
