1
# Copyright (C) 2005 Canonical Ltd
1
# Copyright (C) 2005-2011 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
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
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
17
"""Tree classes, representing directory at point in time.
21
from cStringIO import StringIO
24
from bzrlib.trace import mutter, note
25
from bzrlib.errors import BzrError, BzrCheckError
26
from bzrlib.inventory import Inventory
27
from bzrlib.osutils import appendpath, fingerprint_file
22
from bzrlib.lazy_import import lazy_import
23
lazy_import(globals(), """
27
conflicts as _mod_conflicts,
34
revision as _mod_revision,
40
from bzrlib.decorators import needs_read_lock
41
from bzrlib.inter import InterObject
42
from bzrlib.symbol_versioning import (
29
48
class Tree(object):
30
49
"""Abstract file tree.
32
51
There are several subclasses:
34
53
* `WorkingTree` exists as files on disk editable by the user.
36
55
* `RevisionTree` is a tree as recorded at some point in the past.
40
Trees contain an `Inventory` object, and also know how to retrieve
41
file texts mentioned in the inventory, either from a working
42
directory or from a store.
44
It is possible for trees to contain files that are not described
45
in their inventory or vice versa; for this use `filenames()`.
47
57
Trees can be compared, etc, regardless of whether they are working
48
58
trees or versioned trees.
61
def changes_from(self, other, want_unchanged=False, specific_files=None,
62
extra_trees=None, require_versioned=False, include_root=False,
63
want_unversioned=False):
64
"""Return a TreeDelta of the changes from other to this tree.
66
:param other: A tree to compare with.
67
:param specific_files: An optional list of file paths to restrict the
68
comparison to. When mapping filenames to ids, all matches in all
69
trees (including optional extra_trees) are used, and all children of
70
matched directories are included.
71
:param want_unchanged: An optional boolean requesting the inclusion of
72
unchanged entries in the result.
73
:param extra_trees: An optional list of additional trees to use when
74
mapping the contents of specific_files (paths) to file_ids.
75
:param require_versioned: An optional boolean (defaults to False). When
76
supplied and True all the 'specific_files' must be versioned, or
77
a PathsNotVersionedError will be thrown.
78
:param want_unversioned: Scan for unversioned paths.
80
The comparison will be performed by an InterTree object looked up on
83
# Martin observes that Tree.changes_from returns a TreeDelta and this
84
# may confuse people, because the class name of the returned object is
85
# a synonym of the object referenced in the method name.
86
return InterTree.get(other, self).compare(
87
want_unchanged=want_unchanged,
88
specific_files=specific_files,
89
extra_trees=extra_trees,
90
require_versioned=require_versioned,
91
include_root=include_root,
92
want_unversioned=want_unversioned,
95
def iter_changes(self, from_tree, include_unchanged=False,
96
specific_files=None, pb=None, extra_trees=None,
97
require_versioned=True, want_unversioned=False):
98
"""See InterTree.iter_changes"""
99
intertree = InterTree.get(from_tree, self)
100
return intertree.iter_changes(include_unchanged, specific_files, pb,
101
extra_trees, require_versioned, want_unversioned=want_unversioned)
104
"""Get a list of the conflicts in the tree.
106
Each conflict is an instance of bzrlib.conflicts.Conflict.
108
return _mod_conflicts.ConflictList()
111
"""For trees that can have unversioned files, return all such paths."""
114
def get_parent_ids(self):
115
"""Get the parent ids for this tree.
117
:return: a list of parent ids. [] is returned to indicate
118
a tree with no parents.
119
:raises: BzrError if the parents are not known.
121
raise NotImplementedError(self.get_parent_ids)
51
123
def has_filename(self, filename):
52
124
"""True if the tree has given filename."""
53
raise NotImplementedError()
125
raise NotImplementedError(self.has_filename)
55
127
def has_id(self, file_id):
56
return self.inventory.has_id(file_id)
128
raise NotImplementedError(self.has_id)
130
def __contains__(self, file_id):
131
return self.has_id(file_id)
58
133
def has_or_had_id(self, file_id):
59
if file_id == self.inventory.root.file_id:
61
return self.inventory.has_id(file_id)
66
return iter(self.inventory)
134
raise NotImplementedError(self.has_or_had_id)
136
def is_ignored(self, filename):
137
"""Check whether the filename is ignored by this tree.
139
:param filename: The relative filename within the tree.
140
:return: True if the filename is ignored.
144
def all_file_ids(self):
145
"""Iterate through all file ids, including ids for missing files."""
146
raise NotImplementedError(self.all_file_ids)
68
148
def id2path(self, file_id):
69
return self.inventory.id2path(file_id)
149
"""Return the path for a file id.
153
raise NotImplementedError(self.id2path)
155
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
156
"""Walk the tree in 'by_dir' order.
158
This will yield each entry in the tree as a (path, entry) tuple.
159
The order that they are yielded is:
161
Directories are walked in a depth-first lexicographical order,
162
however, whenever a directory is reached, all of its direct child
163
nodes are yielded in lexicographical order before yielding the
166
For example, in the tree::
176
The yield order (ignoring root) would be::
178
a, f, a/b, a/d, a/b/c, a/d/e, f/g
180
:param yield_parents: If True, yield the parents from the root leading
181
down to specific_file_ids that have been requested. This has no
182
impact if specific_file_ids is None.
184
raise NotImplementedError(self.iter_entries_by_dir)
186
def list_files(self, include_root=False, from_dir=None, recursive=True):
187
"""List all files in this tree.
189
:param include_root: Whether to include the entry for the tree root
190
:param from_dir: Directory under which to list files
191
:param recursive: Whether to list files recursively
192
:return: iterator over tuples of (path, versioned, kind, file_id,
195
raise NotImplementedError(self.list_files)
197
def iter_references(self):
198
if self.supports_tree_reference():
199
for path, entry in self.iter_entries_by_dir():
200
if entry.kind == 'tree-reference':
201
yield path, entry.file_id
71
203
def kind(self, file_id):
72
raise NotImplementedError("subclasses must implement kind")
74
def _get_inventory(self):
75
return self._inventory
204
raise NotImplementedError("Tree subclass %s must implement kind"
205
% self.__class__.__name__)
207
def stored_kind(self, file_id):
208
"""File kind stored for this file_id.
210
May not match kind on disk for working trees. Always available
211
for versioned files, even when the file itself is missing.
213
return self.kind(file_id)
215
def path_content_summary(self, path):
216
"""Get a summary of the information about path.
218
All the attributes returned are for the canonical form, not the
219
convenient form (if content filters are in use.)
221
:param path: A relative path within the tree.
222
:return: A tuple containing kind, size, exec, sha1-or-link.
223
Kind is always present (see tree.kind()).
224
size is present if kind is file and the size of the
225
canonical form can be cheaply determined, None otherwise.
226
exec is None unless kind is file and the platform supports the 'x'
228
sha1-or-link is the link target if kind is symlink, or the sha1 if
229
it can be obtained without reading the file.
231
raise NotImplementedError(self.path_content_summary)
233
def get_reference_revision(self, file_id, path=None):
234
raise NotImplementedError("Tree subclass %s must implement "
235
"get_reference_revision"
236
% self.__class__.__name__)
238
def _comparison_data(self, entry, path):
239
"""Return a tuple of kind, executable, stat_value for a file.
241
entry may be None if there is no inventory entry for the file, but
242
path must always be supplied.
244
kind is None if there is no file present (even if an inventory id is
245
present). executable is False for non-file entries.
247
raise NotImplementedError(self._comparison_data)
249
def _file_size(self, entry, stat_value):
250
raise NotImplementedError(self._file_size)
252
def get_file(self, file_id, path=None):
253
"""Return a file object for the file file_id in the tree.
255
If both file_id and path are defined, it is implementation defined as
256
to which one is used.
258
raise NotImplementedError(self.get_file)
260
def get_file_with_stat(self, file_id, path=None):
261
"""Get a file handle and stat object for file_id.
263
The default implementation returns (self.get_file, None) for backwards
266
:param file_id: The file id to read.
267
:param path: The path of the file, if it is known.
268
:return: A tuple (file_handle, stat_value_or_None). If the tree has
269
no stat facility, or need for a stat cache feedback during commit,
270
it may return None for the second element of the tuple.
272
return (self.get_file(file_id, path), None)
274
def get_file_text(self, file_id, path=None):
275
"""Return the byte content of a file.
277
:param file_id: The file_id of the file.
278
:param path: The path of the file.
279
If both file_id and path are supplied, an implementation may use
282
my_file = self.get_file(file_id, path)
284
return my_file.read()
288
def get_file_lines(self, file_id, path=None):
289
"""Return the content of a file, as lines.
291
:param file_id: The file_id of the file.
292
:param path: The path of the file.
294
If both file_id and path are supplied, an implementation may use
297
return osutils.split_lines(self.get_file_text(file_id, path))
299
def get_file_sha1(self, file_id, path=None, stat_value=None):
300
"""Return the SHA1 file for a file.
302
:param file_id: The handle for this file.
303
:param path: The path that this file can be found at.
304
These must point to the same object.
305
:param stat_value: Optional stat value for the object
307
raise NotImplementedError(self.get_file_sha1)
309
def get_file_mtime(self, file_id, path=None):
310
"""Return the modification time for a file.
312
:param file_id: The handle for this file.
313
:param path: The path that this file can be found at.
314
These must point to the same object.
316
raise NotImplementedError(self.get_file_mtime)
318
def get_file_size(self, file_id):
319
"""Return the size of a file in bytes.
321
This applies only to regular files. If invoked on directories or
322
symlinks, it will return None.
323
:param file_id: The file-id of the file
325
raise NotImplementedError(self.get_file_size)
77
327
def get_file_by_path(self, path):
78
return self.get_file(self._inventory.path2id(path))
80
inventory = property(_get_inventory,
81
doc="Inventory of this Tree")
328
raise NotImplementedError(self.get_file_by_path)
330
def is_executable(self, file_id, path=None):
331
"""Check if a file is executable.
333
:param file_id: The handle for this file.
334
:param path: The path that this file can be found at.
335
These must point to the same object.
337
raise NotImplementedError(self.is_executable)
339
def iter_files_bytes(self, desired_files):
340
"""Iterate through file contents.
342
Files will not necessarily be returned in the order they occur in
343
desired_files. No specific order is guaranteed.
345
Yields pairs of identifier, bytes_iterator. identifier is an opaque
346
value supplied by the caller as part of desired_files. It should
347
uniquely identify the file version in the caller's context. (Examples:
348
an index number or a TreeTransform trans_id.)
350
bytes_iterator is an iterable of bytestrings for the file. The
351
kind of iterable and length of the bytestrings are unspecified, but for
352
this implementation, it is a tuple containing a single bytestring with
353
the complete text of the file.
355
:param desired_files: a list of (file_id, identifier) pairs
357
for file_id, identifier in desired_files:
358
# We wrap the string in a tuple so that we can return an iterable
359
# of bytestrings. (Technically, a bytestring is also an iterable
360
# of bytestrings, but iterating through each character is not
362
cur_file = (self.get_file_text(file_id),)
363
yield identifier, cur_file
365
def get_symlink_target(self, file_id, path=None):
366
"""Get the target for a given file_id.
368
It is assumed that the caller already knows that file_id is referencing
370
:param file_id: Handle for the symlink entry.
371
:param path: The path of the file.
372
If both file_id and path are supplied, an implementation may use
374
:return: The path the symlink points to.
376
raise NotImplementedError(self.get_symlink_target)
378
def get_root_id(self):
379
"""Return the file_id for the root of this tree."""
380
raise NotImplementedError(self.get_root_id)
382
def annotate_iter(self, file_id,
383
default_revision=_mod_revision.CURRENT_REVISION):
384
"""Return an iterator of revision_id, line tuples.
386
For working trees (and mutable trees in general), the special
387
revision_id 'current:' will be used for lines that are new in this
388
tree, e.g. uncommitted changes.
389
:param file_id: The file to produce an annotated version from
390
:param default_revision: For lines that don't match a basis, mark them
391
with this revision id. Not all implementations will make use of
394
raise NotImplementedError(self.annotate_iter)
396
def _get_plan_merge_data(self, file_id, other, base):
397
from bzrlib import versionedfile
398
vf = versionedfile._PlanMergeVersionedFile(file_id)
399
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
400
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
402
last_revision_base = None
404
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
405
return vf, last_revision_a, last_revision_b, last_revision_base
407
def plan_file_merge(self, file_id, other, base=None):
408
"""Generate a merge plan based on annotations.
410
If the file contains uncommitted changes in this tree, they will be
411
attributed to the 'current:' pseudo-revision. If the file contains
412
uncommitted changes in the other tree, they will be assigned to the
413
'other:' pseudo-revision.
415
data = self._get_plan_merge_data(file_id, other, base)
416
vf, last_revision_a, last_revision_b, last_revision_base = data
417
return vf.plan_merge(last_revision_a, last_revision_b,
420
def plan_file_lca_merge(self, file_id, other, base=None):
421
"""Generate a merge plan based lca-newness.
423
If the file contains uncommitted changes in this tree, they will be
424
attributed to the 'current:' pseudo-revision. If the file contains
425
uncommitted changes in the other tree, they will be assigned to the
426
'other:' pseudo-revision.
428
data = self._get_plan_merge_data(file_id, other, base)
429
vf, last_revision_a, last_revision_b, last_revision_base = data
430
return vf.plan_lca_merge(last_revision_a, last_revision_b,
433
def _iter_parent_trees(self):
434
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
435
for revision_id in self.get_parent_ids():
437
yield self.revision_tree(revision_id)
438
except errors.NoSuchRevisionInTree:
439
yield self.repository.revision_tree(revision_id)
441
def _get_file_revision(self, file_id, vf, tree_revision):
442
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
444
if getattr(self, '_repository', None) is None:
445
last_revision = tree_revision
446
parent_keys = [(file_id, t.get_file_revision(file_id)) for t in
447
self._iter_parent_trees()]
448
vf.add_lines((file_id, last_revision), parent_keys,
449
self.get_file_lines(file_id))
450
repo = self.branch.repository
453
last_revision = self.get_file_revision(file_id)
454
base_vf = self._repository.texts
455
if base_vf not in vf.fallback_versionedfiles:
456
vf.fallback_versionedfiles.append(base_vf)
83
459
def _check_retrieved(self, ie, f):
86
fp = fingerprint_file(f)
462
fp = osutils.fingerprint_file(f)
89
if ie.text_size != None:
465
if ie.text_size is not None:
90
466
if ie.text_size != fp['size']:
91
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
467
raise errors.BzrError(
468
"mismatched size for file %r in %r" %
469
(ie.file_id, self._store),
92
470
["inventory expects %d bytes" % ie.text_size,
93
471
"file is actually %d bytes" % fp['size'],
94
472
"store is probably damaged/corrupt"])
96
474
if ie.text_sha1 != fp['sha1']:
97
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
475
raise errors.BzrError("wrong SHA-1 for file %r in %r" %
476
(ie.file_id, self._store),
98
477
["inventory expects %s" % ie.text_sha1,
99
478
"file is actually %s" % fp['sha1'],
100
479
"store is probably damaged/corrupt"])
103
def print_file(self, file_id):
104
"""Print file with id `file_id` to stdout."""
106
sys.stdout.write(self.get_file_text(file_id))
481
def path2id(self, path):
482
"""Return the id for path in this tree."""
483
raise NotImplementedError(self.path2id)
485
def paths2ids(self, paths, trees=[], require_versioned=True):
486
"""Return all the ids that can be reached by walking from paths.
488
Each path is looked up in this tree and any extras provided in
489
trees, and this is repeated recursively: the children in an extra tree
490
of a directory that has been renamed under a provided path in this tree
491
are all returned, even if none exist under a provided path in this
492
tree, and vice versa.
494
:param paths: An iterable of paths to start converting to ids from.
495
Alternatively, if paths is None, no ids should be calculated and None
496
will be returned. This is offered to make calling the api unconditional
497
for code that *might* take a list of files.
498
:param trees: Additional trees to consider.
499
:param require_versioned: If False, do not raise NotVersionedError if
500
an element of paths is not versioned in this tree and all of trees.
502
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
504
def iter_children(self, file_id):
505
entry = self.iter_entries_by_dir([file_id]).next()[1]
506
for child in getattr(entry, 'children', {}).itervalues():
108
509
def lock_read(self):
510
"""Lock this tree for multiple read only operations.
512
:return: A bzrlib.lock.LogicalLockResult.
516
def revision_tree(self, revision_id):
517
"""Obtain a revision tree for the revision revision_id.
519
The intention of this method is to allow access to possibly cached
520
tree data. Implementors of this method should raise NoSuchRevision if
521
the tree is not locally available, even if they could obtain the
522
tree via a repository or some other means. Callers are responsible
523
for finding the ultimate source for a revision tree.
525
:param revision_id: The revision_id of the requested tree.
527
:raises: NoSuchRevision if the tree cannot be obtained.
529
raise errors.NoSuchRevisionInTree(self, revision_id)
532
"""What files are present in this tree and unknown.
534
:return: an iterator over the unknown files.
111
538
def unlock(self):
114
541
def filter_unversioned_files(self, paths):
115
"""Filter out paths that are not versioned.
542
"""Filter out paths that are versioned.
544
:return: set of paths.
546
raise NotImplementedError(self.filter_unversioned_files)
548
def walkdirs(self, prefix=""):
549
"""Walk the contents of this tree from path down.
551
This yields all the data about the contents of a directory at a time.
552
After each directory has been yielded, if the caller has mutated the
553
list to exclude some directories, they are then not descended into.
555
The data yielded is of the form:
556
((directory-relpath, directory-path-from-root, directory-fileid),
557
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
558
versioned_kind), ...]),
559
- directory-relpath is the containing dirs relpath from prefix
560
- directory-path-from-root is the containing dirs path from /
561
- directory-fileid is the id of the directory if it is versioned.
562
- relpath is the relative path within the subtree being walked.
563
- basename is the basename
564
- kind is the kind of the file now. If unknonwn then the file is not
565
present within the tree - but it may be recorded as versioned. See
567
- lstat is the stat data *if* the file was statted.
568
- path_from_tree_root is the path from the root of the tree.
569
- file_id is the file_id if the entry is versioned.
570
- versioned_kind is the kind of the file as last recorded in the
571
versioning system. If 'unknown' the file is not versioned.
572
One of 'kind' and 'versioned_kind' must not be 'unknown'.
574
:param prefix: Start walking from prefix within the tree rather than
575
at the root. This allows one to walk a subtree but get paths that are
576
relative to a tree rooted higher up.
577
:return: an iterator over the directory data.
579
raise NotImplementedError(self.walkdirs)
581
def supports_content_filtering(self):
584
def _content_filter_stack(self, path=None, file_id=None):
585
"""The stack of content filters for a path if filtering is supported.
587
Readers will be applied in first-to-last order.
588
Writers will be applied in last-to-first order.
589
Either the path or the file-id needs to be provided.
591
:param path: path relative to the root of the tree
593
:param file_id: file_id or None if unknown
594
:return: the list of filters - [] if there are none
596
filter_pref_names = filters._get_registered_names()
597
if len(filter_pref_names) == 0:
600
path = self.id2path(file_id)
601
prefs = self.iter_search_rules([path], filter_pref_names).next()
602
stk = filters._get_filter_stack_for(prefs)
603
if 'filters' in debug.debug_flags:
604
trace.note("*** %s content-filter: %s => %r" % (path,prefs,stk))
607
def _content_filter_stack_provider(self):
608
"""A function that returns a stack of ContentFilters.
610
The function takes a path (relative to the top of the tree) and a
611
file-id as parameters.
613
:return: None if content filtering is not supported by this tree.
615
if self.supports_content_filtering():
616
return lambda path, file_id: \
617
self._content_filter_stack(path, file_id)
621
def iter_search_rules(self, path_names, pref_names=None,
622
_default_searcher=None):
623
"""Find the preferences for filenames in a tree.
625
:param path_names: an iterable of paths to find attributes for.
626
Paths are given relative to the root of the tree.
627
:param pref_names: the list of preferences to lookup - None for all
628
:param _default_searcher: private parameter to assist testing - don't use
629
:return: an iterator of tuple sequences, one per path-name.
630
See _RulesSearcher.get_items for details on the tuple sequence.
632
if _default_searcher is None:
633
_default_searcher = rules._per_user_searcher
634
searcher = self._get_rules_searcher(_default_searcher)
635
if searcher is not None:
636
if pref_names is not None:
637
for path in path_names:
638
yield searcher.get_selected_items(path, pref_names)
640
for path in path_names:
641
yield searcher.get_items(path)
643
def _get_rules_searcher(self, default_searcher):
644
"""Get the RulesSearcher for this tree given the default one."""
645
searcher = default_searcher
649
class InventoryTree(Tree):
650
"""A tree that relies on an inventory for its metadata.
652
Trees contain an `Inventory` object, and also know how to retrieve
653
file texts mentioned in the inventory, either from a working
654
directory or from a store.
656
It is possible for trees to contain files that are not described
657
in their inventory or vice versa; for this use `filenames()`.
659
Subclasses should set the _inventory attribute, which is considered
660
private to external API users.
663
def get_canonical_inventory_paths(self, paths):
664
"""Like get_canonical_inventory_path() but works on multiple items.
666
:param paths: A sequence of paths relative to the root of the tree.
667
:return: A list of paths, with each item the corresponding input path
668
adjusted to account for existing elements that match case
671
return list(self._yield_canonical_inventory_paths(paths))
673
def get_canonical_inventory_path(self, path):
674
"""Returns the first inventory item that case-insensitively matches path.
676
If a path matches exactly, it is returned. If no path matches exactly
677
but more than one path matches case-insensitively, it is implementation
678
defined which is returned.
680
If no path matches case-insensitively, the input path is returned, but
681
with as many path entries that do exist changed to their canonical
684
If you need to resolve many names from the same tree, you should
685
use get_canonical_inventory_paths() to avoid O(N) behaviour.
687
:param path: A paths relative to the root of the tree.
688
:return: The input path adjusted to account for existing elements
689
that match case insensitively.
691
return self._yield_canonical_inventory_paths([path]).next()
693
def _yield_canonical_inventory_paths(self, paths):
695
# First, if the path as specified exists exactly, just use it.
696
if self.path2id(path) is not None:
700
cur_id = self.get_root_id()
702
bit_iter = iter(path.split("/"))
706
for child in self.iter_children(cur_id):
708
# XXX: it seem like if the child is known to be in the
709
# tree, we shouldn't need to go from its id back to
710
# its path -- mbp 2010-02-11
712
# XXX: it seems like we could be more efficient
713
# by just directly looking up the original name and
714
# only then searching all children; also by not
715
# chopping paths so much. -- mbp 2010-02-11
716
child_base = os.path.basename(self.id2path(child))
717
if (child_base == elt):
718
# if we found an exact match, we can stop now; if
719
# we found an approximate match we need to keep
720
# searching because there might be an exact match
723
new_path = osutils.pathjoin(cur_path, child_base)
725
elif child_base.lower() == lelt:
727
new_path = osutils.pathjoin(cur_path, child_base)
728
except errors.NoSuchId:
729
# before a change is committed we can see this error...
734
# got to the end of this directory and no entries matched.
735
# Return what matched so far, plus the rest as specified.
736
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
741
def _get_inventory(self):
742
return self._inventory
744
inventory = property(_get_inventory,
745
doc="Inventory of this Tree")
748
def path2id(self, path):
749
"""Return the id for path in this tree."""
750
return self._inventory.path2id(path)
752
def id2path(self, file_id):
753
"""Return the path for a file id.
757
return self.inventory.id2path(file_id)
759
def has_id(self, file_id):
760
return self.inventory.has_id(file_id)
762
def has_or_had_id(self, file_id):
763
return self.inventory.has_id(file_id)
765
def all_file_ids(self):
766
return set(self.inventory)
768
@deprecated_method(deprecated_in((2, 4, 0)))
770
return iter(self.inventory)
772
def filter_unversioned_files(self, paths):
773
"""Filter out paths that are versioned.
117
775
:return: set of paths.
119
777
# NB: we specifically *don't* call self.has_filename, because for
120
# WorkingTrees that can indicate files that exist on disk but that
778
# WorkingTrees that can indicate files that exist on disk but that
121
779
# are not versioned.
122
780
pred = self.inventory.has_filename
123
781
return set((p for p in paths if not pred(p)))
126
class RevisionTree(Tree):
127
"""Tree viewing a previous revision.
129
File text can be retrieved from the text store.
131
TODO: Some kind of `__repr__` method, but a good one
132
probably means knowing the branch and revision number,
133
or at least passing a description to the constructor.
136
def __init__(self, branch, inv, revision_id):
137
self._branch = branch
138
self._weave_store = branch.weave_store
139
self._inventory = inv
140
self._revision_id = revision_id
142
def get_weave(self, file_id):
143
return self._weave_store.get_weave(file_id,
144
self._branch.get_transaction())
146
def get_file_lines(self, file_id):
147
ie = self._inventory[file_id]
148
weave = self.get_weave(file_id)
149
return weave.get_lines(ie.revision)
151
def get_file_text(self, file_id):
152
return ''.join(self.get_file_lines(file_id))
154
def get_file(self, file_id):
155
return StringIO(self.get_file_text(file_id))
157
def get_file_size(self, file_id):
158
return self._inventory[file_id].text_size
160
def get_file_sha1(self, file_id):
161
ie = self._inventory[file_id]
162
if ie.kind == "file":
165
def is_executable(self, file_id):
166
ie = self._inventory[file_id]
167
if ie.kind != "file":
169
return self._inventory[file_id].executable
171
def has_filename(self, filename):
172
return bool(self.inventory.path2id(filename))
174
def list_files(self):
175
# The only files returned by this are those from the version
176
for path, entry in self.inventory.iter_entries():
177
yield path, 'V', entry.kind, entry.file_id, entry
179
def get_symlink_target(self, file_id):
180
ie = self._inventory[file_id]
181
return ie.symlink_target;
183
def kind(self, file_id):
184
return self._inventory[file_id].kind
187
self._branch.lock_read()
190
self._branch.unlock()
193
class EmptyTree(Tree):
195
self._inventory = Inventory()
197
def get_symlink_target(self, file_id):
200
def has_filename(self, filename):
203
def kind(self, file_id):
204
assert self._inventory[file_id].kind == "root_directory"
205
return "root_directory"
207
def list_files(self):
210
def __contains__(self, file_id):
211
return file_id in self._inventory
213
def get_file_sha1(self, file_id):
214
assert self._inventory[file_id].kind == "root_directory"
784
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
785
"""Walk the tree in 'by_dir' order.
787
This will yield each entry in the tree as a (path, entry) tuple.
788
The order that they are yielded is:
790
See Tree.iter_entries_by_dir for details.
792
:param yield_parents: If True, yield the parents from the root leading
793
down to specific_file_ids that have been requested. This has no
794
impact if specific_file_ids is None.
796
return self.inventory.iter_entries_by_dir(
797
specific_file_ids=specific_file_ids, yield_parents=yield_parents)
799
def get_file_by_path(self, path):
800
return self.get_file(self._inventory.path2id(path), path)
218
803
######################################################################
273
def find_renames(old_inv, new_inv):
274
for file_id in old_inv:
275
if file_id not in new_inv:
277
old_name = old_inv.id2path(file_id)
278
new_name = new_inv.id2path(file_id)
279
if old_name != new_name:
280
yield (old_name, new_name)
856
def find_ids_across_trees(filenames, trees, require_versioned=True):
857
"""Find the ids corresponding to specified filenames.
859
All matches in all trees will be used, and all children of matched
860
directories will be used.
862
:param filenames: The filenames to find file_ids for (if None, returns
864
:param trees: The trees to find file_ids within
865
:param require_versioned: if true, all specified filenames must occur in
867
:return: a set of file ids for the specified filenames and their children.
871
specified_path_ids = _find_ids_across_trees(filenames, trees,
873
return _find_children_across_trees(specified_path_ids, trees)
876
def _find_ids_across_trees(filenames, trees, require_versioned):
877
"""Find the ids corresponding to specified filenames.
879
All matches in all trees will be used, but subdirectories are not scanned.
881
:param filenames: The filenames to find file_ids for
882
:param trees: The trees to find file_ids within
883
:param require_versioned: if true, all specified filenames must occur in
885
:return: a set of file ids for the specified filenames
888
interesting_ids = set()
889
for tree_path in filenames:
892
file_id = tree.path2id(tree_path)
893
if file_id is not None:
894
interesting_ids.add(file_id)
897
not_versioned.append(tree_path)
898
if len(not_versioned) > 0 and require_versioned:
899
raise errors.PathsNotVersionedError(not_versioned)
900
return interesting_ids
903
def _find_children_across_trees(specified_ids, trees):
904
"""Return a set including specified ids and their children.
906
All matches in all trees will be used.
908
:param trees: The trees to find file_ids within
909
:return: a set containing all specified ids and their children
911
interesting_ids = set(specified_ids)
912
pending = interesting_ids
913
# now handle children of interesting ids
914
# we loop so that we handle all children of each id in both trees
915
while len(pending) > 0:
917
for file_id in pending:
919
if not tree.has_or_had_id(file_id):
921
for child_id in tree.iter_children(file_id):
922
if child_id not in interesting_ids:
923
new_pending.add(child_id)
924
interesting_ids.update(new_pending)
925
pending = new_pending
926
return interesting_ids
929
class InterTree(InterObject):
930
"""This class represents operations taking place between two Trees.
932
Its instances have methods like 'compare' and contain references to the
933
source and target trees these operations are to be carried out on.
935
Clients of bzrlib should not need to use InterTree directly, rather they
936
should use the convenience methods on Tree such as 'Tree.compare()' which
937
will pass through to InterTree as appropriate.
940
# Formats that will be used to test this InterTree. If both are
941
# None, this InterTree will not be tested (e.g. because a complex
943
_matching_from_tree_format = None
944
_matching_to_tree_format = None
949
def is_compatible(kls, source, target):
950
# The default implementation is naive and uses the public API, so
951
# it works for all trees.
954
def _changes_from_entries(self, source_entry, target_entry,
955
source_path=None, target_path=None):
956
"""Generate a iter_changes tuple between source_entry and target_entry.
958
:param source_entry: An inventory entry from self.source, or None.
959
:param target_entry: An inventory entry from self.target, or None.
960
:param source_path: The path of source_entry, if known. If not known
961
it will be looked up.
962
:param target_path: The path of target_entry, if known. If not known
963
it will be looked up.
964
:return: A tuple, item 0 of which is an iter_changes result tuple, and
965
item 1 is True if there are any changes in the result tuple.
967
if source_entry is None:
968
if target_entry is None:
970
file_id = target_entry.file_id
972
file_id = source_entry.file_id
973
if source_entry is not None:
974
source_versioned = True
975
source_name = source_entry.name
976
source_parent = source_entry.parent_id
977
if source_path is None:
978
source_path = self.source.id2path(file_id)
979
source_kind, source_executable, source_stat = \
980
self.source._comparison_data(source_entry, source_path)
982
source_versioned = False
986
source_executable = None
987
if target_entry is not None:
988
target_versioned = True
989
target_name = target_entry.name
990
target_parent = target_entry.parent_id
991
if target_path is None:
992
target_path = self.target.id2path(file_id)
993
target_kind, target_executable, target_stat = \
994
self.target._comparison_data(target_entry, target_path)
996
target_versioned = False
1000
target_executable = None
1001
versioned = (source_versioned, target_versioned)
1002
kind = (source_kind, target_kind)
1003
changed_content = False
1004
if source_kind != target_kind:
1005
changed_content = True
1006
elif source_kind == 'file':
1007
if (self.source.get_file_sha1(file_id, source_path, source_stat) !=
1008
self.target.get_file_sha1(file_id, target_path, target_stat)):
1009
changed_content = True
1010
elif source_kind == 'symlink':
1011
if (self.source.get_symlink_target(file_id) !=
1012
self.target.get_symlink_target(file_id)):
1013
changed_content = True
1014
# XXX: Yes, the indentation below is wrong. But fixing it broke
1015
# test_merge.TestMergerEntriesLCAOnDisk.
1016
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
1017
# the fix from bzr.dev -- vila 2009026
1018
elif source_kind == 'tree-reference':
1019
if (self.source.get_reference_revision(file_id, source_path)
1020
!= self.target.get_reference_revision(file_id, target_path)):
1021
changed_content = True
1022
parent = (source_parent, target_parent)
1023
name = (source_name, target_name)
1024
executable = (source_executable, target_executable)
1025
if (changed_content is not False or versioned[0] != versioned[1]
1026
or parent[0] != parent[1] or name[0] != name[1] or
1027
executable[0] != executable[1]):
1031
return (file_id, (source_path, target_path), changed_content,
1032
versioned, parent, name, kind, executable), changes
1035
def compare(self, want_unchanged=False, specific_files=None,
1036
extra_trees=None, require_versioned=False, include_root=False,
1037
want_unversioned=False):
1038
"""Return the changes from source to target.
1040
:return: A TreeDelta.
1041
:param specific_files: An optional list of file paths to restrict the
1042
comparison to. When mapping filenames to ids, all matches in all
1043
trees (including optional extra_trees) are used, and all children of
1044
matched directories are included.
1045
:param want_unchanged: An optional boolean requesting the inclusion of
1046
unchanged entries in the result.
1047
:param extra_trees: An optional list of additional trees to use when
1048
mapping the contents of specific_files (paths) to file_ids.
1049
:param require_versioned: An optional boolean (defaults to False). When
1050
supplied and True all the 'specific_files' must be versioned, or
1051
a PathsNotVersionedError will be thrown.
1052
:param want_unversioned: Scan for unversioned paths.
1054
trees = (self.source,)
1055
if extra_trees is not None:
1056
trees = trees + tuple(extra_trees)
1057
# target is usually the newer tree:
1058
specific_file_ids = self.target.paths2ids(specific_files, trees,
1059
require_versioned=require_versioned)
1060
if specific_files and not specific_file_ids:
1061
# All files are unversioned, so just return an empty delta
1062
# _compare_trees would think we want a complete delta
1063
result = delta.TreeDelta()
1064
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1065
result.unversioned = [(path, None,
1066
self.target._comparison_data(fake_entry, path)[0]) for path in
1069
return delta._compare_trees(self.source, self.target, want_unchanged,
1070
specific_files, include_root, extra_trees=extra_trees,
1071
require_versioned=require_versioned,
1072
want_unversioned=want_unversioned)
1074
def iter_changes(self, include_unchanged=False,
1075
specific_files=None, pb=None, extra_trees=[],
1076
require_versioned=True, want_unversioned=False):
1077
"""Generate an iterator of changes between trees.
1079
A tuple is returned:
1080
(file_id, (path_in_source, path_in_target),
1081
changed_content, versioned, parent, name, kind,
1084
Changed_content is True if the file's content has changed. This
1085
includes changes to its kind, and to a symlink's target.
1087
versioned, parent, name, kind, executable are tuples of (from, to).
1088
If a file is missing in a tree, its kind is None.
1090
Iteration is done in parent-to-child order, relative to the target
1093
There is no guarantee that all paths are in sorted order: the
1094
requirement to expand the search due to renames may result in children
1095
that should be found early being found late in the search, after
1096
lexically later results have been returned.
1097
:param require_versioned: Raise errors.PathsNotVersionedError if a
1098
path in the specific_files list is not versioned in one of
1099
source, target or extra_trees.
1100
:param specific_files: An optional list of file paths to restrict the
1101
comparison to. When mapping filenames to ids, all matches in all
1102
trees (including optional extra_trees) are used, and all children
1103
of matched directories are included. The parents in the target tree
1104
of the specific files up to and including the root of the tree are
1105
always evaluated for changes too.
1106
:param want_unversioned: Should unversioned files be returned in the
1107
output. An unversioned file is defined as one with (False, False)
1108
for the versioned pair.
1110
lookup_trees = [self.source]
1112
lookup_trees.extend(extra_trees)
1113
# The ids of items we need to examine to insure delta consistency.
1114
precise_file_ids = set()
1115
changed_file_ids = []
1116
if specific_files == []:
1117
specific_file_ids = []
1119
specific_file_ids = self.target.paths2ids(specific_files,
1120
lookup_trees, require_versioned=require_versioned)
1121
if specific_files is not None:
1122
# reparented or added entries must have their parents included
1123
# so that valid deltas can be created. The seen_parents set
1124
# tracks the parents that we need to have.
1125
# The seen_dirs set tracks directory entries we've yielded.
1126
# After outputting version object in to_entries we set difference
1127
# the two seen sets and start checking parents.
1128
seen_parents = set()
1130
if want_unversioned:
1131
all_unversioned = sorted([(p.split('/'), p) for p in
1132
self.target.extras()
1133
if specific_files is None or
1134
osutils.is_inside_any(specific_files, p)])
1135
all_unversioned = collections.deque(all_unversioned)
1137
all_unversioned = collections.deque()
1139
from_entries_by_dir = list(self.source.iter_entries_by_dir(
1140
specific_file_ids=specific_file_ids))
1141
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
1142
to_entries_by_dir = list(self.target.iter_entries_by_dir(
1143
specific_file_ids=specific_file_ids))
1144
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
1146
# the unversioned path lookup only occurs on real trees - where there
1147
# can be extras. So the fake_entry is solely used to look up
1148
# executable it values when execute is not supported.
1149
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1150
for target_path, target_entry in to_entries_by_dir:
1151
while (all_unversioned and
1152
all_unversioned[0][0] < target_path.split('/')):
1153
unversioned_path = all_unversioned.popleft()
1154
target_kind, target_executable, target_stat = \
1155
self.target._comparison_data(fake_entry, unversioned_path[1])
1156
yield (None, (None, unversioned_path[1]), True, (False, False),
1158
(None, unversioned_path[0][-1]),
1159
(None, target_kind),
1160
(None, target_executable))
1161
source_path, source_entry = from_data.get(target_entry.file_id,
1163
result, changes = self._changes_from_entries(source_entry,
1164
target_entry, source_path=source_path, target_path=target_path)
1165
to_paths[result[0]] = result[1][1]
1170
pb.update('comparing files', entry_count, num_entries)
1171
if changes or include_unchanged:
1172
if specific_file_ids is not None:
1173
new_parent_id = result[4][1]
1174
precise_file_ids.add(new_parent_id)
1175
changed_file_ids.append(result[0])
1177
# Ensure correct behaviour for reparented/added specific files.
1178
if specific_files is not None:
1179
# Record output dirs
1180
if result[6][1] == 'directory':
1181
seen_dirs.add(result[0])
1182
# Record parents of reparented/added entries.
1183
versioned = result[3]
1185
if not versioned[0] or parents[0] != parents[1]:
1186
seen_parents.add(parents[1])
1187
while all_unversioned:
1188
# yield any trailing unversioned paths
1189
unversioned_path = all_unversioned.popleft()
1190
to_kind, to_executable, to_stat = \
1191
self.target._comparison_data(fake_entry, unversioned_path[1])
1192
yield (None, (None, unversioned_path[1]), True, (False, False),
1194
(None, unversioned_path[0][-1]),
1196
(None, to_executable))
1197
# Yield all remaining source paths
1198
for path, from_entry in from_entries_by_dir:
1199
file_id = from_entry.file_id
1200
if file_id in to_paths:
1203
if not self.target.has_id(file_id):
1204
# common case - paths we have not emitted are not present in
1208
to_path = self.target.id2path(file_id)
1211
pb.update('comparing files', entry_count, num_entries)
1212
versioned = (True, False)
1213
parent = (from_entry.parent_id, None)
1214
name = (from_entry.name, None)
1215
from_kind, from_executable, stat_value = \
1216
self.source._comparison_data(from_entry, path)
1217
kind = (from_kind, None)
1218
executable = (from_executable, None)
1219
changed_content = from_kind is not None
1220
# the parent's path is necessarily known at this point.
1221
changed_file_ids.append(file_id)
1222
yield(file_id, (path, to_path), changed_content, versioned, parent,
1223
name, kind, executable)
1224
changed_file_ids = set(changed_file_ids)
1225
if specific_file_ids is not None:
1226
for result in self._handle_precise_ids(precise_file_ids,
1230
def _get_entry(self, tree, file_id):
1231
"""Get an inventory entry from a tree, with missing entries as None.
1233
If the tree raises NotImplementedError on accessing .inventory, then
1234
this is worked around using iter_entries_by_dir on just the file id
1237
:param tree: The tree to lookup the entry in.
1238
:param file_id: The file_id to lookup.
1241
inventory = tree.inventory
1242
except NotImplementedError:
1243
# No inventory available.
1245
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1246
return iterator.next()[1]
1247
except StopIteration:
1251
return inventory[file_id]
1252
except errors.NoSuchId:
1255
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1256
discarded_changes=None):
1257
"""Fill out a partial iter_changes to be consistent.
1259
:param precise_file_ids: The file ids of parents that were seen during
1261
:param changed_file_ids: The file ids of already emitted items.
1262
:param discarded_changes: An optional dict of precalculated
1263
iter_changes items which the partial iter_changes had not output
1265
:return: A generator of iter_changes items to output.
1267
# process parents of things that had changed under the users
1268
# requested paths to prevent incorrect paths or parent ids which
1269
# aren't in the tree.
1270
while precise_file_ids:
1271
precise_file_ids.discard(None)
1272
# Don't emit file_ids twice
1273
precise_file_ids.difference_update(changed_file_ids)
1274
if not precise_file_ids:
1276
# If the there was something at a given output path in source, we
1277
# have to include the entry from source in the delta, or we would
1278
# be putting this entry into a used path.
1280
for parent_id in precise_file_ids:
1282
paths.append(self.target.id2path(parent_id))
1283
except errors.NoSuchId:
1284
# This id has been dragged in from the source by delta
1285
# expansion and isn't present in target at all: we don't
1286
# need to check for path collisions on it.
1289
old_id = self.source.path2id(path)
1290
precise_file_ids.add(old_id)
1291
precise_file_ids.discard(None)
1292
current_ids = precise_file_ids
1293
precise_file_ids = set()
1294
# We have to emit all of precise_file_ids that have been altered.
1295
# We may have to output the children of some of those ids if any
1296
# directories have stopped being directories.
1297
for file_id in current_ids:
1299
if discarded_changes:
1300
result = discarded_changes.get(file_id)
1305
old_entry = self._get_entry(self.source, file_id)
1306
new_entry = self._get_entry(self.target, file_id)
1307
result, changes = self._changes_from_entries(
1308
old_entry, new_entry)
1311
# Get this parents parent to examine.
1312
new_parent_id = result[4][1]
1313
precise_file_ids.add(new_parent_id)
1315
if (result[6][0] == 'directory' and
1316
result[6][1] != 'directory'):
1317
# This stopped being a directory, the old children have
1319
if old_entry is None:
1320
# Reusing a discarded change.
1321
old_entry = self._get_entry(self.source, file_id)
1322
for child in old_entry.children.values():
1323
precise_file_ids.add(child.file_id)
1324
changed_file_ids.add(result[0])
1328
InterTree.register_optimiser(InterTree)
1331
class MultiWalker(object):
1332
"""Walk multiple trees simultaneously, getting combined results."""
1334
# Note: This could be written to not assume you can do out-of-order
1335
# lookups. Instead any nodes that don't match in all trees could be
1336
# marked as 'deferred', and then returned in the final cleanup loop.
1337
# For now, I think it is "nicer" to return things as close to the
1338
# "master_tree" order as we can.
1340
def __init__(self, master_tree, other_trees):
1341
"""Create a new MultiWalker.
1343
All trees being walked must implement "iter_entries_by_dir()", such
1344
that they yield (path, object) tuples, where that object will have a
1345
'.file_id' member, that can be used to check equality.
1347
:param master_tree: All trees will be 'slaved' to the master_tree such
1348
that nodes in master_tree will be used as 'first-pass' sync points.
1349
Any nodes that aren't in master_tree will be merged in a second
1351
:param other_trees: A list of other trees to walk simultaneously.
1353
self._master_tree = master_tree
1354
self._other_trees = other_trees
1356
# Keep track of any nodes that were properly processed just out of
1357
# order, that way we don't return them at the end, we don't have to
1358
# track *all* processed file_ids, just the out-of-order ones
1359
self._out_of_order_processed = set()
1362
def _step_one(iterator):
1363
"""Step an iter_entries_by_dir iterator.
1365
:return: (has_more, path, ie)
1366
If has_more is False, path and ie will be None.
1369
path, ie = iterator.next()
1370
except StopIteration:
1371
return False, None, None
1373
return True, path, ie
1376
def _cmp_path_by_dirblock(path1, path2):
1377
"""Compare two paths based on what directory they are in.
1379
This generates a sort order, such that all children of a directory are
1380
sorted together, and grandchildren are in the same order as the
1381
children appear. But all grandchildren come after all children.
1383
:param path1: first path
1384
:param path2: the second path
1385
:return: negative number if ``path1`` comes first,
1386
0 if paths are equal
1387
and a positive number if ``path2`` sorts first
1389
# Shortcut this special case
1392
# This is stolen from _dirstate_helpers_py.py, only switching it to
1393
# Unicode objects. Consider using encode_utf8() and then using the
1394
# optimized versions, or maybe writing optimized unicode versions.
1395
if not isinstance(path1, unicode):
1396
raise TypeError("'path1' must be a unicode string, not %s: %r"
1397
% (type(path1), path1))
1398
if not isinstance(path2, unicode):
1399
raise TypeError("'path2' must be a unicode string, not %s: %r"
1400
% (type(path2), path2))
1401
return cmp(MultiWalker._path_to_key(path1),
1402
MultiWalker._path_to_key(path2))
1405
def _path_to_key(path):
1406
dirname, basename = osutils.split(path)
1407
return (dirname.split(u'/'), basename)
1409
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1410
"""Lookup an inventory entry by file_id.
1412
This is called when an entry is missing in the normal order.
1413
Generally this is because a file was either renamed, or it was
1414
deleted/added. If the entry was found in the inventory and not in
1415
extra_entries, it will be added to self._out_of_order_processed
1417
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1418
should be filled with entries that were found before they were
1419
used. If file_id is present, it will be removed from the
1421
:param other_tree: The Tree to search, in case we didn't find the entry
1423
:param file_id: The file_id to look for
1424
:return: (path, ie) if found or (None, None) if not present.
1426
if file_id in extra_entries:
1427
return extra_entries.pop(file_id)
1428
# TODO: Is id2path better as the first call, or is
1429
# inventory[file_id] better as a first check?
1431
cur_path = other_tree.id2path(file_id)
1432
except errors.NoSuchId:
1434
if cur_path is None:
1437
self._out_of_order_processed.add(file_id)
1438
cur_ie = other_tree.inventory[file_id]
1439
return (cur_path, cur_ie)
1442
"""Match up the values in the different trees."""
1443
for result in self._walk_master_tree():
1445
self._finish_others()
1446
for result in self._walk_others():
1449
def _walk_master_tree(self):
1450
"""First pass, walk all trees in lock-step.
1452
When we are done, all nodes in the master_tree will have been
1453
processed. _other_walkers, _other_entries, and _others_extra will be
1454
set on 'self' for future processing.
1456
# This iterator has the most "inlining" done, because it tends to touch
1457
# every file in the tree, while the others only hit nodes that don't
1459
master_iterator = self._master_tree.iter_entries_by_dir()
1461
other_walkers = [other.iter_entries_by_dir()
1462
for other in self._other_trees]
1463
other_entries = [self._step_one(walker) for walker in other_walkers]
1464
# Track extra nodes in the other trees
1465
others_extra = [{} for i in xrange(len(self._other_trees))]
1467
master_has_more = True
1468
step_one = self._step_one
1469
lookup_by_file_id = self._lookup_by_file_id
1470
out_of_order_processed = self._out_of_order_processed
1472
while master_has_more:
1473
(master_has_more, path, master_ie) = step_one(master_iterator)
1474
if not master_has_more:
1477
file_id = master_ie.file_id
1479
other_values_append = other_values.append
1480
next_other_entries = []
1481
next_other_entries_append = next_other_entries.append
1482
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1483
if not other_has_more:
1484
other_values_append(lookup_by_file_id(
1485
others_extra[idx], self._other_trees[idx], file_id))
1486
next_other_entries_append((False, None, None))
1487
elif file_id == other_ie.file_id:
1488
# This is the critical code path, as most of the entries
1489
# should match between most trees.
1490
other_values_append((other_path, other_ie))
1491
next_other_entries_append(step_one(other_walkers[idx]))
1493
# This walker did not match, step it until it either
1494
# matches, or we know we are past the current walker.
1495
other_walker = other_walkers[idx]
1496
other_extra = others_extra[idx]
1497
while (other_has_more and
1498
self._cmp_path_by_dirblock(other_path, path) < 0):
1499
other_file_id = other_ie.file_id
1500
if other_file_id not in out_of_order_processed:
1501
other_extra[other_file_id] = (other_path, other_ie)
1502
other_has_more, other_path, other_ie = \
1503
step_one(other_walker)
1504
if other_has_more and other_ie.file_id == file_id:
1505
# We ended up walking to this point, match and step
1507
other_values_append((other_path, other_ie))
1508
other_has_more, other_path, other_ie = \
1509
step_one(other_walker)
1511
# This record isn't in the normal order, see if it
1513
other_values_append(lookup_by_file_id(
1514
other_extra, self._other_trees[idx], file_id))
1515
next_other_entries_append((other_has_more, other_path,
1517
other_entries = next_other_entries
1519
# We've matched all the walkers, yield this datapoint
1520
yield path, file_id, master_ie, other_values
1521
self._other_walkers = other_walkers
1522
self._other_entries = other_entries
1523
self._others_extra = others_extra
1525
def _finish_others(self):
1526
"""Finish walking the other iterators, so we get all entries."""
1527
for idx, info in enumerate(self._other_entries):
1528
other_extra = self._others_extra[idx]
1529
(other_has_more, other_path, other_ie) = info
1530
while other_has_more:
1531
other_file_id = other_ie.file_id
1532
if other_file_id not in self._out_of_order_processed:
1533
other_extra[other_file_id] = (other_path, other_ie)
1534
other_has_more, other_path, other_ie = \
1535
self._step_one(self._other_walkers[idx])
1536
del self._other_entries
1538
def _walk_others(self):
1539
"""Finish up by walking all the 'deferred' nodes."""
1540
# TODO: One alternative would be to grab all possible unprocessed
1541
# file_ids, and then sort by path, and then yield them. That
1542
# might ensure better ordering, in case a caller strictly
1543
# requires parents before children.
1544
for idx, other_extra in enumerate(self._others_extra):
1545
others = sorted(other_extra.itervalues(),
1546
key=lambda x: self._path_to_key(x[0]))
1547
for other_path, other_ie in others:
1548
file_id = other_ie.file_id
1549
# We don't need to check out_of_order_processed here, because
1550
# the lookup_by_file_id will be removing anything processed
1551
# from the extras cache
1552
other_extra.pop(file_id)
1553
other_values = [(None, None) for i in xrange(idx)]
1554
other_values.append((other_path, other_ie))
1555
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1556
alt_idx = alt_idx + idx + 1
1557
alt_extra = self._others_extra[alt_idx]
1558
alt_tree = self._other_trees[alt_idx]
1559
other_values.append(self._lookup_by_file_id(
1560
alt_extra, alt_tree, file_id))
1561
yield other_path, file_id, None, other_values