1
# Copyright (C) 2005-2011 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""Tree classes, representing directory at point in time.
20
from __future__ import absolute_import
22
from .lazy_import import lazy_import
23
lazy_import(globals(), """
27
conflicts as _mod_conflicts,
31
revision as _mod_revision,
35
from breezy.i18n import gettext
43
from .inter import InterObject
50
class FileTimestampUnavailable(errors.BzrError):
52
_fmt = "The filestamp for %(path)s is not available."
56
def __init__(self, path):
60
class TreeEntry(object):
61
"""An entry that implements the minimum interface used by commands.
64
def __eq__(self, other):
65
# yes, this is ugly, TODO: best practice __eq__ style.
66
return (isinstance(other, TreeEntry)
67
and other.__class__ == self.__class__)
69
def kind_character(self):
73
class TreeDirectory(TreeEntry):
74
"""See TreeEntry. This is a directory in a working tree."""
76
def kind_character(self):
80
class TreeFile(TreeEntry):
81
"""See TreeEntry. This is a regular file in a working tree."""
83
def kind_character(self):
87
class TreeLink(TreeEntry):
88
"""See TreeEntry. This is a symlink in a working tree."""
90
def kind_character(self):
94
class TreeReference(TreeEntry):
95
"""See TreeEntry. This is a reference to a nested tree in a working tree."""
97
def kind_character(self):
102
"""Abstract file tree.
104
There are several subclasses:
106
* `WorkingTree` exists as files on disk editable by the user.
108
* `RevisionTree` is a tree as recorded at some point in the past.
110
Trees can be compared, etc, regardless of whether they are working
111
trees or versioned trees.
114
def supports_rename_tracking(self):
115
"""Whether this tree supports rename tracking.
117
This defaults to True, but some implementations may want to override
122
def has_versioned_directories(self):
123
"""Whether this tree can contain explicitly versioned directories.
125
This defaults to True, but some implementations may want to override
130
def changes_from(self, other, want_unchanged=False, specific_files=None,
131
extra_trees=None, require_versioned=False, include_root=False,
132
want_unversioned=False):
133
"""Return a TreeDelta of the changes from other to this tree.
135
:param other: A tree to compare with.
136
:param specific_files: An optional list of file paths to restrict the
137
comparison to. When mapping filenames to ids, all matches in all
138
trees (including optional extra_trees) are used, and all children of
139
matched directories are included.
140
:param want_unchanged: An optional boolean requesting the inclusion of
141
unchanged entries in the result.
142
:param extra_trees: An optional list of additional trees to use when
143
mapping the contents of specific_files (paths) to their identities.
144
:param require_versioned: An optional boolean (defaults to False). When
145
supplied and True all the 'specific_files' must be versioned, or
146
a PathsNotVersionedError will be thrown.
147
:param want_unversioned: Scan for unversioned paths.
149
The comparison will be performed by an InterTree object looked up on
152
# Martin observes that Tree.changes_from returns a TreeDelta and this
153
# may confuse people, because the class name of the returned object is
154
# a synonym of the object referenced in the method name.
155
return InterTree.get(other, self).compare(
156
want_unchanged=want_unchanged,
157
specific_files=specific_files,
158
extra_trees=extra_trees,
159
require_versioned=require_versioned,
160
include_root=include_root,
161
want_unversioned=want_unversioned,
164
def iter_changes(self, from_tree, include_unchanged=False,
165
specific_files=None, pb=None, extra_trees=None,
166
require_versioned=True, want_unversioned=False):
167
"""See InterTree.iter_changes"""
168
intertree = InterTree.get(from_tree, self)
169
return intertree.iter_changes(include_unchanged, specific_files, pb,
170
extra_trees, require_versioned, want_unversioned=want_unversioned)
173
"""Get a list of the conflicts in the tree.
175
Each conflict is an instance of breezy.conflicts.Conflict.
177
return _mod_conflicts.ConflictList()
180
"""For trees that can have unversioned files, return all such paths."""
183
def get_parent_ids(self):
184
"""Get the parent ids for this tree.
186
:return: a list of parent ids. [] is returned to indicate
187
a tree with no parents.
188
:raises: BzrError if the parents are not known.
190
raise NotImplementedError(self.get_parent_ids)
192
def has_filename(self, filename):
193
"""True if the tree has given filename."""
194
raise NotImplementedError(self.has_filename)
196
def is_ignored(self, filename):
197
"""Check whether the filename is ignored by this tree.
199
:param filename: The relative filename within the tree.
200
:return: True if the filename is ignored.
204
def all_file_ids(self):
205
"""Iterate through all file ids, including ids for missing files."""
206
raise NotImplementedError(self.all_file_ids)
208
def all_versioned_paths(self):
209
"""Iterate through all paths, including paths for missing files."""
210
raise NotImplementedError(self.all_versioned_paths)
212
def id2path(self, file_id):
213
"""Return the path for a file id.
217
raise NotImplementedError(self.id2path)
219
def iter_entries_by_dir(self, specific_files=None):
220
"""Walk the tree in 'by_dir' order.
222
This will yield each entry in the tree as a (path, entry) tuple.
223
The order that they are yielded is:
225
Directories are walked in a depth-first lexicographical order,
226
however, whenever a directory is reached, all of its direct child
227
nodes are yielded in lexicographical order before yielding the
230
For example, in the tree::
240
The yield order (ignoring root) would be::
242
a, f, a/b, a/d, a/b/c, a/d/e, f/g
244
raise NotImplementedError(self.iter_entries_by_dir)
246
def iter_child_entries(self, path):
247
"""Iterate over the children of a directory or tree reference.
249
:param path: Path of the directory
250
:raise NoSuchFile: When the path does not exist
251
:return: Iterator over entries in the directory
253
raise NotImplementedError(self.iter_child_entries)
255
def list_files(self, include_root=False, from_dir=None, recursive=True):
256
"""List all files in this tree.
258
:param include_root: Whether to include the entry for the tree root
259
:param from_dir: Directory under which to list files
260
:param recursive: Whether to list files recursively
261
:return: iterator over tuples of (path, versioned, kind, file_id,
264
raise NotImplementedError(self.list_files)
266
def iter_references(self):
267
if self.supports_tree_reference():
268
for path, entry in self.iter_entries_by_dir():
269
if entry.kind == 'tree-reference':
270
yield path, entry.file_id
272
def kind(self, path):
273
raise NotImplementedError("Tree subclass %s must implement kind"
274
% self.__class__.__name__)
276
def stored_kind(self, path):
277
"""File kind stored for this path.
279
May not match kind on disk for working trees. Always available
280
for versioned files, even when the file itself is missing.
282
return self.kind(path)
284
def path_content_summary(self, path):
285
"""Get a summary of the information about path.
287
All the attributes returned are for the canonical form, not the
288
convenient form (if content filters are in use.)
290
:param path: A relative path within the tree.
291
:return: A tuple containing kind, size, exec, sha1-or-link.
292
Kind is always present (see tree.kind()).
293
size is present if kind is file and the size of the
294
canonical form can be cheaply determined, None otherwise.
295
exec is None unless kind is file and the platform supports the 'x'
297
sha1-or-link is the link target if kind is symlink, or the sha1 if
298
it can be obtained without reading the file.
300
raise NotImplementedError(self.path_content_summary)
302
def get_reference_revision(self, path):
303
raise NotImplementedError("Tree subclass %s must implement "
304
"get_reference_revision"
305
% self.__class__.__name__)
307
def _comparison_data(self, entry, path):
308
"""Return a tuple of kind, executable, stat_value for a file.
310
entry may be None if there is no inventory entry for the file, but
311
path must always be supplied.
313
kind is None if there is no file present (even if an inventory id is
314
present). executable is False for non-file entries.
316
raise NotImplementedError(self._comparison_data)
318
def get_file(self, path):
319
"""Return a file object for the file path in the tree.
321
raise NotImplementedError(self.get_file)
323
def get_file_with_stat(self, path):
324
"""Get a file handle and stat object for path.
326
The default implementation returns (self.get_file, None) for backwards
329
:param path: The path of the file.
330
:return: A tuple (file_handle, stat_value_or_None). If the tree has
331
no stat facility, or need for a stat cache feedback during commit,
332
it may return None for the second element of the tuple.
334
return (self.get_file(path), None)
336
def get_file_text(self, path):
337
"""Return the byte content of a file.
339
:param path: The path of the file.
341
:returns: A single byte string for the whole file.
343
with self.get_file(path) as my_file:
344
return my_file.read()
346
def get_file_lines(self, path):
347
"""Return the content of a file, as lines.
349
:param path: The path of the file.
351
return osutils.split_lines(self.get_file_text(path))
353
def get_file_verifier(self, path, stat_value=None):
354
"""Return a verifier for a file.
356
The default implementation returns a sha1.
358
:param path: The path that this file can be found at.
359
These must point to the same object.
360
:param stat_value: Optional stat value for the object
361
:return: Tuple with verifier name and verifier data
363
return ("SHA1", self.get_file_sha1(path, stat_value=stat_value))
365
def get_file_sha1(self, path, stat_value=None):
366
"""Return the SHA1 file for a file.
368
:note: callers should use get_file_verifier instead
369
where possible, as the underlying repository implementation may
370
have quicker access to a non-sha1 verifier.
372
:param path: The path that this file can be found at.
373
:param stat_value: Optional stat value for the object
375
raise NotImplementedError(self.get_file_sha1)
377
def get_file_mtime(self, path):
378
"""Return the modification time for a file.
380
:param path: The path that this file can be found at.
382
raise NotImplementedError(self.get_file_mtime)
384
def get_file_size(self, path):
385
"""Return the size of a file in bytes.
387
This applies only to regular files. If invoked on directories or
388
symlinks, it will return None.
390
raise NotImplementedError(self.get_file_size)
392
def is_executable(self, path):
393
"""Check if a file is executable.
395
:param path: The path that this file can be found at.
397
raise NotImplementedError(self.is_executable)
399
def iter_files_bytes(self, desired_files):
400
"""Iterate through file contents.
402
Files will not necessarily be returned in the order they occur in
403
desired_files. No specific order is guaranteed.
405
Yields pairs of identifier, bytes_iterator. identifier is an opaque
406
value supplied by the caller as part of desired_files. It should
407
uniquely identify the file version in the caller's context. (Examples:
408
an index number or a TreeTransform trans_id.)
410
bytes_iterator is an iterable of bytestrings for the file. The
411
kind of iterable and length of the bytestrings are unspecified, but for
412
this implementation, it is a tuple containing a single bytestring with
413
the complete text of the file.
415
:param desired_files: a list of (path, identifier) pairs
417
for path, identifier in desired_files:
418
# We wrap the string in a tuple so that we can return an iterable
419
# of bytestrings. (Technically, a bytestring is also an iterable
420
# of bytestrings, but iterating through each character is not
422
cur_file = (self.get_file_text(path),)
423
yield identifier, cur_file
425
def get_symlink_target(self, path):
426
"""Get the target for a given path.
428
It is assumed that the caller already knows that path is referencing
430
:param path: The path of the file.
431
:return: The path the symlink points to.
433
raise NotImplementedError(self.get_symlink_target)
435
def get_root_id(self):
436
"""Return the file_id for the root of this tree."""
437
raise NotImplementedError(self.get_root_id)
439
def annotate_iter(self, path,
440
default_revision=_mod_revision.CURRENT_REVISION):
441
"""Return an iterator of revision_id, line tuples.
443
For working trees (and mutable trees in general), the special
444
revision_id 'current:' will be used for lines that are new in this
445
tree, e.g. uncommitted changes.
446
:param path: The file to produce an annotated version from
447
:param default_revision: For lines that don't match a basis, mark them
448
with this revision id. Not all implementations will make use of
451
raise NotImplementedError(self.annotate_iter)
453
def _iter_parent_trees(self):
454
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
455
for revision_id in self.get_parent_ids():
457
yield self.revision_tree(revision_id)
458
except errors.NoSuchRevisionInTree:
459
yield self.repository.revision_tree(revision_id)
461
def path2id(self, path):
462
"""Return the id for path in this tree."""
463
raise NotImplementedError(self.path2id)
465
def is_versioned(self, path):
466
"""Check whether path is versioned.
468
:param path: Path to check
471
return self.path2id(path) is not None
473
def find_related_paths_across_trees(self, paths, trees=[],
474
require_versioned=True):
475
"""Find related paths in tree corresponding to specified filenames in any
478
All matches in all trees will be used, and all children of matched
479
directories will be used.
481
:param paths: The filenames to find related paths for (if None, returns
483
:param trees: The trees to find file_ids within
484
:param require_versioned: if true, all specified filenames must occur in
486
:return: a set of paths for the specified filenames and their children
489
raise NotImplementedError(self.find_related_paths_across_trees)
492
"""Lock this tree for multiple read only operations.
494
:return: A breezy.lock.LogicalLockResult.
496
return lock.LogicalLockResult(self.unlock)
498
def revision_tree(self, revision_id):
499
"""Obtain a revision tree for the revision revision_id.
501
The intention of this method is to allow access to possibly cached
502
tree data. Implementors of this method should raise NoSuchRevision if
503
the tree is not locally available, even if they could obtain the
504
tree via a repository or some other means. Callers are responsible
505
for finding the ultimate source for a revision tree.
507
:param revision_id: The revision_id of the requested tree.
509
:raises: NoSuchRevision if the tree cannot be obtained.
511
raise errors.NoSuchRevisionInTree(self, revision_id)
514
"""What files are present in this tree and unknown.
516
:return: an iterator over the unknown files.
523
def filter_unversioned_files(self, paths):
524
"""Filter out paths that are versioned.
526
:return: set of paths.
528
# NB: we specifically *don't* call self.has_filename, because for
529
# WorkingTrees that can indicate files that exist on disk but that
531
return set(p for p in paths if not self.is_versioned(p))
533
def walkdirs(self, prefix=""):
534
"""Walk the contents of this tree from path down.
536
This yields all the data about the contents of a directory at a time.
537
After each directory has been yielded, if the caller has mutated the
538
list to exclude some directories, they are then not descended into.
540
The data yielded is of the form:
541
((directory-relpath, directory-path-from-root, directory-fileid),
542
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
543
versioned_kind), ...]),
544
- directory-relpath is the containing dirs relpath from prefix
545
- directory-path-from-root is the containing dirs path from /
546
- directory-fileid is the id of the directory if it is versioned.
547
- relpath is the relative path within the subtree being walked.
548
- basename is the basename
549
- kind is the kind of the file now. If unknonwn then the file is not
550
present within the tree - but it may be recorded as versioned. See
552
- lstat is the stat data *if* the file was statted.
553
- path_from_tree_root is the path from the root of the tree.
554
- file_id is the file_id if the entry is versioned.
555
- versioned_kind is the kind of the file as last recorded in the
556
versioning system. If 'unknown' the file is not versioned.
557
One of 'kind' and 'versioned_kind' must not be 'unknown'.
559
:param prefix: Start walking from prefix within the tree rather than
560
at the root. This allows one to walk a subtree but get paths that are
561
relative to a tree rooted higher up.
562
:return: an iterator over the directory data.
564
raise NotImplementedError(self.walkdirs)
566
def supports_content_filtering(self):
569
def _content_filter_stack(self, path=None):
570
"""The stack of content filters for a path if filtering is supported.
572
Readers will be applied in first-to-last order.
573
Writers will be applied in last-to-first order.
574
Either the path or the file-id needs to be provided.
576
:param path: path relative to the root of the tree
578
:return: the list of filters - [] if there are none
580
filter_pref_names = filters._get_registered_names()
581
if len(filter_pref_names) == 0:
583
prefs = next(self.iter_search_rules([path], filter_pref_names))
584
stk = filters._get_filter_stack_for(prefs)
585
if 'filters' in debug.debug_flags:
587
gettext("*** {0} content-filter: {1} => {2!r}").format(path, prefs, stk))
590
def _content_filter_stack_provider(self):
591
"""A function that returns a stack of ContentFilters.
593
The function takes a path (relative to the top of the tree) and a
594
file-id as parameters.
596
:return: None if content filtering is not supported by this tree.
598
if self.supports_content_filtering():
599
return lambda path, file_id: \
600
self._content_filter_stack(path)
604
def iter_search_rules(self, path_names, pref_names=None,
605
_default_searcher=None):
606
"""Find the preferences for filenames in a tree.
608
:param path_names: an iterable of paths to find attributes for.
609
Paths are given relative to the root of the tree.
610
:param pref_names: the list of preferences to lookup - None for all
611
:param _default_searcher: private parameter to assist testing - don't use
612
:return: an iterator of tuple sequences, one per path-name.
613
See _RulesSearcher.get_items for details on the tuple sequence.
615
if _default_searcher is None:
616
_default_searcher = rules._per_user_searcher
617
searcher = self._get_rules_searcher(_default_searcher)
618
if searcher is not None:
619
if pref_names is not None:
620
for path in path_names:
621
yield searcher.get_selected_items(path, pref_names)
623
for path in path_names:
624
yield searcher.get_items(path)
626
def _get_rules_searcher(self, default_searcher):
627
"""Get the RulesSearcher for this tree given the default one."""
628
searcher = default_searcher
631
def archive(self, format, name, root='', subdir=None,
633
"""Create an archive of this tree.
635
:param format: Format name (e.g. 'tar')
636
:param name: target file name
637
:param root: Root directory name (or None)
638
:param subdir: Subdirectory to export (or None)
639
:return: Iterator over archive chunks
641
from .archive import create_archive
642
with self.lock_read():
643
return create_archive(format, self, name, root,
644
subdir, force_mtime=force_mtime)
647
def versionable_kind(cls, kind):
648
"""Check if this tree support versioning a specific file kind."""
649
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
652
class InterTree(InterObject):
653
"""This class represents operations taking place between two Trees.
655
Its instances have methods like 'compare' and contain references to the
656
source and target trees these operations are to be carried out on.
658
Clients of breezy should not need to use InterTree directly, rather they
659
should use the convenience methods on Tree such as 'Tree.compare()' which
660
will pass through to InterTree as appropriate.
663
# Formats that will be used to test this InterTree. If both are
664
# None, this InterTree will not be tested (e.g. because a complex
666
_matching_from_tree_format = None
667
_matching_to_tree_format = None
672
def is_compatible(kls, source, target):
673
# The default implementation is naive and uses the public API, so
674
# it works for all trees.
677
def _changes_from_entries(self, source_entry, target_entry, source_path,
679
"""Generate a iter_changes tuple between source_entry and target_entry.
681
:param source_entry: An inventory entry from self.source, or None.
682
:param target_entry: An inventory entry from self.target, or None.
683
:param source_path: The path of source_entry.
684
:param target_path: The path of target_entry.
685
:return: A tuple, item 0 of which is an iter_changes result tuple, and
686
item 1 is True if there are any changes in the result tuple.
688
if source_entry is None:
689
if target_entry is None:
691
file_id = target_entry.file_id
693
file_id = source_entry.file_id
694
if source_entry is not None:
695
source_versioned = True
696
source_name = source_entry.name
697
source_parent = source_entry.parent_id
698
source_kind, source_executable, source_stat = \
699
self.source._comparison_data(source_entry, source_path)
701
source_versioned = False
705
source_executable = None
706
if target_entry is not None:
707
target_versioned = True
708
target_name = target_entry.name
709
target_parent = target_entry.parent_id
710
target_kind, target_executable, target_stat = \
711
self.target._comparison_data(target_entry, target_path)
713
target_versioned = False
717
target_executable = None
718
versioned = (source_versioned, target_versioned)
719
kind = (source_kind, target_kind)
720
changed_content = False
721
if source_kind != target_kind:
722
changed_content = True
723
elif source_kind == 'file':
724
if not self.file_content_matches(
725
source_path, target_path,
726
source_stat, target_stat):
727
changed_content = True
728
elif source_kind == 'symlink':
729
if (self.source.get_symlink_target(source_path) !=
730
self.target.get_symlink_target(target_path)):
731
changed_content = True
732
elif source_kind == 'tree-reference':
733
if (self.source.get_reference_revision(source_path)
734
!= self.target.get_reference_revision(target_path)):
735
changed_content = True
736
parent = (source_parent, target_parent)
737
name = (source_name, target_name)
738
executable = (source_executable, target_executable)
739
if (changed_content is not False or versioned[0] != versioned[1] or
740
parent[0] != parent[1] or name[0] != name[1] or
741
executable[0] != executable[1]):
745
return (file_id, (source_path, target_path), changed_content,
746
versioned, parent, name, kind, executable), changes
748
def compare(self, want_unchanged=False, specific_files=None,
749
extra_trees=None, require_versioned=False, include_root=False,
750
want_unversioned=False):
751
"""Return the changes from source to target.
753
:return: A TreeDelta.
754
:param specific_files: An optional list of file paths to restrict the
755
comparison to. When mapping filenames to ids, all matches in all
756
trees (including optional extra_trees) are used, and all children of
757
matched directories are included.
758
:param want_unchanged: An optional boolean requesting the inclusion of
759
unchanged entries in the result.
760
:param extra_trees: An optional list of additional trees to use when
761
mapping the contents of specific_files (paths) to file_ids.
762
:param require_versioned: An optional boolean (defaults to False). When
763
supplied and True all the 'specific_files' must be versioned, or
764
a PathsNotVersionedError will be thrown.
765
:param want_unversioned: Scan for unversioned paths.
767
trees = (self.source,)
768
if extra_trees is not None:
769
trees = trees + tuple(extra_trees)
770
with self.lock_read():
771
return delta._compare_trees(self.source, self.target, want_unchanged,
772
specific_files, include_root, extra_trees=extra_trees,
773
require_versioned=require_versioned,
774
want_unversioned=want_unversioned)
776
def iter_changes(self, include_unchanged=False,
777
specific_files=None, pb=None, extra_trees=[],
778
require_versioned=True, want_unversioned=False):
779
"""Generate an iterator of changes between trees.
782
(file_id, (path_in_source, path_in_target),
783
changed_content, versioned, parent, name, kind,
786
Changed_content is True if the file's content has changed. This
787
includes changes to its kind, and to a symlink's target.
789
versioned, parent, name, kind, executable are tuples of (from, to).
790
If a file is missing in a tree, its kind is None.
792
Iteration is done in parent-to-child order, relative to the target
795
There is no guarantee that all paths are in sorted order: the
796
requirement to expand the search due to renames may result in children
797
that should be found early being found late in the search, after
798
lexically later results have been returned.
799
:param require_versioned: Raise errors.PathsNotVersionedError if a
800
path in the specific_files list is not versioned in one of
801
source, target or extra_trees.
802
:param specific_files: An optional list of file paths to restrict the
803
comparison to. When mapping filenames to ids, all matches in all
804
trees (including optional extra_trees) are used, and all children
805
of matched directories are included. The parents in the target tree
806
of the specific files up to and including the root of the tree are
807
always evaluated for changes too.
808
:param want_unversioned: Should unversioned files be returned in the
809
output. An unversioned file is defined as one with (False, False)
810
for the versioned pair.
815
extra_trees = list(extra_trees)
816
# The ids of items we need to examine to insure delta consistency.
817
precise_file_ids = set()
818
changed_file_ids = []
819
if specific_files == []:
820
target_specific_files = []
821
source_specific_files = []
823
target_specific_files = self.target.find_related_paths_across_trees(
824
specific_files, [self.source] + extra_trees,
825
require_versioned=require_versioned)
826
source_specific_files = self.source.find_related_paths_across_trees(
827
specific_files, [self.target] + extra_trees,
828
require_versioned=require_versioned)
829
if specific_files is not None:
830
# reparented or added entries must have their parents included
831
# so that valid deltas can be created. The seen_parents set
832
# tracks the parents that we need to have.
833
# The seen_dirs set tracks directory entries we've yielded.
834
# After outputting version object in to_entries we set difference
835
# the two seen sets and start checking parents.
839
all_unversioned = sorted([(p.split('/'), p) for p in
841
if specific_files is None or
842
osutils.is_inside_any(specific_files, p)])
843
all_unversioned = collections.deque(all_unversioned)
845
all_unversioned = collections.deque()
847
from_entries_by_dir = list(self.source.iter_entries_by_dir(
848
specific_files=source_specific_files))
849
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
850
to_entries_by_dir = list(self.target.iter_entries_by_dir(
851
specific_files=target_specific_files))
852
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
854
# the unversioned path lookup only occurs on real trees - where there
855
# can be extras. So the fake_entry is solely used to look up
856
# executable it values when execute is not supported.
857
fake_entry = TreeFile()
858
for target_path, target_entry in to_entries_by_dir:
859
while (all_unversioned and
860
all_unversioned[0][0] < target_path.split('/')):
861
unversioned_path = all_unversioned.popleft()
862
target_kind, target_executable, target_stat = \
863
self.target._comparison_data(
864
fake_entry, unversioned_path[1])
865
yield (None, (None, unversioned_path[1]), True, (False, False),
867
(None, unversioned_path[0][-1]),
869
(None, target_executable))
870
source_path, source_entry = from_data.get(target_entry.file_id,
872
result, changes = self._changes_from_entries(source_entry,
873
target_entry, source_path=source_path, target_path=target_path)
874
to_paths[result[0]] = result[1][1]
879
pb.update('comparing files', entry_count, num_entries)
880
if changes or include_unchanged:
881
if specific_files is not None:
882
new_parent_id = result[4][1]
883
precise_file_ids.add(new_parent_id)
884
changed_file_ids.append(result[0])
886
# Ensure correct behaviour for reparented/added specific files.
887
if specific_files is not None:
889
if result[6][1] == 'directory':
890
seen_dirs.add(result[0])
891
# Record parents of reparented/added entries.
892
versioned = result[3]
894
if not versioned[0] or parents[0] != parents[1]:
895
seen_parents.add(parents[1])
896
while all_unversioned:
897
# yield any trailing unversioned paths
898
unversioned_path = all_unversioned.popleft()
899
to_kind, to_executable, to_stat = \
900
self.target._comparison_data(fake_entry, unversioned_path[1])
901
yield (None, (None, unversioned_path[1]), True, (False, False),
903
(None, unversioned_path[0][-1]),
905
(None, to_executable))
906
# Yield all remaining source paths
907
for path, from_entry in from_entries_by_dir:
908
file_id = from_entry.file_id
909
if file_id in to_paths:
912
to_path = find_previous_path(self.source, self.target, path)
915
pb.update('comparing files', entry_count, num_entries)
916
versioned = (True, False)
917
parent = (from_entry.parent_id, None)
918
name = (from_entry.name, None)
919
from_kind, from_executable, stat_value = \
920
self.source._comparison_data(from_entry, path)
921
kind = (from_kind, None)
922
executable = (from_executable, None)
923
changed_content = from_kind is not None
924
# the parent's path is necessarily known at this point.
925
changed_file_ids.append(file_id)
926
yield(file_id, (path, to_path), changed_content, versioned, parent,
927
name, kind, executable)
928
changed_file_ids = set(changed_file_ids)
929
if specific_files is not None:
930
for result in self._handle_precise_ids(precise_file_ids,
934
def _get_entry(self, tree, path):
935
"""Get an inventory entry from a tree, with missing entries as None.
937
If the tree raises NotImplementedError on accessing .inventory, then
938
this is worked around using iter_entries_by_dir on just the file id
941
:param tree: The tree to lookup the entry in.
942
:param path: The path to look up
944
# No inventory available.
946
iterator = tree.iter_entries_by_dir(specific_files=[path])
947
return next(iterator)[1]
948
except StopIteration:
951
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
952
discarded_changes=None):
953
"""Fill out a partial iter_changes to be consistent.
955
:param precise_file_ids: The file ids of parents that were seen during
957
:param changed_file_ids: The file ids of already emitted items.
958
:param discarded_changes: An optional dict of precalculated
959
iter_changes items which the partial iter_changes had not output
961
:return: A generator of iter_changes items to output.
963
# process parents of things that had changed under the users
964
# requested paths to prevent incorrect paths or parent ids which
965
# aren't in the tree.
966
while precise_file_ids:
967
precise_file_ids.discard(None)
968
# Don't emit file_ids twice
969
precise_file_ids.difference_update(changed_file_ids)
970
if not precise_file_ids:
972
# If the there was something at a given output path in source, we
973
# have to include the entry from source in the delta, or we would
974
# be putting this entry into a used path.
976
for parent_id in precise_file_ids:
978
paths.append(self.target.id2path(parent_id))
979
except errors.NoSuchId:
980
# This id has been dragged in from the source by delta
981
# expansion and isn't present in target at all: we don't
982
# need to check for path collisions on it.
985
old_id = self.source.path2id(path)
986
precise_file_ids.add(old_id)
987
precise_file_ids.discard(None)
988
current_ids = precise_file_ids
989
precise_file_ids = set()
990
# We have to emit all of precise_file_ids that have been altered.
991
# We may have to output the children of some of those ids if any
992
# directories have stopped being directories.
993
for file_id in current_ids:
995
if discarded_changes:
996
result = discarded_changes.get(file_id)
1002
source_path = self.source.id2path(file_id)
1003
except errors.NoSuchId:
1007
source_entry = self._get_entry(
1008
self.source, source_path)
1010
target_path = self.target.id2path(file_id)
1011
except errors.NoSuchId:
1015
target_entry = self._get_entry(
1016
self.target, target_path)
1017
result, changes = self._changes_from_entries(
1018
source_entry, target_entry, source_path, target_path)
1021
# Get this parents parent to examine.
1022
new_parent_id = result[4][1]
1023
precise_file_ids.add(new_parent_id)
1025
if (result[6][0] == 'directory' and
1026
result[6][1] != 'directory'):
1027
# This stopped being a directory, the old children have
1029
if source_entry is None:
1030
# Reusing a discarded change.
1031
source_entry = self._get_entry(
1032
self.source, result[1][0])
1033
precise_file_ids.update(
1035
for child in self.source.iter_child_entries(result[1][0]))
1036
changed_file_ids.add(result[0])
1039
def file_content_matches(
1040
self, source_path, target_path,
1041
source_stat=None, target_stat=None):
1042
"""Check if two files are the same in the source and target trees.
1044
This only checks that the contents of the files are the same,
1045
it does not touch anything else.
1047
:param source_path: Path of the file in the source tree
1048
:param target_path: Path of the file in the target tree
1049
:param source_file_id: Optional file id of the file in the source tree
1050
:param target_file_id: Optional file id of the file in the target tree
1051
:param source_stat: Optional stat value of the file in the source tree
1052
:param target_stat: Optional stat value of the file in the target tree
1053
:return: Boolean indicating whether the files have the same contents
1055
with self.lock_read():
1056
source_verifier_kind, source_verifier_data = (
1057
self.source.get_file_verifier(source_path, source_stat))
1058
target_verifier_kind, target_verifier_data = (
1059
self.target.get_file_verifier(
1060
target_path, target_stat))
1061
if source_verifier_kind == target_verifier_kind:
1062
return (source_verifier_data == target_verifier_data)
1063
# Fall back to SHA1 for now
1064
if source_verifier_kind != "SHA1":
1065
source_sha1 = self.source.get_file_sha1(
1066
source_path, source_file_id, source_stat)
1068
source_sha1 = source_verifier_data
1069
if target_verifier_kind != "SHA1":
1070
target_sha1 = self.target.get_file_sha1(
1071
target_path, target_file_id, target_stat)
1073
target_sha1 = target_verifier_data
1074
return (source_sha1 == target_sha1)
1077
InterTree.register_optimiser(InterTree)
1080
class MultiWalker(object):
1081
"""Walk multiple trees simultaneously, getting combined results."""
1083
# Note: This could be written to not assume you can do out-of-order
1084
# lookups. Instead any nodes that don't match in all trees could be
1085
# marked as 'deferred', and then returned in the final cleanup loop.
1086
# For now, I think it is "nicer" to return things as close to the
1087
# "master_tree" order as we can.
1089
def __init__(self, master_tree, other_trees):
1090
"""Create a new MultiWalker.
1092
All trees being walked must implement "iter_entries_by_dir()", such
1093
that they yield (path, object) tuples, where that object will have a
1094
'.file_id' member, that can be used to check equality.
1096
:param master_tree: All trees will be 'slaved' to the master_tree such
1097
that nodes in master_tree will be used as 'first-pass' sync points.
1098
Any nodes that aren't in master_tree will be merged in a second
1100
:param other_trees: A list of other trees to walk simultaneously.
1102
self._master_tree = master_tree
1103
self._other_trees = other_trees
1105
# Keep track of any nodes that were properly processed just out of
1106
# order, that way we don't return them at the end, we don't have to
1107
# track *all* processed file_ids, just the out-of-order ones
1108
self._out_of_order_processed = set()
1111
def _step_one(iterator):
1112
"""Step an iter_entries_by_dir iterator.
1114
:return: (has_more, path, ie)
1115
If has_more is False, path and ie will be None.
1118
path, ie = next(iterator)
1119
except StopIteration:
1120
return False, None, None
1122
return True, path, ie
1125
def _lt_path_by_dirblock(path1, path2):
1126
"""Compare two paths based on what directory they are in.
1128
This generates a sort order, such that all children of a directory are
1129
sorted together, and grandchildren are in the same order as the
1130
children appear. But all grandchildren come after all children.
1132
:param path1: first path
1133
:param path2: the second path
1134
:return: negative number if ``path1`` comes first,
1135
0 if paths are equal
1136
and a positive number if ``path2`` sorts first
1138
# Shortcut this special case
1141
# This is stolen from _dirstate_helpers_py.py, only switching it to
1142
# Unicode objects. Consider using encode_utf8() and then using the
1143
# optimized versions, or maybe writing optimized unicode versions.
1144
if not isinstance(path1, text_type):
1145
raise TypeError("'path1' must be a unicode string, not %s: %r"
1146
% (type(path1), path1))
1147
if not isinstance(path2, text_type):
1148
raise TypeError("'path2' must be a unicode string, not %s: %r"
1149
% (type(path2), path2))
1150
return (MultiWalker._path_to_key(path1) <
1151
MultiWalker._path_to_key(path2))
1154
def _path_to_key(path):
1155
dirname, basename = osutils.split(path)
1156
return (dirname.split(u'/'), basename)
1158
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1159
"""Lookup an inventory entry by file_id.
1161
This is called when an entry is missing in the normal order.
1162
Generally this is because a file was either renamed, or it was
1163
deleted/added. If the entry was found in the inventory and not in
1164
extra_entries, it will be added to self._out_of_order_processed
1166
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1167
should be filled with entries that were found before they were
1168
used. If file_id is present, it will be removed from the
1170
:param other_tree: The Tree to search, in case we didn't find the entry
1172
:param file_id: The file_id to look for
1173
:return: (path, ie) if found or (None, None) if not present.
1175
if file_id in extra_entries:
1176
return extra_entries.pop(file_id)
1177
# TODO: Is id2path better as the first call, or is
1178
# inventory[file_id] better as a first check?
1180
cur_path = other_tree.id2path(file_id)
1181
except errors.NoSuchId:
1183
if cur_path is None:
1186
self._out_of_order_processed.add(file_id)
1187
cur_ie = other_tree.root_inventory.get_entry(file_id)
1188
return (cur_path, cur_ie)
1191
"""Match up the values in the different trees."""
1192
for result in self._walk_master_tree():
1194
self._finish_others()
1195
for result in self._walk_others():
1198
def _walk_master_tree(self):
1199
"""First pass, walk all trees in lock-step.
1201
When we are done, all nodes in the master_tree will have been
1202
processed. _other_walkers, _other_entries, and _others_extra will be
1203
set on 'self' for future processing.
1205
# This iterator has the most "inlining" done, because it tends to touch
1206
# every file in the tree, while the others only hit nodes that don't
1208
master_iterator = self._master_tree.iter_entries_by_dir()
1210
other_walkers = [other.iter_entries_by_dir()
1211
for other in self._other_trees]
1212
other_entries = [self._step_one(walker) for walker in other_walkers]
1213
# Track extra nodes in the other trees
1214
others_extra = [{} for _ in range(len(self._other_trees))]
1216
master_has_more = True
1217
step_one = self._step_one
1218
lookup_by_file_id = self._lookup_by_file_id
1219
out_of_order_processed = self._out_of_order_processed
1221
while master_has_more:
1222
(master_has_more, path, master_ie) = step_one(master_iterator)
1223
if not master_has_more:
1226
file_id = master_ie.file_id
1228
other_values_append = other_values.append
1229
next_other_entries = []
1230
next_other_entries_append = next_other_entries.append
1231
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1232
if not other_has_more:
1233
other_values_append(lookup_by_file_id(
1234
others_extra[idx], self._other_trees[idx], file_id))
1235
next_other_entries_append((False, None, None))
1236
elif file_id == other_ie.file_id:
1237
# This is the critical code path, as most of the entries
1238
# should match between most trees.
1239
other_values_append((other_path, other_ie))
1240
next_other_entries_append(step_one(other_walkers[idx]))
1242
# This walker did not match, step it until it either
1243
# matches, or we know we are past the current walker.
1244
other_walker = other_walkers[idx]
1245
other_extra = others_extra[idx]
1246
while (other_has_more and
1247
self._lt_path_by_dirblock(other_path, path)):
1248
other_file_id = other_ie.file_id
1249
if other_file_id not in out_of_order_processed:
1250
other_extra[other_file_id] = (other_path, other_ie)
1251
other_has_more, other_path, other_ie = \
1252
step_one(other_walker)
1253
if other_has_more and other_ie.file_id == file_id:
1254
# We ended up walking to this point, match and step
1256
other_values_append((other_path, other_ie))
1257
other_has_more, other_path, other_ie = \
1258
step_one(other_walker)
1260
# This record isn't in the normal order, see if it
1262
other_values_append(lookup_by_file_id(
1263
other_extra, self._other_trees[idx], file_id))
1264
next_other_entries_append((other_has_more, other_path,
1266
other_entries = next_other_entries
1268
# We've matched all the walkers, yield this datapoint
1269
yield path, file_id, master_ie, other_values
1270
self._other_walkers = other_walkers
1271
self._other_entries = other_entries
1272
self._others_extra = others_extra
1274
def _finish_others(self):
1275
"""Finish walking the other iterators, so we get all entries."""
1276
for idx, info in enumerate(self._other_entries):
1277
other_extra = self._others_extra[idx]
1278
(other_has_more, other_path, other_ie) = info
1279
while other_has_more:
1280
other_file_id = other_ie.file_id
1281
if other_file_id not in self._out_of_order_processed:
1282
other_extra[other_file_id] = (other_path, other_ie)
1283
other_has_more, other_path, other_ie = \
1284
self._step_one(self._other_walkers[idx])
1285
del self._other_entries
1287
def _walk_others(self):
1288
"""Finish up by walking all the 'deferred' nodes."""
1289
# TODO: One alternative would be to grab all possible unprocessed
1290
# file_ids, and then sort by path, and then yield them. That
1291
# might ensure better ordering, in case a caller strictly
1292
# requires parents before children.
1293
for idx, other_extra in enumerate(self._others_extra):
1294
others = sorted(viewvalues(other_extra),
1295
key=lambda x: self._path_to_key(x[0]))
1296
for other_path, other_ie in others:
1297
file_id = other_ie.file_id
1298
# We don't need to check out_of_order_processed here, because
1299
# the lookup_by_file_id will be removing anything processed
1300
# from the extras cache
1301
other_extra.pop(file_id)
1302
other_values = [(None, None)] * idx
1303
other_values.append((other_path, other_ie))
1304
for alt_idx, alt_extra in enumerate(self._others_extra[idx + 1:]):
1305
alt_idx = alt_idx + idx + 1
1306
alt_extra = self._others_extra[alt_idx]
1307
alt_tree = self._other_trees[alt_idx]
1308
other_values.append(self._lookup_by_file_id(
1309
alt_extra, alt_tree, file_id))
1310
yield other_path, file_id, None, other_values
1313
def find_previous_paths(from_tree, to_tree, paths):
1314
"""Find previous tree paths.
1316
:param from_tree: From tree
1317
:param to_tree: To tree
1318
:param paths: Iterable over paths to search for
1319
:return: Dictionary mapping from from_tree paths to paths in to_tree, or
1320
None if there is no equivalent path.
1324
ret[path] = find_previous_path(from_tree, to_tree, path)
1328
def find_previous_path(from_tree, to_tree, path, file_id=None):
1329
"""Find previous tree path.
1331
:param from_tree: From tree
1332
:param to_tree: To tree
1333
:param path: Path to search for
1334
:return: path in to_tree, or None if there is no equivalent path.
1337
file_id = from_tree.path2id(path)
1339
raise errors.NoSuchFile(path)
1341
return to_tree.id2path(file_id)
1342
except errors.NoSuchId:
1346
def get_canonical_path(tree, path, normalize):
1347
"""Find the canonical path of an item, ignoring case.
1349
:param tree: Tree to traverse
1350
:param path: Case-insensitive path to look up
1351
:param normalize: Function to normalize a filename for comparison
1352
:return: The canonical path
1355
cur_id = tree.get_root_id()
1357
bit_iter = iter(path.split("/"))
1358
for elt in bit_iter:
1359
lelt = normalize(elt)
1362
for child in tree.iter_child_entries(cur_path, cur_id):
1364
if child.name == elt:
1365
# if we found an exact match, we can stop now; if
1366
# we found an approximate match we need to keep
1367
# searching because there might be an exact match
1369
cur_id = child.file_id
1370
new_path = osutils.pathjoin(cur_path, child.name)
1372
elif normalize(child.name) == lelt:
1373
cur_id = child.file_id
1374
new_path = osutils.pathjoin(cur_path, child.name)
1375
except errors.NoSuchId:
1376
# before a change is committed we can see this error...
1378
except errors.NotADirectory:
1383
# got to the end of this directory and no entries matched.
1384
# Return what matched so far, plus the rest as specified.
1385
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
added
removed
breezy/tree.py
