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:
586
trace.note(gettext("*** {0} content-filter: {1} => {2!r}").format(path, prefs, stk))
589
def _content_filter_stack_provider(self):
590
"""A function that returns a stack of ContentFilters.
592
The function takes a path (relative to the top of the tree) and a
593
file-id as parameters.
595
:return: None if content filtering is not supported by this tree.
597
if self.supports_content_filtering():
598
return lambda path, file_id: \
599
self._content_filter_stack(path)
603
def iter_search_rules(self, path_names, pref_names=None,
604
_default_searcher=None):
605
"""Find the preferences for filenames in a tree.
607
:param path_names: an iterable of paths to find attributes for.
608
Paths are given relative to the root of the tree.
609
:param pref_names: the list of preferences to lookup - None for all
610
:param _default_searcher: private parameter to assist testing - don't use
611
:return: an iterator of tuple sequences, one per path-name.
612
See _RulesSearcher.get_items for details on the tuple sequence.
614
if _default_searcher is None:
615
_default_searcher = rules._per_user_searcher
616
searcher = self._get_rules_searcher(_default_searcher)
617
if searcher is not None:
618
if pref_names is not None:
619
for path in path_names:
620
yield searcher.get_selected_items(path, pref_names)
622
for path in path_names:
623
yield searcher.get_items(path)
625
def _get_rules_searcher(self, default_searcher):
626
"""Get the RulesSearcher for this tree given the default one."""
627
searcher = default_searcher
630
def archive(self, format, name, root='', subdir=None,
632
"""Create an archive of this tree.
634
:param format: Format name (e.g. 'tar')
635
:param name: target file name
636
:param root: Root directory name (or None)
637
:param subdir: Subdirectory to export (or None)
638
:return: Iterator over archive chunks
640
from .archive import create_archive
641
with self.lock_read():
642
return create_archive(format, self, name, root,
643
subdir, force_mtime=force_mtime)
646
def versionable_kind(cls, kind):
647
"""Check if this tree support versioning a specific file kind."""
648
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
651
class InterTree(InterObject):
652
"""This class represents operations taking place between two Trees.
654
Its instances have methods like 'compare' and contain references to the
655
source and target trees these operations are to be carried out on.
657
Clients of breezy should not need to use InterTree directly, rather they
658
should use the convenience methods on Tree such as 'Tree.compare()' which
659
will pass through to InterTree as appropriate.
662
# Formats that will be used to test this InterTree. If both are
663
# None, this InterTree will not be tested (e.g. because a complex
665
_matching_from_tree_format = None
666
_matching_to_tree_format = None
671
def is_compatible(kls, source, target):
672
# The default implementation is naive and uses the public API, so
673
# it works for all trees.
676
def _changes_from_entries(self, source_entry, target_entry, source_path,
678
"""Generate a iter_changes tuple between source_entry and target_entry.
680
:param source_entry: An inventory entry from self.source, or None.
681
:param target_entry: An inventory entry from self.target, or None.
682
:param source_path: The path of source_entry.
683
:param target_path: The path of target_entry.
684
:return: A tuple, item 0 of which is an iter_changes result tuple, and
685
item 1 is True if there are any changes in the result tuple.
687
if source_entry is None:
688
if target_entry is None:
690
file_id = target_entry.file_id
692
file_id = source_entry.file_id
693
if source_entry is not None:
694
source_versioned = True
695
source_name = source_entry.name
696
source_parent = source_entry.parent_id
697
source_kind, source_executable, source_stat = \
698
self.source._comparison_data(source_entry, source_path)
700
source_versioned = False
704
source_executable = None
705
if target_entry is not None:
706
target_versioned = True
707
target_name = target_entry.name
708
target_parent = target_entry.parent_id
709
target_kind, target_executable, target_stat = \
710
self.target._comparison_data(target_entry, target_path)
712
target_versioned = False
716
target_executable = None
717
versioned = (source_versioned, target_versioned)
718
kind = (source_kind, target_kind)
719
changed_content = False
720
if source_kind != target_kind:
721
changed_content = True
722
elif source_kind == 'file':
723
if not self.file_content_matches(
724
source_path, target_path,
725
source_stat, target_stat):
726
changed_content = True
727
elif source_kind == 'symlink':
728
if (self.source.get_symlink_target(source_path) !=
729
self.target.get_symlink_target(target_path)):
730
changed_content = True
731
elif source_kind == 'tree-reference':
732
if (self.source.get_reference_revision(source_path)
733
!= self.target.get_reference_revision(target_path)):
734
changed_content = True
735
parent = (source_parent, target_parent)
736
name = (source_name, target_name)
737
executable = (source_executable, target_executable)
738
if (changed_content is not False or versioned[0] != versioned[1]
739
or parent[0] != parent[1] or name[0] != name[1] or
740
executable[0] != executable[1]):
744
return (file_id, (source_path, target_path), changed_content,
745
versioned, parent, name, kind, executable), changes
747
def compare(self, want_unchanged=False, specific_files=None,
748
extra_trees=None, require_versioned=False, include_root=False,
749
want_unversioned=False):
750
"""Return the changes from source to target.
752
:return: A TreeDelta.
753
:param specific_files: An optional list of file paths to restrict the
754
comparison to. When mapping filenames to ids, all matches in all
755
trees (including optional extra_trees) are used, and all children of
756
matched directories are included.
757
:param want_unchanged: An optional boolean requesting the inclusion of
758
unchanged entries in the result.
759
:param extra_trees: An optional list of additional trees to use when
760
mapping the contents of specific_files (paths) to file_ids.
761
:param require_versioned: An optional boolean (defaults to False). When
762
supplied and True all the 'specific_files' must be versioned, or
763
a PathsNotVersionedError will be thrown.
764
:param want_unversioned: Scan for unversioned paths.
766
trees = (self.source,)
767
if extra_trees is not None:
768
trees = trees + tuple(extra_trees)
769
with self.lock_read():
770
return delta._compare_trees(self.source, self.target, want_unchanged,
771
specific_files, include_root, extra_trees=extra_trees,
772
require_versioned=require_versioned,
773
want_unversioned=want_unversioned)
775
def iter_changes(self, include_unchanged=False,
776
specific_files=None, pb=None, extra_trees=[],
777
require_versioned=True, want_unversioned=False):
778
"""Generate an iterator of changes between trees.
781
(file_id, (path_in_source, path_in_target),
782
changed_content, versioned, parent, name, kind,
785
Changed_content is True if the file's content has changed. This
786
includes changes to its kind, and to a symlink's target.
788
versioned, parent, name, kind, executable are tuples of (from, to).
789
If a file is missing in a tree, its kind is None.
791
Iteration is done in parent-to-child order, relative to the target
794
There is no guarantee that all paths are in sorted order: the
795
requirement to expand the search due to renames may result in children
796
that should be found early being found late in the search, after
797
lexically later results have been returned.
798
:param require_versioned: Raise errors.PathsNotVersionedError if a
799
path in the specific_files list is not versioned in one of
800
source, target or extra_trees.
801
:param specific_files: An optional list of file paths to restrict the
802
comparison to. When mapping filenames to ids, all matches in all
803
trees (including optional extra_trees) are used, and all children
804
of matched directories are included. The parents in the target tree
805
of the specific files up to and including the root of the tree are
806
always evaluated for changes too.
807
:param want_unversioned: Should unversioned files be returned in the
808
output. An unversioned file is defined as one with (False, False)
809
for the versioned pair.
814
extra_trees = list(extra_trees)
815
# The ids of items we need to examine to insure delta consistency.
816
precise_file_ids = set()
817
changed_file_ids = []
818
if specific_files == []:
819
target_specific_files = []
820
source_specific_files = []
822
target_specific_files = self.target.find_related_paths_across_trees(
823
specific_files, [self.source] + extra_trees,
824
require_versioned=require_versioned)
825
source_specific_files = self.source.find_related_paths_across_trees(
826
specific_files, [self.target] + extra_trees,
827
require_versioned=require_versioned)
828
if specific_files is not None:
829
# reparented or added entries must have their parents included
830
# so that valid deltas can be created. The seen_parents set
831
# tracks the parents that we need to have.
832
# The seen_dirs set tracks directory entries we've yielded.
833
# After outputting version object in to_entries we set difference
834
# the two seen sets and start checking parents.
838
all_unversioned = sorted([(p.split('/'), p) for p in
840
if specific_files is None or
841
osutils.is_inside_any(specific_files, p)])
842
all_unversioned = collections.deque(all_unversioned)
844
all_unversioned = collections.deque()
846
from_entries_by_dir = list(self.source.iter_entries_by_dir(
847
specific_files=source_specific_files))
848
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
849
to_entries_by_dir = list(self.target.iter_entries_by_dir(
850
specific_files=target_specific_files))
851
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
853
# the unversioned path lookup only occurs on real trees - where there
854
# can be extras. So the fake_entry is solely used to look up
855
# executable it values when execute is not supported.
856
fake_entry = TreeFile()
857
for target_path, target_entry in to_entries_by_dir:
858
while (all_unversioned and
859
all_unversioned[0][0] < target_path.split('/')):
860
unversioned_path = all_unversioned.popleft()
861
target_kind, target_executable, target_stat = \
862
self.target._comparison_data(fake_entry, unversioned_path[1])
863
yield (None, (None, unversioned_path[1]), True, (False, False),
865
(None, unversioned_path[0][-1]),
867
(None, target_executable))
868
source_path, source_entry = from_data.get(target_entry.file_id,
870
result, changes = self._changes_from_entries(source_entry,
871
target_entry, source_path=source_path, target_path=target_path)
872
to_paths[result[0]] = result[1][1]
877
pb.update('comparing files', entry_count, num_entries)
878
if changes or include_unchanged:
879
if specific_files is not None:
880
new_parent_id = result[4][1]
881
precise_file_ids.add(new_parent_id)
882
changed_file_ids.append(result[0])
884
# Ensure correct behaviour for reparented/added specific files.
885
if specific_files is not None:
887
if result[6][1] == 'directory':
888
seen_dirs.add(result[0])
889
# Record parents of reparented/added entries.
890
versioned = result[3]
892
if not versioned[0] or parents[0] != parents[1]:
893
seen_parents.add(parents[1])
894
while all_unversioned:
895
# yield any trailing unversioned paths
896
unversioned_path = all_unversioned.popleft()
897
to_kind, to_executable, to_stat = \
898
self.target._comparison_data(fake_entry, unversioned_path[1])
899
yield (None, (None, unversioned_path[1]), True, (False, False),
901
(None, unversioned_path[0][-1]),
903
(None, to_executable))
904
# Yield all remaining source paths
905
for path, from_entry in from_entries_by_dir:
906
file_id = from_entry.file_id
907
if file_id in to_paths:
910
to_path = find_previous_path(self.source, self.target, path)
913
pb.update('comparing files', entry_count, num_entries)
914
versioned = (True, False)
915
parent = (from_entry.parent_id, None)
916
name = (from_entry.name, None)
917
from_kind, from_executable, stat_value = \
918
self.source._comparison_data(from_entry, path)
919
kind = (from_kind, None)
920
executable = (from_executable, None)
921
changed_content = from_kind is not None
922
# the parent's path is necessarily known at this point.
923
changed_file_ids.append(file_id)
924
yield(file_id, (path, to_path), changed_content, versioned, parent,
925
name, kind, executable)
926
changed_file_ids = set(changed_file_ids)
927
if specific_files is not None:
928
for result in self._handle_precise_ids(precise_file_ids,
932
def _get_entry(self, tree, path):
933
"""Get an inventory entry from a tree, with missing entries as None.
935
If the tree raises NotImplementedError on accessing .inventory, then
936
this is worked around using iter_entries_by_dir on just the file id
939
:param tree: The tree to lookup the entry in.
940
:param path: The path to look up
942
# No inventory available.
944
iterator = tree.iter_entries_by_dir(specific_files=[path])
945
return next(iterator)[1]
946
except StopIteration:
949
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
950
discarded_changes=None):
951
"""Fill out a partial iter_changes to be consistent.
953
:param precise_file_ids: The file ids of parents that were seen during
955
:param changed_file_ids: The file ids of already emitted items.
956
:param discarded_changes: An optional dict of precalculated
957
iter_changes items which the partial iter_changes had not output
959
:return: A generator of iter_changes items to output.
961
# process parents of things that had changed under the users
962
# requested paths to prevent incorrect paths or parent ids which
963
# aren't in the tree.
964
while precise_file_ids:
965
precise_file_ids.discard(None)
966
# Don't emit file_ids twice
967
precise_file_ids.difference_update(changed_file_ids)
968
if not precise_file_ids:
970
# If the there was something at a given output path in source, we
971
# have to include the entry from source in the delta, or we would
972
# be putting this entry into a used path.
974
for parent_id in precise_file_ids:
976
paths.append(self.target.id2path(parent_id))
977
except errors.NoSuchId:
978
# This id has been dragged in from the source by delta
979
# expansion and isn't present in target at all: we don't
980
# need to check for path collisions on it.
983
old_id = self.source.path2id(path)
984
precise_file_ids.add(old_id)
985
precise_file_ids.discard(None)
986
current_ids = precise_file_ids
987
precise_file_ids = set()
988
# We have to emit all of precise_file_ids that have been altered.
989
# We may have to output the children of some of those ids if any
990
# directories have stopped being directories.
991
for file_id in current_ids:
993
if discarded_changes:
994
result = discarded_changes.get(file_id)
1000
source_path = self.source.id2path(file_id)
1001
except errors.NoSuchId:
1005
source_entry = self._get_entry(self.source, source_path)
1007
target_path = self.target.id2path(file_id)
1008
except errors.NoSuchId:
1012
target_entry = self._get_entry(self.target, target_path)
1013
result, changes = self._changes_from_entries(
1014
source_entry, target_entry, source_path, target_path)
1017
# Get this parents parent to examine.
1018
new_parent_id = result[4][1]
1019
precise_file_ids.add(new_parent_id)
1021
if (result[6][0] == 'directory' and
1022
result[6][1] != 'directory'):
1023
# This stopped being a directory, the old children have
1025
if source_entry is None:
1026
# Reusing a discarded change.
1027
source_entry = self._get_entry(self.source, result[1][0])
1028
precise_file_ids.update(
1030
for child in self.source.iter_child_entries(result[1][0]))
1031
changed_file_ids.add(result[0])
1034
def file_content_matches(
1035
self, source_path, target_path,
1036
source_stat=None, target_stat=None):
1037
"""Check if two files are the same in the source and target trees.
1039
This only checks that the contents of the files are the same,
1040
it does not touch anything else.
1042
:param source_path: Path of the file in the source tree
1043
:param target_path: Path of the file in the target tree
1044
:param source_file_id: Optional file id of the file in the source tree
1045
:param target_file_id: Optional file id of the file in the target tree
1046
:param source_stat: Optional stat value of the file in the source tree
1047
:param target_stat: Optional stat value of the file in the target tree
1048
:return: Boolean indicating whether the files have the same contents
1050
with self.lock_read():
1051
source_verifier_kind, source_verifier_data = (
1052
self.source.get_file_verifier(source_path, source_stat))
1053
target_verifier_kind, target_verifier_data = (
1054
self.target.get_file_verifier(
1055
target_path, target_stat))
1056
if source_verifier_kind == target_verifier_kind:
1057
return (source_verifier_data == target_verifier_data)
1058
# Fall back to SHA1 for now
1059
if source_verifier_kind != "SHA1":
1060
source_sha1 = self.source.get_file_sha1(
1061
source_path, source_file_id, source_stat)
1063
source_sha1 = source_verifier_data
1064
if target_verifier_kind != "SHA1":
1065
target_sha1 = self.target.get_file_sha1(
1066
target_path, target_file_id, target_stat)
1068
target_sha1 = target_verifier_data
1069
return (source_sha1 == target_sha1)
1071
InterTree.register_optimiser(InterTree)
1074
class MultiWalker(object):
1075
"""Walk multiple trees simultaneously, getting combined results."""
1077
# Note: This could be written to not assume you can do out-of-order
1078
# lookups. Instead any nodes that don't match in all trees could be
1079
# marked as 'deferred', and then returned in the final cleanup loop.
1080
# For now, I think it is "nicer" to return things as close to the
1081
# "master_tree" order as we can.
1083
def __init__(self, master_tree, other_trees):
1084
"""Create a new MultiWalker.
1086
All trees being walked must implement "iter_entries_by_dir()", such
1087
that they yield (path, object) tuples, where that object will have a
1088
'.file_id' member, that can be used to check equality.
1090
:param master_tree: All trees will be 'slaved' to the master_tree such
1091
that nodes in master_tree will be used as 'first-pass' sync points.
1092
Any nodes that aren't in master_tree will be merged in a second
1094
:param other_trees: A list of other trees to walk simultaneously.
1096
self._master_tree = master_tree
1097
self._other_trees = other_trees
1099
# Keep track of any nodes that were properly processed just out of
1100
# order, that way we don't return them at the end, we don't have to
1101
# track *all* processed file_ids, just the out-of-order ones
1102
self._out_of_order_processed = set()
1105
def _step_one(iterator):
1106
"""Step an iter_entries_by_dir iterator.
1108
:return: (has_more, path, ie)
1109
If has_more is False, path and ie will be None.
1112
path, ie = next(iterator)
1113
except StopIteration:
1114
return False, None, None
1116
return True, path, ie
1119
def _lt_path_by_dirblock(path1, path2):
1120
"""Compare two paths based on what directory they are in.
1122
This generates a sort order, such that all children of a directory are
1123
sorted together, and grandchildren are in the same order as the
1124
children appear. But all grandchildren come after all children.
1126
:param path1: first path
1127
:param path2: the second path
1128
:return: negative number if ``path1`` comes first,
1129
0 if paths are equal
1130
and a positive number if ``path2`` sorts first
1132
# Shortcut this special case
1135
# This is stolen from _dirstate_helpers_py.py, only switching it to
1136
# Unicode objects. Consider using encode_utf8() and then using the
1137
# optimized versions, or maybe writing optimized unicode versions.
1138
if not isinstance(path1, text_type):
1139
raise TypeError("'path1' must be a unicode string, not %s: %r"
1140
% (type(path1), path1))
1141
if not isinstance(path2, text_type):
1142
raise TypeError("'path2' must be a unicode string, not %s: %r"
1143
% (type(path2), path2))
1144
return (MultiWalker._path_to_key(path1) <
1145
MultiWalker._path_to_key(path2))
1148
def _path_to_key(path):
1149
dirname, basename = osutils.split(path)
1150
return (dirname.split(u'/'), basename)
1152
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1153
"""Lookup an inventory entry by file_id.
1155
This is called when an entry is missing in the normal order.
1156
Generally this is because a file was either renamed, or it was
1157
deleted/added. If the entry was found in the inventory and not in
1158
extra_entries, it will be added to self._out_of_order_processed
1160
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1161
should be filled with entries that were found before they were
1162
used. If file_id is present, it will be removed from the
1164
:param other_tree: The Tree to search, in case we didn't find the entry
1166
:param file_id: The file_id to look for
1167
:return: (path, ie) if found or (None, None) if not present.
1169
if file_id in extra_entries:
1170
return extra_entries.pop(file_id)
1171
# TODO: Is id2path better as the first call, or is
1172
# inventory[file_id] better as a first check?
1174
cur_path = other_tree.id2path(file_id)
1175
except errors.NoSuchId:
1177
if cur_path is None:
1180
self._out_of_order_processed.add(file_id)
1181
cur_ie = other_tree.root_inventory.get_entry(file_id)
1182
return (cur_path, cur_ie)
1185
"""Match up the values in the different trees."""
1186
for result in self._walk_master_tree():
1188
self._finish_others()
1189
for result in self._walk_others():
1192
def _walk_master_tree(self):
1193
"""First pass, walk all trees in lock-step.
1195
When we are done, all nodes in the master_tree will have been
1196
processed. _other_walkers, _other_entries, and _others_extra will be
1197
set on 'self' for future processing.
1199
# This iterator has the most "inlining" done, because it tends to touch
1200
# every file in the tree, while the others only hit nodes that don't
1202
master_iterator = self._master_tree.iter_entries_by_dir()
1204
other_walkers = [other.iter_entries_by_dir()
1205
for other in self._other_trees]
1206
other_entries = [self._step_one(walker) for walker in other_walkers]
1207
# Track extra nodes in the other trees
1208
others_extra = [{} for _ in range(len(self._other_trees))]
1210
master_has_more = True
1211
step_one = self._step_one
1212
lookup_by_file_id = self._lookup_by_file_id
1213
out_of_order_processed = self._out_of_order_processed
1215
while master_has_more:
1216
(master_has_more, path, master_ie) = step_one(master_iterator)
1217
if not master_has_more:
1220
file_id = master_ie.file_id
1222
other_values_append = other_values.append
1223
next_other_entries = []
1224
next_other_entries_append = next_other_entries.append
1225
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1226
if not other_has_more:
1227
other_values_append(lookup_by_file_id(
1228
others_extra[idx], self._other_trees[idx], file_id))
1229
next_other_entries_append((False, None, None))
1230
elif file_id == other_ie.file_id:
1231
# This is the critical code path, as most of the entries
1232
# should match between most trees.
1233
other_values_append((other_path, other_ie))
1234
next_other_entries_append(step_one(other_walkers[idx]))
1236
# This walker did not match, step it until it either
1237
# matches, or we know we are past the current walker.
1238
other_walker = other_walkers[idx]
1239
other_extra = others_extra[idx]
1240
while (other_has_more and
1241
self._lt_path_by_dirblock(other_path, path)):
1242
other_file_id = other_ie.file_id
1243
if other_file_id not in out_of_order_processed:
1244
other_extra[other_file_id] = (other_path, other_ie)
1245
other_has_more, other_path, other_ie = \
1246
step_one(other_walker)
1247
if other_has_more and other_ie.file_id == file_id:
1248
# We ended up walking to this point, match and step
1250
other_values_append((other_path, other_ie))
1251
other_has_more, other_path, other_ie = \
1252
step_one(other_walker)
1254
# This record isn't in the normal order, see if it
1256
other_values_append(lookup_by_file_id(
1257
other_extra, self._other_trees[idx], file_id))
1258
next_other_entries_append((other_has_more, other_path,
1260
other_entries = next_other_entries
1262
# We've matched all the walkers, yield this datapoint
1263
yield path, file_id, master_ie, other_values
1264
self._other_walkers = other_walkers
1265
self._other_entries = other_entries
1266
self._others_extra = others_extra
1268
def _finish_others(self):
1269
"""Finish walking the other iterators, so we get all entries."""
1270
for idx, info in enumerate(self._other_entries):
1271
other_extra = self._others_extra[idx]
1272
(other_has_more, other_path, other_ie) = info
1273
while other_has_more:
1274
other_file_id = other_ie.file_id
1275
if other_file_id not in self._out_of_order_processed:
1276
other_extra[other_file_id] = (other_path, other_ie)
1277
other_has_more, other_path, other_ie = \
1278
self._step_one(self._other_walkers[idx])
1279
del self._other_entries
1281
def _walk_others(self):
1282
"""Finish up by walking all the 'deferred' nodes."""
1283
# TODO: One alternative would be to grab all possible unprocessed
1284
# file_ids, and then sort by path, and then yield them. That
1285
# might ensure better ordering, in case a caller strictly
1286
# requires parents before children.
1287
for idx, other_extra in enumerate(self._others_extra):
1288
others = sorted(viewvalues(other_extra),
1289
key=lambda x: self._path_to_key(x[0]))
1290
for other_path, other_ie in others:
1291
file_id = other_ie.file_id
1292
# We don't need to check out_of_order_processed here, because
1293
# the lookup_by_file_id will be removing anything processed
1294
# from the extras cache
1295
other_extra.pop(file_id)
1296
other_values = [(None, None)] * idx
1297
other_values.append((other_path, other_ie))
1298
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1299
alt_idx = alt_idx + idx + 1
1300
alt_extra = self._others_extra[alt_idx]
1301
alt_tree = self._other_trees[alt_idx]
1302
other_values.append(self._lookup_by_file_id(
1303
alt_extra, alt_tree, file_id))
1304
yield other_path, file_id, None, other_values
1307
def find_previous_paths(from_tree, to_tree, paths):
1308
"""Find previous tree paths.
1310
:param from_tree: From tree
1311
:param to_tree: To tree
1312
:param paths: Iterable over paths to search for
1313
:return: Dictionary mapping from from_tree paths to paths in to_tree, or
1314
None if there is no equivalent path.
1318
ret[path] = find_previous_path(from_tree, to_tree, path)
1322
def find_previous_path(from_tree, to_tree, path, file_id=None):
1323
"""Find previous tree path.
1325
:param from_tree: From tree
1326
:param to_tree: To tree
1327
:param path: Path to search for
1328
:return: path in to_tree, or None if there is no equivalent path.
1331
file_id = from_tree.path2id(path)
1333
raise errors.NoSuchFile(path)
1335
return to_tree.id2path(file_id)
1336
except errors.NoSuchId:
1340
def get_canonical_path(tree, path, normalize):
1341
"""Find the canonical path of an item, ignoring case.
1343
:param tree: Tree to traverse
1344
:param path: Case-insensitive path to look up
1345
:param normalize: Function to normalize a filename for comparison
1346
:return: The canonical path
1349
cur_id = tree.get_root_id()
1351
bit_iter = iter(path.split("/"))
1352
for elt in bit_iter:
1353
lelt = normalize(elt)
1356
for child in tree.iter_child_entries(cur_path, cur_id):
1358
if child.name == elt:
1359
# if we found an exact match, we can stop now; if
1360
# we found an approximate match we need to keep
1361
# searching because there might be an exact match
1363
cur_id = child.file_id
1364
new_path = osutils.pathjoin(cur_path, child.name)
1366
elif normalize(child.name) == lelt:
1367
cur_id = child.file_id
1368
new_path = osutils.pathjoin(cur_path, child.name)
1369
except errors.NoSuchId:
1370
# before a change is committed we can see this error...
1372
except errors.NotADirectory:
1377
# got to the end of this directory and no entries matched.
1378
# Return what matched so far, plus the rest as specified.
1379
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
added
removed
breezy/tree.py
