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
24
from .lazy_import import lazy_import
25
lazy_import(globals(), """
29
conflicts as _mod_conflicts,
33
revision as _mod_revision,
37
from breezy.i18n import gettext
45
from .inter import InterObject
51
class FileTimestampUnavailable(errors.BzrError):
53
_fmt = "The filestamp for %(path)s is not available."
57
def __init__(self, path):
61
class TreeEntry(object):
62
"""An entry that implements the minimum interface used by commands.
65
def __eq__(self, other):
66
# yes, this is ugly, TODO: best practice __eq__ style.
67
return (isinstance(other, TreeEntry)
68
and other.__class__ == self.__class__)
70
def kind_character(self):
74
class TreeDirectory(TreeEntry):
75
"""See TreeEntry. This is a directory in a working tree."""
77
def kind_character(self):
81
class TreeFile(TreeEntry):
82
"""See TreeEntry. This is a regular file in a working tree."""
84
def kind_character(self):
88
class TreeLink(TreeEntry):
89
"""See TreeEntry. This is a symlink in a working tree."""
91
def kind_character(self):
95
class TreeReference(TreeEntry):
96
"""See TreeEntry. This is a reference to a nested tree in a working tree."""
98
def kind_character(self):
103
"""Abstract file tree.
105
There are several subclasses:
107
* `WorkingTree` exists as files on disk editable by the user.
109
* `RevisionTree` is a tree as recorded at some point in the past.
111
Trees can be compared, etc, regardless of whether they are working
112
trees or versioned trees.
115
def supports_rename_tracking(self):
116
"""Whether this tree supports rename tracking.
118
This defaults to True, but some implementations may want to override
123
def has_versioned_directories(self):
124
"""Whether this tree can contain explicitly versioned directories.
126
This defaults to True, but some implementations may want to override
131
def changes_from(self, other, want_unchanged=False, specific_files=None,
132
extra_trees=None, require_versioned=False, include_root=False,
133
want_unversioned=False):
134
"""Return a TreeDelta of the changes from other to this tree.
136
:param other: A tree to compare with.
137
:param specific_files: An optional list of file paths to restrict the
138
comparison to. When mapping filenames to ids, all matches in all
139
trees (including optional extra_trees) are used, and all children of
140
matched directories are included.
141
:param want_unchanged: An optional boolean requesting the inclusion of
142
unchanged entries in the result.
143
:param extra_trees: An optional list of additional trees to use when
144
mapping the contents of specific_files (paths) to file_ids.
145
:param require_versioned: An optional boolean (defaults to False). When
146
supplied and True all the 'specific_files' must be versioned, or
147
a PathsNotVersionedError will be thrown.
148
:param want_unversioned: Scan for unversioned paths.
150
The comparison will be performed by an InterTree object looked up on
153
# Martin observes that Tree.changes_from returns a TreeDelta and this
154
# may confuse people, because the class name of the returned object is
155
# a synonym of the object referenced in the method name.
156
return InterTree.get(other, self).compare(
157
want_unchanged=want_unchanged,
158
specific_files=specific_files,
159
extra_trees=extra_trees,
160
require_versioned=require_versioned,
161
include_root=include_root,
162
want_unversioned=want_unversioned,
165
def iter_changes(self, from_tree, include_unchanged=False,
166
specific_files=None, pb=None, extra_trees=None,
167
require_versioned=True, want_unversioned=False):
168
"""See InterTree.iter_changes"""
169
intertree = InterTree.get(from_tree, self)
170
return intertree.iter_changes(include_unchanged, specific_files, pb,
171
extra_trees, require_versioned, want_unversioned=want_unversioned)
174
"""Get a list of the conflicts in the tree.
176
Each conflict is an instance of breezy.conflicts.Conflict.
178
return _mod_conflicts.ConflictList()
181
"""For trees that can have unversioned files, return all such paths."""
184
def get_parent_ids(self):
185
"""Get the parent ids for this tree.
187
:return: a list of parent ids. [] is returned to indicate
188
a tree with no parents.
189
:raises: BzrError if the parents are not known.
191
raise NotImplementedError(self.get_parent_ids)
193
def has_filename(self, filename):
194
"""True if the tree has given filename."""
195
raise NotImplementedError(self.has_filename)
197
def is_ignored(self, filename):
198
"""Check whether the filename is ignored by this tree.
200
:param filename: The relative filename within the tree.
201
:return: True if the filename is ignored.
205
def all_file_ids(self):
206
"""Iterate through all file ids, including ids for missing files."""
207
raise NotImplementedError(self.all_file_ids)
209
def all_versioned_paths(self):
210
"""Iterate through all paths, including paths for missing files."""
211
raise NotImplementedError(self.all_versioned_paths)
213
def id2path(self, file_id):
214
"""Return the path for a file id.
218
raise NotImplementedError(self.id2path)
220
def iter_entries_by_dir(self, specific_files=None):
221
"""Walk the tree in 'by_dir' order.
223
This will yield each entry in the tree as a (path, entry) tuple.
224
The order that they are yielded is:
226
Directories are walked in a depth-first lexicographical order,
227
however, whenever a directory is reached, all of its direct child
228
nodes are yielded in lexicographical order before yielding the
231
For example, in the tree::
241
The yield order (ignoring root) would be::
243
a, f, a/b, a/d, a/b/c, a/d/e, f/g
245
raise NotImplementedError(self.iter_entries_by_dir)
247
def iter_child_entries(self, path, file_id=None):
248
"""Iterate over the children of a directory or tree reference.
250
:param path: Path of the directory
251
:param file_id: Optional file id of the directory/tree-reference
252
:raise NoSuchId: When the file_id does not exist
253
:return: Iterator over entries in the directory
255
raise NotImplementedError(self.iter_child_entries)
257
def list_files(self, include_root=False, from_dir=None, recursive=True):
258
"""List all files in this tree.
260
:param include_root: Whether to include the entry for the tree root
261
:param from_dir: Directory under which to list files
262
:param recursive: Whether to list files recursively
263
:return: iterator over tuples of (path, versioned, kind, file_id,
266
raise NotImplementedError(self.list_files)
268
def iter_references(self):
269
if self.supports_tree_reference():
270
for path, entry in self.iter_entries_by_dir():
271
if entry.kind == 'tree-reference':
272
yield path, entry.file_id
274
def kind(self, path, file_id=None):
275
raise NotImplementedError("Tree subclass %s must implement kind"
276
% self.__class__.__name__)
278
def stored_kind(self, path, file_id=None):
279
"""File kind stored for this file_id.
281
May not match kind on disk for working trees. Always available
282
for versioned files, even when the file itself is missing.
284
return self.kind(path, file_id)
286
def path_content_summary(self, path):
287
"""Get a summary of the information about path.
289
All the attributes returned are for the canonical form, not the
290
convenient form (if content filters are in use.)
292
:param path: A relative path within the tree.
293
:return: A tuple containing kind, size, exec, sha1-or-link.
294
Kind is always present (see tree.kind()).
295
size is present if kind is file and the size of the
296
canonical form can be cheaply determined, None otherwise.
297
exec is None unless kind is file and the platform supports the 'x'
299
sha1-or-link is the link target if kind is symlink, or the sha1 if
300
it can be obtained without reading the file.
302
raise NotImplementedError(self.path_content_summary)
304
def get_reference_revision(self, path, file_id=None):
305
raise NotImplementedError("Tree subclass %s must implement "
306
"get_reference_revision"
307
% self.__class__.__name__)
309
def _comparison_data(self, entry, path):
310
"""Return a tuple of kind, executable, stat_value for a file.
312
entry may be None if there is no inventory entry for the file, but
313
path must always be supplied.
315
kind is None if there is no file present (even if an inventory id is
316
present). executable is False for non-file entries.
318
raise NotImplementedError(self._comparison_data)
320
def get_file(self, path, file_id=None):
321
"""Return a file object for the file file_id in the tree.
323
If both file_id and path are defined, it is implementation defined as
324
to which one is used.
326
raise NotImplementedError(self.get_file)
328
def get_file_with_stat(self, path, file_id=None):
329
"""Get a file handle and stat object for file_id.
331
The default implementation returns (self.get_file, None) for backwards
334
:param path: The path of the file.
335
:param file_id: The file id to read, if it is known.
336
:return: A tuple (file_handle, stat_value_or_None). If the tree has
337
no stat facility, or need for a stat cache feedback during commit,
338
it may return None for the second element of the tuple.
340
return (self.get_file(path, file_id), None)
342
def get_file_text(self, path, file_id=None):
343
"""Return the byte content of a file.
345
:param path: The path of the file.
346
:param file_id: The file_id of the file.
348
If both file_id and path are supplied, an implementation may use
351
:returns: A single byte string for the whole file.
353
my_file = self.get_file(path, file_id)
355
return my_file.read()
359
def get_file_lines(self, path, file_id=None):
360
"""Return the content of a file, as lines.
362
:param path: The path of the file.
363
:param file_id: The file_id of the file.
365
If both file_id and path are supplied, an implementation may use
368
return osutils.split_lines(self.get_file_text(path, file_id))
370
def get_file_verifier(self, path, file_id=None, stat_value=None):
371
"""Return a verifier for a file.
373
The default implementation returns a sha1.
375
:param file_id: The handle for this file.
376
:param path: The path that this file can be found at.
377
These must point to the same object.
378
:param stat_value: Optional stat value for the object
379
:return: Tuple with verifier name and verifier data
381
return ("SHA1", self.get_file_sha1(path, file_id,
382
stat_value=stat_value))
384
def get_file_sha1(self, path, file_id=None, stat_value=None):
385
"""Return the SHA1 file for a file.
387
:note: callers should use get_file_verifier instead
388
where possible, as the underlying repository implementation may
389
have quicker access to a non-sha1 verifier.
391
:param path: The path that this file can be found at.
392
:param file_id: The handle for this file.
393
These must point to the same object.
394
:param stat_value: Optional stat value for the object
396
raise NotImplementedError(self.get_file_sha1)
398
def get_file_mtime(self, path, file_id=None):
399
"""Return the modification time for a file.
401
:param path: The path that this file can be found at.
402
:param file_id: The handle for this file.
403
These must point to the same object.
405
raise NotImplementedError(self.get_file_mtime)
407
def get_file_size(self, path, file_id=None):
408
"""Return the size of a file in bytes.
410
This applies only to regular files. If invoked on directories or
411
symlinks, it will return None.
412
:param file_id: The file-id of the file
414
raise NotImplementedError(self.get_file_size)
416
def is_executable(self, path, file_id=None):
417
"""Check if a file is executable.
419
:param path: The path that this file can be found at.
420
:param file_id: The handle for this file.
421
These must point to the same object.
423
raise NotImplementedError(self.is_executable)
425
def iter_files_bytes(self, desired_files):
426
"""Iterate through file contents.
428
Files will not necessarily be returned in the order they occur in
429
desired_files. No specific order is guaranteed.
431
Yields pairs of identifier, bytes_iterator. identifier is an opaque
432
value supplied by the caller as part of desired_files. It should
433
uniquely identify the file version in the caller's context. (Examples:
434
an index number or a TreeTransform trans_id.)
436
bytes_iterator is an iterable of bytestrings for the file. The
437
kind of iterable and length of the bytestrings are unspecified, but for
438
this implementation, it is a tuple containing a single bytestring with
439
the complete text of the file.
441
:param desired_files: a list of (path, identifier) pairs
443
for path, identifier in desired_files:
444
# We wrap the string in a tuple so that we can return an iterable
445
# of bytestrings. (Technically, a bytestring is also an iterable
446
# of bytestrings, but iterating through each character is not
448
cur_file = (self.get_file_text(path),)
449
yield identifier, cur_file
451
def get_symlink_target(self, path, file_id=None):
452
"""Get the target for a given file_id.
454
It is assumed that the caller already knows that file_id is referencing
456
:param file_id: Handle for the symlink entry.
457
:param path: The path of the file.
458
If both file_id and path are supplied, an implementation may use
460
:return: The path the symlink points to.
462
raise NotImplementedError(self.get_symlink_target)
464
def get_root_id(self):
465
"""Return the file_id for the root of this tree."""
466
raise NotImplementedError(self.get_root_id)
468
def annotate_iter(self, path, file_id=None,
469
default_revision=_mod_revision.CURRENT_REVISION):
470
"""Return an iterator of revision_id, line tuples.
472
For working trees (and mutable trees in general), the special
473
revision_id 'current:' will be used for lines that are new in this
474
tree, e.g. uncommitted changes.
475
:param file_id: The file to produce an annotated version from
476
:param default_revision: For lines that don't match a basis, mark them
477
with this revision id. Not all implementations will make use of
480
raise NotImplementedError(self.annotate_iter)
482
def _iter_parent_trees(self):
483
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
484
for revision_id in self.get_parent_ids():
486
yield self.revision_tree(revision_id)
487
except errors.NoSuchRevisionInTree:
488
yield self.repository.revision_tree(revision_id)
490
def path2id(self, path):
491
"""Return the id for path in this tree."""
492
raise NotImplementedError(self.path2id)
494
def is_versioned(self, path):
495
"""Check whether path is versioned.
497
:param path: Path to check
500
return self.path2id(path) is not None
502
def find_related_paths_across_trees(self, paths, trees=[],
503
require_versioned=True):
504
"""Find related paths in tree corresponding to specified filenames in any
507
All matches in all trees will be used, and all children of matched
508
directories will be used.
510
:param paths: The filenames to find related paths for (if None, returns
512
:param trees: The trees to find file_ids within
513
:param require_versioned: if true, all specified filenames must occur in
515
:return: a set of paths for the specified filenames and their children
518
raise NotImplementedError(self.find_related_paths_across_trees)
521
"""Lock this tree for multiple read only operations.
523
:return: A breezy.lock.LogicalLockResult.
525
return lock.LogicalLockResult(self.unlock)
527
def revision_tree(self, revision_id):
528
"""Obtain a revision tree for the revision revision_id.
530
The intention of this method is to allow access to possibly cached
531
tree data. Implementors of this method should raise NoSuchRevision if
532
the tree is not locally available, even if they could obtain the
533
tree via a repository or some other means. Callers are responsible
534
for finding the ultimate source for a revision tree.
536
:param revision_id: The revision_id of the requested tree.
538
:raises: NoSuchRevision if the tree cannot be obtained.
540
raise errors.NoSuchRevisionInTree(self, revision_id)
543
"""What files are present in this tree and unknown.
545
:return: an iterator over the unknown files.
552
def filter_unversioned_files(self, paths):
553
"""Filter out paths that are versioned.
555
:return: set of paths.
557
raise NotImplementedError(self.filter_unversioned_files)
559
def walkdirs(self, prefix=""):
560
"""Walk the contents of this tree from path down.
562
This yields all the data about the contents of a directory at a time.
563
After each directory has been yielded, if the caller has mutated the
564
list to exclude some directories, they are then not descended into.
566
The data yielded is of the form:
567
((directory-relpath, directory-path-from-root, directory-fileid),
568
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
569
versioned_kind), ...]),
570
- directory-relpath is the containing dirs relpath from prefix
571
- directory-path-from-root is the containing dirs path from /
572
- directory-fileid is the id of the directory if it is versioned.
573
- relpath is the relative path within the subtree being walked.
574
- basename is the basename
575
- kind is the kind of the file now. If unknonwn then the file is not
576
present within the tree - but it may be recorded as versioned. See
578
- lstat is the stat data *if* the file was statted.
579
- path_from_tree_root is the path from the root of the tree.
580
- file_id is the file_id if the entry is versioned.
581
- versioned_kind is the kind of the file as last recorded in the
582
versioning system. If 'unknown' the file is not versioned.
583
One of 'kind' and 'versioned_kind' must not be 'unknown'.
585
:param prefix: Start walking from prefix within the tree rather than
586
at the root. This allows one to walk a subtree but get paths that are
587
relative to a tree rooted higher up.
588
:return: an iterator over the directory data.
590
raise NotImplementedError(self.walkdirs)
592
def supports_content_filtering(self):
595
def _content_filter_stack(self, path=None):
596
"""The stack of content filters for a path if filtering is supported.
598
Readers will be applied in first-to-last order.
599
Writers will be applied in last-to-first order.
600
Either the path or the file-id needs to be provided.
602
:param path: path relative to the root of the tree
604
:return: the list of filters - [] if there are none
606
filter_pref_names = filters._get_registered_names()
607
if len(filter_pref_names) == 0:
609
prefs = next(self.iter_search_rules([path], filter_pref_names))
610
stk = filters._get_filter_stack_for(prefs)
611
if 'filters' in debug.debug_flags:
612
trace.note(gettext("*** {0} content-filter: {1} => {2!r}").format(path, prefs, stk))
615
def _content_filter_stack_provider(self):
616
"""A function that returns a stack of ContentFilters.
618
The function takes a path (relative to the top of the tree) and a
619
file-id as parameters.
621
:return: None if content filtering is not supported by this tree.
623
if self.supports_content_filtering():
624
return lambda path, file_id: \
625
self._content_filter_stack(path)
629
def iter_search_rules(self, path_names, pref_names=None,
630
_default_searcher=None):
631
"""Find the preferences for filenames in a tree.
633
:param path_names: an iterable of paths to find attributes for.
634
Paths are given relative to the root of the tree.
635
:param pref_names: the list of preferences to lookup - None for all
636
:param _default_searcher: private parameter to assist testing - don't use
637
:return: an iterator of tuple sequences, one per path-name.
638
See _RulesSearcher.get_items for details on the tuple sequence.
640
if _default_searcher is None:
641
_default_searcher = rules._per_user_searcher
642
searcher = self._get_rules_searcher(_default_searcher)
643
if searcher is not None:
644
if pref_names is not None:
645
for path in path_names:
646
yield searcher.get_selected_items(path, pref_names)
648
for path in path_names:
649
yield searcher.get_items(path)
651
def _get_rules_searcher(self, default_searcher):
652
"""Get the RulesSearcher for this tree given the default one."""
653
searcher = default_searcher
656
def archive(self, name, format=None, root=None, subdir=None):
657
"""Create an archive of this tree.
659
:param name: target file name
660
:param format: Format name (e.g. 'tar')
661
:param root: Root directory name (or None)
662
:param subdir: Subdirectory to export (or None)
663
:return: Iterator over archive chunks
665
with self.lock_read():
666
from .export import get_stream_export_generator
667
return get_stream_export_generator(self, name, format, root,
671
def versionable_kind(cls, kind):
672
"""Check if this tree support versioning a specific file kind."""
673
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
676
class InterTree(InterObject):
677
"""This class represents operations taking place between two Trees.
679
Its instances have methods like 'compare' and contain references to the
680
source and target trees these operations are to be carried out on.
682
Clients of breezy should not need to use InterTree directly, rather they
683
should use the convenience methods on Tree such as 'Tree.compare()' which
684
will pass through to InterTree as appropriate.
687
# Formats that will be used to test this InterTree. If both are
688
# None, this InterTree will not be tested (e.g. because a complex
690
_matching_from_tree_format = None
691
_matching_to_tree_format = None
696
def is_compatible(kls, source, target):
697
# The default implementation is naive and uses the public API, so
698
# it works for all trees.
701
def _changes_from_entries(self, source_entry, target_entry, source_path,
703
"""Generate a iter_changes tuple between source_entry and target_entry.
705
:param source_entry: An inventory entry from self.source, or None.
706
:param target_entry: An inventory entry from self.target, or None.
707
:param source_path: The path of source_entry.
708
:param target_path: The path of target_entry.
709
:return: A tuple, item 0 of which is an iter_changes result tuple, and
710
item 1 is True if there are any changes in the result tuple.
712
if source_entry is None:
713
if target_entry is None:
715
file_id = target_entry.file_id
717
file_id = source_entry.file_id
718
if source_entry is not None:
719
source_versioned = True
720
source_name = source_entry.name
721
source_parent = source_entry.parent_id
722
source_kind, source_executable, source_stat = \
723
self.source._comparison_data(source_entry, source_path)
725
source_versioned = False
729
source_executable = None
730
if target_entry is not None:
731
target_versioned = True
732
target_name = target_entry.name
733
target_parent = target_entry.parent_id
734
target_kind, target_executable, target_stat = \
735
self.target._comparison_data(target_entry, target_path)
737
target_versioned = False
741
target_executable = None
742
versioned = (source_versioned, target_versioned)
743
kind = (source_kind, target_kind)
744
changed_content = False
745
if source_kind != target_kind:
746
changed_content = True
747
elif source_kind == 'file':
748
if not self.file_content_matches(
749
source_path, target_path,
750
file_id, file_id, source_stat, target_stat):
751
changed_content = True
752
elif source_kind == 'symlink':
753
if (self.source.get_symlink_target(source_path, file_id) !=
754
self.target.get_symlink_target(target_path, file_id)):
755
changed_content = True
756
elif source_kind == 'tree-reference':
757
if (self.source.get_reference_revision(source_path, file_id)
758
!= self.target.get_reference_revision(target_path, file_id)):
759
changed_content = True
760
parent = (source_parent, target_parent)
761
name = (source_name, target_name)
762
executable = (source_executable, target_executable)
763
if (changed_content is not False or versioned[0] != versioned[1]
764
or parent[0] != parent[1] or name[0] != name[1] or
765
executable[0] != executable[1]):
769
return (file_id, (source_path, target_path), changed_content,
770
versioned, parent, name, kind, executable), changes
772
def compare(self, want_unchanged=False, specific_files=None,
773
extra_trees=None, require_versioned=False, include_root=False,
774
want_unversioned=False):
775
"""Return the changes from source to target.
777
:return: A TreeDelta.
778
:param specific_files: An optional list of file paths to restrict the
779
comparison to. When mapping filenames to ids, all matches in all
780
trees (including optional extra_trees) are used, and all children of
781
matched directories are included.
782
:param want_unchanged: An optional boolean requesting the inclusion of
783
unchanged entries in the result.
784
:param extra_trees: An optional list of additional trees to use when
785
mapping the contents of specific_files (paths) to file_ids.
786
:param require_versioned: An optional boolean (defaults to False). When
787
supplied and True all the 'specific_files' must be versioned, or
788
a PathsNotVersionedError will be thrown.
789
:param want_unversioned: Scan for unversioned paths.
791
trees = (self.source,)
792
if extra_trees is not None:
793
trees = trees + tuple(extra_trees)
794
with self.lock_read():
795
return delta._compare_trees(self.source, self.target, want_unchanged,
796
specific_files, include_root, extra_trees=extra_trees,
797
require_versioned=require_versioned,
798
want_unversioned=want_unversioned)
800
def iter_changes(self, include_unchanged=False,
801
specific_files=None, pb=None, extra_trees=[],
802
require_versioned=True, want_unversioned=False):
803
"""Generate an iterator of changes between trees.
806
(file_id, (path_in_source, path_in_target),
807
changed_content, versioned, parent, name, kind,
810
Changed_content is True if the file's content has changed. This
811
includes changes to its kind, and to a symlink's target.
813
versioned, parent, name, kind, executable are tuples of (from, to).
814
If a file is missing in a tree, its kind is None.
816
Iteration is done in parent-to-child order, relative to the target
819
There is no guarantee that all paths are in sorted order: the
820
requirement to expand the search due to renames may result in children
821
that should be found early being found late in the search, after
822
lexically later results have been returned.
823
:param require_versioned: Raise errors.PathsNotVersionedError if a
824
path in the specific_files list is not versioned in one of
825
source, target or extra_trees.
826
:param specific_files: An optional list of file paths to restrict the
827
comparison to. When mapping filenames to ids, all matches in all
828
trees (including optional extra_trees) are used, and all children
829
of matched directories are included. The parents in the target tree
830
of the specific files up to and including the root of the tree are
831
always evaluated for changes too.
832
:param want_unversioned: Should unversioned files be returned in the
833
output. An unversioned file is defined as one with (False, False)
834
for the versioned pair.
839
extra_trees = list(extra_trees)
840
# The ids of items we need to examine to insure delta consistency.
841
precise_file_ids = set()
842
changed_file_ids = []
843
if specific_files == []:
844
target_specific_files = []
845
source_specific_files = []
847
target_specific_files = self.target.find_related_paths_across_trees(
848
specific_files, [self.source] + extra_trees,
849
require_versioned=require_versioned)
850
source_specific_files = self.source.find_related_paths_across_trees(
851
specific_files, [self.target] + extra_trees,
852
require_versioned=require_versioned)
853
if specific_files is not None:
854
# reparented or added entries must have their parents included
855
# so that valid deltas can be created. The seen_parents set
856
# tracks the parents that we need to have.
857
# The seen_dirs set tracks directory entries we've yielded.
858
# After outputting version object in to_entries we set difference
859
# the two seen sets and start checking parents.
863
all_unversioned = sorted([(p.split('/'), p) for p in
865
if specific_files is None or
866
osutils.is_inside_any(specific_files, p)])
867
all_unversioned = collections.deque(all_unversioned)
869
all_unversioned = collections.deque()
871
from_entries_by_dir = list(self.source.iter_entries_by_dir(
872
specific_files=source_specific_files))
873
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
874
to_entries_by_dir = list(self.target.iter_entries_by_dir(
875
specific_files=target_specific_files))
876
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
878
# the unversioned path lookup only occurs on real trees - where there
879
# can be extras. So the fake_entry is solely used to look up
880
# executable it values when execute is not supported.
881
fake_entry = TreeFile()
882
for target_path, target_entry in to_entries_by_dir:
883
while (all_unversioned and
884
all_unversioned[0][0] < target_path.split('/')):
885
unversioned_path = all_unversioned.popleft()
886
target_kind, target_executable, target_stat = \
887
self.target._comparison_data(fake_entry, unversioned_path[1])
888
yield (None, (None, unversioned_path[1]), True, (False, False),
890
(None, unversioned_path[0][-1]),
892
(None, target_executable))
893
source_path, source_entry = from_data.get(target_entry.file_id,
895
result, changes = self._changes_from_entries(source_entry,
896
target_entry, source_path=source_path, target_path=target_path)
897
to_paths[result[0]] = result[1][1]
902
pb.update('comparing files', entry_count, num_entries)
903
if changes or include_unchanged:
904
if specific_files is not None:
905
new_parent_id = result[4][1]
906
precise_file_ids.add(new_parent_id)
907
changed_file_ids.append(result[0])
909
# Ensure correct behaviour for reparented/added specific files.
910
if specific_files is not None:
912
if result[6][1] == 'directory':
913
seen_dirs.add(result[0])
914
# Record parents of reparented/added entries.
915
versioned = result[3]
917
if not versioned[0] or parents[0] != parents[1]:
918
seen_parents.add(parents[1])
919
while all_unversioned:
920
# yield any trailing unversioned paths
921
unversioned_path = all_unversioned.popleft()
922
to_kind, to_executable, to_stat = \
923
self.target._comparison_data(fake_entry, unversioned_path[1])
924
yield (None, (None, unversioned_path[1]), True, (False, False),
926
(None, unversioned_path[0][-1]),
928
(None, to_executable))
929
# Yield all remaining source paths
930
for path, from_entry in from_entries_by_dir:
931
file_id = from_entry.file_id
932
if file_id in to_paths:
935
to_path = find_previous_path(self.source, self.target, path)
938
pb.update('comparing files', entry_count, num_entries)
939
versioned = (True, False)
940
parent = (from_entry.parent_id, None)
941
name = (from_entry.name, None)
942
from_kind, from_executable, stat_value = \
943
self.source._comparison_data(from_entry, path)
944
kind = (from_kind, None)
945
executable = (from_executable, None)
946
changed_content = from_kind is not None
947
# the parent's path is necessarily known at this point.
948
changed_file_ids.append(file_id)
949
yield(file_id, (path, to_path), changed_content, versioned, parent,
950
name, kind, executable)
951
changed_file_ids = set(changed_file_ids)
952
if specific_files is not None:
953
for result in self._handle_precise_ids(precise_file_ids,
957
def _get_entry(self, tree, path):
958
"""Get an inventory entry from a tree, with missing entries as None.
960
If the tree raises NotImplementedError on accessing .inventory, then
961
this is worked around using iter_entries_by_dir on just the file id
964
:param tree: The tree to lookup the entry in.
965
:param file_id: The file_id to lookup.
967
# No inventory available.
969
iterator = tree.iter_entries_by_dir(specific_files=[path])
970
return iterator.next()[1]
971
except StopIteration:
974
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
975
discarded_changes=None):
976
"""Fill out a partial iter_changes to be consistent.
978
:param precise_file_ids: The file ids of parents that were seen during
980
:param changed_file_ids: The file ids of already emitted items.
981
:param discarded_changes: An optional dict of precalculated
982
iter_changes items which the partial iter_changes had not output
984
:return: A generator of iter_changes items to output.
986
# process parents of things that had changed under the users
987
# requested paths to prevent incorrect paths or parent ids which
988
# aren't in the tree.
989
while precise_file_ids:
990
precise_file_ids.discard(None)
991
# Don't emit file_ids twice
992
precise_file_ids.difference_update(changed_file_ids)
993
if not precise_file_ids:
995
# If the there was something at a given output path in source, we
996
# have to include the entry from source in the delta, or we would
997
# be putting this entry into a used path.
999
for parent_id in precise_file_ids:
1001
paths.append(self.target.id2path(parent_id))
1002
except errors.NoSuchId:
1003
# This id has been dragged in from the source by delta
1004
# expansion and isn't present in target at all: we don't
1005
# need to check for path collisions on it.
1008
old_id = self.source.path2id(path)
1009
precise_file_ids.add(old_id)
1010
precise_file_ids.discard(None)
1011
current_ids = precise_file_ids
1012
precise_file_ids = set()
1013
# We have to emit all of precise_file_ids that have been altered.
1014
# We may have to output the children of some of those ids if any
1015
# directories have stopped being directories.
1016
for file_id in current_ids:
1018
if discarded_changes:
1019
result = discarded_changes.get(file_id)
1025
source_path = self.source.id2path(file_id)
1026
except errors.NoSuchId:
1030
source_entry = self._get_entry(self.source, source_path)
1032
target_path = self.target.id2path(file_id)
1033
except errors.NoSuchId:
1037
target_entry = self._get_entry(self.target, target_path)
1038
result, changes = self._changes_from_entries(
1039
source_entry, target_entry, source_path, target_path)
1042
# Get this parents parent to examine.
1043
new_parent_id = result[4][1]
1044
precise_file_ids.add(new_parent_id)
1046
if (result[6][0] == 'directory' and
1047
result[6][1] != 'directory'):
1048
# This stopped being a directory, the old children have
1050
if source_entry is None:
1051
# Reusing a discarded change.
1052
source_entry = self._get_entry(self.source, result[1][0])
1053
precise_file_ids.update(
1055
for child in self.source.iter_child_entries(result[1][0]))
1056
changed_file_ids.add(result[0])
1059
def file_content_matches(
1060
self, source_path, target_path,
1061
source_file_id=None, target_file_id=None,
1062
source_stat=None, target_stat=None):
1063
"""Check if two files are the same in the source and target trees.
1065
This only checks that the contents of the files are the same,
1066
it does not touch anything else.
1068
:param source_path: Path of the file in the source tree
1069
:param target_path: Path of the file in the target tree
1070
:param source_file_id: Optional file id of the file in the source tree
1071
:param target_file_id: Optional file id of the file in the target tree
1072
:param source_stat: Optional stat value of the file in the source tree
1073
:param target_stat: Optional stat value of the file in the target tree
1074
:return: Boolean indicating whether the files have the same contents
1076
with self.lock_read():
1077
source_verifier_kind, source_verifier_data = (
1078
self.source.get_file_verifier(
1079
source_path, source_file_id, source_stat))
1080
target_verifier_kind, target_verifier_data = (
1081
self.target.get_file_verifier(
1082
target_path, target_file_id, target_stat))
1083
if source_verifier_kind == target_verifier_kind:
1084
return (source_verifier_data == target_verifier_data)
1085
# Fall back to SHA1 for now
1086
if source_verifier_kind != "SHA1":
1087
source_sha1 = self.source.get_file_sha1(
1088
source_path, source_file_id, source_stat)
1090
source_sha1 = source_verifier_data
1091
if target_verifier_kind != "SHA1":
1092
target_sha1 = self.target.get_file_sha1(
1093
target_path, target_file_id, target_stat)
1095
target_sha1 = target_verifier_data
1096
return (source_sha1 == target_sha1)
1098
InterTree.register_optimiser(InterTree)
1101
class MultiWalker(object):
1102
"""Walk multiple trees simultaneously, getting combined results."""
1104
# Note: This could be written to not assume you can do out-of-order
1105
# lookups. Instead any nodes that don't match in all trees could be
1106
# marked as 'deferred', and then returned in the final cleanup loop.
1107
# For now, I think it is "nicer" to return things as close to the
1108
# "master_tree" order as we can.
1110
def __init__(self, master_tree, other_trees):
1111
"""Create a new MultiWalker.
1113
All trees being walked must implement "iter_entries_by_dir()", such
1114
that they yield (path, object) tuples, where that object will have a
1115
'.file_id' member, that can be used to check equality.
1117
:param master_tree: All trees will be 'slaved' to the master_tree such
1118
that nodes in master_tree will be used as 'first-pass' sync points.
1119
Any nodes that aren't in master_tree will be merged in a second
1121
:param other_trees: A list of other trees to walk simultaneously.
1123
self._master_tree = master_tree
1124
self._other_trees = other_trees
1126
# Keep track of any nodes that were properly processed just out of
1127
# order, that way we don't return them at the end, we don't have to
1128
# track *all* processed file_ids, just the out-of-order ones
1129
self._out_of_order_processed = set()
1132
def _step_one(iterator):
1133
"""Step an iter_entries_by_dir iterator.
1135
:return: (has_more, path, ie)
1136
If has_more is False, path and ie will be None.
1139
path, ie = next(iterator)
1140
except StopIteration:
1141
return False, None, None
1143
return True, path, ie
1146
def _cmp_path_by_dirblock(path1, path2):
1147
"""Compare two paths based on what directory they are in.
1149
This generates a sort order, such that all children of a directory are
1150
sorted together, and grandchildren are in the same order as the
1151
children appear. But all grandchildren come after all children.
1153
:param path1: first path
1154
:param path2: the second path
1155
:return: negative number if ``path1`` comes first,
1156
0 if paths are equal
1157
and a positive number if ``path2`` sorts first
1159
# Shortcut this special case
1162
# This is stolen from _dirstate_helpers_py.py, only switching it to
1163
# Unicode objects. Consider using encode_utf8() and then using the
1164
# optimized versions, or maybe writing optimized unicode versions.
1165
if not isinstance(path1, unicode):
1166
raise TypeError("'path1' must be a unicode string, not %s: %r"
1167
% (type(path1), path1))
1168
if not isinstance(path2, unicode):
1169
raise TypeError("'path2' must be a unicode string, not %s: %r"
1170
% (type(path2), path2))
1171
return cmp(MultiWalker._path_to_key(path1),
1172
MultiWalker._path_to_key(path2))
1175
def _path_to_key(path):
1176
dirname, basename = osutils.split(path)
1177
return (dirname.split(u'/'), basename)
1179
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1180
"""Lookup an inventory entry by file_id.
1182
This is called when an entry is missing in the normal order.
1183
Generally this is because a file was either renamed, or it was
1184
deleted/added. If the entry was found in the inventory and not in
1185
extra_entries, it will be added to self._out_of_order_processed
1187
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1188
should be filled with entries that were found before they were
1189
used. If file_id is present, it will be removed from the
1191
:param other_tree: The Tree to search, in case we didn't find the entry
1193
:param file_id: The file_id to look for
1194
:return: (path, ie) if found or (None, None) if not present.
1196
if file_id in extra_entries:
1197
return extra_entries.pop(file_id)
1198
# TODO: Is id2path better as the first call, or is
1199
# inventory[file_id] better as a first check?
1201
cur_path = other_tree.id2path(file_id)
1202
except errors.NoSuchId:
1204
if cur_path is None:
1207
self._out_of_order_processed.add(file_id)
1208
cur_ie = other_tree.root_inventory.get_entry(file_id)
1209
return (cur_path, cur_ie)
1212
"""Match up the values in the different trees."""
1213
for result in self._walk_master_tree():
1215
self._finish_others()
1216
for result in self._walk_others():
1219
def _walk_master_tree(self):
1220
"""First pass, walk all trees in lock-step.
1222
When we are done, all nodes in the master_tree will have been
1223
processed. _other_walkers, _other_entries, and _others_extra will be
1224
set on 'self' for future processing.
1226
# This iterator has the most "inlining" done, because it tends to touch
1227
# every file in the tree, while the others only hit nodes that don't
1229
master_iterator = self._master_tree.iter_entries_by_dir()
1231
other_walkers = [other.iter_entries_by_dir()
1232
for other in self._other_trees]
1233
other_entries = [self._step_one(walker) for walker in other_walkers]
1234
# Track extra nodes in the other trees
1235
others_extra = [{} for _ in range(len(self._other_trees))]
1237
master_has_more = True
1238
step_one = self._step_one
1239
lookup_by_file_id = self._lookup_by_file_id
1240
out_of_order_processed = self._out_of_order_processed
1242
while master_has_more:
1243
(master_has_more, path, master_ie) = step_one(master_iterator)
1244
if not master_has_more:
1247
file_id = master_ie.file_id
1249
other_values_append = other_values.append
1250
next_other_entries = []
1251
next_other_entries_append = next_other_entries.append
1252
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1253
if not other_has_more:
1254
other_values_append(lookup_by_file_id(
1255
others_extra[idx], self._other_trees[idx], file_id))
1256
next_other_entries_append((False, None, None))
1257
elif file_id == other_ie.file_id:
1258
# This is the critical code path, as most of the entries
1259
# should match between most trees.
1260
other_values_append((other_path, other_ie))
1261
next_other_entries_append(step_one(other_walkers[idx]))
1263
# This walker did not match, step it until it either
1264
# matches, or we know we are past the current walker.
1265
other_walker = other_walkers[idx]
1266
other_extra = others_extra[idx]
1267
while (other_has_more and
1268
self._cmp_path_by_dirblock(other_path, path) < 0):
1269
other_file_id = other_ie.file_id
1270
if other_file_id not in out_of_order_processed:
1271
other_extra[other_file_id] = (other_path, other_ie)
1272
other_has_more, other_path, other_ie = \
1273
step_one(other_walker)
1274
if other_has_more and other_ie.file_id == file_id:
1275
# We ended up walking to this point, match and step
1277
other_values_append((other_path, other_ie))
1278
other_has_more, other_path, other_ie = \
1279
step_one(other_walker)
1281
# This record isn't in the normal order, see if it
1283
other_values_append(lookup_by_file_id(
1284
other_extra, self._other_trees[idx], file_id))
1285
next_other_entries_append((other_has_more, other_path,
1287
other_entries = next_other_entries
1289
# We've matched all the walkers, yield this datapoint
1290
yield path, file_id, master_ie, other_values
1291
self._other_walkers = other_walkers
1292
self._other_entries = other_entries
1293
self._others_extra = others_extra
1295
def _finish_others(self):
1296
"""Finish walking the other iterators, so we get all entries."""
1297
for idx, info in enumerate(self._other_entries):
1298
other_extra = self._others_extra[idx]
1299
(other_has_more, other_path, other_ie) = info
1300
while other_has_more:
1301
other_file_id = other_ie.file_id
1302
if other_file_id not in self._out_of_order_processed:
1303
other_extra[other_file_id] = (other_path, other_ie)
1304
other_has_more, other_path, other_ie = \
1305
self._step_one(self._other_walkers[idx])
1306
del self._other_entries
1308
def _walk_others(self):
1309
"""Finish up by walking all the 'deferred' nodes."""
1310
# TODO: One alternative would be to grab all possible unprocessed
1311
# file_ids, and then sort by path, and then yield them. That
1312
# might ensure better ordering, in case a caller strictly
1313
# requires parents before children.
1314
for idx, other_extra in enumerate(self._others_extra):
1315
others = sorted(viewvalues(other_extra),
1316
key=lambda x: self._path_to_key(x[0]))
1317
for other_path, other_ie in others:
1318
file_id = other_ie.file_id
1319
# We don't need to check out_of_order_processed here, because
1320
# the lookup_by_file_id will be removing anything processed
1321
# from the extras cache
1322
other_extra.pop(file_id)
1323
other_values = [(None, None)] * idx
1324
other_values.append((other_path, other_ie))
1325
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1326
alt_idx = alt_idx + idx + 1
1327
alt_extra = self._others_extra[alt_idx]
1328
alt_tree = self._other_trees[alt_idx]
1329
other_values.append(self._lookup_by_file_id(
1330
alt_extra, alt_tree, file_id))
1331
yield other_path, file_id, None, other_values
1334
def find_previous_paths(from_tree, to_tree, paths):
1335
"""Find previous tree paths.
1337
:param from_tree: From tree
1338
:param to_tree: To tree
1339
:param paths: Iterable over paths to search for
1340
:return: Dictionary mapping from from_tree paths to paths in to_tree, or
1341
None if there is no equivalent path.
1345
ret[path] = find_previous_path(from_tree, to_tree, path)
1349
def find_previous_path(from_tree, to_tree, path, file_id=None):
1350
"""Find previous tree path.
1352
:param from_tree: From tree
1353
:param to_tree: To tree
1354
:param path: Path to search for
1355
:return: path in to_tree, or None if there is no equivalent path.
1358
file_id = from_tree.path2id(path)
1360
raise errors.NoSuchFile(path)
1362
return to_tree.id2path(file_id)
1363
except errors.NoSuchId:
added
removed
breezy/tree.py
