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
with self.get_file(path, file_id) as my_file:
354
return my_file.read()
356
def get_file_lines(self, path, file_id=None):
357
"""Return the content of a file, as lines.
359
:param path: The path of the file.
360
:param file_id: The file_id of the file.
362
If both file_id and path are supplied, an implementation may use
365
return osutils.split_lines(self.get_file_text(path, file_id))
367
def get_file_verifier(self, path, file_id=None, stat_value=None):
368
"""Return a verifier for a file.
370
The default implementation returns a sha1.
372
:param file_id: The handle for this file.
373
:param path: The path that this file can be found at.
374
These must point to the same object.
375
:param stat_value: Optional stat value for the object
376
:return: Tuple with verifier name and verifier data
378
return ("SHA1", self.get_file_sha1(path, file_id,
379
stat_value=stat_value))
381
def get_file_sha1(self, path, file_id=None, stat_value=None):
382
"""Return the SHA1 file for a file.
384
:note: callers should use get_file_verifier instead
385
where possible, as the underlying repository implementation may
386
have quicker access to a non-sha1 verifier.
388
:param path: The path that this file can be found at.
389
:param file_id: The handle for this file.
390
These must point to the same object.
391
:param stat_value: Optional stat value for the object
393
raise NotImplementedError(self.get_file_sha1)
395
def get_file_mtime(self, path, file_id=None):
396
"""Return the modification time for a file.
398
:param path: The path that this file can be found at.
399
:param file_id: The handle for this file.
400
These must point to the same object.
402
raise NotImplementedError(self.get_file_mtime)
404
def get_file_size(self, path, file_id=None):
405
"""Return the size of a file in bytes.
407
This applies only to regular files. If invoked on directories or
408
symlinks, it will return None.
409
:param file_id: The file-id of the file
411
raise NotImplementedError(self.get_file_size)
413
def is_executable(self, path, file_id=None):
414
"""Check if a file is executable.
416
:param path: The path that this file can be found at.
417
:param file_id: The handle for this file.
418
These must point to the same object.
420
raise NotImplementedError(self.is_executable)
422
def iter_files_bytes(self, desired_files):
423
"""Iterate through file contents.
425
Files will not necessarily be returned in the order they occur in
426
desired_files. No specific order is guaranteed.
428
Yields pairs of identifier, bytes_iterator. identifier is an opaque
429
value supplied by the caller as part of desired_files. It should
430
uniquely identify the file version in the caller's context. (Examples:
431
an index number or a TreeTransform trans_id.)
433
bytes_iterator is an iterable of bytestrings for the file. The
434
kind of iterable and length of the bytestrings are unspecified, but for
435
this implementation, it is a tuple containing a single bytestring with
436
the complete text of the file.
438
:param desired_files: a list of (path, identifier) pairs
440
for path, identifier in desired_files:
441
# We wrap the string in a tuple so that we can return an iterable
442
# of bytestrings. (Technically, a bytestring is also an iterable
443
# of bytestrings, but iterating through each character is not
445
cur_file = (self.get_file_text(path),)
446
yield identifier, cur_file
448
def get_symlink_target(self, path, file_id=None):
449
"""Get the target for a given file_id.
451
It is assumed that the caller already knows that file_id is referencing
453
:param file_id: Handle for the symlink entry.
454
:param path: The path of the file.
455
If both file_id and path are supplied, an implementation may use
457
:return: The path the symlink points to.
459
raise NotImplementedError(self.get_symlink_target)
461
def get_root_id(self):
462
"""Return the file_id for the root of this tree."""
463
raise NotImplementedError(self.get_root_id)
465
def annotate_iter(self, path, file_id=None,
466
default_revision=_mod_revision.CURRENT_REVISION):
467
"""Return an iterator of revision_id, line tuples.
469
For working trees (and mutable trees in general), the special
470
revision_id 'current:' will be used for lines that are new in this
471
tree, e.g. uncommitted changes.
472
:param file_id: The file to produce an annotated version from
473
:param default_revision: For lines that don't match a basis, mark them
474
with this revision id. Not all implementations will make use of
477
raise NotImplementedError(self.annotate_iter)
479
def _iter_parent_trees(self):
480
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
481
for revision_id in self.get_parent_ids():
483
yield self.revision_tree(revision_id)
484
except errors.NoSuchRevisionInTree:
485
yield self.repository.revision_tree(revision_id)
487
def path2id(self, path):
488
"""Return the id for path in this tree."""
489
raise NotImplementedError(self.path2id)
491
def is_versioned(self, path):
492
"""Check whether path is versioned.
494
:param path: Path to check
497
return self.path2id(path) is not None
499
def find_related_paths_across_trees(self, paths, trees=[],
500
require_versioned=True):
501
"""Find related paths in tree corresponding to specified filenames in any
504
All matches in all trees will be used, and all children of matched
505
directories will be used.
507
:param paths: The filenames to find related paths for (if None, returns
509
:param trees: The trees to find file_ids within
510
:param require_versioned: if true, all specified filenames must occur in
512
:return: a set of paths for the specified filenames and their children
515
raise NotImplementedError(self.find_related_paths_across_trees)
518
"""Lock this tree for multiple read only operations.
520
:return: A breezy.lock.LogicalLockResult.
522
return lock.LogicalLockResult(self.unlock)
524
def revision_tree(self, revision_id):
525
"""Obtain a revision tree for the revision revision_id.
527
The intention of this method is to allow access to possibly cached
528
tree data. Implementors of this method should raise NoSuchRevision if
529
the tree is not locally available, even if they could obtain the
530
tree via a repository or some other means. Callers are responsible
531
for finding the ultimate source for a revision tree.
533
:param revision_id: The revision_id of the requested tree.
535
:raises: NoSuchRevision if the tree cannot be obtained.
537
raise errors.NoSuchRevisionInTree(self, revision_id)
540
"""What files are present in this tree and unknown.
542
:return: an iterator over the unknown files.
549
def filter_unversioned_files(self, paths):
550
"""Filter out paths that are versioned.
552
:return: set of paths.
554
# NB: we specifically *don't* call self.has_filename, because for
555
# WorkingTrees that can indicate files that exist on disk but that
557
return set(p for p in paths if not self.is_versioned(p))
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, format, name, root='', subdir=None,
658
"""Create an archive of this tree.
660
:param name: target file name
661
:param format: Format name (e.g. 'tar')
662
:param root: Root directory name (or None)
663
:param subdir: Subdirectory to export (or None)
664
:param per_file_timestamps: Whether to set the timestamp
665
for each file to the last changed time.
666
:return: Iterator over archive chunks
668
from .archive import create_archive
669
with self.lock_read():
670
return create_archive(format, self, name, root,
671
subdir, force_mtime=force_mtime)
674
def versionable_kind(cls, kind):
675
"""Check if this tree support versioning a specific file kind."""
676
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
679
class InterTree(InterObject):
680
"""This class represents operations taking place between two Trees.
682
Its instances have methods like 'compare' and contain references to the
683
source and target trees these operations are to be carried out on.
685
Clients of breezy should not need to use InterTree directly, rather they
686
should use the convenience methods on Tree such as 'Tree.compare()' which
687
will pass through to InterTree as appropriate.
690
# Formats that will be used to test this InterTree. If both are
691
# None, this InterTree will not be tested (e.g. because a complex
693
_matching_from_tree_format = None
694
_matching_to_tree_format = None
699
def is_compatible(kls, source, target):
700
# The default implementation is naive and uses the public API, so
701
# it works for all trees.
704
def _changes_from_entries(self, source_entry, target_entry, source_path,
706
"""Generate a iter_changes tuple between source_entry and target_entry.
708
:param source_entry: An inventory entry from self.source, or None.
709
:param target_entry: An inventory entry from self.target, or None.
710
:param source_path: The path of source_entry.
711
:param target_path: The path of target_entry.
712
:return: A tuple, item 0 of which is an iter_changes result tuple, and
713
item 1 is True if there are any changes in the result tuple.
715
if source_entry is None:
716
if target_entry is None:
718
file_id = target_entry.file_id
720
file_id = source_entry.file_id
721
if source_entry is not None:
722
source_versioned = True
723
source_name = source_entry.name
724
source_parent = source_entry.parent_id
725
source_kind, source_executable, source_stat = \
726
self.source._comparison_data(source_entry, source_path)
728
source_versioned = False
732
source_executable = None
733
if target_entry is not None:
734
target_versioned = True
735
target_name = target_entry.name
736
target_parent = target_entry.parent_id
737
target_kind, target_executable, target_stat = \
738
self.target._comparison_data(target_entry, target_path)
740
target_versioned = False
744
target_executable = None
745
versioned = (source_versioned, target_versioned)
746
kind = (source_kind, target_kind)
747
changed_content = False
748
if source_kind != target_kind:
749
changed_content = True
750
elif source_kind == 'file':
751
if not self.file_content_matches(
752
source_path, target_path,
753
file_id, file_id, source_stat, target_stat):
754
changed_content = True
755
elif source_kind == 'symlink':
756
if (self.source.get_symlink_target(source_path, file_id) !=
757
self.target.get_symlink_target(target_path, file_id)):
758
changed_content = True
759
elif source_kind == 'tree-reference':
760
if (self.source.get_reference_revision(source_path, file_id)
761
!= self.target.get_reference_revision(target_path, file_id)):
762
changed_content = True
763
parent = (source_parent, target_parent)
764
name = (source_name, target_name)
765
executable = (source_executable, target_executable)
766
if (changed_content is not False or versioned[0] != versioned[1]
767
or parent[0] != parent[1] or name[0] != name[1] or
768
executable[0] != executable[1]):
772
return (file_id, (source_path, target_path), changed_content,
773
versioned, parent, name, kind, executable), changes
775
def compare(self, want_unchanged=False, specific_files=None,
776
extra_trees=None, require_versioned=False, include_root=False,
777
want_unversioned=False):
778
"""Return the changes from source to target.
780
:return: A TreeDelta.
781
:param specific_files: An optional list of file paths to restrict the
782
comparison to. When mapping filenames to ids, all matches in all
783
trees (including optional extra_trees) are used, and all children of
784
matched directories are included.
785
:param want_unchanged: An optional boolean requesting the inclusion of
786
unchanged entries in the result.
787
:param extra_trees: An optional list of additional trees to use when
788
mapping the contents of specific_files (paths) to file_ids.
789
:param require_versioned: An optional boolean (defaults to False). When
790
supplied and True all the 'specific_files' must be versioned, or
791
a PathsNotVersionedError will be thrown.
792
:param want_unversioned: Scan for unversioned paths.
794
trees = (self.source,)
795
if extra_trees is not None:
796
trees = trees + tuple(extra_trees)
797
with self.lock_read():
798
return delta._compare_trees(self.source, self.target, want_unchanged,
799
specific_files, include_root, extra_trees=extra_trees,
800
require_versioned=require_versioned,
801
want_unversioned=want_unversioned)
803
def iter_changes(self, include_unchanged=False,
804
specific_files=None, pb=None, extra_trees=[],
805
require_versioned=True, want_unversioned=False):
806
"""Generate an iterator of changes between trees.
809
(file_id, (path_in_source, path_in_target),
810
changed_content, versioned, parent, name, kind,
813
Changed_content is True if the file's content has changed. This
814
includes changes to its kind, and to a symlink's target.
816
versioned, parent, name, kind, executable are tuples of (from, to).
817
If a file is missing in a tree, its kind is None.
819
Iteration is done in parent-to-child order, relative to the target
822
There is no guarantee that all paths are in sorted order: the
823
requirement to expand the search due to renames may result in children
824
that should be found early being found late in the search, after
825
lexically later results have been returned.
826
:param require_versioned: Raise errors.PathsNotVersionedError if a
827
path in the specific_files list is not versioned in one of
828
source, target or extra_trees.
829
:param specific_files: An optional list of file paths to restrict the
830
comparison to. When mapping filenames to ids, all matches in all
831
trees (including optional extra_trees) are used, and all children
832
of matched directories are included. The parents in the target tree
833
of the specific files up to and including the root of the tree are
834
always evaluated for changes too.
835
:param want_unversioned: Should unversioned files be returned in the
836
output. An unversioned file is defined as one with (False, False)
837
for the versioned pair.
842
extra_trees = list(extra_trees)
843
# The ids of items we need to examine to insure delta consistency.
844
precise_file_ids = set()
845
changed_file_ids = []
846
if specific_files == []:
847
target_specific_files = []
848
source_specific_files = []
850
target_specific_files = self.target.find_related_paths_across_trees(
851
specific_files, [self.source] + extra_trees,
852
require_versioned=require_versioned)
853
source_specific_files = self.source.find_related_paths_across_trees(
854
specific_files, [self.target] + extra_trees,
855
require_versioned=require_versioned)
856
if specific_files is not None:
857
# reparented or added entries must have their parents included
858
# so that valid deltas can be created. The seen_parents set
859
# tracks the parents that we need to have.
860
# The seen_dirs set tracks directory entries we've yielded.
861
# After outputting version object in to_entries we set difference
862
# the two seen sets and start checking parents.
866
all_unversioned = sorted([(p.split('/'), p) for p in
868
if specific_files is None or
869
osutils.is_inside_any(specific_files, p)])
870
all_unversioned = collections.deque(all_unversioned)
872
all_unversioned = collections.deque()
874
from_entries_by_dir = list(self.source.iter_entries_by_dir(
875
specific_files=source_specific_files))
876
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
877
to_entries_by_dir = list(self.target.iter_entries_by_dir(
878
specific_files=target_specific_files))
879
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
881
# the unversioned path lookup only occurs on real trees - where there
882
# can be extras. So the fake_entry is solely used to look up
883
# executable it values when execute is not supported.
884
fake_entry = TreeFile()
885
for target_path, target_entry in to_entries_by_dir:
886
while (all_unversioned and
887
all_unversioned[0][0] < target_path.split('/')):
888
unversioned_path = all_unversioned.popleft()
889
target_kind, target_executable, target_stat = \
890
self.target._comparison_data(fake_entry, unversioned_path[1])
891
yield (None, (None, unversioned_path[1]), True, (False, False),
893
(None, unversioned_path[0][-1]),
895
(None, target_executable))
896
source_path, source_entry = from_data.get(target_entry.file_id,
898
result, changes = self._changes_from_entries(source_entry,
899
target_entry, source_path=source_path, target_path=target_path)
900
to_paths[result[0]] = result[1][1]
905
pb.update('comparing files', entry_count, num_entries)
906
if changes or include_unchanged:
907
if specific_files is not None:
908
new_parent_id = result[4][1]
909
precise_file_ids.add(new_parent_id)
910
changed_file_ids.append(result[0])
912
# Ensure correct behaviour for reparented/added specific files.
913
if specific_files is not None:
915
if result[6][1] == 'directory':
916
seen_dirs.add(result[0])
917
# Record parents of reparented/added entries.
918
versioned = result[3]
920
if not versioned[0] or parents[0] != parents[1]:
921
seen_parents.add(parents[1])
922
while all_unversioned:
923
# yield any trailing unversioned paths
924
unversioned_path = all_unversioned.popleft()
925
to_kind, to_executable, to_stat = \
926
self.target._comparison_data(fake_entry, unversioned_path[1])
927
yield (None, (None, unversioned_path[1]), True, (False, False),
929
(None, unversioned_path[0][-1]),
931
(None, to_executable))
932
# Yield all remaining source paths
933
for path, from_entry in from_entries_by_dir:
934
file_id = from_entry.file_id
935
if file_id in to_paths:
938
to_path = find_previous_path(self.source, self.target, path)
941
pb.update('comparing files', entry_count, num_entries)
942
versioned = (True, False)
943
parent = (from_entry.parent_id, None)
944
name = (from_entry.name, None)
945
from_kind, from_executable, stat_value = \
946
self.source._comparison_data(from_entry, path)
947
kind = (from_kind, None)
948
executable = (from_executable, None)
949
changed_content = from_kind is not None
950
# the parent's path is necessarily known at this point.
951
changed_file_ids.append(file_id)
952
yield(file_id, (path, to_path), changed_content, versioned, parent,
953
name, kind, executable)
954
changed_file_ids = set(changed_file_ids)
955
if specific_files is not None:
956
for result in self._handle_precise_ids(precise_file_ids,
960
def _get_entry(self, tree, path):
961
"""Get an inventory entry from a tree, with missing entries as None.
963
If the tree raises NotImplementedError on accessing .inventory, then
964
this is worked around using iter_entries_by_dir on just the file id
967
:param tree: The tree to lookup the entry in.
968
:param file_id: The file_id to lookup.
970
# No inventory available.
972
iterator = tree.iter_entries_by_dir(specific_files=[path])
973
return iterator.next()[1]
974
except StopIteration:
977
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
978
discarded_changes=None):
979
"""Fill out a partial iter_changes to be consistent.
981
:param precise_file_ids: The file ids of parents that were seen during
983
:param changed_file_ids: The file ids of already emitted items.
984
:param discarded_changes: An optional dict of precalculated
985
iter_changes items which the partial iter_changes had not output
987
:return: A generator of iter_changes items to output.
989
# process parents of things that had changed under the users
990
# requested paths to prevent incorrect paths or parent ids which
991
# aren't in the tree.
992
while precise_file_ids:
993
precise_file_ids.discard(None)
994
# Don't emit file_ids twice
995
precise_file_ids.difference_update(changed_file_ids)
996
if not precise_file_ids:
998
# If the there was something at a given output path in source, we
999
# have to include the entry from source in the delta, or we would
1000
# be putting this entry into a used path.
1002
for parent_id in precise_file_ids:
1004
paths.append(self.target.id2path(parent_id))
1005
except errors.NoSuchId:
1006
# This id has been dragged in from the source by delta
1007
# expansion and isn't present in target at all: we don't
1008
# need to check for path collisions on it.
1011
old_id = self.source.path2id(path)
1012
precise_file_ids.add(old_id)
1013
precise_file_ids.discard(None)
1014
current_ids = precise_file_ids
1015
precise_file_ids = set()
1016
# We have to emit all of precise_file_ids that have been altered.
1017
# We may have to output the children of some of those ids if any
1018
# directories have stopped being directories.
1019
for file_id in current_ids:
1021
if discarded_changes:
1022
result = discarded_changes.get(file_id)
1028
source_path = self.source.id2path(file_id)
1029
except errors.NoSuchId:
1033
source_entry = self._get_entry(self.source, source_path)
1035
target_path = self.target.id2path(file_id)
1036
except errors.NoSuchId:
1040
target_entry = self._get_entry(self.target, target_path)
1041
result, changes = self._changes_from_entries(
1042
source_entry, target_entry, source_path, target_path)
1045
# Get this parents parent to examine.
1046
new_parent_id = result[4][1]
1047
precise_file_ids.add(new_parent_id)
1049
if (result[6][0] == 'directory' and
1050
result[6][1] != 'directory'):
1051
# This stopped being a directory, the old children have
1053
if source_entry is None:
1054
# Reusing a discarded change.
1055
source_entry = self._get_entry(self.source, result[1][0])
1056
precise_file_ids.update(
1058
for child in self.source.iter_child_entries(result[1][0]))
1059
changed_file_ids.add(result[0])
1062
def file_content_matches(
1063
self, source_path, target_path,
1064
source_file_id=None, target_file_id=None,
1065
source_stat=None, target_stat=None):
1066
"""Check if two files are the same in the source and target trees.
1068
This only checks that the contents of the files are the same,
1069
it does not touch anything else.
1071
:param source_path: Path of the file in the source tree
1072
:param target_path: Path of the file in the target tree
1073
:param source_file_id: Optional file id of the file in the source tree
1074
:param target_file_id: Optional file id of the file in the target tree
1075
:param source_stat: Optional stat value of the file in the source tree
1076
:param target_stat: Optional stat value of the file in the target tree
1077
:return: Boolean indicating whether the files have the same contents
1079
with self.lock_read():
1080
source_verifier_kind, source_verifier_data = (
1081
self.source.get_file_verifier(
1082
source_path, source_file_id, source_stat))
1083
target_verifier_kind, target_verifier_data = (
1084
self.target.get_file_verifier(
1085
target_path, target_file_id, target_stat))
1086
if source_verifier_kind == target_verifier_kind:
1087
return (source_verifier_data == target_verifier_data)
1088
# Fall back to SHA1 for now
1089
if source_verifier_kind != "SHA1":
1090
source_sha1 = self.source.get_file_sha1(
1091
source_path, source_file_id, source_stat)
1093
source_sha1 = source_verifier_data
1094
if target_verifier_kind != "SHA1":
1095
target_sha1 = self.target.get_file_sha1(
1096
target_path, target_file_id, target_stat)
1098
target_sha1 = target_verifier_data
1099
return (source_sha1 == target_sha1)
1101
InterTree.register_optimiser(InterTree)
1104
class MultiWalker(object):
1105
"""Walk multiple trees simultaneously, getting combined results."""
1107
# Note: This could be written to not assume you can do out-of-order
1108
# lookups. Instead any nodes that don't match in all trees could be
1109
# marked as 'deferred', and then returned in the final cleanup loop.
1110
# For now, I think it is "nicer" to return things as close to the
1111
# "master_tree" order as we can.
1113
def __init__(self, master_tree, other_trees):
1114
"""Create a new MultiWalker.
1116
All trees being walked must implement "iter_entries_by_dir()", such
1117
that they yield (path, object) tuples, where that object will have a
1118
'.file_id' member, that can be used to check equality.
1120
:param master_tree: All trees will be 'slaved' to the master_tree such
1121
that nodes in master_tree will be used as 'first-pass' sync points.
1122
Any nodes that aren't in master_tree will be merged in a second
1124
:param other_trees: A list of other trees to walk simultaneously.
1126
self._master_tree = master_tree
1127
self._other_trees = other_trees
1129
# Keep track of any nodes that were properly processed just out of
1130
# order, that way we don't return them at the end, we don't have to
1131
# track *all* processed file_ids, just the out-of-order ones
1132
self._out_of_order_processed = set()
1135
def _step_one(iterator):
1136
"""Step an iter_entries_by_dir iterator.
1138
:return: (has_more, path, ie)
1139
If has_more is False, path and ie will be None.
1142
path, ie = next(iterator)
1143
except StopIteration:
1144
return False, None, None
1146
return True, path, ie
1149
def _cmp_path_by_dirblock(path1, path2):
1150
"""Compare two paths based on what directory they are in.
1152
This generates a sort order, such that all children of a directory are
1153
sorted together, and grandchildren are in the same order as the
1154
children appear. But all grandchildren come after all children.
1156
:param path1: first path
1157
:param path2: the second path
1158
:return: negative number if ``path1`` comes first,
1159
0 if paths are equal
1160
and a positive number if ``path2`` sorts first
1162
# Shortcut this special case
1165
# This is stolen from _dirstate_helpers_py.py, only switching it to
1166
# Unicode objects. Consider using encode_utf8() and then using the
1167
# optimized versions, or maybe writing optimized unicode versions.
1168
if not isinstance(path1, unicode):
1169
raise TypeError("'path1' must be a unicode string, not %s: %r"
1170
% (type(path1), path1))
1171
if not isinstance(path2, unicode):
1172
raise TypeError("'path2' must be a unicode string, not %s: %r"
1173
% (type(path2), path2))
1174
return cmp(MultiWalker._path_to_key(path1),
1175
MultiWalker._path_to_key(path2))
1178
def _path_to_key(path):
1179
dirname, basename = osutils.split(path)
1180
return (dirname.split(u'/'), basename)
1182
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1183
"""Lookup an inventory entry by file_id.
1185
This is called when an entry is missing in the normal order.
1186
Generally this is because a file was either renamed, or it was
1187
deleted/added. If the entry was found in the inventory and not in
1188
extra_entries, it will be added to self._out_of_order_processed
1190
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1191
should be filled with entries that were found before they were
1192
used. If file_id is present, it will be removed from the
1194
:param other_tree: The Tree to search, in case we didn't find the entry
1196
:param file_id: The file_id to look for
1197
:return: (path, ie) if found or (None, None) if not present.
1199
if file_id in extra_entries:
1200
return extra_entries.pop(file_id)
1201
# TODO: Is id2path better as the first call, or is
1202
# inventory[file_id] better as a first check?
1204
cur_path = other_tree.id2path(file_id)
1205
except errors.NoSuchId:
1207
if cur_path is None:
1210
self._out_of_order_processed.add(file_id)
1211
cur_ie = other_tree.root_inventory.get_entry(file_id)
1212
return (cur_path, cur_ie)
1215
"""Match up the values in the different trees."""
1216
for result in self._walk_master_tree():
1218
self._finish_others()
1219
for result in self._walk_others():
1222
def _walk_master_tree(self):
1223
"""First pass, walk all trees in lock-step.
1225
When we are done, all nodes in the master_tree will have been
1226
processed. _other_walkers, _other_entries, and _others_extra will be
1227
set on 'self' for future processing.
1229
# This iterator has the most "inlining" done, because it tends to touch
1230
# every file in the tree, while the others only hit nodes that don't
1232
master_iterator = self._master_tree.iter_entries_by_dir()
1234
other_walkers = [other.iter_entries_by_dir()
1235
for other in self._other_trees]
1236
other_entries = [self._step_one(walker) for walker in other_walkers]
1237
# Track extra nodes in the other trees
1238
others_extra = [{} for _ in range(len(self._other_trees))]
1240
master_has_more = True
1241
step_one = self._step_one
1242
lookup_by_file_id = self._lookup_by_file_id
1243
out_of_order_processed = self._out_of_order_processed
1245
while master_has_more:
1246
(master_has_more, path, master_ie) = step_one(master_iterator)
1247
if not master_has_more:
1250
file_id = master_ie.file_id
1252
other_values_append = other_values.append
1253
next_other_entries = []
1254
next_other_entries_append = next_other_entries.append
1255
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1256
if not other_has_more:
1257
other_values_append(lookup_by_file_id(
1258
others_extra[idx], self._other_trees[idx], file_id))
1259
next_other_entries_append((False, None, None))
1260
elif file_id == other_ie.file_id:
1261
# This is the critical code path, as most of the entries
1262
# should match between most trees.
1263
other_values_append((other_path, other_ie))
1264
next_other_entries_append(step_one(other_walkers[idx]))
1266
# This walker did not match, step it until it either
1267
# matches, or we know we are past the current walker.
1268
other_walker = other_walkers[idx]
1269
other_extra = others_extra[idx]
1270
while (other_has_more and
1271
self._cmp_path_by_dirblock(other_path, path) < 0):
1272
other_file_id = other_ie.file_id
1273
if other_file_id not in out_of_order_processed:
1274
other_extra[other_file_id] = (other_path, other_ie)
1275
other_has_more, other_path, other_ie = \
1276
step_one(other_walker)
1277
if other_has_more and other_ie.file_id == file_id:
1278
# We ended up walking to this point, match and step
1280
other_values_append((other_path, other_ie))
1281
other_has_more, other_path, other_ie = \
1282
step_one(other_walker)
1284
# This record isn't in the normal order, see if it
1286
other_values_append(lookup_by_file_id(
1287
other_extra, self._other_trees[idx], file_id))
1288
next_other_entries_append((other_has_more, other_path,
1290
other_entries = next_other_entries
1292
# We've matched all the walkers, yield this datapoint
1293
yield path, file_id, master_ie, other_values
1294
self._other_walkers = other_walkers
1295
self._other_entries = other_entries
1296
self._others_extra = others_extra
1298
def _finish_others(self):
1299
"""Finish walking the other iterators, so we get all entries."""
1300
for idx, info in enumerate(self._other_entries):
1301
other_extra = self._others_extra[idx]
1302
(other_has_more, other_path, other_ie) = info
1303
while other_has_more:
1304
other_file_id = other_ie.file_id
1305
if other_file_id not in self._out_of_order_processed:
1306
other_extra[other_file_id] = (other_path, other_ie)
1307
other_has_more, other_path, other_ie = \
1308
self._step_one(self._other_walkers[idx])
1309
del self._other_entries
1311
def _walk_others(self):
1312
"""Finish up by walking all the 'deferred' nodes."""
1313
# TODO: One alternative would be to grab all possible unprocessed
1314
# file_ids, and then sort by path, and then yield them. That
1315
# might ensure better ordering, in case a caller strictly
1316
# requires parents before children.
1317
for idx, other_extra in enumerate(self._others_extra):
1318
others = sorted(viewvalues(other_extra),
1319
key=lambda x: self._path_to_key(x[0]))
1320
for other_path, other_ie in others:
1321
file_id = other_ie.file_id
1322
# We don't need to check out_of_order_processed here, because
1323
# the lookup_by_file_id will be removing anything processed
1324
# from the extras cache
1325
other_extra.pop(file_id)
1326
other_values = [(None, None)] * idx
1327
other_values.append((other_path, other_ie))
1328
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1329
alt_idx = alt_idx + idx + 1
1330
alt_extra = self._others_extra[alt_idx]
1331
alt_tree = self._other_trees[alt_idx]
1332
other_values.append(self._lookup_by_file_id(
1333
alt_extra, alt_tree, file_id))
1334
yield other_path, file_id, None, other_values
1337
def find_previous_paths(from_tree, to_tree, paths):
1338
"""Find previous tree paths.
1340
:param from_tree: From tree
1341
:param to_tree: To tree
1342
:param paths: Iterable over paths to search for
1343
:return: Dictionary mapping from from_tree paths to paths in to_tree, or
1344
None if there is no equivalent path.
1348
ret[path] = find_previous_path(from_tree, to_tree, path)
1352
def find_previous_path(from_tree, to_tree, path, file_id=None):
1353
"""Find previous tree path.
1355
:param from_tree: From tree
1356
:param to_tree: To tree
1357
:param path: Path to search for
1358
:return: path in to_tree, or None if there is no equivalent path.
1361
file_id = from_tree.path2id(path)
1363
raise errors.NoSuchFile(path)
1365
return to_tree.id2path(file_id)
1366
except errors.NoSuchId:
added
removed
breezy/tree.py
