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.bzr import (
40
from breezy.i18n import gettext
48
from .inter import InterObject
54
class FileTimestampUnavailable(errors.BzrError):
56
_fmt = "The filestamp for %(path)s is not available."
60
def __init__(self, path):
64
class TreeEntry(object):
65
"""An entry that implements the minimum interface used by commands.
68
def __eq__(self, other):
69
# yes, this is ugly, TODO: best practice __eq__ style.
70
return (isinstance(other, TreeEntry)
71
and other.__class__ == self.__class__)
73
def kind_character(self):
77
class TreeDirectory(TreeEntry):
78
"""See TreeEntry. This is a directory in a working tree."""
80
def kind_character(self):
84
class TreeFile(TreeEntry):
85
"""See TreeEntry. This is a regular file in a working tree."""
87
def kind_character(self):
91
class TreeLink(TreeEntry):
92
"""See TreeEntry. This is a symlink in a working tree."""
94
def kind_character(self):
99
"""Abstract file tree.
101
There are several subclasses:
103
* `WorkingTree` exists as files on disk editable by the user.
105
* `RevisionTree` is a tree as recorded at some point in the past.
107
Trees can be compared, etc, regardless of whether they are working
108
trees or versioned trees.
111
def has_versioned_directories(self):
112
"""Whether this tree can contain explicitly versioned directories.
114
This defaults to True, but some implementations may want to override
119
def changes_from(self, other, want_unchanged=False, specific_files=None,
120
extra_trees=None, require_versioned=False, include_root=False,
121
want_unversioned=False):
122
"""Return a TreeDelta of the changes from other to this tree.
124
:param other: A tree to compare with.
125
:param specific_files: An optional list of file paths to restrict the
126
comparison to. When mapping filenames to ids, all matches in all
127
trees (including optional extra_trees) are used, and all children of
128
matched directories are included.
129
:param want_unchanged: An optional boolean requesting the inclusion of
130
unchanged entries in the result.
131
:param extra_trees: An optional list of additional trees to use when
132
mapping the contents of specific_files (paths) to file_ids.
133
:param require_versioned: An optional boolean (defaults to False). When
134
supplied and True all the 'specific_files' must be versioned, or
135
a PathsNotVersionedError will be thrown.
136
:param want_unversioned: Scan for unversioned paths.
138
The comparison will be performed by an InterTree object looked up on
141
# Martin observes that Tree.changes_from returns a TreeDelta and this
142
# may confuse people, because the class name of the returned object is
143
# a synonym of the object referenced in the method name.
144
return InterTree.get(other, self).compare(
145
want_unchanged=want_unchanged,
146
specific_files=specific_files,
147
extra_trees=extra_trees,
148
require_versioned=require_versioned,
149
include_root=include_root,
150
want_unversioned=want_unversioned,
153
def iter_changes(self, from_tree, include_unchanged=False,
154
specific_files=None, pb=None, extra_trees=None,
155
require_versioned=True, want_unversioned=False):
156
"""See InterTree.iter_changes"""
157
intertree = InterTree.get(from_tree, self)
158
return intertree.iter_changes(include_unchanged, specific_files, pb,
159
extra_trees, require_versioned, want_unversioned=want_unversioned)
162
"""Get a list of the conflicts in the tree.
164
Each conflict is an instance of breezy.conflicts.Conflict.
166
return _mod_conflicts.ConflictList()
169
"""For trees that can have unversioned files, return all such paths."""
172
def get_parent_ids(self):
173
"""Get the parent ids for this tree.
175
:return: a list of parent ids. [] is returned to indicate
176
a tree with no parents.
177
:raises: BzrError if the parents are not known.
179
raise NotImplementedError(self.get_parent_ids)
181
def has_filename(self, filename):
182
"""True if the tree has given filename."""
183
raise NotImplementedError(self.has_filename)
185
def is_ignored(self, filename):
186
"""Check whether the filename is ignored by this tree.
188
:param filename: The relative filename within the tree.
189
:return: True if the filename is ignored.
193
def all_file_ids(self):
194
"""Iterate through all file ids, including ids for missing files."""
195
raise NotImplementedError(self.all_file_ids)
197
def all_versioned_paths(self):
198
"""Iterate through all paths, including paths for missing files."""
199
raise NotImplementedError(self.all_versioned_paths)
201
def id2path(self, file_id):
202
"""Return the path for a file id.
206
raise NotImplementedError(self.id2path)
208
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
209
"""Walk the tree in 'by_dir' order.
211
This will yield each entry in the tree as a (path, entry) tuple.
212
The order that they are yielded is:
214
Directories are walked in a depth-first lexicographical order,
215
however, whenever a directory is reached, all of its direct child
216
nodes are yielded in lexicographical order before yielding the
219
For example, in the tree::
229
The yield order (ignoring root) would be::
231
a, f, a/b, a/d, a/b/c, a/d/e, f/g
233
:param yield_parents: If True, yield the parents from the root leading
234
down to specific_file_ids that have been requested. This has no
235
impact if specific_file_ids is None.
237
raise NotImplementedError(self.iter_entries_by_dir)
239
def iter_child_entries(self, path, file_id=None):
240
"""Iterate over the children of a directory or tree reference.
242
:param path: Path of the directory
243
:param file_id: Optional file id of the directory/tree-reference
244
:raise NoSuchId: When the file_id does not exist
245
:return: Iterator over entries in the directory
247
raise NotImplementedError(self.iter_child_entries)
249
def list_files(self, include_root=False, from_dir=None, recursive=True):
250
"""List all files in this tree.
252
:param include_root: Whether to include the entry for the tree root
253
:param from_dir: Directory under which to list files
254
:param recursive: Whether to list files recursively
255
:return: iterator over tuples of (path, versioned, kind, file_id,
258
raise NotImplementedError(self.list_files)
260
def iter_references(self):
261
if self.supports_tree_reference():
262
for path, entry in self.iter_entries_by_dir():
263
if entry.kind == 'tree-reference':
264
yield path, entry.file_id
266
def kind(self, path, file_id=None):
267
raise NotImplementedError("Tree subclass %s must implement kind"
268
% self.__class__.__name__)
270
def stored_kind(self, path, file_id=None):
271
"""File kind stored for this file_id.
273
May not match kind on disk for working trees. Always available
274
for versioned files, even when the file itself is missing.
276
return self.kind(path, file_id)
278
def path_content_summary(self, path):
279
"""Get a summary of the information about path.
281
All the attributes returned are for the canonical form, not the
282
convenient form (if content filters are in use.)
284
:param path: A relative path within the tree.
285
:return: A tuple containing kind, size, exec, sha1-or-link.
286
Kind is always present (see tree.kind()).
287
size is present if kind is file and the size of the
288
canonical form can be cheaply determined, None otherwise.
289
exec is None unless kind is file and the platform supports the 'x'
291
sha1-or-link is the link target if kind is symlink, or the sha1 if
292
it can be obtained without reading the file.
294
raise NotImplementedError(self.path_content_summary)
296
def get_reference_revision(self, path, file_id=None):
297
raise NotImplementedError("Tree subclass %s must implement "
298
"get_reference_revision"
299
% self.__class__.__name__)
301
def _comparison_data(self, entry, path):
302
"""Return a tuple of kind, executable, stat_value for a file.
304
entry may be None if there is no inventory entry for the file, but
305
path must always be supplied.
307
kind is None if there is no file present (even if an inventory id is
308
present). executable is False for non-file entries.
310
raise NotImplementedError(self._comparison_data)
312
def _file_size(self, entry, stat_value):
313
raise NotImplementedError(self._file_size)
315
def get_file(self, path, file_id=None):
316
"""Return a file object for the file file_id in the tree.
318
If both file_id and path are defined, it is implementation defined as
319
to which one is used.
321
raise NotImplementedError(self.get_file)
323
def get_file_with_stat(self, path, file_id=None):
324
"""Get a file handle and stat object for file_id.
326
The default implementation returns (self.get_file, None) for backwards
329
:param path: The path of the file.
330
:param file_id: The file id to read, if it is known.
331
:return: A tuple (file_handle, stat_value_or_None). If the tree has
332
no stat facility, or need for a stat cache feedback during commit,
333
it may return None for the second element of the tuple.
335
return (self.get_file(path, file_id), None)
337
def get_file_text(self, path, file_id=None):
338
"""Return the byte content of a file.
340
:param path: The path of the file.
341
:param file_id: The file_id of the file.
343
If both file_id and path are supplied, an implementation may use
346
:returns: A single byte string for the whole file.
348
my_file = self.get_file(path, file_id)
350
return my_file.read()
354
def get_file_lines(self, path, file_id=None):
355
"""Return the content of a file, as lines.
357
:param path: The path of the file.
358
:param file_id: The file_id of the file.
360
If both file_id and path are supplied, an implementation may use
363
return osutils.split_lines(self.get_file_text(path, file_id))
365
def get_file_verifier(self, path, file_id=None, stat_value=None):
366
"""Return a verifier for a file.
368
The default implementation returns a sha1.
370
:param file_id: The handle for this file.
371
:param path: The path that this file can be found at.
372
These must point to the same object.
373
:param stat_value: Optional stat value for the object
374
:return: Tuple with verifier name and verifier data
376
return ("SHA1", self.get_file_sha1(path, file_id,
377
stat_value=stat_value))
379
def get_file_sha1(self, path, file_id=None, stat_value=None):
380
"""Return the SHA1 file for a file.
382
:note: callers should use get_file_verifier instead
383
where possible, as the underlying repository implementation may
384
have quicker access to a non-sha1 verifier.
386
:param path: The path that this file can be found at.
387
:param file_id: The handle for this file.
388
These must point to the same object.
389
:param stat_value: Optional stat value for the object
391
raise NotImplementedError(self.get_file_sha1)
393
def get_file_mtime(self, path, file_id=None):
394
"""Return the modification time for a file.
396
:param path: The path that this file can be found at.
397
:param file_id: The handle for this file.
398
These must point to the same object.
400
raise NotImplementedError(self.get_file_mtime)
402
def get_file_size(self, path, file_id=None):
403
"""Return the size of a file in bytes.
405
This applies only to regular files. If invoked on directories or
406
symlinks, it will return None.
407
:param file_id: The file-id of the file
409
raise NotImplementedError(self.get_file_size)
411
def is_executable(self, path, file_id=None):
412
"""Check if a file is executable.
414
:param path: The path that this file can be found at.
415
:param file_id: The handle for this file.
416
These must point to the same object.
418
raise NotImplementedError(self.is_executable)
420
def iter_files_bytes(self, desired_files):
421
"""Iterate through file contents.
423
Files will not necessarily be returned in the order they occur in
424
desired_files. No specific order is guaranteed.
426
Yields pairs of identifier, bytes_iterator. identifier is an opaque
427
value supplied by the caller as part of desired_files. It should
428
uniquely identify the file version in the caller's context. (Examples:
429
an index number or a TreeTransform trans_id.)
431
bytes_iterator is an iterable of bytestrings for the file. The
432
kind of iterable and length of the bytestrings are unspecified, but for
433
this implementation, it is a tuple containing a single bytestring with
434
the complete text of the file.
436
:param desired_files: a list of (file_id, identifier) pairs
438
for file_id, identifier in desired_files:
439
# We wrap the string in a tuple so that we can return an iterable
440
# of bytestrings. (Technically, a bytestring is also an iterable
441
# of bytestrings, but iterating through each character is not
443
# TODO(jelmer): Pass paths into iter_files_bytes
444
path = self.id2path(file_id)
445
cur_file = (self.get_file_text(path, file_id),)
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 _get_plan_merge_data(self, file_id, other, base):
480
from .bzr import versionedfile
481
vf = versionedfile._PlanMergeVersionedFile(file_id)
482
last_revision_a = self._get_file_revision(
483
self.id2path(file_id), file_id, vf, 'this:')
484
last_revision_b = other._get_file_revision(
485
other.id2path(file_id), file_id, vf, 'other:')
487
last_revision_base = None
489
last_revision_base = base._get_file_revision(
490
base.id2path(file_id), file_id, vf, 'base:')
491
return vf, last_revision_a, last_revision_b, last_revision_base
493
def plan_file_merge(self, file_id, other, base=None):
494
"""Generate a merge plan based on annotations.
496
If the file contains uncommitted changes in this tree, they will be
497
attributed to the 'current:' pseudo-revision. If the file contains
498
uncommitted changes in the other tree, they will be assigned to the
499
'other:' pseudo-revision.
501
data = self._get_plan_merge_data(file_id, other, base)
502
vf, last_revision_a, last_revision_b, last_revision_base = data
503
return vf.plan_merge(last_revision_a, last_revision_b,
506
def plan_file_lca_merge(self, file_id, other, base=None):
507
"""Generate a merge plan based lca-newness.
509
If the file contains uncommitted changes in this tree, they will be
510
attributed to the 'current:' pseudo-revision. If the file contains
511
uncommitted changes in the other tree, they will be assigned to the
512
'other:' pseudo-revision.
514
data = self._get_plan_merge_data(file_id, other, base)
515
vf, last_revision_a, last_revision_b, last_revision_base = data
516
return vf.plan_lca_merge(last_revision_a, last_revision_b,
519
def _iter_parent_trees(self):
520
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
521
for revision_id in self.get_parent_ids():
523
yield self.revision_tree(revision_id)
524
except errors.NoSuchRevisionInTree:
525
yield self.repository.revision_tree(revision_id)
527
def _get_file_revision(self, path, file_id, vf, tree_revision):
528
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
529
if getattr(self, '_repository', None) is None:
530
last_revision = tree_revision
531
parent_keys = [(file_id, t.get_file_revision(path, file_id)) for t in
532
self._iter_parent_trees()]
533
vf.add_lines((file_id, last_revision), parent_keys,
534
self.get_file_lines(path, file_id))
535
repo = self.branch.repository
538
last_revision = self.get_file_revision(path, file_id)
539
base_vf = self._repository.texts
540
if base_vf not in vf.fallback_versionedfiles:
541
vf.fallback_versionedfiles.append(base_vf)
544
def _check_retrieved(self, ie, f):
547
fp = osutils.fingerprint_file(f)
550
if ie.text_size is not None:
551
if ie.text_size != fp['size']:
552
raise errors.BzrError(
553
"mismatched size for file %r in %r" %
554
(ie.file_id, self._store),
555
["inventory expects %d bytes" % ie.text_size,
556
"file is actually %d bytes" % fp['size'],
557
"store is probably damaged/corrupt"])
559
if ie.text_sha1 != fp['sha1']:
560
raise errors.BzrError("wrong SHA-1 for file %r in %r" %
561
(ie.file_id, self._store),
562
["inventory expects %s" % ie.text_sha1,
563
"file is actually %s" % fp['sha1'],
564
"store is probably damaged/corrupt"])
566
def path2id(self, path):
567
"""Return the id for path in this tree."""
568
raise NotImplementedError(self.path2id)
570
def is_versioned(self, path):
571
"""Check whether path is versioned.
573
:param path: Path to check
576
return self.path2id(path) is not None
578
def paths2ids(self, paths, trees=[], require_versioned=True):
579
"""Return all the ids that can be reached by walking from paths.
581
Each path is looked up in this tree and any extras provided in
582
trees, and this is repeated recursively: the children in an extra tree
583
of a directory that has been renamed under a provided path in this tree
584
are all returned, even if none exist under a provided path in this
585
tree, and vice versa.
587
:param paths: An iterable of paths to start converting to ids from.
588
Alternatively, if paths is None, no ids should be calculated and None
589
will be returned. This is offered to make calling the api unconditional
590
for code that *might* take a list of files.
591
:param trees: Additional trees to consider.
592
:param require_versioned: If False, do not raise NotVersionedError if
593
an element of paths is not versioned in this tree and all of trees.
595
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
597
def iter_children(self, file_id):
598
"""Iterate over the file ids of the children of an entry.
600
:param file_id: File id of the entry
601
:return: Iterator over child file ids.
603
raise NotImplementedError(self.iter_children)
606
"""Lock this tree for multiple read only operations.
608
:return: A breezy.lock.LogicalLockResult.
610
return lock.LogicalLockResult(self.unlock)
612
def revision_tree(self, revision_id):
613
"""Obtain a revision tree for the revision revision_id.
615
The intention of this method is to allow access to possibly cached
616
tree data. Implementors of this method should raise NoSuchRevision if
617
the tree is not locally available, even if they could obtain the
618
tree via a repository or some other means. Callers are responsible
619
for finding the ultimate source for a revision tree.
621
:param revision_id: The revision_id of the requested tree.
623
:raises: NoSuchRevision if the tree cannot be obtained.
625
raise errors.NoSuchRevisionInTree(self, revision_id)
628
"""What files are present in this tree and unknown.
630
:return: an iterator over the unknown files.
637
def filter_unversioned_files(self, paths):
638
"""Filter out paths that are versioned.
640
:return: set of paths.
642
raise NotImplementedError(self.filter_unversioned_files)
644
def walkdirs(self, prefix=""):
645
"""Walk the contents of this tree from path down.
647
This yields all the data about the contents of a directory at a time.
648
After each directory has been yielded, if the caller has mutated the
649
list to exclude some directories, they are then not descended into.
651
The data yielded is of the form:
652
((directory-relpath, directory-path-from-root, directory-fileid),
653
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
654
versioned_kind), ...]),
655
- directory-relpath is the containing dirs relpath from prefix
656
- directory-path-from-root is the containing dirs path from /
657
- directory-fileid is the id of the directory if it is versioned.
658
- relpath is the relative path within the subtree being walked.
659
- basename is the basename
660
- kind is the kind of the file now. If unknonwn then the file is not
661
present within the tree - but it may be recorded as versioned. See
663
- lstat is the stat data *if* the file was statted.
664
- path_from_tree_root is the path from the root of the tree.
665
- file_id is the file_id if the entry is versioned.
666
- versioned_kind is the kind of the file as last recorded in the
667
versioning system. If 'unknown' the file is not versioned.
668
One of 'kind' and 'versioned_kind' must not be 'unknown'.
670
:param prefix: Start walking from prefix within the tree rather than
671
at the root. This allows one to walk a subtree but get paths that are
672
relative to a tree rooted higher up.
673
:return: an iterator over the directory data.
675
raise NotImplementedError(self.walkdirs)
677
def supports_content_filtering(self):
680
def _content_filter_stack(self, path=None, file_id=None):
681
"""The stack of content filters for a path if filtering is supported.
683
Readers will be applied in first-to-last order.
684
Writers will be applied in last-to-first order.
685
Either the path or the file-id needs to be provided.
687
:param path: path relative to the root of the tree
689
:param file_id: file_id or None if unknown
690
:return: the list of filters - [] if there are none
692
filter_pref_names = filters._get_registered_names()
693
if len(filter_pref_names) == 0:
696
path = self.id2path(file_id)
697
prefs = next(self.iter_search_rules([path], filter_pref_names))
698
stk = filters._get_filter_stack_for(prefs)
699
if 'filters' in debug.debug_flags:
700
trace.note(gettext("*** {0} content-filter: {1} => {2!r}").format(path, prefs, stk))
703
def _content_filter_stack_provider(self):
704
"""A function that returns a stack of ContentFilters.
706
The function takes a path (relative to the top of the tree) and a
707
file-id as parameters.
709
:return: None if content filtering is not supported by this tree.
711
if self.supports_content_filtering():
712
return lambda path, file_id: \
713
self._content_filter_stack(path, file_id)
717
def iter_search_rules(self, path_names, pref_names=None,
718
_default_searcher=None):
719
"""Find the preferences for filenames in a tree.
721
:param path_names: an iterable of paths to find attributes for.
722
Paths are given relative to the root of the tree.
723
:param pref_names: the list of preferences to lookup - None for all
724
:param _default_searcher: private parameter to assist testing - don't use
725
:return: an iterator of tuple sequences, one per path-name.
726
See _RulesSearcher.get_items for details on the tuple sequence.
728
if _default_searcher is None:
729
_default_searcher = rules._per_user_searcher
730
searcher = self._get_rules_searcher(_default_searcher)
731
if searcher is not None:
732
if pref_names is not None:
733
for path in path_names:
734
yield searcher.get_selected_items(path, pref_names)
736
for path in path_names:
737
yield searcher.get_items(path)
739
def _get_rules_searcher(self, default_searcher):
740
"""Get the RulesSearcher for this tree given the default one."""
741
searcher = default_searcher
745
def find_ids_across_trees(filenames, trees, require_versioned=True):
746
"""Find the ids corresponding to specified filenames.
748
All matches in all trees will be used, and all children of matched
749
directories will be used.
751
:param filenames: The filenames to find file_ids for (if None, returns
753
:param trees: The trees to find file_ids within
754
:param require_versioned: if true, all specified filenames must occur in
756
:return: a set of file ids for the specified filenames and their children.
760
specified_path_ids = _find_ids_across_trees(filenames, trees,
762
return _find_children_across_trees(specified_path_ids, trees)
765
def _find_ids_across_trees(filenames, trees, require_versioned):
766
"""Find the ids corresponding to specified filenames.
768
All matches in all trees will be used, but subdirectories are not scanned.
770
:param filenames: The filenames to find file_ids for
771
:param trees: The trees to find file_ids within
772
:param require_versioned: if true, all specified filenames must occur in
774
:return: a set of file ids for the specified filenames
777
interesting_ids = set()
778
for tree_path in filenames:
781
file_id = tree.path2id(tree_path)
782
if file_id is not None:
783
interesting_ids.add(file_id)
786
not_versioned.append(tree_path)
787
if len(not_versioned) > 0 and require_versioned:
788
raise errors.PathsNotVersionedError(not_versioned)
789
return interesting_ids
792
def _find_children_across_trees(specified_ids, trees):
793
"""Return a set including specified ids and their children.
795
All matches in all trees will be used.
797
:param trees: The trees to find file_ids within
798
:return: a set containing all specified ids and their children
800
interesting_ids = set(specified_ids)
801
pending = interesting_ids
802
# now handle children of interesting ids
803
# we loop so that we handle all children of each id in both trees
804
while len(pending) > 0:
806
for file_id in pending:
808
if not tree.has_or_had_id(file_id):
810
for child_id in tree.iter_children(file_id):
811
if child_id not in interesting_ids:
812
new_pending.add(child_id)
813
interesting_ids.update(new_pending)
814
pending = new_pending
815
return interesting_ids
818
class InterTree(InterObject):
819
"""This class represents operations taking place between two Trees.
821
Its instances have methods like 'compare' and contain references to the
822
source and target trees these operations are to be carried out on.
824
Clients of breezy should not need to use InterTree directly, rather they
825
should use the convenience methods on Tree such as 'Tree.compare()' which
826
will pass through to InterTree as appropriate.
829
# Formats that will be used to test this InterTree. If both are
830
# None, this InterTree will not be tested (e.g. because a complex
832
_matching_from_tree_format = None
833
_matching_to_tree_format = None
838
def is_compatible(kls, source, target):
839
# The default implementation is naive and uses the public API, so
840
# it works for all trees.
843
def _changes_from_entries(self, source_entry, target_entry,
844
source_path=None, target_path=None):
845
"""Generate a iter_changes tuple between source_entry and target_entry.
847
:param source_entry: An inventory entry from self.source, or None.
848
:param target_entry: An inventory entry from self.target, or None.
849
:param source_path: The path of source_entry, if known. If not known
850
it will be looked up.
851
:param target_path: The path of target_entry, if known. If not known
852
it will be looked up.
853
:return: A tuple, item 0 of which is an iter_changes result tuple, and
854
item 1 is True if there are any changes in the result tuple.
856
if source_entry is None:
857
if target_entry is None:
859
file_id = target_entry.file_id
861
file_id = source_entry.file_id
862
if source_entry is not None:
863
source_versioned = True
864
source_name = source_entry.name
865
source_parent = source_entry.parent_id
866
if source_path is None:
867
source_path = self.source.id2path(file_id)
868
source_kind, source_executable, source_stat = \
869
self.source._comparison_data(source_entry, source_path)
871
source_versioned = False
875
source_executable = None
876
if target_entry is not None:
877
target_versioned = True
878
target_name = target_entry.name
879
target_parent = target_entry.parent_id
880
if target_path is None:
881
target_path = self.target.id2path(file_id)
882
target_kind, target_executable, target_stat = \
883
self.target._comparison_data(target_entry, target_path)
885
target_versioned = False
889
target_executable = None
890
versioned = (source_versioned, target_versioned)
891
kind = (source_kind, target_kind)
892
changed_content = False
893
if source_kind != target_kind:
894
changed_content = True
895
elif source_kind == 'file':
896
if not self.file_content_matches(file_id, file_id, source_path,
897
target_path, source_stat, target_stat):
898
changed_content = True
899
elif source_kind == 'symlink':
900
if (self.source.get_symlink_target(source_path, file_id) !=
901
self.target.get_symlink_target(target_path, file_id)):
902
changed_content = True
903
elif source_kind == 'tree-reference':
904
if (self.source.get_reference_revision(source_path, file_id)
905
!= self.target.get_reference_revision(target_path, file_id)):
906
changed_content = True
907
parent = (source_parent, target_parent)
908
name = (source_name, target_name)
909
executable = (source_executable, target_executable)
910
if (changed_content is not False or versioned[0] != versioned[1]
911
or parent[0] != parent[1] or name[0] != name[1] or
912
executable[0] != executable[1]):
916
return (file_id, (source_path, target_path), changed_content,
917
versioned, parent, name, kind, executable), changes
919
def compare(self, want_unchanged=False, specific_files=None,
920
extra_trees=None, require_versioned=False, include_root=False,
921
want_unversioned=False):
922
"""Return the changes from source to target.
924
:return: A TreeDelta.
925
:param specific_files: An optional list of file paths to restrict the
926
comparison to. When mapping filenames to ids, all matches in all
927
trees (including optional extra_trees) are used, and all children of
928
matched directories are included.
929
:param want_unchanged: An optional boolean requesting the inclusion of
930
unchanged entries in the result.
931
:param extra_trees: An optional list of additional trees to use when
932
mapping the contents of specific_files (paths) to file_ids.
933
:param require_versioned: An optional boolean (defaults to False). When
934
supplied and True all the 'specific_files' must be versioned, or
935
a PathsNotVersionedError will be thrown.
936
:param want_unversioned: Scan for unversioned paths.
938
trees = (self.source,)
939
if extra_trees is not None:
940
trees = trees + tuple(extra_trees)
941
with self.lock_read():
942
# target is usually the newer tree:
943
specific_file_ids = self.target.paths2ids(specific_files, trees,
944
require_versioned=require_versioned)
945
if specific_files and not specific_file_ids:
946
# All files are unversioned, so just return an empty delta
947
# _compare_trees would think we want a complete delta
948
result = delta.TreeDelta()
949
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
950
result.unversioned = [(path, None,
951
self.target._comparison_data(fake_entry, path)[0]) for path in
954
return delta._compare_trees(self.source, self.target, want_unchanged,
955
specific_files, include_root, extra_trees=extra_trees,
956
require_versioned=require_versioned,
957
want_unversioned=want_unversioned)
959
def iter_changes(self, include_unchanged=False,
960
specific_files=None, pb=None, extra_trees=[],
961
require_versioned=True, want_unversioned=False):
962
"""Generate an iterator of changes between trees.
965
(file_id, (path_in_source, path_in_target),
966
changed_content, versioned, parent, name, kind,
969
Changed_content is True if the file's content has changed. This
970
includes changes to its kind, and to a symlink's target.
972
versioned, parent, name, kind, executable are tuples of (from, to).
973
If a file is missing in a tree, its kind is None.
975
Iteration is done in parent-to-child order, relative to the target
978
There is no guarantee that all paths are in sorted order: the
979
requirement to expand the search due to renames may result in children
980
that should be found early being found late in the search, after
981
lexically later results have been returned.
982
:param require_versioned: Raise errors.PathsNotVersionedError if a
983
path in the specific_files list is not versioned in one of
984
source, target or extra_trees.
985
:param specific_files: An optional list of file paths to restrict the
986
comparison to. When mapping filenames to ids, all matches in all
987
trees (including optional extra_trees) are used, and all children
988
of matched directories are included. The parents in the target tree
989
of the specific files up to and including the root of the tree are
990
always evaluated for changes too.
991
:param want_unversioned: Should unversioned files be returned in the
992
output. An unversioned file is defined as one with (False, False)
993
for the versioned pair.
995
lookup_trees = [self.source]
997
lookup_trees.extend(extra_trees)
998
# The ids of items we need to examine to insure delta consistency.
999
precise_file_ids = set()
1000
changed_file_ids = []
1001
if specific_files == []:
1002
specific_file_ids = []
1004
specific_file_ids = self.target.paths2ids(specific_files,
1005
lookup_trees, require_versioned=require_versioned)
1006
if specific_files is not None:
1007
# reparented or added entries must have their parents included
1008
# so that valid deltas can be created. The seen_parents set
1009
# tracks the parents that we need to have.
1010
# The seen_dirs set tracks directory entries we've yielded.
1011
# After outputting version object in to_entries we set difference
1012
# the two seen sets and start checking parents.
1013
seen_parents = set()
1015
if want_unversioned:
1016
all_unversioned = sorted([(p.split('/'), p) for p in
1017
self.target.extras()
1018
if specific_files is None or
1019
osutils.is_inside_any(specific_files, p)])
1020
all_unversioned = collections.deque(all_unversioned)
1022
all_unversioned = collections.deque()
1024
from_entries_by_dir = list(self.source.iter_entries_by_dir(
1025
specific_file_ids=specific_file_ids))
1026
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
1027
to_entries_by_dir = list(self.target.iter_entries_by_dir(
1028
specific_file_ids=specific_file_ids))
1029
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
1031
# the unversioned path lookup only occurs on real trees - where there
1032
# can be extras. So the fake_entry is solely used to look up
1033
# executable it values when execute is not supported.
1034
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1035
for target_path, target_entry in to_entries_by_dir:
1036
while (all_unversioned and
1037
all_unversioned[0][0] < target_path.split('/')):
1038
unversioned_path = all_unversioned.popleft()
1039
target_kind, target_executable, target_stat = \
1040
self.target._comparison_data(fake_entry, unversioned_path[1])
1041
yield (None, (None, unversioned_path[1]), True, (False, False),
1043
(None, unversioned_path[0][-1]),
1044
(None, target_kind),
1045
(None, target_executable))
1046
source_path, source_entry = from_data.get(target_entry.file_id,
1048
result, changes = self._changes_from_entries(source_entry,
1049
target_entry, source_path=source_path, target_path=target_path)
1050
to_paths[result[0]] = result[1][1]
1055
pb.update('comparing files', entry_count, num_entries)
1056
if changes or include_unchanged:
1057
if specific_file_ids is not None:
1058
new_parent_id = result[4][1]
1059
precise_file_ids.add(new_parent_id)
1060
changed_file_ids.append(result[0])
1062
# Ensure correct behaviour for reparented/added specific files.
1063
if specific_files is not None:
1064
# Record output dirs
1065
if result[6][1] == 'directory':
1066
seen_dirs.add(result[0])
1067
# Record parents of reparented/added entries.
1068
versioned = result[3]
1070
if not versioned[0] or parents[0] != parents[1]:
1071
seen_parents.add(parents[1])
1072
while all_unversioned:
1073
# yield any trailing unversioned paths
1074
unversioned_path = all_unversioned.popleft()
1075
to_kind, to_executable, to_stat = \
1076
self.target._comparison_data(fake_entry, unversioned_path[1])
1077
yield (None, (None, unversioned_path[1]), True, (False, False),
1079
(None, unversioned_path[0][-1]),
1081
(None, to_executable))
1082
# Yield all remaining source paths
1083
for path, from_entry in from_entries_by_dir:
1084
file_id = from_entry.file_id
1085
if file_id in to_paths:
1088
if not self.target.has_id(file_id):
1089
# common case - paths we have not emitted are not present in
1093
to_path = self.target.id2path(file_id)
1096
pb.update('comparing files', entry_count, num_entries)
1097
versioned = (True, False)
1098
parent = (from_entry.parent_id, None)
1099
name = (from_entry.name, None)
1100
from_kind, from_executable, stat_value = \
1101
self.source._comparison_data(from_entry, path)
1102
kind = (from_kind, None)
1103
executable = (from_executable, None)
1104
changed_content = from_kind is not None
1105
# the parent's path is necessarily known at this point.
1106
changed_file_ids.append(file_id)
1107
yield(file_id, (path, to_path), changed_content, versioned, parent,
1108
name, kind, executable)
1109
changed_file_ids = set(changed_file_ids)
1110
if specific_file_ids is not None:
1111
for result in self._handle_precise_ids(precise_file_ids,
1115
def _get_entry(self, tree, file_id):
1116
"""Get an inventory entry from a tree, with missing entries as None.
1118
If the tree raises NotImplementedError on accessing .inventory, then
1119
this is worked around using iter_entries_by_dir on just the file id
1122
:param tree: The tree to lookup the entry in.
1123
:param file_id: The file_id to lookup.
1126
inventory = tree.root_inventory
1127
except (AttributeError, NotImplementedError):
1128
# No inventory available.
1130
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1131
return iterator.next()[1]
1132
except StopIteration:
1136
return inventory[file_id]
1137
except errors.NoSuchId:
1140
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1141
discarded_changes=None):
1142
"""Fill out a partial iter_changes to be consistent.
1144
:param precise_file_ids: The file ids of parents that were seen during
1146
:param changed_file_ids: The file ids of already emitted items.
1147
:param discarded_changes: An optional dict of precalculated
1148
iter_changes items which the partial iter_changes had not output
1150
:return: A generator of iter_changes items to output.
1152
# process parents of things that had changed under the users
1153
# requested paths to prevent incorrect paths or parent ids which
1154
# aren't in the tree.
1155
while precise_file_ids:
1156
precise_file_ids.discard(None)
1157
# Don't emit file_ids twice
1158
precise_file_ids.difference_update(changed_file_ids)
1159
if not precise_file_ids:
1161
# If the there was something at a given output path in source, we
1162
# have to include the entry from source in the delta, or we would
1163
# be putting this entry into a used path.
1165
for parent_id in precise_file_ids:
1167
paths.append(self.target.id2path(parent_id))
1168
except errors.NoSuchId:
1169
# This id has been dragged in from the source by delta
1170
# expansion and isn't present in target at all: we don't
1171
# need to check for path collisions on it.
1174
old_id = self.source.path2id(path)
1175
precise_file_ids.add(old_id)
1176
precise_file_ids.discard(None)
1177
current_ids = precise_file_ids
1178
precise_file_ids = set()
1179
# We have to emit all of precise_file_ids that have been altered.
1180
# We may have to output the children of some of those ids if any
1181
# directories have stopped being directories.
1182
for file_id in current_ids:
1184
if discarded_changes:
1185
result = discarded_changes.get(file_id)
1190
old_entry = self._get_entry(self.source, file_id)
1191
new_entry = self._get_entry(self.target, file_id)
1192
result, changes = self._changes_from_entries(
1193
old_entry, new_entry)
1196
# Get this parents parent to examine.
1197
new_parent_id = result[4][1]
1198
precise_file_ids.add(new_parent_id)
1200
if (result[6][0] == 'directory' and
1201
result[6][1] != 'directory'):
1202
# This stopped being a directory, the old children have
1204
if old_entry is None:
1205
# Reusing a discarded change.
1206
old_entry = self._get_entry(self.source, file_id)
1207
precise_file_ids.update(
1208
self.source.iter_children(file_id))
1209
changed_file_ids.add(result[0])
1212
def file_content_matches(
1213
self, source_file_id, target_file_id, source_path=None,
1214
target_path=None, source_stat=None, target_stat=None):
1215
"""Check if two files are the same in the source and target trees.
1217
This only checks that the contents of the files are the same,
1218
it does not touch anything else.
1220
:param source_file_id: File id of the file in the source tree
1221
:param target_file_id: File id of the file in the target tree
1222
:param source_path: Path of the file in the source tree
1223
:param target_path: Path of the file in the target tree
1224
:param source_stat: Optional stat value of the file in the source tree
1225
:param target_stat: Optional stat value of the file in the target tree
1226
:return: Boolean indicating whether the files have the same contents
1228
with self.lock_read():
1229
if source_path is None:
1230
source_path = self.source.id2path(source_file_id)
1231
if target_path is None:
1232
target_path = self.target.id2path(target_file_id)
1233
source_verifier_kind, source_verifier_data = (
1234
self.source.get_file_verifier(
1235
source_path, source_file_id, source_stat))
1236
target_verifier_kind, target_verifier_data = (
1237
self.target.get_file_verifier(
1238
target_path, target_file_id, target_stat))
1239
if source_verifier_kind == target_verifier_kind:
1240
return (source_verifier_data == target_verifier_data)
1241
# Fall back to SHA1 for now
1242
if source_verifier_kind != "SHA1":
1243
source_sha1 = self.source.get_file_sha1(
1244
source_path, source_file_id, source_stat)
1246
source_sha1 = source_verifier_data
1247
if target_verifier_kind != "SHA1":
1248
target_sha1 = self.target.get_file_sha1(
1249
target_path, target_file_id, target_stat)
1251
target_sha1 = target_verifier_data
1252
return (source_sha1 == target_sha1)
1254
InterTree.register_optimiser(InterTree)
1257
class MultiWalker(object):
1258
"""Walk multiple trees simultaneously, getting combined results."""
1260
# Note: This could be written to not assume you can do out-of-order
1261
# lookups. Instead any nodes that don't match in all trees could be
1262
# marked as 'deferred', and then returned in the final cleanup loop.
1263
# For now, I think it is "nicer" to return things as close to the
1264
# "master_tree" order as we can.
1266
def __init__(self, master_tree, other_trees):
1267
"""Create a new MultiWalker.
1269
All trees being walked must implement "iter_entries_by_dir()", such
1270
that they yield (path, object) tuples, where that object will have a
1271
'.file_id' member, that can be used to check equality.
1273
:param master_tree: All trees will be 'slaved' to the master_tree such
1274
that nodes in master_tree will be used as 'first-pass' sync points.
1275
Any nodes that aren't in master_tree will be merged in a second
1277
:param other_trees: A list of other trees to walk simultaneously.
1279
self._master_tree = master_tree
1280
self._other_trees = other_trees
1282
# Keep track of any nodes that were properly processed just out of
1283
# order, that way we don't return them at the end, we don't have to
1284
# track *all* processed file_ids, just the out-of-order ones
1285
self._out_of_order_processed = set()
1288
def _step_one(iterator):
1289
"""Step an iter_entries_by_dir iterator.
1291
:return: (has_more, path, ie)
1292
If has_more is False, path and ie will be None.
1295
path, ie = next(iterator)
1296
except StopIteration:
1297
return False, None, None
1299
return True, path, ie
1302
def _cmp_path_by_dirblock(path1, path2):
1303
"""Compare two paths based on what directory they are in.
1305
This generates a sort order, such that all children of a directory are
1306
sorted together, and grandchildren are in the same order as the
1307
children appear. But all grandchildren come after all children.
1309
:param path1: first path
1310
:param path2: the second path
1311
:return: negative number if ``path1`` comes first,
1312
0 if paths are equal
1313
and a positive number if ``path2`` sorts first
1315
# Shortcut this special case
1318
# This is stolen from _dirstate_helpers_py.py, only switching it to
1319
# Unicode objects. Consider using encode_utf8() and then using the
1320
# optimized versions, or maybe writing optimized unicode versions.
1321
if not isinstance(path1, unicode):
1322
raise TypeError("'path1' must be a unicode string, not %s: %r"
1323
% (type(path1), path1))
1324
if not isinstance(path2, unicode):
1325
raise TypeError("'path2' must be a unicode string, not %s: %r"
1326
% (type(path2), path2))
1327
return cmp(MultiWalker._path_to_key(path1),
1328
MultiWalker._path_to_key(path2))
1331
def _path_to_key(path):
1332
dirname, basename = osutils.split(path)
1333
return (dirname.split(u'/'), basename)
1335
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1336
"""Lookup an inventory entry by file_id.
1338
This is called when an entry is missing in the normal order.
1339
Generally this is because a file was either renamed, or it was
1340
deleted/added. If the entry was found in the inventory and not in
1341
extra_entries, it will be added to self._out_of_order_processed
1343
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1344
should be filled with entries that were found before they were
1345
used. If file_id is present, it will be removed from the
1347
:param other_tree: The Tree to search, in case we didn't find the entry
1349
:param file_id: The file_id to look for
1350
:return: (path, ie) if found or (None, None) if not present.
1352
if file_id in extra_entries:
1353
return extra_entries.pop(file_id)
1354
# TODO: Is id2path better as the first call, or is
1355
# inventory[file_id] better as a first check?
1357
cur_path = other_tree.id2path(file_id)
1358
except errors.NoSuchId:
1360
if cur_path is None:
1363
self._out_of_order_processed.add(file_id)
1364
cur_ie = other_tree.root_inventory[file_id]
1365
return (cur_path, cur_ie)
1368
"""Match up the values in the different trees."""
1369
for result in self._walk_master_tree():
1371
self._finish_others()
1372
for result in self._walk_others():
1375
def _walk_master_tree(self):
1376
"""First pass, walk all trees in lock-step.
1378
When we are done, all nodes in the master_tree will have been
1379
processed. _other_walkers, _other_entries, and _others_extra will be
1380
set on 'self' for future processing.
1382
# This iterator has the most "inlining" done, because it tends to touch
1383
# every file in the tree, while the others only hit nodes that don't
1385
master_iterator = self._master_tree.iter_entries_by_dir()
1387
other_walkers = [other.iter_entries_by_dir()
1388
for other in self._other_trees]
1389
other_entries = [self._step_one(walker) for walker in other_walkers]
1390
# Track extra nodes in the other trees
1391
others_extra = [{} for _ in range(len(self._other_trees))]
1393
master_has_more = True
1394
step_one = self._step_one
1395
lookup_by_file_id = self._lookup_by_file_id
1396
out_of_order_processed = self._out_of_order_processed
1398
while master_has_more:
1399
(master_has_more, path, master_ie) = step_one(master_iterator)
1400
if not master_has_more:
1403
file_id = master_ie.file_id
1405
other_values_append = other_values.append
1406
next_other_entries = []
1407
next_other_entries_append = next_other_entries.append
1408
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1409
if not other_has_more:
1410
other_values_append(lookup_by_file_id(
1411
others_extra[idx], self._other_trees[idx], file_id))
1412
next_other_entries_append((False, None, None))
1413
elif file_id == other_ie.file_id:
1414
# This is the critical code path, as most of the entries
1415
# should match between most trees.
1416
other_values_append((other_path, other_ie))
1417
next_other_entries_append(step_one(other_walkers[idx]))
1419
# This walker did not match, step it until it either
1420
# matches, or we know we are past the current walker.
1421
other_walker = other_walkers[idx]
1422
other_extra = others_extra[idx]
1423
while (other_has_more and
1424
self._cmp_path_by_dirblock(other_path, path) < 0):
1425
other_file_id = other_ie.file_id
1426
if other_file_id not in out_of_order_processed:
1427
other_extra[other_file_id] = (other_path, other_ie)
1428
other_has_more, other_path, other_ie = \
1429
step_one(other_walker)
1430
if other_has_more and other_ie.file_id == file_id:
1431
# We ended up walking to this point, match and step
1433
other_values_append((other_path, other_ie))
1434
other_has_more, other_path, other_ie = \
1435
step_one(other_walker)
1437
# This record isn't in the normal order, see if it
1439
other_values_append(lookup_by_file_id(
1440
other_extra, self._other_trees[idx], file_id))
1441
next_other_entries_append((other_has_more, other_path,
1443
other_entries = next_other_entries
1445
# We've matched all the walkers, yield this datapoint
1446
yield path, file_id, master_ie, other_values
1447
self._other_walkers = other_walkers
1448
self._other_entries = other_entries
1449
self._others_extra = others_extra
1451
def _finish_others(self):
1452
"""Finish walking the other iterators, so we get all entries."""
1453
for idx, info in enumerate(self._other_entries):
1454
other_extra = self._others_extra[idx]
1455
(other_has_more, other_path, other_ie) = info
1456
while other_has_more:
1457
other_file_id = other_ie.file_id
1458
if other_file_id not in self._out_of_order_processed:
1459
other_extra[other_file_id] = (other_path, other_ie)
1460
other_has_more, other_path, other_ie = \
1461
self._step_one(self._other_walkers[idx])
1462
del self._other_entries
1464
def _walk_others(self):
1465
"""Finish up by walking all the 'deferred' nodes."""
1466
# TODO: One alternative would be to grab all possible unprocessed
1467
# file_ids, and then sort by path, and then yield them. That
1468
# might ensure better ordering, in case a caller strictly
1469
# requires parents before children.
1470
for idx, other_extra in enumerate(self._others_extra):
1471
others = sorted(viewvalues(other_extra),
1472
key=lambda x: self._path_to_key(x[0]))
1473
for other_path, other_ie in others:
1474
file_id = other_ie.file_id
1475
# We don't need to check out_of_order_processed here, because
1476
# the lookup_by_file_id will be removing anything processed
1477
# from the extras cache
1478
other_extra.pop(file_id)
1479
other_values = [(None, None)] * idx
1480
other_values.append((other_path, other_ie))
1481
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1482
alt_idx = alt_idx + idx + 1
1483
alt_extra = self._others_extra[alt_idx]
1484
alt_tree = self._other_trees[alt_idx]
1485
other_values.append(self._lookup_by_file_id(
1486
alt_extra, alt_tree, file_id))
1487
yield other_path, file_id, None, other_values
added
removed
breezy/tree.py
