1
# Copyright (C) 2005 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Tree classes, representing directory at point in time.
21
from collections import deque
22
from cStringIO import StringIO
26
conflicts as _mod_conflicts,
29
revision as _mod_revision,
33
from bzrlib.decorators import needs_read_lock
34
from bzrlib.errors import BzrError, BzrCheckError
35
from bzrlib import errors
36
from bzrlib.inventory import Inventory, InventoryFile
37
from bzrlib.inter import InterObject
38
from bzrlib.osutils import fingerprint_file
39
import bzrlib.revision
40
from bzrlib.trace import mutter, note
44
"""Abstract file tree.
46
There are several subclasses:
48
* `WorkingTree` exists as files on disk editable by the user.
50
* `RevisionTree` is a tree as recorded at some point in the past.
52
Trees contain an `Inventory` object, and also know how to retrieve
53
file texts mentioned in the inventory, either from a working
54
directory or from a store.
56
It is possible for trees to contain files that are not described
57
in their inventory or vice versa; for this use `filenames()`.
59
Trees can be compared, etc, regardless of whether they are working
60
trees or versioned trees.
63
def changes_from(self, other, want_unchanged=False, specific_files=None,
64
extra_trees=None, require_versioned=False, include_root=False,
65
want_unversioned=False):
66
"""Return a TreeDelta of the changes from other to this tree.
68
:param other: A tree to compare with.
69
:param specific_files: An optional list of file paths to restrict the
70
comparison to. When mapping filenames to ids, all matches in all
71
trees (including optional extra_trees) are used, and all children of
72
matched directories are included.
73
:param want_unchanged: An optional boolean requesting the inclusion of
74
unchanged entries in the result.
75
:param extra_trees: An optional list of additional trees to use when
76
mapping the contents of specific_files (paths) to file_ids.
77
:param require_versioned: An optional boolean (defaults to False). When
78
supplied and True all the 'specific_files' must be versioned, or
79
a PathsNotVersionedError will be thrown.
80
:param want_unversioned: Scan for unversioned paths.
82
The comparison will be performed by an InterTree object looked up on
85
# Martin observes that Tree.changes_from returns a TreeDelta and this
86
# may confuse people, because the class name of the returned object is
87
# a synonym of the object referenced in the method name.
88
return InterTree.get(other, self).compare(
89
want_unchanged=want_unchanged,
90
specific_files=specific_files,
91
extra_trees=extra_trees,
92
require_versioned=require_versioned,
93
include_root=include_root,
94
want_unversioned=want_unversioned,
97
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
98
def _iter_changes(self, *args, **kwargs):
99
return self.iter_changes(*args, **kwargs)
101
def iter_changes(self, from_tree, include_unchanged=False,
102
specific_files=None, pb=None, extra_trees=None,
103
require_versioned=True, want_unversioned=False):
104
intertree = InterTree.get(from_tree, self)
105
return intertree.iter_changes(include_unchanged, specific_files, pb,
106
extra_trees, require_versioned, want_unversioned=want_unversioned)
109
"""Get a list of the conflicts in the tree.
111
Each conflict is an instance of bzrlib.conflicts.Conflict.
113
return _mod_conflicts.ConflictList()
116
"""For trees that can have unversioned files, return all such paths."""
119
def get_parent_ids(self):
120
"""Get the parent ids for this tree.
122
:return: a list of parent ids. [] is returned to indicate
123
a tree with no parents.
124
:raises: BzrError if the parents are not known.
126
raise NotImplementedError(self.get_parent_ids)
128
def has_filename(self, filename):
129
"""True if the tree has given filename."""
130
raise NotImplementedError(self.has_filename)
132
def has_id(self, file_id):
133
return self.inventory.has_id(file_id)
135
def __contains__(self, file_id):
136
return self.has_id(file_id)
138
def has_or_had_id(self, file_id):
139
if file_id == self.inventory.root.file_id:
141
return self.inventory.has_id(file_id)
143
def is_ignored(self, filename):
144
"""Check whether the filename is ignored by this tree.
146
:param filename: The relative filename within the tree.
147
:return: True if the filename is ignored.
152
return iter(self.inventory)
154
def all_file_ids(self):
155
"""Iterate through all file ids, including ids for missing files."""
156
return set(self.inventory)
158
def id2path(self, file_id):
159
"""Return the path for a file id.
163
return self.inventory.id2path(file_id)
165
def is_control_filename(self, filename):
166
"""True if filename is the name of a control file in this tree.
168
:param filename: A filename within the tree. This is a relative path
169
from the root of this tree.
171
This is true IF and ONLY IF the filename is part of the meta data
172
that bzr controls in this tree. I.E. a random .bzr directory placed
173
on disk will not be a control file for this tree.
175
return self.bzrdir.is_control_filename(filename)
178
def iter_entries_by_dir(self, specific_file_ids=None):
179
"""Walk the tree in 'by_dir' order.
181
This will yield each entry in the tree as a (path, entry) tuple.
182
The order that they are yielded is:
184
Directories are walked in a depth-first lexicographical order,
185
however, whenever a directory is reached, all of its direct child
186
nodes are yielded in lexicographical order before yielding the
189
For example, in the tree::
199
The yield order (ignoring root) would be::
200
a, f, a/b, a/d, a/b/c, a/d/e, f/g
202
return self.inventory.iter_entries_by_dir(
203
specific_file_ids=specific_file_ids)
205
def iter_references(self):
206
for path, entry in self.iter_entries_by_dir():
207
if entry.kind == 'tree-reference':
208
yield path, entry.file_id
210
def kind(self, file_id):
211
raise NotImplementedError("Tree subclass %s must implement kind"
212
% self.__class__.__name__)
214
def stored_kind(self, file_id):
215
"""File kind stored for this file_id.
217
May not match kind on disk for working trees. Always available
218
for versioned files, even when the file itself is missing.
220
return self.kind(file_id)
222
def path_content_summary(self, path):
223
"""Get a summary of the information about path.
225
:param path: A relative path within the tree.
226
:return: A tuple containing kind, size, exec, sha1-or-link.
227
Kind is always present (see tree.kind()).
228
size is present if kind is file, None otherwise.
229
exec is None unless kind is file and the platform supports the 'x'
231
sha1-or-link is the link target if kind is symlink, or the sha1 if
232
it can be obtained without reading the file.
234
raise NotImplementedError(self.path_content_summary)
236
def get_reference_revision(self, file_id, path=None):
237
raise NotImplementedError("Tree subclass %s must implement "
238
"get_reference_revision"
239
% self.__class__.__name__)
241
def _comparison_data(self, entry, path):
242
"""Return a tuple of kind, executable, stat_value for a file.
244
entry may be None if there is no inventory entry for the file, but
245
path must always be supplied.
247
kind is None if there is no file present (even if an inventory id is
248
present). executable is False for non-file entries.
250
raise NotImplementedError(self._comparison_data)
252
def _file_size(self, entry, stat_value):
253
raise NotImplementedError(self._file_size)
255
def _get_inventory(self):
256
return self._inventory
258
def get_file(self, file_id, path=None):
259
"""Return a file object for the file file_id in the tree.
261
If both file_id and path are defined, it is implementation defined as
262
to which one is used.
264
raise NotImplementedError(self.get_file)
266
def get_file_text(self, file_id, path=None):
267
"""Return the byte content of a file.
269
:param file_id: The file_id of the file.
270
:param path: The path of the file.
271
If both file_id and path are supplied, an implementation may use
274
my_file = self.get_file(file_id, path)
276
return my_file.read()
280
def get_file_lines(self, file_id, path=None):
281
"""Return the content of a file, as lines.
283
:param file_id: The file_id of the file.
284
:param path: The path of the file.
285
If both file_id and path are supplied, an implementation may use
288
osutils.split_lines is always used, rather than calling file.readlines,
289
because the behavior of file.readlines may differ by platform.
291
return osutils.split_lines(self.get_file_text(file_id, path))
293
def get_file_mtime(self, file_id, path=None):
294
"""Return the modification time for a file.
296
:param file_id: The handle for this file.
297
:param path: The path that this file can be found at.
298
These must point to the same object.
300
raise NotImplementedError(self.get_file_mtime)
302
def get_file_size(self, file_id):
303
"""Return the size of a file in bytes.
305
This applies only to regular files. If invoked on directories or
306
symlinks, it will return None.
307
:param file_id: The file-id of the file
309
raise NotImplementedError(self.get_file_size)
311
def get_file_by_path(self, path):
312
return self.get_file(self._inventory.path2id(path), path)
314
def iter_files_bytes(self, desired_files):
315
"""Iterate through file contents.
317
Files will not necessarily be returned in the order they occur in
318
desired_files. No specific order is guaranteed.
320
Yields pairs of identifier, bytes_iterator. identifier is an opaque
321
value supplied by the caller as part of desired_files. It should
322
uniquely identify the file version in the caller's context. (Examples:
323
an index number or a TreeTransform trans_id.)
325
bytes_iterator is an iterable of bytestrings for the file. The
326
kind of iterable and length of the bytestrings are unspecified, but for
327
this implementation, it is a tuple containing a single bytestring with
328
the complete text of the file.
330
:param desired_files: a list of (file_id, identifier) pairs
332
for file_id, identifier in desired_files:
333
# We wrap the string in a tuple so that we can return an iterable
334
# of bytestrings. (Technically, a bytestring is also an iterable
335
# of bytestrings, but iterating through each character is not
337
cur_file = (self.get_file_text(file_id),)
338
yield identifier, cur_file
340
def get_symlink_target(self, file_id):
341
"""Get the target for a given file_id.
343
It is assumed that the caller already knows that file_id is referencing
345
:param file_id: Handle for the symlink entry.
346
:return: The path the symlink points to.
348
raise NotImplementedError(self.get_symlink_target)
350
def get_root_id(self):
351
"""Return the file_id for the root of this tree."""
352
raise NotImplementedError(self.get_root_id)
354
def annotate_iter(self, file_id,
355
default_revision=_mod_revision.CURRENT_REVISION):
356
"""Return an iterator of revision_id, line tuples.
358
For working trees (and mutable trees in general), the special
359
revision_id 'current:' will be used for lines that are new in this
360
tree, e.g. uncommitted changes.
361
:param file_id: The file to produce an annotated version from
362
:param default_revision: For lines that don't match a basis, mark them
363
with this revision id. Not all implementations will make use of
366
raise NotImplementedError(self.annotate_iter)
368
def _get_plan_merge_data(self, file_id, other, base):
369
from bzrlib import merge, versionedfile
370
vf = versionedfile._PlanMergeVersionedFile(file_id)
371
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
372
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
374
last_revision_base = None
376
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
377
return vf, last_revision_a, last_revision_b, last_revision_base
379
def plan_file_merge(self, file_id, other, base=None):
380
"""Generate a merge plan based on annotations.
382
If the file contains uncommitted changes in this tree, they will be
383
attributed to the 'current:' pseudo-revision. If the file contains
384
uncommitted changes in the other tree, they will be assigned to the
385
'other:' pseudo-revision.
387
data = self._get_plan_merge_data(file_id, other, base)
388
vf, last_revision_a, last_revision_b, last_revision_base = data
389
return vf.plan_merge(last_revision_a, last_revision_b,
392
def plan_file_lca_merge(self, file_id, other, base=None):
393
"""Generate a merge plan based lca-newness.
395
If the file contains uncommitted changes in this tree, they will be
396
attributed to the 'current:' pseudo-revision. If the file contains
397
uncommitted changes in the other tree, they will be assigned to the
398
'other:' pseudo-revision.
400
data = self._get_plan_merge_data(file_id, other, base)
401
vf, last_revision_a, last_revision_b, last_revision_base = data
402
return vf.plan_lca_merge(last_revision_a, last_revision_b,
405
def _iter_parent_trees(self):
406
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
407
for revision_id in self.get_parent_ids():
409
yield self.revision_tree(revision_id)
410
except errors.NoSuchRevisionInTree:
411
yield self.repository.revision_tree(revision_id)
414
def _file_revision(revision_tree, file_id):
415
"""Determine the revision associated with a file in a given tree."""
416
revision_tree.lock_read()
418
return revision_tree.inventory[file_id].revision
420
revision_tree.unlock()
422
def _get_file_revision(self, file_id, vf, tree_revision):
423
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
425
if getattr(self, '_repository', None) is None:
426
last_revision = tree_revision
427
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
428
self._iter_parent_trees()]
429
vf.add_lines((file_id, last_revision), parent_keys,
430
self.get_file(file_id).readlines())
431
repo = self.branch.repository
434
last_revision = self._file_revision(self, file_id)
435
base_vf = self._repository.texts
436
if base_vf not in vf.fallback_versionedfiles:
437
vf.fallback_versionedfiles.append(base_vf)
440
inventory = property(_get_inventory,
441
doc="Inventory of this Tree")
443
def _check_retrieved(self, ie, f):
446
fp = fingerprint_file(f)
449
if ie.text_size is not None:
450
if ie.text_size != fp['size']:
451
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
452
["inventory expects %d bytes" % ie.text_size,
453
"file is actually %d bytes" % fp['size'],
454
"store is probably damaged/corrupt"])
456
if ie.text_sha1 != fp['sha1']:
457
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
458
["inventory expects %s" % ie.text_sha1,
459
"file is actually %s" % fp['sha1'],
460
"store is probably damaged/corrupt"])
463
def path2id(self, path):
464
"""Return the id for path in this tree."""
465
return self._inventory.path2id(path)
467
def paths2ids(self, paths, trees=[], require_versioned=True):
468
"""Return all the ids that can be reached by walking from paths.
470
Each path is looked up in this tree and any extras provided in
471
trees, and this is repeated recursively: the children in an extra tree
472
of a directory that has been renamed under a provided path in this tree
473
are all returned, even if none exist under a provided path in this
474
tree, and vice versa.
476
:param paths: An iterable of paths to start converting to ids from.
477
Alternatively, if paths is None, no ids should be calculated and None
478
will be returned. This is offered to make calling the api unconditional
479
for code that *might* take a list of files.
480
:param trees: Additional trees to consider.
481
:param require_versioned: If False, do not raise NotVersionedError if
482
an element of paths is not versioned in this tree and all of trees.
484
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
486
def iter_children(self, file_id):
487
entry = self.iter_entries_by_dir([file_id]).next()[1]
488
for child in getattr(entry, 'children', {}).itervalues():
491
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
492
def print_file(self, file_id):
493
"""Print file with id `file_id` to stdout."""
495
sys.stdout.write(self.get_file_text(file_id))
500
def revision_tree(self, revision_id):
501
"""Obtain a revision tree for the revision revision_id.
503
The intention of this method is to allow access to possibly cached
504
tree data. Implementors of this method should raise NoSuchRevision if
505
the tree is not locally available, even if they could obtain the
506
tree via a repository or some other means. Callers are responsible
507
for finding the ultimate source for a revision tree.
509
:param revision_id: The revision_id of the requested tree.
511
:raises: NoSuchRevision if the tree cannot be obtained.
513
raise errors.NoSuchRevisionInTree(self, revision_id)
516
"""What files are present in this tree and unknown.
518
:return: an iterator over the unknown files.
525
def filter_unversioned_files(self, paths):
526
"""Filter out paths that are versioned.
528
:return: set of paths.
530
# NB: we specifically *don't* call self.has_filename, because for
531
# WorkingTrees that can indicate files that exist on disk but that
533
pred = self.inventory.has_filename
534
return set((p for p in paths if not pred(p)))
536
def walkdirs(self, prefix=""):
537
"""Walk the contents of this tree from path down.
539
This yields all the data about the contents of a directory at a time.
540
After each directory has been yielded, if the caller has mutated the
541
list to exclude some directories, they are then not descended into.
543
The data yielded is of the form:
544
((directory-relpath, directory-path-from-root, directory-fileid),
545
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
546
versioned_kind), ...]),
547
- directory-relpath is the containing dirs relpath from prefix
548
- directory-path-from-root is the containing dirs path from /
549
- directory-fileid is the id of the directory if it is versioned.
550
- relpath is the relative path within the subtree being walked.
551
- basename is the basename
552
- kind is the kind of the file now. If unknonwn then the file is not
553
present within the tree - but it may be recorded as versioned. See
555
- lstat is the stat data *if* the file was statted.
556
- path_from_tree_root is the path from the root of the tree.
557
- file_id is the file_id if the entry is versioned.
558
- versioned_kind is the kind of the file as last recorded in the
559
versioning system. If 'unknown' the file is not versioned.
560
One of 'kind' and 'versioned_kind' must not be 'unknown'.
562
:param prefix: Start walking from prefix within the tree rather than
563
at the root. This allows one to walk a subtree but get paths that are
564
relative to a tree rooted higher up.
565
:return: an iterator over the directory data.
567
raise NotImplementedError(self.walkdirs)
569
def iter_search_rules(self, path_names, pref_names=None,
570
_default_searcher=rules._per_user_searcher):
571
"""Find the preferences for filenames in a tree.
573
:param path_names: an iterable of paths to find attributes for.
574
Paths are given relative to the root of the tree.
575
:param pref_names: the list of preferences to lookup - None for all
576
:param _default_searcher: private parameter to assist testing - don't use
577
:return: an iterator of tuple sequences, one per path-name.
578
See _RulesSearcher.get_items for details on the tuple sequence.
580
searcher = self._get_rules_searcher(_default_searcher)
581
if searcher is not None:
582
if pref_names is not None:
583
for path in path_names:
584
yield searcher.get_selected_items(path, pref_names)
586
for path in path_names:
587
yield searcher.get_items(path)
590
def _get_rules_searcher(self, default_searcher):
591
"""Get the RulesSearcher for this tree given the default one."""
592
searcher = default_searcher
596
class EmptyTree(Tree):
599
self._inventory = Inventory(root_id=None)
600
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
601
' use repository.revision_tree instead.',
602
DeprecationWarning, stacklevel=2)
604
def get_parent_ids(self):
607
def get_symlink_target(self, file_id):
610
def has_filename(self, filename):
613
def kind(self, file_id):
616
def list_files(self, include_root=False):
619
def __contains__(self, file_id):
620
return (file_id in self._inventory)
622
def get_file_sha1(self, file_id, path=None, stat_value=None):
626
######################################################################
629
# TODO: Merge these two functions into a single one that can operate
630
# on either a whole tree or a set of files.
632
# TODO: Return the diff in order by filename, not by category or in
633
# random order. Can probably be done by lock-stepping through the
634
# filenames from both trees.
637
def file_status(filename, old_tree, new_tree):
638
"""Return single-letter status, old and new names for a file.
640
The complexity here is in deciding how to represent renames;
641
many complex cases are possible.
643
old_inv = old_tree.inventory
644
new_inv = new_tree.inventory
645
new_id = new_inv.path2id(filename)
646
old_id = old_inv.path2id(filename)
648
if not new_id and not old_id:
649
# easy: doesn't exist in either; not versioned at all
650
if new_tree.is_ignored(filename):
651
return 'I', None, None
653
return '?', None, None
655
# There is now a file of this name, great.
658
# There is no longer a file of this name, but we can describe
659
# what happened to the file that used to have
660
# this name. There are two possibilities: either it was
661
# deleted entirely, or renamed.
662
if new_inv.has_id(old_id):
663
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
665
return 'D', old_inv.id2path(old_id), None
667
# if the file_id is new in this revision, it is added
668
if new_id and not old_inv.has_id(new_id):
671
# if there used to be a file of this name, but that ID has now
672
# disappeared, it is deleted
673
if old_id and not new_inv.has_id(old_id):
680
def find_renames(old_inv, new_inv):
681
for file_id in old_inv:
682
if file_id not in new_inv:
684
old_name = old_inv.id2path(file_id)
685
new_name = new_inv.id2path(file_id)
686
if old_name != new_name:
687
yield (old_name, new_name)
690
def find_ids_across_trees(filenames, trees, require_versioned=True):
691
"""Find the ids corresponding to specified filenames.
693
All matches in all trees will be used, and all children of matched
694
directories will be used.
696
:param filenames: The filenames to find file_ids for (if None, returns
698
:param trees: The trees to find file_ids within
699
:param require_versioned: if true, all specified filenames must occur in
701
:return: a set of file ids for the specified filenames and their children.
705
specified_path_ids = _find_ids_across_trees(filenames, trees,
707
return _find_children_across_trees(specified_path_ids, trees)
710
def _find_ids_across_trees(filenames, trees, require_versioned):
711
"""Find the ids corresponding to specified filenames.
713
All matches in all trees will be used, but subdirectories are not scanned.
715
:param filenames: The filenames to find file_ids for
716
:param trees: The trees to find file_ids within
717
:param require_versioned: if true, all specified filenames must occur in
719
:return: a set of file ids for the specified filenames
722
interesting_ids = set()
723
for tree_path in filenames:
726
file_id = tree.path2id(tree_path)
727
if file_id is not None:
728
interesting_ids.add(file_id)
731
not_versioned.append(tree_path)
732
if len(not_versioned) > 0 and require_versioned:
733
raise errors.PathsNotVersionedError(not_versioned)
734
return interesting_ids
737
def _find_children_across_trees(specified_ids, trees):
738
"""Return a set including specified ids and their children.
740
All matches in all trees will be used.
742
:param trees: The trees to find file_ids within
743
:return: a set containing all specified ids and their children
745
interesting_ids = set(specified_ids)
746
pending = interesting_ids
747
# now handle children of interesting ids
748
# we loop so that we handle all children of each id in both trees
749
while len(pending) > 0:
751
for file_id in pending:
753
if not tree.has_id(file_id):
755
for child_id in tree.iter_children(file_id):
756
if child_id not in interesting_ids:
757
new_pending.add(child_id)
758
interesting_ids.update(new_pending)
759
pending = new_pending
760
return interesting_ids
763
class InterTree(InterObject):
764
"""This class represents operations taking place between two Trees.
766
Its instances have methods like 'compare' and contain references to the
767
source and target trees these operations are to be carried out on.
769
Clients of bzrlib should not need to use InterTree directly, rather they
770
should use the convenience methods on Tree such as 'Tree.compare()' which
771
will pass through to InterTree as appropriate.
777
def compare(self, want_unchanged=False, specific_files=None,
778
extra_trees=None, require_versioned=False, include_root=False,
779
want_unversioned=False):
780
"""Return the changes from source to target.
782
:return: A TreeDelta.
783
:param specific_files: An optional list of file paths to restrict the
784
comparison to. When mapping filenames to ids, all matches in all
785
trees (including optional extra_trees) are used, and all children of
786
matched directories are included.
787
:param want_unchanged: An optional boolean requesting the inclusion of
788
unchanged entries in the result.
789
:param extra_trees: An optional list of additional trees to use when
790
mapping the contents of specific_files (paths) to file_ids.
791
:param require_versioned: An optional boolean (defaults to False). When
792
supplied and True all the 'specific_files' must be versioned, or
793
a PathsNotVersionedError will be thrown.
794
:param want_unversioned: Scan for unversioned paths.
796
# NB: show_status depends on being able to pass in non-versioned files
797
# and report them as unknown
798
trees = (self.source,)
799
if extra_trees is not None:
800
trees = trees + tuple(extra_trees)
801
# target is usually the newer tree:
802
specific_file_ids = self.target.paths2ids(specific_files, trees,
803
require_versioned=require_versioned)
804
if specific_files and not specific_file_ids:
805
# All files are unversioned, so just return an empty delta
806
# _compare_trees would think we want a complete delta
807
result = delta.TreeDelta()
808
fake_entry = InventoryFile('unused', 'unused', 'unused')
809
result.unversioned = [(path, None,
810
self.target._comparison_data(fake_entry, path)[0]) for path in
813
return delta._compare_trees(self.source, self.target, want_unchanged,
814
specific_files, include_root, extra_trees=extra_trees,
815
require_versioned=require_versioned,
816
want_unversioned=want_unversioned)
818
def iter_changes(self, include_unchanged=False,
819
specific_files=None, pb=None, extra_trees=[],
820
require_versioned=True, want_unversioned=False):
821
"""Generate an iterator of changes between trees.
824
(file_id, (path_in_source, path_in_target),
825
changed_content, versioned, parent, name, kind,
828
Changed_content is True if the file's content has changed. This
829
includes changes to its kind, and to a symlink's target.
831
versioned, parent, name, kind, executable are tuples of (from, to).
832
If a file is missing in a tree, its kind is None.
834
Iteration is done in parent-to-child order, relative to the target
837
There is no guarantee that all paths are in sorted order: the
838
requirement to expand the search due to renames may result in children
839
that should be found early being found late in the search, after
840
lexically later results have been returned.
841
:param require_versioned: Raise errors.PathsNotVersionedError if a
842
path in the specific_files list is not versioned in one of
843
source, target or extra_trees.
844
:param want_unversioned: Should unversioned files be returned in the
845
output. An unversioned file is defined as one with (False, False)
846
for the versioned pair.
849
lookup_trees = [self.source]
851
lookup_trees.extend(extra_trees)
852
if specific_files == []:
853
specific_file_ids = []
855
specific_file_ids = self.target.paths2ids(specific_files,
856
lookup_trees, require_versioned=require_versioned)
858
all_unversioned = sorted([(p.split('/'), p) for p in
860
if specific_files is None or
861
osutils.is_inside_any(specific_files, p)])
862
all_unversioned = deque(all_unversioned)
864
all_unversioned = deque()
866
from_entries_by_dir = list(self.source.iter_entries_by_dir(
867
specific_file_ids=specific_file_ids))
868
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
869
to_entries_by_dir = list(self.target.iter_entries_by_dir(
870
specific_file_ids=specific_file_ids))
871
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
873
# the unversioned path lookup only occurs on real trees - where there
874
# can be extras. So the fake_entry is solely used to look up
875
# executable it values when execute is not supported.
876
fake_entry = InventoryFile('unused', 'unused', 'unused')
877
for to_path, to_entry in to_entries_by_dir:
878
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
879
unversioned_path = all_unversioned.popleft()
880
to_kind, to_executable, to_stat = \
881
self.target._comparison_data(fake_entry, unversioned_path[1])
882
yield (None, (None, unversioned_path[1]), True, (False, False),
884
(None, unversioned_path[0][-1]),
886
(None, to_executable))
887
file_id = to_entry.file_id
888
to_paths[file_id] = to_path
890
changed_content = False
891
from_path, from_entry = from_data.get(file_id, (None, None))
892
from_versioned = (from_entry is not None)
893
if from_entry is not None:
894
from_versioned = True
895
from_name = from_entry.name
896
from_parent = from_entry.parent_id
897
from_kind, from_executable, from_stat = \
898
self.source._comparison_data(from_entry, from_path)
901
from_versioned = False
905
from_executable = None
906
versioned = (from_versioned, True)
907
to_kind, to_executable, to_stat = \
908
self.target._comparison_data(to_entry, to_path)
909
kind = (from_kind, to_kind)
910
if kind[0] != kind[1]:
911
changed_content = True
912
elif from_kind == 'file':
913
from_size = self.source._file_size(from_entry, from_stat)
914
to_size = self.target._file_size(to_entry, to_stat)
915
if from_size != to_size:
916
changed_content = True
917
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
918
self.target.get_file_sha1(file_id, to_path, to_stat)):
919
changed_content = True
920
elif from_kind == 'symlink':
921
if (self.source.get_symlink_target(file_id) !=
922
self.target.get_symlink_target(file_id)):
923
changed_content = True
924
elif from_kind == 'tree-reference':
925
if (self.source.get_reference_revision(file_id, from_path)
926
!= self.target.get_reference_revision(file_id, to_path)):
927
changed_content = True
928
parent = (from_parent, to_entry.parent_id)
929
name = (from_name, to_entry.name)
930
executable = (from_executable, to_executable)
932
pb.update('comparing files', entry_count, num_entries)
933
if (changed_content is not False or versioned[0] != versioned[1]
934
or parent[0] != parent[1] or name[0] != name[1] or
935
executable[0] != executable[1] or include_unchanged):
936
yield (file_id, (from_path, to_path), changed_content,
937
versioned, parent, name, kind, executable)
939
while all_unversioned:
940
# yield any trailing unversioned paths
941
unversioned_path = all_unversioned.popleft()
942
to_kind, to_executable, to_stat = \
943
self.target._comparison_data(fake_entry, unversioned_path[1])
944
yield (None, (None, unversioned_path[1]), True, (False, False),
946
(None, unversioned_path[0][-1]),
948
(None, to_executable))
950
def get_to_path(to_entry):
951
if to_entry.parent_id is None:
952
to_path = '' # the root
954
if to_entry.parent_id not in to_paths:
956
return get_to_path(self.target.inventory[to_entry.parent_id])
957
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
959
to_paths[to_entry.file_id] = to_path
962
for path, from_entry in from_entries_by_dir:
963
file_id = from_entry.file_id
964
if file_id in to_paths:
967
if not file_id in self.target.all_file_ids():
968
# common case - paths we have not emitted are not present in
972
to_path = get_to_path(self.target.inventory[file_id])
975
pb.update('comparing files', entry_count, num_entries)
976
versioned = (True, False)
977
parent = (from_entry.parent_id, None)
978
name = (from_entry.name, None)
979
from_kind, from_executable, stat_value = \
980
self.source._comparison_data(from_entry, path)
981
kind = (from_kind, None)
982
executable = (from_executable, None)
983
changed_content = from_kind is not None
984
# the parent's path is necessarily known at this point.
985
yield(file_id, (path, to_path), changed_content, versioned, parent,
986
name, kind, executable)
989
class MultiWalker(object):
990
"""Walk multiple trees simultaneously, getting combined results."""
992
# Note: This could be written to not assume you can do out-of-order
993
# lookups. Instead any nodes that don't match in all trees could be
994
# marked as 'deferred', and then returned in the final cleanup loop.
995
# For now, I think it is "nicer" to return things as close to the
996
# "master_tree" order as we can.
998
def __init__(self, master_tree, other_trees):
999
"""Create a new MultiWalker.
1001
All trees being walked must implement "iter_entries_by_dir()", such
1002
that they yield (path, object) tuples, where that object will have a
1003
'.file_id' member, that can be used to check equality.
1005
:param master_tree: All trees will be 'slaved' to the master_tree such
1006
that nodes in master_tree will be used as 'first-pass' sync points.
1007
Any nodes that aren't in master_tree will be merged in a second
1009
:param other_trees: A list of other trees to walk simultaneously.
1011
self._master_tree = master_tree
1012
self._other_trees = other_trees
1014
# Keep track of any nodes that were properly processed just out of
1015
# order, that way we don't return them at the end, we don't have to
1016
# track *all* processed file_ids, just the out-of-order ones
1017
self._out_of_order_processed = set()
1020
def _step_one(iterator):
1021
"""Step an iter_entries_by_dir iterator.
1023
:return: (has_more, path, ie)
1024
If has_more is False, path and ie will be None.
1027
path, ie = iterator.next()
1028
except StopIteration:
1029
return False, None, None
1031
return True, path, ie
1034
def _cmp_path_by_dirblock(path1, path2):
1035
"""Compare two paths based on what directory they are in.
1037
This generates a sort order, such that all children of a directory are
1038
sorted together, and grandchildren are in the same order as the
1039
children appear. But all grandchildren come after all children.
1041
:param path1: first path
1042
:param path2: the second path
1043
:return: negative number if ``path1`` comes first,
1044
0 if paths are equal
1045
and a positive number if ``path2`` sorts first
1047
# Shortcut this special case
1050
# This is stolen from _dirstate_helpers_py.py, only switching it to
1051
# Unicode objects. Consider using encode_utf8() and then using the
1052
# optimized versions, or maybe writing optimized unicode versions.
1053
if not isinstance(path1, unicode):
1054
raise TypeError("'path1' must be a unicode string, not %s: %r"
1055
% (type(path1), path1))
1056
if not isinstance(path2, unicode):
1057
raise TypeError("'path2' must be a unicode string, not %s: %r"
1058
% (type(path2), path2))
1059
return cmp(MultiWalker._path_to_key(path1),
1060
MultiWalker._path_to_key(path2))
1063
def _path_to_key(path):
1064
dirname, basename = osutils.split(path)
1065
return (dirname.split(u'/'), basename)
1067
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1068
"""Lookup an inventory entry by file_id.
1070
This is called when an entry is missing in the normal order.
1071
Generally this is because a file was either renamed, or it was
1072
deleted/added. If the entry was found in the inventory and not in
1073
extra_entries, it will be added to self._out_of_order_processed
1075
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1076
should be filled with entries that were found before they were
1077
used. If file_id is present, it will be removed from the
1079
:param other_tree: The Tree to search, in case we didn't find the entry
1081
:param file_id: The file_id to look for
1082
:return: (path, ie) if found or (None, None) if not present.
1084
if file_id in extra_entries:
1085
return extra_entries.pop(file_id)
1086
# TODO: Is id2path better as the first call, or is
1087
# inventory[file_id] better as a first check?
1089
cur_path = other_tree.id2path(file_id)
1090
except errors.NoSuchId:
1092
if cur_path is None:
1095
self._out_of_order_processed.add(file_id)
1096
cur_ie = other_tree.inventory[file_id]
1097
return (cur_path, cur_ie)
1100
"""Match up the values in the different trees."""
1101
for result in self._walk_master_tree():
1103
self._finish_others()
1104
for result in self._walk_others():
1107
def _walk_master_tree(self):
1108
"""First pass, walk all trees in lock-step.
1110
When we are done, all nodes in the master_tree will have been
1111
processed. _other_walkers, _other_entries, and _others_extra will be
1112
set on 'self' for future processing.
1114
# This iterator has the most "inlining" done, because it tends to touch
1115
# every file in the tree, while the others only hit nodes that don't
1117
master_iterator = self._master_tree.iter_entries_by_dir()
1119
other_walkers = [other.iter_entries_by_dir()
1120
for other in self._other_trees]
1121
other_entries = [self._step_one(walker) for walker in other_walkers]
1122
# Track extra nodes in the other trees
1123
others_extra = [{} for i in xrange(len(self._other_trees))]
1125
master_has_more = True
1126
step_one = self._step_one
1127
lookup_by_file_id = self._lookup_by_file_id
1128
out_of_order_processed = self._out_of_order_processed
1130
while master_has_more:
1131
(master_has_more, path, master_ie) = step_one(master_iterator)
1132
if not master_has_more:
1135
file_id = master_ie.file_id
1137
other_values_append = other_values.append
1138
next_other_entries = []
1139
next_other_entries_append = next_other_entries.append
1140
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1141
if not other_has_more:
1142
other_values_append(lookup_by_file_id(
1143
others_extra[idx], self._other_trees[idx], file_id))
1144
next_other_entries_append((False, None, None))
1145
elif file_id == other_ie.file_id:
1146
# This is the critical code path, as most of the entries
1147
# should match between most trees.
1148
other_values_append((other_path, other_ie))
1149
next_other_entries_append(step_one(other_walkers[idx]))
1151
# This walker did not match, step it until it either
1152
# matches, or we know we are past the current walker.
1153
other_walker = other_walkers[idx]
1154
other_extra = others_extra[idx]
1155
while (other_has_more and
1156
self._cmp_path_by_dirblock(other_path, path) < 0):
1157
other_file_id = other_ie.file_id
1158
if other_file_id not in out_of_order_processed:
1159
other_extra[other_file_id] = (other_path, other_ie)
1160
other_has_more, other_path, other_ie = \
1161
step_one(other_walker)
1162
if other_has_more and other_ie.file_id == file_id:
1163
# We ended up walking to this point, match and step
1165
other_values_append((other_path, other_ie))
1166
other_has_more, other_path, other_ie = \
1167
step_one(other_walker)
1169
# This record isn't in the normal order, see if it
1171
other_values_append(lookup_by_file_id(
1172
other_extra, self._other_trees[idx], file_id))
1173
next_other_entries_append((other_has_more, other_path,
1175
other_entries = next_other_entries
1177
# We've matched all the walkers, yield this datapoint
1178
yield path, file_id, master_ie, other_values
1179
self._other_walkers = other_walkers
1180
self._other_entries = other_entries
1181
self._others_extra = others_extra
1183
def _finish_others(self):
1184
"""Finish walking the other iterators, so we get all entries."""
1185
for idx, info in enumerate(self._other_entries):
1186
other_extra = self._others_extra[idx]
1187
(other_has_more, other_path, other_ie) = info
1188
while other_has_more:
1189
other_file_id = other_ie.file_id
1190
if other_file_id not in self._out_of_order_processed:
1191
other_extra[other_file_id] = (other_path, other_ie)
1192
other_has_more, other_path, other_ie = \
1193
self._step_one(self._other_walkers[idx])
1194
del self._other_entries
1196
def _walk_others(self):
1197
"""Finish up by walking all the 'deferred' nodes."""
1198
# TODO: One alternative would be to grab all possible unprocessed
1199
# file_ids, and then sort by path, and then yield them. That
1200
# might ensure better ordering, in case a caller strictly
1201
# requires parents before children.
1202
for idx, other_extra in enumerate(self._others_extra):
1203
others = sorted(other_extra.itervalues(),
1204
key=lambda x: self._path_to_key(x[0]))
1205
for other_path, other_ie in others:
1206
file_id = other_ie.file_id
1207
# We don't need to check out_of_order_processed here, because
1208
# the lookup_by_file_id will be removing anything processed
1209
# from the extras cache
1210
other_extra.pop(file_id)
1211
other_values = [(None, None) for i in xrange(idx)]
1212
other_values.append((other_path, other_ie))
1213
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1214
alt_idx = alt_idx + idx + 1
1215
alt_extra = self._others_extra[alt_idx]
1216
alt_tree = self._other_trees[alt_idx]
1217
other_values.append(self._lookup_by_file_id(
1218
alt_extra, alt_tree, file_id))
1219
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
