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
28
revision as _mod_revision,
29
conflicts as _mod_conflicts,
32
from bzrlib.decorators import needs_read_lock
33
from bzrlib.errors import BzrError, BzrCheckError
34
from bzrlib import errors
35
from bzrlib.inventory import Inventory, InventoryFile
36
from bzrlib.inter import InterObject
37
from bzrlib.osutils import fingerprint_file
38
import bzrlib.revision
39
from bzrlib.trace import mutter, note
43
"""Abstract file tree.
45
There are several subclasses:
47
* `WorkingTree` exists as files on disk editable by the user.
49
* `RevisionTree` is a tree as recorded at some point in the past.
51
Trees contain an `Inventory` object, and also know how to retrieve
52
file texts mentioned in the inventory, either from a working
53
directory or from a store.
55
It is possible for trees to contain files that are not described
56
in their inventory or vice versa; for this use `filenames()`.
58
Trees can be compared, etc, regardless of whether they are working
59
trees or versioned trees.
62
def changes_from(self, other, want_unchanged=False, specific_files=None,
63
extra_trees=None, require_versioned=False, include_root=False,
64
want_unversioned=False):
65
"""Return a TreeDelta of the changes from other to this tree.
67
:param other: A tree to compare with.
68
:param specific_files: An optional list of file paths to restrict the
69
comparison to. When mapping filenames to ids, all matches in all
70
trees (including optional extra_trees) are used, and all children of
71
matched directories are included.
72
:param want_unchanged: An optional boolean requesting the inclusion of
73
unchanged entries in the result.
74
:param extra_trees: An optional list of additional trees to use when
75
mapping the contents of specific_files (paths) to file_ids.
76
:param require_versioned: An optional boolean (defaults to False). When
77
supplied and True all the 'specific_files' must be versioned, or
78
a PathsNotVersionedError will be thrown.
79
:param want_unversioned: Scan for unversioned paths.
81
The comparison will be performed by an InterTree object looked up on
84
# Martin observes that Tree.changes_from returns a TreeDelta and this
85
# may confuse people, because the class name of the returned object is
86
# a synonym of the object referenced in the method name.
87
return InterTree.get(other, self).compare(
88
want_unchanged=want_unchanged,
89
specific_files=specific_files,
90
extra_trees=extra_trees,
91
require_versioned=require_versioned,
92
include_root=include_root,
93
want_unversioned=want_unversioned,
96
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
97
def _iter_changes(self, *args, **kwargs):
98
return self.iter_changes(*args, **kwargs)
100
def iter_changes(self, from_tree, include_unchanged=False,
101
specific_files=None, pb=None, extra_trees=None,
102
require_versioned=True, want_unversioned=False):
103
intertree = InterTree.get(from_tree, self)
104
return intertree.iter_changes(include_unchanged, specific_files, pb,
105
extra_trees, require_versioned, want_unversioned=want_unversioned)
108
"""Get a list of the conflicts in the tree.
110
Each conflict is an instance of bzrlib.conflicts.Conflict.
112
return _mod_conflicts.ConflictList()
115
"""For trees that can have unversioned files, return all such paths."""
118
def get_parent_ids(self):
119
"""Get the parent ids for this tree.
121
:return: a list of parent ids. [] is returned to indicate
122
a tree with no parents.
123
:raises: BzrError if the parents are not known.
125
raise NotImplementedError(self.get_parent_ids)
127
def has_filename(self, filename):
128
"""True if the tree has given filename."""
129
raise NotImplementedError(self.has_filename)
131
def has_id(self, file_id):
132
return self.inventory.has_id(file_id)
134
__contains__ = has_id
136
def has_or_had_id(self, file_id):
137
if file_id == self.inventory.root.file_id:
139
return self.inventory.has_id(file_id)
141
def is_ignored(self, filename):
142
"""Check whether the filename is ignored by this tree.
144
:param filename: The relative filename within the tree.
145
:return: True if the filename is ignored.
150
return iter(self.inventory)
152
def all_file_ids(self):
153
"""Iterate through all file ids, including ids for missing files."""
154
return set(self.inventory)
156
def id2path(self, file_id):
157
"""Return the path for a file id.
161
return self.inventory.id2path(file_id)
163
def is_control_filename(self, filename):
164
"""True if filename is the name of a control file in this tree.
166
:param filename: A filename within the tree. This is a relative path
167
from the root of this tree.
169
This is true IF and ONLY IF the filename is part of the meta data
170
that bzr controls in this tree. I.E. a random .bzr directory placed
171
on disk will not be a control file for this tree.
173
return self.bzrdir.is_control_filename(filename)
176
def iter_entries_by_dir(self, specific_file_ids=None):
177
"""Walk the tree in 'by_dir' order.
179
This will yield each entry in the tree as a (path, entry) tuple. The
180
order that they are yielded is: the contents of a directory are
181
preceeded by the parent of a directory, and all the contents of a
182
directory are grouped together.
184
return self.inventory.iter_entries_by_dir(
185
specific_file_ids=specific_file_ids)
187
def iter_references(self):
188
for path, entry in self.iter_entries_by_dir():
189
if entry.kind == 'tree-reference':
190
yield path, entry.file_id
192
def kind(self, file_id):
193
raise NotImplementedError("Tree subclass %s must implement kind"
194
% self.__class__.__name__)
196
def stored_kind(self, file_id):
197
"""File kind stored for this file_id.
199
May not match kind on disk for working trees. Always available
200
for versioned files, even when the file itself is missing.
202
return self.kind(file_id)
204
def path_content_summary(self, path):
205
"""Get a summary of the information about path.
207
:param path: A relative path within the tree.
208
:return: A tuple containing kind, size, exec, sha1-or-link.
209
Kind is always present (see tree.kind()).
210
size is present if kind is file, None otherwise.
211
exec is None unless kind is file and the platform supports the 'x'
213
sha1-or-link is the link target if kind is symlink, or the sha1 if
214
it can be obtained without reading the file.
216
raise NotImplementedError(self.path_content_summary)
218
def get_reference_revision(self, file_id, path=None):
219
raise NotImplementedError("Tree subclass %s must implement "
220
"get_reference_revision"
221
% self.__class__.__name__)
223
def _comparison_data(self, entry, path):
224
"""Return a tuple of kind, executable, stat_value for a file.
226
entry may be None if there is no inventory entry for the file, but
227
path must always be supplied.
229
kind is None if there is no file present (even if an inventory id is
230
present). executable is False for non-file entries.
232
raise NotImplementedError(self._comparison_data)
234
def _file_size(self, entry, stat_value):
235
raise NotImplementedError(self._file_size)
237
def _get_inventory(self):
238
return self._inventory
240
def get_file(self, file_id, path=None):
241
"""Return a file object for the file file_id in the tree.
243
If both file_id and path are defined, it is implementation defined as
244
to which one is used.
246
raise NotImplementedError(self.get_file)
248
def get_file_mtime(self, file_id, path=None):
249
"""Return the modification time for a file.
251
:param file_id: The handle for this file.
252
:param path: The path that this file can be found at.
253
These must point to the same object.
255
raise NotImplementedError(self.get_file_mtime)
257
def get_file_size(self, file_id):
258
"""Return the size of a file in bytes.
260
This applies only to regular files. If invoked on directories or
261
symlinks, it will return None.
262
:param file_id: The file-id of the file
264
raise NotImplementedError(self.get_file_size)
266
def get_file_by_path(self, path):
267
return self.get_file(self._inventory.path2id(path), path)
269
def iter_files_bytes(self, desired_files):
270
"""Iterate through file contents.
272
Files will not necessarily be returned in the order they occur in
273
desired_files. No specific order is guaranteed.
275
Yields pairs of identifier, bytes_iterator. identifier is an opaque
276
value supplied by the caller as part of desired_files. It should
277
uniquely identify the file version in the caller's context. (Examples:
278
an index number or a TreeTransform trans_id.)
280
bytes_iterator is an iterable of bytestrings for the file. The
281
kind of iterable and length of the bytestrings are unspecified, but for
282
this implementation, it is a tuple containing a single bytestring with
283
the complete text of the file.
285
:param desired_files: a list of (file_id, identifier) pairs
287
for file_id, identifier in desired_files:
288
# We wrap the string in a tuple so that we can return an iterable
289
# of bytestrings. (Technically, a bytestring is also an iterable
290
# of bytestrings, but iterating through each character is not
292
cur_file = (self.get_file_text(file_id),)
293
yield identifier, cur_file
295
def get_symlink_target(self, file_id):
296
"""Get the target for a given file_id.
298
It is assumed that the caller already knows that file_id is referencing
300
:param file_id: Handle for the symlink entry.
301
:return: The path the symlink points to.
303
raise NotImplementedError(self.get_symlink_target)
305
def get_root_id(self):
306
"""Return the file_id for the root of this tree."""
307
raise NotImplementedError(self.get_root_id)
309
def annotate_iter(self, file_id,
310
default_revision=_mod_revision.CURRENT_REVISION):
311
"""Return an iterator of revision_id, line tuples.
313
For working trees (and mutable trees in general), the special
314
revision_id 'current:' will be used for lines that are new in this
315
tree, e.g. uncommitted changes.
316
:param file_id: The file to produce an annotated version from
317
:param default_revision: For lines that don't match a basis, mark them
318
with this revision id. Not all implementations will make use of
321
raise NotImplementedError(self.annotate_iter)
323
def _get_plan_merge_data(self, file_id, other, base):
324
from bzrlib import merge, versionedfile
325
vf = versionedfile._PlanMergeVersionedFile(file_id)
326
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
327
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
329
last_revision_base = None
331
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
332
return vf, last_revision_a, last_revision_b, last_revision_base
334
def plan_file_merge(self, file_id, other, base=None):
335
"""Generate a merge plan based on annotations.
337
If the file contains uncommitted changes in this tree, they will be
338
attributed to the 'current:' pseudo-revision. If the file contains
339
uncommitted changes in the other tree, they will be assigned to the
340
'other:' pseudo-revision.
342
data = self._get_plan_merge_data(file_id, other, base)
343
vf, last_revision_a, last_revision_b, last_revision_base = data
344
return vf.plan_merge(last_revision_a, last_revision_b,
347
def plan_file_lca_merge(self, file_id, other, base=None):
348
"""Generate a merge plan based lca-newness.
350
If the file contains uncommitted changes in this tree, they will be
351
attributed to the 'current:' pseudo-revision. If the file contains
352
uncommitted changes in the other tree, they will be assigned to the
353
'other:' pseudo-revision.
355
data = self._get_plan_merge_data(file_id, other, base)
356
vf, last_revision_a, last_revision_b, last_revision_base = data
357
return vf.plan_lca_merge(last_revision_a, last_revision_b,
360
def _get_file_revision(self, file_id, vf, tree_revision):
361
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
362
def file_revision(revision_tree):
363
revision_tree.lock_read()
365
return revision_tree.inventory[file_id].revision
367
revision_tree.unlock()
369
def iter_parent_trees():
370
for revision_id in self.get_parent_ids():
372
yield self.revision_tree(revision_id)
374
yield self.repository.revision_tree(revision_id)
376
if getattr(self, '_repository', None) is None:
377
last_revision = tree_revision
378
parent_keys = [(file_id, file_revision(t)) for t in
380
vf.add_lines((file_id, last_revision), parent_keys,
381
self.get_file(file_id).readlines())
382
repo = self.branch.repository
385
last_revision = file_revision(self)
386
base_vf = self._repository.texts
387
if base_vf not in vf.fallback_versionedfiles:
388
vf.fallback_versionedfiles.append(base_vf)
391
inventory = property(_get_inventory,
392
doc="Inventory of this Tree")
394
def _check_retrieved(self, ie, f):
397
fp = fingerprint_file(f)
400
if ie.text_size is not None:
401
if ie.text_size != fp['size']:
402
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
403
["inventory expects %d bytes" % ie.text_size,
404
"file is actually %d bytes" % fp['size'],
405
"store is probably damaged/corrupt"])
407
if ie.text_sha1 != fp['sha1']:
408
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
409
["inventory expects %s" % ie.text_sha1,
410
"file is actually %s" % fp['sha1'],
411
"store is probably damaged/corrupt"])
414
def path2id(self, path):
415
"""Return the id for path in this tree."""
416
return self._inventory.path2id(path)
418
def paths2ids(self, paths, trees=[], require_versioned=True):
419
"""Return all the ids that can be reached by walking from paths.
421
Each path is looked up in this tree and any extras provided in
422
trees, and this is repeated recursively: the children in an extra tree
423
of a directory that has been renamed under a provided path in this tree
424
are all returned, even if none exist under a provided path in this
425
tree, and vice versa.
427
:param paths: An iterable of paths to start converting to ids from.
428
Alternatively, if paths is None, no ids should be calculated and None
429
will be returned. This is offered to make calling the api unconditional
430
for code that *might* take a list of files.
431
:param trees: Additional trees to consider.
432
:param require_versioned: If False, do not raise NotVersionedError if
433
an element of paths is not versioned in this tree and all of trees.
435
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
437
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
438
def print_file(self, file_id):
439
"""Print file with id `file_id` to stdout."""
441
sys.stdout.write(self.get_file_text(file_id))
446
def revision_tree(self, revision_id):
447
"""Obtain a revision tree for the revision revision_id.
449
The intention of this method is to allow access to possibly cached
450
tree data. Implementors of this method should raise NoSuchRevision if
451
the tree is not locally available, even if they could obtain the
452
tree via a repository or some other means. Callers are responsible
453
for finding the ultimate source for a revision tree.
455
:param revision_id: The revision_id of the requested tree.
457
:raises: NoSuchRevision if the tree cannot be obtained.
459
raise errors.NoSuchRevisionInTree(self, revision_id)
462
"""What files are present in this tree and unknown.
464
:return: an iterator over the unknown files.
471
def filter_unversioned_files(self, paths):
472
"""Filter out paths that are versioned.
474
:return: set of paths.
476
# NB: we specifically *don't* call self.has_filename, because for
477
# WorkingTrees that can indicate files that exist on disk but that
479
pred = self.inventory.has_filename
480
return set((p for p in paths if not pred(p)))
482
def walkdirs(self, prefix=""):
483
"""Walk the contents of this tree from path down.
485
This yields all the data about the contents of a directory at a time.
486
After each directory has been yielded, if the caller has mutated the
487
list to exclude some directories, they are then not descended into.
489
The data yielded is of the form:
490
((directory-relpath, directory-path-from-root, directory-fileid),
491
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
492
versioned_kind), ...]),
493
- directory-relpath is the containing dirs relpath from prefix
494
- directory-path-from-root is the containing dirs path from /
495
- directory-fileid is the id of the directory if it is versioned.
496
- relpath is the relative path within the subtree being walked.
497
- basename is the basename
498
- kind is the kind of the file now. If unknonwn then the file is not
499
present within the tree - but it may be recorded as versioned. See
501
- lstat is the stat data *if* the file was statted.
502
- path_from_tree_root is the path from the root of the tree.
503
- file_id is the file_id if the entry is versioned.
504
- versioned_kind is the kind of the file as last recorded in the
505
versioning system. If 'unknown' the file is not versioned.
506
One of 'kind' and 'versioned_kind' must not be 'unknown'.
508
:param prefix: Start walking from prefix within the tree rather than
509
at the root. This allows one to walk a subtree but get paths that are
510
relative to a tree rooted higher up.
511
:return: an iterator over the directory data.
513
raise NotImplementedError(self.walkdirs)
516
class EmptyTree(Tree):
519
self._inventory = Inventory(root_id=None)
520
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
521
' use repository.revision_tree instead.',
522
DeprecationWarning, stacklevel=2)
524
def get_parent_ids(self):
527
def get_symlink_target(self, file_id):
530
def has_filename(self, filename):
533
def kind(self, file_id):
536
def list_files(self, include_root=False):
539
def __contains__(self, file_id):
540
return (file_id in self._inventory)
542
def get_file_sha1(self, file_id, path=None, stat_value=None):
546
######################################################################
549
# TODO: Merge these two functions into a single one that can operate
550
# on either a whole tree or a set of files.
552
# TODO: Return the diff in order by filename, not by category or in
553
# random order. Can probably be done by lock-stepping through the
554
# filenames from both trees.
557
def file_status(filename, old_tree, new_tree):
558
"""Return single-letter status, old and new names for a file.
560
The complexity here is in deciding how to represent renames;
561
many complex cases are possible.
563
old_inv = old_tree.inventory
564
new_inv = new_tree.inventory
565
new_id = new_inv.path2id(filename)
566
old_id = old_inv.path2id(filename)
568
if not new_id and not old_id:
569
# easy: doesn't exist in either; not versioned at all
570
if new_tree.is_ignored(filename):
571
return 'I', None, None
573
return '?', None, None
575
# There is now a file of this name, great.
578
# There is no longer a file of this name, but we can describe
579
# what happened to the file that used to have
580
# this name. There are two possibilities: either it was
581
# deleted entirely, or renamed.
582
if new_inv.has_id(old_id):
583
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
585
return 'D', old_inv.id2path(old_id), None
587
# if the file_id is new in this revision, it is added
588
if new_id and not old_inv.has_id(new_id):
591
# if there used to be a file of this name, but that ID has now
592
# disappeared, it is deleted
593
if old_id and not new_inv.has_id(old_id):
600
def find_renames(old_inv, new_inv):
601
for file_id in old_inv:
602
if file_id not in new_inv:
604
old_name = old_inv.id2path(file_id)
605
new_name = new_inv.id2path(file_id)
606
if old_name != new_name:
607
yield (old_name, new_name)
610
def find_ids_across_trees(filenames, trees, require_versioned=True):
611
"""Find the ids corresponding to specified filenames.
613
All matches in all trees will be used, and all children of matched
614
directories will be used.
616
:param filenames: The filenames to find file_ids for (if None, returns
618
:param trees: The trees to find file_ids within
619
:param require_versioned: if true, all specified filenames must occur in
621
:return: a set of file ids for the specified filenames and their children.
625
specified_path_ids = _find_ids_across_trees(filenames, trees,
627
return _find_children_across_trees(specified_path_ids, trees)
630
def _find_ids_across_trees(filenames, trees, require_versioned):
631
"""Find the ids corresponding to specified filenames.
633
All matches in all trees will be used, but subdirectories are not scanned.
635
:param filenames: The filenames to find file_ids for
636
:param trees: The trees to find file_ids within
637
:param require_versioned: if true, all specified filenames must occur in
639
:return: a set of file ids for the specified filenames
642
interesting_ids = set()
643
for tree_path in filenames:
646
file_id = tree.path2id(tree_path)
647
if file_id is not None:
648
interesting_ids.add(file_id)
651
not_versioned.append(tree_path)
652
if len(not_versioned) > 0 and require_versioned:
653
raise errors.PathsNotVersionedError(not_versioned)
654
return interesting_ids
657
def _find_children_across_trees(specified_ids, trees):
658
"""Return a set including specified ids and their children.
660
All matches in all trees will be used.
662
:param trees: The trees to find file_ids within
663
:return: a set containing all specified ids and their children
665
interesting_ids = set(specified_ids)
666
pending = interesting_ids
667
# now handle children of interesting ids
668
# we loop so that we handle all children of each id in both trees
669
while len(pending) > 0:
671
for file_id in pending:
673
if not tree.has_id(file_id):
675
entry = tree.inventory[file_id]
676
for child in getattr(entry, 'children', {}).itervalues():
677
if child.file_id not in interesting_ids:
678
new_pending.add(child.file_id)
679
interesting_ids.update(new_pending)
680
pending = new_pending
681
return interesting_ids
684
class InterTree(InterObject):
685
"""This class represents operations taking place between two Trees.
687
Its instances have methods like 'compare' and contain references to the
688
source and target trees these operations are to be carried out on.
690
Clients of bzrlib should not need to use InterTree directly, rather they
691
should use the convenience methods on Tree such as 'Tree.compare()' which
692
will pass through to InterTree as appropriate.
698
def compare(self, want_unchanged=False, specific_files=None,
699
extra_trees=None, require_versioned=False, include_root=False,
700
want_unversioned=False):
701
"""Return the changes from source to target.
703
:return: A TreeDelta.
704
:param specific_files: An optional list of file paths to restrict the
705
comparison to. When mapping filenames to ids, all matches in all
706
trees (including optional extra_trees) are used, and all children of
707
matched directories are included.
708
:param want_unchanged: An optional boolean requesting the inclusion of
709
unchanged entries in the result.
710
:param extra_trees: An optional list of additional trees to use when
711
mapping the contents of specific_files (paths) to file_ids.
712
:param require_versioned: An optional boolean (defaults to False). When
713
supplied and True all the 'specific_files' must be versioned, or
714
a PathsNotVersionedError will be thrown.
715
:param want_unversioned: Scan for unversioned paths.
717
# NB: show_status depends on being able to pass in non-versioned files
718
# and report them as unknown
719
trees = (self.source,)
720
if extra_trees is not None:
721
trees = trees + tuple(extra_trees)
722
# target is usually the newer tree:
723
specific_file_ids = self.target.paths2ids(specific_files, trees,
724
require_versioned=require_versioned)
725
if specific_files and not specific_file_ids:
726
# All files are unversioned, so just return an empty delta
727
# _compare_trees would think we want a complete delta
728
result = delta.TreeDelta()
729
fake_entry = InventoryFile('unused', 'unused', 'unused')
730
result.unversioned = [(path, None,
731
self.target._comparison_data(fake_entry, path)[0]) for path in
734
return delta._compare_trees(self.source, self.target, want_unchanged,
735
specific_files, include_root, extra_trees=extra_trees,
736
require_versioned=require_versioned,
737
want_unversioned=want_unversioned)
739
def iter_changes(self, include_unchanged=False,
740
specific_files=None, pb=None, extra_trees=[],
741
require_versioned=True, want_unversioned=False):
742
"""Generate an iterator of changes between trees.
745
(file_id, (path_in_source, path_in_target),
746
changed_content, versioned, parent, name, kind,
749
Changed_content is True if the file's content has changed. This
750
includes changes to its kind, and to a symlink's target.
752
versioned, parent, name, kind, executable are tuples of (from, to).
753
If a file is missing in a tree, its kind is None.
755
Iteration is done in parent-to-child order, relative to the target
758
There is no guarantee that all paths are in sorted order: the
759
requirement to expand the search due to renames may result in children
760
that should be found early being found late in the search, after
761
lexically later results have been returned.
762
:param require_versioned: Raise errors.PathsNotVersionedError if a
763
path in the specific_files list is not versioned in one of
764
source, target or extra_trees.
765
:param want_unversioned: Should unversioned files be returned in the
766
output. An unversioned file is defined as one with (False, False)
767
for the versioned pair.
770
lookup_trees = [self.source]
772
lookup_trees.extend(extra_trees)
773
if specific_files == []:
774
specific_file_ids = []
776
specific_file_ids = self.target.paths2ids(specific_files,
777
lookup_trees, require_versioned=require_versioned)
779
all_unversioned = sorted([(p.split('/'), p) for p in
781
if specific_files is None or
782
osutils.is_inside_any(specific_files, p)])
783
all_unversioned = deque(all_unversioned)
785
all_unversioned = deque()
787
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
788
specific_file_ids=specific_file_ids))
789
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
790
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
791
specific_file_ids=specific_file_ids))
792
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
794
# the unversioned path lookup only occurs on real trees - where there
795
# can be extras. So the fake_entry is solely used to look up
796
# executable it values when execute is not supported.
797
fake_entry = InventoryFile('unused', 'unused', 'unused')
798
for to_path, to_entry in to_entries_by_dir:
799
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
800
unversioned_path = all_unversioned.popleft()
801
to_kind, to_executable, to_stat = \
802
self.target._comparison_data(fake_entry, unversioned_path[1])
803
yield (None, (None, unversioned_path[1]), True, (False, False),
805
(None, unversioned_path[0][-1]),
807
(None, to_executable))
808
file_id = to_entry.file_id
809
to_paths[file_id] = to_path
811
changed_content = False
812
from_path, from_entry = from_data.get(file_id, (None, None))
813
from_versioned = (from_entry is not None)
814
if from_entry is not None:
815
from_versioned = True
816
from_name = from_entry.name
817
from_parent = from_entry.parent_id
818
from_kind, from_executable, from_stat = \
819
self.source._comparison_data(from_entry, from_path)
822
from_versioned = False
826
from_executable = None
827
versioned = (from_versioned, True)
828
to_kind, to_executable, to_stat = \
829
self.target._comparison_data(to_entry, to_path)
830
kind = (from_kind, to_kind)
831
if kind[0] != kind[1]:
832
changed_content = True
833
elif from_kind == 'file':
834
from_size = self.source._file_size(from_entry, from_stat)
835
to_size = self.target._file_size(to_entry, to_stat)
836
if from_size != to_size:
837
changed_content = True
838
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
839
self.target.get_file_sha1(file_id, to_path, to_stat)):
840
changed_content = True
841
elif from_kind == 'symlink':
842
if (self.source.get_symlink_target(file_id) !=
843
self.target.get_symlink_target(file_id)):
844
changed_content = True
845
elif from_kind == 'tree-reference':
846
if (self.source.get_reference_revision(file_id, from_path)
847
!= self.target.get_reference_revision(file_id, to_path)):
848
changed_content = True
849
parent = (from_parent, to_entry.parent_id)
850
name = (from_name, to_entry.name)
851
executable = (from_executable, to_executable)
853
pb.update('comparing files', entry_count, num_entries)
854
if (changed_content is not False or versioned[0] != versioned[1]
855
or parent[0] != parent[1] or name[0] != name[1] or
856
executable[0] != executable[1] or include_unchanged):
857
yield (file_id, (from_path, to_path), changed_content,
858
versioned, parent, name, kind, executable)
860
while all_unversioned:
861
# yield any trailing unversioned paths
862
unversioned_path = all_unversioned.popleft()
863
to_kind, to_executable, to_stat = \
864
self.target._comparison_data(fake_entry, unversioned_path[1])
865
yield (None, (None, unversioned_path[1]), True, (False, False),
867
(None, unversioned_path[0][-1]),
869
(None, to_executable))
871
def get_to_path(to_entry):
872
if to_entry.parent_id is None:
873
to_path = '' # the root
875
if to_entry.parent_id not in to_paths:
877
return get_to_path(self.target.inventory[to_entry.parent_id])
878
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
880
to_paths[to_entry.file_id] = to_path
883
for path, from_entry in from_entries_by_dir:
884
file_id = from_entry.file_id
885
if file_id in to_paths:
888
if not file_id in self.target.inventory:
889
# common case - paths we have not emitted are not present in
893
to_path = get_to_path(self.target.inventory[file_id])
896
pb.update('comparing files', entry_count, num_entries)
897
versioned = (True, False)
898
parent = (from_entry.parent_id, None)
899
name = (from_entry.name, None)
900
from_kind, from_executable, stat_value = \
901
self.source._comparison_data(from_entry, path)
902
kind = (from_kind, None)
903
executable = (from_executable, None)
904
changed_content = True
905
# the parent's path is necessarily known at this point.
906
yield(file_id, (path, to_path), changed_content, versioned, parent,
907
name, kind, executable)
910
class MultiWalker(object):
911
"""Walk multiple trees simultaneously, getting combined results."""
913
# Note: This could be written to not assume you can do out-of-order
914
# lookups. Instead any nodes that don't match in all trees could be
915
# marked as 'deferred', and then returned in the final cleanup loop.
916
# For now, I think it is "nicer" to return things as close to the
917
# "master_tree" order as we can.
919
def __init__(self, master_tree, other_trees):
920
"""Create a new MultiWalker.
922
All trees being walked must implement "iter_entries_by_dir()", such
923
that they yield (path, object) tuples, where that object will have a
924
'.file_id' member, that can be used to check equality.
926
:param master_tree: All trees will be 'slaved' to the master_tree such
927
that nodes in master_tree will be used as 'first-pass' sync points.
928
Any nodes that aren't in master_tree will be merged in a second
930
:param other_trees: A list of other trees to walk simultaneously.
932
self._master_tree = master_tree
933
self._other_trees = other_trees
935
# Keep track of any nodes that were properly processed just out of
936
# order, that way we don't return them at the end, we don't have to
937
# track *all* processed file_ids, just the out-of-order ones
938
self._out_of_order_processed = set()
941
def _step_one(iterator):
942
"""Step an iter_entries_by_dir iterator.
944
:return: (has_more, path, ie)
945
If has_more is False, path and ie will be None.
948
path, ie = iterator.next()
949
except StopIteration:
950
return False, None, None
952
return True, path, ie
955
def _cmp_path_by_dirblock(path1, path2):
956
"""Compare two paths based on what directory they are in.
958
This generates a sort order, such that all children of a directory are
959
sorted together, and grandchildren are in the same order as the
960
children appear. But all grandchildren come after all children.
962
:param path1: first path
963
:param path2: the second path
964
:return: negative number if ``path1`` comes first,
966
and a positive number if ``path2`` sorts first
968
# Shortcut this special case
971
# This is stolen from _dirstate_helpers_py.py, only switching it to
972
# Unicode objects. Consider using encode_utf8() and then using the
973
# optimized versions, or maybe writing optimized unicode versions.
974
if not isinstance(path1, unicode):
975
raise TypeError("'path1' must be a unicode string, not %s: %r"
976
% (type(path1), path1))
977
if not isinstance(path2, unicode):
978
raise TypeError("'path2' must be a unicode string, not %s: %r"
979
% (type(path2), path2))
980
return cmp(MultiWalker._path_to_key(path1),
981
MultiWalker._path_to_key(path2))
984
def _path_to_key(path):
985
dirname, basename = osutils.split(path)
986
return (dirname.split(u'/'), basename)
988
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
989
"""Lookup an inventory entry by file_id.
991
This is called when an entry is missing in the normal order.
992
Generally this is because a file was either renamed, or it was
993
deleted/added. If the entry was found in the inventory and not in
994
extra_entries, it will be added to self._out_of_order_processed
996
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
997
should be filled with entries that were found before they were
998
used. If file_id is present, it will be removed from the
1000
:param other_tree: The Tree to search, in case we didn't find the entry
1002
:param file_id: The file_id to look for
1003
:return: (path, ie) if found or (None, None) if not present.
1005
if file_id in extra_entries:
1006
return extra_entries.pop(file_id)
1007
# TODO: Is id2path better as the first call, or is
1008
# inventory[file_id] better as a first check?
1010
cur_path = other_tree.id2path(file_id)
1011
except errors.NoSuchId:
1013
if cur_path is None:
1016
self._out_of_order_processed.add(file_id)
1017
cur_ie = other_tree.inventory[file_id]
1018
return (cur_path, cur_ie)
1021
"""Match up the values in the different trees."""
1022
for result in self._walk_master_tree():
1024
self._finish_others()
1025
for result in self._walk_others():
1028
def _walk_master_tree(self):
1029
"""First pass, walk all trees in lock-step.
1031
When we are done, all nodes in the master_tree will have been
1032
processed. _other_walkers, _other_entries, and _others_extra will be
1033
set on 'self' for future processing.
1035
# This iterator has the most "inlining" done, because it tends to touch
1036
# every file in the tree, while the others only hit nodes that don't
1038
master_iterator = self._master_tree.iter_entries_by_dir()
1040
other_walkers = [other.iter_entries_by_dir()
1041
for other in self._other_trees]
1042
other_entries = [self._step_one(walker) for walker in other_walkers]
1043
# Track extra nodes in the other trees
1044
others_extra = [{} for i in xrange(len(self._other_trees))]
1046
master_has_more = True
1047
step_one = self._step_one
1048
lookup_by_file_id = self._lookup_by_file_id
1049
out_of_order_processed = self._out_of_order_processed
1051
while master_has_more:
1052
(master_has_more, path, master_ie) = step_one(master_iterator)
1053
if not master_has_more:
1056
file_id = master_ie.file_id
1058
other_values_append = other_values.append
1059
next_other_entries = []
1060
next_other_entries_append = next_other_entries.append
1061
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1062
if not other_has_more:
1063
other_values_append(lookup_by_file_id(
1064
others_extra[idx], self._other_trees[idx], file_id))
1065
next_other_entries_append((False, None, None))
1066
elif file_id == other_ie.file_id:
1067
# This is the critical code path, as most of the entries
1068
# should match between most trees.
1069
other_values_append((other_path, other_ie))
1070
next_other_entries_append(step_one(other_walkers[idx]))
1072
# This walker did not match, step it until it either
1073
# matches, or we know we are past the current walker.
1074
other_walker = other_walkers[idx]
1075
other_extra = others_extra[idx]
1076
while (other_has_more and
1077
self._cmp_path_by_dirblock(other_path, path) < 0):
1078
other_file_id = other_ie.file_id
1079
if other_file_id not in out_of_order_processed:
1080
other_extra[other_file_id] = (other_path, other_ie)
1081
other_has_more, other_path, other_ie = \
1082
step_one(other_walker)
1083
if other_has_more and other_ie.file_id == file_id:
1084
# We ended up walking to this point, match and step
1086
other_values_append((other_path, other_ie))
1087
other_has_more, other_path, other_ie = \
1088
step_one(other_walker)
1090
# This record isn't in the normal order, see if it
1092
other_values_append(lookup_by_file_id(
1093
other_extra, self._other_trees[idx], file_id))
1094
next_other_entries_append((other_has_more, other_path,
1096
other_entries = next_other_entries
1098
# We've matched all the walkers, yield this datapoint
1099
yield path, file_id, master_ie, other_values
1100
self._other_walkers = other_walkers
1101
self._other_entries = other_entries
1102
self._others_extra = others_extra
1104
def _finish_others(self):
1105
"""Finish walking the other iterators, so we get all entries."""
1106
for idx, info in enumerate(self._other_entries):
1107
other_extra = self._others_extra[idx]
1108
(other_has_more, other_path, other_ie) = info
1109
while other_has_more:
1110
other_file_id = other_ie.file_id
1111
if other_file_id not in self._out_of_order_processed:
1112
other_extra[other_file_id] = (other_path, other_ie)
1113
other_has_more, other_path, other_ie = \
1114
self._step_one(self._other_walkers[idx])
1115
del self._other_entries
1117
def _walk_others(self):
1118
"""Finish up by walking all the 'deferred' nodes."""
1119
# TODO: One alternative would be to grab all possible unprocessed
1120
# file_ids, and then sort by path, and then yield them. That
1121
# might ensure better ordering, in case a caller strictly
1122
# requires parents before children.
1123
for idx, other_extra in enumerate(self._others_extra):
1124
others = sorted(other_extra.itervalues(),
1125
key=lambda x: self._path_to_key(x[0]))
1126
for other_path, other_ie in others:
1127
file_id = other_ie.file_id
1128
# We don't need to check out_of_order_processed here, because
1129
# the lookup_by_file_id will be removing anything processed
1130
# from the extras cache
1131
other_extra.pop(file_id)
1132
other_values = [(None, None) for i in xrange(idx)]
1133
other_values.append((other_path, other_ie))
1134
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1135
alt_idx = alt_idx + idx + 1
1136
alt_extra = self._others_extra[alt_idx]
1137
alt_tree = self._other_trees[alt_idx]
1138
other_values.append(self._lookup_by_file_id(
1139
alt_extra, alt_tree, file_id))
1140
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
