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
__contains__ = has_id
137
def has_or_had_id(self, file_id):
138
if file_id == self.inventory.root.file_id:
140
return self.inventory.has_id(file_id)
142
def is_ignored(self, filename):
143
"""Check whether the filename is ignored by this tree.
145
:param filename: The relative filename within the tree.
146
:return: True if the filename is ignored.
151
return iter(self.inventory)
153
def all_file_ids(self):
154
"""Iterate through all file ids, including ids for missing files."""
155
return set(self.inventory)
157
def id2path(self, file_id):
158
"""Return the path for a file id.
162
return self.inventory.id2path(file_id)
164
def is_control_filename(self, filename):
165
"""True if filename is the name of a control file in this tree.
167
:param filename: A filename within the tree. This is a relative path
168
from the root of this tree.
170
This is true IF and ONLY IF the filename is part of the meta data
171
that bzr controls in this tree. I.E. a random .bzr directory placed
172
on disk will not be a control file for this tree.
174
return self.bzrdir.is_control_filename(filename)
177
def iter_entries_by_dir(self, specific_file_ids=None):
178
"""Walk the tree in 'by_dir' order.
180
This will yield each entry in the tree as a (path, entry) tuple.
181
The order that they are yielded is:
183
Directories are walked in a depth-first lexicographical order,
184
however, whenever a directory is reached, all of its direct child
185
nodes are yielded in lexicographical order before yielding the
188
For example, in the tree::
198
The yield order (ignoring root) would be::
199
a, f, a/b, a/d, a/b/c, a/d/e, f/g
201
return self.inventory.iter_entries_by_dir(
202
specific_file_ids=specific_file_ids)
204
def iter_references(self):
205
for path, entry in self.iter_entries_by_dir():
206
if entry.kind == 'tree-reference':
207
yield path, entry.file_id
209
def kind(self, file_id):
210
raise NotImplementedError("Tree subclass %s must implement kind"
211
% self.__class__.__name__)
213
def stored_kind(self, file_id):
214
"""File kind stored for this file_id.
216
May not match kind on disk for working trees. Always available
217
for versioned files, even when the file itself is missing.
219
return self.kind(file_id)
221
def path_content_summary(self, path):
222
"""Get a summary of the information about path.
224
:param path: A relative path within the tree.
225
:return: A tuple containing kind, size, exec, sha1-or-link.
226
Kind is always present (see tree.kind()).
227
size is present if kind is file, None otherwise.
228
exec is None unless kind is file and the platform supports the 'x'
230
sha1-or-link is the link target if kind is symlink, or the sha1 if
231
it can be obtained without reading the file.
233
raise NotImplementedError(self.path_content_summary)
235
def get_reference_revision(self, file_id, path=None):
236
raise NotImplementedError("Tree subclass %s must implement "
237
"get_reference_revision"
238
% self.__class__.__name__)
240
def _comparison_data(self, entry, path):
241
"""Return a tuple of kind, executable, stat_value for a file.
243
entry may be None if there is no inventory entry for the file, but
244
path must always be supplied.
246
kind is None if there is no file present (even if an inventory id is
247
present). executable is False for non-file entries.
249
raise NotImplementedError(self._comparison_data)
251
def _file_size(self, entry, stat_value):
252
raise NotImplementedError(self._file_size)
254
def _get_inventory(self):
255
return self._inventory
257
def get_file(self, file_id, path=None):
258
"""Return a file object for the file file_id in the tree.
260
If both file_id and path are defined, it is implementation defined as
261
to which one is used.
263
raise NotImplementedError(self.get_file)
265
def get_file_mtime(self, file_id, path=None):
266
"""Return the modification time for a file.
268
:param file_id: The handle for this file.
269
:param path: The path that this file can be found at.
270
These must point to the same object.
272
raise NotImplementedError(self.get_file_mtime)
274
def get_file_size(self, file_id):
275
"""Return the size of a file in bytes.
277
This applies only to regular files. If invoked on directories or
278
symlinks, it will return None.
279
:param file_id: The file-id of the file
281
raise NotImplementedError(self.get_file_size)
283
def get_file_by_path(self, path):
284
return self.get_file(self._inventory.path2id(path), path)
286
def iter_files_bytes(self, desired_files):
287
"""Iterate through file contents.
289
Files will not necessarily be returned in the order they occur in
290
desired_files. No specific order is guaranteed.
292
Yields pairs of identifier, bytes_iterator. identifier is an opaque
293
value supplied by the caller as part of desired_files. It should
294
uniquely identify the file version in the caller's context. (Examples:
295
an index number or a TreeTransform trans_id.)
297
bytes_iterator is an iterable of bytestrings for the file. The
298
kind of iterable and length of the bytestrings are unspecified, but for
299
this implementation, it is a tuple containing a single bytestring with
300
the complete text of the file.
302
:param desired_files: a list of (file_id, identifier) pairs
304
for file_id, identifier in desired_files:
305
# We wrap the string in a tuple so that we can return an iterable
306
# of bytestrings. (Technically, a bytestring is also an iterable
307
# of bytestrings, but iterating through each character is not
309
cur_file = (self.get_file_text(file_id),)
310
yield identifier, cur_file
312
def get_symlink_target(self, file_id):
313
"""Get the target for a given file_id.
315
It is assumed that the caller already knows that file_id is referencing
317
:param file_id: Handle for the symlink entry.
318
:return: The path the symlink points to.
320
raise NotImplementedError(self.get_symlink_target)
322
def get_root_id(self):
323
"""Return the file_id for the root of this tree."""
324
raise NotImplementedError(self.get_root_id)
326
def annotate_iter(self, file_id,
327
default_revision=_mod_revision.CURRENT_REVISION):
328
"""Return an iterator of revision_id, line tuples.
330
For working trees (and mutable trees in general), the special
331
revision_id 'current:' will be used for lines that are new in this
332
tree, e.g. uncommitted changes.
333
:param file_id: The file to produce an annotated version from
334
:param default_revision: For lines that don't match a basis, mark them
335
with this revision id. Not all implementations will make use of
338
raise NotImplementedError(self.annotate_iter)
340
def _get_plan_merge_data(self, file_id, other, base):
341
from bzrlib import merge, versionedfile
342
vf = versionedfile._PlanMergeVersionedFile(file_id)
343
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
344
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
346
last_revision_base = None
348
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
349
return vf, last_revision_a, last_revision_b, last_revision_base
351
def plan_file_merge(self, file_id, other, base=None):
352
"""Generate a merge plan based on annotations.
354
If the file contains uncommitted changes in this tree, they will be
355
attributed to the 'current:' pseudo-revision. If the file contains
356
uncommitted changes in the other tree, they will be assigned to the
357
'other:' pseudo-revision.
359
data = self._get_plan_merge_data(file_id, other, base)
360
vf, last_revision_a, last_revision_b, last_revision_base = data
361
return vf.plan_merge(last_revision_a, last_revision_b,
364
def plan_file_lca_merge(self, file_id, other, base=None):
365
"""Generate a merge plan based lca-newness.
367
If the file contains uncommitted changes in this tree, they will be
368
attributed to the 'current:' pseudo-revision. If the file contains
369
uncommitted changes in the other tree, they will be assigned to the
370
'other:' pseudo-revision.
372
data = self._get_plan_merge_data(file_id, other, base)
373
vf, last_revision_a, last_revision_b, last_revision_base = data
374
return vf.plan_lca_merge(last_revision_a, last_revision_b,
377
def _iter_parent_trees(self):
378
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
379
for revision_id in self.get_parent_ids():
381
yield self.revision_tree(revision_id)
382
except errors.NoSuchRevisionInTree:
383
yield self.repository.revision_tree(revision_id)
386
def _file_revision(revision_tree, file_id):
387
"""Determine the revision associated with a file in a given tree."""
388
revision_tree.lock_read()
390
return revision_tree.inventory[file_id].revision
392
revision_tree.unlock()
394
def _get_file_revision(self, file_id, vf, tree_revision):
395
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
397
if getattr(self, '_repository', None) is None:
398
last_revision = tree_revision
399
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
400
self._iter_parent_trees()]
401
vf.add_lines((file_id, last_revision), parent_keys,
402
self.get_file(file_id).readlines())
403
repo = self.branch.repository
406
last_revision = self._file_revision(self, file_id)
407
base_vf = self._repository.texts
408
if base_vf not in vf.fallback_versionedfiles:
409
vf.fallback_versionedfiles.append(base_vf)
412
inventory = property(_get_inventory,
413
doc="Inventory of this Tree")
415
def _check_retrieved(self, ie, f):
418
fp = fingerprint_file(f)
421
if ie.text_size is not None:
422
if ie.text_size != fp['size']:
423
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
424
["inventory expects %d bytes" % ie.text_size,
425
"file is actually %d bytes" % fp['size'],
426
"store is probably damaged/corrupt"])
428
if ie.text_sha1 != fp['sha1']:
429
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
430
["inventory expects %s" % ie.text_sha1,
431
"file is actually %s" % fp['sha1'],
432
"store is probably damaged/corrupt"])
435
def path2id(self, path):
436
"""Return the id for path in this tree."""
437
return self._inventory.path2id(path)
439
def paths2ids(self, paths, trees=[], require_versioned=True):
440
"""Return all the ids that can be reached by walking from paths.
442
Each path is looked up in this tree and any extras provided in
443
trees, and this is repeated recursively: the children in an extra tree
444
of a directory that has been renamed under a provided path in this tree
445
are all returned, even if none exist under a provided path in this
446
tree, and vice versa.
448
:param paths: An iterable of paths to start converting to ids from.
449
Alternatively, if paths is None, no ids should be calculated and None
450
will be returned. This is offered to make calling the api unconditional
451
for code that *might* take a list of files.
452
:param trees: Additional trees to consider.
453
:param require_versioned: If False, do not raise NotVersionedError if
454
an element of paths is not versioned in this tree and all of trees.
456
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
458
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
459
def print_file(self, file_id):
460
"""Print file with id `file_id` to stdout."""
462
sys.stdout.write(self.get_file_text(file_id))
467
def revision_tree(self, revision_id):
468
"""Obtain a revision tree for the revision revision_id.
470
The intention of this method is to allow access to possibly cached
471
tree data. Implementors of this method should raise NoSuchRevision if
472
the tree is not locally available, even if they could obtain the
473
tree via a repository or some other means. Callers are responsible
474
for finding the ultimate source for a revision tree.
476
:param revision_id: The revision_id of the requested tree.
478
:raises: NoSuchRevision if the tree cannot be obtained.
480
raise errors.NoSuchRevisionInTree(self, revision_id)
483
"""What files are present in this tree and unknown.
485
:return: an iterator over the unknown files.
492
def filter_unversioned_files(self, paths):
493
"""Filter out paths that are versioned.
495
:return: set of paths.
497
# NB: we specifically *don't* call self.has_filename, because for
498
# WorkingTrees that can indicate files that exist on disk but that
500
pred = self.inventory.has_filename
501
return set((p for p in paths if not pred(p)))
503
def walkdirs(self, prefix=""):
504
"""Walk the contents of this tree from path down.
506
This yields all the data about the contents of a directory at a time.
507
After each directory has been yielded, if the caller has mutated the
508
list to exclude some directories, they are then not descended into.
510
The data yielded is of the form:
511
((directory-relpath, directory-path-from-root, directory-fileid),
512
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
513
versioned_kind), ...]),
514
- directory-relpath is the containing dirs relpath from prefix
515
- directory-path-from-root is the containing dirs path from /
516
- directory-fileid is the id of the directory if it is versioned.
517
- relpath is the relative path within the subtree being walked.
518
- basename is the basename
519
- kind is the kind of the file now. If unknonwn then the file is not
520
present within the tree - but it may be recorded as versioned. See
522
- lstat is the stat data *if* the file was statted.
523
- path_from_tree_root is the path from the root of the tree.
524
- file_id is the file_id if the entry is versioned.
525
- versioned_kind is the kind of the file as last recorded in the
526
versioning system. If 'unknown' the file is not versioned.
527
One of 'kind' and 'versioned_kind' must not be 'unknown'.
529
:param prefix: Start walking from prefix within the tree rather than
530
at the root. This allows one to walk a subtree but get paths that are
531
relative to a tree rooted higher up.
532
:return: an iterator over the directory data.
534
raise NotImplementedError(self.walkdirs)
536
def iter_search_rules(self, path_names, pref_names=None,
537
_default_searcher=rules._per_user_searcher):
538
"""Find the preferences for filenames in a tree.
540
:param path_names: an iterable of paths to find attributes for.
541
Paths are given relative to the root of the tree.
542
:param pref_names: the list of preferences to lookup - None for all
543
:param _default_searcher: private parameter to assist testing - don't use
544
:return: an iterator of tuple sequences, one per path-name.
545
See _RulesSearcher.get_items for details on the tuple sequence.
547
searcher = self._get_rules_searcher(_default_searcher)
548
if searcher is not None:
549
if pref_names is not None:
550
for path in path_names:
551
yield searcher.get_selected_items(path, pref_names)
553
for path in path_names:
554
yield searcher.get_items(path)
557
def _get_rules_searcher(self, default_searcher):
558
"""Get the RulesSearcher for this tree given the default one."""
559
searcher = default_searcher
560
file_id = self.path2id(rules.RULES_TREE_FILENAME)
561
if file_id is not None:
562
ini_file = self.get_file(file_id)
563
searcher = rules._StackedRulesSearcher(
564
[rules._IniBasedRulesSearcher(ini_file), default_searcher])
568
class EmptyTree(Tree):
571
self._inventory = Inventory(root_id=None)
572
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
573
' use repository.revision_tree instead.',
574
DeprecationWarning, stacklevel=2)
576
def get_parent_ids(self):
579
def get_symlink_target(self, file_id):
582
def has_filename(self, filename):
585
def kind(self, file_id):
588
def list_files(self, include_root=False):
591
def __contains__(self, file_id):
592
return (file_id in self._inventory)
594
def get_file_sha1(self, file_id, path=None, stat_value=None):
598
######################################################################
601
# TODO: Merge these two functions into a single one that can operate
602
# on either a whole tree or a set of files.
604
# TODO: Return the diff in order by filename, not by category or in
605
# random order. Can probably be done by lock-stepping through the
606
# filenames from both trees.
609
def file_status(filename, old_tree, new_tree):
610
"""Return single-letter status, old and new names for a file.
612
The complexity here is in deciding how to represent renames;
613
many complex cases are possible.
615
old_inv = old_tree.inventory
616
new_inv = new_tree.inventory
617
new_id = new_inv.path2id(filename)
618
old_id = old_inv.path2id(filename)
620
if not new_id and not old_id:
621
# easy: doesn't exist in either; not versioned at all
622
if new_tree.is_ignored(filename):
623
return 'I', None, None
625
return '?', None, None
627
# There is now a file of this name, great.
630
# There is no longer a file of this name, but we can describe
631
# what happened to the file that used to have
632
# this name. There are two possibilities: either it was
633
# deleted entirely, or renamed.
634
if new_inv.has_id(old_id):
635
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
637
return 'D', old_inv.id2path(old_id), None
639
# if the file_id is new in this revision, it is added
640
if new_id and not old_inv.has_id(new_id):
643
# if there used to be a file of this name, but that ID has now
644
# disappeared, it is deleted
645
if old_id and not new_inv.has_id(old_id):
652
def find_renames(old_inv, new_inv):
653
for file_id in old_inv:
654
if file_id not in new_inv:
656
old_name = old_inv.id2path(file_id)
657
new_name = new_inv.id2path(file_id)
658
if old_name != new_name:
659
yield (old_name, new_name)
662
def find_ids_across_trees(filenames, trees, require_versioned=True):
663
"""Find the ids corresponding to specified filenames.
665
All matches in all trees will be used, and all children of matched
666
directories will be used.
668
:param filenames: The filenames to find file_ids for (if None, returns
670
:param trees: The trees to find file_ids within
671
:param require_versioned: if true, all specified filenames must occur in
673
:return: a set of file ids for the specified filenames and their children.
677
specified_path_ids = _find_ids_across_trees(filenames, trees,
679
return _find_children_across_trees(specified_path_ids, trees)
682
def _find_ids_across_trees(filenames, trees, require_versioned):
683
"""Find the ids corresponding to specified filenames.
685
All matches in all trees will be used, but subdirectories are not scanned.
687
:param filenames: The filenames to find file_ids for
688
:param trees: The trees to find file_ids within
689
:param require_versioned: if true, all specified filenames must occur in
691
:return: a set of file ids for the specified filenames
694
interesting_ids = set()
695
for tree_path in filenames:
698
file_id = tree.path2id(tree_path)
699
if file_id is not None:
700
interesting_ids.add(file_id)
703
not_versioned.append(tree_path)
704
if len(not_versioned) > 0 and require_versioned:
705
raise errors.PathsNotVersionedError(not_versioned)
706
return interesting_ids
709
def _find_children_across_trees(specified_ids, trees):
710
"""Return a set including specified ids and their children.
712
All matches in all trees will be used.
714
:param trees: The trees to find file_ids within
715
:return: a set containing all specified ids and their children
717
interesting_ids = set(specified_ids)
718
pending = interesting_ids
719
# now handle children of interesting ids
720
# we loop so that we handle all children of each id in both trees
721
while len(pending) > 0:
723
for file_id in pending:
725
if not tree.has_id(file_id):
727
entry = tree.inventory[file_id]
728
for child in getattr(entry, 'children', {}).itervalues():
729
if child.file_id not in interesting_ids:
730
new_pending.add(child.file_id)
731
interesting_ids.update(new_pending)
732
pending = new_pending
733
return interesting_ids
736
class InterTree(InterObject):
737
"""This class represents operations taking place between two Trees.
739
Its instances have methods like 'compare' and contain references to the
740
source and target trees these operations are to be carried out on.
742
Clients of bzrlib should not need to use InterTree directly, rather they
743
should use the convenience methods on Tree such as 'Tree.compare()' which
744
will pass through to InterTree as appropriate.
750
def compare(self, want_unchanged=False, specific_files=None,
751
extra_trees=None, require_versioned=False, include_root=False,
752
want_unversioned=False):
753
"""Return the changes from source to target.
755
:return: A TreeDelta.
756
:param specific_files: An optional list of file paths to restrict the
757
comparison to. When mapping filenames to ids, all matches in all
758
trees (including optional extra_trees) are used, and all children of
759
matched directories are included.
760
:param want_unchanged: An optional boolean requesting the inclusion of
761
unchanged entries in the result.
762
:param extra_trees: An optional list of additional trees to use when
763
mapping the contents of specific_files (paths) to file_ids.
764
:param require_versioned: An optional boolean (defaults to False). When
765
supplied and True all the 'specific_files' must be versioned, or
766
a PathsNotVersionedError will be thrown.
767
:param want_unversioned: Scan for unversioned paths.
769
# NB: show_status depends on being able to pass in non-versioned files
770
# and report them as unknown
771
trees = (self.source,)
772
if extra_trees is not None:
773
trees = trees + tuple(extra_trees)
774
# target is usually the newer tree:
775
specific_file_ids = self.target.paths2ids(specific_files, trees,
776
require_versioned=require_versioned)
777
if specific_files and not specific_file_ids:
778
# All files are unversioned, so just return an empty delta
779
# _compare_trees would think we want a complete delta
780
result = delta.TreeDelta()
781
fake_entry = InventoryFile('unused', 'unused', 'unused')
782
result.unversioned = [(path, None,
783
self.target._comparison_data(fake_entry, path)[0]) for path in
786
return delta._compare_trees(self.source, self.target, want_unchanged,
787
specific_files, include_root, extra_trees=extra_trees,
788
require_versioned=require_versioned,
789
want_unversioned=want_unversioned)
791
def iter_changes(self, include_unchanged=False,
792
specific_files=None, pb=None, extra_trees=[],
793
require_versioned=True, want_unversioned=False):
794
"""Generate an iterator of changes between trees.
797
(file_id, (path_in_source, path_in_target),
798
changed_content, versioned, parent, name, kind,
801
Changed_content is True if the file's content has changed. This
802
includes changes to its kind, and to a symlink's target.
804
versioned, parent, name, kind, executable are tuples of (from, to).
805
If a file is missing in a tree, its kind is None.
807
Iteration is done in parent-to-child order, relative to the target
810
There is no guarantee that all paths are in sorted order: the
811
requirement to expand the search due to renames may result in children
812
that should be found early being found late in the search, after
813
lexically later results have been returned.
814
:param require_versioned: Raise errors.PathsNotVersionedError if a
815
path in the specific_files list is not versioned in one of
816
source, target or extra_trees.
817
:param want_unversioned: Should unversioned files be returned in the
818
output. An unversioned file is defined as one with (False, False)
819
for the versioned pair.
822
lookup_trees = [self.source]
824
lookup_trees.extend(extra_trees)
825
if specific_files == []:
826
specific_file_ids = []
828
specific_file_ids = self.target.paths2ids(specific_files,
829
lookup_trees, require_versioned=require_versioned)
831
all_unversioned = sorted([(p.split('/'), p) for p in
833
if specific_files is None or
834
osutils.is_inside_any(specific_files, p)])
835
all_unversioned = deque(all_unversioned)
837
all_unversioned = deque()
839
from_entries_by_dir = list(self.source.inventory.iter_entries_by_dir(
840
specific_file_ids=specific_file_ids))
841
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
842
to_entries_by_dir = list(self.target.inventory.iter_entries_by_dir(
843
specific_file_ids=specific_file_ids))
844
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
846
# the unversioned path lookup only occurs on real trees - where there
847
# can be extras. So the fake_entry is solely used to look up
848
# executable it values when execute is not supported.
849
fake_entry = InventoryFile('unused', 'unused', 'unused')
850
for to_path, to_entry in to_entries_by_dir:
851
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
852
unversioned_path = all_unversioned.popleft()
853
to_kind, to_executable, to_stat = \
854
self.target._comparison_data(fake_entry, unversioned_path[1])
855
yield (None, (None, unversioned_path[1]), True, (False, False),
857
(None, unversioned_path[0][-1]),
859
(None, to_executable))
860
file_id = to_entry.file_id
861
to_paths[file_id] = to_path
863
changed_content = False
864
from_path, from_entry = from_data.get(file_id, (None, None))
865
from_versioned = (from_entry is not None)
866
if from_entry is not None:
867
from_versioned = True
868
from_name = from_entry.name
869
from_parent = from_entry.parent_id
870
from_kind, from_executable, from_stat = \
871
self.source._comparison_data(from_entry, from_path)
874
from_versioned = False
878
from_executable = None
879
versioned = (from_versioned, True)
880
to_kind, to_executable, to_stat = \
881
self.target._comparison_data(to_entry, to_path)
882
kind = (from_kind, to_kind)
883
if kind[0] != kind[1]:
884
changed_content = True
885
elif from_kind == 'file':
886
from_size = self.source._file_size(from_entry, from_stat)
887
to_size = self.target._file_size(to_entry, to_stat)
888
if from_size != to_size:
889
changed_content = True
890
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
891
self.target.get_file_sha1(file_id, to_path, to_stat)):
892
changed_content = True
893
elif from_kind == 'symlink':
894
if (self.source.get_symlink_target(file_id) !=
895
self.target.get_symlink_target(file_id)):
896
changed_content = True
897
elif from_kind == 'tree-reference':
898
if (self.source.get_reference_revision(file_id, from_path)
899
!= self.target.get_reference_revision(file_id, to_path)):
900
changed_content = True
901
parent = (from_parent, to_entry.parent_id)
902
name = (from_name, to_entry.name)
903
executable = (from_executable, to_executable)
905
pb.update('comparing files', entry_count, num_entries)
906
if (changed_content is not False or versioned[0] != versioned[1]
907
or parent[0] != parent[1] or name[0] != name[1] or
908
executable[0] != executable[1] or include_unchanged):
909
yield (file_id, (from_path, to_path), changed_content,
910
versioned, parent, name, kind, executable)
912
while all_unversioned:
913
# yield any trailing unversioned paths
914
unversioned_path = all_unversioned.popleft()
915
to_kind, to_executable, to_stat = \
916
self.target._comparison_data(fake_entry, unversioned_path[1])
917
yield (None, (None, unversioned_path[1]), True, (False, False),
919
(None, unversioned_path[0][-1]),
921
(None, to_executable))
923
def get_to_path(to_entry):
924
if to_entry.parent_id is None:
925
to_path = '' # the root
927
if to_entry.parent_id not in to_paths:
929
return get_to_path(self.target.inventory[to_entry.parent_id])
930
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
932
to_paths[to_entry.file_id] = to_path
935
for path, from_entry in from_entries_by_dir:
936
file_id = from_entry.file_id
937
if file_id in to_paths:
940
if not file_id in self.target.inventory:
941
# common case - paths we have not emitted are not present in
945
to_path = get_to_path(self.target.inventory[file_id])
948
pb.update('comparing files', entry_count, num_entries)
949
versioned = (True, False)
950
parent = (from_entry.parent_id, None)
951
name = (from_entry.name, None)
952
from_kind, from_executable, stat_value = \
953
self.source._comparison_data(from_entry, path)
954
kind = (from_kind, None)
955
executable = (from_executable, None)
956
changed_content = True
957
# the parent's path is necessarily known at this point.
958
yield(file_id, (path, to_path), changed_content, versioned, parent,
959
name, kind, executable)
962
class MultiWalker(object):
963
"""Walk multiple trees simultaneously, getting combined results."""
965
# Note: This could be written to not assume you can do out-of-order
966
# lookups. Instead any nodes that don't match in all trees could be
967
# marked as 'deferred', and then returned in the final cleanup loop.
968
# For now, I think it is "nicer" to return things as close to the
969
# "master_tree" order as we can.
971
def __init__(self, master_tree, other_trees):
972
"""Create a new MultiWalker.
974
All trees being walked must implement "iter_entries_by_dir()", such
975
that they yield (path, object) tuples, where that object will have a
976
'.file_id' member, that can be used to check equality.
978
:param master_tree: All trees will be 'slaved' to the master_tree such
979
that nodes in master_tree will be used as 'first-pass' sync points.
980
Any nodes that aren't in master_tree will be merged in a second
982
:param other_trees: A list of other trees to walk simultaneously.
984
self._master_tree = master_tree
985
self._other_trees = other_trees
987
# Keep track of any nodes that were properly processed just out of
988
# order, that way we don't return them at the end, we don't have to
989
# track *all* processed file_ids, just the out-of-order ones
990
self._out_of_order_processed = set()
993
def _step_one(iterator):
994
"""Step an iter_entries_by_dir iterator.
996
:return: (has_more, path, ie)
997
If has_more is False, path and ie will be None.
1000
path, ie = iterator.next()
1001
except StopIteration:
1002
return False, None, None
1004
return True, path, ie
1007
def _cmp_path_by_dirblock(path1, path2):
1008
"""Compare two paths based on what directory they are in.
1010
This generates a sort order, such that all children of a directory are
1011
sorted together, and grandchildren are in the same order as the
1012
children appear. But all grandchildren come after all children.
1014
:param path1: first path
1015
:param path2: the second path
1016
:return: negative number if ``path1`` comes first,
1017
0 if paths are equal
1018
and a positive number if ``path2`` sorts first
1020
# Shortcut this special case
1023
# This is stolen from _dirstate_helpers_py.py, only switching it to
1024
# Unicode objects. Consider using encode_utf8() and then using the
1025
# optimized versions, or maybe writing optimized unicode versions.
1026
if not isinstance(path1, unicode):
1027
raise TypeError("'path1' must be a unicode string, not %s: %r"
1028
% (type(path1), path1))
1029
if not isinstance(path2, unicode):
1030
raise TypeError("'path2' must be a unicode string, not %s: %r"
1031
% (type(path2), path2))
1032
return cmp(MultiWalker._path_to_key(path1),
1033
MultiWalker._path_to_key(path2))
1036
def _path_to_key(path):
1037
dirname, basename = osutils.split(path)
1038
return (dirname.split(u'/'), basename)
1040
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1041
"""Lookup an inventory entry by file_id.
1043
This is called when an entry is missing in the normal order.
1044
Generally this is because a file was either renamed, or it was
1045
deleted/added. If the entry was found in the inventory and not in
1046
extra_entries, it will be added to self._out_of_order_processed
1048
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1049
should be filled with entries that were found before they were
1050
used. If file_id is present, it will be removed from the
1052
:param other_tree: The Tree to search, in case we didn't find the entry
1054
:param file_id: The file_id to look for
1055
:return: (path, ie) if found or (None, None) if not present.
1057
if file_id in extra_entries:
1058
return extra_entries.pop(file_id)
1059
# TODO: Is id2path better as the first call, or is
1060
# inventory[file_id] better as a first check?
1062
cur_path = other_tree.id2path(file_id)
1063
except errors.NoSuchId:
1065
if cur_path is None:
1068
self._out_of_order_processed.add(file_id)
1069
cur_ie = other_tree.inventory[file_id]
1070
return (cur_path, cur_ie)
1073
"""Match up the values in the different trees."""
1074
for result in self._walk_master_tree():
1076
self._finish_others()
1077
for result in self._walk_others():
1080
def _walk_master_tree(self):
1081
"""First pass, walk all trees in lock-step.
1083
When we are done, all nodes in the master_tree will have been
1084
processed. _other_walkers, _other_entries, and _others_extra will be
1085
set on 'self' for future processing.
1087
# This iterator has the most "inlining" done, because it tends to touch
1088
# every file in the tree, while the others only hit nodes that don't
1090
master_iterator = self._master_tree.iter_entries_by_dir()
1092
other_walkers = [other.iter_entries_by_dir()
1093
for other in self._other_trees]
1094
other_entries = [self._step_one(walker) for walker in other_walkers]
1095
# Track extra nodes in the other trees
1096
others_extra = [{} for i in xrange(len(self._other_trees))]
1098
master_has_more = True
1099
step_one = self._step_one
1100
lookup_by_file_id = self._lookup_by_file_id
1101
out_of_order_processed = self._out_of_order_processed
1103
while master_has_more:
1104
(master_has_more, path, master_ie) = step_one(master_iterator)
1105
if not master_has_more:
1108
file_id = master_ie.file_id
1110
other_values_append = other_values.append
1111
next_other_entries = []
1112
next_other_entries_append = next_other_entries.append
1113
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1114
if not other_has_more:
1115
other_values_append(lookup_by_file_id(
1116
others_extra[idx], self._other_trees[idx], file_id))
1117
next_other_entries_append((False, None, None))
1118
elif file_id == other_ie.file_id:
1119
# This is the critical code path, as most of the entries
1120
# should match between most trees.
1121
other_values_append((other_path, other_ie))
1122
next_other_entries_append(step_one(other_walkers[idx]))
1124
# This walker did not match, step it until it either
1125
# matches, or we know we are past the current walker.
1126
other_walker = other_walkers[idx]
1127
other_extra = others_extra[idx]
1128
while (other_has_more and
1129
self._cmp_path_by_dirblock(other_path, path) < 0):
1130
other_file_id = other_ie.file_id
1131
if other_file_id not in out_of_order_processed:
1132
other_extra[other_file_id] = (other_path, other_ie)
1133
other_has_more, other_path, other_ie = \
1134
step_one(other_walker)
1135
if other_has_more and other_ie.file_id == file_id:
1136
# We ended up walking to this point, match and step
1138
other_values_append((other_path, other_ie))
1139
other_has_more, other_path, other_ie = \
1140
step_one(other_walker)
1142
# This record isn't in the normal order, see if it
1144
other_values_append(lookup_by_file_id(
1145
other_extra, self._other_trees[idx], file_id))
1146
next_other_entries_append((other_has_more, other_path,
1148
other_entries = next_other_entries
1150
# We've matched all the walkers, yield this datapoint
1151
yield path, file_id, master_ie, other_values
1152
self._other_walkers = other_walkers
1153
self._other_entries = other_entries
1154
self._others_extra = others_extra
1156
def _finish_others(self):
1157
"""Finish walking the other iterators, so we get all entries."""
1158
for idx, info in enumerate(self._other_entries):
1159
other_extra = self._others_extra[idx]
1160
(other_has_more, other_path, other_ie) = info
1161
while other_has_more:
1162
other_file_id = other_ie.file_id
1163
if other_file_id not in self._out_of_order_processed:
1164
other_extra[other_file_id] = (other_path, other_ie)
1165
other_has_more, other_path, other_ie = \
1166
self._step_one(self._other_walkers[idx])
1167
del self._other_entries
1169
def _walk_others(self):
1170
"""Finish up by walking all the 'deferred' nodes."""
1171
# TODO: One alternative would be to grab all possible unprocessed
1172
# file_ids, and then sort by path, and then yield them. That
1173
# might ensure better ordering, in case a caller strictly
1174
# requires parents before children.
1175
for idx, other_extra in enumerate(self._others_extra):
1176
others = sorted(other_extra.itervalues(),
1177
key=lambda x: self._path_to_key(x[0]))
1178
for other_path, other_ie in others:
1179
file_id = other_ie.file_id
1180
# We don't need to check out_of_order_processed here, because
1181
# the lookup_by_file_id will be removing anything processed
1182
# from the extras cache
1183
other_extra.pop(file_id)
1184
other_values = [(None, None) for i in xrange(idx)]
1185
other_values.append((other_path, other_ie))
1186
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1187
alt_idx = alt_idx + idx + 1
1188
alt_extra = self._others_extra[alt_idx]
1189
alt_tree = self._other_trees[alt_idx]
1190
other_values.append(self._lookup_by_file_id(
1191
alt_extra, alt_tree, file_id))
1192
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
