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.symbol_versioning import deprecated_function, deprecated_in
41
from bzrlib.trace import mutter, note
45
"""Abstract file tree.
47
There are several subclasses:
49
* `WorkingTree` exists as files on disk editable by the user.
51
* `RevisionTree` is a tree as recorded at some point in the past.
53
Trees contain an `Inventory` object, and also know how to retrieve
54
file texts mentioned in the inventory, either from a working
55
directory or from a store.
57
It is possible for trees to contain files that are not described
58
in their inventory or vice versa; for this use `filenames()`.
60
Trees can be compared, etc, regardless of whether they are working
61
trees or versioned trees.
64
def changes_from(self, other, want_unchanged=False, specific_files=None,
65
extra_trees=None, require_versioned=False, include_root=False,
66
want_unversioned=False):
67
"""Return a TreeDelta of the changes from other to this tree.
69
:param other: A tree to compare with.
70
:param specific_files: An optional list of file paths to restrict the
71
comparison to. When mapping filenames to ids, all matches in all
72
trees (including optional extra_trees) are used, and all children of
73
matched directories are included.
74
:param want_unchanged: An optional boolean requesting the inclusion of
75
unchanged entries in the result.
76
:param extra_trees: An optional list of additional trees to use when
77
mapping the contents of specific_files (paths) to file_ids.
78
:param require_versioned: An optional boolean (defaults to False). When
79
supplied and True all the 'specific_files' must be versioned, or
80
a PathsNotVersionedError will be thrown.
81
:param want_unversioned: Scan for unversioned paths.
83
The comparison will be performed by an InterTree object looked up on
86
# Martin observes that Tree.changes_from returns a TreeDelta and this
87
# may confuse people, because the class name of the returned object is
88
# a synonym of the object referenced in the method name.
89
return InterTree.get(other, self).compare(
90
want_unchanged=want_unchanged,
91
specific_files=specific_files,
92
extra_trees=extra_trees,
93
require_versioned=require_versioned,
94
include_root=include_root,
95
want_unversioned=want_unversioned,
98
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
99
def _iter_changes(self, *args, **kwargs):
100
return self.iter_changes(*args, **kwargs)
102
def iter_changes(self, from_tree, include_unchanged=False,
103
specific_files=None, pb=None, extra_trees=None,
104
require_versioned=True, want_unversioned=False):
105
intertree = InterTree.get(from_tree, self)
106
return intertree.iter_changes(include_unchanged, specific_files, pb,
107
extra_trees, require_versioned, want_unversioned=want_unversioned)
110
"""Get a list of the conflicts in the tree.
112
Each conflict is an instance of bzrlib.conflicts.Conflict.
114
return _mod_conflicts.ConflictList()
117
"""For trees that can have unversioned files, return all such paths."""
120
def get_parent_ids(self):
121
"""Get the parent ids for this tree.
123
:return: a list of parent ids. [] is returned to indicate
124
a tree with no parents.
125
:raises: BzrError if the parents are not known.
127
raise NotImplementedError(self.get_parent_ids)
129
def has_filename(self, filename):
130
"""True if the tree has given filename."""
131
raise NotImplementedError(self.has_filename)
133
def has_id(self, file_id):
134
return self.inventory.has_id(file_id)
136
def __contains__(self, file_id):
137
return self.has_id(file_id)
139
def has_or_had_id(self, file_id):
140
if file_id == self.inventory.root.file_id:
142
return self.inventory.has_id(file_id)
144
def is_ignored(self, filename):
145
"""Check whether the filename is ignored by this tree.
147
:param filename: The relative filename within the tree.
148
:return: True if the filename is ignored.
153
return iter(self.inventory)
155
def all_file_ids(self):
156
"""Iterate through all file ids, including ids for missing files."""
157
return set(self.inventory)
159
def id2path(self, file_id):
160
"""Return the path for a file id.
164
return self.inventory.id2path(file_id)
166
def is_control_filename(self, filename):
167
"""True if filename is the name of a control file in this tree.
169
:param filename: A filename within the tree. This is a relative path
170
from the root of this tree.
172
This is true IF and ONLY IF the filename is part of the meta data
173
that bzr controls in this tree. I.E. a random .bzr directory placed
174
on disk will not be a control file for this tree.
176
return self.bzrdir.is_control_filename(filename)
179
def iter_entries_by_dir(self, specific_file_ids=None):
180
"""Walk the tree in 'by_dir' order.
182
This will yield each entry in the tree as a (path, entry) tuple.
183
The order that they are yielded is:
185
Directories are walked in a depth-first lexicographical order,
186
however, whenever a directory is reached, all of its direct child
187
nodes are yielded in lexicographical order before yielding the
190
For example, in the tree::
200
The yield order (ignoring root) would be::
201
a, f, a/b, a/d, a/b/c, a/d/e, f/g
203
return self.inventory.iter_entries_by_dir(
204
specific_file_ids=specific_file_ids)
206
def iter_references(self):
207
for path, entry in self.iter_entries_by_dir():
208
if entry.kind == 'tree-reference':
209
yield path, entry.file_id
211
def kind(self, file_id):
212
raise NotImplementedError("Tree subclass %s must implement kind"
213
% self.__class__.__name__)
215
def stored_kind(self, file_id):
216
"""File kind stored for this file_id.
218
May not match kind on disk for working trees. Always available
219
for versioned files, even when the file itself is missing.
221
return self.kind(file_id)
223
def path_content_summary(self, path):
224
"""Get a summary of the information about path.
226
:param path: A relative path within the tree.
227
:return: A tuple containing kind, size, exec, sha1-or-link.
228
Kind is always present (see tree.kind()).
229
size is present if kind is file, None otherwise.
230
exec is None unless kind is file and the platform supports the 'x'
232
sha1-or-link is the link target if kind is symlink, or the sha1 if
233
it can be obtained without reading the file.
235
raise NotImplementedError(self.path_content_summary)
237
def get_reference_revision(self, file_id, path=None):
238
raise NotImplementedError("Tree subclass %s must implement "
239
"get_reference_revision"
240
% self.__class__.__name__)
242
def _comparison_data(self, entry, path):
243
"""Return a tuple of kind, executable, stat_value for a file.
245
entry may be None if there is no inventory entry for the file, but
246
path must always be supplied.
248
kind is None if there is no file present (even if an inventory id is
249
present). executable is False for non-file entries.
251
raise NotImplementedError(self._comparison_data)
253
def _file_size(self, entry, stat_value):
254
raise NotImplementedError(self._file_size)
256
def _get_inventory(self):
257
return self._inventory
259
def get_file(self, file_id, path=None):
260
"""Return a file object for the file file_id in the tree.
262
If both file_id and path are defined, it is implementation defined as
263
to which one is used.
265
raise NotImplementedError(self.get_file)
267
def get_file_mtime(self, file_id, path=None):
268
"""Return the modification time for a file.
270
:param file_id: The handle for this file.
271
:param path: The path that this file can be found at.
272
These must point to the same object.
274
raise NotImplementedError(self.get_file_mtime)
276
def get_file_size(self, file_id):
277
"""Return the size of a file in bytes.
279
This applies only to regular files. If invoked on directories or
280
symlinks, it will return None.
281
:param file_id: The file-id of the file
283
raise NotImplementedError(self.get_file_size)
285
def get_file_by_path(self, path):
286
return self.get_file(self._inventory.path2id(path), path)
288
def iter_files_bytes(self, desired_files):
289
"""Iterate through file contents.
291
Files will not necessarily be returned in the order they occur in
292
desired_files. No specific order is guaranteed.
294
Yields pairs of identifier, bytes_iterator. identifier is an opaque
295
value supplied by the caller as part of desired_files. It should
296
uniquely identify the file version in the caller's context. (Examples:
297
an index number or a TreeTransform trans_id.)
299
bytes_iterator is an iterable of bytestrings for the file. The
300
kind of iterable and length of the bytestrings are unspecified, but for
301
this implementation, it is a tuple containing a single bytestring with
302
the complete text of the file.
304
:param desired_files: a list of (file_id, identifier) pairs
306
for file_id, identifier in desired_files:
307
# We wrap the string in a tuple so that we can return an iterable
308
# of bytestrings. (Technically, a bytestring is also an iterable
309
# of bytestrings, but iterating through each character is not
311
cur_file = (self.get_file_text(file_id),)
312
yield identifier, cur_file
314
def get_symlink_target(self, file_id):
315
"""Get the target for a given file_id.
317
It is assumed that the caller already knows that file_id is referencing
319
:param file_id: Handle for the symlink entry.
320
:return: The path the symlink points to.
322
raise NotImplementedError(self.get_symlink_target)
324
def get_root_id(self):
325
"""Return the file_id for the root of this tree."""
326
raise NotImplementedError(self.get_root_id)
328
def annotate_iter(self, file_id,
329
default_revision=_mod_revision.CURRENT_REVISION):
330
"""Return an iterator of revision_id, line tuples.
332
For working trees (and mutable trees in general), the special
333
revision_id 'current:' will be used for lines that are new in this
334
tree, e.g. uncommitted changes.
335
:param file_id: The file to produce an annotated version from
336
:param default_revision: For lines that don't match a basis, mark them
337
with this revision id. Not all implementations will make use of
340
raise NotImplementedError(self.annotate_iter)
342
def _get_plan_merge_data(self, file_id, other, base):
343
from bzrlib import merge, versionedfile
344
vf = versionedfile._PlanMergeVersionedFile(file_id)
345
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
346
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
348
last_revision_base = None
350
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
351
return vf, last_revision_a, last_revision_b, last_revision_base
353
def plan_file_merge(self, file_id, other, base=None):
354
"""Generate a merge plan based on annotations.
356
If the file contains uncommitted changes in this tree, they will be
357
attributed to the 'current:' pseudo-revision. If the file contains
358
uncommitted changes in the other tree, they will be assigned to the
359
'other:' pseudo-revision.
361
data = self._get_plan_merge_data(file_id, other, base)
362
vf, last_revision_a, last_revision_b, last_revision_base = data
363
return vf.plan_merge(last_revision_a, last_revision_b,
366
def plan_file_lca_merge(self, file_id, other, base=None):
367
"""Generate a merge plan based lca-newness.
369
If the file contains uncommitted changes in this tree, they will be
370
attributed to the 'current:' pseudo-revision. If the file contains
371
uncommitted changes in the other tree, they will be assigned to the
372
'other:' pseudo-revision.
374
data = self._get_plan_merge_data(file_id, other, base)
375
vf, last_revision_a, last_revision_b, last_revision_base = data
376
return vf.plan_lca_merge(last_revision_a, last_revision_b,
379
def _iter_parent_trees(self):
380
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
381
for revision_id in self.get_parent_ids():
383
yield self.revision_tree(revision_id)
384
except errors.NoSuchRevisionInTree:
385
yield self.repository.revision_tree(revision_id)
388
def _file_revision(revision_tree, file_id):
389
"""Determine the revision associated with a file in a given tree."""
390
revision_tree.lock_read()
392
return revision_tree.inventory[file_id].revision
394
revision_tree.unlock()
396
def _get_file_revision(self, file_id, vf, tree_revision):
397
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
399
if getattr(self, '_repository', None) is None:
400
last_revision = tree_revision
401
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
402
self._iter_parent_trees()]
403
vf.add_lines((file_id, last_revision), parent_keys,
404
self.get_file(file_id).readlines())
405
repo = self.branch.repository
408
last_revision = self._file_revision(self, file_id)
409
base_vf = self._repository.texts
410
if base_vf not in vf.fallback_versionedfiles:
411
vf.fallback_versionedfiles.append(base_vf)
414
inventory = property(_get_inventory,
415
doc="Inventory of this Tree")
417
def _check_retrieved(self, ie, f):
420
fp = fingerprint_file(f)
423
if ie.text_size is not None:
424
if ie.text_size != fp['size']:
425
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
426
["inventory expects %d bytes" % ie.text_size,
427
"file is actually %d bytes" % fp['size'],
428
"store is probably damaged/corrupt"])
430
if ie.text_sha1 != fp['sha1']:
431
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
432
["inventory expects %s" % ie.text_sha1,
433
"file is actually %s" % fp['sha1'],
434
"store is probably damaged/corrupt"])
437
def path2id(self, path):
438
"""Return the id for path in this tree."""
439
return self._inventory.path2id(path)
441
def paths2ids(self, paths, trees=[], require_versioned=True):
442
"""Return all the ids that can be reached by walking from paths.
444
Each path is looked up in this tree and any extras provided in
445
trees, and this is repeated recursively: the children in an extra tree
446
of a directory that has been renamed under a provided path in this tree
447
are all returned, even if none exist under a provided path in this
448
tree, and vice versa.
450
:param paths: An iterable of paths to start converting to ids from.
451
Alternatively, if paths is None, no ids should be calculated and None
452
will be returned. This is offered to make calling the api unconditional
453
for code that *might* take a list of files.
454
:param trees: Additional trees to consider.
455
:param require_versioned: If False, do not raise NotVersionedError if
456
an element of paths is not versioned in this tree and all of trees.
458
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
460
def iter_children(self, file_id):
461
entry = self.iter_entries_by_dir([file_id]).next()[1]
462
for child in getattr(entry, 'children', {}).itervalues():
465
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
466
def print_file(self, file_id):
467
"""Print file with id `file_id` to stdout."""
469
sys.stdout.write(self.get_file_text(file_id))
474
def revision_tree(self, revision_id):
475
"""Obtain a revision tree for the revision revision_id.
477
The intention of this method is to allow access to possibly cached
478
tree data. Implementors of this method should raise NoSuchRevision if
479
the tree is not locally available, even if they could obtain the
480
tree via a repository or some other means. Callers are responsible
481
for finding the ultimate source for a revision tree.
483
:param revision_id: The revision_id of the requested tree.
485
:raises: NoSuchRevision if the tree cannot be obtained.
487
raise errors.NoSuchRevisionInTree(self, revision_id)
490
"""What files are present in this tree and unknown.
492
:return: an iterator over the unknown files.
499
def filter_unversioned_files(self, paths):
500
"""Filter out paths that are versioned.
502
:return: set of paths.
504
# NB: we specifically *don't* call self.has_filename, because for
505
# WorkingTrees that can indicate files that exist on disk but that
507
pred = self.inventory.has_filename
508
return set((p for p in paths if not pred(p)))
510
def walkdirs(self, prefix=""):
511
"""Walk the contents of this tree from path down.
513
This yields all the data about the contents of a directory at a time.
514
After each directory has been yielded, if the caller has mutated the
515
list to exclude some directories, they are then not descended into.
517
The data yielded is of the form:
518
((directory-relpath, directory-path-from-root, directory-fileid),
519
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
520
versioned_kind), ...]),
521
- directory-relpath is the containing dirs relpath from prefix
522
- directory-path-from-root is the containing dirs path from /
523
- directory-fileid is the id of the directory if it is versioned.
524
- relpath is the relative path within the subtree being walked.
525
- basename is the basename
526
- kind is the kind of the file now. If unknonwn then the file is not
527
present within the tree - but it may be recorded as versioned. See
529
- lstat is the stat data *if* the file was statted.
530
- path_from_tree_root is the path from the root of the tree.
531
- file_id is the file_id if the entry is versioned.
532
- versioned_kind is the kind of the file as last recorded in the
533
versioning system. If 'unknown' the file is not versioned.
534
One of 'kind' and 'versioned_kind' must not be 'unknown'.
536
:param prefix: Start walking from prefix within the tree rather than
537
at the root. This allows one to walk a subtree but get paths that are
538
relative to a tree rooted higher up.
539
:return: an iterator over the directory data.
541
raise NotImplementedError(self.walkdirs)
543
def iter_search_rules(self, path_names, pref_names=None,
544
_default_searcher=rules._per_user_searcher):
545
"""Find the preferences for filenames in a tree.
547
:param path_names: an iterable of paths to find attributes for.
548
Paths are given relative to the root of the tree.
549
:param pref_names: the list of preferences to lookup - None for all
550
:param _default_searcher: private parameter to assist testing - don't use
551
:return: an iterator of tuple sequences, one per path-name.
552
See _RulesSearcher.get_items for details on the tuple sequence.
554
searcher = self._get_rules_searcher(_default_searcher)
555
if searcher is not None:
556
if pref_names is not None:
557
for path in path_names:
558
yield searcher.get_selected_items(path, pref_names)
560
for path in path_names:
561
yield searcher.get_items(path)
564
def _get_rules_searcher(self, default_searcher):
565
"""Get the RulesSearcher for this tree given the default one."""
566
searcher = default_searcher
570
class EmptyTree(Tree):
573
self._inventory = Inventory(root_id=None)
574
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
575
' use repository.revision_tree instead.',
576
DeprecationWarning, stacklevel=2)
578
def get_parent_ids(self):
581
def get_symlink_target(self, file_id):
584
def has_filename(self, filename):
587
def kind(self, file_id):
590
def list_files(self, include_root=False):
593
def __contains__(self, file_id):
594
return (file_id in self._inventory)
596
def get_file_sha1(self, file_id, path=None, stat_value=None):
600
######################################################################
603
# TODO: Merge these two functions into a single one that can operate
604
# on either a whole tree or a set of files.
606
# TODO: Return the diff in order by filename, not by category or in
607
# random order. Can probably be done by lock-stepping through the
608
# filenames from both trees.
611
def file_status(filename, old_tree, new_tree):
612
"""Return single-letter status, old and new names for a file.
614
The complexity here is in deciding how to represent renames;
615
many complex cases are possible.
617
old_inv = old_tree.inventory
618
new_inv = new_tree.inventory
619
new_id = new_inv.path2id(filename)
620
old_id = old_inv.path2id(filename)
622
if not new_id and not old_id:
623
# easy: doesn't exist in either; not versioned at all
624
if new_tree.is_ignored(filename):
625
return 'I', None, None
627
return '?', None, None
629
# There is now a file of this name, great.
632
# There is no longer a file of this name, but we can describe
633
# what happened to the file that used to have
634
# this name. There are two possibilities: either it was
635
# deleted entirely, or renamed.
636
if new_inv.has_id(old_id):
637
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
639
return 'D', old_inv.id2path(old_id), None
641
# if the file_id is new in this revision, it is added
642
if new_id and not old_inv.has_id(new_id):
645
# if there used to be a file of this name, but that ID has now
646
# disappeared, it is deleted
647
if old_id and not new_inv.has_id(old_id):
653
@deprecated_function(deprecated_in((1, 9, 0)))
654
def find_renames(old_inv, new_inv):
655
for file_id in old_inv:
656
if file_id not in new_inv:
658
old_name = old_inv.id2path(file_id)
659
new_name = new_inv.id2path(file_id)
660
if old_name != new_name:
661
yield (old_name, new_name)
664
def find_ids_across_trees(filenames, trees, require_versioned=True):
665
"""Find the ids corresponding to specified filenames.
667
All matches in all trees will be used, and all children of matched
668
directories will be used.
670
:param filenames: The filenames to find file_ids for (if None, returns
672
:param trees: The trees to find file_ids within
673
:param require_versioned: if true, all specified filenames must occur in
675
:return: a set of file ids for the specified filenames and their children.
679
specified_path_ids = _find_ids_across_trees(filenames, trees,
681
return _find_children_across_trees(specified_path_ids, trees)
684
def _find_ids_across_trees(filenames, trees, require_versioned):
685
"""Find the ids corresponding to specified filenames.
687
All matches in all trees will be used, but subdirectories are not scanned.
689
:param filenames: The filenames to find file_ids for
690
:param trees: The trees to find file_ids within
691
:param require_versioned: if true, all specified filenames must occur in
693
:return: a set of file ids for the specified filenames
696
interesting_ids = set()
697
for tree_path in filenames:
700
file_id = tree.path2id(tree_path)
701
if file_id is not None:
702
interesting_ids.add(file_id)
705
not_versioned.append(tree_path)
706
if len(not_versioned) > 0 and require_versioned:
707
raise errors.PathsNotVersionedError(not_versioned)
708
return interesting_ids
711
def _find_children_across_trees(specified_ids, trees):
712
"""Return a set including specified ids and their children.
714
All matches in all trees will be used.
716
:param trees: The trees to find file_ids within
717
:return: a set containing all specified ids and their children
719
interesting_ids = set(specified_ids)
720
pending = interesting_ids
721
# now handle children of interesting ids
722
# we loop so that we handle all children of each id in both trees
723
while len(pending) > 0:
725
for file_id in pending:
727
if not tree.has_id(file_id):
729
for child_id in tree.iter_children(file_id):
730
if child_id not in interesting_ids:
731
new_pending.add(child_id)
732
interesting_ids.update(new_pending)
733
pending = new_pending
734
return interesting_ids
737
class InterTree(InterObject):
738
"""This class represents operations taking place between two Trees.
740
Its instances have methods like 'compare' and contain references to the
741
source and target trees these operations are to be carried out on.
743
Clients of bzrlib should not need to use InterTree directly, rather they
744
should use the convenience methods on Tree such as 'Tree.compare()' which
745
will pass through to InterTree as appropriate.
751
def compare(self, want_unchanged=False, specific_files=None,
752
extra_trees=None, require_versioned=False, include_root=False,
753
want_unversioned=False):
754
"""Return the changes from source to target.
756
:return: A TreeDelta.
757
:param specific_files: An optional list of file paths to restrict the
758
comparison to. When mapping filenames to ids, all matches in all
759
trees (including optional extra_trees) are used, and all children of
760
matched directories are included.
761
:param want_unchanged: An optional boolean requesting the inclusion of
762
unchanged entries in the result.
763
:param extra_trees: An optional list of additional trees to use when
764
mapping the contents of specific_files (paths) to file_ids.
765
:param require_versioned: An optional boolean (defaults to False). When
766
supplied and True all the 'specific_files' must be versioned, or
767
a PathsNotVersionedError will be thrown.
768
:param want_unversioned: Scan for unversioned paths.
770
# NB: show_status depends on being able to pass in non-versioned files
771
# and report them as unknown
772
trees = (self.source,)
773
if extra_trees is not None:
774
trees = trees + tuple(extra_trees)
775
# target is usually the newer tree:
776
specific_file_ids = self.target.paths2ids(specific_files, trees,
777
require_versioned=require_versioned)
778
if specific_files and not specific_file_ids:
779
# All files are unversioned, so just return an empty delta
780
# _compare_trees would think we want a complete delta
781
result = delta.TreeDelta()
782
fake_entry = InventoryFile('unused', 'unused', 'unused')
783
result.unversioned = [(path, None,
784
self.target._comparison_data(fake_entry, path)[0]) for path in
787
return delta._compare_trees(self.source, self.target, want_unchanged,
788
specific_files, include_root, extra_trees=extra_trees,
789
require_versioned=require_versioned,
790
want_unversioned=want_unversioned)
792
def iter_changes(self, include_unchanged=False,
793
specific_files=None, pb=None, extra_trees=[],
794
require_versioned=True, want_unversioned=False):
795
"""Generate an iterator of changes between trees.
798
(file_id, (path_in_source, path_in_target),
799
changed_content, versioned, parent, name, kind,
802
Changed_content is True if the file's content has changed. This
803
includes changes to its kind, and to a symlink's target.
805
versioned, parent, name, kind, executable are tuples of (from, to).
806
If a file is missing in a tree, its kind is None.
808
Iteration is done in parent-to-child order, relative to the target
811
There is no guarantee that all paths are in sorted order: the
812
requirement to expand the search due to renames may result in children
813
that should be found early being found late in the search, after
814
lexically later results have been returned.
815
:param require_versioned: Raise errors.PathsNotVersionedError if a
816
path in the specific_files list is not versioned in one of
817
source, target or extra_trees.
818
:param want_unversioned: Should unversioned files be returned in the
819
output. An unversioned file is defined as one with (False, False)
820
for the versioned pair.
823
lookup_trees = [self.source]
825
lookup_trees.extend(extra_trees)
826
if specific_files == []:
827
specific_file_ids = []
829
specific_file_ids = self.target.paths2ids(specific_files,
830
lookup_trees, require_versioned=require_versioned)
832
all_unversioned = sorted([(p.split('/'), p) for p in
834
if specific_files is None or
835
osutils.is_inside_any(specific_files, p)])
836
all_unversioned = deque(all_unversioned)
838
all_unversioned = deque()
840
from_entries_by_dir = list(self.source.iter_entries_by_dir(
841
specific_file_ids=specific_file_ids))
842
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
843
to_entries_by_dir = list(self.target.iter_entries_by_dir(
844
specific_file_ids=specific_file_ids))
845
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
847
# the unversioned path lookup only occurs on real trees - where there
848
# can be extras. So the fake_entry is solely used to look up
849
# executable it values when execute is not supported.
850
fake_entry = InventoryFile('unused', 'unused', 'unused')
851
for to_path, to_entry in to_entries_by_dir:
852
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
853
unversioned_path = all_unversioned.popleft()
854
to_kind, to_executable, to_stat = \
855
self.target._comparison_data(fake_entry, unversioned_path[1])
856
yield (None, (None, unversioned_path[1]), True, (False, False),
858
(None, unversioned_path[0][-1]),
860
(None, to_executable))
861
file_id = to_entry.file_id
862
to_paths[file_id] = to_path
864
changed_content = False
865
from_path, from_entry = from_data.get(file_id, (None, None))
866
from_versioned = (from_entry is not None)
867
if from_entry is not None:
868
from_versioned = True
869
from_name = from_entry.name
870
from_parent = from_entry.parent_id
871
from_kind, from_executable, from_stat = \
872
self.source._comparison_data(from_entry, from_path)
875
from_versioned = False
879
from_executable = None
880
versioned = (from_versioned, True)
881
to_kind, to_executable, to_stat = \
882
self.target._comparison_data(to_entry, to_path)
883
kind = (from_kind, to_kind)
884
if kind[0] != kind[1]:
885
changed_content = True
886
elif from_kind == 'file':
887
from_size = self.source._file_size(from_entry, from_stat)
888
to_size = self.target._file_size(to_entry, to_stat)
889
if from_size != to_size:
890
changed_content = True
891
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
892
self.target.get_file_sha1(file_id, to_path, to_stat)):
893
changed_content = True
894
elif from_kind == 'symlink':
895
if (self.source.get_symlink_target(file_id) !=
896
self.target.get_symlink_target(file_id)):
897
changed_content = True
898
elif from_kind == 'tree-reference':
899
if (self.source.get_reference_revision(file_id, from_path)
900
!= self.target.get_reference_revision(file_id, to_path)):
901
changed_content = True
902
parent = (from_parent, to_entry.parent_id)
903
name = (from_name, to_entry.name)
904
executable = (from_executable, to_executable)
906
pb.update('comparing files', entry_count, num_entries)
907
if (changed_content is not False or versioned[0] != versioned[1]
908
or parent[0] != parent[1] or name[0] != name[1] or
909
executable[0] != executable[1] or include_unchanged):
910
yield (file_id, (from_path, to_path), changed_content,
911
versioned, parent, name, kind, executable)
913
while all_unversioned:
914
# yield any trailing unversioned paths
915
unversioned_path = all_unversioned.popleft()
916
to_kind, to_executable, to_stat = \
917
self.target._comparison_data(fake_entry, unversioned_path[1])
918
yield (None, (None, unversioned_path[1]), True, (False, False),
920
(None, unversioned_path[0][-1]),
922
(None, to_executable))
924
def get_to_path(to_entry):
925
if to_entry.parent_id is None:
926
to_path = '' # the root
928
if to_entry.parent_id not in to_paths:
930
return get_to_path(self.target.inventory[to_entry.parent_id])
931
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
933
to_paths[to_entry.file_id] = to_path
936
for path, from_entry in from_entries_by_dir:
937
file_id = from_entry.file_id
938
if file_id in to_paths:
941
if not file_id in self.target.all_file_ids():
942
# common case - paths we have not emitted are not present in
946
to_path = get_to_path(self.target.inventory[file_id])
949
pb.update('comparing files', entry_count, num_entries)
950
versioned = (True, False)
951
parent = (from_entry.parent_id, None)
952
name = (from_entry.name, None)
953
from_kind, from_executable, stat_value = \
954
self.source._comparison_data(from_entry, path)
955
kind = (from_kind, None)
956
executable = (from_executable, None)
957
changed_content = from_kind is not None
958
# the parent's path is necessarily known at this point.
959
yield(file_id, (path, to_path), changed_content, versioned, parent,
960
name, kind, executable)
963
class MultiWalker(object):
964
"""Walk multiple trees simultaneously, getting combined results."""
966
# Note: This could be written to not assume you can do out-of-order
967
# lookups. Instead any nodes that don't match in all trees could be
968
# marked as 'deferred', and then returned in the final cleanup loop.
969
# For now, I think it is "nicer" to return things as close to the
970
# "master_tree" order as we can.
972
def __init__(self, master_tree, other_trees):
973
"""Create a new MultiWalker.
975
All trees being walked must implement "iter_entries_by_dir()", such
976
that they yield (path, object) tuples, where that object will have a
977
'.file_id' member, that can be used to check equality.
979
:param master_tree: All trees will be 'slaved' to the master_tree such
980
that nodes in master_tree will be used as 'first-pass' sync points.
981
Any nodes that aren't in master_tree will be merged in a second
983
:param other_trees: A list of other trees to walk simultaneously.
985
self._master_tree = master_tree
986
self._other_trees = other_trees
988
# Keep track of any nodes that were properly processed just out of
989
# order, that way we don't return them at the end, we don't have to
990
# track *all* processed file_ids, just the out-of-order ones
991
self._out_of_order_processed = set()
994
def _step_one(iterator):
995
"""Step an iter_entries_by_dir iterator.
997
:return: (has_more, path, ie)
998
If has_more is False, path and ie will be None.
1001
path, ie = iterator.next()
1002
except StopIteration:
1003
return False, None, None
1005
return True, path, ie
1008
def _cmp_path_by_dirblock(path1, path2):
1009
"""Compare two paths based on what directory they are in.
1011
This generates a sort order, such that all children of a directory are
1012
sorted together, and grandchildren are in the same order as the
1013
children appear. But all grandchildren come after all children.
1015
:param path1: first path
1016
:param path2: the second path
1017
:return: negative number if ``path1`` comes first,
1018
0 if paths are equal
1019
and a positive number if ``path2`` sorts first
1021
# Shortcut this special case
1024
# This is stolen from _dirstate_helpers_py.py, only switching it to
1025
# Unicode objects. Consider using encode_utf8() and then using the
1026
# optimized versions, or maybe writing optimized unicode versions.
1027
if not isinstance(path1, unicode):
1028
raise TypeError("'path1' must be a unicode string, not %s: %r"
1029
% (type(path1), path1))
1030
if not isinstance(path2, unicode):
1031
raise TypeError("'path2' must be a unicode string, not %s: %r"
1032
% (type(path2), path2))
1033
return cmp(MultiWalker._path_to_key(path1),
1034
MultiWalker._path_to_key(path2))
1037
def _path_to_key(path):
1038
dirname, basename = osutils.split(path)
1039
return (dirname.split(u'/'), basename)
1041
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1042
"""Lookup an inventory entry by file_id.
1044
This is called when an entry is missing in the normal order.
1045
Generally this is because a file was either renamed, or it was
1046
deleted/added. If the entry was found in the inventory and not in
1047
extra_entries, it will be added to self._out_of_order_processed
1049
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1050
should be filled with entries that were found before they were
1051
used. If file_id is present, it will be removed from the
1053
:param other_tree: The Tree to search, in case we didn't find the entry
1055
:param file_id: The file_id to look for
1056
:return: (path, ie) if found or (None, None) if not present.
1058
if file_id in extra_entries:
1059
return extra_entries.pop(file_id)
1060
# TODO: Is id2path better as the first call, or is
1061
# inventory[file_id] better as a first check?
1063
cur_path = other_tree.id2path(file_id)
1064
except errors.NoSuchId:
1066
if cur_path is None:
1069
self._out_of_order_processed.add(file_id)
1070
cur_ie = other_tree.inventory[file_id]
1071
return (cur_path, cur_ie)
1074
"""Match up the values in the different trees."""
1075
for result in self._walk_master_tree():
1077
self._finish_others()
1078
for result in self._walk_others():
1081
def _walk_master_tree(self):
1082
"""First pass, walk all trees in lock-step.
1084
When we are done, all nodes in the master_tree will have been
1085
processed. _other_walkers, _other_entries, and _others_extra will be
1086
set on 'self' for future processing.
1088
# This iterator has the most "inlining" done, because it tends to touch
1089
# every file in the tree, while the others only hit nodes that don't
1091
master_iterator = self._master_tree.iter_entries_by_dir()
1093
other_walkers = [other.iter_entries_by_dir()
1094
for other in self._other_trees]
1095
other_entries = [self._step_one(walker) for walker in other_walkers]
1096
# Track extra nodes in the other trees
1097
others_extra = [{} for i in xrange(len(self._other_trees))]
1099
master_has_more = True
1100
step_one = self._step_one
1101
lookup_by_file_id = self._lookup_by_file_id
1102
out_of_order_processed = self._out_of_order_processed
1104
while master_has_more:
1105
(master_has_more, path, master_ie) = step_one(master_iterator)
1106
if not master_has_more:
1109
file_id = master_ie.file_id
1111
other_values_append = other_values.append
1112
next_other_entries = []
1113
next_other_entries_append = next_other_entries.append
1114
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1115
if not other_has_more:
1116
other_values_append(lookup_by_file_id(
1117
others_extra[idx], self._other_trees[idx], file_id))
1118
next_other_entries_append((False, None, None))
1119
elif file_id == other_ie.file_id:
1120
# This is the critical code path, as most of the entries
1121
# should match between most trees.
1122
other_values_append((other_path, other_ie))
1123
next_other_entries_append(step_one(other_walkers[idx]))
1125
# This walker did not match, step it until it either
1126
# matches, or we know we are past the current walker.
1127
other_walker = other_walkers[idx]
1128
other_extra = others_extra[idx]
1129
while (other_has_more and
1130
self._cmp_path_by_dirblock(other_path, path) < 0):
1131
other_file_id = other_ie.file_id
1132
if other_file_id not in out_of_order_processed:
1133
other_extra[other_file_id] = (other_path, other_ie)
1134
other_has_more, other_path, other_ie = \
1135
step_one(other_walker)
1136
if other_has_more and other_ie.file_id == file_id:
1137
# We ended up walking to this point, match and step
1139
other_values_append((other_path, other_ie))
1140
other_has_more, other_path, other_ie = \
1141
step_one(other_walker)
1143
# This record isn't in the normal order, see if it
1145
other_values_append(lookup_by_file_id(
1146
other_extra, self._other_trees[idx], file_id))
1147
next_other_entries_append((other_has_more, other_path,
1149
other_entries = next_other_entries
1151
# We've matched all the walkers, yield this datapoint
1152
yield path, file_id, master_ie, other_values
1153
self._other_walkers = other_walkers
1154
self._other_entries = other_entries
1155
self._others_extra = others_extra
1157
def _finish_others(self):
1158
"""Finish walking the other iterators, so we get all entries."""
1159
for idx, info in enumerate(self._other_entries):
1160
other_extra = self._others_extra[idx]
1161
(other_has_more, other_path, other_ie) = info
1162
while other_has_more:
1163
other_file_id = other_ie.file_id
1164
if other_file_id not in self._out_of_order_processed:
1165
other_extra[other_file_id] = (other_path, other_ie)
1166
other_has_more, other_path, other_ie = \
1167
self._step_one(self._other_walkers[idx])
1168
del self._other_entries
1170
def _walk_others(self):
1171
"""Finish up by walking all the 'deferred' nodes."""
1172
# TODO: One alternative would be to grab all possible unprocessed
1173
# file_ids, and then sort by path, and then yield them. That
1174
# might ensure better ordering, in case a caller strictly
1175
# requires parents before children.
1176
for idx, other_extra in enumerate(self._others_extra):
1177
others = sorted(other_extra.itervalues(),
1178
key=lambda x: self._path_to_key(x[0]))
1179
for other_path, other_ie in others:
1180
file_id = other_ie.file_id
1181
# We don't need to check out_of_order_processed here, because
1182
# the lookup_by_file_id will be removing anything processed
1183
# from the extras cache
1184
other_extra.pop(file_id)
1185
other_values = [(None, None) for i in xrange(idx)]
1186
other_values.append((other_path, other_ie))
1187
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1188
alt_idx = alt_idx + idx + 1
1189
alt_extra = self._others_extra[alt_idx]
1190
alt_tree = self._other_trees[alt_idx]
1191
other_values.append(self._lookup_by_file_id(
1192
alt_extra, alt_tree, file_id))
1193
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
