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,
31
revision as _mod_revision,
35
from bzrlib.decorators import needs_read_lock
36
from bzrlib.errors import BzrError, BzrCheckError
37
from bzrlib import errors
38
from bzrlib.inventory import Inventory, InventoryFile
39
from bzrlib.inter import InterObject
40
from bzrlib.osutils import fingerprint_file
41
import bzrlib.revision
42
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
43
from bzrlib.trace import mutter, note
47
"""Abstract file tree.
49
There are several subclasses:
51
* `WorkingTree` exists as files on disk editable by the user.
53
* `RevisionTree` is a tree as recorded at some point in the past.
55
Trees contain an `Inventory` object, and also know how to retrieve
56
file texts mentioned in the inventory, either from a working
57
directory or from a store.
59
It is possible for trees to contain files that are not described
60
in their inventory or vice versa; for this use `filenames()`.
62
Trees can be compared, etc, regardless of whether they are working
63
trees or versioned trees.
66
def changes_from(self, other, want_unchanged=False, specific_files=None,
67
extra_trees=None, require_versioned=False, include_root=False,
68
want_unversioned=False):
69
"""Return a TreeDelta of the changes from other to this tree.
71
:param other: A tree to compare with.
72
:param specific_files: An optional list of file paths to restrict the
73
comparison to. When mapping filenames to ids, all matches in all
74
trees (including optional extra_trees) are used, and all children of
75
matched directories are included.
76
:param want_unchanged: An optional boolean requesting the inclusion of
77
unchanged entries in the result.
78
:param extra_trees: An optional list of additional trees to use when
79
mapping the contents of specific_files (paths) to file_ids.
80
:param require_versioned: An optional boolean (defaults to False). When
81
supplied and True all the 'specific_files' must be versioned, or
82
a PathsNotVersionedError will be thrown.
83
:param want_unversioned: Scan for unversioned paths.
85
The comparison will be performed by an InterTree object looked up on
88
# Martin observes that Tree.changes_from returns a TreeDelta and this
89
# may confuse people, because the class name of the returned object is
90
# a synonym of the object referenced in the method name.
91
return InterTree.get(other, self).compare(
92
want_unchanged=want_unchanged,
93
specific_files=specific_files,
94
extra_trees=extra_trees,
95
require_versioned=require_versioned,
96
include_root=include_root,
97
want_unversioned=want_unversioned,
100
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
101
def _iter_changes(self, *args, **kwargs):
102
return self.iter_changes(*args, **kwargs)
104
def iter_changes(self, from_tree, include_unchanged=False,
105
specific_files=None, pb=None, extra_trees=None,
106
require_versioned=True, want_unversioned=False):
107
intertree = InterTree.get(from_tree, self)
108
return intertree.iter_changes(include_unchanged, specific_files, pb,
109
extra_trees, require_versioned, want_unversioned=want_unversioned)
112
"""Get a list of the conflicts in the tree.
114
Each conflict is an instance of bzrlib.conflicts.Conflict.
116
return _mod_conflicts.ConflictList()
119
"""For trees that can have unversioned files, return all such paths."""
122
def get_parent_ids(self):
123
"""Get the parent ids for this tree.
125
:return: a list of parent ids. [] is returned to indicate
126
a tree with no parents.
127
:raises: BzrError if the parents are not known.
129
raise NotImplementedError(self.get_parent_ids)
131
def has_filename(self, filename):
132
"""True if the tree has given filename."""
133
raise NotImplementedError(self.has_filename)
135
def has_id(self, file_id):
136
return self.inventory.has_id(file_id)
138
def __contains__(self, file_id):
139
return self.has_id(file_id)
141
def has_or_had_id(self, file_id):
142
if file_id == self.inventory.root.file_id:
144
return self.inventory.has_id(file_id)
146
def is_ignored(self, filename):
147
"""Check whether the filename is ignored by this tree.
149
:param filename: The relative filename within the tree.
150
:return: True if the filename is ignored.
155
return iter(self.inventory)
157
def all_file_ids(self):
158
"""Iterate through all file ids, including ids for missing files."""
159
return set(self.inventory)
161
def id2path(self, file_id):
162
"""Return the path for a file id.
166
return self.inventory.id2path(file_id)
168
def is_control_filename(self, filename):
169
"""True if filename is the name of a control file in this tree.
171
:param filename: A filename within the tree. This is a relative path
172
from the root of this tree.
174
This is true IF and ONLY IF the filename is part of the meta data
175
that bzr controls in this tree. I.E. a random .bzr directory placed
176
on disk will not be a control file for this tree.
178
return self.bzrdir.is_control_filename(filename)
181
def iter_entries_by_dir(self, specific_file_ids=None):
182
"""Walk the tree in 'by_dir' order.
184
This will yield each entry in the tree as a (path, entry) tuple.
185
The order that they are yielded is:
187
Directories are walked in a depth-first lexicographical order,
188
however, whenever a directory is reached, all of its direct child
189
nodes are yielded in lexicographical order before yielding the
192
For example, in the tree::
202
The yield order (ignoring root) would be::
203
a, f, a/b, a/d, a/b/c, a/d/e, f/g
205
return self.inventory.iter_entries_by_dir(
206
specific_file_ids=specific_file_ids)
208
def iter_references(self):
209
for path, entry in self.iter_entries_by_dir():
210
if entry.kind == 'tree-reference':
211
yield path, entry.file_id
213
def kind(self, file_id):
214
raise NotImplementedError("Tree subclass %s must implement kind"
215
% self.__class__.__name__)
217
def stored_kind(self, file_id):
218
"""File kind stored for this file_id.
220
May not match kind on disk for working trees. Always available
221
for versioned files, even when the file itself is missing.
223
return self.kind(file_id)
225
def path_content_summary(self, path):
226
"""Get a summary of the information about path.
228
:param path: A relative path within the tree.
229
:return: A tuple containing kind, size, exec, sha1-or-link.
230
Kind is always present (see tree.kind()).
231
size is present if kind is file, None otherwise.
232
exec is None unless kind is file and the platform supports the 'x'
234
sha1-or-link is the link target if kind is symlink, or the sha1 if
235
it can be obtained without reading the file.
237
raise NotImplementedError(self.path_content_summary)
239
def get_reference_revision(self, file_id, path=None):
240
raise NotImplementedError("Tree subclass %s must implement "
241
"get_reference_revision"
242
% self.__class__.__name__)
244
def _comparison_data(self, entry, path):
245
"""Return a tuple of kind, executable, stat_value for a file.
247
entry may be None if there is no inventory entry for the file, but
248
path must always be supplied.
250
kind is None if there is no file present (even if an inventory id is
251
present). executable is False for non-file entries.
253
raise NotImplementedError(self._comparison_data)
255
def _file_size(self, entry, stat_value):
256
raise NotImplementedError(self._file_size)
258
def _get_inventory(self):
259
return self._inventory
261
def get_file(self, file_id, path=None):
262
"""Return a file object for the file file_id in the tree.
264
If both file_id and path are defined, it is implementation defined as
265
to which one is used.
267
raise NotImplementedError(self.get_file)
269
def get_file_text(self, file_id, path=None):
270
"""Return the byte content of a file.
272
:param file_id: The file_id of the file.
273
:param path: The path of the file.
274
If both file_id and path are supplied, an implementation may use
277
my_file = self.get_file(file_id, path)
279
return my_file.read()
283
def get_file_lines(self, file_id, path=None):
284
"""Return the content of a file, as lines.
286
:param file_id: The file_id of the file.
287
:param path: The path of the file.
288
If both file_id and path are supplied, an implementation may use
291
return osutils.split_lines(self.get_file_text(file_id, path))
293
def get_file_mtime(self, file_id, path=None):
294
"""Return the modification time for a file.
296
:param file_id: The handle for this file.
297
:param path: The path that this file can be found at.
298
These must point to the same object.
300
raise NotImplementedError(self.get_file_mtime)
302
def get_file_size(self, file_id):
303
"""Return the size of a file in bytes.
305
This applies only to regular files. If invoked on directories or
306
symlinks, it will return None.
307
:param file_id: The file-id of the file
309
raise NotImplementedError(self.get_file_size)
311
def get_file_by_path(self, path):
312
return self.get_file(self._inventory.path2id(path), path)
314
def get_special_file(self, type):
315
"""Get a file special to Bazaar.
317
:type: a type of XXX maps to a file path .bzrXXX
318
:return: a file-like object or None if the file does not exist
321
file_id = self.path2id(path)
325
return self.get_file(file_id, path)
327
def iter_files_bytes(self, desired_files):
328
"""Iterate through file contents.
330
Files will not necessarily be returned in the order they occur in
331
desired_files. No specific order is guaranteed.
333
Yields pairs of identifier, bytes_iterator. identifier is an opaque
334
value supplied by the caller as part of desired_files. It should
335
uniquely identify the file version in the caller's context. (Examples:
336
an index number or a TreeTransform trans_id.)
338
bytes_iterator is an iterable of bytestrings for the file. The
339
kind of iterable and length of the bytestrings are unspecified, but for
340
this implementation, it is a tuple containing a single bytestring with
341
the complete text of the file.
343
:param desired_files: a list of (file_id, identifier) pairs
345
for file_id, identifier in desired_files:
346
# We wrap the string in a tuple so that we can return an iterable
347
# of bytestrings. (Technically, a bytestring is also an iterable
348
# of bytestrings, but iterating through each character is not
350
cur_file = (self.get_file_text(file_id),)
351
yield identifier, cur_file
353
def get_symlink_target(self, file_id):
354
"""Get the target for a given file_id.
356
It is assumed that the caller already knows that file_id is referencing
358
:param file_id: Handle for the symlink entry.
359
:return: The path the symlink points to.
361
raise NotImplementedError(self.get_symlink_target)
363
def get_root_id(self):
364
"""Return the file_id for the root of this tree."""
365
raise NotImplementedError(self.get_root_id)
367
def annotate_iter(self, file_id,
368
default_revision=_mod_revision.CURRENT_REVISION):
369
"""Return an iterator of revision_id, line tuples.
371
For working trees (and mutable trees in general), the special
372
revision_id 'current:' will be used for lines that are new in this
373
tree, e.g. uncommitted changes.
374
:param file_id: The file to produce an annotated version from
375
:param default_revision: For lines that don't match a basis, mark them
376
with this revision id. Not all implementations will make use of
379
raise NotImplementedError(self.annotate_iter)
381
def _get_plan_merge_data(self, file_id, other, base):
382
from bzrlib import merge, versionedfile
383
vf = versionedfile._PlanMergeVersionedFile(file_id)
384
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
385
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
387
last_revision_base = None
389
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
390
return vf, last_revision_a, last_revision_b, last_revision_base
392
def plan_file_merge(self, file_id, other, base=None):
393
"""Generate a merge plan based on annotations.
395
If the file contains uncommitted changes in this tree, they will be
396
attributed to the 'current:' pseudo-revision. If the file contains
397
uncommitted changes in the other tree, they will be assigned to the
398
'other:' pseudo-revision.
400
data = self._get_plan_merge_data(file_id, other, base)
401
vf, last_revision_a, last_revision_b, last_revision_base = data
402
return vf.plan_merge(last_revision_a, last_revision_b,
405
def plan_file_lca_merge(self, file_id, other, base=None):
406
"""Generate a merge plan based lca-newness.
408
If the file contains uncommitted changes in this tree, they will be
409
attributed to the 'current:' pseudo-revision. If the file contains
410
uncommitted changes in the other tree, they will be assigned to the
411
'other:' pseudo-revision.
413
data = self._get_plan_merge_data(file_id, other, base)
414
vf, last_revision_a, last_revision_b, last_revision_base = data
415
return vf.plan_lca_merge(last_revision_a, last_revision_b,
418
def _iter_parent_trees(self):
419
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
420
for revision_id in self.get_parent_ids():
422
yield self.revision_tree(revision_id)
423
except errors.NoSuchRevisionInTree:
424
yield self.repository.revision_tree(revision_id)
427
def _file_revision(revision_tree, file_id):
428
"""Determine the revision associated with a file in a given tree."""
429
revision_tree.lock_read()
431
return revision_tree.inventory[file_id].revision
433
revision_tree.unlock()
435
def _get_file_revision(self, file_id, vf, tree_revision):
436
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
438
if getattr(self, '_repository', None) is None:
439
last_revision = tree_revision
440
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
441
self._iter_parent_trees()]
442
vf.add_lines((file_id, last_revision), parent_keys,
443
self.get_file(file_id).readlines())
444
repo = self.branch.repository
447
last_revision = self._file_revision(self, file_id)
448
base_vf = self._repository.texts
449
if base_vf not in vf.fallback_versionedfiles:
450
vf.fallback_versionedfiles.append(base_vf)
453
inventory = property(_get_inventory,
454
doc="Inventory of this Tree")
456
def _check_retrieved(self, ie, f):
459
fp = fingerprint_file(f)
462
if ie.text_size is not None:
463
if ie.text_size != fp['size']:
464
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
465
["inventory expects %d bytes" % ie.text_size,
466
"file is actually %d bytes" % fp['size'],
467
"store is probably damaged/corrupt"])
469
if ie.text_sha1 != fp['sha1']:
470
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
471
["inventory expects %s" % ie.text_sha1,
472
"file is actually %s" % fp['sha1'],
473
"store is probably damaged/corrupt"])
476
def path2id(self, path):
477
"""Return the id for path in this tree."""
478
return self._inventory.path2id(path)
480
def paths2ids(self, paths, trees=[], require_versioned=True):
481
"""Return all the ids that can be reached by walking from paths.
483
Each path is looked up in this tree and any extras provided in
484
trees, and this is repeated recursively: the children in an extra tree
485
of a directory that has been renamed under a provided path in this tree
486
are all returned, even if none exist under a provided path in this
487
tree, and vice versa.
489
:param paths: An iterable of paths to start converting to ids from.
490
Alternatively, if paths is None, no ids should be calculated and None
491
will be returned. This is offered to make calling the api unconditional
492
for code that *might* take a list of files.
493
:param trees: Additional trees to consider.
494
:param require_versioned: If False, do not raise NotVersionedError if
495
an element of paths is not versioned in this tree and all of trees.
497
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
499
def iter_children(self, file_id):
500
entry = self.iter_entries_by_dir([file_id]).next()[1]
501
for child in getattr(entry, 'children', {}).itervalues():
504
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
505
def print_file(self, file_id):
506
"""Print file with id `file_id` to stdout."""
508
sys.stdout.write(self.get_file_text(file_id))
513
def revision_tree(self, revision_id):
514
"""Obtain a revision tree for the revision revision_id.
516
The intention of this method is to allow access to possibly cached
517
tree data. Implementors of this method should raise NoSuchRevision if
518
the tree is not locally available, even if they could obtain the
519
tree via a repository or some other means. Callers are responsible
520
for finding the ultimate source for a revision tree.
522
:param revision_id: The revision_id of the requested tree.
524
:raises: NoSuchRevision if the tree cannot be obtained.
526
raise errors.NoSuchRevisionInTree(self, revision_id)
529
"""What files are present in this tree and unknown.
531
:return: an iterator over the unknown files.
538
def filter_unversioned_files(self, paths):
539
"""Filter out paths that are versioned.
541
:return: set of paths.
543
# NB: we specifically *don't* call self.has_filename, because for
544
# WorkingTrees that can indicate files that exist on disk but that
546
pred = self.inventory.has_filename
547
return set((p for p in paths if not pred(p)))
549
def walkdirs(self, prefix=""):
550
"""Walk the contents of this tree from path down.
552
This yields all the data about the contents of a directory at a time.
553
After each directory has been yielded, if the caller has mutated the
554
list to exclude some directories, they are then not descended into.
556
The data yielded is of the form:
557
((directory-relpath, directory-path-from-root, directory-fileid),
558
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
559
versioned_kind), ...]),
560
- directory-relpath is the containing dirs relpath from prefix
561
- directory-path-from-root is the containing dirs path from /
562
- directory-fileid is the id of the directory if it is versioned.
563
- relpath is the relative path within the subtree being walked.
564
- basename is the basename
565
- kind is the kind of the file now. If unknonwn then the file is not
566
present within the tree - but it may be recorded as versioned. See
568
- lstat is the stat data *if* the file was statted.
569
- path_from_tree_root is the path from the root of the tree.
570
- file_id is the file_id if the entry is versioned.
571
- versioned_kind is the kind of the file as last recorded in the
572
versioning system. If 'unknown' the file is not versioned.
573
One of 'kind' and 'versioned_kind' must not be 'unknown'.
575
:param prefix: Start walking from prefix within the tree rather than
576
at the root. This allows one to walk a subtree but get paths that are
577
relative to a tree rooted higher up.
578
:return: an iterator over the directory data.
580
raise NotImplementedError(self.walkdirs)
582
def _content_filter_stack(self, path=None, file_id=None):
583
"""The stack of content filters for a path.
585
Readers will be applied in first-to-last order.
586
Writers will be applied in last-to-first order.
587
Either the path or the file-id needs to be provided.
589
:param path: path relative to the root of the tree
591
:param file_id: file_id or None if unknown
592
:return: the list of filters - [] if there are none
594
filter_pref_names = filters._get_registered_names()
595
if len(filter_pref_names) == 0:
598
path = self.id2path(file_id)
599
prefs = self.iter_search_rules([path], filter_pref_names).next()
600
stk = filters._get_filter_stack_for(prefs)
601
if 'filters' in debug.debug_flags:
602
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
605
def iter_search_rules(self, path_names, pref_names=None,
606
_default_searcher=rules._per_user_searcher):
607
"""Find the preferences for filenames in a tree.
609
:param path_names: an iterable of paths to find attributes for.
610
Paths are given relative to the root of the tree.
611
:param pref_names: the list of preferences to lookup - None for all
612
:param _default_searcher: private parameter to assist testing - don't use
613
:return: an iterator of tuple sequences, one per path-name.
614
See _RulesSearcher.get_items for details on the tuple sequence.
616
searcher = self._get_rules_searcher(_default_searcher)
617
if searcher is not None:
618
if pref_names is not None:
619
for path in path_names:
620
yield searcher.get_selected_items(path, pref_names)
622
for path in path_names:
623
yield searcher.get_items(path)
626
def _get_rules_searcher(self, default_searcher):
627
"""Get the RulesSearcher for this tree given the default one."""
628
searcher = default_searcher
632
class EmptyTree(Tree):
635
self._inventory = Inventory(root_id=None)
636
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
637
' use repository.revision_tree instead.',
638
DeprecationWarning, stacklevel=2)
640
def get_parent_ids(self):
643
def get_symlink_target(self, file_id):
646
def has_filename(self, filename):
649
def kind(self, file_id):
652
def list_files(self, include_root=False):
655
def __contains__(self, file_id):
656
return (file_id in self._inventory)
658
def get_file_sha1(self, file_id, path=None, stat_value=None):
662
######################################################################
665
# TODO: Merge these two functions into a single one that can operate
666
# on either a whole tree or a set of files.
668
# TODO: Return the diff in order by filename, not by category or in
669
# random order. Can probably be done by lock-stepping through the
670
# filenames from both trees.
673
def file_status(filename, old_tree, new_tree):
674
"""Return single-letter status, old and new names for a file.
676
The complexity here is in deciding how to represent renames;
677
many complex cases are possible.
679
old_inv = old_tree.inventory
680
new_inv = new_tree.inventory
681
new_id = new_inv.path2id(filename)
682
old_id = old_inv.path2id(filename)
684
if not new_id and not old_id:
685
# easy: doesn't exist in either; not versioned at all
686
if new_tree.is_ignored(filename):
687
return 'I', None, None
689
return '?', None, None
691
# There is now a file of this name, great.
694
# There is no longer a file of this name, but we can describe
695
# what happened to the file that used to have
696
# this name. There are two possibilities: either it was
697
# deleted entirely, or renamed.
698
if new_inv.has_id(old_id):
699
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
701
return 'D', old_inv.id2path(old_id), None
703
# if the file_id is new in this revision, it is added
704
if new_id and not old_inv.has_id(new_id):
707
# if there used to be a file of this name, but that ID has now
708
# disappeared, it is deleted
709
if old_id and not new_inv.has_id(old_id):
715
@deprecated_function(deprecated_in((1, 9, 0)))
716
def find_renames(old_inv, new_inv):
717
for file_id in old_inv:
718
if file_id not in new_inv:
720
old_name = old_inv.id2path(file_id)
721
new_name = new_inv.id2path(file_id)
722
if old_name != new_name:
723
yield (old_name, new_name)
726
def find_ids_across_trees(filenames, trees, require_versioned=True):
727
"""Find the ids corresponding to specified filenames.
729
All matches in all trees will be used, and all children of matched
730
directories will be used.
732
:param filenames: The filenames to find file_ids for (if None, returns
734
:param trees: The trees to find file_ids within
735
:param require_versioned: if true, all specified filenames must occur in
737
:return: a set of file ids for the specified filenames and their children.
741
specified_path_ids = _find_ids_across_trees(filenames, trees,
743
return _find_children_across_trees(specified_path_ids, trees)
746
def _find_ids_across_trees(filenames, trees, require_versioned):
747
"""Find the ids corresponding to specified filenames.
749
All matches in all trees will be used, but subdirectories are not scanned.
751
:param filenames: The filenames to find file_ids for
752
:param trees: The trees to find file_ids within
753
:param require_versioned: if true, all specified filenames must occur in
755
:return: a set of file ids for the specified filenames
758
interesting_ids = set()
759
for tree_path in filenames:
762
file_id = tree.path2id(tree_path)
763
if file_id is not None:
764
interesting_ids.add(file_id)
767
not_versioned.append(tree_path)
768
if len(not_versioned) > 0 and require_versioned:
769
raise errors.PathsNotVersionedError(not_versioned)
770
return interesting_ids
773
def _find_children_across_trees(specified_ids, trees):
774
"""Return a set including specified ids and their children.
776
All matches in all trees will be used.
778
:param trees: The trees to find file_ids within
779
:return: a set containing all specified ids and their children
781
interesting_ids = set(specified_ids)
782
pending = interesting_ids
783
# now handle children of interesting ids
784
# we loop so that we handle all children of each id in both trees
785
while len(pending) > 0:
787
for file_id in pending:
789
if not tree.has_id(file_id):
791
for child_id in tree.iter_children(file_id):
792
if child_id not in interesting_ids:
793
new_pending.add(child_id)
794
interesting_ids.update(new_pending)
795
pending = new_pending
796
return interesting_ids
799
class InterTree(InterObject):
800
"""This class represents operations taking place between two Trees.
802
Its instances have methods like 'compare' and contain references to the
803
source and target trees these operations are to be carried out on.
805
Clients of bzrlib should not need to use InterTree directly, rather they
806
should use the convenience methods on Tree such as 'Tree.compare()' which
807
will pass through to InterTree as appropriate.
813
def compare(self, want_unchanged=False, specific_files=None,
814
extra_trees=None, require_versioned=False, include_root=False,
815
want_unversioned=False):
816
"""Return the changes from source to target.
818
:return: A TreeDelta.
819
:param specific_files: An optional list of file paths to restrict the
820
comparison to. When mapping filenames to ids, all matches in all
821
trees (including optional extra_trees) are used, and all children of
822
matched directories are included.
823
:param want_unchanged: An optional boolean requesting the inclusion of
824
unchanged entries in the result.
825
:param extra_trees: An optional list of additional trees to use when
826
mapping the contents of specific_files (paths) to file_ids.
827
:param require_versioned: An optional boolean (defaults to False). When
828
supplied and True all the 'specific_files' must be versioned, or
829
a PathsNotVersionedError will be thrown.
830
:param want_unversioned: Scan for unversioned paths.
832
# NB: show_status depends on being able to pass in non-versioned files
833
# and report them as unknown
834
trees = (self.source,)
835
if extra_trees is not None:
836
trees = trees + tuple(extra_trees)
837
# target is usually the newer tree:
838
specific_file_ids = self.target.paths2ids(specific_files, trees,
839
require_versioned=require_versioned)
840
if specific_files and not specific_file_ids:
841
# All files are unversioned, so just return an empty delta
842
# _compare_trees would think we want a complete delta
843
result = delta.TreeDelta()
844
fake_entry = InventoryFile('unused', 'unused', 'unused')
845
result.unversioned = [(path, None,
846
self.target._comparison_data(fake_entry, path)[0]) for path in
849
return delta._compare_trees(self.source, self.target, want_unchanged,
850
specific_files, include_root, extra_trees=extra_trees,
851
require_versioned=require_versioned,
852
want_unversioned=want_unversioned)
854
def iter_changes(self, include_unchanged=False,
855
specific_files=None, pb=None, extra_trees=[],
856
require_versioned=True, want_unversioned=False):
857
"""Generate an iterator of changes between trees.
860
(file_id, (path_in_source, path_in_target),
861
changed_content, versioned, parent, name, kind,
864
Changed_content is True if the file's content has changed. This
865
includes changes to its kind, and to a symlink's target.
867
versioned, parent, name, kind, executable are tuples of (from, to).
868
If a file is missing in a tree, its kind is None.
870
Iteration is done in parent-to-child order, relative to the target
873
There is no guarantee that all paths are in sorted order: the
874
requirement to expand the search due to renames may result in children
875
that should be found early being found late in the search, after
876
lexically later results have been returned.
877
:param require_versioned: Raise errors.PathsNotVersionedError if a
878
path in the specific_files list is not versioned in one of
879
source, target or extra_trees.
880
:param want_unversioned: Should unversioned files be returned in the
881
output. An unversioned file is defined as one with (False, False)
882
for the versioned pair.
885
lookup_trees = [self.source]
887
lookup_trees.extend(extra_trees)
888
if specific_files == []:
889
specific_file_ids = []
891
specific_file_ids = self.target.paths2ids(specific_files,
892
lookup_trees, require_versioned=require_versioned)
894
all_unversioned = sorted([(p.split('/'), p) for p in
896
if specific_files is None or
897
osutils.is_inside_any(specific_files, p)])
898
all_unversioned = deque(all_unversioned)
900
all_unversioned = deque()
902
from_entries_by_dir = list(self.source.iter_entries_by_dir(
903
specific_file_ids=specific_file_ids))
904
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
905
to_entries_by_dir = list(self.target.iter_entries_by_dir(
906
specific_file_ids=specific_file_ids))
907
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
909
# the unversioned path lookup only occurs on real trees - where there
910
# can be extras. So the fake_entry is solely used to look up
911
# executable it values when execute is not supported.
912
fake_entry = InventoryFile('unused', 'unused', 'unused')
913
for to_path, to_entry in to_entries_by_dir:
914
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
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))
923
file_id = to_entry.file_id
924
to_paths[file_id] = to_path
926
changed_content = False
927
from_path, from_entry = from_data.get(file_id, (None, None))
928
from_versioned = (from_entry is not None)
929
if from_entry is not None:
930
from_versioned = True
931
from_name = from_entry.name
932
from_parent = from_entry.parent_id
933
from_kind, from_executable, from_stat = \
934
self.source._comparison_data(from_entry, from_path)
937
from_versioned = False
941
from_executable = None
942
versioned = (from_versioned, True)
943
to_kind, to_executable, to_stat = \
944
self.target._comparison_data(to_entry, to_path)
945
kind = (from_kind, to_kind)
946
if kind[0] != kind[1]:
947
changed_content = True
948
elif from_kind == 'file':
949
if (self.source.get_file_sha1(file_id, from_path, from_stat) !=
950
self.target.get_file_sha1(file_id, to_path, to_stat)):
951
changed_content = True
952
elif from_kind == 'symlink':
953
if (self.source.get_symlink_target(file_id) !=
954
self.target.get_symlink_target(file_id)):
955
changed_content = True
956
elif from_kind == 'tree-reference':
957
if (self.source.get_reference_revision(file_id, from_path)
958
!= self.target.get_reference_revision(file_id, to_path)):
959
changed_content = True
960
parent = (from_parent, to_entry.parent_id)
961
name = (from_name, to_entry.name)
962
executable = (from_executable, to_executable)
964
pb.update('comparing files', entry_count, num_entries)
965
if (changed_content is not False or versioned[0] != versioned[1]
966
or parent[0] != parent[1] or name[0] != name[1] or
967
executable[0] != executable[1] or include_unchanged):
968
yield (file_id, (from_path, to_path), changed_content,
969
versioned, parent, name, kind, executable)
971
while all_unversioned:
972
# yield any trailing unversioned paths
973
unversioned_path = all_unversioned.popleft()
974
to_kind, to_executable, to_stat = \
975
self.target._comparison_data(fake_entry, unversioned_path[1])
976
yield (None, (None, unversioned_path[1]), True, (False, False),
978
(None, unversioned_path[0][-1]),
980
(None, to_executable))
982
def get_to_path(to_entry):
983
if to_entry.parent_id is None:
984
to_path = '' # the root
986
if to_entry.parent_id not in to_paths:
988
return get_to_path(self.target.inventory[to_entry.parent_id])
989
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
991
to_paths[to_entry.file_id] = to_path
994
for path, from_entry in from_entries_by_dir:
995
file_id = from_entry.file_id
996
if file_id in to_paths:
999
if not file_id in self.target.all_file_ids():
1000
# common case - paths we have not emitted are not present in
1004
to_path = get_to_path(self.target.inventory[file_id])
1007
pb.update('comparing files', entry_count, num_entries)
1008
versioned = (True, False)
1009
parent = (from_entry.parent_id, None)
1010
name = (from_entry.name, None)
1011
from_kind, from_executable, stat_value = \
1012
self.source._comparison_data(from_entry, path)
1013
kind = (from_kind, None)
1014
executable = (from_executable, None)
1015
changed_content = from_kind is not None
1016
# the parent's path is necessarily known at this point.
1017
yield(file_id, (path, to_path), changed_content, versioned, parent,
1018
name, kind, executable)
1021
class MultiWalker(object):
1022
"""Walk multiple trees simultaneously, getting combined results."""
1024
# Note: This could be written to not assume you can do out-of-order
1025
# lookups. Instead any nodes that don't match in all trees could be
1026
# marked as 'deferred', and then returned in the final cleanup loop.
1027
# For now, I think it is "nicer" to return things as close to the
1028
# "master_tree" order as we can.
1030
def __init__(self, master_tree, other_trees):
1031
"""Create a new MultiWalker.
1033
All trees being walked must implement "iter_entries_by_dir()", such
1034
that they yield (path, object) tuples, where that object will have a
1035
'.file_id' member, that can be used to check equality.
1037
:param master_tree: All trees will be 'slaved' to the master_tree such
1038
that nodes in master_tree will be used as 'first-pass' sync points.
1039
Any nodes that aren't in master_tree will be merged in a second
1041
:param other_trees: A list of other trees to walk simultaneously.
1043
self._master_tree = master_tree
1044
self._other_trees = other_trees
1046
# Keep track of any nodes that were properly processed just out of
1047
# order, that way we don't return them at the end, we don't have to
1048
# track *all* processed file_ids, just the out-of-order ones
1049
self._out_of_order_processed = set()
1052
def _step_one(iterator):
1053
"""Step an iter_entries_by_dir iterator.
1055
:return: (has_more, path, ie)
1056
If has_more is False, path and ie will be None.
1059
path, ie = iterator.next()
1060
except StopIteration:
1061
return False, None, None
1063
return True, path, ie
1066
def _cmp_path_by_dirblock(path1, path2):
1067
"""Compare two paths based on what directory they are in.
1069
This generates a sort order, such that all children of a directory are
1070
sorted together, and grandchildren are in the same order as the
1071
children appear. But all grandchildren come after all children.
1073
:param path1: first path
1074
:param path2: the second path
1075
:return: negative number if ``path1`` comes first,
1076
0 if paths are equal
1077
and a positive number if ``path2`` sorts first
1079
# Shortcut this special case
1082
# This is stolen from _dirstate_helpers_py.py, only switching it to
1083
# Unicode objects. Consider using encode_utf8() and then using the
1084
# optimized versions, or maybe writing optimized unicode versions.
1085
if not isinstance(path1, unicode):
1086
raise TypeError("'path1' must be a unicode string, not %s: %r"
1087
% (type(path1), path1))
1088
if not isinstance(path2, unicode):
1089
raise TypeError("'path2' must be a unicode string, not %s: %r"
1090
% (type(path2), path2))
1091
return cmp(MultiWalker._path_to_key(path1),
1092
MultiWalker._path_to_key(path2))
1095
def _path_to_key(path):
1096
dirname, basename = osutils.split(path)
1097
return (dirname.split(u'/'), basename)
1099
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1100
"""Lookup an inventory entry by file_id.
1102
This is called when an entry is missing in the normal order.
1103
Generally this is because a file was either renamed, or it was
1104
deleted/added. If the entry was found in the inventory and not in
1105
extra_entries, it will be added to self._out_of_order_processed
1107
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1108
should be filled with entries that were found before they were
1109
used. If file_id is present, it will be removed from the
1111
:param other_tree: The Tree to search, in case we didn't find the entry
1113
:param file_id: The file_id to look for
1114
:return: (path, ie) if found or (None, None) if not present.
1116
if file_id in extra_entries:
1117
return extra_entries.pop(file_id)
1118
# TODO: Is id2path better as the first call, or is
1119
# inventory[file_id] better as a first check?
1121
cur_path = other_tree.id2path(file_id)
1122
except errors.NoSuchId:
1124
if cur_path is None:
1127
self._out_of_order_processed.add(file_id)
1128
cur_ie = other_tree.inventory[file_id]
1129
return (cur_path, cur_ie)
1132
"""Match up the values in the different trees."""
1133
for result in self._walk_master_tree():
1135
self._finish_others()
1136
for result in self._walk_others():
1139
def _walk_master_tree(self):
1140
"""First pass, walk all trees in lock-step.
1142
When we are done, all nodes in the master_tree will have been
1143
processed. _other_walkers, _other_entries, and _others_extra will be
1144
set on 'self' for future processing.
1146
# This iterator has the most "inlining" done, because it tends to touch
1147
# every file in the tree, while the others only hit nodes that don't
1149
master_iterator = self._master_tree.iter_entries_by_dir()
1151
other_walkers = [other.iter_entries_by_dir()
1152
for other in self._other_trees]
1153
other_entries = [self._step_one(walker) for walker in other_walkers]
1154
# Track extra nodes in the other trees
1155
others_extra = [{} for i in xrange(len(self._other_trees))]
1157
master_has_more = True
1158
step_one = self._step_one
1159
lookup_by_file_id = self._lookup_by_file_id
1160
out_of_order_processed = self._out_of_order_processed
1162
while master_has_more:
1163
(master_has_more, path, master_ie) = step_one(master_iterator)
1164
if not master_has_more:
1167
file_id = master_ie.file_id
1169
other_values_append = other_values.append
1170
next_other_entries = []
1171
next_other_entries_append = next_other_entries.append
1172
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1173
if not other_has_more:
1174
other_values_append(lookup_by_file_id(
1175
others_extra[idx], self._other_trees[idx], file_id))
1176
next_other_entries_append((False, None, None))
1177
elif file_id == other_ie.file_id:
1178
# This is the critical code path, as most of the entries
1179
# should match between most trees.
1180
other_values_append((other_path, other_ie))
1181
next_other_entries_append(step_one(other_walkers[idx]))
1183
# This walker did not match, step it until it either
1184
# matches, or we know we are past the current walker.
1185
other_walker = other_walkers[idx]
1186
other_extra = others_extra[idx]
1187
while (other_has_more and
1188
self._cmp_path_by_dirblock(other_path, path) < 0):
1189
other_file_id = other_ie.file_id
1190
if other_file_id not in out_of_order_processed:
1191
other_extra[other_file_id] = (other_path, other_ie)
1192
other_has_more, other_path, other_ie = \
1193
step_one(other_walker)
1194
if other_has_more and other_ie.file_id == file_id:
1195
# We ended up walking to this point, match and step
1197
other_values_append((other_path, other_ie))
1198
other_has_more, other_path, other_ie = \
1199
step_one(other_walker)
1201
# This record isn't in the normal order, see if it
1203
other_values_append(lookup_by_file_id(
1204
other_extra, self._other_trees[idx], file_id))
1205
next_other_entries_append((other_has_more, other_path,
1207
other_entries = next_other_entries
1209
# We've matched all the walkers, yield this datapoint
1210
yield path, file_id, master_ie, other_values
1211
self._other_walkers = other_walkers
1212
self._other_entries = other_entries
1213
self._others_extra = others_extra
1215
def _finish_others(self):
1216
"""Finish walking the other iterators, so we get all entries."""
1217
for idx, info in enumerate(self._other_entries):
1218
other_extra = self._others_extra[idx]
1219
(other_has_more, other_path, other_ie) = info
1220
while other_has_more:
1221
other_file_id = other_ie.file_id
1222
if other_file_id not in self._out_of_order_processed:
1223
other_extra[other_file_id] = (other_path, other_ie)
1224
other_has_more, other_path, other_ie = \
1225
self._step_one(self._other_walkers[idx])
1226
del self._other_entries
1228
def _walk_others(self):
1229
"""Finish up by walking all the 'deferred' nodes."""
1230
# TODO: One alternative would be to grab all possible unprocessed
1231
# file_ids, and then sort by path, and then yield them. That
1232
# might ensure better ordering, in case a caller strictly
1233
# requires parents before children.
1234
for idx, other_extra in enumerate(self._others_extra):
1235
others = sorted(other_extra.itervalues(),
1236
key=lambda x: self._path_to_key(x[0]))
1237
for other_path, other_ie in others:
1238
file_id = other_ie.file_id
1239
# We don't need to check out_of_order_processed here, because
1240
# the lookup_by_file_id will be removing anything processed
1241
# from the extras cache
1242
other_extra.pop(file_id)
1243
other_values = [(None, None) for i in xrange(idx)]
1244
other_values.append((other_path, other_ie))
1245
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1246
alt_idx = alt_idx + idx + 1
1247
alt_extra = self._others_extra[alt_idx]
1248
alt_tree = self._other_trees[alt_idx]
1249
other_values.append(self._lookup_by_file_id(
1250
alt_extra, alt_tree, file_id))
1251
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
