1
# Copyright (C) 2005, 2009 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, NoSuchId
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 iter_files_bytes(self, desired_files):
315
"""Iterate through file contents.
317
Files will not necessarily be returned in the order they occur in
318
desired_files. No specific order is guaranteed.
320
Yields pairs of identifier, bytes_iterator. identifier is an opaque
321
value supplied by the caller as part of desired_files. It should
322
uniquely identify the file version in the caller's context. (Examples:
323
an index number or a TreeTransform trans_id.)
325
bytes_iterator is an iterable of bytestrings for the file. The
326
kind of iterable and length of the bytestrings are unspecified, but for
327
this implementation, it is a tuple containing a single bytestring with
328
the complete text of the file.
330
:param desired_files: a list of (file_id, identifier) pairs
332
for file_id, identifier in desired_files:
333
# We wrap the string in a tuple so that we can return an iterable
334
# of bytestrings. (Technically, a bytestring is also an iterable
335
# of bytestrings, but iterating through each character is not
337
cur_file = (self.get_file_text(file_id),)
338
yield identifier, cur_file
340
def get_symlink_target(self, file_id):
341
"""Get the target for a given file_id.
343
It is assumed that the caller already knows that file_id is referencing
345
:param file_id: Handle for the symlink entry.
346
:return: The path the symlink points to.
348
raise NotImplementedError(self.get_symlink_target)
350
def get_canonical_inventory_paths(self, paths):
351
"""Like get_canonical_inventory_path() but works on multiple items.
353
:param paths: A sequence of paths relative to the root of the tree.
354
:return: A list of paths, with each item the corresponding input path
355
adjusted to account for existing elements that match case
358
return list(self._yield_canonical_inventory_paths(paths))
360
def get_canonical_inventory_path(self, path):
361
"""Returns the first inventory item that case-insensitively matches path.
363
If a path matches exactly, it is returned. If no path matches exactly
364
but more than one path matches case-insensitively, it is implementation
365
defined which is returned.
367
If no path matches case-insensitively, the input path is returned, but
368
with as many path entries that do exist changed to their canonical
371
If you need to resolve many names from the same tree, you should
372
use get_canonical_inventory_paths() to avoid O(N) behaviour.
374
:param path: A paths relative to the root of the tree.
375
:return: The input path adjusted to account for existing elements
376
that match case insensitively.
378
return self._yield_canonical_inventory_paths([path]).next()
380
def _yield_canonical_inventory_paths(self, paths):
382
# First, if the path as specified exists exactly, just use it.
383
if self.path2id(path) is not None:
387
cur_id = self.get_root_id()
389
bit_iter = iter(path.split("/"))
392
for child in self.iter_children(cur_id):
394
child_base = os.path.basename(self.id2path(child))
395
if child_base.lower() == lelt:
397
cur_path = osutils.pathjoin(cur_path, child_base)
400
# before a change is committed we can see this error...
403
# got to the end of this directory and no entries matched.
404
# Return what matched so far, plus the rest as specified.
405
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
410
def get_root_id(self):
411
"""Return the file_id for the root of this tree."""
412
raise NotImplementedError(self.get_root_id)
414
def annotate_iter(self, file_id,
415
default_revision=_mod_revision.CURRENT_REVISION):
416
"""Return an iterator of revision_id, line tuples.
418
For working trees (and mutable trees in general), the special
419
revision_id 'current:' will be used for lines that are new in this
420
tree, e.g. uncommitted changes.
421
:param file_id: The file to produce an annotated version from
422
:param default_revision: For lines that don't match a basis, mark them
423
with this revision id. Not all implementations will make use of
426
raise NotImplementedError(self.annotate_iter)
428
def _get_plan_merge_data(self, file_id, other, base):
429
from bzrlib import merge, versionedfile
430
vf = versionedfile._PlanMergeVersionedFile(file_id)
431
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
432
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
434
last_revision_base = None
436
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
437
return vf, last_revision_a, last_revision_b, last_revision_base
439
def plan_file_merge(self, file_id, other, base=None):
440
"""Generate a merge plan based on annotations.
442
If the file contains uncommitted changes in this tree, they will be
443
attributed to the 'current:' pseudo-revision. If the file contains
444
uncommitted changes in the other tree, they will be assigned to the
445
'other:' pseudo-revision.
447
data = self._get_plan_merge_data(file_id, other, base)
448
vf, last_revision_a, last_revision_b, last_revision_base = data
449
return vf.plan_merge(last_revision_a, last_revision_b,
452
def plan_file_lca_merge(self, file_id, other, base=None):
453
"""Generate a merge plan based lca-newness.
455
If the file contains uncommitted changes in this tree, they will be
456
attributed to the 'current:' pseudo-revision. If the file contains
457
uncommitted changes in the other tree, they will be assigned to the
458
'other:' pseudo-revision.
460
data = self._get_plan_merge_data(file_id, other, base)
461
vf, last_revision_a, last_revision_b, last_revision_base = data
462
return vf.plan_lca_merge(last_revision_a, last_revision_b,
465
def _iter_parent_trees(self):
466
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
467
for revision_id in self.get_parent_ids():
469
yield self.revision_tree(revision_id)
470
except errors.NoSuchRevisionInTree:
471
yield self.repository.revision_tree(revision_id)
474
def _file_revision(revision_tree, file_id):
475
"""Determine the revision associated with a file in a given tree."""
476
revision_tree.lock_read()
478
return revision_tree.inventory[file_id].revision
480
revision_tree.unlock()
482
def _get_file_revision(self, file_id, vf, tree_revision):
483
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
485
if getattr(self, '_repository', None) is None:
486
last_revision = tree_revision
487
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
488
self._iter_parent_trees()]
489
vf.add_lines((file_id, last_revision), parent_keys,
490
self.get_file(file_id).readlines())
491
repo = self.branch.repository
494
last_revision = self._file_revision(self, file_id)
495
base_vf = self._repository.texts
496
if base_vf not in vf.fallback_versionedfiles:
497
vf.fallback_versionedfiles.append(base_vf)
500
inventory = property(_get_inventory,
501
doc="Inventory of this Tree")
503
def _check_retrieved(self, ie, f):
506
fp = fingerprint_file(f)
509
if ie.text_size is not None:
510
if ie.text_size != fp['size']:
511
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
512
["inventory expects %d bytes" % ie.text_size,
513
"file is actually %d bytes" % fp['size'],
514
"store is probably damaged/corrupt"])
516
if ie.text_sha1 != fp['sha1']:
517
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
518
["inventory expects %s" % ie.text_sha1,
519
"file is actually %s" % fp['sha1'],
520
"store is probably damaged/corrupt"])
523
def path2id(self, path):
524
"""Return the id for path in this tree."""
525
return self._inventory.path2id(path)
527
def paths2ids(self, paths, trees=[], require_versioned=True):
528
"""Return all the ids that can be reached by walking from paths.
530
Each path is looked up in this tree and any extras provided in
531
trees, and this is repeated recursively: the children in an extra tree
532
of a directory that has been renamed under a provided path in this tree
533
are all returned, even if none exist under a provided path in this
534
tree, and vice versa.
536
:param paths: An iterable of paths to start converting to ids from.
537
Alternatively, if paths is None, no ids should be calculated and None
538
will be returned. This is offered to make calling the api unconditional
539
for code that *might* take a list of files.
540
:param trees: Additional trees to consider.
541
:param require_versioned: If False, do not raise NotVersionedError if
542
an element of paths is not versioned in this tree and all of trees.
544
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
546
def iter_children(self, file_id):
547
entry = self.iter_entries_by_dir([file_id]).next()[1]
548
for child in getattr(entry, 'children', {}).itervalues():
551
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
552
def print_file(self, file_id):
553
"""Print file with id `file_id` to stdout."""
555
sys.stdout.write(self.get_file_text(file_id))
560
def revision_tree(self, revision_id):
561
"""Obtain a revision tree for the revision revision_id.
563
The intention of this method is to allow access to possibly cached
564
tree data. Implementors of this method should raise NoSuchRevision if
565
the tree is not locally available, even if they could obtain the
566
tree via a repository or some other means. Callers are responsible
567
for finding the ultimate source for a revision tree.
569
:param revision_id: The revision_id of the requested tree.
571
:raises: NoSuchRevision if the tree cannot be obtained.
573
raise errors.NoSuchRevisionInTree(self, revision_id)
576
"""What files are present in this tree and unknown.
578
:return: an iterator over the unknown files.
585
def filter_unversioned_files(self, paths):
586
"""Filter out paths that are versioned.
588
:return: set of paths.
590
# NB: we specifically *don't* call self.has_filename, because for
591
# WorkingTrees that can indicate files that exist on disk but that
593
pred = self.inventory.has_filename
594
return set((p for p in paths if not pred(p)))
596
def walkdirs(self, prefix=""):
597
"""Walk the contents of this tree from path down.
599
This yields all the data about the contents of a directory at a time.
600
After each directory has been yielded, if the caller has mutated the
601
list to exclude some directories, they are then not descended into.
603
The data yielded is of the form:
604
((directory-relpath, directory-path-from-root, directory-fileid),
605
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
606
versioned_kind), ...]),
607
- directory-relpath is the containing dirs relpath from prefix
608
- directory-path-from-root is the containing dirs path from /
609
- directory-fileid is the id of the directory if it is versioned.
610
- relpath is the relative path within the subtree being walked.
611
- basename is the basename
612
- kind is the kind of the file now. If unknonwn then the file is not
613
present within the tree - but it may be recorded as versioned. See
615
- lstat is the stat data *if* the file was statted.
616
- path_from_tree_root is the path from the root of the tree.
617
- file_id is the file_id if the entry is versioned.
618
- versioned_kind is the kind of the file as last recorded in the
619
versioning system. If 'unknown' the file is not versioned.
620
One of 'kind' and 'versioned_kind' must not be 'unknown'.
622
:param prefix: Start walking from prefix within the tree rather than
623
at the root. This allows one to walk a subtree but get paths that are
624
relative to a tree rooted higher up.
625
:return: an iterator over the directory data.
627
raise NotImplementedError(self.walkdirs)
629
def supports_content_filtering(self):
632
def _content_filter_stack(self, path=None, file_id=None):
633
"""The stack of content filters for a path if filtering is supported.
635
Readers will be applied in first-to-last order.
636
Writers will be applied in last-to-first order.
637
Either the path or the file-id needs to be provided.
639
:param path: path relative to the root of the tree
641
:param file_id: file_id or None if unknown
642
:return: the list of filters - [] if there are none
644
filter_pref_names = filters._get_registered_names()
645
if len(filter_pref_names) == 0:
648
path = self.id2path(file_id)
649
prefs = self.iter_search_rules([path], filter_pref_names).next()
650
stk = filters._get_filter_stack_for(prefs)
651
if 'filters' in debug.debug_flags:
652
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
655
def _content_filter_stack_provider(self):
656
"""A function that returns a stack of ContentFilters.
658
The function takes a path (relative to the top of the tree) and a
659
file-id as parameters.
661
:return: None if content filtering is not supported by this tree.
663
if self.supports_content_filtering():
664
return lambda path, file_id: \
665
self._content_filter_stack(path, file_id)
669
def iter_search_rules(self, path_names, pref_names=None,
670
_default_searcher=rules._per_user_searcher):
671
"""Find the preferences for filenames in a tree.
673
:param path_names: an iterable of paths to find attributes for.
674
Paths are given relative to the root of the tree.
675
:param pref_names: the list of preferences to lookup - None for all
676
:param _default_searcher: private parameter to assist testing - don't use
677
:return: an iterator of tuple sequences, one per path-name.
678
See _RulesSearcher.get_items for details on the tuple sequence.
680
searcher = self._get_rules_searcher(_default_searcher)
681
if searcher is not None:
682
if pref_names is not None:
683
for path in path_names:
684
yield searcher.get_selected_items(path, pref_names)
686
for path in path_names:
687
yield searcher.get_items(path)
690
def _get_rules_searcher(self, default_searcher):
691
"""Get the RulesSearcher for this tree given the default one."""
692
searcher = default_searcher
696
######################################################################
699
# TODO: Merge these two functions into a single one that can operate
700
# on either a whole tree or a set of files.
702
# TODO: Return the diff in order by filename, not by category or in
703
# random order. Can probably be done by lock-stepping through the
704
# filenames from both trees.
707
def file_status(filename, old_tree, new_tree):
708
"""Return single-letter status, old and new names for a file.
710
The complexity here is in deciding how to represent renames;
711
many complex cases are possible.
713
old_inv = old_tree.inventory
714
new_inv = new_tree.inventory
715
new_id = new_inv.path2id(filename)
716
old_id = old_inv.path2id(filename)
718
if not new_id and not old_id:
719
# easy: doesn't exist in either; not versioned at all
720
if new_tree.is_ignored(filename):
721
return 'I', None, None
723
return '?', None, None
725
# There is now a file of this name, great.
728
# There is no longer a file of this name, but we can describe
729
# what happened to the file that used to have
730
# this name. There are two possibilities: either it was
731
# deleted entirely, or renamed.
732
if new_inv.has_id(old_id):
733
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
735
return 'D', old_inv.id2path(old_id), None
737
# if the file_id is new in this revision, it is added
738
if new_id and not old_inv.has_id(new_id):
741
# if there used to be a file of this name, but that ID has now
742
# disappeared, it is deleted
743
if old_id and not new_inv.has_id(old_id):
749
@deprecated_function(deprecated_in((1, 9, 0)))
750
def find_renames(old_inv, new_inv):
751
for file_id in old_inv:
752
if file_id not in new_inv:
754
old_name = old_inv.id2path(file_id)
755
new_name = new_inv.id2path(file_id)
756
if old_name != new_name:
757
yield (old_name, new_name)
760
def find_ids_across_trees(filenames, trees, require_versioned=True):
761
"""Find the ids corresponding to specified filenames.
763
All matches in all trees will be used, and all children of matched
764
directories will be used.
766
:param filenames: The filenames to find file_ids for (if None, returns
768
:param trees: The trees to find file_ids within
769
:param require_versioned: if true, all specified filenames must occur in
771
:return: a set of file ids for the specified filenames and their children.
775
specified_path_ids = _find_ids_across_trees(filenames, trees,
777
return _find_children_across_trees(specified_path_ids, trees)
780
def _find_ids_across_trees(filenames, trees, require_versioned):
781
"""Find the ids corresponding to specified filenames.
783
All matches in all trees will be used, but subdirectories are not scanned.
785
:param filenames: The filenames to find file_ids for
786
:param trees: The trees to find file_ids within
787
:param require_versioned: if true, all specified filenames must occur in
789
:return: a set of file ids for the specified filenames
792
interesting_ids = set()
793
for tree_path in filenames:
796
file_id = tree.path2id(tree_path)
797
if file_id is not None:
798
interesting_ids.add(file_id)
801
not_versioned.append(tree_path)
802
if len(not_versioned) > 0 and require_versioned:
803
raise errors.PathsNotVersionedError(not_versioned)
804
return interesting_ids
807
def _find_children_across_trees(specified_ids, trees):
808
"""Return a set including specified ids and their children.
810
All matches in all trees will be used.
812
:param trees: The trees to find file_ids within
813
:return: a set containing all specified ids and their children
815
interesting_ids = set(specified_ids)
816
pending = interesting_ids
817
# now handle children of interesting ids
818
# we loop so that we handle all children of each id in both trees
819
while len(pending) > 0:
821
for file_id in pending:
823
if not tree.has_id(file_id):
825
for child_id in tree.iter_children(file_id):
826
if child_id not in interesting_ids:
827
new_pending.add(child_id)
828
interesting_ids.update(new_pending)
829
pending = new_pending
830
return interesting_ids
833
class InterTree(InterObject):
834
"""This class represents operations taking place between two Trees.
836
Its instances have methods like 'compare' and contain references to the
837
source and target trees these operations are to be carried out on.
839
Clients of bzrlib should not need to use InterTree directly, rather they
840
should use the convenience methods on Tree such as 'Tree.compare()' which
841
will pass through to InterTree as appropriate.
847
def compare(self, want_unchanged=False, specific_files=None,
848
extra_trees=None, require_versioned=False, include_root=False,
849
want_unversioned=False):
850
"""Return the changes from source to target.
852
:return: A TreeDelta.
853
:param specific_files: An optional list of file paths to restrict the
854
comparison to. When mapping filenames to ids, all matches in all
855
trees (including optional extra_trees) are used, and all children of
856
matched directories are included.
857
:param want_unchanged: An optional boolean requesting the inclusion of
858
unchanged entries in the result.
859
:param extra_trees: An optional list of additional trees to use when
860
mapping the contents of specific_files (paths) to file_ids.
861
:param require_versioned: An optional boolean (defaults to False). When
862
supplied and True all the 'specific_files' must be versioned, or
863
a PathsNotVersionedError will be thrown.
864
:param want_unversioned: Scan for unversioned paths.
866
# NB: show_status depends on being able to pass in non-versioned files
867
# and report them as unknown
868
trees = (self.source,)
869
if extra_trees is not None:
870
trees = trees + tuple(extra_trees)
871
# target is usually the newer tree:
872
specific_file_ids = self.target.paths2ids(specific_files, trees,
873
require_versioned=require_versioned)
874
if specific_files and not specific_file_ids:
875
# All files are unversioned, so just return an empty delta
876
# _compare_trees would think we want a complete delta
877
result = delta.TreeDelta()
878
fake_entry = InventoryFile('unused', 'unused', 'unused')
879
result.unversioned = [(path, None,
880
self.target._comparison_data(fake_entry, path)[0]) for path in
883
return delta._compare_trees(self.source, self.target, want_unchanged,
884
specific_files, include_root, extra_trees=extra_trees,
885
require_versioned=require_versioned,
886
want_unversioned=want_unversioned)
888
def iter_changes(self, include_unchanged=False,
889
specific_files=None, pb=None, extra_trees=[],
890
require_versioned=True, want_unversioned=False):
891
"""Generate an iterator of changes between trees.
894
(file_id, (path_in_source, path_in_target),
895
changed_content, versioned, parent, name, kind,
898
Changed_content is True if the file's content has changed. This
899
includes changes to its kind, and to a symlink's target.
901
versioned, parent, name, kind, executable are tuples of (from, to).
902
If a file is missing in a tree, its kind is None.
904
Iteration is done in parent-to-child order, relative to the target
907
There is no guarantee that all paths are in sorted order: the
908
requirement to expand the search due to renames may result in children
909
that should be found early being found late in the search, after
910
lexically later results have been returned.
911
:param require_versioned: Raise errors.PathsNotVersionedError if a
912
path in the specific_files list is not versioned in one of
913
source, target or extra_trees.
914
:param want_unversioned: Should unversioned files be returned in the
915
output. An unversioned file is defined as one with (False, False)
916
for the versioned pair.
919
lookup_trees = [self.source]
921
lookup_trees.extend(extra_trees)
922
if specific_files == []:
923
specific_file_ids = []
925
specific_file_ids = self.target.paths2ids(specific_files,
926
lookup_trees, require_versioned=require_versioned)
928
all_unversioned = sorted([(p.split('/'), p) for p in
930
if specific_files is None or
931
osutils.is_inside_any(specific_files, p)])
932
all_unversioned = deque(all_unversioned)
934
all_unversioned = deque()
936
from_entries_by_dir = list(self.source.iter_entries_by_dir(
937
specific_file_ids=specific_file_ids))
938
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
939
to_entries_by_dir = list(self.target.iter_entries_by_dir(
940
specific_file_ids=specific_file_ids))
941
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
943
# the unversioned path lookup only occurs on real trees - where there
944
# can be extras. So the fake_entry is solely used to look up
945
# executable it values when execute is not supported.
946
fake_entry = InventoryFile('unused', 'unused', 'unused')
947
for to_path, to_entry in to_entries_by_dir:
948
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
949
unversioned_path = all_unversioned.popleft()
950
to_kind, to_executable, to_stat = \
951
self.target._comparison_data(fake_entry, unversioned_path[1])
952
yield (None, (None, unversioned_path[1]), True, (False, False),
954
(None, unversioned_path[0][-1]),
956
(None, to_executable))
957
file_id = to_entry.file_id
958
to_paths[file_id] = to_path
960
changed_content = False
961
from_path, from_entry = from_data.get(file_id, (None, None))
962
from_versioned = (from_entry is not None)
963
if from_entry is not None:
964
from_versioned = True
965
from_name = from_entry.name
966
from_parent = from_entry.parent_id
967
from_kind, from_executable, from_stat = \
968
self.source._comparison_data(from_entry, from_path)
971
from_versioned = False
975
from_executable = None
976
versioned = (from_versioned, True)
977
to_kind, to_executable, to_stat = \
978
self.target._comparison_data(to_entry, to_path)
979
kind = (from_kind, to_kind)
980
if kind[0] != kind[1]:
981
changed_content = True
982
elif from_kind == 'file':
983
if (self.source.get_file_sha1(file_id, from_path, from_stat) !=
984
self.target.get_file_sha1(file_id, to_path, to_stat)):
985
changed_content = True
986
elif from_kind == 'symlink':
987
if (self.source.get_symlink_target(file_id) !=
988
self.target.get_symlink_target(file_id)):
989
changed_content = True
990
elif from_kind == 'tree-reference':
991
if (self.source.get_reference_revision(file_id, from_path)
992
!= self.target.get_reference_revision(file_id, to_path)):
993
changed_content = True
994
parent = (from_parent, to_entry.parent_id)
995
name = (from_name, to_entry.name)
996
executable = (from_executable, to_executable)
998
pb.update('comparing files', entry_count, num_entries)
999
if (changed_content is not False or versioned[0] != versioned[1]
1000
or parent[0] != parent[1] or name[0] != name[1] or
1001
executable[0] != executable[1] or include_unchanged):
1002
yield (file_id, (from_path, to_path), changed_content,
1003
versioned, parent, name, kind, executable)
1005
while all_unversioned:
1006
# yield any trailing unversioned paths
1007
unversioned_path = all_unversioned.popleft()
1008
to_kind, to_executable, to_stat = \
1009
self.target._comparison_data(fake_entry, unversioned_path[1])
1010
yield (None, (None, unversioned_path[1]), True, (False, False),
1012
(None, unversioned_path[0][-1]),
1014
(None, to_executable))
1016
def get_to_path(to_entry):
1017
if to_entry.parent_id is None:
1018
to_path = '' # the root
1020
if to_entry.parent_id not in to_paths:
1022
return get_to_path(self.target.inventory[to_entry.parent_id])
1023
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
1025
to_paths[to_entry.file_id] = to_path
1028
for path, from_entry in from_entries_by_dir:
1029
file_id = from_entry.file_id
1030
if file_id in to_paths:
1033
if not file_id in self.target.all_file_ids():
1034
# common case - paths we have not emitted are not present in
1038
to_path = get_to_path(self.target.inventory[file_id])
1041
pb.update('comparing files', entry_count, num_entries)
1042
versioned = (True, False)
1043
parent = (from_entry.parent_id, None)
1044
name = (from_entry.name, None)
1045
from_kind, from_executable, stat_value = \
1046
self.source._comparison_data(from_entry, path)
1047
kind = (from_kind, None)
1048
executable = (from_executable, None)
1049
changed_content = from_kind is not None
1050
# the parent's path is necessarily known at this point.
1051
yield(file_id, (path, to_path), changed_content, versioned, parent,
1052
name, kind, executable)
1055
class MultiWalker(object):
1056
"""Walk multiple trees simultaneously, getting combined results."""
1058
# Note: This could be written to not assume you can do out-of-order
1059
# lookups. Instead any nodes that don't match in all trees could be
1060
# marked as 'deferred', and then returned in the final cleanup loop.
1061
# For now, I think it is "nicer" to return things as close to the
1062
# "master_tree" order as we can.
1064
def __init__(self, master_tree, other_trees):
1065
"""Create a new MultiWalker.
1067
All trees being walked must implement "iter_entries_by_dir()", such
1068
that they yield (path, object) tuples, where that object will have a
1069
'.file_id' member, that can be used to check equality.
1071
:param master_tree: All trees will be 'slaved' to the master_tree such
1072
that nodes in master_tree will be used as 'first-pass' sync points.
1073
Any nodes that aren't in master_tree will be merged in a second
1075
:param other_trees: A list of other trees to walk simultaneously.
1077
self._master_tree = master_tree
1078
self._other_trees = other_trees
1080
# Keep track of any nodes that were properly processed just out of
1081
# order, that way we don't return them at the end, we don't have to
1082
# track *all* processed file_ids, just the out-of-order ones
1083
self._out_of_order_processed = set()
1086
def _step_one(iterator):
1087
"""Step an iter_entries_by_dir iterator.
1089
:return: (has_more, path, ie)
1090
If has_more is False, path and ie will be None.
1093
path, ie = iterator.next()
1094
except StopIteration:
1095
return False, None, None
1097
return True, path, ie
1100
def _cmp_path_by_dirblock(path1, path2):
1101
"""Compare two paths based on what directory they are in.
1103
This generates a sort order, such that all children of a directory are
1104
sorted together, and grandchildren are in the same order as the
1105
children appear. But all grandchildren come after all children.
1107
:param path1: first path
1108
:param path2: the second path
1109
:return: negative number if ``path1`` comes first,
1110
0 if paths are equal
1111
and a positive number if ``path2`` sorts first
1113
# Shortcut this special case
1116
# This is stolen from _dirstate_helpers_py.py, only switching it to
1117
# Unicode objects. Consider using encode_utf8() and then using the
1118
# optimized versions, or maybe writing optimized unicode versions.
1119
if not isinstance(path1, unicode):
1120
raise TypeError("'path1' must be a unicode string, not %s: %r"
1121
% (type(path1), path1))
1122
if not isinstance(path2, unicode):
1123
raise TypeError("'path2' must be a unicode string, not %s: %r"
1124
% (type(path2), path2))
1125
return cmp(MultiWalker._path_to_key(path1),
1126
MultiWalker._path_to_key(path2))
1129
def _path_to_key(path):
1130
dirname, basename = osutils.split(path)
1131
return (dirname.split(u'/'), basename)
1133
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1134
"""Lookup an inventory entry by file_id.
1136
This is called when an entry is missing in the normal order.
1137
Generally this is because a file was either renamed, or it was
1138
deleted/added. If the entry was found in the inventory and not in
1139
extra_entries, it will be added to self._out_of_order_processed
1141
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1142
should be filled with entries that were found before they were
1143
used. If file_id is present, it will be removed from the
1145
:param other_tree: The Tree to search, in case we didn't find the entry
1147
:param file_id: The file_id to look for
1148
:return: (path, ie) if found or (None, None) if not present.
1150
if file_id in extra_entries:
1151
return extra_entries.pop(file_id)
1152
# TODO: Is id2path better as the first call, or is
1153
# inventory[file_id] better as a first check?
1155
cur_path = other_tree.id2path(file_id)
1156
except errors.NoSuchId:
1158
if cur_path is None:
1161
self._out_of_order_processed.add(file_id)
1162
cur_ie = other_tree.inventory[file_id]
1163
return (cur_path, cur_ie)
1166
"""Match up the values in the different trees."""
1167
for result in self._walk_master_tree():
1169
self._finish_others()
1170
for result in self._walk_others():
1173
def _walk_master_tree(self):
1174
"""First pass, walk all trees in lock-step.
1176
When we are done, all nodes in the master_tree will have been
1177
processed. _other_walkers, _other_entries, and _others_extra will be
1178
set on 'self' for future processing.
1180
# This iterator has the most "inlining" done, because it tends to touch
1181
# every file in the tree, while the others only hit nodes that don't
1183
master_iterator = self._master_tree.iter_entries_by_dir()
1185
other_walkers = [other.iter_entries_by_dir()
1186
for other in self._other_trees]
1187
other_entries = [self._step_one(walker) for walker in other_walkers]
1188
# Track extra nodes in the other trees
1189
others_extra = [{} for i in xrange(len(self._other_trees))]
1191
master_has_more = True
1192
step_one = self._step_one
1193
lookup_by_file_id = self._lookup_by_file_id
1194
out_of_order_processed = self._out_of_order_processed
1196
while master_has_more:
1197
(master_has_more, path, master_ie) = step_one(master_iterator)
1198
if not master_has_more:
1201
file_id = master_ie.file_id
1203
other_values_append = other_values.append
1204
next_other_entries = []
1205
next_other_entries_append = next_other_entries.append
1206
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1207
if not other_has_more:
1208
other_values_append(lookup_by_file_id(
1209
others_extra[idx], self._other_trees[idx], file_id))
1210
next_other_entries_append((False, None, None))
1211
elif file_id == other_ie.file_id:
1212
# This is the critical code path, as most of the entries
1213
# should match between most trees.
1214
other_values_append((other_path, other_ie))
1215
next_other_entries_append(step_one(other_walkers[idx]))
1217
# This walker did not match, step it until it either
1218
# matches, or we know we are past the current walker.
1219
other_walker = other_walkers[idx]
1220
other_extra = others_extra[idx]
1221
while (other_has_more and
1222
self._cmp_path_by_dirblock(other_path, path) < 0):
1223
other_file_id = other_ie.file_id
1224
if other_file_id not in out_of_order_processed:
1225
other_extra[other_file_id] = (other_path, other_ie)
1226
other_has_more, other_path, other_ie = \
1227
step_one(other_walker)
1228
if other_has_more and other_ie.file_id == file_id:
1229
# We ended up walking to this point, match and step
1231
other_values_append((other_path, other_ie))
1232
other_has_more, other_path, other_ie = \
1233
step_one(other_walker)
1235
# This record isn't in the normal order, see if it
1237
other_values_append(lookup_by_file_id(
1238
other_extra, self._other_trees[idx], file_id))
1239
next_other_entries_append((other_has_more, other_path,
1241
other_entries = next_other_entries
1243
# We've matched all the walkers, yield this datapoint
1244
yield path, file_id, master_ie, other_values
1245
self._other_walkers = other_walkers
1246
self._other_entries = other_entries
1247
self._others_extra = others_extra
1249
def _finish_others(self):
1250
"""Finish walking the other iterators, so we get all entries."""
1251
for idx, info in enumerate(self._other_entries):
1252
other_extra = self._others_extra[idx]
1253
(other_has_more, other_path, other_ie) = info
1254
while other_has_more:
1255
other_file_id = other_ie.file_id
1256
if other_file_id not in self._out_of_order_processed:
1257
other_extra[other_file_id] = (other_path, other_ie)
1258
other_has_more, other_path, other_ie = \
1259
self._step_one(self._other_walkers[idx])
1260
del self._other_entries
1262
def _walk_others(self):
1263
"""Finish up by walking all the 'deferred' nodes."""
1264
# TODO: One alternative would be to grab all possible unprocessed
1265
# file_ids, and then sort by path, and then yield them. That
1266
# might ensure better ordering, in case a caller strictly
1267
# requires parents before children.
1268
for idx, other_extra in enumerate(self._others_extra):
1269
others = sorted(other_extra.itervalues(),
1270
key=lambda x: self._path_to_key(x[0]))
1271
for other_path, other_ie in others:
1272
file_id = other_ie.file_id
1273
# We don't need to check out_of_order_processed here, because
1274
# the lookup_by_file_id will be removing anything processed
1275
# from the extras cache
1276
other_extra.pop(file_id)
1277
other_values = [(None, None) for i in xrange(idx)]
1278
other_values.append((other_path, other_ie))
1279
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1280
alt_idx = alt_idx + idx + 1
1281
alt_extra = self._others_extra[alt_idx]
1282
alt_tree = self._other_trees[alt_idx]
1283
other_values.append(self._lookup_by_file_id(
1284
alt_extra, alt_tree, file_id))
1285
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
