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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""Tree classes, representing directory at point in time.
21
from collections import deque
25
conflicts as _mod_conflicts,
30
revision as _mod_revision,
33
from bzrlib.decorators import needs_read_lock
34
from bzrlib.errors import BzrError, NoSuchId
35
from bzrlib import errors
36
from bzrlib.inventory import 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 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
def iter_changes(self, from_tree, include_unchanged=False,
99
specific_files=None, pb=None, extra_trees=None,
100
require_versioned=True, want_unversioned=False):
101
intertree = InterTree.get(from_tree, self)
102
return intertree.iter_changes(include_unchanged, specific_files, pb,
103
extra_trees, require_versioned, want_unversioned=want_unversioned)
106
"""Get a list of the conflicts in the tree.
108
Each conflict is an instance of bzrlib.conflicts.Conflict.
110
return _mod_conflicts.ConflictList()
113
"""For trees that can have unversioned files, return all such paths."""
116
def get_parent_ids(self):
117
"""Get the parent ids for this tree.
119
:return: a list of parent ids. [] is returned to indicate
120
a tree with no parents.
121
:raises: BzrError if the parents are not known.
123
raise NotImplementedError(self.get_parent_ids)
125
def has_filename(self, filename):
126
"""True if the tree has given filename."""
127
raise NotImplementedError(self.has_filename)
129
def has_id(self, file_id):
130
return self.inventory.has_id(file_id)
132
def __contains__(self, file_id):
133
return self.has_id(file_id)
135
def has_or_had_id(self, file_id):
136
return self.inventory.has_id(file_id)
138
def is_ignored(self, filename):
139
"""Check whether the filename is ignored by this tree.
141
:param filename: The relative filename within the tree.
142
:return: True if the filename is ignored.
147
return iter(self.inventory)
149
def all_file_ids(self):
150
"""Iterate through all file ids, including ids for missing files."""
151
return set(self.inventory)
153
def id2path(self, file_id):
154
"""Return the path for a file id.
158
return self.inventory.id2path(file_id)
160
def is_control_filename(self, filename):
161
"""True if filename is the name of a control file in this tree.
163
:param filename: A filename within the tree. This is a relative path
164
from the root of this tree.
166
This is true IF and ONLY IF the filename is part of the meta data
167
that bzr controls in this tree. I.E. a random .bzr directory placed
168
on disk will not be a control file for this tree.
170
return self.bzrdir.is_control_filename(filename)
173
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
174
"""Walk the tree in 'by_dir' order.
176
This will yield each entry in the tree as a (path, entry) tuple.
177
The order that they are yielded is:
179
Directories are walked in a depth-first lexicographical order,
180
however, whenever a directory is reached, all of its direct child
181
nodes are yielded in lexicographical order before yielding the
184
For example, in the tree::
194
The yield order (ignoring root) would be::
195
a, f, a/b, a/d, a/b/c, a/d/e, f/g
197
:param yield_parents: If True, yield the parents from the root leading
198
down to specific_file_ids that have been requested. This has no
199
impact if specific_file_ids is None.
201
return self.inventory.iter_entries_by_dir(
202
specific_file_ids=specific_file_ids, yield_parents=yield_parents)
204
def iter_references(self):
205
if self.supports_tree_reference():
206
for path, entry in self.iter_entries_by_dir():
207
if entry.kind == 'tree-reference':
208
yield path, entry.file_id
210
def kind(self, file_id):
211
raise NotImplementedError("Tree subclass %s must implement kind"
212
% self.__class__.__name__)
214
def stored_kind(self, file_id):
215
"""File kind stored for this file_id.
217
May not match kind on disk for working trees. Always available
218
for versioned files, even when the file itself is missing.
220
return self.kind(file_id)
222
def path_content_summary(self, path):
223
"""Get a summary of the information about path.
225
:param path: A relative path within the tree.
226
:return: A tuple containing kind, size, exec, sha1-or-link.
227
Kind is always present (see tree.kind()).
228
size is present if kind is file, None otherwise.
229
exec is None unless kind is file and the platform supports the 'x'
231
sha1-or-link is the link target if kind is symlink, or the sha1 if
232
it can be obtained without reading the file.
234
raise NotImplementedError(self.path_content_summary)
236
def get_reference_revision(self, file_id, path=None):
237
raise NotImplementedError("Tree subclass %s must implement "
238
"get_reference_revision"
239
% self.__class__.__name__)
241
def _comparison_data(self, entry, path):
242
"""Return a tuple of kind, executable, stat_value for a file.
244
entry may be None if there is no inventory entry for the file, but
245
path must always be supplied.
247
kind is None if there is no file present (even if an inventory id is
248
present). executable is False for non-file entries.
250
raise NotImplementedError(self._comparison_data)
252
def _file_size(self, entry, stat_value):
253
raise NotImplementedError(self._file_size)
255
def _get_inventory(self):
256
return self._inventory
258
def get_file(self, file_id, path=None):
259
"""Return a file object for the file file_id in the tree.
261
If both file_id and path are defined, it is implementation defined as
262
to which one is used.
264
raise NotImplementedError(self.get_file)
266
def get_file_with_stat(self, file_id, path=None):
267
"""Get a file handle and stat object for file_id.
269
The default implementation returns (self.get_file, None) for backwards
272
:param file_id: The file id to read.
273
:param path: The path of the file, if it is known.
274
:return: A tuple (file_handle, stat_value_or_None). If the tree has
275
no stat facility, or need for a stat cache feedback during commit,
276
it may return None for the second element of the tuple.
278
return (self.get_file(file_id, path), None)
280
def get_file_text(self, file_id, path=None):
281
"""Return the byte content of a file.
283
:param file_id: The file_id of the file.
284
:param path: The path of the file.
285
If both file_id and path are supplied, an implementation may use
288
my_file = self.get_file(file_id, path)
290
return my_file.read()
294
def get_file_lines(self, file_id, path=None):
295
"""Return the content of a file, as lines.
297
:param file_id: The file_id of the file.
298
:param path: The path of the file.
299
If both file_id and path are supplied, an implementation may use
302
return osutils.split_lines(self.get_file_text(file_id, path))
304
def get_file_mtime(self, file_id, path=None):
305
"""Return the modification time for a file.
307
:param file_id: The handle for this file.
308
:param path: The path that this file can be found at.
309
These must point to the same object.
311
raise NotImplementedError(self.get_file_mtime)
313
def get_file_size(self, file_id):
314
"""Return the size of a file in bytes.
316
This applies only to regular files. If invoked on directories or
317
symlinks, it will return None.
318
:param file_id: The file-id of the file
320
raise NotImplementedError(self.get_file_size)
322
def get_file_by_path(self, path):
323
return self.get_file(self._inventory.path2id(path), path)
325
def iter_files_bytes(self, desired_files):
326
"""Iterate through file contents.
328
Files will not necessarily be returned in the order they occur in
329
desired_files. No specific order is guaranteed.
331
Yields pairs of identifier, bytes_iterator. identifier is an opaque
332
value supplied by the caller as part of desired_files. It should
333
uniquely identify the file version in the caller's context. (Examples:
334
an index number or a TreeTransform trans_id.)
336
bytes_iterator is an iterable of bytestrings for the file. The
337
kind of iterable and length of the bytestrings are unspecified, but for
338
this implementation, it is a tuple containing a single bytestring with
339
the complete text of the file.
341
:param desired_files: a list of (file_id, identifier) pairs
343
for file_id, identifier in desired_files:
344
# We wrap the string in a tuple so that we can return an iterable
345
# of bytestrings. (Technically, a bytestring is also an iterable
346
# of bytestrings, but iterating through each character is not
348
cur_file = (self.get_file_text(file_id),)
349
yield identifier, cur_file
351
def get_symlink_target(self, file_id):
352
"""Get the target for a given file_id.
354
It is assumed that the caller already knows that file_id is referencing
356
:param file_id: Handle for the symlink entry.
357
:return: The path the symlink points to.
359
raise NotImplementedError(self.get_symlink_target)
361
def get_canonical_inventory_paths(self, paths):
362
"""Like get_canonical_inventory_path() but works on multiple items.
364
:param paths: A sequence of paths relative to the root of the tree.
365
:return: A list of paths, with each item the corresponding input path
366
adjusted to account for existing elements that match case
369
return list(self._yield_canonical_inventory_paths(paths))
371
def get_canonical_inventory_path(self, path):
372
"""Returns the first inventory item that case-insensitively matches path.
374
If a path matches exactly, it is returned. If no path matches exactly
375
but more than one path matches case-insensitively, it is implementation
376
defined which is returned.
378
If no path matches case-insensitively, the input path is returned, but
379
with as many path entries that do exist changed to their canonical
382
If you need to resolve many names from the same tree, you should
383
use get_canonical_inventory_paths() to avoid O(N) behaviour.
385
:param path: A paths relative to the root of the tree.
386
:return: The input path adjusted to account for existing elements
387
that match case insensitively.
389
return self._yield_canonical_inventory_paths([path]).next()
391
def _yield_canonical_inventory_paths(self, paths):
393
# First, if the path as specified exists exactly, just use it.
394
if self.path2id(path) is not None:
398
cur_id = self.get_root_id()
400
bit_iter = iter(path.split("/"))
403
for child in self.iter_children(cur_id):
405
child_base = os.path.basename(self.id2path(child))
406
if child_base.lower() == lelt:
408
cur_path = osutils.pathjoin(cur_path, child_base)
411
# before a change is committed we can see this error...
414
# got to the end of this directory and no entries matched.
415
# Return what matched so far, plus the rest as specified.
416
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
421
def get_root_id(self):
422
"""Return the file_id for the root of this tree."""
423
raise NotImplementedError(self.get_root_id)
425
def annotate_iter(self, file_id,
426
default_revision=_mod_revision.CURRENT_REVISION):
427
"""Return an iterator of revision_id, line tuples.
429
For working trees (and mutable trees in general), the special
430
revision_id 'current:' will be used for lines that are new in this
431
tree, e.g. uncommitted changes.
432
:param file_id: The file to produce an annotated version from
433
:param default_revision: For lines that don't match a basis, mark them
434
with this revision id. Not all implementations will make use of
437
raise NotImplementedError(self.annotate_iter)
439
def _get_plan_merge_data(self, file_id, other, base):
440
from bzrlib import versionedfile
441
vf = versionedfile._PlanMergeVersionedFile(file_id)
442
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
443
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
445
last_revision_base = None
447
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
448
return vf, last_revision_a, last_revision_b, last_revision_base
450
def plan_file_merge(self, file_id, other, base=None):
451
"""Generate a merge plan based on annotations.
453
If the file contains uncommitted changes in this tree, they will be
454
attributed to the 'current:' pseudo-revision. If the file contains
455
uncommitted changes in the other tree, they will be assigned to the
456
'other:' pseudo-revision.
458
data = self._get_plan_merge_data(file_id, other, base)
459
vf, last_revision_a, last_revision_b, last_revision_base = data
460
return vf.plan_merge(last_revision_a, last_revision_b,
463
def plan_file_lca_merge(self, file_id, other, base=None):
464
"""Generate a merge plan based lca-newness.
466
If the file contains uncommitted changes in this tree, they will be
467
attributed to the 'current:' pseudo-revision. If the file contains
468
uncommitted changes in the other tree, they will be assigned to the
469
'other:' pseudo-revision.
471
data = self._get_plan_merge_data(file_id, other, base)
472
vf, last_revision_a, last_revision_b, last_revision_base = data
473
return vf.plan_lca_merge(last_revision_a, last_revision_b,
476
def _iter_parent_trees(self):
477
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
478
for revision_id in self.get_parent_ids():
480
yield self.revision_tree(revision_id)
481
except errors.NoSuchRevisionInTree:
482
yield self.repository.revision_tree(revision_id)
485
def _file_revision(revision_tree, file_id):
486
"""Determine the revision associated with a file in a given tree."""
487
revision_tree.lock_read()
489
return revision_tree.inventory[file_id].revision
491
revision_tree.unlock()
493
def _get_file_revision(self, file_id, vf, tree_revision):
494
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
496
if getattr(self, '_repository', None) is None:
497
last_revision = tree_revision
498
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
499
self._iter_parent_trees()]
500
vf.add_lines((file_id, last_revision), parent_keys,
501
self.get_file(file_id).readlines())
502
repo = self.branch.repository
505
last_revision = self._file_revision(self, file_id)
506
base_vf = self._repository.texts
507
if base_vf not in vf.fallback_versionedfiles:
508
vf.fallback_versionedfiles.append(base_vf)
511
inventory = property(_get_inventory,
512
doc="Inventory of this Tree")
514
def _check_retrieved(self, ie, f):
517
fp = fingerprint_file(f)
520
if ie.text_size is not None:
521
if ie.text_size != fp['size']:
522
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
523
["inventory expects %d bytes" % ie.text_size,
524
"file is actually %d bytes" % fp['size'],
525
"store is probably damaged/corrupt"])
527
if ie.text_sha1 != fp['sha1']:
528
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
529
["inventory expects %s" % ie.text_sha1,
530
"file is actually %s" % fp['sha1'],
531
"store is probably damaged/corrupt"])
534
def path2id(self, path):
535
"""Return the id for path in this tree."""
536
return self._inventory.path2id(path)
538
def paths2ids(self, paths, trees=[], require_versioned=True):
539
"""Return all the ids that can be reached by walking from paths.
541
Each path is looked up in this tree and any extras provided in
542
trees, and this is repeated recursively: the children in an extra tree
543
of a directory that has been renamed under a provided path in this tree
544
are all returned, even if none exist under a provided path in this
545
tree, and vice versa.
547
:param paths: An iterable of paths to start converting to ids from.
548
Alternatively, if paths is None, no ids should be calculated and None
549
will be returned. This is offered to make calling the api unconditional
550
for code that *might* take a list of files.
551
:param trees: Additional trees to consider.
552
:param require_versioned: If False, do not raise NotVersionedError if
553
an element of paths is not versioned in this tree and all of trees.
555
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
557
def iter_children(self, file_id):
558
entry = self.iter_entries_by_dir([file_id]).next()[1]
559
for child in getattr(entry, 'children', {}).itervalues():
565
def revision_tree(self, revision_id):
566
"""Obtain a revision tree for the revision revision_id.
568
The intention of this method is to allow access to possibly cached
569
tree data. Implementors of this method should raise NoSuchRevision if
570
the tree is not locally available, even if they could obtain the
571
tree via a repository or some other means. Callers are responsible
572
for finding the ultimate source for a revision tree.
574
:param revision_id: The revision_id of the requested tree.
576
:raises: NoSuchRevision if the tree cannot be obtained.
578
raise errors.NoSuchRevisionInTree(self, revision_id)
581
"""What files are present in this tree and unknown.
583
:return: an iterator over the unknown files.
590
def filter_unversioned_files(self, paths):
591
"""Filter out paths that are versioned.
593
:return: set of paths.
595
# NB: we specifically *don't* call self.has_filename, because for
596
# WorkingTrees that can indicate files that exist on disk but that
598
pred = self.inventory.has_filename
599
return set((p for p in paths if not pred(p)))
601
def walkdirs(self, prefix=""):
602
"""Walk the contents of this tree from path down.
604
This yields all the data about the contents of a directory at a time.
605
After each directory has been yielded, if the caller has mutated the
606
list to exclude some directories, they are then not descended into.
608
The data yielded is of the form:
609
((directory-relpath, directory-path-from-root, directory-fileid),
610
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
611
versioned_kind), ...]),
612
- directory-relpath is the containing dirs relpath from prefix
613
- directory-path-from-root is the containing dirs path from /
614
- directory-fileid is the id of the directory if it is versioned.
615
- relpath is the relative path within the subtree being walked.
616
- basename is the basename
617
- kind is the kind of the file now. If unknonwn then the file is not
618
present within the tree - but it may be recorded as versioned. See
620
- lstat is the stat data *if* the file was statted.
621
- path_from_tree_root is the path from the root of the tree.
622
- file_id is the file_id if the entry is versioned.
623
- versioned_kind is the kind of the file as last recorded in the
624
versioning system. If 'unknown' the file is not versioned.
625
One of 'kind' and 'versioned_kind' must not be 'unknown'.
627
:param prefix: Start walking from prefix within the tree rather than
628
at the root. This allows one to walk a subtree but get paths that are
629
relative to a tree rooted higher up.
630
:return: an iterator over the directory data.
632
raise NotImplementedError(self.walkdirs)
634
def supports_content_filtering(self):
637
def _content_filter_stack(self, path=None, file_id=None):
638
"""The stack of content filters for a path if filtering is supported.
640
Readers will be applied in first-to-last order.
641
Writers will be applied in last-to-first order.
642
Either the path or the file-id needs to be provided.
644
:param path: path relative to the root of the tree
646
:param file_id: file_id or None if unknown
647
:return: the list of filters - [] if there are none
649
filter_pref_names = filters._get_registered_names()
650
if len(filter_pref_names) == 0:
653
path = self.id2path(file_id)
654
prefs = self.iter_search_rules([path], filter_pref_names).next()
655
stk = filters._get_filter_stack_for(prefs)
656
if 'filters' in debug.debug_flags:
657
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
660
def _content_filter_stack_provider(self):
661
"""A function that returns a stack of ContentFilters.
663
The function takes a path (relative to the top of the tree) and a
664
file-id as parameters.
666
:return: None if content filtering is not supported by this tree.
668
if self.supports_content_filtering():
669
return lambda path, file_id: \
670
self._content_filter_stack(path, file_id)
674
def iter_search_rules(self, path_names, pref_names=None,
675
_default_searcher=None):
676
"""Find the preferences for filenames in a tree.
678
:param path_names: an iterable of paths to find attributes for.
679
Paths are given relative to the root of the tree.
680
:param pref_names: the list of preferences to lookup - None for all
681
:param _default_searcher: private parameter to assist testing - don't use
682
:return: an iterator of tuple sequences, one per path-name.
683
See _RulesSearcher.get_items for details on the tuple sequence.
685
if _default_searcher is None:
686
_default_searcher = rules._per_user_searcher
687
searcher = self._get_rules_searcher(_default_searcher)
688
if searcher is not None:
689
if pref_names is not None:
690
for path in path_names:
691
yield searcher.get_selected_items(path, pref_names)
693
for path in path_names:
694
yield searcher.get_items(path)
697
def _get_rules_searcher(self, default_searcher):
698
"""Get the RulesSearcher for this tree given the default one."""
699
searcher = default_searcher
703
######################################################################
706
# TODO: Merge these two functions into a single one that can operate
707
# on either a whole tree or a set of files.
709
# TODO: Return the diff in order by filename, not by category or in
710
# random order. Can probably be done by lock-stepping through the
711
# filenames from both trees.
714
def file_status(filename, old_tree, new_tree):
715
"""Return single-letter status, old and new names for a file.
717
The complexity here is in deciding how to represent renames;
718
many complex cases are possible.
720
old_inv = old_tree.inventory
721
new_inv = new_tree.inventory
722
new_id = new_inv.path2id(filename)
723
old_id = old_inv.path2id(filename)
725
if not new_id and not old_id:
726
# easy: doesn't exist in either; not versioned at all
727
if new_tree.is_ignored(filename):
728
return 'I', None, None
730
return '?', None, None
732
# There is now a file of this name, great.
735
# There is no longer a file of this name, but we can describe
736
# what happened to the file that used to have
737
# this name. There are two possibilities: either it was
738
# deleted entirely, or renamed.
739
if new_inv.has_id(old_id):
740
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
742
return 'D', old_inv.id2path(old_id), None
744
# if the file_id is new in this revision, it is added
745
if new_id and not old_inv.has_id(new_id):
748
# if there used to be a file of this name, but that ID has now
749
# disappeared, it is deleted
750
if old_id and not new_inv.has_id(old_id):
756
@deprecated_function(deprecated_in((1, 9, 0)))
757
def find_renames(old_inv, new_inv):
758
for file_id in old_inv:
759
if file_id not in new_inv:
761
old_name = old_inv.id2path(file_id)
762
new_name = new_inv.id2path(file_id)
763
if old_name != new_name:
764
yield (old_name, new_name)
767
def find_ids_across_trees(filenames, trees, require_versioned=True):
768
"""Find the ids corresponding to specified filenames.
770
All matches in all trees will be used, and all children of matched
771
directories will be used.
773
:param filenames: The filenames to find file_ids for (if None, returns
775
:param trees: The trees to find file_ids within
776
:param require_versioned: if true, all specified filenames must occur in
778
:return: a set of file ids for the specified filenames and their children.
782
specified_path_ids = _find_ids_across_trees(filenames, trees,
784
return _find_children_across_trees(specified_path_ids, trees)
787
def _find_ids_across_trees(filenames, trees, require_versioned):
788
"""Find the ids corresponding to specified filenames.
790
All matches in all trees will be used, but subdirectories are not scanned.
792
:param filenames: The filenames to find file_ids for
793
:param trees: The trees to find file_ids within
794
:param require_versioned: if true, all specified filenames must occur in
796
:return: a set of file ids for the specified filenames
799
interesting_ids = set()
800
for tree_path in filenames:
803
file_id = tree.path2id(tree_path)
804
if file_id is not None:
805
interesting_ids.add(file_id)
808
not_versioned.append(tree_path)
809
if len(not_versioned) > 0 and require_versioned:
810
raise errors.PathsNotVersionedError(not_versioned)
811
return interesting_ids
814
def _find_children_across_trees(specified_ids, trees):
815
"""Return a set including specified ids and their children.
817
All matches in all trees will be used.
819
:param trees: The trees to find file_ids within
820
:return: a set containing all specified ids and their children
822
interesting_ids = set(specified_ids)
823
pending = interesting_ids
824
# now handle children of interesting ids
825
# we loop so that we handle all children of each id in both trees
826
while len(pending) > 0:
828
for file_id in pending:
830
if not tree.has_or_had_id(file_id):
832
for child_id in tree.iter_children(file_id):
833
if child_id not in interesting_ids:
834
new_pending.add(child_id)
835
interesting_ids.update(new_pending)
836
pending = new_pending
837
return interesting_ids
840
class InterTree(InterObject):
841
"""This class represents operations taking place between two Trees.
843
Its instances have methods like 'compare' and contain references to the
844
source and target trees these operations are to be carried out on.
846
Clients of bzrlib should not need to use InterTree directly, rather they
847
should use the convenience methods on Tree such as 'Tree.compare()' which
848
will pass through to InterTree as appropriate.
853
def _changes_from_entries(self, source_entry, target_entry,
854
source_path=None, target_path=None):
855
"""Generate a iter_changes tuple between source_entry and target_entry.
857
:param source_entry: An inventory entry from self.source, or None.
858
:param target_entry: An inventory entry from self.target, or None.
859
:param source_path: The path of source_entry, if known. If not known
860
it will be looked up.
861
:param target_path: The path of target_entry, if known. If not known
862
it will be looked up.
863
:return: A tuple, item 0 of which is an iter_changes result tuple, and
864
item 1 is True if there are any changes in the result tuple.
866
if source_entry is None:
867
if target_entry is None:
869
file_id = target_entry.file_id
871
file_id = source_entry.file_id
872
if source_entry is not None:
873
source_versioned = True
874
source_name = source_entry.name
875
source_parent = source_entry.parent_id
876
if source_path is None:
877
source_path = self.source.id2path(file_id)
878
source_kind, source_executable, source_stat = \
879
self.source._comparison_data(source_entry, source_path)
881
source_versioned = False
885
source_executable = None
886
if target_entry is not None:
887
target_versioned = True
888
target_name = target_entry.name
889
target_parent = target_entry.parent_id
890
if target_path is None:
891
target_path = self.target.id2path(file_id)
892
target_kind, target_executable, target_stat = \
893
self.target._comparison_data(target_entry, target_path)
895
target_versioned = False
899
target_executable = None
900
versioned = (source_versioned, target_versioned)
901
kind = (source_kind, target_kind)
902
changed_content = False
903
if source_kind != target_kind:
904
changed_content = True
905
elif source_kind == 'file':
906
if (self.source.get_file_sha1(file_id, source_path, source_stat) !=
907
self.target.get_file_sha1(file_id, target_path, target_stat)):
908
changed_content = True
909
elif source_kind == 'symlink':
910
if (self.source.get_symlink_target(file_id) !=
911
self.target.get_symlink_target(file_id)):
912
changed_content = True
913
# XXX: Yes, the indentation below is wrong. But fixing it broke
914
# test_merge.TestMergerEntriesLCAOnDisk.
915
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
916
# the fix from bzr.dev -- vila 2009026
917
elif source_kind == 'tree-reference':
918
if (self.source.get_reference_revision(file_id, source_path)
919
!= self.target.get_reference_revision(file_id, target_path)):
920
changed_content = True
921
parent = (source_parent, target_parent)
922
name = (source_name, target_name)
923
executable = (source_executable, target_executable)
924
if (changed_content is not False or versioned[0] != versioned[1]
925
or parent[0] != parent[1] or name[0] != name[1] or
926
executable[0] != executable[1]):
930
return (file_id, (source_path, target_path), changed_content,
931
versioned, parent, name, kind, executable), changes
934
def compare(self, want_unchanged=False, specific_files=None,
935
extra_trees=None, require_versioned=False, include_root=False,
936
want_unversioned=False):
937
"""Return the changes from source to target.
939
:return: A TreeDelta.
940
:param specific_files: An optional list of file paths to restrict the
941
comparison to. When mapping filenames to ids, all matches in all
942
trees (including optional extra_trees) are used, and all children of
943
matched directories are included.
944
:param want_unchanged: An optional boolean requesting the inclusion of
945
unchanged entries in the result.
946
:param extra_trees: An optional list of additional trees to use when
947
mapping the contents of specific_files (paths) to file_ids.
948
:param require_versioned: An optional boolean (defaults to False). When
949
supplied and True all the 'specific_files' must be versioned, or
950
a PathsNotVersionedError will be thrown.
951
:param want_unversioned: Scan for unversioned paths.
953
# NB: show_status depends on being able to pass in non-versioned files
954
# and report them as unknown
955
trees = (self.source,)
956
if extra_trees is not None:
957
trees = trees + tuple(extra_trees)
958
# target is usually the newer tree:
959
specific_file_ids = self.target.paths2ids(specific_files, trees,
960
require_versioned=require_versioned)
961
if specific_files and not specific_file_ids:
962
# All files are unversioned, so just return an empty delta
963
# _compare_trees would think we want a complete delta
964
result = delta.TreeDelta()
965
fake_entry = InventoryFile('unused', 'unused', 'unused')
966
result.unversioned = [(path, None,
967
self.target._comparison_data(fake_entry, path)[0]) for path in
970
return delta._compare_trees(self.source, self.target, want_unchanged,
971
specific_files, include_root, extra_trees=extra_trees,
972
require_versioned=require_versioned,
973
want_unversioned=want_unversioned)
975
def iter_changes(self, include_unchanged=False,
976
specific_files=None, pb=None, extra_trees=[],
977
require_versioned=True, want_unversioned=False):
978
"""Generate an iterator of changes between trees.
981
(file_id, (path_in_source, path_in_target),
982
changed_content, versioned, parent, name, kind,
985
Changed_content is True if the file's content has changed. This
986
includes changes to its kind, and to a symlink's target.
988
versioned, parent, name, kind, executable are tuples of (from, to).
989
If a file is missing in a tree, its kind is None.
991
Iteration is done in parent-to-child order, relative to the target
994
There is no guarantee that all paths are in sorted order: the
995
requirement to expand the search due to renames may result in children
996
that should be found early being found late in the search, after
997
lexically later results have been returned.
998
:param require_versioned: Raise errors.PathsNotVersionedError if a
999
path in the specific_files list is not versioned in one of
1000
source, target or extra_trees.
1001
:param specific_files: An optional list of file paths to restrict the
1002
comparison to. When mapping filenames to ids, all matches in all
1003
trees (including optional extra_trees) are used, and all children
1004
of matched directories are included. The parents in the target tree
1005
of the specific files up to and including the root of the tree are
1006
always evaluated for changes too.
1007
:param want_unversioned: Should unversioned files be returned in the
1008
output. An unversioned file is defined as one with (False, False)
1009
for the versioned pair.
1011
lookup_trees = [self.source]
1013
lookup_trees.extend(extra_trees)
1014
# The ids of items we need to examine to insure delta consistency.
1015
precise_file_ids = set()
1016
changed_file_ids = []
1017
if specific_files == []:
1018
specific_file_ids = []
1020
specific_file_ids = self.target.paths2ids(specific_files,
1021
lookup_trees, require_versioned=require_versioned)
1022
if specific_files is not None:
1023
# reparented or added entries must have their parents included
1024
# so that valid deltas can be created. The seen_parents set
1025
# tracks the parents that we need to have.
1026
# The seen_dirs set tracks directory entries we've yielded.
1027
# After outputting version object in to_entries we set difference
1028
# the two seen sets and start checking parents.
1029
seen_parents = set()
1031
if want_unversioned:
1032
all_unversioned = sorted([(p.split('/'), p) for p in
1033
self.target.extras()
1034
if specific_files is None or
1035
osutils.is_inside_any(specific_files, p)])
1036
all_unversioned = deque(all_unversioned)
1038
all_unversioned = deque()
1040
from_entries_by_dir = list(self.source.iter_entries_by_dir(
1041
specific_file_ids=specific_file_ids))
1042
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
1043
to_entries_by_dir = list(self.target.iter_entries_by_dir(
1044
specific_file_ids=specific_file_ids))
1045
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
1047
# the unversioned path lookup only occurs on real trees - where there
1048
# can be extras. So the fake_entry is solely used to look up
1049
# executable it values when execute is not supported.
1050
fake_entry = InventoryFile('unused', 'unused', 'unused')
1051
for target_path, target_entry in to_entries_by_dir:
1052
while (all_unversioned and
1053
all_unversioned[0][0] < target_path.split('/')):
1054
unversioned_path = all_unversioned.popleft()
1055
target_kind, target_executable, target_stat = \
1056
self.target._comparison_data(fake_entry, unversioned_path[1])
1057
yield (None, (None, unversioned_path[1]), True, (False, False),
1059
(None, unversioned_path[0][-1]),
1060
(None, target_kind),
1061
(None, target_executable))
1062
source_path, source_entry = from_data.get(target_entry.file_id,
1064
result, changes = self._changes_from_entries(source_entry,
1065
target_entry, source_path=source_path, target_path=target_path)
1066
to_paths[result[0]] = result[1][1]
1071
pb.update('comparing files', entry_count, num_entries)
1072
if changes or include_unchanged:
1073
if specific_file_ids is not None:
1074
new_parent_id = result[4][1]
1075
precise_file_ids.add(new_parent_id)
1076
changed_file_ids.append(result[0])
1078
# Ensure correct behaviour for reparented/added specific files.
1079
if specific_files is not None:
1080
# Record output dirs
1081
if result[6][1] == 'directory':
1082
seen_dirs.add(result[0])
1083
# Record parents of reparented/added entries.
1084
versioned = result[3]
1086
if not versioned[0] or parents[0] != parents[1]:
1087
seen_parents.add(parents[1])
1088
while all_unversioned:
1089
# yield any trailing unversioned paths
1090
unversioned_path = all_unversioned.popleft()
1091
to_kind, to_executable, to_stat = \
1092
self.target._comparison_data(fake_entry, unversioned_path[1])
1093
yield (None, (None, unversioned_path[1]), True, (False, False),
1095
(None, unversioned_path[0][-1]),
1097
(None, to_executable))
1098
# Yield all remaining source paths
1099
for path, from_entry in from_entries_by_dir:
1100
file_id = from_entry.file_id
1101
if file_id in to_paths:
1104
if file_id not in self.target.all_file_ids():
1105
# common case - paths we have not emitted are not present in
1109
to_path = self.target.id2path(file_id)
1112
pb.update('comparing files', entry_count, num_entries)
1113
versioned = (True, False)
1114
parent = (from_entry.parent_id, None)
1115
name = (from_entry.name, None)
1116
from_kind, from_executable, stat_value = \
1117
self.source._comparison_data(from_entry, path)
1118
kind = (from_kind, None)
1119
executable = (from_executable, None)
1120
changed_content = from_kind is not None
1121
# the parent's path is necessarily known at this point.
1122
changed_file_ids.append(file_id)
1123
yield(file_id, (path, to_path), changed_content, versioned, parent,
1124
name, kind, executable)
1125
changed_file_ids = set(changed_file_ids)
1126
if specific_file_ids is not None:
1127
for result in self._handle_precise_ids(precise_file_ids,
1131
def _get_entry(self, tree, file_id):
1132
"""Get an inventory entry from a tree, with missing entries as None.
1134
If the tree raises NotImplementedError on accessing .inventory, then
1135
this is worked around using iter_entries_by_dir on just the file id
1138
:param tree: The tree to lookup the entry in.
1139
:param file_id: The file_id to lookup.
1142
inventory = tree.inventory
1143
except NotImplementedError:
1144
# No inventory available.
1146
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1147
return iterator.next()[1]
1148
except StopIteration:
1152
return inventory[file_id]
1153
except errors.NoSuchId:
1156
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1157
discarded_changes=None):
1158
"""Fill out a partial iter_changes to be consistent.
1160
:param precise_file_ids: The file ids of parents that were seen during
1162
:param changed_file_ids: The file ids of already emitted items.
1163
:param discarded_changes: An optional dict of precalculated
1164
iter_changes items which the partial iter_changes had not output
1166
:return: A generator of iter_changes items to output.
1168
# process parents of things that had changed under the users
1169
# requested paths to prevent incorrect paths or parent ids which
1170
# aren't in the tree.
1171
while precise_file_ids:
1172
precise_file_ids.discard(None)
1173
# Don't emit file_ids twice
1174
precise_file_ids.difference_update(changed_file_ids)
1175
if not precise_file_ids:
1177
# If the there was something at a given output path in source, we
1178
# have to include the entry from source in the delta, or we would
1179
# be putting this entry into a used path.
1181
for parent_id in precise_file_ids:
1183
paths.append(self.target.id2path(parent_id))
1184
except errors.NoSuchId:
1185
# This id has been dragged in from the source by delta
1186
# expansion and isn't present in target at all: we don't
1187
# need to check for path collisions on it.
1190
old_id = self.source.path2id(path)
1191
precise_file_ids.add(old_id)
1192
precise_file_ids.discard(None)
1193
current_ids = precise_file_ids
1194
precise_file_ids = set()
1195
# We have to emit all of precise_file_ids that have been altered.
1196
# We may have to output the children of some of those ids if any
1197
# directories have stopped being directories.
1198
for file_id in current_ids:
1200
if discarded_changes:
1201
result = discarded_changes.get(file_id)
1206
old_entry = self._get_entry(self.source, file_id)
1207
new_entry = self._get_entry(self.target, file_id)
1208
result, changes = self._changes_from_entries(
1209
old_entry, new_entry)
1212
# Get this parents parent to examine.
1213
new_parent_id = result[4][1]
1214
precise_file_ids.add(new_parent_id)
1216
if (result[6][0] == 'directory' and
1217
result[6][1] != 'directory'):
1218
# This stopped being a directory, the old children have
1220
if old_entry is None:
1221
# Reusing a discarded change.
1222
old_entry = self._get_entry(self.source, file_id)
1223
for child in old_entry.children.values():
1224
precise_file_ids.add(child.file_id)
1225
changed_file_ids.add(result[0])
1229
class MultiWalker(object):
1230
"""Walk multiple trees simultaneously, getting combined results."""
1232
# Note: This could be written to not assume you can do out-of-order
1233
# lookups. Instead any nodes that don't match in all trees could be
1234
# marked as 'deferred', and then returned in the final cleanup loop.
1235
# For now, I think it is "nicer" to return things as close to the
1236
# "master_tree" order as we can.
1238
def __init__(self, master_tree, other_trees):
1239
"""Create a new MultiWalker.
1241
All trees being walked must implement "iter_entries_by_dir()", such
1242
that they yield (path, object) tuples, where that object will have a
1243
'.file_id' member, that can be used to check equality.
1245
:param master_tree: All trees will be 'slaved' to the master_tree such
1246
that nodes in master_tree will be used as 'first-pass' sync points.
1247
Any nodes that aren't in master_tree will be merged in a second
1249
:param other_trees: A list of other trees to walk simultaneously.
1251
self._master_tree = master_tree
1252
self._other_trees = other_trees
1254
# Keep track of any nodes that were properly processed just out of
1255
# order, that way we don't return them at the end, we don't have to
1256
# track *all* processed file_ids, just the out-of-order ones
1257
self._out_of_order_processed = set()
1260
def _step_one(iterator):
1261
"""Step an iter_entries_by_dir iterator.
1263
:return: (has_more, path, ie)
1264
If has_more is False, path and ie will be None.
1267
path, ie = iterator.next()
1268
except StopIteration:
1269
return False, None, None
1271
return True, path, ie
1274
def _cmp_path_by_dirblock(path1, path2):
1275
"""Compare two paths based on what directory they are in.
1277
This generates a sort order, such that all children of a directory are
1278
sorted together, and grandchildren are in the same order as the
1279
children appear. But all grandchildren come after all children.
1281
:param path1: first path
1282
:param path2: the second path
1283
:return: negative number if ``path1`` comes first,
1284
0 if paths are equal
1285
and a positive number if ``path2`` sorts first
1287
# Shortcut this special case
1290
# This is stolen from _dirstate_helpers_py.py, only switching it to
1291
# Unicode objects. Consider using encode_utf8() and then using the
1292
# optimized versions, or maybe writing optimized unicode versions.
1293
if not isinstance(path1, unicode):
1294
raise TypeError("'path1' must be a unicode string, not %s: %r"
1295
% (type(path1), path1))
1296
if not isinstance(path2, unicode):
1297
raise TypeError("'path2' must be a unicode string, not %s: %r"
1298
% (type(path2), path2))
1299
return cmp(MultiWalker._path_to_key(path1),
1300
MultiWalker._path_to_key(path2))
1303
def _path_to_key(path):
1304
dirname, basename = osutils.split(path)
1305
return (dirname.split(u'/'), basename)
1307
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1308
"""Lookup an inventory entry by file_id.
1310
This is called when an entry is missing in the normal order.
1311
Generally this is because a file was either renamed, or it was
1312
deleted/added. If the entry was found in the inventory and not in
1313
extra_entries, it will be added to self._out_of_order_processed
1315
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1316
should be filled with entries that were found before they were
1317
used. If file_id is present, it will be removed from the
1319
:param other_tree: The Tree to search, in case we didn't find the entry
1321
:param file_id: The file_id to look for
1322
:return: (path, ie) if found or (None, None) if not present.
1324
if file_id in extra_entries:
1325
return extra_entries.pop(file_id)
1326
# TODO: Is id2path better as the first call, or is
1327
# inventory[file_id] better as a first check?
1329
cur_path = other_tree.id2path(file_id)
1330
except errors.NoSuchId:
1332
if cur_path is None:
1335
self._out_of_order_processed.add(file_id)
1336
cur_ie = other_tree.inventory[file_id]
1337
return (cur_path, cur_ie)
1340
"""Match up the values in the different trees."""
1341
for result in self._walk_master_tree():
1343
self._finish_others()
1344
for result in self._walk_others():
1347
def _walk_master_tree(self):
1348
"""First pass, walk all trees in lock-step.
1350
When we are done, all nodes in the master_tree will have been
1351
processed. _other_walkers, _other_entries, and _others_extra will be
1352
set on 'self' for future processing.
1354
# This iterator has the most "inlining" done, because it tends to touch
1355
# every file in the tree, while the others only hit nodes that don't
1357
master_iterator = self._master_tree.iter_entries_by_dir()
1359
other_walkers = [other.iter_entries_by_dir()
1360
for other in self._other_trees]
1361
other_entries = [self._step_one(walker) for walker in other_walkers]
1362
# Track extra nodes in the other trees
1363
others_extra = [{} for i in xrange(len(self._other_trees))]
1365
master_has_more = True
1366
step_one = self._step_one
1367
lookup_by_file_id = self._lookup_by_file_id
1368
out_of_order_processed = self._out_of_order_processed
1370
while master_has_more:
1371
(master_has_more, path, master_ie) = step_one(master_iterator)
1372
if not master_has_more:
1375
file_id = master_ie.file_id
1377
other_values_append = other_values.append
1378
next_other_entries = []
1379
next_other_entries_append = next_other_entries.append
1380
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1381
if not other_has_more:
1382
other_values_append(lookup_by_file_id(
1383
others_extra[idx], self._other_trees[idx], file_id))
1384
next_other_entries_append((False, None, None))
1385
elif file_id == other_ie.file_id:
1386
# This is the critical code path, as most of the entries
1387
# should match between most trees.
1388
other_values_append((other_path, other_ie))
1389
next_other_entries_append(step_one(other_walkers[idx]))
1391
# This walker did not match, step it until it either
1392
# matches, or we know we are past the current walker.
1393
other_walker = other_walkers[idx]
1394
other_extra = others_extra[idx]
1395
while (other_has_more and
1396
self._cmp_path_by_dirblock(other_path, path) < 0):
1397
other_file_id = other_ie.file_id
1398
if other_file_id not in out_of_order_processed:
1399
other_extra[other_file_id] = (other_path, other_ie)
1400
other_has_more, other_path, other_ie = \
1401
step_one(other_walker)
1402
if other_has_more and other_ie.file_id == file_id:
1403
# We ended up walking to this point, match and step
1405
other_values_append((other_path, other_ie))
1406
other_has_more, other_path, other_ie = \
1407
step_one(other_walker)
1409
# This record isn't in the normal order, see if it
1411
other_values_append(lookup_by_file_id(
1412
other_extra, self._other_trees[idx], file_id))
1413
next_other_entries_append((other_has_more, other_path,
1415
other_entries = next_other_entries
1417
# We've matched all the walkers, yield this datapoint
1418
yield path, file_id, master_ie, other_values
1419
self._other_walkers = other_walkers
1420
self._other_entries = other_entries
1421
self._others_extra = others_extra
1423
def _finish_others(self):
1424
"""Finish walking the other iterators, so we get all entries."""
1425
for idx, info in enumerate(self._other_entries):
1426
other_extra = self._others_extra[idx]
1427
(other_has_more, other_path, other_ie) = info
1428
while other_has_more:
1429
other_file_id = other_ie.file_id
1430
if other_file_id not in self._out_of_order_processed:
1431
other_extra[other_file_id] = (other_path, other_ie)
1432
other_has_more, other_path, other_ie = \
1433
self._step_one(self._other_walkers[idx])
1434
del self._other_entries
1436
def _walk_others(self):
1437
"""Finish up by walking all the 'deferred' nodes."""
1438
# TODO: One alternative would be to grab all possible unprocessed
1439
# file_ids, and then sort by path, and then yield them. That
1440
# might ensure better ordering, in case a caller strictly
1441
# requires parents before children.
1442
for idx, other_extra in enumerate(self._others_extra):
1443
others = sorted(other_extra.itervalues(),
1444
key=lambda x: self._path_to_key(x[0]))
1445
for other_path, other_ie in others:
1446
file_id = other_ie.file_id
1447
# We don't need to check out_of_order_processed here, because
1448
# the lookup_by_file_id will be removing anything processed
1449
# from the extras cache
1450
other_extra.pop(file_id)
1451
other_values = [(None, None) for i in xrange(idx)]
1452
other_values.append((other_path, other_ie))
1453
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1454
alt_idx = alt_idx + idx + 1
1455
alt_extra = self._others_extra[alt_idx]
1456
alt_tree = self._other_trees[alt_idx]
1457
other_values.append(self._lookup_by_file_id(
1458
alt_extra, alt_tree, file_id))
1459
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
