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
All the attributes returned are for the canonical form, not the
226
convenient form (if content filters are in use.)
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 and the size of the
232
canonical form can be cheaply determined, None otherwise.
233
exec is None unless kind is file and the platform supports the 'x'
235
sha1-or-link is the link target if kind is symlink, or the sha1 if
236
it can be obtained without reading the file.
238
raise NotImplementedError(self.path_content_summary)
240
def get_reference_revision(self, file_id, path=None):
241
raise NotImplementedError("Tree subclass %s must implement "
242
"get_reference_revision"
243
% self.__class__.__name__)
245
def _comparison_data(self, entry, path):
246
"""Return a tuple of kind, executable, stat_value for a file.
248
entry may be None if there is no inventory entry for the file, but
249
path must always be supplied.
251
kind is None if there is no file present (even if an inventory id is
252
present). executable is False for non-file entries.
254
raise NotImplementedError(self._comparison_data)
256
def _file_size(self, entry, stat_value):
257
raise NotImplementedError(self._file_size)
259
def _get_inventory(self):
260
return self._inventory
262
def get_file(self, file_id, path=None):
263
"""Return a file object for the file file_id in the tree.
265
If both file_id and path are defined, it is implementation defined as
266
to which one is used.
268
raise NotImplementedError(self.get_file)
270
def get_file_with_stat(self, file_id, path=None):
271
"""Get a file handle and stat object for file_id.
273
The default implementation returns (self.get_file, None) for backwards
276
:param file_id: The file id to read.
277
:param path: The path of the file, if it is known.
278
:return: A tuple (file_handle, stat_value_or_None). If the tree has
279
no stat facility, or need for a stat cache feedback during commit,
280
it may return None for the second element of the tuple.
282
return (self.get_file(file_id, path), None)
284
def get_file_text(self, file_id, path=None):
285
"""Return the byte content of a file.
287
:param file_id: The file_id of the file.
288
:param path: The path of the file.
289
If both file_id and path are supplied, an implementation may use
292
my_file = self.get_file(file_id, path)
294
return my_file.read()
298
def get_file_lines(self, file_id, path=None):
299
"""Return the content of a file, as lines.
301
:param file_id: The file_id of the file.
302
:param path: The path of the file.
303
If both file_id and path are supplied, an implementation may use
306
return osutils.split_lines(self.get_file_text(file_id, path))
308
def get_file_mtime(self, file_id, path=None):
309
"""Return the modification time for a file.
311
:param file_id: The handle for this file.
312
:param path: The path that this file can be found at.
313
These must point to the same object.
315
raise NotImplementedError(self.get_file_mtime)
317
def get_file_size(self, file_id):
318
"""Return the size of a file in bytes.
320
This applies only to regular files. If invoked on directories or
321
symlinks, it will return None.
322
:param file_id: The file-id of the file
324
raise NotImplementedError(self.get_file_size)
326
def get_file_by_path(self, path):
327
return self.get_file(self._inventory.path2id(path), path)
329
def iter_files_bytes(self, desired_files):
330
"""Iterate through file contents.
332
Files will not necessarily be returned in the order they occur in
333
desired_files. No specific order is guaranteed.
335
Yields pairs of identifier, bytes_iterator. identifier is an opaque
336
value supplied by the caller as part of desired_files. It should
337
uniquely identify the file version in the caller's context. (Examples:
338
an index number or a TreeTransform trans_id.)
340
bytes_iterator is an iterable of bytestrings for the file. The
341
kind of iterable and length of the bytestrings are unspecified, but for
342
this implementation, it is a tuple containing a single bytestring with
343
the complete text of the file.
345
:param desired_files: a list of (file_id, identifier) pairs
347
for file_id, identifier in desired_files:
348
# We wrap the string in a tuple so that we can return an iterable
349
# of bytestrings. (Technically, a bytestring is also an iterable
350
# of bytestrings, but iterating through each character is not
352
cur_file = (self.get_file_text(file_id),)
353
yield identifier, cur_file
355
def get_symlink_target(self, file_id):
356
"""Get the target for a given file_id.
358
It is assumed that the caller already knows that file_id is referencing
360
:param file_id: Handle for the symlink entry.
361
:return: The path the symlink points to.
363
raise NotImplementedError(self.get_symlink_target)
365
def get_canonical_inventory_paths(self, paths):
366
"""Like get_canonical_inventory_path() but works on multiple items.
368
:param paths: A sequence of paths relative to the root of the tree.
369
:return: A list of paths, with each item the corresponding input path
370
adjusted to account for existing elements that match case
373
return list(self._yield_canonical_inventory_paths(paths))
375
def get_canonical_inventory_path(self, path):
376
"""Returns the first inventory item that case-insensitively matches path.
378
If a path matches exactly, it is returned. If no path matches exactly
379
but more than one path matches case-insensitively, it is implementation
380
defined which is returned.
382
If no path matches case-insensitively, the input path is returned, but
383
with as many path entries that do exist changed to their canonical
386
If you need to resolve many names from the same tree, you should
387
use get_canonical_inventory_paths() to avoid O(N) behaviour.
389
:param path: A paths relative to the root of the tree.
390
:return: The input path adjusted to account for existing elements
391
that match case insensitively.
393
return self._yield_canonical_inventory_paths([path]).next()
395
def _yield_canonical_inventory_paths(self, paths):
397
# First, if the path as specified exists exactly, just use it.
398
if self.path2id(path) is not None:
402
cur_id = self.get_root_id()
404
bit_iter = iter(path.split("/"))
407
for child in self.iter_children(cur_id):
409
child_base = os.path.basename(self.id2path(child))
410
if child_base.lower() == lelt:
412
cur_path = osutils.pathjoin(cur_path, child_base)
415
# before a change is committed we can see this error...
418
# got to the end of this directory and no entries matched.
419
# Return what matched so far, plus the rest as specified.
420
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
425
def get_root_id(self):
426
"""Return the file_id for the root of this tree."""
427
raise NotImplementedError(self.get_root_id)
429
def annotate_iter(self, file_id,
430
default_revision=_mod_revision.CURRENT_REVISION):
431
"""Return an iterator of revision_id, line tuples.
433
For working trees (and mutable trees in general), the special
434
revision_id 'current:' will be used for lines that are new in this
435
tree, e.g. uncommitted changes.
436
:param file_id: The file to produce an annotated version from
437
:param default_revision: For lines that don't match a basis, mark them
438
with this revision id. Not all implementations will make use of
441
raise NotImplementedError(self.annotate_iter)
443
def _get_plan_merge_data(self, file_id, other, base):
444
from bzrlib import versionedfile
445
vf = versionedfile._PlanMergeVersionedFile(file_id)
446
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
447
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
449
last_revision_base = None
451
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
452
return vf, last_revision_a, last_revision_b, last_revision_base
454
def plan_file_merge(self, file_id, other, base=None):
455
"""Generate a merge plan based on annotations.
457
If the file contains uncommitted changes in this tree, they will be
458
attributed to the 'current:' pseudo-revision. If the file contains
459
uncommitted changes in the other tree, they will be assigned to the
460
'other:' pseudo-revision.
462
data = self._get_plan_merge_data(file_id, other, base)
463
vf, last_revision_a, last_revision_b, last_revision_base = data
464
return vf.plan_merge(last_revision_a, last_revision_b,
467
def plan_file_lca_merge(self, file_id, other, base=None):
468
"""Generate a merge plan based lca-newness.
470
If the file contains uncommitted changes in this tree, they will be
471
attributed to the 'current:' pseudo-revision. If the file contains
472
uncommitted changes in the other tree, they will be assigned to the
473
'other:' pseudo-revision.
475
data = self._get_plan_merge_data(file_id, other, base)
476
vf, last_revision_a, last_revision_b, last_revision_base = data
477
return vf.plan_lca_merge(last_revision_a, last_revision_b,
480
def _iter_parent_trees(self):
481
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
482
for revision_id in self.get_parent_ids():
484
yield self.revision_tree(revision_id)
485
except errors.NoSuchRevisionInTree:
486
yield self.repository.revision_tree(revision_id)
489
def _file_revision(revision_tree, file_id):
490
"""Determine the revision associated with a file in a given tree."""
491
revision_tree.lock_read()
493
return revision_tree.inventory[file_id].revision
495
revision_tree.unlock()
497
def _get_file_revision(self, file_id, vf, tree_revision):
498
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
500
if getattr(self, '_repository', None) is None:
501
last_revision = tree_revision
502
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
503
self._iter_parent_trees()]
504
vf.add_lines((file_id, last_revision), parent_keys,
505
self.get_file(file_id).readlines())
506
repo = self.branch.repository
509
last_revision = self._file_revision(self, file_id)
510
base_vf = self._repository.texts
511
if base_vf not in vf.fallback_versionedfiles:
512
vf.fallback_versionedfiles.append(base_vf)
515
inventory = property(_get_inventory,
516
doc="Inventory of this Tree")
518
def _check_retrieved(self, ie, f):
521
fp = fingerprint_file(f)
524
if ie.text_size is not None:
525
if ie.text_size != fp['size']:
526
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
527
["inventory expects %d bytes" % ie.text_size,
528
"file is actually %d bytes" % fp['size'],
529
"store is probably damaged/corrupt"])
531
if ie.text_sha1 != fp['sha1']:
532
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
533
["inventory expects %s" % ie.text_sha1,
534
"file is actually %s" % fp['sha1'],
535
"store is probably damaged/corrupt"])
538
def path2id(self, path):
539
"""Return the id for path in this tree."""
540
return self._inventory.path2id(path)
542
def paths2ids(self, paths, trees=[], require_versioned=True):
543
"""Return all the ids that can be reached by walking from paths.
545
Each path is looked up in this tree and any extras provided in
546
trees, and this is repeated recursively: the children in an extra tree
547
of a directory that has been renamed under a provided path in this tree
548
are all returned, even if none exist under a provided path in this
549
tree, and vice versa.
551
:param paths: An iterable of paths to start converting to ids from.
552
Alternatively, if paths is None, no ids should be calculated and None
553
will be returned. This is offered to make calling the api unconditional
554
for code that *might* take a list of files.
555
:param trees: Additional trees to consider.
556
:param require_versioned: If False, do not raise NotVersionedError if
557
an element of paths is not versioned in this tree and all of trees.
559
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
561
def iter_children(self, file_id):
562
entry = self.iter_entries_by_dir([file_id]).next()[1]
563
for child in getattr(entry, 'children', {}).itervalues():
569
def revision_tree(self, revision_id):
570
"""Obtain a revision tree for the revision revision_id.
572
The intention of this method is to allow access to possibly cached
573
tree data. Implementors of this method should raise NoSuchRevision if
574
the tree is not locally available, even if they could obtain the
575
tree via a repository or some other means. Callers are responsible
576
for finding the ultimate source for a revision tree.
578
:param revision_id: The revision_id of the requested tree.
580
:raises: NoSuchRevision if the tree cannot be obtained.
582
raise errors.NoSuchRevisionInTree(self, revision_id)
585
"""What files are present in this tree and unknown.
587
:return: an iterator over the unknown files.
594
def filter_unversioned_files(self, paths):
595
"""Filter out paths that are versioned.
597
:return: set of paths.
599
# NB: we specifically *don't* call self.has_filename, because for
600
# WorkingTrees that can indicate files that exist on disk but that
602
pred = self.inventory.has_filename
603
return set((p for p in paths if not pred(p)))
605
def walkdirs(self, prefix=""):
606
"""Walk the contents of this tree from path down.
608
This yields all the data about the contents of a directory at a time.
609
After each directory has been yielded, if the caller has mutated the
610
list to exclude some directories, they are then not descended into.
612
The data yielded is of the form:
613
((directory-relpath, directory-path-from-root, directory-fileid),
614
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
615
versioned_kind), ...]),
616
- directory-relpath is the containing dirs relpath from prefix
617
- directory-path-from-root is the containing dirs path from /
618
- directory-fileid is the id of the directory if it is versioned.
619
- relpath is the relative path within the subtree being walked.
620
- basename is the basename
621
- kind is the kind of the file now. If unknonwn then the file is not
622
present within the tree - but it may be recorded as versioned. See
624
- lstat is the stat data *if* the file was statted.
625
- path_from_tree_root is the path from the root of the tree.
626
- file_id is the file_id if the entry is versioned.
627
- versioned_kind is the kind of the file as last recorded in the
628
versioning system. If 'unknown' the file is not versioned.
629
One of 'kind' and 'versioned_kind' must not be 'unknown'.
631
:param prefix: Start walking from prefix within the tree rather than
632
at the root. This allows one to walk a subtree but get paths that are
633
relative to a tree rooted higher up.
634
:return: an iterator over the directory data.
636
raise NotImplementedError(self.walkdirs)
638
def supports_content_filtering(self):
641
def _content_filter_stack(self, path=None, file_id=None):
642
"""The stack of content filters for a path if filtering is supported.
644
Readers will be applied in first-to-last order.
645
Writers will be applied in last-to-first order.
646
Either the path or the file-id needs to be provided.
648
:param path: path relative to the root of the tree
650
:param file_id: file_id or None if unknown
651
:return: the list of filters - [] if there are none
653
filter_pref_names = filters._get_registered_names()
654
if len(filter_pref_names) == 0:
657
path = self.id2path(file_id)
658
prefs = self.iter_search_rules([path], filter_pref_names).next()
659
stk = filters._get_filter_stack_for(prefs)
660
if 'filters' in debug.debug_flags:
661
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
664
def _content_filter_stack_provider(self):
665
"""A function that returns a stack of ContentFilters.
667
The function takes a path (relative to the top of the tree) and a
668
file-id as parameters.
670
:return: None if content filtering is not supported by this tree.
672
if self.supports_content_filtering():
673
return lambda path, file_id: \
674
self._content_filter_stack(path, file_id)
678
def iter_search_rules(self, path_names, pref_names=None,
679
_default_searcher=None):
680
"""Find the preferences for filenames in a tree.
682
:param path_names: an iterable of paths to find attributes for.
683
Paths are given relative to the root of the tree.
684
:param pref_names: the list of preferences to lookup - None for all
685
:param _default_searcher: private parameter to assist testing - don't use
686
:return: an iterator of tuple sequences, one per path-name.
687
See _RulesSearcher.get_items for details on the tuple sequence.
689
if _default_searcher is None:
690
_default_searcher = rules._per_user_searcher
691
searcher = self._get_rules_searcher(_default_searcher)
692
if searcher is not None:
693
if pref_names is not None:
694
for path in path_names:
695
yield searcher.get_selected_items(path, pref_names)
697
for path in path_names:
698
yield searcher.get_items(path)
700
def _get_rules_searcher(self, default_searcher):
701
"""Get the RulesSearcher for this tree given the default one."""
702
searcher = default_searcher
706
######################################################################
709
# TODO: Merge these two functions into a single one that can operate
710
# on either a whole tree or a set of files.
712
# TODO: Return the diff in order by filename, not by category or in
713
# random order. Can probably be done by lock-stepping through the
714
# filenames from both trees.
717
def file_status(filename, old_tree, new_tree):
718
"""Return single-letter status, old and new names for a file.
720
The complexity here is in deciding how to represent renames;
721
many complex cases are possible.
723
old_inv = old_tree.inventory
724
new_inv = new_tree.inventory
725
new_id = new_inv.path2id(filename)
726
old_id = old_inv.path2id(filename)
728
if not new_id and not old_id:
729
# easy: doesn't exist in either; not versioned at all
730
if new_tree.is_ignored(filename):
731
return 'I', None, None
733
return '?', None, None
735
# There is now a file of this name, great.
738
# There is no longer a file of this name, but we can describe
739
# what happened to the file that used to have
740
# this name. There are two possibilities: either it was
741
# deleted entirely, or renamed.
742
if new_inv.has_id(old_id):
743
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
745
return 'D', old_inv.id2path(old_id), None
747
# if the file_id is new in this revision, it is added
748
if new_id and not old_inv.has_id(new_id):
751
# if there used to be a file of this name, but that ID has now
752
# disappeared, it is deleted
753
if old_id and not new_inv.has_id(old_id):
759
@deprecated_function(deprecated_in((1, 9, 0)))
760
def find_renames(old_inv, new_inv):
761
for file_id in old_inv:
762
if file_id not in new_inv:
764
old_name = old_inv.id2path(file_id)
765
new_name = new_inv.id2path(file_id)
766
if old_name != new_name:
767
yield (old_name, new_name)
770
def find_ids_across_trees(filenames, trees, require_versioned=True):
771
"""Find the ids corresponding to specified filenames.
773
All matches in all trees will be used, and all children of matched
774
directories will be used.
776
:param filenames: The filenames to find file_ids for (if None, returns
778
:param trees: The trees to find file_ids within
779
:param require_versioned: if true, all specified filenames must occur in
781
:return: a set of file ids for the specified filenames and their children.
785
specified_path_ids = _find_ids_across_trees(filenames, trees,
787
return _find_children_across_trees(specified_path_ids, trees)
790
def _find_ids_across_trees(filenames, trees, require_versioned):
791
"""Find the ids corresponding to specified filenames.
793
All matches in all trees will be used, but subdirectories are not scanned.
795
:param filenames: The filenames to find file_ids for
796
:param trees: The trees to find file_ids within
797
:param require_versioned: if true, all specified filenames must occur in
799
:return: a set of file ids for the specified filenames
802
interesting_ids = set()
803
for tree_path in filenames:
806
file_id = tree.path2id(tree_path)
807
if file_id is not None:
808
interesting_ids.add(file_id)
811
not_versioned.append(tree_path)
812
if len(not_versioned) > 0 and require_versioned:
813
raise errors.PathsNotVersionedError(not_versioned)
814
return interesting_ids
817
def _find_children_across_trees(specified_ids, trees):
818
"""Return a set including specified ids and their children.
820
All matches in all trees will be used.
822
:param trees: The trees to find file_ids within
823
:return: a set containing all specified ids and their children
825
interesting_ids = set(specified_ids)
826
pending = interesting_ids
827
# now handle children of interesting ids
828
# we loop so that we handle all children of each id in both trees
829
while len(pending) > 0:
831
for file_id in pending:
833
if not tree.has_or_had_id(file_id):
835
for child_id in tree.iter_children(file_id):
836
if child_id not in interesting_ids:
837
new_pending.add(child_id)
838
interesting_ids.update(new_pending)
839
pending = new_pending
840
return interesting_ids
843
class InterTree(InterObject):
844
"""This class represents operations taking place between two Trees.
846
Its instances have methods like 'compare' and contain references to the
847
source and target trees these operations are to be carried out on.
849
Clients of bzrlib should not need to use InterTree directly, rather they
850
should use the convenience methods on Tree such as 'Tree.compare()' which
851
will pass through to InterTree as appropriate.
856
def _changes_from_entries(self, source_entry, target_entry,
857
source_path=None, target_path=None):
858
"""Generate a iter_changes tuple between source_entry and target_entry.
860
:param source_entry: An inventory entry from self.source, or None.
861
:param target_entry: An inventory entry from self.target, or None.
862
:param source_path: The path of source_entry, if known. If not known
863
it will be looked up.
864
:param target_path: The path of target_entry, if known. If not known
865
it will be looked up.
866
:return: A tuple, item 0 of which is an iter_changes result tuple, and
867
item 1 is True if there are any changes in the result tuple.
869
if source_entry is None:
870
if target_entry is None:
872
file_id = target_entry.file_id
874
file_id = source_entry.file_id
875
if source_entry is not None:
876
source_versioned = True
877
source_name = source_entry.name
878
source_parent = source_entry.parent_id
879
if source_path is None:
880
source_path = self.source.id2path(file_id)
881
source_kind, source_executable, source_stat = \
882
self.source._comparison_data(source_entry, source_path)
884
source_versioned = False
888
source_executable = None
889
if target_entry is not None:
890
target_versioned = True
891
target_name = target_entry.name
892
target_parent = target_entry.parent_id
893
if target_path is None:
894
target_path = self.target.id2path(file_id)
895
target_kind, target_executable, target_stat = \
896
self.target._comparison_data(target_entry, target_path)
898
target_versioned = False
902
target_executable = None
903
versioned = (source_versioned, target_versioned)
904
kind = (source_kind, target_kind)
905
changed_content = False
906
if source_kind != target_kind:
907
changed_content = True
908
elif source_kind == 'file':
909
if (self.source.get_file_sha1(file_id, source_path, source_stat) !=
910
self.target.get_file_sha1(file_id, target_path, target_stat)):
911
changed_content = True
912
elif source_kind == 'symlink':
913
if (self.source.get_symlink_target(file_id) !=
914
self.target.get_symlink_target(file_id)):
915
changed_content = True
916
# XXX: Yes, the indentation below is wrong. But fixing it broke
917
# test_merge.TestMergerEntriesLCAOnDisk.
918
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
919
# the fix from bzr.dev -- vila 2009026
920
elif source_kind == 'tree-reference':
921
if (self.source.get_reference_revision(file_id, source_path)
922
!= self.target.get_reference_revision(file_id, target_path)):
923
changed_content = True
924
parent = (source_parent, target_parent)
925
name = (source_name, target_name)
926
executable = (source_executable, target_executable)
927
if (changed_content is not False or versioned[0] != versioned[1]
928
or parent[0] != parent[1] or name[0] != name[1] or
929
executable[0] != executable[1]):
933
return (file_id, (source_path, target_path), changed_content,
934
versioned, parent, name, kind, executable), changes
937
def compare(self, want_unchanged=False, specific_files=None,
938
extra_trees=None, require_versioned=False, include_root=False,
939
want_unversioned=False):
940
"""Return the changes from source to target.
942
:return: A TreeDelta.
943
:param specific_files: An optional list of file paths to restrict the
944
comparison to. When mapping filenames to ids, all matches in all
945
trees (including optional extra_trees) are used, and all children of
946
matched directories are included.
947
:param want_unchanged: An optional boolean requesting the inclusion of
948
unchanged entries in the result.
949
:param extra_trees: An optional list of additional trees to use when
950
mapping the contents of specific_files (paths) to file_ids.
951
:param require_versioned: An optional boolean (defaults to False). When
952
supplied and True all the 'specific_files' must be versioned, or
953
a PathsNotVersionedError will be thrown.
954
:param want_unversioned: Scan for unversioned paths.
956
# NB: show_status depends on being able to pass in non-versioned files
957
# and report them as unknown
958
trees = (self.source,)
959
if extra_trees is not None:
960
trees = trees + tuple(extra_trees)
961
# target is usually the newer tree:
962
specific_file_ids = self.target.paths2ids(specific_files, trees,
963
require_versioned=require_versioned)
964
if specific_files and not specific_file_ids:
965
# All files are unversioned, so just return an empty delta
966
# _compare_trees would think we want a complete delta
967
result = delta.TreeDelta()
968
fake_entry = InventoryFile('unused', 'unused', 'unused')
969
result.unversioned = [(path, None,
970
self.target._comparison_data(fake_entry, path)[0]) for path in
973
return delta._compare_trees(self.source, self.target, want_unchanged,
974
specific_files, include_root, extra_trees=extra_trees,
975
require_versioned=require_versioned,
976
want_unversioned=want_unversioned)
978
def iter_changes(self, include_unchanged=False,
979
specific_files=None, pb=None, extra_trees=[],
980
require_versioned=True, want_unversioned=False):
981
"""Generate an iterator of changes between trees.
984
(file_id, (path_in_source, path_in_target),
985
changed_content, versioned, parent, name, kind,
988
Changed_content is True if the file's content has changed. This
989
includes changes to its kind, and to a symlink's target.
991
versioned, parent, name, kind, executable are tuples of (from, to).
992
If a file is missing in a tree, its kind is None.
994
Iteration is done in parent-to-child order, relative to the target
997
There is no guarantee that all paths are in sorted order: the
998
requirement to expand the search due to renames may result in children
999
that should be found early being found late in the search, after
1000
lexically later results have been returned.
1001
:param require_versioned: Raise errors.PathsNotVersionedError if a
1002
path in the specific_files list is not versioned in one of
1003
source, target or extra_trees.
1004
:param specific_files: An optional list of file paths to restrict the
1005
comparison to. When mapping filenames to ids, all matches in all
1006
trees (including optional extra_trees) are used, and all children
1007
of matched directories are included. The parents in the target tree
1008
of the specific files up to and including the root of the tree are
1009
always evaluated for changes too.
1010
:param want_unversioned: Should unversioned files be returned in the
1011
output. An unversioned file is defined as one with (False, False)
1012
for the versioned pair.
1014
lookup_trees = [self.source]
1016
lookup_trees.extend(extra_trees)
1017
# The ids of items we need to examine to insure delta consistency.
1018
precise_file_ids = set()
1019
changed_file_ids = []
1020
if specific_files == []:
1021
specific_file_ids = []
1023
specific_file_ids = self.target.paths2ids(specific_files,
1024
lookup_trees, require_versioned=require_versioned)
1025
if specific_files is not None:
1026
# reparented or added entries must have their parents included
1027
# so that valid deltas can be created. The seen_parents set
1028
# tracks the parents that we need to have.
1029
# The seen_dirs set tracks directory entries we've yielded.
1030
# After outputting version object in to_entries we set difference
1031
# the two seen sets and start checking parents.
1032
seen_parents = set()
1034
if want_unversioned:
1035
all_unversioned = sorted([(p.split('/'), p) for p in
1036
self.target.extras()
1037
if specific_files is None or
1038
osutils.is_inside_any(specific_files, p)])
1039
all_unversioned = deque(all_unversioned)
1041
all_unversioned = deque()
1043
from_entries_by_dir = list(self.source.iter_entries_by_dir(
1044
specific_file_ids=specific_file_ids))
1045
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
1046
to_entries_by_dir = list(self.target.iter_entries_by_dir(
1047
specific_file_ids=specific_file_ids))
1048
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
1050
# the unversioned path lookup only occurs on real trees - where there
1051
# can be extras. So the fake_entry is solely used to look up
1052
# executable it values when execute is not supported.
1053
fake_entry = InventoryFile('unused', 'unused', 'unused')
1054
for target_path, target_entry in to_entries_by_dir:
1055
while (all_unversioned and
1056
all_unversioned[0][0] < target_path.split('/')):
1057
unversioned_path = all_unversioned.popleft()
1058
target_kind, target_executable, target_stat = \
1059
self.target._comparison_data(fake_entry, unversioned_path[1])
1060
yield (None, (None, unversioned_path[1]), True, (False, False),
1062
(None, unversioned_path[0][-1]),
1063
(None, target_kind),
1064
(None, target_executable))
1065
source_path, source_entry = from_data.get(target_entry.file_id,
1067
result, changes = self._changes_from_entries(source_entry,
1068
target_entry, source_path=source_path, target_path=target_path)
1069
to_paths[result[0]] = result[1][1]
1074
pb.update('comparing files', entry_count, num_entries)
1075
if changes or include_unchanged:
1076
if specific_file_ids is not None:
1077
new_parent_id = result[4][1]
1078
precise_file_ids.add(new_parent_id)
1079
changed_file_ids.append(result[0])
1081
# Ensure correct behaviour for reparented/added specific files.
1082
if specific_files is not None:
1083
# Record output dirs
1084
if result[6][1] == 'directory':
1085
seen_dirs.add(result[0])
1086
# Record parents of reparented/added entries.
1087
versioned = result[3]
1089
if not versioned[0] or parents[0] != parents[1]:
1090
seen_parents.add(parents[1])
1091
while all_unversioned:
1092
# yield any trailing unversioned paths
1093
unversioned_path = all_unversioned.popleft()
1094
to_kind, to_executable, to_stat = \
1095
self.target._comparison_data(fake_entry, unversioned_path[1])
1096
yield (None, (None, unversioned_path[1]), True, (False, False),
1098
(None, unversioned_path[0][-1]),
1100
(None, to_executable))
1101
# Yield all remaining source paths
1102
for path, from_entry in from_entries_by_dir:
1103
file_id = from_entry.file_id
1104
if file_id in to_paths:
1107
if file_id not in self.target.all_file_ids():
1108
# common case - paths we have not emitted are not present in
1112
to_path = self.target.id2path(file_id)
1115
pb.update('comparing files', entry_count, num_entries)
1116
versioned = (True, False)
1117
parent = (from_entry.parent_id, None)
1118
name = (from_entry.name, None)
1119
from_kind, from_executable, stat_value = \
1120
self.source._comparison_data(from_entry, path)
1121
kind = (from_kind, None)
1122
executable = (from_executable, None)
1123
changed_content = from_kind is not None
1124
# the parent's path is necessarily known at this point.
1125
changed_file_ids.append(file_id)
1126
yield(file_id, (path, to_path), changed_content, versioned, parent,
1127
name, kind, executable)
1128
changed_file_ids = set(changed_file_ids)
1129
if specific_file_ids is not None:
1130
for result in self._handle_precise_ids(precise_file_ids,
1134
def _get_entry(self, tree, file_id):
1135
"""Get an inventory entry from a tree, with missing entries as None.
1137
If the tree raises NotImplementedError on accessing .inventory, then
1138
this is worked around using iter_entries_by_dir on just the file id
1141
:param tree: The tree to lookup the entry in.
1142
:param file_id: The file_id to lookup.
1145
inventory = tree.inventory
1146
except NotImplementedError:
1147
# No inventory available.
1149
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1150
return iterator.next()[1]
1151
except StopIteration:
1155
return inventory[file_id]
1156
except errors.NoSuchId:
1159
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1160
discarded_changes=None):
1161
"""Fill out a partial iter_changes to be consistent.
1163
:param precise_file_ids: The file ids of parents that were seen during
1165
:param changed_file_ids: The file ids of already emitted items.
1166
:param discarded_changes: An optional dict of precalculated
1167
iter_changes items which the partial iter_changes had not output
1169
:return: A generator of iter_changes items to output.
1171
# process parents of things that had changed under the users
1172
# requested paths to prevent incorrect paths or parent ids which
1173
# aren't in the tree.
1174
while precise_file_ids:
1175
precise_file_ids.discard(None)
1176
# Don't emit file_ids twice
1177
precise_file_ids.difference_update(changed_file_ids)
1178
if not precise_file_ids:
1180
# If the there was something at a given output path in source, we
1181
# have to include the entry from source in the delta, or we would
1182
# be putting this entry into a used path.
1184
for parent_id in precise_file_ids:
1186
paths.append(self.target.id2path(parent_id))
1187
except errors.NoSuchId:
1188
# This id has been dragged in from the source by delta
1189
# expansion and isn't present in target at all: we don't
1190
# need to check for path collisions on it.
1193
old_id = self.source.path2id(path)
1194
precise_file_ids.add(old_id)
1195
precise_file_ids.discard(None)
1196
current_ids = precise_file_ids
1197
precise_file_ids = set()
1198
# We have to emit all of precise_file_ids that have been altered.
1199
# We may have to output the children of some of those ids if any
1200
# directories have stopped being directories.
1201
for file_id in current_ids:
1203
if discarded_changes:
1204
result = discarded_changes.get(file_id)
1209
old_entry = self._get_entry(self.source, file_id)
1210
new_entry = self._get_entry(self.target, file_id)
1211
result, changes = self._changes_from_entries(
1212
old_entry, new_entry)
1215
# Get this parents parent to examine.
1216
new_parent_id = result[4][1]
1217
precise_file_ids.add(new_parent_id)
1219
if (result[6][0] == 'directory' and
1220
result[6][1] != 'directory'):
1221
# This stopped being a directory, the old children have
1223
if old_entry is None:
1224
# Reusing a discarded change.
1225
old_entry = self._get_entry(self.source, file_id)
1226
for child in old_entry.children.values():
1227
precise_file_ids.add(child.file_id)
1228
changed_file_ids.add(result[0])
1232
class MultiWalker(object):
1233
"""Walk multiple trees simultaneously, getting combined results."""
1235
# Note: This could be written to not assume you can do out-of-order
1236
# lookups. Instead any nodes that don't match in all trees could be
1237
# marked as 'deferred', and then returned in the final cleanup loop.
1238
# For now, I think it is "nicer" to return things as close to the
1239
# "master_tree" order as we can.
1241
def __init__(self, master_tree, other_trees):
1242
"""Create a new MultiWalker.
1244
All trees being walked must implement "iter_entries_by_dir()", such
1245
that they yield (path, object) tuples, where that object will have a
1246
'.file_id' member, that can be used to check equality.
1248
:param master_tree: All trees will be 'slaved' to the master_tree such
1249
that nodes in master_tree will be used as 'first-pass' sync points.
1250
Any nodes that aren't in master_tree will be merged in a second
1252
:param other_trees: A list of other trees to walk simultaneously.
1254
self._master_tree = master_tree
1255
self._other_trees = other_trees
1257
# Keep track of any nodes that were properly processed just out of
1258
# order, that way we don't return them at the end, we don't have to
1259
# track *all* processed file_ids, just the out-of-order ones
1260
self._out_of_order_processed = set()
1263
def _step_one(iterator):
1264
"""Step an iter_entries_by_dir iterator.
1266
:return: (has_more, path, ie)
1267
If has_more is False, path and ie will be None.
1270
path, ie = iterator.next()
1271
except StopIteration:
1272
return False, None, None
1274
return True, path, ie
1277
def _cmp_path_by_dirblock(path1, path2):
1278
"""Compare two paths based on what directory they are in.
1280
This generates a sort order, such that all children of a directory are
1281
sorted together, and grandchildren are in the same order as the
1282
children appear. But all grandchildren come after all children.
1284
:param path1: first path
1285
:param path2: the second path
1286
:return: negative number if ``path1`` comes first,
1287
0 if paths are equal
1288
and a positive number if ``path2`` sorts first
1290
# Shortcut this special case
1293
# This is stolen from _dirstate_helpers_py.py, only switching it to
1294
# Unicode objects. Consider using encode_utf8() and then using the
1295
# optimized versions, or maybe writing optimized unicode versions.
1296
if not isinstance(path1, unicode):
1297
raise TypeError("'path1' must be a unicode string, not %s: %r"
1298
% (type(path1), path1))
1299
if not isinstance(path2, unicode):
1300
raise TypeError("'path2' must be a unicode string, not %s: %r"
1301
% (type(path2), path2))
1302
return cmp(MultiWalker._path_to_key(path1),
1303
MultiWalker._path_to_key(path2))
1306
def _path_to_key(path):
1307
dirname, basename = osutils.split(path)
1308
return (dirname.split(u'/'), basename)
1310
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1311
"""Lookup an inventory entry by file_id.
1313
This is called when an entry is missing in the normal order.
1314
Generally this is because a file was either renamed, or it was
1315
deleted/added. If the entry was found in the inventory and not in
1316
extra_entries, it will be added to self._out_of_order_processed
1318
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1319
should be filled with entries that were found before they were
1320
used. If file_id is present, it will be removed from the
1322
:param other_tree: The Tree to search, in case we didn't find the entry
1324
:param file_id: The file_id to look for
1325
:return: (path, ie) if found or (None, None) if not present.
1327
if file_id in extra_entries:
1328
return extra_entries.pop(file_id)
1329
# TODO: Is id2path better as the first call, or is
1330
# inventory[file_id] better as a first check?
1332
cur_path = other_tree.id2path(file_id)
1333
except errors.NoSuchId:
1335
if cur_path is None:
1338
self._out_of_order_processed.add(file_id)
1339
cur_ie = other_tree.inventory[file_id]
1340
return (cur_path, cur_ie)
1343
"""Match up the values in the different trees."""
1344
for result in self._walk_master_tree():
1346
self._finish_others()
1347
for result in self._walk_others():
1350
def _walk_master_tree(self):
1351
"""First pass, walk all trees in lock-step.
1353
When we are done, all nodes in the master_tree will have been
1354
processed. _other_walkers, _other_entries, and _others_extra will be
1355
set on 'self' for future processing.
1357
# This iterator has the most "inlining" done, because it tends to touch
1358
# every file in the tree, while the others only hit nodes that don't
1360
master_iterator = self._master_tree.iter_entries_by_dir()
1362
other_walkers = [other.iter_entries_by_dir()
1363
for other in self._other_trees]
1364
other_entries = [self._step_one(walker) for walker in other_walkers]
1365
# Track extra nodes in the other trees
1366
others_extra = [{} for i in xrange(len(self._other_trees))]
1368
master_has_more = True
1369
step_one = self._step_one
1370
lookup_by_file_id = self._lookup_by_file_id
1371
out_of_order_processed = self._out_of_order_processed
1373
while master_has_more:
1374
(master_has_more, path, master_ie) = step_one(master_iterator)
1375
if not master_has_more:
1378
file_id = master_ie.file_id
1380
other_values_append = other_values.append
1381
next_other_entries = []
1382
next_other_entries_append = next_other_entries.append
1383
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1384
if not other_has_more:
1385
other_values_append(lookup_by_file_id(
1386
others_extra[idx], self._other_trees[idx], file_id))
1387
next_other_entries_append((False, None, None))
1388
elif file_id == other_ie.file_id:
1389
# This is the critical code path, as most of the entries
1390
# should match between most trees.
1391
other_values_append((other_path, other_ie))
1392
next_other_entries_append(step_one(other_walkers[idx]))
1394
# This walker did not match, step it until it either
1395
# matches, or we know we are past the current walker.
1396
other_walker = other_walkers[idx]
1397
other_extra = others_extra[idx]
1398
while (other_has_more and
1399
self._cmp_path_by_dirblock(other_path, path) < 0):
1400
other_file_id = other_ie.file_id
1401
if other_file_id not in out_of_order_processed:
1402
other_extra[other_file_id] = (other_path, other_ie)
1403
other_has_more, other_path, other_ie = \
1404
step_one(other_walker)
1405
if other_has_more and other_ie.file_id == file_id:
1406
# We ended up walking to this point, match and step
1408
other_values_append((other_path, other_ie))
1409
other_has_more, other_path, other_ie = \
1410
step_one(other_walker)
1412
# This record isn't in the normal order, see if it
1414
other_values_append(lookup_by_file_id(
1415
other_extra, self._other_trees[idx], file_id))
1416
next_other_entries_append((other_has_more, other_path,
1418
other_entries = next_other_entries
1420
# We've matched all the walkers, yield this datapoint
1421
yield path, file_id, master_ie, other_values
1422
self._other_walkers = other_walkers
1423
self._other_entries = other_entries
1424
self._others_extra = others_extra
1426
def _finish_others(self):
1427
"""Finish walking the other iterators, so we get all entries."""
1428
for idx, info in enumerate(self._other_entries):
1429
other_extra = self._others_extra[idx]
1430
(other_has_more, other_path, other_ie) = info
1431
while other_has_more:
1432
other_file_id = other_ie.file_id
1433
if other_file_id not in self._out_of_order_processed:
1434
other_extra[other_file_id] = (other_path, other_ie)
1435
other_has_more, other_path, other_ie = \
1436
self._step_one(self._other_walkers[idx])
1437
del self._other_entries
1439
def _walk_others(self):
1440
"""Finish up by walking all the 'deferred' nodes."""
1441
# TODO: One alternative would be to grab all possible unprocessed
1442
# file_ids, and then sort by path, and then yield them. That
1443
# might ensure better ordering, in case a caller strictly
1444
# requires parents before children.
1445
for idx, other_extra in enumerate(self._others_extra):
1446
others = sorted(other_extra.itervalues(),
1447
key=lambda x: self._path_to_key(x[0]))
1448
for other_path, other_ie in others:
1449
file_id = other_ie.file_id
1450
# We don't need to check out_of_order_processed here, because
1451
# the lookup_by_file_id will be removing anything processed
1452
# from the extras cache
1453
other_extra.pop(file_id)
1454
other_values = [(None, None) for i in xrange(idx)]
1455
other_values.append((other_path, other_ie))
1456
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1457
alt_idx = alt_idx + idx + 1
1458
alt_extra = self._others_extra[alt_idx]
1459
alt_tree = self._other_trees[alt_idx]
1460
other_values.append(self._lookup_by_file_id(
1461
alt_extra, alt_tree, file_id))
1462
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
