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
22
from cStringIO import StringIO
26
conflicts as _mod_conflicts,
31
revision as _mod_revision,
35
from bzrlib.decorators import needs_read_lock
36
from bzrlib.errors import BzrError, BzrCheckError, NoSuchId
37
from bzrlib import errors
38
from bzrlib.inventory import Inventory, InventoryFile
39
from bzrlib.inter import InterObject
40
from bzrlib.osutils import fingerprint_file
41
import bzrlib.revision
42
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
43
from bzrlib.trace import mutter, note
47
"""Abstract file tree.
49
There are several subclasses:
51
* `WorkingTree` exists as files on disk editable by the user.
53
* `RevisionTree` is a tree as recorded at some point in the past.
55
Trees contain an `Inventory` object, and also know how to retrieve
56
file texts mentioned in the inventory, either from a working
57
directory or from a store.
59
It is possible for trees to contain files that are not described
60
in their inventory or vice versa; for this use `filenames()`.
62
Trees can be compared, etc, regardless of whether they are working
63
trees or versioned trees.
66
def changes_from(self, other, want_unchanged=False, specific_files=None,
67
extra_trees=None, require_versioned=False, include_root=False,
68
want_unversioned=False):
69
"""Return a TreeDelta of the changes from other to this tree.
71
:param other: A tree to compare with.
72
:param specific_files: An optional list of file paths to restrict the
73
comparison to. When mapping filenames to ids, all matches in all
74
trees (including optional extra_trees) are used, and all children of
75
matched directories are included.
76
:param want_unchanged: An optional boolean requesting the inclusion of
77
unchanged entries in the result.
78
:param extra_trees: An optional list of additional trees to use when
79
mapping the contents of specific_files (paths) to file_ids.
80
:param require_versioned: An optional boolean (defaults to False). When
81
supplied and True all the 'specific_files' must be versioned, or
82
a PathsNotVersionedError will be thrown.
83
:param want_unversioned: Scan for unversioned paths.
85
The comparison will be performed by an InterTree object looked up on
88
# Martin observes that Tree.changes_from returns a TreeDelta and this
89
# may confuse people, because the class name of the returned object is
90
# a synonym of the object referenced in the method name.
91
return InterTree.get(other, self).compare(
92
want_unchanged=want_unchanged,
93
specific_files=specific_files,
94
extra_trees=extra_trees,
95
require_versioned=require_versioned,
96
include_root=include_root,
97
want_unversioned=want_unversioned,
100
def iter_changes(self, from_tree, include_unchanged=False,
101
specific_files=None, pb=None, extra_trees=None,
102
require_versioned=True, want_unversioned=False):
103
intertree = InterTree.get(from_tree, self)
104
return intertree.iter_changes(include_unchanged, specific_files, pb,
105
extra_trees, require_versioned, want_unversioned=want_unversioned)
109
def has_changes(self, from_tree):
110
"""Quickly check that the tree contains at least one change.
112
:return: True if a change is found. False otherwise
114
changes = self.iter_changes(from_tree)
116
change = changes.next()
117
# Exclude root (talk about black magic... --vila 20090629)
118
if change[4] == (None, None):
119
change = changes.next()
121
except StopIteration:
126
"""Get a list of the conflicts in the tree.
128
Each conflict is an instance of bzrlib.conflicts.Conflict.
130
return _mod_conflicts.ConflictList()
133
"""For trees that can have unversioned files, return all such paths."""
136
def get_parent_ids(self):
137
"""Get the parent ids for this tree.
139
:return: a list of parent ids. [] is returned to indicate
140
a tree with no parents.
141
:raises: BzrError if the parents are not known.
143
raise NotImplementedError(self.get_parent_ids)
145
def has_filename(self, filename):
146
"""True if the tree has given filename."""
147
raise NotImplementedError(self.has_filename)
149
def has_id(self, file_id):
150
return self.inventory.has_id(file_id)
152
def __contains__(self, file_id):
153
return self.has_id(file_id)
155
def has_or_had_id(self, file_id):
156
if file_id == self.inventory.root.file_id:
158
return self.inventory.has_id(file_id)
160
def is_ignored(self, filename):
161
"""Check whether the filename is ignored by this tree.
163
:param filename: The relative filename within the tree.
164
:return: True if the filename is ignored.
169
return iter(self.inventory)
171
def all_file_ids(self):
172
"""Iterate through all file ids, including ids for missing files."""
173
return set(self.inventory)
175
def id2path(self, file_id):
176
"""Return the path for a file id.
180
return self.inventory.id2path(file_id)
182
def is_control_filename(self, filename):
183
"""True if filename is the name of a control file in this tree.
185
:param filename: A filename within the tree. This is a relative path
186
from the root of this tree.
188
This is true IF and ONLY IF the filename is part of the meta data
189
that bzr controls in this tree. I.E. a random .bzr directory placed
190
on disk will not be a control file for this tree.
192
return self.bzrdir.is_control_filename(filename)
195
def iter_entries_by_dir(self, specific_file_ids=None):
196
"""Walk the tree in 'by_dir' order.
198
This will yield each entry in the tree as a (path, entry) tuple.
199
The order that they are yielded is:
201
Directories are walked in a depth-first lexicographical order,
202
however, whenever a directory is reached, all of its direct child
203
nodes are yielded in lexicographical order before yielding the
206
For example, in the tree::
216
The yield order (ignoring root) would be::
217
a, f, a/b, a/d, a/b/c, a/d/e, f/g
219
return self.inventory.iter_entries_by_dir(
220
specific_file_ids=specific_file_ids)
222
def iter_references(self):
223
if self.supports_tree_reference():
224
for path, entry in self.iter_entries_by_dir():
225
if entry.kind == 'tree-reference':
226
yield path, entry.file_id
228
def kind(self, file_id):
229
raise NotImplementedError("Tree subclass %s must implement kind"
230
% self.__class__.__name__)
232
def stored_kind(self, file_id):
233
"""File kind stored for this file_id.
235
May not match kind on disk for working trees. Always available
236
for versioned files, even when the file itself is missing.
238
return self.kind(file_id)
240
def path_content_summary(self, path):
241
"""Get a summary of the information about path.
243
:param path: A relative path within the tree.
244
:return: A tuple containing kind, size, exec, sha1-or-link.
245
Kind is always present (see tree.kind()).
246
size is present if kind is file, None otherwise.
247
exec is None unless kind is file and the platform supports the 'x'
249
sha1-or-link is the link target if kind is symlink, or the sha1 if
250
it can be obtained without reading the file.
252
raise NotImplementedError(self.path_content_summary)
254
def get_reference_revision(self, file_id, path=None):
255
raise NotImplementedError("Tree subclass %s must implement "
256
"get_reference_revision"
257
% self.__class__.__name__)
259
def _comparison_data(self, entry, path):
260
"""Return a tuple of kind, executable, stat_value for a file.
262
entry may be None if there is no inventory entry for the file, but
263
path must always be supplied.
265
kind is None if there is no file present (even if an inventory id is
266
present). executable is False for non-file entries.
268
raise NotImplementedError(self._comparison_data)
270
def _file_size(self, entry, stat_value):
271
raise NotImplementedError(self._file_size)
273
def _get_inventory(self):
274
return self._inventory
276
def get_file(self, file_id, path=None):
277
"""Return a file object for the file file_id in the tree.
279
If both file_id and path are defined, it is implementation defined as
280
to which one is used.
282
raise NotImplementedError(self.get_file)
284
def get_file_with_stat(self, file_id, path=None):
285
"""Get a file handle and stat object for file_id.
287
The default implementation returns (self.get_file, None) for backwards
290
:param file_id: The file id to read.
291
:param path: The path of the file, if it is known.
292
:return: A tuple (file_handle, stat_value_or_None). If the tree has
293
no stat facility, or need for a stat cache feedback during commit,
294
it may return None for the second element of the tuple.
296
return (self.get_file(file_id, path), None)
298
def get_file_text(self, file_id, path=None):
299
"""Return the byte content of a file.
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
my_file = self.get_file(file_id, path)
308
return my_file.read()
312
def get_file_lines(self, file_id, path=None):
313
"""Return the content of a file, as lines.
315
:param file_id: The file_id of the file.
316
:param path: The path of the file.
317
If both file_id and path are supplied, an implementation may use
320
return osutils.split_lines(self.get_file_text(file_id, path))
322
def get_file_mtime(self, file_id, path=None):
323
"""Return the modification time for a file.
325
:param file_id: The handle for this file.
326
:param path: The path that this file can be found at.
327
These must point to the same object.
329
raise NotImplementedError(self.get_file_mtime)
331
def get_file_size(self, file_id):
332
"""Return the size of a file in bytes.
334
This applies only to regular files. If invoked on directories or
335
symlinks, it will return None.
336
:param file_id: The file-id of the file
338
raise NotImplementedError(self.get_file_size)
340
def get_file_by_path(self, path):
341
return self.get_file(self._inventory.path2id(path), path)
343
def iter_files_bytes(self, desired_files):
344
"""Iterate through file contents.
346
Files will not necessarily be returned in the order they occur in
347
desired_files. No specific order is guaranteed.
349
Yields pairs of identifier, bytes_iterator. identifier is an opaque
350
value supplied by the caller as part of desired_files. It should
351
uniquely identify the file version in the caller's context. (Examples:
352
an index number or a TreeTransform trans_id.)
354
bytes_iterator is an iterable of bytestrings for the file. The
355
kind of iterable and length of the bytestrings are unspecified, but for
356
this implementation, it is a tuple containing a single bytestring with
357
the complete text of the file.
359
:param desired_files: a list of (file_id, identifier) pairs
361
for file_id, identifier in desired_files:
362
# We wrap the string in a tuple so that we can return an iterable
363
# of bytestrings. (Technically, a bytestring is also an iterable
364
# of bytestrings, but iterating through each character is not
366
cur_file = (self.get_file_text(file_id),)
367
yield identifier, cur_file
369
def get_symlink_target(self, file_id):
370
"""Get the target for a given file_id.
372
It is assumed that the caller already knows that file_id is referencing
374
:param file_id: Handle for the symlink entry.
375
:return: The path the symlink points to.
377
raise NotImplementedError(self.get_symlink_target)
379
def get_canonical_inventory_paths(self, paths):
380
"""Like get_canonical_inventory_path() but works on multiple items.
382
:param paths: A sequence of paths relative to the root of the tree.
383
:return: A list of paths, with each item the corresponding input path
384
adjusted to account for existing elements that match case
387
return list(self._yield_canonical_inventory_paths(paths))
389
def get_canonical_inventory_path(self, path):
390
"""Returns the first inventory item that case-insensitively matches path.
392
If a path matches exactly, it is returned. If no path matches exactly
393
but more than one path matches case-insensitively, it is implementation
394
defined which is returned.
396
If no path matches case-insensitively, the input path is returned, but
397
with as many path entries that do exist changed to their canonical
400
If you need to resolve many names from the same tree, you should
401
use get_canonical_inventory_paths() to avoid O(N) behaviour.
403
:param path: A paths relative to the root of the tree.
404
:return: The input path adjusted to account for existing elements
405
that match case insensitively.
407
return self._yield_canonical_inventory_paths([path]).next()
409
def _yield_canonical_inventory_paths(self, paths):
411
# First, if the path as specified exists exactly, just use it.
412
if self.path2id(path) is not None:
416
cur_id = self.get_root_id()
418
bit_iter = iter(path.split("/"))
421
for child in self.iter_children(cur_id):
423
child_base = os.path.basename(self.id2path(child))
424
if child_base.lower() == lelt:
426
cur_path = osutils.pathjoin(cur_path, child_base)
429
# before a change is committed we can see this error...
432
# got to the end of this directory and no entries matched.
433
# Return what matched so far, plus the rest as specified.
434
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
439
def get_root_id(self):
440
"""Return the file_id for the root of this tree."""
441
raise NotImplementedError(self.get_root_id)
443
def annotate_iter(self, file_id,
444
default_revision=_mod_revision.CURRENT_REVISION):
445
"""Return an iterator of revision_id, line tuples.
447
For working trees (and mutable trees in general), the special
448
revision_id 'current:' will be used for lines that are new in this
449
tree, e.g. uncommitted changes.
450
:param file_id: The file to produce an annotated version from
451
:param default_revision: For lines that don't match a basis, mark them
452
with this revision id. Not all implementations will make use of
455
raise NotImplementedError(self.annotate_iter)
457
def _get_plan_merge_data(self, file_id, other, base):
458
from bzrlib import merge, versionedfile
459
vf = versionedfile._PlanMergeVersionedFile(file_id)
460
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
461
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
463
last_revision_base = None
465
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
466
return vf, last_revision_a, last_revision_b, last_revision_base
468
def plan_file_merge(self, file_id, other, base=None):
469
"""Generate a merge plan based on annotations.
471
If the file contains uncommitted changes in this tree, they will be
472
attributed to the 'current:' pseudo-revision. If the file contains
473
uncommitted changes in the other tree, they will be assigned to the
474
'other:' pseudo-revision.
476
data = self._get_plan_merge_data(file_id, other, base)
477
vf, last_revision_a, last_revision_b, last_revision_base = data
478
return vf.plan_merge(last_revision_a, last_revision_b,
481
def plan_file_lca_merge(self, file_id, other, base=None):
482
"""Generate a merge plan based lca-newness.
484
If the file contains uncommitted changes in this tree, they will be
485
attributed to the 'current:' pseudo-revision. If the file contains
486
uncommitted changes in the other tree, they will be assigned to the
487
'other:' pseudo-revision.
489
data = self._get_plan_merge_data(file_id, other, base)
490
vf, last_revision_a, last_revision_b, last_revision_base = data
491
return vf.plan_lca_merge(last_revision_a, last_revision_b,
494
def _iter_parent_trees(self):
495
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
496
for revision_id in self.get_parent_ids():
498
yield self.revision_tree(revision_id)
499
except errors.NoSuchRevisionInTree:
500
yield self.repository.revision_tree(revision_id)
503
def _file_revision(revision_tree, file_id):
504
"""Determine the revision associated with a file in a given tree."""
505
revision_tree.lock_read()
507
return revision_tree.inventory[file_id].revision
509
revision_tree.unlock()
511
def _get_file_revision(self, file_id, vf, tree_revision):
512
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
514
if getattr(self, '_repository', None) is None:
515
last_revision = tree_revision
516
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
517
self._iter_parent_trees()]
518
vf.add_lines((file_id, last_revision), parent_keys,
519
self.get_file(file_id).readlines())
520
repo = self.branch.repository
523
last_revision = self._file_revision(self, file_id)
524
base_vf = self._repository.texts
525
if base_vf not in vf.fallback_versionedfiles:
526
vf.fallback_versionedfiles.append(base_vf)
529
inventory = property(_get_inventory,
530
doc="Inventory of this Tree")
532
def _check_retrieved(self, ie, f):
535
fp = fingerprint_file(f)
538
if ie.text_size is not None:
539
if ie.text_size != fp['size']:
540
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
541
["inventory expects %d bytes" % ie.text_size,
542
"file is actually %d bytes" % fp['size'],
543
"store is probably damaged/corrupt"])
545
if ie.text_sha1 != fp['sha1']:
546
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
547
["inventory expects %s" % ie.text_sha1,
548
"file is actually %s" % fp['sha1'],
549
"store is probably damaged/corrupt"])
552
def path2id(self, path):
553
"""Return the id for path in this tree."""
554
return self._inventory.path2id(path)
556
def paths2ids(self, paths, trees=[], require_versioned=True):
557
"""Return all the ids that can be reached by walking from paths.
559
Each path is looked up in this tree and any extras provided in
560
trees, and this is repeated recursively: the children in an extra tree
561
of a directory that has been renamed under a provided path in this tree
562
are all returned, even if none exist under a provided path in this
563
tree, and vice versa.
565
:param paths: An iterable of paths to start converting to ids from.
566
Alternatively, if paths is None, no ids should be calculated and None
567
will be returned. This is offered to make calling the api unconditional
568
for code that *might* take a list of files.
569
:param trees: Additional trees to consider.
570
:param require_versioned: If False, do not raise NotVersionedError if
571
an element of paths is not versioned in this tree and all of trees.
573
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
575
def iter_children(self, file_id):
576
entry = self.iter_entries_by_dir([file_id]).next()[1]
577
for child in getattr(entry, 'children', {}).itervalues():
583
def revision_tree(self, revision_id):
584
"""Obtain a revision tree for the revision revision_id.
586
The intention of this method is to allow access to possibly cached
587
tree data. Implementors of this method should raise NoSuchRevision if
588
the tree is not locally available, even if they could obtain the
589
tree via a repository or some other means. Callers are responsible
590
for finding the ultimate source for a revision tree.
592
:param revision_id: The revision_id of the requested tree.
594
:raises: NoSuchRevision if the tree cannot be obtained.
596
raise errors.NoSuchRevisionInTree(self, revision_id)
599
"""What files are present in this tree and unknown.
601
:return: an iterator over the unknown files.
608
def filter_unversioned_files(self, paths):
609
"""Filter out paths that are versioned.
611
:return: set of paths.
613
# NB: we specifically *don't* call self.has_filename, because for
614
# WorkingTrees that can indicate files that exist on disk but that
616
pred = self.inventory.has_filename
617
return set((p for p in paths if not pred(p)))
619
def walkdirs(self, prefix=""):
620
"""Walk the contents of this tree from path down.
622
This yields all the data about the contents of a directory at a time.
623
After each directory has been yielded, if the caller has mutated the
624
list to exclude some directories, they are then not descended into.
626
The data yielded is of the form:
627
((directory-relpath, directory-path-from-root, directory-fileid),
628
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
629
versioned_kind), ...]),
630
- directory-relpath is the containing dirs relpath from prefix
631
- directory-path-from-root is the containing dirs path from /
632
- directory-fileid is the id of the directory if it is versioned.
633
- relpath is the relative path within the subtree being walked.
634
- basename is the basename
635
- kind is the kind of the file now. If unknonwn then the file is not
636
present within the tree - but it may be recorded as versioned. See
638
- lstat is the stat data *if* the file was statted.
639
- path_from_tree_root is the path from the root of the tree.
640
- file_id is the file_id if the entry is versioned.
641
- versioned_kind is the kind of the file as last recorded in the
642
versioning system. If 'unknown' the file is not versioned.
643
One of 'kind' and 'versioned_kind' must not be 'unknown'.
645
:param prefix: Start walking from prefix within the tree rather than
646
at the root. This allows one to walk a subtree but get paths that are
647
relative to a tree rooted higher up.
648
:return: an iterator over the directory data.
650
raise NotImplementedError(self.walkdirs)
652
def supports_content_filtering(self):
655
def _content_filter_stack(self, path=None, file_id=None):
656
"""The stack of content filters for a path if filtering is supported.
658
Readers will be applied in first-to-last order.
659
Writers will be applied in last-to-first order.
660
Either the path or the file-id needs to be provided.
662
:param path: path relative to the root of the tree
664
:param file_id: file_id or None if unknown
665
:return: the list of filters - [] if there are none
667
filter_pref_names = filters._get_registered_names()
668
if len(filter_pref_names) == 0:
671
path = self.id2path(file_id)
672
prefs = self.iter_search_rules([path], filter_pref_names).next()
673
stk = filters._get_filter_stack_for(prefs)
674
if 'filters' in debug.debug_flags:
675
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
678
def _content_filter_stack_provider(self):
679
"""A function that returns a stack of ContentFilters.
681
The function takes a path (relative to the top of the tree) and a
682
file-id as parameters.
684
:return: None if content filtering is not supported by this tree.
686
if self.supports_content_filtering():
687
return lambda path, file_id: \
688
self._content_filter_stack(path, file_id)
692
def iter_search_rules(self, path_names, pref_names=None,
693
_default_searcher=None):
694
"""Find the preferences for filenames in a tree.
696
:param path_names: an iterable of paths to find attributes for.
697
Paths are given relative to the root of the tree.
698
:param pref_names: the list of preferences to lookup - None for all
699
:param _default_searcher: private parameter to assist testing - don't use
700
:return: an iterator of tuple sequences, one per path-name.
701
See _RulesSearcher.get_items for details on the tuple sequence.
703
if _default_searcher is None:
704
_default_searcher = rules._per_user_searcher
705
searcher = self._get_rules_searcher(_default_searcher)
706
if searcher is not None:
707
if pref_names is not None:
708
for path in path_names:
709
yield searcher.get_selected_items(path, pref_names)
711
for path in path_names:
712
yield searcher.get_items(path)
715
def _get_rules_searcher(self, default_searcher):
716
"""Get the RulesSearcher for this tree given the default one."""
717
searcher = default_searcher
721
######################################################################
724
# TODO: Merge these two functions into a single one that can operate
725
# on either a whole tree or a set of files.
727
# TODO: Return the diff in order by filename, not by category or in
728
# random order. Can probably be done by lock-stepping through the
729
# filenames from both trees.
732
def file_status(filename, old_tree, new_tree):
733
"""Return single-letter status, old and new names for a file.
735
The complexity here is in deciding how to represent renames;
736
many complex cases are possible.
738
old_inv = old_tree.inventory
739
new_inv = new_tree.inventory
740
new_id = new_inv.path2id(filename)
741
old_id = old_inv.path2id(filename)
743
if not new_id and not old_id:
744
# easy: doesn't exist in either; not versioned at all
745
if new_tree.is_ignored(filename):
746
return 'I', None, None
748
return '?', None, None
750
# There is now a file of this name, great.
753
# There is no longer a file of this name, but we can describe
754
# what happened to the file that used to have
755
# this name. There are two possibilities: either it was
756
# deleted entirely, or renamed.
757
if new_inv.has_id(old_id):
758
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
760
return 'D', old_inv.id2path(old_id), None
762
# if the file_id is new in this revision, it is added
763
if new_id and not old_inv.has_id(new_id):
766
# if there used to be a file of this name, but that ID has now
767
# disappeared, it is deleted
768
if old_id and not new_inv.has_id(old_id):
774
@deprecated_function(deprecated_in((1, 9, 0)))
775
def find_renames(old_inv, new_inv):
776
for file_id in old_inv:
777
if file_id not in new_inv:
779
old_name = old_inv.id2path(file_id)
780
new_name = new_inv.id2path(file_id)
781
if old_name != new_name:
782
yield (old_name, new_name)
785
def find_ids_across_trees(filenames, trees, require_versioned=True):
786
"""Find the ids corresponding to specified filenames.
788
All matches in all trees will be used, and all children of matched
789
directories will be used.
791
:param filenames: The filenames to find file_ids for (if None, returns
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 and their children.
800
specified_path_ids = _find_ids_across_trees(filenames, trees,
802
return _find_children_across_trees(specified_path_ids, trees)
805
def _find_ids_across_trees(filenames, trees, require_versioned):
806
"""Find the ids corresponding to specified filenames.
808
All matches in all trees will be used, but subdirectories are not scanned.
810
:param filenames: The filenames to find file_ids for
811
:param trees: The trees to find file_ids within
812
:param require_versioned: if true, all specified filenames must occur in
814
:return: a set of file ids for the specified filenames
817
interesting_ids = set()
818
for tree_path in filenames:
821
file_id = tree.path2id(tree_path)
822
if file_id is not None:
823
interesting_ids.add(file_id)
826
not_versioned.append(tree_path)
827
if len(not_versioned) > 0 and require_versioned:
828
raise errors.PathsNotVersionedError(not_versioned)
829
return interesting_ids
832
def _find_children_across_trees(specified_ids, trees):
833
"""Return a set including specified ids and their children.
835
All matches in all trees will be used.
837
:param trees: The trees to find file_ids within
838
:return: a set containing all specified ids and their children
840
interesting_ids = set(specified_ids)
841
pending = interesting_ids
842
# now handle children of interesting ids
843
# we loop so that we handle all children of each id in both trees
844
while len(pending) > 0:
846
for file_id in pending:
848
if not tree.has_id(file_id):
850
for child_id in tree.iter_children(file_id):
851
if child_id not in interesting_ids:
852
new_pending.add(child_id)
853
interesting_ids.update(new_pending)
854
pending = new_pending
855
return interesting_ids
858
class InterTree(InterObject):
859
"""This class represents operations taking place between two Trees.
861
Its instances have methods like 'compare' and contain references to the
862
source and target trees these operations are to be carried out on.
864
Clients of bzrlib should not need to use InterTree directly, rather they
865
should use the convenience methods on Tree such as 'Tree.compare()' which
866
will pass through to InterTree as appropriate.
872
def compare(self, want_unchanged=False, specific_files=None,
873
extra_trees=None, require_versioned=False, include_root=False,
874
want_unversioned=False):
875
"""Return the changes from source to target.
877
:return: A TreeDelta.
878
:param specific_files: An optional list of file paths to restrict the
879
comparison to. When mapping filenames to ids, all matches in all
880
trees (including optional extra_trees) are used, and all children of
881
matched directories are included.
882
:param want_unchanged: An optional boolean requesting the inclusion of
883
unchanged entries in the result.
884
:param extra_trees: An optional list of additional trees to use when
885
mapping the contents of specific_files (paths) to file_ids.
886
:param require_versioned: An optional boolean (defaults to False). When
887
supplied and True all the 'specific_files' must be versioned, or
888
a PathsNotVersionedError will be thrown.
889
:param want_unversioned: Scan for unversioned paths.
891
# NB: show_status depends on being able to pass in non-versioned files
892
# and report them as unknown
893
trees = (self.source,)
894
if extra_trees is not None:
895
trees = trees + tuple(extra_trees)
896
# target is usually the newer tree:
897
specific_file_ids = self.target.paths2ids(specific_files, trees,
898
require_versioned=require_versioned)
899
if specific_files and not specific_file_ids:
900
# All files are unversioned, so just return an empty delta
901
# _compare_trees would think we want a complete delta
902
result = delta.TreeDelta()
903
fake_entry = InventoryFile('unused', 'unused', 'unused')
904
result.unversioned = [(path, None,
905
self.target._comparison_data(fake_entry, path)[0]) for path in
908
return delta._compare_trees(self.source, self.target, want_unchanged,
909
specific_files, include_root, extra_trees=extra_trees,
910
require_versioned=require_versioned,
911
want_unversioned=want_unversioned)
913
def iter_changes(self, include_unchanged=False,
914
specific_files=None, pb=None, extra_trees=[],
915
require_versioned=True, want_unversioned=False):
916
"""Generate an iterator of changes between trees.
919
(file_id, (path_in_source, path_in_target),
920
changed_content, versioned, parent, name, kind,
923
Changed_content is True if the file's content has changed. This
924
includes changes to its kind, and to a symlink's target.
926
versioned, parent, name, kind, executable are tuples of (from, to).
927
If a file is missing in a tree, its kind is None.
929
Iteration is done in parent-to-child order, relative to the target
932
There is no guarantee that all paths are in sorted order: the
933
requirement to expand the search due to renames may result in children
934
that should be found early being found late in the search, after
935
lexically later results have been returned.
936
:param require_versioned: Raise errors.PathsNotVersionedError if a
937
path in the specific_files list is not versioned in one of
938
source, target or extra_trees.
939
:param want_unversioned: Should unversioned files be returned in the
940
output. An unversioned file is defined as one with (False, False)
941
for the versioned pair.
943
lookup_trees = [self.source]
945
lookup_trees.extend(extra_trees)
946
if specific_files == []:
947
specific_file_ids = []
949
specific_file_ids = self.target.paths2ids(specific_files,
950
lookup_trees, require_versioned=require_versioned)
952
all_unversioned = sorted([(p.split('/'), p) for p in
954
if specific_files is None or
955
osutils.is_inside_any(specific_files, p)])
956
all_unversioned = deque(all_unversioned)
958
all_unversioned = deque()
960
from_entries_by_dir = list(self.source.iter_entries_by_dir(
961
specific_file_ids=specific_file_ids))
962
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
963
to_entries_by_dir = list(self.target.iter_entries_by_dir(
964
specific_file_ids=specific_file_ids))
965
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
967
# the unversioned path lookup only occurs on real trees - where there
968
# can be extras. So the fake_entry is solely used to look up
969
# executable it values when execute is not supported.
970
fake_entry = InventoryFile('unused', 'unused', 'unused')
971
for to_path, to_entry in to_entries_by_dir:
972
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
973
unversioned_path = all_unversioned.popleft()
974
to_kind, to_executable, to_stat = \
975
self.target._comparison_data(fake_entry, unversioned_path[1])
976
yield (None, (None, unversioned_path[1]), True, (False, False),
978
(None, unversioned_path[0][-1]),
980
(None, to_executable))
981
file_id = to_entry.file_id
982
to_paths[file_id] = to_path
984
changed_content = False
985
from_path, from_entry = from_data.get(file_id, (None, None))
986
from_versioned = (from_entry is not None)
987
if from_entry is not None:
988
from_versioned = True
989
from_name = from_entry.name
990
from_parent = from_entry.parent_id
991
from_kind, from_executable, from_stat = \
992
self.source._comparison_data(from_entry, from_path)
995
from_versioned = False
999
from_executable = None
1000
versioned = (from_versioned, True)
1001
to_kind, to_executable, to_stat = \
1002
self.target._comparison_data(to_entry, to_path)
1003
kind = (from_kind, to_kind)
1004
if kind[0] != kind[1]:
1005
changed_content = True
1006
elif from_kind == 'file':
1007
if (self.source.get_file_sha1(file_id, from_path, from_stat) !=
1008
self.target.get_file_sha1(file_id, to_path, to_stat)):
1009
changed_content = True
1010
elif from_kind == 'symlink':
1011
if (self.source.get_symlink_target(file_id) !=
1012
self.target.get_symlink_target(file_id)):
1013
changed_content = True
1014
# XXX: Yes, the indentation below is wrong. But fixing it broke
1015
# test_merge.TestMergerEntriesLCAOnDisk.
1016
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
1017
# the fix from bzr.dev -- vila 2009026
1018
elif from_kind == 'tree-reference':
1019
if (self.source.get_reference_revision(file_id, from_path)
1020
!= self.target.get_reference_revision(file_id, to_path)):
1021
changed_content = True
1022
parent = (from_parent, to_entry.parent_id)
1023
name = (from_name, to_entry.name)
1024
executable = (from_executable, to_executable)
1026
pb.update('comparing files', entry_count, num_entries)
1027
if (changed_content is not False or versioned[0] != versioned[1]
1028
or parent[0] != parent[1] or name[0] != name[1] or
1029
executable[0] != executable[1] or include_unchanged):
1030
yield (file_id, (from_path, to_path), changed_content,
1031
versioned, parent, name, kind, executable)
1033
while all_unversioned:
1034
# yield any trailing unversioned paths
1035
unversioned_path = all_unversioned.popleft()
1036
to_kind, to_executable, to_stat = \
1037
self.target._comparison_data(fake_entry, unversioned_path[1])
1038
yield (None, (None, unversioned_path[1]), True, (False, False),
1040
(None, unversioned_path[0][-1]),
1042
(None, to_executable))
1044
def get_to_path(to_entry):
1045
if to_entry.parent_id is None:
1046
to_path = '' # the root
1048
if to_entry.parent_id not in to_paths:
1050
return get_to_path(self.target.inventory[to_entry.parent_id])
1051
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
1053
to_paths[to_entry.file_id] = to_path
1056
for path, from_entry in from_entries_by_dir:
1057
file_id = from_entry.file_id
1058
if file_id in to_paths:
1061
if not file_id in self.target.all_file_ids():
1062
# common case - paths we have not emitted are not present in
1066
to_path = get_to_path(self.target.inventory[file_id])
1069
pb.update('comparing files', entry_count, num_entries)
1070
versioned = (True, False)
1071
parent = (from_entry.parent_id, None)
1072
name = (from_entry.name, None)
1073
from_kind, from_executable, stat_value = \
1074
self.source._comparison_data(from_entry, path)
1075
kind = (from_kind, None)
1076
executable = (from_executable, None)
1077
changed_content = from_kind is not None
1078
# the parent's path is necessarily known at this point.
1079
yield(file_id, (path, to_path), changed_content, versioned, parent,
1080
name, kind, executable)
1083
class MultiWalker(object):
1084
"""Walk multiple trees simultaneously, getting combined results."""
1086
# Note: This could be written to not assume you can do out-of-order
1087
# lookups. Instead any nodes that don't match in all trees could be
1088
# marked as 'deferred', and then returned in the final cleanup loop.
1089
# For now, I think it is "nicer" to return things as close to the
1090
# "master_tree" order as we can.
1092
def __init__(self, master_tree, other_trees):
1093
"""Create a new MultiWalker.
1095
All trees being walked must implement "iter_entries_by_dir()", such
1096
that they yield (path, object) tuples, where that object will have a
1097
'.file_id' member, that can be used to check equality.
1099
:param master_tree: All trees will be 'slaved' to the master_tree such
1100
that nodes in master_tree will be used as 'first-pass' sync points.
1101
Any nodes that aren't in master_tree will be merged in a second
1103
:param other_trees: A list of other trees to walk simultaneously.
1105
self._master_tree = master_tree
1106
self._other_trees = other_trees
1108
# Keep track of any nodes that were properly processed just out of
1109
# order, that way we don't return them at the end, we don't have to
1110
# track *all* processed file_ids, just the out-of-order ones
1111
self._out_of_order_processed = set()
1114
def _step_one(iterator):
1115
"""Step an iter_entries_by_dir iterator.
1117
:return: (has_more, path, ie)
1118
If has_more is False, path and ie will be None.
1121
path, ie = iterator.next()
1122
except StopIteration:
1123
return False, None, None
1125
return True, path, ie
1128
def _cmp_path_by_dirblock(path1, path2):
1129
"""Compare two paths based on what directory they are in.
1131
This generates a sort order, such that all children of a directory are
1132
sorted together, and grandchildren are in the same order as the
1133
children appear. But all grandchildren come after all children.
1135
:param path1: first path
1136
:param path2: the second path
1137
:return: negative number if ``path1`` comes first,
1138
0 if paths are equal
1139
and a positive number if ``path2`` sorts first
1141
# Shortcut this special case
1144
# This is stolen from _dirstate_helpers_py.py, only switching it to
1145
# Unicode objects. Consider using encode_utf8() and then using the
1146
# optimized versions, or maybe writing optimized unicode versions.
1147
if not isinstance(path1, unicode):
1148
raise TypeError("'path1' must be a unicode string, not %s: %r"
1149
% (type(path1), path1))
1150
if not isinstance(path2, unicode):
1151
raise TypeError("'path2' must be a unicode string, not %s: %r"
1152
% (type(path2), path2))
1153
return cmp(MultiWalker._path_to_key(path1),
1154
MultiWalker._path_to_key(path2))
1157
def _path_to_key(path):
1158
dirname, basename = osutils.split(path)
1159
return (dirname.split(u'/'), basename)
1161
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1162
"""Lookup an inventory entry by file_id.
1164
This is called when an entry is missing in the normal order.
1165
Generally this is because a file was either renamed, or it was
1166
deleted/added. If the entry was found in the inventory and not in
1167
extra_entries, it will be added to self._out_of_order_processed
1169
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1170
should be filled with entries that were found before they were
1171
used. If file_id is present, it will be removed from the
1173
:param other_tree: The Tree to search, in case we didn't find the entry
1175
:param file_id: The file_id to look for
1176
:return: (path, ie) if found or (None, None) if not present.
1178
if file_id in extra_entries:
1179
return extra_entries.pop(file_id)
1180
# TODO: Is id2path better as the first call, or is
1181
# inventory[file_id] better as a first check?
1183
cur_path = other_tree.id2path(file_id)
1184
except errors.NoSuchId:
1186
if cur_path is None:
1189
self._out_of_order_processed.add(file_id)
1190
cur_ie = other_tree.inventory[file_id]
1191
return (cur_path, cur_ie)
1194
"""Match up the values in the different trees."""
1195
for result in self._walk_master_tree():
1197
self._finish_others()
1198
for result in self._walk_others():
1201
def _walk_master_tree(self):
1202
"""First pass, walk all trees in lock-step.
1204
When we are done, all nodes in the master_tree will have been
1205
processed. _other_walkers, _other_entries, and _others_extra will be
1206
set on 'self' for future processing.
1208
# This iterator has the most "inlining" done, because it tends to touch
1209
# every file in the tree, while the others only hit nodes that don't
1211
master_iterator = self._master_tree.iter_entries_by_dir()
1213
other_walkers = [other.iter_entries_by_dir()
1214
for other in self._other_trees]
1215
other_entries = [self._step_one(walker) for walker in other_walkers]
1216
# Track extra nodes in the other trees
1217
others_extra = [{} for i in xrange(len(self._other_trees))]
1219
master_has_more = True
1220
step_one = self._step_one
1221
lookup_by_file_id = self._lookup_by_file_id
1222
out_of_order_processed = self._out_of_order_processed
1224
while master_has_more:
1225
(master_has_more, path, master_ie) = step_one(master_iterator)
1226
if not master_has_more:
1229
file_id = master_ie.file_id
1231
other_values_append = other_values.append
1232
next_other_entries = []
1233
next_other_entries_append = next_other_entries.append
1234
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1235
if not other_has_more:
1236
other_values_append(lookup_by_file_id(
1237
others_extra[idx], self._other_trees[idx], file_id))
1238
next_other_entries_append((False, None, None))
1239
elif file_id == other_ie.file_id:
1240
# This is the critical code path, as most of the entries
1241
# should match between most trees.
1242
other_values_append((other_path, other_ie))
1243
next_other_entries_append(step_one(other_walkers[idx]))
1245
# This walker did not match, step it until it either
1246
# matches, or we know we are past the current walker.
1247
other_walker = other_walkers[idx]
1248
other_extra = others_extra[idx]
1249
while (other_has_more and
1250
self._cmp_path_by_dirblock(other_path, path) < 0):
1251
other_file_id = other_ie.file_id
1252
if other_file_id not in out_of_order_processed:
1253
other_extra[other_file_id] = (other_path, other_ie)
1254
other_has_more, other_path, other_ie = \
1255
step_one(other_walker)
1256
if other_has_more and other_ie.file_id == file_id:
1257
# We ended up walking to this point, match and step
1259
other_values_append((other_path, other_ie))
1260
other_has_more, other_path, other_ie = \
1261
step_one(other_walker)
1263
# This record isn't in the normal order, see if it
1265
other_values_append(lookup_by_file_id(
1266
other_extra, self._other_trees[idx], file_id))
1267
next_other_entries_append((other_has_more, other_path,
1269
other_entries = next_other_entries
1271
# We've matched all the walkers, yield this datapoint
1272
yield path, file_id, master_ie, other_values
1273
self._other_walkers = other_walkers
1274
self._other_entries = other_entries
1275
self._others_extra = others_extra
1277
def _finish_others(self):
1278
"""Finish walking the other iterators, so we get all entries."""
1279
for idx, info in enumerate(self._other_entries):
1280
other_extra = self._others_extra[idx]
1281
(other_has_more, other_path, other_ie) = info
1282
while other_has_more:
1283
other_file_id = other_ie.file_id
1284
if other_file_id not in self._out_of_order_processed:
1285
other_extra[other_file_id] = (other_path, other_ie)
1286
other_has_more, other_path, other_ie = \
1287
self._step_one(self._other_walkers[idx])
1288
del self._other_entries
1290
def _walk_others(self):
1291
"""Finish up by walking all the 'deferred' nodes."""
1292
# TODO: One alternative would be to grab all possible unprocessed
1293
# file_ids, and then sort by path, and then yield them. That
1294
# might ensure better ordering, in case a caller strictly
1295
# requires parents before children.
1296
for idx, other_extra in enumerate(self._others_extra):
1297
others = sorted(other_extra.itervalues(),
1298
key=lambda x: self._path_to_key(x[0]))
1299
for other_path, other_ie in others:
1300
file_id = other_ie.file_id
1301
# We don't need to check out_of_order_processed here, because
1302
# the lookup_by_file_id will be removing anything processed
1303
# from the extras cache
1304
other_extra.pop(file_id)
1305
other_values = [(None, None) for i in xrange(idx)]
1306
other_values.append((other_path, other_ie))
1307
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1308
alt_idx = alt_idx + idx + 1
1309
alt_extra = self._others_extra[alt_idx]
1310
alt_tree = self._other_trees[alt_idx]
1311
other_values.append(self._lookup_by_file_id(
1312
alt_extra, alt_tree, file_id))
1313
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
