1
# Copyright (C) 2005, 2009 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Tree classes, representing directory at point in time.
21
from collections import deque
22
from cStringIO import StringIO
26
conflicts as _mod_conflicts,
29
revision as _mod_revision,
33
from bzrlib.decorators import needs_read_lock
34
from bzrlib.errors import BzrError, BzrCheckError, NoSuchId
35
from bzrlib import errors
36
from bzrlib.inventory import Inventory, 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 mutter, 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
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
99
def _iter_changes(self, *args, **kwargs):
100
return self.iter_changes(*args, **kwargs)
102
def iter_changes(self, from_tree, include_unchanged=False,
103
specific_files=None, pb=None, extra_trees=None,
104
require_versioned=True, want_unversioned=False):
105
intertree = InterTree.get(from_tree, self)
106
return intertree.iter_changes(include_unchanged, specific_files, pb,
107
extra_trees, require_versioned, want_unversioned=want_unversioned)
110
"""Get a list of the conflicts in the tree.
112
Each conflict is an instance of bzrlib.conflicts.Conflict.
114
return _mod_conflicts.ConflictList()
117
"""For trees that can have unversioned files, return all such paths."""
120
def get_parent_ids(self):
121
"""Get the parent ids for this tree.
123
:return: a list of parent ids. [] is returned to indicate
124
a tree with no parents.
125
:raises: BzrError if the parents are not known.
127
raise NotImplementedError(self.get_parent_ids)
129
def has_filename(self, filename):
130
"""True if the tree has given filename."""
131
raise NotImplementedError(self.has_filename)
133
def has_id(self, file_id):
134
return self.inventory.has_id(file_id)
136
def __contains__(self, file_id):
137
return self.has_id(file_id)
139
def has_or_had_id(self, file_id):
140
if file_id == self.inventory.root.file_id:
142
return self.inventory.has_id(file_id)
144
def is_ignored(self, filename):
145
"""Check whether the filename is ignored by this tree.
147
:param filename: The relative filename within the tree.
148
:return: True if the filename is ignored.
153
return iter(self.inventory)
155
def all_file_ids(self):
156
"""Iterate through all file ids, including ids for missing files."""
157
return set(self.inventory)
159
def id2path(self, file_id):
160
"""Return the path for a file id.
164
return self.inventory.id2path(file_id)
166
def is_control_filename(self, filename):
167
"""True if filename is the name of a control file in this tree.
169
:param filename: A filename within the tree. This is a relative path
170
from the root of this tree.
172
This is true IF and ONLY IF the filename is part of the meta data
173
that bzr controls in this tree. I.E. a random .bzr directory placed
174
on disk will not be a control file for this tree.
176
return self.bzrdir.is_control_filename(filename)
179
def iter_entries_by_dir(self, specific_file_ids=None):
180
"""Walk the tree in 'by_dir' order.
182
This will yield each entry in the tree as a (path, entry) tuple.
183
The order that they are yielded is:
185
Directories are walked in a depth-first lexicographical order,
186
however, whenever a directory is reached, all of its direct child
187
nodes are yielded in lexicographical order before yielding the
190
For example, in the tree::
200
The yield order (ignoring root) would be::
201
a, f, a/b, a/d, a/b/c, a/d/e, f/g
203
return self.inventory.iter_entries_by_dir(
204
specific_file_ids=specific_file_ids)
206
def iter_references(self):
207
for path, entry in self.iter_entries_by_dir():
208
if entry.kind == 'tree-reference':
209
yield path, entry.file_id
211
def kind(self, file_id):
212
raise NotImplementedError("Tree subclass %s must implement kind"
213
% self.__class__.__name__)
215
def stored_kind(self, file_id):
216
"""File kind stored for this file_id.
218
May not match kind on disk for working trees. Always available
219
for versioned files, even when the file itself is missing.
221
return self.kind(file_id)
223
def path_content_summary(self, path):
224
"""Get a summary of the information about path.
226
:param path: A relative path within the tree.
227
:return: A tuple containing kind, size, exec, sha1-or-link.
228
Kind is always present (see tree.kind()).
229
size is present if kind is file, None otherwise.
230
exec is None unless kind is file and the platform supports the 'x'
232
sha1-or-link is the link target if kind is symlink, or the sha1 if
233
it can be obtained without reading the file.
235
raise NotImplementedError(self.path_content_summary)
237
def get_reference_revision(self, file_id, path=None):
238
raise NotImplementedError("Tree subclass %s must implement "
239
"get_reference_revision"
240
% self.__class__.__name__)
242
def _comparison_data(self, entry, path):
243
"""Return a tuple of kind, executable, stat_value for a file.
245
entry may be None if there is no inventory entry for the file, but
246
path must always be supplied.
248
kind is None if there is no file present (even if an inventory id is
249
present). executable is False for non-file entries.
251
raise NotImplementedError(self._comparison_data)
253
def _file_size(self, entry, stat_value):
254
raise NotImplementedError(self._file_size)
256
def _get_inventory(self):
257
return self._inventory
259
def get_file(self, file_id, path=None):
260
"""Return a file object for the file file_id in the tree.
262
If both file_id and path are defined, it is implementation defined as
263
to which one is used.
265
raise NotImplementedError(self.get_file)
267
def get_file_text(self, file_id, path=None):
268
"""Return the byte content of a file.
270
:param file_id: The file_id of the file.
271
:param path: The path of the file.
272
If both file_id and path are supplied, an implementation may use
275
my_file = self.get_file(file_id, path)
277
return my_file.read()
281
def get_file_lines(self, file_id, path=None):
282
"""Return the content of a file, as lines.
284
:param file_id: The file_id of the file.
285
:param path: The path of the file.
286
If both file_id and path are supplied, an implementation may use
289
return osutils.split_lines(self.get_file_text(file_id, path))
291
def get_file_mtime(self, file_id, path=None):
292
"""Return the modification time for a file.
294
:param file_id: The handle for this file.
295
:param path: The path that this file can be found at.
296
These must point to the same object.
298
raise NotImplementedError(self.get_file_mtime)
300
def get_file_size(self, file_id):
301
"""Return the size of a file in bytes.
303
This applies only to regular files. If invoked on directories or
304
symlinks, it will return None.
305
:param file_id: The file-id of the file
307
raise NotImplementedError(self.get_file_size)
309
def get_file_by_path(self, path):
310
return self.get_file(self._inventory.path2id(path), path)
312
def iter_files_bytes(self, desired_files):
313
"""Iterate through file contents.
315
Files will not necessarily be returned in the order they occur in
316
desired_files. No specific order is guaranteed.
318
Yields pairs of identifier, bytes_iterator. identifier is an opaque
319
value supplied by the caller as part of desired_files. It should
320
uniquely identify the file version in the caller's context. (Examples:
321
an index number or a TreeTransform trans_id.)
323
bytes_iterator is an iterable of bytestrings for the file. The
324
kind of iterable and length of the bytestrings are unspecified, but for
325
this implementation, it is a tuple containing a single bytestring with
326
the complete text of the file.
328
:param desired_files: a list of (file_id, identifier) pairs
330
for file_id, identifier in desired_files:
331
# We wrap the string in a tuple so that we can return an iterable
332
# of bytestrings. (Technically, a bytestring is also an iterable
333
# of bytestrings, but iterating through each character is not
335
cur_file = (self.get_file_text(file_id),)
336
yield identifier, cur_file
338
def get_symlink_target(self, file_id):
339
"""Get the target for a given file_id.
341
It is assumed that the caller already knows that file_id is referencing
343
:param file_id: Handle for the symlink entry.
344
:return: The path the symlink points to.
346
raise NotImplementedError(self.get_symlink_target)
348
def get_canonical_inventory_paths(self, paths):
349
"""Like get_canonical_inventory_path() but works on multiple items.
351
:param paths: A sequence of paths relative to the root of the tree.
352
:return: A list of paths, with each item the corresponding input path
353
adjusted to account for existing elements that match case
356
return list(self._yield_canonical_inventory_paths(paths))
358
def get_canonical_inventory_path(self, path):
359
"""Returns the first inventory item that case-insensitively matches path.
361
If a path matches exactly, it is returned. If no path matches exactly
362
but more than one path matches case-insensitively, it is implementation
363
defined which is returned.
365
If no path matches case-insensitively, the input path is returned, but
366
with as many path entries that do exist changed to their canonical
369
If you need to resolve many names from the same tree, you should
370
use get_canonical_inventory_paths() to avoid O(N) behaviour.
372
:param path: A paths relative to the root of the tree.
373
:return: The input path adjusted to account for existing elements
374
that match case insensitively.
376
return self._yield_canonical_inventory_paths([path]).next()
378
def _yield_canonical_inventory_paths(self, paths):
380
# First, if the path as specified exists exactly, just use it.
381
if self.path2id(path) is not None:
385
cur_id = self.get_root_id()
387
bit_iter = iter(path.split("/"))
390
for child in self.iter_children(cur_id):
392
child_base = os.path.basename(self.id2path(child))
393
if child_base.lower() == lelt:
395
cur_path = osutils.pathjoin(cur_path, child_base)
398
# before a change is committed we can see this error...
401
# got to the end of this directory and no entries matched.
402
# Return what matched so far, plus the rest as specified.
403
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
408
def get_root_id(self):
409
"""Return the file_id for the root of this tree."""
410
raise NotImplementedError(self.get_root_id)
412
def annotate_iter(self, file_id,
413
default_revision=_mod_revision.CURRENT_REVISION):
414
"""Return an iterator of revision_id, line tuples.
416
For working trees (and mutable trees in general), the special
417
revision_id 'current:' will be used for lines that are new in this
418
tree, e.g. uncommitted changes.
419
:param file_id: The file to produce an annotated version from
420
:param default_revision: For lines that don't match a basis, mark them
421
with this revision id. Not all implementations will make use of
424
raise NotImplementedError(self.annotate_iter)
426
def _get_plan_merge_data(self, file_id, other, base):
427
from bzrlib import merge, versionedfile
428
vf = versionedfile._PlanMergeVersionedFile(file_id)
429
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
430
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
432
last_revision_base = None
434
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
435
return vf, last_revision_a, last_revision_b, last_revision_base
437
def plan_file_merge(self, file_id, other, base=None):
438
"""Generate a merge plan based on annotations.
440
If the file contains uncommitted changes in this tree, they will be
441
attributed to the 'current:' pseudo-revision. If the file contains
442
uncommitted changes in the other tree, they will be assigned to the
443
'other:' pseudo-revision.
445
data = self._get_plan_merge_data(file_id, other, base)
446
vf, last_revision_a, last_revision_b, last_revision_base = data
447
return vf.plan_merge(last_revision_a, last_revision_b,
450
def plan_file_lca_merge(self, file_id, other, base=None):
451
"""Generate a merge plan based lca-newness.
453
If the file contains uncommitted changes in this tree, they will be
454
attributed to the 'current:' pseudo-revision. If the file contains
455
uncommitted changes in the other tree, they will be assigned to the
456
'other:' pseudo-revision.
458
data = self._get_plan_merge_data(file_id, other, base)
459
vf, last_revision_a, last_revision_b, last_revision_base = data
460
return vf.plan_lca_merge(last_revision_a, last_revision_b,
463
def _iter_parent_trees(self):
464
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
465
for revision_id in self.get_parent_ids():
467
yield self.revision_tree(revision_id)
468
except errors.NoSuchRevisionInTree:
469
yield self.repository.revision_tree(revision_id)
472
def _file_revision(revision_tree, file_id):
473
"""Determine the revision associated with a file in a given tree."""
474
revision_tree.lock_read()
476
return revision_tree.inventory[file_id].revision
478
revision_tree.unlock()
480
def _get_file_revision(self, file_id, vf, tree_revision):
481
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
483
if getattr(self, '_repository', None) is None:
484
last_revision = tree_revision
485
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
486
self._iter_parent_trees()]
487
vf.add_lines((file_id, last_revision), parent_keys,
488
self.get_file(file_id).readlines())
489
repo = self.branch.repository
492
last_revision = self._file_revision(self, file_id)
493
base_vf = self._repository.texts
494
if base_vf not in vf.fallback_versionedfiles:
495
vf.fallback_versionedfiles.append(base_vf)
498
inventory = property(_get_inventory,
499
doc="Inventory of this Tree")
501
def _check_retrieved(self, ie, f):
504
fp = fingerprint_file(f)
507
if ie.text_size is not None:
508
if ie.text_size != fp['size']:
509
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
510
["inventory expects %d bytes" % ie.text_size,
511
"file is actually %d bytes" % fp['size'],
512
"store is probably damaged/corrupt"])
514
if ie.text_sha1 != fp['sha1']:
515
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
516
["inventory expects %s" % ie.text_sha1,
517
"file is actually %s" % fp['sha1'],
518
"store is probably damaged/corrupt"])
521
def path2id(self, path):
522
"""Return the id for path in this tree."""
523
return self._inventory.path2id(path)
525
def paths2ids(self, paths, trees=[], require_versioned=True):
526
"""Return all the ids that can be reached by walking from paths.
528
Each path is looked up in this tree and any extras provided in
529
trees, and this is repeated recursively: the children in an extra tree
530
of a directory that has been renamed under a provided path in this tree
531
are all returned, even if none exist under a provided path in this
532
tree, and vice versa.
534
:param paths: An iterable of paths to start converting to ids from.
535
Alternatively, if paths is None, no ids should be calculated and None
536
will be returned. This is offered to make calling the api unconditional
537
for code that *might* take a list of files.
538
:param trees: Additional trees to consider.
539
:param require_versioned: If False, do not raise NotVersionedError if
540
an element of paths is not versioned in this tree and all of trees.
542
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
544
def iter_children(self, file_id):
545
entry = self.iter_entries_by_dir([file_id]).next()[1]
546
for child in getattr(entry, 'children', {}).itervalues():
549
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
550
def print_file(self, file_id):
551
"""Print file with id `file_id` to stdout."""
553
sys.stdout.write(self.get_file_text(file_id))
558
def revision_tree(self, revision_id):
559
"""Obtain a revision tree for the revision revision_id.
561
The intention of this method is to allow access to possibly cached
562
tree data. Implementors of this method should raise NoSuchRevision if
563
the tree is not locally available, even if they could obtain the
564
tree via a repository or some other means. Callers are responsible
565
for finding the ultimate source for a revision tree.
567
:param revision_id: The revision_id of the requested tree.
569
:raises: NoSuchRevision if the tree cannot be obtained.
571
raise errors.NoSuchRevisionInTree(self, revision_id)
574
"""What files are present in this tree and unknown.
576
:return: an iterator over the unknown files.
583
def filter_unversioned_files(self, paths):
584
"""Filter out paths that are versioned.
586
:return: set of paths.
588
# NB: we specifically *don't* call self.has_filename, because for
589
# WorkingTrees that can indicate files that exist on disk but that
591
pred = self.inventory.has_filename
592
return set((p for p in paths if not pred(p)))
594
def walkdirs(self, prefix=""):
595
"""Walk the contents of this tree from path down.
597
This yields all the data about the contents of a directory at a time.
598
After each directory has been yielded, if the caller has mutated the
599
list to exclude some directories, they are then not descended into.
601
The data yielded is of the form:
602
((directory-relpath, directory-path-from-root, directory-fileid),
603
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
604
versioned_kind), ...]),
605
- directory-relpath is the containing dirs relpath from prefix
606
- directory-path-from-root is the containing dirs path from /
607
- directory-fileid is the id of the directory if it is versioned.
608
- relpath is the relative path within the subtree being walked.
609
- basename is the basename
610
- kind is the kind of the file now. If unknonwn then the file is not
611
present within the tree - but it may be recorded as versioned. See
613
- lstat is the stat data *if* the file was statted.
614
- path_from_tree_root is the path from the root of the tree.
615
- file_id is the file_id if the entry is versioned.
616
- versioned_kind is the kind of the file as last recorded in the
617
versioning system. If 'unknown' the file is not versioned.
618
One of 'kind' and 'versioned_kind' must not be 'unknown'.
620
:param prefix: Start walking from prefix within the tree rather than
621
at the root. This allows one to walk a subtree but get paths that are
622
relative to a tree rooted higher up.
623
:return: an iterator over the directory data.
625
raise NotImplementedError(self.walkdirs)
627
def supports_content_filtering(self):
630
def iter_search_rules(self, path_names, pref_names=None,
631
_default_searcher=rules._per_user_searcher):
632
"""Find the preferences for filenames in a tree.
634
:param path_names: an iterable of paths to find attributes for.
635
Paths are given relative to the root of the tree.
636
:param pref_names: the list of preferences to lookup - None for all
637
:param _default_searcher: private parameter to assist testing - don't use
638
:return: an iterator of tuple sequences, one per path-name.
639
See _RulesSearcher.get_items for details on the tuple sequence.
641
searcher = self._get_rules_searcher(_default_searcher)
642
if searcher is not None:
643
if pref_names is not None:
644
for path in path_names:
645
yield searcher.get_selected_items(path, pref_names)
647
for path in path_names:
648
yield searcher.get_items(path)
651
def _get_rules_searcher(self, default_searcher):
652
"""Get the RulesSearcher for this tree given the default one."""
653
searcher = default_searcher
657
######################################################################
660
# TODO: Merge these two functions into a single one that can operate
661
# on either a whole tree or a set of files.
663
# TODO: Return the diff in order by filename, not by category or in
664
# random order. Can probably be done by lock-stepping through the
665
# filenames from both trees.
668
def file_status(filename, old_tree, new_tree):
669
"""Return single-letter status, old and new names for a file.
671
The complexity here is in deciding how to represent renames;
672
many complex cases are possible.
674
old_inv = old_tree.inventory
675
new_inv = new_tree.inventory
676
new_id = new_inv.path2id(filename)
677
old_id = old_inv.path2id(filename)
679
if not new_id and not old_id:
680
# easy: doesn't exist in either; not versioned at all
681
if new_tree.is_ignored(filename):
682
return 'I', None, None
684
return '?', None, None
686
# There is now a file of this name, great.
689
# There is no longer a file of this name, but we can describe
690
# what happened to the file that used to have
691
# this name. There are two possibilities: either it was
692
# deleted entirely, or renamed.
693
if new_inv.has_id(old_id):
694
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
696
return 'D', old_inv.id2path(old_id), None
698
# if the file_id is new in this revision, it is added
699
if new_id and not old_inv.has_id(new_id):
702
# if there used to be a file of this name, but that ID has now
703
# disappeared, it is deleted
704
if old_id and not new_inv.has_id(old_id):
710
@deprecated_function(deprecated_in((1, 9, 0)))
711
def find_renames(old_inv, new_inv):
712
for file_id in old_inv:
713
if file_id not in new_inv:
715
old_name = old_inv.id2path(file_id)
716
new_name = new_inv.id2path(file_id)
717
if old_name != new_name:
718
yield (old_name, new_name)
721
def find_ids_across_trees(filenames, trees, require_versioned=True):
722
"""Find the ids corresponding to specified filenames.
724
All matches in all trees will be used, and all children of matched
725
directories will be used.
727
:param filenames: The filenames to find file_ids for (if None, returns
729
:param trees: The trees to find file_ids within
730
:param require_versioned: if true, all specified filenames must occur in
732
:return: a set of file ids for the specified filenames and their children.
736
specified_path_ids = _find_ids_across_trees(filenames, trees,
738
return _find_children_across_trees(specified_path_ids, trees)
741
def _find_ids_across_trees(filenames, trees, require_versioned):
742
"""Find the ids corresponding to specified filenames.
744
All matches in all trees will be used, but subdirectories are not scanned.
746
:param filenames: The filenames to find file_ids for
747
:param trees: The trees to find file_ids within
748
:param require_versioned: if true, all specified filenames must occur in
750
:return: a set of file ids for the specified filenames
753
interesting_ids = set()
754
for tree_path in filenames:
757
file_id = tree.path2id(tree_path)
758
if file_id is not None:
759
interesting_ids.add(file_id)
762
not_versioned.append(tree_path)
763
if len(not_versioned) > 0 and require_versioned:
764
raise errors.PathsNotVersionedError(not_versioned)
765
return interesting_ids
768
def _find_children_across_trees(specified_ids, trees):
769
"""Return a set including specified ids and their children.
771
All matches in all trees will be used.
773
:param trees: The trees to find file_ids within
774
:return: a set containing all specified ids and their children
776
interesting_ids = set(specified_ids)
777
pending = interesting_ids
778
# now handle children of interesting ids
779
# we loop so that we handle all children of each id in both trees
780
while len(pending) > 0:
782
for file_id in pending:
784
if not tree.has_id(file_id):
786
for child_id in tree.iter_children(file_id):
787
if child_id not in interesting_ids:
788
new_pending.add(child_id)
789
interesting_ids.update(new_pending)
790
pending = new_pending
791
return interesting_ids
794
class InterTree(InterObject):
795
"""This class represents operations taking place between two Trees.
797
Its instances have methods like 'compare' and contain references to the
798
source and target trees these operations are to be carried out on.
800
Clients of bzrlib should not need to use InterTree directly, rather they
801
should use the convenience methods on Tree such as 'Tree.compare()' which
802
will pass through to InterTree as appropriate.
808
def compare(self, want_unchanged=False, specific_files=None,
809
extra_trees=None, require_versioned=False, include_root=False,
810
want_unversioned=False):
811
"""Return the changes from source to target.
813
:return: A TreeDelta.
814
:param specific_files: An optional list of file paths to restrict the
815
comparison to. When mapping filenames to ids, all matches in all
816
trees (including optional extra_trees) are used, and all children of
817
matched directories are included.
818
:param want_unchanged: An optional boolean requesting the inclusion of
819
unchanged entries in the result.
820
:param extra_trees: An optional list of additional trees to use when
821
mapping the contents of specific_files (paths) to file_ids.
822
:param require_versioned: An optional boolean (defaults to False). When
823
supplied and True all the 'specific_files' must be versioned, or
824
a PathsNotVersionedError will be thrown.
825
:param want_unversioned: Scan for unversioned paths.
827
# NB: show_status depends on being able to pass in non-versioned files
828
# and report them as unknown
829
trees = (self.source,)
830
if extra_trees is not None:
831
trees = trees + tuple(extra_trees)
832
# target is usually the newer tree:
833
specific_file_ids = self.target.paths2ids(specific_files, trees,
834
require_versioned=require_versioned)
835
if specific_files and not specific_file_ids:
836
# All files are unversioned, so just return an empty delta
837
# _compare_trees would think we want a complete delta
838
result = delta.TreeDelta()
839
fake_entry = InventoryFile('unused', 'unused', 'unused')
840
result.unversioned = [(path, None,
841
self.target._comparison_data(fake_entry, path)[0]) for path in
844
return delta._compare_trees(self.source, self.target, want_unchanged,
845
specific_files, include_root, extra_trees=extra_trees,
846
require_versioned=require_versioned,
847
want_unversioned=want_unversioned)
849
def iter_changes(self, include_unchanged=False,
850
specific_files=None, pb=None, extra_trees=[],
851
require_versioned=True, want_unversioned=False):
852
"""Generate an iterator of changes between trees.
855
(file_id, (path_in_source, path_in_target),
856
changed_content, versioned, parent, name, kind,
859
Changed_content is True if the file's content has changed. This
860
includes changes to its kind, and to a symlink's target.
862
versioned, parent, name, kind, executable are tuples of (from, to).
863
If a file is missing in a tree, its kind is None.
865
Iteration is done in parent-to-child order, relative to the target
868
There is no guarantee that all paths are in sorted order: the
869
requirement to expand the search due to renames may result in children
870
that should be found early being found late in the search, after
871
lexically later results have been returned.
872
:param require_versioned: Raise errors.PathsNotVersionedError if a
873
path in the specific_files list is not versioned in one of
874
source, target or extra_trees.
875
:param want_unversioned: Should unversioned files be returned in the
876
output. An unversioned file is defined as one with (False, False)
877
for the versioned pair.
880
lookup_trees = [self.source]
882
lookup_trees.extend(extra_trees)
883
if specific_files == []:
884
specific_file_ids = []
886
specific_file_ids = self.target.paths2ids(specific_files,
887
lookup_trees, require_versioned=require_versioned)
889
all_unversioned = sorted([(p.split('/'), p) for p in
891
if specific_files is None or
892
osutils.is_inside_any(specific_files, p)])
893
all_unversioned = deque(all_unversioned)
895
all_unversioned = deque()
897
from_entries_by_dir = list(self.source.iter_entries_by_dir(
898
specific_file_ids=specific_file_ids))
899
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
900
to_entries_by_dir = list(self.target.iter_entries_by_dir(
901
specific_file_ids=specific_file_ids))
902
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
904
# the unversioned path lookup only occurs on real trees - where there
905
# can be extras. So the fake_entry is solely used to look up
906
# executable it values when execute is not supported.
907
fake_entry = InventoryFile('unused', 'unused', 'unused')
908
for to_path, to_entry in to_entries_by_dir:
909
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
910
unversioned_path = all_unversioned.popleft()
911
to_kind, to_executable, to_stat = \
912
self.target._comparison_data(fake_entry, unversioned_path[1])
913
yield (None, (None, unversioned_path[1]), True, (False, False),
915
(None, unversioned_path[0][-1]),
917
(None, to_executable))
918
file_id = to_entry.file_id
919
to_paths[file_id] = to_path
921
changed_content = False
922
from_path, from_entry = from_data.get(file_id, (None, None))
923
from_versioned = (from_entry is not None)
924
if from_entry is not None:
925
from_versioned = True
926
from_name = from_entry.name
927
from_parent = from_entry.parent_id
928
from_kind, from_executable, from_stat = \
929
self.source._comparison_data(from_entry, from_path)
932
from_versioned = False
936
from_executable = None
937
versioned = (from_versioned, True)
938
to_kind, to_executable, to_stat = \
939
self.target._comparison_data(to_entry, to_path)
940
kind = (from_kind, to_kind)
941
if kind[0] != kind[1]:
942
changed_content = True
943
elif from_kind == 'file':
944
from_size = self.source._file_size(from_entry, from_stat)
945
to_size = self.target._file_size(to_entry, to_stat)
946
if from_size != to_size:
947
changed_content = True
948
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
949
self.target.get_file_sha1(file_id, to_path, to_stat)):
950
changed_content = True
951
elif from_kind == 'symlink':
952
if (self.source.get_symlink_target(file_id) !=
953
self.target.get_symlink_target(file_id)):
954
changed_content = True
955
elif from_kind == 'tree-reference':
956
if (self.source.get_reference_revision(file_id, from_path)
957
!= self.target.get_reference_revision(file_id, to_path)):
958
changed_content = True
959
parent = (from_parent, to_entry.parent_id)
960
name = (from_name, to_entry.name)
961
executable = (from_executable, to_executable)
963
pb.update('comparing files', entry_count, num_entries)
964
if (changed_content is not False or versioned[0] != versioned[1]
965
or parent[0] != parent[1] or name[0] != name[1] or
966
executable[0] != executable[1] or include_unchanged):
967
yield (file_id, (from_path, to_path), changed_content,
968
versioned, parent, name, kind, executable)
970
while all_unversioned:
971
# yield any trailing unversioned paths
972
unversioned_path = all_unversioned.popleft()
973
to_kind, to_executable, to_stat = \
974
self.target._comparison_data(fake_entry, unversioned_path[1])
975
yield (None, (None, unversioned_path[1]), True, (False, False),
977
(None, unversioned_path[0][-1]),
979
(None, to_executable))
981
def get_to_path(to_entry):
982
if to_entry.parent_id is None:
983
to_path = '' # the root
985
if to_entry.parent_id not in to_paths:
987
return get_to_path(self.target.inventory[to_entry.parent_id])
988
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
990
to_paths[to_entry.file_id] = to_path
993
for path, from_entry in from_entries_by_dir:
994
file_id = from_entry.file_id
995
if file_id in to_paths:
998
if not file_id in self.target.all_file_ids():
999
# common case - paths we have not emitted are not present in
1003
to_path = get_to_path(self.target.inventory[file_id])
1006
pb.update('comparing files', entry_count, num_entries)
1007
versioned = (True, False)
1008
parent = (from_entry.parent_id, None)
1009
name = (from_entry.name, None)
1010
from_kind, from_executable, stat_value = \
1011
self.source._comparison_data(from_entry, path)
1012
kind = (from_kind, None)
1013
executable = (from_executable, None)
1014
changed_content = from_kind is not None
1015
# the parent's path is necessarily known at this point.
1016
yield(file_id, (path, to_path), changed_content, versioned, parent,
1017
name, kind, executable)
1020
class MultiWalker(object):
1021
"""Walk multiple trees simultaneously, getting combined results."""
1023
# Note: This could be written to not assume you can do out-of-order
1024
# lookups. Instead any nodes that don't match in all trees could be
1025
# marked as 'deferred', and then returned in the final cleanup loop.
1026
# For now, I think it is "nicer" to return things as close to the
1027
# "master_tree" order as we can.
1029
def __init__(self, master_tree, other_trees):
1030
"""Create a new MultiWalker.
1032
All trees being walked must implement "iter_entries_by_dir()", such
1033
that they yield (path, object) tuples, where that object will have a
1034
'.file_id' member, that can be used to check equality.
1036
:param master_tree: All trees will be 'slaved' to the master_tree such
1037
that nodes in master_tree will be used as 'first-pass' sync points.
1038
Any nodes that aren't in master_tree will be merged in a second
1040
:param other_trees: A list of other trees to walk simultaneously.
1042
self._master_tree = master_tree
1043
self._other_trees = other_trees
1045
# Keep track of any nodes that were properly processed just out of
1046
# order, that way we don't return them at the end, we don't have to
1047
# track *all* processed file_ids, just the out-of-order ones
1048
self._out_of_order_processed = set()
1051
def _step_one(iterator):
1052
"""Step an iter_entries_by_dir iterator.
1054
:return: (has_more, path, ie)
1055
If has_more is False, path and ie will be None.
1058
path, ie = iterator.next()
1059
except StopIteration:
1060
return False, None, None
1062
return True, path, ie
1065
def _cmp_path_by_dirblock(path1, path2):
1066
"""Compare two paths based on what directory they are in.
1068
This generates a sort order, such that all children of a directory are
1069
sorted together, and grandchildren are in the same order as the
1070
children appear. But all grandchildren come after all children.
1072
:param path1: first path
1073
:param path2: the second path
1074
:return: negative number if ``path1`` comes first,
1075
0 if paths are equal
1076
and a positive number if ``path2`` sorts first
1078
# Shortcut this special case
1081
# This is stolen from _dirstate_helpers_py.py, only switching it to
1082
# Unicode objects. Consider using encode_utf8() and then using the
1083
# optimized versions, or maybe writing optimized unicode versions.
1084
if not isinstance(path1, unicode):
1085
raise TypeError("'path1' must be a unicode string, not %s: %r"
1086
% (type(path1), path1))
1087
if not isinstance(path2, unicode):
1088
raise TypeError("'path2' must be a unicode string, not %s: %r"
1089
% (type(path2), path2))
1090
return cmp(MultiWalker._path_to_key(path1),
1091
MultiWalker._path_to_key(path2))
1094
def _path_to_key(path):
1095
dirname, basename = osutils.split(path)
1096
return (dirname.split(u'/'), basename)
1098
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1099
"""Lookup an inventory entry by file_id.
1101
This is called when an entry is missing in the normal order.
1102
Generally this is because a file was either renamed, or it was
1103
deleted/added. If the entry was found in the inventory and not in
1104
extra_entries, it will be added to self._out_of_order_processed
1106
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1107
should be filled with entries that were found before they were
1108
used. If file_id is present, it will be removed from the
1110
:param other_tree: The Tree to search, in case we didn't find the entry
1112
:param file_id: The file_id to look for
1113
:return: (path, ie) if found or (None, None) if not present.
1115
if file_id in extra_entries:
1116
return extra_entries.pop(file_id)
1117
# TODO: Is id2path better as the first call, or is
1118
# inventory[file_id] better as a first check?
1120
cur_path = other_tree.id2path(file_id)
1121
except errors.NoSuchId:
1123
if cur_path is None:
1126
self._out_of_order_processed.add(file_id)
1127
cur_ie = other_tree.inventory[file_id]
1128
return (cur_path, cur_ie)
1131
"""Match up the values in the different trees."""
1132
for result in self._walk_master_tree():
1134
self._finish_others()
1135
for result in self._walk_others():
1138
def _walk_master_tree(self):
1139
"""First pass, walk all trees in lock-step.
1141
When we are done, all nodes in the master_tree will have been
1142
processed. _other_walkers, _other_entries, and _others_extra will be
1143
set on 'self' for future processing.
1145
# This iterator has the most "inlining" done, because it tends to touch
1146
# every file in the tree, while the others only hit nodes that don't
1148
master_iterator = self._master_tree.iter_entries_by_dir()
1150
other_walkers = [other.iter_entries_by_dir()
1151
for other in self._other_trees]
1152
other_entries = [self._step_one(walker) for walker in other_walkers]
1153
# Track extra nodes in the other trees
1154
others_extra = [{} for i in xrange(len(self._other_trees))]
1156
master_has_more = True
1157
step_one = self._step_one
1158
lookup_by_file_id = self._lookup_by_file_id
1159
out_of_order_processed = self._out_of_order_processed
1161
while master_has_more:
1162
(master_has_more, path, master_ie) = step_one(master_iterator)
1163
if not master_has_more:
1166
file_id = master_ie.file_id
1168
other_values_append = other_values.append
1169
next_other_entries = []
1170
next_other_entries_append = next_other_entries.append
1171
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1172
if not other_has_more:
1173
other_values_append(lookup_by_file_id(
1174
others_extra[idx], self._other_trees[idx], file_id))
1175
next_other_entries_append((False, None, None))
1176
elif file_id == other_ie.file_id:
1177
# This is the critical code path, as most of the entries
1178
# should match between most trees.
1179
other_values_append((other_path, other_ie))
1180
next_other_entries_append(step_one(other_walkers[idx]))
1182
# This walker did not match, step it until it either
1183
# matches, or we know we are past the current walker.
1184
other_walker = other_walkers[idx]
1185
other_extra = others_extra[idx]
1186
while (other_has_more and
1187
self._cmp_path_by_dirblock(other_path, path) < 0):
1188
other_file_id = other_ie.file_id
1189
if other_file_id not in out_of_order_processed:
1190
other_extra[other_file_id] = (other_path, other_ie)
1191
other_has_more, other_path, other_ie = \
1192
step_one(other_walker)
1193
if other_has_more and other_ie.file_id == file_id:
1194
# We ended up walking to this point, match and step
1196
other_values_append((other_path, other_ie))
1197
other_has_more, other_path, other_ie = \
1198
step_one(other_walker)
1200
# This record isn't in the normal order, see if it
1202
other_values_append(lookup_by_file_id(
1203
other_extra, self._other_trees[idx], file_id))
1204
next_other_entries_append((other_has_more, other_path,
1206
other_entries = next_other_entries
1208
# We've matched all the walkers, yield this datapoint
1209
yield path, file_id, master_ie, other_values
1210
self._other_walkers = other_walkers
1211
self._other_entries = other_entries
1212
self._others_extra = others_extra
1214
def _finish_others(self):
1215
"""Finish walking the other iterators, so we get all entries."""
1216
for idx, info in enumerate(self._other_entries):
1217
other_extra = self._others_extra[idx]
1218
(other_has_more, other_path, other_ie) = info
1219
while other_has_more:
1220
other_file_id = other_ie.file_id
1221
if other_file_id not in self._out_of_order_processed:
1222
other_extra[other_file_id] = (other_path, other_ie)
1223
other_has_more, other_path, other_ie = \
1224
self._step_one(self._other_walkers[idx])
1225
del self._other_entries
1227
def _walk_others(self):
1228
"""Finish up by walking all the 'deferred' nodes."""
1229
# TODO: One alternative would be to grab all possible unprocessed
1230
# file_ids, and then sort by path, and then yield them. That
1231
# might ensure better ordering, in case a caller strictly
1232
# requires parents before children.
1233
for idx, other_extra in enumerate(self._others_extra):
1234
others = sorted(other_extra.itervalues(),
1235
key=lambda x: self._path_to_key(x[0]))
1236
for other_path, other_ie in others:
1237
file_id = other_ie.file_id
1238
# We don't need to check out_of_order_processed here, because
1239
# the lookup_by_file_id will be removing anything processed
1240
# from the extras cache
1241
other_extra.pop(file_id)
1242
other_values = [(None, None) for i in xrange(idx)]
1243
other_values.append((other_path, other_ie))
1244
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1245
alt_idx = alt_idx + idx + 1
1246
alt_extra = self._others_extra[alt_idx]
1247
alt_tree = self._other_trees[alt_idx]
1248
other_values.append(self._lookup_by_file_id(
1249
alt_extra, alt_tree, file_id))
1250
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
