1
# Copyright (C) 2005 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Tree classes, representing directory at point in time.
21
from collections import deque
22
from cStringIO import StringIO
26
conflicts as _mod_conflicts,
31
revision as _mod_revision,
35
from bzrlib.decorators import needs_read_lock
36
from bzrlib.errors import BzrError, BzrCheckError
37
from bzrlib import errors
38
from bzrlib.inventory import Inventory, InventoryFile
39
from bzrlib.inter import InterObject
40
from bzrlib.osutils import fingerprint_file
41
import bzrlib.revision
42
from bzrlib.symbol_versioning import deprecated_function, deprecated_in
43
from bzrlib.trace import mutter, note
47
"""Abstract file tree.
49
There are several subclasses:
51
* `WorkingTree` exists as files on disk editable by the user.
53
* `RevisionTree` is a tree as recorded at some point in the past.
55
Trees contain an `Inventory` object, and also know how to retrieve
56
file texts mentioned in the inventory, either from a working
57
directory or from a store.
59
It is possible for trees to contain files that are not described
60
in their inventory or vice versa; for this use `filenames()`.
62
Trees can be compared, etc, regardless of whether they are working
63
trees or versioned trees.
66
def changes_from(self, other, want_unchanged=False, specific_files=None,
67
extra_trees=None, require_versioned=False, include_root=False,
68
want_unversioned=False):
69
"""Return a TreeDelta of the changes from other to this tree.
71
:param other: A tree to compare with.
72
:param specific_files: An optional list of file paths to restrict the
73
comparison to. When mapping filenames to ids, all matches in all
74
trees (including optional extra_trees) are used, and all children of
75
matched directories are included.
76
:param want_unchanged: An optional boolean requesting the inclusion of
77
unchanged entries in the result.
78
:param extra_trees: An optional list of additional trees to use when
79
mapping the contents of specific_files (paths) to file_ids.
80
:param require_versioned: An optional boolean (defaults to False). When
81
supplied and True all the 'specific_files' must be versioned, or
82
a PathsNotVersionedError will be thrown.
83
:param want_unversioned: Scan for unversioned paths.
85
The comparison will be performed by an InterTree object looked up on
88
# Martin observes that Tree.changes_from returns a TreeDelta and this
89
# may confuse people, because the class name of the returned object is
90
# a synonym of the object referenced in the method name.
91
return InterTree.get(other, self).compare(
92
want_unchanged=want_unchanged,
93
specific_files=specific_files,
94
extra_trees=extra_trees,
95
require_versioned=require_versioned,
96
include_root=include_root,
97
want_unversioned=want_unversioned,
100
@symbol_versioning.deprecated_method(symbol_versioning.one_three)
101
def _iter_changes(self, *args, **kwargs):
102
return self.iter_changes(*args, **kwargs)
104
def iter_changes(self, from_tree, include_unchanged=False,
105
specific_files=None, pb=None, extra_trees=None,
106
require_versioned=True, want_unversioned=False):
107
intertree = InterTree.get(from_tree, self)
108
return intertree.iter_changes(include_unchanged, specific_files, pb,
109
extra_trees, require_versioned, want_unversioned=want_unversioned)
112
"""Get a list of the conflicts in the tree.
114
Each conflict is an instance of bzrlib.conflicts.Conflict.
116
return _mod_conflicts.ConflictList()
119
"""For trees that can have unversioned files, return all such paths."""
122
def get_parent_ids(self):
123
"""Get the parent ids for this tree.
125
:return: a list of parent ids. [] is returned to indicate
126
a tree with no parents.
127
:raises: BzrError if the parents are not known.
129
raise NotImplementedError(self.get_parent_ids)
131
def has_filename(self, filename):
132
"""True if the tree has given filename."""
133
raise NotImplementedError(self.has_filename)
135
def has_id(self, file_id):
136
return self.inventory.has_id(file_id)
138
def __contains__(self, file_id):
139
return self.has_id(file_id)
141
def has_or_had_id(self, file_id):
142
if file_id == self.inventory.root.file_id:
144
return self.inventory.has_id(file_id)
146
def is_ignored(self, filename):
147
"""Check whether the filename is ignored by this tree.
149
:param filename: The relative filename within the tree.
150
:return: True if the filename is ignored.
155
return iter(self.inventory)
157
def all_file_ids(self):
158
"""Iterate through all file ids, including ids for missing files."""
159
return set(self.inventory)
161
def id2path(self, file_id):
162
"""Return the path for a file id.
166
return self.inventory.id2path(file_id)
168
def is_control_filename(self, filename):
169
"""True if filename is the name of a control file in this tree.
171
:param filename: A filename within the tree. This is a relative path
172
from the root of this tree.
174
This is true IF and ONLY IF the filename is part of the meta data
175
that bzr controls in this tree. I.E. a random .bzr directory placed
176
on disk will not be a control file for this tree.
178
return self.bzrdir.is_control_filename(filename)
181
def iter_entries_by_dir(self, specific_file_ids=None):
182
"""Walk the tree in 'by_dir' order.
184
This will yield each entry in the tree as a (path, entry) tuple.
185
The order that they are yielded is:
187
Directories are walked in a depth-first lexicographical order,
188
however, whenever a directory is reached, all of its direct child
189
nodes are yielded in lexicographical order before yielding the
192
For example, in the tree::
202
The yield order (ignoring root) would be::
203
a, f, a/b, a/d, a/b/c, a/d/e, f/g
205
return self.inventory.iter_entries_by_dir(
206
specific_file_ids=specific_file_ids)
208
def iter_references(self):
209
for path, entry in self.iter_entries_by_dir():
210
if entry.kind == 'tree-reference':
211
yield path, entry.file_id
213
def kind(self, file_id):
214
raise NotImplementedError("Tree subclass %s must implement kind"
215
% self.__class__.__name__)
217
def stored_kind(self, file_id):
218
"""File kind stored for this file_id.
220
May not match kind on disk for working trees. Always available
221
for versioned files, even when the file itself is missing.
223
return self.kind(file_id)
225
def path_content_summary(self, path):
226
"""Get a summary of the information about path.
228
:param path: A relative path within the tree.
229
:return: A tuple containing kind, size, exec, sha1-or-link.
230
Kind is always present (see tree.kind()).
231
size is present if kind is file, None otherwise.
232
exec is None unless kind is file and the platform supports the 'x'
234
sha1-or-link is the link target if kind is symlink, or the sha1 if
235
it can be obtained without reading the file.
237
raise NotImplementedError(self.path_content_summary)
239
def get_reference_revision(self, file_id, path=None):
240
raise NotImplementedError("Tree subclass %s must implement "
241
"get_reference_revision"
242
% self.__class__.__name__)
244
def _comparison_data(self, entry, path):
245
"""Return a tuple of kind, executable, stat_value for a file.
247
entry may be None if there is no inventory entry for the file, but
248
path must always be supplied.
250
kind is None if there is no file present (even if an inventory id is
251
present). executable is False for non-file entries.
253
raise NotImplementedError(self._comparison_data)
255
def _file_size(self, entry, stat_value):
256
raise NotImplementedError(self._file_size)
258
def _get_inventory(self):
259
return self._inventory
261
def get_file(self, file_id, path=None):
262
"""Return a file object for the file file_id in the tree.
264
If both file_id and path are defined, it is implementation defined as
265
to which one is used.
267
raise NotImplementedError(self.get_file)
269
def get_file_text(self, file_id, path=None):
270
"""Return the byte content of a file.
272
:param file_id: The file_id of the file.
273
:param path: The path of the file.
274
If both file_id and path are supplied, an implementation may use
277
my_file = self.get_file(file_id, path)
279
return my_file.read()
283
def get_file_lines(self, file_id, path=None):
284
"""Return the content of a file, as lines.
286
:param file_id: The file_id of the file.
287
:param path: The path of the file.
288
If both file_id and path are supplied, an implementation may use
291
return osutils.split_lines(self.get_file_text(file_id, path))
293
def get_file_mtime(self, file_id, path=None):
294
"""Return the modification time for a file.
296
:param file_id: The handle for this file.
297
:param path: The path that this file can be found at.
298
These must point to the same object.
300
raise NotImplementedError(self.get_file_mtime)
302
def get_file_size(self, file_id):
303
"""Return the size of a file in bytes.
305
This applies only to regular files. If invoked on directories or
306
symlinks, it will return None.
307
:param file_id: The file-id of the file
309
raise NotImplementedError(self.get_file_size)
311
def get_file_by_path(self, path):
312
return self.get_file(self._inventory.path2id(path), path)
314
def iter_files_bytes(self, desired_files):
315
"""Iterate through file contents.
317
Files will not necessarily be returned in the order they occur in
318
desired_files. No specific order is guaranteed.
320
Yields pairs of identifier, bytes_iterator. identifier is an opaque
321
value supplied by the caller as part of desired_files. It should
322
uniquely identify the file version in the caller's context. (Examples:
323
an index number or a TreeTransform trans_id.)
325
bytes_iterator is an iterable of bytestrings for the file. The
326
kind of iterable and length of the bytestrings are unspecified, but for
327
this implementation, it is a tuple containing a single bytestring with
328
the complete text of the file.
330
:param desired_files: a list of (file_id, identifier) pairs
332
for file_id, identifier in desired_files:
333
# We wrap the string in a tuple so that we can return an iterable
334
# of bytestrings. (Technically, a bytestring is also an iterable
335
# of bytestrings, but iterating through each character is not
337
cur_file = (self.get_file_text(file_id),)
338
yield identifier, cur_file
340
def get_symlink_target(self, file_id):
341
"""Get the target for a given file_id.
343
It is assumed that the caller already knows that file_id is referencing
345
:param file_id: Handle for the symlink entry.
346
:return: The path the symlink points to.
348
raise NotImplementedError(self.get_symlink_target)
350
def get_root_id(self):
351
"""Return the file_id for the root of this tree."""
352
raise NotImplementedError(self.get_root_id)
354
def annotate_iter(self, file_id,
355
default_revision=_mod_revision.CURRENT_REVISION):
356
"""Return an iterator of revision_id, line tuples.
358
For working trees (and mutable trees in general), the special
359
revision_id 'current:' will be used for lines that are new in this
360
tree, e.g. uncommitted changes.
361
:param file_id: The file to produce an annotated version from
362
:param default_revision: For lines that don't match a basis, mark them
363
with this revision id. Not all implementations will make use of
366
raise NotImplementedError(self.annotate_iter)
368
def _get_plan_merge_data(self, file_id, other, base):
369
from bzrlib import merge, versionedfile
370
vf = versionedfile._PlanMergeVersionedFile(file_id)
371
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
372
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
374
last_revision_base = None
376
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
377
return vf, last_revision_a, last_revision_b, last_revision_base
379
def plan_file_merge(self, file_id, other, base=None):
380
"""Generate a merge plan based on annotations.
382
If the file contains uncommitted changes in this tree, they will be
383
attributed to the 'current:' pseudo-revision. If the file contains
384
uncommitted changes in the other tree, they will be assigned to the
385
'other:' pseudo-revision.
387
data = self._get_plan_merge_data(file_id, other, base)
388
vf, last_revision_a, last_revision_b, last_revision_base = data
389
return vf.plan_merge(last_revision_a, last_revision_b,
392
def plan_file_lca_merge(self, file_id, other, base=None):
393
"""Generate a merge plan based lca-newness.
395
If the file contains uncommitted changes in this tree, they will be
396
attributed to the 'current:' pseudo-revision. If the file contains
397
uncommitted changes in the other tree, they will be assigned to the
398
'other:' pseudo-revision.
400
data = self._get_plan_merge_data(file_id, other, base)
401
vf, last_revision_a, last_revision_b, last_revision_base = data
402
return vf.plan_lca_merge(last_revision_a, last_revision_b,
405
def _iter_parent_trees(self):
406
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
407
for revision_id in self.get_parent_ids():
409
yield self.revision_tree(revision_id)
410
except errors.NoSuchRevisionInTree:
411
yield self.repository.revision_tree(revision_id)
414
def _file_revision(revision_tree, file_id):
415
"""Determine the revision associated with a file in a given tree."""
416
revision_tree.lock_read()
418
return revision_tree.inventory[file_id].revision
420
revision_tree.unlock()
422
def _get_file_revision(self, file_id, vf, tree_revision):
423
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
425
if getattr(self, '_repository', None) is None:
426
last_revision = tree_revision
427
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
428
self._iter_parent_trees()]
429
vf.add_lines((file_id, last_revision), parent_keys,
430
self.get_file(file_id).readlines())
431
repo = self.branch.repository
434
last_revision = self._file_revision(self, file_id)
435
base_vf = self._repository.texts
436
if base_vf not in vf.fallback_versionedfiles:
437
vf.fallback_versionedfiles.append(base_vf)
440
inventory = property(_get_inventory,
441
doc="Inventory of this Tree")
443
def _check_retrieved(self, ie, f):
446
fp = fingerprint_file(f)
449
if ie.text_size is not None:
450
if ie.text_size != fp['size']:
451
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
452
["inventory expects %d bytes" % ie.text_size,
453
"file is actually %d bytes" % fp['size'],
454
"store is probably damaged/corrupt"])
456
if ie.text_sha1 != fp['sha1']:
457
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
458
["inventory expects %s" % ie.text_sha1,
459
"file is actually %s" % fp['sha1'],
460
"store is probably damaged/corrupt"])
463
def path2id(self, path):
464
"""Return the id for path in this tree."""
465
return self._inventory.path2id(path)
467
def paths2ids(self, paths, trees=[], require_versioned=True):
468
"""Return all the ids that can be reached by walking from paths.
470
Each path is looked up in this tree and any extras provided in
471
trees, and this is repeated recursively: the children in an extra tree
472
of a directory that has been renamed under a provided path in this tree
473
are all returned, even if none exist under a provided path in this
474
tree, and vice versa.
476
:param paths: An iterable of paths to start converting to ids from.
477
Alternatively, if paths is None, no ids should be calculated and None
478
will be returned. This is offered to make calling the api unconditional
479
for code that *might* take a list of files.
480
:param trees: Additional trees to consider.
481
:param require_versioned: If False, do not raise NotVersionedError if
482
an element of paths is not versioned in this tree and all of trees.
484
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
486
def iter_children(self, file_id):
487
entry = self.iter_entries_by_dir([file_id]).next()[1]
488
for child in getattr(entry, 'children', {}).itervalues():
491
@symbol_versioning.deprecated_method(symbol_versioning.one_six)
492
def print_file(self, file_id):
493
"""Print file with id `file_id` to stdout."""
495
sys.stdout.write(self.get_file_text(file_id))
500
def revision_tree(self, revision_id):
501
"""Obtain a revision tree for the revision revision_id.
503
The intention of this method is to allow access to possibly cached
504
tree data. Implementors of this method should raise NoSuchRevision if
505
the tree is not locally available, even if they could obtain the
506
tree via a repository or some other means. Callers are responsible
507
for finding the ultimate source for a revision tree.
509
:param revision_id: The revision_id of the requested tree.
511
:raises: NoSuchRevision if the tree cannot be obtained.
513
raise errors.NoSuchRevisionInTree(self, revision_id)
516
"""What files are present in this tree and unknown.
518
:return: an iterator over the unknown files.
525
def filter_unversioned_files(self, paths):
526
"""Filter out paths that are versioned.
528
:return: set of paths.
530
# NB: we specifically *don't* call self.has_filename, because for
531
# WorkingTrees that can indicate files that exist on disk but that
533
pred = self.inventory.has_filename
534
return set((p for p in paths if not pred(p)))
536
def walkdirs(self, prefix=""):
537
"""Walk the contents of this tree from path down.
539
This yields all the data about the contents of a directory at a time.
540
After each directory has been yielded, if the caller has mutated the
541
list to exclude some directories, they are then not descended into.
543
The data yielded is of the form:
544
((directory-relpath, directory-path-from-root, directory-fileid),
545
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
546
versioned_kind), ...]),
547
- directory-relpath is the containing dirs relpath from prefix
548
- directory-path-from-root is the containing dirs path from /
549
- directory-fileid is the id of the directory if it is versioned.
550
- relpath is the relative path within the subtree being walked.
551
- basename is the basename
552
- kind is the kind of the file now. If unknonwn then the file is not
553
present within the tree - but it may be recorded as versioned. See
555
- lstat is the stat data *if* the file was statted.
556
- path_from_tree_root is the path from the root of the tree.
557
- file_id is the file_id if the entry is versioned.
558
- versioned_kind is the kind of the file as last recorded in the
559
versioning system. If 'unknown' the file is not versioned.
560
One of 'kind' and 'versioned_kind' must not be 'unknown'.
562
:param prefix: Start walking from prefix within the tree rather than
563
at the root. This allows one to walk a subtree but get paths that are
564
relative to a tree rooted higher up.
565
:return: an iterator over the directory data.
567
raise NotImplementedError(self.walkdirs)
569
def supports_content_filtering(self):
572
def _content_filter_stack(self, path=None, file_id=None):
573
"""The stack of content filters for a path if filtering is supported.
575
Readers will be applied in first-to-last order.
576
Writers will be applied in last-to-first order.
577
Either the path or the file-id needs to be provided.
579
:param path: path relative to the root of the tree
581
:param file_id: file_id or None if unknown
582
:return: the list of filters - [] if there are none
584
filter_pref_names = filters._get_registered_names()
585
if len(filter_pref_names) == 0:
588
path = self.id2path(file_id)
589
prefs = self.iter_search_rules([path], filter_pref_names).next()
590
stk = filters._get_filter_stack_for(prefs)
591
if 'filters' in debug.debug_flags:
592
note("*** %s content-filter: %s => %r" % (path,prefs,stk))
595
def _content_filter_stack_provider(self):
596
"""A function that returns a stack of ContentFilters.
598
The function takes a path (relative to the top of the tree) and a
599
file-id as parameters.
601
:return: None if content filtering is not supported by this tree.
603
if self.supports_content_filtering():
604
return lambda path, file_id: \
605
self._content_filter_stack(path, file_id)
609
def iter_search_rules(self, path_names, pref_names=None,
610
_default_searcher=rules._per_user_searcher):
611
"""Find the preferences for filenames in a tree.
613
:param path_names: an iterable of paths to find attributes for.
614
Paths are given relative to the root of the tree.
615
:param pref_names: the list of preferences to lookup - None for all
616
:param _default_searcher: private parameter to assist testing - don't use
617
:return: an iterator of tuple sequences, one per path-name.
618
See _RulesSearcher.get_items for details on the tuple sequence.
620
searcher = self._get_rules_searcher(_default_searcher)
621
if searcher is not None:
622
if pref_names is not None:
623
for path in path_names:
624
yield searcher.get_selected_items(path, pref_names)
626
for path in path_names:
627
yield searcher.get_items(path)
630
def _get_rules_searcher(self, default_searcher):
631
"""Get the RulesSearcher for this tree given the default one."""
632
searcher = default_searcher
636
class EmptyTree(Tree):
639
self._inventory = Inventory(root_id=None)
640
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
641
' use repository.revision_tree instead.',
642
DeprecationWarning, stacklevel=2)
644
def get_parent_ids(self):
647
def get_symlink_target(self, file_id):
650
def has_filename(self, filename):
653
def kind(self, file_id):
656
def list_files(self, include_root=False):
659
def __contains__(self, file_id):
660
return (file_id in self._inventory)
662
def get_file_sha1(self, file_id, path=None, stat_value=None):
666
######################################################################
669
# TODO: Merge these two functions into a single one that can operate
670
# on either a whole tree or a set of files.
672
# TODO: Return the diff in order by filename, not by category or in
673
# random order. Can probably be done by lock-stepping through the
674
# filenames from both trees.
677
def file_status(filename, old_tree, new_tree):
678
"""Return single-letter status, old and new names for a file.
680
The complexity here is in deciding how to represent renames;
681
many complex cases are possible.
683
old_inv = old_tree.inventory
684
new_inv = new_tree.inventory
685
new_id = new_inv.path2id(filename)
686
old_id = old_inv.path2id(filename)
688
if not new_id and not old_id:
689
# easy: doesn't exist in either; not versioned at all
690
if new_tree.is_ignored(filename):
691
return 'I', None, None
693
return '?', None, None
695
# There is now a file of this name, great.
698
# There is no longer a file of this name, but we can describe
699
# what happened to the file that used to have
700
# this name. There are two possibilities: either it was
701
# deleted entirely, or renamed.
702
if new_inv.has_id(old_id):
703
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
705
return 'D', old_inv.id2path(old_id), None
707
# if the file_id is new in this revision, it is added
708
if new_id and not old_inv.has_id(new_id):
711
# if there used to be a file of this name, but that ID has now
712
# disappeared, it is deleted
713
if old_id and not new_inv.has_id(old_id):
719
@deprecated_function(deprecated_in((1, 9, 0)))
720
def find_renames(old_inv, new_inv):
721
for file_id in old_inv:
722
if file_id not in new_inv:
724
old_name = old_inv.id2path(file_id)
725
new_name = new_inv.id2path(file_id)
726
if old_name != new_name:
727
yield (old_name, new_name)
730
def find_ids_across_trees(filenames, trees, require_versioned=True):
731
"""Find the ids corresponding to specified filenames.
733
All matches in all trees will be used, and all children of matched
734
directories will be used.
736
:param filenames: The filenames to find file_ids for (if None, returns
738
:param trees: The trees to find file_ids within
739
:param require_versioned: if true, all specified filenames must occur in
741
:return: a set of file ids for the specified filenames and their children.
745
specified_path_ids = _find_ids_across_trees(filenames, trees,
747
return _find_children_across_trees(specified_path_ids, trees)
750
def _find_ids_across_trees(filenames, trees, require_versioned):
751
"""Find the ids corresponding to specified filenames.
753
All matches in all trees will be used, but subdirectories are not scanned.
755
:param filenames: The filenames to find file_ids for
756
:param trees: The trees to find file_ids within
757
:param require_versioned: if true, all specified filenames must occur in
759
:return: a set of file ids for the specified filenames
762
interesting_ids = set()
763
for tree_path in filenames:
766
file_id = tree.path2id(tree_path)
767
if file_id is not None:
768
interesting_ids.add(file_id)
771
not_versioned.append(tree_path)
772
if len(not_versioned) > 0 and require_versioned:
773
raise errors.PathsNotVersionedError(not_versioned)
774
return interesting_ids
777
def _find_children_across_trees(specified_ids, trees):
778
"""Return a set including specified ids and their children.
780
All matches in all trees will be used.
782
:param trees: The trees to find file_ids within
783
:return: a set containing all specified ids and their children
785
interesting_ids = set(specified_ids)
786
pending = interesting_ids
787
# now handle children of interesting ids
788
# we loop so that we handle all children of each id in both trees
789
while len(pending) > 0:
791
for file_id in pending:
793
if not tree.has_id(file_id):
795
for child_id in tree.iter_children(file_id):
796
if child_id not in interesting_ids:
797
new_pending.add(child_id)
798
interesting_ids.update(new_pending)
799
pending = new_pending
800
return interesting_ids
803
class InterTree(InterObject):
804
"""This class represents operations taking place between two Trees.
806
Its instances have methods like 'compare' and contain references to the
807
source and target trees these operations are to be carried out on.
809
Clients of bzrlib should not need to use InterTree directly, rather they
810
should use the convenience methods on Tree such as 'Tree.compare()' which
811
will pass through to InterTree as appropriate.
817
def compare(self, want_unchanged=False, specific_files=None,
818
extra_trees=None, require_versioned=False, include_root=False,
819
want_unversioned=False):
820
"""Return the changes from source to target.
822
:return: A TreeDelta.
823
:param specific_files: An optional list of file paths to restrict the
824
comparison to. When mapping filenames to ids, all matches in all
825
trees (including optional extra_trees) are used, and all children of
826
matched directories are included.
827
:param want_unchanged: An optional boolean requesting the inclusion of
828
unchanged entries in the result.
829
:param extra_trees: An optional list of additional trees to use when
830
mapping the contents of specific_files (paths) to file_ids.
831
:param require_versioned: An optional boolean (defaults to False). When
832
supplied and True all the 'specific_files' must be versioned, or
833
a PathsNotVersionedError will be thrown.
834
:param want_unversioned: Scan for unversioned paths.
836
# NB: show_status depends on being able to pass in non-versioned files
837
# and report them as unknown
838
trees = (self.source,)
839
if extra_trees is not None:
840
trees = trees + tuple(extra_trees)
841
# target is usually the newer tree:
842
specific_file_ids = self.target.paths2ids(specific_files, trees,
843
require_versioned=require_versioned)
844
if specific_files and not specific_file_ids:
845
# All files are unversioned, so just return an empty delta
846
# _compare_trees would think we want a complete delta
847
result = delta.TreeDelta()
848
fake_entry = InventoryFile('unused', 'unused', 'unused')
849
result.unversioned = [(path, None,
850
self.target._comparison_data(fake_entry, path)[0]) for path in
853
return delta._compare_trees(self.source, self.target, want_unchanged,
854
specific_files, include_root, extra_trees=extra_trees,
855
require_versioned=require_versioned,
856
want_unversioned=want_unversioned)
858
def iter_changes(self, include_unchanged=False,
859
specific_files=None, pb=None, extra_trees=[],
860
require_versioned=True, want_unversioned=False):
861
"""Generate an iterator of changes between trees.
864
(file_id, (path_in_source, path_in_target),
865
changed_content, versioned, parent, name, kind,
868
Changed_content is True if the file's content has changed. This
869
includes changes to its kind, and to a symlink's target.
871
versioned, parent, name, kind, executable are tuples of (from, to).
872
If a file is missing in a tree, its kind is None.
874
Iteration is done in parent-to-child order, relative to the target
877
There is no guarantee that all paths are in sorted order: the
878
requirement to expand the search due to renames may result in children
879
that should be found early being found late in the search, after
880
lexically later results have been returned.
881
:param require_versioned: Raise errors.PathsNotVersionedError if a
882
path in the specific_files list is not versioned in one of
883
source, target or extra_trees.
884
:param want_unversioned: Should unversioned files be returned in the
885
output. An unversioned file is defined as one with (False, False)
886
for the versioned pair.
889
lookup_trees = [self.source]
891
lookup_trees.extend(extra_trees)
892
if specific_files == []:
893
specific_file_ids = []
895
specific_file_ids = self.target.paths2ids(specific_files,
896
lookup_trees, require_versioned=require_versioned)
898
all_unversioned = sorted([(p.split('/'), p) for p in
900
if specific_files is None or
901
osutils.is_inside_any(specific_files, p)])
902
all_unversioned = deque(all_unversioned)
904
all_unversioned = deque()
906
from_entries_by_dir = list(self.source.iter_entries_by_dir(
907
specific_file_ids=specific_file_ids))
908
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
909
to_entries_by_dir = list(self.target.iter_entries_by_dir(
910
specific_file_ids=specific_file_ids))
911
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
913
# the unversioned path lookup only occurs on real trees - where there
914
# can be extras. So the fake_entry is solely used to look up
915
# executable it values when execute is not supported.
916
fake_entry = InventoryFile('unused', 'unused', 'unused')
917
for to_path, to_entry in to_entries_by_dir:
918
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
919
unversioned_path = all_unversioned.popleft()
920
to_kind, to_executable, to_stat = \
921
self.target._comparison_data(fake_entry, unversioned_path[1])
922
yield (None, (None, unversioned_path[1]), True, (False, False),
924
(None, unversioned_path[0][-1]),
926
(None, to_executable))
927
file_id = to_entry.file_id
928
to_paths[file_id] = to_path
930
changed_content = False
931
from_path, from_entry = from_data.get(file_id, (None, None))
932
from_versioned = (from_entry is not None)
933
if from_entry is not None:
934
from_versioned = True
935
from_name = from_entry.name
936
from_parent = from_entry.parent_id
937
from_kind, from_executable, from_stat = \
938
self.source._comparison_data(from_entry, from_path)
941
from_versioned = False
945
from_executable = None
946
versioned = (from_versioned, True)
947
to_kind, to_executable, to_stat = \
948
self.target._comparison_data(to_entry, to_path)
949
kind = (from_kind, to_kind)
950
if kind[0] != kind[1]:
951
changed_content = True
952
elif from_kind == 'file':
953
if (self.source.get_file_sha1(file_id, from_path, from_stat) !=
954
self.target.get_file_sha1(file_id, to_path, to_stat)):
955
changed_content = True
956
elif from_kind == 'symlink':
957
if (self.source.get_symlink_target(file_id) !=
958
self.target.get_symlink_target(file_id)):
959
changed_content = True
960
elif from_kind == 'tree-reference':
961
if (self.source.get_reference_revision(file_id, from_path)
962
!= self.target.get_reference_revision(file_id, to_path)):
963
changed_content = True
964
parent = (from_parent, to_entry.parent_id)
965
name = (from_name, to_entry.name)
966
executable = (from_executable, to_executable)
968
pb.update('comparing files', entry_count, num_entries)
969
if (changed_content is not False or versioned[0] != versioned[1]
970
or parent[0] != parent[1] or name[0] != name[1] or
971
executable[0] != executable[1] or include_unchanged):
972
yield (file_id, (from_path, to_path), changed_content,
973
versioned, parent, name, kind, executable)
975
while all_unversioned:
976
# yield any trailing unversioned paths
977
unversioned_path = all_unversioned.popleft()
978
to_kind, to_executable, to_stat = \
979
self.target._comparison_data(fake_entry, unversioned_path[1])
980
yield (None, (None, unversioned_path[1]), True, (False, False),
982
(None, unversioned_path[0][-1]),
984
(None, to_executable))
986
def get_to_path(to_entry):
987
if to_entry.parent_id is None:
988
to_path = '' # the root
990
if to_entry.parent_id not in to_paths:
992
return get_to_path(self.target.inventory[to_entry.parent_id])
993
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
995
to_paths[to_entry.file_id] = to_path
998
for path, from_entry in from_entries_by_dir:
999
file_id = from_entry.file_id
1000
if file_id in to_paths:
1003
if not file_id in self.target.all_file_ids():
1004
# common case - paths we have not emitted are not present in
1008
to_path = get_to_path(self.target.inventory[file_id])
1011
pb.update('comparing files', entry_count, num_entries)
1012
versioned = (True, False)
1013
parent = (from_entry.parent_id, None)
1014
name = (from_entry.name, None)
1015
from_kind, from_executable, stat_value = \
1016
self.source._comparison_data(from_entry, path)
1017
kind = (from_kind, None)
1018
executable = (from_executable, None)
1019
changed_content = from_kind is not None
1020
# the parent's path is necessarily known at this point.
1021
yield(file_id, (path, to_path), changed_content, versioned, parent,
1022
name, kind, executable)
1025
class MultiWalker(object):
1026
"""Walk multiple trees simultaneously, getting combined results."""
1028
# Note: This could be written to not assume you can do out-of-order
1029
# lookups. Instead any nodes that don't match in all trees could be
1030
# marked as 'deferred', and then returned in the final cleanup loop.
1031
# For now, I think it is "nicer" to return things as close to the
1032
# "master_tree" order as we can.
1034
def __init__(self, master_tree, other_trees):
1035
"""Create a new MultiWalker.
1037
All trees being walked must implement "iter_entries_by_dir()", such
1038
that they yield (path, object) tuples, where that object will have a
1039
'.file_id' member, that can be used to check equality.
1041
:param master_tree: All trees will be 'slaved' to the master_tree such
1042
that nodes in master_tree will be used as 'first-pass' sync points.
1043
Any nodes that aren't in master_tree will be merged in a second
1045
:param other_trees: A list of other trees to walk simultaneously.
1047
self._master_tree = master_tree
1048
self._other_trees = other_trees
1050
# Keep track of any nodes that were properly processed just out of
1051
# order, that way we don't return them at the end, we don't have to
1052
# track *all* processed file_ids, just the out-of-order ones
1053
self._out_of_order_processed = set()
1056
def _step_one(iterator):
1057
"""Step an iter_entries_by_dir iterator.
1059
:return: (has_more, path, ie)
1060
If has_more is False, path and ie will be None.
1063
path, ie = iterator.next()
1064
except StopIteration:
1065
return False, None, None
1067
return True, path, ie
1070
def _cmp_path_by_dirblock(path1, path2):
1071
"""Compare two paths based on what directory they are in.
1073
This generates a sort order, such that all children of a directory are
1074
sorted together, and grandchildren are in the same order as the
1075
children appear. But all grandchildren come after all children.
1077
:param path1: first path
1078
:param path2: the second path
1079
:return: negative number if ``path1`` comes first,
1080
0 if paths are equal
1081
and a positive number if ``path2`` sorts first
1083
# Shortcut this special case
1086
# This is stolen from _dirstate_helpers_py.py, only switching it to
1087
# Unicode objects. Consider using encode_utf8() and then using the
1088
# optimized versions, or maybe writing optimized unicode versions.
1089
if not isinstance(path1, unicode):
1090
raise TypeError("'path1' must be a unicode string, not %s: %r"
1091
% (type(path1), path1))
1092
if not isinstance(path2, unicode):
1093
raise TypeError("'path2' must be a unicode string, not %s: %r"
1094
% (type(path2), path2))
1095
return cmp(MultiWalker._path_to_key(path1),
1096
MultiWalker._path_to_key(path2))
1099
def _path_to_key(path):
1100
dirname, basename = osutils.split(path)
1101
return (dirname.split(u'/'), basename)
1103
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1104
"""Lookup an inventory entry by file_id.
1106
This is called when an entry is missing in the normal order.
1107
Generally this is because a file was either renamed, or it was
1108
deleted/added. If the entry was found in the inventory and not in
1109
extra_entries, it will be added to self._out_of_order_processed
1111
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1112
should be filled with entries that were found before they were
1113
used. If file_id is present, it will be removed from the
1115
:param other_tree: The Tree to search, in case we didn't find the entry
1117
:param file_id: The file_id to look for
1118
:return: (path, ie) if found or (None, None) if not present.
1120
if file_id in extra_entries:
1121
return extra_entries.pop(file_id)
1122
# TODO: Is id2path better as the first call, or is
1123
# inventory[file_id] better as a first check?
1125
cur_path = other_tree.id2path(file_id)
1126
except errors.NoSuchId:
1128
if cur_path is None:
1131
self._out_of_order_processed.add(file_id)
1132
cur_ie = other_tree.inventory[file_id]
1133
return (cur_path, cur_ie)
1136
"""Match up the values in the different trees."""
1137
for result in self._walk_master_tree():
1139
self._finish_others()
1140
for result in self._walk_others():
1143
def _walk_master_tree(self):
1144
"""First pass, walk all trees in lock-step.
1146
When we are done, all nodes in the master_tree will have been
1147
processed. _other_walkers, _other_entries, and _others_extra will be
1148
set on 'self' for future processing.
1150
# This iterator has the most "inlining" done, because it tends to touch
1151
# every file in the tree, while the others only hit nodes that don't
1153
master_iterator = self._master_tree.iter_entries_by_dir()
1155
other_walkers = [other.iter_entries_by_dir()
1156
for other in self._other_trees]
1157
other_entries = [self._step_one(walker) for walker in other_walkers]
1158
# Track extra nodes in the other trees
1159
others_extra = [{} for i in xrange(len(self._other_trees))]
1161
master_has_more = True
1162
step_one = self._step_one
1163
lookup_by_file_id = self._lookup_by_file_id
1164
out_of_order_processed = self._out_of_order_processed
1166
while master_has_more:
1167
(master_has_more, path, master_ie) = step_one(master_iterator)
1168
if not master_has_more:
1171
file_id = master_ie.file_id
1173
other_values_append = other_values.append
1174
next_other_entries = []
1175
next_other_entries_append = next_other_entries.append
1176
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1177
if not other_has_more:
1178
other_values_append(lookup_by_file_id(
1179
others_extra[idx], self._other_trees[idx], file_id))
1180
next_other_entries_append((False, None, None))
1181
elif file_id == other_ie.file_id:
1182
# This is the critical code path, as most of the entries
1183
# should match between most trees.
1184
other_values_append((other_path, other_ie))
1185
next_other_entries_append(step_one(other_walkers[idx]))
1187
# This walker did not match, step it until it either
1188
# matches, or we know we are past the current walker.
1189
other_walker = other_walkers[idx]
1190
other_extra = others_extra[idx]
1191
while (other_has_more and
1192
self._cmp_path_by_dirblock(other_path, path) < 0):
1193
other_file_id = other_ie.file_id
1194
if other_file_id not in out_of_order_processed:
1195
other_extra[other_file_id] = (other_path, other_ie)
1196
other_has_more, other_path, other_ie = \
1197
step_one(other_walker)
1198
if other_has_more and other_ie.file_id == file_id:
1199
# We ended up walking to this point, match and step
1201
other_values_append((other_path, other_ie))
1202
other_has_more, other_path, other_ie = \
1203
step_one(other_walker)
1205
# This record isn't in the normal order, see if it
1207
other_values_append(lookup_by_file_id(
1208
other_extra, self._other_trees[idx], file_id))
1209
next_other_entries_append((other_has_more, other_path,
1211
other_entries = next_other_entries
1213
# We've matched all the walkers, yield this datapoint
1214
yield path, file_id, master_ie, other_values
1215
self._other_walkers = other_walkers
1216
self._other_entries = other_entries
1217
self._others_extra = others_extra
1219
def _finish_others(self):
1220
"""Finish walking the other iterators, so we get all entries."""
1221
for idx, info in enumerate(self._other_entries):
1222
other_extra = self._others_extra[idx]
1223
(other_has_more, other_path, other_ie) = info
1224
while other_has_more:
1225
other_file_id = other_ie.file_id
1226
if other_file_id not in self._out_of_order_processed:
1227
other_extra[other_file_id] = (other_path, other_ie)
1228
other_has_more, other_path, other_ie = \
1229
self._step_one(self._other_walkers[idx])
1230
del self._other_entries
1232
def _walk_others(self):
1233
"""Finish up by walking all the 'deferred' nodes."""
1234
# TODO: One alternative would be to grab all possible unprocessed
1235
# file_ids, and then sort by path, and then yield them. That
1236
# might ensure better ordering, in case a caller strictly
1237
# requires parents before children.
1238
for idx, other_extra in enumerate(self._others_extra):
1239
others = sorted(other_extra.itervalues(),
1240
key=lambda x: self._path_to_key(x[0]))
1241
for other_path, other_ie in others:
1242
file_id = other_ie.file_id
1243
# We don't need to check out_of_order_processed here, because
1244
# the lookup_by_file_id will be removing anything processed
1245
# from the extras cache
1246
other_extra.pop(file_id)
1247
other_values = [(None, None) for i in xrange(idx)]
1248
other_values.append((other_path, other_ie))
1249
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1250
alt_idx = alt_idx + idx + 1
1251
alt_extra = self._others_extra[alt_idx]
1252
alt_tree = self._other_trees[alt_idx]
1253
other_values.append(self._lookup_by_file_id(
1254
alt_extra, alt_tree, file_id))
1255
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
