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
def iter_changes(self, from_tree, include_unchanged=False,
99
specific_files=None, pb=None, extra_trees=None,
100
require_versioned=True, want_unversioned=False):
101
intertree = InterTree.get(from_tree, self)
102
return intertree.iter_changes(include_unchanged, specific_files, pb,
103
extra_trees, require_versioned, want_unversioned=want_unversioned)
106
"""Get a list of the conflicts in the tree.
108
Each conflict is an instance of bzrlib.conflicts.Conflict.
110
return _mod_conflicts.ConflictList()
113
"""For trees that can have unversioned files, return all such paths."""
116
def get_parent_ids(self):
117
"""Get the parent ids for this tree.
119
:return: a list of parent ids. [] is returned to indicate
120
a tree with no parents.
121
:raises: BzrError if the parents are not known.
123
raise NotImplementedError(self.get_parent_ids)
125
def has_filename(self, filename):
126
"""True if the tree has given filename."""
127
raise NotImplementedError(self.has_filename)
129
def has_id(self, file_id):
130
return self.inventory.has_id(file_id)
132
def __contains__(self, file_id):
133
return self.has_id(file_id)
135
def has_or_had_id(self, file_id):
136
if file_id == self.inventory.root.file_id:
138
return self.inventory.has_id(file_id)
140
def is_ignored(self, filename):
141
"""Check whether the filename is ignored by this tree.
143
:param filename: The relative filename within the tree.
144
:return: True if the filename is ignored.
149
return iter(self.inventory)
151
def all_file_ids(self):
152
"""Iterate through all file ids, including ids for missing files."""
153
return set(self.inventory)
155
def id2path(self, file_id):
156
"""Return the path for a file id.
160
return self.inventory.id2path(file_id)
162
def is_control_filename(self, filename):
163
"""True if filename is the name of a control file in this tree.
165
:param filename: A filename within the tree. This is a relative path
166
from the root of this tree.
168
This is true IF and ONLY IF the filename is part of the meta data
169
that bzr controls in this tree. I.E. a random .bzr directory placed
170
on disk will not be a control file for this tree.
172
return self.bzrdir.is_control_filename(filename)
175
def iter_entries_by_dir(self, specific_file_ids=None):
176
"""Walk the tree in 'by_dir' order.
178
This will yield each entry in the tree as a (path, entry) tuple.
179
The order that they are yielded is:
181
Directories are walked in a depth-first lexicographical order,
182
however, whenever a directory is reached, all of its direct child
183
nodes are yielded in lexicographical order before yielding the
186
For example, in the tree::
196
The yield order (ignoring root) would be::
197
a, f, a/b, a/d, a/b/c, a/d/e, f/g
199
return self.inventory.iter_entries_by_dir(
200
specific_file_ids=specific_file_ids)
202
def iter_references(self):
203
for path, entry in self.iter_entries_by_dir():
204
if entry.kind == 'tree-reference':
205
yield path, entry.file_id
207
def kind(self, file_id):
208
raise NotImplementedError("Tree subclass %s must implement kind"
209
% self.__class__.__name__)
211
def stored_kind(self, file_id):
212
"""File kind stored for this file_id.
214
May not match kind on disk for working trees. Always available
215
for versioned files, even when the file itself is missing.
217
return self.kind(file_id)
219
def path_content_summary(self, path):
220
"""Get a summary of the information about path.
222
:param path: A relative path within the tree.
223
:return: A tuple containing kind, size, exec, sha1-or-link.
224
Kind is always present (see tree.kind()).
225
size is present if kind is file, None otherwise.
226
exec is None unless kind is file and the platform supports the 'x'
228
sha1-or-link is the link target if kind is symlink, or the sha1 if
229
it can be obtained without reading the file.
231
raise NotImplementedError(self.path_content_summary)
233
def get_reference_revision(self, file_id, path=None):
234
raise NotImplementedError("Tree subclass %s must implement "
235
"get_reference_revision"
236
% self.__class__.__name__)
238
def _comparison_data(self, entry, path):
239
"""Return a tuple of kind, executable, stat_value for a file.
241
entry may be None if there is no inventory entry for the file, but
242
path must always be supplied.
244
kind is None if there is no file present (even if an inventory id is
245
present). executable is False for non-file entries.
247
raise NotImplementedError(self._comparison_data)
249
def _file_size(self, entry, stat_value):
250
raise NotImplementedError(self._file_size)
252
def _get_inventory(self):
253
return self._inventory
255
def get_file(self, file_id, path=None):
256
"""Return a file object for the file file_id in the tree.
258
If both file_id and path are defined, it is implementation defined as
259
to which one is used.
261
raise NotImplementedError(self.get_file)
263
def get_file_text(self, file_id, path=None):
264
"""Return the byte content of a file.
266
:param file_id: The file_id of the file.
267
:param path: The path of the file.
268
If both file_id and path are supplied, an implementation may use
271
my_file = self.get_file(file_id, path)
273
return my_file.read()
277
def get_file_lines(self, file_id, path=None):
278
"""Return the content of a file, as lines.
280
:param file_id: The file_id of the file.
281
:param path: The path of the file.
282
If both file_id and path are supplied, an implementation may use
285
return osutils.split_lines(self.get_file_text(file_id, path))
287
def get_file_mtime(self, file_id, path=None):
288
"""Return the modification time for a file.
290
:param file_id: The handle for this file.
291
:param path: The path that this file can be found at.
292
These must point to the same object.
294
raise NotImplementedError(self.get_file_mtime)
296
def get_file_size(self, file_id):
297
"""Return the size of a file in bytes.
299
This applies only to regular files. If invoked on directories or
300
symlinks, it will return None.
301
:param file_id: The file-id of the file
303
raise NotImplementedError(self.get_file_size)
305
def get_file_by_path(self, path):
306
return self.get_file(self._inventory.path2id(path), path)
308
def iter_files_bytes(self, desired_files):
309
"""Iterate through file contents.
311
Files will not necessarily be returned in the order they occur in
312
desired_files. No specific order is guaranteed.
314
Yields pairs of identifier, bytes_iterator. identifier is an opaque
315
value supplied by the caller as part of desired_files. It should
316
uniquely identify the file version in the caller's context. (Examples:
317
an index number or a TreeTransform trans_id.)
319
bytes_iterator is an iterable of bytestrings for the file. The
320
kind of iterable and length of the bytestrings are unspecified, but for
321
this implementation, it is a tuple containing a single bytestring with
322
the complete text of the file.
324
:param desired_files: a list of (file_id, identifier) pairs
326
for file_id, identifier in desired_files:
327
# We wrap the string in a tuple so that we can return an iterable
328
# of bytestrings. (Technically, a bytestring is also an iterable
329
# of bytestrings, but iterating through each character is not
331
cur_file = (self.get_file_text(file_id),)
332
yield identifier, cur_file
334
def get_symlink_target(self, file_id):
335
"""Get the target for a given file_id.
337
It is assumed that the caller already knows that file_id is referencing
339
:param file_id: Handle for the symlink entry.
340
:return: The path the symlink points to.
342
raise NotImplementedError(self.get_symlink_target)
344
def get_canonical_inventory_paths(self, paths):
345
"""Like get_canonical_inventory_path() but works on multiple items.
347
:param paths: A sequence of paths relative to the root of the tree.
348
:return: A list of paths, with each item the corresponding input path
349
adjusted to account for existing elements that match case
352
return list(self._yield_canonical_inventory_paths(paths))
354
def get_canonical_inventory_path(self, path):
355
"""Returns the first inventory item that case-insensitively matches path.
357
If a path matches exactly, it is returned. If no path matches exactly
358
but more than one path matches case-insensitively, it is implementation
359
defined which is returned.
361
If no path matches case-insensitively, the input path is returned, but
362
with as many path entries that do exist changed to their canonical
365
If you need to resolve many names from the same tree, you should
366
use get_canonical_inventory_paths() to avoid O(N) behaviour.
368
:param path: A paths relative to the root of the tree.
369
:return: The input path adjusted to account for existing elements
370
that match case insensitively.
372
return self._yield_canonical_inventory_paths([path]).next()
374
def _yield_canonical_inventory_paths(self, paths):
376
# First, if the path as specified exists exactly, just use it.
377
if self.path2id(path) is not None:
381
cur_id = self.get_root_id()
383
bit_iter = iter(path.split("/"))
386
for child in self.iter_children(cur_id):
388
child_base = os.path.basename(self.id2path(child))
389
if child_base.lower() == lelt:
391
cur_path = osutils.pathjoin(cur_path, child_base)
394
# before a change is committed we can see this error...
397
# got to the end of this directory and no entries matched.
398
# Return what matched so far, plus the rest as specified.
399
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
404
def get_root_id(self):
405
"""Return the file_id for the root of this tree."""
406
raise NotImplementedError(self.get_root_id)
408
def annotate_iter(self, file_id,
409
default_revision=_mod_revision.CURRENT_REVISION):
410
"""Return an iterator of revision_id, line tuples.
412
For working trees (and mutable trees in general), the special
413
revision_id 'current:' will be used for lines that are new in this
414
tree, e.g. uncommitted changes.
415
:param file_id: The file to produce an annotated version from
416
:param default_revision: For lines that don't match a basis, mark them
417
with this revision id. Not all implementations will make use of
420
raise NotImplementedError(self.annotate_iter)
422
def _get_plan_merge_data(self, file_id, other, base):
423
from bzrlib import merge, versionedfile
424
vf = versionedfile._PlanMergeVersionedFile(file_id)
425
last_revision_a = self._get_file_revision(file_id, vf, 'this:')
426
last_revision_b = other._get_file_revision(file_id, vf, 'other:')
428
last_revision_base = None
430
last_revision_base = base._get_file_revision(file_id, vf, 'base:')
431
return vf, last_revision_a, last_revision_b, last_revision_base
433
def plan_file_merge(self, file_id, other, base=None):
434
"""Generate a merge plan based on annotations.
436
If the file contains uncommitted changes in this tree, they will be
437
attributed to the 'current:' pseudo-revision. If the file contains
438
uncommitted changes in the other tree, they will be assigned to the
439
'other:' pseudo-revision.
441
data = self._get_plan_merge_data(file_id, other, base)
442
vf, last_revision_a, last_revision_b, last_revision_base = data
443
return vf.plan_merge(last_revision_a, last_revision_b,
446
def plan_file_lca_merge(self, file_id, other, base=None):
447
"""Generate a merge plan based lca-newness.
449
If the file contains uncommitted changes in this tree, they will be
450
attributed to the 'current:' pseudo-revision. If the file contains
451
uncommitted changes in the other tree, they will be assigned to the
452
'other:' pseudo-revision.
454
data = self._get_plan_merge_data(file_id, other, base)
455
vf, last_revision_a, last_revision_b, last_revision_base = data
456
return vf.plan_lca_merge(last_revision_a, last_revision_b,
459
def _iter_parent_trees(self):
460
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
461
for revision_id in self.get_parent_ids():
463
yield self.revision_tree(revision_id)
464
except errors.NoSuchRevisionInTree:
465
yield self.repository.revision_tree(revision_id)
468
def _file_revision(revision_tree, file_id):
469
"""Determine the revision associated with a file in a given tree."""
470
revision_tree.lock_read()
472
return revision_tree.inventory[file_id].revision
474
revision_tree.unlock()
476
def _get_file_revision(self, file_id, vf, tree_revision):
477
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
479
if getattr(self, '_repository', None) is None:
480
last_revision = tree_revision
481
parent_keys = [(file_id, self._file_revision(t, file_id)) for t in
482
self._iter_parent_trees()]
483
vf.add_lines((file_id, last_revision), parent_keys,
484
self.get_file(file_id).readlines())
485
repo = self.branch.repository
488
last_revision = self._file_revision(self, file_id)
489
base_vf = self._repository.texts
490
if base_vf not in vf.fallback_versionedfiles:
491
vf.fallback_versionedfiles.append(base_vf)
494
inventory = property(_get_inventory,
495
doc="Inventory of this Tree")
497
def _check_retrieved(self, ie, f):
500
fp = fingerprint_file(f)
503
if ie.text_size is not None:
504
if ie.text_size != fp['size']:
505
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
506
["inventory expects %d bytes" % ie.text_size,
507
"file is actually %d bytes" % fp['size'],
508
"store is probably damaged/corrupt"])
510
if ie.text_sha1 != fp['sha1']:
511
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
512
["inventory expects %s" % ie.text_sha1,
513
"file is actually %s" % fp['sha1'],
514
"store is probably damaged/corrupt"])
517
def path2id(self, path):
518
"""Return the id for path in this tree."""
519
return self._inventory.path2id(path)
521
def paths2ids(self, paths, trees=[], require_versioned=True):
522
"""Return all the ids that can be reached by walking from paths.
524
Each path is looked up in this tree and any extras provided in
525
trees, and this is repeated recursively: the children in an extra tree
526
of a directory that has been renamed under a provided path in this tree
527
are all returned, even if none exist under a provided path in this
528
tree, and vice versa.
530
:param paths: An iterable of paths to start converting to ids from.
531
Alternatively, if paths is None, no ids should be calculated and None
532
will be returned. This is offered to make calling the api unconditional
533
for code that *might* take a list of files.
534
:param trees: Additional trees to consider.
535
:param require_versioned: If False, do not raise NotVersionedError if
536
an element of paths is not versioned in this tree and all of trees.
538
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
540
def iter_children(self, file_id):
541
entry = self.iter_entries_by_dir([file_id]).next()[1]
542
for child in getattr(entry, 'children', {}).itervalues():
548
def revision_tree(self, revision_id):
549
"""Obtain a revision tree for the revision revision_id.
551
The intention of this method is to allow access to possibly cached
552
tree data. Implementors of this method should raise NoSuchRevision if
553
the tree is not locally available, even if they could obtain the
554
tree via a repository or some other means. Callers are responsible
555
for finding the ultimate source for a revision tree.
557
:param revision_id: The revision_id of the requested tree.
559
:raises: NoSuchRevision if the tree cannot be obtained.
561
raise errors.NoSuchRevisionInTree(self, revision_id)
564
"""What files are present in this tree and unknown.
566
:return: an iterator over the unknown files.
573
def filter_unversioned_files(self, paths):
574
"""Filter out paths that are versioned.
576
:return: set of paths.
578
# NB: we specifically *don't* call self.has_filename, because for
579
# WorkingTrees that can indicate files that exist on disk but that
581
pred = self.inventory.has_filename
582
return set((p for p in paths if not pred(p)))
584
def walkdirs(self, prefix=""):
585
"""Walk the contents of this tree from path down.
587
This yields all the data about the contents of a directory at a time.
588
After each directory has been yielded, if the caller has mutated the
589
list to exclude some directories, they are then not descended into.
591
The data yielded is of the form:
592
((directory-relpath, directory-path-from-root, directory-fileid),
593
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
594
versioned_kind), ...]),
595
- directory-relpath is the containing dirs relpath from prefix
596
- directory-path-from-root is the containing dirs path from /
597
- directory-fileid is the id of the directory if it is versioned.
598
- relpath is the relative path within the subtree being walked.
599
- basename is the basename
600
- kind is the kind of the file now. If unknonwn then the file is not
601
present within the tree - but it may be recorded as versioned. See
603
- lstat is the stat data *if* the file was statted.
604
- path_from_tree_root is the path from the root of the tree.
605
- file_id is the file_id if the entry is versioned.
606
- versioned_kind is the kind of the file as last recorded in the
607
versioning system. If 'unknown' the file is not versioned.
608
One of 'kind' and 'versioned_kind' must not be 'unknown'.
610
:param prefix: Start walking from prefix within the tree rather than
611
at the root. This allows one to walk a subtree but get paths that are
612
relative to a tree rooted higher up.
613
:return: an iterator over the directory data.
615
raise NotImplementedError(self.walkdirs)
617
def supports_content_filtering(self):
620
def iter_search_rules(self, path_names, pref_names=None,
621
_default_searcher=rules._per_user_searcher):
622
"""Find the preferences for filenames in a tree.
624
:param path_names: an iterable of paths to find attributes for.
625
Paths are given relative to the root of the tree.
626
:param pref_names: the list of preferences to lookup - None for all
627
:param _default_searcher: private parameter to assist testing - don't use
628
:return: an iterator of tuple sequences, one per path-name.
629
See _RulesSearcher.get_items for details on the tuple sequence.
631
searcher = self._get_rules_searcher(_default_searcher)
632
if searcher is not None:
633
if pref_names is not None:
634
for path in path_names:
635
yield searcher.get_selected_items(path, pref_names)
637
for path in path_names:
638
yield searcher.get_items(path)
641
def _get_rules_searcher(self, default_searcher):
642
"""Get the RulesSearcher for this tree given the default one."""
643
searcher = default_searcher
647
######################################################################
650
# TODO: Merge these two functions into a single one that can operate
651
# on either a whole tree or a set of files.
653
# TODO: Return the diff in order by filename, not by category or in
654
# random order. Can probably be done by lock-stepping through the
655
# filenames from both trees.
658
def file_status(filename, old_tree, new_tree):
659
"""Return single-letter status, old and new names for a file.
661
The complexity here is in deciding how to represent renames;
662
many complex cases are possible.
664
old_inv = old_tree.inventory
665
new_inv = new_tree.inventory
666
new_id = new_inv.path2id(filename)
667
old_id = old_inv.path2id(filename)
669
if not new_id and not old_id:
670
# easy: doesn't exist in either; not versioned at all
671
if new_tree.is_ignored(filename):
672
return 'I', None, None
674
return '?', None, None
676
# There is now a file of this name, great.
679
# There is no longer a file of this name, but we can describe
680
# what happened to the file that used to have
681
# this name. There are two possibilities: either it was
682
# deleted entirely, or renamed.
683
if new_inv.has_id(old_id):
684
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
686
return 'D', old_inv.id2path(old_id), None
688
# if the file_id is new in this revision, it is added
689
if new_id and not old_inv.has_id(new_id):
692
# if there used to be a file of this name, but that ID has now
693
# disappeared, it is deleted
694
if old_id and not new_inv.has_id(old_id):
700
@deprecated_function(deprecated_in((1, 9, 0)))
701
def find_renames(old_inv, new_inv):
702
for file_id in old_inv:
703
if file_id not in new_inv:
705
old_name = old_inv.id2path(file_id)
706
new_name = new_inv.id2path(file_id)
707
if old_name != new_name:
708
yield (old_name, new_name)
711
def find_ids_across_trees(filenames, trees, require_versioned=True):
712
"""Find the ids corresponding to specified filenames.
714
All matches in all trees will be used, and all children of matched
715
directories will be used.
717
:param filenames: The filenames to find file_ids for (if None, returns
719
:param trees: The trees to find file_ids within
720
:param require_versioned: if true, all specified filenames must occur in
722
:return: a set of file ids for the specified filenames and their children.
726
specified_path_ids = _find_ids_across_trees(filenames, trees,
728
return _find_children_across_trees(specified_path_ids, trees)
731
def _find_ids_across_trees(filenames, trees, require_versioned):
732
"""Find the ids corresponding to specified filenames.
734
All matches in all trees will be used, but subdirectories are not scanned.
736
:param filenames: The filenames to find file_ids for
737
:param trees: The trees to find file_ids within
738
:param require_versioned: if true, all specified filenames must occur in
740
:return: a set of file ids for the specified filenames
743
interesting_ids = set()
744
for tree_path in filenames:
747
file_id = tree.path2id(tree_path)
748
if file_id is not None:
749
interesting_ids.add(file_id)
752
not_versioned.append(tree_path)
753
if len(not_versioned) > 0 and require_versioned:
754
raise errors.PathsNotVersionedError(not_versioned)
755
return interesting_ids
758
def _find_children_across_trees(specified_ids, trees):
759
"""Return a set including specified ids and their children.
761
All matches in all trees will be used.
763
:param trees: The trees to find file_ids within
764
:return: a set containing all specified ids and their children
766
interesting_ids = set(specified_ids)
767
pending = interesting_ids
768
# now handle children of interesting ids
769
# we loop so that we handle all children of each id in both trees
770
while len(pending) > 0:
772
for file_id in pending:
774
if not tree.has_id(file_id):
776
for child_id in tree.iter_children(file_id):
777
if child_id not in interesting_ids:
778
new_pending.add(child_id)
779
interesting_ids.update(new_pending)
780
pending = new_pending
781
return interesting_ids
784
class InterTree(InterObject):
785
"""This class represents operations taking place between two Trees.
787
Its instances have methods like 'compare' and contain references to the
788
source and target trees these operations are to be carried out on.
790
Clients of bzrlib should not need to use InterTree directly, rather they
791
should use the convenience methods on Tree such as 'Tree.compare()' which
792
will pass through to InterTree as appropriate.
798
def compare(self, want_unchanged=False, specific_files=None,
799
extra_trees=None, require_versioned=False, include_root=False,
800
want_unversioned=False):
801
"""Return the changes from source to target.
803
:return: A TreeDelta.
804
:param specific_files: An optional list of file paths to restrict the
805
comparison to. When mapping filenames to ids, all matches in all
806
trees (including optional extra_trees) are used, and all children of
807
matched directories are included.
808
:param want_unchanged: An optional boolean requesting the inclusion of
809
unchanged entries in the result.
810
:param extra_trees: An optional list of additional trees to use when
811
mapping the contents of specific_files (paths) to file_ids.
812
:param require_versioned: An optional boolean (defaults to False). When
813
supplied and True all the 'specific_files' must be versioned, or
814
a PathsNotVersionedError will be thrown.
815
:param want_unversioned: Scan for unversioned paths.
817
# NB: show_status depends on being able to pass in non-versioned files
818
# and report them as unknown
819
trees = (self.source,)
820
if extra_trees is not None:
821
trees = trees + tuple(extra_trees)
822
# target is usually the newer tree:
823
specific_file_ids = self.target.paths2ids(specific_files, trees,
824
require_versioned=require_versioned)
825
if specific_files and not specific_file_ids:
826
# All files are unversioned, so just return an empty delta
827
# _compare_trees would think we want a complete delta
828
result = delta.TreeDelta()
829
fake_entry = InventoryFile('unused', 'unused', 'unused')
830
result.unversioned = [(path, None,
831
self.target._comparison_data(fake_entry, path)[0]) for path in
834
return delta._compare_trees(self.source, self.target, want_unchanged,
835
specific_files, include_root, extra_trees=extra_trees,
836
require_versioned=require_versioned,
837
want_unversioned=want_unversioned)
839
def iter_changes(self, include_unchanged=False,
840
specific_files=None, pb=None, extra_trees=[],
841
require_versioned=True, want_unversioned=False):
842
"""Generate an iterator of changes between trees.
845
(file_id, (path_in_source, path_in_target),
846
changed_content, versioned, parent, name, kind,
849
Changed_content is True if the file's content has changed. This
850
includes changes to its kind, and to a symlink's target.
852
versioned, parent, name, kind, executable are tuples of (from, to).
853
If a file is missing in a tree, its kind is None.
855
Iteration is done in parent-to-child order, relative to the target
858
There is no guarantee that all paths are in sorted order: the
859
requirement to expand the search due to renames may result in children
860
that should be found early being found late in the search, after
861
lexically later results have been returned.
862
:param require_versioned: Raise errors.PathsNotVersionedError if a
863
path in the specific_files list is not versioned in one of
864
source, target or extra_trees.
865
:param want_unversioned: Should unversioned files be returned in the
866
output. An unversioned file is defined as one with (False, False)
867
for the versioned pair.
870
lookup_trees = [self.source]
872
lookup_trees.extend(extra_trees)
873
if specific_files == []:
874
specific_file_ids = []
876
specific_file_ids = self.target.paths2ids(specific_files,
877
lookup_trees, require_versioned=require_versioned)
879
all_unversioned = sorted([(p.split('/'), p) for p in
881
if specific_files is None or
882
osutils.is_inside_any(specific_files, p)])
883
all_unversioned = deque(all_unversioned)
885
all_unversioned = deque()
887
from_entries_by_dir = list(self.source.iter_entries_by_dir(
888
specific_file_ids=specific_file_ids))
889
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
890
to_entries_by_dir = list(self.target.iter_entries_by_dir(
891
specific_file_ids=specific_file_ids))
892
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
894
# the unversioned path lookup only occurs on real trees - where there
895
# can be extras. So the fake_entry is solely used to look up
896
# executable it values when execute is not supported.
897
fake_entry = InventoryFile('unused', 'unused', 'unused')
898
for to_path, to_entry in to_entries_by_dir:
899
while all_unversioned and all_unversioned[0][0] < to_path.split('/'):
900
unversioned_path = all_unversioned.popleft()
901
to_kind, to_executable, to_stat = \
902
self.target._comparison_data(fake_entry, unversioned_path[1])
903
yield (None, (None, unversioned_path[1]), True, (False, False),
905
(None, unversioned_path[0][-1]),
907
(None, to_executable))
908
file_id = to_entry.file_id
909
to_paths[file_id] = to_path
911
changed_content = False
912
from_path, from_entry = from_data.get(file_id, (None, None))
913
from_versioned = (from_entry is not None)
914
if from_entry is not None:
915
from_versioned = True
916
from_name = from_entry.name
917
from_parent = from_entry.parent_id
918
from_kind, from_executable, from_stat = \
919
self.source._comparison_data(from_entry, from_path)
922
from_versioned = False
926
from_executable = None
927
versioned = (from_versioned, True)
928
to_kind, to_executable, to_stat = \
929
self.target._comparison_data(to_entry, to_path)
930
kind = (from_kind, to_kind)
931
if kind[0] != kind[1]:
932
changed_content = True
933
elif from_kind == 'file':
934
from_size = self.source._file_size(from_entry, from_stat)
935
to_size = self.target._file_size(to_entry, to_stat)
936
if from_size != to_size:
937
changed_content = True
938
elif (self.source.get_file_sha1(file_id, from_path, from_stat) !=
939
self.target.get_file_sha1(file_id, to_path, to_stat)):
940
changed_content = True
941
elif from_kind == 'symlink':
942
if (self.source.get_symlink_target(file_id) !=
943
self.target.get_symlink_target(file_id)):
944
changed_content = True
945
elif from_kind == 'tree-reference':
946
if (self.source.get_reference_revision(file_id, from_path)
947
!= self.target.get_reference_revision(file_id, to_path)):
948
changed_content = True
949
parent = (from_parent, to_entry.parent_id)
950
name = (from_name, to_entry.name)
951
executable = (from_executable, to_executable)
953
pb.update('comparing files', entry_count, num_entries)
954
if (changed_content is not False or versioned[0] != versioned[1]
955
or parent[0] != parent[1] or name[0] != name[1] or
956
executable[0] != executable[1] or include_unchanged):
957
yield (file_id, (from_path, to_path), changed_content,
958
versioned, parent, name, kind, executable)
960
while all_unversioned:
961
# yield any trailing unversioned paths
962
unversioned_path = all_unversioned.popleft()
963
to_kind, to_executable, to_stat = \
964
self.target._comparison_data(fake_entry, unversioned_path[1])
965
yield (None, (None, unversioned_path[1]), True, (False, False),
967
(None, unversioned_path[0][-1]),
969
(None, to_executable))
971
def get_to_path(to_entry):
972
if to_entry.parent_id is None:
973
to_path = '' # the root
975
if to_entry.parent_id not in to_paths:
977
return get_to_path(self.target.inventory[to_entry.parent_id])
978
to_path = osutils.pathjoin(to_paths[to_entry.parent_id],
980
to_paths[to_entry.file_id] = to_path
983
for path, from_entry in from_entries_by_dir:
984
file_id = from_entry.file_id
985
if file_id in to_paths:
988
if not file_id in self.target.all_file_ids():
989
# common case - paths we have not emitted are not present in
993
to_path = get_to_path(self.target.inventory[file_id])
996
pb.update('comparing files', entry_count, num_entries)
997
versioned = (True, False)
998
parent = (from_entry.parent_id, None)
999
name = (from_entry.name, None)
1000
from_kind, from_executable, stat_value = \
1001
self.source._comparison_data(from_entry, path)
1002
kind = (from_kind, None)
1003
executable = (from_executable, None)
1004
changed_content = from_kind is not None
1005
# the parent's path is necessarily known at this point.
1006
yield(file_id, (path, to_path), changed_content, versioned, parent,
1007
name, kind, executable)
1010
class MultiWalker(object):
1011
"""Walk multiple trees simultaneously, getting combined results."""
1013
# Note: This could be written to not assume you can do out-of-order
1014
# lookups. Instead any nodes that don't match in all trees could be
1015
# marked as 'deferred', and then returned in the final cleanup loop.
1016
# For now, I think it is "nicer" to return things as close to the
1017
# "master_tree" order as we can.
1019
def __init__(self, master_tree, other_trees):
1020
"""Create a new MultiWalker.
1022
All trees being walked must implement "iter_entries_by_dir()", such
1023
that they yield (path, object) tuples, where that object will have a
1024
'.file_id' member, that can be used to check equality.
1026
:param master_tree: All trees will be 'slaved' to the master_tree such
1027
that nodes in master_tree will be used as 'first-pass' sync points.
1028
Any nodes that aren't in master_tree will be merged in a second
1030
:param other_trees: A list of other trees to walk simultaneously.
1032
self._master_tree = master_tree
1033
self._other_trees = other_trees
1035
# Keep track of any nodes that were properly processed just out of
1036
# order, that way we don't return them at the end, we don't have to
1037
# track *all* processed file_ids, just the out-of-order ones
1038
self._out_of_order_processed = set()
1041
def _step_one(iterator):
1042
"""Step an iter_entries_by_dir iterator.
1044
:return: (has_more, path, ie)
1045
If has_more is False, path and ie will be None.
1048
path, ie = iterator.next()
1049
except StopIteration:
1050
return False, None, None
1052
return True, path, ie
1055
def _cmp_path_by_dirblock(path1, path2):
1056
"""Compare two paths based on what directory they are in.
1058
This generates a sort order, such that all children of a directory are
1059
sorted together, and grandchildren are in the same order as the
1060
children appear. But all grandchildren come after all children.
1062
:param path1: first path
1063
:param path2: the second path
1064
:return: negative number if ``path1`` comes first,
1065
0 if paths are equal
1066
and a positive number if ``path2`` sorts first
1068
# Shortcut this special case
1071
# This is stolen from _dirstate_helpers_py.py, only switching it to
1072
# Unicode objects. Consider using encode_utf8() and then using the
1073
# optimized versions, or maybe writing optimized unicode versions.
1074
if not isinstance(path1, unicode):
1075
raise TypeError("'path1' must be a unicode string, not %s: %r"
1076
% (type(path1), path1))
1077
if not isinstance(path2, unicode):
1078
raise TypeError("'path2' must be a unicode string, not %s: %r"
1079
% (type(path2), path2))
1080
return cmp(MultiWalker._path_to_key(path1),
1081
MultiWalker._path_to_key(path2))
1084
def _path_to_key(path):
1085
dirname, basename = osutils.split(path)
1086
return (dirname.split(u'/'), basename)
1088
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1089
"""Lookup an inventory entry by file_id.
1091
This is called when an entry is missing in the normal order.
1092
Generally this is because a file was either renamed, or it was
1093
deleted/added. If the entry was found in the inventory and not in
1094
extra_entries, it will be added to self._out_of_order_processed
1096
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1097
should be filled with entries that were found before they were
1098
used. If file_id is present, it will be removed from the
1100
:param other_tree: The Tree to search, in case we didn't find the entry
1102
:param file_id: The file_id to look for
1103
:return: (path, ie) if found or (None, None) if not present.
1105
if file_id in extra_entries:
1106
return extra_entries.pop(file_id)
1107
# TODO: Is id2path better as the first call, or is
1108
# inventory[file_id] better as a first check?
1110
cur_path = other_tree.id2path(file_id)
1111
except errors.NoSuchId:
1113
if cur_path is None:
1116
self._out_of_order_processed.add(file_id)
1117
cur_ie = other_tree.inventory[file_id]
1118
return (cur_path, cur_ie)
1121
"""Match up the values in the different trees."""
1122
for result in self._walk_master_tree():
1124
self._finish_others()
1125
for result in self._walk_others():
1128
def _walk_master_tree(self):
1129
"""First pass, walk all trees in lock-step.
1131
When we are done, all nodes in the master_tree will have been
1132
processed. _other_walkers, _other_entries, and _others_extra will be
1133
set on 'self' for future processing.
1135
# This iterator has the most "inlining" done, because it tends to touch
1136
# every file in the tree, while the others only hit nodes that don't
1138
master_iterator = self._master_tree.iter_entries_by_dir()
1140
other_walkers = [other.iter_entries_by_dir()
1141
for other in self._other_trees]
1142
other_entries = [self._step_one(walker) for walker in other_walkers]
1143
# Track extra nodes in the other trees
1144
others_extra = [{} for i in xrange(len(self._other_trees))]
1146
master_has_more = True
1147
step_one = self._step_one
1148
lookup_by_file_id = self._lookup_by_file_id
1149
out_of_order_processed = self._out_of_order_processed
1151
while master_has_more:
1152
(master_has_more, path, master_ie) = step_one(master_iterator)
1153
if not master_has_more:
1156
file_id = master_ie.file_id
1158
other_values_append = other_values.append
1159
next_other_entries = []
1160
next_other_entries_append = next_other_entries.append
1161
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1162
if not other_has_more:
1163
other_values_append(lookup_by_file_id(
1164
others_extra[idx], self._other_trees[idx], file_id))
1165
next_other_entries_append((False, None, None))
1166
elif file_id == other_ie.file_id:
1167
# This is the critical code path, as most of the entries
1168
# should match between most trees.
1169
other_values_append((other_path, other_ie))
1170
next_other_entries_append(step_one(other_walkers[idx]))
1172
# This walker did not match, step it until it either
1173
# matches, or we know we are past the current walker.
1174
other_walker = other_walkers[idx]
1175
other_extra = others_extra[idx]
1176
while (other_has_more and
1177
self._cmp_path_by_dirblock(other_path, path) < 0):
1178
other_file_id = other_ie.file_id
1179
if other_file_id not in out_of_order_processed:
1180
other_extra[other_file_id] = (other_path, other_ie)
1181
other_has_more, other_path, other_ie = \
1182
step_one(other_walker)
1183
if other_has_more and other_ie.file_id == file_id:
1184
# We ended up walking to this point, match and step
1186
other_values_append((other_path, other_ie))
1187
other_has_more, other_path, other_ie = \
1188
step_one(other_walker)
1190
# This record isn't in the normal order, see if it
1192
other_values_append(lookup_by_file_id(
1193
other_extra, self._other_trees[idx], file_id))
1194
next_other_entries_append((other_has_more, other_path,
1196
other_entries = next_other_entries
1198
# We've matched all the walkers, yield this datapoint
1199
yield path, file_id, master_ie, other_values
1200
self._other_walkers = other_walkers
1201
self._other_entries = other_entries
1202
self._others_extra = others_extra
1204
def _finish_others(self):
1205
"""Finish walking the other iterators, so we get all entries."""
1206
for idx, info in enumerate(self._other_entries):
1207
other_extra = self._others_extra[idx]
1208
(other_has_more, other_path, other_ie) = info
1209
while other_has_more:
1210
other_file_id = other_ie.file_id
1211
if other_file_id not in self._out_of_order_processed:
1212
other_extra[other_file_id] = (other_path, other_ie)
1213
other_has_more, other_path, other_ie = \
1214
self._step_one(self._other_walkers[idx])
1215
del self._other_entries
1217
def _walk_others(self):
1218
"""Finish up by walking all the 'deferred' nodes."""
1219
# TODO: One alternative would be to grab all possible unprocessed
1220
# file_ids, and then sort by path, and then yield them. That
1221
# might ensure better ordering, in case a caller strictly
1222
# requires parents before children.
1223
for idx, other_extra in enumerate(self._others_extra):
1224
others = sorted(other_extra.itervalues(),
1225
key=lambda x: self._path_to_key(x[0]))
1226
for other_path, other_ie in others:
1227
file_id = other_ie.file_id
1228
# We don't need to check out_of_order_processed here, because
1229
# the lookup_by_file_id will be removing anything processed
1230
# from the extras cache
1231
other_extra.pop(file_id)
1232
other_values = [(None, None) for i in xrange(idx)]
1233
other_values.append((other_path, other_ie))
1234
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1235
alt_idx = alt_idx + idx + 1
1236
alt_extra = self._others_extra[alt_idx]
1237
alt_tree = self._other_trees[alt_idx]
1238
other_values.append(self._lookup_by_file_id(
1239
alt_extra, alt_tree, file_id))
1240
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
