84
99
specific_files=specific_files,
85
100
extra_trees=extra_trees,
86
101
require_versioned=require_versioned,
87
include_root=include_root
102
include_root=include_root,
103
want_unversioned=want_unversioned,
90
def _iter_changes(self, from_tree, include_unchanged=False,
91
specific_file_ids=None, pb=None):
106
def iter_changes(self, from_tree, include_unchanged=False,
107
specific_files=None, pb=None, extra_trees=None,
108
require_versioned=True, want_unversioned=False):
109
"""See InterTree.iter_changes"""
92
110
intertree = InterTree.get(from_tree, self)
93
return intertree._iter_changes(from_tree, self, include_unchanged,
94
specific_file_ids, pb)
111
return intertree.iter_changes(include_unchanged, specific_files, pb,
112
extra_trees, require_versioned, want_unversioned=want_unversioned)
96
114
def conflicts(self):
97
115
"""Get a list of the conflicts in the tree.
99
117
Each conflict is an instance of bzrlib.conflicts.Conflict.
119
return _mod_conflicts.ConflictList()
122
"""For trees that can have unversioned files, return all such paths."""
103
125
def get_parent_ids(self):
104
"""Get the parent ids for this tree.
126
"""Get the parent ids for this tree.
106
128
:return: a list of parent ids. [] is returned to indicate
107
129
a tree with no parents.
108
130
:raises: BzrError if the parents are not known.
110
132
raise NotImplementedError(self.get_parent_ids)
112
134
def has_filename(self, filename):
113
135
"""True if the tree has given filename."""
114
raise NotImplementedError()
136
raise NotImplementedError(self.has_filename)
116
138
def has_id(self, file_id):
117
return self.inventory.has_id(file_id)
139
raise NotImplementedError(self.has_id)
119
__contains__ = has_id
141
@deprecated_method(deprecated_in((2, 4, 0)))
142
def __contains__(self, file_id):
143
return self.has_id(file_id)
121
145
def has_or_had_id(self, file_id):
122
if file_id == self.inventory.root.file_id:
124
return self.inventory.has_id(file_id)
127
return iter(self.inventory)
146
raise NotImplementedError(self.has_or_had_id)
148
def is_ignored(self, filename):
149
"""Check whether the filename is ignored by this tree.
151
:param filename: The relative filename within the tree.
152
:return: True if the filename is ignored.
156
def all_file_ids(self):
157
"""Iterate through all file ids, including ids for missing files."""
158
raise NotImplementedError(self.all_file_ids)
129
160
def id2path(self, file_id):
130
return self.inventory.id2path(file_id)
132
def is_control_filename(self, filename):
133
"""True if filename is the name of a control file in this tree.
135
:param filename: A filename within the tree. This is a relative path
136
from the root of this tree.
138
This is true IF and ONLY IF the filename is part of the meta data
139
that bzr controls in this tree. I.E. a random .bzr directory placed
140
on disk will not be a control file for this tree.
161
"""Return the path for a file id.
142
return self.bzrdir.is_control_filename(filename)
165
raise NotImplementedError(self.id2path)
145
def iter_entries_by_dir(self, specific_file_ids=None):
167
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
146
168
"""Walk the tree in 'by_dir' order.
148
This will yield each entry in the tree as a (path, entry) tuple. The
149
order that they are yielded is: the contents of a directory are
150
preceeded by the parent of a directory, and all the contents of a
151
directory are grouped together.
153
return self.inventory.iter_entries_by_dir(
154
specific_file_ids=specific_file_ids)
170
This will yield each entry in the tree as a (path, entry) tuple.
171
The order that they are yielded is:
173
Directories are walked in a depth-first lexicographical order,
174
however, whenever a directory is reached, all of its direct child
175
nodes are yielded in lexicographical order before yielding the
178
For example, in the tree::
188
The yield order (ignoring root) would be::
190
a, f, a/b, a/d, a/b/c, a/d/e, f/g
192
:param yield_parents: If True, yield the parents from the root leading
193
down to specific_file_ids that have been requested. This has no
194
impact if specific_file_ids is None.
196
raise NotImplementedError(self.iter_entries_by_dir)
198
def list_files(self, include_root=False, from_dir=None, recursive=True):
199
"""List all files in this tree.
201
:param include_root: Whether to include the entry for the tree root
202
:param from_dir: Directory under which to list files
203
:param recursive: Whether to list files recursively
204
:return: iterator over tuples of (path, versioned, kind, file_id,
207
raise NotImplementedError(self.list_files)
209
def iter_references(self):
210
if self.supports_tree_reference():
211
for path, entry in self.iter_entries_by_dir():
212
if entry.kind == 'tree-reference':
213
yield path, entry.file_id
156
215
def kind(self, file_id):
157
raise NotImplementedError("subclasses must implement kind")
216
raise NotImplementedError("Tree subclass %s must implement kind"
217
% self.__class__.__name__)
219
def stored_kind(self, file_id):
220
"""File kind stored for this file_id.
222
May not match kind on disk for working trees. Always available
223
for versioned files, even when the file itself is missing.
225
return self.kind(file_id)
227
def path_content_summary(self, path):
228
"""Get a summary of the information about path.
230
All the attributes returned are for the canonical form, not the
231
convenient form (if content filters are in use.)
233
:param path: A relative path within the tree.
234
:return: A tuple containing kind, size, exec, sha1-or-link.
235
Kind is always present (see tree.kind()).
236
size is present if kind is file and the size of the
237
canonical form can be cheaply determined, None otherwise.
238
exec is None unless kind is file and the platform supports the 'x'
240
sha1-or-link is the link target if kind is symlink, or the sha1 if
241
it can be obtained without reading the file.
243
raise NotImplementedError(self.path_content_summary)
245
def get_reference_revision(self, file_id, path=None):
246
raise NotImplementedError("Tree subclass %s must implement "
247
"get_reference_revision"
248
% self.__class__.__name__)
159
250
def _comparison_data(self, entry, path):
160
251
"""Return a tuple of kind, executable, stat_value for a file.
170
261
def _file_size(self, entry, stat_value):
171
262
raise NotImplementedError(self._file_size)
173
def _get_inventory(self):
174
return self._inventory
176
def get_file(self, file_id):
177
"""Return a file object for the file file_id in the tree."""
264
def get_file(self, file_id, path=None):
265
"""Return a file object for the file file_id in the tree.
267
If both file_id and path are defined, it is implementation defined as
268
to which one is used.
178
270
raise NotImplementedError(self.get_file)
180
def get_file_by_path(self, path):
181
return self.get_file(self._inventory.path2id(path))
183
def annotate_iter(self, file_id):
184
"""Return an iterator of revision_id, line tuples
272
def get_file_with_stat(self, file_id, path=None):
273
"""Get a file handle and stat object for file_id.
275
The default implementation returns (self.get_file, None) for backwards
278
:param file_id: The file id to read.
279
:param path: The path of the file, if it is known.
280
:return: A tuple (file_handle, stat_value_or_None). If the tree has
281
no stat facility, or need for a stat cache feedback during commit,
282
it may return None for the second element of the tuple.
284
return (self.get_file(file_id, path), None)
286
def get_file_text(self, file_id, path=None):
287
"""Return the byte content of a file.
289
:param file_id: The file_id of the file.
290
:param path: The path of the file.
292
If both file_id and path are supplied, an implementation may use
295
:returns: A single byte string for the whole file.
297
my_file = self.get_file(file_id, path)
299
return my_file.read()
303
def get_file_lines(self, file_id, path=None):
304
"""Return the content of a file, as lines.
306
:param file_id: The file_id of the file.
307
:param path: The path of the file.
309
If both file_id and path are supplied, an implementation may use
312
return osutils.split_lines(self.get_file_text(file_id, path))
314
def get_file_verifier(self, file_id, path=None, stat_value=None):
315
"""Return a verifier for a file.
317
The default implementation returns a sha1.
319
:param file_id: The handle for this file.
320
:param path: The path that this file can be found at.
321
These must point to the same object.
322
:param stat_value: Optional stat value for the object
323
:return: Tuple with verifier name and verifier data
325
return ("SHA1", self.get_file_sha1(file_id, path=path,
326
stat_value=stat_value))
328
def get_file_sha1(self, file_id, path=None, stat_value=None):
329
"""Return the SHA1 file for a file.
331
:note: callers should use get_file_verifier instead
332
where possible, as the underlying repository implementation may
333
have quicker access to a non-sha1 verifier.
335
:param file_id: The handle for this file.
336
:param path: The path that this file can be found at.
337
These must point to the same object.
338
:param stat_value: Optional stat value for the object
340
raise NotImplementedError(self.get_file_sha1)
342
def get_file_mtime(self, file_id, path=None):
343
"""Return the modification time for a file.
345
:param file_id: The handle for this file.
346
:param path: The path that this file can be found at.
347
These must point to the same object.
349
raise NotImplementedError(self.get_file_mtime)
351
def get_file_size(self, file_id):
352
"""Return the size of a file in bytes.
354
This applies only to regular files. If invoked on directories or
355
symlinks, it will return None.
356
:param file_id: The file-id of the file
358
raise NotImplementedError(self.get_file_size)
360
def is_executable(self, file_id, path=None):
361
"""Check if a file is executable.
363
:param file_id: The handle for this file.
364
:param path: The path that this file can be found at.
365
These must point to the same object.
367
raise NotImplementedError(self.is_executable)
369
def iter_files_bytes(self, desired_files):
370
"""Iterate through file contents.
372
Files will not necessarily be returned in the order they occur in
373
desired_files. No specific order is guaranteed.
375
Yields pairs of identifier, bytes_iterator. identifier is an opaque
376
value supplied by the caller as part of desired_files. It should
377
uniquely identify the file version in the caller's context. (Examples:
378
an index number or a TreeTransform trans_id.)
380
bytes_iterator is an iterable of bytestrings for the file. The
381
kind of iterable and length of the bytestrings are unspecified, but for
382
this implementation, it is a tuple containing a single bytestring with
383
the complete text of the file.
385
:param desired_files: a list of (file_id, identifier) pairs
387
for file_id, identifier in desired_files:
388
# We wrap the string in a tuple so that we can return an iterable
389
# of bytestrings. (Technically, a bytestring is also an iterable
390
# of bytestrings, but iterating through each character is not
392
cur_file = (self.get_file_text(file_id),)
393
yield identifier, cur_file
395
def get_symlink_target(self, file_id, path=None):
396
"""Get the target for a given file_id.
398
It is assumed that the caller already knows that file_id is referencing
400
:param file_id: Handle for the symlink entry.
401
:param path: The path of the file.
402
If both file_id and path are supplied, an implementation may use
404
:return: The path the symlink points to.
406
raise NotImplementedError(self.get_symlink_target)
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.
186
416
For working trees (and mutable trees in general), the special
187
417
revision_id 'current:' will be used for lines that are new in this
188
418
tree, e.g. uncommitted changes.
189
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
191
424
raise NotImplementedError(self.annotate_iter)
193
inventory = property(_get_inventory,
194
doc="Inventory of this Tree")
426
def _get_plan_merge_data(self, file_id, other, base):
427
from bzrlib import 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)
471
def _get_file_revision(self, file_id, vf, tree_revision):
472
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
474
if getattr(self, '_repository', None) is None:
475
last_revision = tree_revision
476
parent_keys = [(file_id, t.get_file_revision(file_id)) for t in
477
self._iter_parent_trees()]
478
vf.add_lines((file_id, last_revision), parent_keys,
479
self.get_file_lines(file_id))
480
repo = self.branch.repository
483
last_revision = self.get_file_revision(file_id)
484
base_vf = self._repository.texts
485
if base_vf not in vf.fallback_versionedfiles:
486
vf.fallback_versionedfiles.append(base_vf)
196
489
def _check_retrieved(self, ie, f):
197
490
if not __debug__:
199
fp = fingerprint_file(f)
492
fp = osutils.fingerprint_file(f)
202
495
if ie.text_size is not None:
203
496
if ie.text_size != fp['size']:
204
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
497
raise errors.BzrError(
498
"mismatched size for file %r in %r" %
499
(ie.file_id, self._store),
205
500
["inventory expects %d bytes" % ie.text_size,
206
501
"file is actually %d bytes" % fp['size'],
207
502
"store is probably damaged/corrupt"])
209
504
if ie.text_sha1 != fp['sha1']:
210
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
505
raise errors.BzrError("wrong SHA-1 for file %r in %r" %
506
(ie.file_id, self._store),
211
507
["inventory expects %s" % ie.text_sha1,
212
508
"file is actually %s" % fp['sha1'],
213
509
"store is probably damaged/corrupt"])
215
511
def path2id(self, path):
216
512
"""Return the id for path in this tree."""
217
return self._inventory.path2id(path)
219
def print_file(self, file_id):
220
"""Print file with id `file_id` to stdout."""
222
sys.stdout.write(self.get_file_text(file_id))
513
raise NotImplementedError(self.path2id)
515
def paths2ids(self, paths, trees=[], require_versioned=True):
516
"""Return all the ids that can be reached by walking from paths.
518
Each path is looked up in this tree and any extras provided in
519
trees, and this is repeated recursively: the children in an extra tree
520
of a directory that has been renamed under a provided path in this tree
521
are all returned, even if none exist under a provided path in this
522
tree, and vice versa.
524
:param paths: An iterable of paths to start converting to ids from.
525
Alternatively, if paths is None, no ids should be calculated and None
526
will be returned. This is offered to make calling the api unconditional
527
for code that *might* take a list of files.
528
:param trees: Additional trees to consider.
529
:param require_versioned: If False, do not raise NotVersionedError if
530
an element of paths is not versioned in this tree and all of trees.
532
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
534
def iter_children(self, file_id):
535
entry = self.iter_entries_by_dir([file_id]).next()[1]
536
for child in getattr(entry, 'children', {}).itervalues():
224
539
def lock_read(self):
540
"""Lock this tree for multiple read only operations.
542
:return: A bzrlib.lock.LogicalLockResult.
227
546
def revision_tree(self, revision_id):
294
609
raise NotImplementedError(self.walkdirs)
297
class EmptyTree(Tree):
300
self._inventory = Inventory(root_id=None)
301
symbol_versioning.warn('EmptyTree is deprecated as of bzr 0.9 please'
302
' use repository.revision_tree instead.',
303
DeprecationWarning, stacklevel=2)
305
def get_parent_ids(self):
308
def get_symlink_target(self, file_id):
311
def has_filename(self, filename):
611
def supports_content_filtering(self):
314
def kind(self, file_id):
315
assert self._inventory[file_id].kind == "directory"
318
def list_files(self, include_root=False):
321
def __contains__(self, file_id):
322
return (file_id in self._inventory)
324
def get_file_sha1(self, file_id, path=None, stat_value=None):
328
######################################################################
331
# TODO: Merge these two functions into a single one that can operate
332
# on either a whole tree or a set of files.
334
# TODO: Return the diff in order by filename, not by category or in
335
# random order. Can probably be done by lock-stepping through the
336
# filenames from both trees.
339
def file_status(filename, old_tree, new_tree):
340
"""Return single-letter status, old and new names for a file.
342
The complexity here is in deciding how to represent renames;
343
many complex cases are possible.
614
def _content_filter_stack(self, path=None, file_id=None):
615
"""The stack of content filters for a path if filtering is supported.
617
Readers will be applied in first-to-last order.
618
Writers will be applied in last-to-first order.
619
Either the path or the file-id needs to be provided.
621
:param path: path relative to the root of the tree
623
:param file_id: file_id or None if unknown
624
:return: the list of filters - [] if there are none
626
filter_pref_names = filters._get_registered_names()
627
if len(filter_pref_names) == 0:
630
path = self.id2path(file_id)
631
prefs = self.iter_search_rules([path], filter_pref_names).next()
632
stk = filters._get_filter_stack_for(prefs)
633
if 'filters' in debug.debug_flags:
634
trace.note(gettext("*** {0} content-filter: {1} => {2!r}").format(path,prefs,stk))
637
def _content_filter_stack_provider(self):
638
"""A function that returns a stack of ContentFilters.
640
The function takes a path (relative to the top of the tree) and a
641
file-id as parameters.
643
:return: None if content filtering is not supported by this tree.
645
if self.supports_content_filtering():
646
return lambda path, file_id: \
647
self._content_filter_stack(path, file_id)
651
def iter_search_rules(self, path_names, pref_names=None,
652
_default_searcher=None):
653
"""Find the preferences for filenames in a tree.
655
:param path_names: an iterable of paths to find attributes for.
656
Paths are given relative to the root of the tree.
657
:param pref_names: the list of preferences to lookup - None for all
658
:param _default_searcher: private parameter to assist testing - don't use
659
:return: an iterator of tuple sequences, one per path-name.
660
See _RulesSearcher.get_items for details on the tuple sequence.
662
if _default_searcher is None:
663
_default_searcher = rules._per_user_searcher
664
searcher = self._get_rules_searcher(_default_searcher)
665
if searcher is not None:
666
if pref_names is not None:
667
for path in path_names:
668
yield searcher.get_selected_items(path, pref_names)
670
for path in path_names:
671
yield searcher.get_items(path)
673
def _get_rules_searcher(self, default_searcher):
674
"""Get the RulesSearcher for this tree given the default one."""
675
searcher = default_searcher
679
class InventoryTree(Tree):
680
"""A tree that relies on an inventory for its metadata.
682
Trees contain an `Inventory` object, and also know how to retrieve
683
file texts mentioned in the inventory, either from a working
684
directory or from a store.
686
It is possible for trees to contain files that are not described
687
in their inventory or vice versa; for this use `filenames()`.
689
Subclasses should set the _inventory attribute, which is considered
690
private to external API users.
345
old_inv = old_tree.inventory
346
new_inv = new_tree.inventory
347
new_id = new_inv.path2id(filename)
348
old_id = old_inv.path2id(filename)
350
if not new_id and not old_id:
351
# easy: doesn't exist in either; not versioned at all
352
if new_tree.is_ignored(filename):
353
return 'I', None, None
355
return '?', None, None
357
# There is now a file of this name, great.
360
# There is no longer a file of this name, but we can describe
361
# what happened to the file that used to have
362
# this name. There are two possibilities: either it was
363
# deleted entirely, or renamed.
365
if new_inv.has_id(old_id):
366
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
368
return 'D', old_inv.id2path(old_id), None
370
# if the file_id is new in this revision, it is added
371
if new_id and not old_inv.has_id(new_id):
374
# if there used to be a file of this name, but that ID has now
375
# disappeared, it is deleted
376
if old_id and not new_inv.has_id(old_id):
383
def find_renames(old_inv, new_inv):
384
for file_id in old_inv:
385
if file_id not in new_inv:
387
old_name = old_inv.id2path(file_id)
388
new_name = new_inv.id2path(file_id)
389
if old_name != new_name:
390
yield (old_name, new_name)
693
def get_canonical_inventory_paths(self, paths):
694
"""Like get_canonical_inventory_path() but works on multiple items.
696
:param paths: A sequence of paths relative to the root of the tree.
697
:return: A list of paths, with each item the corresponding input path
698
adjusted to account for existing elements that match case
701
return list(self._yield_canonical_inventory_paths(paths))
703
def get_canonical_inventory_path(self, path):
704
"""Returns the first inventory item that case-insensitively matches path.
706
If a path matches exactly, it is returned. If no path matches exactly
707
but more than one path matches case-insensitively, it is implementation
708
defined which is returned.
710
If no path matches case-insensitively, the input path is returned, but
711
with as many path entries that do exist changed to their canonical
714
If you need to resolve many names from the same tree, you should
715
use get_canonical_inventory_paths() to avoid O(N) behaviour.
717
:param path: A paths relative to the root of the tree.
718
:return: The input path adjusted to account for existing elements
719
that match case insensitively.
721
return self._yield_canonical_inventory_paths([path]).next()
723
def _yield_canonical_inventory_paths(self, paths):
725
# First, if the path as specified exists exactly, just use it.
726
if self.path2id(path) is not None:
730
cur_id = self.get_root_id()
732
bit_iter = iter(path.split("/"))
736
for child in self.iter_children(cur_id):
738
# XXX: it seem like if the child is known to be in the
739
# tree, we shouldn't need to go from its id back to
740
# its path -- mbp 2010-02-11
742
# XXX: it seems like we could be more efficient
743
# by just directly looking up the original name and
744
# only then searching all children; also by not
745
# chopping paths so much. -- mbp 2010-02-11
746
child_base = os.path.basename(self.id2path(child))
747
if (child_base == elt):
748
# if we found an exact match, we can stop now; if
749
# we found an approximate match we need to keep
750
# searching because there might be an exact match
753
new_path = osutils.pathjoin(cur_path, child_base)
755
elif child_base.lower() == lelt:
757
new_path = osutils.pathjoin(cur_path, child_base)
758
except errors.NoSuchId:
759
# before a change is committed we can see this error...
764
# got to the end of this directory and no entries matched.
765
# Return what matched so far, plus the rest as specified.
766
cur_path = osutils.pathjoin(cur_path, elt, *list(bit_iter))
771
def _get_inventory(self):
772
return self._inventory
774
inventory = property(_get_inventory,
775
doc="Inventory of this Tree")
777
def _unpack_file_id(self, file_id):
778
"""Find the inventory and inventory file id for a tree file id.
780
:param file_id: The tree file id, as bytestring or tuple
781
:return: Inventory and inventory file id
783
if isinstance(file_id, tuple):
784
if len(file_id) != 1:
785
raise ValueError("nested trees not yet supported: %r" % file_id)
787
return self.inventory, file_id
790
def path2id(self, path):
791
"""Return the id for path in this tree."""
792
return self._path2inv_file_id(path)[1]
794
def _path2inv_file_id(self, path):
795
"""Lookup a inventory and inventory file id by path.
797
:param path: Path to look up
798
:return: tuple with inventory and inventory file id
800
return self.inventory, self.inventory.path2id(path)
802
def id2path(self, file_id):
803
"""Return the path for a file id.
807
inventory, file_id = self._unpack_file_id(file_id)
808
return inventory.id2path(file_id)
810
def has_id(self, file_id):
811
inventory, file_id = self._unpack_file_id(file_id)
812
return inventory.has_id(file_id)
814
def has_or_had_id(self, file_id):
815
inventory, file_id = self._unpack_file_id(file_id)
816
return inventory.has_id(file_id)
818
def all_file_ids(self):
820
[entry.file_id for path, entry in self.iter_entries_by_dir()])
822
@deprecated_method(deprecated_in((2, 4, 0)))
824
return iter(self.all_file_ids())
826
def filter_unversioned_files(self, paths):
827
"""Filter out paths that are versioned.
829
:return: set of paths.
831
# NB: we specifically *don't* call self.has_filename, because for
832
# WorkingTrees that can indicate files that exist on disk but that
834
return set((p for p in paths if self.path2id(p) is None))
837
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
838
"""Walk the tree in 'by_dir' order.
840
This will yield each entry in the tree as a (path, entry) tuple.
841
The order that they are yielded is:
843
See Tree.iter_entries_by_dir for details.
845
:param yield_parents: If True, yield the parents from the root leading
846
down to specific_file_ids that have been requested. This has no
847
impact if specific_file_ids is None.
849
if specific_file_ids is None:
850
inventory_file_ids = None
852
inventory_file_ids = []
853
for tree_file_id in specific_file_ids:
854
inventory, inv_file_id = self._unpack_file_id(tree_file_id)
855
if not inventory is self.inventory: # for now
856
raise AssertionError("%r != %r" % (
857
inventory, self.inventory))
858
inventory_file_ids.append(inv_file_id)
859
return self.inventory.iter_entries_by_dir(
860
specific_file_ids=inventory_file_ids, yield_parents=yield_parents)
862
@deprecated_method(deprecated_in((2, 5, 0)))
863
def get_file_by_path(self, path):
864
return self.get_file(self.path2id(path), path)
393
867
def find_ids_across_trees(filenames, trees, require_versioned=True):
394
868
"""Find the ids corresponding to specified filenames.
396
870
All matches in all trees will be used, and all children of matched
397
871
directories will be used.
408
882
specified_path_ids = _find_ids_across_trees(filenames, trees,
409
883
require_versioned)
410
884
return _find_children_across_trees(specified_path_ids, trees)
411
# specified_ids = [id for path, id in _find_path_ids_across_trees(filenames, trees, require_versioned)]
412
# return _find_children_across_trees(specified_ids, trees)
414
def find_path_ids_across_trees(filenames, trees, require_versioned=True):
415
"""Find the paths and ids corresponding to specified filenames.
417
All matches in all trees will be used, and all children of matched
418
directories will be included
420
:param filenames: The filenames to find file_ids for
421
:param trees: The trees to find file_ids within
422
:param require_versioned: if true, all specified filenames must occur in
424
:return: a set of (path, file ids) for the specified filenames and their
425
children. The returned path is the path of the id in the first tree
426
that contains it. This matters when files have been moved
430
# This function needs to know the ids for filenames in all trees, then
431
# search for those same files and children in all the other trees.
432
# it is complicated by the same path in two trees being able to have
433
# different ids, which might both be present in both trees.
434
# consider two trees, which have had 'mv foo bar' and 'mv baz foo' done
435
# in this case, a diff of 'foo' should should changes to both the current
436
# 'bar' and the current 'foo' which was baz. Its arguable that if
437
# the situation is 'mv parent/foo bar' and 'mv baz parent/foo', that
438
# we should return the current bar and the current parent/foo' - at the
439
# moment we do, but we loop around all ids and all trees: I*T checks.
441
# Updating this algorithm to be fast in the common case:
442
# nothing has moved, all files have the same id in parent, child and there
443
# are only two trees (or one is working tree and the others are parents).
444
# walk the dirstate. as we find each path, gather the paths of that
445
# id in all trees. add a mapping from the id to the path in those trees.
446
# now lookup children by id, again in all trees; for these trees that
447
# nothing has moved in, the id->path mapping will allow us to find the
448
# parent trivially. To answer 'has anything been moved' in one of the
449
# dirstate parent trees though, we will need to stare harder at it.
451
# Now, given a path index, that is trivial for any one tree, and given
452
# that we can ask for additional data from a dirstate tree, its a single
453
# pass, though it will require scanning the entire tree to find paths
454
# that were at the current location.
455
# ideal results?: There are three things: tree, path, id. Pathologically
456
# we can have completely disjoint ids for each tree; but we cannot have
457
# disjoin paths for each tree, except if we scan each tree for the
458
# different ids from other trees.
460
specified_path_ids = _find_ids_across_trees(filenames, trees,
462
return _find_path_id_children_across_trees(specified_path_ids, trees)
465
887
def _find_ids_across_trees(filenames, trees, require_versioned):
466
888
"""Find the ids corresponding to specified filenames.
468
890
All matches in all trees will be used, but subdirectories are not scanned.
470
892
:param filenames: The filenames to find file_ids for
471
893
:param trees: The trees to find file_ids within
472
894
:param require_versioned: if true, all specified filenames must occur in
473
895
at least one tree.
474
:return: a set of (path, file ids) for the specified filenames
896
:return: a set of file ids for the specified filenames
476
898
not_versioned = []
477
899
interesting_ids = set()
522
943
Its instances have methods like 'compare' and contain references to the
523
944
source and target trees these operations are to be carried out on.
525
clients of bzrlib should not need to use InterTree directly, rather they
946
Clients of bzrlib should not need to use InterTree directly, rather they
526
947
should use the convenience methods on Tree such as 'Tree.compare()' which
527
948
will pass through to InterTree as appropriate.
951
# Formats that will be used to test this InterTree. If both are
952
# None, this InterTree will not be tested (e.g. because a complex
954
_matching_from_tree_format = None
955
_matching_to_tree_format = None
960
def is_compatible(kls, source, target):
961
# The default implementation is naive and uses the public API, so
962
# it works for all trees.
965
def _changes_from_entries(self, source_entry, target_entry,
966
source_path=None, target_path=None):
967
"""Generate a iter_changes tuple between source_entry and target_entry.
969
:param source_entry: An inventory entry from self.source, or None.
970
:param target_entry: An inventory entry from self.target, or None.
971
:param source_path: The path of source_entry, if known. If not known
972
it will be looked up.
973
:param target_path: The path of target_entry, if known. If not known
974
it will be looked up.
975
:return: A tuple, item 0 of which is an iter_changes result tuple, and
976
item 1 is True if there are any changes in the result tuple.
978
if source_entry is None:
979
if target_entry is None:
981
file_id = target_entry.file_id
983
file_id = source_entry.file_id
984
if source_entry is not None:
985
source_versioned = True
986
source_name = source_entry.name
987
source_parent = source_entry.parent_id
988
if source_path is None:
989
source_path = self.source.id2path(file_id)
990
source_kind, source_executable, source_stat = \
991
self.source._comparison_data(source_entry, source_path)
993
source_versioned = False
997
source_executable = None
998
if target_entry is not None:
999
target_versioned = True
1000
target_name = target_entry.name
1001
target_parent = target_entry.parent_id
1002
if target_path is None:
1003
target_path = self.target.id2path(file_id)
1004
target_kind, target_executable, target_stat = \
1005
self.target._comparison_data(target_entry, target_path)
1007
target_versioned = False
1009
target_parent = None
1011
target_executable = None
1012
versioned = (source_versioned, target_versioned)
1013
kind = (source_kind, target_kind)
1014
changed_content = False
1015
if source_kind != target_kind:
1016
changed_content = True
1017
elif source_kind == 'file':
1018
if not self.file_content_matches(file_id, file_id, source_path,
1019
target_path, source_stat, target_stat):
1020
changed_content = True
1021
elif source_kind == 'symlink':
1022
if (self.source.get_symlink_target(file_id) !=
1023
self.target.get_symlink_target(file_id)):
1024
changed_content = True
1025
# XXX: Yes, the indentation below is wrong. But fixing it broke
1026
# test_merge.TestMergerEntriesLCAOnDisk.
1027
# test_nested_tree_subtree_renamed_and_modified. We'll wait for
1028
# the fix from bzr.dev -- vila 2009026
1029
elif source_kind == 'tree-reference':
1030
if (self.source.get_reference_revision(file_id, source_path)
1031
!= self.target.get_reference_revision(file_id, target_path)):
1032
changed_content = True
1033
parent = (source_parent, target_parent)
1034
name = (source_name, target_name)
1035
executable = (source_executable, target_executable)
1036
if (changed_content is not False or versioned[0] != versioned[1]
1037
or parent[0] != parent[1] or name[0] != name[1] or
1038
executable[0] != executable[1]):
1042
return (file_id, (source_path, target_path), changed_content,
1043
versioned, parent, name, kind, executable), changes
532
1045
@needs_read_lock
533
1046
def compare(self, want_unchanged=False, specific_files=None,
534
extra_trees=None, require_versioned=False, include_root=False):
1047
extra_trees=None, require_versioned=False, include_root=False,
1048
want_unversioned=False):
535
1049
"""Return the changes from source to target.
537
1051
:return: A TreeDelta.
546
1060
:param require_versioned: An optional boolean (defaults to False). When
547
1061
supplied and True all the 'specific_files' must be versioned, or
548
1062
a PathsNotVersionedError will be thrown.
1063
:param want_unversioned: Scan for unversioned paths.
550
# NB: show_status depends on being able to pass in non-versioned files
551
# and report them as unknown
552
trees = (self.source, self.target)
1065
trees = (self.source,)
553
1066
if extra_trees is not None:
554
1067
trees = trees + tuple(extra_trees)
555
specific_file_ids = find_ids_across_trees(specific_files,
556
trees, require_versioned=require_versioned)
1068
# target is usually the newer tree:
1069
specific_file_ids = self.target.paths2ids(specific_files, trees,
1070
require_versioned=require_versioned)
557
1071
if specific_files and not specific_file_ids:
558
1072
# All files are unversioned, so just return an empty delta
559
1073
# _compare_trees would think we want a complete delta
560
return delta.TreeDelta()
1074
result = delta.TreeDelta()
1075
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1076
result.unversioned = [(path, None,
1077
self.target._comparison_data(fake_entry, path)[0]) for path in
561
1080
return delta._compare_trees(self.source, self.target, want_unchanged,
562
specific_file_ids, include_root)
1081
specific_files, include_root, extra_trees=extra_trees,
1082
require_versioned=require_versioned,
1083
want_unversioned=want_unversioned)
564
def _iter_changes(self, from_tree, to_tree, include_unchanged,
565
specific_file_ids, pb):
1085
def iter_changes(self, include_unchanged=False,
1086
specific_files=None, pb=None, extra_trees=[],
1087
require_versioned=True, want_unversioned=False):
566
1088
"""Generate an iterator of changes between trees.
568
1090
A tuple is returned:
569
(file_id, path, changed_content, versioned, parent, name, kind,
1091
(file_id, (path_in_source, path_in_target),
1092
changed_content, versioned, parent, name, kind,
572
Path is relative to the to_tree. changed_content is True if the file's
573
content has changed. This includes changes to its kind, and to
1095
Changed_content is True if the file's content has changed. This
1096
includes changes to its kind, and to a symlink's target.
576
1098
versioned, parent, name, kind, executable are tuples of (from, to).
577
1099
If a file is missing in a tree, its kind is None.
579
Iteration is done in parent-to-child order, relative to the to_tree.
1101
Iteration is done in parent-to-child order, relative to the target
1104
There is no guarantee that all paths are in sorted order: the
1105
requirement to expand the search due to renames may result in children
1106
that should be found early being found late in the search, after
1107
lexically later results have been returned.
1108
:param require_versioned: Raise errors.PathsNotVersionedError if a
1109
path in the specific_files list is not versioned in one of
1110
source, target or extra_trees.
1111
:param specific_files: An optional list of file paths to restrict the
1112
comparison to. When mapping filenames to ids, all matches in all
1113
trees (including optional extra_trees) are used, and all children
1114
of matched directories are included. The parents in the target tree
1115
of the specific files up to and including the root of the tree are
1116
always evaluated for changes too.
1117
:param want_unversioned: Should unversioned files be returned in the
1118
output. An unversioned file is defined as one with (False, False)
1119
for the versioned pair.
1121
lookup_trees = [self.source]
1123
lookup_trees.extend(extra_trees)
1124
# The ids of items we need to examine to insure delta consistency.
1125
precise_file_ids = set()
1126
changed_file_ids = []
1127
if specific_files == []:
1128
specific_file_ids = []
1130
specific_file_ids = self.target.paths2ids(specific_files,
1131
lookup_trees, require_versioned=require_versioned)
1132
if specific_files is not None:
1133
# reparented or added entries must have their parents included
1134
# so that valid deltas can be created. The seen_parents set
1135
# tracks the parents that we need to have.
1136
# The seen_dirs set tracks directory entries we've yielded.
1137
# After outputting version object in to_entries we set difference
1138
# the two seen sets and start checking parents.
1139
seen_parents = set()
1141
if want_unversioned:
1142
all_unversioned = sorted([(p.split('/'), p) for p in
1143
self.target.extras()
1144
if specific_files is None or
1145
osutils.is_inside_any(specific_files, p)])
1146
all_unversioned = collections.deque(all_unversioned)
1148
all_unversioned = collections.deque()
582
from_entries_by_dir = list(from_tree.inventory.iter_entries_by_dir(
1150
from_entries_by_dir = list(self.source.iter_entries_by_dir(
583
1151
specific_file_ids=specific_file_ids))
584
1152
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
585
to_entries_by_dir = list(to_tree.inventory.iter_entries_by_dir(
1153
to_entries_by_dir = list(self.target.iter_entries_by_dir(
586
1154
specific_file_ids=specific_file_ids))
587
1155
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
589
for to_path, to_entry in to_entries_by_dir:
590
file_id = to_entry.file_id
591
to_paths[file_id] = to_path
1157
# the unversioned path lookup only occurs on real trees - where there
1158
# can be extras. So the fake_entry is solely used to look up
1159
# executable it values when execute is not supported.
1160
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
1161
for target_path, target_entry in to_entries_by_dir:
1162
while (all_unversioned and
1163
all_unversioned[0][0] < target_path.split('/')):
1164
unversioned_path = all_unversioned.popleft()
1165
target_kind, target_executable, target_stat = \
1166
self.target._comparison_data(fake_entry, unversioned_path[1])
1167
yield (None, (None, unversioned_path[1]), True, (False, False),
1169
(None, unversioned_path[0][-1]),
1170
(None, target_kind),
1171
(None, target_executable))
1172
source_path, source_entry = from_data.get(target_entry.file_id,
1174
result, changes = self._changes_from_entries(source_entry,
1175
target_entry, source_path=source_path, target_path=target_path)
1176
to_paths[result[0]] = result[1][1]
592
1177
entry_count += 1
593
changed_content = False
594
from_path, from_entry = from_data.get(file_id, (None, None))
595
from_versioned = (from_entry is not None)
596
if from_entry is not None:
597
from_versioned = True
598
from_name = from_entry.name
599
from_parent = from_entry.parent_id
600
from_kind, from_executable, from_stat = \
601
from_tree._comparison_data(from_entry, from_path)
602
1179
entry_count += 1
604
from_versioned = False
608
from_executable = None
609
versioned = (from_versioned, True)
1181
pb.update('comparing files', entry_count, num_entries)
1182
if changes or include_unchanged:
1183
if specific_file_ids is not None:
1184
new_parent_id = result[4][1]
1185
precise_file_ids.add(new_parent_id)
1186
changed_file_ids.append(result[0])
1188
# Ensure correct behaviour for reparented/added specific files.
1189
if specific_files is not None:
1190
# Record output dirs
1191
if result[6][1] == 'directory':
1192
seen_dirs.add(result[0])
1193
# Record parents of reparented/added entries.
1194
versioned = result[3]
1196
if not versioned[0] or parents[0] != parents[1]:
1197
seen_parents.add(parents[1])
1198
while all_unversioned:
1199
# yield any trailing unversioned paths
1200
unversioned_path = all_unversioned.popleft()
610
1201
to_kind, to_executable, to_stat = \
611
to_tree._comparison_data(to_entry, to_path)
612
kind = (from_kind, to_kind)
613
if kind[0] != kind[1]:
614
changed_content = True
615
elif from_kind == 'file':
616
from_size = from_tree._file_size(from_entry, from_stat)
617
to_size = to_tree._file_size(to_entry, to_stat)
618
if from_size != to_size:
619
changed_content = True
620
elif (from_tree.get_file_sha1(file_id, from_path, from_stat) !=
621
to_tree.get_file_sha1(file_id, to_path, to_stat)):
622
changed_content = True
623
elif from_kind == 'symlink':
624
if (from_tree.get_symlink_target(file_id) !=
625
to_tree.get_symlink_target(file_id)):
626
changed_content = True
627
parent = (from_parent, to_entry.parent_id)
628
name = (from_name, to_entry.name)
629
executable = (from_executable, to_executable)
631
pb.update('comparing files', entry_count, num_entries)
632
if (changed_content is not False or versioned[0] != versioned[1]
633
or parent[0] != parent[1] or name[0] != name[1] or
634
executable[0] != executable[1] or include_unchanged):
635
yield (file_id, to_path, changed_content, versioned, parent,
636
name, kind, executable)
638
def get_to_path(from_entry):
639
if from_entry.parent_id is None:
642
if from_entry.parent_id not in to_paths:
643
get_to_path(from_tree.inventory[from_entry.parent_id])
644
to_path = osutils.pathjoin(to_paths[from_entry.parent_id],
646
to_paths[from_entry.file_id] = to_path
1202
self.target._comparison_data(fake_entry, unversioned_path[1])
1203
yield (None, (None, unversioned_path[1]), True, (False, False),
1205
(None, unversioned_path[0][-1]),
1207
(None, to_executable))
1208
# Yield all remaining source paths
649
1209
for path, from_entry in from_entries_by_dir:
650
1210
file_id = from_entry.file_id
651
1211
if file_id in to_paths:
653
to_path = get_to_path(from_entry)
1214
if not self.target.has_id(file_id):
1215
# common case - paths we have not emitted are not present in
1219
to_path = self.target.id2path(file_id)
654
1220
entry_count += 1
655
1221
if pb is not None:
656
1222
pb.update('comparing files', entry_count, num_entries)
658
1224
parent = (from_entry.parent_id, None)
659
1225
name = (from_entry.name, None)
660
1226
from_kind, from_executable, stat_value = \
661
from_tree._comparison_data(from_entry, path)
1227
self.source._comparison_data(from_entry, path)
662
1228
kind = (from_kind, None)
663
1229
executable = (from_executable, None)
664
changed_content = True
1230
changed_content = from_kind is not None
665
1231
# the parent's path is necessarily known at this point.
666
yield(file_id, to_path, changed_content, versioned, parent,
1232
changed_file_ids.append(file_id)
1233
yield(file_id, (path, to_path), changed_content, versioned, parent,
667
1234
name, kind, executable)
670
# This was deprecated before 0.12, but did not have an official warning
671
@symbol_versioning.deprecated_function(symbol_versioning.zero_twelve)
672
def RevisionTree(*args, **kwargs):
673
"""RevisionTree has moved to bzrlib.revisiontree.RevisionTree()
675
Accessing it as bzrlib.tree.RevisionTree has been deprecated as of
678
from bzrlib.revisiontree import RevisionTree as _RevisionTree
679
return _RevisionTree(*args, **kwargs)
1235
changed_file_ids = set(changed_file_ids)
1236
if specific_file_ids is not None:
1237
for result in self._handle_precise_ids(precise_file_ids,
1241
def _get_entry(self, tree, file_id):
1242
"""Get an inventory entry from a tree, with missing entries as None.
1244
If the tree raises NotImplementedError on accessing .inventory, then
1245
this is worked around using iter_entries_by_dir on just the file id
1248
:param tree: The tree to lookup the entry in.
1249
:param file_id: The file_id to lookup.
1252
inventory = tree.inventory
1253
except NotImplementedError:
1254
# No inventory available.
1256
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1257
return iterator.next()[1]
1258
except StopIteration:
1262
return inventory[file_id]
1263
except errors.NoSuchId:
1266
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1267
discarded_changes=None):
1268
"""Fill out a partial iter_changes to be consistent.
1270
:param precise_file_ids: The file ids of parents that were seen during
1272
:param changed_file_ids: The file ids of already emitted items.
1273
:param discarded_changes: An optional dict of precalculated
1274
iter_changes items which the partial iter_changes had not output
1276
:return: A generator of iter_changes items to output.
1278
# process parents of things that had changed under the users
1279
# requested paths to prevent incorrect paths or parent ids which
1280
# aren't in the tree.
1281
while precise_file_ids:
1282
precise_file_ids.discard(None)
1283
# Don't emit file_ids twice
1284
precise_file_ids.difference_update(changed_file_ids)
1285
if not precise_file_ids:
1287
# If the there was something at a given output path in source, we
1288
# have to include the entry from source in the delta, or we would
1289
# be putting this entry into a used path.
1291
for parent_id in precise_file_ids:
1293
paths.append(self.target.id2path(parent_id))
1294
except errors.NoSuchId:
1295
# This id has been dragged in from the source by delta
1296
# expansion and isn't present in target at all: we don't
1297
# need to check for path collisions on it.
1300
old_id = self.source.path2id(path)
1301
precise_file_ids.add(old_id)
1302
precise_file_ids.discard(None)
1303
current_ids = precise_file_ids
1304
precise_file_ids = set()
1305
# We have to emit all of precise_file_ids that have been altered.
1306
# We may have to output the children of some of those ids if any
1307
# directories have stopped being directories.
1308
for file_id in current_ids:
1310
if discarded_changes:
1311
result = discarded_changes.get(file_id)
1316
old_entry = self._get_entry(self.source, file_id)
1317
new_entry = self._get_entry(self.target, file_id)
1318
result, changes = self._changes_from_entries(
1319
old_entry, new_entry)
1322
# Get this parents parent to examine.
1323
new_parent_id = result[4][1]
1324
precise_file_ids.add(new_parent_id)
1326
if (result[6][0] == 'directory' and
1327
result[6][1] != 'directory'):
1328
# This stopped being a directory, the old children have
1330
if old_entry is None:
1331
# Reusing a discarded change.
1332
old_entry = self._get_entry(self.source, file_id)
1333
for child in old_entry.children.values():
1334
precise_file_ids.add(child.file_id)
1335
changed_file_ids.add(result[0])
1339
def file_content_matches(self, source_file_id, target_file_id,
1340
source_path=None, target_path=None, source_stat=None, target_stat=None):
1341
"""Check if two files are the same in the source and target trees.
1343
This only checks that the contents of the files are the same,
1344
it does not touch anything else.
1346
:param source_file_id: File id of the file in the source tree
1347
:param target_file_id: File id of the file in the target tree
1348
:param source_path: Path of the file in the source tree
1349
:param target_path: Path of the file in the target tree
1350
:param source_stat: Optional stat value of the file in the source tree
1351
:param target_stat: Optional stat value of the file in the target tree
1352
:return: Boolean indicating whether the files have the same contents
1354
source_verifier_kind, source_verifier_data = self.source.get_file_verifier(
1355
source_file_id, source_path, source_stat)
1356
target_verifier_kind, target_verifier_data = self.target.get_file_verifier(
1357
target_file_id, target_path, target_stat)
1358
if source_verifier_kind == target_verifier_kind:
1359
return (source_verifier_data == target_verifier_data)
1360
# Fall back to SHA1 for now
1361
if source_verifier_kind != "SHA1":
1362
source_sha1 = self.source.get_file_sha1(source_file_id,
1363
source_path, source_stat)
1365
source_sha1 = source_verifier_data
1366
if target_verifier_kind != "SHA1":
1367
target_sha1 = self.target.get_file_sha1(target_file_id,
1368
target_path, target_stat)
1370
target_sha1 = target_verifier_data
1371
return (source_sha1 == target_sha1)
1373
InterTree.register_optimiser(InterTree)
1376
class MultiWalker(object):
1377
"""Walk multiple trees simultaneously, getting combined results."""
1379
# Note: This could be written to not assume you can do out-of-order
1380
# lookups. Instead any nodes that don't match in all trees could be
1381
# marked as 'deferred', and then returned in the final cleanup loop.
1382
# For now, I think it is "nicer" to return things as close to the
1383
# "master_tree" order as we can.
1385
def __init__(self, master_tree, other_trees):
1386
"""Create a new MultiWalker.
1388
All trees being walked must implement "iter_entries_by_dir()", such
1389
that they yield (path, object) tuples, where that object will have a
1390
'.file_id' member, that can be used to check equality.
1392
:param master_tree: All trees will be 'slaved' to the master_tree such
1393
that nodes in master_tree will be used as 'first-pass' sync points.
1394
Any nodes that aren't in master_tree will be merged in a second
1396
:param other_trees: A list of other trees to walk simultaneously.
1398
self._master_tree = master_tree
1399
self._other_trees = other_trees
1401
# Keep track of any nodes that were properly processed just out of
1402
# order, that way we don't return them at the end, we don't have to
1403
# track *all* processed file_ids, just the out-of-order ones
1404
self._out_of_order_processed = set()
1407
def _step_one(iterator):
1408
"""Step an iter_entries_by_dir iterator.
1410
:return: (has_more, path, ie)
1411
If has_more is False, path and ie will be None.
1414
path, ie = iterator.next()
1415
except StopIteration:
1416
return False, None, None
1418
return True, path, ie
1421
def _cmp_path_by_dirblock(path1, path2):
1422
"""Compare two paths based on what directory they are in.
1424
This generates a sort order, such that all children of a directory are
1425
sorted together, and grandchildren are in the same order as the
1426
children appear. But all grandchildren come after all children.
1428
:param path1: first path
1429
:param path2: the second path
1430
:return: negative number if ``path1`` comes first,
1431
0 if paths are equal
1432
and a positive number if ``path2`` sorts first
1434
# Shortcut this special case
1437
# This is stolen from _dirstate_helpers_py.py, only switching it to
1438
# Unicode objects. Consider using encode_utf8() and then using the
1439
# optimized versions, or maybe writing optimized unicode versions.
1440
if not isinstance(path1, unicode):
1441
raise TypeError("'path1' must be a unicode string, not %s: %r"
1442
% (type(path1), path1))
1443
if not isinstance(path2, unicode):
1444
raise TypeError("'path2' must be a unicode string, not %s: %r"
1445
% (type(path2), path2))
1446
return cmp(MultiWalker._path_to_key(path1),
1447
MultiWalker._path_to_key(path2))
1450
def _path_to_key(path):
1451
dirname, basename = osutils.split(path)
1452
return (dirname.split(u'/'), basename)
1454
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1455
"""Lookup an inventory entry by file_id.
1457
This is called when an entry is missing in the normal order.
1458
Generally this is because a file was either renamed, or it was
1459
deleted/added. If the entry was found in the inventory and not in
1460
extra_entries, it will be added to self._out_of_order_processed
1462
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1463
should be filled with entries that were found before they were
1464
used. If file_id is present, it will be removed from the
1466
:param other_tree: The Tree to search, in case we didn't find the entry
1468
:param file_id: The file_id to look for
1469
:return: (path, ie) if found or (None, None) if not present.
1471
if file_id in extra_entries:
1472
return extra_entries.pop(file_id)
1473
# TODO: Is id2path better as the first call, or is
1474
# inventory[file_id] better as a first check?
1476
cur_path = other_tree.id2path(file_id)
1477
except errors.NoSuchId:
1479
if cur_path is None:
1482
self._out_of_order_processed.add(file_id)
1483
cur_ie = other_tree.inventory[file_id]
1484
return (cur_path, cur_ie)
1487
"""Match up the values in the different trees."""
1488
for result in self._walk_master_tree():
1490
self._finish_others()
1491
for result in self._walk_others():
1494
def _walk_master_tree(self):
1495
"""First pass, walk all trees in lock-step.
1497
When we are done, all nodes in the master_tree will have been
1498
processed. _other_walkers, _other_entries, and _others_extra will be
1499
set on 'self' for future processing.
1501
# This iterator has the most "inlining" done, because it tends to touch
1502
# every file in the tree, while the others only hit nodes that don't
1504
master_iterator = self._master_tree.iter_entries_by_dir()
1506
other_walkers = [other.iter_entries_by_dir()
1507
for other in self._other_trees]
1508
other_entries = [self._step_one(walker) for walker in other_walkers]
1509
# Track extra nodes in the other trees
1510
others_extra = [{} for i in xrange(len(self._other_trees))]
1512
master_has_more = True
1513
step_one = self._step_one
1514
lookup_by_file_id = self._lookup_by_file_id
1515
out_of_order_processed = self._out_of_order_processed
1517
while master_has_more:
1518
(master_has_more, path, master_ie) = step_one(master_iterator)
1519
if not master_has_more:
1522
file_id = master_ie.file_id
1524
other_values_append = other_values.append
1525
next_other_entries = []
1526
next_other_entries_append = next_other_entries.append
1527
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1528
if not other_has_more:
1529
other_values_append(lookup_by_file_id(
1530
others_extra[idx], self._other_trees[idx], file_id))
1531
next_other_entries_append((False, None, None))
1532
elif file_id == other_ie.file_id:
1533
# This is the critical code path, as most of the entries
1534
# should match between most trees.
1535
other_values_append((other_path, other_ie))
1536
next_other_entries_append(step_one(other_walkers[idx]))
1538
# This walker did not match, step it until it either
1539
# matches, or we know we are past the current walker.
1540
other_walker = other_walkers[idx]
1541
other_extra = others_extra[idx]
1542
while (other_has_more and
1543
self._cmp_path_by_dirblock(other_path, path) < 0):
1544
other_file_id = other_ie.file_id
1545
if other_file_id not in out_of_order_processed:
1546
other_extra[other_file_id] = (other_path, other_ie)
1547
other_has_more, other_path, other_ie = \
1548
step_one(other_walker)
1549
if other_has_more and other_ie.file_id == file_id:
1550
# We ended up walking to this point, match and step
1552
other_values_append((other_path, other_ie))
1553
other_has_more, other_path, other_ie = \
1554
step_one(other_walker)
1556
# This record isn't in the normal order, see if it
1558
other_values_append(lookup_by_file_id(
1559
other_extra, self._other_trees[idx], file_id))
1560
next_other_entries_append((other_has_more, other_path,
1562
other_entries = next_other_entries
1564
# We've matched all the walkers, yield this datapoint
1565
yield path, file_id, master_ie, other_values
1566
self._other_walkers = other_walkers
1567
self._other_entries = other_entries
1568
self._others_extra = others_extra
1570
def _finish_others(self):
1571
"""Finish walking the other iterators, so we get all entries."""
1572
for idx, info in enumerate(self._other_entries):
1573
other_extra = self._others_extra[idx]
1574
(other_has_more, other_path, other_ie) = info
1575
while other_has_more:
1576
other_file_id = other_ie.file_id
1577
if other_file_id not in self._out_of_order_processed:
1578
other_extra[other_file_id] = (other_path, other_ie)
1579
other_has_more, other_path, other_ie = \
1580
self._step_one(self._other_walkers[idx])
1581
del self._other_entries
1583
def _walk_others(self):
1584
"""Finish up by walking all the 'deferred' nodes."""
1585
# TODO: One alternative would be to grab all possible unprocessed
1586
# file_ids, and then sort by path, and then yield them. That
1587
# might ensure better ordering, in case a caller strictly
1588
# requires parents before children.
1589
for idx, other_extra in enumerate(self._others_extra):
1590
others = sorted(other_extra.itervalues(),
1591
key=lambda x: self._path_to_key(x[0]))
1592
for other_path, other_ie in others:
1593
file_id = other_ie.file_id
1594
# We don't need to check out_of_order_processed here, because
1595
# the lookup_by_file_id will be removing anything processed
1596
# from the extras cache
1597
other_extra.pop(file_id)
1598
other_values = [(None, None) for i in xrange(idx)]
1599
other_values.append((other_path, other_ie))
1600
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1601
alt_idx = alt_idx + idx + 1
1602
alt_extra = self._others_extra[alt_idx]
1603
alt_tree = self._other_trees[alt_idx]
1604
other_values.append(self._lookup_by_file_id(
1605
alt_extra, alt_tree, file_id))
1606
yield other_path, file_id, None, other_values
added
removed
bzrlib/tree.py
