1
# Copyright (C) 2005 Canonical Ltd
1
# Copyright (C) 2005-2011 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
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
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
17
"""Tree classes, representing directory at point in time.
20
from __future__ import absolute_import
21
from cStringIO import StringIO
24
from bzrlib.trace import mutter, note
25
from bzrlib.errors import BzrError, BzrCheckError
26
from bzrlib.inventory import Inventory
27
from bzrlib.osutils import appendpath, fingerprint_file
24
from .lazy_import import lazy_import
25
lazy_import(globals(), """
29
conflicts as _mod_conflicts,
33
revision as _mod_revision,
37
from breezy.bzr import (
40
from breezy.i18n import gettext
48
from .inter import InterObject
54
class FileTimestampUnavailable(errors.BzrError):
56
_fmt = "The filestamp for %(path)s is not available."
60
def __init__(self, path):
29
64
class Tree(object):
30
65
"""Abstract file tree.
32
67
There are several subclasses:
34
69
* `WorkingTree` exists as files on disk editable by the user.
36
71
* `RevisionTree` is a tree as recorded at some point in the past.
40
Trees contain an `Inventory` object, and also know how to retrieve
41
file texts mentioned in the inventory, either from a working
42
directory or from a store.
44
It is possible for trees to contain files that are not described
45
in their inventory or vice versa; for this use `filenames()`.
47
73
Trees can be compared, etc, regardless of whether they are working
48
74
trees or versioned trees.
77
def has_versioned_directories(self):
78
"""Whether this tree can contain explicitly versioned directories.
80
This defaults to True, but some implementations may want to override
85
def changes_from(self, other, want_unchanged=False, specific_files=None,
86
extra_trees=None, require_versioned=False, include_root=False,
87
want_unversioned=False):
88
"""Return a TreeDelta of the changes from other to this tree.
90
:param other: A tree to compare with.
91
:param specific_files: An optional list of file paths to restrict the
92
comparison to. When mapping filenames to ids, all matches in all
93
trees (including optional extra_trees) are used, and all children of
94
matched directories are included.
95
:param want_unchanged: An optional boolean requesting the inclusion of
96
unchanged entries in the result.
97
:param extra_trees: An optional list of additional trees to use when
98
mapping the contents of specific_files (paths) to file_ids.
99
:param require_versioned: An optional boolean (defaults to False). When
100
supplied and True all the 'specific_files' must be versioned, or
101
a PathsNotVersionedError will be thrown.
102
:param want_unversioned: Scan for unversioned paths.
104
The comparison will be performed by an InterTree object looked up on
107
# Martin observes that Tree.changes_from returns a TreeDelta and this
108
# may confuse people, because the class name of the returned object is
109
# a synonym of the object referenced in the method name.
110
return InterTree.get(other, self).compare(
111
want_unchanged=want_unchanged,
112
specific_files=specific_files,
113
extra_trees=extra_trees,
114
require_versioned=require_versioned,
115
include_root=include_root,
116
want_unversioned=want_unversioned,
119
def iter_changes(self, from_tree, include_unchanged=False,
120
specific_files=None, pb=None, extra_trees=None,
121
require_versioned=True, want_unversioned=False):
122
"""See InterTree.iter_changes"""
123
intertree = InterTree.get(from_tree, self)
124
return intertree.iter_changes(include_unchanged, specific_files, pb,
125
extra_trees, require_versioned, want_unversioned=want_unversioned)
128
"""Get a list of the conflicts in the tree.
130
Each conflict is an instance of breezy.conflicts.Conflict.
132
return _mod_conflicts.ConflictList()
135
"""For trees that can have unversioned files, return all such paths."""
138
def get_parent_ids(self):
139
"""Get the parent ids for this tree.
141
:return: a list of parent ids. [] is returned to indicate
142
a tree with no parents.
143
:raises: BzrError if the parents are not known.
145
raise NotImplementedError(self.get_parent_ids)
51
147
def has_filename(self, filename):
52
148
"""True if the tree has given filename."""
53
raise NotImplementedError()
149
raise NotImplementedError(self.has_filename)
55
151
def has_id(self, file_id):
56
return self.inventory.has_id(file_id)
152
raise NotImplementedError(self.has_id)
58
154
def has_or_had_id(self, file_id):
59
if file_id == self.inventory.root.file_id:
61
return self.inventory.has_id(file_id)
66
return iter(self.inventory)
155
raise NotImplementedError(self.has_or_had_id)
157
def is_ignored(self, filename):
158
"""Check whether the filename is ignored by this tree.
160
:param filename: The relative filename within the tree.
161
:return: True if the filename is ignored.
165
def all_file_ids(self):
166
"""Iterate through all file ids, including ids for missing files."""
167
raise NotImplementedError(self.all_file_ids)
68
169
def id2path(self, file_id):
69
return self.inventory.id2path(file_id)
71
def kind(self, file_id):
72
raise NotImplementedError("subclasses must implement kind")
74
def _get_inventory(self):
75
return self._inventory
77
def get_file_by_path(self, path):
78
return self.get_file(self._inventory.path2id(path))
80
inventory = property(_get_inventory,
81
doc="Inventory of this Tree")
170
"""Return the path for a file id.
174
raise NotImplementedError(self.id2path)
176
def iter_entries_by_dir(self, specific_file_ids=None, yield_parents=False):
177
"""Walk the tree in 'by_dir' order.
179
This will yield each entry in the tree as a (path, entry) tuple.
180
The order that they are yielded is:
182
Directories are walked in a depth-first lexicographical order,
183
however, whenever a directory is reached, all of its direct child
184
nodes are yielded in lexicographical order before yielding the
187
For example, in the tree::
197
The yield order (ignoring root) would be::
199
a, f, a/b, a/d, a/b/c, a/d/e, f/g
201
:param yield_parents: If True, yield the parents from the root leading
202
down to specific_file_ids that have been requested. This has no
203
impact if specific_file_ids is None.
205
raise NotImplementedError(self.iter_entries_by_dir)
207
def iter_child_entries(self, file_id, path=None):
208
"""Iterate over the children of a directory or tree reference.
210
:param file_id: File id of the directory/tree-reference
211
:param path: Optional path of the directory
212
:raise NoSuchId: When the file_id does not exist
213
:return: Iterator over entries in the directory
215
raise NotImplementedError(self.iter_child_entries)
217
def list_files(self, include_root=False, from_dir=None, recursive=True):
218
"""List all files in this tree.
220
:param include_root: Whether to include the entry for the tree root
221
:param from_dir: Directory under which to list files
222
:param recursive: Whether to list files recursively
223
:return: iterator over tuples of (path, versioned, kind, file_id,
226
raise NotImplementedError(self.list_files)
228
def iter_references(self):
229
if self.supports_tree_reference():
230
for path, entry in self.iter_entries_by_dir():
231
if entry.kind == 'tree-reference':
232
yield path, entry.file_id
234
def kind(self, path, file_id=None):
235
raise NotImplementedError("Tree subclass %s must implement kind"
236
% self.__class__.__name__)
238
def stored_kind(self, path, file_id=None):
239
"""File kind stored for this file_id.
241
May not match kind on disk for working trees. Always available
242
for versioned files, even when the file itself is missing.
244
return self.kind(path, file_id)
246
def path_content_summary(self, path):
247
"""Get a summary of the information about path.
249
All the attributes returned are for the canonical form, not the
250
convenient form (if content filters are in use.)
252
:param path: A relative path within the tree.
253
:return: A tuple containing kind, size, exec, sha1-or-link.
254
Kind is always present (see tree.kind()).
255
size is present if kind is file and the size of the
256
canonical form can be cheaply determined, None otherwise.
257
exec is None unless kind is file and the platform supports the 'x'
259
sha1-or-link is the link target if kind is symlink, or the sha1 if
260
it can be obtained without reading the file.
262
raise NotImplementedError(self.path_content_summary)
264
def get_reference_revision(self, path, file_id=None):
265
raise NotImplementedError("Tree subclass %s must implement "
266
"get_reference_revision"
267
% self.__class__.__name__)
269
def _comparison_data(self, entry, path):
270
"""Return a tuple of kind, executable, stat_value for a file.
272
entry may be None if there is no inventory entry for the file, but
273
path must always be supplied.
275
kind is None if there is no file present (even if an inventory id is
276
present). executable is False for non-file entries.
278
raise NotImplementedError(self._comparison_data)
280
def _file_size(self, entry, stat_value):
281
raise NotImplementedError(self._file_size)
283
def get_file(self, path, file_id=None):
284
"""Return a file object for the file file_id in the tree.
286
If both file_id and path are defined, it is implementation defined as
287
to which one is used.
289
raise NotImplementedError(self.get_file)
291
def get_file_with_stat(self, path, file_id=None):
292
"""Get a file handle and stat object for file_id.
294
The default implementation returns (self.get_file, None) for backwards
297
:param path: The path of the file.
298
:param file_id: The file id to read, if it is known.
299
:return: A tuple (file_handle, stat_value_or_None). If the tree has
300
no stat facility, or need for a stat cache feedback during commit,
301
it may return None for the second element of the tuple.
303
return (self.get_file(path, file_id), None)
305
def get_file_text(self, path, file_id=None):
306
"""Return the byte content of a file.
308
:param path: The path of the file.
309
:param file_id: The file_id of the file.
311
If both file_id and path are supplied, an implementation may use
314
:returns: A single byte string for the whole file.
316
my_file = self.get_file(path, file_id)
318
return my_file.read()
322
def get_file_lines(self, path, file_id=None):
323
"""Return the content of a file, as lines.
325
:param path: The path of the file.
326
:param file_id: The file_id of the file.
328
If both file_id and path are supplied, an implementation may use
331
return osutils.split_lines(self.get_file_text(path, file_id))
333
def get_file_verifier(self, path, file_id=None, stat_value=None):
334
"""Return a verifier for a file.
336
The default implementation returns a sha1.
338
:param file_id: The handle for this file.
339
:param path: The path that this file can be found at.
340
These must point to the same object.
341
:param stat_value: Optional stat value for the object
342
:return: Tuple with verifier name and verifier data
344
return ("SHA1", self.get_file_sha1(path, file_id,
345
stat_value=stat_value))
347
def get_file_sha1(self, path, file_id=None, stat_value=None):
348
"""Return the SHA1 file for a file.
350
:note: callers should use get_file_verifier instead
351
where possible, as the underlying repository implementation may
352
have quicker access to a non-sha1 verifier.
354
:param path: The path that this file can be found at.
355
:param file_id: The handle for this file.
356
These must point to the same object.
357
:param stat_value: Optional stat value for the object
359
raise NotImplementedError(self.get_file_sha1)
361
def get_file_mtime(self, path, file_id=None):
362
"""Return the modification time for a file.
364
:param path: The path that this file can be found at.
365
:param file_id: The handle for this file.
366
These must point to the same object.
368
raise NotImplementedError(self.get_file_mtime)
370
def get_file_size(self, path, file_id=None):
371
"""Return the size of a file in bytes.
373
This applies only to regular files. If invoked on directories or
374
symlinks, it will return None.
375
:param file_id: The file-id of the file
377
raise NotImplementedError(self.get_file_size)
379
def is_executable(self, path, file_id=None):
380
"""Check if a file is executable.
382
:param path: The path that this file can be found at.
383
:param file_id: The handle for this file.
384
These must point to the same object.
386
raise NotImplementedError(self.is_executable)
388
def iter_files_bytes(self, desired_files):
389
"""Iterate through file contents.
391
Files will not necessarily be returned in the order they occur in
392
desired_files. No specific order is guaranteed.
394
Yields pairs of identifier, bytes_iterator. identifier is an opaque
395
value supplied by the caller as part of desired_files. It should
396
uniquely identify the file version in the caller's context. (Examples:
397
an index number or a TreeTransform trans_id.)
399
bytes_iterator is an iterable of bytestrings for the file. The
400
kind of iterable and length of the bytestrings are unspecified, but for
401
this implementation, it is a tuple containing a single bytestring with
402
the complete text of the file.
404
:param desired_files: a list of (file_id, identifier) pairs
406
for file_id, identifier in desired_files:
407
# We wrap the string in a tuple so that we can return an iterable
408
# of bytestrings. (Technically, a bytestring is also an iterable
409
# of bytestrings, but iterating through each character is not
411
# TODO(jelmer): Pass paths into iter_files_bytes
412
path = self.id2path(file_id)
413
cur_file = (self.get_file_text(path, file_id),)
414
yield identifier, cur_file
416
def get_symlink_target(self, path, file_id=None):
417
"""Get the target for a given file_id.
419
It is assumed that the caller already knows that file_id is referencing
421
:param file_id: Handle for the symlink entry.
422
:param path: The path of the file.
423
If both file_id and path are supplied, an implementation may use
425
:return: The path the symlink points to.
427
raise NotImplementedError(self.get_symlink_target)
429
def get_root_id(self):
430
"""Return the file_id for the root of this tree."""
431
raise NotImplementedError(self.get_root_id)
433
def annotate_iter(self, path, file_id=None,
434
default_revision=_mod_revision.CURRENT_REVISION):
435
"""Return an iterator of revision_id, line tuples.
437
For working trees (and mutable trees in general), the special
438
revision_id 'current:' will be used for lines that are new in this
439
tree, e.g. uncommitted changes.
440
:param file_id: The file to produce an annotated version from
441
:param default_revision: For lines that don't match a basis, mark them
442
with this revision id. Not all implementations will make use of
445
raise NotImplementedError(self.annotate_iter)
447
def _get_plan_merge_data(self, file_id, other, base):
448
from .bzr import versionedfile
449
vf = versionedfile._PlanMergeVersionedFile(file_id)
450
last_revision_a = self._get_file_revision(
451
self.id2path(file_id), file_id, vf, 'this:')
452
last_revision_b = other._get_file_revision(
453
other.id2path(file_id), file_id, vf, 'other:')
455
last_revision_base = None
457
last_revision_base = base._get_file_revision(
458
base.id2path(file_id), file_id, vf, 'base:')
459
return vf, last_revision_a, last_revision_b, last_revision_base
461
def plan_file_merge(self, file_id, other, base=None):
462
"""Generate a merge plan based on annotations.
464
If the file contains uncommitted changes in this tree, they will be
465
attributed to the 'current:' pseudo-revision. If the file contains
466
uncommitted changes in the other tree, they will be assigned to the
467
'other:' pseudo-revision.
469
data = self._get_plan_merge_data(file_id, other, base)
470
vf, last_revision_a, last_revision_b, last_revision_base = data
471
return vf.plan_merge(last_revision_a, last_revision_b,
474
def plan_file_lca_merge(self, file_id, other, base=None):
475
"""Generate a merge plan based lca-newness.
477
If the file contains uncommitted changes in this tree, they will be
478
attributed to the 'current:' pseudo-revision. If the file contains
479
uncommitted changes in the other tree, they will be assigned to the
480
'other:' pseudo-revision.
482
data = self._get_plan_merge_data(file_id, other, base)
483
vf, last_revision_a, last_revision_b, last_revision_base = data
484
return vf.plan_lca_merge(last_revision_a, last_revision_b,
487
def _iter_parent_trees(self):
488
"""Iterate through parent trees, defaulting to Tree.revision_tree."""
489
for revision_id in self.get_parent_ids():
491
yield self.revision_tree(revision_id)
492
except errors.NoSuchRevisionInTree:
493
yield self.repository.revision_tree(revision_id)
495
def _get_file_revision(self, path, file_id, vf, tree_revision):
496
"""Ensure that file_id, tree_revision is in vf to plan the merge."""
497
if getattr(self, '_repository', None) is None:
498
last_revision = tree_revision
499
parent_keys = [(file_id, t.get_file_revision(path, file_id)) for t in
500
self._iter_parent_trees()]
501
vf.add_lines((file_id, last_revision), parent_keys,
502
self.get_file_lines(path, file_id))
503
repo = self.branch.repository
506
last_revision = self.get_file_revision(path, file_id)
507
base_vf = self._repository.texts
508
if base_vf not in vf.fallback_versionedfiles:
509
vf.fallback_versionedfiles.append(base_vf)
83
512
def _check_retrieved(self, ie, f):
86
fp = fingerprint_file(f)
515
fp = osutils.fingerprint_file(f)
89
if ie.text_size != None:
518
if ie.text_size is not None:
90
519
if ie.text_size != fp['size']:
91
raise BzrError("mismatched size for file %r in %r" % (ie.file_id, self._store),
520
raise errors.BzrError(
521
"mismatched size for file %r in %r" %
522
(ie.file_id, self._store),
92
523
["inventory expects %d bytes" % ie.text_size,
93
524
"file is actually %d bytes" % fp['size'],
94
525
"store is probably damaged/corrupt"])
96
527
if ie.text_sha1 != fp['sha1']:
97
raise BzrError("wrong SHA-1 for file %r in %r" % (ie.file_id, self._store),
528
raise errors.BzrError("wrong SHA-1 for file %r in %r" %
529
(ie.file_id, self._store),
98
530
["inventory expects %s" % ie.text_sha1,
99
531
"file is actually %s" % fp['sha1'],
100
532
"store is probably damaged/corrupt"])
103
def print_file(self, file_id):
104
"""Print file with id `file_id` to stdout."""
106
sys.stdout.write(self.get_file_text(file_id))
534
def path2id(self, path):
535
"""Return the id for path in this tree."""
536
raise NotImplementedError(self.path2id)
538
def paths2ids(self, paths, trees=[], require_versioned=True):
539
"""Return all the ids that can be reached by walking from paths.
541
Each path is looked up in this tree and any extras provided in
542
trees, and this is repeated recursively: the children in an extra tree
543
of a directory that has been renamed under a provided path in this tree
544
are all returned, even if none exist under a provided path in this
545
tree, and vice versa.
547
:param paths: An iterable of paths to start converting to ids from.
548
Alternatively, if paths is None, no ids should be calculated and None
549
will be returned. This is offered to make calling the api unconditional
550
for code that *might* take a list of files.
551
:param trees: Additional trees to consider.
552
:param require_versioned: If False, do not raise NotVersionedError if
553
an element of paths is not versioned in this tree and all of trees.
555
return find_ids_across_trees(paths, [self] + list(trees), require_versioned)
557
def iter_children(self, file_id):
558
"""Iterate over the file ids of the children of an entry.
560
:param file_id: File id of the entry
561
:return: Iterator over child file ids.
563
raise NotImplementedError(self.iter_children)
108
565
def lock_read(self):
566
"""Lock this tree for multiple read only operations.
568
:return: A breezy.lock.LogicalLockResult.
570
return lock.LogicalLockResult(self.unlock)
572
def revision_tree(self, revision_id):
573
"""Obtain a revision tree for the revision revision_id.
575
The intention of this method is to allow access to possibly cached
576
tree data. Implementors of this method should raise NoSuchRevision if
577
the tree is not locally available, even if they could obtain the
578
tree via a repository or some other means. Callers are responsible
579
for finding the ultimate source for a revision tree.
581
:param revision_id: The revision_id of the requested tree.
583
:raises: NoSuchRevision if the tree cannot be obtained.
585
raise errors.NoSuchRevisionInTree(self, revision_id)
588
"""What files are present in this tree and unknown.
590
:return: an iterator over the unknown files.
111
594
def unlock(self):
114
597
def filter_unversioned_files(self, paths):
115
"""Filter out paths that are not versioned.
598
"""Filter out paths that are versioned.
117
600
:return: set of paths.
119
# NB: we specifically *don't* call self.has_filename, because for
120
# WorkingTrees that can indicate files that exist on disk but that
122
pred = self.inventory.has_filename
123
return set((p for p in paths if not pred(p)))
126
class RevisionTree(Tree):
127
"""Tree viewing a previous revision.
129
File text can be retrieved from the text store.
131
TODO: Some kind of `__repr__` method, but a good one
132
probably means knowing the branch and revision number,
133
or at least passing a description to the constructor.
136
def __init__(self, branch, inv, revision_id):
137
self._branch = branch
138
self._weave_store = branch.weave_store
139
self._inventory = inv
140
self._revision_id = revision_id
142
def get_weave(self, file_id):
143
return self._weave_store.get_weave(file_id,
144
self._branch.get_transaction())
146
def get_file_lines(self, file_id):
147
ie = self._inventory[file_id]
148
weave = self.get_weave(file_id)
149
return weave.get_lines(ie.revision)
151
def get_file_text(self, file_id):
152
return ''.join(self.get_file_lines(file_id))
154
def get_file(self, file_id):
155
return StringIO(self.get_file_text(file_id))
157
def get_file_size(self, file_id):
158
return self._inventory[file_id].text_size
160
def get_file_sha1(self, file_id):
161
ie = self._inventory[file_id]
162
if ie.kind == "file":
165
def is_executable(self, file_id):
166
ie = self._inventory[file_id]
167
if ie.kind != "file":
169
return self._inventory[file_id].executable
171
def has_filename(self, filename):
172
return bool(self.inventory.path2id(filename))
174
def list_files(self):
175
# The only files returned by this are those from the version
176
for path, entry in self.inventory.iter_entries():
177
yield path, 'V', entry.kind, entry.file_id, entry
179
def get_symlink_target(self, file_id):
180
ie = self._inventory[file_id]
181
return ie.symlink_target;
183
def kind(self, file_id):
184
return self._inventory[file_id].kind
187
self._branch.lock_read()
190
self._branch.unlock()
193
class EmptyTree(Tree):
195
self._inventory = Inventory()
197
def get_symlink_target(self, file_id):
200
def has_filename(self, filename):
602
raise NotImplementedError(self.filter_unversioned_files)
604
def walkdirs(self, prefix=""):
605
"""Walk the contents of this tree from path down.
607
This yields all the data about the contents of a directory at a time.
608
After each directory has been yielded, if the caller has mutated the
609
list to exclude some directories, they are then not descended into.
611
The data yielded is of the form:
612
((directory-relpath, directory-path-from-root, directory-fileid),
613
[(relpath, basename, kind, lstat, path_from_tree_root, file_id,
614
versioned_kind), ...]),
615
- directory-relpath is the containing dirs relpath from prefix
616
- directory-path-from-root is the containing dirs path from /
617
- directory-fileid is the id of the directory if it is versioned.
618
- relpath is the relative path within the subtree being walked.
619
- basename is the basename
620
- kind is the kind of the file now. If unknonwn then the file is not
621
present within the tree - but it may be recorded as versioned. See
623
- lstat is the stat data *if* the file was statted.
624
- path_from_tree_root is the path from the root of the tree.
625
- file_id is the file_id if the entry is versioned.
626
- versioned_kind is the kind of the file as last recorded in the
627
versioning system. If 'unknown' the file is not versioned.
628
One of 'kind' and 'versioned_kind' must not be 'unknown'.
630
:param prefix: Start walking from prefix within the tree rather than
631
at the root. This allows one to walk a subtree but get paths that are
632
relative to a tree rooted higher up.
633
:return: an iterator over the directory data.
635
raise NotImplementedError(self.walkdirs)
637
def supports_content_filtering(self):
203
def kind(self, file_id):
204
assert self._inventory[file_id].kind == "root_directory"
205
return "root_directory"
207
def list_files(self):
210
def __contains__(self, file_id):
211
return file_id in self._inventory
213
def get_file_sha1(self, file_id):
214
assert self._inventory[file_id].kind == "root_directory"
640
def _content_filter_stack(self, path=None, file_id=None):
641
"""The stack of content filters for a path if filtering is supported.
643
Readers will be applied in first-to-last order.
644
Writers will be applied in last-to-first order.
645
Either the path or the file-id needs to be provided.
647
:param path: path relative to the root of the tree
649
:param file_id: file_id or None if unknown
650
:return: the list of filters - [] if there are none
652
filter_pref_names = filters._get_registered_names()
653
if len(filter_pref_names) == 0:
656
path = self.id2path(file_id)
657
prefs = next(self.iter_search_rules([path], filter_pref_names))
658
stk = filters._get_filter_stack_for(prefs)
659
if 'filters' in debug.debug_flags:
660
trace.note(gettext("*** {0} content-filter: {1} => {2!r}").format(path, prefs, stk))
663
def _content_filter_stack_provider(self):
664
"""A function that returns a stack of ContentFilters.
666
The function takes a path (relative to the top of the tree) and a
667
file-id as parameters.
669
:return: None if content filtering is not supported by this tree.
671
if self.supports_content_filtering():
672
return lambda path, file_id: \
673
self._content_filter_stack(path, file_id)
677
def iter_search_rules(self, path_names, pref_names=None,
678
_default_searcher=None):
679
"""Find the preferences for filenames in a tree.
681
:param path_names: an iterable of paths to find attributes for.
682
Paths are given relative to the root of the tree.
683
:param pref_names: the list of preferences to lookup - None for all
684
:param _default_searcher: private parameter to assist testing - don't use
685
:return: an iterator of tuple sequences, one per path-name.
686
See _RulesSearcher.get_items for details on the tuple sequence.
688
if _default_searcher is None:
689
_default_searcher = rules._per_user_searcher
690
searcher = self._get_rules_searcher(_default_searcher)
691
if searcher is not None:
692
if pref_names is not None:
693
for path in path_names:
694
yield searcher.get_selected_items(path, pref_names)
696
for path in path_names:
697
yield searcher.get_items(path)
699
def _get_rules_searcher(self, default_searcher):
700
"""Get the RulesSearcher for this tree given the default one."""
701
searcher = default_searcher
705
def find_ids_across_trees(filenames, trees, require_versioned=True):
706
"""Find the ids corresponding to specified filenames.
708
All matches in all trees will be used, and all children of matched
709
directories will be used.
711
:param filenames: The filenames to find file_ids for (if None, returns
713
:param trees: The trees to find file_ids within
714
:param require_versioned: if true, all specified filenames must occur in
716
:return: a set of file ids for the specified filenames and their children.
218
######################################################################
221
# TODO: Merge these two functions into a single one that can operate
222
# on either a whole tree or a set of files.
224
# TODO: Return the diff in order by filename, not by category or in
225
# random order. Can probably be done by lock-stepping through the
226
# filenames from both trees.
229
def file_status(filename, old_tree, new_tree):
230
"""Return single-letter status, old and new names for a file.
232
The complexity here is in deciding how to represent renames;
233
many complex cases are possible.
235
old_inv = old_tree.inventory
236
new_inv = new_tree.inventory
237
new_id = new_inv.path2id(filename)
238
old_id = old_inv.path2id(filename)
240
if not new_id and not old_id:
241
# easy: doesn't exist in either; not versioned at all
242
if new_tree.is_ignored(filename):
243
return 'I', None, None
245
return '?', None, None
247
# There is now a file of this name, great.
250
# There is no longer a file of this name, but we can describe
251
# what happened to the file that used to have
252
# this name. There are two possibilities: either it was
253
# deleted entirely, or renamed.
255
if new_inv.has_id(old_id):
256
return 'X', old_inv.id2path(old_id), new_inv.id2path(old_id)
258
return 'D', old_inv.id2path(old_id), None
260
# if the file_id is new in this revision, it is added
261
if new_id and not old_inv.has_id(new_id):
264
# if there used to be a file of this name, but that ID has now
265
# disappeared, it is deleted
266
if old_id and not new_inv.has_id(old_id):
273
def find_renames(old_inv, new_inv):
274
for file_id in old_inv:
275
if file_id not in new_inv:
277
old_name = old_inv.id2path(file_id)
278
new_name = new_inv.id2path(file_id)
279
if old_name != new_name:
280
yield (old_name, new_name)
720
specified_path_ids = _find_ids_across_trees(filenames, trees,
722
return _find_children_across_trees(specified_path_ids, trees)
725
def _find_ids_across_trees(filenames, trees, require_versioned):
726
"""Find the ids corresponding to specified filenames.
728
All matches in all trees will be used, but subdirectories are not scanned.
730
:param filenames: The filenames to find file_ids for
731
:param trees: The trees to find file_ids within
732
:param require_versioned: if true, all specified filenames must occur in
734
:return: a set of file ids for the specified filenames
737
interesting_ids = set()
738
for tree_path in filenames:
741
file_id = tree.path2id(tree_path)
742
if file_id is not None:
743
interesting_ids.add(file_id)
746
not_versioned.append(tree_path)
747
if len(not_versioned) > 0 and require_versioned:
748
raise errors.PathsNotVersionedError(not_versioned)
749
return interesting_ids
752
def _find_children_across_trees(specified_ids, trees):
753
"""Return a set including specified ids and their children.
755
All matches in all trees will be used.
757
:param trees: The trees to find file_ids within
758
:return: a set containing all specified ids and their children
760
interesting_ids = set(specified_ids)
761
pending = interesting_ids
762
# now handle children of interesting ids
763
# we loop so that we handle all children of each id in both trees
764
while len(pending) > 0:
766
for file_id in pending:
768
if not tree.has_or_had_id(file_id):
770
for child_id in tree.iter_children(file_id):
771
if child_id not in interesting_ids:
772
new_pending.add(child_id)
773
interesting_ids.update(new_pending)
774
pending = new_pending
775
return interesting_ids
778
class InterTree(InterObject):
779
"""This class represents operations taking place between two Trees.
781
Its instances have methods like 'compare' and contain references to the
782
source and target trees these operations are to be carried out on.
784
Clients of breezy should not need to use InterTree directly, rather they
785
should use the convenience methods on Tree such as 'Tree.compare()' which
786
will pass through to InterTree as appropriate.
789
# Formats that will be used to test this InterTree. If both are
790
# None, this InterTree will not be tested (e.g. because a complex
792
_matching_from_tree_format = None
793
_matching_to_tree_format = None
798
def is_compatible(kls, source, target):
799
# The default implementation is naive and uses the public API, so
800
# it works for all trees.
803
def _changes_from_entries(self, source_entry, target_entry,
804
source_path=None, target_path=None):
805
"""Generate a iter_changes tuple between source_entry and target_entry.
807
:param source_entry: An inventory entry from self.source, or None.
808
:param target_entry: An inventory entry from self.target, or None.
809
:param source_path: The path of source_entry, if known. If not known
810
it will be looked up.
811
:param target_path: The path of target_entry, if known. If not known
812
it will be looked up.
813
:return: A tuple, item 0 of which is an iter_changes result tuple, and
814
item 1 is True if there are any changes in the result tuple.
816
if source_entry is None:
817
if target_entry is None:
819
file_id = target_entry.file_id
821
file_id = source_entry.file_id
822
if source_entry is not None:
823
source_versioned = True
824
source_name = source_entry.name
825
source_parent = source_entry.parent_id
826
if source_path is None:
827
source_path = self.source.id2path(file_id)
828
source_kind, source_executable, source_stat = \
829
self.source._comparison_data(source_entry, source_path)
831
source_versioned = False
835
source_executable = None
836
if target_entry is not None:
837
target_versioned = True
838
target_name = target_entry.name
839
target_parent = target_entry.parent_id
840
if target_path is None:
841
target_path = self.target.id2path(file_id)
842
target_kind, target_executable, target_stat = \
843
self.target._comparison_data(target_entry, target_path)
845
target_versioned = False
849
target_executable = None
850
versioned = (source_versioned, target_versioned)
851
kind = (source_kind, target_kind)
852
changed_content = False
853
if source_kind != target_kind:
854
changed_content = True
855
elif source_kind == 'file':
856
if not self.file_content_matches(file_id, file_id, source_path,
857
target_path, source_stat, target_stat):
858
changed_content = True
859
elif source_kind == 'symlink':
860
if (self.source.get_symlink_target(source_path, file_id) !=
861
self.target.get_symlink_target(target_path, file_id)):
862
changed_content = True
863
elif source_kind == 'tree-reference':
864
if (self.source.get_reference_revision(source_path, file_id)
865
!= self.target.get_reference_revision(target_path, file_id)):
866
changed_content = True
867
parent = (source_parent, target_parent)
868
name = (source_name, target_name)
869
executable = (source_executable, target_executable)
870
if (changed_content is not False or versioned[0] != versioned[1]
871
or parent[0] != parent[1] or name[0] != name[1] or
872
executable[0] != executable[1]):
876
return (file_id, (source_path, target_path), changed_content,
877
versioned, parent, name, kind, executable), changes
879
def compare(self, want_unchanged=False, specific_files=None,
880
extra_trees=None, require_versioned=False, include_root=False,
881
want_unversioned=False):
882
"""Return the changes from source to target.
884
:return: A TreeDelta.
885
:param specific_files: An optional list of file paths to restrict the
886
comparison to. When mapping filenames to ids, all matches in all
887
trees (including optional extra_trees) are used, and all children of
888
matched directories are included.
889
:param want_unchanged: An optional boolean requesting the inclusion of
890
unchanged entries in the result.
891
:param extra_trees: An optional list of additional trees to use when
892
mapping the contents of specific_files (paths) to file_ids.
893
:param require_versioned: An optional boolean (defaults to False). When
894
supplied and True all the 'specific_files' must be versioned, or
895
a PathsNotVersionedError will be thrown.
896
:param want_unversioned: Scan for unversioned paths.
898
trees = (self.source,)
899
if extra_trees is not None:
900
trees = trees + tuple(extra_trees)
901
with self.lock_read():
902
# target is usually the newer tree:
903
specific_file_ids = self.target.paths2ids(specific_files, trees,
904
require_versioned=require_versioned)
905
if specific_files and not specific_file_ids:
906
# All files are unversioned, so just return an empty delta
907
# _compare_trees would think we want a complete delta
908
result = delta.TreeDelta()
909
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
910
result.unversioned = [(path, None,
911
self.target._comparison_data(fake_entry, path)[0]) for path in
914
return delta._compare_trees(self.source, self.target, want_unchanged,
915
specific_files, include_root, extra_trees=extra_trees,
916
require_versioned=require_versioned,
917
want_unversioned=want_unversioned)
919
def iter_changes(self, include_unchanged=False,
920
specific_files=None, pb=None, extra_trees=[],
921
require_versioned=True, want_unversioned=False):
922
"""Generate an iterator of changes between trees.
925
(file_id, (path_in_source, path_in_target),
926
changed_content, versioned, parent, name, kind,
929
Changed_content is True if the file's content has changed. This
930
includes changes to its kind, and to a symlink's target.
932
versioned, parent, name, kind, executable are tuples of (from, to).
933
If a file is missing in a tree, its kind is None.
935
Iteration is done in parent-to-child order, relative to the target
938
There is no guarantee that all paths are in sorted order: the
939
requirement to expand the search due to renames may result in children
940
that should be found early being found late in the search, after
941
lexically later results have been returned.
942
:param require_versioned: Raise errors.PathsNotVersionedError if a
943
path in the specific_files list is not versioned in one of
944
source, target or extra_trees.
945
:param specific_files: An optional list of file paths to restrict the
946
comparison to. When mapping filenames to ids, all matches in all
947
trees (including optional extra_trees) are used, and all children
948
of matched directories are included. The parents in the target tree
949
of the specific files up to and including the root of the tree are
950
always evaluated for changes too.
951
:param want_unversioned: Should unversioned files be returned in the
952
output. An unversioned file is defined as one with (False, False)
953
for the versioned pair.
955
lookup_trees = [self.source]
957
lookup_trees.extend(extra_trees)
958
# The ids of items we need to examine to insure delta consistency.
959
precise_file_ids = set()
960
changed_file_ids = []
961
if specific_files == []:
962
specific_file_ids = []
964
specific_file_ids = self.target.paths2ids(specific_files,
965
lookup_trees, require_versioned=require_versioned)
966
if specific_files is not None:
967
# reparented or added entries must have their parents included
968
# so that valid deltas can be created. The seen_parents set
969
# tracks the parents that we need to have.
970
# The seen_dirs set tracks directory entries we've yielded.
971
# After outputting version object in to_entries we set difference
972
# the two seen sets and start checking parents.
976
all_unversioned = sorted([(p.split('/'), p) for p in
978
if specific_files is None or
979
osutils.is_inside_any(specific_files, p)])
980
all_unversioned = collections.deque(all_unversioned)
982
all_unversioned = collections.deque()
984
from_entries_by_dir = list(self.source.iter_entries_by_dir(
985
specific_file_ids=specific_file_ids))
986
from_data = dict((e.file_id, (p, e)) for p, e in from_entries_by_dir)
987
to_entries_by_dir = list(self.target.iter_entries_by_dir(
988
specific_file_ids=specific_file_ids))
989
num_entries = len(from_entries_by_dir) + len(to_entries_by_dir)
991
# the unversioned path lookup only occurs on real trees - where there
992
# can be extras. So the fake_entry is solely used to look up
993
# executable it values when execute is not supported.
994
fake_entry = inventory.InventoryFile('unused', 'unused', 'unused')
995
for target_path, target_entry in to_entries_by_dir:
996
while (all_unversioned and
997
all_unversioned[0][0] < target_path.split('/')):
998
unversioned_path = all_unversioned.popleft()
999
target_kind, target_executable, target_stat = \
1000
self.target._comparison_data(fake_entry, unversioned_path[1])
1001
yield (None, (None, unversioned_path[1]), True, (False, False),
1003
(None, unversioned_path[0][-1]),
1004
(None, target_kind),
1005
(None, target_executable))
1006
source_path, source_entry = from_data.get(target_entry.file_id,
1008
result, changes = self._changes_from_entries(source_entry,
1009
target_entry, source_path=source_path, target_path=target_path)
1010
to_paths[result[0]] = result[1][1]
1015
pb.update('comparing files', entry_count, num_entries)
1016
if changes or include_unchanged:
1017
if specific_file_ids is not None:
1018
new_parent_id = result[4][1]
1019
precise_file_ids.add(new_parent_id)
1020
changed_file_ids.append(result[0])
1022
# Ensure correct behaviour for reparented/added specific files.
1023
if specific_files is not None:
1024
# Record output dirs
1025
if result[6][1] == 'directory':
1026
seen_dirs.add(result[0])
1027
# Record parents of reparented/added entries.
1028
versioned = result[3]
1030
if not versioned[0] or parents[0] != parents[1]:
1031
seen_parents.add(parents[1])
1032
while all_unversioned:
1033
# yield any trailing unversioned paths
1034
unversioned_path = all_unversioned.popleft()
1035
to_kind, to_executable, to_stat = \
1036
self.target._comparison_data(fake_entry, unversioned_path[1])
1037
yield (None, (None, unversioned_path[1]), True, (False, False),
1039
(None, unversioned_path[0][-1]),
1041
(None, to_executable))
1042
# Yield all remaining source paths
1043
for path, from_entry in from_entries_by_dir:
1044
file_id = from_entry.file_id
1045
if file_id in to_paths:
1048
if not self.target.has_id(file_id):
1049
# common case - paths we have not emitted are not present in
1053
to_path = self.target.id2path(file_id)
1056
pb.update('comparing files', entry_count, num_entries)
1057
versioned = (True, False)
1058
parent = (from_entry.parent_id, None)
1059
name = (from_entry.name, None)
1060
from_kind, from_executable, stat_value = \
1061
self.source._comparison_data(from_entry, path)
1062
kind = (from_kind, None)
1063
executable = (from_executable, None)
1064
changed_content = from_kind is not None
1065
# the parent's path is necessarily known at this point.
1066
changed_file_ids.append(file_id)
1067
yield(file_id, (path, to_path), changed_content, versioned, parent,
1068
name, kind, executable)
1069
changed_file_ids = set(changed_file_ids)
1070
if specific_file_ids is not None:
1071
for result in self._handle_precise_ids(precise_file_ids,
1075
def _get_entry(self, tree, file_id):
1076
"""Get an inventory entry from a tree, with missing entries as None.
1078
If the tree raises NotImplementedError on accessing .inventory, then
1079
this is worked around using iter_entries_by_dir on just the file id
1082
:param tree: The tree to lookup the entry in.
1083
:param file_id: The file_id to lookup.
1086
inventory = tree.root_inventory
1087
except NotImplementedError:
1088
# No inventory available.
1090
iterator = tree.iter_entries_by_dir(specific_file_ids=[file_id])
1091
return iterator.next()[1]
1092
except StopIteration:
1096
return inventory[file_id]
1097
except errors.NoSuchId:
1100
def _handle_precise_ids(self, precise_file_ids, changed_file_ids,
1101
discarded_changes=None):
1102
"""Fill out a partial iter_changes to be consistent.
1104
:param precise_file_ids: The file ids of parents that were seen during
1106
:param changed_file_ids: The file ids of already emitted items.
1107
:param discarded_changes: An optional dict of precalculated
1108
iter_changes items which the partial iter_changes had not output
1110
:return: A generator of iter_changes items to output.
1112
# process parents of things that had changed under the users
1113
# requested paths to prevent incorrect paths or parent ids which
1114
# aren't in the tree.
1115
while precise_file_ids:
1116
precise_file_ids.discard(None)
1117
# Don't emit file_ids twice
1118
precise_file_ids.difference_update(changed_file_ids)
1119
if not precise_file_ids:
1121
# If the there was something at a given output path in source, we
1122
# have to include the entry from source in the delta, or we would
1123
# be putting this entry into a used path.
1125
for parent_id in precise_file_ids:
1127
paths.append(self.target.id2path(parent_id))
1128
except errors.NoSuchId:
1129
# This id has been dragged in from the source by delta
1130
# expansion and isn't present in target at all: we don't
1131
# need to check for path collisions on it.
1134
old_id = self.source.path2id(path)
1135
precise_file_ids.add(old_id)
1136
precise_file_ids.discard(None)
1137
current_ids = precise_file_ids
1138
precise_file_ids = set()
1139
# We have to emit all of precise_file_ids that have been altered.
1140
# We may have to output the children of some of those ids if any
1141
# directories have stopped being directories.
1142
for file_id in current_ids:
1144
if discarded_changes:
1145
result = discarded_changes.get(file_id)
1150
old_entry = self._get_entry(self.source, file_id)
1151
new_entry = self._get_entry(self.target, file_id)
1152
result, changes = self._changes_from_entries(
1153
old_entry, new_entry)
1156
# Get this parents parent to examine.
1157
new_parent_id = result[4][1]
1158
precise_file_ids.add(new_parent_id)
1160
if (result[6][0] == 'directory' and
1161
result[6][1] != 'directory'):
1162
# This stopped being a directory, the old children have
1164
if old_entry is None:
1165
# Reusing a discarded change.
1166
old_entry = self._get_entry(self.source, file_id)
1167
precise_file_ids.update(
1168
self.source.iter_children(file_id))
1169
changed_file_ids.add(result[0])
1172
def file_content_matches(
1173
self, source_file_id, target_file_id, source_path=None,
1174
target_path=None, source_stat=None, target_stat=None):
1175
"""Check if two files are the same in the source and target trees.
1177
This only checks that the contents of the files are the same,
1178
it does not touch anything else.
1180
:param source_file_id: File id of the file in the source tree
1181
:param target_file_id: File id of the file in the target tree
1182
:param source_path: Path of the file in the source tree
1183
:param target_path: Path of the file in the target tree
1184
:param source_stat: Optional stat value of the file in the source tree
1185
:param target_stat: Optional stat value of the file in the target tree
1186
:return: Boolean indicating whether the files have the same contents
1188
with self.lock_read():
1189
if source_path is None:
1190
source_path = self.source.id2path(source_file_id)
1191
if target_path is None:
1192
target_path = self.target.id2path(target_file_id)
1193
source_verifier_kind, source_verifier_data = (
1194
self.source.get_file_verifier(
1195
source_path, source_file_id, source_stat))
1196
target_verifier_kind, target_verifier_data = (
1197
self.target.get_file_verifier(
1198
target_path, target_file_id, target_stat))
1199
if source_verifier_kind == target_verifier_kind:
1200
return (source_verifier_data == target_verifier_data)
1201
# Fall back to SHA1 for now
1202
if source_verifier_kind != "SHA1":
1203
source_sha1 = self.source.get_file_sha1(
1204
source_path, source_file_id, source_stat)
1206
source_sha1 = source_verifier_data
1207
if target_verifier_kind != "SHA1":
1208
target_sha1 = self.target.get_file_sha1(
1209
target_path, target_file_id, target_stat)
1211
target_sha1 = target_verifier_data
1212
return (source_sha1 == target_sha1)
1214
InterTree.register_optimiser(InterTree)
1217
class MultiWalker(object):
1218
"""Walk multiple trees simultaneously, getting combined results."""
1220
# Note: This could be written to not assume you can do out-of-order
1221
# lookups. Instead any nodes that don't match in all trees could be
1222
# marked as 'deferred', and then returned in the final cleanup loop.
1223
# For now, I think it is "nicer" to return things as close to the
1224
# "master_tree" order as we can.
1226
def __init__(self, master_tree, other_trees):
1227
"""Create a new MultiWalker.
1229
All trees being walked must implement "iter_entries_by_dir()", such
1230
that they yield (path, object) tuples, where that object will have a
1231
'.file_id' member, that can be used to check equality.
1233
:param master_tree: All trees will be 'slaved' to the master_tree such
1234
that nodes in master_tree will be used as 'first-pass' sync points.
1235
Any nodes that aren't in master_tree will be merged in a second
1237
:param other_trees: A list of other trees to walk simultaneously.
1239
self._master_tree = master_tree
1240
self._other_trees = other_trees
1242
# Keep track of any nodes that were properly processed just out of
1243
# order, that way we don't return them at the end, we don't have to
1244
# track *all* processed file_ids, just the out-of-order ones
1245
self._out_of_order_processed = set()
1248
def _step_one(iterator):
1249
"""Step an iter_entries_by_dir iterator.
1251
:return: (has_more, path, ie)
1252
If has_more is False, path and ie will be None.
1255
path, ie = next(iterator)
1256
except StopIteration:
1257
return False, None, None
1259
return True, path, ie
1262
def _cmp_path_by_dirblock(path1, path2):
1263
"""Compare two paths based on what directory they are in.
1265
This generates a sort order, such that all children of a directory are
1266
sorted together, and grandchildren are in the same order as the
1267
children appear. But all grandchildren come after all children.
1269
:param path1: first path
1270
:param path2: the second path
1271
:return: negative number if ``path1`` comes first,
1272
0 if paths are equal
1273
and a positive number if ``path2`` sorts first
1275
# Shortcut this special case
1278
# This is stolen from _dirstate_helpers_py.py, only switching it to
1279
# Unicode objects. Consider using encode_utf8() and then using the
1280
# optimized versions, or maybe writing optimized unicode versions.
1281
if not isinstance(path1, unicode):
1282
raise TypeError("'path1' must be a unicode string, not %s: %r"
1283
% (type(path1), path1))
1284
if not isinstance(path2, unicode):
1285
raise TypeError("'path2' must be a unicode string, not %s: %r"
1286
% (type(path2), path2))
1287
return cmp(MultiWalker._path_to_key(path1),
1288
MultiWalker._path_to_key(path2))
1291
def _path_to_key(path):
1292
dirname, basename = osutils.split(path)
1293
return (dirname.split(u'/'), basename)
1295
def _lookup_by_file_id(self, extra_entries, other_tree, file_id):
1296
"""Lookup an inventory entry by file_id.
1298
This is called when an entry is missing in the normal order.
1299
Generally this is because a file was either renamed, or it was
1300
deleted/added. If the entry was found in the inventory and not in
1301
extra_entries, it will be added to self._out_of_order_processed
1303
:param extra_entries: A dictionary of {file_id: (path, ie)}. This
1304
should be filled with entries that were found before they were
1305
used. If file_id is present, it will be removed from the
1307
:param other_tree: The Tree to search, in case we didn't find the entry
1309
:param file_id: The file_id to look for
1310
:return: (path, ie) if found or (None, None) if not present.
1312
if file_id in extra_entries:
1313
return extra_entries.pop(file_id)
1314
# TODO: Is id2path better as the first call, or is
1315
# inventory[file_id] better as a first check?
1317
cur_path = other_tree.id2path(file_id)
1318
except errors.NoSuchId:
1320
if cur_path is None:
1323
self._out_of_order_processed.add(file_id)
1324
cur_ie = other_tree.root_inventory[file_id]
1325
return (cur_path, cur_ie)
1328
"""Match up the values in the different trees."""
1329
for result in self._walk_master_tree():
1331
self._finish_others()
1332
for result in self._walk_others():
1335
def _walk_master_tree(self):
1336
"""First pass, walk all trees in lock-step.
1338
When we are done, all nodes in the master_tree will have been
1339
processed. _other_walkers, _other_entries, and _others_extra will be
1340
set on 'self' for future processing.
1342
# This iterator has the most "inlining" done, because it tends to touch
1343
# every file in the tree, while the others only hit nodes that don't
1345
master_iterator = self._master_tree.iter_entries_by_dir()
1347
other_walkers = [other.iter_entries_by_dir()
1348
for other in self._other_trees]
1349
other_entries = [self._step_one(walker) for walker in other_walkers]
1350
# Track extra nodes in the other trees
1351
others_extra = [{} for _ in range(len(self._other_trees))]
1353
master_has_more = True
1354
step_one = self._step_one
1355
lookup_by_file_id = self._lookup_by_file_id
1356
out_of_order_processed = self._out_of_order_processed
1358
while master_has_more:
1359
(master_has_more, path, master_ie) = step_one(master_iterator)
1360
if not master_has_more:
1363
file_id = master_ie.file_id
1365
other_values_append = other_values.append
1366
next_other_entries = []
1367
next_other_entries_append = next_other_entries.append
1368
for idx, (other_has_more, other_path, other_ie) in enumerate(other_entries):
1369
if not other_has_more:
1370
other_values_append(lookup_by_file_id(
1371
others_extra[idx], self._other_trees[idx], file_id))
1372
next_other_entries_append((False, None, None))
1373
elif file_id == other_ie.file_id:
1374
# This is the critical code path, as most of the entries
1375
# should match between most trees.
1376
other_values_append((other_path, other_ie))
1377
next_other_entries_append(step_one(other_walkers[idx]))
1379
# This walker did not match, step it until it either
1380
# matches, or we know we are past the current walker.
1381
other_walker = other_walkers[idx]
1382
other_extra = others_extra[idx]
1383
while (other_has_more and
1384
self._cmp_path_by_dirblock(other_path, path) < 0):
1385
other_file_id = other_ie.file_id
1386
if other_file_id not in out_of_order_processed:
1387
other_extra[other_file_id] = (other_path, other_ie)
1388
other_has_more, other_path, other_ie = \
1389
step_one(other_walker)
1390
if other_has_more and other_ie.file_id == file_id:
1391
# We ended up walking to this point, match and step
1393
other_values_append((other_path, other_ie))
1394
other_has_more, other_path, other_ie = \
1395
step_one(other_walker)
1397
# This record isn't in the normal order, see if it
1399
other_values_append(lookup_by_file_id(
1400
other_extra, self._other_trees[idx], file_id))
1401
next_other_entries_append((other_has_more, other_path,
1403
other_entries = next_other_entries
1405
# We've matched all the walkers, yield this datapoint
1406
yield path, file_id, master_ie, other_values
1407
self._other_walkers = other_walkers
1408
self._other_entries = other_entries
1409
self._others_extra = others_extra
1411
def _finish_others(self):
1412
"""Finish walking the other iterators, so we get all entries."""
1413
for idx, info in enumerate(self._other_entries):
1414
other_extra = self._others_extra[idx]
1415
(other_has_more, other_path, other_ie) = info
1416
while other_has_more:
1417
other_file_id = other_ie.file_id
1418
if other_file_id not in self._out_of_order_processed:
1419
other_extra[other_file_id] = (other_path, other_ie)
1420
other_has_more, other_path, other_ie = \
1421
self._step_one(self._other_walkers[idx])
1422
del self._other_entries
1424
def _walk_others(self):
1425
"""Finish up by walking all the 'deferred' nodes."""
1426
# TODO: One alternative would be to grab all possible unprocessed
1427
# file_ids, and then sort by path, and then yield them. That
1428
# might ensure better ordering, in case a caller strictly
1429
# requires parents before children.
1430
for idx, other_extra in enumerate(self._others_extra):
1431
others = sorted(viewvalues(other_extra),
1432
key=lambda x: self._path_to_key(x[0]))
1433
for other_path, other_ie in others:
1434
file_id = other_ie.file_id
1435
# We don't need to check out_of_order_processed here, because
1436
# the lookup_by_file_id will be removing anything processed
1437
# from the extras cache
1438
other_extra.pop(file_id)
1439
other_values = [(None, None)] * idx
1440
other_values.append((other_path, other_ie))
1441
for alt_idx, alt_extra in enumerate(self._others_extra[idx+1:]):
1442
alt_idx = alt_idx + idx + 1
1443
alt_extra = self._others_extra[alt_idx]
1444
alt_tree = self._other_trees[alt_idx]
1445
other_values.append(self._lookup_by_file_id(
1446
alt_extra, alt_tree, file_id))
1447
yield other_path, file_id, None, other_values
added
removed
breezy/tree.py
