1
# Copyright (C) 2005 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""WorkingTree object and friends.
19
A WorkingTree represents the editable working copy of a branch.
20
Operations which represent the WorkingTree are also done here,
21
such as renaming or adding files. The WorkingTree has an inventory
22
which is updated by these operations. A commit produces a
23
new revision based on the workingtree and its inventory.
25
At the moment every WorkingTree has its own branch. Remote
26
WorkingTrees aren't supported.
28
To get a WorkingTree, call Branch.working_tree():
32
# TODO: Don't allow WorkingTrees to be constructed for remote branches if
35
# FIXME: I don't know if writing out the cache from the destructor is really a
36
# good idea, because destructors are considered poor taste in Python, and it's
37
# not predictable when it will be written out.
39
# TODO: Give the workingtree sole responsibility for the working inventory;
40
# remove the variable and references to it from the branch. This may require
41
# updating the commit code so as to update the inventory within the working
42
# copy, and making sure there's only one WorkingTree for any directory on disk.
43
# At the momenthey may alias the inventory and have old copies of it in memory.
45
from copy import deepcopy
50
from bzrlib.branch import (Branch,
55
from bzrlib.errors import (BzrCheckError,
58
WeaveRevisionNotPresent,
62
from bzrlib.inventory import InventoryEntry
63
from bzrlib.osutils import (appendpath,
77
from bzrlib.textui import show_status
79
from bzrlib.trace import mutter
83
def gen_file_id(name):
84
"""Return new file id.
86
This should probably generate proper UUIDs, but for the moment we
87
cope with just randomness because running uuidgen every time is
90
from binascii import hexlify
97
idx = name.rfind('\\')
101
# make it not a hidden file
102
name = name.lstrip('.')
104
# remove any wierd characters; we don't escape them but rather
106
name = re.sub(r'[^\w.]', '', name)
108
s = hexlify(rand_bytes(8))
109
return '-'.join((name, compact_date(time()), s))
113
"""Return a new tree-root file id."""
114
return gen_file_id('TREE_ROOT')
117
class TreeEntry(object):
118
"""An entry that implements the minium interface used by commands.
120
This needs further inspection, it may be better to have
121
InventoryEntries without ids - though that seems wrong. For now,
122
this is a parallel hierarchy to InventoryEntry, and needs to become
123
one of several things: decorates to that hierarchy, children of, or
125
Another note is that these objects are currently only used when there is
126
no InventoryEntry available - i.e. for unversioned objects.
127
Perhaps they should be UnversionedEntry et al. ? - RBC 20051003
130
def __eq__(self, other):
131
# yes, this us ugly, TODO: best practice __eq__ style.
132
return (isinstance(other, TreeEntry)
133
and other.__class__ == self.__class__)
135
def kind_character(self):
139
class TreeDirectory(TreeEntry):
140
"""See TreeEntry. This is a directory in a working tree."""
142
def __eq__(self, other):
143
return (isinstance(other, TreeDirectory)
144
and other.__class__ == self.__class__)
146
def kind_character(self):
150
class TreeFile(TreeEntry):
151
"""See TreeEntry. This is a regular file in a working tree."""
153
def __eq__(self, other):
154
return (isinstance(other, TreeFile)
155
and other.__class__ == self.__class__)
157
def kind_character(self):
161
class TreeLink(TreeEntry):
162
"""See TreeEntry. This is a symlink in a working tree."""
164
def __eq__(self, other):
165
return (isinstance(other, TreeLink)
166
and other.__class__ == self.__class__)
168
def kind_character(self):
172
class WorkingTree(bzrlib.tree.Tree):
173
"""Working copy tree.
175
The inventory is held in the `Branch` working-inventory, and the
176
files are in a directory on disk.
178
It is possible for a `WorkingTree` to have a filename which is
179
not listed in the Inventory and vice versa.
182
def __init__(self, basedir=u'.', branch=None):
183
"""Construct a WorkingTree for basedir.
185
If the branch is not supplied, it is opened automatically.
186
If the branch is supplied, it must be the branch for this basedir.
187
(branch.base is not cross checked, because for remote branches that
188
would be meaningless).
190
from bzrlib.hashcache import HashCache
191
from bzrlib.trace import note, mutter
192
assert isinstance(basedir, basestring), \
193
"base directory %r is not a string" % basedir
195
branch = Branch.open(basedir)
196
assert isinstance(branch, Branch), \
197
"branch %r is not a Branch" % branch
199
self.basedir = realpath(basedir)
201
# update the whole cache up front and write to disk if anything changed;
202
# in the future we might want to do this more selectively
203
# two possible ways offer themselves : in self._unlock, write the cache
204
# if needed, or, when the cache sees a change, append it to the hash
205
# cache file, and have the parser take the most recent entry for a
207
hc = self._hashcache = HashCache(basedir)
215
self._set_inventory(self.read_working_inventory())
217
def _set_inventory(self, inv):
218
self._inventory = inv
219
self.path2id = self._inventory.path2id
222
def open_containing(path=None):
223
"""Open an existing working tree which has its root about path.
225
This probes for a working tree at path and searches upwards from there.
227
Basically we keep looking up until we find the control directory or
228
run into /. If there isn't one, raises NotBranchError.
229
TODO: give this a new exception.
230
If there is one, it is returned, along with the unused portion of path.
236
if path.find('://') != -1:
237
raise NotBranchError(path=path)
242
return WorkingTree(path), tail
243
except NotBranchError:
246
tail = pathjoin(os.path.basename(path), tail)
248
tail = os.path.basename(path)
250
path = os.path.dirname(path)
252
# reached the root, whatever that may be
253
raise NotBranchError(path=path)
256
"""Iterate through file_ids for this tree.
258
file_ids are in a WorkingTree if they are in the working inventory
259
and the working file exists.
261
inv = self._inventory
262
for path, ie in inv.iter_entries():
263
if bzrlib.osutils.lexists(self.abspath(path)):
267
return "<%s of %s>" % (self.__class__.__name__,
268
getattr(self, 'basedir', None))
270
def abspath(self, filename):
271
return pathjoin(self.basedir, filename)
273
def relpath(self, abs):
274
"""Return the local path portion from a given absolute path."""
275
return relpath(self.basedir, abs)
277
def has_filename(self, filename):
278
return bzrlib.osutils.lexists(self.abspath(filename))
280
def get_file(self, file_id):
281
return self.get_file_byname(self.id2path(file_id))
283
def get_file_byname(self, filename):
284
return file(self.abspath(filename), 'rb')
286
def get_root_id(self):
287
"""Return the id of this trees root"""
288
inv = self.read_working_inventory()
289
return inv.root.file_id
291
def _get_store_filename(self, file_id):
292
## XXX: badly named; this is not in the store at all
293
return self.abspath(self.id2path(file_id))
296
def commit(self, *args, **kw):
297
from bzrlib.commit import Commit
298
Commit().commit(self.branch, *args, **kw)
299
self._set_inventory(self.read_working_inventory())
301
def id2abspath(self, file_id):
302
return self.abspath(self.id2path(file_id))
304
def has_id(self, file_id):
305
# files that have been deleted are excluded
306
inv = self._inventory
307
if not inv.has_id(file_id):
309
path = inv.id2path(file_id)
310
return bzrlib.osutils.lexists(self.abspath(path))
312
def has_or_had_id(self, file_id):
313
if file_id == self.inventory.root.file_id:
315
return self.inventory.has_id(file_id)
317
__contains__ = has_id
319
def get_file_size(self, file_id):
320
return os.path.getsize(self.id2abspath(file_id))
323
def get_file_sha1(self, file_id):
324
path = self._inventory.id2path(file_id)
325
return self._hashcache.get_sha1(path)
327
def is_executable(self, file_id):
329
return self._inventory[file_id].executable
331
path = self._inventory.id2path(file_id)
332
mode = os.lstat(self.abspath(path)).st_mode
333
return bool(stat.S_ISREG(mode) and stat.S_IEXEC&mode)
336
def add(self, files, ids=None):
337
"""Make files versioned.
339
Note that the command line normally calls smart_add instead,
340
which can automatically recurse.
342
This adds the files to the inventory, so that they will be
343
recorded by the next commit.
346
List of paths to add, relative to the base of the tree.
349
If set, use these instead of automatically generated ids.
350
Must be the same length as the list of files, but may
351
contain None for ids that are to be autogenerated.
353
TODO: Perhaps have an option to add the ids even if the files do
356
TODO: Perhaps callback with the ids and paths as they're added.
358
# TODO: Re-adding a file that is removed in the working copy
359
# should probably put it back with the previous ID.
360
if isinstance(files, basestring):
361
assert(ids is None or isinstance(ids, basestring))
367
ids = [None] * len(files)
369
assert(len(ids) == len(files))
371
inv = self.read_working_inventory()
372
for f,file_id in zip(files, ids):
373
if is_control_file(f):
374
raise BzrError("cannot add control file %s" % quotefn(f))
379
raise BzrError("cannot add top-level %r" % f)
381
fullpath = normpath(self.abspath(f))
384
kind = file_kind(fullpath)
386
# maybe something better?
387
raise BzrError('cannot add: not a regular file, symlink or directory: %s' % quotefn(f))
389
if not InventoryEntry.versionable_kind(kind):
390
raise BzrError('cannot add: not a versionable file ('
391
'i.e. regular file, symlink or directory): %s' % quotefn(f))
394
file_id = gen_file_id(f)
395
inv.add_path(f, kind=kind, file_id=file_id)
397
mutter("add file %s file_id:{%s} kind=%r" % (f, file_id, kind))
398
self._write_inventory(inv)
401
def add_pending_merge(self, *revision_ids):
402
# TODO: Perhaps should check at this point that the
403
# history of the revision is actually present?
404
p = self.pending_merges()
406
for rev_id in revision_ids:
412
self.set_pending_merges(p)
414
def pending_merges(self):
415
"""Return a list of pending merges.
417
These are revisions that have been merged into the working
418
directory but not yet committed.
420
cfn = self.branch.control_files._rel_controlfilename('pending-merges')
421
if not self.branch.control_files._transport.has(cfn):
424
for l in self.branch.control_files.controlfile('pending-merges', 'r').readlines():
425
p.append(l.rstrip('\n'))
429
def set_pending_merges(self, rev_list):
430
self.branch.control_files.put_utf8('pending-merges', '\n'.join(rev_list))
432
def get_symlink_target(self, file_id):
433
return os.readlink(self.id2abspath(file_id))
435
def file_class(self, filename):
436
if self.path2id(filename):
438
elif self.is_ignored(filename):
444
def list_files(self):
445
"""Recursively list all files as (path, class, kind, id).
447
Lists, but does not descend into unversioned directories.
449
This does not include files that have been deleted in this
452
Skips the control directory.
454
inv = self._inventory
456
def descend(from_dir_relpath, from_dir_id, dp):
460
## TODO: If we find a subdirectory with its own .bzr
461
## directory, then that is a separate tree and we
462
## should exclude it.
463
if bzrlib.BZRDIR == f:
467
fp = appendpath(from_dir_relpath, f)
470
fap = appendpath(dp, f)
472
f_ie = inv.get_child(from_dir_id, f)
475
elif self.is_ignored(fp):
484
raise BzrCheckError("file %r entered as kind %r id %r, "
486
% (fap, f_ie.kind, f_ie.file_id, fk))
488
# make a last minute entry
492
if fk == 'directory':
493
entry = TreeDirectory()
496
elif fk == 'symlink':
501
yield fp, c, fk, (f_ie and f_ie.file_id), entry
503
if fk != 'directory':
507
# don't descend unversioned directories
510
for ff in descend(fp, f_ie.file_id, fap):
513
for f in descend(u'', inv.root.file_id, self.basedir):
517
def move(self, from_paths, to_name):
520
to_name must exist in the inventory.
522
If to_name exists and is a directory, the files are moved into
523
it, keeping their old names.
525
Note that to_name is only the last component of the new name;
526
this doesn't change the directory.
528
This returns a list of (from_path, to_path) pairs for each
532
## TODO: Option to move IDs only
533
assert not isinstance(from_paths, basestring)
535
to_abs = self.abspath(to_name)
536
if not isdir(to_abs):
537
raise BzrError("destination %r is not a directory" % to_abs)
538
if not self.has_filename(to_name):
539
raise BzrError("destination %r not in working directory" % to_abs)
540
to_dir_id = inv.path2id(to_name)
541
if to_dir_id == None and to_name != '':
542
raise BzrError("destination %r is not a versioned directory" % to_name)
543
to_dir_ie = inv[to_dir_id]
544
if to_dir_ie.kind not in ('directory', 'root_directory'):
545
raise BzrError("destination %r is not a directory" % to_abs)
547
to_idpath = inv.get_idpath(to_dir_id)
550
if not self.has_filename(f):
551
raise BzrError("%r does not exist in working tree" % f)
552
f_id = inv.path2id(f)
554
raise BzrError("%r is not versioned" % f)
555
name_tail = splitpath(f)[-1]
556
dest_path = appendpath(to_name, name_tail)
557
if self.has_filename(dest_path):
558
raise BzrError("destination %r already exists" % dest_path)
559
if f_id in to_idpath:
560
raise BzrError("can't move %r to a subdirectory of itself" % f)
562
# OK, so there's a race here, it's possible that someone will
563
# create a file in this interval and then the rename might be
564
# left half-done. But we should have caught most problems.
565
orig_inv = deepcopy(self.inventory)
568
name_tail = splitpath(f)[-1]
569
dest_path = appendpath(to_name, name_tail)
570
result.append((f, dest_path))
571
inv.rename(inv.path2id(f), to_dir_id, name_tail)
573
rename(self.abspath(f), self.abspath(dest_path))
575
raise BzrError("failed to rename %r to %r: %s" %
576
(f, dest_path, e[1]),
577
["rename rolled back"])
579
# restore the inventory on error
580
self._set_inventory(orig_inv)
582
self._write_inventory(inv)
586
def rename_one(self, from_rel, to_rel):
589
This can change the directory or the filename or both.
592
if not self.has_filename(from_rel):
593
raise BzrError("can't rename: old working file %r does not exist" % from_rel)
594
if self.has_filename(to_rel):
595
raise BzrError("can't rename: new working file %r already exists" % to_rel)
597
file_id = inv.path2id(from_rel)
599
raise BzrError("can't rename: old name %r is not versioned" % from_rel)
602
from_parent = entry.parent_id
603
from_name = entry.name
605
if inv.path2id(to_rel):
606
raise BzrError("can't rename: new name %r is already versioned" % to_rel)
608
to_dir, to_tail = os.path.split(to_rel)
609
to_dir_id = inv.path2id(to_dir)
610
if to_dir_id == None and to_dir != '':
611
raise BzrError("can't determine destination directory id for %r" % to_dir)
613
mutter("rename_one:")
614
mutter(" file_id {%s}" % file_id)
615
mutter(" from_rel %r" % from_rel)
616
mutter(" to_rel %r" % to_rel)
617
mutter(" to_dir %r" % to_dir)
618
mutter(" to_dir_id {%s}" % to_dir_id)
620
inv.rename(file_id, to_dir_id, to_tail)
622
from_abs = self.abspath(from_rel)
623
to_abs = self.abspath(to_rel)
625
rename(from_abs, to_abs)
627
inv.rename(file_id, from_parent, from_name)
628
raise BzrError("failed to rename %r to %r: %s"
629
% (from_abs, to_abs, e[1]),
630
["rename rolled back"])
631
self._write_inventory(inv)
635
"""Return all unknown files.
637
These are files in the working directory that are not versioned or
638
control files or ignored.
640
>>> from bzrlib.branch import ScratchBranch
641
>>> b = ScratchBranch(files=['foo', 'foo~'])
642
>>> tree = WorkingTree(b.base, b)
643
>>> map(str, tree.unknowns())
646
>>> list(b.unknowns())
648
>>> tree.remove('foo')
649
>>> list(b.unknowns())
652
for subp in self.extras():
653
if not self.is_ignored(subp):
656
def iter_conflicts(self):
658
for path in (s[0] for s in self.list_files()):
659
stem = get_conflicted_stem(path)
662
if stem not in conflicted:
667
def pull(self, source, overwrite=False):
668
from bzrlib.merge import merge_inner
671
old_revision_history = self.branch.revision_history()
672
count = self.branch.pull(source, overwrite)
673
new_revision_history = self.branch.revision_history()
674
if new_revision_history != old_revision_history:
675
if len(old_revision_history):
676
other_revision = old_revision_history[-1]
678
other_revision = None
679
repository = self.branch.repository
680
merge_inner(self.branch,
681
self.branch.basis_tree(),
682
repository.revision_tree(other_revision))
688
"""Yield all unknown files in this WorkingTree.
690
If there are any unknown directories then only the directory is
691
returned, not all its children. But if there are unknown files
692
under a versioned subdirectory, they are returned.
694
Currently returned depth-first, sorted by name within directories.
696
## TODO: Work from given directory downwards
697
for path, dir_entry in self.inventory.directories():
698
mutter("search for unknowns in %r", path)
699
dirabs = self.abspath(path)
700
if not isdir(dirabs):
701
# e.g. directory deleted
705
for subf in os.listdir(dirabs):
707
and (subf not in dir_entry.children)):
712
subp = appendpath(path, subf)
716
def ignored_files(self):
717
"""Yield list of PATH, IGNORE_PATTERN"""
718
for subp in self.extras():
719
pat = self.is_ignored(subp)
724
def get_ignore_list(self):
725
"""Return list of ignore patterns.
727
Cached in the Tree object after the first call.
729
if hasattr(self, '_ignorelist'):
730
return self._ignorelist
732
l = bzrlib.DEFAULT_IGNORE[:]
733
if self.has_filename(bzrlib.IGNORE_FILENAME):
734
f = self.get_file_byname(bzrlib.IGNORE_FILENAME)
735
l.extend([line.rstrip("\n\r") for line in f.readlines()])
740
def is_ignored(self, filename):
741
r"""Check whether the filename matches an ignore pattern.
743
Patterns containing '/' or '\' need to match the whole path;
744
others match against only the last component.
746
If the file is ignored, returns the pattern which caused it to
747
be ignored, otherwise None. So this can simply be used as a
748
boolean if desired."""
750
# TODO: Use '**' to match directories, and other extended
751
# globbing stuff from cvs/rsync.
753
# XXX: fnmatch is actually not quite what we want: it's only
754
# approximately the same as real Unix fnmatch, and doesn't
755
# treat dotfiles correctly and allows * to match /.
756
# Eventually it should be replaced with something more
759
for pat in self.get_ignore_list():
760
if '/' in pat or '\\' in pat:
762
# as a special case, you can put ./ at the start of a
763
# pattern; this is good to match in the top-level
766
if (pat[:2] == './') or (pat[:2] == '.\\'):
770
if fnmatch.fnmatchcase(filename, newpat):
773
if fnmatch.fnmatchcase(splitpath(filename)[-1], pat):
778
def kind(self, file_id):
779
return file_kind(self.id2abspath(file_id))
782
"""See Branch.lock_read, and WorkingTree.unlock."""
783
return self.branch.lock_read()
785
def lock_write(self):
786
"""See Branch.lock_write, and WorkingTree.unlock."""
787
return self.branch.lock_write()
789
def _basis_inventory_name(self, revision_id):
790
return 'basis-inventory.%s' % revision_id
792
def set_last_revision(self, new_revision, old_revision=None):
793
if old_revision is not None:
795
path = self._basis_inventory_name(old_revision)
796
path = self.branch.control_files._rel_controlfilename(path)
797
self.branch.control_files._transport.delete(path)
801
xml = self.branch.repository.get_inventory_xml(new_revision)
802
path = self._basis_inventory_name(new_revision)
803
self.branch.control_files.put_utf8(path, xml)
804
except WeaveRevisionNotPresent:
807
def read_basis_inventory(self, revision_id):
808
"""Read the cached basis inventory."""
809
path = self._basis_inventory_name(revision_id)
810
return self.branch.control_files.controlfile(path, 'r').read()
813
def read_working_inventory(self):
814
"""Read the working inventory."""
815
# ElementTree does its own conversion from UTF-8, so open in
817
f = self.branch.control_files.controlfile('inventory', 'rb')
818
return bzrlib.xml5.serializer_v5.read_inventory(f)
821
def remove(self, files, verbose=False):
822
"""Remove nominated files from the working inventory..
824
This does not remove their text. This does not run on XXX on what? RBC
826
TODO: Refuse to remove modified files unless --force is given?
828
TODO: Do something useful with directories.
830
TODO: Should this remove the text or not? Tough call; not
831
removing may be useful and the user can just use use rm, and
832
is the opposite of add. Removing it is consistent with most
833
other tools. Maybe an option.
835
## TODO: Normalize names
836
## TODO: Remove nested loops; better scalability
837
if isinstance(files, basestring):
842
# do this before any modifications
846
# TODO: Perhaps make this just a warning, and continue?
847
# This tends to happen when
848
raise NotVersionedError(path=f)
849
mutter("remove inventory entry %s {%s}", quotefn(f), fid)
851
# having remove it, it must be either ignored or unknown
852
if self.is_ignored(f):
856
show_status(new_status, inv[fid].kind, quotefn(f))
859
self._write_inventory(inv)
862
def revert(self, filenames, old_tree=None, backups=True):
863
from bzrlib.merge import merge_inner
865
old_tree = self.branch.basis_tree()
866
merge_inner(self.branch, old_tree,
867
self, ignore_zero=True,
868
backup_files=backups,
869
interesting_files=filenames)
870
if not len(filenames):
871
self.set_pending_merges([])
874
def set_inventory(self, new_inventory_list):
875
from bzrlib.inventory import (Inventory,
880
inv = Inventory(self.get_root_id())
881
for path, file_id, parent, kind in new_inventory_list:
882
name = os.path.basename(path)
885
# fixme, there should be a factory function inv,add_??
886
if kind == 'directory':
887
inv.add(InventoryDirectory(file_id, name, parent))
889
inv.add(InventoryFile(file_id, name, parent))
890
elif kind == 'symlink':
891
inv.add(InventoryLink(file_id, name, parent))
893
raise BzrError("unknown kind %r" % kind)
894
self._write_inventory(inv)
897
def set_root_id(self, file_id):
898
"""Set the root id for this tree."""
899
inv = self.read_working_inventory()
900
orig_root_id = inv.root.file_id
901
del inv._byid[inv.root.file_id]
902
inv.root.file_id = file_id
903
inv._byid[inv.root.file_id] = inv.root
906
if entry.parent_id in (None, orig_root_id):
907
entry.parent_id = inv.root.file_id
908
self._write_inventory(inv)
911
"""See Branch.unlock.
913
WorkingTree locking just uses the Branch locking facilities.
914
This is current because all working trees have an embedded branch
915
within them. IF in the future, we were to make branch data shareable
916
between multiple working trees, i.e. via shared storage, then we
917
would probably want to lock both the local tree, and the branch.
919
if self._hashcache.needs_write:
920
self._hashcache.write()
921
return self.branch.unlock()
924
def _write_inventory(self, inv):
925
"""Write inventory as the current inventory."""
926
from cStringIO import StringIO
927
from bzrlib.atomicfile import AtomicFile
929
bzrlib.xml5.serializer_v5.write_inventory(inv, sio)
931
f = AtomicFile(self.branch.control_files.controlfilename('inventory'))
937
self._set_inventory(inv)
938
mutter('wrote working inventory')
941
CONFLICT_SUFFIXES = ('.THIS', '.BASE', '.OTHER')
942
def get_conflicted_stem(path):
943
for suffix in CONFLICT_SUFFIXES:
944
if path.endswith(suffix):
945
return path[:-len(suffix)]
added
removed
bzrlib/workingtree.py
