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,
61
from bzrlib.inventory import InventoryEntry
62
from bzrlib.osutils import (appendpath,
76
from bzrlib.textui import show_status
78
from bzrlib.trace import mutter
82
def gen_file_id(name):
83
"""Return new file id.
85
This should probably generate proper UUIDs, but for the moment we
86
cope with just randomness because running uuidgen every time is
89
from binascii import hexlify
96
idx = name.rfind('\\')
100
# make it not a hidden file
101
name = name.lstrip('.')
103
# remove any wierd characters; we don't escape them but rather
105
name = re.sub(r'[^\w.]', '', name)
107
s = hexlify(rand_bytes(8))
108
return '-'.join((name, compact_date(time()), s))
112
"""Return a new tree-root file id."""
113
return gen_file_id('TREE_ROOT')
116
class TreeEntry(object):
117
"""An entry that implements the minium interface used by commands.
119
This needs further inspection, it may be better to have
120
InventoryEntries without ids - though that seems wrong. For now,
121
this is a parallel hierarchy to InventoryEntry, and needs to become
122
one of several things: decorates to that hierarchy, children of, or
124
Another note is that these objects are currently only used when there is
125
no InventoryEntry available - i.e. for unversioned objects.
126
Perhaps they should be UnversionedEntry et al. ? - RBC 20051003
129
def __eq__(self, other):
130
# yes, this us ugly, TODO: best practice __eq__ style.
131
return (isinstance(other, TreeEntry)
132
and other.__class__ == self.__class__)
134
def kind_character(self):
138
class TreeDirectory(TreeEntry):
139
"""See TreeEntry. This is a directory in a working tree."""
141
def __eq__(self, other):
142
return (isinstance(other, TreeDirectory)
143
and other.__class__ == self.__class__)
145
def kind_character(self):
149
class TreeFile(TreeEntry):
150
"""See TreeEntry. This is a regular file in a working tree."""
152
def __eq__(self, other):
153
return (isinstance(other, TreeFile)
154
and other.__class__ == self.__class__)
156
def kind_character(self):
160
class TreeLink(TreeEntry):
161
"""See TreeEntry. This is a symlink in a working tree."""
163
def __eq__(self, other):
164
return (isinstance(other, TreeLink)
165
and other.__class__ == self.__class__)
167
def kind_character(self):
171
class WorkingTree(bzrlib.tree.Tree):
172
"""Working copy tree.
174
The inventory is held in the `Branch` working-inventory, and the
175
files are in a directory on disk.
177
It is possible for a `WorkingTree` to have a filename which is
178
not listed in the Inventory and vice versa.
181
def __init__(self, basedir=u'.', branch=None):
182
"""Construct a WorkingTree for basedir.
184
If the branch is not supplied, it is opened automatically.
185
If the branch is supplied, it must be the branch for this basedir.
186
(branch.base is not cross checked, because for remote branches that
187
would be meaningless).
189
from bzrlib.hashcache import HashCache
190
from bzrlib.trace import note, mutter
191
assert isinstance(basedir, basestring), \
192
"base directory %r is not a string" % basedir
194
branch = Branch.open(basedir)
195
assert isinstance(branch, Branch), \
196
"branch %r is not a Branch" % branch
198
self.basedir = realpath(basedir)
200
self._set_inventory(self.read_working_inventory())
202
# update the whole cache up front and write to disk if anything changed;
203
# in the future we might want to do this more selectively
204
# two possible ways offer themselves : in self._unlock, write the cache
205
# if needed, or, when the cache sees a change, append it to the hash
206
# cache file, and have the parser take the most recent entry for a
208
hc = self._hashcache = HashCache(basedir)
216
def _set_inventory(self, inv):
217
self._inventory = inv
218
self.path2id = self._inventory.path2id
221
def open_containing(path=None):
222
"""Open an existing working tree which has its root about path.
224
This probes for a working tree at path and searches upwards from there.
226
Basically we keep looking up until we find the control directory or
227
run into /. If there isn't one, raises NotBranchError.
228
TODO: give this a new exception.
229
If there is one, it is returned, along with the unused portion of path.
235
if path.find('://') != -1:
236
raise NotBranchError(path=path)
241
return WorkingTree(path), tail
242
except NotBranchError:
245
tail = pathjoin(os.path.basename(path), tail)
247
tail = os.path.basename(path)
249
path = os.path.dirname(path)
251
# reached the root, whatever that may be
252
raise NotBranchError(path=path)
255
"""Iterate through file_ids for this tree.
257
file_ids are in a WorkingTree if they are in the working inventory
258
and the working file exists.
260
inv = self._inventory
261
for path, ie in inv.iter_entries():
262
if bzrlib.osutils.lexists(self.abspath(path)):
266
return "<%s of %s>" % (self.__class__.__name__,
267
getattr(self, 'basedir', None))
269
def abspath(self, filename):
270
return pathjoin(self.basedir, filename)
272
def relpath(self, abs):
273
"""Return the local path portion from a given absolute path."""
274
return relpath(self.basedir, abs)
276
def has_filename(self, filename):
277
return bzrlib.osutils.lexists(self.abspath(filename))
279
def get_file(self, file_id):
280
return self.get_file_byname(self.id2path(file_id))
282
def get_file_byname(self, filename):
283
return file(self.abspath(filename), 'rb')
285
def get_root_id(self):
286
"""Return the id of this trees root"""
287
inv = self.read_working_inventory()
288
return inv.root.file_id
290
def _get_store_filename(self, file_id):
291
## XXX: badly named; this is not in the store at all
292
return self.abspath(self.id2path(file_id))
295
def commit(self, *args, **kw):
296
from bzrlib.commit import Commit
297
Commit().commit(self.branch, *args, **kw)
298
self._set_inventory(self.read_working_inventory())
300
def id2abspath(self, file_id):
301
return self.abspath(self.id2path(file_id))
303
def has_id(self, file_id):
304
# files that have been deleted are excluded
305
inv = self._inventory
306
if not inv.has_id(file_id):
308
path = inv.id2path(file_id)
309
return bzrlib.osutils.lexists(self.abspath(path))
311
def has_or_had_id(self, file_id):
312
if file_id == self.inventory.root.file_id:
314
return self.inventory.has_id(file_id)
316
__contains__ = has_id
318
def get_file_size(self, file_id):
319
return os.path.getsize(self.id2abspath(file_id))
321
def get_file_sha1(self, file_id):
322
path = self._inventory.id2path(file_id)
323
return self._hashcache.get_sha1(path)
325
def is_executable(self, file_id):
327
return self._inventory[file_id].executable
329
path = self._inventory.id2path(file_id)
330
mode = os.lstat(self.abspath(path)).st_mode
331
return bool(stat.S_ISREG(mode) and stat.S_IEXEC&mode)
334
def add(self, files, ids=None):
335
"""Make files versioned.
337
Note that the command line normally calls smart_add instead,
338
which can automatically recurse.
340
This adds the files to the inventory, so that they will be
341
recorded by the next commit.
344
List of paths to add, relative to the base of the tree.
347
If set, use these instead of automatically generated ids.
348
Must be the same length as the list of files, but may
349
contain None for ids that are to be autogenerated.
351
TODO: Perhaps have an option to add the ids even if the files do
354
TODO: Perhaps callback with the ids and paths as they're added.
356
# TODO: Re-adding a file that is removed in the working copy
357
# should probably put it back with the previous ID.
358
if isinstance(files, basestring):
359
assert(ids is None or isinstance(ids, basestring))
365
ids = [None] * len(files)
367
assert(len(ids) == len(files))
369
inv = self.read_working_inventory()
370
for f,file_id in zip(files, ids):
371
if is_control_file(f):
372
raise BzrError("cannot add control file %s" % quotefn(f))
377
raise BzrError("cannot add top-level %r" % f)
379
fullpath = normpath(self.abspath(f))
382
kind = file_kind(fullpath)
384
# maybe something better?
385
raise BzrError('cannot add: not a regular file, symlink or directory: %s' % quotefn(f))
387
if not InventoryEntry.versionable_kind(kind):
388
raise BzrError('cannot add: not a versionable file ('
389
'i.e. regular file, symlink or directory): %s' % quotefn(f))
392
file_id = gen_file_id(f)
393
inv.add_path(f, kind=kind, file_id=file_id)
395
mutter("add file %s file_id:{%s} kind=%r" % (f, file_id, kind))
396
self._write_inventory(inv)
399
def add_pending_merge(self, *revision_ids):
400
# TODO: Perhaps should check at this point that the
401
# history of the revision is actually present?
402
p = self.pending_merges()
404
for rev_id in revision_ids:
410
self.set_pending_merges(p)
412
def pending_merges(self):
413
"""Return a list of pending merges.
415
These are revisions that have been merged into the working
416
directory but not yet committed.
418
cfn = self.branch._rel_controlfilename('pending-merges')
419
if not self.branch._transport.has(cfn):
422
for l in self.branch.controlfile('pending-merges', 'r').readlines():
423
p.append(l.rstrip('\n'))
427
def set_pending_merges(self, rev_list):
428
self.branch.put_controlfile('pending-merges', '\n'.join(rev_list))
430
def get_symlink_target(self, file_id):
431
return os.readlink(self.id2abspath(file_id))
433
def file_class(self, filename):
434
if self.path2id(filename):
436
elif self.is_ignored(filename):
442
def list_files(self):
443
"""Recursively list all files as (path, class, kind, id).
445
Lists, but does not descend into unversioned directories.
447
This does not include files that have been deleted in this
450
Skips the control directory.
452
inv = self._inventory
454
def descend(from_dir_relpath, from_dir_id, dp):
458
## TODO: If we find a subdirectory with its own .bzr
459
## directory, then that is a separate tree and we
460
## should exclude it.
461
if bzrlib.BZRDIR == f:
465
fp = appendpath(from_dir_relpath, f)
468
fap = appendpath(dp, f)
470
f_ie = inv.get_child(from_dir_id, f)
473
elif self.is_ignored(fp):
482
raise BzrCheckError("file %r entered as kind %r id %r, "
484
% (fap, f_ie.kind, f_ie.file_id, fk))
486
# make a last minute entry
490
if fk == 'directory':
491
entry = TreeDirectory()
494
elif fk == 'symlink':
499
yield fp, c, fk, (f_ie and f_ie.file_id), entry
501
if fk != 'directory':
505
# don't descend unversioned directories
508
for ff in descend(fp, f_ie.file_id, fap):
511
for f in descend(u'', inv.root.file_id, self.basedir):
515
def move(self, from_paths, to_name):
518
to_name must exist in the inventory.
520
If to_name exists and is a directory, the files are moved into
521
it, keeping their old names.
523
Note that to_name is only the last component of the new name;
524
this doesn't change the directory.
526
This returns a list of (from_path, to_path) pairs for each
530
## TODO: Option to move IDs only
531
assert not isinstance(from_paths, basestring)
533
to_abs = self.abspath(to_name)
534
if not isdir(to_abs):
535
raise BzrError("destination %r is not a directory" % to_abs)
536
if not self.has_filename(to_name):
537
raise BzrError("destination %r not in working directory" % to_abs)
538
to_dir_id = inv.path2id(to_name)
539
if to_dir_id == None and to_name != '':
540
raise BzrError("destination %r is not a versioned directory" % to_name)
541
to_dir_ie = inv[to_dir_id]
542
if to_dir_ie.kind not in ('directory', 'root_directory'):
543
raise BzrError("destination %r is not a directory" % to_abs)
545
to_idpath = inv.get_idpath(to_dir_id)
548
if not self.has_filename(f):
549
raise BzrError("%r does not exist in working tree" % f)
550
f_id = inv.path2id(f)
552
raise BzrError("%r is not versioned" % f)
553
name_tail = splitpath(f)[-1]
554
dest_path = appendpath(to_name, name_tail)
555
if self.has_filename(dest_path):
556
raise BzrError("destination %r already exists" % dest_path)
557
if f_id in to_idpath:
558
raise BzrError("can't move %r to a subdirectory of itself" % f)
560
# OK, so there's a race here, it's possible that someone will
561
# create a file in this interval and then the rename might be
562
# left half-done. But we should have caught most problems.
563
orig_inv = deepcopy(self.inventory)
566
name_tail = splitpath(f)[-1]
567
dest_path = appendpath(to_name, name_tail)
568
result.append((f, dest_path))
569
inv.rename(inv.path2id(f), to_dir_id, name_tail)
571
rename(self.abspath(f), self.abspath(dest_path))
573
raise BzrError("failed to rename %r to %r: %s" %
574
(f, dest_path, e[1]),
575
["rename rolled back"])
577
# restore the inventory on error
578
self._set_inventory(orig_inv)
580
self._write_inventory(inv)
584
def rename_one(self, from_rel, to_rel):
587
This can change the directory or the filename or both.
590
if not self.has_filename(from_rel):
591
raise BzrError("can't rename: old working file %r does not exist" % from_rel)
592
if self.has_filename(to_rel):
593
raise BzrError("can't rename: new working file %r already exists" % to_rel)
595
file_id = inv.path2id(from_rel)
597
raise BzrError("can't rename: old name %r is not versioned" % from_rel)
600
from_parent = entry.parent_id
601
from_name = entry.name
603
if inv.path2id(to_rel):
604
raise BzrError("can't rename: new name %r is already versioned" % to_rel)
606
to_dir, to_tail = os.path.split(to_rel)
607
to_dir_id = inv.path2id(to_dir)
608
if to_dir_id == None and to_dir != '':
609
raise BzrError("can't determine destination directory id for %r" % to_dir)
611
mutter("rename_one:")
612
mutter(" file_id {%s}" % file_id)
613
mutter(" from_rel %r" % from_rel)
614
mutter(" to_rel %r" % to_rel)
615
mutter(" to_dir %r" % to_dir)
616
mutter(" to_dir_id {%s}" % to_dir_id)
618
inv.rename(file_id, to_dir_id, to_tail)
620
from_abs = self.abspath(from_rel)
621
to_abs = self.abspath(to_rel)
623
rename(from_abs, to_abs)
625
inv.rename(file_id, from_parent, from_name)
626
raise BzrError("failed to rename %r to %r: %s"
627
% (from_abs, to_abs, e[1]),
628
["rename rolled back"])
629
self._write_inventory(inv)
633
"""Return all unknown files.
635
These are files in the working directory that are not versioned or
636
control files or ignored.
638
>>> from bzrlib.branch import ScratchBranch
639
>>> b = ScratchBranch(files=['foo', 'foo~'])
640
>>> tree = WorkingTree(b.base, b)
641
>>> map(str, tree.unknowns())
644
>>> list(b.unknowns())
646
>>> tree.remove('foo')
647
>>> list(b.unknowns())
650
for subp in self.extras():
651
if not self.is_ignored(subp):
654
def iter_conflicts(self):
656
for path in (s[0] for s in self.list_files()):
657
stem = get_conflicted_stem(path)
660
if stem not in conflicted:
665
def pull(self, source, overwrite=False):
666
from bzrlib.merge import merge_inner
669
old_revision_history = self.branch.revision_history()
670
count = self.branch.pull(source, overwrite)
671
new_revision_history = self.branch.revision_history()
672
if new_revision_history != old_revision_history:
673
if len(old_revision_history):
674
other_revision = old_revision_history[-1]
676
other_revision = None
677
merge_inner(self.branch,
678
self.branch.basis_tree(),
679
self.branch.revision_tree(other_revision))
685
"""Yield all unknown files in this WorkingTree.
687
If there are any unknown directories then only the directory is
688
returned, not all its children. But if there are unknown files
689
under a versioned subdirectory, they are returned.
691
Currently returned depth-first, sorted by name within directories.
693
## TODO: Work from given directory downwards
694
for path, dir_entry in self.inventory.directories():
695
mutter("search for unknowns in %r", path)
696
dirabs = self.abspath(path)
697
if not isdir(dirabs):
698
# e.g. directory deleted
702
for subf in os.listdir(dirabs):
704
and (subf not in dir_entry.children)):
709
subp = appendpath(path, subf)
713
def ignored_files(self):
714
"""Yield list of PATH, IGNORE_PATTERN"""
715
for subp in self.extras():
716
pat = self.is_ignored(subp)
721
def get_ignore_list(self):
722
"""Return list of ignore patterns.
724
Cached in the Tree object after the first call.
726
if hasattr(self, '_ignorelist'):
727
return self._ignorelist
729
l = bzrlib.DEFAULT_IGNORE[:]
730
if self.has_filename(bzrlib.IGNORE_FILENAME):
731
f = self.get_file_byname(bzrlib.IGNORE_FILENAME)
732
l.extend([line.rstrip("\n\r") for line in f.readlines()])
737
def is_ignored(self, filename):
738
r"""Check whether the filename matches an ignore pattern.
740
Patterns containing '/' or '\' need to match the whole path;
741
others match against only the last component.
743
If the file is ignored, returns the pattern which caused it to
744
be ignored, otherwise None. So this can simply be used as a
745
boolean if desired."""
747
# TODO: Use '**' to match directories, and other extended
748
# globbing stuff from cvs/rsync.
750
# XXX: fnmatch is actually not quite what we want: it's only
751
# approximately the same as real Unix fnmatch, and doesn't
752
# treat dotfiles correctly and allows * to match /.
753
# Eventually it should be replaced with something more
756
for pat in self.get_ignore_list():
757
if '/' in pat or '\\' in pat:
759
# as a special case, you can put ./ at the start of a
760
# pattern; this is good to match in the top-level
763
if (pat[:2] == './') or (pat[:2] == '.\\'):
767
if fnmatch.fnmatchcase(filename, newpat):
770
if fnmatch.fnmatchcase(splitpath(filename)[-1], pat):
775
def kind(self, file_id):
776
return file_kind(self.id2abspath(file_id))
779
"""See Branch.lock_read, and WorkingTree.unlock."""
780
return self.branch.lock_read()
782
def lock_write(self):
783
"""See Branch.lock_write, and WorkingTree.unlock."""
784
return self.branch.lock_write()
786
def _basis_inventory_name(self, revision_id):
787
return 'basis-inventory.%s' % revision_id
789
def set_last_revision(self, new_revision, old_revision=None):
792
path = self._basis_inventory_name(old_revision)
793
path = self.branch._rel_controlfilename(path)
794
self.branch._transport.delete(path)
798
xml = self.branch.get_inventory_xml(new_revision)
799
path = self._basis_inventory_name(new_revision)
800
self.branch.put_controlfile(path, xml)
801
except WeaveRevisionNotPresent:
804
def read_basis_inventory(self, revision_id):
805
"""Read the cached basis inventory."""
806
path = self._basis_inventory_name(revision_id)
807
return self.branch.controlfile(path, 'r').read()
810
def read_working_inventory(self):
811
"""Read the working inventory."""
812
# ElementTree does its own conversion from UTF-8, so open in
814
f = self.branch.controlfile('inventory', 'rb')
815
return bzrlib.xml5.serializer_v5.read_inventory(f)
818
def remove(self, files, verbose=False):
819
"""Remove nominated files from the working inventory..
821
This does not remove their text. This does not run on XXX on what? RBC
823
TODO: Refuse to remove modified files unless --force is given?
825
TODO: Do something useful with directories.
827
TODO: Should this remove the text or not? Tough call; not
828
removing may be useful and the user can just use use rm, and
829
is the opposite of add. Removing it is consistent with most
830
other tools. Maybe an option.
832
## TODO: Normalize names
833
## TODO: Remove nested loops; better scalability
834
if isinstance(files, basestring):
839
# do this before any modifications
843
# TODO: Perhaps make this just a warning, and continue?
844
# This tends to happen when
845
raise NotVersionedError(path=f)
846
mutter("remove inventory entry %s {%s}", quotefn(f), fid)
848
# having remove it, it must be either ignored or unknown
849
if self.is_ignored(f):
853
show_status(new_status, inv[fid].kind, quotefn(f))
856
self._write_inventory(inv)
859
def revert(self, filenames, old_tree=None, backups=True):
860
from bzrlib.merge import merge_inner
862
old_tree = self.branch.basis_tree()
863
merge_inner(self.branch, old_tree,
864
self, ignore_zero=True,
865
backup_files=backups,
866
interesting_files=filenames)
867
if not len(filenames):
868
self.set_pending_merges([])
871
def set_inventory(self, new_inventory_list):
872
from bzrlib.inventory import (Inventory,
877
inv = Inventory(self.get_root_id())
878
for path, file_id, parent, kind in new_inventory_list:
879
name = os.path.basename(path)
882
# fixme, there should be a factory function inv,add_??
883
if kind == 'directory':
884
inv.add(InventoryDirectory(file_id, name, parent))
886
inv.add(InventoryFile(file_id, name, parent))
887
elif kind == 'symlink':
888
inv.add(InventoryLink(file_id, name, parent))
890
raise BzrError("unknown kind %r" % kind)
891
self._write_inventory(inv)
894
def set_root_id(self, file_id):
895
"""Set the root id for this tree."""
896
inv = self.read_working_inventory()
897
orig_root_id = inv.root.file_id
898
del inv._byid[inv.root.file_id]
899
inv.root.file_id = file_id
900
inv._byid[inv.root.file_id] = inv.root
903
if entry.parent_id in (None, orig_root_id):
904
entry.parent_id = inv.root.file_id
905
self._write_inventory(inv)
908
"""See Branch.unlock.
910
WorkingTree locking just uses the Branch locking facilities.
911
This is current because all working trees have an embedded branch
912
within them. IF in the future, we were to make branch data shareable
913
between multiple working trees, i.e. via shared storage, then we
914
would probably want to lock both the local tree, and the branch.
916
return self.branch.unlock()
919
def _write_inventory(self, inv):
920
"""Write inventory as the current inventory."""
921
from cStringIO import StringIO
922
from bzrlib.atomicfile import AtomicFile
924
bzrlib.xml5.serializer_v5.write_inventory(inv, sio)
926
f = AtomicFile(self.branch.controlfilename('inventory'))
932
self._set_inventory(inv)
933
mutter('wrote working inventory')
936
CONFLICT_SUFFIXES = ('.THIS', '.BASE', '.OTHER')
937
def get_conflicted_stem(path):
938
for suffix in CONFLICT_SUFFIXES:
939
if path.endswith(suffix):
940
return path[:-len(suffix)]
added
removed
bzrlib/workingtree.py
