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,
53
from bzrlib.errors import (BzrCheckError,
56
WeaveRevisionNotPresent,
60
from bzrlib.inventory import InventoryEntry
61
from bzrlib.osutils import (appendpath,
75
from bzrlib.textui import show_status
77
from bzrlib.trace import mutter
79
from bzrlib.decorators import needs_read_lock, needs_write_lock
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
# update the whole cache up front and write to disk if anything changed;
201
# in the future we might want to do this more selectively
202
# two possible ways offer themselves : in self._unlock, write the cache
203
# if needed, or, when the cache sees a change, append it to the hash
204
# cache file, and have the parser take the most recent entry for a
206
hc = self._hashcache = HashCache(basedir)
214
self._set_inventory(self.read_working_inventory())
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))
322
def get_file_sha1(self, file_id):
323
path = self._inventory.id2path(file_id)
324
return self._hashcache.get_sha1(path)
326
def is_executable(self, file_id):
328
return self._inventory[file_id].executable
330
path = self._inventory.id2path(file_id)
331
mode = os.lstat(self.abspath(path)).st_mode
332
return bool(stat.S_ISREG(mode) and stat.S_IEXEC&mode)
335
def add(self, files, ids=None):
336
"""Make files versioned.
338
Note that the command line normally calls smart_add instead,
339
which can automatically recurse.
341
This adds the files to the inventory, so that they will be
342
recorded by the next commit.
345
List of paths to add, relative to the base of the tree.
348
If set, use these instead of automatically generated ids.
349
Must be the same length as the list of files, but may
350
contain None for ids that are to be autogenerated.
352
TODO: Perhaps have an option to add the ids even if the files do
355
TODO: Perhaps callback with the ids and paths as they're added.
357
# TODO: Re-adding a file that is removed in the working copy
358
# should probably put it back with the previous ID.
359
if isinstance(files, basestring):
360
assert(ids is None or isinstance(ids, basestring))
366
ids = [None] * len(files)
368
assert(len(ids) == len(files))
370
inv = self.read_working_inventory()
371
for f,file_id in zip(files, ids):
372
if is_control_file(f):
373
raise BzrError("cannot add control file %s" % quotefn(f))
378
raise BzrError("cannot add top-level %r" % f)
380
fullpath = normpath(self.abspath(f))
383
kind = file_kind(fullpath)
385
# maybe something better?
386
raise BzrError('cannot add: not a regular file, symlink or directory: %s' % quotefn(f))
388
if not InventoryEntry.versionable_kind(kind):
389
raise BzrError('cannot add: not a versionable file ('
390
'i.e. regular file, symlink or directory): %s' % quotefn(f))
393
file_id = gen_file_id(f)
394
inv.add_path(f, kind=kind, file_id=file_id)
396
mutter("add file %s file_id:{%s} kind=%r" % (f, file_id, kind))
397
self._write_inventory(inv)
400
def add_pending_merge(self, *revision_ids):
401
# TODO: Perhaps should check at this point that the
402
# history of the revision is actually present?
403
p = self.pending_merges()
405
for rev_id in revision_ids:
411
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.
421
f = self.branch.control_files.controlfile('pending-merges', 'r')
425
for l in f.readlines():
426
p.append(l.rstrip('\n'))
430
def set_pending_merges(self, rev_list):
431
self.branch.control_files.put_utf8('pending-merges', '\n'.join(rev_list))
433
def get_symlink_target(self, file_id):
434
return os.readlink(self.id2abspath(file_id))
436
def file_class(self, filename):
437
if self.path2id(filename):
439
elif self.is_ignored(filename):
445
def list_files(self):
446
"""Recursively list all files as (path, class, kind, id).
448
Lists, but does not descend into unversioned directories.
450
This does not include files that have been deleted in this
453
Skips the control directory.
455
inv = self._inventory
457
def descend(from_dir_relpath, from_dir_id, dp):
461
## TODO: If we find a subdirectory with its own .bzr
462
## directory, then that is a separate tree and we
463
## should exclude it.
464
if bzrlib.BZRDIR == f:
468
fp = appendpath(from_dir_relpath, f)
471
fap = appendpath(dp, f)
473
f_ie = inv.get_child(from_dir_id, f)
476
elif self.is_ignored(fp):
485
raise BzrCheckError("file %r entered as kind %r id %r, "
487
% (fap, f_ie.kind, f_ie.file_id, fk))
489
# make a last minute entry
493
if fk == 'directory':
494
entry = TreeDirectory()
497
elif fk == 'symlink':
502
yield fp, c, fk, (f_ie and f_ie.file_id), entry
504
if fk != 'directory':
508
# don't descend unversioned directories
511
for ff in descend(fp, f_ie.file_id, fap):
514
for f in descend(u'', inv.root.file_id, self.basedir):
518
def move(self, from_paths, to_name):
521
to_name must exist in the inventory.
523
If to_name exists and is a directory, the files are moved into
524
it, keeping their old names.
526
Note that to_name is only the last component of the new name;
527
this doesn't change the directory.
529
This returns a list of (from_path, to_path) pairs for each
533
## TODO: Option to move IDs only
534
assert not isinstance(from_paths, basestring)
536
to_abs = self.abspath(to_name)
537
if not isdir(to_abs):
538
raise BzrError("destination %r is not a directory" % to_abs)
539
if not self.has_filename(to_name):
540
raise BzrError("destination %r not in working directory" % to_abs)
541
to_dir_id = inv.path2id(to_name)
542
if to_dir_id == None and to_name != '':
543
raise BzrError("destination %r is not a versioned directory" % to_name)
544
to_dir_ie = inv[to_dir_id]
545
if to_dir_ie.kind not in ('directory', 'root_directory'):
546
raise BzrError("destination %r is not a directory" % to_abs)
548
to_idpath = inv.get_idpath(to_dir_id)
551
if not self.has_filename(f):
552
raise BzrError("%r does not exist in working tree" % f)
553
f_id = inv.path2id(f)
555
raise BzrError("%r is not versioned" % f)
556
name_tail = splitpath(f)[-1]
557
dest_path = appendpath(to_name, name_tail)
558
if self.has_filename(dest_path):
559
raise BzrError("destination %r already exists" % dest_path)
560
if f_id in to_idpath:
561
raise BzrError("can't move %r to a subdirectory of itself" % f)
563
# OK, so there's a race here, it's possible that someone will
564
# create a file in this interval and then the rename might be
565
# left half-done. But we should have caught most problems.
566
orig_inv = deepcopy(self.inventory)
569
name_tail = splitpath(f)[-1]
570
dest_path = appendpath(to_name, name_tail)
571
result.append((f, dest_path))
572
inv.rename(inv.path2id(f), to_dir_id, name_tail)
574
rename(self.abspath(f), self.abspath(dest_path))
576
raise BzrError("failed to rename %r to %r: %s" %
577
(f, dest_path, e[1]),
578
["rename rolled back"])
580
# restore the inventory on error
581
self._set_inventory(orig_inv)
583
self._write_inventory(inv)
587
def rename_one(self, from_rel, to_rel):
590
This can change the directory or the filename or both.
593
if not self.has_filename(from_rel):
594
raise BzrError("can't rename: old working file %r does not exist" % from_rel)
595
if self.has_filename(to_rel):
596
raise BzrError("can't rename: new working file %r already exists" % to_rel)
598
file_id = inv.path2id(from_rel)
600
raise BzrError("can't rename: old name %r is not versioned" % from_rel)
603
from_parent = entry.parent_id
604
from_name = entry.name
606
if inv.path2id(to_rel):
607
raise BzrError("can't rename: new name %r is already versioned" % to_rel)
609
to_dir, to_tail = os.path.split(to_rel)
610
to_dir_id = inv.path2id(to_dir)
611
if to_dir_id == None and to_dir != '':
612
raise BzrError("can't determine destination directory id for %r" % to_dir)
614
mutter("rename_one:")
615
mutter(" file_id {%s}" % file_id)
616
mutter(" from_rel %r" % from_rel)
617
mutter(" to_rel %r" % to_rel)
618
mutter(" to_dir %r" % to_dir)
619
mutter(" to_dir_id {%s}" % to_dir_id)
621
inv.rename(file_id, to_dir_id, to_tail)
623
from_abs = self.abspath(from_rel)
624
to_abs = self.abspath(to_rel)
626
rename(from_abs, to_abs)
628
inv.rename(file_id, from_parent, from_name)
629
raise BzrError("failed to rename %r to %r: %s"
630
% (from_abs, to_abs, e[1]),
631
["rename rolled back"])
632
self._write_inventory(inv)
636
"""Return all unknown files.
638
These are files in the working directory that are not versioned or
639
control files or ignored.
641
>>> from bzrlib.branch import ScratchBranch
642
>>> b = ScratchBranch(files=['foo', 'foo~'])
643
>>> tree = WorkingTree(b.base, b)
644
>>> map(str, tree.unknowns())
647
>>> list(b.unknowns())
649
>>> tree.remove('foo')
650
>>> list(b.unknowns())
653
for subp in self.extras():
654
if not self.is_ignored(subp):
657
def iter_conflicts(self):
659
for path in (s[0] for s in self.list_files()):
660
stem = get_conflicted_stem(path)
663
if stem not in conflicted:
668
def pull(self, source, overwrite=False):
669
from bzrlib.merge import merge_inner
672
old_revision_history = self.branch.revision_history()
673
count = self.branch.pull(source, overwrite)
674
new_revision_history = self.branch.revision_history()
675
if new_revision_history != old_revision_history:
676
if len(old_revision_history):
677
other_revision = old_revision_history[-1]
679
other_revision = None
680
repository = self.branch.repository
681
merge_inner(self.branch,
682
self.branch.basis_tree(),
683
repository.revision_tree(other_revision))
689
"""Yield all unknown files in this WorkingTree.
691
If there are any unknown directories then only the directory is
692
returned, not all its children. But if there are unknown files
693
under a versioned subdirectory, they are returned.
695
Currently returned depth-first, sorted by name within directories.
697
## TODO: Work from given directory downwards
698
for path, dir_entry in self.inventory.directories():
699
mutter("search for unknowns in %r", path)
700
dirabs = self.abspath(path)
701
if not isdir(dirabs):
702
# e.g. directory deleted
706
for subf in os.listdir(dirabs):
708
and (subf not in dir_entry.children)):
713
subp = appendpath(path, subf)
717
def ignored_files(self):
718
"""Yield list of PATH, IGNORE_PATTERN"""
719
for subp in self.extras():
720
pat = self.is_ignored(subp)
725
def get_ignore_list(self):
726
"""Return list of ignore patterns.
728
Cached in the Tree object after the first call.
730
if hasattr(self, '_ignorelist'):
731
return self._ignorelist
733
l = bzrlib.DEFAULT_IGNORE[:]
734
if self.has_filename(bzrlib.IGNORE_FILENAME):
735
f = self.get_file_byname(bzrlib.IGNORE_FILENAME)
736
l.extend([line.rstrip("\n\r") for line in f.readlines()])
741
def is_ignored(self, filename):
742
r"""Check whether the filename matches an ignore pattern.
744
Patterns containing '/' or '\' need to match the whole path;
745
others match against only the last component.
747
If the file is ignored, returns the pattern which caused it to
748
be ignored, otherwise None. So this can simply be used as a
749
boolean if desired."""
751
# TODO: Use '**' to match directories, and other extended
752
# globbing stuff from cvs/rsync.
754
# XXX: fnmatch is actually not quite what we want: it's only
755
# approximately the same as real Unix fnmatch, and doesn't
756
# treat dotfiles correctly and allows * to match /.
757
# Eventually it should be replaced with something more
760
for pat in self.get_ignore_list():
761
if '/' in pat or '\\' in pat:
763
# as a special case, you can put ./ at the start of a
764
# pattern; this is good to match in the top-level
767
if (pat[:2] == './') or (pat[:2] == '.\\'):
771
if fnmatch.fnmatchcase(filename, newpat):
774
if fnmatch.fnmatchcase(splitpath(filename)[-1], pat):
779
def kind(self, file_id):
780
return file_kind(self.id2abspath(file_id))
783
"""See Branch.lock_read, and WorkingTree.unlock."""
784
return self.branch.lock_read()
786
def lock_write(self):
787
"""See Branch.lock_write, and WorkingTree.unlock."""
788
return self.branch.lock_write()
790
def _basis_inventory_name(self, revision_id):
791
return 'basis-inventory.%s' % revision_id
793
def set_last_revision(self, new_revision, old_revision=None):
794
if old_revision is not None:
796
path = self._basis_inventory_name(old_revision)
797
path = self.branch.control_files._escape(path)
798
self.branch.control_files._transport.delete(path)
802
xml = self.branch.repository.get_inventory_xml(new_revision)
803
path = self._basis_inventory_name(new_revision)
804
self.branch.control_files.put_utf8(path, xml)
805
except WeaveRevisionNotPresent:
808
def read_basis_inventory(self, revision_id):
809
"""Read the cached basis inventory."""
810
path = self._basis_inventory_name(revision_id)
811
return self.branch.control_files.controlfile(path, 'r').read()
814
def read_working_inventory(self):
815
"""Read the working inventory."""
816
# ElementTree does its own conversion from UTF-8, so open in
818
f = self.branch.control_files.controlfile('inventory', 'rb')
819
return bzrlib.xml5.serializer_v5.read_inventory(f)
822
def remove(self, files, verbose=False):
823
"""Remove nominated files from the working inventory..
825
This does not remove their text. This does not run on XXX on what? RBC
827
TODO: Refuse to remove modified files unless --force is given?
829
TODO: Do something useful with directories.
831
TODO: Should this remove the text or not? Tough call; not
832
removing may be useful and the user can just use use rm, and
833
is the opposite of add. Removing it is consistent with most
834
other tools. Maybe an option.
836
## TODO: Normalize names
837
## TODO: Remove nested loops; better scalability
838
if isinstance(files, basestring):
843
# do this before any modifications
847
# TODO: Perhaps make this just a warning, and continue?
848
# This tends to happen when
849
raise NotVersionedError(path=f)
850
mutter("remove inventory entry %s {%s}", quotefn(f), fid)
852
# having remove it, it must be either ignored or unknown
853
if self.is_ignored(f):
857
show_status(new_status, inv[fid].kind, quotefn(f))
860
self._write_inventory(inv)
863
def revert(self, filenames, old_tree=None, backups=True):
864
from bzrlib.merge import merge_inner
866
old_tree = self.branch.basis_tree()
867
merge_inner(self.branch, old_tree,
868
self, ignore_zero=True,
869
backup_files=backups,
870
interesting_files=filenames)
871
if not len(filenames):
872
self.set_pending_merges([])
875
def set_inventory(self, new_inventory_list):
876
from bzrlib.inventory import (Inventory,
881
inv = Inventory(self.get_root_id())
882
for path, file_id, parent, kind in new_inventory_list:
883
name = os.path.basename(path)
886
# fixme, there should be a factory function inv,add_??
887
if kind == 'directory':
888
inv.add(InventoryDirectory(file_id, name, parent))
890
inv.add(InventoryFile(file_id, name, parent))
891
elif kind == 'symlink':
892
inv.add(InventoryLink(file_id, name, parent))
894
raise BzrError("unknown kind %r" % kind)
895
self._write_inventory(inv)
898
def set_root_id(self, file_id):
899
"""Set the root id for this tree."""
900
inv = self.read_working_inventory()
901
orig_root_id = inv.root.file_id
902
del inv._byid[inv.root.file_id]
903
inv.root.file_id = file_id
904
inv._byid[inv.root.file_id] = inv.root
907
if entry.parent_id in (None, orig_root_id):
908
entry.parent_id = inv.root.file_id
909
self._write_inventory(inv)
912
"""See Branch.unlock.
914
WorkingTree locking just uses the Branch locking facilities.
915
This is current because all working trees have an embedded branch
916
within them. IF in the future, we were to make branch data shareable
917
between multiple working trees, i.e. via shared storage, then we
918
would probably want to lock both the local tree, and the branch.
920
# FIXME: We want to write out the hashcache only when the last lock on
921
# this working copy is released. Peeking at the lock count is a bit
922
# of a nasty hack; probably it's better to have a transaction object,
923
# which can do some finalization when it's either successfully or
924
# unsuccessfully completed. (Denys's original patch did that.)
925
if self._hashcache.needs_write and self.branch.control_files._lock_count==1:
926
self._hashcache.write()
927
return self.branch.unlock()
930
def _write_inventory(self, inv):
931
"""Write inventory as the current inventory."""
932
from cStringIO import StringIO
933
from bzrlib.atomicfile import AtomicFile
935
bzrlib.xml5.serializer_v5.write_inventory(inv, sio)
937
f = AtomicFile(self.branch.control_files.controlfilename('inventory'))
943
self._set_inventory(inv)
944
mutter('wrote working inventory')
947
CONFLICT_SUFFIXES = ('.THIS', '.BASE', '.OTHER')
948
def get_conflicted_stem(path):
949
for suffix in CONFLICT_SUFFIXES:
950
if path.endswith(suffix):
951
return path[:-len(suffix)]
added
removed
bzrlib/workingtree.py
