1
# (C) 2005 Canonical Ltd
1
# Copyright (C) 2005, 2006, 2007 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
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# FIXME: This refactoring of the workingtree code doesn't seem to keep
18
# the WorkingTree's copy of the inventory in sync with the branch. The
19
# branch modifies its working inventory when it does a commit to make
20
# missing files permanently removed.
22
# TODO: Maybe also keep the full path of the entry, and the children?
23
# But those depend on its position within a particular inventory, and
24
# it would be nice not to need to hold the backpointer here.
18
26
# This should really be an id randomly assigned when the tree is
19
27
# created, but it's not for now.
20
28
ROOT_ID = "TREE_ROOT"
23
import sys, os.path, types, re
27
from cElementTree import Element, ElementTree, SubElement
29
from elementtree.ElementTree import Element, ElementTree, SubElement
31
from xml import XMLMixin
32
from errors import bailout, BzrError, BzrCheckError
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
49
from bzrlib.errors import (
36
53
from bzrlib.trace import mutter
38
class InventoryEntry(XMLMixin):
56
class InventoryEntry(object):
39
57
"""Description of a versioned file.
41
59
An InventoryEntry has the following fields, which are also
42
60
present in the XML inventory-entry element:
45
* *name*: (only the basename within the directory, must not
47
* *kind*: "directory" or "file"
48
* *directory_id*: (if absent/null means the branch root directory)
49
* *text_sha1*: only for files
50
* *text_size*: in bytes, only for files
51
* *text_id*: identifier for the text version, only for files
53
InventoryEntries can also exist inside a WorkingTree
54
inventory, in which case they are not yet bound to a
55
particular revision of the file. In that case the text_sha1,
56
text_size and text_id are absent.
65
(within the parent directory)
68
file_id of the parent directory, or ROOT_ID
71
the revision_id in which this variation of this file was
75
Indicates that this file should be executable on systems
79
sha-1 of the text of the file
82
size in bytes of the text of the file
84
(reading a version 4 tree created a text_id field.)
59
86
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
63
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
64
>>> for j in i.iter_entries():
89
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
90
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
91
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
92
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
93
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
94
>>> for ix, j in enumerate(i.iter_entries()):
95
... print (j[0] == shouldbe[ix], j[1])
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
Traceback (most recent call last):
72
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
97
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
98
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
99
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
100
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
101
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
102
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
103
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
75
104
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
108
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
109
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
>>> for j in i.iter_entries():
84
... assert i.path2id(j[0])
111
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
112
>>> for path, entry in i.iter_entries():
114
... assert i.path2id(path)
90
121
src/wibble/wibble.c
91
122
>>> i.id2path('2326')
92
123
'src/wibble/wibble.c'
94
TODO: Maybe also keep the full path of the entry, and the children?
95
But those depend on its position within a particular inventory, and
96
it would be nice not to need to hold the backpointer here.
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
126
# Constants returned by describe_change()
128
# TODO: These should probably move to some kind of FileChangeDescription
129
# class; that's like what's inside a TreeDelta but we want to be able to
130
# generate them just for one file at a time.
132
MODIFIED_AND_RENAMED = 'modified and renamed'
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
136
def detect_changes(self, old_entry):
137
"""Return a (text_modified, meta_modified) from this to old_entry.
139
_read_tree_state must have been called on self and old_entry prior to
140
calling detect_changes.
144
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
145
output_to, reverse=False):
146
"""Perform a diff from this to to_entry.
148
text_diff will be used for textual difference calculation.
149
This is a template method, override _diff in child classes.
151
self._read_tree_state(tree.id2path(self.file_id), tree)
153
# cannot diff from one kind to another - you must do a removal
154
# and an addif they do not match.
155
assert self.kind == to_entry.kind
156
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
158
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
161
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
162
output_to, reverse=False):
163
"""Perform a diff between two entries of the same kind."""
165
def find_previous_heads(self, previous_inventories,
166
versioned_file_store,
169
"""Return the revisions and entries that directly precede this.
171
Returned as a map from revision to inventory entry.
173
This is a map containing the file revisions in all parents
174
for which the file exists, and its revision is not a parent of
175
any other. If the file is new, the set will be empty.
177
:param versioned_file_store: A store where ancestry data on this
178
file id can be queried.
179
:param transaction: The transaction that queries to the versioned
180
file store should be completed under.
181
:param entry_vf: The entry versioned file, if its already available.
183
def get_ancestors(weave, entry):
184
return set(weave.get_ancestry(entry.revision))
185
# revision:ie mapping for each ie found in previous_inventories.
187
# revision:ie mapping with one revision for each head.
189
# revision: ancestor list for each head
191
# identify candidate head revision ids.
192
for inv in previous_inventories:
193
if self.file_id in inv:
194
ie = inv[self.file_id]
195
assert ie.file_id == self.file_id
196
if ie.revision in candidates:
197
# same revision value in two different inventories:
198
# correct possible inconsistencies:
199
# * there was a bug in revision updates with 'x' bit
202
if candidates[ie.revision].executable != ie.executable:
203
candidates[ie.revision].executable = False
204
ie.executable = False
205
except AttributeError:
207
# must now be the same.
208
assert candidates[ie.revision] == ie
210
# add this revision as a candidate.
211
candidates[ie.revision] = ie
213
# common case optimisation
214
if len(candidates) == 1:
215
# if there is only one candidate revision found
216
# then we can opening the versioned file to access ancestry:
217
# there cannot be any ancestors to eliminate when there is
218
# only one revision available.
219
heads[ie.revision] = ie
222
# eliminate ancestors amongst the available candidates:
223
# heads are those that are not an ancestor of any other candidate
224
# - this provides convergence at a per-file level.
225
for ie in candidates.values():
226
# may be an ancestor of a known head:
227
already_present = 0 != len(
228
[head for head in heads
229
if ie.revision in head_ancestors[head]])
231
# an ancestor of an analyzed candidate.
233
# not an ancestor of a known head:
234
# load the versioned file for this file id if needed
236
entry_vf = versioned_file_store.get_weave_or_empty(
237
self.file_id, transaction)
238
ancestors = get_ancestors(entry_vf, ie)
239
# may knock something else out:
240
check_heads = list(heads.keys())
241
for head in check_heads:
242
if head in ancestors:
243
# this previously discovered 'head' is not
244
# really a head - its an ancestor of the newly
247
head_ancestors[ie.revision] = ancestors
248
heads[ie.revision] = ie
251
def get_tar_item(self, root, dp, now, tree):
252
"""Get a tarfile item and a file stream for its content."""
253
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
254
# TODO: would be cool to actually set it to the timestamp of the
255
# revision it was last changed
257
fileobj = self._put_in_tar(item, tree)
261
"""Return true if the object this entry represents has textual data.
263
Note that textual data includes binary content.
265
Also note that all entries get weave files created for them.
266
This attribute is primarily used when upgrading from old trees that
267
did not have the weave index for all inventory entries.
271
def __init__(self, file_id, name, parent_id, text_id=None):
106
272
"""Create an InventoryEntry
108
274
The filename must be a single component, relative to the
109
275
parent directory; it cannot be a whole path or relative name.
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
277
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
282
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
117
283
Traceback (most recent call last):
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
284
InvalidEntryName: Invalid entry name: src/hello.c
286
assert isinstance(name, basestring), name
120
287
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
288
raise errors.InvalidEntryName(name=name)
289
self.executable = False
291
self.text_sha1 = None
292
self.text_size = None
123
293
self.file_id = file_id
294
assert isinstance(file_id, (str, None.__class__)), \
295
'bad type %r for %r' % (type(file_id), file_id)
126
297
self.text_id = text_id
127
298
self.parent_id = parent_id
128
if kind == 'directory':
133
raise BzrError("unhandled entry kind %r" % kind)
299
self.symlink_target = None
300
self.reference_revision = None
302
def kind_character(self):
303
"""Return a short kind indicator useful for appending to names."""
304
raise BzrError('unknown kind %r' % self.kind)
306
known_kinds = ('file', 'directory', 'symlink')
308
def _put_in_tar(self, item, tree):
309
"""populate item for stashing in a tar, and return the content stream.
311
If no content is available, return None.
313
raise BzrError("don't know how to export {%s} of kind %r" %
314
(self.file_id, self.kind))
316
def put_on_disk(self, dest, dp, tree):
317
"""Create a representation of self on disk in the prefix dest.
319
This is a template method - implement _put_on_disk in subclasses.
321
fullpath = osutils.pathjoin(dest, dp)
322
self._put_on_disk(fullpath, tree)
323
# mutter(" export {%s} kind %s to %s", self.file_id,
324
# self.kind, fullpath)
326
def _put_on_disk(self, fullpath, tree):
327
"""Put this entry onto disk at fullpath, from tree tree."""
328
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
137
330
def sorted_children(self):
138
l = self.children.items()
331
return sorted(self.children.items())
334
def versionable_kind(kind):
335
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
337
def check(self, checker, rev_id, inv, tree):
338
"""Check this inventory entry is intact.
340
This is a template method, override _check for kind specific
343
:param checker: Check object providing context for the checks;
344
can be used to find out what parts of the repository have already
346
:param rev_id: Revision id from which this InventoryEntry was loaded.
347
Not necessarily the last-changed revision for this file.
348
:param inv: Inventory from which the entry was loaded.
349
:param tree: RevisionTree for this entry.
351
if self.parent_id is not None:
352
if not inv.has_id(self.parent_id):
353
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
354
% (self.parent_id, rev_id))
355
self._check(checker, rev_id, tree)
357
def _check(self, checker, rev_id, tree):
358
"""Check this inventory entry for kind specific errors."""
359
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
144
other = InventoryEntry(self.file_id, self.name, self.kind,
145
self.parent_id, text_id=self.text_id)
146
other.text_sha1 = self.text_sha1
147
other.text_size = self.text_size
363
"""Clone this inventory entry."""
364
raise NotImplementedError
367
def describe_change(old_entry, new_entry):
368
"""Describe the change between old_entry and this.
370
This smells of being an InterInventoryEntry situation, but as its
371
the first one, we're making it a static method for now.
373
An entry with a different parent, or different name is considered
374
to be renamed. Reparenting is an internal detail.
375
Note that renaming the parent does not trigger a rename for the
378
# TODO: Perhaps return an object rather than just a string
379
if old_entry is new_entry:
380
# also the case of both being None
382
elif old_entry is None:
384
elif new_entry is None:
386
if old_entry.kind != new_entry.kind:
388
text_modified, meta_modified = new_entry.detect_changes(old_entry)
389
if text_modified or meta_modified:
393
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
394
if old_entry.parent_id != new_entry.parent_id:
396
elif old_entry.name != new_entry.name:
400
if renamed and not modified:
401
return InventoryEntry.RENAMED
402
if modified and not renamed:
404
if modified and renamed:
405
return InventoryEntry.MODIFIED_AND_RENAMED
151
408
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
409
return ("%s(%r, %r, parent_id=%r, revision=%r)"
153
410
% (self.__class__.__name__,
160
def to_element(self):
161
"""Convert to XML element"""
164
e.set('name', self.name)
165
e.set('file_id', self.file_id)
166
e.set('kind', self.kind)
168
if self.text_size != None:
169
e.set('text_size', '%d' % self.text_size)
171
for f in ['text_id', 'text_sha1']:
176
# to be conservative, we don't externalize the root pointers
177
# for now, leaving them as null in the xml form. in a future
178
# version it will be implied by nested elements.
179
if self.parent_id != ROOT_ID:
180
assert isinstance(self.parent_id, basestring)
181
e.set('parent_id', self.parent_id)
188
def from_element(cls, elt):
189
assert elt.tag == 'entry'
191
## original format inventories don't have a parent_id for
192
## nodes in the root directory, but it's cleaner to use one
194
parent_id = elt.get('parent_id')
195
if parent_id == None:
198
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
199
self.text_id = elt.get('text_id')
200
self.text_sha1 = elt.get('text_sha1')
202
## mutter("read inventoryentry: %r" % (elt.attrib))
204
v = elt.get('text_size')
205
self.text_size = v and int(v)
210
from_element = classmethod(from_element)
212
def __cmp__(self, other):
416
def snapshot(self, revision, path, previous_entries,
417
work_tree, commit_builder):
418
"""Make a snapshot of this entry which may or may not have changed.
420
This means that all its fields are populated, that it has its
421
text stored in the text store or weave.
423
# mutter('new parents of %s are %r', path, previous_entries)
424
self._read_tree_state(path, work_tree)
425
# TODO: Where should we determine whether to reuse a
426
# previous revision id or create a new revision? 20060606
427
if len(previous_entries) == 1:
428
# cannot be unchanged unless there is only one parent file rev.
429
parent_ie = previous_entries.values()[0]
430
if self._unchanged(parent_ie):
431
# mutter("found unchanged entry")
432
self.revision = parent_ie.revision
434
return self._snapshot_into_revision(revision, previous_entries,
435
work_tree, commit_builder)
437
def _snapshot_into_revision(self, revision, previous_entries, work_tree,
439
"""Record this revision unconditionally into a store.
441
The entry's last-changed revision property (`revision`) is updated to
442
that of the new revision.
444
:param revision: id of the new revision that is being recorded.
446
:returns: String description of the commit (e.g. "merged", "modified"), etc.
448
# mutter('new revision {%s} for {%s}', revision, self.file_id)
449
self.revision = revision
450
self._snapshot_text(previous_entries, work_tree, commit_builder)
452
def _snapshot_text(self, file_parents, work_tree, commit_builder):
453
"""Record the 'text' of this entry, whatever form that takes.
455
This default implementation simply adds an empty text.
457
raise NotImplementedError(self._snapshot_text)
459
def __eq__(self, other):
215
460
if not isinstance(other, InventoryEntry):
216
461
return NotImplemented
218
return cmp(self.file_id, other.file_id) \
219
or cmp(self.name, other.name) \
220
or cmp(self.text_sha1, other.text_sha1) \
221
or cmp(self.text_size, other.text_size) \
222
or cmp(self.text_id, other.text_id) \
223
or cmp(self.parent_id, other.parent_id) \
224
or cmp(self.kind, other.kind)
463
return ((self.file_id == other.file_id)
464
and (self.name == other.name)
465
and (other.symlink_target == self.symlink_target)
466
and (self.text_sha1 == other.text_sha1)
467
and (self.text_size == other.text_size)
468
and (self.text_id == other.text_id)
469
and (self.parent_id == other.parent_id)
470
and (self.kind == other.kind)
471
and (self.revision == other.revision)
472
and (self.executable == other.executable)
473
and (self.reference_revision == other.reference_revision)
476
def __ne__(self, other):
477
return not (self == other)
480
raise ValueError('not hashable')
482
def _unchanged(self, previous_ie):
483
"""Has this entry changed relative to previous_ie.
485
This method should be overridden in child classes.
488
# different inv parent
489
if previous_ie.parent_id != self.parent_id:
492
elif previous_ie.name != self.name:
496
def _read_tree_state(self, path, work_tree):
497
"""Populate fields in the inventory entry from the given tree.
499
Note that this should be modified to be a noop on virtual trees
500
as all entries created there are prepopulated.
502
# TODO: Rather than running this manually, we should check the
503
# working sha1 and other expensive properties when they're
504
# first requested, or preload them if they're already known
505
pass # nothing to do by default
507
def _forget_tree_state(self):
228
511
class RootEntry(InventoryEntry):
513
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
514
'text_id', 'parent_id', 'children', 'executable',
515
'revision', 'symlink_target', 'reference_revision']
517
def _check(self, checker, rev_id, tree):
518
"""See InventoryEntry._check"""
229
520
def __init__(self, file_id):
230
521
self.file_id = file_id
231
522
self.children = {}
232
self.kind = 'root_directory'
523
self.kind = 'directory'
233
524
self.parent_id = None
527
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
528
' Please use InventoryDirectory instead.',
529
DeprecationWarning, stacklevel=2)
236
def __cmp__(self, other):
531
def __eq__(self, other):
239
532
if not isinstance(other, RootEntry):
240
533
return NotImplemented
241
return cmp(self.file_id, other.file_id) \
242
or cmp(self.children, other.children)
246
class Inventory(XMLMixin):
535
return (self.file_id == other.file_id) \
536
and (self.children == other.children)
539
class InventoryDirectory(InventoryEntry):
540
"""A directory in an inventory."""
542
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
543
'text_id', 'parent_id', 'children', 'executable',
544
'revision', 'symlink_target', 'reference_revision']
546
def _check(self, checker, rev_id, tree):
547
"""See InventoryEntry._check"""
548
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
549
raise BzrCheckError('directory {%s} has text in revision {%s}'
550
% (self.file_id, rev_id))
553
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
554
other.revision = self.revision
555
# note that children are *not* copied; they're pulled across when
559
def __init__(self, file_id, name, parent_id):
560
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
562
self.kind = 'directory'
564
def kind_character(self):
565
"""See InventoryEntry.kind_character."""
568
def _put_in_tar(self, item, tree):
569
"""See InventoryEntry._put_in_tar."""
570
item.type = tarfile.DIRTYPE
577
def _put_on_disk(self, fullpath, tree):
578
"""See InventoryEntry._put_on_disk."""
581
def _snapshot_text(self, file_parents, work_tree, commit_builder):
582
"""See InventoryEntry._snapshot_text."""
583
commit_builder.modified_directory(self.file_id, file_parents)
586
class InventoryFile(InventoryEntry):
587
"""A file in an inventory."""
589
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
590
'text_id', 'parent_id', 'children', 'executable',
591
'revision', 'symlink_target', 'reference_revision']
593
def _check(self, checker, tree_revision_id, tree):
594
"""See InventoryEntry._check"""
595
t = (self.file_id, self.revision)
596
if t in checker.checked_texts:
597
prev_sha = checker.checked_texts[t]
598
if prev_sha != self.text_sha1:
599
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
600
(self.file_id, tree_revision_id))
602
checker.repeated_text_cnt += 1
605
if self.file_id not in checker.checked_weaves:
606
mutter('check weave {%s}', self.file_id)
607
w = tree.get_weave(self.file_id)
608
# Not passing a progress bar, because it creates a new
609
# progress, which overwrites the current progress,
610
# and doesn't look nice
612
checker.checked_weaves[self.file_id] = True
614
w = tree.get_weave(self.file_id)
616
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
617
checker.checked_text_cnt += 1
618
# We can't check the length, because Weave doesn't store that
619
# information, and the whole point of looking at the weave's
620
# sha1sum is that we don't have to extract the text.
621
if self.text_sha1 != w.get_sha1(self.revision):
622
raise BzrCheckError('text {%s} version {%s} wrong sha1'
623
% (self.file_id, self.revision))
624
checker.checked_texts[t] = self.text_sha1
627
other = InventoryFile(self.file_id, self.name, self.parent_id)
628
other.executable = self.executable
629
other.text_id = self.text_id
630
other.text_sha1 = self.text_sha1
631
other.text_size = self.text_size
632
other.revision = self.revision
635
def detect_changes(self, old_entry):
636
"""See InventoryEntry.detect_changes."""
637
assert self.text_sha1 is not None
638
assert old_entry.text_sha1 is not None
639
text_modified = (self.text_sha1 != old_entry.text_sha1)
640
meta_modified = (self.executable != old_entry.executable)
641
return text_modified, meta_modified
643
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
644
output_to, reverse=False):
645
"""See InventoryEntry._diff."""
647
from_text = tree.get_file(self.file_id).readlines()
649
to_text = to_tree.get_file(to_entry.file_id).readlines()
653
text_diff(from_label, from_text,
654
to_label, to_text, output_to)
656
text_diff(to_label, to_text,
657
from_label, from_text, output_to)
658
except errors.BinaryFile:
660
label_pair = (to_label, from_label)
662
label_pair = (from_label, to_label)
663
print >> output_to, "Binary files %s and %s differ" % label_pair
666
"""See InventoryEntry.has_text."""
669
def __init__(self, file_id, name, parent_id):
670
super(InventoryFile, self).__init__(file_id, name, parent_id)
673
def kind_character(self):
674
"""See InventoryEntry.kind_character."""
677
def _put_in_tar(self, item, tree):
678
"""See InventoryEntry._put_in_tar."""
679
item.type = tarfile.REGTYPE
680
fileobj = tree.get_file(self.file_id)
681
item.size = self.text_size
682
if tree.is_executable(self.file_id):
688
def _put_on_disk(self, fullpath, tree):
689
"""See InventoryEntry._put_on_disk."""
690
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
691
if tree.is_executable(self.file_id):
692
os.chmod(fullpath, 0755)
694
def _read_tree_state(self, path, work_tree):
695
"""See InventoryEntry._read_tree_state."""
696
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
697
# FIXME: 20050930 probe for the text size when getting sha1
698
# in _read_tree_state
699
self.executable = work_tree.is_executable(self.file_id, path=path)
702
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
703
% (self.__class__.__name__,
710
def _forget_tree_state(self):
711
self.text_sha1 = None
713
def _snapshot_text(self, file_parents, work_tree, commit_builder):
714
"""See InventoryEntry._snapshot_text."""
715
def get_content_byte_lines():
716
return work_tree.get_file(self.file_id).readlines()
717
self.text_sha1, self.text_size = commit_builder.modified_file_text(
718
self.file_id, file_parents, get_content_byte_lines, self.text_sha1, self.text_size)
720
def _unchanged(self, previous_ie):
721
"""See InventoryEntry._unchanged."""
722
compatible = super(InventoryFile, self)._unchanged(previous_ie)
723
if self.text_sha1 != previous_ie.text_sha1:
726
# FIXME: 20050930 probe for the text size when getting sha1
727
# in _read_tree_state
728
self.text_size = previous_ie.text_size
729
if self.executable != previous_ie.executable:
734
class InventoryLink(InventoryEntry):
735
"""A file in an inventory."""
737
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
738
'text_id', 'parent_id', 'children', 'executable',
739
'revision', 'symlink_target', 'reference_revision']
741
def _check(self, checker, rev_id, tree):
742
"""See InventoryEntry._check"""
743
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
744
raise BzrCheckError('symlink {%s} has text in revision {%s}'
745
% (self.file_id, rev_id))
746
if self.symlink_target is None:
747
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
748
% (self.file_id, rev_id))
751
other = InventoryLink(self.file_id, self.name, self.parent_id)
752
other.symlink_target = self.symlink_target
753
other.revision = self.revision
756
def detect_changes(self, old_entry):
757
"""See InventoryEntry.detect_changes."""
758
# FIXME: which _modified field should we use ? RBC 20051003
759
text_modified = (self.symlink_target != old_entry.symlink_target)
761
mutter(" symlink target changed")
762
meta_modified = False
763
return text_modified, meta_modified
765
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
766
output_to, reverse=False):
767
"""See InventoryEntry._diff."""
768
from_text = self.symlink_target
769
if to_entry is not None:
770
to_text = to_entry.symlink_target
775
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
778
print >>output_to, '=== target was %r' % self.symlink_target
780
print >>output_to, '=== target is %r' % self.symlink_target
782
def __init__(self, file_id, name, parent_id):
783
super(InventoryLink, self).__init__(file_id, name, parent_id)
784
self.kind = 'symlink'
786
def kind_character(self):
787
"""See InventoryEntry.kind_character."""
790
def _put_in_tar(self, item, tree):
791
"""See InventoryEntry._put_in_tar."""
792
item.type = tarfile.SYMTYPE
796
item.linkname = self.symlink_target
799
def _put_on_disk(self, fullpath, tree):
800
"""See InventoryEntry._put_on_disk."""
802
os.symlink(self.symlink_target, fullpath)
804
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
806
def _read_tree_state(self, path, work_tree):
807
"""See InventoryEntry._read_tree_state."""
808
self.symlink_target = work_tree.get_symlink_target(self.file_id)
810
def _forget_tree_state(self):
811
self.symlink_target = None
813
def _unchanged(self, previous_ie):
814
"""See InventoryEntry._unchanged."""
815
compatible = super(InventoryLink, self)._unchanged(previous_ie)
816
if self.symlink_target != previous_ie.symlink_target:
820
def _snapshot_text(self, file_parents, work_tree, commit_builder):
821
"""See InventoryEntry._snapshot_text."""
822
commit_builder.modified_link(
823
self.file_id, file_parents, self.symlink_target)
826
class TreeReference(InventoryEntry):
828
kind = 'tree-reference'
830
def __init__(self, file_id, name, parent_id, revision=None,
831
reference_revision=None):
832
InventoryEntry.__init__(self, file_id, name, parent_id)
833
self.revision = revision
834
self.reference_revision = reference_revision
837
return TreeReference(self.file_id, self.name, self.parent_id,
838
self.revision, self.reference_revision)
840
def _snapshot_text(self, file_parents, work_tree, commit_builder):
841
commit_builder.modified_reference(self.file_id, file_parents)
843
def _read_tree_state(self, path, work_tree):
844
"""Populate fields in the inventory entry from the given tree.
846
self.reference_revision = work_tree.get_reference_revision(
849
def _forget_tree_state(self):
850
self.reference_revision = None
853
class Inventory(object):
247
854
"""Inventory of versioned files in a tree.
249
856
This describes which file_id is present at each point in the tree,
296
901
The inventory is created with a default root directory, with
299
self.root = RootEntry(ROOT_ID)
904
if root_id is not None:
905
assert root_id.__class__ == str
906
self._set_root(InventoryDirectory(root_id, u'', None))
910
self.revision_id = revision_id
912
def _set_root(self, ie):
300
914
self._byid = {self.root.file_id: self.root}
917
# TODO: jam 20051218 Should copy also copy the revision_id?
918
entries = self.iter_entries()
919
other = Inventory(entries.next()[1].file_id)
920
# copy recursively so we know directories will be added before
921
# their children. There are more efficient ways than this...
922
for path, entry in entries():
923
other.add(entry.copy())
303
926
def __iter__(self):
304
927
return iter(self._byid)
307
929
def __len__(self):
308
930
"""Returns number of entries."""
309
931
return len(self._byid)
312
933
def iter_entries(self, from_dir=None):
313
934
"""Return (path, entry) pairs, in order by name."""
317
elif isinstance(from_dir, basestring):
318
from_dir = self._byid[from_dir]
320
kids = from_dir.children.items()
322
for name, ie in kids:
324
if ie.kind == 'directory':
325
for cn, cie in self.iter_entries(from_dir=ie.file_id):
326
yield os.path.join(name, cn), cie
936
if self.root is None:
940
elif isinstance(from_dir, basestring):
941
from_dir = self._byid[from_dir]
943
# unrolling the recursive called changed the time from
944
# 440ms/663ms (inline/total) to 116ms/116ms
945
children = from_dir.children.items()
947
children = collections.deque(children)
948
stack = [(u'', children)]
950
from_dir_relpath, children = stack[-1]
953
name, ie = children.popleft()
955
# we know that from_dir_relpath never ends in a slash
956
# and 'f' doesn't begin with one, we can do a string op, rather
957
# than the checks of pathjoin(), though this means that all paths
959
path = from_dir_relpath + '/' + name
963
if ie.kind != 'directory':
966
# But do this child first
967
new_children = ie.children.items()
969
new_children = collections.deque(new_children)
970
stack.append((path, new_children))
971
# Break out of inner loop, so that we start outer loop with child
974
# if we finished all children, pop it off the stack
977
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
978
"""Iterate over the entries in a directory first order.
980
This returns all entries for a directory before returning
981
the entries for children of a directory. This is not
982
lexicographically sorted order, and is a hybrid between
983
depth-first and breadth-first.
985
:return: This yields (path, entry) pairs
987
if specific_file_ids:
988
safe = osutils.safe_file_id
989
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
990
# TODO? Perhaps this should return the from_dir so that the root is
991
# yielded? or maybe an option?
993
if self.root is None:
995
# Optimize a common case
996
if specific_file_ids is not None and len(specific_file_ids) == 1:
997
file_id = list(specific_file_ids)[0]
999
yield self.id2path(file_id), self[file_id]
1001
from_dir = self.root
1002
if (specific_file_ids is None or
1003
self.root.file_id in specific_file_ids):
1004
yield u'', self.root
1005
elif isinstance(from_dir, basestring):
1006
from_dir = self._byid[from_dir]
1008
if specific_file_ids is not None:
1009
# TODO: jam 20070302 This could really be done as a loop rather
1010
# than a bunch of recursive calls.
1013
def add_ancestors(file_id):
1014
if file_id not in byid:
1016
parent_id = byid[file_id].parent_id
1017
if parent_id is None:
1019
if parent_id not in parents:
1020
parents.add(parent_id)
1021
add_ancestors(parent_id)
1022
for file_id in specific_file_ids:
1023
add_ancestors(file_id)
1027
stack = [(u'', from_dir)]
1029
cur_relpath, cur_dir = stack.pop()
1032
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1034
child_relpath = cur_relpath + child_name
1036
if (specific_file_ids is None or
1037
child_ie.file_id in specific_file_ids):
1038
yield child_relpath, child_ie
1040
if child_ie.kind == 'directory':
1041
if parents is None or child_ie.file_id in parents:
1042
child_dirs.append((child_relpath+'/', child_ie))
1043
stack.extend(reversed(child_dirs))
1046
"""Return list of (path, ie) for all entries except the root.
1048
This may be faster than iter_entries.
1051
def descend(dir_ie, dir_path):
1052
kids = dir_ie.children.items()
1054
for name, ie in kids:
1055
child_path = osutils.pathjoin(dir_path, name)
1056
accum.append((child_path, ie))
1057
if ie.kind == 'directory':
1058
descend(ie, child_path)
1060
descend(self.root, u'')
330
1063
def directories(self):
331
"""Return (path, entry) pairs for all directories.
1064
"""Return (path, entry) pairs for all directories, including the root.
333
def descend(parent_ie):
334
parent_name = parent_ie.name
335
yield parent_name, parent_ie
337
# directory children in sorted order
339
for ie in parent_ie.children.itervalues():
340
if ie.kind == 'directory':
341
dn.append((ie.name, ie))
1067
def descend(parent_ie, parent_path):
1068
accum.append((parent_path, parent_ie))
344
for name, child_ie in dn:
345
for sub_name, sub_ie in descend(child_ie):
346
yield appendpath(parent_name, sub_name), sub_ie
1070
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
348
for name, ie in descend(self.root):
1073
for name, child_ie in kids:
1074
child_path = osutils.pathjoin(parent_path, name)
1075
descend(child_ie, child_path)
1076
descend(self.root, u'')
353
1079
def __contains__(self, file_id):
354
1080
"""True if this entry contains a file with given id.
356
1082
>>> inv = Inventory()
357
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1083
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1084
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
358
1085
>>> '123' in inv
360
1087
>>> '456' in inv
363
return file_id in self._byid
1090
file_id = osutils.safe_file_id(file_id)
1091
return (file_id in self._byid)
366
1093
def __getitem__(self, file_id):
367
1094
"""Return the entry for given file_id.
369
1096
>>> inv = Inventory()
370
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
1097
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1098
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
371
1099
>>> inv['123123'].name
1102
file_id = osutils.safe_file_id(file_id)
375
1104
return self._byid[file_id]
376
1105
except KeyError:
378
raise BzrError("can't look up file_id None")
380
raise BzrError("file_id {%s} not in inventory" % file_id)
1106
# really we're passing an inventory, not a tree...
1107
raise errors.NoSuchId(self, file_id)
1109
def get_file_kind(self, file_id):
1110
file_id = osutils.safe_file_id(file_id)
1111
return self._byid[file_id].kind
383
1113
def get_child(self, parent_id, filename):
1114
parent_id = osutils.safe_file_id(parent_id)
384
1115
return self[parent_id].children.get(filename)
1117
def _add_child(self, entry):
1118
"""Add an entry to the inventory, without adding it to its parent"""
1119
if entry.file_id in self._byid:
1120
raise BzrError("inventory already contains entry with id {%s}" %
1122
self._byid[entry.file_id] = entry
1123
for child in getattr(entry, 'children', {}).itervalues():
1124
self._add_child(child)
387
1127
def add(self, entry):
388
1128
"""Add entry to inventory.
390
1130
To add a file to a branch ready to be committed, use Branch.add,
1133
Returns the new entry object.
392
1135
if entry.file_id in self._byid:
393
bailout("inventory already contains entry with id {%s}" % entry.file_id)
396
parent = self._byid[entry.parent_id]
398
bailout("parent_id {%s} not in inventory" % entry.parent_id)
400
if parent.children.has_key(entry.name):
401
bailout("%s is already versioned" %
402
appendpath(self.id2path(parent.file_id), entry.name))
404
self._byid[entry.file_id] = entry
405
parent.children[entry.name] = entry
408
def add_path(self, relpath, kind, file_id=None):
1136
raise errors.DuplicateFileId(entry.file_id,
1137
self._byid[entry.file_id])
1139
if entry.parent_id is None:
1140
assert self.root is None and len(self._byid) == 0
1144
parent = self._byid[entry.parent_id]
1146
raise BzrError("parent_id {%s} not in inventory" %
1149
if entry.name in parent.children:
1150
raise BzrError("%s is already versioned" %
1151
osutils.pathjoin(self.id2path(parent.file_id),
1153
parent.children[entry.name] = entry
1154
return self._add_child(entry)
1156
def add_path(self, relpath, kind, file_id=None, parent_id=None):
409
1157
"""Add entry from a path.
411
The immediate parent must already be versioned"""
412
parts = bzrlib.osutils.splitpath(relpath)
1159
The immediate parent must already be versioned.
1161
Returns the new entry object."""
1163
parts = osutils.splitpath(relpath)
413
1165
if len(parts) == 0:
414
bailout("cannot re-add root of inventory")
417
file_id = bzrlib.branch.gen_file_id(relpath)
419
parent_id = self.path2id(parts[:-1])
420
assert parent_id != None
421
ie = InventoryEntry(file_id, parts[-1],
422
kind=kind, parent_id=parent_id)
1167
file_id = generate_ids.gen_root_id()
1169
file_id = osutils.safe_file_id(file_id)
1170
self.root = InventoryDirectory(file_id, '', None)
1171
self._byid = {self.root.file_id: self.root}
1174
parent_path = parts[:-1]
1175
parent_id = self.path2id(parent_path)
1176
if parent_id is None:
1177
raise errors.NotVersionedError(path=parent_path)
1178
ie = make_entry(kind, parts[-1], parent_id, file_id)
423
1179
return self.add(ie)
426
1181
def __delitem__(self, file_id):
427
1182
"""Remove entry by id.
429
1184
>>> inv = Inventory()
430
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1185
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1186
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
431
1187
>>> '123' in inv
433
1189
>>> del inv['123']
434
1190
>>> '123' in inv
1193
file_id = osutils.safe_file_id(file_id)
437
1194
ie = self[file_id]
439
assert self[ie.parent_id].children[ie.name] == ie
1196
assert ie.parent_id is None or \
1197
self[ie.parent_id].children[ie.name] == ie
441
# TODO: Test deleting all children; maybe hoist to a separate
443
if ie.kind == 'directory':
444
for cie in ie.children.values():
445
del self[cie.file_id]
448
1199
del self._byid[file_id]
449
del self[ie.parent_id].children[ie.name]
453
return Set(self._byid)
456
def to_element(self):
457
"""Convert to XML Element"""
458
e = Element('inventory')
460
for path, ie in self.iter_entries():
461
e.append(ie.to_element())
465
def from_element(cls, elt):
466
"""Construct from XML Element
468
>>> inv = Inventory()
469
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
470
>>> elt = inv.to_element()
471
>>> inv2 = Inventory.from_element(elt)
475
assert elt.tag == 'inventory'
478
o.add(InventoryEntry.from_element(e))
481
from_element = classmethod(from_element)
484
def __cmp__(self, other):
1200
if ie.parent_id is not None:
1201
del self[ie.parent_id].children[ie.name]
1203
def __eq__(self, other):
485
1204
"""Compare two sets by comparing their contents.
487
1206
>>> i1 = Inventory()
488
1207
>>> i2 = Inventory()
491
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1210
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1211
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
494
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1214
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1215
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
501
1219
if not isinstance(other, Inventory):
502
1220
return NotImplemented
504
if self.id_set() ^ other.id_set():
507
for file_id in self._byid:
508
c = cmp(self[file_id], other[file_id])
1222
return self._byid == other._byid
1224
def __ne__(self, other):
1225
return not self.__eq__(other)
1228
raise ValueError('not hashable')
1230
def _iter_file_id_parents(self, file_id):
1231
"""Yield the parents of file_id up to the root."""
1232
file_id = osutils.safe_file_id(file_id)
1233
while file_id is not None:
1235
ie = self._byid[file_id]
1237
raise errors.NoSuchId(tree=None, file_id=file_id)
1239
file_id = ie.parent_id
514
1241
def get_idpath(self, file_id):
515
1242
"""Return a list of file_ids for the path to an entry.
added
removed
bzrlib/inventory.py
