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
18
# TODO: Maybe store inventory_id in the file? Not really needed.
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.
21
26
# This should really be an id randomly assigned when the tree is
22
27
# created, but it's not for now.
23
28
ROOT_ID = "TREE_ROOT"
26
import sys, os.path, types, re
30
from cElementTree import Element, ElementTree, SubElement
32
from elementtree.ElementTree import Element, ElementTree, SubElement
34
from xml import XMLMixin
35
from errors import bailout, BzrError
34
from bzrlib.lazy_import import lazy_import
35
lazy_import(globals(), """
38
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
49
from bzrlib.errors import (
53
from bzrlib.symbol_versioning import deprecated_method, zero_ninetyone
39
54
from bzrlib.trace import mutter
41
class InventoryEntry(XMLMixin):
57
class InventoryEntry(object):
42
58
"""Description of a versioned file.
44
60
An InventoryEntry has the following fields, which are also
45
61
present in the XML inventory-entry element:
48
* *name*: (only the basename within the directory, must not
50
* *kind*: "directory" or "file"
51
* *directory_id*: (if absent/null means the branch root directory)
52
* *text_sha1*: only for files
53
* *text_size*: in bytes, only for files
54
* *text_id*: identifier for the text version, only for files
56
InventoryEntries can also exist inside a WorkingTree
57
inventory, in which case they are not yet bound to a
58
particular revision of the file. In that case the text_sha1,
59
text_size and text_id are absent.
66
(within the parent directory)
69
file_id of the parent directory, or ROOT_ID
72
the revision_id in which this variation of this file was
76
Indicates that this file should be executable on systems
80
sha-1 of the text of the file
83
size in bytes of the text of the file
85
(reading a version 4 tree created a text_id field.)
62
87
>>> i = Inventory()
65
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
66
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
67
>>> for j in i.iter_entries():
90
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
91
InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None)
92
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
93
InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None)
94
>>> shouldbe = {0: '', 1: 'src', 2: 'src/hello.c'}
95
>>> for ix, j in enumerate(i.iter_entries()):
96
... print (j[0] == shouldbe[ix], j[1])
70
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
71
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
72
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
73
Traceback (most recent call last):
75
BzrError: ('inventory already contains entry with id {2323}', [])
76
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
77
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
98
(True, InventoryDirectory('TREE_ROOT', u'', parent_id=None, revision=None))
99
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT', revision=None))
100
(True, InventoryFile('2323', 'hello.c', parent_id='123', sha1=None, len=None))
101
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
102
InventoryFile('2324', 'bye.c', parent_id='123', sha1=None, len=None)
103
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
104
InventoryDirectory('2325', 'wibble', parent_id='123', revision=None)
78
105
>>> i.path2id('src/wibble')
82
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
109
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
110
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
84
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
85
>>> for j in i.iter_entries():
87
... assert i.path2id(j[0])
112
InventoryFile('2326', 'wibble.c', parent_id='2325', sha1=None, len=None)
113
>>> for path, entry in i.iter_entries():
115
... assert i.path2id(path)
93
122
src/wibble/wibble.c
94
123
>>> i.id2path('2326')
95
124
'src/wibble/wibble.c'
97
TODO: Maybe also keep the full path of the entry, and the children?
98
But those depend on its position within a particular inventory, and
99
it would be nice not to need to hold the backpointer here.
102
# TODO: split InventoryEntry into subclasses for files,
103
# directories, etc etc.
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
127
# Constants returned by describe_change()
129
# TODO: These should probably move to some kind of FileChangeDescription
130
# class; that's like what's inside a TreeDelta but we want to be able to
131
# generate them just for one file at a time.
133
MODIFIED_AND_RENAMED = 'modified and renamed'
137
def detect_changes(self, old_entry):
138
"""Return a (text_modified, meta_modified) from this to old_entry.
140
_read_tree_state must have been called on self and old_entry prior to
141
calling detect_changes.
145
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff from this to to_entry.
149
text_diff will be used for textual difference calculation.
150
This is a template method, override _diff in child classes.
152
self._read_tree_state(tree.id2path(self.file_id), tree)
154
# cannot diff from one kind to another - you must do a removal
155
# and an addif they do not match.
156
assert self.kind == to_entry.kind
157
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
159
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
162
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
163
output_to, reverse=False):
164
"""Perform a diff between two entries of the same kind."""
166
def parent_candidates(self, previous_inventories):
167
"""Find possible per-file graph parents.
169
This is currently defined by:
170
- Select the last changed revision in the parent inventory.
171
- Do deal with a short lived bug in bzr 0.8's development two entries
172
that have the same last changed but different 'x' bit settings are
175
# revision:ie mapping for each ie found in previous_inventories.
177
# identify candidate head revision ids.
178
for inv in previous_inventories:
179
if self.file_id in inv:
180
ie = inv[self.file_id]
181
assert ie.file_id == self.file_id
182
if ie.revision in candidates:
183
# same revision value in two different inventories:
184
# correct possible inconsistencies:
185
# * there was a bug in revision updates with 'x' bit
188
if candidates[ie.revision].executable != ie.executable:
189
candidates[ie.revision].executable = False
190
ie.executable = False
191
except AttributeError:
193
# must now be the same.
194
assert candidates[ie.revision] == ie
196
# add this revision as a candidate.
197
candidates[ie.revision] = ie
200
@deprecated_method(zero_ninetyone)
201
def find_previous_heads(self, previous_inventories,
202
versioned_file_store,
205
"""Return the revisions and entries that directly precede this.
207
Returned as a map from revision to inventory entry.
209
This is a map containing the file revisions in all parents
210
for which the file exists, and its revision is not a parent of
211
any other. If the file is new, the set will be empty.
213
:param versioned_file_store: A store where ancestry data on this
214
file id can be queried.
215
:param transaction: The transaction that queries to the versioned
216
file store should be completed under.
217
:param entry_vf: The entry versioned file, if its already available.
219
candidates = self.parent_candidates(previous_inventories)
221
# revision:ie mapping with one revision for each head.
223
# common case optimisation
224
if len(candidates) == 1:
225
# if there is only one candidate revision found
226
# then we can avoid opening the versioned file to access ancestry:
227
# there cannot be any ancestors to eliminate when there is
228
# only one revision available.
231
# --- what follows is now encapsulated in repository.get_graph.heads(),
232
# but that is not accessible from here as we have no repository
233
# pointer. Note that the repository.get_graph.heads() call can return
234
# different results *at the moment* because of the kind-changing check
235
# we have in parent_candidates().
237
# eliminate ancestors amongst the available candidates:
238
# heads are those that are not an ancestor of any other candidate
239
# - this provides convergence at a per-file level.
240
def get_ancestors(weave, entry):
241
return set(weave.get_ancestry(entry.revision, topo_sorted=False))
242
# revision: ancestor list for each head
244
for ie in candidates.values():
245
# may be an ancestor of a known head:
246
already_present = 0 != len(
247
[head for head in heads
248
if ie.revision in head_ancestors[head]])
250
# an ancestor of an analyzed candidate.
252
# not an ancestor of a known head:
253
# load the versioned file for this file id if needed
255
entry_vf = versioned_file_store.get_weave_or_empty(
256
self.file_id, transaction)
257
ancestors = get_ancestors(entry_vf, ie)
258
# may knock something else out:
259
check_heads = list(heads.keys())
260
for head in check_heads:
261
if head in ancestors:
262
# this previously discovered 'head' is not
263
# really a head - its an ancestor of the newly
266
head_ancestors[ie.revision] = ancestors
267
heads[ie.revision] = ie
270
def get_tar_item(self, root, dp, now, tree):
271
"""Get a tarfile item and a file stream for its content."""
272
item = tarfile.TarInfo(osutils.pathjoin(root, dp).encode('utf8'))
273
# TODO: would be cool to actually set it to the timestamp of the
274
# revision it was last changed
276
fileobj = self._put_in_tar(item, tree)
280
"""Return true if the object this entry represents has textual data.
282
Note that textual data includes binary content.
284
Also note that all entries get weave files created for them.
285
This attribute is primarily used when upgrading from old trees that
286
did not have the weave index for all inventory entries.
290
def __init__(self, file_id, name, parent_id, text_id=None):
106
291
"""Create an InventoryEntry
108
293
The filename must be a single component, relative to the
109
294
parent directory; it cannot be a whole path or relative name.
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
296
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
301
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
117
302
Traceback (most recent call last):
118
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
303
InvalidEntryName: Invalid entry name: src/hello.c
121
if len(splitpath(name)) != 1:
122
bailout('InventoryEntry name is not a simple filename: %r'
305
assert isinstance(name, basestring), name
306
if '/' in name or '\\' in name:
307
raise errors.InvalidEntryName(name=name)
308
self.executable = False
310
self.text_sha1 = None
311
self.text_size = None
125
312
self.file_id = file_id
313
assert isinstance(file_id, (str, None.__class__)), \
314
'bad type %r for %r' % (type(file_id), file_id)
128
316
self.text_id = text_id
129
317
self.parent_id = parent_id
130
self.text_sha1 = None
131
self.text_size = None
132
if kind == 'directory':
137
raise BzrError("unhandled entry kind %r" % kind)
318
self.symlink_target = None
319
self.reference_revision = None
321
def kind_character(self):
322
"""Return a short kind indicator useful for appending to names."""
323
raise BzrError('unknown kind %r' % self.kind)
325
known_kinds = ('file', 'directory', 'symlink')
327
def _put_in_tar(self, item, tree):
328
"""populate item for stashing in a tar, and return the content stream.
330
If no content is available, return None.
332
raise BzrError("don't know how to export {%s} of kind %r" %
333
(self.file_id, self.kind))
335
def put_on_disk(self, dest, dp, tree):
336
"""Create a representation of self on disk in the prefix dest.
338
This is a template method - implement _put_on_disk in subclasses.
340
fullpath = osutils.pathjoin(dest, dp)
341
self._put_on_disk(fullpath, tree)
342
# mutter(" export {%s} kind %s to %s", self.file_id,
343
# self.kind, fullpath)
345
def _put_on_disk(self, fullpath, tree):
346
"""Put this entry onto disk at fullpath, from tree tree."""
347
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
141
349
def sorted_children(self):
142
l = self.children.items()
350
return sorted(self.children.items())
353
def versionable_kind(kind):
354
return (kind in ('file', 'directory', 'symlink', 'tree-reference'))
356
def check(self, checker, rev_id, inv, tree):
357
"""Check this inventory entry is intact.
359
This is a template method, override _check for kind specific
362
:param checker: Check object providing context for the checks;
363
can be used to find out what parts of the repository have already
365
:param rev_id: Revision id from which this InventoryEntry was loaded.
366
Not necessarily the last-changed revision for this file.
367
:param inv: Inventory from which the entry was loaded.
368
:param tree: RevisionTree for this entry.
370
if self.parent_id is not None:
371
if not inv.has_id(self.parent_id):
372
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
373
% (self.parent_id, rev_id))
374
self._check(checker, rev_id, tree)
376
def _check(self, checker, rev_id, tree):
377
"""Check this inventory entry for kind specific errors."""
378
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
148
other = InventoryEntry(self.file_id, self.name, self.kind,
149
self.parent_id, text_id=self.text_id)
150
other.text_sha1 = self.text_sha1
151
other.text_size = self.text_size
382
"""Clone this inventory entry."""
383
raise NotImplementedError
386
def describe_change(old_entry, new_entry):
387
"""Describe the change between old_entry and this.
389
This smells of being an InterInventoryEntry situation, but as its
390
the first one, we're making it a static method for now.
392
An entry with a different parent, or different name is considered
393
to be renamed. Reparenting is an internal detail.
394
Note that renaming the parent does not trigger a rename for the
397
# TODO: Perhaps return an object rather than just a string
398
if old_entry is new_entry:
399
# also the case of both being None
401
elif old_entry is None:
403
elif new_entry is None:
405
if old_entry.kind != new_entry.kind:
407
text_modified, meta_modified = new_entry.detect_changes(old_entry)
408
if text_modified or meta_modified:
412
# TODO 20060511 (mbp, rbc) factor out 'detect_rename' here.
413
if old_entry.parent_id != new_entry.parent_id:
415
elif old_entry.name != new_entry.name:
419
if renamed and not modified:
420
return InventoryEntry.RENAMED
421
if modified and not renamed:
423
if modified and renamed:
424
return InventoryEntry.MODIFIED_AND_RENAMED
155
427
def __repr__(self):
156
return ("%s(%r, %r, kind=%r, parent_id=%r)"
428
return ("%s(%r, %r, parent_id=%r, revision=%r)"
157
429
% (self.__class__.__name__,
164
def to_element(self):
165
"""Convert to XML element"""
168
e.set('name', self.name)
169
e.set('file_id', self.file_id)
170
e.set('kind', self.kind)
172
if self.text_size != None:
173
e.set('text_size', '%d' % self.text_size)
175
for f in ['text_id', 'text_sha1']:
180
# to be conservative, we don't externalize the root pointers
181
# for now, leaving them as null in the xml form. in a future
182
# version it will be implied by nested elements.
183
if self.parent_id != ROOT_ID:
184
assert isinstance(self.parent_id, basestring)
185
e.set('parent_id', self.parent_id)
192
def from_element(cls, elt):
193
assert elt.tag == 'entry'
195
## original format inventories don't have a parent_id for
196
## nodes in the root directory, but it's cleaner to use one
198
parent_id = elt.get('parent_id')
199
if parent_id == None:
202
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
203
self.text_id = elt.get('text_id')
204
self.text_sha1 = elt.get('text_sha1')
435
def snapshot(self, revision, path, previous_entries,
436
work_tree, commit_builder):
437
"""Make a snapshot of this entry which may or may not have changed.
206
## mutter("read inventoryentry: %r" % (elt.attrib))
208
v = elt.get('text_size')
209
self.text_size = v and int(v)
214
from_element = classmethod(from_element)
216
def __cmp__(self, other):
439
This means that all its fields are populated, that it has its
440
text stored in the text store or weave.
442
:return: True if anything was recorded
444
# cannot be unchanged unless there is only one parent file rev.
445
self._read_tree_state(path, work_tree)
446
if len(previous_entries) == 1:
447
parent_ie = previous_entries.values()[0]
448
if self._unchanged(parent_ie):
449
self.revision = parent_ie.revision
451
self.revision = revision
452
return self._snapshot_text(previous_entries, work_tree, commit_builder)
454
def _snapshot_text(self, file_parents, work_tree, commit_builder):
455
"""Record the 'text' of this entry, whatever form that takes.
457
:return: True if anything was recorded
459
raise NotImplementedError(self._snapshot_text)
461
def __eq__(self, other):
219
462
if not isinstance(other, InventoryEntry):
220
463
return NotImplemented
222
return cmp(self.file_id, other.file_id) \
223
or cmp(self.name, other.name) \
224
or cmp(self.text_sha1, other.text_sha1) \
225
or cmp(self.text_size, other.text_size) \
226
or cmp(self.text_id, other.text_id) \
227
or cmp(self.parent_id, other.parent_id) \
228
or cmp(self.kind, other.kind)
465
return ((self.file_id == other.file_id)
466
and (self.name == other.name)
467
and (other.symlink_target == self.symlink_target)
468
and (self.text_sha1 == other.text_sha1)
469
and (self.text_size == other.text_size)
470
and (self.text_id == other.text_id)
471
and (self.parent_id == other.parent_id)
472
and (self.kind == other.kind)
473
and (self.revision == other.revision)
474
and (self.executable == other.executable)
475
and (self.reference_revision == other.reference_revision)
478
def __ne__(self, other):
479
return not (self == other)
482
raise ValueError('not hashable')
484
def _unchanged(self, previous_ie):
485
"""Has this entry changed relative to previous_ie.
487
This method should be overridden in child classes.
490
# different inv parent
491
if previous_ie.parent_id != self.parent_id:
494
elif previous_ie.name != self.name:
496
elif previous_ie.kind != self.kind:
500
def _read_tree_state(self, path, work_tree):
501
"""Populate fields in the inventory entry from the given tree.
503
Note that this should be modified to be a noop on virtual trees
504
as all entries created there are prepopulated.
506
# TODO: Rather than running this manually, we should check the
507
# working sha1 and other expensive properties when they're
508
# first requested, or preload them if they're already known
509
pass # nothing to do by default
511
def _forget_tree_state(self):
232
515
class RootEntry(InventoryEntry):
517
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
518
'text_id', 'parent_id', 'children', 'executable',
519
'revision', 'symlink_target', 'reference_revision']
521
def _check(self, checker, rev_id, tree):
522
"""See InventoryEntry._check"""
233
524
def __init__(self, file_id):
234
525
self.file_id = file_id
235
526
self.children = {}
236
self.kind = 'root_directory'
527
self.kind = 'directory'
237
528
self.parent_id = None
531
symbol_versioning.warn('RootEntry is deprecated as of bzr 0.10.'
532
' Please use InventoryDirectory instead.',
533
DeprecationWarning, stacklevel=2)
240
def __cmp__(self, other):
535
def __eq__(self, other):
243
536
if not isinstance(other, RootEntry):
244
537
return NotImplemented
245
return cmp(self.file_id, other.file_id) \
246
or cmp(self.children, other.children)
250
class Inventory(XMLMixin):
539
return (self.file_id == other.file_id) \
540
and (self.children == other.children)
543
class InventoryDirectory(InventoryEntry):
544
"""A directory in an inventory."""
546
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
547
'text_id', 'parent_id', 'children', 'executable',
548
'revision', 'symlink_target', 'reference_revision']
550
def _check(self, checker, rev_id, tree):
551
"""See InventoryEntry._check"""
552
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
553
raise BzrCheckError('directory {%s} has text in revision {%s}'
554
% (self.file_id, rev_id))
557
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
558
other.revision = self.revision
559
# note that children are *not* copied; they're pulled across when
563
def __init__(self, file_id, name, parent_id):
564
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
566
self.kind = 'directory'
568
def kind_character(self):
569
"""See InventoryEntry.kind_character."""
572
def _put_in_tar(self, item, tree):
573
"""See InventoryEntry._put_in_tar."""
574
item.type = tarfile.DIRTYPE
581
def _put_on_disk(self, fullpath, tree):
582
"""See InventoryEntry._put_on_disk."""
585
def _snapshot_text(self, file_parents, work_tree, commit_builder):
586
"""See InventoryEntry._snapshot_text."""
587
commit_builder.modified_directory(self.file_id, file_parents)
591
class InventoryFile(InventoryEntry):
592
"""A file in an inventory."""
594
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
595
'text_id', 'parent_id', 'children', 'executable',
596
'revision', 'symlink_target', 'reference_revision']
598
def _check(self, checker, tree_revision_id, tree):
599
"""See InventoryEntry._check"""
600
t = (self.file_id, self.revision)
601
if t in checker.checked_texts:
602
prev_sha = checker.checked_texts[t]
603
if prev_sha != self.text_sha1:
604
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
605
(self.file_id, tree_revision_id))
607
checker.repeated_text_cnt += 1
610
if self.file_id not in checker.checked_weaves:
611
mutter('check weave {%s}', self.file_id)
612
w = tree._get_weave(self.file_id)
613
# Not passing a progress bar, because it creates a new
614
# progress, which overwrites the current progress,
615
# and doesn't look nice
617
checker.checked_weaves[self.file_id] = True
619
w = tree._get_weave(self.file_id)
621
mutter('check version {%s} of {%s}', tree_revision_id, self.file_id)
622
checker.checked_text_cnt += 1
623
# We can't check the length, because Weave doesn't store that
624
# information, and the whole point of looking at the weave's
625
# sha1sum is that we don't have to extract the text.
626
if self.text_sha1 != w.get_sha1(self.revision):
627
raise BzrCheckError('text {%s} version {%s} wrong sha1'
628
% (self.file_id, self.revision))
629
checker.checked_texts[t] = self.text_sha1
632
other = InventoryFile(self.file_id, self.name, self.parent_id)
633
other.executable = self.executable
634
other.text_id = self.text_id
635
other.text_sha1 = self.text_sha1
636
other.text_size = self.text_size
637
other.revision = self.revision
640
def detect_changes(self, old_entry):
641
"""See InventoryEntry.detect_changes."""
642
assert self.text_sha1 is not None
643
assert old_entry.text_sha1 is not None
644
text_modified = (self.text_sha1 != old_entry.text_sha1)
645
meta_modified = (self.executable != old_entry.executable)
646
return text_modified, meta_modified
648
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
649
output_to, reverse=False):
650
"""See InventoryEntry._diff."""
652
from_text = tree.get_file(self.file_id).readlines()
654
to_text = to_tree.get_file(to_entry.file_id).readlines()
658
text_diff(from_label, from_text,
659
to_label, to_text, output_to)
661
text_diff(to_label, to_text,
662
from_label, from_text, output_to)
663
except errors.BinaryFile:
665
label_pair = (to_label, from_label)
667
label_pair = (from_label, to_label)
668
print >> output_to, \
669
("Binary files %s and %s differ" % label_pair).encode('utf8')
672
"""See InventoryEntry.has_text."""
675
def __init__(self, file_id, name, parent_id):
676
super(InventoryFile, self).__init__(file_id, name, parent_id)
679
def kind_character(self):
680
"""See InventoryEntry.kind_character."""
683
def _put_in_tar(self, item, tree):
684
"""See InventoryEntry._put_in_tar."""
685
item.type = tarfile.REGTYPE
686
fileobj = tree.get_file(self.file_id)
687
item.size = self.text_size
688
if tree.is_executable(self.file_id):
694
def _put_on_disk(self, fullpath, tree):
695
"""See InventoryEntry._put_on_disk."""
696
osutils.pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
697
if tree.is_executable(self.file_id):
698
os.chmod(fullpath, 0755)
700
def _read_tree_state(self, path, work_tree):
701
"""See InventoryEntry._read_tree_state."""
702
self.text_sha1 = work_tree.get_file_sha1(self.file_id, path=path)
703
# FIXME: 20050930 probe for the text size when getting sha1
704
# in _read_tree_state
705
self.executable = work_tree.is_executable(self.file_id, path=path)
708
return ("%s(%r, %r, parent_id=%r, sha1=%r, len=%s)"
709
% (self.__class__.__name__,
716
def _forget_tree_state(self):
717
self.text_sha1 = None
719
def snapshot(self, revision, path, previous_entries,
720
work_tree, commit_builder):
721
"""See InventoryEntry.snapshot."""
722
# Note: We use a custom implementation of this method for files
723
# because it's a performance critical part of commit.
725
# If this is the initial commit for this file, we know the sha is
726
# coming later so skip calculating it now (in _read_tree_state())
727
if len(previous_entries) == 0:
728
self.executable = work_tree.is_executable(self.file_id, path=path)
730
self._read_tree_state(path, work_tree)
732
# If nothing is changed from the sole parent, there's nothing to do
733
if len(previous_entries) == 1:
734
parent_ie = previous_entries.values()[0]
735
if self._unchanged(parent_ie):
736
self.revision = parent_ie.revision
739
# Add the file to the repository
740
self.revision = revision
741
def get_content_byte_lines():
742
return work_tree.get_file(self.file_id, path).readlines()
743
self.text_sha1, self.text_size = commit_builder.modified_file_text(
744
self.file_id, previous_entries, get_content_byte_lines,
745
self.text_sha1, self.text_size)
748
def _unchanged(self, previous_ie):
749
"""See InventoryEntry._unchanged."""
750
compatible = super(InventoryFile, self)._unchanged(previous_ie)
751
if self.text_sha1 != previous_ie.text_sha1:
754
# FIXME: 20050930 probe for the text size when getting sha1
755
# in _read_tree_state
756
self.text_size = previous_ie.text_size
757
if self.executable != previous_ie.executable:
762
class InventoryLink(InventoryEntry):
763
"""A file in an inventory."""
765
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
766
'text_id', 'parent_id', 'children', 'executable',
767
'revision', 'symlink_target', 'reference_revision']
769
def _check(self, checker, rev_id, tree):
770
"""See InventoryEntry._check"""
771
if self.text_sha1 is not None or self.text_size is not None or self.text_id is not None:
772
raise BzrCheckError('symlink {%s} has text in revision {%s}'
773
% (self.file_id, rev_id))
774
if self.symlink_target is None:
775
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
776
% (self.file_id, rev_id))
779
other = InventoryLink(self.file_id, self.name, self.parent_id)
780
other.symlink_target = self.symlink_target
781
other.revision = self.revision
784
def detect_changes(self, old_entry):
785
"""See InventoryEntry.detect_changes."""
786
# FIXME: which _modified field should we use ? RBC 20051003
787
text_modified = (self.symlink_target != old_entry.symlink_target)
789
mutter(" symlink target changed")
790
meta_modified = False
791
return text_modified, meta_modified
793
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
794
output_to, reverse=False):
795
"""See InventoryEntry._diff."""
796
from_text = self.symlink_target
797
if to_entry is not None:
798
to_text = to_entry.symlink_target
803
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
806
print >>output_to, '=== target was %r' % self.symlink_target
808
print >>output_to, '=== target is %r' % self.symlink_target
810
def __init__(self, file_id, name, parent_id):
811
super(InventoryLink, self).__init__(file_id, name, parent_id)
812
self.kind = 'symlink'
814
def kind_character(self):
815
"""See InventoryEntry.kind_character."""
818
def _put_in_tar(self, item, tree):
819
"""See InventoryEntry._put_in_tar."""
820
item.type = tarfile.SYMTYPE
824
item.linkname = self.symlink_target
827
def _put_on_disk(self, fullpath, tree):
828
"""See InventoryEntry._put_on_disk."""
830
os.symlink(self.symlink_target, fullpath)
832
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
834
def _read_tree_state(self, path, work_tree):
835
"""See InventoryEntry._read_tree_state."""
836
self.symlink_target = work_tree.get_symlink_target(self.file_id)
838
def _forget_tree_state(self):
839
self.symlink_target = None
841
def _unchanged(self, previous_ie):
842
"""See InventoryEntry._unchanged."""
843
compatible = super(InventoryLink, self)._unchanged(previous_ie)
844
if self.symlink_target != previous_ie.symlink_target:
848
def _snapshot_text(self, file_parents, work_tree, commit_builder):
849
"""See InventoryEntry._snapshot_text."""
850
commit_builder.modified_link(
851
self.file_id, file_parents, self.symlink_target)
855
class TreeReference(InventoryEntry):
857
kind = 'tree-reference'
859
def __init__(self, file_id, name, parent_id, revision=None,
860
reference_revision=None):
861
InventoryEntry.__init__(self, file_id, name, parent_id)
862
self.revision = revision
863
self.reference_revision = reference_revision
866
return TreeReference(self.file_id, self.name, self.parent_id,
867
self.revision, self.reference_revision)
869
def _snapshot_text(self, file_parents, work_tree, commit_builder):
870
commit_builder.modified_reference(self.file_id, file_parents)
873
def _read_tree_state(self, path, work_tree):
874
"""Populate fields in the inventory entry from the given tree.
876
self.reference_revision = work_tree.get_reference_revision(
879
def _forget_tree_state(self):
880
self.reference_revision = None
883
class Inventory(object):
251
884
"""Inventory of versioned files in a tree.
253
886
This describes which file_id is present at each point in the tree,
316
931
The inventory is created with a default root directory, with
319
self.root = RootEntry(ROOT_ID)
934
if root_id is not None:
935
assert root_id.__class__ == str
936
self._set_root(InventoryDirectory(root_id, u'', None))
940
self.revision_id = revision_id
942
def _set_root(self, ie):
320
944
self._byid = {self.root.file_id: self.root}
947
# TODO: jam 20051218 Should copy also copy the revision_id?
948
entries = self.iter_entries()
949
other = Inventory(entries.next()[1].file_id)
950
# copy recursively so we know directories will be added before
951
# their children. There are more efficient ways than this...
952
for path, entry in entries():
953
other.add(entry.copy())
323
956
def __iter__(self):
324
957
return iter(self._byid)
327
959
def __len__(self):
328
960
"""Returns number of entries."""
329
961
return len(self._byid)
332
963
def iter_entries(self, from_dir=None):
333
964
"""Return (path, entry) pairs, in order by name."""
337
elif isinstance(from_dir, basestring):
338
from_dir = self._byid[from_dir]
340
kids = from_dir.children.items()
342
for name, ie in kids:
344
if ie.kind == 'directory':
345
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
yield os.path.join(name, cn), cie
966
if self.root is None:
970
elif isinstance(from_dir, basestring):
971
from_dir = self._byid[from_dir]
973
# unrolling the recursive called changed the time from
974
# 440ms/663ms (inline/total) to 116ms/116ms
975
children = from_dir.children.items()
977
children = collections.deque(children)
978
stack = [(u'', children)]
980
from_dir_relpath, children = stack[-1]
983
name, ie = children.popleft()
985
# we know that from_dir_relpath never ends in a slash
986
# and 'f' doesn't begin with one, we can do a string op, rather
987
# than the checks of pathjoin(), though this means that all paths
989
path = from_dir_relpath + '/' + name
993
if ie.kind != 'directory':
996
# But do this child first
997
new_children = ie.children.items()
999
new_children = collections.deque(new_children)
1000
stack.append((path, new_children))
1001
# Break out of inner loop, so that we start outer loop with child
1004
# if we finished all children, pop it off the stack
1007
def iter_entries_by_dir(self, from_dir=None, specific_file_ids=None):
1008
"""Iterate over the entries in a directory first order.
1010
This returns all entries for a directory before returning
1011
the entries for children of a directory. This is not
1012
lexicographically sorted order, and is a hybrid between
1013
depth-first and breadth-first.
1015
:return: This yields (path, entry) pairs
1017
if specific_file_ids:
1018
safe = osutils.safe_file_id
1019
specific_file_ids = set(safe(fid) for fid in specific_file_ids)
1020
# TODO? Perhaps this should return the from_dir so that the root is
1021
# yielded? or maybe an option?
1022
if from_dir is None:
1023
if self.root is None:
1025
# Optimize a common case
1026
if specific_file_ids is not None and len(specific_file_ids) == 1:
1027
file_id = list(specific_file_ids)[0]
1029
yield self.id2path(file_id), self[file_id]
1031
from_dir = self.root
1032
if (specific_file_ids is None or
1033
self.root.file_id in specific_file_ids):
1034
yield u'', self.root
1035
elif isinstance(from_dir, basestring):
1036
from_dir = self._byid[from_dir]
1038
if specific_file_ids is not None:
1039
# TODO: jam 20070302 This could really be done as a loop rather
1040
# than a bunch of recursive calls.
1043
def add_ancestors(file_id):
1044
if file_id not in byid:
1046
parent_id = byid[file_id].parent_id
1047
if parent_id is None:
1049
if parent_id not in parents:
1050
parents.add(parent_id)
1051
add_ancestors(parent_id)
1052
for file_id in specific_file_ids:
1053
add_ancestors(file_id)
1057
stack = [(u'', from_dir)]
1059
cur_relpath, cur_dir = stack.pop()
1062
for child_name, child_ie in sorted(cur_dir.children.iteritems()):
1064
child_relpath = cur_relpath + child_name
1066
if (specific_file_ids is None or
1067
child_ie.file_id in specific_file_ids):
1068
yield child_relpath, child_ie
1070
if child_ie.kind == 'directory':
1071
if parents is None or child_ie.file_id in parents:
1072
child_dirs.append((child_relpath+'/', child_ie))
1073
stack.extend(reversed(child_dirs))
1075
def make_entry(self, kind, name, parent_id, file_id=None):
1076
"""Simple thunk to bzrlib.inventory.make_entry."""
1077
return make_entry(kind, name, parent_id, file_id)
1080
"""Return list of (path, ie) for all entries except the root.
1082
This may be faster than iter_entries.
1085
def descend(dir_ie, dir_path):
1086
kids = dir_ie.children.items()
1088
for name, ie in kids:
1089
child_path = osutils.pathjoin(dir_path, name)
1090
accum.append((child_path, ie))
1091
if ie.kind == 'directory':
1092
descend(ie, child_path)
1094
descend(self.root, u'')
350
1097
def directories(self):
351
"""Return (path, entry) pairs for all directories.
1098
"""Return (path, entry) pairs for all directories, including the root.
353
def descend(parent_ie):
354
parent_name = parent_ie.name
355
yield parent_name, parent_ie
357
# directory children in sorted order
359
for ie in parent_ie.children.itervalues():
360
if ie.kind == 'directory':
361
dn.append((ie.name, ie))
1101
def descend(parent_ie, parent_path):
1102
accum.append((parent_path, parent_ie))
364
for name, child_ie in dn:
365
for sub_name, sub_ie in descend(child_ie):
366
yield appendpath(parent_name, sub_name), sub_ie
1104
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
368
for name, ie in descend(self.root):
1107
for name, child_ie in kids:
1108
child_path = osutils.pathjoin(parent_path, name)
1109
descend(child_ie, child_path)
1110
descend(self.root, u'')
373
1113
def __contains__(self, file_id):
374
1114
"""True if this entry contains a file with given id.
376
1116
>>> inv = Inventory()
377
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1117
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1118
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
378
1119
>>> '123' in inv
380
1121
>>> '456' in inv
383
return file_id in self._byid
1124
file_id = osutils.safe_file_id(file_id)
1125
return (file_id in self._byid)
386
1127
def __getitem__(self, file_id):
387
1128
"""Return the entry for given file_id.
389
1130
>>> inv = Inventory()
390
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
1131
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
1132
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT', sha1=None, len=None)
391
1133
>>> inv['123123'].name
395
raise BzrError("can't look up file_id None")
1136
file_id = osutils.safe_file_id(file_id)
398
1138
return self._byid[file_id]
399
1139
except KeyError:
400
raise BzrError("file_id {%s} not in inventory" % file_id)
1140
# really we're passing an inventory, not a tree...
1141
raise errors.NoSuchId(self, file_id)
1143
def get_file_kind(self, file_id):
1144
file_id = osutils.safe_file_id(file_id)
1145
return self._byid[file_id].kind
403
1147
def get_child(self, parent_id, filename):
1148
parent_id = osutils.safe_file_id(parent_id)
404
1149
return self[parent_id].children.get(filename)
1151
def _add_child(self, entry):
1152
"""Add an entry to the inventory, without adding it to its parent"""
1153
if entry.file_id in self._byid:
1154
raise BzrError("inventory already contains entry with id {%s}" %
1156
self._byid[entry.file_id] = entry
1157
for child in getattr(entry, 'children', {}).itervalues():
1158
self._add_child(child)
407
1161
def add(self, entry):
408
1162
"""Add entry to inventory.
410
1164
To add a file to a branch ready to be committed, use Branch.add,
1167
Returns the new entry object.
412
1169
if entry.file_id in self._byid:
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
416
parent = self._byid[entry.parent_id]
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
420
if parent.children.has_key(entry.name):
421
bailout("%s is already versioned" %
422
appendpath(self.id2path(parent.file_id), entry.name))
424
self._byid[entry.file_id] = entry
425
parent.children[entry.name] = entry
428
def add_path(self, relpath, kind, file_id=None):
1170
raise errors.DuplicateFileId(entry.file_id,
1171
self._byid[entry.file_id])
1173
if entry.parent_id is None:
1174
assert self.root is None and len(self._byid) == 0
1178
parent = self._byid[entry.parent_id]
1180
raise BzrError("parent_id {%s} not in inventory" %
1183
if entry.name in parent.children:
1184
raise BzrError("%s is already versioned" %
1185
osutils.pathjoin(self.id2path(parent.file_id),
1186
entry.name).encode('utf-8'))
1187
parent.children[entry.name] = entry
1188
return self._add_child(entry)
1190
def add_path(self, relpath, kind, file_id=None, parent_id=None):
429
1191
"""Add entry from a path.
431
The immediate parent must already be versioned"""
432
parts = bzrlib.osutils.splitpath(relpath)
1193
The immediate parent must already be versioned.
1195
Returns the new entry object."""
1197
parts = osutils.splitpath(relpath)
433
1199
if len(parts) == 0:
434
bailout("cannot re-add root of inventory")
437
file_id = bzrlib.branch.gen_file_id(relpath)
439
parent_id = self.path2id(parts[:-1])
440
assert parent_id != None
441
ie = InventoryEntry(file_id, parts[-1],
442
kind=kind, parent_id=parent_id)
1201
file_id = generate_ids.gen_root_id()
1203
file_id = osutils.safe_file_id(file_id)
1204
self.root = InventoryDirectory(file_id, '', None)
1205
self._byid = {self.root.file_id: self.root}
1208
parent_path = parts[:-1]
1209
parent_id = self.path2id(parent_path)
1210
if parent_id is None:
1211
raise errors.NotVersionedError(path=parent_path)
1212
ie = make_entry(kind, parts[-1], parent_id, file_id)
443
1213
return self.add(ie)
446
1215
def __delitem__(self, file_id):
447
1216
"""Remove entry by id.
449
1218
>>> inv = Inventory()
450
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
1219
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
1220
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT', sha1=None, len=None)
451
1221
>>> '123' in inv
453
1223
>>> del inv['123']
454
1224
>>> '123' in inv
1227
file_id = osutils.safe_file_id(file_id)
457
1228
ie = self[file_id]
459
assert self[ie.parent_id].children[ie.name] == ie
1230
assert ie.parent_id is None or \
1231
self[ie.parent_id].children[ie.name] == ie
461
# TODO: Test deleting all children; maybe hoist to a separate
463
if ie.kind == 'directory':
464
for cie in ie.children.values():
465
del self[cie.file_id]
468
1233
del self._byid[file_id]
469
del self[ie.parent_id].children[ie.name]
473
return Set(self._byid)
476
def to_element(self):
477
"""Convert to XML Element"""
478
e = Element('inventory')
480
for path, ie in self.iter_entries():
481
e.append(ie.to_element())
485
def from_element(cls, elt):
486
"""Construct from XML Element
488
>>> inv = Inventory()
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
490
>>> elt = inv.to_element()
491
>>> inv2 = Inventory.from_element(elt)
495
assert elt.tag == 'inventory'
498
o.add(InventoryEntry.from_element(e))
501
from_element = classmethod(from_element)
504
def __cmp__(self, other):
1234
if ie.parent_id is not None:
1235
del self[ie.parent_id].children[ie.name]
1237
def __eq__(self, other):
505
1238
"""Compare two sets by comparing their contents.
507
1240
>>> i1 = Inventory()
508
1241
>>> i2 = Inventory()
511
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1244
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
1245
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
514
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
1248
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
1249
InventoryFile('123', 'foo', parent_id='TREE_ROOT', sha1=None, len=None)
521
1253
if not isinstance(other, Inventory):
522
1254
return NotImplemented
524
if self.id_set() ^ other.id_set():
527
for file_id in self._byid:
528
c = cmp(self[file_id], other[file_id])
1256
return self._byid == other._byid
1258
def __ne__(self, other):
1259
return not self.__eq__(other)
1262
raise ValueError('not hashable')
1264
def _iter_file_id_parents(self, file_id):
1265
"""Yield the parents of file_id up to the root."""
1266
file_id = osutils.safe_file_id(file_id)
1267
while file_id is not None:
1269
ie = self._byid[file_id]
1271
raise errors.NoSuchId(tree=None, file_id=file_id)
1273
file_id = ie.parent_id
534
1275
def get_idpath(self, file_id):
535
1276
"""Return a list of file_ids for the path to an entry.
added
removed
bzrlib/inventory.py
