« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 1544
- compare with another revision
- download diff
- download tarball
- view history from revision 1544
- Committer: Martin Pool
- Date: 2006-01-13 06:31:42 UTC
- Revision ID: mbp@sourcefrog.net-20060113063142-8e706dc1483c69e1
Bump version to 0.8pre
- files added:
- .rsyncexclude
- bzrlib/export
- bzrlib/plugins
- bzrlib/store
- bzrlib/tests
- bzrlib/tests/blackbox
- bzrlib/transport
- bzrlib/ui
- bzrlib/util
- bzrlib/util/configobj
- bzrlib/util/configobj/docs
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- bzrlib/util/elementtree
- bzrlib/util/urlgrabber
- contrib
- contrib/bash
- contrib/emacs
- contrib/zsh
- patches
- tools
- files removed:
- bzrlib/tests.py
- files renamed:
- bzrlib/store.py => bzrlib/store/__init__.py
- files modified:
- .bzrignore
- bzr *
- setup.py *
added
removed
bzrlib/inventory.py
1
#! /usr/bin/env python
2
# -*- coding: UTF-8 -*-
3
1
# (C) 2005 Canonical Ltd
2
#
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
8
7
#
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
13
12
#
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
16
18
"""Inventories map files to their name in a revision."""
19
20
21
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
22
__author__ = "Martin Pool <mbp@canonical.com>"
23
24
import sys, os.path, types
25
from sets import Set
26
27
try:
28
from cElementTree import Element, ElementTree, SubElement
29
except ImportError:
30
from elementtree.ElementTree import Element, ElementTree, SubElement
31
32
from xml import XMLMixin
33
from errors import bailout
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
36
37
class InventoryEntry(XMLMixin):
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.
21
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.
25
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
28
ROOT_ID = "TREE_ROOT"
29
30
31
import os.path
32
import re
33
import sys
34
import tarfile
35
import types
36
37
import bzrlib
38
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
39
pathjoin, sha_strings)
40
from bzrlib.trace import mutter
41
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
42
BzrError, BzrCheckError)
43
44
45
class InventoryEntry(object):
38
46
"""Description of a versioned file.
39
47
40
48
An InventoryEntry has the following fields, which are also
41
49
present in the XML inventory-entry element:
42
50
43
* *file_id*
44
* *name*: (only the basename within the directory, must not
45
contain slashes)
46
* *kind*: "directory" or "file"
47
* *directory_id*: (if absent/null means the branch root directory)
48
* *text_sha1*: only for files
49
* *text_size*: in bytes, only for files
50
* *text_id*: identifier for the text version, only for files
51
52
InventoryEntries can also exist inside a WorkingTree
53
inventory, in which case they are not yet bound to a
54
particular revision of the file. In that case the text_sha1,
55
text_size and text_id are absent.
56
51
file_id
52
53
name
54
(within the parent directory)
55
56
parent_id
57
file_id of the parent directory, or ROOT_ID
58
59
revision
60
the revision_id in which this variation of this file was
61
introduced.
62
63
executable
64
Indicates that this file should be executable on systems
65
that support it.
66
67
text_sha1
68
sha-1 of the text of the file
69
70
text_size
71
size in bytes of the text of the file
72
73
(reading a version 4 tree created a text_id field.)
57
74
58
75
>>> i = Inventory()
59
76
>>> i.path2id('')
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
62
>>> for j in i.iter_entries():
63
... print j
77
'TREE_ROOT'
78
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
79
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
80
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
81
InventoryFile('2323', 'hello.c', parent_id='123')
82
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
83
>>> for ix, j in enumerate(i.iter_entries()):
84
... print (j[0] == shouldbe[ix], j[1])
64
85
...
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
66
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
86
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
87
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
88
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
68
89
Traceback (most recent call last):
69
90
...
70
BzrError: ('inventory already contains entry with id {2323}', [])
71
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
91
BzrError: inventory already contains entry with id {2323}
92
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
93
InventoryFile('2324', 'bye.c', parent_id='123')
94
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
95
InventoryDirectory('2325', 'wibble', parent_id='123')
73
96
>>> i.path2id('src/wibble')
74
97
'2325'
75
98
>>> '2325' in i
76
99
True
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
100
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
101
InventoryFile('2326', 'wibble.c', parent_id='2325')
78
102
>>> i['2326']
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
80
>>> for j in i.iter_entries():
81
... print j[0]
82
... assert i.path2id(j[0])
103
InventoryFile('2326', 'wibble.c', parent_id='2325')
104
>>> for path, entry in i.iter_entries():
105
... print path
106
... assert i.path2id(path)
83
107
...
84
108
src
85
109
src/bye.c
88
112
src/wibble/wibble.c
89
113
>>> i.id2path('2326')
90
114
'src/wibble/wibble.c'
91
92
:todo: Maybe also keep the full path of the entry, and the children?
93
But those depend on its position within a particular inventory, and
94
it would be nice not to need to hold the backpointer here.
95
115
"""
96
def __init__(self, file_id, name, kind='file', text_id=None,
97
parent_id=None):
116
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
118
'text_id', 'parent_id', 'children', 'executable',
119
'revision']
120
121
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
122
weave_store.add_text(self.file_id, self.revision, new_lines, parents,
123
transaction)
124
125
def detect_changes(self, old_entry):
126
"""Return a (text_modified, meta_modified) from this to old_entry.
127
128
_read_tree_state must have been called on self and old_entry prior to
129
calling detect_changes.
130
"""
131
return False, False
132
133
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
134
output_to, reverse=False):
135
"""Perform a diff from this to to_entry.
136
137
text_diff will be used for textual difference calculation.
138
This is a template method, override _diff in child classes.
139
"""
140
self._read_tree_state(tree.id2path(self.file_id), tree)
141
if to_entry:
142
# cannot diff from one kind to another - you must do a removal
143
# and an addif they do not match.
144
assert self.kind == to_entry.kind
145
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
146
to_tree)
147
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
148
output_to, reverse)
149
150
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
151
output_to, reverse=False):
152
"""Perform a diff between two entries of the same kind."""
153
154
def find_previous_heads(self, previous_inventories, entry_weave):
155
"""Return the revisions and entries that directly preceed this.
156
157
Returned as a map from revision to inventory entry.
158
159
This is a map containing the file revisions in all parents
160
for which the file exists, and its revision is not a parent of
161
any other. If the file is new, the set will be empty.
162
"""
163
def get_ancestors(weave, entry):
164
return set(map(weave.idx_to_name,
165
weave.inclusions([weave.lookup(entry.revision)])))
166
heads = {}
167
head_ancestors = {}
168
for inv in previous_inventories:
169
if self.file_id in inv:
170
ie = inv[self.file_id]
171
assert ie.file_id == self.file_id
172
if ie.revision in heads:
173
# fixup logic, there was a bug in revision updates.
174
# with x bit support.
175
try:
176
if heads[ie.revision].executable != ie.executable:
177
heads[ie.revision].executable = False
178
ie.executable = False
179
except AttributeError:
180
pass
181
assert heads[ie.revision] == ie
182
else:
183
# may want to add it.
184
# may already be covered:
185
already_present = 0 != len(
186
[head for head in heads
187
if ie.revision in head_ancestors[head]])
188
if already_present:
189
# an ancestor of a known head.
190
continue
191
# definately a head:
192
ancestors = get_ancestors(entry_weave, ie)
193
# may knock something else out:
194
check_heads = list(heads.keys())
195
for head in check_heads:
196
if head in ancestors:
197
# this head is not really a head
198
heads.pop(head)
199
head_ancestors[ie.revision] = ancestors
200
heads[ie.revision] = ie
201
return heads
202
203
def get_tar_item(self, root, dp, now, tree):
204
"""Get a tarfile item and a file stream for its content."""
205
item = tarfile.TarInfo(pathjoin(root, dp))
206
# TODO: would be cool to actually set it to the timestamp of the
207
# revision it was last changed
208
item.mtime = now
209
fileobj = self._put_in_tar(item, tree)
210
return item, fileobj
211
212
def has_text(self):
213
"""Return true if the object this entry represents has textual data.
214
215
Note that textual data includes binary content.
216
217
Also note that all entries get weave files created for them.
218
This attribute is primarily used when upgrading from old trees that
219
did not have the weave index for all inventory entries.
220
"""
221
return False
222
223
def __init__(self, file_id, name, parent_id, text_id=None):
98
224
"""Create an InventoryEntry
99
225
100
226
The filename must be a single component, relative to the
101
227
parent directory; it cannot be a whole path or relative name.
102
228
103
>>> e = InventoryEntry('123', 'hello.c')
229
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
104
230
>>> e.name
105
231
'hello.c'
106
232
>>> e.file_id
107
233
'123'
108
>>> e = InventoryEntry('123', 'src/hello.c')
234
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
109
235
Traceback (most recent call last):
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
236
InvalidEntryName: Invalid entry name: src/hello.c
111
237
"""
112
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
115
% name)
116
238
assert isinstance(name, basestring), name
239
if '/' in name or '\\' in name:
240
raise InvalidEntryName(name=name)
241
self.executable = False
242
self.revision = None
243
self.text_sha1 = None
244
self.text_size = None
117
245
self.file_id = file_id
118
246
self.name = name
119
assert kind in ['file', 'directory']
120
self.kind = kind
121
247
self.text_id = text_id
122
248
self.parent_id = parent_id
123
self.text_sha1 = None
124
self.text_size = None
249
self.symlink_target = None
250
251
def kind_character(self):
252
"""Return a short kind indicator useful for appending to names."""
253
raise BzrError('unknown kind %r' % self.kind)
254
255
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
256
257
def _put_in_tar(self, item, tree):
258
"""populate item for stashing in a tar, and return the content stream.
259
260
If no content is available, return None.
261
"""
262
raise BzrError("don't know how to export {%s} of kind %r" %
263
(self.file_id, self.kind))
264
265
def put_on_disk(self, dest, dp, tree):
266
"""Create a representation of self on disk in the prefix dest.
267
268
This is a template method - implement _put_on_disk in subclasses.
269
"""
270
fullpath = pathjoin(dest, dp)
271
self._put_on_disk(fullpath, tree)
272
mutter(" export {%s} kind %s to %s", self.file_id,
273
self.kind, fullpath)
274
275
def _put_on_disk(self, fullpath, tree):
276
"""Put this entry onto disk at fullpath, from tree tree."""
277
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
278
279
def sorted_children(self):
280
l = self.children.items()
281
l.sort()
282
return l
283
284
@staticmethod
285
def versionable_kind(kind):
286
return kind in ('file', 'directory', 'symlink')
287
288
def check(self, checker, rev_id, inv, tree):
289
"""Check this inventory entry is intact.
290
291
This is a template method, override _check for kind specific
292
tests.
293
"""
294
if self.parent_id != None:
295
if not inv.has_id(self.parent_id):
296
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
297
% (self.parent_id, rev_id))
298
self._check(checker, rev_id, tree)
299
300
def _check(self, checker, rev_id, tree):
301
"""Check this inventory entry for kind specific errors."""
302
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
303
(self.kind, rev_id))
125
304
126
305
127
306
def copy(self):
128
other = InventoryEntry(self.file_id, self.name, self.kind,
129
self.text_id, self.parent_id)
130
other.text_sha1 = self.text_sha1
131
other.text_size = self.text_size
132
return other
307
"""Clone this inventory entry."""
308
raise NotImplementedError
133
309
310
def _get_snapshot_change(self, previous_entries):
311
if len(previous_entries) > 1:
312
return 'merged'
313
elif len(previous_entries) == 0:
314
return 'added'
315
else:
316
return 'modified/renamed/reparented'
134
317
135
318
def __repr__(self):
136
return ("%s(%r, %r, kind=%r, parent_id=%r)"
319
return ("%s(%r, %r, parent_id=%r)"
137
320
% (self.__class__.__name__,
138
321
self.file_id,
139
322
self.name,
140
self.kind,
141
323
self.parent_id))
142
324
143
144
def to_element(self):
145
"""Convert to XML element"""
146
e = Element('entry')
147
148
e.set('name', self.name)
149
e.set('file_id', self.file_id)
150
e.set('kind', self.kind)
151
152
if self.text_size is not None:
153
e.set('text_size', '%d' % self.text_size)
154
155
for f in ['text_id', 'text_sha1', 'parent_id']:
156
v = getattr(self, f)
157
if v is not None:
158
e.set(f, v)
159
160
e.tail = '\n'
161
162
return e
163
164
165
def from_element(cls, elt):
166
assert elt.tag == 'entry'
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
168
self.text_id = elt.get('text_id')
169
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
171
172
## mutter("read inventoryentry: %r" % (elt.attrib))
173
174
v = elt.get('text_size')
175
self.text_size = v and int(v)
176
177
return self
178
179
180
from_element = classmethod(from_element)
181
182
def __cmp__(self, other):
183
if self is other:
184
return 0
325
def snapshot(self, revision, path, previous_entries,
326
work_tree, weave_store, transaction):
327
"""Make a snapshot of this entry which may or may not have changed.
328
329
This means that all its fields are populated, that it has its
330
text stored in the text store or weave.
331
"""
332
mutter('new parents of %s are %r', path, previous_entries)
333
self._read_tree_state(path, work_tree)
334
if len(previous_entries) == 1:
335
# cannot be unchanged unless there is only one parent file rev.
336
parent_ie = previous_entries.values()[0]
337
if self._unchanged(parent_ie):
338
mutter("found unchanged entry")
339
self.revision = parent_ie.revision
340
return "unchanged"
341
return self.snapshot_revision(revision, previous_entries,
342
work_tree, weave_store, transaction)
343
344
def snapshot_revision(self, revision, previous_entries, work_tree,
345
weave_store, transaction):
346
"""Record this revision unconditionally."""
347
mutter('new revision for {%s}', self.file_id)
348
self.revision = revision
349
change = self._get_snapshot_change(previous_entries)
350
self._snapshot_text(previous_entries, work_tree, weave_store,
351
transaction)
352
return change
353
354
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
355
"""Record the 'text' of this entry, whatever form that takes.
356
357
This default implementation simply adds an empty text.
358
"""
359
mutter('storing file {%s} in revision {%s}',
360
self.file_id, self.revision)
361
self._add_text_to_weave([], file_parents, weave_store, transaction)
362
363
def __eq__(self, other):
185
364
if not isinstance(other, InventoryEntry):
186
365
return NotImplemented
187
366
188
return cmp(self.file_id, other.file_id) \
189
or cmp(self.name, other.name) \
190
or cmp(self.text_sha1, other.text_sha1) \
191
or cmp(self.text_size, other.text_size) \
192
or cmp(self.text_id, other.text_id) \
193
or cmp(self.parent_id, other.parent_id) \
194
or cmp(self.kind, other.kind)
195
196
197
198
class Inventory(XMLMixin):
367
return ((self.file_id == other.file_id)
368
and (self.name == other.name)
369
and (other.symlink_target == self.symlink_target)
370
and (self.text_sha1 == other.text_sha1)
371
and (self.text_size == other.text_size)
372
and (self.text_id == other.text_id)
373
and (self.parent_id == other.parent_id)
374
and (self.kind == other.kind)
375
and (self.revision == other.revision)
376
and (self.executable == other.executable)
377
)
378
379
def __ne__(self, other):
380
return not (self == other)
381
382
def __hash__(self):
383
raise ValueError('not hashable')
384
385
def _unchanged(self, previous_ie):
386
"""Has this entry changed relative to previous_ie.
387
388
This method should be overriden in child classes.
389
"""
390
compatible = True
391
# different inv parent
392
if previous_ie.parent_id != self.parent_id:
393
compatible = False
394
# renamed
395
elif previous_ie.name != self.name:
396
compatible = False
397
return compatible
398
399
def _read_tree_state(self, path, work_tree):
400
"""Populate fields in the inventory entry from the given tree.
401
402
Note that this should be modified to be a noop on virtual trees
403
as all entries created there are prepopulated.
404
"""
405
# TODO: Rather than running this manually, we should check the
406
# working sha1 and other expensive properties when they're
407
# first requested, or preload them if they're already known
408
pass # nothing to do by default
409
410
411
class RootEntry(InventoryEntry):
412
413
def _check(self, checker, rev_id, tree):
414
"""See InventoryEntry._check"""
415
416
def __init__(self, file_id):
417
self.file_id = file_id
418
self.children = {}
419
self.kind = 'root_directory'
420
self.parent_id = None
421
self.name = u''
422
423
def __eq__(self, other):
424
if not isinstance(other, RootEntry):
425
return NotImplemented
426
427
return (self.file_id == other.file_id) \
428
and (self.children == other.children)
429
430
431
class InventoryDirectory(InventoryEntry):
432
"""A directory in an inventory."""
433
434
def _check(self, checker, rev_id, tree):
435
"""See InventoryEntry._check"""
436
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
437
raise BzrCheckError('directory {%s} has text in revision {%s}'
438
% (self.file_id, rev_id))
439
440
def copy(self):
441
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
442
other.revision = self.revision
443
# note that children are *not* copied; they're pulled across when
444
# others are added
445
return other
446
447
def __init__(self, file_id, name, parent_id):
448
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
449
self.children = {}
450
self.kind = 'directory'
451
452
def kind_character(self):
453
"""See InventoryEntry.kind_character."""
454
return '/'
455
456
def _put_in_tar(self, item, tree):
457
"""See InventoryEntry._put_in_tar."""
458
item.type = tarfile.DIRTYPE
459
fileobj = None
460
item.name += '/'
461
item.size = 0
462
item.mode = 0755
463
return fileobj
464
465
def _put_on_disk(self, fullpath, tree):
466
"""See InventoryEntry._put_on_disk."""
467
os.mkdir(fullpath)
468
469
470
class InventoryFile(InventoryEntry):
471
"""A file in an inventory."""
472
473
def _check(self, checker, rev_id, tree):
474
"""See InventoryEntry._check"""
475
revision = self.revision
476
t = (self.file_id, revision)
477
if t in checker.checked_texts:
478
prev_sha = checker.checked_texts[t]
479
if prev_sha != self.text_sha1:
480
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
481
(self.file_id, rev_id))
482
else:
483
checker.repeated_text_cnt += 1
484
return
485
486
if self.file_id not in checker.checked_weaves:
487
mutter('check weave {%s}', self.file_id)
488
w = tree.get_weave(self.file_id)
489
# Not passing a progress bar, because it creates a new
490
# progress, which overwrites the current progress,
491
# and doesn't look nice
492
w.check()
493
checker.checked_weaves[self.file_id] = True
494
else:
495
w = tree.get_weave_prelude(self.file_id)
496
497
mutter('check version {%s} of {%s}', rev_id, self.file_id)
498
checker.checked_text_cnt += 1
499
# We can't check the length, because Weave doesn't store that
500
# information, and the whole point of looking at the weave's
501
# sha1sum is that we don't have to extract the text.
502
if self.text_sha1 != w.get_sha1(self.revision):
503
raise BzrCheckError('text {%s} version {%s} wrong sha1'
504
% (self.file_id, self.revision))
505
checker.checked_texts[t] = self.text_sha1
506
507
def copy(self):
508
other = InventoryFile(self.file_id, self.name, self.parent_id)
509
other.executable = self.executable
510
other.text_id = self.text_id
511
other.text_sha1 = self.text_sha1
512
other.text_size = self.text_size
513
other.revision = self.revision
514
return other
515
516
def detect_changes(self, old_entry):
517
"""See InventoryEntry.detect_changes."""
518
assert self.text_sha1 != None
519
assert old_entry.text_sha1 != None
520
text_modified = (self.text_sha1 != old_entry.text_sha1)
521
meta_modified = (self.executable != old_entry.executable)
522
return text_modified, meta_modified
523
524
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
525
output_to, reverse=False):
526
"""See InventoryEntry._diff."""
527
from_text = tree.get_file(self.file_id).readlines()
528
if to_entry:
529
to_text = to_tree.get_file(to_entry.file_id).readlines()
530
else:
531
to_text = []
532
if not reverse:
533
text_diff(from_label, from_text,
534
to_label, to_text, output_to)
535
else:
536
text_diff(to_label, to_text,
537
from_label, from_text, output_to)
538
539
def has_text(self):
540
"""See InventoryEntry.has_text."""
541
return True
542
543
def __init__(self, file_id, name, parent_id):
544
super(InventoryFile, self).__init__(file_id, name, parent_id)
545
self.kind = 'file'
546
547
def kind_character(self):
548
"""See InventoryEntry.kind_character."""
549
return ''
550
551
def _put_in_tar(self, item, tree):
552
"""See InventoryEntry._put_in_tar."""
553
item.type = tarfile.REGTYPE
554
fileobj = tree.get_file(self.file_id)
555
item.size = self.text_size
556
if tree.is_executable(self.file_id):
557
item.mode = 0755
558
else:
559
item.mode = 0644
560
return fileobj
561
562
def _put_on_disk(self, fullpath, tree):
563
"""See InventoryEntry._put_on_disk."""
564
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
565
if tree.is_executable(self.file_id):
566
os.chmod(fullpath, 0755)
567
568
def _read_tree_state(self, path, work_tree):
569
"""See InventoryEntry._read_tree_state."""
570
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
571
self.executable = work_tree.is_executable(self.file_id)
572
573
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
574
"""See InventoryEntry._snapshot_text."""
575
mutter('storing file {%s} in revision {%s}',
576
self.file_id, self.revision)
577
# special case to avoid diffing on renames or
578
# reparenting
579
if (len(file_parents) == 1
580
and self.text_sha1 == file_parents.values()[0].text_sha1
581
and self.text_size == file_parents.values()[0].text_size):
582
previous_ie = file_parents.values()[0]
583
weave_store.add_identical_text(
584
self.file_id, previous_ie.revision,
585
self.revision, file_parents, transaction)
586
else:
587
new_lines = work_tree.get_file(self.file_id).readlines()
588
self._add_text_to_weave(new_lines, file_parents, weave_store,
589
transaction)
590
self.text_sha1 = sha_strings(new_lines)
591
self.text_size = sum(map(len, new_lines))
592
593
594
def _unchanged(self, previous_ie):
595
"""See InventoryEntry._unchanged."""
596
compatible = super(InventoryFile, self)._unchanged(previous_ie)
597
if self.text_sha1 != previous_ie.text_sha1:
598
compatible = False
599
else:
600
# FIXME: 20050930 probe for the text size when getting sha1
601
# in _read_tree_state
602
self.text_size = previous_ie.text_size
603
if self.executable != previous_ie.executable:
604
compatible = False
605
return compatible
606
607
608
class InventoryLink(InventoryEntry):
609
"""A file in an inventory."""
610
611
__slots__ = ['symlink_target']
612
613
def _check(self, checker, rev_id, tree):
614
"""See InventoryEntry._check"""
615
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
616
raise BzrCheckError('symlink {%s} has text in revision {%s}'
617
% (self.file_id, rev_id))
618
if self.symlink_target == None:
619
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
620
% (self.file_id, rev_id))
621
622
def copy(self):
623
other = InventoryLink(self.file_id, self.name, self.parent_id)
624
other.symlink_target = self.symlink_target
625
other.revision = self.revision
626
return other
627
628
def detect_changes(self, old_entry):
629
"""See InventoryEntry.detect_changes."""
630
# FIXME: which _modified field should we use ? RBC 20051003
631
text_modified = (self.symlink_target != old_entry.symlink_target)
632
if text_modified:
633
mutter(" symlink target changed")
634
meta_modified = False
635
return text_modified, meta_modified
636
637
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
638
output_to, reverse=False):
639
"""See InventoryEntry._diff."""
640
from_text = self.symlink_target
641
if to_entry is not None:
642
to_text = to_entry.symlink_target
643
if reverse:
644
temp = from_text
645
from_text = to_text
646
to_text = temp
647
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
648
else:
649
if not reverse:
650
print >>output_to, '=== target was %r' % self.symlink_target
651
else:
652
print >>output_to, '=== target is %r' % self.symlink_target
653
654
def __init__(self, file_id, name, parent_id):
655
super(InventoryLink, self).__init__(file_id, name, parent_id)
656
self.kind = 'symlink'
657
658
def kind_character(self):
659
"""See InventoryEntry.kind_character."""
660
return ''
661
662
def _put_in_tar(self, item, tree):
663
"""See InventoryEntry._put_in_tar."""
664
item.type = tarfile.SYMTYPE
665
fileobj = None
666
item.size = 0
667
item.mode = 0755
668
item.linkname = self.symlink_target
669
return fileobj
670
671
def _put_on_disk(self, fullpath, tree):
672
"""See InventoryEntry._put_on_disk."""
673
try:
674
os.symlink(self.symlink_target, fullpath)
675
except OSError,e:
676
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
677
678
def _read_tree_state(self, path, work_tree):
679
"""See InventoryEntry._read_tree_state."""
680
self.symlink_target = work_tree.get_symlink_target(self.file_id)
681
682
def _unchanged(self, previous_ie):
683
"""See InventoryEntry._unchanged."""
684
compatible = super(InventoryLink, self)._unchanged(previous_ie)
685
if self.symlink_target != previous_ie.symlink_target:
686
compatible = False
687
return compatible
688
689
690
class Inventory(object):
199
691
"""Inventory of versioned files in a tree.
200
692
201
An Inventory acts like a set of InventoryEntry items. You can
202
also look files up by their file_id or name.
203
204
May be read from and written to a metadata file in a tree. To
205
manipulate the inventory (for example to add a file), it is read
206
in, modified, and then written back out.
693
This describes which file_id is present at each point in the tree,
694
and possibly the SHA-1 or other information about the file.
695
Entries can be looked up either by path or by file_id.
207
696
208
697
The inventory represents a typical unix file tree, with
209
698
directories containing files and subdirectories. We never store
213
702
returned quickly.
214
703
215
704
InventoryEntry objects must not be modified after they are
216
inserted.
705
inserted, other than through the Inventory API.
217
706
218
707
>>> inv = Inventory()
219
>>> inv.write_xml(sys.stdout)
220
<inventory>
221
</inventory>
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
708
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
709
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
223
710
>>> inv['123-123'].name
224
711
'hello.c'
225
>>> for file_id in inv: print file_id
226
...
227
123-123
228
712
229
713
May be treated as an iterator or set to look up file ids:
230
714
237
721
238
722
>>> [x[0] for x in inv.iter_entries()]
239
723
['hello.c']
240
241
>>> inv.write_xml(sys.stdout)
242
<inventory>
243
<entry file_id="123-123" kind="file" name="hello.c" />
244
</inventory>
245
724
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
725
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
726
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
246
727
"""
247
248
## TODO: Clear up handling of files in subdirectories; we probably
249
## do want to be able to just look them up by name but this
250
## probably means gradually walking down the path, looking up as we go.
251
252
## TODO: Make sure only canonical filenames are stored.
253
254
## TODO: Do something sensible about the possible collisions on
255
## case-losing filesystems. Perhaps we should just always forbid
256
## such collisions.
257
258
## _tree should probably just be stored as
259
## InventoryEntry._children on each directory.
260
261
def __init__(self):
728
def __init__(self, root_id=ROOT_ID):
262
729
"""Create or read an inventory.
263
730
264
731
If a working directory is specified, the inventory is read
265
732
from there. If the file is specified, read from that. If not,
266
733
the inventory is created empty.
734
735
The inventory is created with a default root directory, with
736
an id of None.
267
737
"""
268
self._byid = dict()
269
270
# _tree is indexed by parent_id; at each level a map from name
271
# to ie. The None entry is the root.
272
self._tree = {None: {}}
738
# We are letting Branch.initialize() create a unique inventory
739
# root id. Rather than generating a random one here.
740
#if root_id is None:
741
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
742
self.root = RootEntry(root_id)
743
self._byid = {self.root.file_id: self.root}
744
745
746
def copy(self):
747
other = Inventory(self.root.file_id)
748
# copy recursively so we know directories will be added before
749
# their children. There are more efficient ways than this...
750
for path, entry in self.iter_entries():
751
if entry == self.root:
752
continue
753
other.add(entry.copy())
754
return other
273
755
274
756
275
757
def __iter__(self):
281
763
return len(self._byid)
282
764
283
765
284
def iter_entries(self, parent_id=None):
766
def iter_entries(self, from_dir=None):
285
767
"""Return (path, entry) pairs, in order by name."""
286
kids = self._tree[parent_id].items()
768
if from_dir == None:
769
assert self.root
770
from_dir = self.root
771
elif isinstance(from_dir, basestring):
772
from_dir = self._byid[from_dir]
773
774
kids = from_dir.children.items()
287
775
kids.sort()
288
776
for name, ie in kids:
289
777
yield name, ie
290
778
if ie.kind == 'directory':
291
for cn, cie in self.iter_entries(parent_id=ie.file_id):
292
yield joinpath([name, cn]), cie
293
294
295
def directories(self, include_root=True):
296
"""Return (path, entry) pairs for all directories.
297
"""
298
if include_root:
299
yield '', None
300
for path, entry in self.iter_entries():
301
if entry.kind == 'directory':
302
yield path, entry
779
for cn, cie in self.iter_entries(from_dir=ie.file_id):
780
yield pathjoin(name, cn), cie
781
782
783
def entries(self):
784
"""Return list of (path, ie) for all entries except the root.
785
786
This may be faster than iter_entries.
787
"""
788
accum = []
789
def descend(dir_ie, dir_path):
790
kids = dir_ie.children.items()
791
kids.sort()
792
for name, ie in kids:
793
child_path = pathjoin(dir_path, name)
794
accum.append((child_path, ie))
795
if ie.kind == 'directory':
796
descend(ie, child_path)
797
798
descend(self.root, u'')
799
return accum
800
801
802
def directories(self):
803
"""Return (path, entry) pairs for all directories, including the root.
804
"""
805
accum = []
806
def descend(parent_ie, parent_path):
807
accum.append((parent_path, parent_ie))
808
809
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
810
kids.sort()
811
812
for name, child_ie in kids:
813
child_path = pathjoin(parent_path, name)
814
descend(child_ie, child_path)
815
descend(self.root, u'')
816
return accum
303
817
304
818
305
819
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
309
310
311
312
# TODO: return all paths and entries
313
314
315
820
def __contains__(self, file_id):
316
821
"""True if this entry contains a file with given id.
317
822
318
823
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
824
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
825
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
320
826
>>> '123' in inv
321
827
True
322
828
>>> '456' in inv
329
835
"""Return the entry for given file_id.
330
836
331
837
>>> inv = Inventory()
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
838
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
839
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
333
840
>>> inv['123123'].name
334
841
'hello.c'
335
842
"""
336
return self._byid[file_id]
843
try:
844
return self._byid[file_id]
845
except KeyError:
846
if file_id == None:
847
raise BzrError("can't look up file_id None")
848
else:
849
raise BzrError("file_id {%s} not in inventory" % file_id)
850
851
852
def get_file_kind(self, file_id):
853
return self._byid[file_id].kind
854
855
def get_child(self, parent_id, filename):
856
return self[parent_id].children.get(filename)
337
857
338
858
339
859
def add(self, entry):
340
860
"""Add entry to inventory.
341
861
342
862
To add a file to a branch ready to be committed, use Branch.add,
343
which calls this."""
344
if entry.file_id in self:
345
bailout("inventory already contains entry with id {%s}" % entry.file_id)
346
347
if entry.parent_id != None:
348
if entry.parent_id not in self:
349
bailout("parent_id %s of new entry not found in inventory"
350
% entry.parent_id)
351
352
if self._tree[entry.parent_id].has_key(entry.name):
353
bailout("%s is already versioned"
354
% appendpath(self.id2path(entry.parent_id), entry.name))
863
which calls this.
864
865
Returns the new entry object.
866
"""
867
if entry.file_id in self._byid:
868
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
869
870
if entry.parent_id == ROOT_ID or entry.parent_id is None:
871
entry.parent_id = self.root.file_id
872
873
try:
874
parent = self._byid[entry.parent_id]
875
except KeyError:
876
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
877
878
if parent.children.has_key(entry.name):
879
raise BzrError("%s is already versioned" %
880
pathjoin(self.id2path(parent.file_id), entry.name))
355
881
356
882
self._byid[entry.file_id] = entry
357
self._tree[entry.parent_id][entry.name] = entry
358
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
883
parent.children[entry.name] = entry
884
return entry
885
886
887
def add_path(self, relpath, kind, file_id=None):
888
"""Add entry from a path.
889
890
The immediate parent must already be versioned.
891
892
Returns the new entry object."""
893
from bzrlib.workingtree import gen_file_id
894
895
parts = bzrlib.osutils.splitpath(relpath)
896
if len(parts) == 0:
897
raise BzrError("cannot re-add root of inventory")
898
899
if file_id == None:
900
file_id = gen_file_id(relpath)
901
902
parent_path = parts[:-1]
903
parent_id = self.path2id(parent_path)
904
if parent_id == None:
905
raise NotVersionedError(path=parent_path)
906
if kind == 'directory':
907
ie = InventoryDirectory(file_id, parts[-1], parent_id)
908
elif kind == 'file':
909
ie = InventoryFile(file_id, parts[-1], parent_id)
910
elif kind == 'symlink':
911
ie = InventoryLink(file_id, parts[-1], parent_id)
912
else:
913
raise BzrError("unknown kind %r" % kind)
914
return self.add(ie)
361
915
362
916
363
917
def __delitem__(self, file_id):
364
918
"""Remove entry by id.
365
919
366
920
>>> inv = Inventory()
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
921
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
922
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
368
923
>>> '123' in inv
369
924
True
370
925
>>> del inv['123']
373
928
"""
374
929
ie = self[file_id]
375
930
376
assert self._tree[ie.parent_id][ie.name] == ie
931
assert self[ie.parent_id].children[ie.name] == ie
377
932
378
933
# TODO: Test deleting all children; maybe hoist to a separate
379
934
# deltree method?
380
935
if ie.kind == 'directory':
381
for cie in self._tree[file_id].values():
936
for cie in ie.children.values():
382
937
del self[cie.file_id]
383
del self._tree[file_id]
938
del ie.children
384
939
385
940
del self._byid[file_id]
386
del self._tree[ie.parent_id][ie.name]
387
388
389
def id_set(self):
390
return Set(self._byid)
391
392
393
def to_element(self):
394
"""Convert to XML Element"""
395
e = Element('inventory')
396
e.text = '\n'
397
for path, ie in self.iter_entries():
398
e.append(ie.to_element())
399
return e
400
401
402
def from_element(cls, elt):
403
"""Construct from XML Element
404
405
>>> inv = Inventory()
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
407
>>> elt = inv.to_element()
408
>>> inv2 = Inventory.from_element(elt)
409
>>> inv2 == inv
410
True
411
"""
412
assert elt.tag == 'inventory'
413
o = cls()
414
for e in elt:
415
o.add(InventoryEntry.from_element(e))
416
return o
417
418
from_element = classmethod(from_element)
419
420
421
def __cmp__(self, other):
941
del self[ie.parent_id].children[ie.name]
942
943
944
def __eq__(self, other):
422
945
"""Compare two sets by comparing their contents.
423
946
424
947
>>> i1 = Inventory()
425
948
>>> i2 = Inventory()
426
949
>>> i1 == i2
427
950
True
428
>>> i1.add(InventoryEntry('123', 'foo'))
951
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
952
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
429
953
>>> i1 == i2
430
954
False
431
>>> i2.add(InventoryEntry('123', 'foo'))
955
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
956
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
432
957
>>> i1 == i2
433
958
True
434
959
"""
435
if self is other:
436
return 0
437
438
960
if not isinstance(other, Inventory):
439
961
return NotImplemented
440
962
441
if self.id_set() ^ other.id_set():
442
return 1
443
444
for file_id in self._byid:
445
c = cmp(self[file_id], other[file_id])
446
if c: return c
447
448
return 0
449
450
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
963
if len(self._byid) != len(other._byid):
964
# shortcut: obviously not the same
965
return False
966
967
return self._byid == other._byid
968
969
970
def __ne__(self, other):
971
return not self.__eq__(other)
972
973
974
def __hash__(self):
975
raise ValueError('not hashable')
976
977
978
def get_idpath(self, file_id):
979
"""Return a list of file_ids for the path to an entry.
980
981
The list contains one element for each directory followed by
982
the id of the file itself. So the length of the returned list
983
is equal to the depth of the file in the tree, counting the
984
root directory as depth 1.
985
"""
453
986
p = []
454
987
while file_id != None:
455
ie = self[file_id]
456
p = [ie.name] + p
988
try:
989
ie = self._byid[file_id]
990
except KeyError:
991
raise BzrError("file_id {%s} not found in inventory" % file_id)
992
p.insert(0, ie.file_id)
457
993
file_id = ie.parent_id
458
return joinpath(p)
994
return p
995
996
997
def id2path(self, file_id):
998
"""Return as a list the path to file_id.
999
1000
>>> i = Inventory()
1001
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
1002
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
1003
>>> print i.id2path('foo-id')
1004
src/foo.c
1005
"""
1006
# get all names, skipping root
1007
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
1008
if p:
1009
return pathjoin(*p)
1010
else:
1011
return ''
459
1012
460
1013
461
1014
467
1020
468
1021
This returns the entry of the last component in the path,
469
1022
which may be either a file or a directory.
1023
1024
Returns None iff the path is not found.
470
1025
"""
471
assert isinstance(name, types.StringTypes)
472
473
parent_id = None
474
for f in splitpath(name):
1026
if isinstance(name, types.StringTypes):
1027
name = splitpath(name)
1028
1029
mutter("lookup path %r" % name)
1030
1031
parent = self.root
1032
for f in name:
475
1033
try:
476
cie = self._tree[parent_id][f]
1034
cie = parent.children[f]
477
1035
assert cie.name == f
478
parent_id = cie.file_id
1036
assert cie.parent_id == parent.file_id
1037
parent = cie
479
1038
except KeyError:
480
1039
# or raise an error?
481
1040
return None
482
1041
483
return parent_id
484
485
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
1042
return parent.file_id
488
1043
489
1044
490
1045
def has_filename(self, names):
492
1047
493
1048
494
1049
def has_id(self, file_id):
495
assert isinstance(file_id, str)
496
1050
return self._byid.has_key(file_id)
497
1051
498
1052
499
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)
1053
def rename(self, file_id, new_parent_id, new_name):
1054
"""Move a file within the inventory.
1055
1056
This can change either the name, or the parent, or both.
1057
1058
This does not move the working file."""
1059
if not is_valid_name(new_name):
1060
raise BzrError("not an acceptable filename: %r" % new_name)
1061
1062
new_parent = self._byid[new_parent_id]
1063
if new_name in new_parent.children:
1064
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1065
1066
new_parent_idpath = self.get_idpath(new_parent_id)
1067
if file_id in new_parent_idpath:
1068
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1069
% (self.id2path(file_id), self.id2path(new_parent_id)))
1070
1071
file_ie = self._byid[file_id]
1072
old_parent = self._byid[file_ie.parent_id]
1073
1074
# TODO: Don't leave things messed up if this fails
1075
1076
del old_parent.children[file_ie.name]
1077
new_parent.children[new_name] = file_ie
1078
1079
file_ie.name = new_name
1080
file_ie.parent_id = new_parent_id
1081
1082
1083
1084
1085
_NAME_RE = None
1086
1087
def is_valid_name(name):
1088
global _NAME_RE
1089
if _NAME_RE == None:
1090
_NAME_RE = re.compile(r'^[^/\\]+$')
1091
1092
return bool(_NAME_RE.match(name))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1