« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 1185.16.85
- compare with another revision
- download diff
- download tarball
- view history from revision 1185.16.85
- Committer: mbp at sourcefrog
- Date: 2005-10-20 05:32:20 UTC
- mto: (1185.41.1 bzr.sftp) (1534.1.1 integration) (1495.1.4) (1505.1.18 bound-branch) (1508.1.15) (1526.1.3 run_tests_twice_for_i18n) (1540.1.3 bzr.dev (Main development branch)) (1685.1.1 bzr-encoding) (1553.5.1 bzr.dev (Main development branch)) (1608.2.1 bzr.mbp.escape-stores)
- mto: This revision was merged to the branch mainline in revision 1484.
- Revision ID: mbp@sourcefrog.net-20051020053220-98d5bce475edc359
- rules for using destructors
- files added:
- .rsyncexclude
- bzrlib/plugins
- bzrlib/selftest
- bzrlib/store
- bzrlib/transport
- bzrlib/util
- bzrlib/util/effbot
- bzrlib/util/effbot/org
- 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
- elementtree/ => bzrlib/util/elementtree/
- 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
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
34
36
35
37
import bzrlib
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
38
from bzrlib.osutils import (pumpfile, quotefn, splitpath, joinpath,
39
appendpath, sha_strings)
37
40
from bzrlib.trace import mutter
38
39
class InventoryEntry(XMLMixin):
41
from bzrlib.errors import (NotVersionedError, InvalidEntryName,
42
BzrError, BzrCheckError)
43
44
45
class InventoryEntry(object):
40
46
"""Description of a versioned file.
41
47
42
48
An InventoryEntry has the following fields, which are also
43
49
present in the XML inventory-entry element:
44
50
45
* *file_id*
46
* *name*: (only the basename within the directory, must not
47
contain slashes)
48
* *kind*: "directory" or "file"
49
* *directory_id*: (if absent/null means the branch root directory)
50
* *text_sha1*: only for files
51
* *text_size*: in bytes, only for files
52
* *text_id*: identifier for the text version, only for files
53
54
InventoryEntries can also exist inside a WorkingTree
55
inventory, in which case they are not yet bound to a
56
particular revision of the file. In that case the text_sha1,
57
text_size and text_id are absent.
58
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.)
59
74
60
75
>>> i = Inventory()
61
76
>>> i.path2id('')
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
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')
64
82
>>> for j in i.iter_entries():
65
83
... print j
66
84
...
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
85
('src', InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
86
('src/hello.c', InventoryFile('2323', 'hello.c', parent_id='123'))
87
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
70
88
Traceback (most recent call last):
71
89
...
72
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
90
BzrError: inventory already contains entry with id {2323}
91
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
92
InventoryFile('2324', 'bye.c', parent_id='123')
93
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
94
InventoryDirectory('2325', 'wibble', parent_id='123')
75
95
>>> i.path2id('src/wibble')
76
96
'2325'
77
97
>>> '2325' in i
78
98
True
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
99
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
100
InventoryFile('2326', 'wibble.c', parent_id='2325')
80
101
>>> i['2326']
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
>>> for j in i.iter_entries():
83
... print j[0]
84
... assert i.path2id(j[0])
102
InventoryFile('2326', 'wibble.c', parent_id='2325')
103
>>> for path, entry in i.iter_entries():
104
... print path.replace('\\\\', '/') # for win32 os.sep
105
... assert i.path2id(path)
85
106
...
86
107
src
87
108
src/bye.c
88
109
src/hello.c
89
110
src/wibble
90
111
src/wibble/wibble.c
91
>>> i.id2path('2326')
112
>>> i.id2path('2326').replace('\\\\', '/')
92
113
'src/wibble/wibble.c'
93
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.
97
114
"""
98
def __init__(self, file_id, name, kind='file', text_id=None,
99
parent_id=None):
115
116
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
117
'text_id', 'parent_id', 'children', 'executable',
118
'revision']
119
120
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
121
weave_store.add_text(self.file_id, self.revision, new_lines, parents,
122
transaction)
123
124
def detect_changes(self, old_entry):
125
"""Return a (text_modified, meta_modified) from this to old_entry.
126
127
_read_tree_state must have been called on self and old_entry prior to
128
calling detect_changes.
129
"""
130
return False, False
131
132
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
133
output_to, reverse=False):
134
"""Perform a diff from this to to_entry.
135
136
text_diff will be used for textual difference calculation.
137
This is a template method, override _diff in child classes.
138
"""
139
self._read_tree_state(tree.id2path(self.file_id), tree)
140
if to_entry:
141
# cannot diff from one kind to another - you must do a removal
142
# and an addif they do not match.
143
assert self.kind == to_entry.kind
144
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
145
to_tree)
146
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
147
output_to, reverse)
148
149
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
150
output_to, reverse=False):
151
"""Perform a diff between two entries of the same kind."""
152
153
def find_previous_heads(self, previous_inventories, entry_weave):
154
"""Return the revisions and entries that directly preceed this.
155
156
Returned as a map from revision to inventory entry.
157
158
This is a map containing the file revisions in all parents
159
for which the file exists, and its revision is not a parent of
160
any other. If the file is new, the set will be empty.
161
"""
162
def get_ancestors(weave, entry):
163
return set(map(weave.idx_to_name,
164
weave.inclusions([weave.lookup(entry.revision)])))
165
heads = {}
166
head_ancestors = {}
167
for inv in previous_inventories:
168
if self.file_id in inv:
169
ie = inv[self.file_id]
170
assert ie.file_id == self.file_id
171
if ie.revision in heads:
172
# fixup logic, there was a bug in revision updates.
173
# with x bit support.
174
try:
175
if heads[ie.revision].executable != ie.executable:
176
heads[ie.revision].executable = False
177
ie.executable = False
178
except AttributeError:
179
pass
180
assert heads[ie.revision] == ie
181
else:
182
# may want to add it.
183
# may already be covered:
184
already_present = 0 != len(
185
[head for head in heads
186
if ie.revision in head_ancestors[head]])
187
if already_present:
188
# an ancestor of a known head.
189
continue
190
# definately a head:
191
ancestors = get_ancestors(entry_weave, ie)
192
# may knock something else out:
193
check_heads = list(heads.keys())
194
for head in check_heads:
195
if head in ancestors:
196
# this head is not really a head
197
heads.pop(head)
198
head_ancestors[ie.revision] = ancestors
199
heads[ie.revision] = ie
200
return heads
201
202
def get_tar_item(self, root, dp, now, tree):
203
"""Get a tarfile item and a file stream for its content."""
204
item = tarfile.TarInfo(os.path.join(root, dp))
205
# TODO: would be cool to actually set it to the timestamp of the
206
# revision it was last changed
207
item.mtime = now
208
fileobj = self._put_in_tar(item, tree)
209
return item, fileobj
210
211
def has_text(self):
212
"""Return true if the object this entry represents has textual data.
213
214
Note that textual data includes binary content.
215
216
Also note that all entries get weave files created for them.
217
This attribute is primarily used when upgrading from old trees that
218
did not have the weave index for all inventory entries.
219
"""
220
return False
221
222
def __init__(self, file_id, name, parent_id, text_id=None):
100
223
"""Create an InventoryEntry
101
224
102
225
The filename must be a single component, relative to the
103
226
parent directory; it cannot be a whole path or relative name.
104
227
105
>>> e = InventoryEntry('123', 'hello.c')
228
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
106
229
>>> e.name
107
230
'hello.c'
108
231
>>> e.file_id
109
232
'123'
110
>>> e = InventoryEntry('123', 'src/hello.c')
233
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
111
234
Traceback (most recent call last):
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
235
InvalidEntryName: Invalid entry name: src/hello.c
113
236
"""
114
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
117
% name)
118
237
assert isinstance(name, basestring), name
238
if '/' in name or '\\' in name:
239
raise InvalidEntryName(name=name)
240
self.executable = False
241
self.revision = None
242
self.text_sha1 = None
243
self.text_size = None
119
244
self.file_id = file_id
120
245
self.name = name
121
assert kind in ['file', 'directory']
122
self.kind = kind
123
246
self.text_id = text_id
124
247
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
248
self.symlink_target = None
249
250
def kind_character(self):
251
"""Return a short kind indicator useful for appending to names."""
252
raise BzrError('unknown kind %r' % self.kind)
253
254
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
255
256
def _put_in_tar(self, item, tree):
257
"""populate item for stashing in a tar, and return the content stream.
258
259
If no content is available, return None.
260
"""
261
raise BzrError("don't know how to export {%s} of kind %r" %
262
(self.file_id, self.kind))
263
264
def put_on_disk(self, dest, dp, tree):
265
"""Create a representation of self on disk in the prefix dest.
266
267
This is a template method - implement _put_on_disk in subclasses.
268
"""
269
fullpath = appendpath(dest, dp)
270
self._put_on_disk(fullpath, tree)
271
mutter(" export {%s} kind %s to %s" % (self.file_id, self.kind, fullpath))
272
273
def _put_on_disk(self, fullpath, tree):
274
"""Put this entry onto disk at fullpath, from tree tree."""
275
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
276
277
def sorted_children(self):
278
l = self.children.items()
279
l.sort()
280
return l
281
282
@staticmethod
283
def versionable_kind(kind):
284
return kind in ('file', 'directory', 'symlink')
285
286
def check(self, checker, rev_id, inv, tree):
287
"""Check this inventory entry is intact.
288
289
This is a template method, override _check for kind specific
290
tests.
291
"""
292
if self.parent_id != None:
293
if not inv.has_id(self.parent_id):
294
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
295
% (self.parent_id, rev_id))
296
self._check(checker, rev_id, tree)
297
298
def _check(self, checker, rev_id, tree):
299
"""Check this inventory entry for kind specific errors."""
300
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
301
(self.kind, rev_id))
127
302
128
303
129
304
def copy(self):
130
other = InventoryEntry(self.file_id, self.name, self.kind,
131
self.text_id, self.parent_id)
132
other.text_sha1 = self.text_sha1
133
other.text_size = self.text_size
134
return other
305
"""Clone this inventory entry."""
306
raise NotImplementedError
135
307
308
def _get_snapshot_change(self, previous_entries):
309
if len(previous_entries) > 1:
310
return 'merged'
311
elif len(previous_entries) == 0:
312
return 'added'
313
else:
314
return 'modified/renamed/reparented'
136
315
137
316
def __repr__(self):
138
return ("%s(%r, %r, kind=%r, parent_id=%r)"
317
return ("%s(%r, %r, parent_id=%r)"
139
318
% (self.__class__.__name__,
140
319
self.file_id,
141
320
self.name,
142
self.kind,
143
321
self.parent_id))
144
322
145
146
def to_element(self):
147
"""Convert to XML element"""
148
e = Element('entry')
149
150
e.set('name', self.name)
151
e.set('file_id', self.file_id)
152
e.set('kind', self.kind)
153
154
if self.text_size is not None:
155
e.set('text_size', '%d' % self.text_size)
156
157
for f in ['text_id', 'text_sha1', 'parent_id']:
158
v = getattr(self, f)
159
if v is not None:
160
e.set(f, v)
161
162
e.tail = '\n'
163
164
return e
165
166
167
def from_element(cls, elt):
168
assert elt.tag == 'entry'
169
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
170
self.text_id = elt.get('text_id')
171
self.text_sha1 = elt.get('text_sha1')
172
self.parent_id = elt.get('parent_id')
173
174
## mutter("read inventoryentry: %r" % (elt.attrib))
175
176
v = elt.get('text_size')
177
self.text_size = v and int(v)
178
179
return self
180
181
182
from_element = classmethod(from_element)
183
184
def __cmp__(self, other):
185
if self is other:
186
return 0
323
def snapshot(self, revision, path, previous_entries,
324
work_tree, weave_store, transaction):
325
"""Make a snapshot of this entry which may or may not have changed.
326
327
This means that all its fields are populated, that it has its
328
text stored in the text store or weave.
329
"""
330
mutter('new parents of %s are %r', path, previous_entries)
331
self._read_tree_state(path, work_tree)
332
if len(previous_entries) == 1:
333
# cannot be unchanged unless there is only one parent file rev.
334
parent_ie = previous_entries.values()[0]
335
if self._unchanged(parent_ie):
336
mutter("found unchanged entry")
337
self.revision = parent_ie.revision
338
return "unchanged"
339
return self.snapshot_revision(revision, previous_entries,
340
work_tree, weave_store, transaction)
341
342
def snapshot_revision(self, revision, previous_entries, work_tree,
343
weave_store, transaction):
344
"""Record this revision unconditionally."""
345
mutter('new revision for {%s}', self.file_id)
346
self.revision = revision
347
change = self._get_snapshot_change(previous_entries)
348
self._snapshot_text(previous_entries, work_tree, weave_store,
349
transaction)
350
return change
351
352
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
353
"""Record the 'text' of this entry, whatever form that takes.
354
355
This default implementation simply adds an empty text.
356
"""
357
mutter('storing file {%s} in revision {%s}',
358
self.file_id, self.revision)
359
self._add_text_to_weave([], file_parents, weave_store, transaction)
360
361
def __eq__(self, other):
187
362
if not isinstance(other, InventoryEntry):
188
363
return NotImplemented
189
364
190
return cmp(self.file_id, other.file_id) \
191
or cmp(self.name, other.name) \
192
or cmp(self.text_sha1, other.text_sha1) \
193
or cmp(self.text_size, other.text_size) \
194
or cmp(self.text_id, other.text_id) \
195
or cmp(self.parent_id, other.parent_id) \
196
or cmp(self.kind, other.kind)
197
198
199
200
class Inventory(XMLMixin):
365
return ((self.file_id == other.file_id)
366
and (self.name == other.name)
367
and (other.symlink_target == self.symlink_target)
368
and (self.text_sha1 == other.text_sha1)
369
and (self.text_size == other.text_size)
370
and (self.text_id == other.text_id)
371
and (self.parent_id == other.parent_id)
372
and (self.kind == other.kind)
373
and (self.revision == other.revision)
374
and (self.executable == other.executable)
375
)
376
377
def __ne__(self, other):
378
return not (self == other)
379
380
def __hash__(self):
381
raise ValueError('not hashable')
382
383
def _unchanged(self, previous_ie):
384
"""Has this entry changed relative to previous_ie.
385
386
This method should be overriden in child classes.
387
"""
388
compatible = True
389
# different inv parent
390
if previous_ie.parent_id != self.parent_id:
391
compatible = False
392
# renamed
393
elif previous_ie.name != self.name:
394
compatible = False
395
return compatible
396
397
def _read_tree_state(self, path, work_tree):
398
"""Populate fields in the inventory entry from the given tree.
399
400
Note that this should be modified to be a noop on virtual trees
401
as all entries created there are prepopulated.
402
"""
403
# TODO: Rather than running this manually, we should check the
404
# working sha1 and other expensive properties when they're
405
# first requested, or preload them if they're already known
406
pass # nothing to do by default
407
408
409
class RootEntry(InventoryEntry):
410
411
def _check(self, checker, rev_id, tree):
412
"""See InventoryEntry._check"""
413
414
def __init__(self, file_id):
415
self.file_id = file_id
416
self.children = {}
417
self.kind = 'root_directory'
418
self.parent_id = None
419
self.name = ''
420
421
def __eq__(self, other):
422
if not isinstance(other, RootEntry):
423
return NotImplemented
424
425
return (self.file_id == other.file_id) \
426
and (self.children == other.children)
427
428
429
class InventoryDirectory(InventoryEntry):
430
"""A directory in an inventory."""
431
432
def _check(self, checker, rev_id, tree):
433
"""See InventoryEntry._check"""
434
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
435
raise BzrCheckError('directory {%s} has text in revision {%s}'
436
% (self.file_id, rev_id))
437
438
def copy(self):
439
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
440
other.revision = self.revision
441
# note that children are *not* copied; they're pulled across when
442
# others are added
443
return other
444
445
def __init__(self, file_id, name, parent_id):
446
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
447
self.children = {}
448
self.kind = 'directory'
449
450
def kind_character(self):
451
"""See InventoryEntry.kind_character."""
452
return '/'
453
454
def _put_in_tar(self, item, tree):
455
"""See InventoryEntry._put_in_tar."""
456
item.type = tarfile.DIRTYPE
457
fileobj = None
458
item.name += '/'
459
item.size = 0
460
item.mode = 0755
461
return fileobj
462
463
def _put_on_disk(self, fullpath, tree):
464
"""See InventoryEntry._put_on_disk."""
465
os.mkdir(fullpath)
466
467
468
class InventoryFile(InventoryEntry):
469
"""A file in an inventory."""
470
471
def _check(self, checker, rev_id, tree):
472
"""See InventoryEntry._check"""
473
revision = self.revision
474
t = (self.file_id, revision)
475
if t in checker.checked_texts:
476
prev_sha = checker.checked_texts[t]
477
if prev_sha != self.text_sha1:
478
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
479
(self.file_id, rev_id))
480
else:
481
checker.repeated_text_cnt += 1
482
return
483
mutter('check version {%s} of {%s}', rev_id, self.file_id)
484
file_lines = tree.get_file_lines(self.file_id)
485
checker.checked_text_cnt += 1
486
if self.text_size != sum(map(len, file_lines)):
487
raise BzrCheckError('text {%s} wrong size' % self.text_id)
488
if self.text_sha1 != sha_strings(file_lines):
489
raise BzrCheckError('text {%s} wrong sha1' % self.text_id)
490
checker.checked_texts[t] = self.text_sha1
491
492
def copy(self):
493
other = InventoryFile(self.file_id, self.name, self.parent_id)
494
other.executable = self.executable
495
other.text_id = self.text_id
496
other.text_sha1 = self.text_sha1
497
other.text_size = self.text_size
498
other.revision = self.revision
499
return other
500
501
def detect_changes(self, old_entry):
502
"""See InventoryEntry.detect_changes."""
503
assert self.text_sha1 != None
504
assert old_entry.text_sha1 != None
505
text_modified = (self.text_sha1 != old_entry.text_sha1)
506
meta_modified = (self.executable != old_entry.executable)
507
return text_modified, meta_modified
508
509
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
510
output_to, reverse=False):
511
"""See InventoryEntry._diff."""
512
from_text = tree.get_file(self.file_id).readlines()
513
if to_entry:
514
to_text = to_tree.get_file(to_entry.file_id).readlines()
515
else:
516
to_text = []
517
if not reverse:
518
text_diff(from_label, from_text,
519
to_label, to_text, output_to)
520
else:
521
text_diff(to_label, to_text,
522
from_label, from_text, output_to)
523
524
def has_text(self):
525
"""See InventoryEntry.has_text."""
526
return True
527
528
def __init__(self, file_id, name, parent_id):
529
super(InventoryFile, self).__init__(file_id, name, parent_id)
530
self.kind = 'file'
531
532
def kind_character(self):
533
"""See InventoryEntry.kind_character."""
534
return ''
535
536
def _put_in_tar(self, item, tree):
537
"""See InventoryEntry._put_in_tar."""
538
item.type = tarfile.REGTYPE
539
fileobj = tree.get_file(self.file_id)
540
item.size = self.text_size
541
if tree.is_executable(self.file_id):
542
item.mode = 0755
543
else:
544
item.mode = 0644
545
return fileobj
546
547
def _put_on_disk(self, fullpath, tree):
548
"""See InventoryEntry._put_on_disk."""
549
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
550
if tree.is_executable(self.file_id):
551
os.chmod(fullpath, 0755)
552
553
def _read_tree_state(self, path, work_tree):
554
"""See InventoryEntry._read_tree_state."""
555
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
556
self.executable = work_tree.is_executable(self.file_id)
557
558
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
559
"""See InventoryEntry._snapshot_text."""
560
mutter('storing file {%s} in revision {%s}',
561
self.file_id, self.revision)
562
# special case to avoid diffing on renames or
563
# reparenting
564
if (len(file_parents) == 1
565
and self.text_sha1 == file_parents.values()[0].text_sha1
566
and self.text_size == file_parents.values()[0].text_size):
567
previous_ie = file_parents.values()[0]
568
weave_store.add_identical_text(
569
self.file_id, previous_ie.revision,
570
self.revision, file_parents, transaction)
571
else:
572
new_lines = work_tree.get_file(self.file_id).readlines()
573
self._add_text_to_weave(new_lines, file_parents, weave_store,
574
transaction)
575
self.text_sha1 = sha_strings(new_lines)
576
self.text_size = sum(map(len, new_lines))
577
578
579
def _unchanged(self, previous_ie):
580
"""See InventoryEntry._unchanged."""
581
compatible = super(InventoryFile, self)._unchanged(previous_ie)
582
if self.text_sha1 != previous_ie.text_sha1:
583
compatible = False
584
else:
585
# FIXME: 20050930 probe for the text size when getting sha1
586
# in _read_tree_state
587
self.text_size = previous_ie.text_size
588
if self.executable != previous_ie.executable:
589
compatible = False
590
return compatible
591
592
593
class InventoryLink(InventoryEntry):
594
"""A file in an inventory."""
595
596
__slots__ = ['symlink_target']
597
598
def _check(self, checker, rev_id, tree):
599
"""See InventoryEntry._check"""
600
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
601
raise BzrCheckError('symlink {%s} has text in revision {%s}'
602
% (self.file_id, rev_id))
603
if self.symlink_target == None:
604
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
605
% (self.file_id, rev_id))
606
607
def copy(self):
608
other = InventoryLink(self.file_id, self.name, self.parent_id)
609
other.symlink_target = self.symlink_target
610
other.revision = self.revision
611
return other
612
613
def detect_changes(self, old_entry):
614
"""See InventoryEntry.detect_changes."""
615
# FIXME: which _modified field should we use ? RBC 20051003
616
text_modified = (self.symlink_target != old_entry.symlink_target)
617
if text_modified:
618
mutter(" symlink target changed")
619
meta_modified = False
620
return text_modified, meta_modified
621
622
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
623
output_to, reverse=False):
624
"""See InventoryEntry._diff."""
625
from_text = self.symlink_target
626
if to_entry is not None:
627
to_text = to_entry.symlink_target
628
if reverse:
629
temp = from_text
630
from_text = to_text
631
to_text = temp
632
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
633
else:
634
if not reverse:
635
print >>output_to, '=== target was %r' % self.symlink_target
636
else:
637
print >>output_to, '=== target is %r' % self.symlink_target
638
639
def __init__(self, file_id, name, parent_id):
640
super(InventoryLink, self).__init__(file_id, name, parent_id)
641
self.kind = 'symlink'
642
643
def kind_character(self):
644
"""See InventoryEntry.kind_character."""
645
return ''
646
647
def _put_in_tar(self, item, tree):
648
"""See InventoryEntry._put_in_tar."""
649
iterm.type = tarfile.SYMTYPE
650
fileobj = None
651
item.size = 0
652
item.mode = 0755
653
item.linkname = self.symlink_target
654
return fileobj
655
656
def _put_on_disk(self, fullpath, tree):
657
"""See InventoryEntry._put_on_disk."""
658
try:
659
os.symlink(self.symlink_target, fullpath)
660
except OSError,e:
661
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
662
663
def _read_tree_state(self, path, work_tree):
664
"""See InventoryEntry._read_tree_state."""
665
self.symlink_target = work_tree.get_symlink_target(self.file_id)
666
667
def _unchanged(self, previous_ie):
668
"""See InventoryEntry._unchanged."""
669
compatible = super(InventoryLink, self)._unchanged(previous_ie)
670
if self.symlink_target != previous_ie.symlink_target:
671
compatible = False
672
return compatible
673
674
675
class Inventory(object):
201
676
"""Inventory of versioned files in a tree.
202
677
203
An Inventory acts like a set of InventoryEntry items. You can
204
also look files up by their file_id or name.
205
206
May be read from and written to a metadata file in a tree. To
207
manipulate the inventory (for example to add a file), it is read
208
in, modified, and then written back out.
678
This describes which file_id is present at each point in the tree,
679
and possibly the SHA-1 or other information about the file.
680
Entries can be looked up either by path or by file_id.
209
681
210
682
The inventory represents a typical unix file tree, with
211
683
directories containing files and subdirectories. We never store
215
687
returned quickly.
216
688
217
689
InventoryEntry objects must not be modified after they are
218
inserted.
690
inserted, other than through the Inventory API.
219
691
220
692
>>> inv = Inventory()
221
>>> inv.write_xml(sys.stdout)
222
<inventory>
223
</inventory>
224
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
693
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
694
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT')
225
695
>>> inv['123-123'].name
226
696
'hello.c'
227
>>> for file_id in inv: print file_id
228
...
229
123-123
230
697
231
698
May be treated as an iterator or set to look up file ids:
232
699
239
706
240
707
>>> [x[0] for x in inv.iter_entries()]
241
708
['hello.c']
242
243
>>> inv.write_xml(sys.stdout)
244
<inventory>
245
<entry file_id="123-123" kind="file" name="hello.c" />
246
</inventory>
247
709
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
710
>>> inv.add(InventoryFile('123-123', 'hello.c', ROOT_ID))
711
InventoryFile('123-123', 'hello.c', parent_id='TREE_ROOT-12345678-12345678')
248
712
"""
249
250
## TODO: Clear up handling of files in subdirectories; we probably
251
## do want to be able to just look them up by name but this
252
## probably means gradually walking down the path, looking up as we go.
253
254
## TODO: Make sure only canonical filenames are stored.
255
256
## TODO: Do something sensible about the possible collisions on
257
## case-losing filesystems. Perhaps we should just always forbid
258
## such collisions.
259
260
## _tree should probably just be stored as
261
## InventoryEntry._children on each directory.
262
263
def __init__(self):
713
def __init__(self, root_id=ROOT_ID):
264
714
"""Create or read an inventory.
265
715
266
716
If a working directory is specified, the inventory is read
267
717
from there. If the file is specified, read from that. If not,
268
718
the inventory is created empty.
719
720
The inventory is created with a default root directory, with
721
an id of None.
269
722
"""
270
self._byid = dict()
271
272
# _tree is indexed by parent_id; at each level a map from name
273
# to ie. The None entry is the root.
274
self._tree = {None: {}}
723
# We are letting Branch.initialize() create a unique inventory
724
# root id. Rather than generating a random one here.
725
#if root_id is None:
726
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
727
self.root = RootEntry(root_id)
728
self._byid = {self.root.file_id: self.root}
729
730
731
def copy(self):
732
other = Inventory(self.root.file_id)
733
# copy recursively so we know directories will be added before
734
# their children. There are more efficient ways than this...
735
for path, entry in self.iter_entries():
736
if entry == self.root:
737
continue
738
other.add(entry.copy())
739
return other
275
740
276
741
277
742
def __iter__(self):
283
748
return len(self._byid)
284
749
285
750
286
def iter_entries(self, parent_id=None):
751
def iter_entries(self, from_dir=None):
287
752
"""Return (path, entry) pairs, in order by name."""
288
kids = self._tree[parent_id].items()
753
if from_dir == None:
754
assert self.root
755
from_dir = self.root
756
elif isinstance(from_dir, basestring):
757
from_dir = self._byid[from_dir]
758
759
kids = from_dir.children.items()
289
760
kids.sort()
290
761
for name, ie in kids:
291
762
yield name, ie
292
763
if ie.kind == 'directory':
293
for cn, cie in self.iter_entries(parent_id=ie.file_id):
294
yield joinpath([name, cn]), cie
295
296
297
def directories(self, include_root=True):
298
"""Return (path, entry) pairs for all directories.
299
"""
300
if include_root:
301
yield '', None
302
for path, entry in self.iter_entries():
303
if entry.kind == 'directory':
304
yield path, entry
764
for cn, cie in self.iter_entries(from_dir=ie.file_id):
765
yield os.path.join(name, cn), cie
766
767
768
def entries(self):
769
"""Return list of (path, ie) for all entries except the root.
770
771
This may be faster than iter_entries.
772
"""
773
accum = []
774
def descend(dir_ie, dir_path):
775
kids = dir_ie.children.items()
776
kids.sort()
777
for name, ie in kids:
778
child_path = os.path.join(dir_path, name)
779
accum.append((child_path, ie))
780
if ie.kind == 'directory':
781
descend(ie, child_path)
782
783
descend(self.root, '')
784
return accum
785
786
787
def directories(self):
788
"""Return (path, entry) pairs for all directories, including the root.
789
"""
790
accum = []
791
def descend(parent_ie, parent_path):
792
accum.append((parent_path, parent_ie))
793
794
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
795
kids.sort()
796
797
for name, child_ie in kids:
798
child_path = os.path.join(parent_path, name)
799
descend(child_ie, child_path)
800
descend(self.root, '')
801
return accum
305
802
306
803
307
804
308
def children(self, parent_id):
309
"""Return entries that are direct children of parent_id."""
310
return self._tree[parent_id]
311
312
313
314
# TODO: return all paths and entries
315
316
317
805
def __contains__(self, file_id):
318
806
"""True if this entry contains a file with given id.
319
807
320
808
>>> inv = Inventory()
321
>>> inv.add(InventoryEntry('123', 'foo.c'))
809
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
810
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
322
811
>>> '123' in inv
323
812
True
324
813
>>> '456' in inv
331
820
"""Return the entry for given file_id.
332
821
333
822
>>> inv = Inventory()
334
>>> inv.add(InventoryEntry('123123', 'hello.c'))
823
>>> inv.add(InventoryFile('123123', 'hello.c', ROOT_ID))
824
InventoryFile('123123', 'hello.c', parent_id='TREE_ROOT')
335
825
>>> inv['123123'].name
336
826
'hello.c'
337
827
"""
338
return self._byid[file_id]
828
try:
829
return self._byid[file_id]
830
except KeyError:
831
if file_id == None:
832
raise BzrError("can't look up file_id None")
833
else:
834
raise BzrError("file_id {%s} not in inventory" % file_id)
835
836
837
def get_file_kind(self, file_id):
838
return self._byid[file_id].kind
839
840
def get_child(self, parent_id, filename):
841
return self[parent_id].children.get(filename)
339
842
340
843
341
844
def add(self, entry):
342
845
"""Add entry to inventory.
343
846
344
847
To add a file to a branch ready to be committed, use Branch.add,
345
which calls this."""
346
if entry.file_id in self:
347
bailout("inventory already contains entry with id {%s}" % entry.file_id)
348
349
if entry.parent_id != None:
350
if entry.parent_id not in self:
351
bailout("parent_id %s of new entry not found in inventory"
352
% entry.parent_id)
353
354
if self._tree[entry.parent_id].has_key(entry.name):
355
bailout("%s is already versioned"
356
% appendpath(self.id2path(entry.parent_id), entry.name))
848
which calls this.
849
850
Returns the new entry object.
851
"""
852
if entry.file_id in self._byid:
853
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
854
855
if entry.parent_id == ROOT_ID or entry.parent_id is None:
856
entry.parent_id = self.root.file_id
857
858
try:
859
parent = self._byid[entry.parent_id]
860
except KeyError:
861
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
862
863
if parent.children.has_key(entry.name):
864
raise BzrError("%s is already versioned" %
865
appendpath(self.id2path(parent.file_id), entry.name))
357
866
358
867
self._byid[entry.file_id] = entry
359
self._tree[entry.parent_id][entry.name] = entry
360
361
if entry.kind == 'directory':
362
self._tree[entry.file_id] = {}
868
parent.children[entry.name] = entry
869
return entry
363
870
364
871
365
872
def add_path(self, relpath, kind, file_id=None):
366
873
"""Add entry from a path.
367
874
368
The immediate parent must already be versioned"""
875
The immediate parent must already be versioned.
876
877
Returns the new entry object."""
878
from bzrlib.branch import gen_file_id
879
369
880
parts = bzrlib.osutils.splitpath(relpath)
370
881
if len(parts) == 0:
371
bailout("cannot re-add root of inventory")
372
373
if file_id is None:
374
file_id = bzrlib.branch.gen_file_id(relpath)
375
376
parent_id = self.path2id(parts[:-1])
377
ie = InventoryEntry(file_id, parts[-1],
378
kind=kind, parent_id=parent_id)
882
raise BzrError("cannot re-add root of inventory")
883
884
if file_id == None:
885
file_id = gen_file_id(relpath)
886
887
parent_path = parts[:-1]
888
parent_id = self.path2id(parent_path)
889
if parent_id == None:
890
raise NotVersionedError(path=parent_path)
891
if kind == 'directory':
892
ie = InventoryDirectory(file_id, parts[-1], parent_id)
893
elif kind == 'file':
894
ie = InventoryFile(file_id, parts[-1], parent_id)
895
elif kind == 'symlink':
896
ie = InventoryLink(file_id, parts[-1], parent_id)
897
else:
898
raise BzrError("unknown kind %r" % kind)
379
899
return self.add(ie)
380
900
381
901
383
903
"""Remove entry by id.
384
904
385
905
>>> inv = Inventory()
386
>>> inv.add(InventoryEntry('123', 'foo.c'))
906
>>> inv.add(InventoryFile('123', 'foo.c', ROOT_ID))
907
InventoryFile('123', 'foo.c', parent_id='TREE_ROOT')
387
908
>>> '123' in inv
388
909
True
389
910
>>> del inv['123']
392
913
"""
393
914
ie = self[file_id]
394
915
395
assert self._tree[ie.parent_id][ie.name] == ie
916
assert self[ie.parent_id].children[ie.name] == ie
396
917
397
918
# TODO: Test deleting all children; maybe hoist to a separate
398
919
# deltree method?
399
920
if ie.kind == 'directory':
400
for cie in self._tree[file_id].values():
921
for cie in ie.children.values():
401
922
del self[cie.file_id]
402
del self._tree[file_id]
923
del ie.children
403
924
404
925
del self._byid[file_id]
405
del self._tree[ie.parent_id][ie.name]
406
407
408
def id_set(self):
409
return Set(self._byid)
410
411
412
def to_element(self):
413
"""Convert to XML Element"""
414
e = Element('inventory')
415
e.text = '\n'
416
for path, ie in self.iter_entries():
417
e.append(ie.to_element())
418
return e
419
420
421
def from_element(cls, elt):
422
"""Construct from XML Element
423
424
>>> inv = Inventory()
425
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
426
>>> elt = inv.to_element()
427
>>> inv2 = Inventory.from_element(elt)
428
>>> inv2 == inv
429
True
430
"""
431
assert elt.tag == 'inventory'
432
o = cls()
433
for e in elt:
434
o.add(InventoryEntry.from_element(e))
435
return o
436
437
from_element = classmethod(from_element)
438
439
440
def __cmp__(self, other):
926
del self[ie.parent_id].children[ie.name]
927
928
929
def __eq__(self, other):
441
930
"""Compare two sets by comparing their contents.
442
931
443
932
>>> i1 = Inventory()
444
933
>>> i2 = Inventory()
445
934
>>> i1 == i2
446
935
True
447
>>> i1.add(InventoryEntry('123', 'foo'))
936
>>> i1.add(InventoryFile('123', 'foo', ROOT_ID))
937
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
448
938
>>> i1 == i2
449
939
False
450
>>> i2.add(InventoryEntry('123', 'foo'))
940
>>> i2.add(InventoryFile('123', 'foo', ROOT_ID))
941
InventoryFile('123', 'foo', parent_id='TREE_ROOT')
451
942
>>> i1 == i2
452
943
True
453
944
"""
454
if self is other:
455
return 0
456
457
945
if not isinstance(other, Inventory):
458
946
return NotImplemented
459
947
460
if self.id_set() ^ other.id_set():
461
return 1
462
463
for file_id in self._byid:
464
c = cmp(self[file_id], other[file_id])
465
if c: return c
466
467
return 0
468
469
470
def id2path(self, file_id):
471
"""Return as a list the path to file_id."""
948
if len(self._byid) != len(other._byid):
949
# shortcut: obviously not the same
950
return False
951
952
return self._byid == other._byid
953
954
955
def __ne__(self, other):
956
return not self.__eq__(other)
957
958
959
def __hash__(self):
960
raise ValueError('not hashable')
961
962
963
def get_idpath(self, file_id):
964
"""Return a list of file_ids for the path to an entry.
965
966
The list contains one element for each directory followed by
967
the id of the file itself. So the length of the returned list
968
is equal to the depth of the file in the tree, counting the
969
root directory as depth 1.
970
"""
472
971
p = []
473
972
while file_id != None:
474
ie = self[file_id]
475
p = [ie.name] + p
973
try:
974
ie = self._byid[file_id]
975
except KeyError:
976
raise BzrError("file_id {%s} not found in inventory" % file_id)
977
p.insert(0, ie.file_id)
476
978
file_id = ie.parent_id
477
return joinpath(p)
979
return p
980
981
982
def id2path(self, file_id):
983
"""Return as a list the path to file_id.
984
985
>>> i = Inventory()
986
>>> e = i.add(InventoryDirectory('src-id', 'src', ROOT_ID))
987
>>> e = i.add(InventoryFile('foo-id', 'foo.c', parent_id='src-id'))
988
>>> print i.id2path('foo-id').replace(os.sep, '/')
989
src/foo.c
990
"""
991
# get all names, skipping root
992
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
993
return os.sep.join(p)
478
994
479
995
480
996
486
1002
487
1003
This returns the entry of the last component in the path,
488
1004
which may be either a file or a directory.
1005
1006
Returns None iff the path is not found.
489
1007
"""
490
1008
if isinstance(name, types.StringTypes):
491
1009
name = splitpath(name)
492
1010
493
parent_id = None
1011
mutter("lookup path %r" % name)
1012
1013
parent = self.root
494
1014
for f in name:
495
1015
try:
496
cie = self._tree[parent_id][f]
1016
cie = parent.children[f]
497
1017
assert cie.name == f
498
parent_id = cie.file_id
1018
assert cie.parent_id == parent.file_id
1019
parent = cie
499
1020
except KeyError:
500
1021
# or raise an error?
501
1022
return None
502
1023
503
return parent_id
504
505
506
def get_child(self, parent_id, child_name):
507
return self._tree[parent_id].get(child_name)
1024
return parent.file_id
508
1025
509
1026
510
1027
def has_filename(self, names):
512
1029
513
1030
514
1031
def has_id(self, file_id):
515
assert isinstance(file_id, str)
516
1032
return self._byid.has_key(file_id)
517
1033
518
1034
519
520
521
522
if __name__ == '__main__':
523
import doctest, inventory
524
doctest.testmod(inventory)
1035
def rename(self, file_id, new_parent_id, new_name):
1036
"""Move a file within the inventory.
1037
1038
This can change either the name, or the parent, or both.
1039
1040
This does not move the working file."""
1041
if not is_valid_name(new_name):
1042
raise BzrError("not an acceptable filename: %r" % new_name)
1043
1044
new_parent = self._byid[new_parent_id]
1045
if new_name in new_parent.children:
1046
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
1047
1048
new_parent_idpath = self.get_idpath(new_parent_id)
1049
if file_id in new_parent_idpath:
1050
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
1051
% (self.id2path(file_id), self.id2path(new_parent_id)))
1052
1053
file_ie = self._byid[file_id]
1054
old_parent = self._byid[file_ie.parent_id]
1055
1056
# TODO: Don't leave things messed up if this fails
1057
1058
del old_parent.children[file_ie.name]
1059
new_parent.children[new_name] = file_ie
1060
1061
file_ie.name = new_name
1062
file_ie.parent_id = new_parent_id
1063
1064
1065
1066
1067
_NAME_RE = None
1068
1069
def is_valid_name(name):
1070
global _NAME_RE
1071
if _NAME_RE == None:
1072
_NAME_RE = re.compile(r'^[^/\\]+$')
1073
1074
return bool(_NAME_RE.match(name))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1