1
# Copyright (C) 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Read in a bundle stream, and process it into a BundleReader object."""
20
from cStringIO import StringIO
28
from bzrlib.bundle import apply_bundle
29
from bzrlib.errors import (TestamentMismatch, BzrError,
30
MalformedHeader, MalformedPatches, NotABundle)
31
from bzrlib.inventory import (Inventory, InventoryEntry,
32
InventoryDirectory, InventoryFile,
34
from bzrlib.osutils import sha_file, sha_string, pathjoin
35
from bzrlib.revision import Revision, NULL_REVISION
36
from bzrlib.testament import StrictTestament
37
from bzrlib.trace import mutter, warning
38
import bzrlib.transport
39
from bzrlib.tree import Tree
40
import bzrlib.urlutils
41
from bzrlib.xml5 import serializer_v5
44
class RevisionInfo(object):
45
"""Gets filled out for each revision object that is read.
47
def __init__(self, revision_id):
48
self.revision_id = revision_id
54
self.inventory_sha1 = None
56
self.parent_ids = None
59
self.properties = None
60
self.tree_actions = None
63
return pprint.pformat(self.__dict__)
65
def as_revision(self):
66
rev = Revision(revision_id=self.revision_id,
67
committer=self.committer,
68
timestamp=float(self.timestamp),
69
timezone=int(self.timezone),
70
inventory_sha1=self.inventory_sha1,
71
message='\n'.join(self.message))
74
rev.parent_ids.extend(self.parent_ids)
77
for property in self.properties:
78
key_end = property.find(': ')
79
assert key_end is not None
80
key = property[:key_end].encode('utf-8')
81
value = property[key_end+2:].encode('utf-8')
82
rev.properties[key] = value
87
class BundleInfo(object):
88
"""This contains the meta information. Stuff that allows you to
89
recreate the revision or inventory XML.
96
# A list of RevisionInfo objects
99
# The next entries are created during complete_info() and
100
# other post-read functions.
102
# A list of real Revision objects
103
self.real_revisions = []
105
self.timestamp = None
109
return pprint.pformat(self.__dict__)
111
def complete_info(self):
112
"""This makes sure that all information is properly
113
split up, based on the assumptions that can be made
114
when information is missing.
116
from bzrlib.timestamp import unpack_highres_date
117
# Put in all of the guessable information.
118
if not self.timestamp and self.date:
119
self.timestamp, self.timezone = unpack_highres_date(self.date)
121
self.real_revisions = []
122
for rev in self.revisions:
123
if rev.timestamp is None:
124
if rev.date is not None:
125
rev.timestamp, rev.timezone = \
126
unpack_highres_date(rev.date)
128
rev.timestamp = self.timestamp
129
rev.timezone = self.timezone
130
if rev.message is None and self.message:
131
rev.message = self.message
132
if rev.committer is None and self.committer:
133
rev.committer = self.committer
134
self.real_revisions.append(rev.as_revision())
136
def get_base(self, revision):
137
revision_info = self.get_revision_info(revision.revision_id)
138
if revision_info.base_id is not None:
139
if revision_info.base_id == NULL_REVISION:
142
return revision_info.base_id
143
if len(revision.parent_ids) == 0:
144
# There is no base listed, and
145
# the lowest revision doesn't have a parent
146
# so this is probably against the empty tree
147
# and thus base truly is None
150
return revision.parent_ids[-1]
152
def _get_target(self):
153
"""Return the target revision."""
154
if len(self.real_revisions) > 0:
155
return self.real_revisions[0].revision_id
156
elif len(self.revisions) > 0:
157
return self.revisions[0].revision_id
160
target = property(_get_target, doc='The target revision id')
162
def get_revision(self, revision_id):
163
for r in self.real_revisions:
164
if r.revision_id == revision_id:
166
raise KeyError(revision_id)
168
def get_revision_info(self, revision_id):
169
for r in self.revisions:
170
if r.revision_id == revision_id:
172
raise KeyError(revision_id)
174
def revision_tree(self, repository, revision_id, base=None):
175
revision_id = osutils.safe_revision_id(revision_id)
176
revision = self.get_revision(revision_id)
177
base = self.get_base(revision)
178
assert base != revision_id
179
self._validate_references_from_repository(repository)
180
revision_info = self.get_revision_info(revision_id)
181
inventory_revision_id = revision_id
182
bundle_tree = BundleTree(repository.revision_tree(base),
183
inventory_revision_id)
184
self._update_tree(bundle_tree, revision_id)
186
inv = bundle_tree.inventory
187
self._validate_inventory(inv, revision_id)
188
self._validate_revision(inv, revision_id)
192
def _validate_references_from_repository(self, repository):
193
"""Now that we have a repository which should have some of the
194
revisions we care about, go through and validate all of them
199
def add_sha(d, revision_id, sha1):
200
if revision_id is None:
202
raise BzrError('A Null revision should always'
203
'have a null sha1 hash')
206
# This really should have been validated as part
207
# of _validate_revisions but lets do it again
208
if sha1 != d[revision_id]:
209
raise BzrError('** Revision %r referenced with 2 different'
210
' sha hashes %s != %s' % (revision_id,
211
sha1, d[revision_id]))
213
d[revision_id] = sha1
215
# All of the contained revisions were checked
216
# in _validate_revisions
218
for rev_info in self.revisions:
219
checked[rev_info.revision_id] = True
220
add_sha(rev_to_sha, rev_info.revision_id, rev_info.sha1)
222
for (rev, rev_info) in zip(self.real_revisions, self.revisions):
223
add_sha(inv_to_sha, rev_info.revision_id, rev_info.inventory_sha1)
227
for revision_id, sha1 in rev_to_sha.iteritems():
228
if repository.has_revision(revision_id):
229
testament = StrictTestament.from_revision(repository,
231
local_sha1 = self._testament_sha1_from_revision(repository,
233
if sha1 != local_sha1:
234
raise BzrError('sha1 mismatch. For revision id {%s}'
235
'local: %s, bundle: %s' % (revision_id, local_sha1, sha1))
238
elif revision_id not in checked:
239
missing[revision_id] = sha1
241
for inv_id, sha1 in inv_to_sha.iteritems():
242
if repository.has_revision(inv_id):
243
# Note: branch.get_inventory_sha1() just returns the value that
244
# is stored in the revision text, and that value may be out
245
# of date. This is bogus, because that means we aren't
246
# validating the actual text, just that we wrote and read the
247
# string. But for now, what the hell.
248
local_sha1 = repository.get_inventory_sha1(inv_id)
249
if sha1 != local_sha1:
250
raise BzrError('sha1 mismatch. For inventory id {%s}'
251
'local: %s, bundle: %s' %
252
(inv_id, local_sha1, sha1))
257
# I don't know if this is an error yet
258
warning('Not all revision hashes could be validated.'
259
' Unable validate %d hashes' % len(missing))
260
mutter('Verified %d sha hashes for the bundle.' % count)
262
def _validate_inventory(self, inv, revision_id):
263
"""At this point we should have generated the BundleTree,
264
so build up an inventory, and make sure the hashes match.
267
assert inv is not None
269
# Now we should have a complete inventory entry.
270
s = serializer_v5.write_inventory_to_string(inv)
272
# Target revision is the last entry in the real_revisions list
273
rev = self.get_revision(revision_id)
274
assert rev.revision_id == revision_id
275
if sha1 != rev.inventory_sha1:
276
open(',,bogus-inv', 'wb').write(s)
277
warning('Inventory sha hash mismatch for revision %s. %s'
278
' != %s' % (revision_id, sha1, rev.inventory_sha1))
280
def _validate_revision(self, inventory, revision_id):
281
"""Make sure all revision entries match their checksum."""
283
# This is a mapping from each revision id to it's sha hash
286
rev = self.get_revision(revision_id)
287
rev_info = self.get_revision_info(revision_id)
288
assert rev.revision_id == rev_info.revision_id
289
assert rev.revision_id == revision_id
290
sha1 = self._testament_sha1(rev, inventory)
291
if sha1 != rev_info.sha1:
292
raise TestamentMismatch(rev.revision_id, rev_info.sha1, sha1)
293
if rev.revision_id in rev_to_sha1:
294
raise BzrError('Revision {%s} given twice in the list'
296
rev_to_sha1[rev.revision_id] = sha1
298
def _update_tree(self, bundle_tree, revision_id):
299
"""This fills out a BundleTree based on the information
302
:param bundle_tree: A BundleTree to update with the new information.
305
def get_rev_id(last_changed, path, kind):
306
if last_changed is not None:
307
# last_changed will be a Unicode string because of how it was
308
# read. Convert it back to utf8.
309
changed_revision_id = osutils.safe_revision_id(last_changed,
312
changed_revision_id = revision_id
313
bundle_tree.note_last_changed(path, changed_revision_id)
314
return changed_revision_id
316
def extra_info(info, new_path):
319
for info_item in info:
321
name, value = info_item.split(':', 1)
323
raise 'Value %r has no colon' % info_item
324
if name == 'last-changed':
326
elif name == 'executable':
327
assert value in ('yes', 'no'), value
328
val = (value == 'yes')
329
bundle_tree.note_executable(new_path, val)
330
elif name == 'target':
331
bundle_tree.note_target(new_path, value)
332
elif name == 'encoding':
334
return last_changed, encoding
336
def do_patch(path, lines, encoding):
337
if encoding is not None:
338
assert encoding == 'base64'
339
patch = base64.decodestring(''.join(lines))
341
patch = ''.join(lines)
342
bundle_tree.note_patch(path, patch)
344
def renamed(kind, extra, lines):
345
info = extra.split(' // ')
347
raise BzrError('renamed action lines need both a from and to'
350
if info[1].startswith('=> '):
351
new_path = info[1][3:]
355
bundle_tree.note_rename(old_path, new_path)
356
last_modified, encoding = extra_info(info[2:], new_path)
357
revision = get_rev_id(last_modified, new_path, kind)
359
do_patch(new_path, lines, encoding)
361
def removed(kind, extra, lines):
362
info = extra.split(' // ')
364
# TODO: in the future we might allow file ids to be
365
# given for removed entries
366
raise BzrError('removed action lines should only have the path'
369
bundle_tree.note_deletion(path)
371
def added(kind, extra, lines):
372
info = extra.split(' // ')
374
raise BzrError('add action lines require the path and file id'
377
raise BzrError('add action lines have fewer than 5 entries.'
380
if not info[1].startswith('file-id:'):
381
raise BzrError('The file-id should follow the path for an add'
383
# This will be Unicode because of how the stream is read. Turn it
384
# back into a utf8 file_id
385
file_id = osutils.safe_file_id(info[1][8:], warn=False)
387
bundle_tree.note_id(file_id, path, kind)
388
# this will be overridden in extra_info if executable is specified.
389
bundle_tree.note_executable(path, False)
390
last_changed, encoding = extra_info(info[2:], path)
391
revision = get_rev_id(last_changed, path, kind)
392
if kind == 'directory':
394
do_patch(path, lines, encoding)
396
def modified(kind, extra, lines):
397
info = extra.split(' // ')
399
raise BzrError('modified action lines have at least'
400
'the path in them: %r' % extra)
403
last_modified, encoding = extra_info(info[1:], path)
404
revision = get_rev_id(last_modified, path, kind)
406
do_patch(path, lines, encoding)
414
for action_line, lines in \
415
self.get_revision_info(revision_id).tree_actions:
416
first = action_line.find(' ')
418
raise BzrError('Bogus action line'
419
' (no opening space): %r' % action_line)
420
second = action_line.find(' ', first+1)
422
raise BzrError('Bogus action line'
423
' (missing second space): %r' % action_line)
424
action = action_line[:first]
425
kind = action_line[first+1:second]
426
if kind not in ('file', 'directory', 'symlink'):
427
raise BzrError('Bogus action line'
428
' (invalid object kind %r): %r' % (kind, action_line))
429
extra = action_line[second+1:]
431
if action not in valid_actions:
432
raise BzrError('Bogus action line'
433
' (unrecognized action): %r' % action_line)
434
valid_actions[action](kind, extra, lines)
436
def install_revisions(self, target_repo):
437
"""Install revisions and return the target revision"""
438
apply_bundle.install_bundle(target_repo, self)
442
class BundleTree(Tree):
443
def __init__(self, base_tree, revision_id):
444
self.base_tree = base_tree
445
self._renamed = {} # Mapping from old_path => new_path
446
self._renamed_r = {} # new_path => old_path
447
self._new_id = {} # new_path => new_id
448
self._new_id_r = {} # new_id => new_path
449
self._kinds = {} # new_id => kind
450
self._last_changed = {} # new_id => revision_id
451
self._executable = {} # new_id => executable value
453
self._targets = {} # new path => new symlink target
455
self.contents_by_id = True
456
self.revision_id = revision_id
457
self._inventory = None
460
return pprint.pformat(self.__dict__)
462
def note_rename(self, old_path, new_path):
463
"""A file/directory has been renamed from old_path => new_path"""
464
assert new_path not in self._renamed
465
assert old_path not in self._renamed_r
466
self._renamed[new_path] = old_path
467
self._renamed_r[old_path] = new_path
469
def note_id(self, new_id, new_path, kind='file'):
470
"""Files that don't exist in base need a new id."""
471
self._new_id[new_path] = new_id
472
self._new_id_r[new_id] = new_path
473
self._kinds[new_id] = kind
475
def note_last_changed(self, file_id, revision_id):
476
if (file_id in self._last_changed
477
and self._last_changed[file_id] != revision_id):
478
raise BzrError('Mismatched last-changed revision for file_id {%s}'
479
': %s != %s' % (file_id,
480
self._last_changed[file_id],
482
self._last_changed[file_id] = revision_id
484
def note_patch(self, new_path, patch):
485
"""There is a patch for a given filename."""
486
self.patches[new_path] = patch
488
def note_target(self, new_path, target):
489
"""The symlink at the new path has the given target"""
490
self._targets[new_path] = target
492
def note_deletion(self, old_path):
493
"""The file at old_path has been deleted."""
494
self.deleted.append(old_path)
496
def note_executable(self, new_path, executable):
497
self._executable[new_path] = executable
499
def old_path(self, new_path):
500
"""Get the old_path (path in the base_tree) for the file at new_path"""
501
assert new_path[:1] not in ('\\', '/')
502
old_path = self._renamed.get(new_path)
503
if old_path is not None:
505
dirname,basename = os.path.split(new_path)
506
# dirname is not '' doesn't work, because
507
# dirname may be a unicode entry, and is
508
# requires the objects to be identical
510
old_dir = self.old_path(dirname)
514
old_path = pathjoin(old_dir, basename)
517
#If the new path wasn't in renamed, the old one shouldn't be in
519
if old_path in self._renamed_r:
523
def new_path(self, old_path):
524
"""Get the new_path (path in the target_tree) for the file at old_path
527
assert old_path[:1] not in ('\\', '/')
528
new_path = self._renamed_r.get(old_path)
529
if new_path is not None:
531
if new_path in self._renamed:
533
dirname,basename = os.path.split(old_path)
535
new_dir = self.new_path(dirname)
539
new_path = pathjoin(new_dir, basename)
542
#If the old path wasn't in renamed, the new one shouldn't be in
544
if new_path in self._renamed:
548
def path2id(self, path):
549
"""Return the id of the file present at path in the target tree."""
550
file_id = self._new_id.get(path)
551
if file_id is not None:
553
old_path = self.old_path(path)
556
if old_path in self.deleted:
558
if getattr(self.base_tree, 'path2id', None) is not None:
559
return self.base_tree.path2id(old_path)
561
return self.base_tree.inventory.path2id(old_path)
563
def id2path(self, file_id):
564
"""Return the new path in the target tree of the file with id file_id"""
565
path = self._new_id_r.get(file_id)
568
old_path = self.base_tree.id2path(file_id)
571
if old_path in self.deleted:
573
return self.new_path(old_path)
575
def old_contents_id(self, file_id):
576
"""Return the id in the base_tree for the given file_id.
577
Return None if the file did not exist in base.
579
if self.contents_by_id:
580
if self.base_tree.has_id(file_id):
584
new_path = self.id2path(file_id)
585
return self.base_tree.path2id(new_path)
587
def get_file(self, file_id):
588
"""Return a file-like object containing the new contents of the
589
file given by file_id.
591
TODO: It might be nice if this actually generated an entry
592
in the text-store, so that the file contents would
595
base_id = self.old_contents_id(file_id)
596
if (base_id is not None and
597
base_id != self.base_tree.inventory.root.file_id):
598
patch_original = self.base_tree.get_file(base_id)
600
patch_original = None
601
file_patch = self.patches.get(self.id2path(file_id))
602
if file_patch is None:
603
if (patch_original is None and
604
self.get_kind(file_id) == 'directory'):
606
assert patch_original is not None, "None: %s" % file_id
607
return patch_original
609
assert not file_patch.startswith('\\'), \
610
'Malformed patch for %s, %r' % (file_id, file_patch)
611
return patched_file(file_patch, patch_original)
613
def get_symlink_target(self, file_id):
614
new_path = self.id2path(file_id)
616
return self._targets[new_path]
618
return self.base_tree.get_symlink_target(file_id)
620
def get_kind(self, file_id):
621
if file_id in self._kinds:
622
return self._kinds[file_id]
623
return self.base_tree.inventory[file_id].kind
625
def is_executable(self, file_id):
626
path = self.id2path(file_id)
627
if path in self._executable:
628
return self._executable[path]
630
return self.base_tree.inventory[file_id].executable
632
def get_last_changed(self, file_id):
633
path = self.id2path(file_id)
634
if path in self._last_changed:
635
return self._last_changed[path]
636
return self.base_tree.inventory[file_id].revision
638
def get_size_and_sha1(self, file_id):
639
"""Return the size and sha1 hash of the given file id.
640
If the file was not locally modified, this is extracted
641
from the base_tree. Rather than re-reading the file.
643
new_path = self.id2path(file_id)
646
if new_path not in self.patches:
647
# If the entry does not have a patch, then the
648
# contents must be the same as in the base_tree
649
ie = self.base_tree.inventory[file_id]
650
if ie.text_size is None:
651
return ie.text_size, ie.text_sha1
652
return int(ie.text_size), ie.text_sha1
653
fileobj = self.get_file(file_id)
654
content = fileobj.read()
655
return len(content), sha_string(content)
657
def _get_inventory(self):
658
"""Build up the inventory entry for the BundleTree.
660
This need to be called before ever accessing self.inventory
662
from os.path import dirname, basename
664
assert self.base_tree is not None
665
base_inv = self.base_tree.inventory
666
inv = Inventory(None, self.revision_id)
668
def add_entry(file_id):
669
path = self.id2path(file_id)
675
parent_path = dirname(path)
676
parent_id = self.path2id(parent_path)
678
kind = self.get_kind(file_id)
679
revision_id = self.get_last_changed(file_id)
681
name = basename(path)
682
if kind == 'directory':
683
ie = InventoryDirectory(file_id, name, parent_id)
685
ie = InventoryFile(file_id, name, parent_id)
686
ie.executable = self.is_executable(file_id)
687
elif kind == 'symlink':
688
ie = InventoryLink(file_id, name, parent_id)
689
ie.symlink_target = self.get_symlink_target(file_id)
690
ie.revision = revision_id
692
if kind in ('directory', 'symlink'):
693
ie.text_size, ie.text_sha1 = None, None
695
ie.text_size, ie.text_sha1 = self.get_size_and_sha1(file_id)
696
if (ie.text_size is None) and (kind == 'file'):
697
raise BzrError('Got a text_size of None for file_id %r' % file_id)
700
sorted_entries = self.sorted_path_id()
701
for path, file_id in sorted_entries:
706
# Have to overload the inherited inventory property
707
# because _get_inventory is only called in the parent.
708
# Reading the docs, property() objects do not use
709
# overloading, they use the function as it was defined
711
inventory = property(_get_inventory)
714
for path, entry in self.inventory.iter_entries():
717
def sorted_path_id(self):
719
for result in self._new_id.iteritems():
721
for id in self.base_tree:
722
path = self.id2path(id)
725
paths.append((path, id))
730
def patched_file(file_patch, original):
731
"""Produce a file-like object with the patched version of a text"""
732
from bzrlib.patches import iter_patched
733
from bzrlib.iterablefile import IterableFile
735
return IterableFile(())
736
# string.splitlines(True) also splits on '\r', but the iter_patched code
737
# only expects to iterate over '\n' style lines
738
return IterableFile(iter_patched(original,
739
StringIO(file_patch).readlines()))
added
removed
bzrlib/bundle/bundle_data.py
