« back to all changes in this revision
Viewing changes to bzrlib/bundle/read_bundle.py
- browse files at revision 1732.2.5
- compare with another revision
- compare with revision 2221
- download diff
- download tarball
- view history from revision 1732.2.5
- Committer: Martin Pool
- Date: 2006-06-10 23:16:19 UTC
- mfrom: (1759 +trunk)
- mto: This revision was merged to the branch mainline in revision 1761.
- Revision ID: mbp@sourcefrog.net-20060610231619-05b997deeb005d02
[merge] bzr.dev
- files added:
- bzrlib/benchmarks/bench_bench.py
- bzrlib/bundle
- bzrlib/bundle/old
- bzrlib/bundle/serializer
- bzrlib/tests/test_patches_data
- files removed:
- bzrlib/util/configobj/validate.py
- files modified:
- BRANCH.TODO
added
removed
bzrlib/bundle/read_bundle.py
1
#!/usr/bin/env python
2
"""\
3
Read in a bundle stream, and process it into a BundleReader object.
4
"""
5
6
import base64
7
from cStringIO import StringIO
8
import os
9
import pprint
10
11
from bzrlib.errors import (TestamentMismatch, BzrError,
12
MalformedHeader, MalformedPatches, NotABundle)
13
from bzrlib.bundle.common import get_header, header_str
14
from bzrlib.inventory import (Inventory, InventoryEntry,
15
InventoryDirectory, InventoryFile,
16
InventoryLink)
17
from bzrlib.osutils import sha_file, sha_string
18
from bzrlib.revision import Revision, NULL_REVISION
19
from bzrlib.testament import StrictTestament
20
from bzrlib.trace import mutter, warning
21
from bzrlib.tree import Tree
22
from bzrlib.xml5 import serializer_v5
23
24
25
class RevisionInfo(object):
26
"""Gets filled out for each revision object that is read.
27
"""
28
def __init__(self, revision_id):
29
self.revision_id = revision_id
30
self.sha1 = None
31
self.committer = None
32
self.date = None
33
self.timestamp = None
34
self.timezone = None
35
self.inventory_sha1 = None
36
37
self.parent_ids = None
38
self.base_id = None
39
self.message = None
40
self.properties = None
41
self.tree_actions = None
42
43
def __str__(self):
44
return pprint.pformat(self.__dict__)
45
46
def as_revision(self):
47
rev = Revision(revision_id=self.revision_id,
48
committer=self.committer,
49
timestamp=float(self.timestamp),
50
timezone=int(self.timezone),
51
inventory_sha1=self.inventory_sha1,
52
message='\n'.join(self.message))
53
54
if self.parent_ids:
55
rev.parent_ids.extend(self.parent_ids)
56
57
if self.properties:
58
for property in self.properties:
59
key_end = property.find(': ')
60
assert key_end is not None
61
key = property[:key_end].encode('utf-8')
62
value = property[key_end+2:].encode('utf-8')
63
rev.properties[key] = value
64
65
return rev
66
67
68
class BundleInfo(object):
69
"""This contains the meta information. Stuff that allows you to
70
recreate the revision or inventory XML.
71
"""
72
def __init__(self):
73
self.committer = None
74
self.date = None
75
self.message = None
76
77
# A list of RevisionInfo objects
78
self.revisions = []
79
80
# The next entries are created during complete_info() and
81
# other post-read functions.
82
83
# A list of real Revision objects
84
self.real_revisions = []
85
86
self.timestamp = None
87
self.timezone = None
88
89
def __str__(self):
90
return pprint.pformat(self.__dict__)
91
92
def complete_info(self):
93
"""This makes sure that all information is properly
94
split up, based on the assumptions that can be made
95
when information is missing.
96
"""
97
from bzrlib.bundle.common import unpack_highres_date
98
# Put in all of the guessable information.
99
if not self.timestamp and self.date:
100
self.timestamp, self.timezone = unpack_highres_date(self.date)
101
102
self.real_revisions = []
103
for rev in self.revisions:
104
if rev.timestamp is None:
105
if rev.date is not None:
106
rev.timestamp, rev.timezone = \
107
unpack_highres_date(rev.date)
108
else:
109
rev.timestamp = self.timestamp
110
rev.timezone = self.timezone
111
if rev.message is None and self.message:
112
rev.message = self.message
113
if rev.committer is None and self.committer:
114
rev.committer = self.committer
115
self.real_revisions.append(rev.as_revision())
116
117
def get_base(self, revision):
118
revision_info = self.get_revision_info(revision.revision_id)
119
if revision_info.base_id is not None:
120
if revision_info.base_id == NULL_REVISION:
121
return None
122
else:
123
return revision_info.base_id
124
if len(revision.parent_ids) == 0:
125
# There is no base listed, and
126
# the lowest revision doesn't have a parent
127
# so this is probably against the empty tree
128
# and thus base truly is None
129
return None
130
else:
131
return revision.parent_ids[-1]
132
133
def _get_target(self):
134
"""Return the target revision."""
135
if len(self.real_revisions) > 0:
136
return self.real_revisions[0].revision_id
137
elif len(self.revisions) > 0:
138
return self.revisions[0].revision_id
139
return None
140
141
target = property(_get_target, doc='The target revision id')
142
143
def get_revision(self, revision_id):
144
for r in self.real_revisions:
145
if r.revision_id == revision_id:
146
return r
147
raise KeyError(revision_id)
148
149
def get_revision_info(self, revision_id):
150
for r in self.revisions:
151
if r.revision_id == revision_id:
152
return r
153
raise KeyError(revision_id)
154
155
156
class BundleReader(object):
157
"""This class reads in a bundle from a file, and returns
158
a Bundle object, which can then be applied against a tree.
159
"""
160
def __init__(self, from_file):
161
"""Read in the bundle from the file.
162
163
:param from_file: A file-like object (must have iterator support).
164
"""
165
object.__init__(self)
166
self.from_file = iter(from_file)
167
self._next_line = None
168
169
self.info = BundleInfo()
170
# We put the actual inventory ids in the footer, so that the patch
171
# is easier to read for humans.
172
# Unfortunately, that means we need to read everything before we
173
# can create a proper bundle.
174
self._read()
175
self._validate()
176
177
def _read(self):
178
self._read_header()
179
while self._next_line is not None:
180
self._read_revision_header()
181
if self._next_line is None:
182
break
183
self._read_patches()
184
self._read_footer()
185
186
def _validate(self):
187
"""Make sure that the information read in makes sense
188
and passes appropriate checksums.
189
"""
190
# Fill in all the missing blanks for the revisions
191
# and generate the real_revisions list.
192
self.info.complete_info()
193
194
def _validate_revision(self, inventory, revision_id):
195
"""Make sure all revision entries match their checksum."""
196
197
# This is a mapping from each revision id to it's sha hash
198
rev_to_sha1 = {}
199
200
rev = self.info.get_revision(revision_id)
201
rev_info = self.info.get_revision_info(revision_id)
202
assert rev.revision_id == rev_info.revision_id
203
assert rev.revision_id == revision_id
204
sha1 = StrictTestament(rev, inventory).as_sha1()
205
if sha1 != rev_info.sha1:
206
raise TestamentMismatch(rev.revision_id, rev_info.sha1, sha1)
207
if rev_to_sha1.has_key(rev.revision_id):
208
raise BzrError('Revision {%s} given twice in the list'
209
% (rev.revision_id))
210
rev_to_sha1[rev.revision_id] = sha1
211
212
def _validate_references_from_repository(self, repository):
213
"""Now that we have a repository which should have some of the
214
revisions we care about, go through and validate all of them
215
that we can.
216
"""
217
rev_to_sha = {}
218
inv_to_sha = {}
219
def add_sha(d, revision_id, sha1):
220
if revision_id is None:
221
if sha1 is not None:
222
raise BzrError('A Null revision should always'
223
'have a null sha1 hash')
224
return
225
if revision_id in d:
226
# This really should have been validated as part
227
# of _validate_revisions but lets do it again
228
if sha1 != d[revision_id]:
229
raise BzrError('** Revision %r referenced with 2 different'
230
' sha hashes %s != %s' % (revision_id,
231
sha1, d[revision_id]))
232
else:
233
d[revision_id] = sha1
234
235
# All of the contained revisions were checked
236
# in _validate_revisions
237
checked = {}
238
for rev_info in self.info.revisions:
239
checked[rev_info.revision_id] = True
240
add_sha(rev_to_sha, rev_info.revision_id, rev_info.sha1)
241
242
for (rev, rev_info) in zip(self.info.real_revisions, self.info.revisions):
243
add_sha(inv_to_sha, rev_info.revision_id, rev_info.inventory_sha1)
244
245
count = 0
246
missing = {}
247
for revision_id, sha1 in rev_to_sha.iteritems():
248
if repository.has_revision(revision_id):
249
testament = StrictTestament.from_revision(repository,
250
revision_id)
251
local_sha1 = testament.as_sha1()
252
if sha1 != local_sha1:
253
raise BzrError('sha1 mismatch. For revision id {%s}'
254
'local: %s, bundle: %s' % (revision_id, local_sha1, sha1))
255
else:
256
count += 1
257
elif revision_id not in checked:
258
missing[revision_id] = sha1
259
260
for inv_id, sha1 in inv_to_sha.iteritems():
261
if repository.has_revision(inv_id):
262
# Note: branch.get_inventory_sha1() just returns the value that
263
# is stored in the revision text, and that value may be out
264
# of date. This is bogus, because that means we aren't
265
# validating the actual text, just that we wrote and read the
266
# string. But for now, what the hell.
267
local_sha1 = repository.get_inventory_sha1(inv_id)
268
if sha1 != local_sha1:
269
raise BzrError('sha1 mismatch. For inventory id {%s}'
270
'local: %s, bundle: %s' %
271
(inv_id, local_sha1, sha1))
272
else:
273
count += 1
274
275
if len(missing) > 0:
276
# I don't know if this is an error yet
277
warning('Not all revision hashes could be validated.'
278
' Unable validate %d hashes' % len(missing))
279
mutter('Verified %d sha hashes for the bundle.' % count)
280
281
def _validate_inventory(self, inv, revision_id):
282
"""At this point we should have generated the BundleTree,
283
so build up an inventory, and make sure the hashes match.
284
"""
285
286
assert inv is not None
287
288
# Now we should have a complete inventory entry.
289
s = serializer_v5.write_inventory_to_string(inv)
290
sha1 = sha_string(s)
291
# Target revision is the last entry in the real_revisions list
292
rev = self.info.get_revision(revision_id)
293
assert rev.revision_id == revision_id
294
if sha1 != rev.inventory_sha1:
295
open(',,bogus-inv', 'wb').write(s)
296
warning('Inventory sha hash mismatch for revision %s. %s'
297
' != %s' % (revision_id, sha1, rev.inventory_sha1))
298
299
def get_bundle(self, repository):
300
"""Return the meta information, and a Bundle tree which can
301
be used to populate the local stores and working tree, respectively.
302
"""
303
return self.info, self.revision_tree(repository, self.info.target)
304
305
def revision_tree(self, repository, revision_id, base=None):
306
revision = self.info.get_revision(revision_id)
307
base = self.info.get_base(revision)
308
assert base != revision_id
309
self._validate_references_from_repository(repository)
310
revision_info = self.info.get_revision_info(revision_id)
311
inventory_revision_id = revision_id
312
bundle_tree = BundleTree(repository.revision_tree(base),
313
inventory_revision_id)
314
self._update_tree(bundle_tree, revision_id)
315
316
inv = bundle_tree.inventory
317
self._validate_inventory(inv, revision_id)
318
self._validate_revision(inv, revision_id)
319
320
return bundle_tree
321
322
def _next(self):
323
"""yield the next line, but secretly
324
keep 1 extra line for peeking.
325
"""
326
for line in self.from_file:
327
last = self._next_line
328
self._next_line = line
329
if last is not None:
330
#mutter('yielding line: %r' % last)
331
yield last
332
last = self._next_line
333
self._next_line = None
334
#mutter('yielding line: %r' % last)
335
yield last
336
337
def _read_header(self):
338
"""Read the bzr header"""
339
header = get_header()
340
found = False
341
for line in self._next():
342
if found:
343
# not all mailers will keep trailing whitespace
344
if line == '#\n':
345
line = '# \n'
346
if (not line.startswith('# ') or not line.endswith('\n')
347
or line[2:-1].decode('utf-8') != header[0]):
348
raise MalformedHeader('Found a header, but it'
349
' was improperly formatted')
350
header.pop(0) # We read this line.
351
if not header:
352
break # We found everything.
353
elif (line.startswith('#') and line.endswith('\n')):
354
line = line[1:-1].strip().decode('utf-8')
355
if line[:len(header_str)] == header_str:
356
if line == header[0]:
357
found = True
358
else:
359
raise MalformedHeader('Found what looks like'
360
' a header, but did not match')
361
header.pop(0)
362
else:
363
raise NotABundle('Did not find an opening header')
364
365
def _read_revision_header(self):
366
self.info.revisions.append(RevisionInfo(None))
367
for line in self._next():
368
# The bzr header is terminated with a blank line
369
# which does not start with '#'
370
if line is None or line == '\n':
371
break
372
self._handle_next(line)
373
374
def _read_next_entry(self, line, indent=1):
375
"""Read in a key-value pair
376
"""
377
if not line.startswith('#'):
378
raise MalformedHeader('Bzr header did not start with #')
379
line = line[1:-1].decode('utf-8') # Remove the '#' and '\n'
380
if line[:indent] == ' '*indent:
381
line = line[indent:]
382
if not line:
383
return None, None# Ignore blank lines
384
385
loc = line.find(': ')
386
if loc != -1:
387
key = line[:loc]
388
value = line[loc+2:]
389
if not value:
390
value = self._read_many(indent=indent+2)
391
elif line[-1:] == ':':
392
key = line[:-1]
393
value = self._read_many(indent=indent+2)
394
else:
395
raise MalformedHeader('While looking for key: value pairs,'
396
' did not find the colon %r' % (line))
397
398
key = key.replace(' ', '_')
399
#mutter('found %s: %s' % (key, value))
400
return key, value
401
402
def _handle_next(self, line):
403
if line is None:
404
return
405
key, value = self._read_next_entry(line, indent=1)
406
mutter('_handle_next %r => %r' % (key, value))
407
if key is None:
408
return
409
410
revision_info = self.info.revisions[-1]
411
if hasattr(revision_info, key):
412
if getattr(revision_info, key) is None:
413
setattr(revision_info, key, value)
414
else:
415
raise MalformedHeader('Duplicated Key: %s' % key)
416
else:
417
# What do we do with a key we don't recognize
418
raise MalformedHeader('Unknown Key: "%s"' % key)
419
420
def _read_many(self, indent):
421
"""If a line ends with no entry, that means that it should be
422
followed with multiple lines of values.
423
424
This detects the end of the list, because it will be a line that
425
does not start properly indented.
426
"""
427
values = []
428
start = '#' + (' '*indent)
429
430
if self._next_line is None or self._next_line[:len(start)] != start:
431
return values
432
433
for line in self._next():
434
values.append(line[len(start):-1].decode('utf-8'))
435
if self._next_line is None or self._next_line[:len(start)] != start:
436
break
437
return values
438
439
def _read_one_patch(self):
440
"""Read in one patch, return the complete patch, along with
441
the next line.
442
443
:return: action, lines, do_continue
444
"""
445
#mutter('_read_one_patch: %r' % self._next_line)
446
# Peek and see if there are no patches
447
if self._next_line is None or self._next_line.startswith('#'):
448
return None, [], False
449
450
first = True
451
lines = []
452
for line in self._next():
453
if first:
454
if not line.startswith('==='):
455
raise MalformedPatches('The first line of all patches'
456
' should be a bzr meta line "==="'
457
': %r' % line)
458
action = line[4:-1].decode('utf-8')
459
elif line.startswith('... '):
460
action += line[len('... '):-1].decode('utf-8')
461
462
if (self._next_line is not None and
463
self._next_line.startswith('===')):
464
return action, lines, True
465
elif self._next_line is None or self._next_line.startswith('#'):
466
return action, lines, False
467
468
if first:
469
first = False
470
elif not line.startswith('... '):
471
lines.append(line)
472
473
return action, lines, False
474
475
def _read_patches(self):
476
do_continue = True
477
revision_actions = []
478
while do_continue:
479
action, lines, do_continue = self._read_one_patch()
480
if action is not None:
481
revision_actions.append((action, lines))
482
assert self.info.revisions[-1].tree_actions is None
483
self.info.revisions[-1].tree_actions = revision_actions
484
485
def _read_footer(self):
486
"""Read the rest of the meta information.
487
488
:param first_line: The previous step iterates past what it
489
can handle. That extra line is given here.
490
"""
491
for line in self._next():
492
self._handle_next(line)
493
if not self._next_line.startswith('#'):
494
self._next().next()
495
break
496
if self._next_line is None:
497
break
498
499
def _update_tree(self, bundle_tree, revision_id):
500
"""This fills out a BundleTree based on the information
501
that was read in.
502
503
:param bundle_tree: A BundleTree to update with the new information.
504
"""
505
506
def get_rev_id(last_changed, path, kind):
507
if last_changed is not None:
508
changed_revision_id = last_changed.decode('utf-8')
509
else:
510
changed_revision_id = revision_id
511
bundle_tree.note_last_changed(path, changed_revision_id)
512
return changed_revision_id
513
514
def extra_info(info, new_path):
515
last_changed = None
516
encoding = None
517
for info_item in info:
518
try:
519
name, value = info_item.split(':', 1)
520
except ValueError:
521
raise 'Value %r has no colon' % info_item
522
if name == 'last-changed':
523
last_changed = value
524
elif name == 'executable':
525
assert value in ('yes', 'no'), value
526
val = (value == 'yes')
527
bundle_tree.note_executable(new_path, val)
528
elif name == 'target':
529
bundle_tree.note_target(new_path, value)
530
elif name == 'encoding':
531
encoding = value
532
return last_changed, encoding
533
534
def do_patch(path, lines, encoding):
535
if encoding is not None:
536
assert encoding == 'base64'
537
patch = base64.decodestring(''.join(lines))
538
else:
539
patch = ''.join(lines)
540
bundle_tree.note_patch(path, patch)
541
542
def renamed(kind, extra, lines):
543
info = extra.split(' // ')
544
if len(info) < 2:
545
raise BzrError('renamed action lines need both a from and to'
546
': %r' % extra)
547
old_path = info[0]
548
if info[1].startswith('=> '):
549
new_path = info[1][3:]
550
else:
551
new_path = info[1]
552
553
bundle_tree.note_rename(old_path, new_path)
554
last_modified, encoding = extra_info(info[2:], new_path)
555
revision = get_rev_id(last_modified, new_path, kind)
556
if lines:
557
do_patch(new_path, lines, encoding)
558
559
def removed(kind, extra, lines):
560
info = extra.split(' // ')
561
if len(info) > 1:
562
# TODO: in the future we might allow file ids to be
563
# given for removed entries
564
raise BzrError('removed action lines should only have the path'
565
': %r' % extra)
566
path = info[0]
567
bundle_tree.note_deletion(path)
568
569
def added(kind, extra, lines):
570
info = extra.split(' // ')
571
if len(info) <= 1:
572
raise BzrError('add action lines require the path and file id'
573
': %r' % extra)
574
elif len(info) > 5:
575
raise BzrError('add action lines have fewer than 5 entries.'
576
': %r' % extra)
577
path = info[0]
578
if not info[1].startswith('file-id:'):
579
raise BzrError('The file-id should follow the path for an add'
580
': %r' % extra)
581
file_id = info[1][8:]
582
583
bundle_tree.note_id(file_id, path, kind)
584
# this will be overridden in extra_info if executable is specified.
585
bundle_tree.note_executable(path, False)
586
last_changed, encoding = extra_info(info[2:], path)
587
revision = get_rev_id(last_changed, path, kind)
588
if kind == 'directory':
589
return
590
do_patch(path, lines, encoding)
591
592
def modified(kind, extra, lines):
593
info = extra.split(' // ')
594
if len(info) < 1:
595
raise BzrError('modified action lines have at least'
596
'the path in them: %r' % extra)
597
path = info[0]
598
599
last_modified, encoding = extra_info(info[1:], path)
600
revision = get_rev_id(last_modified, path, kind)
601
if lines:
602
do_patch(path, lines, encoding)
603
604
valid_actions = {
605
'renamed':renamed,
606
'removed':removed,
607
'added':added,
608
'modified':modified
609
}
610
for action_line, lines in \
611
self.info.get_revision_info(revision_id).tree_actions:
612
first = action_line.find(' ')
613
if first == -1:
614
raise BzrError('Bogus action line'
615
' (no opening space): %r' % action_line)
616
second = action_line.find(' ', first+1)
617
if second == -1:
618
raise BzrError('Bogus action line'
619
' (missing second space): %r' % action_line)
620
action = action_line[:first]
621
kind = action_line[first+1:second]
622
if kind not in ('file', 'directory', 'symlink'):
623
raise BzrError('Bogus action line'
624
' (invalid object kind %r): %r' % (kind, action_line))
625
extra = action_line[second+1:]
626
627
if action not in valid_actions:
628
raise BzrError('Bogus action line'
629
' (unrecognized action): %r' % action_line)
630
valid_actions[action](kind, extra, lines)
631
632
633
class BundleTree(Tree):
634
def __init__(self, base_tree, revision_id):
635
self.base_tree = base_tree
636
self._renamed = {} # Mapping from old_path => new_path
637
self._renamed_r = {} # new_path => old_path
638
self._new_id = {} # new_path => new_id
639
self._new_id_r = {} # new_id => new_path
640
self._kinds = {} # new_id => kind
641
self._last_changed = {} # new_id => revision_id
642
self._executable = {} # new_id => executable value
643
self.patches = {}
644
self._targets = {} # new path => new symlink target
645
self.deleted = []
646
self.contents_by_id = True
647
self.revision_id = revision_id
648
self._inventory = None
649
650
def __str__(self):
651
return pprint.pformat(self.__dict__)
652
653
def note_rename(self, old_path, new_path):
654
"""A file/directory has been renamed from old_path => new_path"""
655
assert not self._renamed.has_key(new_path)
656
assert not self._renamed_r.has_key(old_path)
657
self._renamed[new_path] = old_path
658
self._renamed_r[old_path] = new_path
659
660
def note_id(self, new_id, new_path, kind='file'):
661
"""Files that don't exist in base need a new id."""
662
self._new_id[new_path] = new_id
663
self._new_id_r[new_id] = new_path
664
self._kinds[new_id] = kind
665
666
def note_last_changed(self, file_id, revision_id):
667
if (self._last_changed.has_key(file_id)
668
and self._last_changed[file_id] != revision_id):
669
raise BzrError('Mismatched last-changed revision for file_id {%s}'
670
': %s != %s' % (file_id,
671
self._last_changed[file_id],
672
revision_id))
673
self._last_changed[file_id] = revision_id
674
675
def note_patch(self, new_path, patch):
676
"""There is a patch for a given filename."""
677
self.patches[new_path] = patch
678
679
def note_target(self, new_path, target):
680
"""The symlink at the new path has the given target"""
681
self._targets[new_path] = target
682
683
def note_deletion(self, old_path):
684
"""The file at old_path has been deleted."""
685
self.deleted.append(old_path)
686
687
def note_executable(self, new_path, executable):
688
self._executable[new_path] = executable
689
690
def old_path(self, new_path):
691
"""Get the old_path (path in the base_tree) for the file at new_path"""
692
assert new_path[:1] not in ('\\', '/')
693
old_path = self._renamed.get(new_path)
694
if old_path is not None:
695
return old_path
696
dirname,basename = os.path.split(new_path)
697
# dirname is not '' doesn't work, because
698
# dirname may be a unicode entry, and is
699
# requires the objects to be identical
700
if dirname != '':
701
old_dir = self.old_path(dirname)
702
if old_dir is None:
703
old_path = None
704
else:
705
old_path = os.path.join(old_dir, basename)
706
else:
707
old_path = new_path
708
#If the new path wasn't in renamed, the old one shouldn't be in
709
#renamed_r
710
if self._renamed_r.has_key(old_path):
711
return None
712
return old_path
713
714
def new_path(self, old_path):
715
"""Get the new_path (path in the target_tree) for the file at old_path
716
in the base tree.
717
"""
718
assert old_path[:1] not in ('\\', '/')
719
new_path = self._renamed_r.get(old_path)
720
if new_path is not None:
721
return new_path
722
if self._renamed.has_key(new_path):
723
return None
724
dirname,basename = os.path.split(old_path)
725
if dirname != '':
726
new_dir = self.new_path(dirname)
727
if new_dir is None:
728
new_path = None
729
else:
730
new_path = os.path.join(new_dir, basename)
731
else:
732
new_path = old_path
733
#If the old path wasn't in renamed, the new one shouldn't be in
734
#renamed_r
735
if self._renamed.has_key(new_path):
736
return None
737
return new_path
738
739
def path2id(self, path):
740
"""Return the id of the file present at path in the target tree."""
741
file_id = self._new_id.get(path)
742
if file_id is not None:
743
return file_id
744
old_path = self.old_path(path)
745
if old_path is None:
746
return None
747
if old_path in self.deleted:
748
return None
749
if hasattr(self.base_tree, 'path2id'):
750
return self.base_tree.path2id(old_path)
751
else:
752
return self.base_tree.inventory.path2id(old_path)
753
754
def id2path(self, file_id):
755
"""Return the new path in the target tree of the file with id file_id"""
756
path = self._new_id_r.get(file_id)
757
if path is not None:
758
return path
759
old_path = self.base_tree.id2path(file_id)
760
if old_path is None:
761
return None
762
if old_path in self.deleted:
763
return None
764
return self.new_path(old_path)
765
766
def old_contents_id(self, file_id):
767
"""Return the id in the base_tree for the given file_id.
768
Return None if the file did not exist in base.
769
"""
770
if self.contents_by_id:
771
if self.base_tree.has_id(file_id):
772
return file_id
773
else:
774
return None
775
new_path = self.id2path(file_id)
776
return self.base_tree.path2id(new_path)
777
778
def get_file(self, file_id):
779
"""Return a file-like object containing the new contents of the
780
file given by file_id.
781
782
TODO: It might be nice if this actually generated an entry
783
in the text-store, so that the file contents would
784
then be cached.
785
"""
786
base_id = self.old_contents_id(file_id)
787
if base_id is not None:
788
patch_original = self.base_tree.get_file(base_id)
789
else:
790
patch_original = None
791
file_patch = self.patches.get(self.id2path(file_id))
792
if file_patch is None:
793
if (patch_original is None and
794
self.get_kind(file_id) == 'directory'):
795
return StringIO()
796
assert patch_original is not None, "None: %s" % file_id
797
return patch_original
798
799
assert not file_patch.startswith('\\'), \
800
'Malformed patch for %s, %r' % (file_id, file_patch)
801
return patched_file(file_patch, patch_original)
802
803
def get_symlink_target(self, file_id):
804
new_path = self.id2path(file_id)
805
try:
806
return self._targets[new_path]
807
except KeyError:
808
return self.base_tree.get_symlink_target(file_id)
809
810
def get_kind(self, file_id):
811
if file_id in self._kinds:
812
return self._kinds[file_id]
813
return self.base_tree.inventory[file_id].kind
814
815
def is_executable(self, file_id):
816
path = self.id2path(file_id)
817
if path in self._executable:
818
return self._executable[path]
819
else:
820
return self.base_tree.inventory[file_id].executable
821
822
def get_last_changed(self, file_id):
823
path = self.id2path(file_id)
824
if path in self._last_changed:
825
return self._last_changed[path]
826
return self.base_tree.inventory[file_id].revision
827
828
def get_size_and_sha1(self, file_id):
829
"""Return the size and sha1 hash of the given file id.
830
If the file was not locally modified, this is extracted
831
from the base_tree. Rather than re-reading the file.
832
"""
833
new_path = self.id2path(file_id)
834
if new_path is None:
835
return None, None
836
if new_path not in self.patches:
837
# If the entry does not have a patch, then the
838
# contents must be the same as in the base_tree
839
ie = self.base_tree.inventory[file_id]
840
if ie.text_size is None:
841
return ie.text_size, ie.text_sha1
842
return int(ie.text_size), ie.text_sha1
843
fileobj = self.get_file(file_id)
844
content = fileobj.read()
845
return len(content), sha_string(content)
846
847
def _get_inventory(self):
848
"""Build up the inventory entry for the BundleTree.
849
850
This need to be called before ever accessing self.inventory
851
"""
852
from os.path import dirname, basename
853
854
assert self.base_tree is not None
855
base_inv = self.base_tree.inventory
856
root_id = base_inv.root.file_id
857
try:
858
# New inventories have a unique root_id
859
inv = Inventory(root_id, self.revision_id)
860
except TypeError:
861
inv = Inventory(revision_id=self.revision_id)
862
863
def add_entry(file_id):
864
path = self.id2path(file_id)
865
if path is None:
866
return
867
parent_path = dirname(path)
868
if parent_path == u'':
869
parent_id = root_id
870
else:
871
parent_id = self.path2id(parent_path)
872
873
kind = self.get_kind(file_id)
874
revision_id = self.get_last_changed(file_id)
875
876
name = basename(path)
877
if kind == 'directory':
878
ie = InventoryDirectory(file_id, name, parent_id)
879
elif kind == 'file':
880
ie = InventoryFile(file_id, name, parent_id)
881
ie.executable = self.is_executable(file_id)
882
elif kind == 'symlink':
883
ie = InventoryLink(file_id, name, parent_id)
884
ie.symlink_target = self.get_symlink_target(file_id)
885
ie.revision = revision_id
886
887
if kind in ('directory', 'symlink'):
888
ie.text_size, ie.text_sha1 = None, None
889
else:
890
ie.text_size, ie.text_sha1 = self.get_size_and_sha1(file_id)
891
if (ie.text_size is None) and (kind == 'file'):
892
raise BzrError('Got a text_size of None for file_id %r' % file_id)
893
inv.add(ie)
894
895
sorted_entries = self.sorted_path_id()
896
for path, file_id in sorted_entries:
897
if file_id == inv.root.file_id:
898
continue
899
add_entry(file_id)
900
901
return inv
902
903
# Have to overload the inherited inventory property
904
# because _get_inventory is only called in the parent.
905
# Reading the docs, property() objects do not use
906
# overloading, they use the function as it was defined
907
# at that instant
908
inventory = property(_get_inventory)
909
910
def __iter__(self):
911
for path, entry in self.inventory.iter_entries():
912
yield entry.file_id
913
914
def sorted_path_id(self):
915
paths = []
916
for result in self._new_id.iteritems():
917
paths.append(result)
918
for id in self.base_tree:
919
path = self.id2path(id)
920
if path is None:
921
continue
922
paths.append((path, id))
923
paths.sort()
924
return paths
925
926
927
def patched_file(file_patch, original):
928
"""Produce a file-like object with the patched version of a text"""
929
from bzrlib.patches import iter_patched
930
from bzrlib.iterablefile import IterableFile
931
if file_patch == "":
932
return IterableFile(())
933
return IterableFile(iter_patched(original, file_patch.splitlines(True)))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1