36
45
An InventoryEntry has the following fields, which are also
37
46
present in the XML inventory-entry element:
40
* *name*: (only the basename within the directory, must not
42
* *kind*: "directory" or "file"
43
* *directory_id*: (if absent/null means the branch root directory)
44
* *text_sha1*: only for files
45
* *text_size*: in bytes, only for files
46
* *text_id*: identifier for the text version, only for files
48
InventoryEntries can also exist inside a WorkingTree
49
inventory, in which case they are not yet bound to a
50
particular revision of the file. In that case the text_sha1,
51
text_size and text_id are absent.
51
(within the parent directory)
54
file_id of the parent directory, or ROOT_ID
57
the revision_id in which this variation of this file was
61
Indicates that this file should be executable on systems
65
sha-1 of the text of the file
68
size in bytes of the text of the file
70
(reading a version 4 tree created a text_id field.)
54
72
>>> i = Inventory()
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
59
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
60
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
75
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
76
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
77
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
78
InventoryFile('2323', 'hello.c', parent_id='123')
61
79
>>> for j in i.iter_entries():
64
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
65
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
66
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
82
('src', InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
83
('src/hello.c', InventoryFile('2323', 'hello.c', parent_id='123'))
84
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
67
85
Traceback (most recent call last):
69
87
BzrError: inventory already contains entry with id {2323}
70
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
71
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
72
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
73
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
88
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
89
InventoryFile('2324', 'bye.c', parent_id='123')
90
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
91
InventoryDirectory('2325', 'wibble', parent_id='123')
74
92
>>> i.path2id('src/wibble')
78
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
96
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
97
InventoryFile('2326', 'wibble.c', parent_id='2325')
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
>>> for j in i.iter_entries():
84
... assert i.path2id(j[0])
99
InventoryFile('2326', 'wibble.c', parent_id='2325')
100
>>> for path, entry in i.iter_entries():
101
... print path.replace('\\\\', '/') # for win32 os.sep
102
... assert i.path2id(path)
90
108
src/wibble/wibble.c
109
>>> i.id2path('2326').replace('\\\\', '/')
92
110
'src/wibble/wibble.c'
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.
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
102
113
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
103
'text_id', 'parent_id', 'children',
104
'text_version', 'entry_version', ]
107
def __init__(self, file_id, name, kind, parent_id, text_id=None):
114
'text_id', 'parent_id', 'children', 'executable',
117
def _add_text_to_weave(self, new_lines, parents, weave_store):
118
weave_store.add_text(self.file_id, self.revision, new_lines, parents)
120
def detect_changes(self, old_entry):
121
"""Return a (text_modified, meta_modified) from this to old_entry.
123
_read_tree_state must have been called on self and old_entry prior to
124
calling detect_changes.
128
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
129
output_to, reverse=False):
130
"""Perform a diff from this to to_entry.
132
text_diff will be used for textual difference calculation.
133
This is a template method, override _diff in child classes.
135
self._read_tree_state(tree.id2path(self.file_id), tree)
137
# cannot diff from one kind to another - you must do a removal
138
# and an addif they do not match.
139
assert self.kind == to_entry.kind
140
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
142
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
145
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
146
output_to, reverse=False):
147
"""Perform a diff between two entries of the same kind."""
149
def find_previous_heads(self, previous_inventories, entry_weave):
150
"""Return the revisions and entries that directly preceed this.
152
Returned as a map from revision to inventory entry.
154
This is a map containing the file revisions in all parents
155
for which the file exists, and its revision is not a parent of
156
any other. If the file is new, the set will be empty.
158
def get_ancestors(weave, entry):
159
return set(map(weave.idx_to_name,
160
weave.inclusions([weave.lookup(entry.revision)])))
163
for inv in previous_inventories:
164
if self.file_id in inv:
165
ie = inv[self.file_id]
166
assert ie.file_id == self.file_id
167
if ie.revision in heads:
168
assert heads[ie.revision] == ie
170
# may want to add it.
171
# may already be covered:
172
already_present = 0 != len(
173
[head for head in heads
174
if ie.revision in head_ancestors[head]])
176
# an ancestor of a known head.
179
ancestors = get_ancestors(entry_weave, ie)
180
# may knock something else out:
181
check_heads = list(heads.keys())
182
for head in check_heads:
183
if head in ancestors:
184
# this head is not really a head
186
head_ancestors[ie.revision] = ancestors
187
heads[ie.revision] = ie
190
def get_tar_item(self, root, dp, now, tree):
191
"""Get a tarfile item and a file stream for its content."""
192
item = tarfile.TarInfo(os.path.join(root, dp))
193
# TODO: would be cool to actually set it to the timestamp of the
194
# revision it was last changed
196
fileobj = self._put_in_tar(item, tree)
200
"""Return true if the object this entry represents has textual data.
202
Note that textual data includes binary content.
204
Also note that all entries get weave files created for them.
205
This attribute is primarily used when upgrading from old trees that
206
did not have the weave index for all inventory entries.
210
def __init__(self, file_id, name, parent_id, text_id=None):
108
211
"""Create an InventoryEntry
110
213
The filename must be a single component, relative to the
111
214
parent directory; it cannot be a whole path or relative name.
113
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
216
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
118
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
221
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
119
222
Traceback (most recent call last):
120
223
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
123
226
if '/' in name or '\\' in name:
124
227
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
126
self.text_version = None
127
self.entry_version = None
229
self.executable = False
128
231
self.text_sha1 = None
129
232
self.text_size = None
130
233
self.file_id = file_id
133
235
self.text_id = text_id
134
236
self.parent_id = parent_id
135
if kind == 'directory':
140
raise BzrError("unhandled entry kind %r" % kind)
237
self.symlink_target = None
239
def kind_character(self):
240
"""Return a short kind indicator useful for appending to names."""
241
raise BzrError('unknown kind %r' % self.kind)
243
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
245
def _put_in_tar(self, item, tree):
246
"""populate item for stashing in a tar, and return the content stream.
248
If no content is available, return None.
250
raise BzrError("don't know how to export {%s} of kind %r" %
251
(self.file_id, self.kind))
253
def put_on_disk(self, dest, dp, tree):
254
"""Create a representation of self on disk in the prefix dest.
256
This is a template method - implement _put_on_disk in subclasses.
258
fullpath = appendpath(dest, dp)
259
self._put_on_disk(fullpath, tree)
260
mutter(" export {%s} kind %s to %s" % (self.file_id, self.kind, fullpath))
262
def _put_on_disk(self, fullpath, tree):
263
"""Put this entry onto disk at fullpath, from tree tree."""
264
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
144
266
def sorted_children(self):
145
267
l = self.children.items()
272
def versionable_kind(kind):
273
return kind in ('file', 'directory', 'symlink')
275
def check(self, checker, rev_id, inv, tree):
276
"""Check this inventory entry is intact.
278
This is a template method, override _check for kind specific
281
if self.parent_id != None:
282
if not inv.has_id(self.parent_id):
283
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
284
% (self.parent_id, rev_id))
285
self._check(checker, rev_id, tree)
287
def _check(self, checker, rev_id, tree):
288
"""Check this inventory entry for kind specific errors."""
289
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
151
other = InventoryEntry(self.file_id, self.name, self.kind,
152
self.parent_id, text_id=self.text_id)
153
other.text_sha1 = self.text_sha1
154
other.text_size = self.text_size
155
# note that children are *not* copied; they're pulled across when
294
"""Clone this inventory entry."""
295
raise NotImplementedError
297
def _get_snapshot_change(self, previous_entries):
298
if len(previous_entries) > 1:
300
elif len(previous_entries) == 0:
303
return 'modified/renamed/reparented'
160
305
def __repr__(self):
161
return ("%s(%r, %r, kind=%r, parent_id=%r)"
306
return ("%s(%r, %r, parent_id=%r)"
162
307
% (self.__class__.__name__,
312
def snapshot(self, revision, path, previous_entries,
313
work_tree, weave_store):
314
"""Make a snapshot of this entry which may or may not have changed.
316
This means that all its fields are populated, that it has its
317
text stored in the text store or weave.
319
mutter('new parents of %s are %r', path, previous_entries)
320
self._read_tree_state(path, work_tree)
321
if len(previous_entries) == 1:
322
# cannot be unchanged unless there is only one parent file rev.
323
parent_ie = previous_entries.values()[0]
324
if self._unchanged(parent_ie):
325
mutter("found unchanged entry")
326
self.revision = parent_ie.revision
328
return self.snapshot_revision(revision, previous_entries,
329
work_tree, weave_store)
331
def snapshot_revision(self, revision, previous_entries, work_tree,
333
"""Record this revision unconditionally."""
334
mutter('new revision for {%s}', self.file_id)
335
self.revision = revision
336
change = self._get_snapshot_change(previous_entries)
337
self._snapshot_text(previous_entries, work_tree, weave_store)
340
def _snapshot_text(self, file_parents, work_tree, weave_store):
341
"""Record the 'text' of this entry, whatever form that takes.
343
This default implementation simply adds an empty text.
345
mutter('storing file {%s} in revision {%s}',
346
self.file_id, self.revision)
347
self._add_text_to_weave([], file_parents, weave_store)
169
349
def __eq__(self, other):
170
350
if not isinstance(other, InventoryEntry):
171
351
return NotImplemented
173
return (self.file_id == other.file_id) \
174
and (self.name == other.name) \
175
and (self.text_sha1 == other.text_sha1) \
176
and (self.text_size == other.text_size) \
177
and (self.text_id == other.text_id) \
178
and (self.parent_id == other.parent_id) \
179
and (self.kind == other.kind) \
180
and (self.text_version == other.text_version) \
181
and (self.entry_version == other.entry_version)
353
return ((self.file_id == other.file_id)
354
and (self.name == other.name)
355
and (other.symlink_target == self.symlink_target)
356
and (self.text_sha1 == other.text_sha1)
357
and (self.text_size == other.text_size)
358
and (self.text_id == other.text_id)
359
and (self.parent_id == other.parent_id)
360
and (self.kind == other.kind)
361
and (self.revision == other.revision)
362
and (self.executable == other.executable)
184
365
def __ne__(self, other):
185
366
return not (self == other)
205
410
and (self.children == other.children)
413
class InventoryDirectory(InventoryEntry):
414
"""A directory in an inventory."""
416
def _check(self, checker, rev_id, tree):
417
"""See InventoryEntry._check"""
418
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
419
raise BzrCheckError('directory {%s} has text in revision {%s}'
420
% (self.file_id, rev_id))
423
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
424
other.revision = self.revision
425
# note that children are *not* copied; they're pulled across when
429
def __init__(self, file_id, name, parent_id):
430
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
432
self.kind = 'directory'
434
def kind_character(self):
435
"""See InventoryEntry.kind_character."""
438
def _put_in_tar(self, item, tree):
439
"""See InventoryEntry._put_in_tar."""
440
item.type = tarfile.DIRTYPE
447
def _put_on_disk(self, fullpath, tree):
448
"""See InventoryEntry._put_on_disk."""
452
class InventoryFile(InventoryEntry):
453
"""A file in an inventory."""
455
def _check(self, checker, rev_id, tree):
456
"""See InventoryEntry._check"""
457
revision = self.revision
458
t = (self.file_id, revision)
459
if t in checker.checked_texts:
460
prev_sha = checker.checked_texts[t]
461
if prev_sha != self.text_sha1:
462
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
463
(self.file_id, rev_id))
465
checker.repeated_text_cnt += 1
467
mutter('check version {%s} of {%s}', rev_id, self.file_id)
468
file_lines = tree.get_file_lines(self.file_id)
469
checker.checked_text_cnt += 1
470
if self.text_size != sum(map(len, file_lines)):
471
raise BzrCheckError('text {%s} wrong size' % self.text_id)
472
if self.text_sha1 != sha_strings(file_lines):
473
raise BzrCheckError('text {%s} wrong sha1' % self.text_id)
474
checker.checked_texts[t] = self.text_sha1
477
other = InventoryFile(self.file_id, self.name, self.parent_id)
478
other.executable = self.executable
479
other.text_id = self.text_id
480
other.text_sha1 = self.text_sha1
481
other.text_size = self.text_size
482
other.revision = self.revision
485
def detect_changes(self, old_entry):
486
"""See InventoryEntry.detect_changes."""
487
assert self.text_sha1 != None
488
assert old_entry.text_sha1 != None
489
text_modified = (self.text_sha1 != old_entry.text_sha1)
490
meta_modified = (self.executable != old_entry.executable)
491
return text_modified, meta_modified
493
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
494
output_to, reverse=False):
495
"""See InventoryEntry._diff."""
496
from_text = tree.get_file(self.file_id).readlines()
498
to_text = to_tree.get_file(to_entry.file_id).readlines()
502
text_diff(from_label, from_text,
503
to_label, to_text, output_to)
505
text_diff(to_label, to_text,
506
from_label, from_text, output_to)
509
"""See InventoryEntry.has_text."""
512
def __init__(self, file_id, name, parent_id):
513
super(InventoryFile, self).__init__(file_id, name, parent_id)
516
def kind_character(self):
517
"""See InventoryEntry.kind_character."""
520
def _put_in_tar(self, item, tree):
521
"""See InventoryEntry._put_in_tar."""
522
item.type = tarfile.REGTYPE
523
fileobj = tree.get_file(self.file_id)
524
item.size = self.text_size
525
if tree.is_executable(self.file_id):
531
def _put_on_disk(self, fullpath, tree):
532
"""See InventoryEntry._put_on_disk."""
533
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
534
if tree.is_executable(self.file_id):
535
os.chmod(fullpath, 0755)
537
def _read_tree_state(self, path, work_tree):
538
"""See InventoryEntry._read_tree_state."""
539
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
540
self.executable = work_tree.is_executable(self.file_id)
542
def _snapshot_text(self, file_parents, work_tree, weave_store):
543
"""See InventoryEntry._snapshot_text."""
544
mutter('storing file {%s} in revision {%s}',
545
self.file_id, self.revision)
546
# special case to avoid diffing on renames or
548
if (len(file_parents) == 1
549
and self.text_sha1 == file_parents.values()[0].text_sha1
550
and self.text_size == file_parents.values()[0].text_size):
551
previous_ie = file_parents.values()[0]
552
weave_store.add_identical_text(
553
self.file_id, previous_ie.revision,
554
self.revision, file_parents)
556
new_lines = work_tree.get_file(self.file_id).readlines()
557
self._add_text_to_weave(new_lines, file_parents, weave_store)
558
self.text_sha1 = sha_strings(new_lines)
559
self.text_size = sum(map(len, new_lines))
562
def _unchanged(self, previous_ie):
563
"""See InventoryEntry._unchanged."""
564
compatible = super(InventoryFile, self)._unchanged(previous_ie)
565
if self.text_sha1 != previous_ie.text_sha1:
568
# FIXME: 20050930 probe for the text size when getting sha1
569
# in _read_tree_state
570
self.text_size = previous_ie.text_size
574
class InventoryLink(InventoryEntry):
575
"""A file in an inventory."""
577
__slots__ = ['symlink_target']
579
def _check(self, checker, rev_id, tree):
580
"""See InventoryEntry._check"""
581
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
582
raise BzrCheckError('symlink {%s} has text in revision {%s}'
583
% (self.file_id, rev_id))
584
if self.symlink_target == None:
585
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
586
% (self.file_id, rev_id))
589
other = InventoryLink(self.file_id, self.name, self.parent_id)
590
other.symlink_target = self.symlink_target
591
other.revision = self.revision
594
def detect_changes(self, old_entry):
595
"""See InventoryEntry.detect_changes."""
596
# FIXME: which _modified field should we use ? RBC 20051003
597
text_modified = (self.symlink_target != old_entry.symlink_target)
599
mutter(" symlink target changed")
600
meta_modified = False
601
return text_modified, meta_modified
603
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
604
output_to, reverse=False):
605
"""See InventoryEntry._diff."""
606
from_text = self.symlink_target
607
if to_entry is not None:
608
to_text = to_entry.symlink_target
613
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
616
print >>output_to, '=== target was %r' % self.symlink_target
618
print >>output_to, '=== target is %r' % self.symlink_target
620
def __init__(self, file_id, name, parent_id):
621
super(InventoryLink, self).__init__(file_id, name, parent_id)
622
self.kind = 'symlink'
624
def kind_character(self):
625
"""See InventoryEntry.kind_character."""
628
def _put_in_tar(self, item, tree):
629
"""See InventoryEntry._put_in_tar."""
630
iterm.type = tarfile.SYMTYPE
634
item.linkname = self.symlink_target
637
def _put_on_disk(self, fullpath, tree):
638
"""See InventoryEntry._put_on_disk."""
640
os.symlink(self.symlink_target, fullpath)
642
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
644
def _read_tree_state(self, path, work_tree):
645
"""See InventoryEntry._read_tree_state."""
646
self.symlink_target = work_tree.get_symlink_target(self.file_id)
648
def _unchanged(self, previous_ie):
649
"""See InventoryEntry._unchanged."""
650
compatible = super(InventoryLink, self)._unchanged(previous_ie)
651
if self.symlink_target != previous_ie.symlink_target:
209
656
class Inventory(object):
210
657
"""Inventory of versioned files in a tree.