37
48
An InventoryEntry has the following fields, which are also
38
49
present in the XML inventory-entry element:
41
* *name*: (only the basename within the directory, must not
43
* *kind*: "directory" or "file"
44
* *directory_id*: (if absent/null means the branch root directory)
45
* *text_sha1*: only for files
46
* *text_size*: in bytes, only for files
47
* *text_id*: identifier for the text version, only for files
49
InventoryEntries can also exist inside a WorkingTree
50
inventory, in which case they are not yet bound to a
51
particular revision of the file. In that case the text_sha1,
52
text_size and text_id are absent.
54
(within the parent directory)
57
file_id of the parent directory, or ROOT_ID
60
the revision_id in which this variation of this file was
64
Indicates that this file should be executable on systems
68
sha-1 of the text of the file
71
size in bytes of the text of the file
73
(reading a version 4 tree created a text_id field.)
55
75
>>> i = Inventory()
58
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
59
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
60
>>> for j in i.iter_entries():
78
>>> i.add(InventoryDirectory('123', 'src', ROOT_ID))
79
InventoryDirectory('123', 'src', parent_id='TREE_ROOT')
80
>>> i.add(InventoryFile('2323', 'hello.c', parent_id='123'))
81
InventoryFile('2323', 'hello.c', parent_id='123')
82
>>> shouldbe = {0: 'src', 1: pathjoin('src','hello.c')}
83
>>> for ix, j in enumerate(i.iter_entries()):
84
... print (j[0] == shouldbe[ix], j[1])
63
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
64
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
65
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
86
(True, InventoryDirectory('123', 'src', parent_id='TREE_ROOT'))
87
(True, InventoryFile('2323', 'hello.c', parent_id='123'))
88
>>> i.add(InventoryFile('2323', 'bye.c', '123'))
66
89
Traceback (most recent call last):
68
91
BzrError: inventory already contains entry with id {2323}
69
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
70
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
92
>>> i.add(InventoryFile('2324', 'bye.c', '123'))
93
InventoryFile('2324', 'bye.c', parent_id='123')
94
>>> i.add(InventoryDirectory('2325', 'wibble', '123'))
95
InventoryDirectory('2325', 'wibble', parent_id='123')
71
96
>>> i.path2id('src/wibble')
75
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
100
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
101
InventoryFile('2326', 'wibble.c', parent_id='2325')
77
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
78
>>> for j in i.iter_entries():
80
... assert i.path2id(j[0])
103
InventoryFile('2326', 'wibble.c', parent_id='2325')
104
>>> for path, entry in i.iter_entries():
106
... assert i.path2id(path)
86
112
src/wibble/wibble.c
87
113
>>> i.id2path('2326')
88
114
'src/wibble/wibble.c'
90
TODO: Maybe also keep the full path of the entry, and the children?
91
But those depend on its position within a particular inventory, and
92
it would be nice not to need to hold the backpointer here.
95
# TODO: split InventoryEntry into subclasses for files,
96
# directories, etc etc.
98
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
99
'text_id', 'parent_id', 'children', ]
101
def __init__(self, file_id, name, kind, parent_id, text_id=None):
118
'text_id', 'parent_id', 'children', 'executable',
121
def _add_text_to_weave(self, new_lines, parents, weave_store, transaction):
122
versionedfile = weave_store.get_weave(self.file_id, transaction)
123
versionedfile.add_lines(self.revision, parents, new_lines)
125
def detect_changes(self, old_entry):
126
"""Return a (text_modified, meta_modified) from this to old_entry.
128
_read_tree_state must have been called on self and old_entry prior to
129
calling detect_changes.
133
def diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
134
output_to, reverse=False):
135
"""Perform a diff from this to to_entry.
137
text_diff will be used for textual difference calculation.
138
This is a template method, override _diff in child classes.
140
self._read_tree_state(tree.id2path(self.file_id), tree)
142
# cannot diff from one kind to another - you must do a removal
143
# and an addif they do not match.
144
assert self.kind == to_entry.kind
145
to_entry._read_tree_state(to_tree.id2path(to_entry.file_id),
147
self._diff(text_diff, from_label, tree, to_label, to_entry, to_tree,
150
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
151
output_to, reverse=False):
152
"""Perform a diff between two entries of the same kind."""
154
def find_previous_heads(self, previous_inventories, entry_weave):
155
"""Return the revisions and entries that directly preceed this.
157
Returned as a map from revision to inventory entry.
159
This is a map containing the file revisions in all parents
160
for which the file exists, and its revision is not a parent of
161
any other. If the file is new, the set will be empty.
163
def get_ancestors(weave, entry):
164
return set(weave.get_ancestry(entry.revision))
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.
175
if heads[ie.revision].executable != ie.executable:
176
heads[ie.revision].executable = False
177
ie.executable = False
178
except AttributeError:
180
assert heads[ie.revision] == ie
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]])
188
# an ancestor of a known 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
198
head_ancestors[ie.revision] = ancestors
199
heads[ie.revision] = ie
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(pathjoin(root, dp))
205
# TODO: would be cool to actually set it to the timestamp of the
206
# revision it was last changed
208
fileobj = self._put_in_tar(item, tree)
212
"""Return true if the object this entry represents has textual data.
214
Note that textual data includes binary content.
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.
222
def __init__(self, file_id, name, parent_id, text_id=None):
102
223
"""Create an InventoryEntry
104
225
The filename must be a single component, relative to the
105
226
parent directory; it cannot be a whole path or relative name.
107
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
228
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
112
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
233
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
113
234
Traceback (most recent call last):
114
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
235
InvalidEntryName: Invalid entry name: src/hello.c
237
assert isinstance(name, basestring), name
116
238
if '/' in name or '\\' in name:
117
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
239
raise InvalidEntryName(name=name)
240
self.executable = False
119
242
self.text_sha1 = None
120
243
self.text_size = None
122
244
self.file_id = file_id
125
246
self.text_id = text_id
126
247
self.parent_id = parent_id
127
if kind == 'directory':
132
raise BzrError("unhandled entry kind %r" % kind)
248
self.symlink_target = None
250
def kind_character(self):
251
"""Return a short kind indicator useful for appending to names."""
252
raise BzrError('unknown kind %r' % self.kind)
254
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
256
def _put_in_tar(self, item, tree):
257
"""populate item for stashing in a tar, and return the content stream.
259
If no content is available, return None.
261
raise BzrError("don't know how to export {%s} of kind %r" %
262
(self.file_id, self.kind))
264
def put_on_disk(self, dest, dp, tree):
265
"""Create a representation of self on disk in the prefix dest.
267
This is a template method - implement _put_on_disk in subclasses.
269
fullpath = pathjoin(dest, dp)
270
self._put_on_disk(fullpath, tree)
271
mutter(" export {%s} kind %s to %s", self.file_id,
274
def _put_on_disk(self, fullpath, tree):
275
"""Put this entry onto disk at fullpath, from tree tree."""
276
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
136
278
def sorted_children(self):
137
279
l = self.children.items()
284
def versionable_kind(kind):
285
return kind in ('file', 'directory', 'symlink')
287
def check(self, checker, rev_id, inv, tree):
288
"""Check this inventory entry is intact.
290
This is a template method, override _check for kind specific
293
if self.parent_id != None:
294
if not inv.has_id(self.parent_id):
295
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
296
% (self.parent_id, rev_id))
297
self._check(checker, rev_id, tree)
299
def _check(self, checker, rev_id, tree):
300
"""Check this inventory entry for kind specific errors."""
301
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
143
other = InventoryEntry(self.file_id, self.name, self.kind,
144
self.parent_id, text_id=self.text_id)
145
other.text_sha1 = self.text_sha1
146
other.text_size = self.text_size
147
# note that children are *not* copied; they're pulled across when
306
"""Clone this inventory entry."""
307
raise NotImplementedError
309
def _get_snapshot_change(self, previous_entries):
310
if len(previous_entries) > 1:
312
elif len(previous_entries) == 0:
315
return 'modified/renamed/reparented'
152
317
def __repr__(self):
153
return ("%s(%r, %r, kind=%r, parent_id=%r)"
318
return ("%s(%r, %r, parent_id=%r)"
154
319
% (self.__class__.__name__,
161
def to_element(self):
162
"""Convert to XML element"""
163
from bzrlib.xml import Element
167
e.set('name', self.name)
168
e.set('file_id', self.file_id)
169
e.set('kind', self.kind)
171
if self.text_size != None:
172
e.set('text_size', '%d' % self.text_size)
174
for f in ['text_id', 'text_sha1']:
179
# to be conservative, we don't externalize the root pointers
180
# for now, leaving them as null in the xml form. in a future
181
# version it will be implied by nested elements.
182
if self.parent_id != ROOT_ID:
183
assert isinstance(self.parent_id, basestring)
184
e.set('parent_id', self.parent_id)
191
def from_element(cls, elt):
192
assert elt.tag == 'entry'
194
## original format inventories don't have a parent_id for
195
## nodes in the root directory, but it's cleaner to use one
197
parent_id = elt.get('parent_id')
198
if parent_id == None:
201
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
202
self.text_id = elt.get('text_id')
203
self.text_sha1 = elt.get('text_sha1')
205
## mutter("read inventoryentry: %r" % (elt.attrib))
207
v = elt.get('text_size')
208
self.text_size = v and int(v)
213
from_element = classmethod(from_element)
324
def snapshot(self, revision, path, previous_entries,
325
work_tree, weave_store, transaction):
326
"""Make a snapshot of this entry which may or may not have changed.
328
This means that all its fields are populated, that it has its
329
text stored in the text store or weave.
331
mutter('new parents of %s are %r', path, previous_entries)
332
self._read_tree_state(path, work_tree)
333
if len(previous_entries) == 1:
334
# cannot be unchanged unless there is only one parent file rev.
335
parent_ie = previous_entries.values()[0]
336
if self._unchanged(parent_ie):
337
mutter("found unchanged entry")
338
self.revision = parent_ie.revision
340
return self.snapshot_revision(revision, previous_entries,
341
work_tree, weave_store, transaction)
343
def snapshot_revision(self, revision, previous_entries, work_tree,
344
weave_store, transaction):
345
"""Record this revision unconditionally."""
346
mutter('new revision for {%s}', self.file_id)
347
self.revision = revision
348
change = self._get_snapshot_change(previous_entries)
349
self._snapshot_text(previous_entries, work_tree, weave_store,
353
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
354
"""Record the 'text' of this entry, whatever form that takes.
356
This default implementation simply adds an empty text.
358
mutter('storing file {%s} in revision {%s}',
359
self.file_id, self.revision)
360
self._add_text_to_weave([], file_parents.keys(), weave_store, transaction)
215
362
def __eq__(self, other):
216
363
if not isinstance(other, InventoryEntry):
217
364
return NotImplemented
219
return (self.file_id == other.file_id) \
220
and (self.name == other.name) \
221
and (self.text_sha1 == other.text_sha1) \
222
and (self.text_size == other.text_size) \
223
and (self.text_id == other.text_id) \
224
and (self.parent_id == other.parent_id) \
225
and (self.kind == other.kind)
366
return ((self.file_id == other.file_id)
367
and (self.name == other.name)
368
and (other.symlink_target == self.symlink_target)
369
and (self.text_sha1 == other.text_sha1)
370
and (self.text_size == other.text_size)
371
and (self.text_id == other.text_id)
372
and (self.parent_id == other.parent_id)
373
and (self.kind == other.kind)
374
and (self.revision == other.revision)
375
and (self.executable == other.executable)
228
378
def __ne__(self, other):
229
379
return not (self == other)
249
430
and (self.children == other.children)
433
class InventoryDirectory(InventoryEntry):
434
"""A directory in an inventory."""
436
def _check(self, checker, rev_id, tree):
437
"""See InventoryEntry._check"""
438
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
439
raise BzrCheckError('directory {%s} has text in revision {%s}'
440
% (self.file_id, rev_id))
443
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
444
other.revision = self.revision
445
# note that children are *not* copied; they're pulled across when
449
def __init__(self, file_id, name, parent_id):
450
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
452
self.kind = 'directory'
454
def kind_character(self):
455
"""See InventoryEntry.kind_character."""
458
def _put_in_tar(self, item, tree):
459
"""See InventoryEntry._put_in_tar."""
460
item.type = tarfile.DIRTYPE
467
def _put_on_disk(self, fullpath, tree):
468
"""See InventoryEntry._put_on_disk."""
472
class InventoryFile(InventoryEntry):
473
"""A file in an inventory."""
475
def _check(self, checker, rev_id, tree):
476
"""See InventoryEntry._check"""
477
revision = self.revision
478
t = (self.file_id, revision)
479
if t in checker.checked_texts:
480
prev_sha = checker.checked_texts[t]
481
if prev_sha != self.text_sha1:
482
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
483
(self.file_id, rev_id))
485
checker.repeated_text_cnt += 1
488
if self.file_id not in checker.checked_weaves:
489
mutter('check weave {%s}', self.file_id)
490
w = tree.get_weave(self.file_id)
491
# Not passing a progress bar, because it creates a new
492
# progress, which overwrites the current progress,
493
# and doesn't look nice
495
checker.checked_weaves[self.file_id] = True
497
w = tree.get_weave(self.file_id)
499
mutter('check version {%s} of {%s}', rev_id, self.file_id)
500
checker.checked_text_cnt += 1
501
# We can't check the length, because Weave doesn't store that
502
# information, and the whole point of looking at the weave's
503
# sha1sum is that we don't have to extract the text.
504
if self.text_sha1 != w.get_sha1(self.revision):
505
raise BzrCheckError('text {%s} version {%s} wrong sha1'
506
% (self.file_id, self.revision))
507
checker.checked_texts[t] = self.text_sha1
510
other = InventoryFile(self.file_id, self.name, self.parent_id)
511
other.executable = self.executable
512
other.text_id = self.text_id
513
other.text_sha1 = self.text_sha1
514
other.text_size = self.text_size
515
other.revision = self.revision
518
def detect_changes(self, old_entry):
519
"""See InventoryEntry.detect_changes."""
520
assert self.text_sha1 != None
521
assert old_entry.text_sha1 != None
522
text_modified = (self.text_sha1 != old_entry.text_sha1)
523
meta_modified = (self.executable != old_entry.executable)
524
return text_modified, meta_modified
526
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
527
output_to, reverse=False):
528
"""See InventoryEntry._diff."""
529
from_text = tree.get_file(self.file_id).readlines()
531
to_text = to_tree.get_file(to_entry.file_id).readlines()
535
text_diff(from_label, from_text,
536
to_label, to_text, output_to)
538
text_diff(to_label, to_text,
539
from_label, from_text, output_to)
542
"""See InventoryEntry.has_text."""
545
def __init__(self, file_id, name, parent_id):
546
super(InventoryFile, self).__init__(file_id, name, parent_id)
549
def kind_character(self):
550
"""See InventoryEntry.kind_character."""
553
def _put_in_tar(self, item, tree):
554
"""See InventoryEntry._put_in_tar."""
555
item.type = tarfile.REGTYPE
556
fileobj = tree.get_file(self.file_id)
557
item.size = self.text_size
558
if tree.is_executable(self.file_id):
564
def _put_on_disk(self, fullpath, tree):
565
"""See InventoryEntry._put_on_disk."""
566
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
567
if tree.is_executable(self.file_id):
568
os.chmod(fullpath, 0755)
570
def _read_tree_state(self, path, work_tree):
571
"""See InventoryEntry._read_tree_state."""
572
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
573
self.executable = work_tree.is_executable(self.file_id)
575
def _forget_tree_state(self):
576
self.text_sha1 = None
577
self.executable = None
579
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
580
"""See InventoryEntry._snapshot_text."""
581
mutter('storing file {%s} in revision {%s}',
582
self.file_id, self.revision)
583
# special case to avoid diffing on renames or
585
if (len(file_parents) == 1
586
and self.text_sha1 == file_parents.values()[0].text_sha1
587
and self.text_size == file_parents.values()[0].text_size):
588
previous_ie = file_parents.values()[0]
589
versionedfile = weave_store.get_weave(self.file_id, transaction)
590
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents.keys())
592
new_lines = work_tree.get_file(self.file_id).readlines()
593
self._add_text_to_weave(new_lines, file_parents.keys(), weave_store,
595
self.text_sha1 = sha_strings(new_lines)
596
self.text_size = sum(map(len, new_lines))
599
def _unchanged(self, previous_ie):
600
"""See InventoryEntry._unchanged."""
601
compatible = super(InventoryFile, self)._unchanged(previous_ie)
602
if self.text_sha1 != previous_ie.text_sha1:
605
# FIXME: 20050930 probe for the text size when getting sha1
606
# in _read_tree_state
607
self.text_size = previous_ie.text_size
608
if self.executable != previous_ie.executable:
613
class InventoryLink(InventoryEntry):
614
"""A file in an inventory."""
616
__slots__ = ['symlink_target']
618
def _check(self, checker, rev_id, tree):
619
"""See InventoryEntry._check"""
620
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
621
raise BzrCheckError('symlink {%s} has text in revision {%s}'
622
% (self.file_id, rev_id))
623
if self.symlink_target == None:
624
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
625
% (self.file_id, rev_id))
628
other = InventoryLink(self.file_id, self.name, self.parent_id)
629
other.symlink_target = self.symlink_target
630
other.revision = self.revision
633
def detect_changes(self, old_entry):
634
"""See InventoryEntry.detect_changes."""
635
# FIXME: which _modified field should we use ? RBC 20051003
636
text_modified = (self.symlink_target != old_entry.symlink_target)
638
mutter(" symlink target changed")
639
meta_modified = False
640
return text_modified, meta_modified
642
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
643
output_to, reverse=False):
644
"""See InventoryEntry._diff."""
645
from_text = self.symlink_target
646
if to_entry is not None:
647
to_text = to_entry.symlink_target
652
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
655
print >>output_to, '=== target was %r' % self.symlink_target
657
print >>output_to, '=== target is %r' % self.symlink_target
659
def __init__(self, file_id, name, parent_id):
660
super(InventoryLink, self).__init__(file_id, name, parent_id)
661
self.kind = 'symlink'
663
def kind_character(self):
664
"""See InventoryEntry.kind_character."""
667
def _put_in_tar(self, item, tree):
668
"""See InventoryEntry._put_in_tar."""
669
item.type = tarfile.SYMTYPE
673
item.linkname = self.symlink_target
676
def _put_on_disk(self, fullpath, tree):
677
"""See InventoryEntry._put_on_disk."""
679
os.symlink(self.symlink_target, fullpath)
681
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
683
def _read_tree_state(self, path, work_tree):
684
"""See InventoryEntry._read_tree_state."""
685
self.symlink_target = work_tree.get_symlink_target(self.file_id)
687
def _forget_tree_state(self):
688
self.symlink_target = None
690
def _unchanged(self, previous_ie):
691
"""See InventoryEntry._unchanged."""
692
compatible = super(InventoryLink, self)._unchanged(previous_ie)
693
if self.symlink_target != previous_ie.symlink_target:
253
698
class Inventory(object):
254
699
"""Inventory of versioned files in a tree.
added
removed
bzrlib/inventory.py
