36
48
An InventoryEntry has the following fields, which are also
37
49
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.
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.)
54
75
>>> i = Inventory()
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
59
>>> 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])
62
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
63
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
64
>>> 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'))
65
89
Traceback (most recent call last):
67
91
BzrError: inventory already contains entry with id {2323}
68
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
>>> 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')
70
96
>>> i.path2id('src/wibble')
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
100
>>> i.add(InventoryFile('2326', 'wibble.c', '2325'))
101
InventoryFile('2326', 'wibble.c', parent_id='2325')
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
77
>>> for j in i.iter_entries():
79
... 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)
85
112
src/wibble/wibble.c
86
113
>>> i.id2path('2326')
87
114
'src/wibble/wibble.c'
89
TODO: Maybe also keep the full path of the entry, and the children?
90
But those depend on its position within a particular inventory, and
91
it would be nice not to need to hold the backpointer here.
94
# TODO: split InventoryEntry into subclasses for files,
95
# directories, etc etc.
97
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
98
'text_id', 'parent_id', 'children', ]
100
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(map(weave.idx_to_name,
165
weave.inclusions([weave.lookup(entry.revision)])))
168
for inv in previous_inventories:
169
if self.file_id in inv:
170
ie = inv[self.file_id]
171
assert ie.file_id == self.file_id
172
if ie.revision in heads:
173
# fixup logic, there was a bug in revision updates.
174
# with x bit support.
176
if heads[ie.revision].executable != ie.executable:
177
heads[ie.revision].executable = False
178
ie.executable = False
179
except AttributeError:
181
assert heads[ie.revision] == ie
183
# may want to add it.
184
# may already be covered:
185
already_present = 0 != len(
186
[head for head in heads
187
if ie.revision in head_ancestors[head]])
189
# an ancestor of a known head.
192
ancestors = get_ancestors(entry_weave, ie)
193
# may knock something else out:
194
check_heads = list(heads.keys())
195
for head in check_heads:
196
if head in ancestors:
197
# this head is not really a head
199
head_ancestors[ie.revision] = ancestors
200
heads[ie.revision] = ie
203
def get_tar_item(self, root, dp, now, tree):
204
"""Get a tarfile item and a file stream for its content."""
205
item = tarfile.TarInfo(pathjoin(root, dp))
206
# TODO: would be cool to actually set it to the timestamp of the
207
# revision it was last changed
209
fileobj = self._put_in_tar(item, tree)
213
"""Return true if the object this entry represents has textual data.
215
Note that textual data includes binary content.
217
Also note that all entries get weave files created for them.
218
This attribute is primarily used when upgrading from old trees that
219
did not have the weave index for all inventory entries.
223
def __init__(self, file_id, name, parent_id, text_id=None):
101
224
"""Create an InventoryEntry
103
226
The filename must be a single component, relative to the
104
227
parent directory; it cannot be a whole path or relative name.
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
229
>>> e = InventoryFile('123', 'hello.c', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
234
>>> e = InventoryFile('123', 'src/hello.c', ROOT_ID)
112
235
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
236
InvalidEntryName: Invalid entry name: src/hello.c
238
assert isinstance(name, basestring), name
115
239
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
240
raise InvalidEntryName(name=name)
241
self.executable = False
118
243
self.text_sha1 = None
119
244
self.text_size = None
121
245
self.file_id = file_id
124
247
self.text_id = text_id
125
248
self.parent_id = parent_id
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
249
self.symlink_target = None
251
def kind_character(self):
252
"""Return a short kind indicator useful for appending to names."""
253
raise BzrError('unknown kind %r' % self.kind)
255
known_kinds = ('file', 'directory', 'symlink', 'root_directory')
257
def _put_in_tar(self, item, tree):
258
"""populate item for stashing in a tar, and return the content stream.
260
If no content is available, return None.
262
raise BzrError("don't know how to export {%s} of kind %r" %
263
(self.file_id, self.kind))
265
def put_on_disk(self, dest, dp, tree):
266
"""Create a representation of self on disk in the prefix dest.
268
This is a template method - implement _put_on_disk in subclasses.
270
fullpath = pathjoin(dest, dp)
271
self._put_on_disk(fullpath, tree)
272
mutter(" export {%s} kind %s to %s", self.file_id,
275
def _put_on_disk(self, fullpath, tree):
276
"""Put this entry onto disk at fullpath, from tree tree."""
277
raise BzrError("don't know how to export {%s} of kind %r" % (self.file_id, self.kind))
135
279
def sorted_children(self):
136
280
l = self.children.items()
285
def versionable_kind(kind):
286
return kind in ('file', 'directory', 'symlink')
288
def check(self, checker, rev_id, inv, tree):
289
"""Check this inventory entry is intact.
291
This is a template method, override _check for kind specific
294
if self.parent_id != None:
295
if not inv.has_id(self.parent_id):
296
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
297
% (self.parent_id, rev_id))
298
self._check(checker, rev_id, tree)
300
def _check(self, checker, rev_id, tree):
301
"""Check this inventory entry for kind specific errors."""
302
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
142
other = InventoryEntry(self.file_id, self.name, self.kind,
143
self.parent_id, text_id=self.text_id)
144
other.text_sha1 = self.text_sha1
145
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
307
"""Clone this inventory entry."""
308
raise NotImplementedError
310
def _get_snapshot_change(self, previous_entries):
311
if len(previous_entries) > 1:
313
elif len(previous_entries) == 0:
316
return 'modified/renamed/reparented'
151
318
def __repr__(self):
152
return ("%s(%r, %r, kind=%r, parent_id=%r)"
319
return ("%s(%r, %r, parent_id=%r)"
153
320
% (self.__class__.__name__,
160
def to_element(self):
161
"""Convert to XML element"""
162
from bzrlib.xml import Element
166
e.set('name', self.name)
167
e.set('file_id', self.file_id)
168
e.set('kind', self.kind)
170
if self.text_size != None:
171
e.set('text_size', '%d' % self.text_size)
173
for f in ['text_id', 'text_sha1']:
178
# to be conservative, we don't externalize the root pointers
179
# for now, leaving them as null in the xml form. in a future
180
# version it will be implied by nested elements.
181
if self.parent_id != ROOT_ID:
182
assert isinstance(self.parent_id, basestring)
183
e.set('parent_id', self.parent_id)
190
def from_element(cls, elt):
191
assert elt.tag == 'entry'
193
## original format inventories don't have a parent_id for
194
## nodes in the root directory, but it's cleaner to use one
196
parent_id = elt.get('parent_id')
197
if parent_id == None:
200
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
201
self.text_id = elt.get('text_id')
202
self.text_sha1 = elt.get('text_sha1')
204
## mutter("read inventoryentry: %r" % (elt.attrib))
206
v = elt.get('text_size')
207
self.text_size = v and int(v)
212
from_element = classmethod(from_element)
325
def snapshot(self, revision, path, previous_entries,
326
work_tree, weave_store, transaction):
327
"""Make a snapshot of this entry which may or may not have changed.
329
This means that all its fields are populated, that it has its
330
text stored in the text store or weave.
332
mutter('new parents of %s are %r', path, previous_entries)
333
self._read_tree_state(path, work_tree)
334
if len(previous_entries) == 1:
335
# cannot be unchanged unless there is only one parent file rev.
336
parent_ie = previous_entries.values()[0]
337
if self._unchanged(parent_ie):
338
mutter("found unchanged entry")
339
self.revision = parent_ie.revision
341
return self.snapshot_revision(revision, previous_entries,
342
work_tree, weave_store, transaction)
344
def snapshot_revision(self, revision, previous_entries, work_tree,
345
weave_store, transaction):
346
"""Record this revision unconditionally."""
347
mutter('new revision for {%s}', self.file_id)
348
self.revision = revision
349
change = self._get_snapshot_change(previous_entries)
350
self._snapshot_text(previous_entries, work_tree, weave_store,
354
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
355
"""Record the 'text' of this entry, whatever form that takes.
357
This default implementation simply adds an empty text.
359
mutter('storing file {%s} in revision {%s}',
360
self.file_id, self.revision)
361
self._add_text_to_weave([], file_parents, weave_store, transaction)
214
363
def __eq__(self, other):
215
364
if not isinstance(other, InventoryEntry):
216
365
return NotImplemented
218
return (self.file_id == other.file_id) \
219
and (self.name == other.name) \
220
and (self.text_sha1 == other.text_sha1) \
221
and (self.text_size == other.text_size) \
222
and (self.text_id == other.text_id) \
223
and (self.parent_id == other.parent_id) \
224
and (self.kind == other.kind)
367
return ((self.file_id == other.file_id)
368
and (self.name == other.name)
369
and (other.symlink_target == self.symlink_target)
370
and (self.text_sha1 == other.text_sha1)
371
and (self.text_size == other.text_size)
372
and (self.text_id == other.text_id)
373
and (self.parent_id == other.parent_id)
374
and (self.kind == other.kind)
375
and (self.revision == other.revision)
376
and (self.executable == other.executable)
227
379
def __ne__(self, other):
228
380
return not (self == other)
248
431
and (self.children == other.children)
434
class InventoryDirectory(InventoryEntry):
435
"""A directory in an inventory."""
437
def _check(self, checker, rev_id, tree):
438
"""See InventoryEntry._check"""
439
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
440
raise BzrCheckError('directory {%s} has text in revision {%s}'
441
% (self.file_id, rev_id))
444
other = InventoryDirectory(self.file_id, self.name, self.parent_id)
445
other.revision = self.revision
446
# note that children are *not* copied; they're pulled across when
450
def __init__(self, file_id, name, parent_id):
451
super(InventoryDirectory, self).__init__(file_id, name, parent_id)
453
self.kind = 'directory'
455
def kind_character(self):
456
"""See InventoryEntry.kind_character."""
459
def _put_in_tar(self, item, tree):
460
"""See InventoryEntry._put_in_tar."""
461
item.type = tarfile.DIRTYPE
468
def _put_on_disk(self, fullpath, tree):
469
"""See InventoryEntry._put_on_disk."""
473
class InventoryFile(InventoryEntry):
474
"""A file in an inventory."""
476
def _check(self, checker, rev_id, tree):
477
"""See InventoryEntry._check"""
478
revision = self.revision
479
t = (self.file_id, revision)
480
if t in checker.checked_texts:
481
prev_sha = checker.checked_texts[t]
482
if prev_sha != self.text_sha1:
483
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
484
(self.file_id, rev_id))
486
checker.repeated_text_cnt += 1
489
if self.file_id not in checker.checked_weaves:
490
mutter('check weave {%s}', self.file_id)
491
w = tree.get_weave(self.file_id)
492
# Not passing a progress bar, because it creates a new
493
# progress, which overwrites the current progress,
494
# and doesn't look nice
496
checker.checked_weaves[self.file_id] = True
498
w = tree.get_weave_prelude(self.file_id)
500
mutter('check version {%s} of {%s}', rev_id, self.file_id)
501
checker.checked_text_cnt += 1
502
# We can't check the length, because Weave doesn't store that
503
# information, and the whole point of looking at the weave's
504
# sha1sum is that we don't have to extract the text.
505
if self.text_sha1 != w.get_sha1(self.revision):
506
raise BzrCheckError('text {%s} version {%s} wrong sha1'
507
% (self.file_id, self.revision))
508
checker.checked_texts[t] = self.text_sha1
511
other = InventoryFile(self.file_id, self.name, self.parent_id)
512
other.executable = self.executable
513
other.text_id = self.text_id
514
other.text_sha1 = self.text_sha1
515
other.text_size = self.text_size
516
other.revision = self.revision
519
def detect_changes(self, old_entry):
520
"""See InventoryEntry.detect_changes."""
521
assert self.text_sha1 != None
522
assert old_entry.text_sha1 != None
523
text_modified = (self.text_sha1 != old_entry.text_sha1)
524
meta_modified = (self.executable != old_entry.executable)
525
return text_modified, meta_modified
527
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
528
output_to, reverse=False):
529
"""See InventoryEntry._diff."""
530
from_text = tree.get_file(self.file_id).readlines()
532
to_text = to_tree.get_file(to_entry.file_id).readlines()
536
text_diff(from_label, from_text,
537
to_label, to_text, output_to)
539
text_diff(to_label, to_text,
540
from_label, from_text, output_to)
543
"""See InventoryEntry.has_text."""
546
def __init__(self, file_id, name, parent_id):
547
super(InventoryFile, self).__init__(file_id, name, parent_id)
550
def kind_character(self):
551
"""See InventoryEntry.kind_character."""
554
def _put_in_tar(self, item, tree):
555
"""See InventoryEntry._put_in_tar."""
556
item.type = tarfile.REGTYPE
557
fileobj = tree.get_file(self.file_id)
558
item.size = self.text_size
559
if tree.is_executable(self.file_id):
565
def _put_on_disk(self, fullpath, tree):
566
"""See InventoryEntry._put_on_disk."""
567
pumpfile(tree.get_file(self.file_id), file(fullpath, 'wb'))
568
if tree.is_executable(self.file_id):
569
os.chmod(fullpath, 0755)
571
def _read_tree_state(self, path, work_tree):
572
"""See InventoryEntry._read_tree_state."""
573
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
574
self.executable = work_tree.is_executable(self.file_id)
576
def _forget_tree_state(self):
577
self.text_sha1 = None
578
self.executable = None
580
def _snapshot_text(self, file_parents, work_tree, weave_store, transaction):
581
"""See InventoryEntry._snapshot_text."""
582
mutter('storing file {%s} in revision {%s}',
583
self.file_id, self.revision)
584
# special case to avoid diffing on renames or
586
if (len(file_parents) == 1
587
and self.text_sha1 == file_parents.values()[0].text_sha1
588
and self.text_size == file_parents.values()[0].text_size):
589
previous_ie = file_parents.values()[0]
590
versionedfile = weave_store.get_weave(self.file_id, transaction)
591
versionedfile.clone_text(self.revision, previous_ie.revision, file_parents)
593
new_lines = work_tree.get_file(self.file_id).readlines()
594
self._add_text_to_weave(new_lines, file_parents, weave_store,
596
self.text_sha1 = sha_strings(new_lines)
597
self.text_size = sum(map(len, new_lines))
600
def _unchanged(self, previous_ie):
601
"""See InventoryEntry._unchanged."""
602
compatible = super(InventoryFile, self)._unchanged(previous_ie)
603
if self.text_sha1 != previous_ie.text_sha1:
606
# FIXME: 20050930 probe for the text size when getting sha1
607
# in _read_tree_state
608
self.text_size = previous_ie.text_size
609
if self.executable != previous_ie.executable:
614
class InventoryLink(InventoryEntry):
615
"""A file in an inventory."""
617
__slots__ = ['symlink_target']
619
def _check(self, checker, rev_id, tree):
620
"""See InventoryEntry._check"""
621
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
622
raise BzrCheckError('symlink {%s} has text in revision {%s}'
623
% (self.file_id, rev_id))
624
if self.symlink_target == None:
625
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
626
% (self.file_id, rev_id))
629
other = InventoryLink(self.file_id, self.name, self.parent_id)
630
other.symlink_target = self.symlink_target
631
other.revision = self.revision
634
def detect_changes(self, old_entry):
635
"""See InventoryEntry.detect_changes."""
636
# FIXME: which _modified field should we use ? RBC 20051003
637
text_modified = (self.symlink_target != old_entry.symlink_target)
639
mutter(" symlink target changed")
640
meta_modified = False
641
return text_modified, meta_modified
643
def _diff(self, text_diff, from_label, tree, to_label, to_entry, to_tree,
644
output_to, reverse=False):
645
"""See InventoryEntry._diff."""
646
from_text = self.symlink_target
647
if to_entry is not None:
648
to_text = to_entry.symlink_target
653
print >>output_to, '=== target changed %r => %r' % (from_text, to_text)
656
print >>output_to, '=== target was %r' % self.symlink_target
658
print >>output_to, '=== target is %r' % self.symlink_target
660
def __init__(self, file_id, name, parent_id):
661
super(InventoryLink, self).__init__(file_id, name, parent_id)
662
self.kind = 'symlink'
664
def kind_character(self):
665
"""See InventoryEntry.kind_character."""
668
def _put_in_tar(self, item, tree):
669
"""See InventoryEntry._put_in_tar."""
670
item.type = tarfile.SYMTYPE
674
item.linkname = self.symlink_target
677
def _put_on_disk(self, fullpath, tree):
678
"""See InventoryEntry._put_on_disk."""
680
os.symlink(self.symlink_target, fullpath)
682
raise BzrError("Failed to create symlink %r -> %r, error: %s" % (fullpath, self.symlink_target, e))
684
def _read_tree_state(self, path, work_tree):
685
"""See InventoryEntry._read_tree_state."""
686
self.symlink_target = work_tree.get_symlink_target(self.file_id)
688
def _forget_tree_state(self):
689
self.symlink_target = None
691
def _unchanged(self, previous_ie):
692
"""See InventoryEntry._unchanged."""
693
compatible = super(InventoryLink, self)._unchanged(previous_ie)
694
if self.symlink_target != previous_ie.symlink_target:
252
699
class Inventory(object):
253
700
"""Inventory of versioned files in a tree.
added
removed
bzrlib/inventory.py
