15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
# TODO: Maybe also keep the full path of the entry, and the children?
19
# But those depend on its position within a particular inventory, and
20
# it would be nice not to need to hold the backpointer here.
22
# TODO: Perhaps split InventoryEntry into subclasses for files,
23
# directories, etc etc.
18
# TODO: Maybe store inventory_id in the file? Not really needed.
26
21
# This should really be an id randomly assigned when the tree is
31
26
import sys, os.path, types, re
30
from cElementTree import Element, ElementTree, SubElement
32
from elementtree.ElementTree import Element, ElementTree, SubElement
34
from xml import XMLMixin
35
from errors import bailout, BzrError
34
from bzrlib.errors import BzrError, BzrCheckError
36
38
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
37
39
from bzrlib.trace import mutter
38
from bzrlib.errors import NotVersionedError
41
class InventoryEntry(object):
41
class InventoryEntry(XMLMixin):
42
42
"""Description of a versioned file.
44
44
An InventoryEntry has the following fields, which are also
45
45
present in the XML inventory-entry element:
50
(within the parent directory)
56
file_id of the parent directory, or ROOT_ID
59
the revision_id in which the name or parent of this file was
63
sha-1 of the text of the file
66
size in bytes of the text of the file
69
the revision_id in which the text of this file was introduced
71
(reading a version 4 tree created a text_id field.)
48
* *name*: (only the basename within the directory, must not
50
* *kind*: "directory" or "file"
51
* *directory_id*: (if absent/null means the branch root directory)
52
* *text_sha1*: only for files
53
* *text_size*: in bytes, only for files
54
* *text_id*: identifier for the text version, only for files
56
InventoryEntries can also exist inside a WorkingTree
57
inventory, in which case they are not yet bound to a
58
particular revision of the file. In that case the text_sha1,
59
text_size and text_id are absent.
73
62
>>> i = Inventory()
76
65
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
77
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
78
66
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
79
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
80
67
>>> for j in i.iter_entries():
85
72
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
86
73
Traceback (most recent call last):
88
BzrError: inventory already contains entry with id {2323}
75
BzrError: ('inventory already contains entry with id {2323}', [])
89
76
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
90
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
91
77
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
92
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
93
78
>>> i.path2id('src/wibble')
97
82
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
98
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
100
84
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
101
85
>>> for j in i.iter_entries():
109
93
src/wibble/wibble.c
110
94
>>> i.id2path('2326')
111
95
'src/wibble/wibble.c'
97
:todo: Maybe also keep the full path of the entry, and the children?
98
But those depend on its position within a particular inventory, and
99
it would be nice not to need to hold the backpointer here.
102
# TODO: split InventoryEntry into subclasses for files,
103
# directories, etc etc.
114
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
115
'text_id', 'parent_id', 'children',
116
'text_version', 'entry_version', ]
119
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
120
106
"""Create an InventoryEntry
130
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
131
117
Traceback (most recent call last):
132
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
118
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
134
assert isinstance(name, basestring), name
135
if '/' in name or '\\' in name:
136
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
138
self.text_version = None
139
self.entry_version = None
140
self.text_sha1 = None
141
self.text_size = None
121
if len(splitpath(name)) != 1:
122
bailout('InventoryEntry name is not a simple filename: %r'
142
125
self.file_id = file_id
145
128
self.text_id = text_id
146
129
self.parent_id = parent_id
130
self.text_sha1 = None
131
self.text_size = None
147
132
if kind == 'directory':
148
133
self.children = {}
149
134
elif kind == 'file':
163
148
other = InventoryEntry(self.file_id, self.name, self.kind,
165
other.text_id = self.text_id
149
self.parent_id, text_id=self.text_id)
166
150
other.text_sha1 = self.text_sha1
167
151
other.text_size = self.text_size
168
# note that children are *not* copied; they're pulled across when
182
def __eq__(self, other):
164
def to_element(self):
165
"""Convert to XML element"""
168
e.set('name', self.name)
169
e.set('file_id', self.file_id)
170
e.set('kind', self.kind)
172
if self.text_size != None:
173
e.set('text_size', '%d' % self.text_size)
175
for f in ['text_id', 'text_sha1']:
180
# to be conservative, we don't externalize the root pointers
181
# for now, leaving them as null in the xml form. in a future
182
# version it will be implied by nested elements.
183
if self.parent_id != ROOT_ID:
184
assert isinstance(self.parent_id, basestring)
185
e.set('parent_id', self.parent_id)
192
def from_element(cls, elt):
193
assert elt.tag == 'entry'
195
## original format inventories don't have a parent_id for
196
## nodes in the root directory, but it's cleaner to use one
198
parent_id = elt.get('parent_id')
199
if parent_id == None:
202
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
203
self.text_id = elt.get('text_id')
204
self.text_sha1 = elt.get('text_sha1')
206
## mutter("read inventoryentry: %r" % (elt.attrib))
208
v = elt.get('text_size')
209
self.text_size = v and int(v)
214
from_element = classmethod(from_element)
216
def __cmp__(self, other):
183
219
if not isinstance(other, InventoryEntry):
184
220
return NotImplemented
186
return (self.file_id == other.file_id) \
187
and (self.name == other.name) \
188
and (self.text_sha1 == other.text_sha1) \
189
and (self.text_size == other.text_size) \
190
and (self.text_id == other.text_id) \
191
and (self.parent_id == other.parent_id) \
192
and (self.kind == other.kind) \
193
and (self.text_version == other.text_version) \
194
and (self.entry_version == other.entry_version)
197
def __ne__(self, other):
198
return not (self == other)
201
raise ValueError('not hashable')
222
return cmp(self.file_id, other.file_id) \
223
or cmp(self.name, other.name) \
224
or cmp(self.text_sha1, other.text_sha1) \
225
or cmp(self.text_size, other.text_size) \
226
or cmp(self.text_id, other.text_id) \
227
or cmp(self.parent_id, other.parent_id) \
228
or cmp(self.kind, other.kind)
210
237
self.parent_id = None
213
def __eq__(self, other):
240
def __cmp__(self, other):
214
243
if not isinstance(other, RootEntry):
215
244
return NotImplemented
217
return (self.file_id == other.file_id) \
218
and (self.children == other.children)
222
class Inventory(object):
245
return cmp(self.file_id, other.file_id) \
246
or cmp(self.children, other.children)
250
class Inventory(XMLMixin):
223
251
"""Inventory of versioned files in a tree.
225
253
This describes which file_id is present at each point in the tree,
237
265
inserted, other than through the Inventory API.
239
267
>>> inv = Inventory()
268
>>> inv.write_xml(sys.stdout)
240
271
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
241
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT')
242
272
>>> inv['123-123'].name
254
284
>>> [x[0] for x in inv.iter_entries()]
256
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
257
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
258
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT-12345678-12345678')
287
>>> inv.write_xml(sys.stdout)
289
<entry file_id="123-123" kind="file" name="hello.c" />
260
def __init__(self, root_id=ROOT_ID):
294
## TODO: Make sure only canonical filenames are stored.
296
## TODO: Do something sensible about the possible collisions on
297
## case-losing filesystems. Perhaps we should just always forbid
300
## TODO: No special cases for root, rather just give it a file id
301
## like everything else.
303
## TODO: Probably change XML serialization to use nesting rather
304
## than parent_id pointers.
306
## TODO: Perhaps hold the ElementTree in memory and work directly
307
## on that rather than converting into Python objects every time?
261
310
"""Create or read an inventory.
263
312
If a working directory is specified, the inventory is read
267
316
The inventory is created with a default root directory, with
270
# We are letting Branch(init=True) create a unique inventory
271
# root id. Rather than generating a random one here.
273
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
274
self.root = RootEntry(root_id)
319
self.root = RootEntry(ROOT_ID)
275
320
self._byid = {self.root.file_id: self.root}
279
other = Inventory(self.root.file_id)
280
for entry in self._byid.itervalues():
281
if entry == self.root:
283
other.add(entry.copy())
287
323
def __iter__(self):
288
324
return iter(self._byid)
308
344
if ie.kind == 'directory':
309
345
for cn, cie in self.iter_entries(from_dir=ie.file_id):
310
yield os.path.join(name, cn), cie
314
"""Return list of (path, ie) for all entries except the root.
316
This may be faster than iter_entries.
346
yield '/'.join((name, cn)), cie
350
def directories(self):
351
"""Return (path, entry) pairs for all directories.
319
def descend(dir_ie, dir_path):
320
kids = dir_ie.children.items()
322
for name, ie in kids:
323
child_path = os.path.join(dir_path, name)
324
accum.append((child_path, ie))
353
def descend(parent_ie):
354
parent_name = parent_ie.name
355
yield parent_name, parent_ie
357
# directory children in sorted order
359
for ie in parent_ie.children.itervalues():
325
360
if ie.kind == 'directory':
326
descend(ie, child_path)
328
descend(self.root, '')
332
def directories(self):
333
"""Return (path, entry) pairs for all directories, including the root.
336
def descend(parent_ie, parent_path):
337
accum.append((parent_path, parent_ie))
361
dn.append((ie.name, ie))
339
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
364
for name, child_ie in dn:
365
for sub_name, sub_ie in descend(child_ie):
366
yield appendpath(parent_name, sub_name), sub_ie
342
for name, child_ie in kids:
343
child_path = os.path.join(parent_path, name)
344
descend(child_ie, child_path)
345
descend(self.root, '')
368
for name, ie in descend(self.root):
367
389
>>> inv = Inventory()
368
390
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
369
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
370
391
>>> inv['123123'].name
395
raise BzrError("can't look up file_id None")
374
398
return self._byid[file_id]
377
raise BzrError("can't look up file_id None")
379
raise BzrError("file_id {%s} not in inventory" % file_id)
382
def get_file_kind(self, file_id):
383
return self._byid[file_id].kind
400
raise BzrError("file_id {%s} not in inventory" % file_id)
385
403
def get_child(self, parent_id, filename):
386
404
return self[parent_id].children.get(filename)
390
408
"""Add entry to inventory.
392
410
To add a file to a branch ready to be committed, use Branch.add,
395
Returns the new entry object.
397
412
if entry.file_id in self._byid:
398
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
400
if entry.parent_id == ROOT_ID or entry.parent_id is None:
401
entry.parent_id = self.root.file_id
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
404
416
parent = self._byid[entry.parent_id]
406
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
408
420
if parent.children.has_key(entry.name):
409
raise BzrError("%s is already versioned" %
421
bailout("%s is already versioned" %
410
422
appendpath(self.id2path(parent.file_id), entry.name))
412
424
self._byid[entry.file_id] = entry
413
425
parent.children[entry.name] = entry
417
428
def add_path(self, relpath, kind, file_id=None):
418
429
"""Add entry from a path.
420
The immediate parent must already be versioned.
422
Returns the new entry object."""
423
from bzrlib.branch import gen_file_id
431
The immediate parent must already be versioned"""
425
432
parts = bzrlib.osutils.splitpath(relpath)
426
433
if len(parts) == 0:
427
raise BzrError("cannot re-add root of inventory")
434
bailout("cannot re-add root of inventory")
429
436
if file_id == None:
430
file_id = gen_file_id(relpath)
432
parent_path = parts[:-1]
433
parent_id = self.path2id(parent_path)
434
if parent_id == None:
435
raise NotVersionedError(parent_path)
437
file_id = bzrlib.branch.gen_file_id(relpath)
439
parent_id = self.path2id(parts[:-1])
440
assert parent_id != None
437
441
ie = InventoryEntry(file_id, parts[-1],
438
442
kind=kind, parent_id=parent_id)
439
443
return self.add(ie)
466
469
del self[ie.parent_id].children[ie.name]
469
def __eq__(self, other):
473
return Set(self._byid)
476
def to_element(self):
477
"""Convert to XML Element"""
478
e = Element('inventory')
480
for path, ie in self.iter_entries():
481
e.append(ie.to_element())
485
def from_element(cls, elt):
486
"""Construct from XML Element
488
>>> inv = Inventory()
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
490
>>> elt = inv.to_element()
491
>>> inv2 = Inventory.from_element(elt)
495
assert elt.tag == 'inventory'
498
o.add(InventoryEntry.from_element(e))
501
from_element = classmethod(from_element)
504
def __cmp__(self, other):
470
505
"""Compare two sets by comparing their contents.
472
507
>>> i1 = Inventory()
476
511
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
477
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
480
514
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
481
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
485
521
if not isinstance(other, Inventory):
486
522
return NotImplemented
488
if len(self._byid) != len(other._byid):
489
# shortcut: obviously not the same
492
return self._byid == other._byid
495
def __ne__(self, other):
496
return not (self == other)
500
raise ValueError('not hashable')
524
if self.id_set() ^ other.id_set():
527
for file_id in self._byid:
528
c = cmp(self[file_id], other[file_id])
503
534
def get_idpath(self, file_id):
574
605
This does not move the working file."""
575
606
if not is_valid_name(new_name):
576
raise BzrError("not an acceptable filename: %r" % new_name)
607
bailout("not an acceptable filename: %r" % new_name)
578
609
new_parent = self._byid[new_parent_id]
579
610
if new_name in new_parent.children:
580
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
611
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
582
613
new_parent_idpath = self.get_idpath(new_parent_id)
583
614
if file_id in new_parent_idpath:
584
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
585
616
% (self.id2path(file_id), self.id2path(new_parent_id)))
587
618
file_ie = self._byid[file_id]
added
removed
bzrlib/inventory.py
