15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
# TODO: Maybe store inventory_id in the file? Not really needed.
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.
21
26
# This should really be an id randomly assigned when the tree is
26
31
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
38
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
39
37
from bzrlib.trace import mutter
38
from bzrlib.errors import NotVersionedError
41
class InventoryEntry(XMLMixin):
41
class InventoryEntry(object):
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:
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.
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.)
62
73
>>> i = Inventory()
65
76
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
77
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
66
78
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
79
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
67
80
>>> for j in i.iter_entries():
72
85
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
73
86
Traceback (most recent call last):
75
BzrError: ('inventory already contains entry with id {2323}', [])
88
BzrError: inventory already contains entry with id {2323}
76
89
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
90
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
77
91
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
92
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
78
93
>>> i.path2id('src/wibble')
82
97
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
98
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
84
100
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
85
101
>>> for j in i.iter_entries():
93
109
src/wibble/wibble.c
94
110
>>> i.id2path('2326')
95
111
'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', ]
105
119
def __init__(self, file_id, name, kind, parent_id, text_id=None):
106
120
"""Create an InventoryEntry
116
130
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
117
131
Traceback (most recent call last):
118
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
132
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
121
if len(splitpath(name)) != 1:
122
bailout('InventoryEntry name is not a simple filename: %r'
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
125
142
self.file_id = file_id
128
145
self.text_id = text_id
129
146
self.parent_id = parent_id
130
self.text_sha1 = None
131
self.text_size = None
132
147
if kind == 'directory':
133
148
self.children = {}
134
149
elif kind == 'file':
148
163
other = InventoryEntry(self.file_id, self.name, self.kind,
149
self.parent_id, text_id=self.text_id)
165
other.text_id = self.text_id
150
166
other.text_sha1 = self.text_sha1
151
167
other.text_size = self.text_size
168
# note that children are *not* copied; they're pulled across when
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):
182
def __eq__(self, other):
219
183
if not isinstance(other, InventoryEntry):
220
184
return NotImplemented
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)
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')
237
210
self.parent_id = None
240
def __cmp__(self, other):
213
def __eq__(self, other):
243
214
if not isinstance(other, RootEntry):
244
215
return NotImplemented
245
return cmp(self.file_id, other.file_id) \
246
or cmp(self.children, other.children)
250
class Inventory(XMLMixin):
217
return (self.file_id == other.file_id) \
218
and (self.children == other.children)
222
class Inventory(object):
251
223
"""Inventory of versioned files in a tree.
253
225
This describes which file_id is present at each point in the tree,
265
237
inserted, other than through the Inventory API.
267
239
>>> inv = Inventory()
268
>>> inv.write_xml(sys.stdout)
271
240
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
241
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT')
272
242
>>> inv['123-123'].name
284
254
>>> [x[0] for x in inv.iter_entries()]
287
>>> inv.write_xml(sys.stdout)
289
<entry file_id="123-123" kind="file" name="hello.c" />
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')
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?
260
def __init__(self, root_id=ROOT_ID):
310
261
"""Create or read an inventory.
312
263
If a working directory is specified, the inventory is read
316
267
The inventory is created with a default root directory, with
319
self.root = RootEntry(ROOT_ID)
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)
320
275
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())
323
287
def __iter__(self):
324
288
return iter(self._byid)
344
308
if ie.kind == 'directory':
345
309
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
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.
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))
325
if ie.kind == 'directory':
326
descend(ie, child_path)
328
descend(self.root, '')
350
332
def directories(self):
351
"""Return (path, entry) pairs for all directories.
333
"""Return (path, entry) pairs for all directories, including the root.
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():
360
if ie.kind == 'directory':
361
dn.append((ie.name, ie))
336
def descend(parent_ie, parent_path):
337
accum.append((parent_path, parent_ie))
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
339
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
368
for name, ie in descend(self.root):
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, '')
389
367
>>> inv = Inventory()
390
368
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
369
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
391
370
>>> inv['123123'].name
395
raise BzrError("can't look up file_id None")
398
374
return self._byid[file_id]
400
raise BzrError("file_id {%s} not in inventory" % 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
403
385
def get_child(self, parent_id, filename):
404
386
return self[parent_id].children.get(filename)
408
390
"""Add entry to inventory.
410
392
To add a file to a branch ready to be committed, use Branch.add,
395
Returns the new entry object.
412
397
if entry.file_id in self._byid:
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
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
416
404
parent = self._byid[entry.parent_id]
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
406
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
420
408
if parent.children.has_key(entry.name):
421
bailout("%s is already versioned" %
409
raise BzrError("%s is already versioned" %
422
410
appendpath(self.id2path(parent.file_id), entry.name))
424
412
self._byid[entry.file_id] = entry
425
413
parent.children[entry.name] = entry
428
417
def add_path(self, relpath, kind, file_id=None):
429
418
"""Add entry from a path.
431
The immediate parent must already be versioned"""
420
The immediate parent must already be versioned.
422
Returns the new entry object."""
423
from bzrlib.branch import gen_file_id
432
425
parts = bzrlib.osutils.splitpath(relpath)
433
426
if len(parts) == 0:
434
bailout("cannot re-add root of inventory")
427
raise BzrError("cannot re-add root of inventory")
436
429
if file_id == None:
437
file_id = bzrlib.branch.gen_file_id(relpath)
439
parent_id = self.path2id(parts[:-1])
440
assert parent_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)
441
437
ie = InventoryEntry(file_id, parts[-1],
442
438
kind=kind, parent_id=parent_id)
443
439
return self.add(ie)
469
466
del self[ie.parent_id].children[ie.name]
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):
469
def __eq__(self, other):
505
470
"""Compare two sets by comparing their contents.
507
472
>>> i1 = Inventory()
511
476
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
477
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
514
480
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
481
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
521
485
if not isinstance(other, Inventory):
522
486
return NotImplemented
524
if self.id_set() ^ other.id_set():
527
for file_id in self._byid:
528
c = cmp(self[file_id], other[file_id])
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')
534
503
def get_idpath(self, file_id):
605
574
This does not move the working file."""
606
575
if not is_valid_name(new_name):
607
bailout("not an acceptable filename: %r" % new_name)
576
raise BzrError("not an acceptable filename: %r" % new_name)
609
578
new_parent = self._byid[new_parent_id]
610
579
if new_name in new_parent.children:
611
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
580
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
613
582
new_parent_idpath = self.get_idpath(new_parent_id)
614
583
if file_id in new_parent_idpath:
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
584
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
616
585
% (self.id2path(file_id), self.id2path(new_parent_id)))
618
587
file_ie = self._byid[file_id]
added
removed
bzrlib/inventory.py
