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
23
28
ROOT_ID = "TREE_ROOT"
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
38
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
37
from bzrlib.errors import BzrError, BzrCheckError
39
from bzrlib.osutils import quotefn, splitpath, joinpath, appendpath
39
40
from bzrlib.trace import mutter
41
class InventoryEntry(XMLMixin):
41
from bzrlib.errors import NotVersionedError
44
class InventoryEntry(object):
42
45
"""Description of a versioned file.
44
47
An InventoryEntry has the following fields, which are also
45
48
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.
53
(within the parent directory)
59
file_id of the parent directory, or ROOT_ID
62
the revision_id in which the name or parent of this file was
66
sha-1 of the text of the file
69
size in bytes of the text of the file
72
the revision_id in which the text of this file was introduced
74
(reading a version 4 tree created a text_id field.)
62
76
>>> i = Inventory()
65
79
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
80
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
66
81
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
82
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
67
83
>>> for j in i.iter_entries():
72
88
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
73
89
Traceback (most recent call last):
75
BzrError: ('inventory already contains entry with id {2323}', [])
91
BzrError: inventory already contains entry with id {2323}
76
92
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
93
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
77
94
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
95
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
78
96
>>> i.path2id('src/wibble')
82
100
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
101
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
84
103
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
85
>>> for j in i.iter_entries():
87
... assert i.path2id(j[0])
104
>>> for path, entry in i.iter_entries():
105
... print path.replace('\\\\', '/') # for win32 os.sep
106
... assert i.path2id(path)
93
112
src/wibble/wibble.c
113
>>> i.id2path('2326').replace('\\\\', '/')
95
114
'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.
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
118
'text_id', 'parent_id', 'children',
119
'text_version', 'name_version', ]
105
122
def __init__(self, file_id, name, kind, parent_id, text_id=None):
106
123
"""Create an InventoryEntry
116
133
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
117
134
Traceback (most recent call last):
118
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
135
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
121
if len(splitpath(name)) != 1:
122
bailout('InventoryEntry name is not a simple filename: %r'
137
assert isinstance(name, basestring), name
138
if '/' in name or '\\' in name:
139
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
141
self.text_version = None
142
self.name_version = None
143
self.text_sha1 = None
144
self.text_size = None
125
145
self.file_id = file_id
128
148
self.text_id = text_id
129
149
self.parent_id = parent_id
130
self.text_sha1 = None
131
self.text_size = None
132
150
if kind == 'directory':
133
151
self.children = {}
134
152
elif kind == 'file':
148
166
other = InventoryEntry(self.file_id, self.name, self.kind,
149
self.parent_id, text_id=self.text_id)
168
other.text_id = self.text_id
150
169
other.text_sha1 = self.text_sha1
151
170
other.text_size = self.text_size
171
other.text_version = self.text_version
172
other.name_version = self.name_version
173
# 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):
187
def __eq__(self, other):
219
188
if not isinstance(other, InventoryEntry):
220
189
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)
191
return (self.file_id == other.file_id) \
192
and (self.name == other.name) \
193
and (self.text_sha1 == other.text_sha1) \
194
and (self.text_size == other.text_size) \
195
and (self.text_id == other.text_id) \
196
and (self.parent_id == other.parent_id) \
197
and (self.kind == other.kind) \
198
and (self.text_version == other.text_version) \
199
and (self.name_version == other.name_version)
202
def __ne__(self, other):
203
return not (self == other)
206
raise ValueError('not hashable')
237
215
self.parent_id = None
240
def __cmp__(self, other):
218
def __eq__(self, other):
243
219
if not isinstance(other, RootEntry):
244
220
return NotImplemented
245
return cmp(self.file_id, other.file_id) \
246
or cmp(self.children, other.children)
250
class Inventory(XMLMixin):
222
return (self.file_id == other.file_id) \
223
and (self.children == other.children)
227
class Inventory(object):
251
228
"""Inventory of versioned files in a tree.
253
230
This describes which file_id is present at each point in the tree,
265
242
inserted, other than through the Inventory API.
267
244
>>> inv = Inventory()
268
>>> inv.write_xml(sys.stdout)
271
245
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
246
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT')
272
247
>>> inv['123-123'].name
284
259
>>> [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" />
261
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
262
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
263
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?
265
def __init__(self, root_id=ROOT_ID):
310
266
"""Create or read an inventory.
312
268
If a working directory is specified, the inventory is read
316
272
The inventory is created with a default root directory, with
319
self.root = RootEntry(ROOT_ID)
275
# We are letting Branch.initialize() create a unique inventory
276
# root id. Rather than generating a random one here.
278
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
279
self.root = RootEntry(root_id)
320
280
self._byid = {self.root.file_id: self.root}
284
other = Inventory(self.root.file_id)
285
# copy recursively so we know directories will be added before
286
# their children. There are more efficient ways than this...
287
for path, entry in self.iter_entries():
288
if entry == self.root:
290
other.add(entry.copy())
323
294
def __iter__(self):
324
295
return iter(self._byid)
344
315
if ie.kind == 'directory':
345
316
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
317
yield os.path.join(name, cn), cie
321
"""Return list of (path, ie) for all entries except the root.
323
This may be faster than iter_entries.
326
def descend(dir_ie, dir_path):
327
kids = dir_ie.children.items()
329
for name, ie in kids:
330
child_path = os.path.join(dir_path, name)
331
accum.append((child_path, ie))
332
if ie.kind == 'directory':
333
descend(ie, child_path)
335
descend(self.root, '')
350
339
def directories(self):
351
"""Return (path, entry) pairs for all directories.
340
"""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))
343
def descend(parent_ie, parent_path):
344
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
346
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
368
for name, ie in descend(self.root):
349
for name, child_ie in kids:
350
child_path = os.path.join(parent_path, name)
351
descend(child_ie, child_path)
352
descend(self.root, '')
389
374
>>> inv = Inventory()
390
375
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
376
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
391
377
>>> inv['123123'].name
395
raise BzrError("can't look up file_id None")
398
381
return self._byid[file_id]
400
raise BzrError("file_id {%s} not in inventory" % file_id)
384
raise BzrError("can't look up file_id None")
386
raise BzrError("file_id {%s} not in inventory" % file_id)
389
def get_file_kind(self, file_id):
390
return self._byid[file_id].kind
403
392
def get_child(self, parent_id, filename):
404
393
return self[parent_id].children.get(filename)
408
397
"""Add entry to inventory.
410
399
To add a file to a branch ready to be committed, use Branch.add,
402
Returns the new entry object.
412
404
if entry.file_id in self._byid:
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
405
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
407
if entry.parent_id == ROOT_ID or entry.parent_id is None:
408
entry.parent_id = self.root.file_id
416
411
parent = self._byid[entry.parent_id]
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
413
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
420
415
if parent.children.has_key(entry.name):
421
bailout("%s is already versioned" %
416
raise BzrError("%s is already versioned" %
422
417
appendpath(self.id2path(parent.file_id), entry.name))
424
419
self._byid[entry.file_id] = entry
425
420
parent.children[entry.name] = entry
428
424
def add_path(self, relpath, kind, file_id=None):
429
425
"""Add entry from a path.
431
The immediate parent must already be versioned"""
427
The immediate parent must already be versioned.
429
Returns the new entry object."""
430
from bzrlib.branch import gen_file_id
432
432
parts = bzrlib.osutils.splitpath(relpath)
433
433
if len(parts) == 0:
434
bailout("cannot re-add root of inventory")
434
raise BzrError("cannot re-add root of inventory")
436
436
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
437
file_id = gen_file_id(relpath)
439
parent_path = parts[:-1]
440
parent_id = self.path2id(parent_path)
441
if parent_id == None:
442
raise NotVersionedError(parent_path)
441
444
ie = InventoryEntry(file_id, parts[-1],
442
445
kind=kind, parent_id=parent_id)
443
446
return self.add(ie)
469
473
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):
476
def __eq__(self, other):
505
477
"""Compare two sets by comparing their contents.
507
479
>>> i1 = Inventory()
511
483
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
484
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
514
487
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
488
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
521
492
if not isinstance(other, Inventory):
522
493
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])
495
if len(self._byid) != len(other._byid):
496
# shortcut: obviously not the same
499
return self._byid == other._byid
502
def __ne__(self, other):
503
return not self.__eq__(other)
507
raise ValueError('not hashable')
534
510
def get_idpath(self, file_id):
605
581
This does not move the working file."""
606
582
if not is_valid_name(new_name):
607
bailout("not an acceptable filename: %r" % new_name)
583
raise BzrError("not an acceptable filename: %r" % new_name)
609
585
new_parent = self._byid[new_parent_id]
610
586
if new_name in new_parent.children:
611
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
587
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
613
589
new_parent_idpath = self.get_idpath(new_parent_id)
614
590
if file_id in new_parent_idpath:
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
591
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
616
592
% (self.id2path(file_id), self.id2path(new_parent_id)))
618
594
file_ie = self._byid[file_id]
added
removed
bzrlib/inventory.py
