14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Inventories map files to their name in a revision."""
19
# TODO: Maybe store inventory_id in the file? Not really needed.
21
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
22
__author__ = "Martin Pool <mbp@canonical.com>"
24
import sys, os.path, types
28
from cElementTree import Element, ElementTree, SubElement
30
from elementtree.ElementTree import Element, ElementTree, SubElement
32
from xml import XMLMixin
33
from errors import bailout
18
# This should really be an id randomly assigned when the tree is
19
# created, but it's not for now.
23
import sys, os.path, types, re
26
from bzrlib.errors import BzrError, BzrCheckError
36
28
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
37
29
from bzrlib.trace import mutter
30
from bzrlib.errors import NotVersionedError
39
class InventoryEntry(XMLMixin):
33
class InventoryEntry(object):
40
34
"""Description of a versioned file.
42
36
An InventoryEntry has the following fields, which are also
60
54
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
59
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
60
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
64
61
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
64
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
65
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
66
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
67
Traceback (most recent call last):
72
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
69
BzrError: inventory already contains entry with id {2323}
70
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
71
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
72
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
73
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
75
74
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
78
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
81
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
82
>>> for j in i.iter_entries():
91
91
>>> i.id2path('2326')
92
92
'src/wibble/wibble.c'
94
:todo: Maybe also keep the full path of the entry, and the children?
94
TODO: Maybe also keep the full path of the entry, and the children?
95
95
But those depend on its position within a particular inventory, and
96
96
it would be nice not to need to hold the backpointer here.
98
def __init__(self, file_id, name, kind='file', text_id=None,
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
102
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
103
'text_id', 'parent_id', 'children', ]
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
100
106
"""Create an InventoryEntry
102
108
The filename must be a single component, relative to the
103
109
parent directory; it cannot be a whole path or relative name.
105
>>> e = InventoryEntry('123', 'hello.c')
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
117
Traceback (most recent call last):
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
120
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
123
self.text_sha1 = None
124
self.text_size = None
119
126
self.file_id = file_id
121
assert kind in ['file', 'directory']
123
129
self.text_id = text_id
124
130
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
127
131
if kind == 'directory':
128
132
self.children = {}
136
raise BzrError("unhandled entry kind %r" % kind)
140
def sorted_children(self):
141
l = self.children.items()
132
147
other = InventoryEntry(self.file_id, self.name, self.kind,
133
self.text_id, self.parent_id)
148
self.parent_id, text_id=self.text_id)
134
149
other.text_sha1 = self.text_sha1
135
150
other.text_size = self.text_size
151
# note that children are *not* copied; they're pulled across when
148
165
def to_element(self):
149
166
"""Convert to XML element"""
167
from bzrlib.xml import Element
150
169
e = Element('entry')
152
171
e.set('name', self.name)
153
172
e.set('file_id', self.file_id)
154
173
e.set('kind', self.kind)
156
if self.text_size is not None:
175
if self.text_size != None:
157
176
e.set('text_size', '%d' % self.text_size)
159
for f in ['text_id', 'text_sha1', 'parent_id']:
178
for f in ['text_id', 'text_sha1']:
160
179
v = getattr(self, f)
183
# to be conservative, we don't externalize the root pointers
184
# for now, leaving them as null in the xml form. in a future
185
# version it will be implied by nested elements.
186
if self.parent_id != ROOT_ID:
187
assert isinstance(self.parent_id, basestring)
188
e.set('parent_id', self.parent_id)
169
195
def from_element(cls, elt):
170
196
assert elt.tag == 'entry'
171
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
198
## original format inventories don't have a parent_id for
199
## nodes in the root directory, but it's cleaner to use one
201
parent_id = elt.get('parent_id')
202
if parent_id == None:
205
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
172
206
self.text_id = elt.get('text_id')
173
207
self.text_sha1 = elt.get('text_sha1')
174
self.parent_id = elt.get('parent_id')
176
209
## mutter("read inventoryentry: %r" % (elt.attrib))
184
217
from_element = classmethod(from_element)
186
def __cmp__(self, other):
219
def __eq__(self, other):
189
220
if not isinstance(other, InventoryEntry):
190
221
return NotImplemented
192
return cmp(self.file_id, other.file_id) \
193
or cmp(self.name, other.name) \
194
or cmp(self.text_sha1, other.text_sha1) \
195
or cmp(self.text_size, other.text_size) \
196
or cmp(self.text_id, other.text_id) \
197
or cmp(self.parent_id, other.parent_id) \
198
or cmp(self.kind, other.kind)
202
class Inventory(XMLMixin):
223
return (self.file_id == other.file_id) \
224
and (self.name == other.name) \
225
and (self.text_sha1 == other.text_sha1) \
226
and (self.text_size == other.text_size) \
227
and (self.text_id == other.text_id) \
228
and (self.parent_id == other.parent_id) \
229
and (self.kind == other.kind)
232
def __ne__(self, other):
233
return not (self == other)
236
raise ValueError('not hashable')
240
class RootEntry(InventoryEntry):
241
def __init__(self, file_id):
242
self.file_id = file_id
244
self.kind = 'root_directory'
245
self.parent_id = None
248
def __eq__(self, other):
249
if not isinstance(other, RootEntry):
250
return NotImplemented
252
return (self.file_id == other.file_id) \
253
and (self.children == other.children)
257
class Inventory(object):
203
258
"""Inventory of versioned files in a tree.
205
An Inventory acts like a set of InventoryEntry items. You can
206
also look files up by their file_id or name.
208
May be read from and written to a metadata file in a tree. To
209
manipulate the inventory (for example to add a file), it is read
210
in, modified, and then written back out.
260
This describes which file_id is present at each point in the tree,
261
and possibly the SHA-1 or other information about the file.
262
Entries can be looked up either by path or by file_id.
212
264
The inventory represents a typical unix file tree, with
213
265
directories containing files and subdirectories. We never store
217
269
returned quickly.
219
271
InventoryEntry objects must not be modified after they are
272
inserted, other than through the Inventory API.
222
274
>>> inv = Inventory()
223
>>> inv.write_xml(sys.stdout)
226
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
275
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
276
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT')
227
277
>>> inv['123-123'].name
239
289
>>> [x[0] for x in inv.iter_entries()]
242
>>> inv.write_xml(sys.stdout)
244
<entry file_id="123-123" kind="file" name="hello.c" />
291
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
292
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
293
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT-12345678-12345678')
249
## TODO: Make sure only canonical filenames are stored.
251
## TODO: Do something sensible about the possible collisions on
252
## case-losing filesystems. Perhaps we should just always forbid
255
## TODO: No special cases for root, rather just give it a file id
256
## like everything else.
258
## TODO: Probably change XML serialization to use nesting
295
def __init__(self, root_id=ROOT_ID):
261
296
"""Create or read an inventory.
263
298
If a working directory is specified, the inventory is read
264
299
from there. If the file is specified, read from that. If not,
265
300
the inventory is created empty.
302
The inventory is created with a default root directory, with
267
self._root = InventoryEntry(None, '', kind='directory')
268
self._byid = {None: self._root}
305
# We are letting Branch(init=True) create a unique inventory
306
# root id. Rather than generating a random one here.
308
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
309
self.root = RootEntry(root_id)
310
self._byid = {self.root.file_id: self.root}
271
313
def __iter__(self):
277
319
return len(self._byid)
280
def iter_entries(self, parent_id=None):
322
def iter_entries(self, from_dir=None):
281
323
"""Return (path, entry) pairs, in order by name."""
282
kids = self[parent_id].children.items()
327
elif isinstance(from_dir, basestring):
328
from_dir = self._byid[from_dir]
330
kids = from_dir.children.items()
284
332
for name, ie in kids:
286
334
if ie.kind == 'directory':
287
for cn, cie in self.iter_entries(parent_id=ie.file_id):
288
yield joinpath([name, cn]), cie
291
def directories(self, include_root=True):
292
"""Return (path, entry) pairs for all directories.
296
for path, entry in self.iter_entries():
297
if entry.kind == 'directory':
335
for cn, cie in self.iter_entries(from_dir=ie.file_id):
336
yield os.path.join(name, cn), cie
340
"""Return list of (path, ie) for all entries except the root.
342
This may be faster than iter_entries.
345
def descend(dir_ie, dir_path):
346
kids = dir_ie.children.items()
348
for name, ie in kids:
349
child_path = os.path.join(dir_path, name)
350
accum.append((child_path, ie))
351
if ie.kind == 'directory':
352
descend(ie, child_path)
354
descend(self.root, '')
358
def directories(self):
359
"""Return (path, entry) pairs for all directories, including the root.
362
def descend(parent_ie, parent_path):
363
accum.append((parent_path, parent_ie))
365
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
368
for name, child_ie in kids:
369
child_path = os.path.join(parent_path, name)
370
descend(child_ie, child_path)
371
descend(self.root, '')
316
391
"""Return the entry for given file_id.
318
393
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123123', 'hello.c'))
394
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
395
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
320
396
>>> inv['123123'].name
323
return self._byid[file_id]
400
return self._byid[file_id]
403
raise BzrError("can't look up file_id None")
405
raise BzrError("file_id {%s} not in inventory" % file_id)
408
def get_file_kind(self, file_id):
409
return self._byid[file_id].kind
326
411
def get_child(self, parent_id, filename):
327
if parent_id == None:
328
return self._root.children.get(filename)
330
return self[parent_id].children.get(filename)
412
return self[parent_id].children.get(filename)
333
415
def add(self, entry):
336
418
To add a file to a branch ready to be committed, use Branch.add,
337
419
which calls this."""
338
420
if entry.file_id in self._byid:
339
bailout("inventory already contains entry with id {%s}" % entry.file_id)
341
parent = self._byid[entry.parent_id]
342
if parent.kind != 'directory':
343
bailout("attempt to add under non-directory {%s}" % parent.file_id)
421
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
423
if entry.parent_id == ROOT_ID or entry.parent_id is None:
424
entry.parent_id = self.root.file_id
427
parent = self._byid[entry.parent_id]
429
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
345
431
if parent.children.has_key(entry.name):
346
bailout("%s is already versioned" %
432
raise BzrError("%s is already versioned" %
347
433
appendpath(self.id2path(parent.file_id), entry.name))
349
435
self._byid[entry.file_id] = entry
350
436
parent.children[entry.name] = entry
353
440
def add_path(self, relpath, kind, file_id=None):
354
441
"""Add entry from a path.
356
443
The immediate parent must already be versioned"""
444
from bzrlib.branch import gen_file_id
357
446
parts = bzrlib.osutils.splitpath(relpath)
358
447
if len(parts) == 0:
359
bailout("cannot re-add root of inventory")
362
file_id = bzrlib.branch.gen_file_id(relpath)
364
parent_id = self.path2id(parts[:-1])
448
raise BzrError("cannot re-add root of inventory")
451
file_id = gen_file_id(relpath)
453
parent_path = parts[:-1]
454
parent_id = self.path2id(parent_path)
455
if parent_id == None:
456
raise NotVersionedError(parent_path)
365
458
ie = InventoryEntry(file_id, parts[-1],
366
459
kind=kind, parent_id=parent_id)
367
460
return self.add(ie)
393
487
del self[ie.parent_id].children[ie.name]
397
return Set(self._byid)
400
490
def to_element(self):
401
491
"""Convert to XML Element"""
492
from bzrlib.xml import Element
402
494
e = Element('inventory')
496
if self.root.file_id not in (None, ROOT_ID):
497
e.set('file_id', self.root.file_id)
404
498
for path, ie in self.iter_entries():
405
499
e.append(ie.to_element())
409
503
def from_element(cls, elt):
410
504
"""Construct from XML Element
412
506
>>> inv = Inventory()
413
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
507
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
508
InventoryEntry('foo.c-123981239', 'foo.c', kind='file', parent_id='TREE_ROOT')
414
509
>>> elt = inv.to_element()
415
510
>>> inv2 = Inventory.from_element(elt)
514
# XXXX: doctest doesn't run this properly under python2.3
419
515
assert elt.tag == 'inventory'
516
root_id = elt.get('file_id') or ROOT_ID
422
o.add(InventoryEntry.from_element(e))
519
ie = InventoryEntry.from_element(e)
520
if ie.parent_id == ROOT_ID:
521
ie.parent_id = root_id
425
525
from_element = classmethod(from_element)
428
def __cmp__(self, other):
528
def __eq__(self, other):
429
529
"""Compare two sets by comparing their contents.
431
531
>>> i1 = Inventory()
432
532
>>> i2 = Inventory()
435
>>> i1.add(InventoryEntry('123', 'foo'))
535
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
536
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
438
>>> i2.add(InventoryEntry('123', 'foo'))
539
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
540
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
445
544
if not isinstance(other, Inventory):
446
545
return NotImplemented
448
if self.id_set() ^ other.id_set():
451
for file_id in self._byid:
452
c = cmp(self[file_id], other[file_id])
458
def id2path(self, file_id):
459
"""Return as a list the path to file_id."""
547
if len(self._byid) != len(other._byid):
548
# shortcut: obviously not the same
551
return self._byid == other._byid
554
def __ne__(self, other):
555
return not (self == other)
559
raise ValueError('not hashable')
563
def get_idpath(self, file_id):
564
"""Return a list of file_ids for the path to an entry.
566
The list contains one element for each directory followed by
567
the id of the file itself. So the length of the returned list
568
is equal to the depth of the file in the tree, counting the
569
root directory as depth 1.
461
572
while file_id != None:
574
ie = self._byid[file_id]
576
raise BzrError("file_id {%s} not found in inventory" % file_id)
577
p.insert(0, ie.file_id)
464
578
file_id = ie.parent_id
582
def id2path(self, file_id):
583
"""Return as a list the path to file_id."""
585
# get all names, skipping root
586
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
587
return os.sep.join(p)
498
625
def has_id(self, file_id):
499
assert isinstance(file_id, str)
500
626
return self._byid.has_key(file_id)
506
if __name__ == '__main__':
507
import doctest, inventory
508
doctest.testmod(inventory)
629
def rename(self, file_id, new_parent_id, new_name):
630
"""Move a file within the inventory.
632
This can change either the name, or the parent, or both.
634
This does not move the working file."""
635
if not is_valid_name(new_name):
636
raise BzrError("not an acceptable filename: %r" % new_name)
638
new_parent = self._byid[new_parent_id]
639
if new_name in new_parent.children:
640
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
642
new_parent_idpath = self.get_idpath(new_parent_id)
643
if file_id in new_parent_idpath:
644
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
645
% (self.id2path(file_id), self.id2path(new_parent_id)))
647
file_ie = self._byid[file_id]
648
old_parent = self._byid[file_ie.parent_id]
650
# TODO: Don't leave things messed up if this fails
652
del old_parent.children[file_ie.name]
653
new_parent.children[new_name] = file_ie
655
file_ie.name = new_name
656
file_ie.parent_id = new_parent_id
663
def is_valid_name(name):
666
_NAME_RE = re.compile(r'^[^/\\]+$')
668
return bool(_NAME_RE.match(name))