14
15
# along with this program; if not, write to the Free Software
15
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
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
28
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
29
from bzrlib.trace import mutter
31
class InventoryEntry(object):
18
"""Inventories map files to their name in a revision."""
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
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
37
class InventoryEntry(XMLMixin):
32
38
"""Description of a versioned file.
34
40
An InventoryEntry has the following fields, which are also
52
58
>>> i = Inventory()
55
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
56
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
57
62
>>> for j in i.iter_entries():
60
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
61
66
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
62
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
63
68
Traceback (most recent call last):
65
BzrError: inventory already contains entry with id {2323}
66
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
67
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
70
BzrError: ('inventory already contains entry with id {2323}', [])
71
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
68
73
>>> i.path2id('src/wibble')
72
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
74
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
75
80
>>> for j in i.iter_entries():
84
89
>>> i.id2path('2326')
85
90
'src/wibble/wibble.c'
87
TODO: Maybe also keep the full path of the entry, and the children?
92
:todo: Maybe also keep the full path of the entry, and the children?
88
93
But those depend on its position within a particular inventory, and
89
94
it would be nice not to need to hold the backpointer here.
92
# TODO: split InventoryEntry into subclasses for files,
93
# directories, etc etc.
95
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
96
'text_id', 'parent_id', 'children', ]
98
def __init__(self, file_id, name, kind, parent_id, text_id=None):
96
def __init__(self, file_id, name, kind='file', text_id=None,
99
98
"""Create an InventoryEntry
101
100
The filename must be a single component, relative to the
102
101
parent directory; it cannot be a whole path or relative name.
104
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
103
>>> e = InventoryEntry('123', 'hello.c')
109
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
108
>>> e = InventoryEntry('123', 'src/hello.c')
110
109
Traceback (most recent call last):
111
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
113
if '/' in name or '\\' in name:
114
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
116
self.text_sha1 = None
117
self.text_size = None
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
119
117
self.file_id = file_id
119
assert kind in ['file', 'directory']
122
121
self.text_id = text_id
123
122
self.parent_id = parent_id
124
if kind == 'directory':
129
raise BzrError("unhandled entry kind %r" % kind)
133
def sorted_children(self):
134
l = self.children.items()
123
self.text_sha1 = None
124
self.text_size = None
140
128
other = InventoryEntry(self.file_id, self.name, self.kind,
141
self.parent_id, text_id=self.text_id)
129
self.text_id, self.parent_id)
142
130
other.text_sha1 = self.text_sha1
143
131
other.text_size = self.text_size
144
# note that children are *not* copied; they're pulled across when
158
144
def to_element(self):
159
145
"""Convert to XML element"""
160
from bzrlib.xml import Element
162
146
e = Element('entry')
164
148
e.set('name', self.name)
165
149
e.set('file_id', self.file_id)
166
150
e.set('kind', self.kind)
168
if self.text_size != None:
152
if self.text_size is not None:
169
153
e.set('text_size', '%d' % self.text_size)
171
for f in ['text_id', 'text_sha1']:
155
for f in ['text_id', 'text_sha1', 'parent_id']:
172
156
v = getattr(self, f)
176
# to be conservative, we don't externalize the root pointers
177
# for now, leaving them as null in the xml form. in a future
178
# version it will be implied by nested elements.
179
if self.parent_id != ROOT_ID:
180
assert isinstance(self.parent_id, basestring)
181
e.set('parent_id', self.parent_id)
188
165
def from_element(cls, elt):
189
166
assert elt.tag == 'entry'
191
## original format inventories don't have a parent_id for
192
## nodes in the root directory, but it's cleaner to use one
194
parent_id = elt.get('parent_id')
195
if parent_id == None:
198
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
199
168
self.text_id = elt.get('text_id')
200
169
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
202
172
## mutter("read inventoryentry: %r" % (elt.attrib))
210
180
from_element = classmethod(from_element)
212
def __eq__(self, other):
182
def __cmp__(self, other):
213
185
if not isinstance(other, InventoryEntry):
214
186
return NotImplemented
216
return (self.file_id == other.file_id) \
217
and (self.name == other.name) \
218
and (self.text_sha1 == other.text_sha1) \
219
and (self.text_size == other.text_size) \
220
and (self.text_id == other.text_id) \
221
and (self.parent_id == other.parent_id) \
222
and (self.kind == other.kind)
225
def __ne__(self, other):
226
return not (self == other)
229
raise ValueError('not hashable')
233
class RootEntry(InventoryEntry):
234
def __init__(self, file_id):
235
self.file_id = file_id
237
self.kind = 'root_directory'
238
self.parent_id = None
241
def __eq__(self, other):
242
if not isinstance(other, RootEntry):
243
return NotImplemented
245
return (self.file_id == other.file_id) \
246
and (self.children == other.children)
250
class Inventory(object):
188
return cmp(self.file_id, other.file_id) \
189
or cmp(self.name, other.name) \
190
or cmp(self.text_sha1, other.text_sha1) \
191
or cmp(self.text_size, other.text_size) \
192
or cmp(self.text_id, other.text_id) \
193
or cmp(self.parent_id, other.parent_id) \
194
or cmp(self.kind, other.kind)
198
class Inventory(XMLMixin):
251
199
"""Inventory of versioned files in a tree.
253
This describes which file_id is present at each point in the tree,
254
and possibly the SHA-1 or other information about the file.
255
Entries can be looked up either by path or by file_id.
201
An Inventory acts like a set of InventoryEntry items. You can
202
also look files up by their file_id or name.
204
May be read from and written to a metadata file in a tree. To
205
manipulate the inventory (for example to add a file), it is read
206
in, modified, and then written back out.
257
208
The inventory represents a typical unix file tree, with
258
209
directories containing files and subdirectories. We never store
262
213
returned quickly.
264
215
InventoryEntry objects must not be modified after they are
265
inserted, other than through the Inventory API.
267
218
>>> inv = Inventory()
268
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
219
>>> inv.write_xml(sys.stdout)
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
269
223
>>> inv['123-123'].name
225
>>> for file_id in inv: print file_id
272
229
May be treated as an iterator or set to look up file ids:
281
238
>>> [x[0] for x in inv.iter_entries()]
283
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
284
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
241
>>> inv.write_xml(sys.stdout)
243
<entry file_id="123-123" kind="file" name="hello.c" />
286
def __init__(self, root_id=ROOT_ID):
248
## TODO: Clear up handling of files in subdirectories; we probably
249
## do want to be able to just look them up by name but this
250
## probably means gradually walking down the path, looking up as we go.
252
## TODO: Make sure only canonical filenames are stored.
254
## TODO: Do something sensible about the possible collisions on
255
## case-losing filesystems. Perhaps we should just always forbid
258
## _tree should probably just be stored as
259
## InventoryEntry._children on each directory.
287
262
"""Create or read an inventory.
289
264
If a working directory is specified, the inventory is read
290
265
from there. If the file is specified, read from that. If not,
291
266
the inventory is created empty.
293
The inventory is created with a default root directory, with
296
# We are letting Branch(init=True) create a unique inventory
297
# root id. Rather than generating a random one here.
299
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
300
self.root = RootEntry(root_id)
301
self._byid = {self.root.file_id: self.root}
270
# _tree is indexed by parent_id; at each level a map from name
271
# to ie. The None entry is the root.
272
self._tree = {None: {}}
304
275
def __iter__(self):
310
281
return len(self._byid)
313
def iter_entries(self, from_dir=None):
284
def iter_entries(self, parent_id=None):
314
285
"""Return (path, entry) pairs, in order by name."""
318
elif isinstance(from_dir, basestring):
319
from_dir = self._byid[from_dir]
321
kids = from_dir.children.items()
286
kids = self._tree[parent_id].items()
323
288
for name, ie in kids:
325
290
if ie.kind == 'directory':
326
for cn, cie in self.iter_entries(from_dir=ie.file_id):
327
yield os.path.join(name, cn), cie
331
"""Return list of (path, ie) for all entries except the root.
333
This may be faster than iter_entries.
336
def descend(dir_ie, dir_path):
337
kids = dir_ie.children.items()
339
for name, ie in kids:
340
child_path = os.path.join(dir_path, name)
341
accum.append((child_path, ie))
342
if ie.kind == 'directory':
343
descend(ie, child_path)
345
descend(self.root, '')
349
def directories(self):
350
"""Return (path, entry) pairs for all directories, including the root.
353
def descend(parent_ie, parent_path):
354
accum.append((parent_path, parent_ie))
356
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
359
for name, child_ie in kids:
360
child_path = os.path.join(parent_path, name)
361
descend(child_ie, child_path)
362
descend(self.root, '')
291
for cn, cie in self.iter_entries(parent_id=ie.file_id):
292
yield joinpath([name, cn]), cie
295
def directories(self, include_root=True):
296
"""Return (path, entry) pairs for all directories.
300
for path, entry in self.iter_entries():
301
if entry.kind == 'directory':
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
312
# TODO: return all paths and entries
367
315
def __contains__(self, file_id):
368
316
"""True if this entry contains a file with given id.
370
318
>>> inv = Inventory()
371
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
381
329
"""Return the entry for given file_id.
383
331
>>> inv = Inventory()
384
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
385
333
>>> inv['123123'].name
389
return self._byid[file_id]
392
raise BzrError("can't look up file_id None")
394
raise BzrError("file_id {%s} not in inventory" % file_id)
397
def get_file_kind(self, file_id):
398
return self._byid[file_id].kind
400
def get_child(self, parent_id, filename):
401
return self[parent_id].children.get(filename)
336
return self._byid[file_id]
404
339
def add(self, entry):
407
342
To add a file to a branch ready to be committed, use Branch.add,
408
343
which calls this."""
409
if entry.file_id in self._byid:
410
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
412
if entry.parent_id == ROOT_ID or entry.parent_id is None:
413
entry.parent_id = self.root.file_id
416
parent = self._byid[entry.parent_id]
418
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
420
if parent.children.has_key(entry.name):
421
raise BzrError("%s is already versioned" %
422
appendpath(self.id2path(parent.file_id), entry.name))
344
if entry.file_id in self:
345
bailout("inventory already contains entry with id {%s}" % entry.file_id)
347
if entry.parent_id != None:
348
if entry.parent_id not in self:
349
bailout("parent_id %s of new entry not found in inventory"
352
if self._tree[entry.parent_id].has_key(entry.name):
353
bailout("%s is already versioned"
354
% appendpath(self.id2path(entry.parent_id), entry.name))
424
356
self._byid[entry.file_id] = entry
425
parent.children[entry.name] = entry
428
def add_path(self, relpath, kind, file_id=None):
429
"""Add entry from a path.
431
The immediate parent must already be versioned"""
432
from bzrlib.errors import NotVersionedError
434
parts = bzrlib.osutils.splitpath(relpath)
436
raise BzrError("cannot re-add root of inventory")
439
from bzrlib.branch import gen_file_id
440
file_id = gen_file_id(relpath)
442
parent_path = parts[:-1]
443
parent_id = self.path2id(parent_path)
444
if parent_id == None:
445
raise NotVersionedError(parent_path)
447
ie = InventoryEntry(file_id, parts[-1],
448
kind=kind, parent_id=parent_id)
357
self._tree[entry.parent_id][entry.name] = entry
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
452
363
def __delitem__(self, file_id):
453
364
"""Remove entry by id.
455
366
>>> inv = Inventory()
456
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
459
370
>>> del inv['123']
463
374
ie = self[file_id]
465
assert self[ie.parent_id].children[ie.name] == ie
376
assert self._tree[ie.parent_id][ie.name] == ie
467
378
# TODO: Test deleting all children; maybe hoist to a separate
468
379
# deltree method?
469
380
if ie.kind == 'directory':
470
for cie in ie.children.values():
381
for cie in self._tree[file_id].values():
471
382
del self[cie.file_id]
383
del self._tree[file_id]
474
385
del self._byid[file_id]
475
del self[ie.parent_id].children[ie.name]
386
del self._tree[ie.parent_id][ie.name]
390
return Set(self._byid)
478
393
def to_element(self):
479
394
"""Convert to XML Element"""
480
from bzrlib.xml import Element
482
395
e = Element('inventory')
484
if self.root.file_id not in (None, ROOT_ID):
485
e.set('file_id', self.root.file_id)
486
397
for path, ie in self.iter_entries():
487
398
e.append(ie.to_element())
491
402
def from_element(cls, elt):
492
403
"""Construct from XML Element
494
405
>>> inv = Inventory()
495
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
496
407
>>> elt = inv.to_element()
497
408
>>> inv2 = Inventory.from_element(elt)
501
# XXXX: doctest doesn't run this properly under python2.3
502
412
assert elt.tag == 'inventory'
503
root_id = elt.get('file_id') or ROOT_ID
506
ie = InventoryEntry.from_element(e)
507
if ie.parent_id == ROOT_ID:
508
ie.parent_id = root_id
415
o.add(InventoryEntry.from_element(e))
512
418
from_element = classmethod(from_element)
515
def __eq__(self, other):
421
def __cmp__(self, other):
516
422
"""Compare two sets by comparing their contents.
518
424
>>> i1 = Inventory()
519
425
>>> i2 = Inventory()
522
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
428
>>> i1.add(InventoryEntry('123', 'foo'))
525
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
431
>>> i2.add(InventoryEntry('123', 'foo'))
529
438
if not isinstance(other, Inventory):
530
439
return NotImplemented
532
if len(self._byid) != len(other._byid):
533
# shortcut: obviously not the same
536
return self._byid == other._byid
539
def __ne__(self, other):
540
return not (self == other)
544
raise ValueError('not hashable')
548
def get_idpath(self, file_id):
549
"""Return a list of file_ids for the path to an entry.
551
The list contains one element for each directory followed by
552
the id of the file itself. So the length of the returned list
553
is equal to the depth of the file in the tree, counting the
554
root directory as depth 1.
441
if self.id_set() ^ other.id_set():
444
for file_id in self._byid:
445
c = cmp(self[file_id], other[file_id])
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
557
454
while file_id != None:
559
ie = self._byid[file_id]
561
raise BzrError("file_id {%s} not found in inventory" % file_id)
562
p.insert(0, ie.file_id)
563
457
file_id = ie.parent_id
567
def id2path(self, file_id):
568
"""Return as a list the path to file_id."""
570
# get all names, skipping root
571
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
572
return os.sep.join(p)
582
468
This returns the entry of the last component in the path,
583
469
which may be either a file or a directory.
585
Returns None iff the path is not found.
587
if isinstance(name, types.StringTypes):
588
name = splitpath(name)
590
mutter("lookup path %r" % name)
471
assert isinstance(name, types.StringTypes)
474
for f in splitpath(name):
595
cie = parent.children[f]
476
cie = self._tree[parent_id][f]
596
477
assert cie.name == f
597
assert cie.parent_id == parent.file_id
478
parent_id = cie.file_id
600
480
# or raise an error?
603
return parent.file_id
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
606
490
def has_filename(self, names):
610
494
def has_id(self, file_id):
495
assert isinstance(file_id, str)
611
496
return self._byid.has_key(file_id)
614
def rename(self, file_id, new_parent_id, new_name):
615
"""Move a file within the inventory.
617
This can change either the name, or the parent, or both.
619
This does not move the working file."""
620
if not is_valid_name(new_name):
621
raise BzrError("not an acceptable filename: %r" % new_name)
623
new_parent = self._byid[new_parent_id]
624
if new_name in new_parent.children:
625
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
627
new_parent_idpath = self.get_idpath(new_parent_id)
628
if file_id in new_parent_idpath:
629
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
630
% (self.id2path(file_id), self.id2path(new_parent_id)))
632
file_ie = self._byid[file_id]
633
old_parent = self._byid[file_ie.parent_id]
635
# TODO: Don't leave things messed up if this fails
637
del old_parent.children[file_ie.name]
638
new_parent.children[new_name] = file_ie
640
file_ie.name = new_name
641
file_ie.parent_id = new_parent_id
646
_NAME_RE = re.compile(r'^[^/\\]+$')
648
def is_valid_name(name):
649
return bool(_NAME_RE.match(name))
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)