15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
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
27
from xml import XMLMixin
28
from ElementTree import ElementTree, Element
29
from errors import bailout
30
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
31
from trace import mutter
33
class InventoryEntry(XMLMixin):
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
30
from bzrlib.errors import NotVersionedError
33
class InventoryEntry(object):
34
34
"""Description of a versioned file.
36
36
An InventoryEntry has the following fields, which are also
54
54
>>> i = Inventory()
56
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
57
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
58
59
>>> for j in i.iter_entries():
61
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
62
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
62
63
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
63
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
64
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
64
65
Traceback (most recent call last):
66
BzrError: ('inventory already contains entry with id {2323}', [])
67
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
68
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
67
BzrError: inventory already contains entry with id {2323}
68
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
69
70
>>> i.path2id('src/wibble')
73
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
75
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
76
77
>>> for j in i.iter_entries():
85
86
>>> i.id2path('2326')
86
87
'src/wibble/wibble.c'
88
:todo: Maybe also keep the full path of the entry, and the children?
89
TODO: Maybe also keep the full path of the entry, and the children?
89
90
But those depend on its position within a particular inventory, and
90
91
it would be nice not to need to hold the backpointer here.
92
def __init__(self, file_id, name, kind='file', text_id=None,
94
# TODO: split InventoryEntry into subclasses for files,
95
# directories, etc etc.
97
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
98
'text_id', 'parent_id', 'children', ]
100
def __init__(self, file_id, name, kind, parent_id, text_id=None):
94
101
"""Create an InventoryEntry
96
103
The filename must be a single component, relative to the
97
104
parent directory; it cannot be a whole path or relative name.
99
>>> e = InventoryEntry('123', 'hello.c')
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
104
>>> e = InventoryEntry('123', 'src/hello.c')
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
105
112
Traceback (most recent call last):
106
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
109
if len(splitpath(name)) != 1:
110
bailout('InventoryEntry name is not a simple filename: %r'
115
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
118
self.text_sha1 = None
119
self.text_size = None
113
121
self.file_id = file_id
115
assert kind in ['file', 'directory']
117
124
self.text_id = text_id
118
125
self.parent_id = parent_id
119
self.text_sha1 = None
120
self.text_size = None
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
135
def sorted_children(self):
136
l = self.children.items()
124
142
other = InventoryEntry(self.file_id, self.name, self.kind,
125
self.text_id, self.parent_id)
143
self.parent_id, text_id=self.text_id)
126
144
other.text_sha1 = self.text_sha1
127
145
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
140
160
def to_element(self):
141
161
"""Convert to XML element"""
162
from bzrlib.xml import Element
142
164
e = Element('entry')
144
166
e.set('name', self.name)
145
167
e.set('file_id', self.file_id)
146
168
e.set('kind', self.kind)
148
if self.text_size is not None:
170
if self.text_size != None:
149
171
e.set('text_size', '%d' % self.text_size)
151
for f in ['text_id', 'text_sha1', 'parent_id']:
173
for f in ['text_id', 'text_sha1']:
152
174
v = getattr(self, f)
178
# to be conservative, we don't externalize the root pointers
179
# for now, leaving them as null in the xml form. in a future
180
# version it will be implied by nested elements.
181
if self.parent_id != ROOT_ID:
182
assert isinstance(self.parent_id, basestring)
183
e.set('parent_id', self.parent_id)
161
190
def from_element(cls, elt):
162
191
assert elt.tag == 'entry'
163
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
193
## original format inventories don't have a parent_id for
194
## nodes in the root directory, but it's cleaner to use one
196
parent_id = elt.get('parent_id')
197
if parent_id == None:
200
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
164
201
self.text_id = elt.get('text_id')
165
202
self.text_sha1 = elt.get('text_sha1')
166
self.parent_id = elt.get('parent_id')
168
204
## mutter("read inventoryentry: %r" % (elt.attrib))
176
212
from_element = classmethod(from_element)
178
def __cmp__(self, other):
214
def __eq__(self, other):
181
215
if not isinstance(other, InventoryEntry):
182
216
return NotImplemented
184
return cmp(self.file_id, other.file_id) \
185
or cmp(self.name, other.name) \
186
or cmp(self.text_sha1, other.text_sha1) \
187
or cmp(self.text_size, other.text_size) \
188
or cmp(self.text_id, other.text_id) \
189
or cmp(self.parent_id, other.parent_id) \
190
or cmp(self.kind, other.kind)
194
class Inventory(XMLMixin):
218
return (self.file_id == other.file_id) \
219
and (self.name == other.name) \
220
and (self.text_sha1 == other.text_sha1) \
221
and (self.text_size == other.text_size) \
222
and (self.text_id == other.text_id) \
223
and (self.parent_id == other.parent_id) \
224
and (self.kind == other.kind)
227
def __ne__(self, other):
228
return not (self == other)
231
raise ValueError('not hashable')
235
class RootEntry(InventoryEntry):
236
def __init__(self, file_id):
237
self.file_id = file_id
239
self.kind = 'root_directory'
240
self.parent_id = None
243
def __eq__(self, other):
244
if not isinstance(other, RootEntry):
245
return NotImplemented
247
return (self.file_id == other.file_id) \
248
and (self.children == other.children)
252
class Inventory(object):
195
253
"""Inventory of versioned files in a tree.
197
An Inventory acts like a set of InventoryEntry items. You can
198
also look files up by their file_id or name.
200
May be read from and written to a metadata file in a tree. To
201
manipulate the inventory (for example to add a file), it is read
202
in, modified, and then written back out.
255
This describes which file_id is present at each point in the tree,
256
and possibly the SHA-1 or other information about the file.
257
Entries can be looked up either by path or by file_id.
204
259
The inventory represents a typical unix file tree, with
205
260
directories containing files and subdirectories. We never store
209
264
returned quickly.
211
266
InventoryEntry objects must not be modified after they are
267
inserted, other than through the Inventory API.
214
269
>>> inv = Inventory()
215
>>> inv.write_xml(sys.stdout)
218
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
270
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
219
271
>>> inv['123-123'].name
221
>>> for file_id in inv: print file_id
225
274
May be treated as an iterator or set to look up file ids:
234
283
>>> [x[0] for x in inv.iter_entries()]
237
>>> inv.write_xml(sys.stdout)
239
<entry file_id="123-123" kind="file" name="hello.c" />
285
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
286
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
244
## TODO: Clear up handling of files in subdirectories; we probably
245
## do want to be able to just look them up by name but this
246
## probably means gradually walking down the path, looking up as we go.
248
## TODO: Make sure only canonical filenames are stored.
250
## TODO: Do something sensible about the possible collisions on
251
## case-losing filesystems. Perhaps we should just always forbid
254
## _tree should probably just be stored as
255
## InventoryEntry._children on each directory.
288
def __init__(self, root_id=ROOT_ID):
258
289
"""Create or read an inventory.
260
291
If a working directory is specified, the inventory is read
261
292
from there. If the file is specified, read from that. If not,
262
293
the inventory is created empty.
295
The inventory is created with a default root directory, with
266
# _tree is indexed by parent_id; at each level a map from name
267
# to ie. The None entry is the root.
268
self._tree = {None: {}}
298
# We are letting Branch(init=True) create a unique inventory
299
# root id. Rather than generating a random one here.
301
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
302
self.root = RootEntry(root_id)
303
self._byid = {self.root.file_id: self.root}
271
306
def __iter__(self):
277
312
return len(self._byid)
280
def iter_entries(self, parent_id=None):
315
def iter_entries(self, from_dir=None):
281
316
"""Return (path, entry) pairs, in order by name."""
282
kids = self._tree[parent_id].items()
320
elif isinstance(from_dir, basestring):
321
from_dir = self._byid[from_dir]
323
kids = from_dir.children.items()
284
325
for name, ie in kids:
286
327
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':
328
for cn, cie in self.iter_entries(from_dir=ie.file_id):
329
yield os.path.join(name, cn), cie
333
"""Return list of (path, ie) for all entries except the root.
335
This may be faster than iter_entries.
338
def descend(dir_ie, dir_path):
339
kids = dir_ie.children.items()
341
for name, ie in kids:
342
child_path = os.path.join(dir_path, name)
343
accum.append((child_path, ie))
344
if ie.kind == 'directory':
345
descend(ie, child_path)
347
descend(self.root, '')
351
def directories(self):
352
"""Return (path, entry) pairs for all directories, including the root.
355
def descend(parent_ie, parent_path):
356
accum.append((parent_path, parent_ie))
358
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
361
for name, child_ie in kids:
362
child_path = os.path.join(parent_path, name)
363
descend(child_ie, child_path)
364
descend(self.root, '')
302
def children(self, parent_id):
303
"""Return entries that are direct children of parent_id."""
304
return self._tree[parent_id]
308
# TODO: return all paths and entries
311
369
def __contains__(self, file_id):
312
370
"""True if this entry contains a file with given id.
314
372
>>> inv = Inventory()
315
>>> inv.add(InventoryEntry('123', 'foo.c'))
373
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
325
383
"""Return the entry for given file_id.
327
385
>>> inv = Inventory()
328
>>> inv.add(InventoryEntry('123123', 'hello.c'))
386
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
329
387
>>> inv['123123'].name
332
return self._byid[file_id]
391
return self._byid[file_id]
394
raise BzrError("can't look up file_id None")
396
raise BzrError("file_id {%s} not in inventory" % file_id)
399
def get_file_kind(self, file_id):
400
return self._byid[file_id].kind
402
def get_child(self, parent_id, filename):
403
return self[parent_id].children.get(filename)
335
406
def add(self, entry):
338
409
To add a file to a branch ready to be committed, use Branch.add,
339
410
which calls this."""
340
if entry.file_id in self:
341
bailout("inventory already contains entry with id {%s}" % entry.file_id)
343
if entry.parent_id != None:
344
if entry.parent_id not in self:
345
bailout("parent_id %s of new entry not found in inventory"
348
if self._tree[entry.parent_id].has_key(entry.name):
349
bailout("%s is already versioned"
350
% appendpath(self.id2path(entry.parent_id), entry.name))
411
if entry.file_id in self._byid:
412
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
414
if entry.parent_id == ROOT_ID or entry.parent_id is None:
415
entry.parent_id = self.root.file_id
418
parent = self._byid[entry.parent_id]
420
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
422
if parent.children.has_key(entry.name):
423
raise BzrError("%s is already versioned" %
424
appendpath(self.id2path(parent.file_id), entry.name))
352
426
self._byid[entry.file_id] = entry
353
self._tree[entry.parent_id][entry.name] = entry
355
if entry.kind == 'directory':
356
self._tree[entry.file_id] = {}
427
parent.children[entry.name] = entry
430
def add_path(self, relpath, kind, file_id=None):
431
"""Add entry from a path.
433
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
436
parts = bzrlib.osutils.splitpath(relpath)
438
raise BzrError("cannot re-add root of inventory")
441
file_id = gen_file_id(relpath)
443
parent_path = parts[:-1]
444
parent_id = self.path2id(parent_path)
445
if parent_id == None:
446
raise NotVersionedError(parent_path)
448
ie = InventoryEntry(file_id, parts[-1],
449
kind=kind, parent_id=parent_id)
359
453
def __delitem__(self, file_id):
360
454
"""Remove entry by id.
362
456
>>> inv = Inventory()
363
>>> inv.add(InventoryEntry('123', 'foo.c'))
457
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
366
460
>>> del inv['123']
370
464
ie = self[file_id]
372
assert self._tree[ie.parent_id][ie.name] == ie
466
assert self[ie.parent_id].children[ie.name] == ie
374
468
# TODO: Test deleting all children; maybe hoist to a separate
375
469
# deltree method?
376
470
if ie.kind == 'directory':
377
for cie in self._tree[file_id].values():
471
for cie in ie.children.values():
378
472
del self[cie.file_id]
379
del self._tree[file_id]
381
475
del self._byid[file_id]
382
del self._tree[ie.parent_id][ie.name]
386
return Set(self._byid)
476
del self[ie.parent_id].children[ie.name]
389
479
def to_element(self):
390
480
"""Convert to XML Element"""
481
from bzrlib.xml import Element
391
483
e = Element('inventory')
485
if self.root.file_id not in (None, ROOT_ID):
486
e.set('file_id', self.root.file_id)
393
487
for path, ie in self.iter_entries():
394
488
e.append(ie.to_element())
398
492
def from_element(cls, elt):
399
493
"""Construct from XML Element
401
495
>>> inv = Inventory()
402
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
496
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
403
497
>>> elt = inv.to_element()
404
498
>>> inv2 = Inventory.from_element(elt)
502
# XXXX: doctest doesn't run this properly under python2.3
408
503
assert elt.tag == 'inventory'
504
root_id = elt.get('file_id') or ROOT_ID
411
o.add(InventoryEntry.from_element(e))
507
ie = InventoryEntry.from_element(e)
508
if ie.parent_id == ROOT_ID:
509
ie.parent_id = root_id
414
513
from_element = classmethod(from_element)
417
def __cmp__(self, other):
516
def __eq__(self, other):
418
517
"""Compare two sets by comparing their contents.
420
519
>>> i1 = Inventory()
421
520
>>> i2 = Inventory()
424
>>> i1.add(InventoryEntry('123', 'foo'))
523
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
427
>>> i2.add(InventoryEntry('123', 'foo'))
526
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
434
530
if not isinstance(other, Inventory):
435
531
return NotImplemented
437
if self.id_set() ^ other.id_set():
440
for file_id in self._byid:
441
c = cmp(self[file_id], other[file_id])
447
def id2path(self, file_id):
448
"""Return as a list the path to file_id."""
533
if len(self._byid) != len(other._byid):
534
# shortcut: obviously not the same
537
return self._byid == other._byid
540
def __ne__(self, other):
541
return not (self == other)
545
raise ValueError('not hashable')
549
def get_idpath(self, file_id):
550
"""Return a list of file_ids for the path to an entry.
552
The list contains one element for each directory followed by
553
the id of the file itself. So the length of the returned list
554
is equal to the depth of the file in the tree, counting the
555
root directory as depth 1.
450
558
while file_id != None:
560
ie = self._byid[file_id]
562
raise BzrError("file_id {%s} not found in inventory" % file_id)
563
p.insert(0, ie.file_id)
453
564
file_id = ie.parent_id
568
def id2path(self, file_id):
569
"""Return as a list the path to file_id."""
571
# get all names, skipping root
572
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
573
return os.sep.join(p)
464
583
This returns the entry of the last component in the path,
465
584
which may be either a file or a directory.
586
Returns None iff the path is not found.
467
assert isinstance(name, types.StringTypes)
470
for f in splitpath(name):
588
if isinstance(name, types.StringTypes):
589
name = splitpath(name)
591
mutter("lookup path %r" % name)
472
cie = self._tree[parent_id][f]
596
cie = parent.children[f]
473
597
assert cie.name == f
474
parent_id = cie.file_id
598
assert cie.parent_id == parent.file_id
476
601
# or raise an error?
482
def get_child(self, parent_id, child_name):
483
return self._tree[parent_id].get(child_name)
604
return parent.file_id
486
607
def has_filename(self, names):
490
611
def has_id(self, file_id):
491
assert isinstance(file_id, str)
492
612
return self._byid.has_key(file_id)
496
if __name__ == '__main__':
497
import doctest, inventory
498
doctest.testmod(inventory)
615
def rename(self, file_id, new_parent_id, new_name):
616
"""Move a file within the inventory.
618
This can change either the name, or the parent, or both.
620
This does not move the working file."""
621
if not is_valid_name(new_name):
622
raise BzrError("not an acceptable filename: %r" % new_name)
624
new_parent = self._byid[new_parent_id]
625
if new_name in new_parent.children:
626
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
628
new_parent_idpath = self.get_idpath(new_parent_id)
629
if file_id in new_parent_idpath:
630
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
631
% (self.id2path(file_id), self.id2path(new_parent_id)))
633
file_ie = self._byid[file_id]
634
old_parent = self._byid[file_ie.parent_id]
636
# TODO: Don't leave things messed up if this fails
638
del old_parent.children[file_ie.name]
639
new_parent.children[new_name] = file_ie
641
file_ie.name = new_name
642
file_ie.parent_id = new_parent_id
647
_NAME_RE = re.compile(r'^[^/\\]+$')
649
def is_valid_name(name):
650
return bool(_NAME_RE.match(name))
added
removed
bzrlib/inventory.py
