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>"
18
# This should really be an id randomly assigned when the tree is
19
# created, but it's not for now.
24
23
import sys, os.path, types, re
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
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
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
64
59
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
62
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
63
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
64
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
65
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'))
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'))
75
70
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
81
76
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
77
>>> for j in i.iter_entries():
91
86
>>> i.id2path('2326')
92
87
'src/wibble/wibble.c'
94
: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?
95
90
But those depend on its position within a particular inventory, and
96
91
it would be nice not to need to hold the backpointer here.
98
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):
100
101
"""Create an InventoryEntry
102
103
The filename must be a single component, relative to the
103
104
parent directory; it cannot be a whole path or relative name.
105
>>> e = InventoryEntry('123', 'hello.c')
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
112
Traceback (most recent call last):
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
115
if len(splitpath(name)) != 1:
116
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
119
121
self.file_id = file_id
121
assert kind in ['file', 'directory']
123
124
self.text_id = text_id
124
125
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
127
126
if kind == 'directory':
128
127
self.children = {}
131
raise BzrError("unhandled entry kind %r" % kind)
131
135
def sorted_children(self):
138
142
other = InventoryEntry(self.file_id, self.name, self.kind,
139
self.text_id, self.parent_id)
143
self.parent_id, text_id=self.text_id)
140
144
other.text_sha1 = self.text_sha1
141
145
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
154
160
def to_element(self):
155
161
"""Convert to XML element"""
162
from bzrlib.xml import Element
156
164
e = Element('entry')
158
166
e.set('name', self.name)
159
167
e.set('file_id', self.file_id)
160
168
e.set('kind', self.kind)
162
if self.text_size is not None:
170
if self.text_size != None:
163
171
e.set('text_size', '%d' % self.text_size)
165
for f in ['text_id', 'text_sha1', 'parent_id']:
173
for f in ['text_id', 'text_sha1']:
166
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)
175
190
def from_element(cls, elt):
176
191
assert elt.tag == 'entry'
177
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)
178
201
self.text_id = elt.get('text_id')
179
202
self.text_sha1 = elt.get('text_sha1')
180
self.parent_id = elt.get('parent_id')
182
204
## mutter("read inventoryentry: %r" % (elt.attrib))
190
212
from_element = classmethod(from_element)
192
def __cmp__(self, other):
214
def __eq__(self, other):
195
215
if not isinstance(other, InventoryEntry):
196
216
return NotImplemented
198
return cmp(self.file_id, other.file_id) \
199
or cmp(self.name, other.name) \
200
or cmp(self.text_sha1, other.text_sha1) \
201
or cmp(self.text_size, other.text_size) \
202
or cmp(self.text_id, other.text_id) \
203
or cmp(self.parent_id, other.parent_id) \
204
or cmp(self.kind, other.kind)
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')
213
240
self.parent_id = None
216
def __cmp__(self, other):
243
def __eq__(self, other):
219
244
if not isinstance(other, RootEntry):
220
245
return NotImplemented
221
return cmp(self.file_id, other.file_id) \
222
or cmp(self.children, other.children)
226
class Inventory(XMLMixin):
247
return (self.file_id == other.file_id) \
248
and (self.children == other.children)
252
class Inventory(object):
227
253
"""Inventory of versioned files in a tree.
229
An Inventory acts like a set of InventoryEntry items. You can
230
also look files up by their file_id or name.
232
May be read from and written to a metadata file in a tree. To
233
manipulate the inventory (for example to add a file), it is read
234
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.
236
259
The inventory represents a typical unix file tree, with
237
260
directories containing files and subdirectories. We never store
263
283
>>> [x[0] for x in inv.iter_entries()]
266
>>> inv.write_xml(sys.stdout)
268
<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))
273
## TODO: Make sure only canonical filenames are stored.
275
## TODO: Do something sensible about the possible collisions on
276
## case-losing filesystems. Perhaps we should just always forbid
279
## TODO: No special cases for root, rather just give it a file id
280
## like everything else.
282
## TODO: Probably change XML serialization to use nesting
288
def __init__(self, root_id=ROOT_ID):
285
289
"""Create or read an inventory.
287
291
If a working directory is specified, the inventory is read
291
295
The inventory is created with a default root directory, with
294
self.root = RootEntry(None)
295
self._byid = {None: self.root}
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}
298
306
def __iter__(self):
319
327
if ie.kind == 'directory':
320
328
for cn, cie in self.iter_entries(from_dir=ie.file_id):
321
yield '/'.join((name, cn)), cie
325
def directories(self, from_dir=None):
326
"""Return (path, entry) pairs for all directories.
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.
328
def descend(parent_ie):
329
parent_name = parent_ie.name
330
yield parent_name, parent_ie
332
# directory children in sorted order
334
for ie in parent_ie.children.itervalues():
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))
335
344
if ie.kind == 'directory':
336
dn.append((ie.name, ie))
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))
339
for name, child_ie in dn:
340
for sub_name, sub_ie in descend(child_ie):
341
yield appendpath(parent_name, sub_name), sub_ie
358
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
343
for name, ie in descend(self.root):
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, '')
362
383
"""Return the entry for given file_id.
364
385
>>> inv = Inventory()
365
>>> inv.add(InventoryEntry('123123', 'hello.c'))
386
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
366
387
>>> inv['123123'].name
369
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
372
402
def get_child(self, parent_id, filename):
373
403
return self[parent_id].children.get(filename)
379
409
To add a file to a branch ready to be committed, use Branch.add,
380
410
which calls this."""
381
411
if entry.file_id in self._byid:
382
bailout("inventory already contains entry with id {%s}" % entry.file_id)
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
385
418
parent = self._byid[entry.parent_id]
387
bailout("parent_id %r not in inventory" % entry.parent_id)
420
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
389
422
if parent.children.has_key(entry.name):
390
bailout("%s is already versioned" %
423
raise BzrError("%s is already versioned" %
391
424
appendpath(self.id2path(parent.file_id), entry.name))
393
426
self._byid[entry.file_id] = entry
398
431
"""Add entry from a path.
400
433
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
401
436
parts = bzrlib.osutils.splitpath(relpath)
402
437
if len(parts) == 0:
403
bailout("cannot re-add root of inventory")
406
file_id = bzrlib.branch.gen_file_id(relpath)
408
parent_id = self.path2id(parts[:-1])
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)
409
448
ie = InventoryEntry(file_id, parts[-1],
410
449
kind=kind, parent_id=parent_id)
411
450
return self.add(ie)
437
476
del self[ie.parent_id].children[ie.name]
441
return Set(self._byid)
444
479
def to_element(self):
445
480
"""Convert to XML Element"""
481
from bzrlib.xml import Element
446
483
e = Element('inventory')
485
if self.root.file_id not in (None, ROOT_ID):
486
e.set('file_id', self.root.file_id)
448
487
for path, ie in self.iter_entries():
449
488
e.append(ie.to_element())
453
492
def from_element(cls, elt):
454
493
"""Construct from XML Element
456
495
>>> inv = Inventory()
457
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
496
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
458
497
>>> elt = inv.to_element()
459
498
>>> inv2 = Inventory.from_element(elt)
502
# XXXX: doctest doesn't run this properly under python2.3
463
503
assert elt.tag == 'inventory'
504
root_id = elt.get('file_id') or ROOT_ID
466
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
469
513
from_element = classmethod(from_element)
472
def __cmp__(self, other):
516
def __eq__(self, other):
473
517
"""Compare two sets by comparing their contents.
475
519
>>> i1 = Inventory()
476
520
>>> i2 = Inventory()
479
>>> i1.add(InventoryEntry('123', 'foo'))
523
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
482
>>> i2.add(InventoryEntry('123', 'foo'))
526
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
489
530
if not isinstance(other, Inventory):
490
531
return NotImplemented
492
if self.id_set() ^ other.id_set():
495
for file_id in self._byid:
496
c = cmp(self[file_id], other[file_id])
502
def id2path(self, file_id):
503
"""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.
505
558
while file_id != None:
506
ie = self._byid[file_id]
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)
508
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)
519
583
This returns the entry of the last component in the path,
520
584
which may be either a file or a directory.
586
Returns None iff the path is not found.
522
588
if isinstance(name, types.StringTypes):
523
589
name = splitpath(name)
591
mutter("lookup path %r" % name)
528
596
cie = parent.children[f]
529
597
assert cie.name == f
598
assert cie.parent_id == parent.file_id
532
601
# or raise an error?
551
620
This does not move the working file."""
552
621
if not is_valid_name(new_name):
553
bailout("not an acceptable filename: %r" % new_name)
622
raise BzrError("not an acceptable filename: %r" % new_name)
555
624
new_parent = self._byid[new_parent_id]
556
625
if new_name in new_parent.children:
557
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
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)))
559
633
file_ie = self._byid[file_id]
560
634
old_parent = self._byid[file_ie.parent_id]