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
18
# This should really be an id randomly assigned when the tree is
19
# created, but it's not for now.
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>"
23
24
import sys, os.path, types, re
26
28
from cElementTree import Element, ElementTree, SubElement
28
30
from elementtree.ElementTree import Element, ElementTree, SubElement
30
32
from xml import XMLMixin
31
from errors import bailout, BzrError, BzrCheckError
33
from errors import bailout
34
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
58
60
>>> i = Inventory()
61
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
62
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
63
64
>>> for j in i.iter_entries():
66
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
67
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
68
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
69
70
Traceback (most recent call last):
71
72
BzrError: ('inventory already contains entry with id {2323}', [])
72
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
73
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
73
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
74
75
>>> i.path2id('src/wibble')
78
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
80
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
81
82
>>> for j in i.iter_entries():
90
91
>>> i.id2path('2326')
91
92
'src/wibble/wibble.c'
93
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?
94
95
But those depend on its position within a particular inventory, and
95
96
it would be nice not to need to hold the backpointer here.
98
# TODO: split InventoryEntry into subclasses for files,
99
# directories, etc etc.
104
def __init__(self, file_id, name, kind, parent_id, text_id=None):
98
def __init__(self, file_id, name, kind='file', text_id=None,
105
100
"""Create an InventoryEntry
107
102
The filename must be a single component, relative to the
108
103
parent directory; it cannot be a whole path or relative name.
110
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
105
>>> e = InventoryEntry('123', 'hello.c')
115
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
116
111
Traceback (most recent call last):
117
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
119
if '/' in name or '\\' in name:
120
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
122
119
self.file_id = file_id
121
assert kind in ['file', 'directory']
125
123
self.text_id = text_id
126
124
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
127
127
if kind == 'directory':
128
128
self.children = {}
132
raise BzrError("unhandled entry kind %r" % kind)
136
131
def sorted_children(self):
143
138
other = InventoryEntry(self.file_id, self.name, self.kind,
144
self.parent_id, text_id=self.text_id)
139
self.text_id, self.parent_id)
145
140
other.text_sha1 = self.text_sha1
146
141
other.text_size = self.text_size
147
# note that children are *not* copied; they're pulled across when
166
159
e.set('file_id', self.file_id)
167
160
e.set('kind', self.kind)
169
if self.text_size != None:
162
if self.text_size is not None:
170
163
e.set('text_size', '%d' % self.text_size)
172
for f in ['text_id', 'text_sha1']:
165
for f in ['text_id', 'text_sha1', 'parent_id']:
173
166
v = getattr(self, f)
177
# to be conservative, we don't externalize the root pointers
178
# for now, leaving them as null in the xml form. in a future
179
# version it will be implied by nested elements.
180
if self.parent_id != ROOT_ID:
181
assert isinstance(self.parent_id, basestring)
182
e.set('parent_id', self.parent_id)
189
175
def from_element(cls, elt):
190
176
assert elt.tag == 'entry'
192
## original format inventories don't have a parent_id for
193
## nodes in the root directory, but it's cleaner to use one
195
parent_id = elt.get('parent_id')
196
if parent_id == None:
199
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
177
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
200
178
self.text_id = elt.get('text_id')
201
179
self.text_sha1 = elt.get('text_sha1')
180
self.parent_id = elt.get('parent_id')
203
182
## mutter("read inventoryentry: %r" % (elt.attrib))
211
190
from_element = classmethod(from_element)
213
def __eq__(self, other):
192
def __cmp__(self, other):
214
195
if not isinstance(other, InventoryEntry):
215
196
return NotImplemented
217
return (self.file_id == other.file_id) \
218
and (self.name == other.name) \
219
and (self.text_sha1 == other.text_sha1) \
220
and (self.text_size == other.text_size) \
221
and (self.text_id == other.text_id) \
222
and (self.parent_id == other.parent_id) \
223
and (self.kind == other.kind)
226
def __ne__(self, other):
227
return not (self == other)
230
raise ValueError('not hashable')
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)
239
213
self.parent_id = None
242
def __eq__(self, other):
216
def __cmp__(self, other):
243
219
if not isinstance(other, RootEntry):
244
220
return NotImplemented
246
return (self.file_id == other.file_id) \
247
and (self.children == other.children)
221
return cmp(self.file_id, other.file_id) \
222
or cmp(self.children, other.children)
251
226
class Inventory(XMLMixin):
252
227
"""Inventory of versioned files in a tree.
254
This describes which file_id is present at each point in the tree,
255
and possibly the SHA-1 or other information about the file.
256
Entries can be looked up either by path or by file_id.
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.
258
236
The inventory represents a typical unix file tree, with
259
237
directories containing files and subdirectories. We never store
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
294
284
def __init__(self):
295
285
"""Create or read an inventory.
301
291
The inventory is created with a default root directory, with
304
self.root = RootEntry(ROOT_ID)
305
self._byid = {self.root.file_id: self.root}
294
self.root = RootEntry(None)
295
self._byid = {None: self.root}
308
298
def __iter__(self):
329
319
if ie.kind == 'directory':
330
320
for cn, cie in self.iter_entries(from_dir=ie.file_id):
331
yield os.path.join(name, cn), cie
335
"""Return list of (path, ie) for all entries except the root.
337
This may be faster than iter_entries.
321
yield '/'.join((name, cn)), cie
325
def directories(self, from_dir=None):
326
"""Return (path, entry) pairs for all directories.
340
def descend(dir_ie, dir_path):
341
kids = dir_ie.children.items()
343
for name, ie in kids:
344
child_path = os.path.join(dir_path, name)
345
accum.append((child_path, ie))
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():
346
335
if ie.kind == 'directory':
347
descend(ie, child_path)
349
descend(self.root, '')
353
def directories(self):
354
"""Return (path, entry) pairs for all directories, including the root.
357
def descend(parent_ie, parent_path):
358
accum.append((parent_path, parent_ie))
336
dn.append((ie.name, ie))
360
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
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
363
for name, child_ie in kids:
364
child_path = os.path.join(parent_path, name)
365
descend(child_ie, child_path)
366
descend(self.root, '')
343
for name, ie in descend(self.root):
385
362
"""Return the entry for given file_id.
387
364
>>> inv = Inventory()
388
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
365
>>> inv.add(InventoryEntry('123123', 'hello.c'))
389
366
>>> inv['123123'].name
393
return self._byid[file_id]
396
raise BzrError("can't look up file_id None")
398
raise BzrError("file_id {%s} not in inventory" % file_id)
401
def get_file_kind(self, file_id):
402
return self._byid[file_id].kind
369
return self._byid[file_id]
404
372
def get_child(self, parent_id, filename):
405
373
return self[parent_id].children.get(filename)
417
385
parent = self._byid[entry.parent_id]
419
bailout("parent_id {%s} not in inventory" % entry.parent_id)
387
bailout("parent_id %r not in inventory" % entry.parent_id)
421
389
if parent.children.has_key(entry.name):
422
390
bailout("%s is already versioned" %
434
402
if len(parts) == 0:
435
403
bailout("cannot re-add root of inventory")
438
406
file_id = bzrlib.branch.gen_file_id(relpath)
440
408
parent_id = self.path2id(parts[:-1])
441
assert parent_id != None
442
409
ie = InventoryEntry(file_id, parts[-1],
443
410
kind=kind, parent_id=parent_id)
444
411
return self.add(ie)
483
454
"""Construct from XML Element
485
456
>>> inv = Inventory()
486
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
457
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
487
458
>>> elt = inv.to_element()
488
459
>>> inv2 = Inventory.from_element(elt)
498
469
from_element = classmethod(from_element)
501
def __eq__(self, other):
472
def __cmp__(self, other):
502
473
"""Compare two sets by comparing their contents.
504
475
>>> i1 = Inventory()
505
476
>>> i2 = Inventory()
508
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
479
>>> i1.add(InventoryEntry('123', 'foo'))
511
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
482
>>> i2.add(InventoryEntry('123', 'foo'))
515
489
if not isinstance(other, Inventory):
516
490
return NotImplemented
518
if len(self._byid) != len(other._byid):
519
# shortcut: obviously not the same
522
return self._byid == other._byid
525
def __ne__(self, other):
526
return not (self == other)
530
raise ValueError('not hashable')
534
def get_idpath(self, file_id):
535
"""Return a list of file_ids for the path to an entry.
537
The list contains one element for each directory followed by
538
the id of the file itself. So the length of the returned list
539
is equal to the depth of the file in the tree, counting the
540
root directory as depth 1.
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."""
543
505
while file_id != None:
545
ie = self._byid[file_id]
547
bailout("file_id {%s} not found in inventory" % file_id)
548
p.insert(0, ie.file_id)
506
ie = self._byid[file_id]
549
508
file_id = ie.parent_id
553
def id2path(self, file_id):
554
"""Return as a list the path to file_id."""
556
# get all names, skipping root
557
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
558
return os.sep.join(p)
568
519
This returns the entry of the last component in the path,
569
520
which may be either a file or a directory.
571
Returns None iff the path is not found.
573
522
if isinstance(name, types.StringTypes):
574
523
name = splitpath(name)
576
mutter("lookup path %r" % name)
581
528
cie = parent.children[f]
582
529
assert cie.name == f
583
assert cie.parent_id == parent.file_id
586
532
# or raise an error?
610
556
if new_name in new_parent.children:
611
557
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
613
new_parent_idpath = self.get_idpath(new_parent_id)
614
if file_id in new_parent_idpath:
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
616
% (self.id2path(file_id), self.id2path(new_parent_id)))
618
559
file_ie = self._byid[file_id]
619
560
old_parent = self._byid[file_ie.parent_id]