19
19
# TODO: Maybe store inventory_id in the file? Not really needed.
21
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
22
21
__author__ = "Martin Pool <mbp@canonical.com>"
24
import sys, os.path, types
24
# This should really be an id randomly assigned when the tree is
25
# created, but it's not for now.
29
import sys, os.path, types, re
25
30
from sets import Set
30
35
from elementtree.ElementTree import Element, ElementTree, SubElement
32
37
from xml import XMLMixin
33
from errors import bailout
38
from errors import bailout, BzrError
36
41
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
60
65
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
68
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
69
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
64
70
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
73
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
74
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
75
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
76
Traceback (most recent call last):
72
78
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'))
79
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
80
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
75
81
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
85
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
81
87
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
88
>>> for j in i.iter_entries():
95
101
But those depend on its position within a particular inventory, and
96
102
it would be nice not to need to hold the backpointer here.
98
def __init__(self, file_id, name, kind='file', text_id=None,
105
# TODO: split InventoryEntry into subclasses for files,
106
# directories, etc etc.
108
def __init__(self, file_id, name, kind, parent_id, text_id=None):
100
109
"""Create an InventoryEntry
102
111
The filename must be a single component, relative to the
103
112
parent directory; it cannot be a whole path or relative name.
105
>>> e = InventoryEntry('123', 'hello.c')
114
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
119
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
120
Traceback (most recent call last):
112
121
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
126
135
self.text_size = None
127
136
if kind == 'directory':
128
137
self.children = {}
139
assert kind == 'file'
142
def sorted_children(self):
143
l = self.children.items()
132
149
other = InventoryEntry(self.file_id, self.name, self.kind,
133
self.text_id, self.parent_id)
150
self.parent_id, text_id=self.text_id)
134
151
other.text_sha1 = self.text_sha1
135
152
other.text_size = self.text_size
153
170
e.set('file_id', self.file_id)
154
171
e.set('kind', self.kind)
156
if self.text_size is not None:
173
if self.text_size != None:
157
174
e.set('text_size', '%d' % self.text_size)
159
for f in ['text_id', 'text_sha1', 'parent_id']:
176
for f in ['text_id', 'text_sha1']:
160
177
v = getattr(self, f)
181
# to be conservative, we don't externalize the root pointers
182
# for now, leaving them as null in the xml form. in a future
183
# version it will be implied by nested elements.
184
if self.parent_id != ROOT_ID:
185
assert isinstance(self.parent_id, basestring)
186
e.set('parent_id', self.parent_id)
169
193
def from_element(cls, elt):
170
194
assert elt.tag == 'entry'
171
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
196
## original format inventories don't have a parent_id for
197
## nodes in the root directory, but it's cleaner to use one
199
parent_id = elt.get('parent_id')
200
if parent_id == None:
203
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
172
204
self.text_id = elt.get('text_id')
173
205
self.text_sha1 = elt.get('text_sha1')
174
self.parent_id = elt.get('parent_id')
176
207
## mutter("read inventoryentry: %r" % (elt.attrib))
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 __cmp__(self, other):
244
if not isinstance(other, RootEntry):
245
return NotImplemented
246
return cmp(self.file_id, other.file_id) \
247
or cmp(self.children, other.children)
202
251
class Inventory(XMLMixin):
203
252
"""Inventory of versioned files in a tree.
217
266
returned quickly.
219
268
InventoryEntry objects must not be modified after they are
269
inserted, other than through the Inventory API.
222
271
>>> inv = Inventory()
223
272
>>> inv.write_xml(sys.stdout)
226
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
275
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
227
276
>>> inv['123-123'].name
263
312
If a working directory is specified, the inventory is read
264
313
from there. If the file is specified, read from that. If not,
265
314
the inventory is created empty.
316
The inventory is created with a default root directory, with
267
self._root = InventoryEntry(None, '', kind='directory')
268
self._byid = {None: self._root}
319
self.root = RootEntry(ROOT_ID)
320
self._byid = {self.root.file_id: self.root}
271
323
def __iter__(self):
277
329
return len(self._byid)
280
def iter_entries(self, parent_id=None):
332
def iter_entries(self, from_dir=None):
281
333
"""Return (path, entry) pairs, in order by name."""
282
kids = self[parent_id].children.items()
337
elif isinstance(from_dir, basestring):
338
from_dir = self._byid[from_dir]
340
kids = from_dir.children.items()
284
342
for name, ie in kids:
286
344
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):
345
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
yield '/'.join((name, cn)), cie
350
def directories(self):
292
351
"""Return (path, entry) pairs for all directories.
296
for path, entry in self.iter_entries():
297
if entry.kind == 'directory':
353
def descend(parent_ie):
354
parent_name = parent_ie.name
355
yield parent_name, parent_ie
357
# directory children in sorted order
359
for ie in parent_ie.children.itervalues():
360
if ie.kind == 'directory':
361
dn.append((ie.name, ie))
364
for name, child_ie in dn:
365
for sub_name, sub_ie in descend(child_ie):
366
yield appendpath(parent_name, sub_name), sub_ie
368
for name, ie in descend(self.root):
316
387
"""Return the entry for given file_id.
318
389
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123123', 'hello.c'))
390
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
320
391
>>> inv['123123'].name
323
return self._byid[file_id]
395
raise BzrError("can't look up file_id None")
398
return self._byid[file_id]
400
raise BzrError("file_id {%s} not in inventory" % file_id)
326
403
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)
404
return self[parent_id].children.get(filename)
333
407
def add(self, entry):
338
412
if entry.file_id in self._byid:
339
413
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)
416
parent = self._byid[entry.parent_id]
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
345
420
if parent.children.has_key(entry.name):
346
421
bailout("%s is already versioned" %
358
433
if len(parts) == 0:
359
434
bailout("cannot re-add root of inventory")
362
437
file_id = bzrlib.branch.gen_file_id(relpath)
364
439
parent_id = self.path2id(parts[:-1])
440
assert parent_id != None
365
441
ie = InventoryEntry(file_id, parts[-1],
366
442
kind=kind, parent_id=parent_id)
367
443
return self.add(ie)
410
486
"""Construct from XML Element
412
488
>>> inv = Inventory()
413
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
414
490
>>> elt = inv.to_element()
415
491
>>> inv2 = Inventory.from_element(elt)
432
508
>>> i2 = Inventory()
435
>>> i1.add(InventoryEntry('123', 'foo'))
511
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
438
>>> i2.add(InventoryEntry('123', 'foo'))
514
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
458
def id2path(self, file_id):
459
"""Return as a list the path to file_id."""
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.
461
543
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)
464
549
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:]]
475
568
This returns the entry of the last component in the path,
476
569
which may be either a file or a directory.
571
Returns None iff the path is not found.
478
573
if isinstance(name, types.StringTypes):
479
574
name = splitpath(name)
576
mutter("lookup path %r" % name)
484
581
cie = parent.children[f]
485
582
assert cie.name == f
583
assert cie.parent_id == parent.file_id
488
586
# or raise an error?
498
596
def has_id(self, file_id):
499
assert isinstance(file_id, str)
500
597
return self._byid.has_key(file_id)
506
if __name__ == '__main__':
507
import doctest, inventory
508
doctest.testmod(inventory)
600
def rename(self, file_id, new_parent_id, new_name):
601
"""Move a file within the inventory.
603
This can change either the name, or the parent, or both.
605
This does not move the working file."""
606
if not is_valid_name(new_name):
607
bailout("not an acceptable filename: %r" % new_name)
609
new_parent = self._byid[new_parent_id]
610
if new_name in new_parent.children:
611
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
file_ie = self._byid[file_id]
619
old_parent = self._byid[file_ie.parent_id]
621
# TODO: Don't leave things messed up if this fails
623
del old_parent.children[file_ie.name]
624
new_parent.children[new_name] = file_ie
626
file_ie.name = new_name
627
file_ie.parent_id = new_parent_id
632
_NAME_RE = re.compile(r'^[^/\\]+$')
634
def is_valid_name(name):
635
return bool(_NAME_RE.match(name))