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>"
24
import sys, os.path, types
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
25
24
from sets import Set
30
29
from elementtree.ElementTree import Element, ElementTree, SubElement
32
31
from xml import XMLMixin
33
from errors import bailout
32
from errors import bailout, BzrError, BzrCheckError
36
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
60
59
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
62
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
63
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
64
64
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
70
Traceback (most recent call last):
72
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'))
73
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
75
75
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
81
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
82
>>> for j in i.iter_entries():
91
91
>>> i.id2path('2326')
92
92
'src/wibble/wibble.c'
94
: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?
95
95
But those depend on its position within a particular inventory, and
96
96
it would be nice not to need to hold the backpointer here.
98
def __init__(self, file_id, name, kind='file', text_id=None,
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
100
106
"""Create an InventoryEntry
102
108
The filename must be a single component, relative to the
103
109
parent directory; it cannot be a whole path or relative name.
105
>>> e = InventoryEntry('123', 'hello.c')
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
117
Traceback (most recent call last):
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
120
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
119
123
self.file_id = file_id
121
assert kind in ['file', 'directory']
123
126
self.text_id = text_id
124
127
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
127
128
if kind == 'directory':
128
129
self.children = {}
133
raise BzrError("unhandled entry kind %r" % kind)
137
def sorted_children(self):
138
l = self.children.items()
132
144
other = InventoryEntry(self.file_id, self.name, self.kind,
133
self.text_id, self.parent_id)
145
self.parent_id, text_id=self.text_id)
134
146
other.text_sha1 = self.text_sha1
135
147
other.text_size = self.text_size
153
165
e.set('file_id', self.file_id)
154
166
e.set('kind', self.kind)
156
if self.text_size is not None:
168
if self.text_size != None:
157
169
e.set('text_size', '%d' % self.text_size)
159
for f in ['text_id', 'text_sha1', 'parent_id']:
171
for f in ['text_id', 'text_sha1']:
160
172
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)
169
188
def from_element(cls, elt):
170
189
assert elt.tag == 'entry'
171
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
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)
172
199
self.text_id = elt.get('text_id')
173
200
self.text_sha1 = elt.get('text_sha1')
174
self.parent_id = elt.get('parent_id')
176
202
## mutter("read inventoryentry: %r" % (elt.attrib))
228
class RootEntry(InventoryEntry):
229
def __init__(self, file_id):
230
self.file_id = file_id
232
self.kind = 'root_directory'
233
self.parent_id = None
236
def __cmp__(self, other):
239
if not isinstance(other, RootEntry):
240
return NotImplemented
241
return cmp(self.file_id, other.file_id) \
242
or cmp(self.children, other.children)
202
246
class Inventory(XMLMixin):
203
247
"""Inventory of versioned files in a tree.
205
An Inventory acts like a set of InventoryEntry items. You can
206
also look files up by their file_id or name.
208
May be read from and written to a metadata file in a tree. To
209
manipulate the inventory (for example to add a file), it is read
210
in, modified, and then written back out.
249
This describes which file_id is present at each point in the tree,
250
and possibly the SHA-1 or other information about the file.
251
Entries can be looked up either by path or by file_id.
212
253
The inventory represents a typical unix file tree, with
213
254
directories containing files and subdirectories. We never store
217
258
returned quickly.
219
260
InventoryEntry objects must not be modified after they are
261
inserted, other than through the Inventory API.
222
263
>>> inv = Inventory()
223
264
>>> inv.write_xml(sys.stdout)
226
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
267
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
227
268
>>> inv['123-123'].name
249
## TODO: Make sure only canonical filenames are stored.
251
## TODO: Do something sensible about the possible collisions on
252
## case-losing filesystems. Perhaps we should just always forbid
255
## TODO: No special cases for root, rather just give it a file id
256
## like everything else.
258
## TODO: Probably change XML serialization to use nesting
260
289
def __init__(self):
261
290
"""Create or read an inventory.
263
292
If a working directory is specified, the inventory is read
264
293
from there. If the file is specified, read from that. If not,
265
294
the inventory is created empty.
296
The inventory is created with a default root directory, with
267
self._root = InventoryEntry(None, '', kind='directory')
268
self._byid = {None: self._root}
299
self.root = RootEntry(ROOT_ID)
300
self._byid = {self.root.file_id: self.root}
271
303
def __iter__(self):
277
309
return len(self._byid)
280
def iter_entries(self, parent_id=None):
312
def iter_entries(self, from_dir=None):
281
313
"""Return (path, entry) pairs, in order by name."""
282
kids = self[parent_id].children.items()
317
elif isinstance(from_dir, basestring):
318
from_dir = self._byid[from_dir]
320
kids = from_dir.children.items()
284
322
for name, ie in kids:
286
324
if ie.kind == 'directory':
287
for cn, cie in self.iter_entries(parent_id=ie.file_id):
288
yield joinpath([name, cn]), cie
325
for cn, cie in self.iter_entries(from_dir=ie.file_id):
326
yield os.path.join(name, cn), cie
291
330
def directories(self):
292
331
"""Return (path, entry) pairs for all directories.
295
for path, entry in self.iter_entries():
296
if entry.kind == 'directory':
333
def descend(parent_ie):
334
parent_name = parent_ie.name
335
yield parent_name, parent_ie
337
# directory children in sorted order
339
for ie in parent_ie.children.itervalues():
340
if ie.kind == 'directory':
341
dn.append((ie.name, ie))
344
for name, child_ie in dn:
345
for sub_name, sub_ie in descend(child_ie):
346
yield appendpath(parent_name, sub_name), sub_ie
348
for name, ie in descend(self.root):
315
367
"""Return the entry for given file_id.
317
369
>>> inv = Inventory()
318
>>> inv.add(InventoryEntry('123123', 'hello.c'))
370
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
319
371
>>> inv['123123'].name
322
return self._byid[file_id]
375
return self._byid[file_id]
378
raise BzrError("can't look up file_id None")
380
raise BzrError("file_id {%s} not in inventory" % file_id)
383
def get_file_kind(self, file_id):
384
return self._byid[file_id].kind
325
386
def get_child(self, parent_id, filename):
326
if parent_id == None:
327
return self._root.children.get(filename)
329
return self[parent_id].children.get(filename)
387
return self[parent_id].children.get(filename)
332
390
def add(self, entry):
337
395
if entry.file_id in self._byid:
338
396
bailout("inventory already contains entry with id {%s}" % entry.file_id)
340
parent = self._byid[entry.parent_id]
341
if parent.kind != 'directory':
342
bailout("attempt to add under non-directory {%s}" % parent.file_id)
399
parent = self._byid[entry.parent_id]
401
bailout("parent_id {%s} not in inventory" % entry.parent_id)
344
403
if parent.children.has_key(entry.name):
345
404
bailout("%s is already versioned" %
357
416
if len(parts) == 0:
358
417
bailout("cannot re-add root of inventory")
361
420
file_id = bzrlib.branch.gen_file_id(relpath)
363
422
parent_id = self.path2id(parts[:-1])
423
assert parent_id != None
364
424
ie = InventoryEntry(file_id, parts[-1],
365
425
kind=kind, parent_id=parent_id)
366
426
return self.add(ie)
409
469
"""Construct from XML Element
411
471
>>> inv = Inventory()
412
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
472
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
413
473
>>> elt = inv.to_element()
414
474
>>> inv2 = Inventory.from_element(elt)
431
491
>>> i2 = Inventory()
434
>>> i1.add(InventoryEntry('123', 'foo'))
494
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
437
>>> i2.add(InventoryEntry('123', 'foo'))
497
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
457
def id2path(self, file_id):
458
"""Return as a list the path to file_id."""
517
def get_idpath(self, file_id):
518
"""Return a list of file_ids for the path to an entry.
520
The list contains one element for each directory followed by
521
the id of the file itself. So the length of the returned list
522
is equal to the depth of the file in the tree, counting the
523
root directory as depth 1.
460
526
while file_id != None:
528
ie = self._byid[file_id]
530
bailout("file_id {%s} not found in inventory" % file_id)
531
p.insert(0, ie.file_id)
463
532
file_id = ie.parent_id
536
def id2path(self, file_id):
537
"""Return as a list the path to file_id."""
539
# get all names, skipping root
540
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
541
return os.sep.join(p)
474
551
This returns the entry of the last component in the path,
475
552
which may be either a file or a directory.
554
Returns None iff the path is not found.
477
556
if isinstance(name, types.StringTypes):
478
557
name = splitpath(name)
559
mutter("lookup path %r" % name)
483
564
cie = parent.children[f]
484
565
assert cie.name == f
566
assert cie.parent_id == parent.file_id
487
569
# or raise an error?
497
579
def has_id(self, file_id):
498
assert isinstance(file_id, str)
499
580
return self._byid.has_key(file_id)
505
if __name__ == '__main__':
506
import doctest, inventory
507
doctest.testmod(inventory)
583
def rename(self, file_id, new_parent_id, new_name):
584
"""Move a file within the inventory.
586
This can change either the name, or the parent, or both.
588
This does not move the working file."""
589
if not is_valid_name(new_name):
590
bailout("not an acceptable filename: %r" % new_name)
592
new_parent = self._byid[new_parent_id]
593
if new_name in new_parent.children:
594
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
596
new_parent_idpath = self.get_idpath(new_parent_id)
597
if file_id in new_parent_idpath:
598
bailout("cannot move directory %r into a subdirectory of itself, %r"
599
% (self.id2path(file_id), self.id2path(new_parent_id)))
601
file_ie = self._byid[file_id]
602
old_parent = self._byid[file_ie.parent_id]
604
# TODO: Don't leave things messed up if this fails
606
del old_parent.children[file_ie.name]
607
new_parent.children[new_name] = file_ie
609
file_ie.name = new_name
610
file_ie.parent_id = new_parent_id
615
_NAME_RE = re.compile(r'^[^/\\]+$')
617
def is_valid_name(name):
618
return bool(_NAME_RE.match(name))