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
# TODO: Maybe also keep the full path of the entry, and the children?
19
# But those depend on its position within a particular inventory, and
20
# it would be nice not to need to hold the backpointer here.
22
# TODO: Perhaps split InventoryEntry into subclasses for files,
23
# directories, etc etc.
26
# This should really be an id randomly assigned when the tree is
27
# created, but it's not for now.
31
import sys, os.path, types, re
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
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
34
from bzrlib.errors import BzrError, BzrCheckError
36
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
37
37
from bzrlib.trace import mutter
38
from bzrlib.errors import NotVersionedError
41
class InventoryEntry(object):
39
class InventoryEntry(XMLMixin):
42
40
"""Description of a versioned file.
44
42
An InventoryEntry has the following fields, which are also
45
43
present in the XML inventory-entry element:
50
(within the parent directory)
56
file_id of the parent directory, or ROOT_ID
59
the revision_id in which the name or parent of this file was
63
sha-1 of the text of the file
66
size in bytes of the text of the file
69
the revision_id in which the text of this file was introduced
71
(reading a version 4 tree created a text_id field.)
46
* *name*: (only the basename within the directory, must not
48
* *kind*: "directory" or "file"
49
* *directory_id*: (if absent/null means the branch root directory)
50
* *text_sha1*: only for files
51
* *text_size*: in bytes, only for files
52
* *text_id*: identifier for the text version, only for files
54
InventoryEntries can also exist inside a WorkingTree
55
inventory, in which case they are not yet bound to a
56
particular revision of the file. In that case the text_sha1,
57
text_size and text_id are absent.
73
60
>>> i = Inventory()
76
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
77
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
78
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
79
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
80
64
>>> for j in i.iter_entries():
83
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
84
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
85
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
86
70
Traceback (most recent call last):
88
BzrError: inventory already contains entry with id {2323}
89
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
90
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
91
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
92
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
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'))
93
75
>>> i.path2id('src/wibble')
97
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
98
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
100
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
101
82
>>> for j in i.iter_entries():
109
90
src/wibble/wibble.c
110
91
>>> i.id2path('2326')
111
92
'src/wibble/wibble.c'
94
:todo: Maybe also keep the full path of the entry, and the children?
95
But those depend on its position within a particular inventory, and
96
it would be nice not to need to hold the backpointer here.
114
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
115
'text_id', 'parent_id', 'children',
116
'text_version', 'entry_version', ]
119
def __init__(self, file_id, name, kind, parent_id, text_id=None):
98
def __init__(self, file_id, name, kind='file', text_id=None,
120
100
"""Create an InventoryEntry
122
102
The filename must be a single component, relative to the
123
103
parent directory; it cannot be a whole path or relative name.
125
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
105
>>> e = InventoryEntry('123', 'hello.c')
130
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
131
111
Traceback (most recent call last):
132
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
134
assert isinstance(name, basestring), name
135
if '/' in name or '\\' in name:
136
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
138
self.text_version = None
139
self.entry_version = None
140
self.text_sha1 = None
141
self.text_size = None
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
142
119
self.file_id = file_id
121
assert kind in ['file', 'directory']
145
123
self.text_id = text_id
146
124
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
147
127
if kind == 'directory':
148
128
self.children = {}
152
raise BzrError("unhandled entry kind %r" % kind)
156
def sorted_children(self):
157
l = self.children.items()
163
132
other = InventoryEntry(self.file_id, self.name, self.kind,
165
other.text_id = self.text_id
133
self.text_id, self.parent_id)
166
134
other.text_sha1 = self.text_sha1
167
135
other.text_size = self.text_size
168
other.text_version = self.text_version
169
# note that children are *not* copied; they're pulled across when
183
def __eq__(self, other):
148
def to_element(self):
149
"""Convert to XML element"""
152
e.set('name', self.name)
153
e.set('file_id', self.file_id)
154
e.set('kind', self.kind)
156
if self.text_size is not None:
157
e.set('text_size', '%d' % self.text_size)
159
for f in ['text_id', 'text_sha1', 'parent_id']:
169
def from_element(cls, elt):
170
assert elt.tag == 'entry'
171
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
172
self.text_id = elt.get('text_id')
173
self.text_sha1 = elt.get('text_sha1')
174
self.parent_id = elt.get('parent_id')
176
## mutter("read inventoryentry: %r" % (elt.attrib))
178
v = elt.get('text_size')
179
self.text_size = v and int(v)
184
from_element = classmethod(from_element)
186
def __cmp__(self, other):
184
189
if not isinstance(other, InventoryEntry):
185
190
return NotImplemented
187
return (self.file_id == other.file_id) \
188
and (self.name == other.name) \
189
and (self.text_sha1 == other.text_sha1) \
190
and (self.text_size == other.text_size) \
191
and (self.text_id == other.text_id) \
192
and (self.parent_id == other.parent_id) \
193
and (self.kind == other.kind) \
194
and (self.text_version == other.text_version) \
195
and (self.entry_version == other.entry_version)
198
def __ne__(self, other):
199
return not (self == other)
202
raise ValueError('not hashable')
206
class RootEntry(InventoryEntry):
207
def __init__(self, file_id):
208
self.file_id = file_id
210
self.kind = 'root_directory'
211
self.parent_id = None
214
def __eq__(self, other):
215
if not isinstance(other, RootEntry):
216
return NotImplemented
218
return (self.file_id == other.file_id) \
219
and (self.children == other.children)
223
class Inventory(object):
192
return cmp(self.file_id, other.file_id) \
193
or cmp(self.name, other.name) \
194
or cmp(self.text_sha1, other.text_sha1) \
195
or cmp(self.text_size, other.text_size) \
196
or cmp(self.text_id, other.text_id) \
197
or cmp(self.parent_id, other.parent_id) \
198
or cmp(self.kind, other.kind)
202
class Inventory(XMLMixin):
224
203
"""Inventory of versioned files in a tree.
226
This describes which file_id is present at each point in the tree,
227
and possibly the SHA-1 or other information about the file.
228
Entries can be looked up either by path or by file_id.
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.
230
212
The inventory represents a typical unix file tree, with
231
213
directories containing files and subdirectories. We never store
255
239
>>> [x[0] for x in inv.iter_entries()]
257
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
258
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
259
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT-12345678-12345678')
242
>>> inv.write_xml(sys.stdout)
244
<entry file_id="123-123" kind="file" name="hello.c" />
261
def __init__(self, root_id=ROOT_ID):
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
262
261
"""Create or read an inventory.
264
263
If a working directory is specified, the inventory is read
265
264
from there. If the file is specified, read from that. If not,
266
265
the inventory is created empty.
268
The inventory is created with a default root directory, with
271
# We are letting Branch(init=True) create a unique inventory
272
# root id. Rather than generating a random one here.
274
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
275
self.root = RootEntry(root_id)
276
self._byid = {self.root.file_id: self.root}
280
other = Inventory(self.root.file_id)
281
# copy recursively so we know directories will be added before
282
# their children. There are more efficient ways than this...
283
for path, entry in self.iter_entries():
284
if entry == self.root:
286
other.add(entry.copy())
267
self._root = InventoryEntry(None, '', kind='directory')
268
self._byid = {None: self._root}
290
271
def __iter__(self):
296
277
return len(self._byid)
299
def iter_entries(self, from_dir=None):
280
def iter_entries(self, parent_id=None):
300
281
"""Return (path, entry) pairs, in order by name."""
304
elif isinstance(from_dir, basestring):
305
from_dir = self._byid[from_dir]
307
kids = from_dir.children.items()
282
kids = self[parent_id].children.items()
309
284
for name, ie in kids:
311
286
if ie.kind == 'directory':
312
for cn, cie in self.iter_entries(from_dir=ie.file_id):
313
yield os.path.join(name, cn), cie
317
"""Return list of (path, ie) for all entries except the root.
319
This may be faster than iter_entries.
322
def descend(dir_ie, dir_path):
323
kids = dir_ie.children.items()
325
for name, ie in kids:
326
child_path = os.path.join(dir_path, name)
327
accum.append((child_path, ie))
328
if ie.kind == 'directory':
329
descend(ie, child_path)
331
descend(self.root, '')
335
def directories(self):
336
"""Return (path, entry) pairs for all directories, including the root.
339
def descend(parent_ie, parent_path):
340
accum.append((parent_path, parent_ie))
342
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
345
for name, child_ie in kids:
346
child_path = os.path.join(parent_path, name)
347
descend(child_ie, child_path)
348
descend(self.root, '')
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':
368
316
"""Return the entry for given file_id.
370
318
>>> inv = Inventory()
371
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
372
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
319
>>> inv.add(InventoryEntry('123123', 'hello.c'))
373
320
>>> inv['123123'].name
377
return self._byid[file_id]
380
raise BzrError("can't look up file_id None")
382
raise BzrError("file_id {%s} not in inventory" % file_id)
385
def get_file_kind(self, file_id):
386
return self._byid[file_id].kind
323
return self._byid[file_id]
388
326
def get_child(self, parent_id, filename):
389
return self[parent_id].children.get(filename)
327
if parent_id == None:
328
return self._root.children.get(filename)
330
return self[parent_id].children.get(filename)
392
333
def add(self, entry):
393
334
"""Add entry to inventory.
395
336
To add a file to a branch ready to be committed, use Branch.add,
398
Returns the new entry object.
400
if entry.file_id in self._byid:
401
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
403
if entry.parent_id == ROOT_ID or entry.parent_id is None:
404
entry.parent_id = self.root.file_id
407
parent = self._byid[entry.parent_id]
409
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
411
if parent.children.has_key(entry.name):
412
raise BzrError("%s is already versioned" %
413
appendpath(self.id2path(parent.file_id), entry.name))
338
if entry.file_id in self:
339
bailout("inventory already contains entry with id {%s}" % entry.file_id)
341
if entry.parent_id != None:
342
if entry.parent_id not in self:
343
bailout("parent_id %s of new entry not found in inventory"
345
# TODO: parent must be a directory
347
if self[entry.parent_id].children.has_key(entry.name):
348
bailout("%s is already versioned"
349
% appendpath(self.id2path(entry.parent_id), entry.name))
415
351
self._byid[entry.file_id] = entry
416
parent.children[entry.name] = entry
352
self[entry.parent_id].children[entry.name] = entry
420
355
def add_path(self, relpath, kind, file_id=None):
421
356
"""Add entry from a path.
423
The immediate parent must already be versioned.
425
Returns the new entry object."""
426
from bzrlib.branch import gen_file_id
358
The immediate parent must already be versioned"""
428
359
parts = bzrlib.osutils.splitpath(relpath)
429
360
if len(parts) == 0:
430
raise BzrError("cannot re-add root of inventory")
433
file_id = gen_file_id(relpath)
435
parent_path = parts[:-1]
436
parent_id = self.path2id(parent_path)
437
if parent_id == None:
438
raise NotVersionedError(parent_path)
361
bailout("cannot re-add root of inventory")
364
file_id = bzrlib.branch.gen_file_id(relpath)
366
parent_id = self.path2id(parts[:-1])
440
367
ie = InventoryEntry(file_id, parts[-1],
441
368
kind=kind, parent_id=parent_id)
442
369
return self.add(ie)
469
395
del self[ie.parent_id].children[ie.name]
472
def __eq__(self, other):
399
return Set(self._byid)
402
def to_element(self):
403
"""Convert to XML Element"""
404
e = Element('inventory')
406
for path, ie in self.iter_entries():
407
e.append(ie.to_element())
411
def from_element(cls, elt):
412
"""Construct from XML Element
414
>>> inv = Inventory()
415
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
416
>>> elt = inv.to_element()
417
>>> inv2 = Inventory.from_element(elt)
421
assert elt.tag == 'inventory'
424
o.add(InventoryEntry.from_element(e))
427
from_element = classmethod(from_element)
430
def __cmp__(self, other):
473
431
"""Compare two sets by comparing their contents.
475
433
>>> i1 = Inventory()
476
434
>>> i2 = Inventory()
479
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
480
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
437
>>> i1.add(InventoryEntry('123', 'foo'))
483
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
484
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
440
>>> i2.add(InventoryEntry('123', 'foo'))
488
447
if not isinstance(other, Inventory):
489
448
return NotImplemented
491
if len(self._byid) != len(other._byid):
492
# shortcut: obviously not the same
495
return self._byid == other._byid
498
def __ne__(self, other):
499
return not (self == other)
503
raise ValueError('not hashable')
506
def get_idpath(self, file_id):
507
"""Return a list of file_ids for the path to an entry.
509
The list contains one element for each directory followed by
510
the id of the file itself. So the length of the returned list
511
is equal to the depth of the file in the tree, counting the
512
root directory as depth 1.
450
if self.id_set() ^ other.id_set():
453
for file_id in self._byid:
454
c = cmp(self[file_id], other[file_id])
460
def id2path(self, file_id):
461
"""Return as a list the path to file_id."""
515
463
while file_id != None:
517
ie = self._byid[file_id]
519
raise BzrError("file_id {%s} not found in inventory" % file_id)
520
p.insert(0, ie.file_id)
521
466
file_id = ie.parent_id
525
def id2path(self, file_id):
526
"""Return as a list the path to file_id."""
528
# get all names, skipping root
529
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
530
return os.sep.join(p)
568
500
def has_id(self, file_id):
501
assert isinstance(file_id, str)
569
502
return self._byid.has_key(file_id)
572
def rename(self, file_id, new_parent_id, new_name):
573
"""Move a file within the inventory.
575
This can change either the name, or the parent, or both.
577
This does not move the working file."""
578
if not is_valid_name(new_name):
579
raise BzrError("not an acceptable filename: %r" % new_name)
581
new_parent = self._byid[new_parent_id]
582
if new_name in new_parent.children:
583
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
585
new_parent_idpath = self.get_idpath(new_parent_id)
586
if file_id in new_parent_idpath:
587
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
588
% (self.id2path(file_id), self.id2path(new_parent_id)))
590
file_ie = self._byid[file_id]
591
old_parent = self._byid[file_ie.parent_id]
593
# TODO: Don't leave things messed up if this fails
595
del old_parent.children[file_ie.name]
596
new_parent.children[new_name] = file_ie
598
file_ie.name = new_name
599
file_ie.parent_id = new_parent_id
606
def is_valid_name(name):
609
_NAME_RE = re.compile(r'^[^/\\]+$')
611
return bool(_NAME_RE.match(name))
508
if __name__ == '__main__':
509
import doctest, inventory
510
doctest.testmod(inventory)