15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
"""Inventories map files to their name in a revision."""
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 osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
37
class InventoryEntry(XMLMixin):
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
34
from bzrlib.errors import BzrError, BzrCheckError
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
37
from bzrlib.trace import mutter
38
from bzrlib.errors import NotVersionedError
41
class InventoryEntry(object):
38
42
"""Description of a versioned file.
40
44
An InventoryEntry has the following fields, which are also
41
45
present in the XML inventory-entry element:
44
* *name*: (only the basename within the directory, must not
46
* *kind*: "directory" or "file"
47
* *directory_id*: (if absent/null means the branch root directory)
48
* *text_sha1*: only for files
49
* *text_size*: in bytes, only for files
50
* *text_id*: identifier for the text version, only for files
52
InventoryEntries can also exist inside a WorkingTree
53
inventory, in which case they are not yet bound to a
54
particular revision of the file. In that case the text_sha1,
55
text_size and text_id are absent.
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.)
58
73
>>> i = Inventory()
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
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
80
>>> for j in i.iter_entries():
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
83
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
66
84
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
85
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
68
86
Traceback (most recent call last):
70
BzrError: ('inventory already contains entry with id {2323}', [])
71
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
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')
73
93
>>> i.path2id('src/wibble')
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
97
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
98
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
79
100
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
80
101
>>> for j in i.iter_entries():
88
109
src/wibble/wibble.c
89
110
>>> i.id2path('2326')
90
111
'src/wibble/wibble.c'
92
:todo: Maybe also keep the full path of the entry, and the children?
93
But those depend on its position within a particular inventory, and
94
it would be nice not to need to hold the backpointer here.
96
def __init__(self, file_id, name, kind='file', text_id=None,
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
120
"""Create an InventoryEntry
100
122
The filename must be a single component, relative to the
101
123
parent directory; it cannot be a whole path or relative name.
103
>>> e = InventoryEntry('123', 'hello.c')
125
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
108
>>> e = InventoryEntry('123', 'src/hello.c')
130
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
109
131
Traceback (most recent call last):
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
132
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
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
117
142
self.file_id = file_id
119
assert kind in ['file', 'directory']
121
145
self.text_id = text_id
122
146
self.parent_id = parent_id
123
self.text_sha1 = None
124
self.text_size = None
147
if kind == 'directory':
152
raise BzrError("unhandled entry kind %r" % kind)
156
def sorted_children(self):
157
l = self.children.items()
128
163
other = InventoryEntry(self.file_id, self.name, self.kind,
129
self.text_id, self.parent_id)
165
other.text_id = self.text_id
130
166
other.text_sha1 = self.text_sha1
131
167
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
144
def to_element(self):
145
"""Convert to XML element"""
148
e.set('name', self.name)
149
e.set('file_id', self.file_id)
150
e.set('kind', self.kind)
152
if self.text_size is not None:
153
e.set('text_size', '%d' % self.text_size)
155
for f in ['text_id', 'text_sha1', 'parent_id']:
165
def from_element(cls, elt):
166
assert elt.tag == 'entry'
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
168
self.text_id = elt.get('text_id')
169
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
172
## mutter("read inventoryentry: %r" % (elt.attrib))
174
v = elt.get('text_size')
175
self.text_size = v and int(v)
180
from_element = classmethod(from_element)
182
def __cmp__(self, other):
183
def __eq__(self, other):
185
184
if not isinstance(other, InventoryEntry):
186
185
return NotImplemented
188
return cmp(self.file_id, other.file_id) \
189
or cmp(self.name, other.name) \
190
or cmp(self.text_sha1, other.text_sha1) \
191
or cmp(self.text_size, other.text_size) \
192
or cmp(self.text_id, other.text_id) \
193
or cmp(self.parent_id, other.parent_id) \
194
or cmp(self.kind, other.kind)
198
class Inventory(XMLMixin):
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):
199
224
"""Inventory of versioned files in a tree.
201
An Inventory acts like a set of InventoryEntry items. You can
202
also look files up by their file_id or name.
204
May be read from and written to a metadata file in a tree. To
205
manipulate the inventory (for example to add a file), it is read
206
in, modified, and then written back out.
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.
208
230
The inventory represents a typical unix file tree, with
209
231
directories containing files and subdirectories. We never store
238
255
>>> [x[0] for x in inv.iter_entries()]
241
>>> inv.write_xml(sys.stdout)
243
<entry file_id="123-123" kind="file" name="hello.c" />
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')
248
## TODO: Clear up handling of files in subdirectories; we probably
249
## do want to be able to just look them up by name but this
250
## probably means gradually walking down the path, looking up as we go.
252
## TODO: Make sure only canonical filenames are stored.
254
## TODO: Do something sensible about the possible collisions on
255
## case-losing filesystems. Perhaps we should just always forbid
258
## _tree should probably just be stored as
259
## InventoryEntry._children on each directory.
261
def __init__(self, root_id=ROOT_ID):
262
262
"""Create or read an inventory.
264
264
If a working directory is specified, the inventory is read
265
265
from there. If the file is specified, read from that. If not,
266
266
the inventory is created empty.
268
The inventory is created with a default root directory, with
270
# _tree is indexed by parent_id; at each level a map from name
271
# to ie. The None entry is the root.
272
self._tree = {None: {}}
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())
275
290
def __iter__(self):
281
296
return len(self._byid)
284
def iter_entries(self, parent_id=None):
299
def iter_entries(self, from_dir=None):
285
300
"""Return (path, entry) pairs, in order by name."""
286
kids = self._tree[parent_id].items()
304
elif isinstance(from_dir, basestring):
305
from_dir = self._byid[from_dir]
307
kids = from_dir.children.items()
288
309
for name, ie in kids:
290
311
if ie.kind == 'directory':
291
for cn, cie in self.iter_entries(parent_id=ie.file_id):
292
yield joinpath([name, cn]), cie
295
def directories(self, include_root=True):
296
"""Return (path, entry) pairs for all directories.
300
for path, entry in self.iter_entries():
301
if entry.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, '')
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
312
# TODO: return all paths and entries
315
353
def __contains__(self, file_id):
316
354
"""True if this entry contains a file with given id.
318
356
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
357
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
358
InventoryEntry('123', 'foo.c', kind='file', parent_id='TREE_ROOT')
329
368
"""Return the entry for given file_id.
331
370
>>> inv = Inventory()
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
371
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
372
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
333
373
>>> inv['123123'].name
336
return self._byid[file_id]
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
388
def get_child(self, parent_id, filename):
389
return self[parent_id].children.get(filename)
339
392
def add(self, entry):
340
393
"""Add entry to inventory.
342
395
To add a file to a branch ready to be committed, use Branch.add,
344
if entry.file_id in self:
345
bailout("inventory already contains entry with id {%s}" % entry.file_id)
347
if entry.parent_id != None:
348
if entry.parent_id not in self:
349
bailout("parent_id %s of new entry not found in inventory"
352
if self._tree[entry.parent_id].has_key(entry.name):
353
bailout("%s is already versioned"
354
% appendpath(self.id2path(entry.parent_id), entry.name))
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))
356
415
self._byid[entry.file_id] = entry
357
self._tree[entry.parent_id][entry.name] = entry
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
416
parent.children[entry.name] = entry
420
def add_path(self, relpath, kind, file_id=None):
421
"""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
428
parts = bzrlib.osutils.splitpath(relpath)
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)
440
ie = InventoryEntry(file_id, parts[-1],
441
kind=kind, parent_id=parent_id)
363
445
def __delitem__(self, file_id):
364
446
"""Remove entry by id.
366
448
>>> inv = Inventory()
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
449
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
450
InventoryEntry('123', 'foo.c', kind='file', parent_id='TREE_ROOT')
370
453
>>> del inv['123']
374
457
ie = self[file_id]
376
assert self._tree[ie.parent_id][ie.name] == ie
459
assert self[ie.parent_id].children[ie.name] == ie
378
461
# TODO: Test deleting all children; maybe hoist to a separate
379
462
# deltree method?
380
463
if ie.kind == 'directory':
381
for cie in self._tree[file_id].values():
464
for cie in ie.children.values():
382
465
del self[cie.file_id]
383
del self._tree[file_id]
385
468
del self._byid[file_id]
386
del self._tree[ie.parent_id][ie.name]
390
return Set(self._byid)
393
def to_element(self):
394
"""Convert to XML Element"""
395
e = Element('inventory')
397
for path, ie in self.iter_entries():
398
e.append(ie.to_element())
402
def from_element(cls, elt):
403
"""Construct from XML Element
405
>>> inv = Inventory()
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
407
>>> elt = inv.to_element()
408
>>> inv2 = Inventory.from_element(elt)
412
assert elt.tag == 'inventory'
415
o.add(InventoryEntry.from_element(e))
418
from_element = classmethod(from_element)
421
def __cmp__(self, other):
469
del self[ie.parent_id].children[ie.name]
472
def __eq__(self, other):
422
473
"""Compare two sets by comparing their contents.
424
475
>>> i1 = Inventory()
425
476
>>> i2 = Inventory()
428
>>> i1.add(InventoryEntry('123', 'foo'))
479
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
480
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
431
>>> i2.add(InventoryEntry('123', 'foo'))
483
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
484
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
438
488
if not isinstance(other, Inventory):
439
489
return NotImplemented
441
if self.id_set() ^ other.id_set():
444
for file_id in self._byid:
445
c = cmp(self[file_id], other[file_id])
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
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.
454
515
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)
457
521
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)
494
568
def has_id(self, file_id):
495
assert isinstance(file_id, str)
496
569
return self._byid.has_key(file_id)
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)
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))