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.
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
37
from bzrlib.errors import BzrError, BzrCheckError
39
from bzrlib.osutils import quotefn, splitpath, joinpath, appendpath
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
40
37
from bzrlib.trace import mutter
41
from bzrlib.errors import NotVersionedError
44
class InventoryEntry(object):
39
class InventoryEntry(XMLMixin):
45
40
"""Description of a versioned file.
47
42
An InventoryEntry has the following fields, which are also
48
43
present in the XML inventory-entry element:
53
(within the parent directory)
59
file_id of the parent directory, or ROOT_ID
62
the revision_id in which the name or parent of this file was
66
sha-1 of the text of the file
69
size in bytes of the text of the file
72
the revision_id in which the text of this file was introduced
74
(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.
76
60
>>> i = Inventory()
79
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
80
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
81
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
82
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'))
83
64
>>> for j in i.iter_entries():
86
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
87
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
88
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
89
70
Traceback (most recent call last):
91
BzrError: inventory already contains entry with id {2323}
92
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
93
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
94
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
95
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'))
96
75
>>> i.path2id('src/wibble')
100
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
101
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
103
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
104
>>> for path, entry in i.iter_entries():
105
... print path.replace('\\\\', '/') # for win32 os.sep
106
... assert i.path2id(path)
82
>>> for j in i.iter_entries():
84
... assert i.path2id(j[0])
112
90
src/wibble/wibble.c
113
>>> i.id2path('2326').replace('\\\\', '/')
114
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.
117
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
118
'text_id', 'parent_id', 'children',
119
'text_version', 'name_version', ]
122
def __init__(self, file_id, name, kind, parent_id, text_id=None):
98
def __init__(self, file_id, name, kind='file', text_id=None,
123
100
"""Create an InventoryEntry
125
102
The filename must be a single component, relative to the
126
103
parent directory; it cannot be a whole path or relative name.
128
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
105
>>> e = InventoryEntry('123', 'hello.c')
133
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
134
111
Traceback (most recent call last):
135
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
137
assert isinstance(name, basestring), name
138
if '/' in name or '\\' in name:
139
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
141
self.text_version = None
142
self.name_version = None
143
self.text_sha1 = None
144
self.text_size = None
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
145
119
self.file_id = file_id
121
assert kind in ['file', 'directory']
148
123
self.text_id = text_id
149
124
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
150
127
if kind == 'directory':
151
128
self.children = {}
155
raise BzrError("unhandled entry kind %r" % kind)
159
def sorted_children(self):
160
l = self.children.items()
166
132
other = InventoryEntry(self.file_id, self.name, self.kind,
168
other.text_id = self.text_id
133
self.text_id, self.parent_id)
169
134
other.text_sha1 = self.text_sha1
170
135
other.text_size = self.text_size
171
other.text_version = self.text_version
172
other.name_version = self.name_version
173
# note that children are *not* copied; they're pulled across when
187
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):
188
189
if not isinstance(other, InventoryEntry):
189
190
return NotImplemented
191
return (self.file_id == other.file_id) \
192
and (self.name == other.name) \
193
and (self.text_sha1 == other.text_sha1) \
194
and (self.text_size == other.text_size) \
195
and (self.text_id == other.text_id) \
196
and (self.parent_id == other.parent_id) \
197
and (self.kind == other.kind) \
198
and (self.text_version == other.text_version) \
199
and (self.name_version == other.name_version)
202
def __ne__(self, other):
203
return not (self == other)
206
raise ValueError('not hashable')
210
class RootEntry(InventoryEntry):
211
def __init__(self, file_id):
212
self.file_id = file_id
214
self.kind = 'root_directory'
215
self.parent_id = None
218
def __eq__(self, other):
219
if not isinstance(other, RootEntry):
220
return NotImplemented
222
return (self.file_id == other.file_id) \
223
and (self.children == other.children)
227
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):
228
203
"""Inventory of versioned files in a tree.
230
This describes which file_id is present at each point in the tree,
231
and possibly the SHA-1 or other information about the file.
232
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.
234
212
The inventory represents a typical unix file tree, with
235
213
directories containing files and subdirectories. We never store
259
239
>>> [x[0] for x in inv.iter_entries()]
261
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
262
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
263
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" />
265
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
266
261
"""Create or read an inventory.
268
263
If a working directory is specified, the inventory is read
269
264
from there. If the file is specified, read from that. If not,
270
265
the inventory is created empty.
272
The inventory is created with a default root directory, with
275
# We are letting Branch.initialize() create a unique inventory
276
# root id. Rather than generating a random one here.
278
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
279
self.root = RootEntry(root_id)
280
self._byid = {self.root.file_id: self.root}
284
other = Inventory(self.root.file_id)
285
# copy recursively so we know directories will be added before
286
# their children. There are more efficient ways than this...
287
for path, entry in self.iter_entries():
288
if entry == self.root:
290
other.add(entry.copy())
267
self._root = InventoryEntry(None, '', kind='directory')
268
self._byid = {None: self._root}
294
271
def __iter__(self):
300
277
return len(self._byid)
303
def iter_entries(self, from_dir=None):
280
def iter_entries(self, parent_id=None):
304
281
"""Return (path, entry) pairs, in order by name."""
308
elif isinstance(from_dir, basestring):
309
from_dir = self._byid[from_dir]
311
kids = from_dir.children.items()
282
kids = self[parent_id].children.items()
313
284
for name, ie in kids:
315
286
if ie.kind == 'directory':
316
for cn, cie in self.iter_entries(from_dir=ie.file_id):
317
yield os.path.join(name, cn), cie
321
"""Return list of (path, ie) for all entries except the root.
323
This may be faster than iter_entries.
326
def descend(dir_ie, dir_path):
327
kids = dir_ie.children.items()
329
for name, ie in kids:
330
child_path = os.path.join(dir_path, name)
331
accum.append((child_path, ie))
332
if ie.kind == 'directory':
333
descend(ie, child_path)
335
descend(self.root, '')
287
for cn, cie in self.iter_entries(parent_id=ie.file_id):
288
yield joinpath([name, cn]), cie
339
291
def directories(self):
340
"""Return (path, entry) pairs for all directories, including the root.
292
"""Return (path, entry) pairs for all directories.
343
def descend(parent_ie, parent_path):
344
accum.append((parent_path, parent_ie))
346
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
349
for name, child_ie in kids:
350
child_path = os.path.join(parent_path, name)
351
descend(child_ie, child_path)
352
descend(self.root, '')
295
for path, entry in self.iter_entries():
296
if entry.kind == 'directory':
372
315
"""Return the entry for given file_id.
374
317
>>> inv = Inventory()
375
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
376
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
318
>>> inv.add(InventoryEntry('123123', 'hello.c'))
377
319
>>> inv['123123'].name
381
return self._byid[file_id]
384
raise BzrError("can't look up file_id None")
386
raise BzrError("file_id {%s} not in inventory" % file_id)
389
def get_file_kind(self, file_id):
390
return self._byid[file_id].kind
322
return self._byid[file_id]
392
325
def get_child(self, parent_id, filename):
393
return self[parent_id].children.get(filename)
326
if parent_id == None:
327
return self._root.children.get(filename)
329
return self[parent_id].children.get(filename)
396
332
def add(self, entry):
397
333
"""Add entry to inventory.
399
335
To add a file to a branch ready to be committed, use Branch.add,
402
Returns the new entry object.
404
337
if entry.file_id in self._byid:
405
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
407
if entry.parent_id == ROOT_ID or entry.parent_id is None:
408
entry.parent_id = self.root.file_id
411
parent = self._byid[entry.parent_id]
413
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
338
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)
415
344
if parent.children.has_key(entry.name):
416
raise BzrError("%s is already versioned" %
345
bailout("%s is already versioned" %
417
346
appendpath(self.id2path(parent.file_id), entry.name))
419
348
self._byid[entry.file_id] = entry
420
349
parent.children[entry.name] = entry
424
352
def add_path(self, relpath, kind, file_id=None):
425
353
"""Add entry from a path.
427
The immediate parent must already be versioned.
429
Returns the new entry object."""
430
from bzrlib.branch import gen_file_id
355
The immediate parent must already be versioned"""
432
356
parts = bzrlib.osutils.splitpath(relpath)
433
357
if len(parts) == 0:
434
raise BzrError("cannot re-add root of inventory")
437
file_id = gen_file_id(relpath)
439
parent_path = parts[:-1]
440
parent_id = self.path2id(parent_path)
441
if parent_id == None:
442
raise NotVersionedError(parent_path)
358
bailout("cannot re-add root of inventory")
361
file_id = bzrlib.branch.gen_file_id(relpath)
363
parent_id = self.path2id(parts[:-1])
444
364
ie = InventoryEntry(file_id, parts[-1],
445
365
kind=kind, parent_id=parent_id)
446
366
return self.add(ie)
473
392
del self[ie.parent_id].children[ie.name]
476
def __eq__(self, other):
396
return Set(self._byid)
399
def to_element(self):
400
"""Convert to XML Element"""
401
e = Element('inventory')
403
for path, ie in self.iter_entries():
404
e.append(ie.to_element())
408
def from_element(cls, elt):
409
"""Construct from XML Element
411
>>> inv = Inventory()
412
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
413
>>> elt = inv.to_element()
414
>>> inv2 = Inventory.from_element(elt)
418
assert elt.tag == 'inventory'
421
o.add(InventoryEntry.from_element(e))
424
from_element = classmethod(from_element)
427
def __cmp__(self, other):
477
428
"""Compare two sets by comparing their contents.
479
430
>>> i1 = Inventory()
480
431
>>> i2 = Inventory()
483
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
484
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
434
>>> i1.add(InventoryEntry('123', 'foo'))
487
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
488
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
437
>>> i2.add(InventoryEntry('123', 'foo'))
492
444
if not isinstance(other, Inventory):
493
445
return NotImplemented
495
if len(self._byid) != len(other._byid):
496
# shortcut: obviously not the same
499
return self._byid == other._byid
502
def __ne__(self, other):
503
return not self.__eq__(other)
507
raise ValueError('not hashable')
510
def get_idpath(self, file_id):
511
"""Return a list of file_ids for the path to an entry.
513
The list contains one element for each directory followed by
514
the id of the file itself. So the length of the returned list
515
is equal to the depth of the file in the tree, counting the
516
root directory as depth 1.
447
if self.id_set() ^ other.id_set():
450
for file_id in self._byid:
451
c = cmp(self[file_id], other[file_id])
457
def id2path(self, file_id):
458
"""Return as a list the path to file_id."""
519
460
while file_id != None:
521
ie = self._byid[file_id]
523
raise BzrError("file_id {%s} not found in inventory" % file_id)
524
p.insert(0, ie.file_id)
461
ie = self._byid[file_id]
525
463
file_id = ie.parent_id
529
def id2path(self, file_id):
530
"""Return as a list the path to file_id."""
532
# get all names, skipping root
533
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
534
return os.sep.join(p)
added
removed
bzrlib/inventory.py
