14
15
# along with this program; if not, write to the Free Software
15
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
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
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
26
from bzrlib.errors import BzrError, BzrCheckError
28
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
29
37
from bzrlib.trace import mutter
30
from bzrlib.errors import NotVersionedError
33
class InventoryEntry(object):
39
class InventoryEntry(XMLMixin):
34
40
"""Description of a versioned file.
36
42
An InventoryEntry has the following fields, which are also
54
60
>>> i = Inventory()
57
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
58
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
59
64
>>> for j in i.iter_entries():
62
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
63
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
64
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
65
70
Traceback (most recent call last):
67
BzrError: inventory already contains entry with id {2323}
68
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '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'))
70
75
>>> i.path2id('src/wibble')
74
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
76
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
77
82
>>> for j in i.iter_entries():
86
91
>>> i.id2path('2326')
87
92
'src/wibble/wibble.c'
89
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?
90
95
But those depend on its position within a particular inventory, and
91
96
it would be nice not to need to hold the backpointer here.
94
# TODO: split InventoryEntry into subclasses for files,
95
# directories, etc etc.
97
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
98
'text_id', 'parent_id', 'children', ]
100
def __init__(self, file_id, name, kind, parent_id, text_id=None):
98
def __init__(self, file_id, name, kind='file', text_id=None,
101
100
"""Create an InventoryEntry
103
102
The filename must be a single component, relative to the
104
103
parent directory; it cannot be a whole path or relative name.
106
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
105
>>> e = InventoryEntry('123', 'hello.c')
111
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
112
111
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
115
if '/' in name or '\\' in name:
116
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
118
self.text_sha1 = None
119
self.text_size = None
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
121
119
self.file_id = file_id
121
assert kind in ['file', 'directory']
124
123
self.text_id = text_id
125
124
self.parent_id = parent_id
126
if kind == 'directory':
131
raise BzrError("unhandled entry kind %r" % kind)
135
def sorted_children(self):
136
l = self.children.items()
125
self.text_sha1 = None
126
self.text_size = None
142
130
other = InventoryEntry(self.file_id, self.name, self.kind,
143
self.parent_id, text_id=self.text_id)
131
self.text_id, self.parent_id)
144
132
other.text_sha1 = self.text_sha1
145
133
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
160
146
def to_element(self):
161
147
"""Convert to XML element"""
162
from bzrlib.xml import Element
164
148
e = Element('entry')
166
150
e.set('name', self.name)
167
151
e.set('file_id', self.file_id)
168
152
e.set('kind', self.kind)
170
if self.text_size != None:
154
if self.text_size is not None:
171
155
e.set('text_size', '%d' % self.text_size)
173
for f in ['text_id', 'text_sha1']:
157
for f in ['text_id', 'text_sha1', 'parent_id']:
174
158
v = getattr(self, f)
178
# to be conservative, we don't externalize the root pointers
179
# for now, leaving them as null in the xml form. in a future
180
# version it will be implied by nested elements.
181
if self.parent_id != ROOT_ID:
182
assert isinstance(self.parent_id, basestring)
183
e.set('parent_id', self.parent_id)
190
167
def from_element(cls, elt):
191
168
assert elt.tag == 'entry'
193
## original format inventories don't have a parent_id for
194
## nodes in the root directory, but it's cleaner to use one
196
parent_id = elt.get('parent_id')
197
if parent_id == None:
200
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
169
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
201
170
self.text_id = elt.get('text_id')
202
171
self.text_sha1 = elt.get('text_sha1')
172
self.parent_id = elt.get('parent_id')
204
174
## mutter("read inventoryentry: %r" % (elt.attrib))
212
182
from_element = classmethod(from_element)
214
def __eq__(self, other):
184
def __cmp__(self, other):
215
187
if not isinstance(other, InventoryEntry):
216
188
return NotImplemented
218
return (self.file_id == other.file_id) \
219
and (self.name == other.name) \
220
and (self.text_sha1 == other.text_sha1) \
221
and (self.text_size == other.text_size) \
222
and (self.text_id == other.text_id) \
223
and (self.parent_id == other.parent_id) \
224
and (self.kind == other.kind)
227
def __ne__(self, other):
228
return not (self == other)
231
raise ValueError('not hashable')
235
class RootEntry(InventoryEntry):
236
def __init__(self, file_id):
237
self.file_id = file_id
239
self.kind = 'root_directory'
240
self.parent_id = None
243
def __eq__(self, other):
244
if not isinstance(other, RootEntry):
245
return NotImplemented
247
return (self.file_id == other.file_id) \
248
and (self.children == other.children)
252
class Inventory(object):
190
return cmp(self.file_id, other.file_id) \
191
or cmp(self.name, other.name) \
192
or cmp(self.text_sha1, other.text_sha1) \
193
or cmp(self.text_size, other.text_size) \
194
or cmp(self.text_id, other.text_id) \
195
or cmp(self.parent_id, other.parent_id) \
196
or cmp(self.kind, other.kind)
200
class Inventory(XMLMixin):
253
201
"""Inventory of versioned files in a tree.
255
This describes which file_id is present at each point in the tree,
256
and possibly the SHA-1 or other information about the file.
257
Entries can be looked up either by path or by file_id.
203
An Inventory acts like a set of InventoryEntry items. You can
204
also look files up by their file_id or name.
206
May be read from and written to a metadata file in a tree. To
207
manipulate the inventory (for example to add a file), it is read
208
in, modified, and then written back out.
259
210
The inventory represents a typical unix file tree, with
260
211
directories containing files and subdirectories. We never store
264
215
returned quickly.
266
217
InventoryEntry objects must not be modified after they are
267
inserted, other than through the Inventory API.
269
220
>>> inv = Inventory()
270
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
221
>>> inv.write_xml(sys.stdout)
224
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
271
225
>>> inv['123-123'].name
227
>>> for file_id in inv: print file_id
274
231
May be treated as an iterator or set to look up file ids:
283
240
>>> [x[0] for x in inv.iter_entries()]
285
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
286
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
243
>>> inv.write_xml(sys.stdout)
245
<entry file_id="123-123" kind="file" name="hello.c" />
288
def __init__(self, root_id=ROOT_ID):
250
## TODO: Clear up handling of files in subdirectories; we probably
251
## do want to be able to just look them up by name but this
252
## probably means gradually walking down the path, looking up as we go.
254
## TODO: Make sure only canonical filenames are stored.
256
## TODO: Do something sensible about the possible collisions on
257
## case-losing filesystems. Perhaps we should just always forbid
260
## _tree should probably just be stored as
261
## InventoryEntry._children on each directory.
289
264
"""Create or read an inventory.
291
266
If a working directory is specified, the inventory is read
292
267
from there. If the file is specified, read from that. If not,
293
268
the inventory is created empty.
295
The inventory is created with a default root directory, with
298
# We are letting Branch(init=True) create a unique inventory
299
# root id. Rather than generating a random one here.
301
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
302
self.root = RootEntry(root_id)
303
self._byid = {self.root.file_id: self.root}
272
# _tree is indexed by parent_id; at each level a map from name
273
# to ie. The None entry is the root.
274
self._tree = {None: {}}
306
277
def __iter__(self):
312
283
return len(self._byid)
315
def iter_entries(self, from_dir=None):
286
def iter_entries(self, parent_id=None):
316
287
"""Return (path, entry) pairs, in order by name."""
320
elif isinstance(from_dir, basestring):
321
from_dir = self._byid[from_dir]
323
kids = from_dir.children.items()
288
kids = self._tree[parent_id].items()
325
290
for name, ie in kids:
327
292
if ie.kind == 'directory':
328
for cn, cie in self.iter_entries(from_dir=ie.file_id):
329
yield os.path.join(name, cn), cie
333
"""Return list of (path, ie) for all entries except the root.
335
This may be faster than iter_entries.
338
def descend(dir_ie, dir_path):
339
kids = dir_ie.children.items()
341
for name, ie in kids:
342
child_path = os.path.join(dir_path, name)
343
accum.append((child_path, ie))
344
if ie.kind == 'directory':
345
descend(ie, child_path)
347
descend(self.root, '')
351
def directories(self):
352
"""Return (path, entry) pairs for all directories, including the root.
355
def descend(parent_ie, parent_path):
356
accum.append((parent_path, parent_ie))
358
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
361
for name, child_ie in kids:
362
child_path = os.path.join(parent_path, name)
363
descend(child_ie, child_path)
364
descend(self.root, '')
293
for cn, cie in self.iter_entries(parent_id=ie.file_id):
294
yield joinpath([name, cn]), cie
297
def directories(self, include_root=True):
298
"""Return (path, entry) pairs for all directories.
302
for path, entry in self.iter_entries():
303
if entry.kind == 'directory':
308
def children(self, parent_id):
309
"""Return entries that are direct children of parent_id."""
310
return self._tree[parent_id]
314
# TODO: return all paths and entries
369
317
def __contains__(self, file_id):
370
318
"""True if this entry contains a file with given id.
372
320
>>> inv = Inventory()
373
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
321
>>> inv.add(InventoryEntry('123', 'foo.c'))
383
331
"""Return the entry for given file_id.
385
333
>>> inv = Inventory()
386
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
334
>>> inv.add(InventoryEntry('123123', 'hello.c'))
387
335
>>> inv['123123'].name
391
return self._byid[file_id]
394
raise BzrError("can't look up file_id None")
396
raise BzrError("file_id {%s} not in inventory" % file_id)
399
def get_file_kind(self, file_id):
400
return self._byid[file_id].kind
402
def get_child(self, parent_id, filename):
403
return self[parent_id].children.get(filename)
338
return self._byid[file_id]
406
341
def add(self, entry):
409
344
To add a file to a branch ready to be committed, use Branch.add,
410
345
which calls this."""
411
if entry.file_id in self._byid:
412
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
414
if entry.parent_id == ROOT_ID or entry.parent_id is None:
415
entry.parent_id = self.root.file_id
418
parent = self._byid[entry.parent_id]
420
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
422
if parent.children.has_key(entry.name):
423
raise BzrError("%s is already versioned" %
424
appendpath(self.id2path(parent.file_id), entry.name))
346
if entry.file_id in self:
347
bailout("inventory already contains entry with id {%s}" % entry.file_id)
349
if entry.parent_id != None:
350
if entry.parent_id not in self:
351
bailout("parent_id %s of new entry not found in inventory"
354
if self._tree[entry.parent_id].has_key(entry.name):
355
bailout("%s is already versioned"
356
% appendpath(self.id2path(entry.parent_id), entry.name))
426
358
self._byid[entry.file_id] = entry
427
parent.children[entry.name] = entry
359
self._tree[entry.parent_id][entry.name] = entry
361
if entry.kind == 'directory':
362
self._tree[entry.file_id] = {}
430
365
def add_path(self, relpath, kind, file_id=None):
431
366
"""Add entry from a path.
433
368
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
436
369
parts = bzrlib.osutils.splitpath(relpath)
437
370
if len(parts) == 0:
438
raise BzrError("cannot re-add root of inventory")
441
file_id = gen_file_id(relpath)
443
parent_path = parts[:-1]
444
parent_id = self.path2id(parent_path)
445
if parent_id == None:
446
raise NotVersionedError(parent_path)
371
bailout("cannot re-add root of inventory")
374
file_id = bzrlib.branch.gen_file_id(relpath)
376
parent_id = self.path2id(parts[:-1])
448
377
ie = InventoryEntry(file_id, parts[-1],
449
378
kind=kind, parent_id=parent_id)
450
379
return self.add(ie)
464
393
ie = self[file_id]
466
assert self[ie.parent_id].children[ie.name] == ie
395
assert self._tree[ie.parent_id][ie.name] == ie
468
397
# TODO: Test deleting all children; maybe hoist to a separate
469
398
# deltree method?
470
399
if ie.kind == 'directory':
471
for cie in ie.children.values():
400
for cie in self._tree[file_id].values():
472
401
del self[cie.file_id]
402
del self._tree[file_id]
475
404
del self._byid[file_id]
476
del self[ie.parent_id].children[ie.name]
405
del self._tree[ie.parent_id][ie.name]
409
return Set(self._byid)
479
412
def to_element(self):
480
413
"""Convert to XML Element"""
481
from bzrlib.xml import Element
483
414
e = Element('inventory')
485
if self.root.file_id not in (None, ROOT_ID):
486
e.set('file_id', self.root.file_id)
487
416
for path, ie in self.iter_entries():
488
417
e.append(ie.to_element())
492
421
def from_element(cls, elt):
493
422
"""Construct from XML Element
495
424
>>> inv = Inventory()
496
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
425
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
497
426
>>> elt = inv.to_element()
498
427
>>> inv2 = Inventory.from_element(elt)
502
# XXXX: doctest doesn't run this properly under python2.3
503
431
assert elt.tag == 'inventory'
504
root_id = elt.get('file_id') or ROOT_ID
507
ie = InventoryEntry.from_element(e)
508
if ie.parent_id == ROOT_ID:
509
ie.parent_id = root_id
434
o.add(InventoryEntry.from_element(e))
513
437
from_element = classmethod(from_element)
516
def __eq__(self, other):
440
def __cmp__(self, other):
517
441
"""Compare two sets by comparing their contents.
519
443
>>> i1 = Inventory()
520
444
>>> i2 = Inventory()
523
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
447
>>> i1.add(InventoryEntry('123', 'foo'))
526
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
450
>>> i2.add(InventoryEntry('123', 'foo'))
530
457
if not isinstance(other, Inventory):
531
458
return NotImplemented
533
if len(self._byid) != len(other._byid):
534
# shortcut: obviously not the same
537
return self._byid == other._byid
540
def __ne__(self, other):
541
return not (self == other)
545
raise ValueError('not hashable')
549
def get_idpath(self, file_id):
550
"""Return a list of file_ids for the path to an entry.
552
The list contains one element for each directory followed by
553
the id of the file itself. So the length of the returned list
554
is equal to the depth of the file in the tree, counting the
555
root directory as depth 1.
460
if self.id_set() ^ other.id_set():
463
for file_id in self._byid:
464
c = cmp(self[file_id], other[file_id])
470
def id2path(self, file_id):
471
"""Return as a list the path to file_id."""
558
473
while file_id != None:
560
ie = self._byid[file_id]
562
raise BzrError("file_id {%s} not found in inventory" % file_id)
563
p.insert(0, ie.file_id)
564
476
file_id = ie.parent_id
568
def id2path(self, file_id):
569
"""Return as a list the path to file_id."""
571
# get all names, skipping root
572
p = [self._byid[fid].name for fid in self.get_idpath(file_id)[1:]]
573
return os.sep.join(p)
583
487
This returns the entry of the last component in the path,
584
488
which may be either a file or a directory.
586
Returns None iff the path is not found.
588
490
if isinstance(name, types.StringTypes):
589
491
name = splitpath(name)
591
mutter("lookup path %r" % name)
596
cie = parent.children[f]
496
cie = self._tree[parent_id][f]
597
497
assert cie.name == f
598
assert cie.parent_id == parent.file_id
498
parent_id = cie.file_id
601
500
# or raise an error?
604
return parent.file_id
506
def get_child(self, parent_id, child_name):
507
return self._tree[parent_id].get(child_name)
607
510
def has_filename(self, names):
611
514
def has_id(self, file_id):
515
assert isinstance(file_id, str)
612
516
return self._byid.has_key(file_id)
615
def rename(self, file_id, new_parent_id, new_name):
616
"""Move a file within the inventory.
618
This can change either the name, or the parent, or both.
620
This does not move the working file."""
621
if not is_valid_name(new_name):
622
raise BzrError("not an acceptable filename: %r" % new_name)
624
new_parent = self._byid[new_parent_id]
625
if new_name in new_parent.children:
626
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
628
new_parent_idpath = self.get_idpath(new_parent_id)
629
if file_id in new_parent_idpath:
630
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
631
% (self.id2path(file_id), self.id2path(new_parent_id)))
633
file_ie = self._byid[file_id]
634
old_parent = self._byid[file_ie.parent_id]
636
# TODO: Don't leave things messed up if this fails
638
del old_parent.children[file_ie.name]
639
new_parent.children[new_name] = file_ie
641
file_ie.name = new_name
642
file_ie.parent_id = new_parent_id
649
def is_valid_name(name):
652
_NAME_RE = re.compile(r'^[^/\\]+$')
654
return bool(_NAME_RE.match(name))
522
if __name__ == '__main__':
523
import doctest, inventory
524
doctest.testmod(inventory)