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
# 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
26
from bzrlib.errors import BzrError, BzrCheckError
28
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
29
from bzrlib.trace import mutter
31
class InventoryEntry(object):
38
32
"""Description of a versioned file.
40
34
An InventoryEntry has the following fields, which are also
58
52
>>> i = Inventory()
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
55
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
56
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
62
57
>>> for j in i.iter_entries():
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
60
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
66
61
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
62
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
68
63
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'))
65
BzrError: inventory already contains entry with id {2323}
66
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
67
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
73
68
>>> i.path2id('src/wibble')
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
72
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
79
74
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
80
75
>>> for j in i.iter_entries():
89
84
>>> i.id2path('2326')
90
85
'src/wibble/wibble.c'
92
:todo: Maybe also keep the full path of the entry, and the children?
87
TODO: Maybe also keep the full path of the entry, and the children?
93
88
But those depend on its position within a particular inventory, and
94
89
it would be nice not to need to hold the backpointer here.
96
def __init__(self, file_id, name, kind='file', text_id=None,
92
# TODO: split InventoryEntry into subclasses for files,
93
# directories, etc etc.
98
def __init__(self, file_id, name, kind, parent_id, text_id=None):
98
99
"""Create an InventoryEntry
100
101
The filename must be a single component, relative to the
101
102
parent directory; it cannot be a whole path or relative name.
103
>>> e = InventoryEntry('123', 'hello.c')
104
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
108
>>> e = InventoryEntry('123', 'src/hello.c')
109
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
109
110
Traceback (most recent call last):
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
111
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
113
if '/' in name or '\\' in name:
114
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
117
116
self.file_id = file_id
119
assert kind in ['file', 'directory']
121
119
self.text_id = text_id
122
120
self.parent_id = parent_id
123
self.text_sha1 = None
124
self.text_size = None
121
if kind == 'directory':
126
raise BzrError("unhandled entry kind %r" % kind)
130
def sorted_children(self):
131
l = self.children.items()
128
137
other = InventoryEntry(self.file_id, self.name, self.kind,
129
self.text_id, self.parent_id)
138
self.parent_id, text_id=self.text_id)
130
139
other.text_sha1 = self.text_sha1
131
140
other.text_size = self.text_size
141
# note that children are *not* copied; they're pulled across when
144
155
def to_element(self):
145
156
"""Convert to XML element"""
157
from bzrlib.xml import Element
146
159
e = Element('entry')
148
161
e.set('name', self.name)
149
162
e.set('file_id', self.file_id)
150
163
e.set('kind', self.kind)
152
if self.text_size is not None:
165
if self.text_size != None:
153
166
e.set('text_size', '%d' % self.text_size)
155
for f in ['text_id', 'text_sha1', 'parent_id']:
168
for f in ['text_id', 'text_sha1']:
156
169
v = getattr(self, f)
173
# to be conservative, we don't externalize the root pointers
174
# for now, leaving them as null in the xml form. in a future
175
# version it will be implied by nested elements.
176
if self.parent_id != ROOT_ID:
177
assert isinstance(self.parent_id, basestring)
178
e.set('parent_id', self.parent_id)
165
185
def from_element(cls, elt):
166
186
assert elt.tag == 'entry'
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
188
## original format inventories don't have a parent_id for
189
## nodes in the root directory, but it's cleaner to use one
191
parent_id = elt.get('parent_id')
192
if parent_id == None:
195
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
168
196
self.text_id = elt.get('text_id')
169
197
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
172
199
## mutter("read inventoryentry: %r" % (elt.attrib))
180
207
from_element = classmethod(from_element)
182
def __cmp__(self, other):
209
def __eq__(self, other):
185
210
if not isinstance(other, InventoryEntry):
186
211
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):
213
return (self.file_id == other.file_id) \
214
and (self.name == other.name) \
215
and (self.text_sha1 == other.text_sha1) \
216
and (self.text_size == other.text_size) \
217
and (self.text_id == other.text_id) \
218
and (self.parent_id == other.parent_id) \
219
and (self.kind == other.kind)
222
def __ne__(self, other):
223
return not (self == other)
226
raise ValueError('not hashable')
230
class RootEntry(InventoryEntry):
231
def __init__(self, file_id):
232
self.file_id = file_id
234
self.kind = 'root_directory'
235
self.parent_id = None
238
def __eq__(self, other):
239
if not isinstance(other, RootEntry):
240
return NotImplemented
242
return (self.file_id == other.file_id) \
243
and (self.children == other.children)
247
class Inventory(object):
199
248
"""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.
250
This describes which file_id is present at each point in the tree,
251
and possibly the SHA-1 or other information about the file.
252
Entries can be looked up either by path or by file_id.
208
254
The inventory represents a typical unix file tree, with
209
255
directories containing files and subdirectories. We never store
213
259
returned quickly.
215
261
InventoryEntry objects must not be modified after they are
262
inserted, other than through the Inventory API.
218
264
>>> inv = Inventory()
219
>>> inv.write_xml(sys.stdout)
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
265
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
223
266
>>> inv['123-123'].name
225
>>> for file_id in inv: print file_id
229
269
May be treated as an iterator or set to look up file ids:
238
278
>>> [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" />
280
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
281
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
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.
283
def __init__(self, root_id=ROOT_ID):
262
284
"""Create or read an inventory.
264
286
If a working directory is specified, the inventory is read
265
287
from there. If the file is specified, read from that. If not,
266
288
the inventory is created empty.
290
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: {}}
293
# We are letting Branch(init=True) create a unique inventory
294
# root id. Rather than generating a random one here.
296
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
297
self.root = RootEntry(root_id)
298
self._byid = {self.root.file_id: self.root}
275
301
def __iter__(self):
281
307
return len(self._byid)
284
def iter_entries(self, parent_id=None):
310
def iter_entries(self, from_dir=None):
285
311
"""Return (path, entry) pairs, in order by name."""
286
kids = self._tree[parent_id].items()
315
elif isinstance(from_dir, basestring):
316
from_dir = self._byid[from_dir]
318
kids = from_dir.children.items()
288
320
for name, ie in kids:
290
322
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':
323
for cn, cie in self.iter_entries(from_dir=ie.file_id):
324
yield os.path.join(name, cn), cie
328
"""Return list of (path, ie) for all entries except the root.
330
This may be faster than iter_entries.
333
def descend(dir_ie, dir_path):
334
kids = dir_ie.children.items()
336
for name, ie in kids:
337
child_path = os.path.join(dir_path, name)
338
accum.append((child_path, ie))
339
if ie.kind == 'directory':
340
descend(ie, child_path)
342
descend(self.root, '')
346
def directories(self):
347
"""Return (path, entry) pairs for all directories, including the root.
350
def descend(parent_ie, parent_path):
351
accum.append((parent_path, parent_ie))
353
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
356
for name, child_ie in kids:
357
child_path = os.path.join(parent_path, name)
358
descend(child_ie, child_path)
359
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
364
def __contains__(self, file_id):
316
365
"""True if this entry contains a file with given id.
318
367
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
368
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
329
378
"""Return the entry for given file_id.
331
380
>>> inv = Inventory()
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
381
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
333
382
>>> inv['123123'].name
336
return self._byid[file_id]
386
return self._byid[file_id]
389
raise BzrError("can't look up file_id None")
391
raise BzrError("file_id {%s} not in inventory" % file_id)
394
def get_file_kind(self, file_id):
395
return self._byid[file_id].kind
397
def get_child(self, parent_id, filename):
398
return self[parent_id].children.get(filename)
339
401
def add(self, entry):
342
404
To add a file to a branch ready to be committed, use Branch.add,
343
405
which calls this."""
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))
406
if entry.file_id in self._byid:
407
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
409
if entry.parent_id == ROOT_ID or entry.parent_id is None:
410
entry.parent_id = self.root.file_id
413
parent = self._byid[entry.parent_id]
415
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
417
if parent.children.has_key(entry.name):
418
raise BzrError("%s is already versioned" %
419
appendpath(self.id2path(parent.file_id), entry.name))
356
421
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] = {}
422
parent.children[entry.name] = entry
425
def add_path(self, relpath, kind, file_id=None):
426
"""Add entry from a path.
428
The immediate parent must already be versioned"""
429
from bzrlib.errors import NotVersionedError
431
parts = bzrlib.osutils.splitpath(relpath)
433
raise BzrError("cannot re-add root of inventory")
436
from bzrlib.branch import gen_file_id
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)
444
ie = InventoryEntry(file_id, parts[-1],
445
kind=kind, parent_id=parent_id)
363
449
def __delitem__(self, file_id):
364
450
"""Remove entry by id.
366
452
>>> inv = Inventory()
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
453
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
370
456
>>> del inv['123']
374
460
ie = self[file_id]
376
assert self._tree[ie.parent_id][ie.name] == ie
462
assert self[ie.parent_id].children[ie.name] == ie
378
464
# TODO: Test deleting all children; maybe hoist to a separate
379
465
# deltree method?
380
466
if ie.kind == 'directory':
381
for cie in self._tree[file_id].values():
467
for cie in ie.children.values():
382
468
del self[cie.file_id]
383
del self._tree[file_id]
385
471
del self._byid[file_id]
386
del self._tree[ie.parent_id][ie.name]
390
return Set(self._byid)
472
del self[ie.parent_id].children[ie.name]
393
475
def to_element(self):
394
476
"""Convert to XML Element"""
477
from bzrlib.xml import Element
395
479
e = Element('inventory')
481
if self.root.file_id not in (None, ROOT_ID):
482
e.set('file_id', self.root.file_id)
397
483
for path, ie in self.iter_entries():
398
484
e.append(ie.to_element())
402
488
def from_element(cls, elt):
403
489
"""Construct from XML Element
405
491
>>> inv = Inventory()
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
492
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
407
493
>>> elt = inv.to_element()
408
494
>>> inv2 = Inventory.from_element(elt)
498
# XXXX: doctest doesn't run this properly under python2.3
412
499
assert elt.tag == 'inventory'
500
root_id = elt.get('file_id') or ROOT_ID
415
o.add(InventoryEntry.from_element(e))
503
ie = InventoryEntry.from_element(e)
504
if ie.parent_id == ROOT_ID:
505
ie.parent_id = root_id
418
509
from_element = classmethod(from_element)
421
def __cmp__(self, other):
512
def __eq__(self, other):
422
513
"""Compare two sets by comparing their contents.
424
515
>>> i1 = Inventory()
425
516
>>> i2 = Inventory()
428
>>> i1.add(InventoryEntry('123', 'foo'))
519
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
431
>>> i2.add(InventoryEntry('123', 'foo'))
522
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
438
526
if not isinstance(other, Inventory):
439
527
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."""
529
if len(self._byid) != len(other._byid):
530
# shortcut: obviously not the same
533
return self._byid == other._byid
536
def __ne__(self, other):
537
return not (self == other)
541
raise ValueError('not hashable')
545
def get_idpath(self, file_id):
546
"""Return a list of file_ids for the path to an entry.
548
The list contains one element for each directory followed by
549
the id of the file itself. So the length of the returned list
550
is equal to the depth of the file in the tree, counting the
551
root directory as depth 1.
454
554
while file_id != None:
556
ie = self._byid[file_id]
558
raise BzrError("file_id {%s} not found in inventory" % file_id)
559
p.insert(0, ie.file_id)
457
560
file_id = ie.parent_id
564
def id2path(self, file_id):
565
"""Return as a list the path to file_id."""
567
# get all names, skipping root
568
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
569
return os.sep.join(p)
468
579
This returns the entry of the last component in the path,
469
580
which may be either a file or a directory.
582
Returns None iff the path is not found.
471
assert isinstance(name, types.StringTypes)
474
for f in splitpath(name):
584
if isinstance(name, types.StringTypes):
585
name = splitpath(name)
587
mutter("lookup path %r" % name)
476
cie = self._tree[parent_id][f]
592
cie = parent.children[f]
477
593
assert cie.name == f
478
parent_id = cie.file_id
594
assert cie.parent_id == parent.file_id
480
597
# or raise an error?
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
600
return parent.file_id
490
603
def has_filename(self, names):
494
607
def has_id(self, file_id):
495
assert isinstance(file_id, str)
496
608
return self._byid.has_key(file_id)
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)
611
def rename(self, file_id, new_parent_id, new_name):
612
"""Move a file within the inventory.
614
This can change either the name, or the parent, or both.
616
This does not move the working file."""
617
if not is_valid_name(new_name):
618
raise BzrError("not an acceptable filename: %r" % new_name)
620
new_parent = self._byid[new_parent_id]
621
if new_name in new_parent.children:
622
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
624
new_parent_idpath = self.get_idpath(new_parent_id)
625
if file_id in new_parent_idpath:
626
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
627
% (self.id2path(file_id), self.id2path(new_parent_id)))
629
file_ie = self._byid[file_id]
630
old_parent = self._byid[file_ie.parent_id]
632
# TODO: Don't leave things messed up if this fails
634
del old_parent.children[file_ie.name]
635
new_parent.children[new_name] = file_ie
637
file_ie.name = new_name
638
file_ie.parent_id = new_parent_id
643
_NAME_RE = re.compile(r'^[^/\\]+$')
645
def is_valid_name(name):
646
return bool(_NAME_RE.match(name))