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."""
20
# TODO: Maybe store inventory_id in the file? Not really needed.
22
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
23
__author__ = "Martin Pool <mbp@canonical.com>"
25
import sys, os.path, types
29
from cElementTree import Element, ElementTree, SubElement
31
from elementtree.ElementTree import Element, ElementTree, SubElement
33
from xml import XMLMixin
34
from errors import bailout
26
from bzrlib.errors import BzrError, BzrCheckError
28
37
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
29
38
from bzrlib.trace import mutter
31
class InventoryEntry(object):
40
class InventoryEntry(XMLMixin):
32
41
"""Description of a versioned file.
34
43
An InventoryEntry has the following fields, which are also
52
61
>>> i = Inventory()
55
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
56
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
63
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
64
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
57
65
>>> for j in i.iter_entries():
60
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
61
69
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
62
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
63
71
Traceback (most recent call last):
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
BzrError: ('inventory already contains entry with id {2323}', [])
74
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
75
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
68
76
>>> i.path2id('src/wibble')
72
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
80
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
74
82
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
75
83
>>> for j in i.iter_entries():
84
92
>>> i.id2path('2326')
85
93
'src/wibble/wibble.c'
87
TODO: Maybe also keep the full path of the entry, and the children?
95
:todo: Maybe also keep the full path of the entry, and the children?
88
96
But those depend on its position within a particular inventory, and
89
97
it would be nice not to need to hold the backpointer here.
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):
99
def __init__(self, file_id, name, kind='file', text_id=None,
99
101
"""Create an InventoryEntry
101
103
The filename must be a single component, relative to the
102
104
parent directory; it cannot be a whole path or relative name.
104
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
106
>>> e = InventoryEntry('123', 'hello.c')
109
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
>>> e = InventoryEntry('123', 'src/hello.c')
110
112
Traceback (most recent call last):
111
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
113
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
113
if '/' in name or '\\' in name:
114
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
116
if len(splitpath(name)) != 1:
117
bailout('InventoryEntry name is not a simple filename: %r'
116
120
self.file_id = file_id
122
assert kind in ['file', 'directory']
119
124
self.text_id = text_id
120
125
self.parent_id = parent_id
121
if kind == 'directory':
126
raise BzrError("unhandled entry kind %r" % kind)
130
def sorted_children(self):
131
l = self.children.items()
126
self.text_sha1 = None
127
self.text_size = None
137
131
other = InventoryEntry(self.file_id, self.name, self.kind,
138
self.parent_id, text_id=self.text_id)
132
self.text_id, self.parent_id)
139
133
other.text_sha1 = self.text_sha1
140
134
other.text_size = self.text_size
141
# note that children are *not* copied; they're pulled across when
155
147
def to_element(self):
156
148
"""Convert to XML element"""
157
from bzrlib.xml import Element
159
149
e = Element('entry')
161
151
e.set('name', self.name)
162
152
e.set('file_id', self.file_id)
163
153
e.set('kind', self.kind)
165
if self.text_size != None:
155
if self.text_size is not None:
166
156
e.set('text_size', '%d' % self.text_size)
168
for f in ['text_id', 'text_sha1']:
158
for f in ['text_id', 'text_sha1', 'parent_id']:
169
159
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)
185
168
def from_element(cls, elt):
186
169
assert elt.tag == 'entry'
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)
170
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
196
171
self.text_id = elt.get('text_id')
197
172
self.text_sha1 = elt.get('text_sha1')
173
self.parent_id = elt.get('parent_id')
199
175
## mutter("read inventoryentry: %r" % (elt.attrib))
207
183
from_element = classmethod(from_element)
209
def __eq__(self, other):
185
def __cmp__(self, other):
210
188
if not isinstance(other, InventoryEntry):
211
189
return NotImplemented
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):
191
return cmp(self.file_id, other.file_id) \
192
or cmp(self.name, other.name) \
193
or cmp(self.text_sha1, other.text_sha1) \
194
or cmp(self.text_size, other.text_size) \
195
or cmp(self.text_id, other.text_id) \
196
or cmp(self.parent_id, other.parent_id) \
197
or cmp(self.kind, other.kind)
201
class Inventory(XMLMixin):
248
202
"""Inventory of versioned files in a tree.
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.
204
An Inventory acts like a set of InventoryEntry items. You can
205
also look files up by their file_id or name.
207
May be read from and written to a metadata file in a tree. To
208
manipulate the inventory (for example to add a file), it is read
209
in, modified, and then written back out.
254
211
The inventory represents a typical unix file tree, with
255
212
directories containing files and subdirectories. We never store
259
216
returned quickly.
261
218
InventoryEntry objects must not be modified after they are
262
inserted, other than through the Inventory API.
264
221
>>> inv = Inventory()
265
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
222
>>> inv.write_xml(sys.stdout)
225
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
266
226
>>> inv['123-123'].name
228
>>> for file_id in inv: print file_id
269
232
May be treated as an iterator or set to look up file ids:
278
241
>>> [x[0] for x in inv.iter_entries()]
244
>>> inv.write_xml(sys.stdout)
246
<entry file_id="123-123" kind="file" name="hello.c" />
251
## TODO: Clear up handling of files in subdirectories; we probably
252
## do want to be able to just look them up by name but this
253
## probably means gradually walking down the path, looking up as we go.
255
## TODO: Make sure only canonical filenames are stored.
257
## TODO: Do something sensible about the possible collisions on
258
## case-losing filesystems. Perhaps we should just always forbid
261
## _tree should probably just be stored as
262
## InventoryEntry._children on each directory.
281
264
def __init__(self):
282
265
"""Create or read an inventory.
284
267
If a working directory is specified, the inventory is read
285
268
from there. If the file is specified, read from that. If not,
286
269
the inventory is created empty.
288
The inventory is created with a default root directory, with
291
self.root = RootEntry(ROOT_ID)
292
self._byid = {self.root.file_id: self.root}
273
# _tree is indexed by parent_id; at each level a map from name
274
# to ie. The None entry is the root.
275
self._tree = {None: {}}
295
278
def __iter__(self):
301
284
return len(self._byid)
304
def iter_entries(self, from_dir=None):
287
def iter_entries(self, parent_id=None):
305
288
"""Return (path, entry) pairs, in order by name."""
309
elif isinstance(from_dir, basestring):
310
from_dir = self._byid[from_dir]
312
kids = from_dir.children.items()
289
kids = self._tree[parent_id].items()
314
291
for name, ie in kids:
316
293
if ie.kind == 'directory':
317
for cn, cie in self.iter_entries(from_dir=ie.file_id):
318
yield os.path.join(name, cn), cie
322
"""Return list of (path, ie) for all entries except the root.
324
This may be faster than iter_entries.
327
def descend(dir_ie, dir_path):
328
kids = dir_ie.children.items()
330
for name, ie in kids:
331
child_path = os.path.join(dir_path, name)
332
accum.append((child_path, ie))
333
if ie.kind == 'directory':
334
descend(ie, child_path)
336
descend(self.root, '')
340
def directories(self):
341
"""Return (path, entry) pairs for all directories, including the root.
344
def descend(parent_ie, parent_path):
345
accum.append((parent_path, parent_ie))
347
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
350
for name, child_ie in kids:
351
child_path = os.path.join(parent_path, name)
352
descend(child_ie, child_path)
353
descend(self.root, '')
294
for cn, cie in self.iter_entries(parent_id=ie.file_id):
295
yield joinpath([name, cn]), cie
298
def directories(self, include_root=True):
299
"""Return (path, entry) pairs for all directories.
303
for path, entry in self.iter_entries():
304
if entry.kind == 'directory':
309
def children(self, parent_id):
310
"""Return entries that are direct children of parent_id."""
311
return self._tree[parent_id]
315
# TODO: return all paths and entries
358
318
def __contains__(self, file_id):
359
319
"""True if this entry contains a file with given id.
361
321
>>> inv = Inventory()
362
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
322
>>> inv.add(InventoryEntry('123', 'foo.c'))
372
332
"""Return the entry for given file_id.
374
334
>>> inv = Inventory()
375
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
335
>>> inv.add(InventoryEntry('123123', 'hello.c'))
376
336
>>> inv['123123'].name
380
return self._byid[file_id]
383
raise BzrError("can't look up file_id None")
385
raise BzrError("file_id {%s} not in inventory" % file_id)
388
def get_file_kind(self, file_id):
389
return self._byid[file_id].kind
391
def get_child(self, parent_id, filename):
392
return self[parent_id].children.get(filename)
339
return self._byid[file_id]
395
342
def add(self, entry):
398
345
To add a file to a branch ready to be committed, use Branch.add,
399
346
which calls this."""
400
if entry.file_id in self._byid:
401
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
404
parent = self._byid[entry.parent_id]
406
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
408
if parent.children.has_key(entry.name):
409
raise BzrError("%s is already versioned" %
410
appendpath(self.id2path(parent.file_id), entry.name))
347
if entry.file_id in self:
348
bailout("inventory already contains entry with id {%s}" % entry.file_id)
350
if entry.parent_id != None:
351
if entry.parent_id not in self:
352
bailout("parent_id %s of new entry not found in inventory"
355
if self._tree[entry.parent_id].has_key(entry.name):
356
bailout("%s is already versioned"
357
% appendpath(self.id2path(entry.parent_id), entry.name))
412
359
self._byid[entry.file_id] = entry
413
parent.children[entry.name] = entry
360
self._tree[entry.parent_id][entry.name] = entry
362
if entry.kind == 'directory':
363
self._tree[entry.file_id] = {}
416
366
def add_path(self, relpath, kind, file_id=None):
417
367
"""Add entry from a path.
419
369
The immediate parent must already be versioned"""
420
from bzrlib.errors import NotVersionedError
422
370
parts = bzrlib.osutils.splitpath(relpath)
423
371
if len(parts) == 0:
424
raise BzrError("cannot re-add root of inventory")
427
from bzrlib.branch import gen_file_id
428
file_id = gen_file_id(relpath)
430
parent_path = parts[:-1]
431
parent_id = self.path2id(parent_path)
432
if parent_id == None:
433
raise NotVersionedError(parent_path)
372
bailout("cannot re-add root of inventory")
375
file_id = bzrlib.branch.gen_file_id(relpath)
377
parent_id = self.path2id(parts[:-1])
435
378
ie = InventoryEntry(file_id, parts[-1],
436
379
kind=kind, parent_id=parent_id)
437
380
return self.add(ie)
451
394
ie = self[file_id]
453
assert self[ie.parent_id].children[ie.name] == ie
396
assert self._tree[ie.parent_id][ie.name] == ie
455
398
# TODO: Test deleting all children; maybe hoist to a separate
456
399
# deltree method?
457
400
if ie.kind == 'directory':
458
for cie in ie.children.values():
401
for cie in self._tree[file_id].values():
459
402
del self[cie.file_id]
403
del self._tree[file_id]
462
405
del self._byid[file_id]
463
del self[ie.parent_id].children[ie.name]
406
del self._tree[ie.parent_id][ie.name]
410
return Set(self._byid)
466
413
def to_element(self):
467
414
"""Convert to XML Element"""
468
from bzrlib.xml import Element
470
415
e = Element('inventory')
472
417
for path, ie in self.iter_entries():
493
438
from_element = classmethod(from_element)
496
def __eq__(self, other):
441
def __cmp__(self, other):
497
442
"""Compare two sets by comparing their contents.
499
444
>>> i1 = Inventory()
500
445
>>> i2 = Inventory()
503
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
448
>>> i1.add(InventoryEntry('123', 'foo'))
506
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
451
>>> i2.add(InventoryEntry('123', 'foo'))
510
458
if not isinstance(other, Inventory):
511
459
return NotImplemented
513
if len(self._byid) != len(other._byid):
514
# shortcut: obviously not the same
517
return self._byid == other._byid
520
def __ne__(self, other):
521
return not (self == other)
525
raise ValueError('not hashable')
529
def get_idpath(self, file_id):
530
"""Return a list of file_ids for the path to an entry.
532
The list contains one element for each directory followed by
533
the id of the file itself. So the length of the returned list
534
is equal to the depth of the file in the tree, counting the
535
root directory as depth 1.
461
if self.id_set() ^ other.id_set():
464
for file_id in self._byid:
465
c = cmp(self[file_id], other[file_id])
471
def id2path(self, file_id):
472
"""Return as a list the path to file_id."""
538
474
while file_id != None:
540
ie = self._byid[file_id]
542
raise BzrError("file_id {%s} not found in inventory" % file_id)
543
p.insert(0, ie.file_id)
544
477
file_id = ie.parent_id
548
def id2path(self, file_id):
549
"""Return as a list the path to file_id."""
551
# get all names, skipping root
552
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
553
return os.sep.join(p)
563
488
This returns the entry of the last component in the path,
564
489
which may be either a file or a directory.
566
Returns None iff the path is not found.
568
491
if isinstance(name, types.StringTypes):
569
492
name = splitpath(name)
571
mutter("lookup path %r" % name)
576
cie = parent.children[f]
497
cie = self._tree[parent_id][f]
577
498
assert cie.name == f
578
assert cie.parent_id == parent.file_id
499
parent_id = cie.file_id
581
501
# or raise an error?
584
return parent.file_id
507
def get_child(self, parent_id, child_name):
508
return self._tree[parent_id].get(child_name)
587
511
def has_filename(self, names):
591
515
def has_id(self, file_id):
516
assert isinstance(file_id, str)
592
517
return self._byid.has_key(file_id)
595
def rename(self, file_id, new_parent_id, new_name):
596
"""Move a file within the inventory.
598
This can change either the name, or the parent, or both.
600
This does not move the working file."""
601
if not is_valid_name(new_name):
602
raise BzrError("not an acceptable filename: %r" % new_name)
604
new_parent = self._byid[new_parent_id]
605
if new_name in new_parent.children:
606
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
608
new_parent_idpath = self.get_idpath(new_parent_id)
609
if file_id in new_parent_idpath:
610
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
611
% (self.id2path(file_id), self.id2path(new_parent_id)))
613
file_ie = self._byid[file_id]
614
old_parent = self._byid[file_ie.parent_id]
616
# TODO: Don't leave things messed up if this fails
618
del old_parent.children[file_ie.name]
619
new_parent.children[new_name] = file_ie
621
file_ie.name = new_name
622
file_ie.parent_id = new_parent_id
627
_NAME_RE = re.compile(r'^[^/\\]+$')
629
def is_valid_name(name):
630
return bool(_NAME_RE.match(name))
523
if __name__ == '__main__':
524
import doctest, inventory
525
doctest.testmod(inventory)