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
24
25
from sets import Set
29
30
from elementtree.ElementTree import Element, ElementTree, SubElement
31
32
from xml import XMLMixin
32
from errors import bailout, BzrError
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
36
from bzrlib.trace import mutter
33
from errors import bailout
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
38
37
class InventoryEntry(XMLMixin):
39
38
"""Description of a versioned file.
59
58
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
63
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
64
62
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
68
66
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
70
68
Traceback (most recent call last):
72
70
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
71
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
75
73
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
81
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
80
>>> for j in i.iter_entries():
91
89
>>> i.id2path('2326')
92
90
'src/wibble/wibble.c'
94
TODO: Maybe also keep the full path of the entry, and the children?
92
:todo: Maybe also keep the full path of the entry, and the children?
95
93
But those depend on its position within a particular inventory, and
96
94
it would be nice not to need to hold the backpointer here.
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
96
def __init__(self, file_id, name, kind='file', text_id=None,
106
98
"""Create an InventoryEntry
108
100
The filename must be a single component, relative to the
109
101
parent directory; it cannot be a whole path or relative name.
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
103
>>> e = InventoryEntry('123', 'hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
108
>>> e = InventoryEntry('123', 'src/hello.c')
117
109
Traceback (most recent call last):
118
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
120
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
123
117
self.file_id = file_id
119
assert kind in ['file', 'directory']
126
121
self.text_id = text_id
127
122
self.parent_id = parent_id
128
if kind == 'directory':
133
raise BzrError("unhandled entry kind %r" % kind)
137
def sorted_children(self):
138
l = self.children.items()
123
self.text_sha1 = None
124
self.text_size = None
144
128
other = InventoryEntry(self.file_id, self.name, self.kind,
145
self.parent_id, text_id=self.text_id)
129
self.text_id, self.parent_id)
146
130
other.text_sha1 = self.text_sha1
147
131
other.text_size = self.text_size
165
149
e.set('file_id', self.file_id)
166
150
e.set('kind', self.kind)
168
if self.text_size != None:
152
if self.text_size is not None:
169
153
e.set('text_size', '%d' % self.text_size)
171
for f in ['text_id', 'text_sha1']:
155
for f in ['text_id', 'text_sha1', 'parent_id']:
172
156
v = getattr(self, f)
176
# to be conservative, we don't externalize the root pointers
177
# for now, leaving them as null in the xml form. in a future
178
# version it will be implied by nested elements.
179
if self.parent_id != ROOT_ID:
180
assert isinstance(self.parent_id, basestring)
181
e.set('parent_id', self.parent_id)
188
165
def from_element(cls, elt):
189
166
assert elt.tag == 'entry'
191
## original format inventories don't have a parent_id for
192
## nodes in the root directory, but it's cleaner to use one
194
parent_id = elt.get('parent_id')
195
if parent_id == None:
198
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
199
168
self.text_id = elt.get('text_id')
200
169
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
202
172
## mutter("read inventoryentry: %r" % (elt.attrib))
228
class RootEntry(InventoryEntry):
229
def __init__(self, file_id):
230
self.file_id = file_id
232
self.kind = 'root_directory'
233
self.parent_id = None
236
def __cmp__(self, other):
239
if not isinstance(other, RootEntry):
240
return NotImplemented
241
return cmp(self.file_id, other.file_id) \
242
or cmp(self.children, other.children)
246
198
class Inventory(XMLMixin):
247
199
"""Inventory of versioned files in a tree.
249
This describes which file_id is present at each point in the tree,
250
and possibly the SHA-1 or other information about the file.
251
Entries can be looked up either by path or by file_id.
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.
253
208
The inventory represents a typical unix file tree, with
254
209
directories containing files and subdirectories. We never store
258
213
returned quickly.
260
215
InventoryEntry objects must not be modified after they are
261
inserted, other than through the Inventory API.
263
218
>>> inv = Inventory()
264
219
>>> inv.write_xml(sys.stdout)
267
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
268
223
>>> inv['123-123'].name
225
>>> for file_id in inv: print file_id
271
229
May be treated as an iterator or set to look up file ids:
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.
289
261
def __init__(self):
290
262
"""Create or read an inventory.
292
264
If a working directory is specified, the inventory is read
293
265
from there. If the file is specified, read from that. If not,
294
266
the inventory is created empty.
296
The inventory is created with a default root directory, with
299
self.root = RootEntry(ROOT_ID)
300
self._byid = {self.root.file_id: self.root}
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: {}}
303
275
def __iter__(self):
309
281
return len(self._byid)
312
def iter_entries(self, from_dir=None):
284
def iter_entries(self, parent_id=None):
313
285
"""Return (path, entry) pairs, in order by name."""
317
elif isinstance(from_dir, basestring):
318
from_dir = self._byid[from_dir]
320
kids = from_dir.children.items()
286
kids = self._tree[parent_id].items()
322
288
for name, ie in kids:
324
290
if ie.kind == 'directory':
325
for cn, cie in self.iter_entries(from_dir=ie.file_id):
326
yield os.path.join(name, cn), cie
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':
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
330
def directories(self):
331
"""Return (path, entry) pairs for all directories.
333
def descend(parent_ie):
334
parent_name = parent_ie.name
335
yield parent_name, parent_ie
337
# directory children in sorted order
339
for ie in parent_ie.children.itervalues():
340
if ie.kind == 'directory':
341
dn.append((ie.name, ie))
344
for name, child_ie in dn:
345
for sub_name, sub_ie in descend(child_ie):
346
yield appendpath(parent_name, sub_name), sub_ie
348
for name, ie in descend(self.root):
312
# TODO: return all paths and entries
353
315
def __contains__(self, file_id):
354
316
"""True if this entry contains a file with given id.
356
318
>>> inv = Inventory()
357
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
367
329
"""Return the entry for given file_id.
369
331
>>> inv = Inventory()
370
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
371
333
>>> inv['123123'].name
375
return self._byid[file_id]
378
raise BzrError("can't look up file_id None")
380
raise BzrError("file_id {%s} not in inventory" % file_id)
383
def get_child(self, parent_id, filename):
384
return self[parent_id].children.get(filename)
336
return self._byid[file_id]
387
339
def add(self, entry):
390
342
To add a file to a branch ready to be committed, use Branch.add,
391
343
which calls this."""
392
if entry.file_id in self._byid:
344
if entry.file_id in self:
393
345
bailout("inventory already contains entry with id {%s}" % entry.file_id)
396
parent = self._byid[entry.parent_id]
398
bailout("parent_id {%s} not in inventory" % entry.parent_id)
400
if parent.children.has_key(entry.name):
401
bailout("%s is already versioned" %
402
appendpath(self.id2path(parent.file_id), entry.name))
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))
404
356
self._byid[entry.file_id] = entry
405
parent.children[entry.name] = entry
408
def add_path(self, relpath, kind, file_id=None):
409
"""Add entry from a path.
411
The immediate parent must already be versioned"""
412
parts = bzrlib.osutils.splitpath(relpath)
414
bailout("cannot re-add root of inventory")
417
file_id = bzrlib.branch.gen_file_id(relpath)
419
parent_id = self.path2id(parts[:-1])
420
assert parent_id != None
421
ie = InventoryEntry(file_id, parts[-1],
422
kind=kind, parent_id=parent_id)
357
self._tree[entry.parent_id][entry.name] = entry
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
426
363
def __delitem__(self, file_id):
427
364
"""Remove entry by id.
429
366
>>> inv = Inventory()
430
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
433
370
>>> del inv['123']
437
374
ie = self[file_id]
439
assert self[ie.parent_id].children[ie.name] == ie
376
assert self._tree[ie.parent_id][ie.name] == ie
441
378
# TODO: Test deleting all children; maybe hoist to a separate
442
379
# deltree method?
443
380
if ie.kind == 'directory':
444
for cie in ie.children.values():
381
for cie in self._tree[file_id].values():
445
382
del self[cie.file_id]
383
del self._tree[file_id]
448
385
del self._byid[file_id]
449
del self[ie.parent_id].children[ie.name]
386
del self._tree[ie.parent_id][ie.name]
452
389
def id_set(self):
466
403
"""Construct from XML Element
468
405
>>> inv = Inventory()
469
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
470
407
>>> elt = inv.to_element()
471
408
>>> inv2 = Inventory.from_element(elt)
488
425
>>> i2 = Inventory()
491
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
428
>>> i1.add(InventoryEntry('123', 'foo'))
494
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
431
>>> i2.add(InventoryEntry('123', 'foo'))
514
def get_idpath(self, file_id):
515
"""Return a list of file_ids for the path to an entry.
517
The list contains one element for each directory followed by
518
the id of the file itself. So the length of the returned list
519
is equal to the depth of the file in the tree, counting the
520
root directory as depth 1.
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
523
454
while file_id != None:
525
ie = self._byid[file_id]
527
bailout("file_id {%s} not found in inventory" % file_id)
528
p.insert(0, ie.file_id)
529
457
file_id = ie.parent_id
533
def id2path(self, file_id):
534
"""Return as a list the path to file_id."""
536
# get all names, skipping root
537
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
538
return os.sep.join(p)
548
468
This returns the entry of the last component in the path,
549
469
which may be either a file or a directory.
551
Returns None iff the path is not found.
553
if isinstance(name, types.StringTypes):
554
name = splitpath(name)
556
mutter("lookup path %r" % name)
471
assert isinstance(name, types.StringTypes)
474
for f in splitpath(name):
561
cie = parent.children[f]
476
cie = self._tree[parent_id][f]
562
477
assert cie.name == f
563
assert cie.parent_id == parent.file_id
478
parent_id = cie.file_id
566
480
# or raise an error?
569
return parent.file_id
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
572
490
def has_filename(self, names):
576
494
def has_id(self, file_id):
495
assert isinstance(file_id, str)
577
496
return self._byid.has_key(file_id)
580
def rename(self, file_id, new_parent_id, new_name):
581
"""Move a file within the inventory.
583
This can change either the name, or the parent, or both.
585
This does not move the working file."""
586
if not is_valid_name(new_name):
587
bailout("not an acceptable filename: %r" % new_name)
589
new_parent = self._byid[new_parent_id]
590
if new_name in new_parent.children:
591
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
593
new_parent_idpath = self.get_idpath(new_parent_id)
594
if file_id in new_parent_idpath:
595
bailout("cannot move directory %r into a subdirectory of itself, %r"
596
% (self.id2path(file_id), self.id2path(new_parent_id)))
598
file_ie = self._byid[file_id]
599
old_parent = self._byid[file_ie.parent_id]
601
# TODO: Don't leave things messed up if this fails
603
del old_parent.children[file_ie.name]
604
new_parent.children[new_name] = file_ie
606
file_ie.name = new_name
607
file_ie.parent_id = new_parent_id
612
_NAME_RE = re.compile(r'^[^/\\]+$')
614
def is_valid_name(name):
615
return bool(_NAME_RE.match(name))
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)