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
17
"""Inventories map files to their name in a revision."""
19
# TODO: Maybe store inventory_id in the file? Not really needed.
21
__author__ = "Martin Pool <mbp@canonical.com>"
18
24
# This should really be an id randomly assigned when the tree is
19
25
# created, but it's not for now.
23
29
import sys, os.path, types, re
33
from cElementTree import Element, ElementTree, SubElement
35
from elementtree.ElementTree import Element, ElementTree, SubElement
37
from xml import XMLMixin
38
from errors import bailout, BzrError
26
from bzrlib.errors import BzrError, BzrCheckError
28
41
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
29
42
from bzrlib.trace import mutter
30
from bzrlib.errors import NotVersionedError
33
class InventoryEntry(object):
44
class InventoryEntry(XMLMixin):
34
45
"""Description of a versioned file.
36
47
An InventoryEntry has the following fields, which are also
64
75
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
65
76
Traceback (most recent call last):
67
BzrError: inventory already contains entry with id {2323}
78
BzrError: ('inventory already contains entry with id {2323}', [])
68
79
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
69
80
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
70
81
>>> i.path2id('src/wibble')
86
97
>>> i.id2path('2326')
87
98
'src/wibble/wibble.c'
89
TODO: Maybe also keep the full path of the entry, and the children?
100
:todo: Maybe also keep the full path of the entry, and the children?
90
101
But those depend on its position within a particular inventory, and
91
102
it would be nice not to need to hold the backpointer here.
94
105
# TODO: split InventoryEntry into subclasses for files,
95
106
# directories, etc etc.
97
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
98
'text_id', 'parent_id', 'children', ]
100
108
def __init__(self, file_id, name, kind, parent_id, text_id=None):
101
109
"""Create an InventoryEntry
111
119
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
112
120
Traceback (most recent call last):
113
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
121
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
124
if len(splitpath(name)) != 1:
125
bailout('InventoryEntry name is not a simple filename: %r'
121
128
self.file_id = file_id
130
assert kind in ['file', 'directory']
124
132
self.text_id = text_id
125
133
self.parent_id = parent_id
134
self.text_sha1 = None
135
self.text_size = None
126
136
if kind == 'directory':
127
137
self.children = {}
131
raise BzrError("unhandled entry kind %r" % kind)
139
assert kind == 'file'
135
142
def sorted_children(self):
143
150
self.parent_id, text_id=self.text_id)
144
151
other.text_sha1 = self.text_sha1
145
152
other.text_size = self.text_size
146
# note that children are *not* copied; they're pulled across when
212
215
from_element = classmethod(from_element)
214
def __eq__(self, other):
217
def __cmp__(self, other):
215
220
if not isinstance(other, InventoryEntry):
216
221
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')
223
return cmp(self.file_id, other.file_id) \
224
or cmp(self.name, other.name) \
225
or cmp(self.text_sha1, other.text_sha1) \
226
or cmp(self.text_size, other.text_size) \
227
or cmp(self.text_id, other.text_id) \
228
or cmp(self.parent_id, other.parent_id) \
229
or cmp(self.kind, other.kind)
240
238
self.parent_id = None
243
def __eq__(self, other):
241
def __cmp__(self, other):
244
244
if not isinstance(other, RootEntry):
245
245
return NotImplemented
247
return (self.file_id == other.file_id) \
248
and (self.children == other.children)
252
class Inventory(object):
246
return cmp(self.file_id, other.file_id) \
247
or cmp(self.children, other.children)
251
class Inventory(XMLMixin):
253
252
"""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.
254
An Inventory acts like a set of InventoryEntry items. You can
255
also look files up by their file_id or name.
257
May be read from and written to a metadata file in a tree. To
258
manipulate the inventory (for example to add a file), it is read
259
in, modified, and then written back out.
259
261
The inventory represents a typical unix file tree, with
260
262
directories containing files and subdirectories. We never store
267
269
inserted, other than through the Inventory API.
269
271
>>> inv = Inventory()
272
>>> inv.write_xml(sys.stdout)
270
275
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
271
276
>>> inv['123-123'].name
283
288
>>> [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))
291
>>> inv.write_xml(sys.stdout)
293
<entry file_id="123-123" kind="file" name="hello.c" />
288
def __init__(self, root_id=ROOT_ID):
298
## TODO: Make sure only canonical filenames are stored.
300
## TODO: Do something sensible about the possible collisions on
301
## case-losing filesystems. Perhaps we should just always forbid
304
## TODO: No special cases for root, rather just give it a file id
305
## like everything else.
307
## TODO: Probably change XML serialization to use nesting
289
310
"""Create or read an inventory.
291
312
If a working directory is specified, the inventory is read
295
316
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)
319
self.root = RootEntry(ROOT_ID)
303
320
self._byid = {self.root.file_id: self.root}
327
344
if ie.kind == 'directory':
328
345
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.
346
yield '/'.join((name, cn)), cie
350
def directories(self):
351
"""Return (path, entry) pairs for all directories.
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))
353
def descend(parent_ie):
354
parent_name = parent_ie.name
355
yield parent_name, parent_ie
357
# directory children in sorted order
359
for ie in parent_ie.children.itervalues():
344
360
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))
361
dn.append((ie.name, ie))
358
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
364
for name, child_ie in dn:
365
for sub_name, sub_ie in descend(child_ie):
366
yield appendpath(parent_name, sub_name), sub_ie
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, '')
368
for name, ie in descend(self.root):
387
391
>>> inv['123123'].name
395
raise BzrError("can't look up file_id None")
391
398
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
400
raise BzrError("file_id {%s} not in inventory" % file_id)
402
403
def get_child(self, parent_id, filename):
403
404
return self[parent_id].children.get(filename)
409
410
To add a file to a branch ready to be committed, use Branch.add,
410
411
which calls this."""
411
412
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
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
418
416
parent = self._byid[entry.parent_id]
420
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
422
420
if parent.children.has_key(entry.name):
423
raise BzrError("%s is already versioned" %
421
bailout("%s is already versioned" %
424
422
appendpath(self.id2path(parent.file_id), entry.name))
426
424
self._byid[entry.file_id] = entry
431
429
"""Add entry from a path.
433
431
The immediate parent must already be versioned"""
434
from bzrlib.branch import gen_file_id
436
432
parts = bzrlib.osutils.splitpath(relpath)
437
433
if len(parts) == 0:
438
raise BzrError("cannot re-add root of inventory")
434
bailout("cannot re-add root of inventory")
440
436
if file_id == None:
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)
437
file_id = bzrlib.branch.gen_file_id(relpath)
439
parent_id = self.path2id(parts[:-1])
440
assert parent_id != None
448
441
ie = InventoryEntry(file_id, parts[-1],
449
442
kind=kind, parent_id=parent_id)
450
443
return self.add(ie)
476
469
del self[ie.parent_id].children[ie.name]
473
return Set(self._byid)
479
476
def to_element(self):
480
477
"""Convert to XML Element"""
481
from bzrlib.xml import Element
483
478
e = Element('inventory')
485
if self.root.file_id not in (None, ROOT_ID):
486
e.set('file_id', self.root.file_id)
487
480
for path, ie in self.iter_entries():
488
481
e.append(ie.to_element())
492
485
def from_element(cls, elt):
493
486
"""Construct from XML Element
495
488
>>> inv = Inventory()
496
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
497
490
>>> elt = inv.to_element()
502
# XXXX: doctest doesn't run this properly under python2.3
503
495
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
498
o.add(InventoryEntry.from_element(e))
513
501
from_element = classmethod(from_element)
516
def __eq__(self, other):
504
def __cmp__(self, other):
517
505
"""Compare two sets by comparing their contents.
519
507
>>> i1 = Inventory()
530
521
if not isinstance(other, Inventory):
531
522
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')
524
if self.id_set() ^ other.id_set():
527
for file_id in self._byid:
528
c = cmp(self[file_id], other[file_id])
549
534
def get_idpath(self, file_id):
569
554
"""Return as a list the path to file_id."""
571
556
# 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)
557
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
620
605
This does not move the working file."""
621
606
if not is_valid_name(new_name):
622
raise BzrError("not an acceptable filename: %r" % new_name)
607
bailout("not an acceptable filename: %r" % new_name)
624
609
new_parent = self._byid[new_parent_id]
625
610
if new_name in new_parent.children:
626
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
611
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
628
613
new_parent_idpath = self.get_idpath(new_parent_id)
629
614
if file_id in new_parent_idpath:
630
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
631
616
% (self.id2path(file_id), self.id2path(new_parent_id)))
633
618
file_ie = self._byid[file_id]