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
27
from cElementTree import Element, ElementTree, SubElement
29
from elementtree.ElementTree import Element, ElementTree, SubElement
31
27
from xml import XMLMixin
32
from errors import bailout, BzrError, BzrCheckError
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
36
from bzrlib.trace import mutter
28
from ElementTree import ElementTree, Element
29
from errors import bailout
30
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
31
from trace import mutter
38
33
class InventoryEntry(XMLMixin):
39
34
"""Description of a versioned file.
59
54
>>> i = Inventory()
62
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
63
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
56
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
57
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
64
58
>>> for j in i.iter_entries():
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
61
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
68
62
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
63
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
70
64
Traceback (most recent call last):
72
66
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'))
67
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
68
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
75
69
>>> i.path2id('src/wibble')
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
73
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
81
75
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
76
>>> for j in i.iter_entries():
91
85
>>> i.id2path('2326')
92
86
'src/wibble/wibble.c'
94
TODO: Maybe also keep the full path of the entry, and the children?
88
:todo: Maybe also keep the full path of the entry, and the children?
95
89
But those depend on its position within a particular inventory, and
96
90
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):
92
def __init__(self, file_id, name, kind='file', text_id=None,
106
94
"""Create an InventoryEntry
108
96
The filename must be a single component, relative to the
109
97
parent directory; it cannot be a whole path or relative name.
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
99
>>> e = InventoryEntry('123', 'hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
104
>>> e = InventoryEntry('123', 'src/hello.c')
117
105
Traceback (most recent call last):
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
106
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)
109
if len(splitpath(name)) != 1:
110
bailout('InventoryEntry name is not a simple filename: %r'
123
113
self.file_id = file_id
115
assert kind in ['file', 'directory']
126
117
self.text_id = text_id
127
118
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()
119
self.text_sha1 = None
120
self.text_size = None
144
124
other = InventoryEntry(self.file_id, self.name, self.kind,
145
self.parent_id, text_id=self.text_id)
125
self.text_id, self.parent_id)
146
126
other.text_sha1 = self.text_sha1
147
127
other.text_size = self.text_size
148
# note that children are *not* copied; they're pulled across when
167
145
e.set('file_id', self.file_id)
168
146
e.set('kind', self.kind)
170
if self.text_size != None:
148
if self.text_size is not None:
171
149
e.set('text_size', '%d' % self.text_size)
173
for f in ['text_id', 'text_sha1']:
151
for f in ['text_id', 'text_sha1', 'parent_id']:
174
152
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
161
def from_element(cls, elt):
191
162
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)
163
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
201
164
self.text_id = elt.get('text_id')
202
165
self.text_sha1 = elt.get('text_sha1')
166
self.parent_id = elt.get('parent_id')
204
168
## mutter("read inventoryentry: %r" % (elt.attrib))
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 __cmp__(self, other):
241
if not isinstance(other, RootEntry):
242
return NotImplemented
243
return cmp(self.file_id, other.file_id) \
244
or cmp(self.children, other.children)
248
194
class Inventory(XMLMixin):
249
195
"""Inventory of versioned files in a tree.
251
This describes which file_id is present at each point in the tree,
252
and possibly the SHA-1 or other information about the file.
253
Entries can be looked up either by path or by file_id.
197
An Inventory acts like a set of InventoryEntry items. You can
198
also look files up by their file_id or name.
200
May be read from and written to a metadata file in a tree. To
201
manipulate the inventory (for example to add a file), it is read
202
in, modified, and then written back out.
255
204
The inventory represents a typical unix file tree, with
256
205
directories containing files and subdirectories. We never store
260
209
returned quickly.
262
211
InventoryEntry objects must not be modified after they are
263
inserted, other than through the Inventory API.
265
214
>>> inv = Inventory()
266
215
>>> inv.write_xml(sys.stdout)
269
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
218
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
270
219
>>> inv['123-123'].name
221
>>> for file_id in inv: print file_id
273
225
May be treated as an iterator or set to look up file ids:
244
## TODO: Clear up handling of files in subdirectories; we probably
245
## do want to be able to just look them up by name but this
246
## probably means gradually walking down the path, looking up as we go.
248
## TODO: Make sure only canonical filenames are stored.
250
## TODO: Do something sensible about the possible collisions on
251
## case-losing filesystems. Perhaps we should just always forbid
254
## _tree should probably just be stored as
255
## InventoryEntry._children on each directory.
291
257
def __init__(self):
292
258
"""Create or read an inventory.
294
260
If a working directory is specified, the inventory is read
295
261
from there. If the file is specified, read from that. If not,
296
262
the inventory is created empty.
298
The inventory is created with a default root directory, with
301
self.root = RootEntry(ROOT_ID)
302
self._byid = {self.root.file_id: self.root}
266
# _tree is indexed by parent_id; at each level a map from name
267
# to ie. The None entry is the root.
268
self._tree = {None: {}}
305
271
def __iter__(self):
311
277
return len(self._byid)
314
def iter_entries(self, from_dir=None):
280
def iter_entries(self, parent_id=None):
315
281
"""Return (path, entry) pairs, in order by name."""
319
elif isinstance(from_dir, basestring):
320
from_dir = self._byid[from_dir]
322
kids = from_dir.children.items()
282
kids = self._tree[parent_id].items()
324
284
for name, ie in kids:
326
286
if ie.kind == 'directory':
327
for cn, cie in self.iter_entries(from_dir=ie.file_id):
328
yield os.path.join(name, cn), cie
287
for cn, cie in self.iter_entries(parent_id=ie.file_id):
288
yield joinpath([name, cn]), cie
291
def directories(self, include_root=True):
292
"""Return (path, entry) pairs for all directories.
296
for path, entry in self.iter_entries():
297
if entry.kind == 'directory':
302
def children(self, parent_id):
303
"""Return entries that are direct children of parent_id."""
304
return self._tree[parent_id]
332
def directories(self):
333
"""Return (path, entry) pairs for all directories.
335
def descend(parent_ie):
336
parent_name = parent_ie.name
337
yield parent_name, parent_ie
339
# directory children in sorted order
341
for ie in parent_ie.children.itervalues():
342
if ie.kind == 'directory':
343
dn.append((ie.name, ie))
346
for name, child_ie in dn:
347
for sub_name, sub_ie in descend(child_ie):
348
yield appendpath(parent_name, sub_name), sub_ie
350
for name, ie in descend(self.root):
308
# TODO: return all paths and entries
355
311
def __contains__(self, file_id):
356
312
"""True if this entry contains a file with given id.
358
314
>>> inv = Inventory()
359
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
315
>>> inv.add(InventoryEntry('123', 'foo.c'))
369
325
"""Return the entry for given file_id.
371
327
>>> inv = Inventory()
372
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
328
>>> inv.add(InventoryEntry('123123', 'hello.c'))
373
329
>>> inv['123123'].name
377
return self._byid[file_id]
380
raise BzrError("can't look up file_id None")
382
raise BzrError("file_id {%s} not in inventory" % file_id)
385
def get_file_kind(self, file_id):
386
return self._byid[file_id].kind
388
def get_child(self, parent_id, filename):
389
return self[parent_id].children.get(filename)
332
return self._byid[file_id]
392
335
def add(self, entry):
395
338
To add a file to a branch ready to be committed, use Branch.add,
396
339
which calls this."""
397
if entry.file_id in self._byid:
340
if entry.file_id in self:
398
341
bailout("inventory already contains entry with id {%s}" % entry.file_id)
401
parent = self._byid[entry.parent_id]
403
bailout("parent_id {%s} not in inventory" % entry.parent_id)
405
if parent.children.has_key(entry.name):
406
bailout("%s is already versioned" %
407
appendpath(self.id2path(parent.file_id), entry.name))
343
if entry.parent_id != None:
344
if entry.parent_id not in self:
345
bailout("parent_id %s of new entry not found in inventory"
348
if self._tree[entry.parent_id].has_key(entry.name):
349
bailout("%s is already versioned"
350
% appendpath(self.id2path(entry.parent_id), entry.name))
409
352
self._byid[entry.file_id] = entry
410
parent.children[entry.name] = entry
413
def add_path(self, relpath, kind, file_id=None):
414
"""Add entry from a path.
416
The immediate parent must already be versioned"""
417
parts = bzrlib.osutils.splitpath(relpath)
419
bailout("cannot re-add root of inventory")
422
file_id = bzrlib.branch.gen_file_id(relpath)
424
parent_id = self.path2id(parts[:-1])
425
assert parent_id != None
426
ie = InventoryEntry(file_id, parts[-1],
427
kind=kind, parent_id=parent_id)
353
self._tree[entry.parent_id][entry.name] = entry
355
if entry.kind == 'directory':
356
self._tree[entry.file_id] = {}
431
359
def __delitem__(self, file_id):
432
360
"""Remove entry by id.
434
362
>>> inv = Inventory()
435
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
363
>>> inv.add(InventoryEntry('123', 'foo.c'))
438
366
>>> del inv['123']
442
370
ie = self[file_id]
444
assert self[ie.parent_id].children[ie.name] == ie
372
assert self._tree[ie.parent_id][ie.name] == ie
446
374
# TODO: Test deleting all children; maybe hoist to a separate
447
375
# deltree method?
448
376
if ie.kind == 'directory':
449
for cie in ie.children.values():
377
for cie in self._tree[file_id].values():
450
378
del self[cie.file_id]
379
del self._tree[file_id]
453
381
del self._byid[file_id]
454
del self[ie.parent_id].children[ie.name]
382
del self._tree[ie.parent_id][ie.name]
457
385
def id_set(self):
471
399
"""Construct from XML Element
473
401
>>> inv = Inventory()
474
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
402
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
475
403
>>> elt = inv.to_element()
476
404
>>> inv2 = Inventory.from_element(elt)
493
421
>>> i2 = Inventory()
496
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
424
>>> i1.add(InventoryEntry('123', 'foo'))
499
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
427
>>> i2.add(InventoryEntry('123', 'foo'))
519
def get_idpath(self, file_id):
520
"""Return a list of file_ids for the path to an entry.
522
The list contains one element for each directory followed by
523
the id of the file itself. So the length of the returned list
524
is equal to the depth of the file in the tree, counting the
525
root directory as depth 1.
447
def id2path(self, file_id):
448
"""Return as a list the path to file_id."""
528
450
while file_id != None:
530
ie = self._byid[file_id]
532
bailout("file_id {%s} not found in inventory" % file_id)
533
p.insert(0, ie.file_id)
534
453
file_id = ie.parent_id
538
def id2path(self, file_id):
539
"""Return as a list the path to file_id."""
541
# get all names, skipping root
542
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
543
return os.sep.join(p)
553
464
This returns the entry of the last component in the path,
554
465
which may be either a file or a directory.
556
Returns None iff the path is not found.
558
if isinstance(name, types.StringTypes):
559
name = splitpath(name)
561
mutter("lookup path %r" % name)
467
assert isinstance(name, types.StringTypes)
470
for f in splitpath(name):
566
cie = parent.children[f]
472
cie = self._tree[parent_id][f]
567
473
assert cie.name == f
568
assert cie.parent_id == parent.file_id
474
parent_id = cie.file_id
571
476
# or raise an error?
574
return parent.file_id
482
def get_child(self, parent_id, child_name):
483
return self._tree[parent_id].get(child_name)
577
486
def has_filename(self, names):
581
490
def has_id(self, file_id):
491
assert isinstance(file_id, str)
582
492
return self._byid.has_key(file_id)
585
def rename(self, file_id, new_parent_id, new_name):
586
"""Move a file within the inventory.
588
This can change either the name, or the parent, or both.
590
This does not move the working file."""
591
if not is_valid_name(new_name):
592
bailout("not an acceptable filename: %r" % new_name)
594
new_parent = self._byid[new_parent_id]
595
if new_name in new_parent.children:
596
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
598
new_parent_idpath = self.get_idpath(new_parent_id)
599
if file_id in new_parent_idpath:
600
bailout("cannot move directory %r into a subdirectory of itself, %r"
601
% (self.id2path(file_id), self.id2path(new_parent_id)))
603
file_ie = self._byid[file_id]
604
old_parent = self._byid[file_ie.parent_id]
606
# TODO: Don't leave things messed up if this fails
608
del old_parent.children[file_ie.name]
609
new_parent.children[new_name] = file_ie
611
file_ie.name = new_name
612
file_ie.parent_id = new_parent_id
617
_NAME_RE = re.compile(r'^[^/\\]+$')
619
def is_valid_name(name):
620
return bool(_NAME_RE.match(name))
496
if __name__ == '__main__':
497
import doctest, inventory
498
doctest.testmod(inventory)