« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 26
- compare with another revision
- download diff
- download tarball
- view history from revision 26
- Committer: mbp at sourcefrog
- Date: 2005-03-12 07:34:32 UTC
- Revision ID: mbp@sourcefrog.net-20050312073432-b2c16a55e0d6e9fb
fix StopIteration error on python2.3(?)
- files added:
- doc/roadmap.txt
- files removed:
- .rsyncexclude
- elementtree
- urlgrabber
- files renamed:
- bzrlib/commands.py => bzr.py
- files modified:
- .bzrignore
added
removed
bzrlib/inventory.py
1
# (C) 2005 Canonical Ltd
1
#! /usr/bin/env python
2
# -*- coding: UTF-8 -*-
2
3
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
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
16
17
17
18
# TODO: Maybe store inventory_id in the file? Not really needed.
19
20
21
# This should really be an id randomly assigned when the tree is
22
# created, but it's not for now.
23
ROOT_ID = "TREE_ROOT"
24
25
26
import sys, os.path, types, re
18
"""Inventories map files to their name in a revision."""
19
20
21
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
22
__author__ = "Martin Pool <mbp@canonical.com>"
23
24
import sys, os.path, types
27
25
from sets import Set
28
26
29
27
try:
30
28
from cElementTree import Element, ElementTree, SubElement
31
29
except ImportError:
32
from elementtree.ElementTree import Element, ElementTree, SubElement
30
from elementtree import Element, ElementTree, SubElement
33
31
34
32
from xml import XMLMixin
35
from errors import bailout, BzrError
36
37
import bzrlib
38
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
39
from bzrlib.trace import mutter
33
from errors import bailout
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
40
36
41
37
class InventoryEntry(XMLMixin):
42
38
"""Description of a versioned file.
61
57
62
58
>>> i = Inventory()
63
59
>>> i.path2id('')
64
'TREE_ROOT'
65
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
66
>>> 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'))
67
62
>>> for j in i.iter_entries():
68
63
... print j
69
64
...
70
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
71
66
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
72
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
73
68
Traceback (most recent call last):
74
69
...
75
70
BzrError: ('inventory already contains entry with id {2323}', [])
76
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
77
>>> 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'))
78
73
>>> i.path2id('src/wibble')
79
74
'2325'
80
75
>>> '2325' in i
81
76
True
82
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
83
78
>>> i['2326']
84
79
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
85
80
>>> for j in i.iter_entries():
94
89
>>> i.id2path('2326')
95
90
'src/wibble/wibble.c'
96
91
97
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?
98
93
But those depend on its position within a particular inventory, and
99
94
it would be nice not to need to hold the backpointer here.
100
95
"""
101
102
# TODO: split InventoryEntry into subclasses for files,
103
# directories, etc etc.
104
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,
97
parent_id=None):
106
98
"""Create an InventoryEntry
107
99
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.
110
102
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
103
>>> e = InventoryEntry('123', 'hello.c')
112
104
>>> e.name
113
105
'hello.c'
114
106
>>> e.file_id
115
107
'123'
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'", [])
119
111
"""
124
116
125
117
self.file_id = file_id
126
118
self.name = name
119
assert kind in ['file', 'directory']
127
120
self.kind = kind
128
121
self.text_id = text_id
129
122
self.parent_id = parent_id
130
123
self.text_sha1 = None
131
124
self.text_size = None
132
if kind == 'directory':
133
self.children = {}
134
elif kind == 'file':
135
pass
136
else:
137
raise BzrError("unhandled entry kind %r" % kind)
138
139
140
141
def sorted_children(self):
142
l = self.children.items()
143
l.sort()
144
return l
145
125
146
126
147
127
def copy(self):
148
128
other = InventoryEntry(self.file_id, self.name, self.kind,
149
self.parent_id, text_id=self.text_id)
129
self.text_id, self.parent_id)
150
130
other.text_sha1 = self.text_sha1
151
131
other.text_size = self.text_size
152
132
return other
169
149
e.set('file_id', self.file_id)
170
150
e.set('kind', self.kind)
171
151
172
if self.text_size != None:
152
if self.text_size is not None:
173
153
e.set('text_size', '%d' % self.text_size)
174
154
175
for f in ['text_id', 'text_sha1']:
155
for f in ['text_id', 'text_sha1', 'parent_id']:
176
156
v = getattr(self, f)
177
if v != None:
157
if v is not None:
178
158
e.set(f, v)
179
159
180
# to be conservative, we don't externalize the root pointers
181
# for now, leaving them as null in the xml form. in a future
182
# version it will be implied by nested elements.
183
if self.parent_id != ROOT_ID:
184
assert isinstance(self.parent_id, basestring)
185
e.set('parent_id', self.parent_id)
186
187
160
e.tail = '\n'
188
161
189
162
return e
191
164
192
165
def from_element(cls, elt):
193
166
assert elt.tag == 'entry'
194
195
## original format inventories don't have a parent_id for
196
## nodes in the root directory, but it's cleaner to use one
197
## internally.
198
parent_id = elt.get('parent_id')
199
if parent_id == None:
200
parent_id = ROOT_ID
201
202
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'))
203
168
self.text_id = elt.get('text_id')
204
169
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
205
171
206
172
## mutter("read inventoryentry: %r" % (elt.attrib))
207
173
229
195
230
196
231
197
232
class RootEntry(InventoryEntry):
233
def __init__(self, file_id):
234
self.file_id = file_id
235
self.children = {}
236
self.kind = 'root_directory'
237
self.parent_id = None
238
self.name = ''
239
240
def __cmp__(self, other):
241
if self is other:
242
return 0
243
if not isinstance(other, RootEntry):
244
return NotImplemented
245
return cmp(self.file_id, other.file_id) \
246
or cmp(self.children, other.children)
247
248
249
250
198
class Inventory(XMLMixin):
251
199
"""Inventory of versioned files in a tree.
252
200
253
This describes which file_id is present at each point in the tree,
254
and possibly the SHA-1 or other information about the file.
255
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.
203
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.
256
207
257
208
The inventory represents a typical unix file tree, with
258
209
directories containing files and subdirectories. We never store
262
213
returned quickly.
263
214
264
215
InventoryEntry objects must not be modified after they are
265
inserted, other than through the Inventory API.
216
inserted.
266
217
267
218
>>> inv = Inventory()
268
219
>>> inv.write_xml(sys.stdout)
269
220
<inventory>
270
221
</inventory>
271
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
272
223
>>> inv['123-123'].name
273
224
'hello.c'
225
>>> for file_id in inv: print file_id
226
...
227
123-123
274
228
275
229
May be treated as an iterator or set to look up file ids:
276
230
291
245
292
246
"""
293
247
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.
251
294
252
## TODO: Make sure only canonical filenames are stored.
295
253
296
254
## TODO: Do something sensible about the possible collisions on
297
255
## case-losing filesystems. Perhaps we should just always forbid
298
256
## such collisions.
299
257
300
## TODO: No special cases for root, rather just give it a file id
301
## like everything else.
302
303
## TODO: Probably change XML serialization to use nesting rather
304
## than parent_id pointers.
305
306
## TODO: Perhaps hold the ElementTree in memory and work directly
307
## on that rather than converting into Python objects every time?
258
## _tree should probably just be stored as
259
## InventoryEntry._children on each directory.
308
260
309
261
def __init__(self):
310
262
"""Create or read an inventory.
312
264
If a working directory is specified, the inventory is read
313
265
from there. If the file is specified, read from that. If not,
314
266
the inventory is created empty.
315
316
The inventory is created with a default root directory, with
317
an id of None.
318
267
"""
319
self.root = RootEntry(ROOT_ID)
320
self._byid = {self.root.file_id: self.root}
268
self._byid = dict()
269
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: {}}
321
273
322
274
323
275
def __iter__(self):
329
281
return len(self._byid)
330
282
331
283
332
def iter_entries(self, from_dir=None):
284
def iter_entries(self, parent_id=None):
333
285
"""Return (path, entry) pairs, in order by name."""
334
if from_dir == None:
335
assert self.root
336
from_dir = self.root
337
elif isinstance(from_dir, basestring):
338
from_dir = self._byid[from_dir]
339
340
kids = from_dir.children.items()
286
kids = self._tree[parent_id].items()
341
287
kids.sort()
342
288
for name, ie in kids:
343
289
yield name, ie
344
290
if ie.kind == 'directory':
345
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
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
293
294
295
def directories(self, include_root=True):
296
"""Return (path, entry) pairs for all directories.
297
"""
298
if include_root:
299
yield '', None
300
for path, entry in self.iter_entries():
301
if entry.kind == 'directory':
302
yield path, entry
303
304
305
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
347
309
348
310
349
311
350
def directories(self):
351
"""Return (path, entry) pairs for all directories.
352
"""
353
def descend(parent_ie):
354
parent_name = parent_ie.name
355
yield parent_name, parent_ie
356
357
# directory children in sorted order
358
dn = []
359
for ie in parent_ie.children.itervalues():
360
if ie.kind == 'directory':
361
dn.append((ie.name, ie))
362
dn.sort()
363
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
367
368
for name, ie in descend(self.root):
369
yield name, ie
370
312
# TODO: return all paths and entries
371
313
372
314
373
315
def __contains__(self, file_id):
374
316
"""True if this entry contains a file with given id.
375
317
376
318
>>> inv = Inventory()
377
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
378
320
>>> '123' in inv
379
321
True
380
322
>>> '456' in inv
387
329
"""Return the entry for given file_id.
388
330
389
331
>>> inv = Inventory()
390
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
391
333
>>> inv['123123'].name
392
334
'hello.c'
393
335
"""
394
if file_id == None:
395
raise BzrError("can't look up file_id None")
396
397
try:
398
return self._byid[file_id]
399
except KeyError:
400
raise BzrError("file_id {%s} not in inventory" % file_id)
401
402
403
def get_child(self, parent_id, filename):
404
return self[parent_id].children.get(filename)
336
return self._byid[file_id]
405
337
406
338
407
339
def add(self, entry):
409
341
410
342
To add a file to a branch ready to be committed, use Branch.add,
411
343
which calls this."""
412
if entry.file_id in self._byid:
344
if entry.file_id in self:
413
345
bailout("inventory already contains entry with id {%s}" % entry.file_id)
414
346
415
try:
416
parent = self._byid[entry.parent_id]
417
except KeyError:
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
419
420
if parent.children.has_key(entry.name):
421
bailout("%s is already versioned" %
422
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"
350
% entry.parent_id)
351
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))
423
355
424
356
self._byid[entry.file_id] = entry
425
parent.children[entry.name] = entry
426
427
428
def add_path(self, relpath, kind, file_id=None):
429
"""Add entry from a path.
430
431
The immediate parent must already be versioned"""
432
parts = bzrlib.osutils.splitpath(relpath)
433
if len(parts) == 0:
434
bailout("cannot re-add root of inventory")
435
436
if file_id == None:
437
file_id = bzrlib.branch.gen_file_id(relpath)
438
439
parent_id = self.path2id(parts[:-1])
440
assert parent_id != None
441
ie = InventoryEntry(file_id, parts[-1],
442
kind=kind, parent_id=parent_id)
443
return self.add(ie)
357
self._tree[entry.parent_id][entry.name] = entry
358
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
444
361
445
362
446
363
def __delitem__(self, file_id):
447
364
"""Remove entry by id.
448
365
449
366
>>> inv = Inventory()
450
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
451
368
>>> '123' in inv
452
369
True
453
370
>>> del inv['123']
456
373
"""
457
374
ie = self[file_id]
458
375
459
assert self[ie.parent_id].children[ie.name] == ie
376
assert self._tree[ie.parent_id][ie.name] == ie
460
377
461
378
# TODO: Test deleting all children; maybe hoist to a separate
462
379
# deltree method?
463
380
if ie.kind == 'directory':
464
for cie in ie.children.values():
381
for cie in self._tree[file_id].values():
465
382
del self[cie.file_id]
466
del ie.children
383
del self._tree[file_id]
467
384
468
385
del self._byid[file_id]
469
del self[ie.parent_id].children[ie.name]
386
del self._tree[ie.parent_id][ie.name]
470
387
471
388
472
389
def id_set(self):
486
403
"""Construct from XML Element
487
404
488
405
>>> inv = Inventory()
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
490
407
>>> elt = inv.to_element()
491
408
>>> inv2 = Inventory.from_element(elt)
492
409
>>> inv2 == inv
508
425
>>> i2 = Inventory()
509
426
>>> i1 == i2
510
427
True
511
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
428
>>> i1.add(InventoryEntry('123', 'foo'))
512
429
>>> i1 == i2
513
430
False
514
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
431
>>> i2.add(InventoryEntry('123', 'foo'))
515
432
>>> i1 == i2
516
433
True
517
434
"""
531
448
return 0
532
449
533
450
534
def get_idpath(self, file_id):
535
"""Return a list of file_ids for the path to an entry.
536
537
The list contains one element for each directory followed by
538
the id of the file itself. So the length of the returned list
539
is equal to the depth of the file in the tree, counting the
540
root directory as depth 1.
541
"""
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
542
453
p = []
543
454
while file_id != None:
544
try:
545
ie = self._byid[file_id]
546
except KeyError:
547
bailout("file_id {%s} not found in inventory" % file_id)
548
p.insert(0, ie.file_id)
455
ie = self[file_id]
456
p = [ie.name] + p
549
457
file_id = ie.parent_id
550
return p
551
552
553
def id2path(self, file_id):
554
"""Return as a list the path to file_id."""
555
556
# get all names, skipping root
557
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
558
return os.sep.join(p)
458
return joinpath(p)
559
459
560
460
561
461
567
467
568
468
This returns the entry of the last component in the path,
569
469
which may be either a file or a directory.
570
571
Returns None iff the path is not found.
572
470
"""
573
if isinstance(name, types.StringTypes):
574
name = splitpath(name)
575
576
mutter("lookup path %r" % name)
577
578
parent = self.root
579
for f in name:
471
assert isinstance(name, types.StringTypes)
472
473
parent_id = None
474
for f in splitpath(name):
580
475
try:
581
cie = parent.children[f]
476
cie = self._tree[parent_id][f]
582
477
assert cie.name == f
583
assert cie.parent_id == parent.file_id
584
parent = cie
478
parent_id = cie.file_id
585
479
except KeyError:
586
480
# or raise an error?
587
481
return None
588
482
589
return parent.file_id
483
return parent_id
484
485
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
590
488
591
489
592
490
def has_filename(self, names):
594
492
595
493
596
494
def has_id(self, file_id):
495
assert isinstance(file_id, str)
597
496
return self._byid.has_key(file_id)
598
497
599
498
600
def rename(self, file_id, new_parent_id, new_name):
601
"""Move a file within the inventory.
602
603
This can change either the name, or the parent, or both.
604
605
This does not move the working file."""
606
if not is_valid_name(new_name):
607
bailout("not an acceptable filename: %r" % new_name)
608
609
new_parent = self._byid[new_parent_id]
610
if new_name in new_parent.children:
611
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
612
613
new_parent_idpath = self.get_idpath(new_parent_id)
614
if file_id in new_parent_idpath:
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
616
% (self.id2path(file_id), self.id2path(new_parent_id)))
617
618
file_ie = self._byid[file_id]
619
old_parent = self._byid[file_ie.parent_id]
620
621
# TODO: Don't leave things messed up if this fails
622
623
del old_parent.children[file_ie.name]
624
new_parent.children[new_name] = file_ie
625
626
file_ie.name = new_name
627
file_ie.parent_id = new_parent_id
628
629
630
631
632
_NAME_RE = re.compile(r'^[^/\\]+$')
633
634
def is_valid_name(name):
635
return bool(_NAME_RE.match(name))
499
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1