« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 282
- compare with another revision
- download diff
- download tarball
- view history from revision 282
- Committer: Martin Pool
- Date: 2005-04-26 05:20:17 UTC
- Revision ID: mbp@sourcefrog.net-20050426052016-8445d0f4fec584d0
- move all TODO items into ./TODO
- files added:
- .bzrignore
- doc
- elementtree
- notes
- urlgrabber
- files renamed:
- bzr.py => bzrlib/commands.py
- files modified:
- README
added
removed
bzrlib/inventory.py
1
#! /usr/bin/env python
2
# -*- coding: UTF-8 -*-
1
# (C) 2005 Canonical Ltd
3
2
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
16
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
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
25
27
from sets import Set
26
28
29
try:
30
from cElementTree import Element, ElementTree, SubElement
31
except ImportError:
32
from elementtree.ElementTree import Element, ElementTree, SubElement
33
27
34
from xml import XMLMixin
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
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
32
40
33
41
class InventoryEntry(XMLMixin):
34
42
"""Description of a versioned file.
53
61
54
62
>>> i = Inventory()
55
63
>>> i.path2id('')
56
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
57
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
64
'TREE_ROOT'
65
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
66
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
58
67
>>> for j in i.iter_entries():
59
68
... print j
60
69
...
61
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
70
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
62
71
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
63
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
64
73
Traceback (most recent call last):
65
74
...
66
75
BzrError: ('inventory already contains entry with id {2323}', [])
67
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
68
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
76
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
77
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
69
78
>>> i.path2id('src/wibble')
70
79
'2325'
71
80
>>> '2325' in i
72
81
True
73
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
82
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
74
83
>>> i['2326']
75
84
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
76
85
>>> for j in i.iter_entries():
85
94
>>> i.id2path('2326')
86
95
'src/wibble/wibble.c'
87
96
88
:todo: Maybe also keep the full path of the entry, and the children?
97
TODO: Maybe also keep the full path of the entry, and the children?
89
98
But those depend on its position within a particular inventory, and
90
99
it would be nice not to need to hold the backpointer here.
91
100
"""
92
def __init__(self, file_id, name, kind='file', text_id=None,
93
parent_id=None):
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):
94
106
"""Create an InventoryEntry
95
107
96
108
The filename must be a single component, relative to the
97
109
parent directory; it cannot be a whole path or relative name.
98
110
99
>>> e = InventoryEntry('123', 'hello.c')
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
100
112
>>> e.name
101
113
'hello.c'
102
114
>>> e.file_id
103
115
'123'
104
>>> e = InventoryEntry('123', 'src/hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
105
117
Traceback (most recent call last):
106
118
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
107
119
"""
112
124
113
125
self.file_id = file_id
114
126
self.name = name
115
assert kind in ['file', 'directory']
116
127
self.kind = kind
117
128
self.text_id = text_id
118
129
self.parent_id = parent_id
119
130
self.text_sha1 = None
120
131
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
121
145
122
146
123
147
def copy(self):
124
148
other = InventoryEntry(self.file_id, self.name, self.kind,
125
self.text_id, self.parent_id)
149
self.parent_id, text_id=self.text_id)
126
150
other.text_sha1 = self.text_sha1
127
151
other.text_size = self.text_size
128
152
return other
145
169
e.set('file_id', self.file_id)
146
170
e.set('kind', self.kind)
147
171
148
if self.text_size is not None:
172
if self.text_size != None:
149
173
e.set('text_size', '%d' % self.text_size)
150
174
151
for f in ['text_id', 'text_sha1', 'parent_id']:
175
for f in ['text_id', 'text_sha1']:
152
176
v = getattr(self, f)
153
if v is not None:
177
if v != None:
154
178
e.set(f, v)
155
179
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
156
187
e.tail = '\n'
157
188
158
189
return e
160
191
161
192
def from_element(cls, elt):
162
193
assert elt.tag == 'entry'
163
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
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)
164
203
self.text_id = elt.get('text_id')
165
204
self.text_sha1 = elt.get('text_sha1')
166
self.parent_id = elt.get('parent_id')
167
205
168
206
## mutter("read inventoryentry: %r" % (elt.attrib))
169
207
191
229
192
230
193
231
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
194
250
class Inventory(XMLMixin):
195
251
"""Inventory of versioned files in a tree.
196
252
197
An Inventory acts like a set of InventoryEntry items. You can
198
also look files up by their file_id or name.
199
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.
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.
203
256
204
257
The inventory represents a typical unix file tree, with
205
258
directories containing files and subdirectories. We never store
209
262
returned quickly.
210
263
211
264
InventoryEntry objects must not be modified after they are
212
inserted.
265
inserted, other than through the Inventory API.
213
266
214
267
>>> inv = Inventory()
215
268
>>> inv.write_xml(sys.stdout)
216
269
<inventory>
217
270
</inventory>
218
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
271
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
219
272
>>> inv['123-123'].name
220
273
'hello.c'
221
>>> for file_id in inv: print file_id
222
...
223
123-123
224
274
225
275
May be treated as an iterator or set to look up file ids:
226
276
241
291
242
292
"""
243
293
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.
247
248
294
## TODO: Make sure only canonical filenames are stored.
249
295
250
296
## TODO: Do something sensible about the possible collisions on
251
297
## case-losing filesystems. Perhaps we should just always forbid
252
298
## such collisions.
253
299
254
## _tree should probably just be stored as
255
## InventoryEntry._children on each directory.
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?
256
308
257
309
def __init__(self):
258
310
"""Create or read an inventory.
260
312
If a working directory is specified, the inventory is read
261
313
from there. If the file is specified, read from that. If not,
262
314
the inventory is created empty.
315
316
The inventory is created with a default root directory, with
317
an id of None.
263
318
"""
264
self._byid = dict()
265
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: {}}
319
self.root = RootEntry(ROOT_ID)
320
self._byid = {self.root.file_id: self.root}
269
321
270
322
271
323
def __iter__(self):
277
329
return len(self._byid)
278
330
279
331
280
def iter_entries(self, parent_id=None):
332
def iter_entries(self, from_dir=None):
281
333
"""Return (path, entry) pairs, in order by name."""
282
kids = self._tree[parent_id].items()
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()
283
341
kids.sort()
284
342
for name, ie in kids:
285
343
yield name, ie
286
344
if ie.kind == 'directory':
287
for cn, cie in self.iter_entries(parent_id=ie.file_id):
288
yield joinpath([name, cn]), cie
289
290
291
def directories(self, include_root=True):
345
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
yield os.path.join(name, cn), cie
347
348
349
350
def directories(self):
292
351
"""Return (path, entry) pairs for all directories.
293
352
"""
294
if include_root:
295
yield '', None
296
for path, entry in self.iter_entries():
297
if entry.kind == 'directory':
298
yield path, entry
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
299
370
300
371
301
372
302
def children(self, parent_id):
303
"""Return entries that are direct children of parent_id."""
304
return self._tree[parent_id]
305
306
307
308
# TODO: return all paths and entries
309
310
311
373
def __contains__(self, file_id):
312
374
"""True if this entry contains a file with given id.
313
375
314
376
>>> inv = Inventory()
315
>>> inv.add(InventoryEntry('123', 'foo.c'))
377
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
316
378
>>> '123' in inv
317
379
True
318
380
>>> '456' in inv
325
387
"""Return the entry for given file_id.
326
388
327
389
>>> inv = Inventory()
328
>>> inv.add(InventoryEntry('123123', 'hello.c'))
390
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
329
391
>>> inv['123123'].name
330
392
'hello.c'
331
393
"""
332
return self._byid[file_id]
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)
333
405
334
406
335
407
def add(self, entry):
337
409
338
410
To add a file to a branch ready to be committed, use Branch.add,
339
411
which calls this."""
340
if entry.file_id in self:
412
if entry.file_id in self._byid:
341
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
342
414
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"
346
% entry.parent_id)
347
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))
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))
351
423
352
424
self._byid[entry.file_id] = entry
353
self._tree[entry.parent_id][entry.name] = entry
354
355
if entry.kind == 'directory':
356
self._tree[entry.file_id] = {}
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
444
358
445
359
446
def __delitem__(self, file_id):
360
447
"""Remove entry by id.
361
448
362
449
>>> inv = Inventory()
363
>>> inv.add(InventoryEntry('123', 'foo.c'))
450
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
364
451
>>> '123' in inv
365
452
True
366
453
>>> del inv['123']
369
456
"""
370
457
ie = self[file_id]
371
458
372
assert self._tree[ie.parent_id][ie.name] == ie
459
assert self[ie.parent_id].children[ie.name] == ie
373
460
374
461
# TODO: Test deleting all children; maybe hoist to a separate
375
462
# deltree method?
376
463
if ie.kind == 'directory':
377
for cie in self._tree[file_id].values():
464
for cie in ie.children.values():
378
465
del self[cie.file_id]
379
del self._tree[file_id]
466
del ie.children
380
467
381
468
del self._byid[file_id]
382
del self._tree[ie.parent_id][ie.name]
469
del self[ie.parent_id].children[ie.name]
383
470
384
471
385
472
def id_set(self):
399
486
"""Construct from XML Element
400
487
401
488
>>> inv = Inventory()
402
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
403
490
>>> elt = inv.to_element()
404
491
>>> inv2 = Inventory.from_element(elt)
405
492
>>> inv2 == inv
421
508
>>> i2 = Inventory()
422
509
>>> i1 == i2
423
510
True
424
>>> i1.add(InventoryEntry('123', 'foo'))
511
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
425
512
>>> i1 == i2
426
513
False
427
>>> i2.add(InventoryEntry('123', 'foo'))
514
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
428
515
>>> i1 == i2
429
516
True
430
517
"""
444
531
return 0
445
532
446
533
447
def id2path(self, file_id):
448
"""Return as a list the path to file_id."""
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
"""
449
542
p = []
450
543
while file_id != None:
451
ie = self[file_id]
452
p = [ie.name] + p
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)
453
549
file_id = ie.parent_id
454
return joinpath(p)
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)
455
559
456
560
457
561
463
567
464
568
This returns the entry of the last component in the path,
465
569
which may be either a file or a directory.
570
571
Returns None iff the path is not found.
466
572
"""
467
assert isinstance(name, types.StringTypes)
468
469
parent_id = None
470
for f in splitpath(name):
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
580
try:
472
cie = self._tree[parent_id][f]
581
cie = parent.children[f]
473
582
assert cie.name == f
474
parent_id = cie.file_id
583
assert cie.parent_id == parent.file_id
584
parent = cie
475
585
except KeyError:
476
586
# or raise an error?
477
587
return None
478
588
479
return parent_id
480
481
482
def get_child(self, parent_id, child_name):
483
return self._tree[parent_id].get(child_name)
589
return parent.file_id
484
590
485
591
486
592
def has_filename(self, names):
488
594
489
595
490
596
def has_id(self, file_id):
491
assert isinstance(file_id, str)
492
597
return self._byid.has_key(file_id)
493
598
494
599
495
496
if __name__ == '__main__':
497
import doctest, inventory
498
doctest.testmod(inventory)
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))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1