« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 407
- compare with another revision
- download diff
- download tarball
- view history from revision 407
- Committer: Martin Pool
- Date: 2005-05-09 04:30:00 UTC
- Revision ID: mbp@sourcefrog.net-20050509043000-9429a30ae891433c
- slight improvements for zsh completion
- files added:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- elementtree
- urlgrabber
- files removed:
- doc/faq.txt
- files renamed:
- bzr.py => bzrlib/commands.py
- files modified:
- .bzrignore
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
# This should really be an id randomly assigned when the tree is
19
# created, but it's not for now.
20
ROOT_ID = "TREE_ROOT"
21
22
23
import sys, os.path, types, re
25
24
from sets import Set
26
25
27
26
try:
30
29
from elementtree.ElementTree import Element, ElementTree, SubElement
31
30
32
31
from xml import XMLMixin
33
from errors import bailout
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
32
from errors import bailout, BzrError, BzrCheckError
33
34
import bzrlib
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
36
from bzrlib.trace import mutter
36
37
37
38
class InventoryEntry(XMLMixin):
38
39
"""Description of a versioned file.
57
58
58
59
>>> i = Inventory()
59
60
>>> i.path2id('')
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
61
'TREE_ROOT'
62
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
63
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
62
64
>>> for j in i.iter_entries():
63
65
... print j
64
66
...
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
66
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
68
70
Traceback (most recent call last):
69
71
...
70
72
BzrError: ('inventory already contains entry with id {2323}', [])
71
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
72
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
73
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
73
75
>>> i.path2id('src/wibble')
74
76
'2325'
75
77
>>> '2325' in i
76
78
True
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
78
80
>>> i['2326']
79
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
80
82
>>> for j in i.iter_entries():
89
91
>>> i.id2path('2326')
90
92
'src/wibble/wibble.c'
91
93
92
:todo: Maybe also keep the full path of the entry, and the children?
94
TODO: Maybe also keep the full path of the entry, and the children?
93
95
But those depend on its position within a particular inventory, and
94
96
it would be nice not to need to hold the backpointer here.
95
97
"""
96
def __init__(self, file_id, name, kind='file', text_id=None,
97
parent_id=None):
98
99
# TODO: split InventoryEntry into subclasses for files,
100
# directories, etc etc.
101
102
text_sha1 = None
103
text_size = None
104
105
def __init__(self, file_id, name, kind, parent_id, text_id=None):
98
106
"""Create an InventoryEntry
99
107
100
108
The filename must be a single component, relative to the
101
109
parent directory; it cannot be a whole path or relative name.
102
110
103
>>> e = InventoryEntry('123', 'hello.c')
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
104
112
>>> e.name
105
113
'hello.c'
106
114
>>> e.file_id
107
115
'123'
108
>>> e = InventoryEntry('123', 'src/hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
109
117
Traceback (most recent call last):
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
111
119
"""
112
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
115
% name)
120
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
116
122
117
123
self.file_id = file_id
118
124
self.name = name
119
assert kind in ['file', 'directory']
120
125
self.kind = kind
121
126
self.text_id = text_id
122
127
self.parent_id = parent_id
123
self.text_sha1 = None
124
self.text_size = None
128
if kind == 'directory':
129
self.children = {}
130
elif kind == 'file':
131
pass
132
else:
133
raise BzrError("unhandled entry kind %r" % kind)
134
135
136
137
def sorted_children(self):
138
l = self.children.items()
139
l.sort()
140
return l
125
141
126
142
127
143
def copy(self):
128
144
other = InventoryEntry(self.file_id, self.name, self.kind,
129
self.text_id, self.parent_id)
145
self.parent_id, text_id=self.text_id)
130
146
other.text_sha1 = self.text_sha1
131
147
other.text_size = self.text_size
132
148
return other
149
165
e.set('file_id', self.file_id)
150
166
e.set('kind', self.kind)
151
167
152
if self.text_size is not None:
168
if self.text_size != None:
153
169
e.set('text_size', '%d' % self.text_size)
154
170
155
for f in ['text_id', 'text_sha1', 'parent_id']:
171
for f in ['text_id', 'text_sha1']:
156
172
v = getattr(self, f)
157
if v is not None:
173
if v != None:
158
174
e.set(f, v)
159
175
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)
182
160
183
e.tail = '\n'
161
184
162
185
return e
164
187
165
188
def from_element(cls, elt):
166
189
assert elt.tag == 'entry'
167
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
190
191
## original format inventories don't have a parent_id for
192
## nodes in the root directory, but it's cleaner to use one
193
## internally.
194
parent_id = elt.get('parent_id')
195
if parent_id == None:
196
parent_id = ROOT_ID
197
198
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
168
199
self.text_id = elt.get('text_id')
169
200
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
171
201
172
202
## mutter("read inventoryentry: %r" % (elt.attrib))
173
203
195
225
196
226
197
227
228
class RootEntry(InventoryEntry):
229
def __init__(self, file_id):
230
self.file_id = file_id
231
self.children = {}
232
self.kind = 'root_directory'
233
self.parent_id = None
234
self.name = ''
235
236
def __cmp__(self, other):
237
if self is other:
238
return 0
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)
243
244
245
198
246
class Inventory(XMLMixin):
199
247
"""Inventory of versioned files in a tree.
200
248
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.
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.
207
252
208
253
The inventory represents a typical unix file tree, with
209
254
directories containing files and subdirectories. We never store
213
258
returned quickly.
214
259
215
260
InventoryEntry objects must not be modified after they are
216
inserted.
261
inserted, other than through the Inventory API.
217
262
218
263
>>> inv = Inventory()
219
264
>>> inv.write_xml(sys.stdout)
220
265
<inventory>
221
266
</inventory>
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
267
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
223
268
>>> inv['123-123'].name
224
269
'hello.c'
225
>>> for file_id in inv: print file_id
226
...
227
123-123
228
270
229
271
May be treated as an iterator or set to look up file ids:
230
272
244
286
</inventory>
245
287
246
288
"""
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
252
## TODO: Make sure only canonical filenames are stored.
253
254
## TODO: Do something sensible about the possible collisions on
255
## case-losing filesystems. Perhaps we should just always forbid
256
## such collisions.
257
258
## _tree should probably just be stored as
259
## InventoryEntry._children on each directory.
260
261
289
def __init__(self):
262
290
"""Create or read an inventory.
263
291
264
292
If a working directory is specified, the inventory is read
265
293
from there. If the file is specified, read from that. If not,
266
294
the inventory is created empty.
295
296
The inventory is created with a default root directory, with
297
an id of None.
267
298
"""
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: {}}
299
self.root = RootEntry(ROOT_ID)
300
self._byid = {self.root.file_id: self.root}
273
301
274
302
275
303
def __iter__(self):
281
309
return len(self._byid)
282
310
283
311
284
def iter_entries(self, parent_id=None):
312
def iter_entries(self, from_dir=None):
285
313
"""Return (path, entry) pairs, in order by name."""
286
kids = self._tree[parent_id].items()
314
if from_dir == None:
315
assert self.root
316
from_dir = self.root
317
elif isinstance(from_dir, basestring):
318
from_dir = self._byid[from_dir]
319
320
kids = from_dir.children.items()
287
321
kids.sort()
288
322
for name, ie in kids:
289
323
yield name, ie
290
324
if ie.kind == 'directory':
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):
325
for cn, cie in self.iter_entries(from_dir=ie.file_id):
326
yield os.path.join(name, cn), cie
327
328
329
330
def directories(self):
296
331
"""Return (path, entry) pairs for all directories.
297
332
"""
298
if include_root:
299
yield '', None
300
for path, entry in self.iter_entries():
301
if entry.kind == 'directory':
302
yield path, entry
333
def descend(parent_ie):
334
parent_name = parent_ie.name
335
yield parent_name, parent_ie
336
337
# directory children in sorted order
338
dn = []
339
for ie in parent_ie.children.itervalues():
340
if ie.kind == 'directory':
341
dn.append((ie.name, ie))
342
dn.sort()
343
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
347
348
for name, ie in descend(self.root):
349
yield name, ie
303
350
304
351
305
352
306
def children(self, parent_id):
307
"""Return entries that are direct children of parent_id."""
308
return self._tree[parent_id]
309
310
311
312
# TODO: return all paths and entries
313
314
315
353
def __contains__(self, file_id):
316
354
"""True if this entry contains a file with given id.
317
355
318
356
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
357
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
320
358
>>> '123' in inv
321
359
True
322
360
>>> '456' in inv
329
367
"""Return the entry for given file_id.
330
368
331
369
>>> inv = Inventory()
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
370
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
333
371
>>> inv['123123'].name
334
372
'hello.c'
335
373
"""
336
return self._byid[file_id]
374
try:
375
return self._byid[file_id]
376
except KeyError:
377
if file_id == None:
378
raise BzrError("can't look up file_id None")
379
else:
380
raise BzrError("file_id {%s} not in inventory" % file_id)
381
382
383
def get_child(self, parent_id, filename):
384
return self[parent_id].children.get(filename)
337
385
338
386
339
387
def add(self, entry):
341
389
342
390
To add a file to a branch ready to be committed, use Branch.add,
343
391
which calls this."""
344
if entry.file_id in self:
392
if entry.file_id in self._byid:
345
393
bailout("inventory already contains entry with id {%s}" % entry.file_id)
346
394
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))
395
try:
396
parent = self._byid[entry.parent_id]
397
except KeyError:
398
bailout("parent_id {%s} not in inventory" % entry.parent_id)
399
400
if parent.children.has_key(entry.name):
401
bailout("%s is already versioned" %
402
appendpath(self.id2path(parent.file_id), entry.name))
355
403
356
404
self._byid[entry.file_id] = entry
357
self._tree[entry.parent_id][entry.name] = entry
358
359
if entry.kind == 'directory':
360
self._tree[entry.file_id] = {}
405
parent.children[entry.name] = entry
406
407
408
def add_path(self, relpath, kind, file_id=None):
409
"""Add entry from a path.
410
411
The immediate parent must already be versioned"""
412
parts = bzrlib.osutils.splitpath(relpath)
413
if len(parts) == 0:
414
bailout("cannot re-add root of inventory")
415
416
if file_id == None:
417
file_id = bzrlib.branch.gen_file_id(relpath)
418
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)
423
return self.add(ie)
361
424
362
425
363
426
def __delitem__(self, file_id):
364
427
"""Remove entry by id.
365
428
366
429
>>> inv = Inventory()
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
430
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
368
431
>>> '123' in inv
369
432
True
370
433
>>> del inv['123']
373
436
"""
374
437
ie = self[file_id]
375
438
376
assert self._tree[ie.parent_id][ie.name] == ie
439
assert self[ie.parent_id].children[ie.name] == ie
377
440
378
441
# TODO: Test deleting all children; maybe hoist to a separate
379
442
# deltree method?
380
443
if ie.kind == 'directory':
381
for cie in self._tree[file_id].values():
444
for cie in ie.children.values():
382
445
del self[cie.file_id]
383
del self._tree[file_id]
446
del ie.children
384
447
385
448
del self._byid[file_id]
386
del self._tree[ie.parent_id][ie.name]
449
del self[ie.parent_id].children[ie.name]
387
450
388
451
389
452
def id_set(self):
403
466
"""Construct from XML Element
404
467
405
468
>>> inv = Inventory()
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
469
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
407
470
>>> elt = inv.to_element()
408
471
>>> inv2 = Inventory.from_element(elt)
409
472
>>> inv2 == inv
425
488
>>> i2 = Inventory()
426
489
>>> i1 == i2
427
490
True
428
>>> i1.add(InventoryEntry('123', 'foo'))
491
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
429
492
>>> i1 == i2
430
493
False
431
>>> i2.add(InventoryEntry('123', 'foo'))
494
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
432
495
>>> i1 == i2
433
496
True
434
497
"""
448
511
return 0
449
512
450
513
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
514
def get_idpath(self, file_id):
515
"""Return a list of file_ids for the path to an entry.
516
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.
521
"""
453
522
p = []
454
523
while file_id != None:
455
ie = self[file_id]
456
p = [ie.name] + p
524
try:
525
ie = self._byid[file_id]
526
except KeyError:
527
bailout("file_id {%s} not found in inventory" % file_id)
528
p.insert(0, ie.file_id)
457
529
file_id = ie.parent_id
458
return joinpath(p)
530
return p
531
532
533
def id2path(self, file_id):
534
"""Return as a list the path to file_id."""
535
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)
459
539
460
540
461
541
467
547
468
548
This returns the entry of the last component in the path,
469
549
which may be either a file or a directory.
550
551
Returns None iff the path is not found.
470
552
"""
471
assert isinstance(name, types.StringTypes)
472
473
parent_id = None
474
for f in splitpath(name):
553
if isinstance(name, types.StringTypes):
554
name = splitpath(name)
555
556
mutter("lookup path %r" % name)
557
558
parent = self.root
559
for f in name:
475
560
try:
476
cie = self._tree[parent_id][f]
561
cie = parent.children[f]
477
562
assert cie.name == f
478
parent_id = cie.file_id
563
assert cie.parent_id == parent.file_id
564
parent = cie
479
565
except KeyError:
480
566
# or raise an error?
481
567
return None
482
568
483
return parent_id
484
485
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
569
return parent.file_id
488
570
489
571
490
572
def has_filename(self, names):
492
574
493
575
494
576
def has_id(self, file_id):
495
assert isinstance(file_id, str)
496
577
return self._byid.has_key(file_id)
497
578
498
579
499
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)
580
def rename(self, file_id, new_parent_id, new_name):
581
"""Move a file within the inventory.
582
583
This can change either the name, or the parent, or both.
584
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)
588
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)))
592
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)))
597
598
file_ie = self._byid[file_id]
599
old_parent = self._byid[file_ie.parent_id]
600
601
# TODO: Don't leave things messed up if this fails
602
603
del old_parent.children[file_ie.name]
604
new_parent.children[new_name] = file_ie
605
606
file_ie.name = new_name
607
file_ie.parent_id = new_parent_id
608
609
610
611
612
_NAME_RE = re.compile(r'^[^/\\]+$')
613
614
def is_valid_name(name):
615
return bool(_NAME_RE.match(name))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1