« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 786
- compare with another revision
- download diff
- download tarball
- view history from revision 786
- Committer: Martin Pool
- Date: 2005-06-27 01:22:28 UTC
- Revision ID: mbp@sourcefrog.net-20050627012228-c8650e0499020c41
- fix missing import
- files added:
- .rsyncexclude
- bzrlib/selftest
- contrib
- contrib/bash
- contrib/zsh
- patches
- plugins
- plugins/changeset
- urlgrabber
- files removed:
- bzrlib/tests.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
25
from sets import Set
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
26
24
27
25
try:
28
26
from cElementTree import Element, ElementTree, SubElement
29
27
except ImportError:
30
28
from elementtree.ElementTree import Element, ElementTree, SubElement
31
29
32
from xml import XMLMixin
33
from errors import bailout
30
from bzrlib.xml import XMLMixin
31
from bzrlib.errors import BzrError, BzrCheckError
34
32
35
33
import bzrlib
36
34
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
59
57
60
58
>>> i = Inventory()
61
59
>>> i.path2id('')
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
60
'TREE_ROOT'
61
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
62
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
64
63
>>> for j in i.iter_entries():
65
64
... print j
66
65
...
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
66
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
67
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
68
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
69
Traceback (most recent call last):
71
70
...
72
BzrError: ('inventory already contains entry with id {2323}', [])
73
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
71
BzrError: inventory already contains entry with id {2323}
72
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
73
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
75
74
>>> i.path2id('src/wibble')
76
75
'2325'
77
76
>>> '2325' in i
78
77
True
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
78
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
80
79
>>> i['2326']
81
80
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
81
>>> for j in i.iter_entries():
91
90
>>> i.id2path('2326')
92
91
'src/wibble/wibble.c'
93
92
94
:todo: Maybe also keep the full path of the entry, and the children?
93
TODO: Maybe also keep the full path of the entry, and the children?
95
94
But those depend on its position within a particular inventory, and
96
95
it would be nice not to need to hold the backpointer here.
97
96
"""
98
def __init__(self, file_id, name, kind='file', text_id=None,
99
parent_id=None):
97
98
# TODO: split InventoryEntry into subclasses for files,
99
# directories, etc etc.
100
101
text_sha1 = None
102
text_size = None
103
104
def __init__(self, file_id, name, kind, parent_id, text_id=None):
100
105
"""Create an InventoryEntry
101
106
102
107
The filename must be a single component, relative to the
103
108
parent directory; it cannot be a whole path or relative name.
104
109
105
>>> e = InventoryEntry('123', 'hello.c')
110
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
106
111
>>> e.name
107
112
'hello.c'
108
113
>>> e.file_id
109
114
'123'
110
>>> e = InventoryEntry('123', 'src/hello.c')
115
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
116
Traceback (most recent call last):
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
117
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
113
118
"""
114
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
117
% name)
119
if '/' in name or '\\' in name:
120
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
118
121
119
122
self.file_id = file_id
120
123
self.name = name
121
assert kind in ['file', 'directory']
122
124
self.kind = kind
123
125
self.text_id = text_id
124
126
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
127
if kind == 'directory':
128
self.children = {}
129
elif kind == 'file':
130
pass
131
else:
132
raise BzrError("unhandled entry kind %r" % kind)
133
134
135
136
def sorted_children(self):
137
l = self.children.items()
138
l.sort()
139
return l
127
140
128
141
129
142
def copy(self):
130
143
other = InventoryEntry(self.file_id, self.name, self.kind,
131
self.text_id, self.parent_id)
144
self.parent_id, text_id=self.text_id)
132
145
other.text_sha1 = self.text_sha1
133
146
other.text_size = self.text_size
147
# note that children are *not* copied; they're pulled across when
148
# others are added
134
149
return other
135
150
136
151
151
166
e.set('file_id', self.file_id)
152
167
e.set('kind', self.kind)
153
168
154
if self.text_size is not None:
169
if self.text_size != None:
155
170
e.set('text_size', '%d' % self.text_size)
156
171
157
for f in ['text_id', 'text_sha1', 'parent_id']:
172
for f in ['text_id', 'text_sha1']:
158
173
v = getattr(self, f)
159
if v is not None:
174
if v != None:
160
175
e.set(f, v)
161
176
177
# to be conservative, we don't externalize the root pointers
178
# for now, leaving them as null in the xml form. in a future
179
# version it will be implied by nested elements.
180
if self.parent_id != ROOT_ID:
181
assert isinstance(self.parent_id, basestring)
182
e.set('parent_id', self.parent_id)
183
162
184
e.tail = '\n'
163
185
164
186
return e
166
188
167
189
def from_element(cls, elt):
168
190
assert elt.tag == 'entry'
169
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
191
192
## original format inventories don't have a parent_id for
193
## nodes in the root directory, but it's cleaner to use one
194
## internally.
195
parent_id = elt.get('parent_id')
196
if parent_id == None:
197
parent_id = ROOT_ID
198
199
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
170
200
self.text_id = elt.get('text_id')
171
201
self.text_sha1 = elt.get('text_sha1')
172
self.parent_id = elt.get('parent_id')
173
202
174
203
## mutter("read inventoryentry: %r" % (elt.attrib))
175
204
181
210
182
211
from_element = classmethod(from_element)
183
212
184
def __cmp__(self, other):
185
if self is other:
186
return 0
213
def __eq__(self, other):
187
214
if not isinstance(other, InventoryEntry):
188
215
return NotImplemented
189
216
190
return cmp(self.file_id, other.file_id) \
191
or cmp(self.name, other.name) \
192
or cmp(self.text_sha1, other.text_sha1) \
193
or cmp(self.text_size, other.text_size) \
194
or cmp(self.text_id, other.text_id) \
195
or cmp(self.parent_id, other.parent_id) \
196
or cmp(self.kind, other.kind)
217
return (self.file_id == other.file_id) \
218
and (self.name == other.name) \
219
and (self.text_sha1 == other.text_sha1) \
220
and (self.text_size == other.text_size) \
221
and (self.text_id == other.text_id) \
222
and (self.parent_id == other.parent_id) \
223
and (self.kind == other.kind)
224
225
226
def __ne__(self, other):
227
return not (self == other)
228
229
def __hash__(self):
230
raise ValueError('not hashable')
231
232
233
234
class RootEntry(InventoryEntry):
235
def __init__(self, file_id):
236
self.file_id = file_id
237
self.children = {}
238
self.kind = 'root_directory'
239
self.parent_id = None
240
self.name = ''
241
242
def __eq__(self, other):
243
if not isinstance(other, RootEntry):
244
return NotImplemented
245
246
return (self.file_id == other.file_id) \
247
and (self.children == other.children)
197
248
198
249
199
250
200
251
class Inventory(XMLMixin):
201
252
"""Inventory of versioned files in a tree.
202
253
203
An Inventory acts like a set of InventoryEntry items. You can
204
also look files up by their file_id or name.
205
206
May be read from and written to a metadata file in a tree. To
207
manipulate the inventory (for example to add a file), it is read
208
in, modified, and then written back out.
254
This describes which file_id is present at each point in the tree,
255
and possibly the SHA-1 or other information about the file.
256
Entries can be looked up either by path or by file_id.
209
257
210
258
The inventory represents a typical unix file tree, with
211
259
directories containing files and subdirectories. We never store
215
263
returned quickly.
216
264
217
265
InventoryEntry objects must not be modified after they are
218
inserted.
266
inserted, other than through the Inventory API.
219
267
220
268
>>> inv = Inventory()
221
269
>>> inv.write_xml(sys.stdout)
222
270
<inventory>
223
271
</inventory>
224
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
272
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
225
273
>>> inv['123-123'].name
226
274
'hello.c'
227
>>> for file_id in inv: print file_id
228
...
229
123-123
230
275
231
276
May be treated as an iterator or set to look up file ids:
232
277
246
291
</inventory>
247
292
248
293
"""
249
250
## TODO: Clear up handling of files in subdirectories; we probably
251
## do want to be able to just look them up by name but this
252
## probably means gradually walking down the path, looking up as we go.
253
254
## TODO: Make sure only canonical filenames are stored.
255
256
## TODO: Do something sensible about the possible collisions on
257
## case-losing filesystems. Perhaps we should just always forbid
258
## such collisions.
259
260
## _tree should probably just be stored as
261
## InventoryEntry._children on each directory.
262
263
294
def __init__(self):
264
295
"""Create or read an inventory.
265
296
266
297
If a working directory is specified, the inventory is read
267
298
from there. If the file is specified, read from that. If not,
268
299
the inventory is created empty.
300
301
The inventory is created with a default root directory, with
302
an id of None.
269
303
"""
270
self._byid = dict()
271
272
# _tree is indexed by parent_id; at each level a map from name
273
# to ie. The None entry is the root.
274
self._tree = {None: {}}
304
self.root = RootEntry(ROOT_ID)
305
self._byid = {self.root.file_id: self.root}
275
306
276
307
277
308
def __iter__(self):
283
314
return len(self._byid)
284
315
285
316
286
def iter_entries(self, parent_id=None):
317
def iter_entries(self, from_dir=None):
287
318
"""Return (path, entry) pairs, in order by name."""
288
kids = self._tree[parent_id].items()
319
if from_dir == None:
320
assert self.root
321
from_dir = self.root
322
elif isinstance(from_dir, basestring):
323
from_dir = self._byid[from_dir]
324
325
kids = from_dir.children.items()
289
326
kids.sort()
290
327
for name, ie in kids:
291
328
yield name, ie
292
329
if ie.kind == 'directory':
293
for cn, cie in self.iter_entries(parent_id=ie.file_id):
294
yield joinpath([name, cn]), cie
295
296
297
def directories(self, include_root=True):
298
"""Return (path, entry) pairs for all directories.
299
"""
300
if include_root:
301
yield '', None
302
for path, entry in self.iter_entries():
303
if entry.kind == 'directory':
304
yield path, entry
330
for cn, cie in self.iter_entries(from_dir=ie.file_id):
331
yield os.path.join(name, cn), cie
332
333
334
def entries(self):
335
"""Return list of (path, ie) for all entries except the root.
336
337
This may be faster than iter_entries.
338
"""
339
accum = []
340
def descend(dir_ie, dir_path):
341
kids = dir_ie.children.items()
342
kids.sort()
343
for name, ie in kids:
344
child_path = os.path.join(dir_path, name)
345
accum.append((child_path, ie))
346
if ie.kind == 'directory':
347
descend(ie, child_path)
348
349
descend(self.root, '')
350
return accum
351
352
353
def directories(self):
354
"""Return (path, entry) pairs for all directories, including the root.
355
"""
356
accum = []
357
def descend(parent_ie, parent_path):
358
accum.append((parent_path, parent_ie))
359
360
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
361
kids.sort()
362
363
for name, child_ie in kids:
364
child_path = os.path.join(parent_path, name)
365
descend(child_ie, child_path)
366
descend(self.root, '')
367
return accum
305
368
306
369
307
370
308
def children(self, parent_id):
309
"""Return entries that are direct children of parent_id."""
310
return self._tree[parent_id]
311
312
313
314
# TODO: return all paths and entries
315
316
317
371
def __contains__(self, file_id):
318
372
"""True if this entry contains a file with given id.
319
373
320
374
>>> inv = Inventory()
321
>>> inv.add(InventoryEntry('123', 'foo.c'))
375
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
322
376
>>> '123' in inv
323
377
True
324
378
>>> '456' in inv
331
385
"""Return the entry for given file_id.
332
386
333
387
>>> inv = Inventory()
334
>>> inv.add(InventoryEntry('123123', 'hello.c'))
388
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
335
389
>>> inv['123123'].name
336
390
'hello.c'
337
391
"""
338
return self._byid[file_id]
392
try:
393
return self._byid[file_id]
394
except KeyError:
395
if file_id == None:
396
raise BzrError("can't look up file_id None")
397
else:
398
raise BzrError("file_id {%s} not in inventory" % file_id)
399
400
401
def get_file_kind(self, file_id):
402
return self._byid[file_id].kind
403
404
def get_child(self, parent_id, filename):
405
return self[parent_id].children.get(filename)
339
406
340
407
341
408
def add(self, entry):
343
410
344
411
To add a file to a branch ready to be committed, use Branch.add,
345
412
which calls this."""
346
if entry.file_id in self:
347
bailout("inventory already contains entry with id {%s}" % entry.file_id)
348
349
if entry.parent_id != None:
350
if entry.parent_id not in self:
351
bailout("parent_id %s of new entry not found in inventory"
352
% entry.parent_id)
353
354
if self._tree[entry.parent_id].has_key(entry.name):
355
bailout("%s is already versioned"
356
% appendpath(self.id2path(entry.parent_id), entry.name))
413
if entry.file_id in self._byid:
414
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
415
416
try:
417
parent = self._byid[entry.parent_id]
418
except KeyError:
419
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
420
421
if parent.children.has_key(entry.name):
422
raise BzrError("%s is already versioned" %
423
appendpath(self.id2path(parent.file_id), entry.name))
357
424
358
425
self._byid[entry.file_id] = entry
359
self._tree[entry.parent_id][entry.name] = entry
360
361
if entry.kind == 'directory':
362
self._tree[entry.file_id] = {}
426
parent.children[entry.name] = entry
363
427
364
428
365
429
def add_path(self, relpath, kind, file_id=None):
366
430
"""Add entry from a path.
367
431
368
432
The immediate parent must already be versioned"""
433
from bzrlib.errors import NotVersionedError
434
369
435
parts = bzrlib.osutils.splitpath(relpath)
370
436
if len(parts) == 0:
371
bailout("cannot re-add root of inventory")
437
raise BzrError("cannot re-add root of inventory")
372
438
373
if file_id is None:
439
if file_id == None:
374
440
file_id = bzrlib.branch.gen_file_id(relpath)
375
441
376
parent_id = self.path2id(parts[:-1])
442
parent_path = parts[:-1]
443
parent_id = self.path2id(parent_path)
444
if parent_id == None:
445
raise NotVersionedError(parent_path)
446
377
447
ie = InventoryEntry(file_id, parts[-1],
378
448
kind=kind, parent_id=parent_id)
379
449
return self.add(ie)
383
453
"""Remove entry by id.
384
454
385
455
>>> inv = Inventory()
386
>>> inv.add(InventoryEntry('123', 'foo.c'))
456
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
387
457
>>> '123' in inv
388
458
True
389
459
>>> del inv['123']
392
462
"""
393
463
ie = self[file_id]
394
464
395
assert self._tree[ie.parent_id][ie.name] == ie
465
assert self[ie.parent_id].children[ie.name] == ie
396
466
397
467
# TODO: Test deleting all children; maybe hoist to a separate
398
468
# deltree method?
399
469
if ie.kind == 'directory':
400
for cie in self._tree[file_id].values():
470
for cie in ie.children.values():
401
471
del self[cie.file_id]
402
del self._tree[file_id]
472
del ie.children
403
473
404
474
del self._byid[file_id]
405
del self._tree[ie.parent_id][ie.name]
406
407
408
def id_set(self):
409
return Set(self._byid)
475
del self[ie.parent_id].children[ie.name]
410
476
411
477
412
478
def to_element(self):
422
488
"""Construct from XML Element
423
489
424
490
>>> inv = Inventory()
425
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
491
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
426
492
>>> elt = inv.to_element()
427
493
>>> inv2 = Inventory.from_element(elt)
428
494
>>> inv2 == inv
437
503
from_element = classmethod(from_element)
438
504
439
505
440
def __cmp__(self, other):
506
def __eq__(self, other):
441
507
"""Compare two sets by comparing their contents.
442
508
443
509
>>> i1 = Inventory()
444
510
>>> i2 = Inventory()
445
511
>>> i1 == i2
446
512
True
447
>>> i1.add(InventoryEntry('123', 'foo'))
513
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
448
514
>>> i1 == i2
449
515
False
450
>>> i2.add(InventoryEntry('123', 'foo'))
516
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
451
517
>>> i1 == i2
452
518
True
453
519
"""
454
if self is other:
455
return 0
456
457
520
if not isinstance(other, Inventory):
458
521
return NotImplemented
459
522
460
if self.id_set() ^ other.id_set():
461
return 1
462
463
for file_id in self._byid:
464
c = cmp(self[file_id], other[file_id])
465
if c: return c
466
467
return 0
468
469
470
def id2path(self, file_id):
471
"""Return as a list the path to file_id."""
523
if len(self._byid) != len(other._byid):
524
# shortcut: obviously not the same
525
return False
526
527
return self._byid == other._byid
528
529
530
def __ne__(self, other):
531
return not (self == other)
532
533
534
def __hash__(self):
535
raise ValueError('not hashable')
536
537
538
539
def get_idpath(self, file_id):
540
"""Return a list of file_ids for the path to an entry.
541
542
The list contains one element for each directory followed by
543
the id of the file itself. So the length of the returned list
544
is equal to the depth of the file in the tree, counting the
545
root directory as depth 1.
546
"""
472
547
p = []
473
548
while file_id != None:
474
ie = self[file_id]
475
p = [ie.name] + p
549
try:
550
ie = self._byid[file_id]
551
except KeyError:
552
raise BzrError("file_id {%s} not found in inventory" % file_id)
553
p.insert(0, ie.file_id)
476
554
file_id = ie.parent_id
477
return joinpath(p)
555
return p
556
557
558
def id2path(self, file_id):
559
"""Return as a list the path to file_id."""
560
561
# get all names, skipping root
562
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
563
return os.sep.join(p)
478
564
479
565
480
566
486
572
487
573
This returns the entry of the last component in the path,
488
574
which may be either a file or a directory.
575
576
Returns None iff the path is not found.
489
577
"""
490
578
if isinstance(name, types.StringTypes):
491
579
name = splitpath(name)
492
580
493
parent_id = None
581
mutter("lookup path %r" % name)
582
583
parent = self.root
494
584
for f in name:
495
585
try:
496
cie = self._tree[parent_id][f]
586
cie = parent.children[f]
497
587
assert cie.name == f
498
parent_id = cie.file_id
588
assert cie.parent_id == parent.file_id
589
parent = cie
499
590
except KeyError:
500
591
# or raise an error?
501
592
return None
502
593
503
return parent_id
504
505
506
def get_child(self, parent_id, child_name):
507
return self._tree[parent_id].get(child_name)
594
return parent.file_id
508
595
509
596
510
597
def has_filename(self, names):
512
599
513
600
514
601
def has_id(self, file_id):
515
assert isinstance(file_id, str)
516
602
return self._byid.has_key(file_id)
517
603
518
604
519
520
521
522
if __name__ == '__main__':
523
import doctest, inventory
524
doctest.testmod(inventory)
605
def rename(self, file_id, new_parent_id, new_name):
606
"""Move a file within the inventory.
607
608
This can change either the name, or the parent, or both.
609
610
This does not move the working file."""
611
if not is_valid_name(new_name):
612
raise BzrError("not an acceptable filename: %r" % new_name)
613
614
new_parent = self._byid[new_parent_id]
615
if new_name in new_parent.children:
616
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
617
618
new_parent_idpath = self.get_idpath(new_parent_id)
619
if file_id in new_parent_idpath:
620
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
621
% (self.id2path(file_id), self.id2path(new_parent_id)))
622
623
file_ie = self._byid[file_id]
624
old_parent = self._byid[file_ie.parent_id]
625
626
# TODO: Don't leave things messed up if this fails
627
628
del old_parent.children[file_ie.name]
629
new_parent.children[new_name] = file_ie
630
631
file_ie.name = new_name
632
file_ie.parent_id = new_parent_id
633
634
635
636
637
_NAME_RE = re.compile(r'^[^/\\]+$')
638
639
def is_valid_name(name):
640
return bool(_NAME_RE.match(name))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1