« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 521
- compare with another revision
- download diff
- download tarball
- view history from revision 521
- Committer: Martin Pool
- Date: 2005-05-17 07:01:47 UTC
- Revision ID: mbp@sourcefrog.net-20050517070147-c38da17418ea6711
- Add patch to give symlink support
- files added:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- patches
- urlgrabber
- files removed:
- bzrlib/tests.py
- files modified:
- .bzrignore
added
removed
bzrlib/inventory.py
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
16
16
17
"""Inventories map files to their name in a revision."""
18
19
# TODO: Maybe store inventory_id in the file? Not really needed.
20
21
__copyright__ = "Copyright (C) 2005 Canonical Ltd."
22
__author__ = "Martin Pool <mbp@canonical.com>"
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
23
22
24
23
import sys, os.path, types, re
25
24
from sets import Set
30
29
from elementtree.ElementTree import Element, ElementTree, SubElement
31
30
32
31
from xml import XMLMixin
33
from errors import bailout
32
from errors import bailout, BzrError, BzrCheckError
34
33
35
34
import bzrlib
36
35
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
59
58
60
59
>>> i = Inventory()
61
60
>>> i.path2id('')
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> 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'))
64
64
>>> for j in i.iter_entries():
65
65
... print j
66
66
...
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
68
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
70
70
Traceback (most recent call last):
71
71
...
72
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'))
73
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
75
75
>>> i.path2id('src/wibble')
76
76
'2325'
77
77
>>> '2325' in i
78
78
True
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
79
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
80
80
>>> i['2326']
81
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
82
82
>>> for j in i.iter_entries():
91
91
>>> i.id2path('2326')
92
92
'src/wibble/wibble.c'
93
93
94
: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?
95
95
But those depend on its position within a particular inventory, and
96
96
it would be nice not to need to hold the backpointer here.
97
97
"""
98
def __init__(self, file_id, name, kind='file', text_id=None,
99
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):
100
106
"""Create an InventoryEntry
101
107
102
108
The filename must be a single component, relative to the
103
109
parent directory; it cannot be a whole path or relative name.
104
110
105
>>> e = InventoryEntry('123', 'hello.c')
111
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
106
112
>>> e.name
107
113
'hello.c'
108
114
>>> e.file_id
109
115
'123'
110
>>> e = InventoryEntry('123', 'src/hello.c')
116
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
111
117
Traceback (most recent call last):
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
118
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
113
119
"""
114
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
117
% name)
120
if '/' in name or '\\' in name:
121
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
118
122
119
123
self.file_id = file_id
120
124
self.name = name
121
assert kind in ['file', 'directory']
122
125
self.kind = kind
123
126
self.text_id = text_id
124
127
self.parent_id = parent_id
125
self.text_sha1 = None
126
self.text_size = None
127
128
if kind == 'directory':
128
129
self.children = {}
130
elif kind == 'file':
131
pass
132
else:
133
raise BzrError("unhandled entry kind %r" % kind)
134
129
135
130
136
131
137
def sorted_children(self):
136
142
137
143
def copy(self):
138
144
other = InventoryEntry(self.file_id, self.name, self.kind,
139
self.text_id, self.parent_id)
145
self.parent_id, text_id=self.text_id)
140
146
other.text_sha1 = self.text_sha1
141
147
other.text_size = self.text_size
148
# note that children are *not* copied; they're pulled across when
149
# others are added
142
150
return other
143
151
144
152
159
167
e.set('file_id', self.file_id)
160
168
e.set('kind', self.kind)
161
169
162
if self.text_size is not None:
170
if self.text_size != None:
163
171
e.set('text_size', '%d' % self.text_size)
164
172
165
for f in ['text_id', 'text_sha1', 'parent_id']:
173
for f in ['text_id', 'text_sha1']:
166
174
v = getattr(self, f)
167
if v is not None:
175
if v != None:
168
176
e.set(f, v)
169
177
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)
184
170
185
e.tail = '\n'
171
186
172
187
return e
174
189
175
190
def from_element(cls, elt):
176
191
assert elt.tag == 'entry'
177
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
192
193
## original format inventories don't have a parent_id for
194
## nodes in the root directory, but it's cleaner to use one
195
## internally.
196
parent_id = elt.get('parent_id')
197
if parent_id == None:
198
parent_id = ROOT_ID
199
200
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
178
201
self.text_id = elt.get('text_id')
179
202
self.text_sha1 = elt.get('text_sha1')
180
self.parent_id = elt.get('parent_id')
181
203
182
204
## mutter("read inventoryentry: %r" % (elt.attrib))
183
205
226
248
class Inventory(XMLMixin):
227
249
"""Inventory of versioned files in a tree.
228
250
229
An Inventory acts like a set of InventoryEntry items. You can
230
also look files up by their file_id or name.
231
232
May be read from and written to a metadata file in a tree. To
233
manipulate the inventory (for example to add a file), it is read
234
in, modified, and then written back out.
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.
235
254
236
255
The inventory represents a typical unix file tree, with
237
256
directories containing files and subdirectories. We never store
247
266
>>> inv.write_xml(sys.stdout)
248
267
<inventory>
249
268
</inventory>
250
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
269
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
251
270
>>> inv['123-123'].name
252
271
'hello.c'
253
272
269
288
</inventory>
270
289
271
290
"""
272
273
## TODO: Make sure only canonical filenames are stored.
274
275
## TODO: Do something sensible about the possible collisions on
276
## case-losing filesystems. Perhaps we should just always forbid
277
## such collisions.
278
279
## TODO: No special cases for root, rather just give it a file id
280
## like everything else.
281
282
## TODO: Probably change XML serialization to use nesting
283
284
291
def __init__(self):
285
292
"""Create or read an inventory.
286
293
291
298
The inventory is created with a default root directory, with
292
299
an id of None.
293
300
"""
294
self.root = RootEntry(None)
295
self._byid = {None: self.root}
301
self.root = RootEntry(ROOT_ID)
302
self._byid = {self.root.file_id: self.root}
296
303
297
304
298
305
def __iter__(self):
318
325
yield name, ie
319
326
if ie.kind == 'directory':
320
327
for cn, cie in self.iter_entries(from_dir=ie.file_id):
321
yield '/'.join((name, cn)), cie
328
yield os.path.join(name, cn), cie
322
329
323
330
324
331
325
def directories(self, from_dir=None):
332
def directories(self):
326
333
"""Return (path, entry) pairs for all directories.
327
334
"""
328
335
def descend(parent_ie):
349
356
"""True if this entry contains a file with given id.
350
357
351
358
>>> inv = Inventory()
352
>>> inv.add(InventoryEntry('123', 'foo.c'))
359
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
353
360
>>> '123' in inv
354
361
True
355
362
>>> '456' in inv
362
369
"""Return the entry for given file_id.
363
370
364
371
>>> inv = Inventory()
365
>>> inv.add(InventoryEntry('123123', 'hello.c'))
372
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
366
373
>>> inv['123123'].name
367
374
'hello.c'
368
375
"""
369
return self._byid[file_id]
370
376
try:
377
return self._byid[file_id]
378
except KeyError:
379
if file_id == None:
380
raise BzrError("can't look up file_id None")
381
else:
382
raise BzrError("file_id {%s} not in inventory" % file_id)
383
384
385
def get_file_kind(self, file_id):
386
return self._byid[file_id].kind
371
387
372
388
def get_child(self, parent_id, filename):
373
389
return self[parent_id].children.get(filename)
384
400
try:
385
401
parent = self._byid[entry.parent_id]
386
402
except KeyError:
387
bailout("parent_id %r not in inventory" % entry.parent_id)
403
bailout("parent_id {%s} not in inventory" % entry.parent_id)
388
404
389
405
if parent.children.has_key(entry.name):
390
406
bailout("%s is already versioned" %
402
418
if len(parts) == 0:
403
419
bailout("cannot re-add root of inventory")
404
420
405
if file_id is None:
421
if file_id == None:
406
422
file_id = bzrlib.branch.gen_file_id(relpath)
407
423
408
424
parent_id = self.path2id(parts[:-1])
425
assert parent_id != None
409
426
ie = InventoryEntry(file_id, parts[-1],
410
427
kind=kind, parent_id=parent_id)
411
428
return self.add(ie)
415
432
"""Remove entry by id.
416
433
417
434
>>> inv = Inventory()
418
>>> inv.add(InventoryEntry('123', 'foo.c'))
435
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
419
436
>>> '123' in inv
420
437
True
421
438
>>> del inv['123']
454
471
"""Construct from XML Element
455
472
456
473
>>> inv = Inventory()
457
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
474
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
458
475
>>> elt = inv.to_element()
459
476
>>> inv2 = Inventory.from_element(elt)
460
477
>>> inv2 == inv
476
493
>>> i2 = Inventory()
477
494
>>> i1 == i2
478
495
True
479
>>> i1.add(InventoryEntry('123', 'foo'))
496
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
480
497
>>> i1 == i2
481
498
False
482
>>> i2.add(InventoryEntry('123', 'foo'))
499
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
483
500
>>> i1 == i2
484
501
True
485
502
"""
505
522
The list contains one element for each directory followed by
506
523
the id of the file itself. So the length of the returned list
507
524
is equal to the depth of the file in the tree, counting the
508
root directory as depth 0.
525
root directory as depth 1.
509
526
"""
510
527
p = []
511
528
while file_id != None:
512
ie = self._byid[file_id]
529
try:
530
ie = self._byid[file_id]
531
except KeyError:
532
bailout("file_id {%s} not found in inventory" % file_id)
513
533
p.insert(0, ie.file_id)
514
534
file_id = ie.parent_id
515
535
return p
517
537
518
538
def id2path(self, file_id):
519
539
"""Return as a list the path to file_id."""
520
p = []
521
while file_id != None:
522
ie = self._byid[file_id]
523
p.insert(0, ie.name)
524
file_id = ie.parent_id
525
return '/'.join(p)
540
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)
526
544
527
545
528
546
534
552
535
553
This returns the entry of the last component in the path,
536
554
which may be either a file or a directory.
555
556
Returns None iff the path is not found.
537
557
"""
538
558
if isinstance(name, types.StringTypes):
539
559
name = splitpath(name)
540
560
541
parent = self[None]
561
mutter("lookup path %r" % name)
562
563
parent = self.root
542
564
for f in name:
543
565
try:
544
566
cie = parent.children[f]
545
567
assert cie.name == f
568
assert cie.parent_id == parent.file_id
546
569
parent = cie
547
570
except KeyError:
548
571
# or raise an error?
595
618
596
619
def is_valid_name(name):
597
620
return bool(_NAME_RE.match(name))
598
599
600
601
if __name__ == '__main__':
602
import doctest, inventory
603
doctest.testmod(inventory)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1