« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 72
- compare with another revision
- download diff
- download tarball
- view history from revision 72
- Committer: mbp at sourcefrog
- Date: 2005-03-23 06:25:55 UTC
- Revision ID: mbp@sourcefrog.net-20050323062555-5489339018d0c043
- import a subset of elementtree for easier installation
- files added:
- bzrlib/tests.py
- files removed:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- patches
- urlgrabber
- 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
# 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
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
24
26
25
27
try:
26
28
from cElementTree import Element, ElementTree, SubElement
28
30
from elementtree.ElementTree import Element, ElementTree, SubElement
29
31
30
32
from xml import XMLMixin
31
from errors import bailout, BzrError, BzrCheckError
33
from errors import bailout
32
34
33
35
import bzrlib
34
36
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
57
59
58
60
>>> i = Inventory()
59
61
>>> i.path2id('')
60
'TREE_ROOT'
61
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
62
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
62
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
63
>>> i.add(InventoryEntry('2323', 'hello.c', parent_id='123'))
63
64
>>> for j in i.iter_entries():
64
65
... print j
65
66
...
66
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
67
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
67
68
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
68
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
69
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
69
70
Traceback (most recent call last):
70
71
...
71
72
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'))
73
>>> i.add(InventoryEntry('2324', 'bye.c', parent_id='123'))
74
>>> i.add(InventoryEntry('2325', 'wibble', parent_id='123', kind='directory'))
74
75
>>> i.path2id('src/wibble')
75
76
'2325'
76
77
>>> '2325' in i
77
78
True
78
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
79
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
79
80
>>> i['2326']
80
81
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
81
82
>>> for j in i.iter_entries():
90
91
>>> i.id2path('2326')
91
92
'src/wibble/wibble.c'
92
93
93
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?
94
95
But those depend on its position within a particular inventory, and
95
96
it would be nice not to need to hold the backpointer here.
96
97
"""
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):
98
def __init__(self, file_id, name, kind='file', text_id=None,
99
parent_id=None):
105
100
"""Create an InventoryEntry
106
101
107
102
The filename must be a single component, relative to the
108
103
parent directory; it cannot be a whole path or relative name.
109
104
110
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
105
>>> e = InventoryEntry('123', 'hello.c')
111
106
>>> e.name
112
107
'hello.c'
113
108
>>> e.file_id
114
109
'123'
115
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
110
>>> e = InventoryEntry('123', 'src/hello.c')
116
111
Traceback (most recent call last):
117
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
112
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
118
113
"""
119
if '/' in name or '\\' in name:
120
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
114
115
if len(splitpath(name)) != 1:
116
bailout('InventoryEntry name is not a simple filename: %r'
117
% name)
121
118
122
119
self.file_id = file_id
123
120
self.name = name
121
assert kind in ['file', 'directory']
124
122
self.kind = kind
125
123
self.text_id = text_id
126
124
self.parent_id = parent_id
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
125
self.text_sha1 = None
126
self.text_size = None
140
127
141
128
142
129
def copy(self):
143
130
other = InventoryEntry(self.file_id, self.name, self.kind,
144
self.parent_id, text_id=self.text_id)
131
self.text_id, self.parent_id)
145
132
other.text_sha1 = self.text_sha1
146
133
other.text_size = self.text_size
147
# note that children are *not* copied; they're pulled across when
148
# others are added
149
134
return other
150
135
151
136
166
151
e.set('file_id', self.file_id)
167
152
e.set('kind', self.kind)
168
153
169
if self.text_size != None:
154
if self.text_size is not None:
170
155
e.set('text_size', '%d' % self.text_size)
171
156
172
for f in ['text_id', 'text_sha1']:
157
for f in ['text_id', 'text_sha1', 'parent_id']:
173
158
v = getattr(self, f)
174
if v != None:
159
if v is not None:
175
160
e.set(f, v)
176
161
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
184
162
e.tail = '\n'
185
163
186
164
return e
188
166
189
167
def from_element(cls, elt):
190
168
assert elt.tag == 'entry'
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)
169
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'))
200
170
self.text_id = elt.get('text_id')
201
171
self.text_sha1 = elt.get('text_sha1')
172
self.parent_id = elt.get('parent_id')
202
173
203
174
## mutter("read inventoryentry: %r" % (elt.attrib))
204
175
210
181
211
182
from_element = classmethod(from_element)
212
183
213
def __eq__(self, other):
184
def __cmp__(self, other):
185
if self is other:
186
return 0
214
187
if not isinstance(other, InventoryEntry):
215
188
return NotImplemented
216
189
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)
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)
248
197
249
198
250
199
251
200
class Inventory(XMLMixin):
252
201
"""Inventory of versioned files in a tree.
253
202
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.
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.
257
209
258
210
The inventory represents a typical unix file tree, with
259
211
directories containing files and subdirectories. We never store
263
215
returned quickly.
264
216
265
217
InventoryEntry objects must not be modified after they are
266
inserted, other than through the Inventory API.
218
inserted.
267
219
268
220
>>> inv = Inventory()
269
221
>>> inv.write_xml(sys.stdout)
270
222
<inventory>
271
223
</inventory>
272
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
224
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
273
225
>>> inv['123-123'].name
274
226
'hello.c'
227
>>> for file_id in inv: print file_id
228
...
229
123-123
275
230
276
231
May be treated as an iterator or set to look up file ids:
277
232
291
246
</inventory>
292
247
293
248
"""
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
294
263
def __init__(self):
295
264
"""Create or read an inventory.
296
265
297
266
If a working directory is specified, the inventory is read
298
267
from there. If the file is specified, read from that. If not,
299
268
the inventory is created empty.
300
301
The inventory is created with a default root directory, with
302
an id of None.
303
269
"""
304
self.root = RootEntry(ROOT_ID)
305
self._byid = {self.root.file_id: self.root}
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: {}}
306
275
307
276
308
277
def __iter__(self):
314
283
return len(self._byid)
315
284
316
285
317
def iter_entries(self, from_dir=None):
286
def iter_entries(self, parent_id=None):
318
287
"""Return (path, entry) pairs, in order by name."""
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()
288
kids = self._tree[parent_id].items()
326
289
kids.sort()
327
290
for name, ie in kids:
328
291
yield name, ie
329
292
if ie.kind == 'directory':
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
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
368
305
369
306
370
307
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
371
317
def __contains__(self, file_id):
372
318
"""True if this entry contains a file with given id.
373
319
374
320
>>> inv = Inventory()
375
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
321
>>> inv.add(InventoryEntry('123', 'foo.c'))
376
322
>>> '123' in inv
377
323
True
378
324
>>> '456' in inv
385
331
"""Return the entry for given file_id.
386
332
387
333
>>> inv = Inventory()
388
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
334
>>> inv.add(InventoryEntry('123123', 'hello.c'))
389
335
>>> inv['123123'].name
390
336
'hello.c'
391
337
"""
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)
338
return self._byid[file_id]
406
339
407
340
408
341
def add(self, entry):
410
343
411
344
To add a file to a branch ready to be committed, use Branch.add,
412
345
which calls this."""
413
if entry.file_id in self._byid:
346
if entry.file_id in self:
414
347
bailout("inventory already contains entry with id {%s}" % entry.file_id)
415
348
416
try:
417
parent = self._byid[entry.parent_id]
418
except KeyError:
419
bailout("parent_id {%s} not in inventory" % entry.parent_id)
420
421
if parent.children.has_key(entry.name):
422
bailout("%s is already versioned" %
423
appendpath(self.id2path(parent.file_id), entry.name))
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))
424
357
425
358
self._byid[entry.file_id] = entry
426
parent.children[entry.name] = entry
359
self._tree[entry.parent_id][entry.name] = entry
360
361
if entry.kind == 'directory':
362
self._tree[entry.file_id] = {}
427
363
428
364
429
365
def add_path(self, relpath, kind, file_id=None):
434
370
if len(parts) == 0:
435
371
bailout("cannot re-add root of inventory")
436
372
437
if file_id == None:
373
if file_id is None:
438
374
file_id = bzrlib.branch.gen_file_id(relpath)
439
375
440
376
parent_id = self.path2id(parts[:-1])
441
assert parent_id != None
442
377
ie = InventoryEntry(file_id, parts[-1],
443
378
kind=kind, parent_id=parent_id)
444
379
return self.add(ie)
448
383
"""Remove entry by id.
449
384
450
385
>>> inv = Inventory()
451
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
386
>>> inv.add(InventoryEntry('123', 'foo.c'))
452
387
>>> '123' in inv
453
388
True
454
389
>>> del inv['123']
457
392
"""
458
393
ie = self[file_id]
459
394
460
assert self[ie.parent_id].children[ie.name] == ie
395
assert self._tree[ie.parent_id][ie.name] == ie
461
396
462
397
# TODO: Test deleting all children; maybe hoist to a separate
463
398
# deltree method?
464
399
if ie.kind == 'directory':
465
for cie in ie.children.values():
400
for cie in self._tree[file_id].values():
466
401
del self[cie.file_id]
467
del ie.children
402
del self._tree[file_id]
468
403
469
404
del self._byid[file_id]
470
del self[ie.parent_id].children[ie.name]
405
del self._tree[ie.parent_id][ie.name]
406
407
408
def id_set(self):
409
return Set(self._byid)
471
410
472
411
473
412
def to_element(self):
483
422
"""Construct from XML Element
484
423
485
424
>>> inv = Inventory()
486
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
425
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
487
426
>>> elt = inv.to_element()
488
427
>>> inv2 = Inventory.from_element(elt)
489
428
>>> inv2 == inv
498
437
from_element = classmethod(from_element)
499
438
500
439
501
def __eq__(self, other):
440
def __cmp__(self, other):
502
441
"""Compare two sets by comparing their contents.
503
442
504
443
>>> i1 = Inventory()
505
444
>>> i2 = Inventory()
506
445
>>> i1 == i2
507
446
True
508
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
447
>>> i1.add(InventoryEntry('123', 'foo'))
509
448
>>> i1 == i2
510
449
False
511
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
450
>>> i2.add(InventoryEntry('123', 'foo'))
512
451
>>> i1 == i2
513
452
True
514
453
"""
454
if self is other:
455
return 0
456
515
457
if not isinstance(other, Inventory):
516
458
return NotImplemented
517
459
518
if len(self._byid) != len(other._byid):
519
# shortcut: obviously not the same
520
return False
521
522
return self._byid == other._byid
523
524
525
def __ne__(self, other):
526
return not (self == other)
527
528
529
def __hash__(self):
530
raise ValueError('not hashable')
531
532
533
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
"""
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."""
542
472
p = []
543
473
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)
474
ie = self[file_id]
475
p = [ie.name] + p
549
476
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)
477
return joinpath(p)
559
478
560
479
561
480
567
486
568
487
This returns the entry of the last component in the path,
569
488
which may be either a file or a directory.
570
571
Returns None iff the path is not found.
572
489
"""
573
490
if isinstance(name, types.StringTypes):
574
491
name = splitpath(name)
575
492
576
mutter("lookup path %r" % name)
577
578
parent = self.root
493
parent_id = None
579
494
for f in name:
580
495
try:
581
cie = parent.children[f]
496
cie = self._tree[parent_id][f]
582
497
assert cie.name == f
583
assert cie.parent_id == parent.file_id
584
parent = cie
498
parent_id = cie.file_id
585
499
except KeyError:
586
500
# or raise an error?
587
501
return None
588
502
589
return parent.file_id
503
return parent_id
504
505
506
def get_child(self, parent_id, child_name):
507
return self._tree[parent_id].get(child_name)
590
508
591
509
592
510
def has_filename(self, names):
594
512
595
513
596
514
def has_id(self, file_id):
515
assert isinstance(file_id, str)
597
516
return self._byid.has_key(file_id)
598
517
599
518
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))
519
520
521
522
if __name__ == '__main__':
523
import doctest, inventory
524
doctest.testmod(inventory)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1