« back to all changes in this revision
Viewing changes to bzrlib/inventory.py
- browse files at revision 549
- compare with another revision
- download diff
- download tarball
- view history from revision 549
- Committer: Martin Pool
- Date: 2005-05-24 21:18:02 UTC
- Revision ID: mbp@sourcefrog.net-20050524211802-87b146b2405e92d2
doc
- files added:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- elementtree
- patches
- 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
30
28
from elementtree.ElementTree import Element, ElementTree, SubElement
31
29
32
30
from xml import XMLMixin
33
from errors import bailout
34
from osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from trace import mutter
31
from errors import bailout, BzrError, BzrCheckError
32
33
import bzrlib
34
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
35
from bzrlib.trace import mutter
36
36
37
37
class InventoryEntry(XMLMixin):
38
38
"""Description of a versioned file.
57
57
58
58
>>> i = Inventory()
59
59
>>> i.path2id('')
60
>>> i.add(InventoryEntry('123', 'src', kind='directory'))
61
>>> 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'))
62
63
>>> for j in i.iter_entries():
63
64
... print j
64
65
...
65
('src', InventoryEntry('123', 'src', kind='directory', parent_id=None))
66
('src', InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT'))
66
67
('src/hello.c', InventoryEntry('2323', 'hello.c', kind='file', parent_id='123'))
67
>>> i.add(InventoryEntry('2323', 'bye.c', parent_id='123'))
68
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
68
69
Traceback (most recent call last):
69
70
...
70
71
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'))
72
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
73
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
73
74
>>> i.path2id('src/wibble')
74
75
'2325'
75
76
>>> '2325' in i
76
77
True
77
>>> i.add(InventoryEntry('2326', 'wibble.c', parent_id='2325'))
78
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
78
79
>>> i['2326']
79
80
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
80
81
>>> for j in i.iter_entries():
89
90
>>> i.id2path('2326')
90
91
'src/wibble/wibble.c'
91
92
92
: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?
93
94
But those depend on its position within a particular inventory, and
94
95
it would be nice not to need to hold the backpointer here.
95
96
"""
96
def __init__(self, file_id, name, kind='file', text_id=None,
97
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):
98
105
"""Create an InventoryEntry
99
106
100
107
The filename must be a single component, relative to the
101
108
parent directory; it cannot be a whole path or relative name.
102
109
103
>>> e = InventoryEntry('123', 'hello.c')
110
>>> e = InventoryEntry('123', 'hello.c', 'file', ROOT_ID)
104
111
>>> e.name
105
112
'hello.c'
106
113
>>> e.file_id
107
114
'123'
108
>>> e = InventoryEntry('123', 'src/hello.c')
115
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
109
116
Traceback (most recent call last):
110
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
117
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
111
118
"""
112
113
if len(splitpath(name)) != 1:
114
bailout('InventoryEntry name is not a simple filename: %r'
115
% name)
119
if '/' in name or '\\' in name:
120
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
116
121
117
122
self.file_id = file_id
118
123
self.name = name
119
assert kind in ['file', 'directory']
120
124
self.kind = kind
121
125
self.text_id = text_id
122
126
self.parent_id = parent_id
123
self.text_sha1 = None
124
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
125
140
126
141
127
142
def copy(self):
128
143
other = InventoryEntry(self.file_id, self.name, self.kind,
129
self.text_id, self.parent_id)
144
self.parent_id, text_id=self.text_id)
130
145
other.text_sha1 = self.text_sha1
131
146
other.text_size = self.text_size
147
# note that children are *not* copied; they're pulled across when
148
# others are added
132
149
return other
133
150
134
151
149
166
e.set('file_id', self.file_id)
150
167
e.set('kind', self.kind)
151
168
152
if self.text_size is not None:
169
if self.text_size != None:
153
170
e.set('text_size', '%d' % self.text_size)
154
171
155
for f in ['text_id', 'text_sha1', 'parent_id']:
172
for f in ['text_id', 'text_sha1']:
156
173
v = getattr(self, f)
157
if v is not None:
174
if v != None:
158
175
e.set(f, v)
159
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
160
184
e.tail = '\n'
161
185
162
186
return e
164
188
165
189
def from_element(cls, elt):
166
190
assert elt.tag == 'entry'
167
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)
168
200
self.text_id = elt.get('text_id')
169
201
self.text_sha1 = elt.get('text_sha1')
170
self.parent_id = elt.get('parent_id')
171
202
172
203
## mutter("read inventoryentry: %r" % (elt.attrib))
173
204
179
210
180
211
from_element = classmethod(from_element)
181
212
182
def __cmp__(self, other):
183
if self is other:
184
return 0
213
def __eq__(self, other):
185
214
if not isinstance(other, InventoryEntry):
186
215
return NotImplemented
187
216
188
return cmp(self.file_id, other.file_id) \
189
or cmp(self.name, other.name) \
190
or cmp(self.text_sha1, other.text_sha1) \
191
or cmp(self.text_size, other.text_size) \
192
or cmp(self.text_id, other.text_id) \
193
or cmp(self.parent_id, other.parent_id) \
194
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)
195
248
196
249
197
250
198
251
class Inventory(XMLMixin):
199
252
"""Inventory of versioned files in a tree.
200
253
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.
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.
207
257
208
258
The inventory represents a typical unix file tree, with
209
259
directories containing files and subdirectories. We never store
213
263
returned quickly.
214
264
215
265
InventoryEntry objects must not be modified after they are
216
inserted.
266
inserted, other than through the Inventory API.
217
267
218
268
>>> inv = Inventory()
219
269
>>> inv.write_xml(sys.stdout)
220
270
<inventory>
221
271
</inventory>
222
>>> inv.add(InventoryEntry('123-123', 'hello.c'))
272
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
223
273
>>> inv['123-123'].name
224
274
'hello.c'
225
>>> for file_id in inv: print file_id
226
...
227
123-123
228
275
229
276
May be treated as an iterator or set to look up file ids:
230
277
244
291
</inventory>
245
292
246
293
"""
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
294
def __init__(self):
262
295
"""Create or read an inventory.
263
296
264
297
If a working directory is specified, the inventory is read
265
298
from there. If the file is specified, read from that. If not,
266
299
the inventory is created empty.
300
301
The inventory is created with a default root directory, with
302
an id of None.
267
303
"""
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: {}}
304
self.root = RootEntry(ROOT_ID)
305
self._byid = {self.root.file_id: self.root}
273
306
274
307
275
308
def __iter__(self):
281
314
return len(self._byid)
282
315
283
316
284
def iter_entries(self, parent_id=None):
317
def iter_entries(self, from_dir=None):
285
318
"""Return (path, entry) pairs, in order by name."""
286
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()
287
326
kids.sort()
288
327
for name, ie in kids:
289
328
yield name, ie
290
329
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):
330
for cn, cie in self.iter_entries(from_dir=ie.file_id):
331
yield os.path.join(name, cn), cie
332
333
334
335
def directories(self):
296
336
"""Return (path, entry) pairs for all directories.
297
337
"""
298
if include_root:
299
yield '', None
300
for path, entry in self.iter_entries():
301
if entry.kind == 'directory':
302
yield path, entry
338
def descend(parent_ie):
339
parent_name = parent_ie.name
340
yield parent_name, parent_ie
341
342
# directory children in sorted order
343
dn = []
344
for ie in parent_ie.children.itervalues():
345
if ie.kind == 'directory':
346
dn.append((ie.name, ie))
347
dn.sort()
348
349
for name, child_ie in dn:
350
for sub_name, sub_ie in descend(child_ie):
351
yield appendpath(parent_name, sub_name), sub_ie
352
353
for name, ie in descend(self.root):
354
yield name, ie
303
355
304
356
305
357
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
358
def __contains__(self, file_id):
316
359
"""True if this entry contains a file with given id.
317
360
318
361
>>> inv = Inventory()
319
>>> inv.add(InventoryEntry('123', 'foo.c'))
362
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
320
363
>>> '123' in inv
321
364
True
322
365
>>> '456' in inv
329
372
"""Return the entry for given file_id.
330
373
331
374
>>> inv = Inventory()
332
>>> inv.add(InventoryEntry('123123', 'hello.c'))
375
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
333
376
>>> inv['123123'].name
334
377
'hello.c'
335
378
"""
336
return self._byid[file_id]
379
try:
380
return self._byid[file_id]
381
except KeyError:
382
if file_id == None:
383
raise BzrError("can't look up file_id None")
384
else:
385
raise BzrError("file_id {%s} not in inventory" % file_id)
386
387
388
def get_file_kind(self, file_id):
389
return self._byid[file_id].kind
390
391
def get_child(self, parent_id, filename):
392
return self[parent_id].children.get(filename)
337
393
338
394
339
395
def add(self, entry):
341
397
342
398
To add a file to a branch ready to be committed, use Branch.add,
343
399
which calls this."""
344
if entry.file_id in self:
400
if entry.file_id in self._byid:
345
401
bailout("inventory already contains entry with id {%s}" % entry.file_id)
346
402
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))
403
try:
404
parent = self._byid[entry.parent_id]
405
except KeyError:
406
bailout("parent_id {%s} not in inventory" % entry.parent_id)
407
408
if parent.children.has_key(entry.name):
409
bailout("%s is already versioned" %
410
appendpath(self.id2path(parent.file_id), entry.name))
355
411
356
412
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] = {}
413
parent.children[entry.name] = entry
414
415
416
def add_path(self, relpath, kind, file_id=None):
417
"""Add entry from a path.
418
419
The immediate parent must already be versioned"""
420
parts = bzrlib.osutils.splitpath(relpath)
421
if len(parts) == 0:
422
bailout("cannot re-add root of inventory")
423
424
if file_id == None:
425
file_id = bzrlib.branch.gen_file_id(relpath)
426
427
parent_id = self.path2id(parts[:-1])
428
assert parent_id != None
429
ie = InventoryEntry(file_id, parts[-1],
430
kind=kind, parent_id=parent_id)
431
return self.add(ie)
361
432
362
433
363
434
def __delitem__(self, file_id):
364
435
"""Remove entry by id.
365
436
366
437
>>> inv = Inventory()
367
>>> inv.add(InventoryEntry('123', 'foo.c'))
438
>>> inv.add(InventoryEntry('123', 'foo.c', 'file', ROOT_ID))
368
439
>>> '123' in inv
369
440
True
370
441
>>> del inv['123']
373
444
"""
374
445
ie = self[file_id]
375
446
376
assert self._tree[ie.parent_id][ie.name] == ie
447
assert self[ie.parent_id].children[ie.name] == ie
377
448
378
449
# TODO: Test deleting all children; maybe hoist to a separate
379
450
# deltree method?
380
451
if ie.kind == 'directory':
381
for cie in self._tree[file_id].values():
452
for cie in ie.children.values():
382
453
del self[cie.file_id]
383
del self._tree[file_id]
454
del ie.children
384
455
385
456
del self._byid[file_id]
386
del self._tree[ie.parent_id][ie.name]
387
388
389
def id_set(self):
390
return Set(self._byid)
457
del self[ie.parent_id].children[ie.name]
391
458
392
459
393
460
def to_element(self):
403
470
"""Construct from XML Element
404
471
405
472
>>> inv = Inventory()
406
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c'))
473
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
407
474
>>> elt = inv.to_element()
408
475
>>> inv2 = Inventory.from_element(elt)
409
476
>>> inv2 == inv
418
485
from_element = classmethod(from_element)
419
486
420
487
421
def __cmp__(self, other):
488
def __eq__(self, other):
422
489
"""Compare two sets by comparing their contents.
423
490
424
491
>>> i1 = Inventory()
425
492
>>> i2 = Inventory()
426
493
>>> i1 == i2
427
494
True
428
>>> i1.add(InventoryEntry('123', 'foo'))
495
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
429
496
>>> i1 == i2
430
497
False
431
>>> i2.add(InventoryEntry('123', 'foo'))
498
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
432
499
>>> i1 == i2
433
500
True
434
501
"""
435
if self is other:
436
return 0
437
438
502
if not isinstance(other, Inventory):
439
503
return NotImplemented
440
504
441
if self.id_set() ^ other.id_set():
442
return 1
443
444
for file_id in self._byid:
445
c = cmp(self[file_id], other[file_id])
446
if c: return c
447
448
return 0
449
450
451
def id2path(self, file_id):
452
"""Return as a list the path to file_id."""
505
if len(self._byid) != len(other._byid):
506
# shortcut: obviously not the same
507
return False
508
509
return self._byid == other._byid
510
511
512
def __ne__(self, other):
513
return not (self == other)
514
515
516
def __hash__(self):
517
raise ValueError('not hashable')
518
519
520
521
def get_idpath(self, file_id):
522
"""Return a list of file_ids for the path to an entry.
523
524
The list contains one element for each directory followed by
525
the id of the file itself. So the length of the returned list
526
is equal to the depth of the file in the tree, counting the
527
root directory as depth 1.
528
"""
453
529
p = []
454
530
while file_id != None:
455
ie = self[file_id]
456
p = [ie.name] + p
531
try:
532
ie = self._byid[file_id]
533
except KeyError:
534
bailout("file_id {%s} not found in inventory" % file_id)
535
p.insert(0, ie.file_id)
457
536
file_id = ie.parent_id
458
return joinpath(p)
537
return p
538
539
540
def id2path(self, file_id):
541
"""Return as a list the path to file_id."""
542
543
# get all names, skipping root
544
p = [self[fid].name for fid in self.get_idpath(file_id)[1:]]
545
return os.sep.join(p)
459
546
460
547
461
548
467
554
468
555
This returns the entry of the last component in the path,
469
556
which may be either a file or a directory.
557
558
Returns None iff the path is not found.
470
559
"""
471
assert isinstance(name, types.StringTypes)
472
473
parent_id = None
474
for f in splitpath(name):
560
if isinstance(name, types.StringTypes):
561
name = splitpath(name)
562
563
mutter("lookup path %r" % name)
564
565
parent = self.root
566
for f in name:
475
567
try:
476
cie = self._tree[parent_id][f]
568
cie = parent.children[f]
477
569
assert cie.name == f
478
parent_id = cie.file_id
570
assert cie.parent_id == parent.file_id
571
parent = cie
479
572
except KeyError:
480
573
# or raise an error?
481
574
return None
482
575
483
return parent_id
484
485
486
def get_child(self, parent_id, child_name):
487
return self._tree[parent_id].get(child_name)
576
return parent.file_id
488
577
489
578
490
579
def has_filename(self, names):
492
581
493
582
494
583
def has_id(self, file_id):
495
assert isinstance(file_id, str)
496
584
return self._byid.has_key(file_id)
497
585
498
586
499
500
if __name__ == '__main__':
501
import doctest, inventory
502
doctest.testmod(inventory)
587
def rename(self, file_id, new_parent_id, new_name):
588
"""Move a file within the inventory.
589
590
This can change either the name, or the parent, or both.
591
592
This does not move the working file."""
593
if not is_valid_name(new_name):
594
bailout("not an acceptable filename: %r" % new_name)
595
596
new_parent = self._byid[new_parent_id]
597
if new_name in new_parent.children:
598
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
599
600
new_parent_idpath = self.get_idpath(new_parent_id)
601
if file_id in new_parent_idpath:
602
bailout("cannot move directory %r into a subdirectory of itself, %r"
603
% (self.id2path(file_id), self.id2path(new_parent_id)))
604
605
file_ie = self._byid[file_id]
606
old_parent = self._byid[file_ie.parent_id]
607
608
# TODO: Don't leave things messed up if this fails
609
610
del old_parent.children[file_ie.name]
611
new_parent.children[new_name] = file_ie
612
613
file_ie.name = new_name
614
file_ie.parent_id = new_parent_id
615
616
617
618
619
_NAME_RE = re.compile(r'^[^/\\]+$')
620
621
def is_valid_name(name):
622
return bool(_NAME_RE.match(name))
Loggerhead is a web-based interface for Breezy
Version: 2.0.1