15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
# TODO: Maybe store inventory_id in the file? Not really needed.
18
# TODO: Maybe also keep the full path of the entry, and the children?
19
# But those depend on its position within a particular inventory, and
20
# it would be nice not to need to hold the backpointer here.
22
# TODO: Perhaps split InventoryEntry into subclasses for files,
23
# directories, etc etc.
21
26
# This should really be an id randomly assigned when the tree is
23
28
ROOT_ID = "TREE_ROOT"
26
import sys, os.path, types, re
30
from cElementTree import Element, ElementTree, SubElement
32
from elementtree.ElementTree import Element, ElementTree, SubElement
34
from xml import XMLMixin
35
from errors import bailout, BzrError
38
from bzrlib.osutils import uuid, quotefn, splitpath, joinpath, appendpath
37
from bzrlib.errors import BzrError, BzrCheckError
39
from bzrlib.osutils import quotefn, splitpath, joinpath, appendpath, sha_strings
39
40
from bzrlib.trace import mutter
41
class InventoryEntry(XMLMixin):
41
from bzrlib.errors import NotVersionedError
44
class InventoryEntry(object):
42
45
"""Description of a versioned file.
44
47
An InventoryEntry has the following fields, which are also
45
48
present in the XML inventory-entry element:
48
* *name*: (only the basename within the directory, must not
50
* *kind*: "directory" or "file"
51
* *directory_id*: (if absent/null means the branch root directory)
52
* *text_sha1*: only for files
53
* *text_size*: in bytes, only for files
54
* *text_id*: identifier for the text version, only for files
56
InventoryEntries can also exist inside a WorkingTree
57
inventory, in which case they are not yet bound to a
58
particular revision of the file. In that case the text_sha1,
59
text_size and text_id are absent.
53
(within the parent directory)
59
file_id of the parent directory, or ROOT_ID
62
the revision_id in which this variationo f this file was
66
sha-1 of the text of the file
69
size in bytes of the text of the file
71
(reading a version 4 tree created a text_id field.)
62
73
>>> i = Inventory()
65
76
>>> i.add(InventoryEntry('123', 'src', 'directory', ROOT_ID))
77
InventoryEntry('123', 'src', kind='directory', parent_id='TREE_ROOT')
66
78
>>> i.add(InventoryEntry('2323', 'hello.c', 'file', parent_id='123'))
79
InventoryEntry('2323', 'hello.c', kind='file', parent_id='123')
67
80
>>> for j in i.iter_entries():
72
85
>>> i.add(InventoryEntry('2323', 'bye.c', 'file', '123'))
73
86
Traceback (most recent call last):
75
BzrError: ('inventory already contains entry with id {2323}', [])
88
BzrError: inventory already contains entry with id {2323}
76
89
>>> i.add(InventoryEntry('2324', 'bye.c', 'file', '123'))
90
InventoryEntry('2324', 'bye.c', kind='file', parent_id='123')
77
91
>>> i.add(InventoryEntry('2325', 'wibble', 'directory', '123'))
92
InventoryEntry('2325', 'wibble', kind='directory', parent_id='123')
78
93
>>> i.path2id('src/wibble')
82
97
>>> i.add(InventoryEntry('2326', 'wibble.c', 'file', '2325'))
98
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
84
100
InventoryEntry('2326', 'wibble.c', kind='file', parent_id='2325')
85
>>> for j in i.iter_entries():
87
... assert i.path2id(j[0])
101
>>> for path, entry in i.iter_entries():
102
... print path.replace('\\\\', '/') # for win32 os.sep
103
... assert i.path2id(path)
93
109
src/wibble/wibble.c
110
>>> i.id2path('2326').replace('\\\\', '/')
95
111
'src/wibble/wibble.c'
97
TODO: Maybe also keep the full path of the entry, and the children?
98
But those depend on its position within a particular inventory, and
99
it would be nice not to need to hold the backpointer here.
102
# TODO: split InventoryEntry into subclasses for files,
103
# directories, etc etc.
114
__slots__ = ['text_sha1', 'text_size', 'file_id', 'name', 'kind',
115
'text_id', 'parent_id', 'children',
116
'revision', 'symlink_target']
105
118
def __init__(self, file_id, name, kind, parent_id, text_id=None):
106
119
"""Create an InventoryEntry
116
129
>>> e = InventoryEntry('123', 'src/hello.c', 'file', ROOT_ID)
117
130
Traceback (most recent call last):
118
BzrError: ("InventoryEntry name is not a simple filename: 'src/hello.c'", [])
131
BzrCheckError: InventoryEntry name 'src/hello.c' is invalid
121
if len(splitpath(name)) != 1:
122
bailout('InventoryEntry name is not a simple filename: %r'
133
assert isinstance(name, basestring), name
134
if '/' in name or '\\' in name:
135
raise BzrCheckError('InventoryEntry name %r is invalid' % name)
138
self.text_sha1 = None
139
self.text_size = None
125
140
self.file_id = file_id
128
143
self.text_id = text_id
129
144
self.parent_id = parent_id
130
self.text_sha1 = None
131
self.text_size = None
145
self.symlink_target = None
132
146
if kind == 'directory':
133
147
self.children = {}
134
148
elif kind == 'file':
150
elif kind == 'symlink':
137
153
raise BzrError("unhandled entry kind %r" % kind)
155
def read_symlink_target(self, path):
156
if self.kind == 'symlink':
158
self.symlink_target = os.readlink(path)
160
raise BzrError("os.readlink error, %s" % e)
141
162
def sorted_children(self):
142
163
l = self.children.items()
167
def check(self, checker, rev_id, inv, tree):
168
if self.parent_id != None:
169
if not inv.has_id(self.parent_id):
170
raise BzrCheckError('missing parent {%s} in inventory for revision {%s}'
171
% (self.parent_id, rev_id))
172
if self.kind == 'file':
173
revision = self.revision
174
t = (self.file_id, revision)
175
if t in checker.checked_texts:
176
prev_sha = checker.checked_texts[t]
177
if prev_sha != self.text_sha1:
178
raise BzrCheckError('mismatched sha1 on {%s} in {%s}' %
179
(self.file_id, rev_id))
181
checker.repeated_text_cnt += 1
183
mutter('check version {%s} of {%s}', rev_id, self.file_id)
184
file_lines = tree.get_file_lines(self.file_id)
185
checker.checked_text_cnt += 1
186
if self.text_size != sum(map(len, file_lines)):
187
raise BzrCheckError('text {%s} wrong size' % self.text_id)
188
if self.text_sha1 != sha_strings(file_lines):
189
raise BzrCheckError('text {%s} wrong sha1' % self.text_id)
190
checker.checked_texts[t] = self.text_sha1
191
elif self.kind == 'directory':
192
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
193
raise BzrCheckError('directory {%s} has text in revision {%s}'
194
% (self.file_id, rev_id))
195
elif self.kind == 'root_directory':
197
elif self.kind == 'symlink':
198
if self.text_sha1 != None or self.text_size != None or self.text_id != None:
199
raise BzrCheckError('symlink {%s} has text in revision {%s}'
200
% (self.file_id, rev_id))
201
if self.symlink_target == None:
202
raise BzrCheckError('symlink {%s} has no target in revision {%s}'
203
% (self.file_id, rev_id))
205
raise BzrCheckError('unknown entry kind %r in revision {%s}' %
148
210
other = InventoryEntry(self.file_id, self.name, self.kind,
149
self.parent_id, text_id=self.text_id)
212
other.text_id = self.text_id
150
213
other.text_sha1 = self.text_sha1
151
214
other.text_size = self.text_size
215
other.symlink_target = self.symlink_target
216
other.revision = self.revision
217
# note that children are *not* copied; they're pulled across when
164
def to_element(self):
165
"""Convert to XML element"""
168
e.set('name', self.name)
169
e.set('file_id', self.file_id)
170
e.set('kind', self.kind)
172
if self.text_size != None:
173
e.set('text_size', '%d' % self.text_size)
175
for f in ['text_id', 'text_sha1']:
180
# to be conservative, we don't externalize the root pointers
181
# for now, leaving them as null in the xml form. in a future
182
# version it will be implied by nested elements.
183
if self.parent_id != ROOT_ID:
184
assert isinstance(self.parent_id, basestring)
185
e.set('parent_id', self.parent_id)
192
def from_element(cls, elt):
193
assert elt.tag == 'entry'
195
## original format inventories don't have a parent_id for
196
## nodes in the root directory, but it's cleaner to use one
198
parent_id = elt.get('parent_id')
199
if parent_id == None:
202
self = cls(elt.get('file_id'), elt.get('name'), elt.get('kind'), parent_id)
203
self.text_id = elt.get('text_id')
204
self.text_sha1 = elt.get('text_sha1')
206
## mutter("read inventoryentry: %r" % (elt.attrib))
208
v = elt.get('text_size')
209
self.text_size = v and int(v)
214
from_element = classmethod(from_element)
216
def __cmp__(self, other):
230
def __eq__(self, other):
219
231
if not isinstance(other, InventoryEntry):
220
232
return NotImplemented
222
return cmp(self.file_id, other.file_id) \
223
or cmp(self.name, other.name) \
224
or cmp(self.text_sha1, other.text_sha1) \
225
or cmp(self.text_size, other.text_size) \
226
or cmp(self.text_id, other.text_id) \
227
or cmp(self.parent_id, other.parent_id) \
228
or cmp(self.kind, other.kind)
234
return (self.file_id == other.file_id) \
235
and (self.name == other.name) \
236
and (other.symlink_target == self.symlink_target) \
237
and (self.text_sha1 == other.text_sha1) \
238
and (self.text_size == other.text_size) \
239
and (self.text_id == other.text_id) \
240
and (self.parent_id == other.parent_id) \
241
and (self.kind == other.kind) \
242
and (self.revision == other.revision)
244
def __ne__(self, other):
245
return not (self == other)
248
raise ValueError('not hashable')
250
def unchanged(self, previous_ie, work_tree):
252
# different inv parent
253
if previous_ie.parent_id != self.parent_id:
256
elif previous_ie.name != self.name:
258
# changed link target
259
elif (hasattr(self,'symlink_target')
260
and self.symlink_target != previous_ie.symlink_target):
262
if self.kind != 'file':
264
self.text_sha1 = work_tree.get_file_sha1(self.file_id)
265
if self.text_sha1 != previous_ie.text_sha1:
268
# FIXME: 20050930 probe for the text size when getting sha1
269
self.text_size = previous_ie.text_size
232
273
class RootEntry(InventoryEntry):
237
278
self.parent_id = None
240
def __cmp__(self, other):
281
def __eq__(self, other):
243
282
if not isinstance(other, RootEntry):
244
283
return NotImplemented
245
return cmp(self.file_id, other.file_id) \
246
or cmp(self.children, other.children)
250
class Inventory(XMLMixin):
285
return (self.file_id == other.file_id) \
286
and (self.children == other.children)
290
class Inventory(object):
251
291
"""Inventory of versioned files in a tree.
253
293
This describes which file_id is present at each point in the tree,
284
322
>>> [x[0] for x in inv.iter_entries()]
287
>>> inv.write_xml(sys.stdout)
289
<entry file_id="123-123" kind="file" name="hello.c" />
324
>>> inv = Inventory('TREE_ROOT-12345678-12345678')
325
>>> inv.add(InventoryEntry('123-123', 'hello.c', 'file', ROOT_ID))
326
InventoryEntry('123-123', 'hello.c', kind='file', parent_id='TREE_ROOT-12345678-12345678')
294
## TODO: Make sure only canonical filenames are stored.
296
## TODO: Do something sensible about the possible collisions on
297
## case-losing filesystems. Perhaps we should just always forbid
300
## TODO: No special cases for root, rather just give it a file id
301
## like everything else.
303
## TODO: Probably change XML serialization to use nesting rather
304
## than parent_id pointers.
306
## TODO: Perhaps hold the ElementTree in memory and work directly
307
## on that rather than converting into Python objects every time?
328
def __init__(self, root_id=ROOT_ID):
310
329
"""Create or read an inventory.
312
331
If a working directory is specified, the inventory is read
316
335
The inventory is created with a default root directory, with
319
self.root = RootEntry(ROOT_ID)
338
# We are letting Branch.initialize() create a unique inventory
339
# root id. Rather than generating a random one here.
341
# root_id = bzrlib.branch.gen_file_id('TREE_ROOT')
342
self.root = RootEntry(root_id)
320
343
self._byid = {self.root.file_id: self.root}
347
other = Inventory(self.root.file_id)
348
# copy recursively so we know directories will be added before
349
# their children. There are more efficient ways than this...
350
for path, entry in self.iter_entries():
351
if entry == self.root:
353
other.add(entry.copy())
323
357
def __iter__(self):
324
358
return iter(self._byid)
344
378
if ie.kind == 'directory':
345
379
for cn, cie in self.iter_entries(from_dir=ie.file_id):
346
380
yield os.path.join(name, cn), cie
384
"""Return list of (path, ie) for all entries except the root.
386
This may be faster than iter_entries.
389
def descend(dir_ie, dir_path):
390
kids = dir_ie.children.items()
392
for name, ie in kids:
393
child_path = os.path.join(dir_path, name)
394
accum.append((child_path, ie))
395
if ie.kind == 'directory':
396
descend(ie, child_path)
398
descend(self.root, '')
350
402
def directories(self):
351
"""Return (path, entry) pairs for all directories.
403
"""Return (path, entry) pairs for all directories, including the root.
353
def descend(parent_ie):
354
parent_name = parent_ie.name
355
yield parent_name, parent_ie
357
# directory children in sorted order
359
for ie in parent_ie.children.itervalues():
360
if ie.kind == 'directory':
361
dn.append((ie.name, ie))
406
def descend(parent_ie, parent_path):
407
accum.append((parent_path, parent_ie))
364
for name, child_ie in dn:
365
for sub_name, sub_ie in descend(child_ie):
366
yield appendpath(parent_name, sub_name), sub_ie
409
kids = [(ie.name, ie) for ie in parent_ie.children.itervalues() if ie.kind == 'directory']
368
for name, ie in descend(self.root):
412
for name, child_ie in kids:
413
child_path = os.path.join(parent_path, name)
414
descend(child_ie, child_path)
415
descend(self.root, '')
389
437
>>> inv = Inventory()
390
438
>>> inv.add(InventoryEntry('123123', 'hello.c', 'file', ROOT_ID))
439
InventoryEntry('123123', 'hello.c', kind='file', parent_id='TREE_ROOT')
391
440
>>> inv['123123'].name
395
raise BzrError("can't look up file_id None")
398
444
return self._byid[file_id]
400
raise BzrError("file_id {%s} not in inventory" % file_id)
447
raise BzrError("can't look up file_id None")
449
raise BzrError("file_id {%s} not in inventory" % file_id)
452
def get_file_kind(self, file_id):
453
return self._byid[file_id].kind
403
455
def get_child(self, parent_id, filename):
404
456
return self[parent_id].children.get(filename)
408
460
"""Add entry to inventory.
410
462
To add a file to a branch ready to be committed, use Branch.add,
465
Returns the new entry object.
412
467
if entry.file_id in self._byid:
413
bailout("inventory already contains entry with id {%s}" % entry.file_id)
468
raise BzrError("inventory already contains entry with id {%s}" % entry.file_id)
470
if entry.parent_id == ROOT_ID or entry.parent_id is None:
471
entry.parent_id = self.root.file_id
416
474
parent = self._byid[entry.parent_id]
418
bailout("parent_id {%s} not in inventory" % entry.parent_id)
476
raise BzrError("parent_id {%s} not in inventory" % entry.parent_id)
420
478
if parent.children.has_key(entry.name):
421
bailout("%s is already versioned" %
479
raise BzrError("%s is already versioned" %
422
480
appendpath(self.id2path(parent.file_id), entry.name))
424
482
self._byid[entry.file_id] = entry
425
483
parent.children[entry.name] = entry
428
487
def add_path(self, relpath, kind, file_id=None):
429
488
"""Add entry from a path.
431
The immediate parent must already be versioned"""
490
The immediate parent must already be versioned.
492
Returns the new entry object."""
493
from bzrlib.branch import gen_file_id
432
495
parts = bzrlib.osutils.splitpath(relpath)
433
496
if len(parts) == 0:
434
bailout("cannot re-add root of inventory")
497
raise BzrError("cannot re-add root of inventory")
436
499
if file_id == None:
437
file_id = bzrlib.branch.gen_file_id(relpath)
439
parent_id = self.path2id(parts[:-1])
440
assert parent_id != None
500
file_id = gen_file_id(relpath)
502
parent_path = parts[:-1]
503
parent_id = self.path2id(parent_path)
504
if parent_id == None:
505
raise NotVersionedError(parent_path)
441
507
ie = InventoryEntry(file_id, parts[-1],
442
508
kind=kind, parent_id=parent_id)
443
509
return self.add(ie)
469
536
del self[ie.parent_id].children[ie.name]
473
return Set(self._byid)
476
def to_element(self):
477
"""Convert to XML Element"""
478
e = Element('inventory')
480
for path, ie in self.iter_entries():
481
e.append(ie.to_element())
485
def from_element(cls, elt):
486
"""Construct from XML Element
488
>>> inv = Inventory()
489
>>> inv.add(InventoryEntry('foo.c-123981239', 'foo.c', 'file', ROOT_ID))
490
>>> elt = inv.to_element()
491
>>> inv2 = Inventory.from_element(elt)
495
assert elt.tag == 'inventory'
498
o.add(InventoryEntry.from_element(e))
501
from_element = classmethod(from_element)
504
def __cmp__(self, other):
539
def __eq__(self, other):
505
540
"""Compare two sets by comparing their contents.
507
542
>>> i1 = Inventory()
511
546
>>> i1.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
547
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
514
550
>>> i2.add(InventoryEntry('123', 'foo', 'file', ROOT_ID))
551
InventoryEntry('123', 'foo', kind='file', parent_id='TREE_ROOT')
521
555
if not isinstance(other, Inventory):
522
556
return NotImplemented
524
if self.id_set() ^ other.id_set():
527
for file_id in self._byid:
528
c = cmp(self[file_id], other[file_id])
558
if len(self._byid) != len(other._byid):
559
# shortcut: obviously not the same
562
return self._byid == other._byid
565
def __ne__(self, other):
566
return not self.__eq__(other)
570
raise ValueError('not hashable')
534
573
def get_idpath(self, file_id):
605
644
This does not move the working file."""
606
645
if not is_valid_name(new_name):
607
bailout("not an acceptable filename: %r" % new_name)
646
raise BzrError("not an acceptable filename: %r" % new_name)
609
648
new_parent = self._byid[new_parent_id]
610
649
if new_name in new_parent.children:
611
bailout("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
650
raise BzrError("%r already exists in %r" % (new_name, self.id2path(new_parent_id)))
613
652
new_parent_idpath = self.get_idpath(new_parent_id)
614
653
if file_id in new_parent_idpath:
615
bailout("cannot move directory %r into a subdirectory of itself, %r"
654
raise BzrError("cannot move directory %r into a subdirectory of itself, %r"
616
655
% (self.id2path(file_id), self.id2path(new_parent_id)))
618
657
file_ie = self._byid[file_id]