47
"""This class represents the abstract storage layout for saving information.
38
"""Store that holds files indexed by unique names.
49
40
Files can be added, but not modified once they are in. Typically
50
41
the hash is used as the name, or something else known to be unique,
44
>>> st = ImmutableScratchStore()
46
>>> st.add(StringIO('hello'), 'aa')
52
You are not allowed to add an id that is already present.
54
Entries can be retrieved as files, which may then be read.
56
>>> st.add(StringIO('goodbye'), '123123')
57
>>> st['123123'].read()
60
TODO: Atomic add by writing to a temporary file and renaming.
62
TODO: Perhaps automatically transform to/from XML in a method?
63
Would just need to tell the constructor what class to
66
TODO: Even within a simple disk store like this, we could
67
gzip the files. But since many are less than one disk
68
block, that might not help a lot.
55
raise NotImplementedError('Children should define their length')
57
def get(self, file_id, suffix=None):
58
"""Returns a file reading from a particular entry.
60
If suffix is present, retrieve the named suffix for file_id.
62
raise NotImplementedError
64
def __getitem__(self, fileid):
65
"""DEPRECATED. Please use .get(file_id) instead."""
66
raise NotImplementedError
68
#def __contains__(self, fileid):
69
# """Deprecated, please use has_id"""
70
# raise NotImplementedError
73
raise NotImplementedError
75
def add(self, f, fileid):
76
"""Add a file object f to the store accessible from the given fileid"""
77
raise NotImplementedError('Children of Store must define their method of adding entries.')
79
def has_id(self, file_id, suffix=None):
80
"""Return True or false for the presence of file_id in the store.
82
suffix, if present, is a per file suffix, i.e. for digital signature
84
raise NotImplementedError
87
"""Return True if this store is able to be listed."""
88
return hasattr(self, "__iter__")
90
def copy_multi(self, other, ids, pb=None, permit_failure=False):
91
"""Copy texts for ids from other into self.
93
If an id is present in self, it is skipped. A count of copied
94
ids is returned, which may be less than len(ids).
96
:param other: Another Store object
97
:param ids: A list of entry ids to be copied
98
:param pb: A ProgressBar object, if none is given, the default will be created.
99
:param permit_failure: Allow missing entries to be ignored
100
:return: (n_copied, [failed]) The number of entries copied successfully,
101
followed by a list of entries which could not be copied (because they
105
pb = bzrlib.ui.ui_factory.progress_bar()
106
pb.update('preparing to copy')
109
ids = list(ids) # get the list for showing a length.
112
if self.has_id(fileid):
115
self._copy_one(fileid, None, other, pb)
116
for suffix in self._suffixes:
118
self._copy_one(fileid, suffix, other, pb)
121
pb.update('copy', count, len(ids))
127
assert count == len(ids)
131
def _copy_one(self, fileid, suffix, other, pb):
132
"""Most generic copy-one object routine.
134
Subclasses can override this to provide an optimised
135
copy between their own instances. Such overriden routines
136
should call this if they have no optimised facility for a
139
f = other.get(fileid, suffix)
140
self.add(f, fileid, suffix)
143
class TransportStore(Store):
144
"""A TransportStore is a Store superclass for Stores that use Transports."""
146
def add(self, f, fileid, suffix=None):
72
def __init__(self, basedir):
73
"""ImmutableStore constructor."""
74
self._basedir = basedir
77
return os.path.join(self._basedir, id)
80
return "%s(%r)" % (self.__class__.__name__, self._basedir)
82
def add(self, f, fileid, compressed=True):
147
83
"""Add contents of a file into the store.
149
f -- A file-like object, or string
85
f -- An open file, or file-like object."""
86
# FIXME: Only works on smallish files
87
# TODO: Can be optimized by copying at the same time as
151
89
mutter("add store entry %r" % (fileid))
153
if suffix is not None:
154
fn = self._relpath(fileid, [suffix])
156
fn = self._relpath(fileid)
157
if self._transport.has(fn):
158
raise BzrError("store %r already contains id %r" % (self._transport.base, fileid))
162
self._transport.mkdir(hash_prefix(fileid)[:-1])
163
except errors.FileExists:
168
def _check_fileid(self, fileid):
169
if not isinstance(fileid, basestring):
170
raise TypeError('Fileids should be a string type: %s %r' % (type(fileid), fileid))
171
if '\\' in fileid or '/' in fileid:
172
raise ValueError("invalid store id %r" % fileid)
174
def has_id(self, fileid, suffix=None):
175
"""See Store.has_id."""
176
if suffix is not None:
177
fn = self._relpath(fileid, [suffix])
179
fn = self._relpath(fileid)
180
return self._transport.has(fn)
182
def _get(self, filename):
183
"""Return an vanilla file stream for clients to read from.
185
This is the body of a template method on 'get', and should be
186
implemented by subclasses.
188
raise NotImplementedError
190
def get(self, fileid, suffix=None):
191
"""See Store.get()."""
192
if suffix is None or suffix == 'gz':
193
fn = self._relpath(fileid)
195
fn = self._relpath(fileid, [suffix])
90
if isinstance(f, types.StringTypes):
95
p = self._path(fileid)
96
if os.access(p, os.F_OK) or os.access(p + '.gz', os.F_OK):
97
bailout("store %r already contains id %r" % (self._basedir, fileid))
100
f = gzip.GzipFile(p + '.gz', 'wb')
101
os.chmod(p + '.gz', 0444)
110
def __contains__(self, fileid):
112
p = self._path(fileid)
113
return (os.access(p, os.R_OK)
114
or os.access(p + '.gz', os.R_OK))
116
# TODO: Guard against the same thing being stored twice, compressed and uncompresse
119
for f in os.listdir(self._basedir):
121
# TODO: case-insensitive?
127
return len(os.listdir(self._basedir))
129
def __getitem__(self, fileid):
130
"""Returns a file reading from a particular entry."""
131
p = self._path(fileid)
198
except errors.NoSuchFile:
199
raise KeyError(fileid)
201
def __init__(self, a_transport, prefixed=False):
202
assert isinstance(a_transport, transport.Transport)
203
super(TransportStore, self).__init__()
204
self._transport = a_transport
205
self._prefixed = prefixed
206
# conflating the .gz extension and user suffixes was a mistake.
207
# RBC 20051017 - TODO SOON, separate them again.
208
self._suffixes = set()
210
def _iter_files_recursive(self):
211
"""Iterate through the files in the transport."""
212
for quoted_relpath in self._transport.iter_files_recursive():
213
yield urllib.unquote(quoted_relpath)
216
for relpath in self._iter_files_recursive():
217
# worst case is one of each suffix.
218
name = os.path.basename(relpath)
219
if name.endswith('.gz'):
222
for count in range(len(self._suffixes)):
223
for suffix in self._suffixes:
224
if name.endswith('.' + suffix):
230
return len(list(self.__iter__()))
232
def _relpath(self, fileid, suffixes=[]):
233
self._check_fileid(fileid)
234
for suffix in suffixes:
235
if not suffix in self._suffixes:
236
raise ValueError("Unregistered suffix %r" % suffix)
237
self._check_fileid(suffix)
239
path = [hash_prefix(fileid) + fileid]
242
path.extend(suffixes)
243
return transport.urlescape('.'.join(path))
246
if self._transport is None:
247
return "%s(None)" % (self.__class__.__name__)
249
return "%s(%r)" % (self.__class__.__name__, self._transport.base)
254
"""Return True if this store is able to be listed."""
255
return self._transport.listable()
257
def register_suffix(self, suffix):
258
"""Register a suffix as being expected in this store."""
259
self._check_fileid(suffix)
260
self._suffixes.add(suffix)
133
return gzip.GzipFile(p + '.gz', 'rb')
135
if e.errno == errno.ENOENT:
262
140
def total_size(self):
263
141
"""Return (count, bytes)
269
for relpath in self._transport.iter_files_recursive():
271
total += self._transport.stat(relpath).st_size
151
total += os.stat(p)[ST_SIZE]
153
total += os.stat(p + '.gz')[ST_SIZE]
273
155
return count, total
276
def ImmutableMemoryStore():
277
return bzrlib.store.text.TextStore(transport.memory.MemoryTransport())
280
class CachedStore(Store):
281
"""A store that caches data locally, to avoid repeated downloads.
282
The precacache method should be used to avoid server round-trips for
160
class ImmutableScratchStore(ImmutableStore):
161
"""Self-destructing test subclass of ImmutableStore.
163
The Store only exists for the lifetime of the Python object.
164
Obviously you should not put anything precious in it.
286
def __init__(self, store, cache_dir):
287
super(CachedStore, self).__init__()
288
self.source_store = store
289
# This clones the source store type with a locally bound
290
# transport. FIXME: it assumes a constructor is == cloning.
291
# clonable store - it might be nicer to actually have a clone()
292
# or something. RBC 20051003
293
self.cache_store = store.__class__(LocalTransport(cache_dir))
296
mutter("Cache add %s" % id)
297
if id not in self.cache_store:
298
self.cache_store.add(self.source_store.get(id), id)
299
return self.cache_store.get(id)
301
def has_id(self, fileid, suffix=None):
302
"""See Store.has_id."""
303
if self.cache_store.has_id(fileid, suffix):
305
if self.source_store.has_id(fileid, suffix):
306
# We could copy at this time
311
def copy_all(store_from, store_to):
312
"""Copy all ids from one store to another."""
313
# TODO: Optional progress indicator
314
if not store_from.listable():
315
raise UnlistableStore(store_from)
316
ids = [f for f in store_from]
317
store_to.copy_multi(store_from, ids)
319
def hash_prefix(file_id):
320
return "%02x/" % (adler32(file_id) & 0xff)
167
ImmutableStore.__init__(self, tempfile.mkdtemp())
170
for f in os.listdir(self._basedir):
171
fpath = os.path.join(self._basedir, f)
172
# needed on windows, and maybe some other filesystems
173
os.chmod(fpath, 0600)
175
os.rmdir(self._basedir)
176
mutter("%r destroyed" % self)
added
removed
bzrlib/store.py
