55
"""This class represents the abstract storage layout for saving information.
38
"""Store that holds files indexed by unique names.
57
40
Files can be added, but not modified once they are in. Typically
58
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.
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):
83
"""Add contents of a file into the store.
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
89
mutter("add store entry %r" % (fileid))
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?
62
126
def __len__(self):
63
raise NotImplementedError('Children should define their length')
65
def get(self, fileid, suffix=None):
66
"""Returns a file reading from a particular entry.
68
If suffix is present, retrieve the named suffix for fileid.
70
raise NotImplementedError
127
return len(os.listdir(self._basedir))
72
129
def __getitem__(self, fileid):
73
"""DEPRECATED. Please use .get(fileid) instead."""
74
raise NotImplementedError
76
#def __contains__(self, fileid):
77
# """Deprecated, please use has_id"""
78
# raise NotImplementedError
81
raise NotImplementedError
83
def add(self, f, fileid):
84
"""Add a file object f to the store accessible from the given fileid"""
85
raise NotImplementedError('Children of Store must define their method of adding entries.')
87
def has_id(self, fileid, suffix=None):
88
"""Return True or false for the presence of fileid in the store.
90
suffix, if present, is a per file suffix, i.e. for digital signature
92
raise NotImplementedError
95
"""Return True if this store is able to be listed."""
96
return (getattr(self, "__iter__", None) is not None)
98
def copy_all_ids(self, store_from, pb=None):
99
"""Copy all the file ids from store_from into self."""
100
if not store_from.listable():
101
raise UnlistableStore(store_from)
103
for count, file_id in enumerate(store_from):
105
pb.update('listing files', count, count)
109
mutter('copy_all ids: %r', ids)
110
self.copy_multi(store_from, ids, pb=pb)
112
def copy_multi(self, other, ids, pb=None, permit_failure=False):
113
"""Copy texts for ids from other into self.
115
If an id is present in self, it is skipped. A count of copied
116
ids is returned, which may be less than len(ids).
118
:param other: Another Store object
119
:param ids: A list of entry ids to be copied
120
:param pb: A ProgressBar object, if none is given, the default will be created.
121
:param permit_failure: Allow missing entries to be ignored
122
:return: (n_copied, [failed]) The number of entries copied successfully,
123
followed by a list of entries which could not be copied (because they
127
pb.update('preparing to copy')
130
ids = [osutils.safe_file_id(i) for i in ids] # get the list for showing a length.
133
if self.has_id(fileid):
136
self._copy_one(fileid, None, other, pb)
137
for suffix in self._suffixes:
139
self._copy_one(fileid, suffix, other, pb)
143
pb.update('copy', count, len(ids))
149
assert count == len(ids)
154
def _copy_one(self, fileid, suffix, other, pb):
155
"""Most generic copy-one object routine.
157
Subclasses can override this to provide an optimised
158
copy between their own instances. Such overriden routines
159
should call this if they have no optimised facility for a
162
mutter('Store._copy_one: %r', fileid)
163
f = other.get(fileid, suffix)
164
self.add(f, fileid, suffix)
167
class TransportStore(Store):
168
"""A TransportStore is a Store superclass for Stores that use Transports."""
170
def add(self, f, fileid, suffix=None):
171
"""Add contents of a file into the store.
173
f -- A file-like object
175
fileid = osutils.safe_file_id(fileid)
176
mutter("add store entry %r", fileid)
177
names = self._id_to_names(fileid, suffix)
178
if self._transport.has_any(names):
179
raise BzrError("store %r already contains id %r"
180
% (self._transport.base, fileid))
182
# Most of the time, just adding the file will work
183
# if we find a time where it fails, (because the dir
184
# doesn't exist), then create the dir, and try again
185
self._add(names[0], f)
187
def _add(self, relpath, f):
188
"""Actually add the file to the given location.
189
This should be overridden by children.
191
raise NotImplementedError('children need to implement this function.')
193
def _check_fileid(self, fileid):
194
if not isinstance(fileid, basestring):
195
raise TypeError('Fileids should be a string type: %s %r' % (type(fileid), fileid))
196
if '\\' in fileid or '/' in fileid:
197
raise ValueError("invalid store id %r" % fileid)
199
def _id_to_names(self, fileid, suffix):
200
"""Return the names in the expected order"""
201
if suffix is not None:
202
fn = self._relpath(fileid, [suffix])
204
fn = self._relpath(fileid)
206
# FIXME RBC 20051128 this belongs in TextStore.
213
def has_id(self, fileid, suffix=None):
214
"""See Store.has_id."""
215
fileid = osutils.safe_file_id(fileid)
216
return self._transport.has_any(self._id_to_names(fileid, suffix))
218
def _get_name(self, fileid, suffix=None):
219
"""A special check, which returns the name of an existing file.
221
This is similar in spirit to 'has_id', but it is designed
222
to return information about which file the store has.
224
for name in self._id_to_names(fileid, suffix=suffix):
225
if self._transport.has(name):
229
def _get(self, filename):
230
"""Return an vanilla file stream for clients to read from.
232
This is the body of a template method on 'get', and should be
233
implemented by subclasses.
235
raise NotImplementedError
237
def get(self, fileid, suffix=None):
238
"""See Store.get()."""
239
fileid = osutils.safe_file_id(fileid)
240
names = self._id_to_names(fileid, suffix)
243
return self._get(name)
244
except errors.NoSuchFile:
246
raise KeyError(fileid)
248
def __init__(self, a_transport, prefixed=False, compressed=False,
249
dir_mode=None, file_mode=None,
251
assert isinstance(a_transport, Transport)
252
super(TransportStore, self).__init__()
253
self._transport = a_transport
254
self._prefixed = prefixed
255
# FIXME RBC 20051128 this belongs in TextStore.
256
self._compressed = compressed
257
self._suffixes = set()
258
self._escaped = escaped
260
# It is okay for these to be None, it just means they
261
# will just use the filesystem defaults
262
self._dir_mode = dir_mode
263
self._file_mode = file_mode
265
def _unescape(self, file_id):
266
"""If filename escaping is enabled for this store, unescape and return the filename."""
268
return urllib.unquote(file_id)
272
def _iter_files_recursive(self):
273
"""Iterate through the files in the transport."""
274
for quoted_relpath in self._transport.iter_files_recursive():
275
# transport iterator always returns quoted paths, regardless of
277
yield urllib.unquote(quoted_relpath)
280
for relpath in self._iter_files_recursive():
281
# worst case is one of each suffix.
282
name = os.path.basename(relpath)
283
if name.endswith('.gz'):
286
for count in range(len(self._suffixes)):
287
for suffix in self._suffixes:
288
if name.endswith('.' + suffix):
291
yield self._unescape(name)
294
return len(list(self.__iter__()))
296
def _relpath(self, fileid, suffixes=None):
297
self._check_fileid(fileid)
299
for suffix in suffixes:
300
if not suffix in self._suffixes:
301
raise ValueError("Unregistered suffix %r" % suffix)
302
self._check_fileid(suffix)
305
fileid = self._escape_file_id(fileid)
307
# hash_prefix adds the '/' separator
308
prefix = self.hash_prefix(fileid, escaped=True)
311
path = prefix + fileid
312
full_path = u'.'.join([path] + suffixes)
313
return urlutils.escape(full_path)
315
def _escape_file_id(self, file_id):
316
"""Turn a file id into a filesystem safe string.
318
This is similar to a plain urllib.quote, except
319
it uses specific safe characters, so that it doesn't
320
have to translate a lot of valid file ids.
322
if not self._escaped:
324
if isinstance(file_id, unicode):
325
file_id = file_id.encode('utf-8')
326
# @ does not get escaped. This is because it is a valid
327
# filesystem character we use all the time, and it looks
328
# a lot better than seeing %40 all the time.
329
safe = "abcdefghijklmnopqrstuvwxyz0123456789-_@,."
330
r = [((c in safe) and c or ('%%%02x' % ord(c)))
334
def hash_prefix(self, fileid, escaped=False):
335
# fileid should be unescaped
336
if not escaped and self._escaped:
337
fileid = self._escape_file_id(fileid)
338
return "%02x/" % (adler32(fileid) & 0xff)
341
if self._transport is None:
342
return "%s(None)" % (self.__class__.__name__)
344
return "%s(%r)" % (self.__class__.__name__, self._transport.base)
349
"""Return True if this store is able to be listed."""
350
return self._transport.listable()
352
def register_suffix(self, suffix):
353
"""Register a suffix as being expected in this store."""
354
self._check_fileid(suffix)
356
raise ValueError('You cannot register the "gz" suffix.')
357
self._suffixes.add(suffix)
130
"""Returns a file reading from a particular entry."""
131
p = self._path(fileid)
133
return gzip.GzipFile(p + '.gz', 'rb')
135
if e.errno == errno.ENOENT:
359
140
def total_size(self):
360
141
"""Return (count, bytes)
added
removed
bzrlib/store.py
