32
33
class StoreError(Exception):
36
"""This class represents the abstract storage layout for saving information.
38
"""Store that holds files indexed by unique names.
40
Files can be added, but not modified once they are in. Typically
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.
39
_max_buffered_requests = 10
41
def __init__(self, transport):
42
assert isinstance(transport, bzrlib.transport.Transport)
43
self._transport = transport
72
def __init__(self, basedir):
73
"""ImmutableStore constructor."""
74
self._basedir = basedir
78
return os.path.join(self._basedir, id)
45
80
def __repr__(self):
46
if self._transport is None:
47
return "%s(None)" % (self.__class__.__name__)
49
return "%s(%r)" % (self.__class__.__name__, self._transport.base)
81
return "%s(%r)" % (self.__class__.__name__, self._basedir)
83
def add(self, f, fileid, compressed=True):
84
"""Add contents of a file into the store.
86
f -- An open file, or file-like object."""
87
# FIXME: Only works on smallish files
88
# TODO: Can be optimized by copying at the same time as
90
mutter("add store entry %r" % (fileid))
91
if isinstance(f, types.StringTypes):
96
p = self._path(fileid)
97
if os.access(p, os.F_OK) or os.access(p + '.gz', os.F_OK):
98
bailout("store %r already contains id %r" % (self._basedir, fileid))
101
f = gzip.GzipFile(p + '.gz', 'wb')
102
os.chmod(p + '.gz', 0444)
111
def __contains__(self, fileid):
113
p = self._path(fileid)
114
return (os.access(p, os.R_OK)
115
or os.access(p + '.gz', os.R_OK))
117
# TODO: Guard against the same thing being stored twice, compressed and uncompresse
120
for f in os.listdir(self._basedir):
122
# TODO: case-insensitive?
53
127
def __len__(self):
54
raise NotImplementedError('Children should define their length')
128
return len(os.listdir(self._basedir))
56
130
def __getitem__(self, fileid):
57
131
"""Returns a file reading from a particular entry."""
58
raise NotImplementedError
60
def __contains__(self, fileid):
62
raise NotImplementedError
65
raise NotImplementedError
67
def add(self, f, fileid):
68
"""Add a file object f to the store accessible from the given fileid"""
69
raise NotImplementedError('Children of Store must define their method of adding entries.')
71
def add_multi(self, entries):
72
"""Add a series of file-like or string objects to the store with the given
75
:param entries: A list of tuples of file,id pairs [(file1, id1), (file2, id2), ...]
76
This could also be a generator yielding (file,id) pairs.
78
for f, fileid in entries:
81
def has(self, fileids):
82
"""Return True/False for each entry in fileids.
84
:param fileids: A List or generator yielding file ids.
85
:return: A generator or list returning True/False for each entry.
87
for fileid in fileids:
132
p = self._path(fileid)
134
return gzip.GzipFile(p + '.gz', 'rb')
136
if e.errno == errno.ENOENT:
93
def get(self, fileids, ignore_missing=False, pb=None):
94
"""Return a set of files, one for each requested entry.
96
:param ignore_missing: If true, return None for entries which do not
98
:return: A list or generator of file-like objects, one for each id.
100
for fileid in fileids:
141
def total_size(self):
142
"""Return (count, bytes)
144
This is the (compressed) size stored on disk, not the size of
109
def copy_multi(self, other, ids, pb=None, permit_failure=False):
110
"""Copy texts for ids from other into self.
112
If an id is present in self, it is skipped. A count of copied
113
ids is returned, which may be less than len(ids).
115
:param other: Another Store object
116
:param ids: A list of entry ids to be copied
117
:param pb: A ProgressBar object, if none is given, the default will be created.
118
:param permit_failure: Allow missing entries to be ignored
119
:return: (n_copied, [failed]) The number of entries copied successfully,
120
followed by a list of entries which could not be copied (because they
124
pb = bzrlib.ui.ui_factory.progress_bar()
126
ids = list(ids) # Make sure we don't have a generator, since we iterate 2 times
127
pb.update('preparing to copy')
129
for file_id, has in zip(ids, self.has(ids)):
131
to_copy.append(file_id)
132
return self._do_copy(other, to_copy, pb, permit_failure=permit_failure)
134
def _do_copy(self, other, to_copy, pb, permit_failure=False):
135
"""This is the standard copying mechanism, just get them one at
136
a time from remote, and store them locally.
138
:param other: Another Store object
139
:param to_copy: A list of entry ids to copy
140
:param pb: A ProgressBar object to display completion status.
141
:param permit_failure: Allow missing entries to be ignored
142
:return: (n_copied, [failed])
143
The number of entries copied, and a list of failed entries.
145
# This should be updated to use add_multi() rather than
146
# the current methods of buffering requests.
147
# One question, is it faster to queue up 1-10 and then copy 1-10
148
# then queue up 11-20, copy 11-20
149
# or to queue up 1-10, copy 1, queue 11, copy 2, etc?
150
# sort of pipeline versus batch.
152
# We can't use self._transport.copy_to because we don't know
153
# whether the local tree is in the same format as other
155
def buffer_requests():
157
buffered_requests = []
158
for fileid in to_copy:
168
buffered_requests.append((f, fileid))
169
if len(buffered_requests) > self._max_buffered_requests:
170
yield buffered_requests.pop(0)
172
pb.update('copy', count, len(to_copy))
174
for req in buffered_requests:
177
pb.update('copy', count, len(to_copy))
179
assert count == len(to_copy)
181
self.add_multi(buffer_requests())
184
return len(to_copy), failed
152
total += os.stat(p)[ST_SIZE]
154
total += os.stat(p + '.gz')[ST_SIZE]
161
class ImmutableScratchStore(ImmutableStore):
162
"""Self-destructing test subclass of ImmutableStore.
164
The Store only exists for the lifetime of the Python object.
165
Obviously you should not put anything precious in it.
168
ImmutableStore.__init__(self, tempfile.mkdtemp())
171
for f in os.listdir(self._basedir):
172
fpath = os.path.join(self._basedir, f)
173
# needed on windows, and maybe some other filesystems
174
os.chmod(fpath, 0600)
176
os.rmdir(self._basedir)
177
mutter("%r destroyed" % self)
added
removed
bzrlib/store.py
