1
# (C) 2005 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
# TODO: Up-front, stat all files in order and remove those which are deleted or
18
# out-of-date. Don't actually re-read them until they're needed. That ought
19
# to bring all the inodes into core so that future stats to them are fast, and
20
# it preserves the nice property that any caller will always get up-to-date
21
# data except in unavoidable cases.
23
# TODO: Perhaps return more details on the file to avoid statting it
24
# again: nonexistent, file type, size, etc
26
# TODO: Perhaps use a Python pickle instead of a text file; might be faster.
30
CACHE_HEADER = "### bzr hashcache v5\n"
35
from bzrlib.osutils import sha_file, pathjoin
36
from bzrlib.trace import mutter, warning
37
from bzrlib.atomicfile import AtomicFile
38
from bzrlib.errors import BzrError
45
def _fingerprint(abspath):
47
fs = os.lstat(abspath)
49
# might be missing, etc
52
if stat.S_ISDIR(fs.st_mode):
55
# we discard any high precision because it's not reliable; perhaps we
56
# could do better on some systems?
57
return (fs.st_size, long(fs.st_mtime),
58
long(fs.st_ctime), fs.st_ino, fs.st_dev, fs.st_mode)
61
class HashCache(object):
62
"""Cache for looking up file SHA-1.
64
Files are considered to match the cached value if the fingerprint
65
of the file has not changed. This includes its mtime, ctime,
66
device number, inode number, and size. This should catch
67
modifications or replacement of the file by a new one.
69
This may not catch modifications that do not change the file's
70
size and that occur within the resolution window of the
71
timestamps. To handle this we specifically do not cache files
72
which have changed since the start of the present second, since
73
they could undetectably change again.
75
This scheme may fail if the machine's clock steps backwards.
78
This does not canonicalize the paths passed in; that should be
82
Indexed by path, points to a two-tuple of the SHA-1 of the file.
86
number of times files have been statted
89
number of times files have been retrieved from the cache, avoiding a
93
number of misses (times files have been completely re-read)
97
def __init__(self, basedir, mode=None):
98
"""Create a hash cache in base dir, and set the file mode to mode."""
99
self.basedir = basedir
103
self.danger_count = 0
104
self.removed_count = 0
105
self.update_count = 0
109
def cache_file_name(self):
110
# FIXME: duplicate path logic here, this should be
111
# something like 'branch.controlfile'.
112
return pathjoin(self.basedir, '.bzr', 'stat-cache')
115
"""Discard all cached information.
117
This does not reset the counters."""
119
self.needs_write = True
124
"""Scan all files and remove entries where the cache entry is obsolete.
126
Obsolete entries are those where the file has been modified or deleted
127
since the entry was inserted.
129
# FIXME optimisation opportunity, on linux [and check other oses]:
130
# rather than iteritems order, stat in inode order.
131
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
134
for inum, path, cache_entry in prep:
135
abspath = pathjoin(self.basedir, path)
136
fp = _fingerprint(abspath)
139
cache_fp = cache_entry[1]
141
if (not fp) or (cache_fp != fp):
142
# not here or not a regular file anymore
143
self.removed_count += 1
144
self.needs_write = True
145
del self._cache[path]
148
def get_sha1(self, path):
149
"""Return the sha1 of a file.
151
abspath = pathjoin(self.basedir, path)
153
file_fp = _fingerprint(abspath)
156
# not a regular file or not existing
157
if path in self._cache:
158
self.removed_count += 1
159
self.needs_write = True
160
del self._cache[path]
163
if path in self._cache:
164
cache_sha1, cache_fp = self._cache[path]
166
cache_sha1, cache_fp = None, None
168
if cache_fp == file_fp:
175
mode = file_fp[FP_MODE_COLUMN]
176
if stat.S_ISREG(mode):
177
digest = sha_file(file(abspath, 'rb', buffering=65000))
178
elif stat.S_ISLNK(mode):
179
digest = sha.new(os.readlink(abspath)).hexdigest()
181
raise BzrError("file %r: unknown file stat mode: %o"%(abspath,mode))
183
now = int(time.time())
184
if file_fp[FP_MTIME_COLUMN] >= now or file_fp[FP_CTIME_COLUMN] >= now:
185
# changed too recently; can't be cached. we can
186
# return the result and it could possibly be cached
189
# the point is that we only want to cache when we are sure that any
190
# subsequent modifications of the file can be detected. If a
191
# modification neither changes the inode, the device, the size, nor
192
# the mode, then we can only distinguish it by time; therefore we
193
# need to let sufficient time elapse before we may cache this entry
194
# again. If we didn't do this, then, for example, a very quick 1
195
# byte replacement in the file might go undetected.
196
self.danger_count += 1
198
self.removed_count += 1
199
self.needs_write = True
200
del self._cache[path]
202
self.update_count += 1
203
self.needs_write = True
204
self._cache[path] = (digest, file_fp)
208
"""Write contents of cache to file."""
209
outf = AtomicFile(self.cache_file_name(), 'wb', new_mode=self._mode)
211
print >>outf, CACHE_HEADER,
213
for path, c in self._cache.iteritems():
214
assert '//' not in path, path
215
outf.write(path.encode('utf-8'))
217
print >>outf, c[0], # hex sha1
219
print >>outf, "%d" % fld,
223
self.needs_write = False
229
"""Reinstate cache from file.
231
Overwrites existing cache.
233
If the cache file has the wrong version marker, this just clears
237
fn = self.cache_file_name()
239
inf = file(fn, 'rb', buffering=65000)
241
mutter("failed to open %s: %s", fn, e)
242
# better write it now so it is valid
243
self.needs_write = True
248
if hdr != CACHE_HEADER:
249
mutter('cache header marker not found at top of %s;'
250
' discarding cache', fn)
251
self.needs_write = True
256
path = l[:pos].decode('utf-8')
257
if path in self._cache:
258
warning('duplicated path %r in cache' % path)
262
fields = l[pos:].split(' ')
264
warning("bad line in hashcache: %r" % l)
269
warning("bad sha1 in hashcache: %r" % sha1)
272
fp = tuple(map(long, fields[1:]))
274
self._cache[path] = (sha1, fp)
276
self.needs_write = False
added
removed
bzrlib/hashcache.py
