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# out-of-date.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
28
CACHE_HEADER = "### bzr hashcache v5\n"
33
from bzrlib.osutils import sha_file
34
from bzrlib.trace import mutter, warning
38
def _fingerprint(abspath):
40
fs = os.lstat(abspath)
42
# might be missing, etc
45
if stat.S_ISDIR(fs.st_mode):
48
# we discard any high precision because it's not reliable; perhaps we
49
# could do better on some systems?
50
return (fs.st_size, long(fs.st_mtime),
51
long(fs.st_ctime), fs.st_ino, fs.st_dev, fs.st_mode)
54
class HashCache(object):
55
"""Cache for looking up file SHA-1.
57
Files are considered to match the cached value if the fingerprint
58
of the file has not changed. This includes its mtime, ctime,
59
device number, inode number, and size. This should catch
60
modifications or replacement of the file by a new one.
62
This may not catch modifications that do not change the file's
63
size and that occur within the resolution window of the
64
timestamps. To handle this we specifically do not cache files
65
which have changed since the start of the present second, since
66
they could undetectably change again.
68
This scheme may fail if the machine's clock steps backwards.
71
This does not canonicalize the paths passed in; that should be
75
Indexed by path, points to a two-tuple of the SHA-1 of the file.
79
number of times files have been statted
82
number of times files have been retrieved from the cache, avoiding a
86
number of misses (times files have been completely re-read)
90
def __init__(self, basedir):
91
self.basedir = basedir
96
self.removed_count = 0
101
def cache_file_name(self):
102
return os.sep.join([self.basedir, '.bzr', 'stat-cache'])
108
"""Discard all cached information.
110
This does not reset the counters."""
112
self.needs_write = True
117
"""Scan all files and remove entries where the cache entry is obsolete.
119
Obsolete entries are those where the file has been modified or deleted
120
since the entry was inserted.
122
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
125
for inum, path, cache_entry in prep:
126
abspath = os.sep.join([self.basedir, path])
127
fp = _fingerprint(abspath)
130
cache_fp = cache_entry[1]
132
if (not fp) or (cache_fp != fp):
133
# not here or not a regular file anymore
134
self.removed_count += 1
135
self.needs_write = True
136
del self._cache[path]
140
def get_sha1(self, path):
141
"""Return the sha1 of a file.
143
abspath = os.sep.join([self.basedir, path])
145
file_fp = _fingerprint(abspath)
148
# not a regular file or not existing
149
if path in self._cache:
150
self.removed_count += 1
151
self.needs_write = True
152
del self._cache[path]
155
if path in self._cache:
156
cache_sha1, cache_fp = self._cache[path]
158
cache_sha1, cache_fp = None, None
160
if cache_fp == file_fp:
167
mode = file_fp[FP_MODE_COLUMN]
168
if stat.S_ISREG(mode):
169
digest = sha_file(file(abspath, 'rb', buffering=65000))
170
elif stat.S_ISLNK(mode):
171
link_target = os.readlink(abspath)
172
digest = sha.new(os.readlink(abspath)).hexdigest()
174
raise BzrError("file %r: unknown file stat mode: %o"%(abspath,mode))
176
now = int(time.time())
177
if file_fp[1] >= now or file_fp[2] >= now:
178
# changed too recently; can't be cached. we can
179
# return the result and it could possibly be cached
181
self.danger_count += 1
183
self.removed_count += 1
184
self.needs_write = True
185
del self._cache[path]
187
self.update_count += 1
188
self.needs_write = True
189
self._cache[path] = (digest, file_fp)
193
"""Write contents of cache to file."""
194
from atomicfile import AtomicFile
196
outf = AtomicFile(self.cache_file_name(), 'wb')
198
print >>outf, CACHE_HEADER,
200
for path, c in self._cache.iteritems():
201
assert '//' not in path, path
202
outf.write(path.encode('utf-8'))
204
print >>outf, c[0], # hex sha1
206
print >>outf, "%d" % fld,
210
self.needs_write = False
218
"""Reinstate cache from file.
220
Overwrites existing cache.
222
If the cache file has the wrong version marker, this just clears
226
fn = self.cache_file_name()
228
inf = file(fn, 'rb', buffering=65000)
230
mutter("failed to open %s: %s" % (fn, e))
231
# better write it now so it is valid
232
self.needs_write = True
237
if hdr != CACHE_HEADER:
238
mutter('cache header marker not found at top of %s; discarding cache'
240
self.needs_write = True
245
path = l[:pos].decode('utf-8')
246
if path in self._cache:
247
warning('duplicated path %r in cache' % path)
251
fields = l[pos:].split(' ')
253
warning("bad line in hashcache: %r" % l)
258
warning("bad sha1 in hashcache: %r" % sha1)
261
fp = tuple(map(long, fields[1:]))
263
self._cache[path] = (sha1, fp)
265
self.needs_write = False
added
removed
bzrlib/hashcache.py
