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# out-of-date.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
36
from bzrlib.trace import mutter, warning
37
from bzrlib.atomicfile import AtomicFile
38
from bzrlib.osutils import pathjoin
43
def _fingerprint(abspath):
45
fs = os.lstat(abspath)
47
# might be missing, etc
50
if stat.S_ISDIR(fs.st_mode):
53
# we discard any high precision because it's not reliable; perhaps we
54
# could do better on some systems?
55
return (fs.st_size, long(fs.st_mtime),
56
long(fs.st_ctime), fs.st_ino, fs.st_dev, fs.st_mode)
59
class HashCache(object):
60
"""Cache for looking up file SHA-1.
62
Files are considered to match the cached value if the fingerprint
63
of the file has not changed. This includes its mtime, ctime,
64
device number, inode number, and size. This should catch
65
modifications or replacement of the file by a new one.
67
This may not catch modifications that do not change the file's
68
size and that occur within the resolution window of the
69
timestamps. To handle this we specifically do not cache files
70
which have changed since the start of the present second, since
71
they could undetectably change again.
73
This scheme may fail if the machine's clock steps backwards.
76
This does not canonicalize the paths passed in; that should be
80
Indexed by path, points to a two-tuple of the SHA-1 of the file.
84
number of times files have been statted
87
number of times files have been retrieved from the cache, avoiding a
91
number of misses (times files have been completely re-read)
95
def __init__(self, basedir):
96
self.basedir = basedir
100
self.danger_count = 0
101
self.removed_count = 0
102
self.update_count = 0
105
def cache_file_name(self):
106
# FIXME: duplicate path logic here, this should be
107
# something like 'branch.controlfile'.
108
return pathjoin(self.basedir, '.bzr', 'stat-cache')
111
"""Discard all cached information.
113
This does not reset the counters."""
115
self.needs_write = True
120
"""Scan all files and remove entries where the cache entry is obsolete.
122
Obsolete entries are those where the file has been modified or deleted
123
since the entry was inserted.
125
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
128
for inum, path, cache_entry in prep:
129
abspath = pathjoin(self.basedir, path)
130
fp = _fingerprint(abspath)
133
cache_fp = cache_entry[1]
135
if (not fp) or (cache_fp != fp):
136
# not here or not a regular file anymore
137
self.removed_count += 1
138
self.needs_write = True
139
del self._cache[path]
142
def get_sha1(self, path):
143
"""Return the sha1 of a file.
145
abspath = pathjoin(self.basedir, path)
147
file_fp = _fingerprint(abspath)
150
# not a regular file or not existing
151
if path in self._cache:
152
self.removed_count += 1
153
self.needs_write = True
154
del self._cache[path]
157
if path in self._cache:
158
cache_sha1, cache_fp = self._cache[path]
160
cache_sha1, cache_fp = None, None
162
if cache_fp == file_fp:
169
mode = file_fp[FP_MODE_COLUMN]
170
if stat.S_ISREG(mode):
171
digest = sha_file(file(abspath, 'rb', buffering=65000))
172
elif stat.S_ISLNK(mode):
173
link_target = os.readlink(abspath)
174
digest = sha.new(os.readlink(abspath)).hexdigest()
176
raise BzrError("file %r: unknown file stat mode: %o"%(abspath,mode))
178
now = int(time.time())
179
if file_fp[1] >= now or file_fp[2] >= now:
180
# changed too recently; can't be cached. we can
181
# return the result and it could possibly be cached
183
self.danger_count += 1
185
self.removed_count += 1
186
self.needs_write = True
187
del self._cache[path]
189
self.update_count += 1
190
self.needs_write = True
191
self._cache[path] = (digest, file_fp)
195
"""Write contents of cache to file."""
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
216
"""Reinstate cache from file.
218
Overwrites existing cache.
220
If the cache file has the wrong version marker, this just clears
224
fn = self.cache_file_name()
226
inf = file(fn, 'rb', buffering=65000)
228
mutter("failed to open %s: %s", fn, e)
229
# better write it now so it is valid
230
self.needs_write = True
235
if hdr != CACHE_HEADER:
236
mutter('cache header marker not found at top of %s;'
237
' discarding cache', fn)
238
self.needs_write = True
243
path = l[:pos].decode('utf-8')
244
if path in self._cache:
245
warning('duplicated path %r in cache' % path)
249
fields = l[pos:].split(' ')
251
warning("bad line in hashcache: %r" % l)
256
warning("bad sha1 in hashcache: %r" % sha1)
259
fp = tuple(map(long, fields[1:]))
261
self._cache[path] = (sha1, fp)
263
self.needs_write = False