14
14
# along with this program; if not, write to the Free Software
15
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# out-of-date.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.
19
# TODO: Perhaps have a way to stat all the files in inode order, and
20
# then remember that they're all fresh for the lifetime of the object?
22
# TODO: Keep track of whether there are in-memory updates that need to
23
25
# TODO: Perhaps return more details on the file to avoid statting it
24
26
# again: nonexistent, file type, size, etc
26
# TODO: Perhaps use a Python pickle instead of a text file; might be faster.
30
31
CACHE_HEADER = "### bzr hashcache v5\n"
35
from bzrlib.osutils import sha_file, pathjoin, safe_unicode
36
from bzrlib.trace import mutter, warning
37
from bzrlib.atomicfile import AtomicFile
38
from bzrlib.errors import BzrError
45
34
def _fingerprint(abspath):
47
38
fs = os.lstat(abspath)
95
86
needs_write = False
97
def __init__(self, root, cache_file_name, mode=None):
98
"""Create a hash cache in base dir, and set the file mode to mode."""
99
self.root = safe_unicode(root)
88
def __init__(self, basedir):
89
self.basedir = basedir
100
90
self.hit_count = 0
101
91
self.miss_count = 0
102
92
self.stat_count = 0
103
93
self.danger_count = 0
104
self.removed_count = 0
105
self.update_count = 0
108
self._cache_file_name = safe_unicode(cache_file_name)
110
98
def cache_file_name(self):
111
return self._cache_file_name
100
return os.path.join(self.basedir, '.bzr', 'stat-cache')
114
106
"""Discard all cached information.
118
110
self.needs_write = True
122
"""Scan all files and remove entries where the cache entry is obsolete.
124
Obsolete entries are those where the file has been modified or deleted
125
since the entry was inserted.
127
# FIXME optimisation opportunity, on linux [and check other oses]:
128
# rather than iteritems order, stat in inode order.
129
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
132
for inum, path, cache_entry in prep:
133
abspath = pathjoin(self.root, path)
134
fp = _fingerprint(abspath)
137
cache_fp = cache_entry[1]
139
if (not fp) or (cache_fp != fp):
140
# not here or not a regular file anymore
141
self.removed_count += 1
142
self.needs_write = True
143
del self._cache[path]
146
114
def get_sha1(self, path):
147
"""Return the sha1 of a file.
115
"""Return the hex SHA-1 of the contents of the file at path.
117
XXX: If the file does not exist or is not a plain file???
149
abspath = pathjoin(self.root, path)
121
from bzrlib.osutils import sha_file
122
from bzrlib.trace import mutter
124
abspath = os.path.join(self.basedir, path)
125
fp = _fingerprint(abspath)
126
c = self._cache.get(path)
128
cache_sha1, cache_fp = c
130
cache_sha1, cache_fp = None, None
150
132
self.stat_count += 1
151
file_fp = _fingerprint(abspath)
154
# not a regular file or not existing
155
if path in self._cache:
156
self.removed_count += 1
157
self.needs_write = True
158
del self._cache[path]
161
if path in self._cache:
162
cache_sha1, cache_fp = self._cache[path]
164
cache_sha1, cache_fp = None, None
166
if cache_fp == file_fp:
137
elif cache_fp and (cache_fp == fp):
167
138
self.hit_count += 1
168
139
return cache_sha1
173
mode = file_fp[FP_MODE_COLUMN]
174
if stat.S_ISREG(mode):
175
digest = sha_file(file(abspath, 'rb', buffering=65000))
176
elif stat.S_ISLNK(mode):
177
digest = sha.new(os.readlink(abspath)).hexdigest()
179
raise BzrError("file %r: unknown file stat mode: %o"%(abspath,mode))
142
digest = sha_file(file(abspath, 'rb'))
181
now = int(time.time())
182
if file_fp[FP_MTIME_COLUMN] >= now or file_fp[FP_CTIME_COLUMN] >= now:
183
# changed too recently; can't be cached. we can
184
# return the result and it could possibly be cached
187
# the point is that we only want to cache when we are sure that any
188
# subsequent modifications of the file can be detected. If a
189
# modification neither changes the inode, the device, the size, nor
190
# the mode, then we can only distinguish it by time; therefore we
191
# need to let sufficient time elapse before we may cache this entry
192
# again. If we didn't do this, then, for example, a very quick 1
193
# byte replacement in the file might go undetected.
194
self.danger_count += 1
196
self.removed_count += 1
144
now = int(time.time())
145
if fp[1] >= now or fp[2] >= now:
146
# changed too recently; can't be cached. we can
147
# return the result and it could possibly be cached
149
self.danger_count += 1
151
mutter("remove outdated entry for %s" % path)
152
self.needs_write = True
153
del self._cache[path]
154
elif (fp != cache_fp) or (digest != cache_sha1):
155
mutter("update entry for %s" % path)
156
mutter(" %r" % (fp,))
157
mutter(" %r" % (cache_fp,))
197
158
self.needs_write = True
198
del self._cache[path]
200
self.update_count += 1
201
self.needs_write = True
202
self._cache[path] = (digest, file_fp)
159
self._cache[path] = (digest, fp)
206
166
"""Write contents of cache to file."""
207
outf = AtomicFile(self.cache_file_name(), 'wb', new_mode=self._mode)
167
from atomicfile import AtomicFile
169
outf = AtomicFile(self.cache_file_name(), 'wb')
209
171
print >>outf, CACHE_HEADER,
234
195
If the cache file has the wrong version marker, this just clears
197
from bzrlib.trace import mutter, warning
238
201
fn = self.cache_file_name()
240
inf = file(fn, 'rb', buffering=65000)
241
204
except IOError, e:
242
mutter("failed to open %s: %s", fn, e)
243
# better write it now so it is valid
244
self.needs_write = True
205
mutter("failed to open %s: %s" % (fn, e))
248
209
hdr = inf.readline()
249
210
if hdr != CACHE_HEADER:
250
mutter('cache header marker not found at top of %s;'
251
' discarding cache', fn)
252
self.needs_write = True
211
mutter('cache header marker not found at top of %s; discarding cache'
added
removed
bzrlib/hashcache.py
