1
# Copyright (C) 2005, 2006 Canonical Ltd
1
# (C) 2005 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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# out-of-date.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.
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
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"
39
filters as _mod_filters,
34
def _fingerprint(abspath):
38
fs = os.lstat(abspath)
40
# might be missing, etc
43
if stat.S_ISDIR(fs.st_mode):
46
# we discard any high precision because it's not reliable; perhaps we
47
# could do better on some systems?
48
return (fs.st_size, long(fs.st_mtime),
49
long(fs.st_ctime), fs.st_ino, fs.st_dev)
51
52
class HashCache(object):
117
110
self.needs_write = True
121
"""Scan all files and remove entries where the cache entry is obsolete.
123
Obsolete entries are those where the file has been modified or deleted
124
since the entry was inserted.
126
# FIXME optimisation opportunity, on linux [and check other oses]:
127
# rather than iteritems order, stat in inode order.
128
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
131
for inum, path, cache_entry in prep:
132
abspath = osutils.pathjoin(self.root, path)
133
fp = self._fingerprint(abspath)
136
cache_fp = cache_entry[1]
138
if (not fp) or (cache_fp != fp):
139
# not here or not a regular file anymore
140
self.removed_count += 1
141
self.needs_write = True
142
del self._cache[path]
144
def get_sha1(self, path, stat_value=None):
145
"""Return the sha1 of a file.
147
if path.__class__ is str:
148
abspath = osutils.pathjoin(self.root_utf8, path)
114
def get_sha1(self, path):
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???
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
150
abspath = osutils.pathjoin(self.root, path)
130
cache_sha1, cache_fp = None, None
151
132
self.stat_count += 1
152
file_fp = self._fingerprint(abspath, stat_value)
155
# not a regular file or not existing
156
if path in self._cache:
157
self.removed_count += 1
158
self.needs_write = True
159
del self._cache[path]
162
if path in self._cache:
163
cache_sha1, cache_fp = self._cache[path]
165
cache_sha1, cache_fp = None, None
167
if cache_fp == file_fp:
168
## mutter("hashcache hit for %s %r -> %s", path, file_fp, cache_sha1)
169
## mutter("now = %s", time.time())
137
elif cache_fp and (cache_fp == fp):
170
138
self.hit_count += 1
171
139
return cache_sha1
175
mode = file_fp[FP_MODE_COLUMN]
176
if stat.S_ISREG(mode):
177
if self._filter_provider is None:
180
filters = self._filter_provider(path=path, file_id=None)
181
digest = self._really_sha1_file(abspath, filters)
182
elif stat.S_ISLNK(mode):
183
target = osutils.readlink(osutils.safe_unicode(abspath))
184
digest = osutils.sha_string(target.encode('UTF-8'))
186
raise errors.BzrError("file %r: unknown file stat mode: %o"
142
digest = sha_file(file(abspath, 'rb', buffering=65000))
189
# window of 3 seconds to allow for 2s resolution on windows,
190
# unsynchronized file servers, etc.
191
cutoff = self._cutoff_time()
192
if file_fp[FP_MTIME_COLUMN] >= cutoff \
193
or file_fp[FP_CTIME_COLUMN] >= cutoff:
194
# changed too recently; can't be cached. we can
195
# return the result and it could possibly be cached
198
# the point is that we only want to cache when we are sure that any
199
# subsequent modifications of the file can be detected. If a
200
# modification neither changes the inode, the device, the size, nor
201
# the mode, then we can only distinguish it by time; therefore we
202
# need to let sufficient time elapse before we may cache this entry
203
# again. If we didn't do this, then, for example, a very quick 1
204
# byte replacement in the file might go undetected.
205
## mutter('%r modified too recently; not caching', path)
206
self.danger_count += 1
208
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,))
209
158
self.needs_write = True
210
del self._cache[path]
212
## mutter('%r added to cache: now=%f, mtime=%d, ctime=%d',
213
## path, time.time(), file_fp[FP_MTIME_COLUMN],
214
## file_fp[FP_CTIME_COLUMN])
215
self.update_count += 1
216
self.needs_write = True
217
self._cache[path] = (digest, file_fp)
220
def _really_sha1_file(self, abspath, filters):
221
"""Calculate the SHA1 of a file by reading the full text"""
222
return _mod_filters.internal_size_sha_file_byname(abspath, filters)[1]
159
self._cache[path] = (digest, fp)
225
166
"""Write contents of cache to file."""
226
outf = atomicfile.AtomicFile(self.cache_file_name(), 'wb',
167
from atomicfile import AtomicFile
169
outf = AtomicFile(self.cache_file_name(), 'wb')
229
outf.write(CACHE_HEADER)
171
print >>outf, CACHE_HEADER,
231
173
for path, c in self._cache.iteritems():
232
line_info = [path.encode('utf-8'), '// ', c[0], ' ']
233
line_info.append(' '.join([str(fld) for fld in c[1]]))
234
line_info.append('\n')
235
outf.write(''.join(line_info))
174
assert '//' not in path, path
175
outf.write(path.encode('utf-8'))
177
print >>outf, c[0], # hex sha1
179
print >>outf, "%d" % fld,
237
183
self.needs_write = False
238
## mutter("write hash cache: %s hits=%d misses=%d stat=%d recent=%d updates=%d",
239
## self.cache_file_name(), self.hit_count, self.miss_count,
241
## self.danger_count, self.update_count)
246
191
"""Reinstate cache from file.
248
193
Overwrites existing cache.
250
If the cache file has the wrong version marker, this just clears
195
If the cache file has the wrong version marker, this just clears
197
from bzrlib.trace import mutter, warning
254
201
fn = self.cache_file_name()
256
203
inf = file(fn, 'rb', buffering=65000)
257
204
except IOError, e:
258
trace.mutter("failed to open %s: %s", fn, e)
259
# better write it now so it is valid
260
self.needs_write = True
205
mutter("failed to open %s: %s" % (fn, e))
263
209
hdr = inf.readline()
264
210
if hdr != CACHE_HEADER:
265
trace.mutter('cache header marker not found at top of %s;'
266
' discarding cache', fn)
267
self.needs_write = True
211
mutter('cache header marker not found at top of %s; discarding cache'
271
216
pos = l.index('// ')
272
217
path = l[:pos].decode('utf-8')
273
218
if path in self._cache:
274
trace.warning('duplicated path %r in cache' % path)
219
warning('duplicated path %r in cache' % path)
278
223
fields = l[pos:].split(' ')
280
trace.warning("bad line in hashcache: %r" % l)
225
warning("bad line in hashcache: %r" % l)
284
229
if len(sha1) != 40:
285
trace.warning("bad sha1 in hashcache: %r" % sha1)
230
warning("bad sha1 in hashcache: %r" % sha1)
288
233
fp = tuple(map(long, fields[1:]))
added
removed
bzrlib/hashcache.py
