« back to all changes in this revision
Viewing changes to bzrlib/hashcache.py
- browse files at revision 861
- compare with another revision
- download diff
- download tarball
- view history from revision 861
- Committer: Martin Pool
- Date: 2005-07-08 02:28:58 UTC
- Revision ID: mbp@sourcefrog.net-20050708022858-f8323aaea09ea4f2
- more hash-cache tests
- files added:
- bzrlib/statcache.py
- files removed:
- bzrlib/intset.py
- files modified:
- .bzrignore
added
removed
bzrlib/hashcache.py
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
16
16
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# out-of-date.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.
22
23
# TODO: Perhaps return more details on the file to avoid statting it
24
# again: nonexistent, file type, size, etc
25
26
27
28
CACHE_HEADER = "### bzr hashcache v5\n"
29
30
import os, stat, time
31
32
from bzrlib.osutils import sha_file
33
from bzrlib.trace import mutter, warning
34
17
18
19
20
CACHE_HEADER = "### bzr statcache v5"
35
21
36
22
37
23
def _fingerprint(abspath):
24
import os, stat
25
38
26
try:
39
27
fs = os.lstat(abspath)
40
28
except OSError:
44
32
if stat.S_ISDIR(fs.st_mode):
45
33
return None
46
34
47
# we discard any high precision because it's not reliable; perhaps we
48
# could do better on some systems?
49
return (fs.st_size, long(fs.st_mtime),
50
long(fs.st_ctime), fs.st_ino, fs.st_dev)
35
return (fs.st_size, fs.st_mtime,
36
fs.st_ctime, fs.st_ino, fs.st_dev)
51
37
52
38
53
39
class HashCache(object):
84
70
miss_count
85
71
number of misses (times files have been completely re-read)
86
72
"""
87
needs_write = False
88
89
73
def __init__(self, basedir):
90
74
self.basedir = basedir
91
75
self.hit_count = 0
92
76
self.miss_count = 0
93
77
self.stat_count = 0
94
78
self.danger_count = 0
95
self.removed_count = 0
96
self.update_count = 0
79
97
80
self._cache = {}
98
81
99
82
100
def cache_file_name(self):
101
return os.sep.join([self.basedir, '.bzr', 'stat-cache'])
102
103
104
105
106
83
def clear(self):
107
84
"""Discard all cached information.
108
85
109
86
This does not reset the counters."""
110
if self._cache:
111
self.needs_write = True
112
self._cache = {}
113
114
115
def scan(self):
116
"""Scan all files and remove entries where the cache entry is obsolete.
117
118
Obsolete entries are those where the file has been modified or deleted
119
since the entry was inserted.
120
"""
121
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
122
prep.sort()
123
124
for inum, path, cache_entry in prep:
125
abspath = os.sep.join([self.basedir, path])
126
fp = _fingerprint(abspath)
127
self.stat_count += 1
128
129
cache_fp = cache_entry[1]
130
131
if (not fp) or (cache_fp != fp):
132
# not here or not a regular file anymore
133
self.removed_count += 1
134
self.needs_write = True
135
del self._cache[path]
136
87
self._cache_sha1 = {}
137
88
138
89
139
90
def get_sha1(self, path):
140
"""Return the sha1 of a file.
91
"""Return the hex SHA-1 of the contents of the file at path.
92
93
XXX: If the file does not exist or is not a plain file???
141
94
"""
142
abspath = os.sep.join([self.basedir, path])
95
96
import os, time
97
from bzrlib.osutils import sha_file
98
99
abspath = os.path.join(self.basedir, path)
100
fp = _fingerprint(abspath)
101
c = self._cache.get(path)
102
if c:
103
cache_sha1, cache_fp = c
104
else:
105
cache_sha1, cache_fp = None, None
106
143
107
self.stat_count += 1
144
file_fp = _fingerprint(abspath)
145
146
if not file_fp:
147
# not a regular file or not existing
148
if path in self._cache:
149
self.removed_count += 1
150
self.needs_write = True
151
del self._cache[path]
152
return None
153
154
if path in self._cache:
155
cache_sha1, cache_fp = self._cache[path]
156
else:
157
cache_sha1, cache_fp = None, None
158
159
if cache_fp == file_fp:
108
109
if not fp:
110
# not a regular file
111
return None
112
elif cache_fp and (cache_fp == fp):
160
113
self.hit_count += 1
161
114
return cache_sha1
162
163
self.miss_count += 1
164
digest = sha_file(file(abspath, 'rb', buffering=65000))
165
166
now = int(time.time())
167
if file_fp[1] >= now or file_fp[2] >= now:
168
# changed too recently; can't be cached. we can
169
# return the result and it could possibly be cached
170
# next time.
171
self.danger_count += 1
172
if cache_fp:
173
self.removed_count += 1
174
self.needs_write = True
175
del self._cache[path]
176
115
else:
177
self.update_count += 1
178
self.needs_write = True
179
self._cache[path] = (digest, file_fp)
180
181
return digest
182
183
184
185
186
def write(self):
116
self.miss_count += 1
117
digest = sha_file(file(abspath, 'rb'))
118
119
now = int(time.time())
120
if fp[1] >= now or fp[2] >= now:
121
# changed too recently; can't be cached. we can
122
# return the result and it could possibly be cached
123
# next time.
124
self.danger_count += 1
125
if cache_fp:
126
del self._cache[path]
127
else:
128
self._cache[path] = (digest, fp)
129
130
return digest
131
132
133
134
def write(self, cachefn):
187
135
"""Write contents of cache to file."""
188
136
from atomicfile import AtomicFile
189
137
190
outf = AtomicFile(self.cache_file_name(), 'wb')
138
outf = AtomicFile(cachefn, 'wb')
191
139
try:
192
print >>outf, CACHE_HEADER,
140
outf.write(CACHE_HEADER + '\n')
193
141
194
142
for path, c in self._cache.iteritems():
195
143
assert '//' not in path, path
197
145
outf.write('// ')
198
146
print >>outf, c[0], # hex sha1
199
147
for fld in c[1]:
200
print >>outf, "%d" % fld,
148
print >>outf, fld,
201
149
print >>outf
202
150
203
151
outf.commit()
204
self.needs_write = False
205
152
finally:
206
153
if not outf.closed:
207
154
outf.abort()
208
155
209
210
211
def read(self):
212
"""Reinstate cache from file.
213
214
Overwrites existing cache.
215
216
If the cache file has the wrong version marker, this just clears
217
the cache."""
218
self._cache = {}
219
220
fn = self.cache_file_name()
221
try:
222
inf = file(fn, 'rb', buffering=65000)
223
except IOError, e:
224
mutter("failed to open %s: %s" % (fn, e))
225
return
226
227
228
hdr = inf.readline()
229
if hdr != CACHE_HEADER:
230
mutter('cache header marker not found at top of %s; discarding cache'
231
% fn)
232
return
233
234
for l in inf:
235
pos = l.index('// ')
236
path = l[:pos].decode('utf-8')
237
if path in self._cache:
238
warning('duplicated path %r in cache' % path)
239
continue
240
241
pos += 3
242
fields = l[pos:].split(' ')
243
if len(fields) != 6:
244
warning("bad line in hashcache: %r" % l)
245
continue
246
247
sha1 = fields[0]
248
if len(sha1) != 40:
249
warning("bad sha1 in hashcache: %r" % sha1)
250
continue
251
252
fp = tuple(map(long, fields[1:]))
253
254
self._cache[path] = (sha1, fp)
255
256
self.needs_write = False
257
258
259
260
Loggerhead is a web-based interface for Breezy
Version: 2.0.1