1
# Copyright (C) 2005, 2006 Canonical Ltd
1
# Copyright (C) 2005, 2006 by 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
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# out-of-date.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
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# out-of-date.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
20
# it preserves the nice property that any caller will always get up-to-date
21
21
# data except in unavoidable cases.
30
30
CACHE_HEADER = "### bzr hashcache v5\n"
39
filters as _mod_filters,
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
41
FP_MTIME_COLUMN = 1
79
75
number of times files have been retrieved from the cache, avoiding a
83
79
number of misses (times files have been completely re-read)
85
81
needs_write = False
87
def __init__(self, root, cache_file_name, mode=None,
88
content_filter_stack_provider=None):
89
"""Create a hash cache in base dir, and set the file mode to mode.
91
:param content_filter_stack_provider: a function that takes a
92
path (relative to the top of the tree) and a file-id as
93
parameters and returns a stack of ContentFilters.
94
If None, no content filtering is performed.
96
self.root = osutils.safe_unicode(root)
97
self.root_utf8 = self.root.encode('utf8') # where is the filesystem encoding ?
83
def __init__(self, root, cache_file_name, mode=None):
84
"""Create a hash cache in base dir, and set the file mode to mode."""
85
self.root = safe_unicode(root)
99
87
self.miss_count = 0
100
88
self.stat_count = 0
103
91
self.update_count = 0
106
self._cache_file_name = osutils.safe_unicode(cache_file_name)
107
self._filter_provider = content_filter_stack_provider
94
self._cache_file_name = safe_unicode(cache_file_name)
109
96
def cache_file_name(self):
110
97
return self._cache_file_name
121
108
"""Scan all files and remove entries where the cache entry is obsolete.
123
110
Obsolete entries are those where the file has been modified or deleted
124
since the entry was inserted.
111
since the entry was inserted.
126
113
# FIXME optimisation opportunity, on linux [and check other oses]:
127
114
# rather than iteritems order, stat in inode order.
128
115
prep = [(ce[1][3], path, ce) for (path, ce) in self._cache.iteritems()]
131
118
for inum, path, cache_entry in prep:
132
abspath = osutils.pathjoin(self.root, path)
119
abspath = pathjoin(self.root, path)
133
120
fp = self._fingerprint(abspath)
134
121
self.stat_count += 1
136
123
cache_fp = cache_entry[1]
138
125
if (not fp) or (cache_fp != fp):
139
126
# not here or not a regular file anymore
140
127
self.removed_count += 1
141
128
self.needs_write = True
142
129
del self._cache[path]
144
def get_sha1(self, path, stat_value=None):
131
def get_sha1(self, path):
145
132
"""Return the sha1 of a file.
147
if path.__class__ is str:
148
abspath = osutils.pathjoin(self.root_utf8, path)
150
abspath = osutils.pathjoin(self.root, path)
134
abspath = pathjoin(self.root, path)
151
135
self.stat_count += 1
152
file_fp = self._fingerprint(abspath, stat_value)
136
file_fp = self._fingerprint(abspath)
155
139
# not a regular file or not existing
156
140
if path in self._cache:
157
141
self.removed_count += 1
158
142
self.needs_write = True
159
143
del self._cache[path]
162
146
if path in self._cache:
163
147
cache_sha1, cache_fp = self._cache[path]
169
153
## mutter("now = %s", time.time())
170
154
self.hit_count += 1
171
155
return cache_sha1
173
157
self.miss_count += 1
175
159
mode = file_fp[FP_MODE_COLUMN]
176
160
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)
161
digest = self._really_sha1_file(abspath)
182
162
elif stat.S_ISLNK(mode):
183
target = osutils.readlink(osutils.safe_unicode(abspath))
184
digest = osutils.sha_string(target.encode('UTF-8'))
163
digest = sha.new(os.readlink(abspath)).hexdigest()
186
raise errors.BzrError("file %r: unknown file stat mode: %o"
165
raise BzrError("file %r: unknown file stat mode: %o"%(abspath,mode))
189
167
# window of 3 seconds to allow for 2s resolution on windows,
190
168
# unsynchronized file servers, etc.
217
195
self._cache[path] = (digest, file_fp)
220
def _really_sha1_file(self, abspath, filters):
198
def _really_sha1_file(self, abspath):
221
199
"""Calculate the SHA1 of a file by reading the full text"""
222
return _mod_filters.internal_size_sha_file_byname(abspath, filters)[1]
200
return sha_file(file(abspath, 'rb', buffering=65000))
225
203
"""Write contents of cache to file."""
226
outf = atomicfile.AtomicFile(self.cache_file_name(), 'wb',
204
outf = AtomicFile(self.cache_file_name(), 'wb', new_mode=self._mode)
229
outf.write(CACHE_HEADER)
206
print >>outf, CACHE_HEADER,
231
208
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))
209
assert '//' not in path, path
210
outf.write(path.encode('utf-8'))
212
print >>outf, c[0], # hex sha1
214
print >>outf, "%d" % fld,
237
217
self.needs_write = False
238
218
## mutter("write hash cache: %s hits=%d misses=%d stat=%d recent=%d updates=%d",
240
220
## self.stat_count,
241
221
## self.danger_count, self.update_count)
246
227
"""Reinstate cache from file.
248
229
Overwrites existing cache.
250
If the cache file has the wrong version marker, this just clears
231
If the cache file has the wrong version marker, this just clears
256
237
inf = file(fn, 'rb', buffering=65000)
257
238
except IOError, e:
258
trace.mutter("failed to open %s: %s", fn, e)
239
mutter("failed to open %s: %s", fn, e)
259
240
# better write it now so it is valid
260
241
self.needs_write = True
263
244
hdr = inf.readline()
264
245
if hdr != CACHE_HEADER:
265
trace.mutter('cache header marker not found at top of %s;'
266
' discarding cache', fn)
246
mutter('cache header marker not found at top of %s;'
247
' discarding cache', fn)
267
248
self.needs_write = True
271
252
pos = l.index('// ')
272
253
path = l[:pos].decode('utf-8')
273
254
if path in self._cache:
274
trace.warning('duplicated path %r in cache' % path)
255
warning('duplicated path %r in cache' % path)
278
259
fields = l[pos:].split(' ')
279
260
if len(fields) != 7:
280
trace.warning("bad line in hashcache: %r" % l)
261
warning("bad line in hashcache: %r" % l)
284
265
if len(sha1) != 40:
285
trace.warning("bad sha1 in hashcache: %r" % sha1)
266
warning("bad sha1 in hashcache: %r" % sha1)
288
269
fp = tuple(map(long, fields[1:]))
298
279
undetectably modified and so can't be cached.
300
281
return int(time.time()) - 3
302
def _fingerprint(self, abspath, stat_value=None):
303
if stat_value is None:
305
stat_value = os.lstat(abspath)
307
# might be missing, etc
309
if stat.S_ISDIR(stat_value.st_mode):
283
def _fingerprint(self, abspath):
285
fs = os.lstat(abspath)
287
# might be missing, etc
289
if stat.S_ISDIR(fs.st_mode):
311
291
# we discard any high precision because it's not reliable; perhaps we
312
292
# could do better on some systems?
313
return (stat_value.st_size, long(stat_value.st_mtime),
314
long(stat_value.st_ctime), stat_value.st_ino,
315
stat_value.st_dev, stat_value.st_mode)
293
return (fs.st_size, long(fs.st_mtime),
294
long(fs.st_ctime), fs.st_ino, fs.st_dev, fs.st_mode)
added
removed
bzrlib/hashcache.py
