1 import metadata, os, stat, struct, tempfile
3 from bup.helpers import *
8 INDEX_HDR = 'BUPI\0\0\0\5'
10 # Time values are handled as integer nanoseconds since the epoch in
11 # memory, but are written as xstat/metadata timespecs. This behavior
12 # matches the existing metadata/xstat/.bupm code.
14 # Record times (mtime, ctime, atime) as xstat/metadata timespecs, and
15 # store all of the times in the index so they won't interfere with the
16 # forthcoming metadata cache.
17 INDEX_SIG = '!QQQqQqQqQIIQII20sHIIQ'
19 ENTLEN = struct.calcsize(INDEX_SIG)
21 FOOTLEN = struct.calcsize(FOOTER_SIG)
23 IX_EXISTS = 0x8000 # file exists on filesystem
24 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
25 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
27 class Error(Exception):
31 class MetaStoreReader:
32 def __init__(self, filename):
33 self._file = open(filename, 'rb')
43 def metadata_at(self, ofs):
45 return metadata.Metadata.read(self._file)
48 class MetaStoreWriter:
49 # For now, we just append to the file, and try to handle any
50 # truncation or corruption somewhat sensibly.
52 def __init__(self, filename):
53 # Map metadata hashes to bupindex.meta offsets.
55 self._filename = filename
56 # FIXME: see how slow this is; does it matter?
57 m_file = open(filename, 'ab+')
62 m = metadata.Metadata.read(m_file)
64 m_encoded = m.encode()
65 self._offsets[m_encoded] = m_off
67 m = metadata.Metadata.read(m_file)
71 log('index metadata in %r appears to be corrupt' % filename)
75 self._file = open(filename, 'ab')
86 def store(self, metadata):
87 meta_encoded = metadata.encode(include_path=False)
88 ofs = self._offsets.get(meta_encoded)
91 ofs = self._file.tell()
92 self._file.write(meta_encoded)
93 self._offsets[meta_encoded] = ofs
98 def __init__(self, ename, parent):
105 (ofs,n) = (f.tell(), len(self.list))
107 count = len(self.list)
108 #log('popping %r with %d entries\n'
109 # % (''.join(self.ename), count))
113 self.parent.count += count + self.count
117 def _golevel(level, f, ename, newentry, metastore, tmax):
118 # close nodes back up the tree
120 default_meta_ofs = metastore.store(metadata.Metadata())
121 while ename[:len(level.ename)] != level.ename:
122 n = BlankNewEntry(level.ename[-1], default_meta_ofs, tmax)
124 (n.children_ofs,n.children_n) = level.write(f)
125 level.parent.list.append(n)
128 # create nodes down the tree
129 while len(level.ename) < len(ename):
130 level = Level(ename[:len(level.ename)+1], level)
132 # are we in precisely the right place?
133 assert(ename == level.ename)
135 BlankNewEntry(ename and level.ename[-1] or None, default_meta_ofs, tmax)
136 (n.children_ofs,n.children_n) = level.write(f)
138 level.parent.list.append(n)
145 def __init__(self, basename, name, meta_ofs, tmax):
146 self.basename = str(basename)
147 self.name = str(name)
148 self.meta_ofs = meta_ofs
150 self.children_ofs = 0
154 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%d,%d,%d,%s/%s,0x%04x,%d,0x%08x/%d)"
155 % (self.name, self.dev, self.ino, self.nlink,
156 self.ctime, self.mtime, self.atime, self.uid, self.gid,
157 self.size, self.mode, self.gitmode,
158 self.flags, self.meta_ofs,
159 self.children_ofs, self.children_n))
163 ctime = xstat.nsecs_to_timespec(self.ctime)
164 mtime = xstat.nsecs_to_timespec(self.mtime)
165 atime = xstat.nsecs_to_timespec(self.atime)
166 return struct.pack(INDEX_SIG,
167 self.dev, self.ino, self.nlink,
171 self.uid, self.gid, self.size, self.mode,
172 self.gitmode, self.sha, self.flags,
173 self.children_ofs, self.children_n,
175 except (DeprecationWarning, struct.error), e:
176 log('pack error: %s (%r)\n' % (e, self))
179 def from_stat(self, st, meta_ofs, tstart, check_device=True):
180 old = (self.dev if check_device else 0,
181 self.ino, self.nlink, self.ctime, self.mtime,
182 self.uid, self.gid, self.size, self.flags & IX_EXISTS)
183 new = (st.st_dev if check_device else 0,
184 st.st_ino, st.st_nlink, st.st_ctime, st.st_mtime,
185 st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
188 self.nlink = st.st_nlink
189 self.ctime = st.st_ctime
190 self.mtime = st.st_mtime
191 self.atime = st.st_atime
194 self.size = st.st_size
195 self.mode = st.st_mode
196 self.flags |= IX_EXISTS
197 self.meta_ofs = meta_ofs
198 # Check that the ctime's "second" is at or after tstart's.
199 ctime_sec_in_ns = xstat.fstime_floor_secs(st.st_ctime) * 10**9
200 if ctime_sec_in_ns >= tstart or old != new \
201 or self.sha == EMPTY_SHA or not self.gitmode:
207 self.uid += 0x100000000
209 self.gid += 0x100000000
210 assert(self.uid >= 0)
211 assert(self.gid >= 0)
212 self.mtime = self._fixup_time(self.mtime)
213 self.ctime = self._fixup_time(self.ctime)
215 def _fixup_time(self, t):
216 if self.tmax != None and t > self.tmax:
222 f = IX_HASHVALID|IX_EXISTS
223 return (self.flags & f) == f
225 def invalidate(self):
226 self.flags &= ~IX_HASHVALID
228 def validate(self, gitmode, sha):
231 assert(gitmode+0 == gitmode)
232 self.gitmode = gitmode
234 self.flags |= IX_HASHVALID|IX_EXISTS
237 return not self.is_deleted()
239 def sha_missing(self):
240 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
242 def is_deleted(self):
243 return (self.flags & IX_EXISTS) == 0
245 def set_deleted(self):
246 if self.flags & IX_EXISTS:
247 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
250 return not self.is_fake()
253 return not self.ctime
256 return (cmp(b.name, a.name)
257 or cmp(a.is_valid(), b.is_valid())
258 or cmp(a.is_fake(), b.is_fake()))
261 f.write(self.basename + '\0' + self.packed())
264 class NewEntry(Entry):
265 def __init__(self, basename, name, tmax, dev, ino, nlink,
267 uid, gid, size, mode, gitmode, sha, flags, meta_ofs,
268 children_ofs, children_n):
269 Entry.__init__(self, basename, name, meta_ofs, tmax)
270 (self.dev, self.ino, self.nlink, self.ctime, self.mtime, self.atime,
271 self.uid, self.gid, self.size, self.mode, self.gitmode, self.sha,
272 self.flags, self.children_ofs, self.children_n
273 ) = (dev, ino, nlink, ctime, mtime, atime, uid, gid,
274 size, mode, gitmode, sha, flags, children_ofs, children_n)
278 class BlankNewEntry(NewEntry):
279 def __init__(self, basename, meta_ofs, tmax):
280 NewEntry.__init__(self, basename, basename, tmax,
281 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
282 0, EMPTY_SHA, 0, meta_ofs, 0, 0)
285 class ExistingEntry(Entry):
286 def __init__(self, parent, basename, name, m, ofs):
287 Entry.__init__(self, basename, name, None, None)
291 (self.dev, self.ino, self.nlink,
292 self.ctime, ctime_ns, self.mtime, mtime_ns, self.atime, atime_ns,
293 self.uid, self.gid, self.size, self.mode, self.gitmode, self.sha,
294 self.flags, self.children_ofs, self.children_n, self.meta_ofs
295 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
296 self.atime = xstat.timespec_to_nsecs((self.atime, atime_ns))
297 self.mtime = xstat.timespec_to_nsecs((self.mtime, mtime_ns))
298 self.ctime = xstat.timespec_to_nsecs((self.ctime, ctime_ns))
300 # effectively, we don't bother messing with IX_SHAMISSING if
301 # not IX_HASHVALID, since it's redundant, and repacking is more
302 # expensive than not repacking.
303 # This is implemented by having sha_missing() check IX_HASHVALID too.
304 def set_sha_missing(self, val):
306 oldval = self.sha_missing() and 1 or 0
308 flag = val and IX_SHAMISSING or 0
309 newflags = (self.flags & (~IX_SHAMISSING)) | flag
310 self.flags = newflags
313 def unset_sha_missing(self, flag):
314 if self.flags & IX_SHAMISSING:
315 self.flags &= ~IX_SHAMISSING
319 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
320 if self.parent and not self.is_valid():
321 self.parent.invalidate()
324 def iter(self, name=None, wantrecurse=None):
326 if dname and not dname.endswith('/'):
328 ofs = self.children_ofs
329 assert(ofs <= len(self._m))
330 assert(self.children_n < 1000000)
331 for i in xrange(self.children_n):
332 eon = self._m.find('\0', ofs)
336 basename = str(buffer(self._m, ofs, eon-ofs))
337 child = ExistingEntry(self, basename, self.name + basename,
340 or child.name.startswith(dname)
341 or child.name.endswith('/') and dname.startswith(child.name)):
342 if not wantrecurse or wantrecurse(child):
343 for e in child.iter(name=name, wantrecurse=wantrecurse):
345 if not name or child.name == name or child.name.startswith(dname):
347 ofs = eon + 1 + ENTLEN
354 def __init__(self, filename):
355 self.filename = filename
357 self.writable = False
361 f = open(filename, 'r+')
363 if e.errno == errno.ENOENT:
368 b = f.read(len(INDEX_HDR))
370 log('warning: %s: header: expected %r, got %r\n'
371 % (filename, INDEX_HDR, b))
373 st = os.fstat(f.fileno())
375 self.m = mmap_readwrite(f)
377 self.count = struct.unpack(FOOTER_SIG,
378 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
384 return int(self.count)
386 def forward_iter(self):
388 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
389 eon = self.m.find('\0', ofs)
393 basename = str(buffer(self.m, ofs, eon-ofs))
394 yield ExistingEntry(None, basename, basename, self.m, eon+1)
395 ofs = eon + 1 + ENTLEN
397 def iter(self, name=None, wantrecurse=None):
398 if len(self.m) > len(INDEX_HDR)+ENTLEN:
400 if dname and not dname.endswith('/'):
402 root = ExistingEntry(None, '/', '/',
403 self.m, len(self.m)-FOOTLEN-ENTLEN)
404 for sub in root.iter(name=name, wantrecurse=wantrecurse):
406 if not dname or dname == root.name:
416 if self.writable and self.m:
421 if self.writable and self.m:
424 self.writable = False
426 def filter(self, prefixes, wantrecurse=None):
427 for (rp, path) in reduce_paths(prefixes):
428 for e in self.iter(rp, wantrecurse=wantrecurse):
429 assert(e.name.startswith(rp))
430 name = path + e.name[len(rp):]
434 # FIXME: this function isn't very generic, because it splits the filename
435 # in an odd way and depends on a terminating '/' to indicate directories.
437 """Split a path into a list of elements of the file system hierarchy."""
439 l = [i+'/' for i in l[:-1]] + l[-1:]
441 l.pop() # extra blank caused by terminating '/'
446 def __init__(self, filename, metastore, tmax):
447 self.rootlevel = self.level = Level([], None)
452 self.filename = filename = realpath(filename)
453 self.metastore = metastore
455 (dir,name) = os.path.split(filename)
456 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
457 self.f = os.fdopen(ffd, 'wb', 65536)
458 self.f.write(INDEX_HDR)
468 os.unlink(self.tmpname)
472 self.level = _golevel(self.level, self.f, [], None,
473 self.metastore, self.tmax)
474 self.count = self.rootlevel.count
477 self.f.write(struct.pack(FOOTER_SIG, self.count))
479 assert(self.level == None)
487 os.rename(self.tmpname, self.filename)
489 def _add(self, ename, entry):
490 if self.lastfile and self.lastfile <= ename:
491 raise Error('%r must come before %r'
492 % (''.join(ename), ''.join(self.lastfile)))
493 self.lastfile = ename
494 self.level = _golevel(self.level, self.f, ename, entry,
495 self.metastore, self.tmax)
497 def add(self, name, st, meta_ofs, hashgen = None):
498 endswith = name.endswith('/')
499 ename = pathsplit(name)
501 #log('add: %r %r\n' % (basename, name))
505 (gitmode, sha) = hashgen(name)
506 flags |= IX_HASHVALID
508 (gitmode, sha) = (0, EMPTY_SHA)
510 isdir = stat.S_ISDIR(st.st_mode)
511 assert(isdir == endswith)
512 e = NewEntry(basename, name, self.tmax,
513 st.st_dev, st.st_ino, st.st_nlink,
514 st.st_ctime, st.st_mtime, st.st_atime,
515 st.st_uid, st.st_gid,
516 st.st_size, st.st_mode, gitmode, sha, flags,
520 meta_ofs = self.metastore.store(metadata.Metadata())
521 e = BlankNewEntry(basename, meta_ofs, self.tmax)
527 def add_ixentry(self, e):
528 e.children_ofs = e.children_n = 0
529 self._add(pathsplit(e.name), e)
531 def new_reader(self):
533 return Reader(self.tmpname)
536 def reduce_paths(paths):
542 if stat.S_ISDIR(st.st_mode):
545 xpaths.append((rp, p))
547 add_error('reduce_paths: %s' % e)
552 for (rp, p) in xpaths:
553 if prev and (prev == rp
554 or (prev.endswith('/') and rp.startswith(prev))):
555 continue # already superceded by previous path
556 paths.append((rp, p))
558 paths.sort(reverse=True)
562 def pfunc(count, total):
563 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
564 def pfinal(count, total):
565 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
566 return merge_iter(iters, 1024, pfunc, pfinal, key='name')