1 import metadata, os, stat, struct, tempfile
3 from bup.helpers import *
8 INDEX_HDR = 'BUPI\0\0\0\6'
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 = '!QQQqQqQqQQII20sHIIQ'
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):
34 self._file = open(filename, 'rb')
44 def metadata_at(self, ofs):
46 return metadata.Metadata.read(self._file)
49 class MetaStoreWriter:
50 # For now, we just append to the file, and try to handle any
51 # truncation or corruption somewhat sensibly.
53 def __init__(self, filename):
54 # Map metadata hashes to bupindex.meta offsets.
56 self._filename = filename
58 # FIXME: see how slow this is; does it matter?
59 m_file = open(filename, 'ab+')
64 m = metadata.Metadata.read(m_file)
66 m_encoded = m.encode()
67 self._offsets[m_encoded] = m_off
69 m = metadata.Metadata.read(m_file)
73 log('index metadata in %r appears to be corrupt' % filename)
77 self._file = open(filename, 'ab')
88 def store(self, metadata):
89 meta_encoded = metadata.encode(include_path=False)
90 ofs = self._offsets.get(meta_encoded)
93 ofs = self._file.tell()
94 self._file.write(meta_encoded)
95 self._offsets[meta_encoded] = ofs
100 def __init__(self, ename, parent):
107 (ofs,n) = (f.tell(), len(self.list))
109 count = len(self.list)
110 #log('popping %r with %d entries\n'
111 # % (''.join(self.ename), count))
115 self.parent.count += count + self.count
119 def _golevel(level, f, ename, newentry, metastore, tmax):
120 # close nodes back up the tree
122 default_meta_ofs = metastore.store(metadata.Metadata())
123 while ename[:len(level.ename)] != level.ename:
124 n = BlankNewEntry(level.ename[-1], default_meta_ofs, tmax)
126 (n.children_ofs,n.children_n) = level.write(f)
127 level.parent.list.append(n)
130 # create nodes down the tree
131 while len(level.ename) < len(ename):
132 level = Level(ename[:len(level.ename)+1], level)
134 # are we in precisely the right place?
135 assert(ename == level.ename)
137 BlankNewEntry(ename and level.ename[-1] or None, default_meta_ofs, tmax)
138 (n.children_ofs,n.children_n) = level.write(f)
140 level.parent.list.append(n)
147 def __init__(self, basename, name, meta_ofs, tmax):
148 self.basename = str(basename)
149 self.name = str(name)
150 self.meta_ofs = meta_ofs
152 self.children_ofs = 0
156 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%d,%s/%s,0x%04x,%d,0x%08x/%d)"
157 % (self.name, self.dev, self.ino, self.nlink,
158 self.ctime, self.mtime, self.atime,
159 self.size, self.mode, self.gitmode,
160 self.flags, self.meta_ofs,
161 self.children_ofs, self.children_n))
165 ctime = xstat.nsecs_to_timespec(self.ctime)
166 mtime = xstat.nsecs_to_timespec(self.mtime)
167 atime = xstat.nsecs_to_timespec(self.atime)
168 return struct.pack(INDEX_SIG,
169 self.dev, self.ino, self.nlink,
173 self.size, self.mode,
174 self.gitmode, self.sha, self.flags,
175 self.children_ofs, self.children_n,
177 except (DeprecationWarning, struct.error) as e:
178 log('pack error: %s (%r)\n' % (e, self))
181 def from_stat(self, st, meta_ofs, tstart, check_device=True):
182 old = (self.dev if check_device else 0,
183 self.ino, self.nlink, self.ctime, self.mtime,
184 self.size, self.flags & IX_EXISTS)
185 new = (st.st_dev if check_device else 0,
186 st.st_ino, st.st_nlink, st.st_ctime, st.st_mtime,
187 st.st_size, IX_EXISTS)
190 self.nlink = st.st_nlink
191 self.ctime = st.st_ctime
192 self.mtime = st.st_mtime
193 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:
206 self.mtime = self._fixup_time(self.mtime)
207 self.ctime = self._fixup_time(self.ctime)
209 def _fixup_time(self, t):
210 if self.tmax != None and t > self.tmax:
216 f = IX_HASHVALID|IX_EXISTS
217 return (self.flags & f) == f
219 def invalidate(self):
220 self.flags &= ~IX_HASHVALID
222 def validate(self, gitmode, sha):
225 assert(gitmode+0 == gitmode)
226 self.gitmode = gitmode
228 self.flags |= IX_HASHVALID|IX_EXISTS
231 return not self.is_deleted()
233 def sha_missing(self):
234 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
236 def is_deleted(self):
237 return (self.flags & IX_EXISTS) == 0
239 def set_deleted(self):
240 if self.flags & IX_EXISTS:
241 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
244 return not self.is_fake()
247 return not self.ctime
250 return (cmp(b.name, a.name)
251 or cmp(a.is_valid(), b.is_valid())
252 or cmp(a.is_fake(), b.is_fake()))
255 f.write(self.basename + '\0' + self.packed())
258 class NewEntry(Entry):
259 def __init__(self, basename, name, tmax, dev, ino, nlink,
261 size, mode, gitmode, sha, flags, meta_ofs,
262 children_ofs, children_n):
263 Entry.__init__(self, basename, name, meta_ofs, tmax)
264 (self.dev, self.ino, self.nlink, self.ctime, self.mtime, self.atime,
265 self.size, self.mode, self.gitmode, self.sha,
266 self.flags, self.children_ofs, self.children_n
267 ) = (dev, ino, nlink, ctime, mtime, atime,
268 size, mode, gitmode, sha, flags, children_ofs, children_n)
272 class BlankNewEntry(NewEntry):
273 def __init__(self, basename, meta_ofs, tmax):
274 NewEntry.__init__(self, basename, basename, tmax,
275 0, 0, 0, 0, 0, 0, 0, 0,
276 0, EMPTY_SHA, 0, meta_ofs, 0, 0)
279 class ExistingEntry(Entry):
280 def __init__(self, parent, basename, name, m, ofs):
281 Entry.__init__(self, basename, name, None, None)
285 (self.dev, self.ino, self.nlink,
286 self.ctime, ctime_ns, self.mtime, mtime_ns, self.atime, atime_ns,
287 self.size, self.mode, self.gitmode, self.sha,
288 self.flags, self.children_ofs, self.children_n, self.meta_ofs
289 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
290 self.atime = xstat.timespec_to_nsecs((self.atime, atime_ns))
291 self.mtime = xstat.timespec_to_nsecs((self.mtime, mtime_ns))
292 self.ctime = xstat.timespec_to_nsecs((self.ctime, ctime_ns))
294 # effectively, we don't bother messing with IX_SHAMISSING if
295 # not IX_HASHVALID, since it's redundant, and repacking is more
296 # expensive than not repacking.
297 # This is implemented by having sha_missing() check IX_HASHVALID too.
298 def set_sha_missing(self, val):
300 oldval = self.sha_missing() and 1 or 0
302 flag = val and IX_SHAMISSING or 0
303 newflags = (self.flags & (~IX_SHAMISSING)) | flag
304 self.flags = newflags
307 def unset_sha_missing(self, flag):
308 if self.flags & IX_SHAMISSING:
309 self.flags &= ~IX_SHAMISSING
313 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
314 if self.parent and not self.is_valid():
315 self.parent.invalidate()
318 def iter(self, name=None, wantrecurse=None):
320 if dname and not dname.endswith('/'):
322 ofs = self.children_ofs
323 assert(ofs <= len(self._m))
324 assert(self.children_n < 1000000)
325 for i in xrange(self.children_n):
326 eon = self._m.find('\0', ofs)
330 basename = str(buffer(self._m, ofs, eon-ofs))
331 child = ExistingEntry(self, basename, self.name + basename,
334 or child.name.startswith(dname)
335 or child.name.endswith('/') and dname.startswith(child.name)):
336 if not wantrecurse or wantrecurse(child):
337 for e in child.iter(name=name, wantrecurse=wantrecurse):
339 if not name or child.name == name or child.name.startswith(dname):
341 ofs = eon + 1 + ENTLEN
348 def __init__(self, filename):
349 self.filename = filename
351 self.writable = False
355 f = open(filename, 'r+')
357 if e.errno == errno.ENOENT:
362 b = f.read(len(INDEX_HDR))
364 log('warning: %s: header: expected %r, got %r\n'
365 % (filename, INDEX_HDR, b))
367 st = os.fstat(f.fileno())
369 self.m = mmap_readwrite(f)
371 self.count = struct.unpack(FOOTER_SIG,
372 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
378 return int(self.count)
380 def forward_iter(self):
382 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
383 eon = self.m.find('\0', ofs)
387 basename = str(buffer(self.m, ofs, eon-ofs))
388 yield ExistingEntry(None, basename, basename, self.m, eon+1)
389 ofs = eon + 1 + ENTLEN
391 def iter(self, name=None, wantrecurse=None):
392 if len(self.m) > len(INDEX_HDR)+ENTLEN:
394 if dname and not dname.endswith('/'):
396 root = ExistingEntry(None, '/', '/',
397 self.m, len(self.m)-FOOTLEN-ENTLEN)
398 for sub in root.iter(name=name, wantrecurse=wantrecurse):
400 if not dname or dname == root.name:
410 if self.writable and self.m:
415 if self.writable and self.m:
418 self.writable = False
420 def filter(self, prefixes, wantrecurse=None):
421 for (rp, path) in reduce_paths(prefixes):
422 for e in self.iter(rp, wantrecurse=wantrecurse):
423 assert(e.name.startswith(rp))
424 name = path + e.name[len(rp):]
428 # FIXME: this function isn't very generic, because it splits the filename
429 # in an odd way and depends on a terminating '/' to indicate directories.
431 """Split a path into a list of elements of the file system hierarchy."""
433 l = [i+'/' for i in l[:-1]] + l[-1:]
435 l.pop() # extra blank caused by terminating '/'
440 def __init__(self, filename, metastore, tmax):
441 self.rootlevel = self.level = Level([], None)
446 self.filename = filename = resolve_parent(filename)
447 self.metastore = metastore
449 (dir,name) = os.path.split(filename)
450 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
451 self.f = os.fdopen(ffd, 'wb', 65536)
452 self.f.write(INDEX_HDR)
462 os.unlink(self.tmpname)
466 self.level = _golevel(self.level, self.f, [], None,
467 self.metastore, self.tmax)
468 self.count = self.rootlevel.count
471 self.f.write(struct.pack(FOOTER_SIG, self.count))
473 assert(self.level == None)
481 os.rename(self.tmpname, self.filename)
483 def _add(self, ename, entry):
484 if self.lastfile and self.lastfile <= ename:
485 raise Error('%r must come before %r'
486 % (''.join(ename), ''.join(self.lastfile)))
487 self.lastfile = ename
488 self.level = _golevel(self.level, self.f, ename, entry,
489 self.metastore, self.tmax)
491 def add(self, name, st, meta_ofs, hashgen = None):
492 endswith = name.endswith('/')
493 ename = pathsplit(name)
495 #log('add: %r %r\n' % (basename, name))
499 (gitmode, sha) = hashgen(name)
500 flags |= IX_HASHVALID
502 (gitmode, sha) = (0, EMPTY_SHA)
504 isdir = stat.S_ISDIR(st.st_mode)
505 assert(isdir == endswith)
506 e = NewEntry(basename, name, self.tmax,
507 st.st_dev, st.st_ino, st.st_nlink,
508 st.st_ctime, st.st_mtime, st.st_atime,
509 st.st_size, st.st_mode, gitmode, sha, flags,
513 meta_ofs = self.metastore.store(metadata.Metadata())
514 e = BlankNewEntry(basename, meta_ofs, self.tmax)
520 def add_ixentry(self, e):
521 e.children_ofs = e.children_n = 0
522 self._add(pathsplit(e.name), e)
524 def new_reader(self):
526 return Reader(self.tmpname)
529 def reduce_paths(paths):
532 rp = resolve_parent(p)
535 if stat.S_ISDIR(st.st_mode):
538 xpaths.append((rp, p))
540 add_error('reduce_paths: %s' % e)
545 for (rp, p) in xpaths:
546 if prev and (prev == rp
547 or (prev.endswith('/') and rp.startswith(prev))):
548 continue # already superceded by previous path
549 paths.append((rp, p))
551 paths.sort(reverse=True)
555 def pfunc(count, total):
556 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
557 def pfinal(count, total):
558 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
559 return merge_iter(iters, 1024, pfunc, pfinal, key='name')