2 from __future__ import absolute_import, print_function
3 import errno, os, stat, struct, tempfile
5 from bup import metadata, xstat
6 from bup._helpers import UINT_MAX, bytescmp
7 from bup.compat import pending_raise, range
8 from bup.helpers import (add_error, log, merge_iter, mmap_readwrite,
9 progress, qprogress, resolve_parent, slashappend)
11 EMPTY_SHA = b'\0' * 20
12 FAKE_SHA = b'\x01' * 20
14 INDEX_HDR = b'BUPI\0\0\0\7'
16 # Time values are handled as integer nanoseconds since the epoch in
17 # memory, but are written as xstat/metadata timespecs. This behavior
18 # matches the existing metadata/xstat/.bupm code.
20 # Record times (mtime, ctime, atime) as xstat/metadata timespecs, and
21 # store all of the times in the index so they won't interfere with the
22 # forthcoming metadata cache.
27 'qQ' # ctime_s, ctime_ns
28 'qQ' # mtime_s, mtime_ns
29 'qQ' # atime_s, atime_ns
39 ENTLEN = struct.calcsize(INDEX_SIG)
41 FOOTLEN = struct.calcsize(FOOTER_SIG)
43 IX_EXISTS = 0x8000 # file exists on filesystem
44 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
45 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
47 class Error(Exception):
51 class MetaStoreReader:
52 def __init__(self, filename):
55 self._file = open(filename, 'rb')
69 def __exit__(self, type, value, traceback):
70 with pending_raise(value, rethrow=True):
73 def metadata_at(self, ofs):
75 return metadata.Metadata.read(self._file)
78 class MetaStoreWriter:
79 # For now, we just append to the file, and try to handle any
80 # truncation or corruption somewhat sensibly.
82 def __init__(self, filename):
84 # Map metadata hashes to bupindex.meta offsets.
86 self._filename = filename
88 # FIXME: see how slow this is; does it matter?
89 m_file = open(filename, 'ab+')
94 m = metadata.Metadata.read(m_file)
96 m_encoded = m.encode()
97 self._offsets[m_encoded] = m_off
99 m = metadata.Metadata.read(m_file)
103 log('index metadata in %r appears to be corrupt\n' % filename)
107 self._file = open(filename, 'ab')
121 def __exit__(self, type, value, traceback):
122 with pending_raise(value, rethrow=False):
125 def store(self, metadata):
126 meta_encoded = metadata.encode(include_path=False)
127 ofs = self._offsets.get(meta_encoded)
130 ofs = self._file.tell()
131 self._file.write(meta_encoded)
132 self._offsets[meta_encoded] = ofs
137 def __init__(self, ename, parent):
144 (ofs,n) = (f.tell(), len(self.list))
146 count = len(self.list)
147 #log('popping %r with %d entries\n'
148 # % (''.join(self.ename), count))
152 self.parent.count += count + self.count
156 def _golevel(level, f, ename, newentry, metastore, tmax):
157 # close nodes back up the tree
159 default_meta_ofs = metastore.store(metadata.Metadata())
160 while ename[:len(level.ename)] != level.ename:
161 n = BlankNewEntry(level.ename[-1], default_meta_ofs, tmax)
163 (n.children_ofs,n.children_n) = level.write(f)
164 level.parent.list.append(n)
167 # create nodes down the tree
168 while len(level.ename) < len(ename):
169 level = Level(ename[:len(level.ename)+1], level)
171 # are we in precisely the right place?
172 assert(ename == level.ename)
174 BlankNewEntry(ename and level.ename[-1] or None, default_meta_ofs, tmax)
175 (n.children_ofs,n.children_n) = level.write(f)
177 level.parent.list.append(n)
184 def __init__(self, basename, name, meta_ofs, tmax):
185 assert basename is None or isinstance(basename, bytes)
186 assert name is None or isinstance(name, bytes)
187 self.basename = basename
189 self.meta_ofs = meta_ofs
191 self.children_ofs = 0
195 return ("(%r,0x%04x,%d,%d,%d,%d,%d,%d,%s/%s,0x%04x,%d,0x%08x/%d)"
196 % (self.name, self.dev, self.ino, self.nlink,
197 self.ctime, self.mtime, self.atime,
198 self.size, self.mode, self.gitmode,
199 self.flags, self.meta_ofs,
200 self.children_ofs, self.children_n))
204 ctime = xstat.nsecs_to_timespec(self.ctime)
205 mtime = xstat.nsecs_to_timespec(self.mtime)
206 atime = xstat.nsecs_to_timespec(self.atime)
207 return struct.pack(INDEX_SIG,
208 self.dev, self.ino, self.nlink,
212 self.size, self.mode,
213 self.gitmode, self.sha, self.flags,
214 self.children_ofs, self.children_n,
216 except (DeprecationWarning, struct.error) as e:
217 log('pack error: %s (%r)\n' % (e, self))
220 def stale(self, st, check_device=True):
221 if self.size != st.st_size:
223 if self.mtime != st.st_mtime:
225 if self.sha == EMPTY_SHA:
229 if self.ctime != st.st_ctime:
231 if self.ino != st.st_ino:
233 if self.nlink != st.st_nlink:
235 if not (self.flags & IX_EXISTS):
237 if check_device and (self.dev != st.st_dev):
241 def update_from_stat(self, st, meta_ofs):
242 # Should only be called when the entry is stale(), and
243 # invalidate() should almost certainly be called afterward.
246 self.nlink = st.st_nlink
247 self.ctime = st.st_ctime
248 self.mtime = st.st_mtime
249 self.atime = st.st_atime
250 self.size = st.st_size
251 self.mode = st.st_mode
252 self.flags |= IX_EXISTS
253 self.meta_ofs = meta_ofs
257 self.mtime = self._fixup_time(self.mtime)
258 self.ctime = self._fixup_time(self.ctime)
260 def _fixup_time(self, t):
261 if self.tmax != None and t > self.tmax:
267 f = IX_HASHVALID|IX_EXISTS
268 return (self.flags & f) == f
270 def invalidate(self):
271 self.flags &= ~IX_HASHVALID
273 def validate(self, gitmode, sha):
276 assert(gitmode+0 == gitmode)
277 self.gitmode = gitmode
279 self.flags |= IX_HASHVALID|IX_EXISTS
282 return not self.is_deleted()
284 def sha_missing(self):
285 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
287 def is_deleted(self):
288 return (self.flags & IX_EXISTS) == 0
290 def set_deleted(self):
291 if self.flags & IX_EXISTS:
292 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
295 return not self.is_fake()
298 return not self.ctime
300 def _cmp(self, other):
301 # Note reversed name ordering
302 bc = bytescmp(other.name, self.name)
305 vc = self.is_valid() - other.is_valid()
308 fc = self.is_fake() - other.is_fake()
313 def __eq__(self, other):
314 return self._cmp(other) == 0
316 def __ne__(self, other):
317 return self._cmp(other) != 0
319 def __lt__(self, other):
320 return self._cmp(other) < 0
322 def __gt__(self, other):
323 return self._cmp(other) > 0
325 def __le__(self, other):
326 return self._cmp(other) <= 0
328 def __ge__(self, other):
329 return self._cmp(other) >= 0
332 f.write(self.basename + b'\0' + self.packed())
335 class NewEntry(Entry):
336 def __init__(self, basename, name, tmax, dev, ino, nlink,
338 size, mode, gitmode, sha, flags, meta_ofs,
339 children_ofs, children_n):
340 Entry.__init__(self, basename, name, meta_ofs, tmax)
341 (self.dev, self.ino, self.nlink, self.ctime, self.mtime, self.atime,
342 self.size, self.mode, self.gitmode, self.sha,
343 self.flags, self.children_ofs, self.children_n
344 ) = (dev, ino, nlink, ctime, mtime, atime,
345 size, mode, gitmode, sha, flags, children_ofs, children_n)
349 class BlankNewEntry(NewEntry):
350 def __init__(self, basename, meta_ofs, tmax):
351 NewEntry.__init__(self, basename, basename, tmax,
352 0, 0, 0, 0, 0, 0, 0, 0,
353 0, EMPTY_SHA, 0, meta_ofs, 0, 0)
356 class ExistingEntry(Entry):
357 def __init__(self, parent, basename, name, m, ofs):
358 Entry.__init__(self, basename, name, None, None)
362 (self.dev, self.ino, self.nlink,
363 self.ctime, ctime_ns, self.mtime, mtime_ns, self.atime, atime_ns,
364 self.size, self.mode, self.gitmode, self.sha,
365 self.flags, self.children_ofs, self.children_n, self.meta_ofs
366 ) = struct.unpack(INDEX_SIG, m[ofs : ofs + ENTLEN])
367 self.atime = xstat.timespec_to_nsecs((self.atime, atime_ns))
368 self.mtime = xstat.timespec_to_nsecs((self.mtime, mtime_ns))
369 self.ctime = xstat.timespec_to_nsecs((self.ctime, ctime_ns))
371 # effectively, we don't bother messing with IX_SHAMISSING if
372 # not IX_HASHVALID, since it's redundant, and repacking is more
373 # expensive than not repacking.
374 # This is implemented by having sha_missing() check IX_HASHVALID too.
375 def set_sha_missing(self, val):
377 oldval = self.sha_missing() and 1 or 0
379 flag = val and IX_SHAMISSING or 0
380 newflags = (self.flags & (~IX_SHAMISSING)) | flag
381 self.flags = newflags
384 def unset_sha_missing(self, flag):
385 if self.flags & IX_SHAMISSING:
386 self.flags &= ~IX_SHAMISSING
390 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
391 if self.parent and not self.is_valid():
392 self.parent.invalidate()
395 def iter(self, name=None, wantrecurse=None):
397 if dname and not dname.endswith(b'/'):
399 ofs = self.children_ofs
400 assert(ofs <= len(self._m))
401 assert(self.children_n <= UINT_MAX) # i.e. python struct 'I'
402 for i in range(self.children_n):
403 eon = self._m.find(b'\0', ofs)
407 basename = self._m[ofs : ofs + (eon - ofs)]
408 child = ExistingEntry(self, basename, self.name + basename,
411 or child.name.startswith(dname)
412 or child.name.endswith(b'/') and dname.startswith(child.name)):
413 if not wantrecurse or wantrecurse(child):
414 for e in child.iter(name=name, wantrecurse=wantrecurse):
416 if not name or child.name == name or child.name.startswith(dname):
418 ofs = eon + 1 + ENTLEN
425 def __init__(self, filename):
427 self.filename = filename
429 self.writable = False
433 f = open(filename, 'rb+')
435 if e.errno == errno.ENOENT:
440 b = f.read(len(INDEX_HDR))
442 log('warning: %s: header: expected %r, got %r\n'
443 % (filename, INDEX_HDR, b))
445 st = os.fstat(f.fileno())
447 self.m = mmap_readwrite(f)
449 self.count = struct.unpack(FOOTER_SIG,
450 self.m[st.st_size - FOOTLEN
456 def __exit__(self, type, value, traceback):
457 with pending_raise(value, rethrow=False):
461 return int(self.count)
463 def forward_iter(self):
465 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
466 eon = self.m.find(b'\0', ofs)
470 basename = self.m[ofs : ofs + (eon - ofs)]
471 yield ExistingEntry(None, basename, basename, self.m, eon+1)
472 ofs = eon + 1 + ENTLEN
474 def iter(self, name=None, wantrecurse=None):
475 if len(self.m) > len(INDEX_HDR)+ENTLEN:
477 if dname and not dname.endswith(b'/'):
479 root = ExistingEntry(None, b'/', b'/',
480 self.m, len(self.m)-FOOTLEN-ENTLEN)
481 for sub in root.iter(name=name, wantrecurse=wantrecurse):
483 if not dname or dname == root.name:
489 def find(self, name):
490 return next((e for e in self.iter(name, wantrecurse=lambda x : True)
498 if self.writable and self.m:
504 if self.writable and self.m:
507 self.writable = False
512 def filter(self, prefixes, wantrecurse=None):
513 for (rp, path) in reduce_paths(prefixes):
515 for e in self.iter(rp, wantrecurse=wantrecurse):
517 assert(e.name.startswith(rp))
518 name = path + e.name[len(rp):]
521 # Always return at least the top for each prefix.
522 # Otherwise something like "save x/y" will produce
523 # nothing if x is up to date.
526 name = path + pe.name[len(rp):]
529 # FIXME: this function isn't very generic, because it splits the filename
530 # in an odd way and depends on a terminating '/' to indicate directories.
532 """Split a path into a list of elements of the file system hierarchy."""
534 l = [i + b'/' for i in l[:-1]] + l[-1:]
536 l.pop() # extra blank caused by terminating '/'
541 def __init__(self, filename, metastore, tmax):
543 self.rootlevel = self.level = Level([], None)
548 self.filename = filename = resolve_parent(filename)
549 self.metastore = metastore
551 (dir,name) = os.path.split(filename)
552 ffd, self.tmpname = tempfile.mkstemp(b'.tmp', filename, dir)
553 self.f = os.fdopen(ffd, 'wb', 65536)
554 self.f.write(INDEX_HDR)
559 def __exit__(self, type, value, traceback):
560 with pending_raise(value, rethrow=False):
569 os.unlink(self.tmpname)
573 self.level = _golevel(self.level, self.f, [], None,
574 self.metastore, self.tmax)
575 self.count = self.rootlevel.count
578 self.f.write(struct.pack(FOOTER_SIG, self.count))
580 assert(self.level == None)
589 os.rename(self.tmpname, self.filename)
594 def _add(self, ename, entry):
595 if self.lastfile and self.lastfile <= ename:
596 raise Error('%r must come before %r'
597 % (''.join(ename), ''.join(self.lastfile)))
598 self.lastfile = ename
599 self.level = _golevel(self.level, self.f, ename, entry,
600 self.metastore, self.tmax)
602 def add(self, name, st, meta_ofs, hashgen = None):
603 endswith = name.endswith(b'/')
604 ename = pathsplit(name)
606 #log('add: %r %r\n' % (basename, name))
610 (gitmode, sha) = hashgen(name)
611 flags |= IX_HASHVALID
613 (gitmode, sha) = (0, EMPTY_SHA)
615 isdir = stat.S_ISDIR(st.st_mode)
616 assert(isdir == endswith)
617 e = NewEntry(basename, name, self.tmax,
618 st.st_dev, st.st_ino, st.st_nlink,
619 st.st_ctime, st.st_mtime, st.st_atime,
620 st.st_size, st.st_mode, gitmode, sha, flags,
624 meta_ofs = self.metastore.store(metadata.Metadata())
625 e = BlankNewEntry(basename, meta_ofs, self.tmax)
631 def add_ixentry(self, e):
632 e.children_ofs = e.children_n = 0
633 self._add(pathsplit(e.name), e)
635 def new_reader(self):
637 return Reader(self.tmpname)
640 def _slashappend_or_add_error(p, caller):
641 """Return p, after ensuring it has a single trailing slash if it names
642 a directory, unless there's an OSError, in which case, call
643 add_error() and return None."""
647 add_error('%s: %s' % (caller, e))
650 if stat.S_ISDIR(st.st_mode):
651 return slashappend(p)
655 def unique_resolved_paths(paths):
656 "Return a collection of unique resolved paths."
657 rps = (_slashappend_or_add_error(resolve_parent(p), 'unique_resolved_paths')
659 return frozenset((x for x in rps if x is not None))
662 def reduce_paths(paths):
665 rp = _slashappend_or_add_error(resolve_parent(p), 'reduce_paths')
667 xpaths.append((rp, slashappend(p) if rp.endswith(b'/') else p))
672 for (rp, p) in xpaths:
673 if prev and (prev == rp
674 or (prev.endswith(b'/') and rp.startswith(prev))):
675 continue # already superceded by previous path
676 paths.append((rp, p))
678 paths.sort(reverse=True)
683 def pfunc(count, total):
684 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
685 def pfinal(count, total):
686 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
687 return merge_iter(iters, 1024, pfunc, pfinal, key='name')