1 import os, stat, struct, tempfile
3 from bup.helpers import *
7 INDEX_HDR = 'BUPI\0\0\0\3'
9 # Use 64-bit integers for mtime/ctime to handle NTFS zero (Y1600) and Y2038.
10 INDEX_SIG = '!QQQqqIIQII20sHII'
12 ENTLEN = struct.calcsize(INDEX_SIG)
14 FOOTLEN = struct.calcsize(FOOTER_SIG)
16 IX_EXISTS = 0x8000 # file exists on filesystem
17 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
18 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
20 class Error(Exception):
25 def __init__(self, ename, parent):
32 (ofs,n) = (f.tell(), len(self.list))
34 count = len(self.list)
35 #log('popping %r with %d entries\n'
36 # % (''.join(self.ename), count))
40 self.parent.count += count + self.count
44 def _golevel(level, f, ename, newentry, tmax):
45 # close nodes back up the tree
47 while ename[:len(level.ename)] != level.ename:
48 n = BlankNewEntry(level.ename[-1], tmax)
50 (n.children_ofs,n.children_n) = level.write(f)
51 level.parent.list.append(n)
54 # create nodes down the tree
55 while len(level.ename) < len(ename):
56 level = Level(ename[:len(level.ename)+1], level)
58 # are we in precisely the right place?
59 assert(ename == level.ename)
60 n = newentry or BlankNewEntry(ename and level.ename[-1] or None, tmax)
61 (n.children_ofs,n.children_n) = level.write(f)
63 level.parent.list.append(n)
70 def __init__(self, basename, name, tmax):
71 self.basename = str(basename)
78 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%d,%d,%s/%s,0x%04x,0x%08x/%d)"
79 % (self.name, self.dev, self.ino, self.nlink,
80 self.ctime, self.mtime, self.uid, self.gid,
81 self.size, self.mode, self.gitmode,
82 self.flags, self.children_ofs, self.children_n))
86 return struct.pack(INDEX_SIG,
87 self.dev, self.ino, self.nlink,
88 self.ctime, self.mtime,
89 self.uid, self.gid, self.size, self.mode,
90 self.gitmode, self.sha, self.flags,
91 self.children_ofs, self.children_n)
92 except (DeprecationWarning, struct.error), e:
93 log('pack error: %s (%r)\n' % (e, self))
96 def from_stat(self, st, tstart):
97 old = (self.dev, self.ino, self.nlink, self.ctime, self.mtime,
98 self.uid, self.gid, self.size, self.flags & IX_EXISTS)
99 new = (st.st_dev, st.st_ino, st.st_nlink,
100 xstat.fstime_floor_secs(st.st_ctime),
101 xstat.fstime_floor_secs(st.st_mtime),
102 st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
105 self.nlink = st.st_nlink
106 self.ctime = xstat.fstime_floor_secs(st.st_ctime)
107 self.mtime = xstat.fstime_floor_secs(st.st_mtime)
110 self.size = st.st_size
111 self.mode = st.st_mode
112 self.flags |= IX_EXISTS
113 if xstat.fstime_floor_secs(st.st_ctime) >= tstart or old != new \
114 or self.sha == EMPTY_SHA or not self.gitmode:
120 self.uid += 0x100000000
122 self.gid += 0x100000000
123 assert(self.uid >= 0)
124 assert(self.gid >= 0)
125 self.mtime = self._fixup_time(self.mtime)
126 self.ctime = self._fixup_time(self.ctime)
128 def _fixup_time(self, t):
129 if self.tmax != None and t > self.tmax:
135 f = IX_HASHVALID|IX_EXISTS
136 return (self.flags & f) == f
138 def invalidate(self):
139 self.flags &= ~IX_HASHVALID
141 def validate(self, gitmode, sha):
144 assert(gitmode+0 == gitmode)
145 self.gitmode = gitmode
147 self.flags |= IX_HASHVALID|IX_EXISTS
150 return not self.is_deleted()
152 def sha_missing(self):
153 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
155 def is_deleted(self):
156 return (self.flags & IX_EXISTS) == 0
158 def set_deleted(self):
159 if self.flags & IX_EXISTS:
160 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
163 return not self.is_fake()
166 return not self.ctime
169 return (cmp(b.name, a.name)
170 or cmp(a.is_valid(), b.is_valid())
171 or cmp(a.is_fake(), b.is_fake()))
174 f.write(self.basename + '\0' + self.packed())
177 class NewEntry(Entry):
178 def __init__(self, basename, name, tmax, dev, ino, nlink, ctime, mtime,
179 uid, gid, size, mode, gitmode, sha, flags,
180 children_ofs, children_n):
181 Entry.__init__(self, basename, name, tmax)
182 (self.dev, self.ino, self.nlink, self.ctime, self.mtime,
183 self.uid, self.gid, self.size, self.mode, self.gitmode, self.sha,
184 self.flags, self.children_ofs, self.children_n
185 ) = (dev, ino, nlink, int(ctime), int(mtime), uid, gid,
186 size, mode, gitmode, sha, flags, children_ofs, children_n)
190 class BlankNewEntry(NewEntry):
191 def __init__(self, basename, tmax):
192 NewEntry.__init__(self, basename, basename, tmax,
193 0, 0, 0, 0, 0, 0, 0, 0, 0,
194 0, EMPTY_SHA, 0, 0, 0)
197 class ExistingEntry(Entry):
198 def __init__(self, parent, basename, name, m, ofs):
199 Entry.__init__(self, basename, name, None)
203 (self.dev, self.ino, self.nlink, self.ctime, self.mtime,
204 self.uid, self.gid, self.size, self.mode, self.gitmode, self.sha,
205 self.flags, self.children_ofs, self.children_n
206 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
208 # effectively, we don't bother messing with IX_SHAMISSING if
209 # not IX_HASHVALID, since it's redundant, and repacking is more
210 # expensive than not repacking.
211 # This is implemented by having sha_missing() check IX_HASHVALID too.
212 def set_sha_missing(self, val):
214 oldval = self.sha_missing() and 1 or 0
216 flag = val and IX_SHAMISSING or 0
217 newflags = (self.flags & (~IX_SHAMISSING)) | flag
218 self.flags = newflags
221 def unset_sha_missing(self, flag):
222 if self.flags & IX_SHAMISSING:
223 self.flags &= ~IX_SHAMISSING
227 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
228 if self.parent and not self.is_valid():
229 self.parent.invalidate()
232 def iter(self, name=None, wantrecurse=None):
234 if dname and not dname.endswith('/'):
236 ofs = self.children_ofs
237 assert(ofs <= len(self._m))
238 assert(self.children_n < 1000000)
239 for i in xrange(self.children_n):
240 eon = self._m.find('\0', ofs)
244 basename = str(buffer(self._m, ofs, eon-ofs))
245 child = ExistingEntry(self, basename, self.name + basename,
248 or child.name.startswith(dname)
249 or child.name.endswith('/') and dname.startswith(child.name)):
250 if not wantrecurse or wantrecurse(child):
251 for e in child.iter(name=name, wantrecurse=wantrecurse):
253 if not name or child.name == name or child.name.startswith(dname):
255 ofs = eon + 1 + ENTLEN
262 def __init__(self, filename):
263 self.filename = filename
265 self.writable = False
269 f = open(filename, 'r+')
271 if e.errno == errno.ENOENT:
276 b = f.read(len(INDEX_HDR))
278 log('warning: %s: header: expected %r, got %r\n'
279 % (filename, INDEX_HDR, b))
281 st = os.fstat(f.fileno())
283 self.m = mmap_readwrite(f)
285 self.count = struct.unpack(FOOTER_SIG,
286 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
292 return int(self.count)
294 def forward_iter(self):
296 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
297 eon = self.m.find('\0', ofs)
301 basename = str(buffer(self.m, ofs, eon-ofs))
302 yield ExistingEntry(None, basename, basename, self.m, eon+1)
303 ofs = eon + 1 + ENTLEN
305 def iter(self, name=None, wantrecurse=None):
306 if len(self.m) > len(INDEX_HDR)+ENTLEN:
308 if dname and not dname.endswith('/'):
310 root = ExistingEntry(None, '/', '/',
311 self.m, len(self.m)-FOOTLEN-ENTLEN)
312 for sub in root.iter(name=name, wantrecurse=wantrecurse):
314 if not dname or dname == root.name:
324 if self.writable and self.m:
329 if self.writable and self.m:
332 self.writable = False
334 def filter(self, prefixes, wantrecurse=None):
335 for (rp, path) in reduce_paths(prefixes):
336 for e in self.iter(rp, wantrecurse=wantrecurse):
337 assert(e.name.startswith(rp))
338 name = path + e.name[len(rp):]
342 # FIXME: this function isn't very generic, because it splits the filename
343 # in an odd way and depends on a terminating '/' to indicate directories.
345 """Split a path into a list of elements of the file system hierarchy."""
347 l = [i+'/' for i in l[:-1]] + l[-1:]
349 l.pop() # extra blank caused by terminating '/'
354 def __init__(self, filename, tmax):
355 self.rootlevel = self.level = Level([], None)
360 self.filename = filename = realpath(filename)
362 (dir,name) = os.path.split(filename)
363 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
364 self.f = os.fdopen(ffd, 'wb', 65536)
365 self.f.write(INDEX_HDR)
375 os.unlink(self.tmpname)
379 self.level = _golevel(self.level, self.f, [], None, self.tmax)
380 self.count = self.rootlevel.count
383 self.f.write(struct.pack(FOOTER_SIG, self.count))
385 assert(self.level == None)
393 os.rename(self.tmpname, self.filename)
395 def _add(self, ename, entry):
396 if self.lastfile and self.lastfile <= ename:
397 raise Error('%r must come before %r'
398 % (''.join(e.name), ''.join(self.lastfile)))
399 self.lastfile = e.name
400 self.level = _golevel(self.level, self.f, ename, entry, self.tmax)
402 def add(self, name, st, hashgen = None):
403 endswith = name.endswith('/')
404 ename = pathsplit(name)
406 #log('add: %r %r\n' % (basename, name))
410 (gitmode, sha) = hashgen(name)
411 flags |= IX_HASHVALID
413 (gitmode, sha) = (0, EMPTY_SHA)
415 isdir = stat.S_ISDIR(st.st_mode)
416 assert(isdir == endswith)
417 e = NewEntry(basename, name, self.tmax,
418 st.st_dev, st.st_ino, st.st_nlink,
419 xstat.fstime_floor_secs(st.st_ctime),
420 xstat.fstime_floor_secs(st.st_mtime),
421 st.st_uid, st.st_gid,
422 st.st_size, st.st_mode, gitmode, sha, flags,
426 e = BlankNewEntry(basename, tmax)
432 def add_ixentry(self, e):
433 e.children_ofs = e.children_n = 0
434 self._add(pathsplit(e.name), e)
436 def new_reader(self):
438 return Reader(self.tmpname)
441 def reduce_paths(paths):
447 if stat.S_ISDIR(st.st_mode):
450 xpaths.append((rp, p))
452 add_error('reduce_paths: %s' % e)
457 for (rp, p) in xpaths:
458 if prev and (prev == rp
459 or (prev.endswith('/') and rp.startswith(prev))):
460 continue # already superceded by previous path
461 paths.append((rp, p))
463 paths.sort(reverse=True)
467 def pfunc(count, total):
468 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
469 def pfinal(count, total):
470 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
471 return merge_iter(iters, 1024, pfunc, pfinal, key='name')