1 import os, stat, struct, tempfile
2 from bup.helpers import *
6 INDEX_HDR = 'BUPI\0\0\0\2'
8 # FIXME: guess I should have used 64-bit integers to store the mtime/ctime.
9 # NTFS mtime=0 corresponds to the year 1600, which can't be stored in a 32-bit
10 # time_t. Next time we update the bupindex format, keep that in mind.
11 INDEX_SIG = '!IIIIIQII20sHII'
13 ENTLEN = struct.calcsize(INDEX_SIG)
15 FOOTLEN = struct.calcsize(FOOTER_SIG)
17 IX_EXISTS = 0x8000 # file exists on filesystem
18 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
19 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
21 class Error(Exception):
26 def __init__(self, ename, parent):
33 (ofs,n) = (f.tell(), len(self.list))
35 count = len(self.list)
36 #log('popping %r with %d entries\n'
37 # % (''.join(self.ename), count))
41 self.parent.count += count + self.count
45 def _golevel(level, f, ename, newentry):
46 # close nodes back up the tree
48 while ename[:len(level.ename)] != level.ename:
49 n = BlankNewEntry(level.ename[-1])
51 (n.children_ofs,n.children_n) = level.write(f)
52 level.parent.list.append(n)
55 # create nodes down the tree
56 while len(level.ename) < len(ename):
57 level = Level(ename[:len(level.ename)+1], level)
59 # are we in precisely the right place?
60 assert(ename == level.ename)
61 n = newentry or BlankNewEntry(ename and level.ename[-1] or None)
62 (n.children_ofs,n.children_n) = level.write(f)
64 level.parent.list.append(n)
71 def __init__(self, basename, name):
72 self.basename = str(basename)
78 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%s/%s,0x%04x,0x%08x/%d)"
79 % (self.name, self.dev,
80 self.ctime, self.mtime, self.uid, self.gid,
81 self.size, oct(self.mode), oct(self.gitmode),
82 self.flags, self.children_ofs, self.children_n))
86 return struct.pack(INDEX_SIG,
87 self.dev, self.ctime, self.mtime,
88 self.uid, self.gid, self.size, self.mode,
89 self.gitmode, self.sha, self.flags,
90 self.children_ofs, self.children_n)
91 except DeprecationWarning, e:
92 log('pack error: %s (%r)\n' % (e, self))
95 def from_stat(self, st, tstart):
96 old = (self.dev, self.ctime, self.mtime,
97 self.uid, self.gid, self.size, self.flags & IX_EXISTS)
99 int(st.st_ctime.approx_secs()),
100 int(st.st_mtime.approx_secs()),
101 st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
103 self.ctime = int(st.st_ctime.approx_secs())
104 self.mtime = int(st.st_mtime.approx_secs())
107 self.size = st.st_size
108 self.mode = st.st_mode
109 self.flags |= IX_EXISTS
110 if int(st.st_ctime.approx_secs()) >= tstart or old != new \
111 or self.sha == EMPTY_SHA or not self.gitmode:
117 self.uid += 0x100000000
119 self.gid += 0x100000000
120 assert(self.uid >= 0)
121 assert(self.gid >= 0)
122 if self.mtime < -0x80000000: # can happen in NTFS on 64-bit linux
124 if self.ctime < -0x80000000:
126 if self.mtime > 0x7fffffff:
127 self.mtime = 0x7fffffff
128 if self.ctime > 0x7fffffff:
129 self.ctime = 0x7fffffff
132 f = IX_HASHVALID|IX_EXISTS
133 return (self.flags & f) == f
135 def invalidate(self):
136 self.flags &= ~IX_HASHVALID
138 def validate(self, gitmode, sha):
141 self.gitmode = gitmode
143 self.flags |= IX_HASHVALID|IX_EXISTS
146 return not self.is_deleted()
148 def sha_missing(self):
149 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
151 def is_deleted(self):
152 return (self.flags & IX_EXISTS) == 0
154 def set_deleted(self):
155 if self.flags & IX_EXISTS:
156 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
159 return not self.is_fake()
162 return not self.ctime
165 return (cmp(a.name, b.name)
166 or -cmp(a.is_valid(), b.is_valid())
167 or -cmp(a.is_fake(), b.is_fake()))
170 f.write(self.basename + '\0' + self.packed())
173 class NewEntry(Entry):
174 def __init__(self, basename, name, dev, ctime, mtime, uid, gid,
175 size, mode, gitmode, sha, flags, children_ofs, children_n):
176 Entry.__init__(self, basename, name)
177 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
178 self.size, self.mode, self.gitmode, self.sha,
179 self.flags, self.children_ofs, self.children_n
180 ) = (dev, int(ctime), int(mtime), uid, gid,
181 size, mode, gitmode, sha, flags, children_ofs, children_n)
185 class BlankNewEntry(NewEntry):
186 def __init__(self, basename):
187 NewEntry.__init__(self, basename, basename,
189 0, EMPTY_SHA, 0, 0, 0)
192 class ExistingEntry(Entry):
193 def __init__(self, parent, basename, name, m, ofs):
194 Entry.__init__(self, basename, name)
198 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
199 self.size, self.mode, self.gitmode, self.sha,
200 self.flags, self.children_ofs, self.children_n
201 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
203 # effectively, we don't bother messing with IX_SHAMISSING if
204 # not IX_HASHVALID, since it's redundant, and repacking is more
205 # expensive than not repacking.
206 # This is implemented by having sha_missing() check IX_HASHVALID too.
207 def set_sha_missing(self, val):
209 oldval = self.sha_missing() and 1 or 0
211 flag = val and IX_SHAMISSING or 0
212 newflags = (self.flags & (~IX_SHAMISSING)) | flag
213 self.flags = newflags
216 def unset_sha_missing(self, flag):
217 if self.flags & IX_SHAMISSING:
218 self.flags &= ~IX_SHAMISSING
222 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
223 if self.parent and not self.is_valid():
224 self.parent.invalidate()
227 def iter(self, name=None, wantrecurse=None):
229 if dname and not dname.endswith('/'):
231 ofs = self.children_ofs
232 assert(ofs <= len(self._m))
233 assert(self.children_n < 1000000)
234 for i in xrange(self.children_n):
235 eon = self._m.find('\0', ofs)
239 basename = str(buffer(self._m, ofs, eon-ofs))
240 child = ExistingEntry(self, basename, self.name + basename,
243 or child.name.startswith(dname)
244 or child.name.endswith('/') and dname.startswith(child.name)):
245 if not wantrecurse or wantrecurse(child):
246 for e in child.iter(name=name, wantrecurse=wantrecurse):
248 if not name or child.name == name or child.name.startswith(dname):
250 ofs = eon + 1 + ENTLEN
257 def __init__(self, filename):
258 self.filename = filename
260 self.writable = False
264 f = open(filename, 'r+')
266 if e.errno == errno.ENOENT:
271 b = f.read(len(INDEX_HDR))
273 log('warning: %s: header: expected %r, got %r'
274 % (filename, INDEX_HDR, b))
276 st = os.fstat(f.fileno())
278 self.m = mmap_readwrite(f)
280 self.count = struct.unpack(FOOTER_SIG,
281 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
287 return int(self.count)
289 def forward_iter(self):
291 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
292 eon = self.m.find('\0', ofs)
296 basename = str(buffer(self.m, ofs, eon-ofs))
297 yield ExistingEntry(None, basename, basename, self.m, eon+1)
298 ofs = eon + 1 + ENTLEN
300 def iter(self, name=None, wantrecurse=None):
301 if len(self.m) > len(INDEX_HDR)+ENTLEN:
303 if dname and not dname.endswith('/'):
305 root = ExistingEntry(None, '/', '/',
306 self.m, len(self.m)-FOOTLEN-ENTLEN)
307 for sub in root.iter(name=name, wantrecurse=wantrecurse):
309 if not dname or dname == root.name:
319 if self.writable and self.m:
324 if self.writable and self.m:
327 self.writable = False
329 def filter(self, prefixes, wantrecurse=None):
330 for (rp, path) in reduce_paths(prefixes):
331 for e in self.iter(rp, wantrecurse=wantrecurse):
332 assert(e.name.startswith(rp))
333 name = path + e.name[len(rp):]
337 # FIXME: this function isn't very generic, because it splits the filename
338 # in an odd way and depends on a terminating '/' to indicate directories.
340 """Split a path into a list of elements of the file system hierarchy."""
342 l = [i+'/' for i in l[:-1]] + l[-1:]
344 l.pop() # extra blank caused by terminating '/'
349 def __init__(self, filename):
350 self.rootlevel = self.level = Level([], None)
355 self.filename = filename = realpath(filename)
356 (dir,name) = os.path.split(filename)
357 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
358 self.f = os.fdopen(ffd, 'wb', 65536)
359 self.f.write(INDEX_HDR)
369 os.unlink(self.tmpname)
373 self.level = _golevel(self.level, self.f, [], None)
374 self.count = self.rootlevel.count
377 self.f.write(struct.pack(FOOTER_SIG, self.count))
379 assert(self.level == None)
387 os.rename(self.tmpname, self.filename)
389 def _add(self, ename, entry):
390 if self.lastfile and self.lastfile <= ename:
391 raise Error('%r must come before %r'
392 % (''.join(e.name), ''.join(self.lastfile)))
393 self.lastfile = e.name
394 self.level = _golevel(self.level, self.f, ename, entry)
396 def add(self, name, st, hashgen = None):
397 endswith = name.endswith('/')
398 ename = pathsplit(name)
400 #log('add: %r %r\n' % (basename, name))
404 (gitmode, sha) = hashgen(name)
405 flags |= IX_HASHVALID
407 (gitmode, sha) = (0, EMPTY_SHA)
409 isdir = stat.S_ISDIR(st.st_mode)
410 assert(isdir == endswith)
411 e = NewEntry(basename, name, st.st_dev,
412 int(st.st_ctime.approx_secs()),
413 int(st.st_mtime.approx_secs()),
414 st.st_uid, st.st_gid,
415 st.st_size, st.st_mode, gitmode, sha, flags,
419 e = BlankNewEntry(basename)
425 def add_ixentry(self, e):
426 e.children_ofs = e.children_n = 0
427 self._add(pathsplit(e.name), e)
429 def new_reader(self):
431 return Reader(self.tmpname)
434 def reduce_paths(paths):
440 if stat.S_ISDIR(st.st_mode):
443 xpaths.append((rp, p))
445 add_error('reduce_paths: %s' % e)
450 for (rp, p) in xpaths:
451 if prev and (prev == rp
452 or (prev.endswith('/') and rp.startswith(prev))):
453 continue # already superceded by previous path
454 paths.append((rp, p))
456 paths.sort(reverse=True)
461 def __init__(self, iters):
465 # FIXME: doesn't remove duplicated entries between iters.
466 # That only happens for parent directories, but will mean the
467 # actual iteration returns fewer entries than this function counts.
468 return sum(len(it) for it in self.iters)
472 l = [iter(it) for it in self.iters]
473 l = [(next(it),it) for it in l]
474 l = filter(lambda x: x[0], l)
478 if not (count % 1024):
479 progress('bup: merging indexes (%d/%d)\r' % (count, total))
484 if e.name != lastname:
491 log('bup: merging indexes (%d/%d), done.\n' % (count, total))