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)
98 new = (st.st_dev, int(st.st_ctime), int(st.st_mtime),
99 st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
101 self.ctime = int(st.st_ctime)
102 self.mtime = int(st.st_mtime)
105 self.size = st.st_size
106 self.mode = st.st_mode
107 self.flags |= IX_EXISTS
108 if int(st.st_ctime) >= tstart or old != new \
109 or self.sha == EMPTY_SHA or not self.gitmode:
115 self.uid += 0x100000000
117 self.gid += 0x100000000
118 assert(self.uid >= 0)
119 assert(self.gid >= 0)
120 if self.mtime < -0x80000000: # can happen in NTFS on 64-bit linux
122 if self.ctime < -0x80000000:
124 if self.mtime > 0x7fffffff:
125 self.mtime = 0x7fffffff
126 if self.ctime > 0x7fffffff:
127 self.ctime = 0x7fffffff
130 f = IX_HASHVALID|IX_EXISTS
131 return (self.flags & f) == f
133 def invalidate(self):
134 self.flags &= ~IX_HASHVALID
136 def validate(self, gitmode, sha):
139 assert(gitmode+0 == gitmode)
140 self.gitmode = gitmode
142 self.flags |= IX_HASHVALID|IX_EXISTS
145 return not self.is_deleted()
147 def sha_missing(self):
148 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
150 def is_deleted(self):
151 return (self.flags & IX_EXISTS) == 0
153 def set_deleted(self):
154 if self.flags & IX_EXISTS:
155 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
158 return not self.is_fake()
161 return not self.ctime
164 return (cmp(b.name, a.name)
165 or cmp(a.is_valid(), b.is_valid())
166 or cmp(a.is_fake(), b.is_fake()))
169 f.write(self.basename + '\0' + self.packed())
172 class NewEntry(Entry):
173 def __init__(self, basename, name, dev, ctime, mtime, uid, gid,
174 size, mode, gitmode, sha, flags, children_ofs, children_n):
175 Entry.__init__(self, basename, name)
176 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
177 self.size, self.mode, self.gitmode, self.sha,
178 self.flags, self.children_ofs, self.children_n
179 ) = (dev, int(ctime), int(mtime), uid, gid,
180 size, mode, gitmode, sha, flags, children_ofs, children_n)
184 class BlankNewEntry(NewEntry):
185 def __init__(self, basename):
186 NewEntry.__init__(self, basename, basename,
188 0, EMPTY_SHA, 0, 0, 0)
191 class ExistingEntry(Entry):
192 def __init__(self, parent, basename, name, m, ofs):
193 Entry.__init__(self, basename, name)
197 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
198 self.size, self.mode, self.gitmode, self.sha,
199 self.flags, self.children_ofs, self.children_n
200 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
202 # effectively, we don't bother messing with IX_SHAMISSING if
203 # not IX_HASHVALID, since it's redundant, and repacking is more
204 # expensive than not repacking.
205 # This is implemented by having sha_missing() check IX_HASHVALID too.
206 def set_sha_missing(self, val):
208 oldval = self.sha_missing() and 1 or 0
210 flag = val and IX_SHAMISSING or 0
211 newflags = (self.flags & (~IX_SHAMISSING)) | flag
212 self.flags = newflags
215 def unset_sha_missing(self, flag):
216 if self.flags & IX_SHAMISSING:
217 self.flags &= ~IX_SHAMISSING
221 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
222 if self.parent and not self.is_valid():
223 self.parent.invalidate()
226 def iter(self, name=None, wantrecurse=None):
228 if dname and not dname.endswith('/'):
230 ofs = self.children_ofs
231 assert(ofs <= len(self._m))
232 assert(self.children_n < 1000000)
233 for i in xrange(self.children_n):
234 eon = self._m.find('\0', ofs)
238 basename = str(buffer(self._m, ofs, eon-ofs))
239 child = ExistingEntry(self, basename, self.name + basename,
242 or child.name.startswith(dname)
243 or child.name.endswith('/') and dname.startswith(child.name)):
244 if not wantrecurse or wantrecurse(child):
245 for e in child.iter(name=name, wantrecurse=wantrecurse):
247 if not name or child.name == name or child.name.startswith(dname):
249 ofs = eon + 1 + ENTLEN
256 def __init__(self, filename):
257 self.filename = filename
259 self.writable = False
263 f = open(filename, 'r+')
265 if e.errno == errno.ENOENT:
270 b = f.read(len(INDEX_HDR))
272 log('warning: %s: header: expected %r, got %r'
273 % (filename, INDEX_HDR, b))
275 st = os.fstat(f.fileno())
277 self.m = mmap_readwrite(f)
279 self.count = struct.unpack(FOOTER_SIG,
280 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
286 return int(self.count)
288 def forward_iter(self):
290 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
291 eon = self.m.find('\0', ofs)
295 basename = str(buffer(self.m, ofs, eon-ofs))
296 yield ExistingEntry(None, basename, basename, self.m, eon+1)
297 ofs = eon + 1 + ENTLEN
299 def iter(self, name=None, wantrecurse=None):
300 if len(self.m) > len(INDEX_HDR)+ENTLEN:
302 if dname and not dname.endswith('/'):
304 root = ExistingEntry(None, '/', '/',
305 self.m, len(self.m)-FOOTLEN-ENTLEN)
306 for sub in root.iter(name=name, wantrecurse=wantrecurse):
308 if not dname or dname == root.name:
318 if self.writable and self.m:
323 if self.writable and self.m:
326 self.writable = False
328 def filter(self, prefixes, wantrecurse=None):
329 for (rp, path) in reduce_paths(prefixes):
330 for e in self.iter(rp, wantrecurse=wantrecurse):
331 assert(e.name.startswith(rp))
332 name = path + e.name[len(rp):]
336 # FIXME: this function isn't very generic, because it splits the filename
337 # in an odd way and depends on a terminating '/' to indicate directories.
339 """Split a path into a list of elements of the file system hierarchy."""
341 l = [i+'/' for i in l[:-1]] + l[-1:]
343 l.pop() # extra blank caused by terminating '/'
348 def __init__(self, filename):
349 self.rootlevel = self.level = Level([], None)
354 self.filename = filename = realpath(filename)
355 (dir,name) = os.path.split(filename)
356 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
357 self.f = os.fdopen(ffd, 'wb', 65536)
358 self.f.write(INDEX_HDR)
368 os.unlink(self.tmpname)
372 self.level = _golevel(self.level, self.f, [], None)
373 self.count = self.rootlevel.count
376 self.f.write(struct.pack(FOOTER_SIG, self.count))
378 assert(self.level == None)
386 os.rename(self.tmpname, self.filename)
388 def _add(self, ename, entry):
389 if self.lastfile and self.lastfile <= ename:
390 raise Error('%r must come before %r'
391 % (''.join(e.name), ''.join(self.lastfile)))
392 self.lastfile = e.name
393 self.level = _golevel(self.level, self.f, ename, entry)
395 def add(self, name, st, hashgen = None):
396 endswith = name.endswith('/')
397 ename = pathsplit(name)
399 #log('add: %r %r\n' % (basename, name))
403 (gitmode, sha) = hashgen(name)
404 flags |= IX_HASHVALID
406 (gitmode, sha) = (0, EMPTY_SHA)
408 isdir = stat.S_ISDIR(st.st_mode)
409 assert(isdir == endswith)
410 e = NewEntry(basename, name, st.st_dev, int(st.st_ctime),
411 int(st.st_mtime), st.st_uid, st.st_gid,
412 st.st_size, st.st_mode, gitmode, sha, flags,
416 e = BlankNewEntry(basename)
422 def add_ixentry(self, e):
423 e.children_ofs = e.children_n = 0
424 self._add(pathsplit(e.name), e)
426 def new_reader(self):
428 return Reader(self.tmpname)
431 def reduce_paths(paths):
437 if stat.S_ISDIR(st.st_mode):
440 xpaths.append((rp, p))
442 add_error('reduce_paths: %s' % e)
447 for (rp, p) in xpaths:
448 if prev and (prev == rp
449 or (prev.endswith('/') and rp.startswith(prev))):
450 continue # already superceded by previous path
451 paths.append((rp, p))
453 paths.sort(reverse=True)
457 def pfunc(count, total):
458 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
459 def pfinal(count, total):
460 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
461 return merge_iter(iters, 1024, pfunc, pfinal, key='name')