1 import os, stat, struct, tempfile
2 from bup.helpers import *
6 INDEX_HDR = 'BUPI\0\0\0\2'
7 INDEX_SIG = '!IIIIIQII20sHII'
8 ENTLEN = struct.calcsize(INDEX_SIG)
10 FOOTLEN = struct.calcsize(FOOTER_SIG)
12 IX_EXISTS = 0x8000 # file exists on filesystem
13 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
14 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
16 class Error(Exception):
21 def __init__(self, ename, parent):
28 (ofs,n) = (f.tell(), len(self.list))
30 count = len(self.list)
31 #log('popping %r with %d entries\n'
32 # % (''.join(self.ename), count))
36 self.parent.count += count + self.count
40 def _golevel(level, f, ename, newentry):
41 # close nodes back up the tree
43 while ename[:len(level.ename)] != level.ename:
44 n = BlankNewEntry(level.ename[-1])
46 (n.children_ofs,n.children_n) = level.write(f)
47 level.parent.list.append(n)
50 # create nodes down the tree
51 while len(level.ename) < len(ename):
52 level = Level(ename[:len(level.ename)+1], level)
54 # are we in precisely the right place?
55 assert(ename == level.ename)
56 n = newentry or BlankNewEntry(ename and level.ename[-1] or None)
57 (n.children_ofs,n.children_n) = level.write(f)
59 level.parent.list.append(n)
66 def __init__(self, basename, name):
67 self.basename = str(basename)
73 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%s/%s,0x%04x,0x%08x/%d)"
74 % (self.name, self.dev,
75 self.ctime, self.mtime, self.uid, self.gid,
76 self.size, oct(self.mode), oct(self.gitmode),
77 self.flags, self.children_ofs, self.children_n))
81 return struct.pack(INDEX_SIG,
82 self.dev, self.ctime, self.mtime,
83 self.uid, self.gid, self.size, self.mode,
84 self.gitmode, self.sha, self.flags,
85 self.children_ofs, self.children_n)
86 except DeprecationWarning, e:
87 log('pack error: %s (%r)\n' % (e, self))
90 def from_stat(self, st, tstart):
91 old = (self.dev, self.ctime, self.mtime,
92 self.uid, self.gid, self.size, self.flags & IX_EXISTS)
93 new = (st.st_dev, int(st.st_ctime), int(st.st_mtime),
94 st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
96 self.ctime = int(st.st_ctime)
97 self.mtime = int(st.st_mtime)
100 self.size = st.st_size
101 self.mode = st.st_mode
102 self.flags |= IX_EXISTS
103 if int(st.st_ctime) >= tstart or old != new \
104 or self.sha == EMPTY_SHA or not self.gitmode:
110 self.uid += 0x100000000
112 self.gid += 0x100000000
113 assert(self.uid >= 0)
114 assert(self.gid >= 0)
117 f = IX_HASHVALID|IX_EXISTS
118 return (self.flags & f) == f
120 def invalidate(self):
121 self.flags &= ~IX_HASHVALID
123 def validate(self, gitmode, sha):
126 self.gitmode = gitmode
128 self.flags |= IX_HASHVALID|IX_EXISTS
131 return not self.is_deleted()
133 def sha_missing(self):
134 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
136 def is_deleted(self):
137 return (self.flags & IX_EXISTS) == 0
139 def set_deleted(self):
140 if self.flags & IX_EXISTS:
141 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
144 return not self.is_fake()
147 return not self.ctime
150 return (cmp(a.name, b.name)
151 or -cmp(a.is_valid(), b.is_valid())
152 or -cmp(a.is_fake(), b.is_fake()))
155 f.write(self.basename + '\0' + self.packed())
158 class NewEntry(Entry):
159 def __init__(self, basename, name, dev, ctime, mtime, uid, gid,
160 size, mode, gitmode, sha, flags, children_ofs, children_n):
161 Entry.__init__(self, basename, name)
162 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
163 self.size, self.mode, self.gitmode, self.sha,
164 self.flags, self.children_ofs, self.children_n
165 ) = (dev, int(ctime), int(mtime), uid, gid,
166 size, mode, gitmode, sha, flags, children_ofs, children_n)
170 class BlankNewEntry(NewEntry):
171 def __init__(self, basename):
172 NewEntry.__init__(self, basename, basename,
174 0, EMPTY_SHA, 0, 0, 0)
177 class ExistingEntry(Entry):
178 def __init__(self, parent, basename, name, m, ofs):
179 Entry.__init__(self, basename, name)
183 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
184 self.size, self.mode, self.gitmode, self.sha,
185 self.flags, self.children_ofs, self.children_n
186 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
188 # effectively, we don't bother messing with IX_SHAMISSING if
189 # not IX_HASHVALID, since it's redundant, and repacking is more
190 # expensive than not repacking.
191 # This is implemented by having sha_missing() check IX_HASHVALID too.
192 def set_sha_missing(self, val):
194 oldval = self.sha_missing() and 1 or 0
196 flag = val and IX_SHAMISSING or 0
197 newflags = (self.flags & (~IX_SHAMISSING)) | flag
198 self.flags = newflags
201 def unset_sha_missing(self, flag):
202 if self.flags & IX_SHAMISSING:
203 self.flags &= ~IX_SHAMISSING
207 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
208 if self.parent and not self.is_valid():
209 self.parent.invalidate()
212 def iter(self, name=None, wantrecurse=None):
214 if dname and not dname.endswith('/'):
216 ofs = self.children_ofs
217 assert(ofs <= len(self._m))
218 assert(self.children_n < 1000000)
219 for i in xrange(self.children_n):
220 eon = self._m.find('\0', ofs)
224 basename = str(buffer(self._m, ofs, eon-ofs))
225 child = ExistingEntry(self, basename, self.name + basename,
228 or child.name.startswith(dname)
229 or child.name.endswith('/') and dname.startswith(child.name)):
230 if not wantrecurse or wantrecurse(child):
231 for e in child.iter(name=name, wantrecurse=wantrecurse):
233 if not name or child.name == name or child.name.startswith(dname):
235 ofs = eon + 1 + ENTLEN
242 def __init__(self, filename):
243 self.filename = filename
245 self.writable = False
249 f = open(filename, 'r+')
251 if e.errno == errno.ENOENT:
256 b = f.read(len(INDEX_HDR))
258 log('warning: %s: header: expected %r, got %r'
259 % (filename, INDEX_HDR, b))
261 st = os.fstat(f.fileno())
263 self.m = mmap_readwrite(f)
265 self.count = struct.unpack(FOOTER_SIG,
266 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
272 return int(self.count)
274 def forward_iter(self):
276 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
277 eon = self.m.find('\0', ofs)
281 basename = str(buffer(self.m, ofs, eon-ofs))
282 yield ExistingEntry(None, basename, basename, self.m, eon+1)
283 ofs = eon + 1 + ENTLEN
285 def iter(self, name=None, wantrecurse=None):
286 if len(self.m) > len(INDEX_HDR)+ENTLEN:
288 if dname and not dname.endswith('/'):
290 root = ExistingEntry(None, '/', '/',
291 self.m, len(self.m)-FOOTLEN-ENTLEN)
292 for sub in root.iter(name=name, wantrecurse=wantrecurse):
294 if not dname or dname == root.name:
304 if self.writable and self.m:
309 if self.writable and self.m:
312 self.writable = False
314 def filter(self, prefixes, wantrecurse=None):
315 for (rp, path) in reduce_paths(prefixes):
316 for e in self.iter(rp, wantrecurse=wantrecurse):
317 assert(e.name.startswith(rp))
318 name = path + e.name[len(rp):]
322 # FIXME: this function isn't very generic, because it splits the filename
323 # in an odd way and depends on a terminating '/' to indicate directories.
325 """Split a path into a list of elements of the file system hierarchy."""
327 l = [i+'/' for i in l[:-1]] + l[-1:]
329 l.pop() # extra blank caused by terminating '/'
334 def __init__(self, filename):
335 self.rootlevel = self.level = Level([], None)
340 self.filename = filename = realpath(filename)
341 (dir,name) = os.path.split(filename)
342 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
343 self.f = os.fdopen(ffd, 'wb', 65536)
344 self.f.write(INDEX_HDR)
354 os.unlink(self.tmpname)
358 self.level = _golevel(self.level, self.f, [], None)
359 self.count = self.rootlevel.count
362 self.f.write(struct.pack(FOOTER_SIG, self.count))
364 assert(self.level == None)
372 os.rename(self.tmpname, self.filename)
374 def _add(self, ename, entry):
375 if self.lastfile and self.lastfile <= ename:
376 raise Error('%r must come before %r'
377 % (''.join(e.name), ''.join(self.lastfile)))
378 self.lastfile = e.name
379 self.level = _golevel(self.level, self.f, ename, entry)
381 def add(self, name, st, hashgen = None):
382 endswith = name.endswith('/')
383 ename = pathsplit(name)
385 #log('add: %r %r\n' % (basename, name))
389 (gitmode, sha) = hashgen(name)
390 flags |= IX_HASHVALID
392 (gitmode, sha) = (0, EMPTY_SHA)
394 isdir = stat.S_ISDIR(st.st_mode)
395 assert(isdir == endswith)
396 e = NewEntry(basename, name, st.st_dev, int(st.st_ctime),
397 int(st.st_mtime), st.st_uid, st.st_gid,
398 st.st_size, st.st_mode, gitmode, sha, flags,
402 e = BlankNewEntry(basename)
408 def add_ixentry(self, e):
409 e.children_ofs = e.children_n = 0
410 self._add(pathsplit(e.name), e)
412 def new_reader(self):
414 return Reader(self.tmpname)
417 def reduce_paths(paths):
423 if stat.S_ISDIR(st.st_mode):
426 xpaths.append((rp, p))
428 add_error('reduce_paths: %s' % e)
433 for (rp, p) in xpaths:
434 if prev and (prev == rp
435 or (prev.endswith('/') and rp.startswith(prev))):
436 continue # already superceded by previous path
437 paths.append((rp, p))
439 paths.sort(reverse=True)
444 def __init__(self, iters):
448 # FIXME: doesn't remove duplicated entries between iters.
449 # That only happens for parent directories, but will mean the
450 # actual iteration returns fewer entries than this function counts.
451 return sum(len(it) for it in self.iters)
455 l = [iter(it) for it in self.iters]
456 l = [(next(it),it) for it in l]
457 l = filter(lambda x: x[0], l)
461 if not (count % 1024):
462 progress('bup: merging indexes (%d/%d)\r' % (count, total))
467 if e.name != lastname:
474 log('bup: merging indexes (%d/%d), done.\n' % (count, total))