1 import os, stat, struct, tempfile
3 from bup.helpers import *
7 INDEX_HDR = 'BUPI\0\0\0\2'
9 # FIXME: guess I should have used 64-bit integers to store the mtime/ctime.
10 # NTFS mtime=0 corresponds to the year 1600, which can't be stored in a 32-bit
11 # time_t. Next time we update the bupindex format, keep that in mind.
12 INDEX_SIG = '!IiiIIQII20sHII'
14 ENTLEN = struct.calcsize(INDEX_SIG)
16 FOOTLEN = struct.calcsize(FOOTER_SIG)
18 IX_EXISTS = 0x8000 # file exists on filesystem
19 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
20 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
22 class Error(Exception):
27 def __init__(self, ename, parent):
34 (ofs,n) = (f.tell(), len(self.list))
36 count = len(self.list)
37 #log('popping %r with %d entries\n'
38 # % (''.join(self.ename), count))
42 self.parent.count += count + self.count
46 def _golevel(level, f, ename, newentry):
47 # close nodes back up the tree
49 while ename[:len(level.ename)] != level.ename:
50 n = BlankNewEntry(level.ename[-1])
52 (n.children_ofs,n.children_n) = level.write(f)
53 level.parent.list.append(n)
56 # create nodes down the tree
57 while len(level.ename) < len(ename):
58 level = Level(ename[:len(level.ename)+1], level)
60 # are we in precisely the right place?
61 assert(ename == level.ename)
62 n = newentry or BlankNewEntry(ename and level.ename[-1] or None)
63 (n.children_ofs,n.children_n) = level.write(f)
65 level.parent.list.append(n)
72 def __init__(self, basename, name):
73 self.basename = str(basename)
79 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%s/%s,0x%04x,0x%08x/%d)"
80 % (self.name, self.dev,
81 self.ctime, self.mtime, self.uid, self.gid,
82 self.size, oct(self.mode), oct(self.gitmode),
83 self.flags, self.children_ofs, self.children_n))
87 return struct.pack(INDEX_SIG,
88 self.dev, 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.ctime, self.mtime,
98 self.uid, self.gid, self.size, self.flags & IX_EXISTS)
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)
104 self.ctime = xstat.fstime_floor_secs(st.st_ctime)
105 self.mtime = xstat.fstime_floor_secs(st.st_mtime)
108 self.size = st.st_size
109 self.mode = st.st_mode
110 self.flags |= IX_EXISTS
111 if xstat.fstime_floor_secs(st.st_ctime) >= tstart or old != new \
112 or self.sha == EMPTY_SHA or not self.gitmode:
118 self.uid += 0x100000000
120 self.gid += 0x100000000
121 assert(self.uid >= 0)
122 assert(self.gid >= 0)
123 if self.mtime < -0x80000000: # can happen in NTFS on 64-bit linux
125 if self.ctime < -0x80000000:
127 if self.mtime > 0x7fffffff:
128 self.mtime = 0x7fffffff
129 if self.ctime > 0x7fffffff:
130 self.ctime = 0x7fffffff
133 f = IX_HASHVALID|IX_EXISTS
134 return (self.flags & f) == f
136 def invalidate(self):
137 self.flags &= ~IX_HASHVALID
139 def validate(self, gitmode, sha):
142 assert(gitmode+0 == gitmode)
143 self.gitmode = gitmode
145 self.flags |= IX_HASHVALID|IX_EXISTS
148 return not self.is_deleted()
150 def sha_missing(self):
151 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
153 def is_deleted(self):
154 return (self.flags & IX_EXISTS) == 0
156 def set_deleted(self):
157 if self.flags & IX_EXISTS:
158 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
161 return not self.is_fake()
164 return not self.ctime
167 return (cmp(b.name, a.name)
168 or cmp(a.is_valid(), b.is_valid())
169 or cmp(a.is_fake(), b.is_fake()))
172 f.write(self.basename + '\0' + self.packed())
175 class NewEntry(Entry):
176 def __init__(self, basename, name, dev, ctime, mtime, uid, gid,
177 size, mode, gitmode, sha, flags, children_ofs, children_n):
178 Entry.__init__(self, basename, name)
179 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
180 self.size, self.mode, self.gitmode, self.sha,
181 self.flags, self.children_ofs, self.children_n
182 ) = (dev, int(ctime), int(mtime), uid, gid,
183 size, mode, gitmode, sha, flags, children_ofs, children_n)
187 class BlankNewEntry(NewEntry):
188 def __init__(self, basename):
189 NewEntry.__init__(self, basename, basename,
191 0, EMPTY_SHA, 0, 0, 0)
194 class ExistingEntry(Entry):
195 def __init__(self, parent, basename, name, m, ofs):
196 Entry.__init__(self, basename, name)
200 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
201 self.size, self.mode, self.gitmode, self.sha,
202 self.flags, self.children_ofs, self.children_n
203 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
205 # effectively, we don't bother messing with IX_SHAMISSING if
206 # not IX_HASHVALID, since it's redundant, and repacking is more
207 # expensive than not repacking.
208 # This is implemented by having sha_missing() check IX_HASHVALID too.
209 def set_sha_missing(self, val):
211 oldval = self.sha_missing() and 1 or 0
213 flag = val and IX_SHAMISSING or 0
214 newflags = (self.flags & (~IX_SHAMISSING)) | flag
215 self.flags = newflags
218 def unset_sha_missing(self, flag):
219 if self.flags & IX_SHAMISSING:
220 self.flags &= ~IX_SHAMISSING
224 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
225 if self.parent and not self.is_valid():
226 self.parent.invalidate()
229 def iter(self, name=None, wantrecurse=None):
231 if dname and not dname.endswith('/'):
233 ofs = self.children_ofs
234 assert(ofs <= len(self._m))
235 assert(self.children_n < 1000000)
236 for i in xrange(self.children_n):
237 eon = self._m.find('\0', ofs)
241 basename = str(buffer(self._m, ofs, eon-ofs))
242 child = ExistingEntry(self, basename, self.name + basename,
245 or child.name.startswith(dname)
246 or child.name.endswith('/') and dname.startswith(child.name)):
247 if not wantrecurse or wantrecurse(child):
248 for e in child.iter(name=name, wantrecurse=wantrecurse):
250 if not name or child.name == name or child.name.startswith(dname):
252 ofs = eon + 1 + ENTLEN
259 def __init__(self, filename):
260 self.filename = filename
262 self.writable = False
266 f = open(filename, 'r+')
268 if e.errno == errno.ENOENT:
273 b = f.read(len(INDEX_HDR))
275 log('warning: %s: header: expected %r, got %r'
276 % (filename, INDEX_HDR, b))
278 st = os.fstat(f.fileno())
280 self.m = mmap_readwrite(f)
282 self.count = struct.unpack(FOOTER_SIG,
283 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
289 return int(self.count)
291 def forward_iter(self):
293 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
294 eon = self.m.find('\0', ofs)
298 basename = str(buffer(self.m, ofs, eon-ofs))
299 yield ExistingEntry(None, basename, basename, self.m, eon+1)
300 ofs = eon + 1 + ENTLEN
302 def iter(self, name=None, wantrecurse=None):
303 if len(self.m) > len(INDEX_HDR)+ENTLEN:
305 if dname and not dname.endswith('/'):
307 root = ExistingEntry(None, '/', '/',
308 self.m, len(self.m)-FOOTLEN-ENTLEN)
309 for sub in root.iter(name=name, wantrecurse=wantrecurse):
311 if not dname or dname == root.name:
321 if self.writable and self.m:
326 if self.writable and self.m:
329 self.writable = False
331 def filter(self, prefixes, wantrecurse=None):
332 for (rp, path) in reduce_paths(prefixes):
333 for e in self.iter(rp, wantrecurse=wantrecurse):
334 assert(e.name.startswith(rp))
335 name = path + e.name[len(rp):]
339 # FIXME: this function isn't very generic, because it splits the filename
340 # in an odd way and depends on a terminating '/' to indicate directories.
342 """Split a path into a list of elements of the file system hierarchy."""
344 l = [i+'/' for i in l[:-1]] + l[-1:]
346 l.pop() # extra blank caused by terminating '/'
351 def __init__(self, filename):
352 self.rootlevel = self.level = Level([], None)
357 self.filename = filename = realpath(filename)
358 (dir,name) = os.path.split(filename)
359 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
360 self.f = os.fdopen(ffd, 'wb', 65536)
361 self.f.write(INDEX_HDR)
371 os.unlink(self.tmpname)
375 self.level = _golevel(self.level, self.f, [], None)
376 self.count = self.rootlevel.count
379 self.f.write(struct.pack(FOOTER_SIG, self.count))
381 assert(self.level == None)
389 os.rename(self.tmpname, self.filename)
391 def _add(self, ename, entry):
392 if self.lastfile and self.lastfile <= ename:
393 raise Error('%r must come before %r'
394 % (''.join(e.name), ''.join(self.lastfile)))
395 self.lastfile = e.name
396 self.level = _golevel(self.level, self.f, ename, entry)
398 def add(self, name, st, hashgen = None):
399 endswith = name.endswith('/')
400 ename = pathsplit(name)
402 #log('add: %r %r\n' % (basename, name))
406 (gitmode, sha) = hashgen(name)
407 flags |= IX_HASHVALID
409 (gitmode, sha) = (0, EMPTY_SHA)
411 isdir = stat.S_ISDIR(st.st_mode)
412 assert(isdir == endswith)
413 e = NewEntry(basename, name, st.st_dev,
414 xstat.fstime_floor_secs(st.st_ctime),
415 xstat.fstime_floor_secs(st.st_mtime),
416 st.st_uid, st.st_gid,
417 st.st_size, st.st_mode, gitmode, sha, flags,
421 e = BlankNewEntry(basename)
427 def add_ixentry(self, e):
428 e.children_ofs = e.children_n = 0
429 self._add(pathsplit(e.name), e)
431 def new_reader(self):
433 return Reader(self.tmpname)
436 def reduce_paths(paths):
442 if stat.S_ISDIR(st.st_mode):
445 xpaths.append((rp, p))
447 add_error('reduce_paths: %s' % e)
452 for (rp, p) in xpaths:
453 if prev and (prev == rp
454 or (prev.endswith('/') and rp.startswith(prev))):
455 continue # already superceded by previous path
456 paths.append((rp, p))
458 paths.sort(reverse=True)
462 def pfunc(count, total):
463 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
464 def pfinal(count, total):
465 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
466 return merge_iter(iters, 1024, pfunc, pfinal, key='name')