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 self.mtime = self._fixup_time(self.mtime)
124 self.ctime = self._fixup_time(self.ctime)
126 def _fixup_time(self, t):
127 if t < -0x80000000: # can happen in NTFS on 64-bit linux
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, dev, ctime, mtime, uid, gid,
179 size, mode, gitmode, sha, flags, children_ofs, children_n):
180 Entry.__init__(self, basename, name)
181 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
182 self.size, self.mode, self.gitmode, self.sha,
183 self.flags, self.children_ofs, self.children_n
184 ) = (dev, int(ctime), int(mtime), uid, gid,
185 size, mode, gitmode, sha, flags, children_ofs, children_n)
189 class BlankNewEntry(NewEntry):
190 def __init__(self, basename):
191 NewEntry.__init__(self, basename, basename,
193 0, EMPTY_SHA, 0, 0, 0)
196 class ExistingEntry(Entry):
197 def __init__(self, parent, basename, name, m, ofs):
198 Entry.__init__(self, basename, name)
202 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
203 self.size, self.mode, self.gitmode, self.sha,
204 self.flags, self.children_ofs, self.children_n
205 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
207 # effectively, we don't bother messing with IX_SHAMISSING if
208 # not IX_HASHVALID, since it's redundant, and repacking is more
209 # expensive than not repacking.
210 # This is implemented by having sha_missing() check IX_HASHVALID too.
211 def set_sha_missing(self, val):
213 oldval = self.sha_missing() and 1 or 0
215 flag = val and IX_SHAMISSING or 0
216 newflags = (self.flags & (~IX_SHAMISSING)) | flag
217 self.flags = newflags
220 def unset_sha_missing(self, flag):
221 if self.flags & IX_SHAMISSING:
222 self.flags &= ~IX_SHAMISSING
226 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
227 if self.parent and not self.is_valid():
228 self.parent.invalidate()
231 def iter(self, name=None, wantrecurse=None):
233 if dname and not dname.endswith('/'):
235 ofs = self.children_ofs
236 assert(ofs <= len(self._m))
237 assert(self.children_n < 1000000)
238 for i in xrange(self.children_n):
239 eon = self._m.find('\0', ofs)
243 basename = str(buffer(self._m, ofs, eon-ofs))
244 child = ExistingEntry(self, basename, self.name + basename,
247 or child.name.startswith(dname)
248 or child.name.endswith('/') and dname.startswith(child.name)):
249 if not wantrecurse or wantrecurse(child):
250 for e in child.iter(name=name, wantrecurse=wantrecurse):
252 if not name or child.name == name or child.name.startswith(dname):
254 ofs = eon + 1 + ENTLEN
261 def __init__(self, filename):
262 self.filename = filename
264 self.writable = False
268 f = open(filename, 'r+')
270 if e.errno == errno.ENOENT:
275 b = f.read(len(INDEX_HDR))
277 log('warning: %s: header: expected %r, got %r'
278 % (filename, INDEX_HDR, b))
280 st = os.fstat(f.fileno())
282 self.m = mmap_readwrite(f)
284 self.count = struct.unpack(FOOTER_SIG,
285 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
291 return int(self.count)
293 def forward_iter(self):
295 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
296 eon = self.m.find('\0', ofs)
300 basename = str(buffer(self.m, ofs, eon-ofs))
301 yield ExistingEntry(None, basename, basename, self.m, eon+1)
302 ofs = eon + 1 + ENTLEN
304 def iter(self, name=None, wantrecurse=None):
305 if len(self.m) > len(INDEX_HDR)+ENTLEN:
307 if dname and not dname.endswith('/'):
309 root = ExistingEntry(None, '/', '/',
310 self.m, len(self.m)-FOOTLEN-ENTLEN)
311 for sub in root.iter(name=name, wantrecurse=wantrecurse):
313 if not dname or dname == root.name:
323 if self.writable and self.m:
328 if self.writable and self.m:
331 self.writable = False
333 def filter(self, prefixes, wantrecurse=None):
334 for (rp, path) in reduce_paths(prefixes):
335 for e in self.iter(rp, wantrecurse=wantrecurse):
336 assert(e.name.startswith(rp))
337 name = path + e.name[len(rp):]
341 # FIXME: this function isn't very generic, because it splits the filename
342 # in an odd way and depends on a terminating '/' to indicate directories.
344 """Split a path into a list of elements of the file system hierarchy."""
346 l = [i+'/' for i in l[:-1]] + l[-1:]
348 l.pop() # extra blank caused by terminating '/'
353 def __init__(self, filename):
354 self.rootlevel = self.level = Level([], None)
359 self.filename = filename = realpath(filename)
360 (dir,name) = os.path.split(filename)
361 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
362 self.f = os.fdopen(ffd, 'wb', 65536)
363 self.f.write(INDEX_HDR)
373 os.unlink(self.tmpname)
377 self.level = _golevel(self.level, self.f, [], None)
378 self.count = self.rootlevel.count
381 self.f.write(struct.pack(FOOTER_SIG, self.count))
383 assert(self.level == None)
391 os.rename(self.tmpname, self.filename)
393 def _add(self, ename, entry):
394 if self.lastfile and self.lastfile <= ename:
395 raise Error('%r must come before %r'
396 % (''.join(e.name), ''.join(self.lastfile)))
397 self.lastfile = e.name
398 self.level = _golevel(self.level, self.f, ename, entry)
400 def add(self, name, st, hashgen = None):
401 endswith = name.endswith('/')
402 ename = pathsplit(name)
404 #log('add: %r %r\n' % (basename, name))
408 (gitmode, sha) = hashgen(name)
409 flags |= IX_HASHVALID
411 (gitmode, sha) = (0, EMPTY_SHA)
413 isdir = stat.S_ISDIR(st.st_mode)
414 assert(isdir == endswith)
415 e = NewEntry(basename, name, st.st_dev,
416 xstat.fstime_floor_secs(st.st_ctime),
417 xstat.fstime_floor_secs(st.st_mtime),
418 st.st_uid, st.st_gid,
419 st.st_size, st.st_mode, gitmode, sha, flags,
423 e = BlankNewEntry(basename)
429 def add_ixentry(self, e):
430 e.children_ofs = e.children_n = 0
431 self._add(pathsplit(e.name), e)
433 def new_reader(self):
435 return Reader(self.tmpname)
438 def reduce_paths(paths):
444 if stat.S_ISDIR(st.st_mode):
447 xpaths.append((rp, p))
449 add_error('reduce_paths: %s' % e)
454 for (rp, p) in xpaths:
455 if prev and (prev == rp
456 or (prev.endswith('/') and rp.startswith(prev))):
457 continue # already superceded by previous path
458 paths.append((rp, p))
460 paths.sort(reverse=True)
464 def pfunc(count, total):
465 qprogress('bup: merging indexes (%d/%d)\r' % (count, total))
466 def pfinal(count, total):
467 progress('bup: merging indexes (%d/%d), done.\n' % (count, total))
468 return merge_iter(iters, 1024, pfunc, pfinal, key='name')