lib/bup/index.py

   1 import os, stat, time, struct, tempfile
   2 from bup.helpers import *
   3
   4 EMPTY_SHA = '\0'*20
   5 FAKE_SHA = '\x01'*20
   6 INDEX_HDR = 'BUPI\0\0\0\2'
   7 INDEX_SIG = '!IIIIIQII20sHII'
   8 ENTLEN = struct.calcsize(INDEX_SIG)
   9 FOOTER_SIG = '!Q'
  10 FOOTLEN = struct.calcsize(FOOTER_SIG)
  11
  12 IX_EXISTS = 0x8000
  13 IX_HASHVALID = 0x4000
  14
  15 class Error(Exception):
  16     pass
  17
  18
  19 class Level:
  20     def __init__(self, ename, parent):
  21         self.parent = parent
  22         self.ename = ename
  23         self.list = []
  24         self.count = 0
  25
  26     def write(self, f):
  27         (ofs,n) = (f.tell(), len(self.list))
  28         if self.list:
  29             count = len(self.list)
  30             #log('popping %r with %d entries\n'
  31             #    % (''.join(self.ename), count))
  32             for e in self.list:
  33                 e.write(f)
  34             if self.parent:
  35                 self.parent.count += count + self.count
  36         return (ofs,n)
  37
  38
  39 def _golevel(level, f, ename, newentry):
  40     # close nodes back up the tree
  41     assert(level)
  42     while ename[:len(level.ename)] != level.ename:
  43         n = BlankNewEntry(level.ename[-1])
  44         n.flags |= IX_EXISTS
  45         (n.children_ofs,n.children_n) = level.write(f)
  46         level.parent.list.append(n)
  47         level = level.parent
  48
  49     # create nodes down the tree
  50     while len(level.ename) < len(ename):
  51         level = Level(ename[:len(level.ename)+1], level)
  52
  53     # are we in precisely the right place?
  54     assert(ename == level.ename)
  55     n = newentry or BlankNewEntry(ename and level.ename[-1] or None)
  56     (n.children_ofs,n.children_n) = level.write(f)
  57     if level.parent:
  58         level.parent.list.append(n)
  59     level = level.parent
  60
  61     return level
  62
  63
  64 class Entry:
  65     def __init__(self, basename, name):
  66         self.basename = str(basename)
  67         self.name = str(name)
  68         self.children_ofs = 0
  69         self.children_n = 0
  70
  71     def __repr__(self):
  72         return ("(%s,0x%04x,%d,%d,%d,%d,%d,%s/%s,0x%04x,0x%08x/%d)"
  73                 % (self.name, self.dev,
  74                    self.ctime, self.mtime, self.uid, self.gid,
  75                    self.size, oct(self.mode), oct(self.gitmode),
  76                    self.flags, self.children_ofs, self.children_n))
  77
  78     def packed(self):
  79         return struct.pack(INDEX_SIG,
  80                            self.dev, self.ctime, self.mtime,
  81                            self.uid, self.gid, self.size, self.mode,
  82                            self.gitmode, self.sha, self.flags,
  83                            self.children_ofs, self.children_n)
  84
  85     def from_stat(self, st, tstart):
  86         old = (self.dev, self.ctime, self.mtime,
  87                self.uid, self.gid, self.size, self.flags & IX_EXISTS)
  88         new = (st.st_dev, int(st.st_ctime), int(st.st_mtime),
  89                st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
  90         self.dev = st.st_dev
  91         self.ctime = int(st.st_ctime)
  92         self.mtime = int(st.st_mtime)
  93         self.uid = st.st_uid
  94         self.gid = st.st_gid
  95         self.size = st.st_size
  96         self.mode = st.st_mode
  97         self.flags |= IX_EXISTS
  98         if int(st.st_ctime) >= tstart or old != new \
  99               or self.sha == EMPTY_SHA or not self.gitmode:
 100             self.invalidate()
 101
 102     def is_valid(self):
 103         f = IX_HASHVALID|IX_EXISTS
 104         return (self.flags & f) == f
 105
 106     def invalidate(self):
 107         self.flags &= ~IX_HASHVALID
 108
 109     def validate(self, gitmode, sha):
 110         assert(sha)
 111         assert(gitmode)
 112         self.gitmode = gitmode
 113         self.sha = sha
 114         self.flags |= IX_HASHVALID|IX_EXISTS
 115
 116     def exists(self):
 117         return not self.is_deleted()
 118
 119     def is_deleted(self):
 120         return (self.flags & IX_EXISTS) == 0
 121
 122     def set_deleted(self):
 123         if self.flags & IX_EXISTS:
 124             self.flags &= ~(IX_EXISTS | IX_HASHVALID)
 125
 126     def is_real(self):
 127         return not self.is_fake()
 128
 129     def is_fake(self):
 130         return not self.ctime
 131
 132     def __cmp__(a, b):
 133         return (cmp(a.name, b.name)
 134                 or -cmp(a.is_valid(), b.is_valid())
 135                 or -cmp(a.is_fake(), b.is_fake()))
 136
 137     def write(self, f):
 138         f.write(self.basename + '\0' + self.packed())
 139
 140
 141 class NewEntry(Entry):
 142     def __init__(self, basename, name, dev, ctime, mtime, uid, gid,
 143                  size, mode, gitmode, sha, flags, children_ofs, children_n):
 144         Entry.__init__(self, basename, name)
 145         (self.dev, self.ctime, self.mtime, self.uid, self.gid,
 146          self.size, self.mode, self.gitmode, self.sha,
 147          self.flags, self.children_ofs, self.children_n
 148          ) = (dev, int(ctime), int(mtime), uid, gid,
 149               size, mode, gitmode, sha, flags, children_ofs, children_n)
 150
 151
 152 class BlankNewEntry(NewEntry):
 153     def __init__(self, basename):
 154         NewEntry.__init__(self, basename, basename,
 155                           0, 0, 0, 0, 0, 0, 0,
 156                           0, EMPTY_SHA, 0, 0, 0)
 157
 158
 159 class ExistingEntry(Entry):
 160     def __init__(self, parent, basename, name, m, ofs):
 161         Entry.__init__(self, basename, name)
 162         self.parent = parent
 163         self._m = m
 164         self._ofs = ofs
 165         (self.dev, self.ctime, self.mtime, self.uid, self.gid,
 166          self.size, self.mode, self.gitmode, self.sha,
 167          self.flags, self.children_ofs, self.children_n
 168          ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
 169
 170     def repack(self):
 171         self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
 172         if self.parent and not self.is_valid():
 173             self.parent.invalidate()
 174             self.parent.repack()
 175
 176     def iter(self, name=None, wantrecurse=None):
 177         dname = name
 178         if dname and not dname.endswith('/'):
 179             dname += '/'
 180         ofs = self.children_ofs
 181         assert(ofs <= len(self._m))
 182         assert(self.children_n < 1000000)
 183         for i in xrange(self.children_n):
 184             eon = self._m.find('\0', ofs)
 185             assert(eon >= 0)
 186             assert(eon >= ofs)
 187             assert(eon > ofs)
 188             basename = str(buffer(self._m, ofs, eon-ofs))
 189             child = ExistingEntry(self, basename, self.name + basename,
 190                                   self._m, eon+1)
 191             if (not dname
 192                  or child.name.startswith(dname)
 193                  or child.name.endswith('/') and dname.startswith(child.name)):
 194                 if not wantrecurse or wantrecurse(child):
 195                     for e in child.iter(name=name, wantrecurse=wantrecurse):
 196                         yield e
 197             if not name or child.name == name or child.name.startswith(dname):
 198                 yield child
 199             ofs = eon + 1 + ENTLEN
 200
 201     def __iter__(self):
 202         return self.iter()
 203
 204
 205 class Reader:
 206     def __init__(self, filename):
 207         self.filename = filename
 208         self.m = ''
 209         self.writable = False
 210         self.count = 0
 211         f = None
 212         try:
 213             f = open(filename, 'r+')
 214         except IOError, e:
 215             if e.errno == errno.ENOENT:
 216                 pass
 217             else:
 218                 raise
 219         if f:
 220             b = f.read(len(INDEX_HDR))
 221             if b != INDEX_HDR:
 222                 log('warning: %s: header: expected %r, got %r'
 223                                  % (filename, INDEX_HDR, b))
 224             else:
 225                 st = os.fstat(f.fileno())
 226                 if st.st_size:
 227                     self.m = mmap_readwrite(f)
 228                     self.writable = True
 229                     self.count = struct.unpack(FOOTER_SIG,
 230                           str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
 231
 232     def __del__(self):
 233         self.close()
 234
 235     def __len__(self):
 236         return int(self.count)
 237
 238     def forward_iter(self):
 239         ofs = len(INDEX_HDR)
 240         while ofs+ENTLEN <= len(self.m)-FOOTLEN:
 241             eon = self.m.find('\0', ofs)
 242             assert(eon >= 0)
 243             assert(eon >= ofs)
 244             assert(eon > ofs)
 245             basename = str(buffer(self.m, ofs, eon-ofs))
 246             yield ExistingEntry(None, basename, basename, self.m, eon+1)
 247             ofs = eon + 1 + ENTLEN
 248
 249     def iter(self, name=None, wantrecurse=None):
 250         if len(self.m) > len(INDEX_HDR)+ENTLEN:
 251             dname = name
 252             if dname and not dname.endswith('/'):
 253                 dname += '/'
 254             root = ExistingEntry(None, '/', '/',
 255                                  self.m, len(self.m)-FOOTLEN-ENTLEN)
 256             for sub in root.iter(name=name, wantrecurse=wantrecurse):
 257                 yield sub
 258             if not dname or dname == root.name:
 259                 yield root
 260
 261     def __iter__(self):
 262         return self.iter()
 263
 264     def exists(self):
 265         return self.m
 266
 267     def save(self):
 268         if self.writable and self.m:
 269             self.m.flush()
 270
 271     def close(self):
 272         self.save()
 273         if self.writable and self.m:
 274             self.m = None
 275             self.writable = False
 276
 277     def filter(self, prefixes, wantrecurse=None):
 278         for (rp, path) in reduce_paths(prefixes):
 279             for e in self.iter(rp, wantrecurse=wantrecurse):
 280                 assert(e.name.startswith(rp))
 281                 name = path + e.name[len(rp):]
 282                 yield (name, e)
 283
 284
 285 class Writer:
 286     def __init__(self, filename):
 287         self.rootlevel = self.level = Level([], None)
 288         self.f = None
 289         self.count = 0
 290         self.lastfile = None
 291         self.filename = None
 292         self.filename = filename = realpath(filename)
 293         (dir,name) = os.path.split(filename)
 294         (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
 295         self.f = os.fdopen(ffd, 'wb', 65536)
 296         self.f.write(INDEX_HDR)
 297
 298     def __del__(self):
 299         self.abort()
 300
 301     def abort(self):
 302         f = self.f
 303         self.f = None
 304         if f:
 305             f.close()
 306             os.unlink(self.tmpname)
 307
 308     def flush(self):
 309         if self.level:
 310             self.level = _golevel(self.level, self.f, [], None)
 311             self.count = self.rootlevel.count
 312             if self.count:
 313                 self.count += 1
 314             self.f.write(struct.pack(FOOTER_SIG, self.count))
 315             self.f.flush()
 316         assert(self.level == None)
 317
 318     def close(self):
 319         self.flush()
 320         f = self.f
 321         self.f = None
 322         if f:
 323             f.close()
 324             os.rename(self.tmpname, self.filename)
 325
 326     def _add(self, ename, entry):
 327         if self.lastfile and self.lastfile <= ename:
 328             raise Error('%r must come before %r'
 329                              % (''.join(e.name), ''.join(self.lastfile)))
 330             self.lastfile = e.name
 331         self.level = _golevel(self.level, self.f, ename, entry)
 332
 333     def add(self, name, st, hashgen = None):
 334         endswith = name.endswith('/')
 335         ename = pathsplit(name)
 336         basename = ename[-1]
 337         #log('add: %r %r\n' % (basename, name))
 338         flags = IX_EXISTS
 339         sha = None
 340         if hashgen:
 341             (gitmode, sha) = hashgen(name)
 342             flags |= IX_HASHVALID
 343         else:
 344             (gitmode, sha) = (0, EMPTY_SHA)
 345         if st:
 346             isdir = stat.S_ISDIR(st.st_mode)
 347             assert(isdir == endswith)
 348             e = NewEntry(basename, name, st.st_dev, int(st.st_ctime),
 349                          int(st.st_mtime), st.st_uid, st.st_gid,
 350                          st.st_size, st.st_mode, gitmode, sha, flags,
 351                          0, 0)
 352         else:
 353             assert(endswith)
 354             e = BlankNewEntry(basename)
 355             e.gitmode = gitmode
 356             e.sha = sha
 357             e.flags = flags
 358         self._add(ename, e)
 359
 360     def add_ixentry(self, e):
 361         e.children_ofs = e.children_n = 0
 362         self._add(pathsplit(e.name), e)
 363
 364     def new_reader(self):
 365         self.flush()
 366         return Reader(self.tmpname)
 367
 368
 369 def reduce_paths(paths):
 370     xpaths = []
 371     for p in paths:
 372         rp = realpath(p)
 373         try:
 374             st = os.lstat(rp)
 375             if stat.S_ISDIR(st.st_mode):
 376                 rp = slashappend(rp)
 377                 p = slashappend(p)
 378         except OSError, e:
 379             if e.errno != errno.ENOENT:
 380                 raise
 381         xpaths.append((rp, p))
 382     xpaths.sort()
 383
 384     paths = []
 385     prev = None
 386     for (rp, p) in xpaths:
 387         if prev and (prev == rp
 388                      or (prev.endswith('/') and rp.startswith(prev))):
 389             continue # already superceded by previous path
 390         paths.append((rp, p))
 391         prev = rp
 392     paths.sort(reverse=True)
 393     return paths
 394
 395
 396 class MergeIter:
 397     def __init__(self, iters):
 398         self.iters = iters
 399
 400     def __len__(self):
 401         # FIXME: doesn't remove duplicated entries between iters.
 402         # That only happens for parent directories, but will mean the
 403         # actual iteration returns fewer entries than this function counts.
 404         return sum(len(it) for it in self.iters)
 405
 406     def __iter__(self):
 407         total = len(self)
 408         l = [iter(it) for it in self.iters]
 409         l = [(next(it),it) for it in l]
 410         l = filter(lambda x: x[0], l)
 411         count = 0
 412         lastname = None
 413         while l:
 414             if not (count % 1024):
 415                 progress('bup: merging indexes (%d/%d)\r' % (count, total))
 416             l.sort()
 417             (e,it) = l.pop()
 418             if not e:
 419                 continue
 420             if e.name != lastname:
 421                 yield e
 422                 lastname = e.name
 423             n = next(it)
 424             if n:
 425                 l.append((n,it))
 426             count += 1
 427         log('bup: merging indexes (%d/%d), done.\n' % (count, total))