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