2 from bup import _helpers
3 from bup.helpers import *
7 extract_bits = _helpers.extract_bits
13 """Wrapper which contains data from multiple index files.
14 Multiple index (.midx) files constitute a wrapper around index (.idx) files
15 and make it possible for bup to expand Git's indexing capabilities to vast
18 def __init__(self, filename):
20 self.force_keep = False
21 assert(filename.endswith('.midx'))
22 self.map = mmap_read(open(filename))
23 if str(self.map[0:4]) != 'MIDX':
24 log('Warning: skipping: invalid MIDX header in %r\n' % filename)
25 self.force_keep = True
26 return self._init_failed()
27 ver = struct.unpack('!I', self.map[4:8])[0]
28 if ver < MIDX_VERSION:
29 log('Warning: ignoring old-style (v%d) midx %r\n'
31 self.force_keep = False # old stuff is boring
32 return self._init_failed()
33 if ver > MIDX_VERSION:
34 log('Warning: ignoring too-new (v%d) midx %r\n'
36 self.force_keep = True # new stuff is exciting
37 return self._init_failed()
39 self.bits = _helpers.firstword(self.map[8:12])
40 self.entries = 2**self.bits
41 self.fanout = buffer(self.map, 12, self.entries*4)
42 self.sha_ofs = 12 + self.entries*4
43 self.nsha = nsha = self._fanget(self.entries-1)
44 self.shatable = buffer(self.map, self.sha_ofs, nsha*20)
45 self.which_ofs = self.sha_ofs + 20*nsha
46 self.whichlist = buffer(self.map, self.which_ofs, nsha*4)
47 self.idxnames = str(self.map[self.which_ofs + 4*nsha:]).split('\0')
49 def _init_failed(self):
52 self.fanout = buffer('\0\0\0\0')
53 self.shatable = buffer('\0'*20)
58 s = self.fanout[start:start+4]
59 return _helpers.firstword(s)
62 return str(self.shatable[i*20:(i+1)*20])
64 def _get_idx_i(self, i):
65 return struct.unpack('!I', self.whichlist[i*4:(i+1)*4])[0]
67 def _get_idxname(self, i):
68 return self.idxnames[self._get_idx_i(i)]
70 def exists(self, hash, want_source=False):
71 """Return nonempty if the object exists in the index files."""
72 global _total_searches, _total_steps
75 el = extract_bits(want, self.bits)
77 start = self._fanget(el-1)
78 startv = el << (32-self.bits)
82 end = self._fanget(el)
83 endv = (el+1) << (32-self.bits)
84 _total_steps += 1 # lookup table is a step
85 hashv = _helpers.firstword(hash)
86 #print '(%08x) %08x %08x %08x' % (extract_bits(want, 32), startv, hashv, endv)
89 #print '! %08x %08x %08x %d - %d' % (startv, hashv, endv, start, end)
90 mid = start + (hashv-startv)*(end-start-1)/(endv-startv)
91 #print ' %08x %08x %08x %d %d %d' % (startv, hashv, endv, start, mid, end)
93 #print ' %08x' % self._num(v)
96 startv = _helpers.firstword(v)
99 endv = _helpers.firstword(v)
101 return want_source and self._get_idxname(mid) or True
105 for i in xrange(self._fanget(self.entries-1)):
106 yield buffer(self.shatable, i*20, 20)
109 return int(self._fanget(self.entries-1))