2 from __future__ import absolute_import, print_function
3 import glob, mmap, os, struct
5 from bup import _helpers
6 from bup.compat import range
7 from bup.helpers import log, mmap_read
8 from bup.io import path_msg
13 extract_bits = _helpers.extract_bits
19 """Wrapper which contains data from multiple index files.
20 Multiple index (.midx) files constitute a wrapper around index (.idx) files
21 and make it possible for bup to expand Git's indexing capabilities to vast
24 def __init__(self, filename):
26 self.force_keep = False
28 assert(filename.endswith(b'.midx'))
29 self.map = mmap_read(open(filename))
30 if self.map[0:4] != b'MIDX':
31 log('Warning: skipping: invalid MIDX header in %r\n'
33 self.force_keep = True
34 return self._init_failed()
35 ver = struct.unpack('!I', self.map[4:8])[0]
36 if ver < MIDX_VERSION:
37 log('Warning: ignoring old-style (v%d) midx %r\n'
38 % (ver, path_msg(filename)))
39 self.force_keep = False # old stuff is boring
40 return self._init_failed()
41 if ver > MIDX_VERSION:
42 log('Warning: ignoring too-new (v%d) midx %r\n'
43 % (ver, path_msg(filename)))
44 self.force_keep = True # new stuff is exciting
45 return self._init_failed()
47 self.bits = _helpers.firstword(self.map[8:12])
48 self.entries = 2**self.bits
50 # fanout len is self.entries * 4
51 self.sha_ofs = self.fanout_ofs + self.entries * 4
52 self.nsha = self._fanget(self.entries - 1)
53 # sha table len is self.nsha * 20
54 self.which_ofs = self.sha_ofs + 20 * self.nsha
55 # which len is self.nsha * 4
56 self.idxnames = self.map[self.which_ofs + 4 * self.nsha:].split(b'\0')
61 def _init_failed(self):
67 if i >= self.entries * 4 or i < 0:
68 raise IndexError('invalid midx index %d' % i)
69 ofs = self.fanout_ofs + i * 4
70 return _helpers.firstword(self.map[ofs : ofs + 4])
73 if i >= self.nsha or i < 0:
74 raise IndexError('invalid midx index %d' % i)
75 ofs = self.sha_ofs + i * 20
76 return self.map[ofs : ofs + 20]
78 def _get_idx_i(self, i):
79 if i >= self.nsha or i < 0:
80 raise IndexError('invalid midx index %d' % i)
81 ofs = self.which_ofs + i * 4
82 return struct.unpack_from('!I', self.map, offset=ofs)[0]
84 def _get_idxname(self, i):
85 return self.idxnames[self._get_idx_i(i)]
88 if self.map is not None:
89 self.fanout = self.shatable = self.whichlist = self.idxnames = None
93 def exists(self, hash, want_source=False):
94 """Return nonempty if the object exists in the index files."""
95 global _total_searches, _total_steps
98 el = extract_bits(want, self.bits)
100 start = self._fanget(el-1)
101 startv = el << (32-self.bits)
105 end = self._fanget(el)
106 endv = (el+1) << (32-self.bits)
107 _total_steps += 1 # lookup table is a step
108 hashv = _helpers.firstword(hash)
109 #print '(%08x) %08x %08x %08x' % (extract_bits(want, 32), startv, hashv, endv)
112 #print '! %08x %08x %08x %d - %d' % (startv, hashv, endv, start, end)
113 mid = start + (hashv - startv) * (end - start - 1) // (endv - startv)
114 #print ' %08x %08x %08x %d %d %d' % (startv, hashv, endv, start, mid, end)
116 #print ' %08x' % self._num(v)
119 startv = _helpers.firstword(v)
122 endv = _helpers.firstword(v)
124 return want_source and self._get_idxname(mid) or True
129 for ofs in range(start, start + self.nsha * 20, 20):
130 yield self.map[ofs : ofs + 20]
133 return int(self.nsha)
136 def clear_midxes(dir=None):
137 for midx in glob.glob(os.path.join(dir, b'*.midx')):