def test_ondisk():
bs = BitSet([10, 11, 30, 50, 80])
st = RamStorage()
f = st.create_file("test")
size = bs.to_disk(f)
f.close()
f = st.open_file("test")
b = OnDiskBitSet(f, 0, size)
assert list(b) == list(bs)
assert b.after(0) == 10
assert b.after(10) == 11
assert b.after(80) is None
assert b.after(99) is None
assert b.before(0) is None
assert b.before(99) == 80
assert b.before(80) == 50
assert b.before(10) is None
f.seek(0)
b = BitSet.from_disk(f, size)
assert list(b) == list(bs)
def test_ondisk():
bs = BitSet([10, 11, 30, 50, 80])
st = RamStorage()
f = st.create_file("test")
size = bs.to_disk(f)
f.close()
f = st.open_file("test")
b = OnDiskBitSet(f, 0, size)
assert list(b) == list(bs)
assert b.after(0) == 10
assert b.after(10) == 11
assert b.after(80) is None
assert b.after(99) is None
assert b.before(0) is None
assert b.before(99) == 80
assert b.before(80) == 50
assert b.before(10) is None
f.seek(0)
b = BitSet.from_disk(f, size)
assert list(b) == list(bs)
def __init__(self, dbfile, basepos, length, doccount):
self._dbfile = dbfile
self._basepos = basepos
self._length = length
self._doccount = doccount
compressed = dbfile.get_byte(basepos + (length - 1))
if compressed:
bbytes = zlib.decompress(dbfile.get(basepos, length - 1))
bitset = BitSet.from_bytes(bbytes)
else:
dbfile.seek(basepos)
bitset = OnDiskBitSet(dbfile, basepos, length - 1)
self._bitset = bitset
def __init__(self, dbfile, basepos, length, doccount):
self._dbfile = dbfile
self._basepos = basepos
self._length = length
self._doccount = doccount
compressed = dbfile.get_byte(basepos + (length - 1))
if compressed:
bbytes = zlib.decompress(dbfile.get(basepos, length - 1))
bitset = BitSet.from_bytes(bbytes)
else:
dbfile.seek(basepos)
bitset = OnDiskBitSet(dbfile, basepos, length - 1)
self._bitset = bitset
class Writer(ColumnWriter):
def __init__(self, dbfile, compressat):
self._dbfile = dbfile
self._compressat = compressat
self._bitset = BitSet()
def __repr__(self):
return "<Bit.Writer>"
def add(self, docnum, value):
if value:
self._bitset.add(docnum)
def finish(self, doccount):
dbfile = self._dbfile
bits = self._bitset.bits
if zlib and len(bits) <= self._compressat:
compressed = zlib.compress(array_tobytes(bits), 3)
dbfile.write(compressed)
dbfile.write_byte(1)
else:
dbfile.write_array(bits)
dbfile.write_byte(0)
class Writer(ColumnWriter):
def __init__(self, dbfile, compressat):
self._dbfile = dbfile
self._compressat = compressat
self._bitset = BitSet()
def __repr__(self):
return "<Bit.Writer>"
def add(self, docnum, value):
if value:
self._bitset.add(docnum)
def finish(self, doccount):
dbfile = self._dbfile
bits = self._bitset.bits
if zlib and len(bits) <= self._compressat:
compressed = zlib.compress(array_tobytes(bits), 3)
dbfile.write(compressed)
dbfile.write_byte(1)
else:
dbfile.write_array(bits)
dbfile.write_byte(0)
def load(self):
if isinstance(self._bitset, OnDiskBitSet):
bs = self._dbfile.get_array(self._basepos, "B",
self._length - 1)
self._bitset = BitSet.from_bytes(bs)
return self
def __init__(self, dbfile, compressat):
self._dbfile = dbfile
self._compressat = compressat
self._bitset = BitSet()
def load(self):
if isinstance(self._bitset, OnDiskBitSet):
bs = self._dbfile.get_array(self._basepos, "B",
self._length - 1)
self._bitset = BitSet.from_bytes(bs)
return self
def __init__(self, dbfile, compressat):
self._dbfile = dbfile
self._compressat = compressat
self._bitset = BitSet()
