def compressed_index_to_file_elias_gamma(index, out_file_name):
print "Compressing with elias gamma..."
from kbp.univ import elias
import struct
file_name = out_file_name + "_elias_gamma"
f = open(file_name, "wb")
for k, v in index.items():
word_len = len(k.encode("utf-8"))
f.write(struct.pack("I", word_len))
f.write(k.encode("utf-8"))
f.write(struct.pack("I", v[0]))
entries = ""
for i in v[1]:
entries += elias.gamma_encode(i + 1)
if len(entries) % 32 != 0:
zeroes = 32 - len(entries) % 32
for i in range(zeroes):
entries += "0"
numbers = []
count = len(entries) / 32
for idx in range(0, count):
numbers.append(int(entries[idx * 32 : (idx + 1) * 32], 2))
f.write(struct.pack("I", count))
for number in numbers:
f.write(struct.pack("I", number))
f.close()
return
def compressed_index_to_file_elias_gamma(index, out_file_name):
print "Compressing with elias gamma..."
from kbp.univ import elias
import struct
file_name = out_file_name + "_elias_gamma"
f = open(file_name, 'wb')
for k, v in index.items():
word_len = len(k.encode("utf-8"))
f.write(struct.pack('I', word_len))
f.write(k.encode("utf-8"))
f.write(struct.pack('I', v[0]))
entries = ""
for i in v[1]:
entries += elias.gamma_encode(i + 1)
if len(entries) % 32 != 0:
zeroes = 32 - len(entries) % 32
for i in range(zeroes):
entries += "0"
numbers = []
count = len(entries) / 32
for idx in range(0, count):
numbers.append(int(entries[idx * 32:(idx + 1) * 32], 2))
f.write(struct.pack('I', count))
for number in numbers:
f.write(struct.pack('I', number))
f.close()
return
#Kabopan - Readable Algorithms. Public Domain, 2007-2009
from kbp.univ.elias import ( \
elias_split, gamma_encode, gamma_decode,
interleaved_gamma_encode, interleaved_gamma_decode,
delta_encode, delta_decode,
omega_encode, omega_decode)
assert elias_split(1) == (0, "")
assert elias_split(14) == (3, "110")
assert gamma_encode(1) == "1"
assert gamma_encode(2) == "010"
assert gamma_encode(14) == "0001" + "110"
assert gamma_decode("000010001") == (17, 9)
assert gamma_decode("00001000100") == (17, 9)
assert interleaved_gamma_encode(14) == "101001"
assert interleaved_gamma_decode("101001") == (14, 6)
assert delta_encode(17) == "001010001"
assert delta_decode("001010001") == (17, 9)
assert [omega_encode(i) for i in range(1, 18)] == ['0', '100', '110', '101000', '101010', '101100', '101110', '1110000',
'1110010', '1110100', '1110110', '1111000', '1111010', '1111100', '1111110', '10100100000', '10100100010']
assert omega_decode('10100100010') == (17, 11)
assert omega_decode('1010010001000') == (17, 11)
