def read_compressed_index_from_file_elias_gamma(out_file_name):
print "Reading compressed index with elias gamma..."
from kbp.univ import elias
import struct
index = dict()
file_name = out_file_name + "_elias_gamma"
f = open(file_name, 'rb')
word_len_bytes = f.read(4)
idx1 = 0
while word_len_bytes:
word_len = struct.unpack('I', word_len_bytes)[0]
word = f.read(word_len).decode("utf8")
freq_bytes = f.read(4)
freq = struct.unpack('I', freq_bytes)[0]
arr_len_bytes = f.read(4)
arr_len = struct.unpack('I', arr_len_bytes)[0]
arr = ""
for i in range(0, arr_len):
string = f.read(4)
number = struct.unpack('I', string)[0]
formated = "{0:b}".format(number)
if len(formated) % 32 != 0:
zeroes = 32 - len(formated) % 32
for i in range(zeroes):
formated = "0" + formated
arr += formated
array = list()
idx = 0
while len(arr) > 0 and "1" in arr:
var = elias.gamma_decode(arr)
idx += 1
array.append(var[0] - 1)
arr = arr[var[1]:]
index[word] = freq, array
word_len_bytes = f.read(4)
idx1 += 1
f.close()
return index
#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)
