def bigram_encode(title):
"""encode a title in bigram form"""
global bigram
result = ''
title = SearchKey.strip_accents(title)
while len(title) >= 2:
if SearchKey.is_valid_character(title[0]):
b = title[0:2]
if b in bigram:
result += bigram[b]
title = title[2:]
else:
result += chr(ord(title[0:1]))
title = title[1:]
else:
#result += '?'
title = title[1:]
if len(title) == 1:
if SearchKey.is_valid_character(title[0]):
result += chr(ord(title[0]))
#else:
# result += '?'
return SearchKey.compact_spaces(result)
def bigram_encode(title):
"""encode a title in bigram form"""
global bigram
result = ''
title = SearchKey.strip_accents(title)
while len(title) >= 2:
if SearchKey.is_valid_character(title[0]):
b = title[0:2]
if b in bigram:
result += bigram[b]
title = title[2:]
else:
result += chr(ord(title[0:1]))
title = title[1:]
else:
#result += '?'
title = title[1:]
if len(title) == 1:
if SearchKey.is_valid_character(title[0]):
result += chr(ord(title[0]))
#else:
# result += '?'
return SearchKey.compact_spaces(result)
def generate_bigram(text):
"""create bigram from pairs of characters"""
global bigram
if len(text) > 2:
try:
if SearchKey.is_valid_character(text[0]) and SearchKey.is_valid_character(text[1]):
bigram[text[0:2]] += 1
except KeyError:
bigram[text[0:2]] = 1
if len(text) > 4:
try:
if SearchKey.is_valid_character(text[2]) and SearchKey.is_valid_character(text[3]):
bigram[text[2:4]] += 1
except KeyError:
bigram[text[2:4]] = 1
def generate_bigram(text):
"""create bigram from pairs of characters"""
global bigram
if len(text) > 2:
try:
if SearchKey.is_valid_character(text[0]) and SearchKey.is_valid_character(text[1]):
bigram[text[0:2]] += 1
except KeyError:
bigram[text[0:2]] = 1
if len(text) > 4:
try:
if SearchKey.is_valid_character(text[2]) and SearchKey.is_valid_character(text[3]):
bigram[text[2:4]] += 1
except KeyError:
bigram[text[2:4]] = 1
