IVcandidates = primary.generate(word)
# if no primary candidates generated
if len(IVcandidates) == 0:
IVcandidates = secondary.generate(word)
# if no secondary candidates generated
if len(IVcandidates) == 0:
class_number = 1
correct_word = '-'
class_numb = 1
else:
prev_tokens = selector.prev_tokens((word, pos), for_prev)
correct_word = selector.choose(prev_tokens, IVcandidates)
for_prev[for_prev.index((word, pos))] = (correct_word, pos)
# if class is correct or NoES
else:
correct_word = '-'
correct[tweet_id][j].append((word, class_number, correct_word))
output.build(splitter.texts, splitter.order, correct)
# agregué class number to correct
# uso tokenizer de language modeling?
# abreviaturas con o sin punto
# para nombres, hago split y que tomen parte??
# check sound....la h?? ejemplo Shanto da santo
# spelling if termina con b, c
# como hago lo del sms.txt
# en secondary agrego algo de reps
# hago memoria de candidatos?
# uso ascii or hardcode?
# risas terminadas en j o empezadas en vocal
