kafiye_üreteci.py

# -*- coding: utf-8 -*-
"""kafiye_üreteci.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1jwglO5r6JVfWsN0sAcPBqTUmCIOaMajE
"""


from trnlp import *
import re

turkish_words = set(line.strip() for line in open('all_NEW_turkish_Words.txt'))

def vowel_extractor(a_word):
  vowel_instance = tuple("aeıioöuüâî")
  all_vowels = [char for char in a_word if char in vowel_instance]
  return tuple(all_vowels)


def syllable_harmony(a_word):
    vowel_instance = re.compile(r'[aeıioöuüâî]')
    consonant_instance = re.compile(r'[bcçdfgğhjklmnprsştvyzxqw]')
    word_temporary = syllabification(a_word)
    harmony = []

    for hece in word_temporary:
        substituted_syllable = re.sub(vowel_instance, 'V', hece)
        substituted_syllable = re.sub(consonant_instance, 'C', substituted_syllable)
        harmony.append(substituted_syllable)


    return tuple(harmony)



all_turkish_words_dict = {item: [
    tuple(syllabification(item)),
    vowel_extractor(item),
    syllable_harmony(item),
    len(syllabification(item))
    ] for item in turkish_words}

def rhyme_generator(input_word):

  input_word = input_word.replace(" ", "").lower()
  input_syllabification = tuple(syllabification(input_word))
  input_vowels = vowel_extractor(input_word)
  input_harmony = syllable_harmony(input_word)
  input_length = len(input_syllabification)

  usable_words = [word for word in turkish_words if len(syllabification(word)) == input_length]

  only_words = [item for item in usable_words if levenshtein_distance(input_word, item) <= 5]


  filtered_step_1 = []

  for item in only_words:
    tuple1_2 = input_vowels
    if item in all_turkish_words_dict:
        tuple2_2 = all_turkish_words_dict[item][1]
        score = sum(1 for i in range(min(len(tuple1_2), len(tuple2_2))) if tuple1_2[i] == tuple2_2[i])
        if score == len(input_vowels):
          filtered_step_1.append(item)
        else:
          pass
    else:
        pass

  filtered_step_2 = []

  for item in filtered_step_1:
    tuple1_1 = input_harmony
    if item in all_turkish_words_dict:
        tuple2_1 = all_turkish_words_dict[item][2]
        score = sum(1 for i in range(min(len(tuple1_1), len(tuple2_1))) if tuple1_1[i] == tuple2_1[i])
        if score <= 1:
          pass
        else:
          filtered_step_2.append(item)
    else:
        pass

  return filtered_step_2

print(rhyme_generator("saksafon"))