| import textdistance | |
| import pickle | |
| with open('model/correction_model.pkl', 'rb') as model_file: | |
| correction_dict = pickle.load(model_file) | |
| def predict_correction(incorrect_word, correction_dict): | |
| if incorrect_word in correction_dict: | |
| return correction_dict[incorrect_word], 0 | |
| else: | |
| best_match = None | |
| best_distance = float('inf') | |
| for correct_word in correction_dict.keys(): | |
| distance = textdistance.levenshtein(incorrect_word, correct_word) | |
| if distance < best_distance: | |
| best_distance = distance | |
| best_match = correct_word | |
| return best_match, best_distance | |
| def correct_text(input_text, correction_dict): | |
| corrected_text = [] | |
| words = input_text.split() | |
| for word in words: | |
| best_correction, distance = predict_correction(word, correction_dict) | |
| if best_correction: | |
| corrected_text.append(best_correction) | |
| else: | |
| corrected_text.append(word) | |
| return ' '.join(corrected_text) | |
| def wordmain (input): | |
| input_text = input | |
| corrected_text = correct_text(input_text, correction_dict) | |
| return corrected_text |