import random
from sklearn.model_selection import train_test_split
import MeCab

mecab = MeCab.Tagger("-Owakati")

def build_vocab(corpus, language="en"):
    vocab = set()
    if language == "en":
        for sentence in corpus:
            tokens = sentence.strip().split()  
            vocab.update(tokens)
    else:  
        for sentence in corpus:
            tokens = mecab.parse(sentence).strip().split()  
            vocab.update(tokens)
    return sorted(vocab)

def preprocess_data(filepath):
    english_sentences = []
    japanese_sentences = []

    with open(filepath, "r", encoding="utf-8") as f:
        for line in f:
            if '\t' in line:
                jp_sentence, en_sentence = line.strip().split("\t") 
                english_sentences.append(en_sentence)
                japanese_sentences.append(jp_sentence)

    return english_sentences, japanese_sentences


def split_dataset(english_sentences, japanese_sentences, train_size=0.8, valid_size=0.1):
    total_size = train_size + valid_size
    en_train_valid, en_test, jp_train_valid, jp_test = train_test_split(english_sentences, japanese_sentences, test_size=1-total_size, random_state=42)
    en_train, en_valid, jp_train, jp_valid = train_test_split(en_train_valid, jp_train_valid, test_size=valid_size/(train_size + valid_size), random_state=42)
    return en_train, en_valid, en_test, jp_train, jp_valid, jp_test


if __name__ == "__main__":
    english_sentences, japanese_sentences = preprocess_data("eng_jpn.txt")
    
    en_vocab = build_vocab(english_sentences, language="en")
    jp_vocab = build_vocab(japanese_sentences, language="jp")

    print(f"English Vocabulary Size: {len(en_vocab)}")
    print(f"Japanese Vocabulary Size: {len(jp_vocab)}")
    
    en_train, en_valid, en_test, jp_train, jp_valid, jp_test = split_dataset(english_sentences, japanese_sentences)
    
    with open("train.txt", "w", encoding="utf-8") as f:
        for jp_sentence, en_sentence in zip(jp_train, en_train):
            f.write(jp_sentence + "\t" + en_sentence + "\n")

    with open("valid.txt", "w", encoding="utf-8") as f:
        for jp_sentence, en_sentence in zip(jp_valid, en_valid):
            f.write(jp_sentence + "\t" + en_sentence + "\n")

    with open("test.txt", "w", encoding="utf-8") as f:
        for jp_sentence, en_sentence in zip(jp_test, en_test):
            f.write(jp_sentence + "\t" + en_sentence + "\n")
    
    print("Dataset splitting complete!")

#English Vocabulary Size: 19920
#Japanese Vocabulary Size: 15086