app.py

import openai
import os 
import gradio as gr
from gtts import gTTS
import tempfile
import re # regular expression

# import requests

### Translation function

def translate_to_japanese(api_key, text): #openai api key
    if not api_key:
        return "Error: API key is missing.", None

    if not text:
        return "Error: Input text is empty.", None

    openai.api_key = api_key

    #### translation 

    messages_translation = [
        {"role": "system", "content": "You are a helpful translator."},
        {"role": "user", "content": f"Translate the following English text to Japanese:\n\n{text}"} #prompt template
    ]

    try:
        response_translation = openai.ChatCompletion.create(
            model="gpt-4o",
            messages=messages_translation,
            max_tokens=300,
            temperature=0.5
        )

        translated_text=response_translation.choices[0].message['content'].strip()

        messages_pronunciation = [
            {"role": "system", "content": "You are a helpful assistant who provides only the phonetic pronunciation (Romaji) of Japanese text, in square brackets."},
            {"role": "user", "content": f"Provide only the Romaji pronunciation (phonetic) for this Japanese text:\n\n{translated_text}"} #prompt template
        ]

        ### pronounciation
        response_pronunciation = openai.ChatCompletion.create(
            model="gpt-4o",
            messages=messages_pronunciation,
            max_tokens=150,
            temperature=0.5
        )

        pronunciation=response_pronunciation.choices[0].message['content'].strip() ### the pron of ___ is konichuwa

        return translated_text, pronunciation

    except openai.error.OpenAIError as e:
        return f"OpenAI API error: {str(e)}", None
    except Exception as e:
        return f"Unexpected error: {str(e)}", None

#### clean pronounciation

def clean_pronunciation(pronunciation_text):
    pronunciation_cleaned=re.sub(r"(?i)(the pronunciation.*?is[:：]*|\n|\"|\')", "", pronunciation_text).strip()
    # Keep only the text inside [ ] if present
    match = re.search(r"\[.*?\]", pronunciation_cleaned)
    if match:
        pronunciation_cleaned = match.group(0)
    return pronunciation_cleaned         ### the balah blah ignore [dsadasdasd]   response: [dsadasdasd] 

#### audio generation function

def generate_audio_from_text(text):
    tts=gTTS(text, lang="ja")
    temp_audio_file=tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") #.wav
    tts.save(temp_audio_file.name)
    return temp_audio_file.name

# ---------------------- MAIN FUNCTION ----------------------

def process_translation(api_key, english_text):
    if not api_key:
        return "Missing API Key", "", "", None, None
    if not english_text:
        return "Please enter English text.", "", "", None, None

    translated_text, pronunciation = translate_to_japanese(api_key, english_text)

    if not pronunciation:
        return translated_text, "", "", None, None

    cleaned_pronunciation = clean_pronunciation(pronunciation) ### konichuwa

    # Create result text file
    result_text = f"English Text: {english_text}\n\nJapanese Translation: {translated_text}\nPronunciation: {cleaned_pronunciation}" ## hello ___ konichuwa
    result_file = tempfile.NamedTemporaryFile(delete=False, suffix=".txt")
    with open(result_file.name, "w", encoding="utf-8") as f:
        f.write(result_text)

    # Generate pronunciation audio
    audio_path = generate_audio_from_text(cleaned_pronunciation)

    return "", translated_text, cleaned_pronunciation, result_file.name, audio_path

### UI design

with gr.Blocks(theme=gr.themes.Ocean()) as app:
    gr.Markdown("# 🇯🇵 English → Japanese Translator with Pronunciation")
    gr.Markdown("Translate English text into **Japanese**, get **Romaji pronunciation**, and listen to it!")

    api_key = gr.Textbox(label="🔑 OpenAI API Key", type="password", placeholder="Enter your OpenAI API key")

    english_text = gr.Textbox(label="📝 Enter English Text", lines=4, placeholder="Type something in English...")

    translate_button = gr.Button("Translate to Japanese 🚀")

    translation_output = gr.Textbox(label="📘 Japanese Translation")

    pronunciation_output = gr.Textbox(label="🔤 Pronunciation (Romaji)")

    audio_output=gr.Audio(label="🔊 Listen to Pronunciation", type="filepath")

    file_output = gr.File(label="⬇️ Download Translation Result")

    translate_button.click(
        fn=process_translation,
        inputs=[api_key, english_text],
        outputs=[gr.Label(), translation_output, pronunciation_output, file_output, audio_output]
    )

app.launch()