import streamlit as st
import requests
import base64
import sounddevice as sd
import soundfile as sf
import tempfile

# ✅ 替換成你的 ngrok 或 render 後端
API_BASE = "https://f876-125-228-143-171.ngrok-free.app "

st.set_page_config(page_title="📇 名片辨識與語音輸入", layout="centered")
st.title("📷 拍照 + 🎙 語音辨識系統")

st.header("📸 名片拍照辨識")
img_file = st.camera_input("請拍攝名片")
if img_file:
    st.image(img_file, caption="名片預覽", use_column_width=True)
    img_bytes = img_file.getvalue()
    base64_img = base64.b64encode(img_bytes).decode("utf-8")
    payload = {"image": f"data:image/jpeg;base64,{base64_img}"}
    with st.spinner("🧠 OCR 處理中..."):
        try:
            res = requests.post(f"{API_BASE}/ocr", json=payload)
            text = res.json().get("text", "")
            st.text_area("📄 辨識結果", value=text, height=200)
        except Exception as e:
            st.error(f"❌ 錯誤：{e}")

st.header("🎤 語音辨識（Whisper）")
duration = st.slider("⏱ 錄音秒數", 2, 10, 4)
if st.button("🎙 開始錄音"):
    with st.spinner("🎧 錄音中..."):
        fs = 16000
        audio = sd.rec(int(duration * fs), samplerate=fs, channels=1)
        sd.wait()
        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
            sf.write(tmp.name, audio, fs)
            tmp.seek(0)
            audio_bytes = tmp.read()
        st.audio(audio_bytes, format="audio/wav")
        with st.spinner("🧠 Whisper 處理中..."):
            try:
                res = requests.post(
                    f"{API_BASE}/whisper",
                    files={"file": ("audio.wav", audio_bytes, "audio/wav")}
                )
                result = res.json().get("text", "")
                st.text_area("📝 語音辨識結果", value=result, height=200)
            except Exception as e:
                st.error(f"❌ Whisper API 錯誤：{e}")