webui.py

import streamlit as st
import os
import tempfile
import zipfile
import shutil
from pdf2image import convert_from_path
from ppt_converter import create_pptx_from_images, reconstruct_pptx_from_zip

# Page configuration
st.set_page_config(
    page_title="PDF/圖片轉可編輯PPTX - 零成本在線版",
    page_icon="📄",
    layout="centered"
)

# Custom CSS for premium aesthetics (Dark Mode / Glassmorphism Feel)
st.markdown("""
    <style>
    /* Main Background & Fonts */
    @import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;600;700&display=swap');
    
    .stApp {
        font-family: 'Inter', -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
    }
    
    /* Title Style */
    .title-container {
        text-align: center;
        padding: 1.5rem 0;
        margin-bottom: 2rem;
        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
        -webkit-background-clip: text;
        -webkit-text-fill-color: transparent;
    }
    
    /* Tab buttons and cards */
    .css-1n76uvr {
        gap: 0rem;
    }
    
    div.stButton > button:first-child {
        background: linear-gradient(135deg, #4f46e5 0%, #7c3aed 100%);
        color: white;
        border: none;
        padding: 0.6rem 2rem;
        border-radius: 8px;
        font-weight: 600;
        letter-spacing: 0.5px;
        transition: all 0.3s ease;
        box-shadow: 0 4px 12px rgba(124, 58, 237, 0.3);
    }
    
    div.stButton > button:first-child:hover {
        transform: translateY(-2px);
        box-shadow: 0 6px 20px rgba(124, 58, 237, 0.4);
        background: linear-gradient(135deg, #4338ca 0%, #6d28d9 100%);
    }
    
    .card {
        background: rgba(255, 255, 255, 0.05);
        border: 1px solid rgba(255, 255, 255, 0.1);
        border-radius: 12px;
        padding: 1.5rem;
        margin-bottom: 1.5rem;
    }
    
    /* Steps styling for Colab */
    .step-number {
        display: inline-flex;
        align-items: center;
        justify-content: center;
        width: 28px;
        height: 28px;
        border-radius: 50%;
        background-color: #4f46e5;
        color: white;
        font-weight: bold;
        margin-right: 8px;
    }
    
    .colab-btn {
        display: inline-block;
        padding: 0.5rem 1.2rem;
        background-color: #f9ab00;
        color: #1e1e1e !important;
        border-radius: 6px;
        font-weight: bold;
        text-decoration: none;
        transition: background-color 0.3s;
        margin: 10px 0;
        box-shadow: 0 4px 10px rgba(249, 171, 0, 0.2);
    }
    .colab-btn:hover {
        background-color: #e39a00;
    }
    </style>
    """, unsafe_allow_html=True)

# Title and introduction
st.markdown("<div class='title-container'><h1>📄 PDF / 圖片 轉 可編輯 PPTX</h1></div>", unsafe_allow_html=True)
st.markdown("""
<div style='text-align: center; margin-top: -20px; margin-bottom: 30px; color: #6b7280; font-size: 1.05rem;'>
將無文字圖層的 AI 扁平簡報（如 NotebookLM）完美轉回 **「圖文分離、背景修復、可編輯」** 的 PowerPoint 檔！
</div>
""", unsafe_allow_html=True)

# Define Tabs
tab1, tab2 = st.tabs(["⚡ 在線極速轉換 (CPU)", "🚀 雲端高精轉換 (Colab GPU 匯入)"])

# -----------------
# Tab 1: Local CPU Mode
# -----------------
with tab1:
    st.markdown("### ⚡ 在線極速轉換")
    st.write("適合一般排版的 PDF 簡報。使用伺服器 CPU 進行 RapidOCR 辨識與 ONNX LaMa 去字修補，隱私安全，一鍵完成。")
    
    uploaded_file = st.file_uploader("選擇上傳您的 PDF 或圖片檔案", type=['pdf', 'png', 'jpg', 'jpeg', 'webp'], key="cpu_uploader")
    
    with st.expander("⚙️ 進階辨識設定"):
        lang_options = ['ch_tra', 'en', 'ja', 'ko', 'ch_sim']
        selected_langs = st.multiselect(
            "選擇 OCR 辨識語言",
            options=lang_options,
            default=['ch_tra', 'en'],
            help="選擇檔案中包含的語言，可大幅提升辨識準確度。"
        )
        high_accuracy = st.toggle(
            "開啟高精度影像增強 (CLAHE)",
            value=False,
            help="開啟後會對圖片進行對比度增強，提升模糊與微小字體的辨識率，但會稍微增加處理時間。"
        )
        
    st.divider()
    
    if uploaded_file is not None:
        if st.button("🚀 開始線上轉換", key="cpu_convert_btn"):
            if not selected_langs:
                st.error("請至少選擇一種辨識語言！")
            else:
                progress_container = st.empty()
                progress_bar = st.progress(0)
                
                with st.spinner('處理中，正在進行 OCR 辨識與背景 AI 修復，請稍候...'):
                    try:
                        with tempfile.TemporaryDirectory() as temp_dir:
                            input_path = os.path.join(temp_dir, uploaded_file.name)
                            output_pptx_path = os.path.join(temp_dir, "output.pptx")
                            
                            # Save uploaded file
                            with open(input_path, "wb") as f:
                                f.write(uploaded_file.getbuffer())
                                
                            ext = os.path.splitext(input_path)[1].lower()
                            image_paths = []
                            
                            if ext == ".pdf":
                                progress_container.info("正在解析 PDF 並轉換為高清圖片...")
                                images = convert_from_path(input_path)
                                total_p = len(images)
                                for idx, img in enumerate(images):
                                    temp_img_path = os.path.join(temp_dir, f"page_{idx}.png")
                                    img.save(temp_img_path, "PNG")
                                    image_paths.append(temp_img_path)
                                    progress_bar.progress(int((idx + 1) / total_p * 20)) # up to 20%
                            elif ext in [".png", ".jpg", ".jpeg", ".webp"]:
                                image_paths = [input_path]
                                progress_bar.progress(20)
                                
                            # Callback for conversion progress
                            def update_conversion_progress(current, total):
                                percent = 20 + int((current / total) * 80)
                                progress_bar.progress(min(percent, 100))
                                progress_container.info(f"正在執行 AI 去字與重組簡報：第 {current} / {total} 頁...")
                                return True
                                
                            # Convert
                            success = create_pptx_from_images(
                                image_paths, 
                                output_pptx_path, 
                                selected_langs, 
                                high_accuracy=high_accuracy,
                                progress_callback=update_conversion_progress
                            )
                            
                            if success:
                                # Read results
                                with open(output_pptx_path, "rb") as f:
                                    pptx_data = f.read()
                                    
                                progress_container.success("🎉 轉換成功！請點擊下方按鈕下載 PPTX。")
                                progress_bar.progress(100)
                                
                                st.download_button(
                                    label="⬇️ 下載可編輯 PPTX 檔案",
                                    data=pptx_data,
                                    file_name=f"{os.path.splitext(uploaded_file.name)[0]}_editable.pptx",
                                    mime="application/vnd.openxmlformats-officedocument.presentationml.presentation",
                                    key="cpu_download"
                                )
                            else:
                                progress_container.error("轉換取消或失敗。")
                                
                    except Exception as e:
                        progress_container.error(f"轉換過程中發生錯誤: {str(e)}")
                        st.exception(e)

# -----------------
# Tab 2: Colab GPU Import Mode
# -----------------
with tab2:
    st.markdown("### 🚀 雲端高精轉換 (Google Colab GPU 分流)")
    st.write("適用於多欄位、包含複雜圖表或超大檔案的簡報。此模式利用多模態大模型進行視覺定位，提供最極致的還原效果，且**完全免費**。")
    
    st.markdown("""
    <div class="card">
        <h4 style="margin-top:0;">💡 簡單三步驟完成高精度轉換：</h4>
        <div style="margin-bottom:10px;">
            <span class="step-number">1</span>
            點擊下方按鈕，在您自己的瀏覽器中打開 Google Colab 筆記本（會自動分配免費的 T4 GPU 算力）：
        </div>
        <div style="text-align:center;">
            <a href="https://colab.research.google.com/" target="_blank" class="colab-btn">💻 開啟 Google Colab 轉換腳本</a>
        </div>
        <div style="margin-bottom:10px; margin-top:10px;">
            <span class="step-number">2</span>
            在 Colab 中上傳您的 PDF 檔案，點擊「全部執行」，等待 1~3 分鐘後下載產生的 <code>Pack.zip</code> 壓縮包。
        </div>
        <div>
            <span class="step-number">3</span>
            將下載的 <code>Pack.zip</code> 拖入下方上傳區，即可在一秒內拼裝並下載您的完美可編輯 PPTX！
        </div>
    </div>
    """, unsafe_allow_html=True)
    
    zip_file = st.file_uploader("請上傳 Colab 產出的 Pack.zip 壓縮包", type=['zip'], key="zip_uploader")
    
    if zip_file is not None:
        if st.button("🎨 一秒拼裝簡報", key="zip_assemble_btn"):
            with st.spinner('正在解壓並映射座標...'):
                try:
                    with tempfile.TemporaryDirectory() as temp_dir:
                        zip_path = os.path.join(temp_dir, zip_file.name)
                        output_pptx_path = os.path.join(temp_dir, "reconstructed.pptx")
                        
                        # Save uploaded zip
                        with open(zip_path, "wb") as f:
                            f.write(zip_file.getbuffer())
                            
                        # Reconstruct
                        success = reconstruct_pptx_from_zip(zip_path, output_pptx_path)
                        
                        if success:
                            with open(output_pptx_path, "rb") as f:
                                pptx_data = f.read()
                                
                            st.success("🎉 拼裝完成！請點擊下方按鈕下載 PPTX。")
                            
                            st.download_button(
                                label="⬇️ 下載已拼裝 PPTX 檔案",
                                data=pptx_data,
                                file_name=f"{os.path.splitext(zip_file.name)[0]}_assembled.pptx",
                                mime="application/vnd.openxmlformats-officedocument.presentationml.presentation",
                                key="zip_download"
                            )
                        else:
                            st.error("組裝失敗，請確認上傳的 ZIP 封包格式是否正確。")
                            
                except Exception as e:
                    st.error(f"組裝過程中發生錯誤: {str(e)}")
                    st.exception(e)

# Footer
st.markdown("""
<div style='text-align: center; margin-top: 50px; font-size: 0.8rem; color: #9ca3af; border-top: 1px solid rgba(255,255,255,0.05); padding-top: 15px;'>
PDF/圖片轉可編輯PPTX系統 • 部署於 Hugging Face Spaces (CPU Basic) • 開源免運維版本
</div>
""", unsafe_allow_html=True)