Upload 5 files

README.md

@@ -1,12 +1,34 @@
 ---
-title: Apec Query System
-emoji: 🔥
-colorFrom: indigo
-colorTo: gray
+title: APEC 綜合查詢系統
+emoji: 🌐
+colorFrom: blue
+colorTo: green
 sdk: gradio
-sdk_version: 6.2.0
+sdk_version: 5.9.1
 app_file: app.py
 pinned: false
+license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🌐 APEC 綜合查詢系統
+
+整合 APEC 提案項目與 FSMM 宣言內容的統一檢索平台。
+
+## 功能特色
+
+- **📊 統計儀表板**：即時顯示提案統計、經濟體分佈、熱門搜尋關鍵字
+- **🍎 FSMM 宣言查詢**：搜尋歷年 FSMM 部長級會議宣言與聲明
+- **📋 APEC 提案查詢**：依年份、經濟體、關鍵字搜尋 APEC 提案
+- **☁️ 上傳更新**：支援批次上傳 PDF/DOCX/HTML 文件自動解析
+
+## 技術架構
+
+- **前端**: Gradio
+- **資料處理**: pandas, pdfminer, python-docx
+- **視覺化**: Plotly
+
+## 使用方式
+
+1. 點擊「統計儀表板」查看整體數據概覽
+2. 點擊儀表板中的年份或經濟體可直接跳轉查詢
+3. 使用搜尋功能輸入關鍵字進行全文檢索

app.py

@@ -0,0 +1,546 @@
+import os
+import json
+import re
+import gradio as gr
+import pandas as pd
+import plotly.express as px
+import plotly.graph_objects as go
+from datetime import datetime
+import threading
+from pdfminer.high_level import extract_text
+from docx import Document
+from bs4 import BeautifulSoup
+
+# Global lock for file access to prevent race conditions during log/load
+data_lock = threading.Lock()
+
+# Config
+DATA_FSMM = "fsmm_data.json"
+DATA_PROPOSALS = "proposals_data.json"
+DATA_QUERIES = "user_queries.json"
+
+# --- FSMM Extraction Logic ---
+def extract_from_pdf(file_path):
+    try:
+        text = extract_text(file_path)
+        return text
+    except Exception as e:
+        print(f"Error reading PDF {file_path}: {e}")
+        return ""
+
+def extract_from_docx(file_path):
+    try:
+        doc = Document(file_path)
+        return "\n".join([para.text for para in doc.paragraphs])
+    except Exception as e:
+        print(f"Error reading DOCX {file_path}: {e}")
+        return ""
+
+def parse_fsmm_filename(filename):
+    match = re.search(r'(\d+)_fsmm_(\w+)', filename)
+    if match:
+        year_short = match.group(1)
+        type_code = match.group(2)
+        year = int(year_short) + 2000
+        
+        type_map = {
+            'jms': 'Joint Ministerial Statement (聯合部長聲明)',
+            'stmt': 'Statement (聲明)',
+            'declaration': 'Declaration (宣言)'
+        }
+        type_name = type_map.get(type_code.lower(), type_code.upper())
+        return year, type_name
+    return None, None
+
+# --- Proposal Extraction Logic ---
+def parse_proposal(file_path):
+    try:
+        with open(file_path, 'r', encoding='utf-8') as f:
+            html_content = f.read()
+        soup = BeautifulSoup(html_content, 'html.parser')
+        data = {'file_id': os.path.basename(file_path).replace('.html', '')}
+        rows = soup.find_all('tr')
+        for row in rows:
+            label_td = row.find('td', class_='ms-formlabel')
+            value_td = row.find('td', class_='ms-formbody')
+            if label_td and value_td:
+                label = label_td.get_text(strip=True).replace(':', '')
+                value = value_td.get_text(separator=' ', strip=True)
+                value = re.sub(r'\s+', ' ', value)
+                data[label] = value
+        return data
+    except Exception as e:
+        print(f"Error parsing proposal {file_path}: {e}")
+        return None
+
+# --- Data Management ---
+def load_json(filepath):
+    with data_lock:
+        if not os.path.exists(filepath):
+            return []
+        try:
+            with open(filepath, 'r', encoding='utf-8') as f:
+                return json.load(f)
+        except:
+            return []
+
+def save_json(filepath, data):
+    with data_lock:
+        with open(filepath, 'w', encoding='utf-8') as f:
+            json.dump(data, f, ensure_ascii=False, indent=2)
+
+def log_query(query):
+    if not query or not str(query).strip():
+        return
+    logs = load_json(DATA_QUERIES)
+    now = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    logs.append({
+        "timestamp": now,
+        "query": str(query).strip()
+    })
+    save_json(DATA_QUERIES, logs)
+
+# --- Upload Logic ---
+def handle_unified_upload(file_objs):
+    if not file_objs:
+        return "請選擇檔案上傳。"
+    
+    # Gradio might return a single file or a list depending on configuration
+    if not isinstance(file_objs, list):
+        file_objs = [file_objs]
+    
+    results = []
+    
+    for file_obj in file_objs:
+        filename = os.path.basename(file_obj.name)
+        
+        if filename.startswith('Proposal_') and filename.endswith('.html'):
+            data = parse_proposal(file_obj.name)
+            if not data:
+                results.append(f"❌ {filename}: 提案解析失敗。")
+                continue
+            
+            proposals = load_json(DATA_PROPOSALS)
+            proposals = [p for p in proposals if p.get('file_id') != data.get('file_id')]
+            proposals.append(data)
+            save_json(DATA_PROPOSALS, proposals)
+            results.append(f"✅ {filename}: 已加入提案 {data.get('Project Title', filename)}")
+            
+        elif '_fsmm_' in filename:
+            year, doc_type = parse_fsmm_filename(filename)
+            if not year:
+                results.append(f"❌ {filename}: 無法解析 FSMM 檔名格式。")
+                continue
+            
+            if filename.endswith('.pdf'): content = extract_from_pdf(file_obj.name)
+            elif filename.endswith('.docx'): content = extract_from_docx(file_obj.name)
+            else:
+                results.append(f"❌ {filename}: 僅支援 .pdf 或 .docx")
+                continue
+            
+            if not content:
+                results.append(f"❌ {filename}: 內容提取失敗。")
+                continue
+            
+            # --- Improved Paragraph Re-assembly ---
+            # Paragraphs are numbered like "1.", "2.", "3." etc.
+            # Each paragraph runs from one number to the next.
+            
+            # First, normalize content by replacing multiple newlines with single space
+            lines = content.split('\n')
+            full_text = ' '.join([l.strip() for l in lines if l.strip()])
+            
+            # Split by numbered paragraph pattern using re.split with capturing group
+            # This splits text like: "1. Text here 2. More text" into ['', '1', '. Text here ', '2', '. More text']
+            parts = re.split(r'\b(\d+)\.\s', full_text)
+            
+            composed_paras = []
+            i = 0
+            # First part (before first number) might be preamble
+            if parts[0].strip() and len(parts[0].strip()) > 20:
+                composed_paras.append(parts[0].strip())
+            
+            # Process numbered paragraphs: parts[1] is number, parts[2] is content, etc.
+            i = 1
+            while i < len(parts) - 1:
+                if parts[i].isdigit():
+                    para_num = parts[i]
+                    para_content = parts[i + 1].strip() if i + 1 < len(parts) else ""
+                    # The content should NOT start with the number again
+                    # Just combine: "1. content"
+                    full_para = f"{para_num}. {para_content}"
+                    composed_paras.append(full_para)
+                    i += 2
+                else:
+                    i += 1
+            
+            new_entries = []
+            for idx, para in enumerate(composed_paras):
+                if len(para) < 10: continue 
+                new_entries.append({
+                    'year': year, 
+                    'type': doc_type, 
+                    'content': para,
+                    'paragraph_index': idx + 1, 
+                    'filename': filename
+                })
+            
+            fsmm_data = load_json(DATA_FSMM)
+            fsmm_data = [item for item in fsmm_data if item['filename'] != filename]
+            fsmm_data.extend(new_entries)
+            save_json(DATA_FSMM, fsmm_data)
+            results.append(f"✅ {filename}: 整合了 {len(new_entries)} 個完整段落。")
+        else:
+            results.append(f"❓ {filename}: 不支援的檔案格式或命名。")
+            
+    return "\n".join(results)
+
+# --- Search Logic ---
+def search_proposals(query, year, economy):
+    log_query(f"Proposals Q:{query}|Y:{year}|E:{economy}")
+    data = load_json(DATA_PROPOSALS)
+    
+    filtered = []
+    q = query.lower() if query else ""
+    y = str(year) if year else ""
+    e = str(economy) if economy else ""
+
+    for item in data:
+        # Keyword search
+        if q:
+            found_q = False
+            for k, v in item.items():
+                if q in str(v).lower():
+                    found_q = True
+                    break
+            if not found_q: continue
+        
+        # Year filter
+        if y and str(item.get('Project Year', '')).strip() != y:
+            continue
+            
+        # Economy filter
+        if e and str(item.get('Proposing Economy(ies)', '')).strip() != e:
+            continue
+            
+        filtered.append(item)
+    
+    if not filtered: return "找不到符合的提案。"
+    
+    html = ""
+    for p in filtered[:20]:
+        html += f"""
+        <div style="border: 1px solid #cbd5e1; padding: 15px; border-radius: 8px; margin-bottom: 15px; background: #fff; box-shadow: 0 1px 3px rgba(0,0,0,0.1);">
+            <div style="font-weight: bold; color: #1e293b; margin-bottom: 8px; font-size: 1.1em;">{p.get('Project Title', '無標題')}</div>
+            <div style="display: flex; gap: 8px; flex-wrap: wrap; font-size: 0.85em; margin-bottom: 10px;">
+                <span style="background: #e2e8f0; padding: 2px 8px; border-radius: 4px;">{p.get('Project No.', '無編號')}</span>
+                <span style="background: #f1f5f9; padding: 2px 8px; border-radius: 4px;">{p.get('Proposing Economy(ies)', '未知經濟體')}</span>
+            </div>
+            <details style="font-size: 0.9em; color: #475569;">
+                <summary style="cursor: pointer; color: #2563eb;">詳細資訊</summary>
+                <div style="margin-top: 10px; display: grid; grid-template-columns: 1fr 2fr; gap:5px;">
+                    {"".join([f"<b>{k}:</b> <div>{v}</div>" for k, v in p.items() if k not in ['Project Title', 'Project No.']])}
+                </div>
+            </details>
+        </div>
+        """
+    return html
+
+def search_fsmm(query, year, doc_type, filename_filter):
+    log_query(f"FSMM {query}")
+    data = load_json(DATA_FSMM)
+    filtered = []
+    q = query.lower() if query else ""
+    
+    for item in data:
+        if q and q not in item['content'].lower(): continue
+        if year and str(item['year']) != year: continue
+        if doc_type and item['type'] != doc_type: continue
+        if filename_filter and item['filename'] != filename_filter: continue
+        filtered.append(item)
+    
+    if not filtered: return "找不到符合的宣言段���。"
+    
+    # Sort by paragraph index if a specific file is selected, otherwise by year
+    if filename_filter:
+        filtered.sort(key=lambda x: x['paragraph_index'])
+    else:
+        filtered.sort(key=lambda x: (x['year'], x['type'], x['paragraph_index']), reverse=True)
+    
+    html = ""
+    for item in filtered[:100]:
+        citation = f"(APEC FSMM, {item['year']}, {item['type']}, Para {item['paragraph_index']})"
+        html += f"""
+        <div style="border: 1px solid #e2e8f0; padding: 15px; border-radius: 8px; margin-bottom: 15px; background: white; box-shadow: 0 1px 3px rgba(0,0,0,0.1);">
+            <div style="display: flex; gap: 8px; margin-bottom: 8px;">
+                <span style="background: #e0e7ff; color: #4338ca; padding: 2px 10px; border-radius: 15px; font-size: 0.8em; font-weight: bold;">{item['year']}</span>
+                <span style="background: #fef3c7; color: #92400e; padding: 2px 10px; border-radius: 15px; font-size: 0.8em; font-weight: bold;">{item['type']}</span>
+                <span style="background: #f1f5f9; color: #475569; padding: 2px 10px; border-radius: 15px; font-size: 0.8em; font-weight: bold;">Para {item['paragraph_index']}</span>
+            </div>
+            <p style="margin-bottom: 12px; line-height: 1.6;">{item['content']}</p>
+            <div style="text-align: right; font-size: 0.85em; color: #64748b; margin-top: 5px;">
+                檔案: {item['filename']} | <code style="background: #f4f4f5; padding: 2px 6px; border-radius: 4px;">{citation}</code>
+            </div>
+        </div>
+        """
+    return html
+
+def get_fsmm_full_text(filename):
+    if not filename: return "請先選擇一個檔案。"
+    data = load_json(DATA_FSMM)
+    doc_paras = [item for item in data if item['filename'] == filename]
+    doc_paras.sort(key=lambda x: x['paragraph_index'])
+    
+    if not doc_paras: return "找不到該檔案的文件內容。"
+    
+    full_text = f"<h3>{filename} 全文內容</h3><hr>"
+    for p in doc_paras:
+        full_text += f"<p style='margin-bottom: 10px; line-height: 1.8;'>[{p['paragraph_index']}] {p['content']}</p>"
+    
+    return full_text
+
+# --- Dashboard Logic ---
+def get_dashboard_stats():
+    # 1. Proposals Stats
+    proposals = load_json(DATA_PROPOSALS)
+    fsmm = load_json(DATA_FSMM)
+    queries = load_json(DATA_QUERIES)
+
+    total_proposals = len(proposals)
+    # FSMM is stored as paragraphs, we need to count unique filenames
+    unique_fsmm_files = len(set(p['filename'] for p in fsmm))
+    total_fsmm_paras = len(fsmm)
+
+    summary_html = f"""
+    <div style="display: flex; gap: 20px; justify-content: space-around; margin-bottom: 20px;">
+        <div style="background: linear-gradient(135deg, #667eea 0%, #764ba2 100%); color: white; padding: 20px; border-radius: 12px; flex: 1; text-align: center; box-shadow: 0 4px 6px rgba(0,0,0,0.1);">
+            <div style="font-size: 0.9em; opacity: 0.9;">總提案數量</div>
+            <div style="font-size: 2.5em; font-weight: bold;">{total_proposals}</div>
+        </div>
+        <div style="background: linear-gradient(135deg, #2af598 0%, #009efd 100%); color: white; padding: 20px; border-radius: 12px; flex: 1; text-align: center; box-shadow: 0 4px 6px rgba(0,0,0,0.1);">
+            <div style="font-size: 0.9em; opacity: 0.9;">FSMM 文件數</div>
+            <div style="font-size: 2.5em; font-weight: bold;">{unique_fsmm_files}</div>
+        </div>
+        <div style="background: linear-gradient(135deg, #ff9a9e 0%, #fecfef 100%); color: #4338ca; padding: 20px; border-radius: 12px; flex: 1; text-align: center; box-shadow: 0 4px 6px rgba(0,0,0,0.1);">
+            <div style="font-size: 0.9em; opacity: 0.9;">FSMM 總段落</div>
+            <div style="font-size: 2.5em; font-weight: bold;">{total_fsmm_paras}</div>
+        </div>
+    </div>
+    """
+
+    # 2. Proposals by Year Chart
+    df_p = pd.DataFrame(proposals)
+    fig_p_year = go.Figure()
+    if not df_p.empty and 'Project Year' in df_p.columns:
+        year_counts = df_p['Project Year'].value_counts().reset_index()
+        year_counts.columns = ['年份', '提案數量']
+        year_counts['年份'] = year_counts['年份'].astype(str)
+        year_counts = year_counts.sort_values('年份')
+        fig_p_year = px.bar(year_counts, x='年份', y='提案數量', title="APEC 提案歷年數量分佈", 
+                            labels={'提案數量': '數量', '年份': '年份'}, color_discrete_sequence=['#6366f1'])
+        fig_p_year.update_layout(template="plotly_white")
+
+    # 3. Proposals by Economy Pie Chart
+    fig_p_econ = go.Figure()
+    if not df_p.empty and 'Proposing Economy(ies)' in df_p.columns:
+        econ_counts = df_p['Proposing Economy(ies)'].value_counts().reset_index()
+        econ_counts.columns = ['經濟體', '提案數量']
+        # Take top 10 and group others
+        if len(econ_counts) > 10:
+            top_10 = econ_counts.head(10)
+            others_sum = econ_counts.iloc[10:]['提案數量'].sum()
+            others_df = pd.DataFrame([{'經濟體': '其他', '提案數量': others_sum}])
+            econ_counts = pd.concat([top_10, others_df])
+        
+        fig_p_econ = px.pie(econ_counts, values='提案數量', names='經濟體', title="提案主辦經濟體分佈",
+                            color_discrete_sequence=px.colors.sequential.RdBu)
+        fig_p_econ.update_layout(template="plotly_white")
+
+    # 4. User Query Stats
+    fig_queries = go.Figure()
+    if queries:
+        df_q = pd.DataFrame(queries)
+        q_counts = df_q['query'].value_counts().head(10).reset_index()
+        q_counts.columns = ['關鍵字', '搜尋次數']
+        fig_queries = px.bar(q_counts, x='搜尋次數', y='關鍵字', orientation='h', title="熱門搜尋關鍵字 TOP 10",
+                             labels={'搜尋次數': '次數', '關鍵字': '關鍵字'}, color='搜尋次數',
+                             color_continuous_scale='Viridis')
+        fig_queries.update_layout(template="plotly_white", yaxis={'categoryorder':'total ascending'})
+
+    # 5. Dataframes for Drill-down
+    df_ye_stats = pd.DataFrame(year_counts) if not df_p.empty and 'Project Year' in df_p.columns else pd.DataFrame()
+    df_ec_stats = pd.DataFrame(econ_counts) if not df_p.empty and 'Proposing Economy(ies)' in df_p.columns else pd.DataFrame()
+    df_qu_stats = pd.DataFrame(q_counts) if queries else pd.DataFrame()
+
+    return summary_html, fig_p_year, fig_p_econ, fig_queries, df_ye_stats, df_ec_stats, df_qu_stats
+
+# --- UI Helpers ---
+def get_p_choices_raw(year_filter=None):
+    data = load_json(DATA_PROPOSALS)
+    years = sorted(list(set(str(it.get('Project Year', '')).strip() for it in data if it.get('Project Year'))), reverse=True)
+    if year_filter:
+        econs = sorted(list(set(str(it.get('Proposing Economy(ies)', '')).strip() 
+                               for it in data if str(it.get('Project Year', '')).strip() == str(year_filter))))
+    else:
+        econs = sorted(list(set(str(it.get('Proposing Economy(ies)', '')).strip() for it in data if it.get('Proposing Economy(ies)'))))
+    return ([""] + years), ([""] + econs)
+
+def get_fsmm_choices_raw():
+    data = load_json(DATA_FSMM)
+    years = sorted(list(set(str(it['year']) for it in data)), reverse=True)
+    types = sorted(list(set(it['type'] for it in data)))
+    filenames = sorted(list(set(it['filename'] for it in data)), reverse=True)
+    return ([""] + years), ([""] + types), ([""] + filenames)
+
+def refresh_p_choices(year_filter=None):
+    years, econs = get_p_choices_raw(year_filter)
+    return gr.update(choices=years), gr.update(choices=econs)
+
+def refresh_fsmm_choices():
+    years, types, files = get_fsmm_choices_raw()
+    return gr.update(choices=years), gr.update(choices=types), gr.update(choices=files)
+
+def handle_drilldown(evt: gr.SelectData):
+    try:
+        selected_val = str(evt.value).strip()
+        print(f"Drill-down triggered: {selected_val}")
+
+        # Check if it's a Year (4 digits)
+        if re.match(r'^\d{4}$', selected_val):
+            print("Detected: Year -> FSMM search")
+            res_fsmm = search_fsmm("", selected_val, "", "")
+            return gr.update(selected="fsmm_tab"), res_fsmm, ""
+        
+        # Check if it's an Economy or keyword (fallback)
+        economies = ["China", "Chinese Taipei", "Australia", "Thailand", "United States", "Korea", "Japan", "Malaysia", "Viet Nam", "Philippines", "Others", "其他"]
+        if selected_val in economies:
+            print("Detected: Economy -> Proposals search")
+            res_p = search_proposals("", "", selected_val)
+            return gr.update(selected="proposal_tab"), "", res_p
+
+        # Fallback: Keyword search on FSMM
+        print("Detected: Keyword -> FSMM search")
+        res_fsmm = search_fsmm(selected_val, "", "", "")
+        return gr.update(selected="fsmm_tab"), res_fsmm, ""
+    except Exception as e:
+        print(f"Error in handle_drilldown: {e}")
+        import traceback
+        traceback.print_exc()
+        return gr.update(), "", ""
+
+# --- UI UI UI ---
+DASHBOARD_CSS = """
+    .drilldown-df table tr:hover {
+        cursor: pointer !important;
+        background-color: #f0f9ff !important;
+        position: relative;
+    }
+    .drilldown-df table tr:hover::after {
+        content: "🖱️ 點擊跳轉查詢";
+        position: absolute;
+        right: 10px;
+        top: 50%;
+        transform: translateY(-50%);
+        font-size: 0.8em;
+        color: #3b82f6;
+        background: #fff;
+        padding: 2px 6px;
+        border-radius: 4px;
+        box-shadow: 0 2px 4px rgba(0,0,0,0.1);
+        pointer-events: none;
+    }
+    .drilldown-df table tr:hover td {
+        color: #2563eb !important;
+        font-weight: bold;
+    }
+"""
+
+with gr.Blocks(title="APEC 綜合查詢系統") as demo:
+    gr.Markdown("# 🌐 APEC 綜合查詢系統")
+    gr.Markdown("整合 APEC 提案項目與 FSMM 宣言宣言內容的統一檢索平台。")
+    
+    with gr.Tabs() as tabs:
+        with gr.Tab("📊 統計儀表板") as dash_tab:
+            gr.Markdown("### 📈 案件統計與搜尋熱度分析")
+            d_refresh_btn = gr.Button("🔄 重新整理數據", variant="primary")
+            d_summary = gr.HTML()
+            with gr.Row():
+                with gr.Column():
+                    d_p_year = gr.Plot(label="歷年分佈")
+                    d_ye_df = gr.Dataframe(label="📅 年度排行榜 (點擊可下鑽查詢)", interactive=False, elem_classes=["drilldown-df"])
+                with gr.Column():
+                    d_p_econ = gr.Plot(label="經濟體分佈")
+                    d_ec_df = gr.Dataframe(label="🌍 經濟體排行榜 (點擊可下鑽查詢)", interactive=False, elem_classes=["drilldown-df"])
+            with gr.Row():
+                with gr.Column():
+                    d_queries = gr.Plot(label="熱門關鍵字")
+                    d_qu_df = gr.Dataframe(label="🔍 熱門關鍵字 (點擊可下鑽查詢)", interactive=False, elem_classes=["drilldown-df"])
+                
+            dash_tab.select(get_dashboard_stats, outputs=[d_summary, d_p_year, d_p_econ, d_queries, d_ye_df, d_ec_df, d_qu_df])
+            d_refresh_btn.click(get_dashboard_stats, outputs=[d_summary, d_p_year, d_p_econ, d_queries, d_ye_df, d_ec_df, d_qu_df])
+
+        with gr.Tab("🍎 FSMM 宣言查詢", id="fsmm_tab") as fsmm_tab:
+            gr.Markdown("### 步驟 1：設定篩選條件 (可選擇年份、類型或特定檔案)")
+            with gr.Row():
+                f_query = gr.Textbox(label="🔍 關鍵字搜尋", placeholder="例如：resilience, climate...")
+                f_year = gr.Dropdown(label="📅 年份", choices=[""], allow_custom_value=True)
+                f_type = gr.Dropdown(label="📝 類型", choices=[""], allow_custom_value=True)
+                f_doc = gr.Dropdown(label="📄 特定檔案 (必選以查看全文)", choices=[""], allow_custom_value=True)
+            
+            with gr.Row():
+                f_btn = gr.Button("🔎 查詢相關段落", variant="primary")
+                f_full_btn = gr.Button("📖 查看所選檔案全文")
+                f_refresh_btn = gr.Button("🔄 重新整理選單")
+                
+            gr.Markdown("---")
+            f_out = gr.HTML(label="輸出區域")
+            
+            # Events
+            f_btn.click(search_fsmm, inputs=[f_query, f_year, f_type, f_doc], outputs=[f_out])
+            f_full_btn.click(get_fsmm_full_text, inputs=[f_doc], outputs=[f_out])
+            f_refresh_btn.click(refresh_fsmm_choices, outputs=[f_year, f_type, f_doc])
+            fsmm_tab.select(refresh_fsmm_choices, outputs=[f_year, f_type, f_doc])
+            
+        with gr.Tab("📊 APEC 提案查詢", id="proposal_tab") as proposal_tab:
+            with gr.Row():
+                p_query = gr.Textbox(label="關鍵字搜尋所有欄位", placeholder="例如：Thailand, agriculture, energy...")
+                p_year = gr.Dropdown(label="📅 年份", choices=[""], allow_custom_value=True)
+                p_econ = gr.Dropdown(label="Proposing Economy", choices=[""], allow_custom_value=True)
+            
+            with gr.Row():
+                p_btn = gr.Button("搜尋提案", variant="primary")
+                p_refresh_btn = gr.Button("🔄 重新整理選單")
+                
+            p_out = gr.HTML()
+            
+            # Events
+            p_btn.click(search_proposals, inputs=[p_query, p_year, p_econ], outputs=[p_out])
+            p_refresh_btn.click(refresh_p_choices, outputs=[p_year, p_econ])
+            p_year.change(lambda y: refresh_p_choices(y)[1], inputs=[p_year], outputs=[p_econ])
+            proposal_tab.select(refresh_p_choices, outputs=[p_year, p_econ])
+            
+        with gr.Tab("☁️ 上傳更新"):
+            gr.Markdown("### 上傳文件自動更新資料庫")
+            gr.Markdown("- **提案 (Proposals)**: 上傳 `Proposal_*.html` 檔案。")
+            gr.Markdown("- **宣言 (FSMM)**: 上傳 `YY_fsmm_*.pdf` 或 `.docx` 檔案。")
+            u_file = gr.File(label="拖曳或選擇多個檔案", file_count="multiple")
+            u_btn = gr.Button("上傳解析")
+            u_status = gr.Textbox(label="處理結果")
+            u_btn.click(handle_unified_upload, inputs=[u_file], outputs=[u_status])
+            
+            # Drill-down Events: outputs=[tabs, f_out, p_out]
+            d_ye_df.select(handle_drilldown, outputs=[tabs, f_out, p_out])
+            d_ec_df.select(handle_drilldown, outputs=[tabs, f_out, p_out])
+            d_qu_df.select(handle_drilldown, outputs=[tabs, f_out, p_out])
+
+    # Initial load of choices for all tabs and dashboard
+    def init_all():
+        fc = refresh_fsmm_choices()
+        pc = refresh_p_choices()
+        db = get_dashboard_stats()
+        return fc + pc + db
+    
+    demo.load(init_all, outputs=[f_year, f_type, f_doc, p_year, p_econ, d_summary, d_p_year, d_p_econ, d_queries, d_ye_df, d_ec_df, d_qu_df])
+
+if __name__ == "__main__":
+    demo.launch(css=DASHBOARD_CSS)

requirements.txt

@@ -0,0 +1,7 @@
+gradio
+pdfminer.six
+python-docx
+pandas
+beautifulsoup4
+plotly
+