Update app.py

app.py

@@ -11,6 +11,7 @@ BASE_DIR = os.path.dirname(os.path.abspath(__file__))
 PAPERS_DIR = os.path.join(BASE_DIR, "papers")
 CONFIG_PATH = os.path.join(BASE_DIR, "config.yaml")
 OUTPUT_DIR = os.path.join(BASE_DIR, "mineru_outputs")
+ZIP_OUTPUT_PATH = os.path.join(BASE_DIR, "mineru_results.zip") # 压缩包路径
 
 os.makedirs(PAPERS_DIR, exist_ok=True)
 
@@ -19,7 +20,7 @@ def get_debug_info():
     now = datetime.now().strftime("%H:%M:%S")
     files = os.listdir(PAPERS_DIR) if os.path.exists(PAPERS_DIR) else "Directory missing"
     
-    # 递归检查输出目录下的内容，看看到底生成了什么
+    # 递归检查输出目录下的内容
     output_detail = "Not generated"
     if os.path.exists(OUTPUT_DIR):
         all_output_items = []
@@ -65,19 +66,16 @@ def run_mineru_parsing():
         env["MINERU_DEVICE_MODE"] = "cpu"
         env["MINERU_VIRTUAL_VRAM_SIZE"] = "8"
 
-        # 尝试使用完整的 mineru 命令
         command = ["mineru", "-p", PAPERS_DIR, "-o", OUTPUT_DIR]
         
-        # 使用 subprocess.run 捕获所有输出
         result = subprocess.run(
             command, 
             env=env, 
             capture_output=True, 
             text=True,
-            timeout=300 # 设置5分钟超时防止卡死
+            timeout=300 
         )
 
-        # 组合 stdout 和 stderr 作为完整的运行日志
         full_log = f"--- STDOUT ---\n{result.stdout}\n\n--- STDERR ---\n{result.stderr}"
         
         if result.returncode == 0:
@@ -91,6 +89,37 @@ def run_mineru_parsing():
         error_log = f"Exception occurred during execution:\n{str(e)}"
         return "❌ 运行异常", get_debug_info(), error_log
 
+def run_final_generation():
+    """执行 python main.py 并压缩结果"""
+    if not os.path.exists(OUTPUT_DIR):
+        return "❌ 请先执行第二步解析", get_debug_info(), "No output folder found.", None
+
+    try:
+        # 1. 运行 python main.py
+        command = [sys.executable, "main.py"]
+        result = subprocess.run(
+            command,
+            capture_output=True,
+            text=True,
+            timeout=600 # 适当增加超时时间
+        )
+
+        full_log = f"--- STDOUT ---\n{result.stdout}\n\n--- STDERR ---\n{result.stderr}"
+
+        if result.returncode != 0:
+            return f"❌ 生成失败 (Exit Code: {result.returncode})", get_debug_info(), full_log, None
+
+        # 2. 压缩 mineru_outputs 文件夹
+        # shutil.make_archive 会自动加上 .zip 后缀，所以 base_name 不带后缀
+        zip_base_name = ZIP_OUTPUT_PATH.replace(".zip", "")
+        shutil.make_archive(zip_base_name, 'zip', OUTPUT_DIR)
+
+        return "✅ 最终生成并压缩完成", get_debug_info(), full_log, ZIP_OUTPUT_PATH
+
+    except Exception as e:
+        error_log = f"Exception occurred during final generation:\n{str(e)}"
+        return "❌ 最终生成异常", get_debug_info(), error_log, None
+
 # --- UI ---
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("# 📑 Mineru PDF 解析调试器")
@@ -110,15 +139,21 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 parse_btn = gr.Button("🚀 Run Mineru (CPU Mode)", variant="primary")
                 parse_status = gr.Textbox(label="运行状态")
 
+            with gr.Group():
+                gr.Markdown("### 3. 最终生成")
+                gen_btn = gr.Button("🔨 执行 main.py 并打包", variant="primary")
+                gen_status = gr.Textbox(label="生成状态")
+                download_file = gr.File(label="下载压缩后的结果", interactive=False)
+
         with gr.Column(scale=1):
             gr.Markdown("### 🔍 实时系统监控")
             debug_view = gr.Textbox(label="文件系统快照", value=get_debug_info(), lines=8, interactive=False)
             
-            gr.Markdown("### 📜 Mineru 终端输出日志")
+            gr.Markdown("### 📜 终端输出日志")
             cmd_logs = gr.Textbox(
                 label="Command Output (Stdout/Stderr)", 
-                placeholder="等待解析任务开始...", 
-                lines=15, 
+                placeholder="等待任务开始...", 
+                lines=20, 
                 interactive=False
             )
             refresh_btn = gr.Button("🔄 刷新状态")
@@ -127,11 +162,16 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     key_btn.click(save_api_key, inputs=key_input, outputs=[parse_status, debug_view])
     pdf_btn.click(save_pdf, inputs=pdf_input, outputs=[parse_status, debug_view])
     
-    # 解析按钮会更新三个地方：状态、文件监控、详细日志
     parse_btn.click(
         fn=run_mineru_parsing, 
         outputs=[parse_status, debug_view, cmd_logs]
     )
+
+    # 最终生成逻辑绑定
+    gen_btn.click(
+        fn=run_final_generation,
+        outputs=[gen_status, debug_view, cmd_logs, download_file]
+    )
     
     refresh_btn.click(get_debug_info, outputs=debug_view)