Update app.py

app.py

@@ -94,11 +94,8 @@ purple_theme = PurpleTheme()
 # --- ⚙️ Backend Logic & Functions ---
 # ==========================================
 def get_debug_info():
-    """Read server file system status"""
     now = datetime.now().strftime("%H:%M:%S")
     files = os.listdir(PAPERS_DIR) if os.path.exists(PAPERS_DIR) else "Directory missing"
-    
-    # Recursively check output directory contents
     output_detail = "Not generated"
     if os.path.exists(OUTPUT_DIR):
         all_output_items = []
@@ -106,7 +103,6 @@ def get_debug_info():
             for name in files_in_out:
                 all_output_items.append(os.path.join(os.path.relpath(root, OUTPUT_DIR), name))
         output_detail = f"Found {len(all_output_items)} files: {all_output_items[:5]}..." if all_output_items else "Directory exists but is EMPTY"
-    
     return f"[{now}] 📁 papers/ Content: {files}\n\n[{now}] 📂 mineru_outputs Status: {output_detail}"
 
 def save_pdf(file):
@@ -125,35 +121,25 @@ def save_api_settings(api_key, api_base_url=None):
         if os.path.exists(CONFIG_PATH):
             with open(CONFIG_PATH, "r", encoding="utf-8") as f:
                 config = yaml.safe_load(f) or {}
-                
-        # Save API Key
         config.setdefault("api_keys", {})["gemini_api_key"] = api_key
-        
-        # Save base URL if not empty
         if api_base_url:
             config["api_base_url"] = api_base_url
-            
-        # Write to YAML
         with open(CONFIG_PATH, "w", encoding="utf-8") as f:
             yaml.dump(config, f, allow_unicode=True)
-            
         success_msg = "✅ Key saved"
         if api_base_url:
             success_msg += ", Base URL updated"
-            
         return success_msg, get_debug_info()
     except Exception as e:
         return f"❌ Error: {str(e)}", get_debug_info()
 
 def run_mineru_parsing_and_dag_gen():
-    """Execute PDF parsing and DAG generation (supports real-time streaming)"""
     if not os.path.exists(PAPERS_DIR) or not any(f.endswith('.pdf') for f in os.listdir(PAPERS_DIR)):
         yield "❌ No PDF file found", get_debug_info(), "No execution logs."
         return
 
     full_log = ""
     try:
-        # ================= Step 1: Mineru Parsing =================
         env = os.environ.copy()
         env["MINERU_FORMULA_ENABLE"] = "false"
         env["MINERU_TABLE_ENABLE"] = "false"
@@ -161,19 +147,12 @@ def run_mineru_parsing_and_dag_gen():
         env["MINERU_VIRTUAL_VRAM_SIZE"] = "8"
 
         command_mineru = ["mineru", "-p", PAPERS_DIR, "-o", OUTPUT_DIR]
-        
         full_log += "--- Mineru Executing ---\n"
         yield "⏳ Executing Mineru parsing...", get_debug_info(), full_log
 
         process_mineru = subprocess.Popen(
-            command_mineru, 
-            env=env, 
-            stdout=subprocess.PIPE, 
-            stderr=subprocess.STDOUT, 
-            text=True,
-            bufsize=1 
+            command_mineru, env=env, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, bufsize=1
         )
-
         for line in iter(process_mineru.stdout.readline, ''):
             full_log += line
             yield "⏳ Executing Mineru parsing...", get_debug_info(), full_log
@@ -181,35 +160,23 @@ def run_mineru_parsing_and_dag_gen():
         returncode_mineru = process_mineru.wait()
 
         if returncode_mineru != 0:
-            status = f"❌ Mineru parsing failed (Exit Code: {returncode_mineru})"
-            yield status, get_debug_info(), full_log
+            yield f"❌ Mineru parsing failed (Exit Code: {returncode_mineru})", get_debug_info(), full_log
             return
 
-        # ================= Step 2: DAG Generation =================
         command_dag = [sys.executable, "gen_dag.py"]
-        
         full_log += "\n--- DAG Gen Executing ---\n"
         yield "⏳ Mineru parsing complete, executing DAG generation...", get_debug_info(), full_log
         
         process_dag = subprocess.Popen(
-            command_dag, 
-            stdout=subprocess.PIPE, 
-            stderr=subprocess.STDOUT, 
-            text=True,
-            bufsize=1
+            command_dag, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, bufsize=1
         )
-
         for line in iter(process_dag.stdout.readline, ''):
             full_log += line
             yield "⏳ Executing DAG generation...", get_debug_info(), full_log
         process_dag.stdout.close()
         returncode_dag = process_dag.wait()
         
-        if returncode_dag == 0:
-            status = "✅ PDF parsing & DAG generation fully completed"
-        else:
-            status = f"❌ DAG generation failed (Exit Code: {returncode_dag})"
-            
+        status = "✅ PDF parsing & DAG generation fully completed" if returncode_dag == 0 else f"❌ DAG generation failed (Exit Code: {returncode_dag})"
         yield status, get_debug_info(), full_log
 
     except Exception as e:
@@ -217,46 +184,35 @@ def run_mineru_parsing_and_dag_gen():
         yield "❌ Execution Exception", get_debug_info(), error_log
 
 def run_final_generation(task_type="all"):
-    """Execute generation scripts and zip results"""
     if not os.path.exists(OUTPUT_DIR):
         return "❌ Please run the parsing step first", get_debug_info(), "No output folder found.", None
 
     scripts_to_run = []
-    if task_type == "ppt":
-        scripts_to_run = ["gen_ppt.py"]
-    elif task_type == "poster":
-        scripts_to_run = ["gen_poster.py"]
-    elif task_type == "pr":
-        scripts_to_run = ["gen_pr.py"]
-    elif task_type == "all":
-        scripts_to_run = ["gen_ppt.py", "gen_poster.py", "gen_pr.py"]
-    else:
-        return "❌ Unknown task type", get_debug_info(), "Invalid task_type.", None
+    if task_type == "ppt": scripts_to_run = ["gen_ppt.py"]
+    elif task_type == "poster": scripts_to_run = ["gen_poster.py"]
+    elif task_type == "pr": scripts_to_run = ["gen_pr.py"]
+    elif task_type == "all": scripts_to_run = ["gen_ppt.py", "gen_poster.py", "gen_pr.py"]
+    else: return "❌ Unknown task type", get_debug_info(), "Invalid task_type.", None
 
     full_log = f"🚀 Preparing to start {len(scripts_to_run)} tasks...\n"
     success = True
 
     def execute_script(script):
         command = [sys.executable, script]
-        result = subprocess.run(command, capture_output=True, text=True, timeout=600)
-        return script, result
+        return script, subprocess.run(command, capture_output=True, text=True, timeout=600)
 
     try:
         with ThreadPoolExecutor(max_workers=len(scripts_to_run)) as executor:
             future_to_script = {executor.submit(execute_script, s): s for s in scripts_to_run}
-            
             for future in as_completed(future_to_script):
                 script_name = future_to_script[future]
                 try:
                     _, result = future.result()
-                    
                     full_log += f"\n================ ✅ Execution Complete: {script_name} ================\n"
                     full_log += f"--- STDOUT ---\n{result.stdout}\n\n--- STDERR ---\n{result.stderr}\n"
-                    
                     if result.returncode != 0:
                         success = False
                         full_log += f"❌ [Error] {script_name} returned non-zero exit code (Exit Code: {result.returncode})\n"
-                
                 except subprocess.TimeoutExpired as e:
                     success = False
                     full_log += f"\n================ ❌ Task Timeout: {script_name} ================\n{str(e)}\n"
@@ -264,16 +220,11 @@ def run_final_generation(task_type="all"):
                     success = False
                     full_log += f"\n================ ❌ Task Exception: {script_name} ================\n{str(e)}\n"
 
-        if not success:
-            return f"❌ {task_type.upper()} contains failed tasks, please check logs", get_debug_info(), full_log, None
+        if not success: return f"❌ {task_type.upper()} contains failed tasks, please check logs", get_debug_info(), full_log, None
 
-        # Zip the mineru_outputs folder
         zip_base_name = ZIP_OUTPUT_PATH.replace(".zip", "")
         shutil.make_archive(zip_base_name, 'zip', OUTPUT_DIR)
-        
-        success_msg = f"✅ {task_type.upper()} generated and zipped successfully"
-        return success_msg, get_debug_info(), full_log, ZIP_OUTPUT_PATH
-
+        return f"✅ {task_type.upper()} generated and zipped successfully", get_debug_info(), full_log, ZIP_OUTPUT_PATH
     except Exception as e:
         error_log = full_log + f"\n[Global Exception] Exception occurred:\n{str(e)}"
         return "❌ Global exception during final generation", get_debug_info(), error_log, None
@@ -294,101 +245,118 @@ body, .gradio-container {
 .dark body, .dark .gradio-container {
     background-color: #1a1a1a !important;
     background-image: linear-gradient(rgba(168, 85, 247, .1) 1px, transparent 1px), linear-gradient(90deg, rgba(168, 85, 247, .1) 1px, transparent 1px) !important;
-    background-size: 40px 40px !important;
 }
 
-#col-container {
-    margin: 0 auto;
-    max-width: 1200px;
-}
-
-#main-title {
-    text-align: center !important;
-    padding: 1.5rem 0 0.5rem 0;
-}
+#col-container { margin: 0 auto; max-width: 1200px; }
 
+#main-title { text-align: center !important; padding: 1.5rem 0 0.5rem 0; }
 #main-title h1 {
-    font-size: 2.6em !important;
-    font-weight: 800 !important;
+    font-size: 2.6em !important; font-weight: 800 !important;
     background: linear-gradient(135deg, #A855F7 0%, #C084FC 50%, #9333EA 100%);
     background-size: 200% 200%;
-    -webkit-background-clip: text;
-    -webkit-text-fill-color: transparent;
-    background-clip: text;
-    animation: gradient-shift 4s ease infinite;
-    letter-spacing: -0.02em;
-}
-
-@keyframes gradient-shift {
-    0%, 100% { background-position: 0% 50%; }
-    50% { background-position: 100% 50%; }
+    -webkit-background-clip: text; -webkit-text-fill-color: transparent;
+    animation: gradient-shift 4s ease infinite; letter-spacing: -0.02em;
 }
 
-#subtitle {
-    text-align: center !important;
-    margin-bottom: 2rem;
-}
-
-#subtitle p {
-    margin: 0 auto;
-    color: #666;
-    font-size: 1.1rem;
-    font-weight: 500;
-}
+@keyframes gradient-shift { 0%, 100% { background-position: 0% 50%; } 50% { background-position: 100% 50%; } }
 
+#subtitle { text-align: center !important; margin-bottom: 2rem; }
+#subtitle p { margin: 0 auto; color: #666; font-size: 1.1rem; font-weight: 500; }
 .dark #subtitle p { color: #DAB2FF; }
 
 /* Glassmorphism Cards */
 .gradio-group {
     background: rgba(255, 255, 255, 0.9) !important;
-    border: 2px solid #E9D5FF !important;
-    border-radius: 12px !important;
+    border: 2px solid #E9D5FF !important; border-radius: 12px !important;
     box-shadow: 0 4px 24px rgba(168, 85, 247, 0.08) !important;
-    backdrop-filter: blur(10px);
-    transition: all 0.3s ease;
-    padding: 15px !important;
+    backdrop-filter: blur(10px); transition: all 0.3s ease; padding: 15px !important;
 }
+.gradio-group:hover { box-shadow: 0 8px 32px rgba(168, 85, 247, 0.12) !important; border-color: #C084FC !important; }
+.dark .gradio-group { background: rgba(30, 30, 30, 0.9) !important; border-color: rgba(168, 85, 247, 0.3) !important; }
 
-.gradio-group:hover {
-    box-shadow: 0 8px 32px rgba(168, 85, 247, 0.12) !important;
-    border-color: #C084FC !important;
+/* ======== 新增：1. 隐藏原生组件，自定义全局拖拽与点击上传区域 ======== */
+#pdf-upload-box {
+    background: transparent !important;
+    border: 2px dashed rgba(192, 132, 252, 0.5) !important;
+    border-radius: 12px !important;
+    min-height: 100px !important;
+    position: relative;
+    overflow: hidden;
+    display: flex !important; align-items: center !important; justify-content: center !important;
+    transition: all 0.3s ease !important;
 }
+#pdf-upload-box:hover { border-color: #A855F7 !important; background: rgba(168, 85, 247, 0.05) !important; }
 
-.dark .gradio-group {
-    background: rgba(30, 30, 30, 0.9) !important;
-    border-color: rgba(168, 85, 247, 0.3) !important;
-}
+/* 将原生的输入框变为全透明但保留点击/拖拽属性 */
+#pdf-upload-box > * { opacity: 0 !important; z-index: 10 !important; position: relative; width: 100%; height: 100%; cursor: pointer;}
 
-/* Terminal Log Style */
-.log-box textarea {
-    font-family: 'IBM Plex Mono', Consolas, monospace !important; 
-    font-size: 13px !important; 
-    background-color: #1e1e1e !important; 
-    color: #DAB2FF !important; 
-    border: 1px solid #C084FC !important;
-    border-radius: 8px !important;
+/* 自定义文字显示 */
+#pdf-upload-box::before {
+    content: "🖱️ 点击此处选择 PDF \A 🌍 或将文件拖拽至全局任意位置";
+    white-space: pre-wrap; text-align: center; color: #7E22CE;
+    font-weight: 600; font-size: 1.1rem; position: absolute;
+    z-index: 1; pointer-events: none; line-height: 1.6;
 }
 
-/* Borderless Status Text */
-.status-text textarea {
-    background-color: transparent !important;
-    border: none !important;
-    box-shadow: none !important;
-    font-weight: 600 !important;
-    color: #6B21A8 !important;
+/* 激活全局拖拽时的魔术全屏放大样式 */
+body.dragging-file #pdf-upload-box {
+    position: fixed !important; top: 0 !important; left: 0 !important;
+    width: 100vw !important; height: 100vh !important;
+    z-index: 99999 !important; background: rgba(250, 245, 255, 0.85) !important;
+    backdrop-filter: blur(8px) !important; border: 6px dashed #A855F7 !important; border-radius: 0 !important;
+}
+body.dragging-file #pdf-upload-box::before {
+    content: "📥 松开鼠标，立即上传 PDF"; font-size: 2.5rem; color: #9333EA; font-weight: 800;
 }
 
-.dark .status-text textarea {
-    color: #C084FC !important;
+/* ======== 新增：2. 等长圆角按钮及悬浮动效 ======== */
+.action-row { display: flex !important; gap: 15px !important; margin-bottom: 10px !important; }
+.action-btn {
+    border-radius: 30px !important;
+    background: linear-gradient(135deg, #A855F7, #9333EA) !important;
+    color: white !important; font-weight: 600 !important; border: none !important;
+    height: 55px !important; flex: 1 !important;
+    transition: transform 0.3s cubic-bezier(0.175, 0.885, 0.32, 1.275), box-shadow 0.3s ease !important;
+    box-shadow: 0 4px 15px rgba(147, 51, 234, 0.2) !important;
+    cursor: pointer !important; display: flex !important; align-items: center !important; justify-content: center !important;
 }
+.action-btn:hover {
+    transform: translateY(-5px) scale(1.03) !important;
+    box-shadow: 0 10px 25px rgba(147, 51, 234, 0.4) !important;
+    background: linear-gradient(135deg, #C084FC, #A855F7) !important;
+}
+.action-btn:active { transform: translateY(2px) scale(0.98) !important; box-shadow: 0 2px 10px rgba(147, 51, 234, 0.2) !important; }
 
-/* Custom Scrollbars */
+/* Terminal Log Style */
+.log-box textarea { font-family: 'IBM Plex Mono', monospace !important; font-size: 13px !important; background-color: #1e1e1e !important; color: #DAB2FF !important; border: 1px solid #C084FC !important; border-radius: 8px !important; }
+.status-text textarea { background-color: transparent !important; border: none !important; box-shadow: none !important; font-weight: 600 !important; color: #6B21A8 !important; }
+.dark .status-text textarea { color: #C084FC !important; }
 ::-webkit-scrollbar { width: 8px; height: 8px; }
 ::-webkit-scrollbar-track { background: rgba(168, 85, 247, 0.05); border-radius: 4px; }
 ::-webkit-scrollbar-thumb { background: linear-gradient(135deg, #A855F7, #C084FC); border-radius: 4px; }
 ::-webkit-scrollbar-thumb:hover { background: linear-gradient(135deg, #9333EA, #A855F7); }
 """
 
+# ======== 新增：全局拖拽监听 JavaScript ========
+global_drag_js = """
+function() {
+    let dragCounter = 0;
+    window.addEventListener('dragenter', (e) => {
+        e.preventDefault(); dragCounter++; document.body.classList.add('dragging-file');
+    });
+    window.addEventListener('dragleave', (e) => {
+        e.preventDefault(); dragCounter--;
+        if(dragCounter === 0) document.body.classList.remove('dragging-file');
+    });
+    window.addEventListener('drop', (e) => {
+        dragCounter = 0;
+        // 延迟移除class以确保Gradio能捕捉到底层组件的drop事件
+        setTimeout(() => document.body.classList.remove('dragging-file'), 150);
+    });
+    window.addEventListener('dragover', (e) => { e.preventDefault(); });
+}
+"""
+
 with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
     with gr.Column(elem_id="col-container"):
         gr.Markdown("# **PaperX / Mineru Parsing Platform**", elem_id="main-title")
@@ -408,16 +376,13 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
                 # 2. Document Parsing
                 with gr.Group(elem_classes="gradio-group"):
                     gr.Markdown("### 📄 2. Document Parsing")
-                    pdf_input = gr.File(label="Drag and drop or click to upload PDF", file_types=[".pdf"])
                     
-                    parse_btn = gr.Button("🚀 Start Mineru & DAG Extraction", variant="primary", size="lg")
+                    # 调整了这里的控件，绑定CSS做UI隐藏和特效
+                    pdf_input = gr.File(label="", file_types=[".pdf"], elem_id="pdf-upload-box")
                     
+                    parse_btn = gr.Button("🚀 Start Mineru & DAG Extraction", variant="primary", size="lg")
                     parse_status = gr.Textbox(
-                        show_label=False, 
-                        placeholder="Waiting for document upload...", 
-                        lines=1, 
-                        interactive=False, 
-                        elem_classes="status-text"
+                        show_label=False, placeholder="Waiting for document upload...", lines=1, interactive=False, elem_classes="status-text"
                     )
 
                 # 3. Asset Generation
@@ -425,10 +390,11 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
                     gr.Markdown("### 🎯 3. Asset Generation")
                     gr.Markdown("Generate final formats based on DAG structure:")
                     
-                    with gr.Row():
-                        gen_ppt_btn = gr.Button("📊 Generate PPT")
-                        gen_poster_btn = gr.Button("🖼️ Generate Poster")
-                        gen_pr_btn = gr.Button("📰 Generate PR Article")
+                    # 使用自定义的 .action-row 和 .action-btn 实现等宽高丝滑悬浮按钮
+                    with gr.Row(elem_classes="action-row"):
+                        gen_ppt_btn = gr.Button("📊 Gen PPT", elem_classes="action-btn")
+                        gen_poster_btn = gr.Button("🖼️ Gen Poster", elem_classes="action-btn")
+                        gen_pr_btn = gr.Button("📰 Gen Article", elem_classes="action-btn")
                     
                     gen_all_btn = gr.Button("✨ Generate All Assets (ALL)", variant="primary")
 
@@ -439,11 +405,7 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
                 with gr.Group(elem_classes="gradio-group"):
                     gr.Markdown("### 📦 Generation Results & Download")
                     gen_status = gr.Textbox(
-                        show_label=False, 
-                        placeholder="No generation task currently...", 
-                        lines=2, 
-                        interactive=False,
-                        elem_classes="status-text"
+                        show_label=False, placeholder="No generation task currently...", lines=2, interactive=False, elem_classes="status-text"
                     )
                     download_file = gr.File(label="📥 Get Final Zip Archive", interactive=False, visible=False)
 
@@ -453,21 +415,13 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
                     with gr.Tabs():
                         with gr.Tab("📜 Terminal Stream"):
                             cmd_logs = gr.Textbox(
-                                label="Stdout / Stderr", 
-                                placeholder="Waiting for task to start...", 
-                                lines=14, 
-                                interactive=False,
-                                elem_classes="log-box" 
+                                label="Stdout / Stderr", placeholder="Waiting for task to start...", lines=14, interactive=False, elem_classes="log-box" 
                             )
                         
                         with gr.Tab("🔍 System Snapshot"):
                             refresh_btn = gr.Button("🔄 Refresh Directory Tree") 
                             debug_view = gr.Textbox(
-                                label="Workspace Files", 
-                                lines=13, 
-                                interactive=False, 
-                                value=get_debug_info(),
-                                elem_classes="log-box"
+                                label="Workspace Files", lines=13, interactive=False, value=get_debug_info(), elem_classes="log-box"
                             )
 
     # ================= LOGIC BINDINGS =================
@@ -476,10 +430,7 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
     pdf_input.upload(fn=save_pdf, inputs=pdf_input, outputs=[parse_status, debug_view])
     pdf_input.clear(fn=lambda: ("ℹ️ File cleared", get_debug_info()), outputs=[parse_status, debug_view])
     
-    parse_btn.click(
-        fn=run_mineru_parsing_and_dag_gen, 
-        outputs=[parse_status, debug_view, cmd_logs]
-    )
+    parse_btn.click(fn=run_mineru_parsing_and_dag_gen, outputs=[parse_status, debug_view, cmd_logs])
 
     def trigger_gen(task):
         status, debug, logs, file_path = run_final_generation(task)
@@ -493,5 +444,8 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
     
     refresh_btn.click(fn=get_debug_info, outputs=debug_view)
 
+    # 在页面加载时注入全局拖拽脚本
+    demo.load(js=global_drag_js)
+
 if __name__ == "__main__":
     demo.launch()