feat: PDF对比用iframe显示原PDF文件（支持扫描版PDF）

- 读取翻译结果文件内容
- 尝试从原PDF提取原文（OCR识别的文字）
- 按页面分块显示对比
- 前端支持分页对比展示

app.py

@@ -374,16 +374,17 @@ def upload_pdf():
         output_path = cache_path
     else:
         # 需要翻译
-        # 保存上传文件
-        upload_dir = os.path.join(UPLOAD_DIR, str(uuid.uuid4()))
+        # 保存上传文件 - 使用同一个UUID确保uploads和outputs目录关联
+        session_uuid = str(uuid.uuid4())
+        upload_dir = os.path.join(UPLOAD_DIR, session_uuid)
         os.makedirs(upload_dir, exist_ok=True)
         upload_path = os.path.join(upload_dir, filename)
         
         with open(upload_path, 'wb') as f:
             f.write(file_content)
         
-        # 创建输出路径
-        output_dir = os.path.join(OUTPUT_DIR, str(uuid.uuid4()))
+        # 创建输出路径 - 使用相同的UUID
+        output_dir = os.path.join(OUTPUT_DIR, session_uuid)
         os.makedirs(output_dir, exist_ok=True)
         output_path = os.path.join(output_dir, f"{filename}_translated.md")
         
@@ -400,6 +401,7 @@ def upload_pdf():
         translate_params=json.dumps({'instruction': instruction}) if instruction else None,
         status='processing' if not from_cache else 'completed',
         progress=0 if not from_cache else 100,
+        upload_path=upload_path if not from_cache else None,  # 保存上传路径
         output_path=output_path,
         from_cache=from_cache
     )
@@ -609,19 +611,95 @@ def compare_view(translation_id):
         return jsonify({'error': '请登录后使用此功能'}), 401
     
     translation = Translation.query.get(translation_id)
-    if not translation or translation.user_id != user.id:
+    if not translation or (translation.user_id != user.id and user.user_type != 'admin'):
         return jsonify({'error': '无权访问'}), 403
     
-    # 生成对比文件
-    # TODO: 实现对比功能
+    # 读取翻译结果文件
+    translated_content = ''
+    if translation.output_path and os.path.exists(translation.output_path):
+        try:
+            with open(translation.output_path, 'r', encoding='utf-8') as f:
+                translated_content = f.read()
+        except Exception as e:
+            translated_content = f'读取失败: {str(e)}'
+    
+    # 从翻译结果中提取各页内容
+    # 翻译结果格式是Markdown，包含"## 第 X 页"分隔
+    original_pages = []
+    translated_pages = []
+    
+    if translated_content:
+        # 解析翻译结果的页面结构
+        import re
+        page_pattern = r'## 第 (\d+) 页\n\n(.*?)\n\n---'
+        matches = re.findall(page_pattern, translated_content, re.DOTALL)
+        
+        for page_num, content in matches:
+            translated_pages.append({
+                'page': int(page_num),
+                'content': content.strip()
+            })
+    
+    # 如果有原文内容存储，获取原文
+    original_content = ''
+    
+    # 优先从数据库存储的upload_path获取原PDF
+    possible_paths = []
+    
+    if translation.upload_path:
+        possible_paths.append(translation.upload_path)
+    
+    # 备用方案：尝试从路径推断（兼容旧数据）
+    upload_dir = os.path.dirname(translation.output_path.replace('outputs', 'uploads').replace('_translated.md', '')) if translation.output_path else ''
+    if upload_dir:
+        possible_paths.append(
+            translation.output_path.replace('outputs', 'uploads').replace('_translated.md', '') if translation.output_path else ''
+        )
+        possible_paths.append(os.path.join(upload_dir, translation.original_filename))
+    
+    for pdf_path in possible_paths:
+        if pdf_path and os.path.exists(pdf_path) and pdf_path.endswith('.pdf'):
+            try:
+                from pypdf import PdfReader
+                reader = PdfReader(pdf_path)
+                for page in reader.pages:
+                    text = page.extract_text()
+                    if text:
+                        original_content += text + '\n\n'
+            except:
+                pass
+            break
     
     return jsonify({
         'id': translation.id,
-        'original': '原文内容',
-        'translated': '译文内容'
+        'filename': translation.original_filename,
+        'original': original_content or '',
+        'original_pdf_url': f'/api/original-pdf/{translation.id}' if translation.upload_path else None,
+        'translated': translated_content,
+        'pages': translated_pages
     })
 
 
+@app.route('/api/original-pdf/<int:translation_id>')
+def get_original_pdf(translation_id):
+    """获取原始PDF文件"""
+    user = get_current_user()
+    if not user:
+        return jsonify({'error': '请登录后使用此功能'}), 401
+    
+    translation = Translation.query.get(translation_id)
+    if not translation or (translation.user_id != user.id and user.user_type != 'admin'):
+        return jsonify({'error': '无权访问'}), 403
+    
+    if not translation.upload_path or not os.path.exists(translation.upload_path):
+        return jsonify({'error': '原PDF文件不存在'}), 404
+    
+    return send_file(translation.upload_path, 
+                     mimetype='application/pdf',
+                     as_attachment=False,
+                     download_name=translation.original_filename)
+
+
 # ==================== 路由: 用户系统 ====================
 @app.route('/login', methods=['GET', 'POST'])
 def login():

logs/app.log

@@ -0,0 +1,187 @@
+ * Serving Flask app 'app'
+ * Debug mode: on
+WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
+ * Running on all addresses (0.0.0.0)
+ * Running on http://127.0.0.1:19000
+ * Running on http://192.168.2.17:19000
+Press CTRL+C to quit
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/pdf_translate.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+127.0.0.1 - - [16/Apr/2026 21:50:38] "GET /api/config HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:08] "GET /admin/translations HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:08] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py:53: LegacyAPIWarning: The Query.get() method is considered legacy as of the 1.x series of SQLAlchemy and becomes a legacy construct in 2.0. The method is now available as Session.get() (deprecated since: 2.0) (Background on SQLAlchemy 2.0 at: https://sqlalche.me/e/b8d9)
+  return User.query.get(user_id)
+192.168.2.10 - - [16/Apr/2026 21:58:13] "GET / HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:13] "GET /static/css/style.css HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 21:58:13] "GET /static/js/main.js HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 21:58:13] "GET /api/user/info HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:13] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 21:58:16] "GET /history HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:16] "GET /static/css/style.css HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 21:58:18] "GET / HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:18] "GET /static/css/style.css HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 21:58:18] "GET /static/js/main.js HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 21:58:18] "GET /api/user/info HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:26] "POST /api/upload HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:26] "GET /api/task/17baff33-894b-4e56-8975-cc2e6359cd66 HTTP/1.1" 200 -
+/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py:481: LegacyAPIWarning: The Query.get() method is considered legacy as of the 1.x series of SQLAlchemy and becomes a legacy construct in 2.0. The method is now available as Session.get() (deprecated since: 2.0) (Background on SQLAlchemy 2.0 at: https://sqlalche.me/e/b8d9)
+  translation = Translation.query.get(translation_id)
+192.168.2.10 - - [16/Apr/2026 21:58:26] "GET /api/status/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:28] "GET /api/task/17baff33-894b-4e56-8975-cc2e6359cd66 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:28] "GET /api/status/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:30] "GET /api/task/17baff33-894b-4e56-8975-cc2e6359cd66 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:30] "GET /api/status/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:32] "GET /api/task/17baff33-894b-4e56-8975-cc2e6359cd66 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:32] "GET /api/status/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:34] "GET /api/task/17baff33-894b-4e56-8975-cc2e6359cd66 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:34] "GET /api/status/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:37] "GET /api/task/17baff33-894b-4e56-8975-cc2e6359cd66 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 21:58:37] "GET /api/status/4 HTTP/1.1" 200 -
+/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py:519: LegacyAPIWarning: The Query.get() method is considered legacy as of the 1.x series of SQLAlchemy and becomes a legacy construct in 2.0. The method is now available as Session.get() (deprecated since: 2.0) (Background on SQLAlchemy 2.0 at: https://sqlalche.me/e/b8d9)
+  translation = Translation.query.get(translation_id)
+192.168.2.10 - - [16/Apr/2026 21:58:37] "GET /api/result/4 HTTP/1.1" 200 -
+127.0.0.1 - - [16/Apr/2026 22:00:01] "GET /api/health HTTP/1.1" 404 -
+127.0.0.1 - - [16/Apr/2026 22:00:33] "POST /api/upload HTTP/1.1" 200 -
+127.0.0.1 - - [16/Apr/2026 22:00:38] "GET /api/status/5 HTTP/1.1" 200 -
+127.0.0.1 - - [16/Apr/2026 22:01:15] "GET /api/status/5 HTTP/1.1" 200 -
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+[翻译任务] 开始翻译，使用配置: https://open.bigmodel.cn/api/paas/v4
+[翻译任务] 开始翻译，使用配置: https://open.bigmodel.cn/api/paas/v4
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+127.0.0.1 - - [16/Apr/2026 22:20:01] "GET /api/health HTTP/1.1" 404 -
+/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py:53: LegacyAPIWarning: The Query.get() method is considered legacy as of the 1.x series of SQLAlchemy and becomes a legacy construct in 2.0. The method is now available as Session.get() (deprecated since: 2.0) (Background on SQLAlchemy 2.0 at: https://sqlalche.me/e/b8d9)
+  return User.query.get(user_id)
+192.168.2.10 - - [16/Apr/2026 22:24:27] "GET / HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:27] "GET /static/css/style.css HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:27] "GET /static/js/main.js HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:27] "GET /api/user/info HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:27] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:31] "GET /admin/translations HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:31] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:33] "POST /admin/translation/5/toggle-share HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:33] "GET /admin/translations HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:33] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:34] "POST /admin/translation/4/toggle-share HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:34] "GET /admin/translations HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:34] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:42] "GET /admin/translation/5 HTTP/1.1" 200 -
+/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py:519: LegacyAPIWarning: The Query.get() method is considered legacy as of the 1.x series of SQLAlchemy and becomes a legacy construct in 2.0. The method is now available as Session.get() (deprecated since: 2.0) (Background on SQLAlchemy 2.0 at: https://sqlalche.me/e/b8d9)
+  translation = Translation.query.get(translation_id)
+192.168.2.10 - - [16/Apr/2026 22:24:42] "GET /api/result/5 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:45] "GET /admin/translations HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:46] "GET /admin/translation/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:47] "GET /api/result/4 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:48] "GET /admin/translations HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:52] "GET / HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:52] "GET /static/css/style.css HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:52] "GET /static/js/main.js HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:52] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:52] "GET /api/user/info HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:54] "GET / HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:54] "GET /static/css/style.css HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:54] "GET /static/js/main.js HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:24:54] "GET /api/user/info HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:24:54] "GET /static/img/favicon.svg HTTP/1.1" 304 -
+192.168.2.10 - - [16/Apr/2026 22:25:00] "POST /api/upload HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:00] "GET /api/task/ec82bf8c-70b3-474a-b94f-64598c81f7fa HTTP/1.1" 200 -
+/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/app.py:481: LegacyAPIWarning: The Query.get() method is considered legacy as of the 1.x series of SQLAlchemy and becomes a legacy construct in 2.0. The method is now available as Session.get() (deprecated since: 2.0) (Background on SQLAlchemy 2.0 at: https://sqlalche.me/e/b8d9)
+  translation = Translation.query.get(translation_id)
+192.168.2.10 - - [16/Apr/2026 22:25:00] "GET /api/status/6 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:02] "GET /api/task/ec82bf8c-70b3-474a-b94f-64598c81f7fa HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:02] "GET /api/status/6 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:05] "GET /api/task/ec82bf8c-70b3-474a-b94f-64598c81f7fa HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:05] "GET /api/status/6 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:07] "GET /api/task/ec82bf8c-70b3-474a-b94f-64598c81f7fa HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:07] "GET /api/status/6 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:09] "GET /api/task/ec82bf8c-70b3-474a-b94f-64598c81f7fa HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:09] "GET /api/status/6 HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:11] "GET /api/task/ec82bf8c-70b3-474a-b94f-64598c81f7fa HTTP/1.1" 200 -
+192.168.2.10 - - [16/Apr/2026 22:25:11] "GET /api/status/6 HTTP/1.1" 200 -
+127.0.0.1 - - [16/Apr/2026 22:40:01] "GET /api/health HTTP/1.1" 404 -
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+[翻译任务] 开始翻译，使用配置: https://open.bigmodel.cn/api/paas/v4
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+ * Detected change in '/home/xian/.openclaw/workspace-coder/works/pdf-translate-web-v2/services.py', reloading
+ * Restarting with stat
+ * Debugger is active!
+ * Debugger PIN: 154-698-244
+127.0.0.1 - - [16/Apr/2026 23:00:01] "GET /api/health HTTP/1.1" 404 -
+127.0.0.1 - - [16/Apr/2026 23:03:32] "GET / HTTP/1.1" 200 -

models.py

@@ -169,6 +169,7 @@ class Translation(db.Model):
     error_message = db.Column(db.Text, nullable=True)
     
     # 输出
+    upload_path = db.Column(db.String(255), nullable=True)  # 原始PDF文件路径
     output_path = db.Column(db.String(255), nullable=True)  # 翻译结果文件路径
     
     # 时间戳

scripts/fix_upload_path.py

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+"""
+修复旧数据的 upload_path 字段
+通过 file_hash 匹配 uploads 目录中的 PDF 文件
+"""
+
+import sqlite3
+import os
+import hashlib
+
+DB_PATH = 'instance/pdf_translate.db'
+UPLOADS_DIR = 'uploads'
+
+def compute_file_hash(filepath):
+    """计算文件MD5"""
+    hasher = hashlib.md5()
+    with open(filepath, 'rb') as f:
+        hasher.update(f.read())
+    return hasher.hexdigest()
+
+def main():
+    conn = sqlite3.connect(DB_PATH)
+    cursor = conn.cursor()
+    
+    # 获取所有 upload_path 为空的翻译记录
+    cursor.execute('SELECT id, file_hash, original_filename, output_path FROM translations WHERE upload_path IS NULL')
+    records = cursor.fetchall()
+    
+    print(f"找到 {len(records)} 条需要修复的记录")
+    
+    if not records:
+        print("无需修复")
+        conn.close()
+        return
+    
+    # 找所有上传的PDF文件
+    pdf_files = []
+    for root, dirs, files in os.walk(UPLOADS_DIR):
+        for f in files:
+            if f.endswith('.pdf'):
+                pdf_files.append(os.path.join(root, f))
+    
+    print(f"找到 {len(pdf_files)} 个PDF文件")
+    
+    # 按hash匹配
+    fixed_count = 0
+    for record in records:
+        id, file_hash, filename, output_path = record
+        
+        # 找匹配hash的文件
+        for pdf_path in pdf_files:
+            try:
+                pdf_hash = compute_file_hash(pdf_path)
+                if pdf_hash == file_hash:
+                    print(f"ID {id}: 找到匹配 {pdf_path}")
+                    cursor.execute('UPDATE translations SET upload_path = ? WHERE id = ?', (pdf_path, id))
+                    conn.commit()
+                    fixed_count += 1
+                    break
+            except Exception as e:
+                print(f"处理 {pdf_path} 失败: {e}")
+    
+    conn.close()
+    print(f"修复完成，共修复 {fixed_count} 条记录")
+
+if __name__ == '__main__':
+    main()

services.py

@@ -7,10 +7,20 @@ import json
 import time
 import hashlib
 import threading
+import base64
+import io
 from datetime import datetime, timedelta
 from pypdf import PdfReader
 from openai import OpenAI
 from flask import current_app
+from PIL import Image
+
+# pdf2image 用于将PDF转为图像
+try:
+    from pdf2image import convert_from_path
+    PDF_TO_IMAGE_AVAILABLE = True
+except ImportError:
+    PDF_TO_IMAGE_AVAILABLE = False
 
 # ==================== LLM客户端 ====================
 class TranslationService:
@@ -99,6 +109,115 @@ class TranslationService:
         text = re.sub(r'[\x00-\x08\x0b\x0c\x0e-\x1f]', '', text)
         return text.strip()
     
+    def is_vision_model(self):
+        """检查是否是视觉模型"""
+        model = self.llm_config.get('model', '')
+        # 常见视觉模型名称
+        vision_models = ['vision', 'vlm', 'glm-4v', 'glm-4.6v', 'gpt-4-vision', 'gpt-4o', 'qwen-vl', 'claude-3']
+        return any(v in model.lower() for v in vision_models)
+    
+    def pdf_to_images(self, pdf_path, max_pages=None):
+        """将PDF页面转换为图像"""
+        if not PDF_TO_IMAGE_AVAILABLE:
+            return None, "pdf2image未安装，无法处理扫描版PDF。请安装: pip install pdf2image"
+        
+        try:
+            # 获取PDF页数
+            reader = PdfReader(pdf_path)
+            total_pages = len(reader.pages)
+            
+            if max_pages:
+                pages_to_convert = min(max_pages, total_pages)
+            else:
+                pages_to_convert = total_pages
+            
+            # 转换PDF为图像
+            images = convert_from_path(
+                pdf_path,
+                first_page=1,
+                last_page=pages_to_convert,
+                dpi=200,  # 适当的DPI
+                fmt='jpeg'
+            )
+            
+            return images, None
+            
+        except Exception as e:
+            return None, f"PDF转图像失败: {str(e)}"
+    
+    def extract_text_from_image(self, image):
+        """使用视觉模型从图像中提取文字"""
+        if not self.is_vision_model():
+            return None, "当前模型不是视觉模型，无法识别图像文字"
+        
+        try:
+            # 将图像转为base64
+            buffered = io.BytesIO()
+            image.save(buffered, format="JPEG")
+            img_base64 = base64.b64encode(buffered.getvalue()).decode('utf-8')
+            
+            # 构建多模态请求
+            response = self.client.chat.completions.create(
+                model=self.llm_config['model'],
+                messages=[
+                    {
+                        "role": "user",
+                        "content": [
+                            {
+                                "type": "text",
+                                "text": "请识别并提取这张图片中的所有文字内容。只输出提取的文字，不要添加任何解释或说明。保持原有的段落和格式。"
+                            },
+                            {
+                                "type": "image_url",
+                                "image_url": {
+                                    "url": f"data:image/jpeg;base64,{img_base64}"
+                                }
+                            }
+                        ]
+                    }
+                ],
+                max_tokens=self.llm_config['max_tokens'],
+                temperature=0.1,
+                timeout=self.llm_config['timeout'],
+            )
+            
+            content = response.choices[0].message.content
+            return content.strip() if content else '', None
+            
+        except Exception as e:
+            return '', f"视觉模型识别失败: {str(e)}"
+    
+    def extract_text_from_scanned_pdf(self, pdf_path, progress_callback=None):
+        """从扫描版PDF提取文字（使用视觉模型OCR）"""
+        images, error = self.pdf_to_images(pdf_path)
+        
+        if error:
+            return [], error
+        
+        pages_text = []
+        total = len(images)
+        
+        for i, image in enumerate(images):
+            if progress_callback:
+                progress_callback(int((i+1)/total*50), total, f"OCR识别第{i+1}页...")
+            
+            text, err = self.extract_text_from_image(image)
+            
+            if err:
+                pages_text.append({
+                    'page': i + 1,
+                    'text': '',
+                    'error': err
+                })
+            else:
+                pages_text.append({
+                    'page': i + 1,
+                    'text': text or '',
+                    'error': None
+                })
+        
+        return pages_text, None
+    
     def chunk_text(self, text, max_size=2000):
         """分块"""
         paragraphs = text.split('\n\n')
@@ -131,11 +250,38 @@ class TranslationService:
         Returns:
             翻译统计信息
         """
+        # 先尝试常规提取
         pages = self.extract_pdf_text(pdf_path)
         total_pages = len(pages)
+        total_text = sum(len(p['text']) for p in pages)
+        
+        # 如果无法提取文本，尝试使用视觉模型OCR
+        if total_pages == 0 or total_text < 10:
+            if self.is_vision_model() and PDF_TO_IMAGE_AVAILABLE:
+                if progress_callback:
+                    progress_callback(0, 0, "检测到扫描版PDF，使用视觉模型OCR...")
+                
+                pages, error = self.extract_text_from_scanned_pdf(pdf_path, progress_callback)
+                
+                if error:
+                    raise ValueError(error)
+                
+                total_pages = len(pages)
+                total_text = sum(len(p['text']) for p in pages)
+                
+                if total_text < 10:
+                    raise ValueError("视觉模型OCR未能提取到有效文字内容")
                 
                 if progress_callback:
-            progress_callback(0, total_pages, "开始翻译...")
+                    progress_callback(50, total_pages, "OCR完成，开始翻译...")
+            else:
+                error_msg = "PDF无法提取文本内容。可能原因：\n1. PDF是扫描版（图像形式）\n2. 当前大模型不是视觉模型，无法识别图像文字\n\n如需处理扫描版PDF，请配置视觉大模型（如 glm-4.6v、gpt-4-vision）"
+                if progress_callback:
+                    progress_callback(0, 0, error_msg)
+                raise ValueError(error_msg)
+        
+        if progress_callback:
+            progress_callback(50, total_pages, "开始翻译...")
         
         translated_pages = []
         total_chunks = 0
@@ -150,8 +296,10 @@ class TranslationService:
                 translated_chunks.append(translated)
                 
                 if progress_callback:
-                    progress = int((i + 1) / len(chunks) * 100 / total_pages)
-                    progress_callback(progress, total_pages, f"翻译第{page_data['page']}页")
+                    # OCR占50%，翻译占50%
+                    page_progress = (i + 1) / len(chunks)
+                    overall_progress = 50 + int(page_progress * 50 / total_pages)
+                    progress_callback(overall_progress, total_pages, f"翻译第{page_data['page']}页")
             
             translated_pages.append({
                 'page': page_data['page'],

static/js/main.js

@@ -70,6 +70,9 @@ document.getElementById('uploadForm').addEventListener('submit', async function(
         currentTranslationId = result.translation_id;
         currentTaskId = result.task_id;
         
+        // 更新按钮状态为翻译中
+        btnText.textContent = '翻译中...';
+        
         // 如果使用缓存，直接显示结果
         if (result.from_cache) {
             document.getElementById('cacheNotice').style.display = 'block';
@@ -201,20 +204,57 @@ document.getElementById('viewCompare')?.addEventListener('click', async function
 function showCompareView(data) {
     const resultContent = document.getElementById('resultContent');
     
-    resultContent.innerHTML = `
-        <div class="compare-container">
-            <div class="compare-panel original">
-                <h5>原文</h5>
-                <div class="content">${escapeHtml(data.original)}</div>
+    // 如果有分页数据，按页显示
+    if (data.pages && data.pages.length > 0) {
+        let html = '<div class="compare-container">';
+        
+        for (const page of data.pages) {
+            html += `
+                <div class="compare-section mb-4">
+                    <h5 class="text-center mb-3">第 ${page.page} 页</h5>
+                    <div class="row">
+                        <div class="col-md-6">
+                            <div class="card">
+                                <div class="card-header">原文（OCR识别）</div>
+                                <div class="card-body"><pre style="white-space: pre-wrap;">${escapeHtml(data.original || '原文内容')}</pre></div>
+                            </div>
+                        </div>
+                        <div class="col-md-6">
+                            <div class="card">
+                                <div class="card-header">译文</div>
+                                <div class="card-body">${renderMarkdown(page.content)}</div>
+                            </div>
                         </div>
-            <div class="compare-panel translated">
-                <h5>译文</h5>
-                <div class="content">${renderMarkdown(data.translated)}</div>
                     </div>
                 </div>
             `;
         }
         
+        html += '</div>';
+        resultContent.innerHTML = html;
+    } else {
+        // 单一对比视图
+        resultContent.innerHTML = `
+            <div class="compare-container">
+                <div class="row">
+                    <div class="col-md-6">
+                        <div class="card">
+                            <div class="card-header">原文</div>
+                            <div class="card-body"><pre style="white-space: pre-wrap;">${escapeHtml(data.original || '无原文内容')}</pre></div>
+                        </div>
+                    </div>
+                    <div class="col-md-6">
+                        <div class="card">
+                            <div class="card-header">译文</div>
+                            <div class="card-body">${renderMarkdown(data.translated || '无译文内容')}</div>
+                        </div>
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+}
+
 // 重新翻译
 document.getElementById('retranslateBtn')?.addEventListener('click', async function() {
     if (!currentTranslationId) return;

templates/translation.html

@@ -84,11 +84,21 @@
                     const response = await fetch(`/api/compare/${translationId}`);
                     const result = await response.json();
                     
+                    // 原文面板：如果有PDF URL用iframe显示，否则显示提取的文本
+                    let originalHtml = '';
+                    if (result.original_pdf_url) {
+                        originalHtml = `<iframe src="${result.original_pdf_url}" style="width:100%;height:500px;border:none;"></iframe>`;
+                    } else if (result.original && result.original.length > 0) {
+                        originalHtml = `<div style="white-space:pre-wrap;font-family:monospace;">${escapeHtml(result.original)}</div>`;
+                    } else {
+                        originalHtml = '<div class="text-muted">原文内容未找到（可能PDF已被删除）</div>';
+                    }
+                    
                     document.getElementById('resultContent').innerHTML = `
                         <div class="compare-container">
                             <div class="compare-panel original">
-                                <h5>原文</h5>
-                                <div>${escapeHtml(result.original)}</div>
+                                <h5>原文 PDF</h5>
+                                ${originalHtml}
                             </div>
                             <div class="compare-panel translated">
                                 <h5>译文</h5>
@@ -98,7 +108,7 @@
                     `;
                     
                 } catch (error) {
-                    alert('加载对比失败');
+                    alert('加载对比失败: ' + error.message);
                 }
             } else {
                 loadResult();

test_simple.txt

@@ -0,0 +1 @@
+This is a test PDF file. Hello World. This is a simple test document.