fix: 修复滚动同步位置计算+添加调试日志

- 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()

static/css/style.css

@@ -75,12 +75,21 @@ body {
     display: grid;
     grid-template-columns: 1fr 1fr;
     gap: 20px;
+    height: 70vh;
 }
 
 .compare-panel {
     padding: 15px;
     border-radius: 5px;
     background-color: #f8f9fa;
+    overflow-y: auto;
+    height: 100%;
+}
+
+.compare-panel h5 {
+    margin-bottom: 10px;
+    padding-bottom: 5px;
+    border-bottom: 1px solid #ddd;
 }
 
 .compare-panel.original {
@@ -91,6 +100,48 @@ body {
     border-left: 4px solid var(--success-color);
 }
 
+.compare-panel iframe {
+    width: 100%;
+    height: calc(100% - 40px);
+    border: none;
+}
+
+#pdfContainer {
+    height: calc(100% - 40px);
+    overflow-y: auto;
+}
+
+#pdfPages {
+    padding: 10px;
+}
+
+.pdf-page-canvas {
+    border: 1px solid #ddd;
+    border-radius: 3px;
+    background: white;
+}
+
+.translated-page {
+    padding-bottom: 30px;
+    border-bottom: 2px dashed #ddd;
+    margin-bottom: 20px;
+}
+
+.translated-page:last-child {
+    border-bottom: none;
+}
+
+.page-header {
+    font-size: 0.9em;
+    padding: 5px 10px;
+    background: #f0f0f0;
+    border-radius: 3px;
+}
+
+.page-content {
+    padding: 10px;
+}
+
 /* 会员卡片 */
 .pricing-card {
     transition: transform 0.3s ease;

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

@@ -7,6 +7,7 @@
     <link rel="icon" href="/static/img/favicon.svg" type="image/svg+xml">
     <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/css/bootstrap.min.css" rel="stylesheet">
     <link href="/static/css/style.css" rel="stylesheet">
+    <script src="https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.11.174/pdf.min.js"></script>
 </head>
 <body>
     <nav class="navbar navbar-expand-lg navbar-dark bg-primary">
@@ -76,6 +77,11 @@
         }
         
         // 切换对比视图
+        let syncScrollEnabled = true;
+        let pdfDoc = null;
+        let pagePositions = [];  // PDF各页位置
+        let translatedPagePositions = [];  // 译文各页位置
+        
         document.getElementById('toggleCompare').addEventListener('click', async function() {
             showCompare = !showCompare;
             
@@ -84,27 +90,220 @@
                     const response = await fetch(`/api/compare/${translationId}`);
                     const result = await response.json();
                     
+                    // 原文面板：如果有PDF URL用PDF.js渲染
+                    let originalHtml = '';
+                    if (result.original_pdf_url) {
+                        originalHtml = '<div id="pdfPages"></div>';
+                    } else if (result.original && result.original.length > 0) {
+                        originalHtml = `<div class="original-text" style="white-space:pre-wrap;font-family:monospace;">${escapeHtml(result.original)}</div>`;
+                    } else {
+                        originalHtml = '<div class="text-muted">原文内容未找到（可能PDF已被删除）</div>';
+                    }
+                    
+                    // 译文按页渲染
+                    let translatedHtml = renderTranslatedByPage(result.translated);
+                    
                     document.getElementById('resultContent').innerHTML = `
                         <div class="compare-container">
-                            <div class="compare-panel original">
-                                <h5>原文</h5>
-                                <div>${escapeHtml(result.original)}</div>
+                            <div class="compare-panel original" id="originalPanel">
+                                <h5>原文 PDF <small class="text-muted">(共${result.pages.length}页)</small></h5>
+                                <div id="pdfContainer">${originalHtml}</div>
                             </div>
-                            <div class="compare-panel translated">
-                                <h5>译文</h5>
-                                <div>${renderMarkdown(result.translated)}</div>
+                            <div class="compare-panel translated" id="translatedPanel">
+                                <h5>译文 <small class="text-muted">(共${result.pages.length}页)</small></h5>
+                                <div class="translated-content">${translatedHtml}</div>
                             </div>
                         </div>
+                        <div class="text-center mt-2">
+                            <small class="text-muted">💡 滚动任意一侧，另一侧自动同步到对应页</small>
+                        </div>
                     `;
                     
+                    // 先计算译文位置（因为已经渲染好了）
+                    setTimeout(calculateTranslatedPositions, 100);
+                    
+                    // 如果有PDF，用PDF.js渲染，完成后计算位置并启用同步
+                    if (result.original_pdf_url) {
+                        renderPDF(result.original_pdf_url).then(() => {
+                            setTimeout(() => {
+                                calculatePdfPositions();
+                                enableSyncScroll();
+                            }, 200);
+                        });
+                    } else {
+                        setTimeout(enableSyncScroll, 300);
+                    }
+                    
                 } catch (error) {
-                    alert('加载对比失败');
+                    alert('加载对比失败: ' + error.message);
                 }
             } else {
                 loadResult();
             }
         });
         
+        // 译文按页渲染
+        function renderTranslatedByPage(content) {
+            if (!content) return '<div class="text-muted">译文加载失败</div>';
+            
+            // 解析分页 - 按 "---" 分隔
+            const parts = content.split(/\n\n---\n\n/);
+            let html = '';
+            
+            parts.forEach((part, index) => {
+                // 提取页码
+                const pageNumMatch = part.match(/## 第 (\d+) 页\n\n/);
+                const pageNum = pageNumMatch ? pageNumMatch[1] : (index + 1);
+                
+                // 移除页码标题
+                let contentOnly = part.replace(/## 第 \d+ 页\n\n/, '').replace(/^# .*\n\n/, '').replace(/^> .*\n\n/, '');
+                
+                html += `
+                    <div class="translated-page" data-page="${pageNum}">
+                        <div class="page-header text-center text-muted mb-3">— 第 ${pageNum} 页 —</div>
+                        <div class="page-content">${renderMarkdown(contentOnly)}</div>
+                    </div>
+                `;
+            });
+            
+            return html;
+        }
+        
+        // PDF.js渲染PDF（返回Promise）
+        async function renderPDF(url) {
+            pdfjsLib.GlobalWorkerOptions.workerSrc = 'https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.11.174/pdf.worker.min.js';
+            
+            const pdf = await pdfjsLib.getDocument(url).promise;
+            pdfDoc = pdf;
+            
+            const container = document.getElementById('pdfPages');
+            container.innerHTML = '';
+            
+            // 渲染所有页面
+            for (let i = 1; i <= pdf.numPages; i++) {
+                // 页码标识
+                const pageNumDiv = document.createElement('div');
+                pageNumDiv.className = 'page-header text-center text-muted mb-2';
+                pageNumDiv.textContent = `— 第 ${i} 页 —`;
+                pageNumDiv.dataset.page = i;
+                container.appendChild(pageNumDiv);
+                
+                const page = await pdf.getPage(i);
+                const canvas = document.createElement('canvas');
+                canvas.className = 'pdf-page-canvas';
+                canvas.style.width = '100%';
+                canvas.style.display = 'block';
+                canvas.style.marginBottom = '20px';
+                canvas.dataset.page = i;
+                
+                const context = canvas.getContext('2d');
+                const viewport = page.getViewport({ scale: 1.5 });
+                
+                canvas.height = viewport.height;
+                canvas.width = viewport.width;
+                
+                await page.render({
+                    canvasContext: context,
+                    viewport: viewport
+                }).promise;
+                
+                container.appendChild(canvas);
+            }
+            
+            return pdf;
+        }
+        
+        // 计算译文各页位置（相对于滚动容器顶部）
+        function calculateTranslatedPositions() {
+            translatedPagePositions = [];
+            const translatedPanel = document.getElementById('translatedPanel');
+            const translatedPages = document.querySelectorAll('.translated-page');
+            
+            translatedPages.forEach(el => {
+                const page = parseInt(el.dataset.page);
+                // 使用getBoundingClientRect计算相对位置
+                const rect = el.getBoundingClientRect();
+                const panelRect = translatedPanel.getBoundingClientRect();
+                // 相对位置 = 元素top - 容器top + 当前scrollTop
+                const relativeTop = rect.top - panelRect.top + translatedPanel.scrollTop;
+                translatedPagePositions.push({
+                    page: page,
+                    top: relativeTop
+                });
+            });
+            
+            console.log('译文页位置:', translatedPagePositions);
+        }
+        
+        // 计算PDF各页位置（相对于滚动容器顶部）
+        function calculatePdfPositions() {
+            pagePositions = [];
+            const originalPanel = document.getElementById('originalPanel');
+            const pdfPages = document.querySelectorAll('#pdfPages .page-header');
+            
+            pdfPages.forEach(el => {
+                const page = parseInt(el.dataset.page);
+                // 使用getBoundingClientRect计算相对位置
+                const rect = el.getBoundingClientRect();
+                const panelRect = originalPanel.getBoundingClientRect();
+                // 相对位置 = 元素top - 容器top + 当前scrollTop
+                const relativeTop = rect.top - panelRect.top + originalPanel.scrollTop;
+                pagePositions.push({
+                    page: page,
+                    top: relativeTop
+                });
+            });
+            
+            console.log('PDF页位置:', pagePositions);
+        }
+        
+        // 根据滚动位置找当前页
+        function findCurrentPage(scrollTop, positions) {
+            for (let i = positions.length - 1; i >= 0; i--) {
+                if (scrollTop >= positions[i].top - 30) {
+                    return positions[i].page;
+                }
+            }
+            return 1;
+        }
+        
+        // 滚动到指定页
+        function scrollToPage(panel, positions, pageNum) {
+            const pos = positions.find(p => p.page === pageNum);
+            if (pos) {
+                panel.scrollTop = pos.top - 10;
+            }
+        }
+        
+        // 滚动同步
+        function enableSyncScroll() {
+            const originalPanel = document.getElementById('originalPanel');
+            const translatedPanel = document.getElementById('translatedPanel');
+            
+            if (!originalPanel || !translatedPanel || pagePositions.length === 0 || translatedPagePositions.length === 0) {
+                console.log('滚动同步未启用: 缺少元素或位置数据');
+                return;
+            }
+            
+            console.log('滚动同步已启用');
+            
+            originalPanel.addEventListener('scroll', function() {
+                if (!syncScrollEnabled) return;
+                const currentPage = findCurrentPage(this.scrollTop, pagePositions);
+                syncScrollEnabled = false;
+                scrollToPage(translatedPanel, translatedPagePositions, currentPage);
+                setTimeout(() => syncScrollEnabled = true, 100);
+            });
+            
+            translatedPanel.addEventListener('scroll', function() {
+                if (!syncScrollEnabled) return;
+                const currentPage = findCurrentPage(this.scrollTop, translatedPagePositions);
+                syncScrollEnabled = false;
+                scrollToPage(originalPanel, pagePositions, currentPage);
+                setTimeout(() => syncScrollEnabled = true, 100);
+            });
+        }
+        
         // 重新翻译
         document.getElementById('retranslateBtn')?.addEventListener('click', async function() {
             const instruction = document.getElementById('retranslateInstruction').value;

test_simple.txt

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