fix(agetn features):

1. Historical multimodal message writing is incorporated into the conversation context;
2. Resolve the issues where csv, json, and txt files cannot be recognized due to encoding problems;
3. File quantity limit;
4. Error details

api/app/controllers/file_storage_controller.py

@@ -91,7 +91,7 @@ async def upload_file(
 
     if file_size > settings.MAX_FILE_SIZE:
         raise HTTPException(
-            status_code=status.HTTP_400_BAD_REQUEST,
+            status_code=status.HTTP_413_CONTENT_TOO_LARGE,
             detail=f"The file size exceeds the {settings.MAX_FILE_SIZE} byte limit"
         )
 

api/app/core/config.py

@@ -97,6 +97,7 @@ class Settings:
 
     # File Upload
     MAX_FILE_SIZE: int = int(os.getenv("MAX_FILE_SIZE", "52428800"))
+    MAX_FILE_COUNT: int = int(os.getenv("MAX_FILE_COUNT", "20"))
     FILE_PATH: str = os.getenv("FILE_PATH", "/files")
     FILE_URL_EXPIRES: int = int(os.getenv("FILE_URL_EXPIRES", "3600"))
 

api/app/main.py

@@ -506,10 +506,13 @@ async def http_exception_handler(request: Request, exc: HTTPException):
         404: "errors.common.not_found",
         405: "errors.common.method_not_allowed",
         409: "errors.common.conflict",
+        413: "errors.common.payload_too_large",
         422: "errors.common.validation_failed",
         429: "errors.common.too_many_requests",
         500: "errors.common.internal_error",
+        502: "errors.common.bad_gateway",
         503: "errors.common.service_unavailable",
+        504: "errors.common.gateway_timeout",
     }
     
     # 如果有对应的翻译键，使用翻译
@@ -534,7 +537,7 @@ async def http_exception_handler(request: Request, exc: HTTPException):
     
     return JSONResponse(
         status_code=exc.status_code,
-        content=fail(code=exc.status_code, msg=translated_message, error=translated_message)
+        content=fail(code=exc.status_code, msg=translated_message, error=exc.detail)
     )
 
 

api/app/schemas/app_schema.py

@@ -151,7 +151,7 @@ class FileUploadConfig(BaseModel):
     document_enabled: bool = Field(default=False)
     document_max_size_mb: int = Field(default=100)
     document_allowed_extensions: List[str] = Field(
-        default=["pdf", "docx", "xlsx", "txt", "csv", "json", "md"]
+        default=["pdf", "docx", "doc", "xlsx", "xls", "txt", "csv", "json", "md"]
     )
     # 视频文件：MP4/MOV/AVI/WebM，最大 500MB
     video_enabled: bool = Field(default=False)
@@ -160,7 +160,15 @@ class FileUploadConfig(BaseModel):
         default=["mp4", "mov"]
     )
     # 最大文件数量
-    max_file_count: int = Field(default=5, ge=1, le=20)
+    max_file_count: int = Field(default=5, ge=1)
+
+    @field_validator("max_file_count")
+    @classmethod
+    def validate_max_file_count(cls, v: int) -> int:
+        from app.core.config import settings
+        if v > settings.MAX_FILE_COUNT:
+            raise ValueError(f"max_file_count 不能超过 {settings.MAX_FILE_COUNT}")
+        return v
 
 
 class OpeningStatementConfig(BaseModel):

api/app/services/app_chat_service.py

@@ -124,7 +124,7 @@ class AppChatService:
             limit=10
         )
         history = [
-            {"role": msg.role, "content": msg.content}
+            {"role": msg.role, "content": [{"type": "text", "text": msg.content}] + msg.meta_data.get("files", [])}
             for msg in messages
         ]
 
@@ -188,12 +188,7 @@ class AppChatService:
             "audio_url": None
         }
         if files:
-            for f in files:
+            human_meta["files"].extend(processed_files)
-                # url = await MultimodalService(self.db).get_file_url(f)
-                human_meta["files"].append({
-                    "type": f.type,
-                    "url": f.url
-                })
 
         # 保存消息
         if audio_url:
@@ -322,7 +317,7 @@ class AppChatService:
                     limit=memory_config.get("max_history", 10)
                 )
                 history = [
-                    {"role": msg.role, "content": msg.content}
+                    {"role": msg.role, "content": [{"type": "text", "text": msg.content}] + msg.meta_data.get("files", [])}
                     for msg in messages
                 ]
 

api/app/services/conversation_service.py

@@ -291,7 +291,7 @@ class ConversationService:
         history = [
             {
                 "role": msg.role,
-                "content": msg.content
+                "content": [{"type": "text", "text": msg.content}] + msg.meta_data.get("files", [])
             }
             for msg in messages
         ]

api/app/services/draft_run_service.py

@@ -658,7 +658,7 @@ class AgentRunService:
                             "total_tokens": 0
                         })
                     },
-                    files=files,
+                    files=processed_files,
                     audio_url=audio_url
                 )
 
@@ -820,6 +820,7 @@ class AgentRunService:
                 conversation_id=conversation_id,
                 max_history=memory_config.get("max_history", 10)
             )
+            print(history)
 
             # 6. 处理多模态文件
             processed_files = None
@@ -904,7 +905,7 @@ class AgentRunService:
                     meta_data={
                         "usage": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": total_tokens}
                     },
-                    files=files,
+                    files=processed_files,
                     audio_url=stream_audio_url
                 )
 
@@ -1182,12 +1183,7 @@ class AgentRunService:
                 "files": []
             }
             if files:
-                for f in files:
+                human_meta["files"].extend(files)
-                    # url = await MultimodalService(self.db).get_file_url(f)
-                    human_meta["files"].append({
-                        "type": f.type,
-                        "url": f.url
-                    })
             # 保存用户消息
             conversation_service.add_message(
                 conversation_id=conv_uuid,

api/app/services/multimodal_service.py

@@ -11,6 +11,8 @@
 import base64
 import io
 import uuid
+import zipfile
+import chardet
 from abc import ABC, abstractmethod
 from typing import List, Dict, Any, Optional
 
@@ -42,12 +44,10 @@ PDF_MIME = ['application/pdf']
 DOC_MIME = [
     'application/msword',
     'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
-    'application/zip'
 ]
 XLSX_MIME = [
     'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
     'application/vnd.ms-excel',
-    'application/zip'
 ]
 CSV_MIME = ['text/csv', 'application/csv']
 JSON_MIME = ['application/json']
@@ -588,12 +588,12 @@ class MultimodalService:
                     file.set_content(file_content)
             file_mime_type = magic.from_buffer(file_content, mime=True)
             if file_mime_type in TEXT_MIME:
-                return file_content.decode("utf-8")
+                return self._decode_text_safe(file_content)
             elif file_mime_type in PDF_MIME:
                 return await self._extract_pdf_text(file_content)
-            elif file_mime_type in DOC_MIME and file.file_type.endswith(('docx', 'doc')):
+            elif self._is_word_file(file_content, file_mime_type):
                 return await self._extract_word_text(file_content)
-            elif file_mime_type in XLSX_MIME and file.file_type.endswith(("xlsx", "xls")):
+            elif self._is_excel_file(file_content, file_mime_type):
                 return await self._extract_xlsx_text(file_content)
             elif file_mime_type in CSV_MIME:
                 return await self._extract_csv_text(file_content)
@@ -647,27 +647,89 @@ class MultimodalService:
             logger.error(f"提取 Excel 文本失败: {e}")
             return f"[Excel 提取失败: {str(e)}]"
 
-    @staticmethod
+    async def _extract_csv_text(self, file_content: bytes) -> str:
-    async def _extract_csv_text(file_content: bytes) -> str:
         """提取 CSV 文本"""
         try:
-            text = file_content.decode('utf-8-sig')
+            text = self._decode_text_safe(file_content)
             reader = csv.reader(io.StringIO(text))
             return '\n'.join('\t'.join(row) for row in reader)
         except Exception as e:
             logger.error(f"提取 CSV 文本失败: {e}")
             return f"[CSV 提取失败: {str(e)}]"
 
-    @staticmethod
+    async def _extract_json_text(self, file_content: bytes) -> str:
-    async def _extract_json_text(file_content: bytes) -> str:
         """提取 JSON 文本"""
         try:
-            data = json.loads(file_content.decode('utf-8'))
+            text = self._decode_text_safe(file_content)
+            data = json.loads(text)
             return json.dumps(data, ensure_ascii=False, indent=2)
         except Exception as e:
             logger.error(f"提取 JSON 文本失败: {e}")
             return f"[JSON 提取失败: {str(e)}]"
 
+    def _is_word_file(self, file_content: bytes, mime_type: str) -> bool:
+        """判断是不是 Word 文件（doc / docx），不依赖后缀"""
+        # 旧版 .doc
+        if mime_type == 'application/msword':
+            return True
+
+        # 新版 .docx（ZIP 内部包含 word/document.xml）
+        header = file_content[:4]
+        if header == b'PK\x03\x04':
+            try:
+                with zipfile.ZipFile(io.BytesIO(file_content)) as zf:
+                    return "word/document.xml" in zf.namelist()
+            except:
+                pass
+
+        return False
+
+    def _is_excel_file(self, file_content: bytes, mime_type: str) -> bool:
+        """判断是不是 Excel 文件（xls / xlsx），不依赖后缀"""
+        # 旧版 .xls
+        if mime_type == 'application/vnd.ms-excel':
+            return True
+
+        # 新版 .xlsx（ZIP 内部包含 xl/workbook.xml）
+        header = file_content[:4]
+        if header == b'PK\x03\x04':
+            try:
+                with zipfile.ZipFile(io.BytesIO(file_content)) as zf:
+                    return "xl/workbook.xml" in zf.namelist()
+            except:
+                pass
+
+        return False
+
+    @staticmethod
+    def _decode_text_safe(file_content: bytes) -> str:
+        """
+        【万能文本解码】
+        自动检测编码，支持 utf-8 / gbk / gb2312 / utf-8-sig / ascii 等
+        永远不报错，永远不乱码
+        """
+        if not file_content:
+            return ""
+
+        # 1. 自动检测文件编码
+        detect = chardet.detect(file_content)
+        encoding = detect.get("encoding", "utf-8").lower()
+
+        # 2. 兼容常见中文编码
+        compatible_encodings = ["utf-8", "gbk", "gb18030", "gb2312", "ascii", "latin-1"]
+
+        # 3. 按优先级尝试解码
+        for enc in [encoding] + compatible_encodings:
+            if not enc:
+                continue
+            try:
+                return file_content.decode(enc.strip())
+            except (UnicodeDecodeError, LookupError):
+                continue
+
+        # 终极兜底
+        return file_content.decode("utf-8", errors="replace")
+
 
 def get_multimodal_service(db: Session) -> MultimodalService:
     """获取多模态服务实例（依赖注入）"""

api/app/services/shared_chat_service.py

@@ -264,7 +264,7 @@ class SharedChatService:
                 limit=memory_config.get("max_history", 10)
             )
             history = [
-                {"role": msg.role, "content": msg.content}
+                {"role": msg.role, "content": [{"type": "text", "text": msg.content}] + msg.meta_data.get("files", [])}
                 for msg in messages
             ]
 
@@ -472,7 +472,7 @@ class SharedChatService:
                     limit=memory_config.get("max_history", 10)
                 )
                 history = [
-                    {"role": msg.role, "content": msg.content}
+                    {"role": msg.role, "content": [{"type": "text", "text": msg.content}] + msg.meta_data.get("files", [])}
                     for msg in messages
                 ]