Feat: output intinity test log (#12097)

### What problem does this PR solve?

Output log to file when run infinity tests.

### Type of change


- [x] New Feature (non-breaking change which adds functionality)

.github/workflows/tests.yml

@@ -247,7 +247,7 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
+          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api > infinity_sdk_test.log
 
       - name: Run frontend api tests against Infinity
         run: |
@@ -256,7 +256,7 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short sdk/python/test/test_frontend_api/get_email.py sdk/python/test/test_frontend_api/test_dataset.py
+          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short sdk/python/test/test_frontend_api/get_email.py sdk/python/test/test_frontend_api/test_dataset.py > infinity_api_test.log
 
       - name: Run http api tests against Infinity
         run: |
@@ -265,7 +265,7 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
+          source .venv/bin/activate && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api > infinity_http_api_test.log
 
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed

api/apps/__init__.py

@@ -108,7 +108,7 @@ def _load_user():
     authorization = request.headers.get("Authorization")
     g.user = None
     if not authorization:
-        return
+        return None
 
     try:
         access_token = str(jwt.loads(authorization))

api/apps/llm_app.py

@@ -25,7 +25,7 @@ from api.utils.api_utils import get_allowed_llm_factories, get_data_error_result
 from common.constants import StatusEnum, LLMType
 from api.db.db_models import TenantLLM
 from rag.utils.base64_image import test_image
-from rag.llm import EmbeddingModel, ChatModel, RerankModel, CvModel, TTSModel, OcrModel
+from rag.llm import EmbeddingModel, ChatModel, RerankModel, CvModel, TTSModel, OcrModel, Seq2txtModel
 
 
 @manager.route("/factories", methods=["GET"])  # noqa: F821
@@ -208,70 +208,83 @@ async def add_llm():
     msg = ""
     mdl_nm = llm["llm_name"].split("___")[0]
     extra = {"provider": factory}
-    if llm["model_type"] == LLMType.EMBEDDING.value:
+    model_type = llm["model_type"]
-        assert factory in EmbeddingModel, f"Embedding model from {factory} is not supported yet."
+    model_api_key = llm["api_key"]
-        mdl = EmbeddingModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
+    model_base_url = llm.get("api_base", "")
-        try:
+    match model_type:
-            arr, tc = mdl.encode(["Test if the api key is available"])
+        case LLMType.EMBEDDING.value:
-            if len(arr[0]) == 0:
+            assert factory in EmbeddingModel, f"Embedding model from {factory} is not supported yet."
-                raise Exception("Fail")
+            mdl = EmbeddingModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
-        except Exception as e:
+            try:
-            msg += f"\nFail to access embedding model({mdl_nm})." + str(e)
+                arr, tc = mdl.encode(["Test if the api key is available"])
-    elif llm["model_type"] == LLMType.CHAT.value:
+                if len(arr[0]) == 0:
-        assert factory in ChatModel, f"Chat model from {factory} is not supported yet."
+                    raise Exception("Fail")
-        mdl = ChatModel[factory](
+            except Exception as e:
-            key=llm["api_key"],
+                msg += f"\nFail to access embedding model({mdl_nm})." + str(e)
-            model_name=mdl_nm,
+        case LLMType.CHAT.value:
-            base_url=llm["api_base"],
+            assert factory in ChatModel, f"Chat model from {factory} is not supported yet."
-            **extra,
+            mdl = ChatModel[factory](
-        )
+                key=model_api_key,
-        try:
+                model_name=mdl_nm,
-            m, tc = await mdl.async_chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {"temperature": 0.9})
+                base_url=model_base_url,
-            if not tc and m.find("**ERROR**:") >= 0:
+                **extra,
-                raise Exception(m)
+            )
-        except Exception as e:
+            try:
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+                m, tc = await mdl.async_chat(None, [{"role": "user", "content": "Hello! How are you doing!"}],
-    elif llm["model_type"] == LLMType.RERANK:
+                                             {"temperature": 0.9})
-        assert factory in RerankModel, f"RE-rank model from {factory} is not supported yet."
+                if not tc and m.find("**ERROR**:") >= 0:
-        try:
+                    raise Exception(m)
-            mdl = RerankModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
+            except Exception as e:
-            arr, tc = mdl.similarity("Hello~ RAGFlower!", ["Hi, there!", "Ohh, my friend!"])
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-            if len(arr) == 0:
+
-                raise Exception("Not known.")
+        case LLMType.RERANK.value:
-        except KeyError:
+            assert factory in RerankModel, f"RE-rank model from {factory} is not supported yet."
-            msg += f"{factory} dose not support this model({factory}/{mdl_nm})"
+            try:
-        except Exception as e:
+                mdl = RerankModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+                arr, tc = mdl.similarity("Hello~ RAGFlower!", ["Hi, there!", "Ohh, my friend!"])
-    elif llm["model_type"] == LLMType.IMAGE2TEXT.value:
+                if len(arr) == 0:
-        assert factory in CvModel, f"Image to text model from {factory} is not supported yet."
+                    raise Exception("Not known.")
-        mdl = CvModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
+            except KeyError:
-        try:
+                msg += f"{factory} dose not support this model({factory}/{mdl_nm})"
-            image_data = test_image
+            except Exception as e:
-            m, tc = mdl.describe(image_data)
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-            if not tc and m.find("**ERROR**:") >= 0:
+
-                raise Exception(m)
+        case LLMType.IMAGE2TEXT.value:
-        except Exception as e:
+            assert factory in CvModel, f"Image to text model from {factory} is not supported yet."
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+            mdl = CvModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
-    elif llm["model_type"] == LLMType.TTS:
+            try:
-        assert factory in TTSModel, f"TTS model from {factory} is not supported yet."
+                image_data = test_image
-        mdl = TTSModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm["api_base"])
+                m, tc = mdl.describe(image_data)
-        try:
+                if not tc and m.find("**ERROR**:") >= 0:
-            for resp in mdl.tts("Hello~ RAGFlower!"):
+                    raise Exception(m)
-                pass
+            except Exception as e:
-        except RuntimeError as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case LLMType.TTS.value:
-    elif llm["model_type"] == LLMType.OCR.value:
+            assert factory in TTSModel, f"TTS model from {factory} is not supported yet."
-        assert factory in OcrModel, f"OCR model from {factory} is not supported yet."
+            mdl = TTSModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
-        try:
+            try:
-            mdl = OcrModel[factory](key=llm["api_key"], model_name=mdl_nm, base_url=llm.get("api_base", ""))
+                for resp in mdl.tts("Hello~ RAGFlower!"):
-            ok, reason = mdl.check_available()
+                    pass
-            if not ok:
+            except RuntimeError as e:
-                raise RuntimeError(reason or "Model not available")
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
-        except Exception as e:
+        case LLMType.OCR.value:
-            msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+            assert factory in OcrModel, f"OCR model from {factory} is not supported yet."
-    else:
+            try:
-        # TODO: check other type of models
+                mdl = OcrModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
-        pass
+                ok, reason = mdl.check_available()
+                if not ok:
+                    raise RuntimeError(reason or "Model not available")
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case LLMType.SPEECH2TEXT:
+            assert factory in Seq2txtModel, f"Speech model from {factory} is not supported yet."
+            try:
+                mdl = Seq2txtModel[factory](key=model_api_key, model_name=mdl_nm, base_url=model_base_url)
+                # TODO: check the availability
+            except Exception as e:
+                msg += f"\nFail to access model({factory}/{mdl_nm})." + str(e)
+        case _:
+            raise RuntimeError(f"Unknown model type: {model_type}")
 
     if msg:
         return get_data_error_result(message=msg)

api/apps/mcp_server_app.py

@@ -326,7 +326,6 @@ async def list_tools() -> Response:
                 try:
                     tools = await asyncio.to_thread(tool_call_session.get_tools, timeout)
                 except Exception as e:
-                    tools = []
                     return get_data_error_result(message=f"MCP list tools error: {e}")
 
                 results[server_key] = []
@@ -428,7 +427,6 @@ async def test_mcp() -> Response:
         try:
             tools = await asyncio.to_thread(tool_call_session.get_tools, timeout)
         except Exception as e:
-            tools = []
             return get_data_error_result(message=f"Test MCP error: {e}")
         finally:
             # PERF: blocking call to close sessions — consider moving to background thread or task queue

api/apps/sdk/chat.py

@@ -287,7 +287,7 @@ def list_chat(tenant_id):
     chats = DialogService.get_list(tenant_id, page_number, items_per_page, orderby, desc, id, name)
     if not chats:
         return get_result(data=[])
-    list_assts = []
+    list_assistants = []
     key_mapping = {
         "parameters": "variables",
         "prologue": "opener",
@@ -321,5 +321,5 @@ def list_chat(tenant_id):
         del res["kb_ids"]
         res["datasets"] = kb_list
         res["avatar"] = res.pop("icon")
-        list_assts.append(res)
+        list_assistants.append(res)
-    return get_result(data=list_assts)
+    return get_result(data=list_assistants)

api/apps/sdk/files.py

@@ -205,7 +205,8 @@ async def create(tenant_id):
         if not FileService.is_parent_folder_exist(pf_id):
             return get_json_result(data=False, message="Parent Folder Doesn't Exist!", code=RetCode.BAD_REQUEST)
         if FileService.query(name=req["name"], parent_id=pf_id):
-            return get_json_result(data=False, message="Duplicated folder name in the same folder.", code=409)
+            return get_json_result(data=False, message="Duplicated folder name in the same folder.",
+                                   code=RetCode.CONFLICT)
 
         if input_file_type == FileType.FOLDER.value:
             file_type = FileType.FOLDER.value
@@ -565,11 +566,13 @@ async def rename(tenant_id):
 
         if file.type != FileType.FOLDER.value and pathlib.Path(req["name"].lower()).suffix != pathlib.Path(
                 file.name.lower()).suffix:
-            return get_json_result(data=False, message="The extension of file can't be changed", code=RetCode.BAD_REQUEST)
+            return get_json_result(data=False, message="The extension of file can't be changed",
+                                   code=RetCode.BAD_REQUEST)
 
         for existing_file in FileService.query(name=req["name"], pf_id=file.parent_id):
             if existing_file.name == req["name"]:
-                return get_json_result(data=False, message="Duplicated file name in the same folder.", code=409)
+                return get_json_result(data=False, message="Duplicated file name in the same folder.",
+                                       code=RetCode.CONFLICT)
 
         if not FileService.update_by_id(req["file_id"], {"name": req["name"]}):
             return get_json_result(message="Database error (File rename)!", code=RetCode.SERVER_ERROR)
@@ -631,9 +634,10 @@ async def get(tenant_id, file_id):
     except Exception as e:
         return server_error_response(e)
 
+
 @manager.route("/file/download/<attachment_id>", methods=["GET"])  # noqa: F821
 @token_required
-async def download_attachment(tenant_id,attachment_id):
+async def download_attachment(tenant_id, attachment_id):
     try:
         ext = request.args.get("ext", "markdown")
         data = await asyncio.to_thread(settings.STORAGE_IMPL.get, tenant_id, attachment_id)
@@ -645,6 +649,7 @@ async def download_attachment(tenant_id,attachment_id):
     except Exception as e:
         return server_error_response(e)
 
+
 @manager.route('/file/mv', methods=['POST'])  # noqa: F821
 @token_required
 async def move(tenant_id):

api/apps/sdk/session.py

@@ -448,7 +448,7 @@ async def chat_completion_openai_like(tenant_id, chat_id):
 @token_required
 async def agents_completion_openai_compatibility(tenant_id, agent_id):
     req = await get_request_json()
-    tiktokenenc = tiktoken.get_encoding("cl100k_base")
+    tiktoken_encode = tiktoken.get_encoding("cl100k_base")
     messages = req.get("messages", [])
     if not messages:
         return get_error_data_result("You must provide at least one message.")
@@ -456,7 +456,7 @@ async def agents_completion_openai_compatibility(tenant_id, agent_id):
         return get_error_data_result(f"You don't own the agent {agent_id}")
 
     filtered_messages = [m for m in messages if m["role"] in ["user", "assistant"]]
-    prompt_tokens = sum(len(tiktokenenc.encode(m["content"])) for m in filtered_messages)
+    prompt_tokens = sum(len(tiktoken_encode.encode(m["content"])) for m in filtered_messages)
     if not filtered_messages:
         return jsonify(
             get_data_openai(
@@ -464,7 +464,7 @@ async def agents_completion_openai_compatibility(tenant_id, agent_id):
                 content="No valid messages found (user or assistant).",
                 finish_reason="stop",
                 model=req.get("model", ""),
-                completion_tokens=len(tiktokenenc.encode("No valid messages found (user or assistant).")),
+                completion_tokens=len(tiktoken_encode.encode("No valid messages found (user or assistant).")),
                 prompt_tokens=prompt_tokens,
             )
         )
@@ -501,6 +501,8 @@ async def agents_completion_openai_compatibility(tenant_id, agent_id):
             ):
             return jsonify(response)
 
+        return None
+
 
 @manager.route("/agents/<agent_id>/completions", methods=["POST"])  # noqa: F821
 @token_required
@@ -920,6 +922,7 @@ async def chatbot_completions(dialog_id):
     async for answer in iframe_completion(dialog_id, **req):
         return get_result(data=answer)
 
+    return None
 
 @manager.route("/chatbots/<dialog_id>/info", methods=["GET"])  # noqa: F821
 async def chatbots_inputs(dialog_id):
@@ -967,6 +970,7 @@ async def agent_bot_completions(agent_id):
     async for answer in agent_completion(objs[0].tenant_id, agent_id, **req):
         return get_result(data=answer)
 
+    return None
 
 @manager.route("/agentbots/<agent_id>/inputs", methods=["GET"])  # noqa: F821
 async def begin_inputs(agent_id):

api/apps/user_app.py

@@ -660,7 +660,7 @@ def user_register(user_id, user):
     tenant_llm = get_init_tenant_llm(user_id)
 
     if not UserService.save(**user):
-        return
+        return None
     TenantService.insert(**tenant)
     UserTenantService.insert(**usr_tenant)
     TenantLLMService.insert_many(tenant_llm)

common/constants.py

@@ -54,6 +54,7 @@ class RetCode(IntEnum, CustomEnum):
     SERVER_ERROR = 500
     FORBIDDEN = 403
     NOT_FOUND = 404
+    CONFLICT = 409
 
 
 class StatusEnum(Enum):

common/data_source/blob_connector.py

@@ -64,15 +64,23 @@ class BlobStorageConnector(LoadConnector, PollConnector):
 
         elif self.bucket_type == BlobType.S3:
             authentication_method = credentials.get("authentication_method", "access_key")
+
             if authentication_method == "access_key":
                 if not all(
                     credentials.get(key)
                     for key in ["aws_access_key_id", "aws_secret_access_key"]
                 ):
                     raise ConnectorMissingCredentialError("Amazon S3")
+
             elif authentication_method == "iam_role":
                 if not credentials.get("aws_role_arn"):
                     raise ConnectorMissingCredentialError("Amazon S3 IAM role ARN is required")
+                
+            elif authentication_method == "assume_role":
+                pass
+
+            else:
+                raise ConnectorMissingCredentialError("Unsupported S3 authentication method")
 
         elif self.bucket_type == BlobType.GOOGLE_CLOUD_STORAGE:
             if not all(
@@ -293,4 +301,4 @@ if __name__ == "__main__":
     except ConnectorMissingCredentialError as e:
         print(f"Error: {e}")
     except Exception as e:
-        print(f"An unexpected error occurred: {e}")
+        print(f"An unexpected error occurred: {e}")

common/data_source/utils.py

@@ -254,18 +254,21 @@ def create_s3_client(bucket_type: BlobType, credentials: dict[str, Any], europea
     elif bucket_type == BlobType.S3:
         authentication_method = credentials.get("authentication_method", "access_key")
 
+        region_name = credentials.get("region") or None
+
         if authentication_method == "access_key":
             session = boto3.Session(
                 aws_access_key_id=credentials["aws_access_key_id"],
                 aws_secret_access_key=credentials["aws_secret_access_key"],
+                region_name=region_name,
             )
-            return session.client("s3")
+            return session.client("s3", region_name=region_name)
 
         elif authentication_method == "iam_role":
             role_arn = credentials["aws_role_arn"]
 
             def _refresh_credentials() -> dict[str, str]:
-                sts_client = boto3.client("sts")
+                sts_client = boto3.client("sts", region_name=credentials.get("region") or None)
                 assumed_role_object = sts_client.assume_role(
                     RoleArn=role_arn,
                     RoleSessionName=f"onyx_blob_storage_{int(datetime.now().timestamp())}",
@@ -285,11 +288,11 @@ def create_s3_client(bucket_type: BlobType, credentials: dict[str, Any], europea
             )
             botocore_session = get_session()
             botocore_session._credentials = refreshable
-            session = boto3.Session(botocore_session=botocore_session)
+            session = boto3.Session(botocore_session=botocore_session, region_name=region_name)
-            return session.client("s3")
+            return session.client("s3", region_name=region_name)
 
         elif authentication_method == "assume_role":
-            return boto3.client("s3")
+            return boto3.client("s3", region_name=region_name)
 
         else:
             raise ValueError("Invalid authentication method for S3.")

common/log_utils.py

@@ -75,9 +75,12 @@ def init_root_logger(logfile_basename: str, log_format: str = "%(asctime)-15s %(
 def log_exception(e, *args):
     logging.exception(e)
     for a in args:
-        if hasattr(a, "text"):
+        try:
-            logging.error(a.text)
+            text = getattr(a, "text")
-            raise Exception(a.text)
+        except Exception:
-        else:
+            text = None
-            logging.error(str(a))
+        if text is not None:
+            logging.error(text)
+            raise Exception(text)
+        logging.error(str(a))
     raise e

common/token_utils.py

@@ -44,23 +44,23 @@ def total_token_count_from_response(resp):
     if resp is None:
         return 0
 
-    if hasattr(resp, "usage") and hasattr(resp.usage, "total_tokens"):
+    try:
-        try:
+        if hasattr(resp, "usage") and hasattr(resp.usage, "total_tokens"):
             return resp.usage.total_tokens
-        except Exception:
+    except Exception:
-            pass
+        pass
 
-    if hasattr(resp, "usage_metadata") and hasattr(resp.usage_metadata, "total_tokens"):
+    try:
-        try:
+        if hasattr(resp, "usage_metadata") and hasattr(resp.usage_metadata, "total_tokens"):
             return resp.usage_metadata.total_tokens
-        except Exception:
+    except Exception:
-            pass
+        pass
 
-    if hasattr(resp, "meta") and hasattr(resp.meta, "billed_units") and hasattr(resp.meta.billed_units, "input_tokens"):
+    try:
-        try:
+        if hasattr(resp, "meta") and hasattr(resp.meta, "billed_units") and hasattr(resp.meta.billed_units, "input_tokens"):
-            return resp.meta.billed_units.input_tokens 
+            return resp.meta.billed_units.input_tokens
-        except Exception:
+    except Exception:
-            pass
+        pass
 
     if isinstance(resp, dict) and 'usage' in resp and 'total_tokens' in resp['usage']:
         try:
@@ -85,4 +85,3 @@ def total_token_count_from_response(resp):
 def truncate(string: str, max_len: int) -> str:
     """Returns truncated text if the length of text exceed max_len."""
     return encoder.decode(encoder.encode(string)[:max_len])
-

web/src/components/chunk-method-dialog/index.tsx

@@ -21,6 +21,7 @@ import { IChangeParserConfigRequestBody } from '@/interfaces/request/document';
 import {
   ChunkMethodItem,
   EnableTocToggle,
+  ImageContextWindow,
   ParseTypeItem,
 } from '@/pages/dataset/dataset-setting/configuration/common-item';
 import { zodResolver } from '@hookform/resolvers/zod';
@@ -119,6 +120,7 @@ export function ChunkMethodDialog({
         auto_questions: z.coerce.number().optional(),
         html4excel: z.boolean().optional(),
         toc_extraction: z.boolean().optional(),
+        image_context_window: z.coerce.number().optional(),
         mineru_parse_method: z.enum(['auto', 'txt', 'ocr']).optional(),
         mineru_formula_enable: z.boolean().optional(),
         mineru_table_enable: z.boolean().optional(),
@@ -364,7 +366,10 @@ export function ChunkMethodDialog({
                   className="space-y-3"
                 >
                   {selectedTag === DocumentParserType.Naive && (
-                    <EnableTocToggle />
+                    <>
+                      <EnableTocToggle />
+                      <ImageContextWindow />
+                    </>
                   )}
                   {showAutoKeywords(selectedTag) && (
                     <>

web/src/components/chunk-method-dialog/use-default-parser-values.ts

@@ -18,6 +18,7 @@ export function useDefaultParserValues() {
       auto_questions: 0,
       html4excel: false,
       toc_extraction: false,
+      image_context_window: 0,
       mineru_parse_method: 'auto',
       mineru_formula_enable: true,
       mineru_table_enable: true,

web/src/components/file-uploader.tsx

@@ -145,6 +145,8 @@ interface FileUploaderProps
    */
   maxFileCount?: DropzoneProps['maxFiles'];
 
+  hideDropzoneOnMaxFileCount?: boolean;
+
   /**
    * Whether the uploader should accept multiple files.
    * @type boolean
@@ -178,6 +180,7 @@ export function FileUploader(props: FileUploaderProps) {
     maxFileCount = 100000000000,
     multiple = false,
     disabled = false,
+    hideDropzoneOnMaxFileCount = false,
     className,
     title,
     description,
@@ -189,6 +192,8 @@ export function FileUploader(props: FileUploaderProps) {
     onChange: onValueChange,
   });
 
+  const reachesMaxFileCount = (files?.length ?? 0) >= maxFileCount;
+
   const onDrop = React.useCallback(
     (acceptedFiles: File[], rejectedFiles: FileRejection[]) => {
       if (!multiple && maxFileCount === 1 && acceptedFiles.length > 1) {
@@ -263,65 +268,68 @@ export function FileUploader(props: FileUploaderProps) {
 
   return (
     <div className="relative flex flex-col gap-6 overflow-hidden">
-      <Dropzone
+      {!(hideDropzoneOnMaxFileCount && reachesMaxFileCount) && (
-        onDrop={onDrop}
+        <Dropzone
-        accept={accept}
+          onDrop={onDrop}
-        maxSize={maxSize}
+          accept={accept}
-        maxFiles={maxFileCount}
+          maxSize={maxSize}
-        multiple={maxFileCount > 1 || multiple}
+          maxFiles={maxFileCount}
-        disabled={isDisabled}
+          multiple={maxFileCount > 1 || multiple}
-      >
+          disabled={isDisabled}
-        {({ getRootProps, getInputProps, isDragActive }) => (
+        >
-          <div
+          {({ getRootProps, getInputProps, isDragActive }) => (
-            {...getRootProps()}
+            <div
-            className={cn(
+              {...getRootProps()}
-              'group relative grid h-72 w-full cursor-pointer place-items-center rounded-lg border border-dashed border-border-default px-5 py-2.5 text-center transition hover:bg-border-button bg-bg-card',
+              className={cn(
-              'ring-offset-background focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2',
+                'group relative grid h-72 w-full cursor-pointer place-items-center rounded-lg border border-dashed border-border-default px-5 py-2.5 text-center transition hover:bg-border-button bg-bg-card',
-              isDragActive && 'border-border-button',
+                'ring-offset-background focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2',
-              isDisabled && 'pointer-events-none opacity-60',
+                isDragActive && 'border-border-button',
-              className,
+                isDisabled && 'pointer-events-none opacity-60',
-            )}
+                className,
-            {...dropzoneProps}
+              )}
-          >
+              {...dropzoneProps}
-            <input {...getInputProps()} />
+            >
-            {isDragActive ? (
+              <input {...getInputProps()} />
-              <div className="flex flex-col items-center justify-center gap-4 sm:px-5">
+              {isDragActive ? (
-                <div className="rounded-full border border-dashed p-3">
+                <div className="flex flex-col items-center justify-center gap-4 sm:px-5">
-                  <Upload
+                  <div className="rounded-full border border-dashed p-3">
-                    className="size-7 text-text-secondary transition-colors group-hover:text-text-primary"
+                    <Upload
-                    aria-hidden="true"
+                      className="size-7 text-text-secondary transition-colors group-hover:text-text-primary"
-                  />
+                      aria-hidden="true"
-                </div>
+                    />
-                <p className="font-medium text-text-secondary">
+                  </div>
-                  Drop the files here
-                </p>
-              </div>
-            ) : (
-              <div className="flex flex-col items-center justify-center gap-4 sm:px-5">
-                <div className="rounded-full border border-dashed p-3">
-                  <Upload
-                    className="size-7 text-text-secondary transition-colors group-hover:text-text-primary"
-                    aria-hidden="true"
-                  />
-                </div>
-                <div className="flex flex-col gap-px">
                   <p className="font-medium text-text-secondary">
-                    {title || t('knowledgeDetails.uploadTitle')}
+                    Drop the files here
-                  </p>
-                  <p className="text-sm text-text-disabled">
-                    {description || t('knowledgeDetails.uploadDescription')}
-                    {/* You can upload
-                    {maxFileCount > 1
-                      ? ` ${maxFileCount === Infinity ? 'multiple' : maxFileCount}
-                      files (up to ${formatBytes(maxSize)} each)`
-                      : ` a file with ${formatBytes(maxSize)}`} */}
                   </p>
                 </div>
-              </div>
+              ) : (
-            )}
+                <div className="flex flex-col items-center justify-center gap-4 sm:px-5">
-          </div>
+                  <div className="rounded-full border border-dashed p-3">
-        )}
+                    <Upload
-      </Dropzone>
+                      className="size-7 text-text-secondary transition-colors group-hover:text-text-primary"
+                      aria-hidden="true"
+                    />
+                  </div>
+                  <div className="flex flex-col gap-px">
+                    <p className="font-medium text-text-secondary">
+                      {title || t('knowledgeDetails.uploadTitle')}
+                    </p>
+                    <p className="text-sm text-text-disabled">
+                      {description || t('knowledgeDetails.uploadDescription')}
+                      {/* You can upload
+                      {maxFileCount > 1
+                        ? ` ${maxFileCount === Infinity ? 'multiple' : maxFileCount}
+                        files (up to ${formatBytes(maxSize)} each)`
+                        : ` a file with ${formatBytes(maxSize)}`} */}
+                    </p>
+                  </div>
+                </div>
+              )}
+            </div>
+          )}
+        </Dropzone>
+      )}
+
       {files?.length ? (
         <div className="h-fit w-full px-3">
           <div className="flex max-h-48 flex-col gap-4 overflow-auto scrollbar-auto">

web/src/components/image/index.tsx

@@ -1,11 +1,10 @@
 import { api_host } from '@/utils/api';
 import classNames from 'classnames';
+import React from 'react';
 import { Popover, PopoverContent, PopoverTrigger } from '../ui/popover';
 
-interface IImage {
+interface IImage extends React.ImgHTMLAttributes<HTMLImageElement> {
   id: string;
-  className?: string;
-  onClick?(): void;
   t?: string | number;
 }
 

web/src/components/list-filter-bar/filter-field.tsx

@@ -0,0 +1,170 @@
+import { Checkbox } from '@/components/ui/checkbox';
+import { ChevronDown, ChevronUp } from 'lucide-react';
+import { memo, useState } from 'react';
+import { ControllerRenderProps, useFormContext } from 'react-hook-form';
+import { FormControl, FormField, FormItem, FormLabel } from '../ui/form';
+import { FilterType } from './interface';
+const handleCheckChange = ({
+  checked,
+  field,
+  item,
+  isNestedField = false,
+  parentId = '',
+}: {
+  checked: boolean;
+  field: ControllerRenderProps<
+    { [x: string]: { [x: string]: any } | string[] },
+    string
+  >;
+  item: FilterType;
+  isNestedField?: boolean;
+  parentId?: string;
+}) => {
+  if (isNestedField && parentId) {
+    const currentValue = field.value || {};
+    const currentParentValues =
+      (currentValue as Record<string, string[]>)[parentId] || [];
+
+    const newParentValues = checked
+      ? [...currentParentValues, item.id.toString()]
+      : currentParentValues.filter(
+          (value: string) => value !== item.id.toString(),
+        );
+
+    const newValue = {
+      ...currentValue,
+      [parentId]: newParentValues,
+    };
+
+    if (newValue[parentId].length === 0) {
+      delete newValue[parentId];
+    }
+
+    return field.onChange(newValue);
+  } else {
+    const list = checked
+      ? [...(Array.isArray(field.value) ? field.value : []), item.id.toString()]
+      : (Array.isArray(field.value) ? field.value : []).filter(
+          (value) => value !== item.id.toString(),
+        );
+    return field.onChange(list);
+  }
+};
+
+const FilterItem = memo(
+  ({
+    item,
+    field,
+    level = 0,
+  }: {
+    item: FilterType;
+    field: ControllerRenderProps<
+      { [x: string]: { [x: string]: any } | string[] },
+      string
+    >;
+    level: number;
+  }) => {
+    return (
+      <div
+        className={`flex items-center justify-between text-text-primary text-xs ${level > 0 ? 'ml-4' : ''}`}
+      >
+        <FormItem className="flex flex-row space-x-3 space-y-0 items-center">
+          <FormControl>
+            <Checkbox
+              checked={field.value?.includes(item.id.toString())}
+              onCheckedChange={(checked: boolean) =>
+                handleCheckChange({ checked, field, item })
+              }
+            />
+          </FormControl>
+          <FormLabel onClick={(e) => e.stopPropagation()}>
+            {item.label}
+          </FormLabel>
+        </FormItem>
+        {item.count !== undefined && (
+          <span className="text-sm">{item.count}</span>
+        )}
+      </div>
+    );
+  },
+);
+
+export const FilterField = memo(
+  ({
+    item,
+    parent,
+    level = 0,
+  }: {
+    item: FilterType;
+    parent: FilterType;
+    level?: number;
+  }) => {
+    const form = useFormContext();
+    const [showAll, setShowAll] = useState(false);
+    const hasNestedList = item.list && item.list.length > 0;
+
+    return (
+      <FormField
+        key={item.id}
+        control={form.control}
+        name={parent.field as string}
+        render={({ field }) => {
+          if (hasNestedList) {
+            return (
+              <div className={`flex flex-col gap-2 ${level > 0 ? 'ml-4' : ''}`}>
+                <div
+                  className="flex items-center justify-between cursor-pointer"
+                  onClick={() => {
+                    setShowAll(!showAll);
+                  }}
+                >
+                  <FormLabel className="text-text-primary">
+                    {item.label}
+                  </FormLabel>
+                  {showAll ? (
+                    <ChevronUp size={12} />
+                  ) : (
+                    <ChevronDown size={12} />
+                  )}
+                </div>
+                {showAll &&
+                  item.list?.map((child) => (
+                    <FilterField
+                      key={child.id}
+                      item={child}
+                      parent={{
+                        ...item,
+                        field: `${parent.field}.${item.field}`,
+                      }}
+                      level={level + 1}
+                    />
+                    // <FilterItem key={child.id} item={child} field={child.field} level={level+1} />
+                    // <div
+                    //   className="flex flex-row space-x-3 space-y-0 items-center"
+                    //   key={child.id}
+                    // >
+                    //   <FormControl>
+                    //     <Checkbox
+                    //       checked={field.value?.includes(child.id.toString())}
+                    //       onCheckedChange={(checked) =>
+                    //         handleCheckChange({ checked, field, item: child })
+                    //       }
+                    //     />
+                    //   </FormControl>
+                    //   <FormLabel onClick={(e) => e.stopPropagation()}>
+                    //     {child.label}
+                    //   </FormLabel>
+                    // </div>
+                  ))}
+              </div>
+            );
+          }
+
+          return <FilterItem item={item} field={field} level={level} />;
+        }}
+      />
+    );
+  },
+);
+
+FilterField.displayName = 'FilterField';

web/src/components/list-filter-bar/filter-popover.tsx

@@ -4,21 +4,27 @@ import {
   PopoverTrigger,
 } from '@/components/ui/popover';
 import { zodResolver } from '@hookform/resolvers/zod';
-import { PropsWithChildren, useCallback, useEffect, useState } from 'react';
+import {
+  PropsWithChildren,
+  useCallback,
+  useEffect,
+  useMemo,
+  useState,
+} from 'react';
 import { useForm } from 'react-hook-form';
 import { ZodArray, ZodString, z } from 'zod';
 
 import { Button } from '@/components/ui/button';
-import { Checkbox } from '@/components/ui/checkbox';
+
 import {
   Form,
-  FormControl,
   FormField,
   FormItem,
   FormLabel,
   FormMessage,
 } from '@/components/ui/form';
 import { t } from 'i18next';
+import { FilterField } from './filter-field';
 import { FilterChange, FilterCollection, FilterValue } from './interface';
 
 export type CheckboxFormMultipleProps = {
@@ -35,29 +41,71 @@ function CheckboxFormMultiple({
   onChange,
   setOpen,
 }: CheckboxFormMultipleProps) {
-  const fieldsDict = filters?.reduce<Record<string, Array<any>>>((pre, cur) => {
+  const [resolvedFilters, setResolvedFilters] =
-    pre[cur.field] = [];
+    useState<FilterCollection[]>(filters);
-    return pre;
-  }, {});
 
-  const FormSchema = z.object(
+  useEffect(() => {
-    filters.reduce<Record<string, ZodArray<ZodString, 'many'>>>((pre, cur) => {
+    if (filters && filters.length > 0) {
-      pre[cur.field] = z.array(z.string());
+      setResolvedFilters(filters);
+    }
+  }, [filters]);
 
-      // .refine((value) => value.some((item) => item), {
+  const fieldsDict = useMemo(() => {
-      //   message: 'You have to select at least one item.',
+    if (resolvedFilters.length === 0) {
-      // });
+      return {};
+    }
+
+    return resolvedFilters.reduce<Record<string, any>>((pre, cur) => {
+      const hasNested = cur.list?.some(
+        (item) => item.list && item.list.length > 0,
+      );
+
+      if (hasNested) {
+        pre[cur.field] = {};
+      } else {
+        pre[cur.field] = [];
+      }
       return pre;
-    }, {}),
+    }, {});
-  );
+  }, [resolvedFilters]);
+
+  const FormSchema = useMemo(() => {
+    if (resolvedFilters.length === 0) {
+      return z.object({});
+    }
+
+    return z.object(
+      resolvedFilters.reduce<
+        Record<
+          string,
+          ZodArray<ZodString, 'many'> | z.ZodObject<any> | z.ZodOptional<any>
+        >
+      >((pre, cur) => {
+        const hasNested = cur.list?.some(
+          (item) => item.list && item.list.length > 0,
+        );
+
+        if (hasNested) {
+          pre[cur.field] = z
+            .record(z.string(), z.array(z.string().optional()).optional())
+            .optional();
+        } else {
+          pre[cur.field] = z.array(z.string().optional()).optional();
+        }
+
+        return pre;
+      }, {}),
+    );
+  }, [resolvedFilters]);
 
   const form = useForm<z.infer<typeof FormSchema>>({
-    resolver: zodResolver(FormSchema),
+    resolver: resolvedFilters.length > 0 ? zodResolver(FormSchema) : undefined,
     defaultValues: fieldsDict,
   });
 
-  function onSubmit(data: z.infer<typeof FormSchema>) {
+  function onSubmit() {
-    onChange?.(data);
+    const formValues = form.getValues();
+    onChange?.({ ...formValues });
     setOpen(false);
   }
 
@@ -67,8 +115,10 @@ function CheckboxFormMultiple({
   }, [fieldsDict, onChange, setOpen]);
 
   useEffect(() => {
-    form.reset(value);
+    if (resolvedFilters.length > 0) {
-  }, [form, value]);
+      form.reset(value || fieldsDict);
+    }
+  }, [form, value, resolvedFilters, fieldsDict]);
 
   return (
     <Form {...form}>
@@ -85,44 +135,21 @@ function CheckboxFormMultiple({
             render={() => (
               <FormItem className="space-y-4">
                 <div>
-                  <FormLabel className="text-base text-text-sub-title-invert">
+                  <FormLabel className="text-text-primary">{x.label}</FormLabel>
-                    {x.label}
-                  </FormLabel>
                 </div>
-                {x.list.map((item) => (
+                {x.list.map((item) => {
-                  <FormField
+                  return (
-                    key={item.id}
+                    <FilterField
-                    control={form.control}
+                      key={item.id}
-                    name={x.field}
+                      item={{ ...item }}
-                    render={({ field }) => {
+                      parent={{
-                      return (
+                        ...x,
-                        <div className="flex items-center justify-between text-text-primary text-xs">
+                        id: x.field,
-                          <FormItem
+                        // field: `${x.field}${item.field ? '.' + item.field : ''}`,
-                            key={item.id}
+                      }}
-                            className="flex flex-row  space-x-3 space-y-0 items-center "
+                    />
-                          >
+                  );
-                            <FormControl>
+                })}
-                              <Checkbox
-                                checked={field.value?.includes(item.id)}
-                                onCheckedChange={(checked) => {
-                                  return checked
-                                    ? field.onChange([...field.value, item.id])
-                                    : field.onChange(
-                                        field.value?.filter(
-                                          (value) => value !== item.id,
-                                        ),
-                                      );
-                                }}
-                              />
-                            </FormControl>
-                            <FormLabel>{item.label}</FormLabel>
-                          </FormItem>
-                          <span className=" text-sm">{item.count}</span>
-                        </div>
-                      );
-                    }}
-                  />
-                ))}
                 <FormMessage />
               </FormItem>
             )}
@@ -137,7 +164,13 @@ function CheckboxFormMultiple({
           >
             {t('common.clear')}
           </Button>
-          <Button type="submit" size={'sm'}>
+          <Button
+            type="submit"
+            onClick={() => {
+              console.log(form.formState.errors, form.getValues());
+            }}
+            size={'sm'}
+          >
             {t('common.submit')}
           </Button>
         </div>

web/src/components/list-filter-bar/index.tsx

@@ -17,6 +17,7 @@ interface IProps {
   onSearchChange?: ChangeEventHandler<HTMLInputElement>;
   showFilter?: boolean;
   leftPanel?: ReactNode;
+  preChildren?: ReactNode;
 }
 
 export const FilterButton = React.forwardRef<
@@ -46,6 +47,7 @@ export const FilterButton = React.forwardRef<
 export default function ListFilterBar({
   title,
   children,
+  preChildren,
   searchString,
   onSearchChange,
   showFilter = true,
@@ -63,7 +65,18 @@ export default function ListFilterBar({
   const filterCount = useMemo(() => {
     return typeof value === 'object' && value !== null
       ? Object.values(value).reduce((pre, cur) => {
-          return pre + cur.length;
+          if (Array.isArray(cur)) {
+            return pre + cur.length;
+          }
+          if (typeof cur === 'object') {
+            return (
+              pre +
+              Object.values(cur).reduce((pre, cur) => {
+                return pre + cur.length;
+              }, 0)
+            );
+          }
+          return pre;
         }, 0)
       : 0;
   }, [value]);
@@ -80,6 +93,7 @@ export default function ListFilterBar({
         {leftPanel || title}
       </div>
       <div className="flex gap-5 items-center">
+        {preChildren}
         {showFilter && (
           <FilterPopover
             value={value}

web/src/components/list-filter-bar/interface.ts

@@ -1,6 +1,9 @@
 export type FilterType = {
   id: string;
+  field?: string;
   label: string | JSX.Element;
+  list?: FilterType[];
+  value?: string | string[];
   count?: number;
 };
 
@@ -10,6 +13,9 @@ export type FilterCollection = {
   list: FilterType[];
 };
 
-export type FilterValue = Record<string, Array<string>>;
+export type FilterValue = Record<
+  string,
+  Array<string> | Record<string, Array<string>>
+>;
 
 export type FilterChange = (value: FilterValue) => void;

web/src/components/next-markdown-content/index.tsx

@@ -5,7 +5,6 @@ import { getExtension } from '@/utils/document-util';
 import DOMPurify from 'dompurify';
 import { memo, useCallback, useEffect, useMemo } from 'react';
 import Markdown from 'react-markdown';
-import reactStringReplace from 'react-string-replace';
 import SyntaxHighlighter from 'react-syntax-highlighter';
 import rehypeKatex from 'rehype-katex';
 import rehypeRaw from 'rehype-raw';
@@ -18,7 +17,6 @@ import { useTranslation } from 'react-i18next';
 import 'katex/dist/katex.min.css'; // `rehype-katex` does not import the CSS for you
 
 import {
-  currentReg,
   preprocessLaTeX,
   replaceTextByOldReg,
   replaceThinkToSection,
@@ -31,6 +29,11 @@ import classNames from 'classnames';
 import { omit } from 'lodash';
 import { pipe } from 'lodash/fp';
 import { CircleAlert } from 'lucide-react';
+import { ImageCarousel } from '../markdown-content/image-carousel';
+import {
+  groupConsecutiveReferences,
+  shouldShowCarousel,
+} from '../markdown-content/reference-utils';
 import { Button } from '../ui/button';
 import {
   HoverCard,
@@ -39,6 +42,19 @@ import {
 } from '../ui/hover-card';
 import styles from './index.less';
 
+// Helper function to convert IReferenceObject to IReference
+const convertReferenceObjectToReference = (
+  referenceObject: IReferenceObject,
+) => {
+  const chunks = Object.values(referenceObject.chunks);
+  const docAggs = Object.values(referenceObject.doc_aggs);
+  return {
+    chunks,
+    doc_aggs: docAggs,
+    total: chunks.length,
+  };
+};
+
 const getChunkIndex = (match: string) => Number(match);
 // TODO: The display of the table is inconsistent with the display previously placed in the MessageItem.
 function MarkdownContent({
@@ -211,47 +227,95 @@ function MarkdownContent({
 
   const renderReference = useCallback(
     (text: string) => {
-      let replacedText = reactStringReplace(text, currentReg, (match, i) => {
+      const groups = groupConsecutiveReferences(text);
-        const chunkIndex = getChunkIndex(match);
+      const elements = [];
+      let lastIndex = 0;
 
-        const { documentUrl, fileExtension, imageId, chunkItem, documentId } =
+      const convertedReference = reference
-          getReferenceInfo(chunkIndex);
+        ? convertReferenceObjectToReference(reference)
+        : null;
 
-        const docType = chunkItem?.doc_type;
+      groups.forEach((group, groupIndex) => {
+        if (group[0].start > lastIndex) {
+          elements.push(text.substring(lastIndex, group[0].start));
+        }
 
-        return showImage(docType) ? (
+        if (
-          <section>
+          convertedReference &&
-            <Image
+          shouldShowCarousel(group, convertedReference)
-              id={imageId}
+        ) {
-              className={styles.referenceInnerChunkImage}
+          elements.push(
-              onClick={
+            <ImageCarousel
-                documentId
+              key={`carousel-${groupIndex}`}
-                  ? handleDocumentButtonClick(
+              group={group}
-                      documentId,
+              reference={convertedReference}
-                      chunkItem,
+              fileThumbnails={fileThumbnails}
-                      fileExtension === 'pdf',
+              onImageClick={handleDocumentButtonClick}
-                      documentUrl,
+            />,
-                    )
+          );
-                  : () => {}
+        } else {
-              }
+          group.forEach((ref) => {
-            ></Image>
+            const chunkIndex = getChunkIndex(ref.id);
-            <span className="text-accent-primary">{imageId}</span>
+            const {
-          </section>
+              documentUrl,
-        ) : (
+              fileExtension,
-          <HoverCard key={i}>
+              imageId,
-            <HoverCardTrigger>
+              chunkItem,
-              <CircleAlert className="size-4 inline-block" />
+              documentId,
-            </HoverCardTrigger>
+            } = getReferenceInfo(chunkIndex);
-            <HoverCardContent className="max-w-3xl">
+            const docType = chunkItem?.doc_type;
-              {renderPopoverContent(chunkIndex)}
+
-            </HoverCardContent>
+            if (showImage(docType)) {
-          </HoverCard>
+              elements.push(
-        );
+                <section key={ref.id}>
+                  <Image
+                    id={imageId}
+                    className={styles.referenceInnerChunkImage}
+                    onClick={
+                      documentId
+                        ? handleDocumentButtonClick(
+                            documentId,
+                            chunkItem,
+                            fileExtension === 'pdf',
+                            documentUrl,
+                          )
+                        : () => {}
+                    }
+                  />
+                  <span className="text-accent-primary"> {imageId}</span>
+                </section>,
+              );
+            } else {
+              elements.push(
+                <HoverCard key={ref.id}>
+                  <HoverCardTrigger>
+                    <CircleAlert className="size-4 inline-block" />
+                  </HoverCardTrigger>
+                  <HoverCardContent className="max-w-3xl">
+                    {renderPopoverContent(chunkIndex)}
+                  </HoverCardContent>
+                </HoverCard>,
+              );
+            }
+          });
+        }
+
+        lastIndex = group[group.length - 1].end;
       });
 
-      return replacedText;
+      if (lastIndex < text.length) {
+        elements.push(text.substring(lastIndex));
+      }
+
+      return elements;
     },
-    [renderPopoverContent, getReferenceInfo, handleDocumentButtonClick],
+    [
+      renderPopoverContent,
+      getReferenceInfo,
+      handleDocumentButtonClick,
+      reference,
+      fileThumbnails,
+    ],
   );
 
   return (

web/src/components/webhook-response-status.tsx

@@ -0,0 +1,19 @@
+import { useTranslation } from 'react-i18next';
+import { RAGFlowFormItem } from './ragflow-form';
+import { Input } from './ui/input';
+
+type WebHookResponseStatusFormFieldProps = {
+  name: string;
+};
+
+export function WebHookResponseStatusFormField({
+  name,
+}: WebHookResponseStatusFormFieldProps) {
+  const { t } = useTranslation();
+
+  return (
+    <RAGFlowFormItem name={name} label={t('flow.webhook.status')}>
+      <Input type="number"></Input>
+    </RAGFlowFormItem>
+  );
+}

web/src/hooks/use-document-request.ts

@@ -124,6 +124,7 @@ export const useFetchDocumentList = () => {
         {
           suffix: filterValue.type,
           run_status: filterValue.run,
+          metadata: filterValue.metadata,
         },
       );
       if (ret.data.code === 0) {
@@ -196,6 +197,7 @@ export const useGetDocumentFilter = (): {
     filter: data?.filter || {
       run_status: {},
       suffix: {},
+      metadata: {},
     },
     onOpenChange: handleOnpenChange,
   };

web/src/interfaces/database/document.ts

@@ -60,4 +60,5 @@ interface GraphRag {
 export type IDocumentInfoFilter = {
   run_status: Record<number, number>;
   suffix: Record<string, number>;
+  metadata: Record<string, Record<string, number>>;
 };

web/src/interfaces/database/knowledge.ts

@@ -116,12 +116,15 @@ export interface ITenantInfo {
   tts_id: string;
 }
 
+export type ChunkDocType = 'image' | 'table';
+
 export interface IChunk {
   available_int: number; // Whether to enable, 0: not enabled, 1: enabled
   chunk_id: string;
   content_with_weight: string;
   doc_id: string;
   doc_name: string;
+  doc_type_kwd?: ChunkDocType;
   image_id: string;
   important_kwd?: string[];
   question_kwd?: string[]; // keywords

web/src/locales/en.ts

@@ -364,6 +364,9 @@ Procedural Memory: Learned skills, habits, and automated procedures.`,
     },
     knowledgeConfiguration: {
       settings: 'Settings',
+      imageContextWindow: 'Image context window',
+      imageContextWindowTip:
+        'Captures N tokens of text above and below the image to provide richer background context for the image chunk.',
       autoMetadata: 'Auto metadata',
       mineruOptions: 'MinerU Options',
       mineruParseMethod: 'Parse Method',
@@ -604,6 +607,12 @@ This auto-tagging feature enhances retrieval by adding another layer of domain-s
         'The document being parsed cannot be deleted',
     },
     chunk: {
+      type: 'Type',
+      docType: {
+        image: 'Image',
+        table: 'Table',
+        text: 'Text',
+      },
       chunk: 'Chunk',
       bulk: 'Bulk',
       selectAll: 'Select all',
@@ -2084,12 +2093,14 @@ Important structured information may include: names, dates, locations, events, k
         schema: 'Schema',
         response: 'Response',
         executionMode: 'Execution mode',
+        executionModeTip:
+          'Accepted Response: The system returns an acknowledgment immediately after the request is validated, while the workflow continues to execute asynchronously in the background. /Final Response: The system returns a response only after the workflow execution is completed.',
         authMethods: 'Authentication methods',
         authType: 'Authentication type',
         limit: 'Request limit',
         per: 'Time period',
         maxBodySize: 'Maximum body size',
-        ipWhitelist: 'Ip whitelist',
+        ipWhitelist: 'IP whitelist',
         tokenHeader: 'Token header',
         tokenValue: 'Token value',
         username: 'Username',
@@ -2109,6 +2120,8 @@ Important structured information may include: names, dates, locations, events, k
         queryParameters: 'Query parameters',
         headerParameters: 'Header parameters',
         requestBodyParameters: 'Request body parameters',
+        streaming: 'Accepted response',
+        immediately: 'Final response',
       },
     },
     llmTools: {

web/src/locales/zh.ts

@@ -1835,6 +1835,8 @@ Tokenizer 会根据所选方式将内容存储为对应的数据结构。`,
         schema: '模式',
         response: '响应',
         executionMode: '执行模式',
+        executionModeTip:
+          'Accepted Response：请求校验通过后立即返回接收成功响应，工作流在后台异步执行；Final Response：系统在工作流执行完成后返回最终处理结果',
         authMethods: '认证方法',
         authType: '认证类型',
         limit: '请求限制',

web/src/pages/agent/form/begin-form/webhook/response.tsx

@@ -1,8 +1,8 @@
 import { Collapse } from '@/components/collapse';
 import { SelectWithSearch } from '@/components/originui/select-with-search';
 import { RAGFlowFormItem } from '@/components/ragflow-form';
-import { Input } from '@/components/ui/input';
 import { Textarea } from '@/components/ui/textarea';
+import { WebHookResponseStatusFormField } from '@/components/webhook-response-status';
 import { WebhookExecutionMode } from '@/pages/agent/constant';
 import { buildOptions } from '@/utils/form';
 import { useFormContext, useWatch } from 'react-hook-form';
@@ -24,19 +24,17 @@ export function WebhookResponse() {
         <RAGFlowFormItem
           name="execution_mode"
           label={t('flow.webhook.executionMode')}
+          tooltip={t('flow.webhook.executionModeTip')}
         >
           <SelectWithSearch
-            options={buildOptions(WebhookExecutionMode)}
+            options={buildOptions(WebhookExecutionMode, t, 'flow.webhook')}
           ></SelectWithSearch>
         </RAGFlowFormItem>
         {executionMode === WebhookExecutionMode.Immediately && (
           <>
-            <RAGFlowFormItem
+            <WebHookResponseStatusFormField
               name={'response.status'}
-              label={t('flow.webhook.status')}
+            ></WebHookResponseStatusFormField>
-            >
-              <Input type="number"></Input>
-            </RAGFlowFormItem>
             {/* <DynamicResponse
               name="response.headers_template"
               label={t('flow.webhook.headersTemplate')}

web/src/pages/agent/form/message-form/index.tsx

@@ -10,6 +10,7 @@ import {
 } from '@/components/ui/form';
 import { RAGFlowSelect } from '@/components/ui/select';
 import { Switch } from '@/components/ui/switch';
+import { WebHookResponseStatusFormField } from '@/components/webhook-response-status';
 import { zodResolver } from '@hookform/resolvers/zod';
 import { X } from 'lucide-react';
 import { memo } from 'react';
@@ -20,6 +21,7 @@ import { ExportFileType } from '../../constant';
 import { INextOperatorForm } from '../../interface';
 import { FormWrapper } from '../components/form-wrapper';
 import { PromptEditor } from '../components/prompt-editor';
+import { useShowWebhookResponseStatus } from './use-show-response-status';
 import { useValues } from './use-values';
 import { useWatchFormChange } from './use-watch-change';
 
@@ -38,6 +40,7 @@ function MessageForm({ node }: INextOperatorForm) {
       .optional(),
     output_format: z.string().optional(),
     auto_play: z.boolean().optional(),
+    status: z.number().optional(),
   });
 
   const form = useForm({
@@ -56,9 +59,14 @@ function MessageForm({ node }: INextOperatorForm) {
     control: form.control,
   });
 
+  const showWebhookResponseStatus = useShowWebhookResponseStatus(form);
+
   return (
     <Form {...form}>
       <FormWrapper>
+        {showWebhookResponseStatus && (
+          <WebHookResponseStatusFormField name="status"></WebHookResponseStatusFormField>
+        )}
         <FormContainer>
           <FormItem>
             <FormLabel tooltip={t('flow.msgTip')}>{t('flow.msg')}</FormLabel>

web/src/pages/agent/form/message-form/use-show-response-status.ts

@@ -0,0 +1,30 @@
+import { isEmpty } from 'lodash';
+import { useEffect, useMemo } from 'react';
+import { UseFormReturn } from 'react-hook-form';
+import {
+  AgentDialogueMode,
+  BeginId,
+  WebhookExecutionMode,
+} from '../../constant';
+import useGraphStore from '../../store';
+import { BeginFormSchemaType } from '../begin-form/schema';
+
+export function useShowWebhookResponseStatus(form: UseFormReturn<any>) {
+  const getNode = useGraphStore((state) => state.getNode);
+
+  const showWebhookResponseStatus = useMemo(() => {
+    const formData: BeginFormSchemaType = getNode(BeginId)?.data.form;
+    return (
+      formData.mode === AgentDialogueMode.Webhook &&
+      formData.execution_mode === WebhookExecutionMode.Streaming
+    );
+  }, []);
+
+  useEffect(() => {
+    if (showWebhookResponseStatus && isEmpty(form.getValues('status'))) {
+      form.setValue('status', 200, { shouldValidate: true, shouldDirty: true });
+    }
+  }, []);
+
+  return showWebhookResponseStatus;
+}

web/src/pages/chunk/parsed-result/add-knowledge/components/knowledge-chunk/components/chunk-card/index.less

@@ -4,6 +4,8 @@
 }
 
 .imagePreview {
+  width: 100%;
+  height: 100%;
   max-width: 50vw;
   max-height: 50vh;
   object-fit: contain;

web/src/pages/chunk/parsed-result/add-knowledge/components/knowledge-chunk/components/chunk-card/index.tsx

@@ -2,17 +2,19 @@ import Image from '@/components/image';
 import { useTheme } from '@/components/theme-provider';
 import { Card } from '@/components/ui/card';
 import { Checkbox } from '@/components/ui/checkbox';
-import {
-  Popover,
-  PopoverContent,
-  PopoverTrigger,
-} from '@/components/ui/popover';
 import { Switch } from '@/components/ui/switch';
+import {
+  Tooltip,
+  TooltipContent,
+  TooltipTrigger,
+} from '@/components/ui/tooltip';
 import { IChunk } from '@/interfaces/database/knowledge';
+import { cn } from '@/lib/utils';
 import { CheckedState } from '@radix-ui/react-checkbox';
 import classNames from 'classnames';
 import DOMPurify from 'dompurify';
 import { useEffect, useState } from 'react';
+import { useTranslation } from 'react-i18next';
 import { ChunkTextMode } from '../../constant';
 import styles from './index.less';
 
@@ -39,6 +41,7 @@ const ChunkCard = ({
   textMode,
   t: imageCacheKey,
 }: IProps) => {
+  const { t } = useTranslation();
   const available = Number(item.available_int);
   const [enabled, setEnabled] = useState(false);
   const { theme } = useTheme();
@@ -63,51 +66,59 @@ const ChunkCard = ({
   useEffect(() => {
     setEnabled(available === 1);
   }, [available]);
-  const [open, setOpen] = useState<boolean>(false);
+
   return (
     <Card
-      className={classNames(styles.chunkCard, {
+      className={classNames('relative flex-none', styles.chunkCard, {
         [`${theme === 'dark' ? styles.cardSelectedDark : styles.cardSelected}`]:
           selected,
       })}
     >
+      <span
+        className="
+        absolute top-0 right-0 px-4 py-1
+        leading-none text-xs text-text-disabled
+        bg-bg-card rounded-bl-2xl rounded-tr-lg
+        border-l-0.5 border-b-0.5 border-border-button"
+      >
+        {t(
+          `chunk.docType.${item.doc_type_kwd ? String(item.doc_type_kwd).toLowerCase() : 'text'}`,
+        )}
+      </span>
+
       <div className="flex items-start justify-between gap-2">
         <Checkbox onCheckedChange={handleCheck} checked={checked}></Checkbox>
+
+        {/* Using <Tooltip> instead of <Popover> to avoid flickering when hovering over the image */}
         {item.image_id && (
-          <Popover open={open}>
+          <Tooltip>
-            <PopoverTrigger
+            <TooltipTrigger>
-              asChild
+              <Image
-              onMouseEnter={() => setOpen(true)}
+                t={imageCacheKey}
-              onMouseLeave={() => setOpen(false)}
+                id={item.image_id}
-            >
+                className={styles.image}
-              <div>
+              />
-                <Image
+            </TooltipTrigger>
-                  t={imageCacheKey}
+            <TooltipContent
-                  id={item.image_id}
-                  className={styles.image}
-                />
-              </div>
-            </PopoverTrigger>
-            <PopoverContent
               className="p-0"
               align={'start'}
               side={'right'}
               sideOffset={-20}
+              tabIndex={-1}
             >
-              <div>
+              <Image
-                <Image
+                t={imageCacheKey}
-                  t={imageCacheKey}
+                id={item.image_id}
-                  id={item.image_id}
+                className={styles.imagePreview}
-                  className={styles.imagePreview}
+              />
-                ></Image>
+            </TooltipContent>
-              </div>
+          </Tooltip>
-            </PopoverContent>
-          </Popover>
         )}
+
         <section
           onDoubleClick={handleContentDoubleClick}
           onClick={handleContentClick}
-          className={styles.content}
+          className={cn(styles.content, 'mt-2')}
         >
           <div
             dangerouslySetInnerHTML={{
@@ -118,7 +129,8 @@ const ChunkCard = ({
             })}
           ></div>
         </section>
-        <div>
+
+        <div className="mt-2">
           <Switch
             checked={enabled}
             onCheckedChange={onChange}

web/src/pages/chunk/parsed-result/add-knowledge/components/knowledge-chunk/components/chunk-creating-modal/index.tsx

@@ -15,6 +15,7 @@ import {
   HoverCardContent,
   HoverCardTrigger,
 } from '@/components/ui/hover-card';
+import { Input } from '@/components/ui/input';
 import { Modal } from '@/components/ui/modal/modal';
 import Space from '@/components/ui/space';
 import { Switch } from '@/components/ui/switch';
@@ -76,6 +77,7 @@ const ChunkCreatingModal: React.FC<IModalProps<any> & kFProps> = ({
   const { removeChunk } = useDeleteChunkByIds();
   const { data } = useFetchChunk(chunkId);
   const { t } = useTranslation();
+  const isEditMode = !!chunkId;
 
   const isTagParser = parserId === 'tag';
   const onSubmit = useCallback(
@@ -144,6 +146,28 @@ const ChunkCreatingModal: React.FC<IModalProps<any> & kFProps> = ({
             )}
           />
 
+          {/* Do not display the type field in create mode */}
+          {isEditMode && (
+            <FormField
+              control={form.control}
+              name="doc_type_kwd"
+              render={({ field }) => (
+                <FormItem>
+                  <FormLabel>{t(`chunk.type`)}</FormLabel>
+                  <FormControl>
+                    <Input
+                      type="text"
+                      value={t(
+                        `chunk.docType.${field.value ? String(field.value).toLowerCase() : 'text'}`,
+                      )}
+                      readOnly
+                    />
+                  </FormControl>
+                </FormItem>
+              )}
+            />
+          )}
+
           <FormField
             control={form.control}
             name="image"
@@ -151,18 +175,18 @@ const ChunkCreatingModal: React.FC<IModalProps<any> & kFProps> = ({
               <FormItem>
                 <FormLabel className="gap-1">{t('chunk.image')}</FormLabel>
 
-                <div className="grid grid-cols-2 gap-4 items-start">
+                <div className="space-y-4">
                   {data?.data?.img_id && (
                     <Image
                       id={data?.data?.img_id}
-                      className="w-full object-contain"
+                      className="mx-auto w-auto max-w-full object-contain max-h-[800px]"
                     />
                   )}
 
                   <div className="col-start-2 col-end-3 only:col-span-2">
                     <FormControl>
                       <FileUploader
-                        className="h-48"
+                        className="h-auto p-6"
                         value={field.value}
                         onValueChange={field.onChange}
                         accept={{
@@ -171,6 +195,7 @@ const ChunkCreatingModal: React.FC<IModalProps<any> & kFProps> = ({
                           'image/webp': [],
                         }}
                         maxFileCount={1}
+                        hideDropzoneOnMaxFileCount
                         title={t('chunk.imageUploaderTitle')}
                         description={<></>}
                       />

web/src/pages/chunk/parsed-result/add-knowledge/components/knowledge-chunk/components/chunk-result-bar/checkbox-sets.tsx

@@ -45,7 +45,7 @@ export default (props: ICheckboxSetProps) => {
   }, [selectedChunkIds]);
 
   return (
-    <div className="flex gap-[40px] py-4 px-2">
+    <div className="flex gap-[40px] py-4 px-2 h-14">
       <div className="flex items-center gap-3 cursor-pointer text-muted-foreground hover:text-text-primary">
         <Checkbox
           id="all_chunks_checkbox"

web/src/pages/dataset/components/metedata/manage-values-modal.tsx

@@ -204,7 +204,8 @@ export const ManageValuesModal = (props: IManageValuesProps) => {
             </div>
           </div>
         )}
-        {metaData.restrictDefinedValues && (
+        {((metaData.restrictDefinedValues && isShowValueSwitch) ||
+          !isShowValueSwitch) && (
           <div className="flex flex-col gap-2">
             <div className="flex justify-between items-center">
               <div>{t('knowledgeDetails.metadata.values')}</div>

web/src/pages/dataset/dataset-setting/configuration/common-item.tsx

@@ -310,6 +310,36 @@ export function EnableTocToggle() {
   );
 }
 
+export function ImageContextWindow() {
+  const { t } = useTranslate('knowledgeConfiguration');
+  const form = useFormContext();
+
+  return (
+    <FormField
+      control={form.control}
+      name="parser_config.image_context_window"
+      render={({ field }) => (
+        <FormItem>
+          <FormControl>
+            <SliderInputFormField
+              {...field}
+              label={t('imageContextWindow')}
+              tooltip={t('imageContextWindowTip')}
+              defaultValue={0}
+              min={0}
+              max={256}
+            />
+          </FormControl>
+          <div className="flex pt-1">
+            <div className="w-1/4"></div>
+            <FormMessage />
+          </div>
+        </FormItem>
+      )}
+    />
+  );
+}
+
 export function OverlappedPercent() {
   return (
     <SliderInputFormField

web/src/pages/dataset/dataset-setting/configuration/naive.tsx

@@ -14,6 +14,7 @@ import {
 import {
   AutoMetadata,
   EnableTocToggle,
+  ImageContextWindow,
   OverlappedPercent,
 } from './common-item';
 
@@ -26,6 +27,7 @@ export function NaiveConfiguration() {
         <DelimiterFormField></DelimiterFormField>
         <ChildrenDelimiterForm />
         <EnableTocToggle />
+        <ImageContextWindow />
         <AutoMetadata />
         <OverlappedPercent />
       </ConfigurationFormContainer>

web/src/pages/dataset/dataset-setting/form-schema.ts

@@ -32,6 +32,7 @@ export const formSchema = z
         tag_kb_ids: z.array(z.string()).nullish(),
         topn_tags: z.number().optional(),
         toc_extraction: z.boolean().optional(),
+        image_context_window: z.number().optional(),
         overlapped_percent: z.number().optional(),
         // MinerU-specific options
         mineru_parse_method: z.enum(['auto', 'txt', 'ocr']).optional(),

web/src/pages/dataset/dataset-setting/index.tsx

@@ -70,6 +70,7 @@ export default function DatasetSettings() {
         html4excel: false,
         topn_tags: 3,
         toc_extraction: false,
+        image_context_window: 0,
         overlapped_percent: 0,
         // MinerU-specific defaults
         mineru_parse_method: 'auto',

web/src/pages/dataset/dataset/generate-button/generate.tsx

@@ -216,12 +216,6 @@ const Generate: React.FC<GenerateProps> = (props) => {
                 ? graphRunData
                 : raptorRunData
             ) as ITraceInfo;
-            console.log(
-              name,
-              'data',
-              data,
-              !data || (!data.progress && data.progress !== 0),
-            );
             return (
               <div key={name}>
                 <MenuItem

web/src/pages/dataset/dataset/use-select-filters.ts

@@ -25,12 +25,33 @@ export function useSelectDatasetFilters() {
       }));
     }
   }, [filter.run_status, t]);
+  const metaDataList = useMemo(() => {
+    if (filter.metadata) {
+      return Object.keys(filter.metadata).map((x) => ({
+        id: x.toString(),
+        field: x.toString(),
+        label: x.toString(),
+        list: Object.keys(filter.metadata[x]).map((y) => ({
+          id: y.toString(),
+          field: y.toString(),
+          label: y.toString(),
+          value: [y],
+          count: filter.metadata[x][y],
+        })),
+        count: Object.keys(filter.metadata[x]).reduce(
+          (acc, cur) => acc + filter.metadata[x][cur],
+          0,
+        ),
+      }));
+    }
+  }, [filter.metadata]);
   const filters: FilterCollection[] = useMemo(() => {
     return [
       { field: 'type', label: 'File Type', list: fileTypes },
       { field: 'run', label: 'Status', list: fileStatus },
+      { field: 'metadata', label: 'metadata', list: metaDataList },
     ] as FilterCollection[];
-  }, [fileStatus, fileTypes]);
+  }, [fileStatus, fileTypes, metaDataList]);
 
   return { filters, onOpenChange };
 }

web/src/pages/next-search/markdown-content/index.tsx

@@ -4,9 +4,8 @@ import { IReference, IReferenceChunk } from '@/interfaces/database/chat';
 import { getExtension } from '@/utils/document-util';
 import { InfoCircleOutlined } from '@ant-design/icons';
 import DOMPurify from 'dompurify';
-import { memo, useCallback, useEffect, useMemo } from 'react';
+import React, { memo, useCallback, useEffect, useMemo } from 'react';
 import Markdown from 'react-markdown';
-import reactStringReplace from 'react-string-replace';
 import SyntaxHighlighter from 'react-syntax-highlighter';
 import rehypeKatex from 'rehype-katex';
 import rehypeRaw from 'rehype-raw';
@@ -18,8 +17,13 @@ import { useTranslation } from 'react-i18next';
 
 import 'katex/dist/katex.min.css'; // `rehype-katex` does not import the CSS for you
 
+import ImageCarousel from '@/components/markdown-content/image-carousel';
+import {
+  groupConsecutiveReferences,
+  shouldShowCarousel,
+  type ReferenceGroup,
+} from '@/components/markdown-content/reference-utils';
 import {
-  currentReg,
   preprocessLaTeX,
   replaceTextByOldReg,
   replaceThinkToSection,
@@ -37,8 +41,6 @@ import classNames from 'classnames';
 import { omit } from 'lodash';
 import { pipe } from 'lodash/fp';
 
-const getChunkIndex = (match: string) => Number(match);
-
 // Defining Tailwind CSS class name constants
 const styles = {
   referenceChunkImage: 'w-[10vw] object-contain',
@@ -86,18 +88,11 @@ const MarkdownContent = ({
     (
       documentId: string,
       chunk: IReferenceChunk,
-      // isPdf: boolean,
+      isPdf: boolean = false,
-      // documentUrl?: string,
+      documentUrl?: string,
     ) =>
       () => {
-        // if (!isPdf) {
-        //   if (!documentUrl) {
-        //     return;
-        //   }
-        //   window.open(documentUrl, '_blank');
-        // } else {
         clickDocumentButton?.(documentId, chunk);
-        // }
       },
     [clickDocumentButton],
   );
@@ -218,43 +213,83 @@ const MarkdownContent = ({
 
   const renderReference = useCallback(
     (text: string) => {
-      let replacedText = reactStringReplace(text, currentReg, (match, i) => {
+      const groups = groupConsecutiveReferences(text);
-        const chunkIndex = getChunkIndex(match);
+      const elements: React.ReactNode[] = [];
+      let lastIndex = 0;
 
-        const { imageId, chunkItem, documentId } = getReferenceInfo(chunkIndex);
+      groups.forEach((group: ReferenceGroup) => {
+        // Add text before the group
+        if (group[0].start > lastIndex) {
+          elements.push(text.slice(lastIndex, group[0].start));
+        }
 
-        const docType = chunkItem?.doc_type;
+        // Determine if this group should be a carousel
+        if (shouldShowCarousel(group, reference)) {
+          // Render carousel for consecutive image group
+          elements.push(
+            <ImageCarousel
+              key={`carousel-${group[0].id}`}
+              group={group}
+              reference={reference}
+              fileThumbnails={fileThumbnails}
+              onImageClick={handleDocumentButtonClick}
+            />,
+          );
+        } else {
+          // Render individual references in the group
+          group.forEach((ref) => {
+            const chunkIndex = Number(ref.id);
+            const { imageId, chunkItem, documentId } =
+              getReferenceInfo(chunkIndex);
+            const docType = chunkItem?.doc_type;
 
-        return showImage(docType) ? (
+            if (showImage(docType)) {
-          <Image
+              elements.push(
-            id={imageId}
+                <Image
-            className={styles.referenceInnerChunkImage}
+                  key={ref.id}
-            onClick={
+                  id={imageId}
-              documentId
+                  className={styles.referenceInnerChunkImage}
-                ? handleDocumentButtonClick(
+                  onClick={
-                    documentId,
+                    documentId
-                    chunkItem,
+                      ? handleDocumentButtonClick(documentId, chunkItem)
-                    // fileExtension === 'pdf',
+                      : () => {}
-                    // documentUrl,
+                  }
-                  )
+                />,
-                : () => {}
+              );
+            } else {
+              elements.push(
+                <Popover key={ref.id}>
+                  <PopoverTrigger>
+                    <InfoCircleOutlined className={styles.referenceIcon} />
+                  </PopoverTrigger>
+                  <PopoverContent className="!w-fit">
+                    {getPopoverContent(chunkIndex)}
+                  </PopoverContent>
+                </Popover>,
+              );
             }
-          ></Image>
+            // Add the original reference text
-        ) : (
+            elements.push(ref.fullMatch);
-          <Popover>
+          });
-            <PopoverTrigger>
+        }
-              <InfoCircleOutlined className={styles.referenceIcon} />
+
-            </PopoverTrigger>
+        lastIndex = group[group.length - 1].end;
-            <PopoverContent className="!w-fit">
-              {getPopoverContent(chunkIndex)}
-            </PopoverContent>
-          </Popover>
-        );
       });
 
-      return replacedText;
+      // Add any remaining text after the last group
+      if (lastIndex < text.length) {
+        elements.push(text.slice(lastIndex));
+      }
+
+      return elements;
     },
-    [getPopoverContent, getReferenceInfo, handleDocumentButtonClick],
+    [
+      reference,
+      fileThumbnails,
+      handleDocumentButtonClick,
+      getReferenceInfo,
+      getPopoverContent,
+    ],
   );
 
   return (
@@ -265,7 +300,11 @@ const MarkdownContent = ({
       components={
         {
           'custom-typography': ({ children }: { children: string }) =>
-            renderReference(children),
+            Array.isArray(renderReference(children))
+              ? renderReference(children).map((element, index) => (
+                  <React.Fragment key={index}>{element}</React.Fragment>
+                ))
+              : renderReference(children),
           code(props: any) {
             const { children, className, ...rest } = props;
             const restProps = omit(rest, 'node');

web/src/pages/user-setting/data-source/component/blob-token-field.tsx

@@ -0,0 +1,247 @@
+import { useEffect, useMemo, useState } from 'react';
+import { useFormContext } from 'react-hook-form';
+
+import { SelectWithSearch } from '@/components/originui/select-with-search';
+import { RAGFlowFormItem } from '@/components/ragflow-form';
+import { Input } from '@/components/ui/input';
+import { Segmented } from '@/components/ui/segmented';
+import { t } from 'i18next';
+
+// UI-only auth modes for S3
+// access_key: Access Key ID + Secret
+// iam_role: only Role ARN
+// assume_role: no input fields (uses environment role)
+type AuthMode = 'access_key' | 'iam_role' | 'assume_role';
+type BlobMode = 's3' | 's3_compatible';
+
+const modeOptions = [
+  { label: 'S3', value: 's3' },
+  { label: 'S3 Compatible', value: 's3_compatible' },
+];
+
+const authOptions = [
+  { label: 'Access Key', value: 'access_key' },
+  { label: 'IAM Role', value: 'iam_role' },
+  { label: 'Assume Role', value: 'assume_role' },
+];
+
+const addressingOptions = [
+  { label: 'Virtual Hosted Style', value: 'virtual' },
+  { label: 'Path Style', value: 'path' },
+];
+
+const deriveInitialAuthMode = (credentials: any): AuthMode => {
+  const authMethod = credentials?.authentication_method;
+  if (authMethod === 'iam_role') return 'iam_role';
+  if (authMethod === 'assume_role') return 'assume_role';
+  if (credentials?.aws_role_arn) return 'iam_role';
+  if (credentials?.aws_access_key_id || credentials?.aws_secret_access_key)
+    return 'access_key';
+  return 'access_key';
+};
+
+const deriveInitialMode = (bucketType?: string): BlobMode =>
+  bucketType === 's3_compatible' ? 's3_compatible' : 's3';
+
+const BlobTokenField = () => {
+  const form = useFormContext();
+  const credentials = form.watch('config.credentials');
+  const watchedBucketType = form.watch('config.bucket_type');
+
+  const [mode, setMode] = useState<BlobMode>(
+    deriveInitialMode(watchedBucketType),
+  );
+  const [authMode, setAuthMode] = useState<AuthMode>(() =>
+    deriveInitialAuthMode(credentials),
+  );
+
+  // Keep bucket_type in sync with UI mode
+  useEffect(() => {
+    const nextMode = deriveInitialMode(watchedBucketType);
+    setMode((prev) => (prev === nextMode ? prev : nextMode));
+  }, [watchedBucketType]);
+
+  useEffect(() => {
+    form.setValue('config.bucket_type', mode, { shouldDirty: true });
+    // Default addressing style for compatible mode
+    if (
+      mode === 's3_compatible' &&
+      !form.getValues('config.credentials.addressing_style')
+    ) {
+      form.setValue('config.credentials.addressing_style', 'virtual', {
+        shouldDirty: false,
+      });
+    }
+    if (mode === 's3_compatible' && authMode !== 'access_key') {
+      setAuthMode('access_key');
+    }
+    // Persist authentication_method for backend
+    const nextAuthMethod: AuthMode =
+      mode === 's3_compatible' ? 'access_key' : authMode;
+    form.setValue('config.credentials.authentication_method', nextAuthMethod, {
+      shouldDirty: true,
+    });
+    // Clear errors for fields that are not relevant in the current mode/auth selection
+    const inactiveFields: string[] = [];
+    if (mode === 's3_compatible') {
+      inactiveFields.push('config.credentials.aws_role_arn');
+    } else {
+      if (authMode === 'iam_role') {
+        inactiveFields.push('config.credentials.aws_access_key_id');
+        inactiveFields.push('config.credentials.aws_secret_access_key');
+      }
+      if (authMode === 'assume_role') {
+        inactiveFields.push('config.credentials.aws_access_key_id');
+        inactiveFields.push('config.credentials.aws_secret_access_key');
+        inactiveFields.push('config.credentials.aws_role_arn');
+      }
+    }
+    if (inactiveFields.length) {
+      form.clearErrors(inactiveFields as any);
+    }
+  }, [form, mode, authMode]);
+
+  const isS3 = mode === 's3';
+  const requiresAccessKey =
+    authMode === 'access_key' || mode === 's3_compatible';
+  const requiresRoleArn = isS3 && authMode === 'iam_role';
+
+  // Help text for assume role (no inputs)
+  const assumeRoleNote = useMemo(
+    () => t('No credentials required. Uses the default environment role.'),
+    [t],
+  );
+
+  return (
+    <div className="flex flex-col gap-4">
+      <div className="flex flex-col gap-2">
+        <div className="text-sm text-text-secondary">Mode</div>
+        <Segmented
+          options={modeOptions}
+          value={mode}
+          onChange={(val) => setMode(val as BlobMode)}
+          className="w-full"
+          itemClassName="flex-1 justify-center"
+        />
+      </div>
+
+      {isS3 && (
+        <div className="flex flex-col gap-2">
+          <div className="text-sm text-text-secondary">Authentication</div>
+          <Segmented
+            options={authOptions}
+            value={authMode}
+            onChange={(val) => setAuthMode(val as AuthMode)}
+            className="w-full"
+            itemClassName="flex-1 justify-center"
+          />
+        </div>
+      )}
+
+      {requiresAccessKey && (
+        <RAGFlowFormItem
+          name="config.credentials.aws_access_key_id"
+          label="AWS Access Key ID"
+          required={requiresAccessKey}
+          rules={{
+            validate: (val) =>
+              requiresAccessKey
+                ? Boolean(val) || 'Access Key ID is required'
+                : true,
+          }}
+        >
+          {(field) => (
+            <Input {...field} placeholder="AKIA..." autoComplete="off" />
+          )}
+        </RAGFlowFormItem>
+      )}
+
+      {requiresAccessKey && (
+        <RAGFlowFormItem
+          name="config.credentials.aws_secret_access_key"
+          label="AWS Secret Access Key"
+          required={requiresAccessKey}
+          rules={{
+            validate: (val) =>
+              requiresAccessKey
+                ? Boolean(val) || 'Secret Access Key is required'
+                : true,
+          }}
+        >
+          {(field) => (
+            <Input
+              {...field}
+              type="password"
+              placeholder="****************"
+              autoComplete="new-password"
+            />
+          )}
+        </RAGFlowFormItem>
+      )}
+
+      {requiresRoleArn && (
+        <RAGFlowFormItem
+          name="config.credentials.aws_role_arn"
+          label="Role ARN"
+          required={requiresRoleArn}
+          tooltip="The role will be assumed by the runtime environment."
+          rules={{
+            validate: (val) =>
+              requiresRoleArn ? Boolean(val) || 'Role ARN is required' : true,
+          }}
+        >
+          {(field) => (
+            <Input
+              {...field}
+              placeholder="arn:aws:iam::123456789012:role/YourRole"
+              autoComplete="off"
+            />
+          )}
+        </RAGFlowFormItem>
+      )}
+
+      {isS3 && authMode === 'assume_role' && (
+        <div className="text-sm text-text-secondary bg-bg-card border border-border-button rounded-md px-3 py-2">
+          {assumeRoleNote}
+        </div>
+      )}
+
+      {mode === 's3_compatible' && (
+        <div className="flex flex-col gap-4">
+          <RAGFlowFormItem
+            name="config.credentials.addressing_style"
+            label="Addressing Style"
+            tooltip={t('setting.S3CompatibleAddressingStyleTip')}
+            required={false}
+          >
+            {(field) => (
+              <SelectWithSearch
+                triggerClassName="!shrink"
+                options={addressingOptions}
+                value={field.value || 'virtual'}
+                onChange={(val) => field.onChange(val)}
+              />
+            )}
+          </RAGFlowFormItem>
+
+          <RAGFlowFormItem
+            name="config.credentials.endpoint_url"
+            label="Endpoint URL"
+            required={false}
+            tooltip={t('setting.S3CompatibleEndpointUrlTip')}
+          >
+            {(field) => (
+              <Input
+                {...field}
+                placeholder="https://fsn1.your-objectstorage.com"
+                autoComplete="off"
+              />
+            )}
+          </RAGFlowFormItem>
+        </div>
+      )}
+    </div>
+  );
+};
+
+export default BlobTokenField;

web/src/pages/user-setting/data-source/contant.tsx

@@ -3,6 +3,8 @@ import SvgIcon from '@/components/svg-icon';
 import { t, TFunction } from 'i18next';
 import { useEffect, useState } from 'react';
 import { useTranslation } from 'react-i18next';
+import { BedrockRegionList } from '../setting-model/constant';
+import BlobTokenField from './component/blob-token-field';
 import BoxTokenField from './component/box-token-field';
 import { ConfluenceIndexingModeField } from './component/confluence-token-field';
 import GmailTokenField from './component/gmail-token-field';
@@ -105,6 +107,11 @@ export const generateDataSourceInfo = (t: TFunction) => {
   };
 };
 
+const awsRegionOptions = BedrockRegionList.map((r) => ({
+  label: r,
+  value: r,
+}));
+
 export const useDataSourceInfo = () => {
   const { t } = useTranslation();
   const [dataSourceInfo, setDataSourceInfo] = useState<IDataSourceInfoMap>(
@@ -222,18 +229,6 @@ export const DataSourceFormFields = {
     },
   ],
   [DataSourceKey.S3]: [
-    {
-      label: 'AWS Access Key ID',
-      name: 'config.credentials.aws_access_key_id',
-      type: FormFieldType.Text,
-      required: true,
-    },
-    {
-      label: 'AWS Secret Access Key',
-      name: 'config.credentials.aws_secret_access_key',
-      type: FormFieldType.Password,
-      required: true,
-    },
     {
       label: 'Bucket Name',
       name: 'config.bucket_name',
@@ -241,39 +236,21 @@ export const DataSourceFormFields = {
       required: true,
     },
     {
-      label: 'Bucket Type',
+      label: 'Region',
-      name: 'config.bucket_type',
+      name: 'config.credentials.region',
       type: FormFieldType.Select,
-      options: [
-        { label: 'S3', value: 's3' },
-        { label: 'S3 Compatible', value: 's3_compatible' },
-      ],
-      required: true,
-    },
-    {
-      label: 'Addressing Style',
-      name: 'config.credentials.addressing_style',
-      type: FormFieldType.Select,
-      options: [
-        { label: 'Virtual Hosted Style', value: 'virtual' },
-        { label: 'Path Style', value: 'path' },
-      ],
       required: false,
-      placeholder: 'Virtual Hosted Style',
+      options: awsRegionOptions,
-      tooltip: t('setting.S3CompatibleAddressingStyleTip'),
+      customValidate: (val: string, formValues: any) => {
-      shouldRender: (formValues: any) => {
+        const credentials = formValues?.config?.credentials || {};
-        return formValues?.config?.bucket_type === 's3_compatible';
+        const bucketType = formValues?.config?.bucket_type || 's3';
-      },
+        const hasAccessKey = Boolean(
-    },
+          credentials.aws_access_key_id || credentials.aws_secret_access_key,
-    {
+        );
-      label: 'Endpoint URL',
+        if (bucketType === 's3' && hasAccessKey) {
-      name: 'config.credentials.endpoint_url',
+          return Boolean(val) || 'Region is required when using access key';
-      type: FormFieldType.Text,
+        }
-      required: false,
+        return true;
-      placeholder: 'https://fsn1.your-objectstorage.com',
-      tooltip: t('setting.S3CompatibleEndpointUrlTip'),
-      shouldRender: (formValues: any) => {
-        return formValues?.config?.bucket_type === 's3_compatible';
       },
     },
     {
@@ -283,6 +260,14 @@ export const DataSourceFormFields = {
       required: false,
       tooltip: t('setting.s3PrefixTip'),
     },
+    {
+      label: 'Credentials',
+      name: 'config.credentials.__blob_token',
+      type: FormFieldType.Custom,
+      hideLabel: true,
+      required: false,
+      render: () => <BlobTokenField />,
+    },
   ],
   [DataSourceKey.NOTION]: [
     {
@@ -700,6 +685,9 @@ export const DataSourceFormDefaultValues = {
       credentials: {
         aws_access_key_id: '',
         aws_secret_access_key: '',
+        region: '',
+        authentication_method: 'access_key',
+        aws_role_arn: '',
         endpoint_url: '',
         addressing_style: 'virtual',
       },

web/src/pages/user-setting/data-source/data-source-detail-page/index.tsx

@@ -10,8 +10,7 @@ import { Input } from '@/components/ui/input';
 import { Separator } from '@/components/ui/separator';
 import { RunningStatus } from '@/constants/knowledge';
 import { t } from 'i18next';
-import { debounce } from 'lodash';
+import { CirclePause, Loader2, Repeat } from 'lucide-react';
-import { CirclePause, Repeat } from 'lucide-react';
 import { useCallback, useEffect, useMemo, useRef, useState } from 'react';
 import { FieldValues } from 'react-hook-form';
 import {
@@ -120,11 +119,11 @@ const SourceDetailPage = () => {
     ];
   }, [detail, runSchedule]);
 
-  const { handleAddOk } = useAddDataSource();
+  const { addLoading, handleAddOk } = useAddDataSource();
 
   const onSubmit = useCallback(() => {
     formRef?.current?.submit();
-  }, [formRef]);
+  }, []);
 
   useEffect(() => {
     if (detail) {
@@ -140,9 +139,7 @@ const SourceDetailPage = () => {
         return {
           ...field,
           horizontal: true,
-          onChange: () => {
+          onChange: undefined,
-            onSubmit();
-          },
         };
       });
       setFields(newFields);
@@ -175,12 +172,23 @@ const SourceDetailPage = () => {
             <DynamicForm.Root
               ref={formRef}
               fields={fields}
-              onSubmit={debounce((data) => {
+              onSubmit={(data) => handleAddOk(data)}
-                handleAddOk(data);
-              }, 500)}
               defaultValues={defaultValues}
             />
           </div>
+          <div className="max-w-[1200px] flex justify-end">
+            <button
+              type="button"
+              onClick={onSubmit}
+              disabled={addLoading}
+              className="flex items-center justify-center min-w-[100px] px-4 py-2 bg-primary text-white rounded-md disabled:opacity-60"
+            >
+              {addLoading && <Loader2 className="mr-2 h-4 w-4 animate-spin" />}
+              {addLoading
+                ? t('modal.loadingText', { defaultValue: 'Submitting...' })
+                : t('modal.okText', { defaultValue: 'Submit' })}
+            </button>
+          </div>
           <section className="flex flex-col gap-2">
             <div className="text-2xl text-text-primary mb-2">
               {t('setting.log')}