feat:add dataprocessing task retry functionality

apiserver/graph/schema/dataprocessing.gql

@@ -128,6 +128,26 @@ query getLogInfo($input: DataProcessDetailsInput){
   }
 }
 
+query dataProcessLogInfoByFileName($input: DataProcessFileLogInput){
+  dataProcess {
+    dataProcessLogInfoByFileName(input: $input) {
+      status
+      data 
+      message
+    }
+  }
+}
+
+query dataProcessRetry($input: DataProcessRetryInput){
+  dataProcess {
+    dataProcessRetry(input: $input) {
+      status
+      data 
+      message
+    }
+  }
+}
+
 mutation createDataProcessTask($input: AddDataProcessInput) {
   dataProcess {
     createDataProcessTask(input: $input) {

apiserver/graph/schema/dataprocessing.graphqls

@@ -21,6 +21,10 @@ type DataProcessQuery {
   checkDataProcessTaskName(input: CheckDataProcessTaskNameInput): DataProcessResponse
   # 日志信息
   getLogInfo(input: DataProcessDetailsInput): DataProcessResponse
+  # 获取文件处理的日志
+  dataProcessLogInfoByFileName(input: DataProcessFileLogInput): DataProcessResponse
+  # 任务重试
+  dataProcessRetry(input: DataProcessRetryInput): DataProcessResponse
 }
 
 
@@ -71,6 +75,7 @@ input LLMConfigItem {
   top_p: String
   max_tokens: String
   prompt_template: String
+  provider: String
 }
 
 input DeleteDataProcessInput {
@@ -86,6 +91,17 @@ input CheckDataProcessTaskNameInput {
   namespace: String!
 }
 
+input DataProcessFileLogInput {
+  id: String!
+  file_name: String!
+  type: String!
+}
+
+input DataProcessRetryInput {
+  id: String!
+  creator: String!
+}
+
 # 数据处理列表分页
 type PaginatedDataProcessItem {
   status: Int!

apiserver/pkg/dataprocessing/dataprocessing.go

@@ -259,3 +259,59 @@ func GetLogInfo(ctx context.Context, c dynamic.Interface, obj *generated.DataPro
 	}
 	return data, nil
 }
+
+func DataProcessLogInfoByFileName(ctx context.Context, c dynamic.Interface, obj *generated.DataProcessQuery, input *generated.DataProcessFileLogInput) (*generated.DataProcessResponse, error) {
+	// prepare http request
+	jsonParams, err := json.Marshal(input)
+	if err != nil {
+		return nil, err
+	}
+	req, err := http.NewRequest("POST", url+"/get-log-by-file-name", bytes.NewBuffer(jsonParams))
+	if err != nil {
+		return nil, err
+	}
+
+	// call dataprocessing server
+	client := http.Client{}
+	resp, err := client.Do(req)
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+
+	// parse http response
+	data := &generated.DataProcessResponse{}
+	err = json.NewDecoder(resp.Body).Decode(data)
+	if err != nil {
+		return nil, err
+	}
+	return data, nil
+}
+
+func DataProcessRetry(ctx context.Context, c dynamic.Interface, obj *generated.DataProcessQuery, input *generated.DataProcessRetryInput) (*generated.DataProcessResponse, error) {
+	// prepare http request
+	jsonParams, err := json.Marshal(input)
+	if err != nil {
+		return nil, err
+	}
+	req, err := http.NewRequest("POST", url+"/retry", bytes.NewBuffer(jsonParams))
+	if err != nil {
+		return nil, err
+	}
+
+	// call dataprocessing server
+	client := http.Client{}
+	resp, err := client.Do(req)
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+
+	// parse http response
+	data := &generated.DataProcessResponse{}
+	err = json.NewDecoder(resp.Body).Decode(data)
+	if err != nil {
+		return nil, err
+	}
+	return data, nil
+}

data-processing/data_manipulation/common/log_tag_const.py

@@ -22,6 +22,7 @@
 MINIO = "Minio"
 MINIO_STORE_PROCESS = "Minio Store Process"
 
+DATA_PROCESS_SERVICE = "Data Process Service"
 DATA_PROCESS_DETAIL = "Data Process Detail"
 
 COMMON_HANDLE = "Common Handle"

data-processing/data_manipulation/common/special_characters.py

@@ -34,6 +34,6 @@
 # whitespaces in unicode can be found here:
 # https://en.wikipedia.org/wiki/Whitespace_character
 VARIOUS_WHITESPACES = {
-    ' ', '	', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ',
+    '	', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ',
     ' ', ' ', ' ', '　', '​', '‌', '‍', '⁠', '', '„'
 }

data-processing/data_manipulation/controller/data_process_controller.py

@@ -153,3 +153,36 @@ async def get_log_info(request):
     )
     return json(res) 
 
+@data_process.route('get-log-by-file-name', methods=['POST'])
+async def get_log_by_file_name(request):
+    """check task name by name and namespace.
+    
+    example for request.json
+    {
+        "id": "01HGWBE48DT3ADE9ZKA62SW4WS",
+        "file_name": "xx.pdf",
+        "type": "qa_split"
+    }
+    """
+    res = data_process_service.get_log_by_file_name(
+        request.json,
+        pool=request.app.config['conn_pool']
+    )
+    return json(res) 
+
+
+@data_process.route('retry', methods=['POST'])
+async def retry(request):
+    """check task name by name and namespace.
+    
+    example for request.json
+    {
+        "id": "01HGWBE48DT3ADE9ZKA62SW4WS"
+    }
+    """
+    res = data_process_service.retry(
+        request.json,
+        pool=request.app.config['conn_pool']
+    )
+    return json(res) 
+