feat(web): add document-extractor node

web/src/i18n/en.ts

@@ -2163,6 +2163,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re
       'memory-write': 'Memory Storage',
       unknown: 'Unknown Node',
       notes: 'Sticky Note',
+      'document-extractor': 'Document Extractor',
 
       clickToConfigure: 'Click to configure node parameters',
       nodeProperties: 'Node Properties',
@@ -2361,6 +2362,9 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re
           placeholder: 'Enter note...',
           removeLink: 'Remove Link',
         },
+        'document-extractor': {
+          file_selector: 'Input Variable',
+        },
         name: 'Key',
         type: 'Type',
         value: 'Value',

web/src/i18n/zh.ts

@@ -2160,6 +2160,7 @@ export const zh = {
       'memory-write': '记忆储存',
       unknown: '未知节点',
       notes: '便签',
+      'document-extractor': '文档提取器',
 
       clickToConfigure: '点击配置节点参数',
       nodeProperties: '节点属性',
@@ -2361,6 +2362,9 @@ export const zh = {
           placeholder: '输入注释...',
           removeLink: '取消链接',
         },
+        'document-extractor': {
+          file_selector: '输入变量',
+        },
         name: '键',
         type: '类型',
         value: '值',

web/src/views/Workflow/components/Properties/CycleVarsList/index.tsx

@@ -1,10 +1,11 @@
-import { type FC } from 'react'
+import { type FC, useMemo } from 'react'
 import { useTranslation } from 'react-i18next';
 import { Form, Select, Input, Button, InputNumber, Flex } from 'antd'
 
 import VariableSelect from '../VariableSelect'
 import type { Suggestion } from '../../Editor/plugin/AutocompletePlugin'
 import RadioGroupBtn from '../RadioGroupBtn'
+import { getChildNodeVariables } from '../hooks/useVariableList'
 
 interface CycleVar {
   name: string;
@@ -44,40 +45,14 @@ const CycleVarsList: FC<CycleVarsListProps> = ({
   const { t } = useTranslation();
   const form = Form.useFormInstance();
 
-  // 获取循环节点的子节点变量
-  const getChildNodeVariables = () => {
+  const availableOptions = useMemo(() => {
     if (!selectedNode || !graphRef?.current || selectedNode.getData()?.type !== 'loop') {
       return options;
     }
+    const childVars = getChildNodeVariables(selectedNode, graphRef);
 
-    const loopNodeId = selectedNode.getData()?.id;
-    const childNodes = graphRef.current.getNodes().filter((node: any) => 
-      node.getData()?.cycle === loopNodeId
-    );
-
-    const childVariables: Suggestion[] = [];
-    childNodes.forEach((childNode: any) => {
-      const childData = childNode.getData();
-      if (childData?.config) {
-        Object.keys(childData.config).forEach(key => {
-          if (childData.config[key]?.defaultValue) {
-            childVariables.push({
-              key: `${childData.id}.${key}`,
-              label: `${childData.name || childData.type}.${key}`,
-              type: 'output',
-              dataType: 'string',
-              value: `${childData.id}.${key}`,
-              nodeData: childData
-            });
-          }
-        });
-      }
-    });
-
-    return [...options, ...childVariables];
-  };
-
-  const availableOptions = getChildNodeVariables();
+    return options.filter(option => !childVars.some(item => item.value === option.value))
+  }, [options, selectedNode, graphRef]);
 
   return (
     <Form.List name={parentName}>

web/src/views/Workflow/components/Properties/hooks/useVariableList.ts

@@ -42,6 +42,9 @@ const NODE_VARIABLES = {
   'memory-read': [
     { label: 'answer', dataType: 'string', field: 'answer' },
     { label: 'intermediate_outputs', dataType: 'array[object]', field: 'intermediate_outputs' }
+  ],
+  'document-extractor': [
+    { label: 'text', dataType: 'string', field: 'text' },
   ]
 } as const;
 
@@ -177,7 +180,8 @@ const hasOutputNodeTypes = [
   'var-aggregator',
   'http-request',
   'tool',
-  'jinja-render'
+  'jinja-render',
+  'document-extractor'
 ];
 
 /**

web/src/views/Workflow/components/Properties/index.tsx

@@ -636,6 +636,8 @@ const Properties: FC<PropertiesProps> = ({
                               size="small"
                               parentName={key}
                               options={getFilteredVariableList(selectedNode?.data?.type, key)}
+                              selectedNode={selectedNode}
+                              graphRef={graphRef}
                             />
                           </Form.Item>
                         )

web/src/views/Workflow/constant.ts

@@ -473,6 +473,16 @@ export const nodeLibrary: NodeLibrary[] = [
           },
         }
       },
+      {
+        type: "document-extractor", icon: codeExecutionIcon,
+        config: {
+          file_selector: {
+            type: 'variableList',
+            placeholder: 'common.pleaseSelect',
+            onFilterVariableNames: ['sys.files']
+          }
+        }
+      },
     ]
   },
 ];
@@ -858,100 +868,6 @@ export interface OutputVariable {
   }>;
 }
 
-/**
- * Output variable definitions for each node type
- * Specifies what variables each node produces
- */
-export const outputVariable: { [key: string]: OutputVariable } = {
-  start: {
-    sys: [
-      { name: "message", type: "string" },
-      { name: "conversation_id", type: "string" },
-      { name: "execution_id", type: "string", },
-      { name: "workspace_id", type: "string" },
-      { name: "user_id", type: "string" },
-    ],
-    define: ['variables']
-  },
-  end: {
-  },
-  llm: {
-    default: [
-      { name: "output", type: "string" },
-    ]
-  },
-  'knowledge-retrieval': {
-    default: [
-      { name: "output", type: "array[object]" },
-    ]
-  },
-  'parameter-extractor': {
-    default: [
-      { name: "__is_success", type: "number" },
-      { name: "__reason", type: "string" },
-    ],
-    define: ['params']
-  },
-  'memory-read': {
-    default: [
-      { name: "answer", type: "string" },
-      { name: "intermediate_outputs", type: "array[object]" },
-    ],
-  },
-  'memory-write': {
-
-  },
-  'if-else': {
-
-  },
-  'question-classifier': {
-    default: [
-      { name: "class_name", type: "string" },
-      // { name: "output", type: "string" },
-    ],
-  },
-  'iteration': {
-    default: [
-      // { name: "item", type: "string" }, // 仅内部使用
-      { name: "output", type: "array[string]" },
-    ],
-  },
-  'loop': {
-    define: ['cycle_vars']
-  },
-  'cycle-start': {
-
-  },
-  'break': {
-
-  },
-  'var-aggregator': {
-    // default: [
-    //   { name: "output", type: "string" },
-    // ],
-    define: ['group_variables']
-  },
-  'assigner': {
-
-  },
-  'http-request': {
-    default: [
-      { name: "body", type: "string" },
-      { name: "status_code", type: "number" },
-    ],
-  },
-  'tool': {
-    default: [
-      { name: "data", type: "string" },
-    ],
-  },
-  'jinja-render': {
-    default: [
-      { name: "output", type: "string" },
-    ],
-  },
-}
-
 /**
  * Default edge attributes configuration
  * Defines visual styling for edges/connections