feat: doc extract struct

web/app/components/workflow/block-selector/constants.tsx

@@ -69,6 +69,11 @@ export const BLOCKS: Block[] = [
     type: BlockEnum.HttpRequest,
     title: 'HTTP Request',
   },
+  {
+    classification: BlockClassificationEnum.Utilities,
+    type: BlockEnum.DocExtractor,
+    title: 'Doc Extractor',
+  },
 ]
 
 export const BLOCK_CLASSIFICATIONS: string[] = [

web/app/components/workflow/constants.ts

@@ -14,6 +14,7 @@ import ToolDefault from './nodes/tool/default'
 import VariableAssignerDefault from './nodes/variable-assigner/default'
 import EndNodeDefault from './nodes/end/default'
 import IterationDefault from './nodes/iteration/default'
+import DocExtractorDefault from './nodes/doc-extractor/default'
 
 type NodesExtraData = {
   author: string
@@ -160,6 +161,16 @@ export const NODES_EXTRA_DATA: Record<BlockEnum, NodesExtraData> = {
     getAvailableNextNodes: ToolDefault.getAvailableNextNodes,
     checkValid: ToolDefault.checkValid,
   },
+  [BlockEnum.DocExtractor]: {
+    author: 'Dify',
+    about: '',
+    availablePrevNodes: [],
+    availableNextNodes: [],
+    getAvailablePrevNodes: DocExtractorDefault.getAvailablePrevNodes,
+    getAvailableNextNodes: DocExtractorDefault.getAvailableNextNodes,
+    checkValid: DocExtractorDefault.checkValid,
+  },
+
 }
 
 export const ALL_CHAT_AVAILABLE_BLOCKS = Object.keys(NODES_EXTRA_DATA).filter(key => key !== BlockEnum.End && key !== BlockEnum.Start) as BlockEnum[]
@@ -274,6 +285,12 @@ export const NODES_INITIAL_DATA = {
     desc: '',
     ...ToolDefault.defaultValue,
   },
+  [BlockEnum.DocExtractor]: {
+    type: BlockEnum.DocExtractor,
+    title: '',
+    desc: '',
+    ...DocExtractorDefault.defaultValue,
+  },
 }
 
 export const NODE_WIDTH = 240

web/app/components/workflow/nodes/_base/components/variable/utils.ts

@@ -235,6 +235,16 @@ const formatItem = (
       break
     }
 
+    case BlockEnum.DocExtractor: {
+      res.vars = [
+        {
+          variable: 'text',
+          type: VarType.string,
+        },
+      ]
+      break
+    }
+
     case 'env': {
       res.vars = data.envList.map((env: EnvironmentVariable) => {
         return {

web/app/components/workflow/nodes/constants.ts

@@ -28,6 +28,8 @@ import ParameterExtractorNode from './parameter-extractor/node'
 import ParameterExtractorPanel from './parameter-extractor/panel'
 import IterationNode from './iteration/node'
 import IterationPanel from './iteration/panel'
+import DocExtractorNode from './doc-extractor/node'
+import DocExtractorPanel from './doc-extractor/panel'
 
 export const NodeComponentMap: Record<string, ComponentType<any>> = {
   [BlockEnum.Start]: StartNode,
@@ -45,6 +47,7 @@ export const NodeComponentMap: Record<string, ComponentType<any>> = {
   [BlockEnum.VariableAggregator]: VariableAssignerNode,
   [BlockEnum.ParameterExtractor]: ParameterExtractorNode,
   [BlockEnum.Iteration]: IterationNode,
+  [BlockEnum.DocExtractor]: DocExtractorNode,
 }
 
 export const PanelComponentMap: Record<string, ComponentType<any>> = {
@@ -63,6 +66,7 @@ export const PanelComponentMap: Record<string, ComponentType<any>> = {
   [BlockEnum.VariableAggregator]: VariableAssignerPanel,
   [BlockEnum.ParameterExtractor]: ParameterExtractorPanel,
   [BlockEnum.Iteration]: IterationPanel,
+  [BlockEnum.DocExtractor]: DocExtractorPanel,
 }
 
 export const CUSTOM_NODE_TYPE = 'custom'

web/app/components/workflow/nodes/doc-extractor/default.ts

@@ -0,0 +1,35 @@
+import { BlockEnum } from '../../types'
+import type { NodeDefault } from '../../types'
+import { type DocExtractorNodeType } from './types'
+import { ALL_CHAT_AVAILABLE_BLOCKS, ALL_COMPLETION_AVAILABLE_BLOCKS } from '@/app/components/workflow/constants'
+const i18nPrefix = 'workflow.errorMsg'
+
+const nodeDefault: NodeDefault<DocExtractorNodeType> = {
+  defaultValue: {
+    variable: [],
+  },
+  getAvailablePrevNodes(isChatMode: boolean) {
+    const nodes = isChatMode
+      ? ALL_CHAT_AVAILABLE_BLOCKS
+      : ALL_COMPLETION_AVAILABLE_BLOCKS.filter(type => type !== BlockEnum.End)
+    return nodes
+  },
+  getAvailableNextNodes(isChatMode: boolean) {
+    const nodes = isChatMode ? ALL_CHAT_AVAILABLE_BLOCKS : ALL_COMPLETION_AVAILABLE_BLOCKS
+    return nodes
+  },
+  checkValid(payload: DocExtractorNodeType, t: any) {
+    let errorMessages = ''
+    const { variable } = payload
+
+    if (!errorMessages && !variable?.length)
+      errorMessages = t(`${i18nPrefix}.fieldRequired`, { field: t('workflow.nodes.assigner.assignedVariable') })
+
+    return {
+      isValid: !errorMessages,
+      errorMessage: errorMessages,
+    }
+  },
+}
+
+export default nodeDefault

web/app/components/workflow/nodes/doc-extractor/node.tsx

@@ -0,0 +1,40 @@
+import type { FC } from 'react'
+import React from 'react'
+import { useNodes } from 'reactflow'
+import { useTranslation } from 'react-i18next'
+import NodeVariableItem from '../variable-assigner/components/node-variable-item'
+import { type DocExtractorNodeType } from './types'
+import { isENV, isSystemVar } from '@/app/components/workflow/nodes/_base/components/variable/utils'
+import { BlockEnum, type Node, type NodeProps } from '@/app/components/workflow/types'
+
+const i18nPrefix = 'workflow.nodes.assigner'
+
+const NodeComponent: FC<NodeProps<DocExtractorNodeType>> = ({
+  data,
+}) => {
+  const { t } = useTranslation()
+
+  const nodes: Node[] = useNodes()
+  const { variable } = data
+
+  if (!variable || variable.length === 0)
+    return null
+
+  const isSystem = isSystemVar(variable)
+  const isEnv = isENV(variable)
+  const node = isSystem ? nodes.find(node => node.data.type === BlockEnum.Start) : nodes.find(node => node.id === variable[0])
+  const varName = isSystem ? `sys.${variable[variable.length - 1]}` : variable.slice(1).join('.')
+  return (
+    <div className='relative px-3'>
+      <div className='mb-1 system-2xs-medium-uppercase text-text-tertiary'>{t(`${i18nPrefix}.assignedVariable`)}</div>
+      <NodeVariableItem
+        node={node as Node}
+        isEnv={isEnv}
+        varName={varName}
+        className='bg-workflow-block-parma-bg'
+      />
+    </div>
+  )
+}
+
+export default React.memo(NodeComponent)

web/app/components/workflow/nodes/doc-extractor/panel.tsx

@@ -0,0 +1,45 @@
+import type { FC } from 'react'
+import React from 'react'
+import { useTranslation } from 'react-i18next'
+import VarReferencePicker from '../_base/components/variable/var-reference-picker'
+import useConfig from './use-config'
+import type { DocExtractorNodeType } from './types'
+import Field from '@/app/components/workflow/nodes/_base/components/field'
+import { type NodePanelProps } from '@/app/components/workflow/types'
+
+const i18nPrefix = 'workflow.nodes.docExtractor'
+
+const Panel: FC<NodePanelProps<DocExtractorNodeType>> = ({
+  id,
+  data,
+}) => {
+  const { t } = useTranslation()
+
+  const {
+    readOnly,
+    inputs,
+    handleVarChanges,
+    filterVar,
+  } = useConfig(id, data)
+
+  return (
+    <div className='mt-2'>
+      <div className='px-4 pb-4 space-y-4'>
+        <Field
+          title={t(`${i18nPrefix}.assignedVariable`)}
+        >
+          <VarReferencePicker
+            readonly={readOnly}
+            nodeId={id}
+            isShowNodeName
+            value={inputs.variable || []}
+            onChange={handleVarChanges}
+            filterVar={filterVar}
+          />
+        </Field>
+      </div>
+    </div>
+  )
+}
+
+export default React.memo(Panel)

web/app/components/workflow/nodes/doc-extractor/types.ts

@@ -0,0 +1,5 @@
+import type { CommonNodeType, ValueSelector } from '@/app/components/workflow/types'
+
+export type DocExtractorNodeType = CommonNodeType & {
+  variable: ValueSelector
+}

web/app/components/workflow/nodes/doc-extractor/use-config.ts

@@ -0,0 +1,35 @@
+import { useCallback } from 'react'
+import produce from 'immer'
+import type { ValueSelector, Var } from '../../types'
+import { VarType } from '../../types'
+import { type DocExtractorNodeType } from './types'
+import useNodeCrud from '@/app/components/workflow/nodes/_base/hooks/use-node-crud'
+import {
+  useNodesReadOnly,
+} from '@/app/components/workflow/hooks'
+
+const useConfig = (id: string, payload: DocExtractorNodeType) => {
+  const { nodesReadOnly: readOnly } = useNodesReadOnly()
+
+  const { inputs, setInputs } = useNodeCrud<DocExtractorNodeType>(id, payload)
+
+  const handleVarChanges = useCallback((variable: ValueSelector | string) => {
+    const newInputs = produce(inputs, (draft) => {
+      draft.variable = variable as ValueSelector
+    })
+    setInputs(newInputs)
+  }, [inputs, setInputs])
+
+  const filterVar = useCallback((varPayload: Var) => {
+    return varPayload.type !== VarType.file
+  }, [])
+
+  return {
+    readOnly,
+    inputs,
+    filterVar,
+    handleVarChanges,
+  }
+}
+
+export default useConfig

web/app/components/workflow/nodes/variable-assigner/components/node-variable-item.tsx

@@ -8,12 +8,14 @@ import type { Node } from '@/app/components/workflow/types'
 import { BlockEnum } from '@/app/components/workflow/types'
 
 type NodeVariableItemProps = {
+  className?: string
   isEnv: boolean
   node: Node
   varName: string
   showBorder?: boolean
 }
 const NodeVariableItem = ({
+  className,
   isEnv,
   node,
   varName,
@@ -22,7 +24,7 @@ const NodeVariableItem = ({
   return (
     <div className={cn(
       'relative flex items-center mt-0.5 h-6 bg-gray-100 rounded-md  px-1 text-xs font-normal text-gray-700',
-      showBorder && '!bg-black/[0.02]',
+      showBorder && '!bg-black/[0.02]', className,
     )}>
       {!isEnv && (
         <div className='flex items-center'>

web/app/components/workflow/types.ts

@@ -25,6 +25,8 @@ export enum BlockEnum {
   Tool = 'tool',
   ParameterExtractor = 'parameter-extractor',
   Iteration = 'iteration',
+  DocExtractor = 'doc-extractor',
+  ListFilter = 'list-filter',
 }
 
 export type Branch = {
@@ -193,6 +195,7 @@ export enum VarType {
   secret = 'secret',
   boolean = 'boolean',
   object = 'object',
+  file = 'file',
   array = 'array',
   arrayString = 'array[string]',
   arrayNumber = 'array[number]',

web/i18n/en-US/workflow.ts

@@ -176,6 +176,7 @@ const translation = {
     'iteration-start': 'Iteration Start',
     'iteration': 'Iteration',
     'parameter-extractor': 'Parameter Extractor',
+    'doc-extractor': 'Doc Extractor',
   },
   blocksAbout: {
     'start': 'Define the initial parameters for launching a workflow',
@@ -192,6 +193,7 @@ const translation = {
     'variable-aggregator': 'Aggregate multi-branch variables into a single variable for unified configuration of downstream nodes.',
     'iteration': 'Perform multiple steps on a list object until all results are outputted.',
     'parameter-extractor': 'Use LLM to extract structured parameters from natural language for tool invocations or HTTP requests.',
+    'doc-extractor': 'TODO',
   },
   operator: {
     zoomIn: 'Zoom In',
@@ -489,6 +491,8 @@ const translation = {
         showAuthor: 'Show Author',
       },
     },
+    docExtractor: {
+    },
   },
   tracing: {
     stopBy: 'Stop by {{user}}',