Revert "Feat/parent child retrieval" (#12095)

web/models/datasets.ts

@@ -1,7 +1,6 @@
 import type { DataSourceNotionPage, DataSourceProvider } from './common'
 import type { AppIconType, AppMode, RetrievalConfig } from '@/types/app'
 import type { Tag } from '@/app/components/base/tag-management/constant'
-import type { IndexingType } from '@/app/components/datasets/create/step-two'
 
 export enum DataSourceType {
   FILE = 'upload_file',
@@ -11,12 +10,6 @@ export enum DataSourceType {
 
 export type DatasetPermission = 'only_me' | 'all_team_members' | 'partial_members'
 
-export enum ChunkingMode {
-  'text' = 'text_model', // General text
-  'qa' = 'qa_model', // General QA
-  'parentChild' = 'hierarchical_model', // Parent-Child
-}
-
 export type DataSet = {
   id: string
   name: string
@@ -25,12 +18,11 @@ export type DataSet = {
   description: string
   permission: DatasetPermission
   data_source_type: DataSourceType
-  indexing_technique: IndexingType
+  indexing_technique: 'high_quality' | 'economy'
   created_by: string
   updated_by: string
   updated_at: number
   app_count: number
-  doc_form: ChunkingMode
   document_count: number
   word_count: number
   provider: string
@@ -103,12 +95,6 @@ export type CustomFile = File & {
   created_at?: number
 }
 
-export type DocumentItem = {
-  id: string
-  name: string
-  extension: string
-}
-
 export type CrawlOptions = {
   crawl_sub_pages: boolean
   only_main_content: boolean
@@ -158,7 +144,7 @@ export type IndexingEstimateResponse = {
   total_price: number
   currency: string
   total_segments: number
-  preview: Array<{ content: string; child_chunks: string[] }>
+  preview: string[]
   qa_preview?: QA[]
 }
 
@@ -184,12 +170,7 @@ export type IndexingStatusBatchResponse = {
   data: IndexingStatusResponse[]
 }
 
-export enum ProcessMode {
-  general = 'custom',
-  parentChild = 'hierarchical',
-}
-
-export type ParentMode = 'full-doc' | 'paragraph'
+export type ProcessMode = 'automatic' | 'custom'
 
 export type ProcessRuleResponse = {
   mode: ProcessMode
@@ -200,8 +181,6 @@ export type ProcessRuleResponse = {
 export type Rules = {
   pre_processing_rules: PreProcessingRule[]
   segmentation: Segmentation
-  parent_mode: ParentMode
-  subchunk_segmentation: Segmentation
 }
 
 export type Limits = {
@@ -216,7 +195,7 @@ export type PreProcessingRule = {
 export type Segmentation = {
   separator: string
   max_tokens: number
-  chunk_overlap?: number
+  chunk_overlap: number
 }
 
 export const DocumentIndexingStatusList = [
@@ -279,14 +258,13 @@ export type InitialDocumentDetail = {
   display_status: DocumentDisplayStatus
   completed_segments?: number
   total_segments?: number
-  doc_form: ChunkingMode
+  doc_form: 'text_model' | 'qa_model'
   doc_language: string
 }
 
 export type SimpleDocumentDetail = InitialDocumentDetail & {
   enabled: boolean
   word_count: number
-  is_qa: boolean // TODO waiting for backend to add this field
   error?: string | null
   archived: boolean
   updated_at: number
@@ -311,7 +289,7 @@ export type DocumentListResponse = {
 export type DocumentReq = {
   original_document_id?: string
   indexing_technique?: string
-  doc_form: ChunkingMode
+  doc_form: 'text_model' | 'qa_model'
   doc_language: string
   process_rule: ProcessRule
 }
@@ -353,7 +331,7 @@ export type NotionPage = {
 }
 
 export type ProcessRule = {
-  mode: ProcessMode
+  mode: string
   rules: Rules
 }
 
@@ -363,11 +341,6 @@ export type createDocumentResponse = {
   documents: InitialDocumentDetail[]
 }
 
-export type PrecessRule = {
-  mode: ProcessMode
-  rules: Rules
-}
-
 export type FullDocumentDetail = SimpleDocumentDetail & {
   batch: string
   created_api_request_id: string
@@ -390,8 +363,6 @@ export type FullDocumentDetail = SimpleDocumentDetail & {
   doc_type?: DocType | null | 'others'
   doc_metadata?: DocMetadata | null
   segment_count: number
-  dataset_process_rule: PrecessRule
-  document_process_rule: ProcessRule
   [key: string]: any
 }
 
@@ -428,12 +399,12 @@ export const SEGMENT_STATUS_LIST = ['waiting', 'completed', 'error', 'indexing']
 export type SegmentStatus = typeof SEGMENT_STATUS_LIST[number]
 
 export type SegmentsQuery = {
-  page?: string
+  last_id?: string
   limit: number
   // status?: SegmentStatus
   hit_count_gte?: number
   keyword?: string
-  enabled?: boolean | 'all'
+  enabled?: boolean
 }
 
 export type SegmentDetailModel = {
@@ -458,8 +429,6 @@ export type SegmentDetailModel = {
   error: string | null
   stopped_at: number
   answer?: string
-  child_chunks?: ChildChunkDetail[]
-  updated_at: number
 }
 
 export type SegmentsResponse = {
@@ -467,8 +436,6 @@ export type SegmentsResponse = {
   has_more: boolean
   limit: number
   total: number
-  total_pages: number
-  page: number
 }
 
 export type HitTestingRecord = {
@@ -481,18 +448,10 @@ export type HitTestingRecord = {
   created_at: number
 }
 
-export type HitTestingChildChunk = {
-  id: string
-  content: string
-  position: number
-  score: number
-}
 export type HitTesting = {
   segment: Segment
-  content: Segment
   score: number
   tsne_position: TsnePosition
-  child_chunks?: HitTestingChildChunk[] | null
 }
 
 export type ExternalKnowledgeBaseHitTesting = {
@@ -571,7 +530,11 @@ export type SegmentUpdater = {
   content: string
   answer?: string
   keywords?: string[]
-  regenerate_child_chunks?: boolean
+}
+
+export enum DocForm {
+  TEXT = 'text_model',
+  QA = 'qa_model',
 }
 
 export type ErrorDocsResponse = {
@@ -616,49 +579,3 @@ export const DEFAULT_WEIGHTED_SCORE = {
     keyword: 0.3,
   },
 }
-
-export type ChildChunkType = 'automatic' | 'customized'
-
-export type ChildChunkDetail = {
-  id: string
-  position: number
-  segment_id: string
-  content: string
-  word_count: number
-  created_at: number
-  updated_at: number
-  type: ChildChunkType
-}
-
-export type ChildSegmentsResponse = {
-  data: ChildChunkDetail[]
-  total: number
-  total_pages: number
-  page: number
-  limit: number
-}
-
-export type UpdateDocumentParams = {
-  datasetId: string
-  documentId: string
-}
-
-// Used in api url
-export enum DocumentActionType {
-  enable = 'enable',
-  disable = 'disable',
-  archive = 'archive',
-  unArchive = 'un_archive',
-  delete = 'delete',
-}
-
-export type UpdateDocumentBatchParams = {
-  datasetId: string
-  documentId?: string
-  documentIds?: string[] | string
-}
-
-export type BatchImportResponse = {
-  job_id: string
-  job_status: string
-}