|
| 1 | +import { and, eq, inArray } from 'drizzle-orm/expressions'; |
| 2 | +import pMap from 'p-map'; |
| 3 | + |
| 4 | +import * as EXPORT_TABLES from '@/database/schemas'; |
| 5 | +import { LobeChatDatabase } from '@/database/type'; |
| 6 | + |
| 7 | +interface BaseTableConfig { |
| 8 | + table: keyof typeof EXPORT_TABLES; |
| 9 | + type: 'base'; |
| 10 | + userField?: string; |
| 11 | +} |
| 12 | + |
| 13 | +interface RelationTableConfig { |
| 14 | + relations: { |
| 15 | + field: string; |
| 16 | + sourceField?: string; |
| 17 | + sourceTable: string; |
| 18 | + }[]; |
| 19 | + table: keyof typeof EXPORT_TABLES; |
| 20 | + type: 'relation'; |
| 21 | +} |
| 22 | + |
| 23 | +// 配置拆分为基础表和关联表 |
| 24 | + |
| 25 | +export const DATA_EXPORT_CONFIG = { |
| 26 | + // 1. 基础表 |
| 27 | + baseTables: [ |
| 28 | + { table: 'users', userField: 'id' }, |
| 29 | + { table: 'userSettings', userField: 'id' }, |
| 30 | + { table: 'userInstalledPlugins' }, |
| 31 | + { table: 'agents' }, |
| 32 | + { table: 'sessionGroups' }, |
| 33 | + { table: 'sessions' }, |
| 34 | + { table: 'topics' }, |
| 35 | + { table: 'threads' }, |
| 36 | + { table: 'messages' }, |
| 37 | + { table: 'files' }, |
| 38 | + { table: 'knowledgeBases' }, |
| 39 | + { table: 'agentsKnowledgeBases' }, |
| 40 | + { table: 'aiProviders' }, |
| 41 | + { table: 'aiModels' }, |
| 42 | + { table: 'asyncTasks' }, |
| 43 | + { table: 'chunks' }, |
| 44 | + { table: 'embeddings' }, |
| 45 | + ] as BaseTableConfig[], |
| 46 | + // 2. 关联表 |
| 47 | + relationTables: [ |
| 48 | + { |
| 49 | + relations: [ |
| 50 | + { sourceField: 'id', sourceTable: 'agents', field: 'agentId' }, |
| 51 | + { sourceField: 'sessionId', sourceTable: 'sessions' }, |
| 52 | + ], |
| 53 | + table: 'agentsToSessions', |
| 54 | + }, |
| 55 | + // { |
| 56 | + // relations: [ |
| 57 | + // { sourceField: 'agentId', sourceTable: 'agents' }, |
| 58 | + // { sourceField: 'fileId', sourceTable: 'files' }, |
| 59 | + // ], |
| 60 | + // table: 'agentsFiles', |
| 61 | + // }, |
| 62 | + // { |
| 63 | + // relations: [{ field: 'sessionId', sourceTable: 'sessions' }], |
| 64 | + // table: 'filesToSessions', |
| 65 | + // }, |
| 66 | + // { |
| 67 | + // relations: [{ field: 'id', sourceTable: 'chunks' }], |
| 68 | + // table: 'fileChunks', |
| 69 | + // }, |
| 70 | + { |
| 71 | + relations: [{ field: 'id', sourceTable: 'messages' }], |
| 72 | + table: 'messagePlugins', |
| 73 | + }, |
| 74 | + { |
| 75 | + relations: [{ field: 'id', sourceTable: 'messages' }], |
| 76 | + table: 'messageTTS', |
| 77 | + }, |
| 78 | + { |
| 79 | + relations: [{ field: 'id', sourceTable: 'messages' }], |
| 80 | + table: 'messageTranslates', |
| 81 | + }, |
| 82 | + { |
| 83 | + relations: [ |
| 84 | + { field: 'messageId', sourceTable: 'messages' }, |
| 85 | + { field: 'fileId', sourceTable: 'files' }, |
| 86 | + ], |
| 87 | + table: 'messagesFiles', |
| 88 | + }, |
| 89 | + { |
| 90 | + relations: [{ field: 'messageId', sourceTable: 'messages' }], |
| 91 | + table: 'messageQueries', |
| 92 | + }, |
| 93 | + { |
| 94 | + relations: [ |
| 95 | + { field: 'messageId', sourceTable: 'messages' }, |
| 96 | + { field: 'chunkId', sourceTable: 'chunks' }, |
| 97 | + ], |
| 98 | + table: 'messageQueryChunks', |
| 99 | + }, |
| 100 | + { |
| 101 | + relations: [ |
| 102 | + { field: 'messageId', sourceTable: 'messages' }, |
| 103 | + { field: 'chunkId', sourceTable: 'chunks' }, |
| 104 | + ], |
| 105 | + table: 'messageChunks', |
| 106 | + }, |
| 107 | + { |
| 108 | + relations: [ |
| 109 | + { field: 'knowledgeBaseId', sourceTable: 'knowledgeBases' }, |
| 110 | + { field: 'fileId', sourceTable: 'files' }, |
| 111 | + ], |
| 112 | + table: 'knowledgeBaseFiles', |
| 113 | + }, |
| 114 | + ] as RelationTableConfig[], |
| 115 | +}; |
| 116 | + |
| 117 | +export class DataExporterRepos { |
| 118 | + constructor( |
| 119 | + private db: LobeChatDatabase, |
| 120 | + private userId: string, |
| 121 | + ) {} |
| 122 | + |
| 123 | + private removeUserId(data: any[]) { |
| 124 | + return data.map((item) => { |
| 125 | + // eslint-disable-next-line @typescript-eslint/no-unused-vars |
| 126 | + const { userId: _, ...rest } = item; |
| 127 | + return rest; |
| 128 | + }); |
| 129 | + } |
| 130 | + |
| 131 | + private async queryTable(config: RelationTableConfig, existingData: Record<string, any[]>) { |
| 132 | + const { table } = config; |
| 133 | + const tableObj = EXPORT_TABLES[table]; |
| 134 | + if (!tableObj) throw new Error(`Table ${table} not found`); |
| 135 | + |
| 136 | + try { |
| 137 | + let where; |
| 138 | + |
| 139 | + const conditions = config.relations.map((relation) => { |
| 140 | + const sourceData = existingData[relation.sourceTable] || []; |
| 141 | + |
| 142 | + const sourceIds = sourceData.map((item) => item[relation.sourceField || 'id']); |
| 143 | + console.log(sourceIds); |
| 144 | + return inArray(tableObj[relation.field], sourceIds); |
| 145 | + }); |
| 146 | + |
| 147 | + where = conditions.length === 1 ? conditions[0] : and(...conditions); |
| 148 | + |
| 149 | + const result = await this.db.query[table].findMany({ where }); |
| 150 | + |
| 151 | + // 只对使用 userId 查询的表移除 userId 字段 |
| 152 | + console.log(`Successfully exported table: ${table}, count: ${result.length}`); |
| 153 | + return config.relations ? result : this.removeUserId(result); |
| 154 | + } catch (error) { |
| 155 | + console.error(`Error querying table ${table}:`, error); |
| 156 | + return []; |
| 157 | + } |
| 158 | + } |
| 159 | + |
| 160 | + private async queryBaseTables(config: BaseTableConfig) { |
| 161 | + const { table } = config; |
| 162 | + const tableObj = EXPORT_TABLES[table]; |
| 163 | + if (!tableObj) throw new Error(`Table ${table} not found`); |
| 164 | + |
| 165 | + try { |
| 166 | + // 如果有关联配置,使用关联查询 |
| 167 | + |
| 168 | + // 默认使用 userId 查询,特殊情况使用 userField |
| 169 | + const userField = config.userField || 'userId'; |
| 170 | + const where = eq(tableObj[userField], this.userId); |
| 171 | + |
| 172 | + const result = await this.db.query[table].findMany({ where }); |
| 173 | + |
| 174 | + // 只对使用 userId 查询的表移除 userId 字段 |
| 175 | + console.log(`Successfully exported table: ${table}, count: ${result.length}`); |
| 176 | + return this.removeUserId(result); |
| 177 | + } catch (error) { |
| 178 | + console.error(`Error querying table ${table}:`, error); |
| 179 | + return []; |
| 180 | + } |
| 181 | + } |
| 182 | + |
| 183 | + async export(concurrency = 10) { |
| 184 | + const result: Record<string, any[]> = {}; |
| 185 | + |
| 186 | + // 1. 首先并发查询所有基础表 |
| 187 | + console.log('Querying base tables...'); |
| 188 | + const baseResults = await pMap( |
| 189 | + DATA_EXPORT_CONFIG.baseTables, |
| 190 | + async (config) => ({ data: await this.queryBaseTables(config), table: config.table }), |
| 191 | + { concurrency }, |
| 192 | + ); |
| 193 | + |
| 194 | + // 更新结果集 |
| 195 | + baseResults.forEach(({ table, data }) => { |
| 196 | + result[table] = data; |
| 197 | + }); |
| 198 | + |
| 199 | + console.log('baseResults:', baseResults); |
| 200 | + // 2. 然后并发查询所有关联表 |
| 201 | + |
| 202 | + console.log('Querying relation tables...'); |
| 203 | + const relationResults = await pMap( |
| 204 | + DATA_EXPORT_CONFIG.relationTables, |
| 205 | + async (config) => ({ |
| 206 | + data: await this.queryTable(config, result), |
| 207 | + table: config.table, |
| 208 | + }), |
| 209 | + { concurrency }, |
| 210 | + ); |
| 211 | + |
| 212 | + // 更新结果集 |
| 213 | + relationResults.forEach(({ table, data }) => { |
| 214 | + result[table] = data; |
| 215 | + }); |
| 216 | + |
| 217 | + return result; |
| 218 | + } |
| 219 | +} |
0 commit comments