|
| 1 | +import { retryFn } from '@fastgpt/global/common/system/utils'; |
| 2 | +import { connectionMongo } from '../../mongo'; |
| 3 | +import { MongoRawTextBufferSchema, bucketName } from './schema'; |
| 4 | +import { addLog } from '../../system/log'; |
| 5 | + |
| 6 | +const getGridBucket = () => { |
| 7 | + return new connectionMongo.mongo.GridFSBucket(connectionMongo.connection.db!, { |
| 8 | + bucketName: bucketName |
| 9 | + }); |
| 10 | +}; |
| 11 | + |
| 12 | +export const addRawTextBuffer = async ({ |
| 13 | + sourceId, |
| 14 | + sourceName, |
| 15 | + text, |
| 16 | + expiredTime |
| 17 | +}: { |
| 18 | + sourceId: string; |
| 19 | + sourceName: string; |
| 20 | + text: string; |
| 21 | + expiredTime: Date; |
| 22 | +}) => { |
| 23 | + const gridBucket = getGridBucket(); |
| 24 | + const metadata = { |
| 25 | + sourceId, |
| 26 | + sourceName, |
| 27 | + expiredTime |
| 28 | + }; |
| 29 | + |
| 30 | + const buffer = Buffer.from(text); |
| 31 | + |
| 32 | + const fileSize = buffer.length; |
| 33 | + // 单块大小:尽可能大,但不超过 14MB,不小于128KB |
| 34 | + const chunkSizeBytes = (() => { |
| 35 | + // 计算理想块大小:文件大小 ÷ 目标块数(10)。 并且每个块需要小于 14MB |
| 36 | + const idealChunkSize = Math.min(Math.ceil(fileSize / 10), 14 * 1024 * 1024); |
| 37 | + |
| 38 | + // 确保块大小至少为128KB |
| 39 | + const minChunkSize = 128 * 1024; // 128KB |
| 40 | + |
| 41 | + // 取理想块大小和最小块大小中的较大值 |
| 42 | + let chunkSize = Math.max(idealChunkSize, minChunkSize); |
| 43 | + |
| 44 | + // 将块大小向上取整到最接近的64KB的倍数,使其更整齐 |
| 45 | + chunkSize = Math.ceil(chunkSize / (64 * 1024)) * (64 * 1024); |
| 46 | + |
| 47 | + return chunkSize; |
| 48 | + })(); |
| 49 | + |
| 50 | + const uploadStream = gridBucket.openUploadStream(sourceId, { |
| 51 | + metadata, |
| 52 | + chunkSizeBytes |
| 53 | + }); |
| 54 | + |
| 55 | + return retryFn(async () => { |
| 56 | + return new Promise((resolve, reject) => { |
| 57 | + uploadStream.end(buffer); |
| 58 | + uploadStream.on('finish', () => { |
| 59 | + resolve(uploadStream.id); |
| 60 | + }); |
| 61 | + uploadStream.on('error', (error) => { |
| 62 | + addLog.error('addRawTextBuffer error', error); |
| 63 | + resolve(''); |
| 64 | + }); |
| 65 | + }); |
| 66 | + }); |
| 67 | +}; |
| 68 | + |
| 69 | +export const getRawTextBuffer = async (sourceId: string) => { |
| 70 | + const gridBucket = getGridBucket(); |
| 71 | + |
| 72 | + return retryFn(async () => { |
| 73 | + const bufferData = await MongoRawTextBufferSchema.findOne( |
| 74 | + { |
| 75 | + 'metadata.sourceId': sourceId |
| 76 | + }, |
| 77 | + '_id metadata' |
| 78 | + ).lean(); |
| 79 | + if (!bufferData) { |
| 80 | + return null; |
| 81 | + } |
| 82 | + |
| 83 | + // Read file content |
| 84 | + const downloadStream = gridBucket.openDownloadStream(bufferData._id); |
| 85 | + const chunks: Buffer[] = []; |
| 86 | + |
| 87 | + return new Promise<{ |
| 88 | + text: string; |
| 89 | + sourceName: string; |
| 90 | + } | null>((resolve, reject) => { |
| 91 | + downloadStream.on('data', (chunk) => { |
| 92 | + chunks.push(chunk); |
| 93 | + }); |
| 94 | + |
| 95 | + downloadStream.on('end', () => { |
| 96 | + const buffer = Buffer.concat(chunks); |
| 97 | + const text = buffer.toString('utf8'); |
| 98 | + resolve({ |
| 99 | + text, |
| 100 | + sourceName: bufferData.metadata?.sourceName || '' |
| 101 | + }); |
| 102 | + }); |
| 103 | + |
| 104 | + downloadStream.on('error', (error) => { |
| 105 | + addLog.error('getRawTextBuffer error', error); |
| 106 | + resolve(null); |
| 107 | + }); |
| 108 | + }); |
| 109 | + }); |
| 110 | +}; |
| 111 | + |
| 112 | +export const deleteRawTextBuffer = async (sourceId: string): Promise<boolean> => { |
| 113 | + const gridBucket = getGridBucket(); |
| 114 | + |
| 115 | + return retryFn(async () => { |
| 116 | + const buffer = await MongoRawTextBufferSchema.findOne({ 'metadata.sourceId': sourceId }); |
| 117 | + if (!buffer) { |
| 118 | + return false; |
| 119 | + } |
| 120 | + |
| 121 | + await gridBucket.delete(buffer._id); |
| 122 | + return true; |
| 123 | + }); |
| 124 | +}; |
| 125 | + |
| 126 | +export const updateRawTextBufferExpiredTime = async ({ |
| 127 | + sourceId, |
| 128 | + expiredTime |
| 129 | +}: { |
| 130 | + sourceId: string; |
| 131 | + expiredTime: Date; |
| 132 | +}) => { |
| 133 | + return retryFn(async () => { |
| 134 | + return MongoRawTextBufferSchema.updateOne( |
| 135 | + { 'metadata.sourceId': sourceId }, |
| 136 | + { $set: { 'metadata.expiredTime': expiredTime } } |
| 137 | + ); |
| 138 | + }); |
| 139 | +}; |
0 commit comments