mirror of
https://github.com/labring/FastGPT.git
synced 2025-07-27 08:25:07 +00:00

* Feat: Images dataset collection (#4941) * New pic (#4858) * 更新数据集相关类型,添加图像文件ID和预览URL支持;优化数据集导入功能,新增图像数据集处理组件;修复部分国际化文本;更新文件上传逻辑以支持新功能。 * 与原先代码的差别 * 新增 V4.9.10 更新说明,支持 PG 设置`systemEnv.hnswMaxScanTuples`参数,优化 LLM stream 调用超时,修复全文检索多知识库排序问题。同时更新数据集索引,移除 datasetId 字段以简化查询。 * 更换成fileId_image逻辑,并增加训练队列匹配的逻辑 * 新增图片集合判断逻辑,优化预览URL生成流程,确保仅在数据集为图片集合时生成预览URL,并添加相关日志输出以便调试。 * Refactor Docker Compose configuration to comment out exposed ports for production environments, update image versions for pgvector, fastgpt, and mcp_server, and enhance Redis service with a health check. Additionally, standardize dataset collection labels in constants and improve internationalization strings across multiple languages. * Enhance TrainingStates component by adding internationalization support for the imageParse training mode and update defaultCounts to include imageParse mode in trainingDetail API. * Enhance dataset import context by adding additional steps for image dataset import process and improve internationalization strings for modal buttons in the useEditTitle hook. * Update DatasetImportContext to conditionally render MyStep component based on data source type, improving the import process for non-image datasets. * Refactor image dataset handling by improving internationalization strings, enhancing error messages, and streamlining the preview URL generation process. * 图片上传到新建的 dataset_collection_images 表,逻辑跟随更改 * 修改了除了controller的其他部分问题 * 把图片数据集的逻辑整合到controller里面 * 补充i18n * 补充i18n * resolve评论:主要是上传逻辑的更改和组件复用 * 图片名称的图标显示 * 修改编译报错的命名问题 * 删除不需要的collectionid部分 * 多余文件的处理和改动一个删除按钮 * 除了loading和统一的imageId,其他都resolve掉的 * 处理图标报错 * 复用了MyPhotoView并采用全部替换的方式将imageFileId变成imageId * 去除不必要文件修改 * 报错和字段修改 * 增加上传成功后删除临时文件的逻辑以及回退一些修改 * 删除path字段,将图片保存到gridfs内,并修改增删等操作的代码 * 修正编译错误 --------- Co-authored-by: archer <545436317@qq.com> * perf: image dataset * feat: insert image * perf: image icon * fix: training state --------- Co-authored-by: Zhuangzai fa <143257420+ctrlz526@users.noreply.github.com> * fix: ts (#4948) * Thirddatasetmd (#4942) * add thirddataset.md * fix thirddataset.md * fix * delete wrong png --------- Co-authored-by: dreamer6680 <146868355@qq.com> * perf: api dataset code * perf: log * add secondary.tsx (#4946) * add secondary.tsx * fix --------- Co-authored-by: dreamer6680 <146868355@qq.com> * perf: multiple menu * perf: i18n * feat: parse queue (#4960) * feat: parse queue * feat: sync parse queue * fix thirddataset.md (#4962) * fix thirddataset-4.png (#4963) * feat: Dataset template import (#4934) * 模版导入部分除了文档还没写 * 修复模版导入的 build 错误 * Document production * compress pictures * Change some constants to variables --------- Co-authored-by: Archer <545436317@qq.com> * perf: template import * doc * llm pargraph * bocha tool * fix: del collection --------- Co-authored-by: Zhuangzai fa <143257420+ctrlz526@users.noreply.github.com> Co-authored-by: dreamer6680 <1468683855@qq.com> Co-authored-by: dreamer6680 <146868355@qq.com>
182 lines
4.6 KiB
TypeScript
182 lines
4.6 KiB
TypeScript
import { retryFn } from '@fastgpt/global/common/system/utils';
|
||
import { connectionMongo } from '../../mongo';
|
||
import { MongoRawTextBufferSchema, bucketName } from './schema';
|
||
import { addLog } from '../../system/log';
|
||
import { setCron } from '../../system/cron';
|
||
import { checkTimerLock } from '../../system/timerLock/utils';
|
||
import { TimerIdEnum } from '../../system/timerLock/constants';
|
||
|
||
const getGridBucket = () => {
|
||
return new connectionMongo.mongo.GridFSBucket(connectionMongo.connection.db!, {
|
||
bucketName: bucketName
|
||
});
|
||
};
|
||
|
||
export const addRawTextBuffer = async ({
|
||
sourceId,
|
||
sourceName,
|
||
text,
|
||
expiredTime
|
||
}: {
|
||
sourceId: string;
|
||
sourceName: string;
|
||
text: string;
|
||
expiredTime: Date;
|
||
}) => {
|
||
const gridBucket = getGridBucket();
|
||
const metadata = {
|
||
sourceId,
|
||
sourceName,
|
||
expiredTime
|
||
};
|
||
|
||
const buffer = Buffer.from(text);
|
||
|
||
const fileSize = buffer.length;
|
||
// 单块大小:尽可能大,但不超过 14MB,不小于128KB
|
||
const chunkSizeBytes = (() => {
|
||
// 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
|
||
const idealChunkSize = Math.min(Math.ceil(fileSize / 10), 14 * 1024 * 1024);
|
||
|
||
// 确保块大小至少为128KB
|
||
const minChunkSize = 128 * 1024; // 128KB
|
||
|
||
// 取理想块大小和最小块大小中的较大值
|
||
let chunkSize = Math.max(idealChunkSize, minChunkSize);
|
||
|
||
// 将块大小向上取整到最接近的64KB的倍数,使其更整齐
|
||
chunkSize = Math.ceil(chunkSize / (64 * 1024)) * (64 * 1024);
|
||
|
||
return chunkSize;
|
||
})();
|
||
|
||
const uploadStream = gridBucket.openUploadStream(sourceId, {
|
||
metadata,
|
||
chunkSizeBytes
|
||
});
|
||
|
||
return retryFn(async () => {
|
||
return new Promise((resolve, reject) => {
|
||
uploadStream.end(buffer);
|
||
uploadStream.on('finish', () => {
|
||
resolve(uploadStream.id);
|
||
});
|
||
uploadStream.on('error', (error) => {
|
||
addLog.error('addRawTextBuffer error', error);
|
||
resolve('');
|
||
});
|
||
});
|
||
});
|
||
};
|
||
|
||
export const getRawTextBuffer = async (sourceId: string) => {
|
||
const gridBucket = getGridBucket();
|
||
|
||
return retryFn(async () => {
|
||
const bufferData = await MongoRawTextBufferSchema.findOne(
|
||
{
|
||
'metadata.sourceId': sourceId
|
||
},
|
||
'_id metadata'
|
||
).lean();
|
||
if (!bufferData) {
|
||
return null;
|
||
}
|
||
|
||
// Read file content
|
||
const downloadStream = gridBucket.openDownloadStream(bufferData._id);
|
||
const chunks: Buffer[] = [];
|
||
|
||
return new Promise<{
|
||
text: string;
|
||
sourceName: string;
|
||
} | null>((resolve, reject) => {
|
||
downloadStream.on('data', (chunk) => {
|
||
chunks.push(chunk);
|
||
});
|
||
|
||
downloadStream.on('end', () => {
|
||
const buffer = Buffer.concat(chunks);
|
||
const text = buffer.toString('utf8');
|
||
resolve({
|
||
text,
|
||
sourceName: bufferData.metadata?.sourceName || ''
|
||
});
|
||
});
|
||
|
||
downloadStream.on('error', (error) => {
|
||
addLog.error('getRawTextBuffer error', error);
|
||
resolve(null);
|
||
});
|
||
});
|
||
});
|
||
};
|
||
|
||
export const deleteRawTextBuffer = async (sourceId: string): Promise<boolean> => {
|
||
const gridBucket = getGridBucket();
|
||
|
||
return retryFn(async () => {
|
||
const buffer = await MongoRawTextBufferSchema.findOne({ 'metadata.sourceId': sourceId });
|
||
if (!buffer) {
|
||
return false;
|
||
}
|
||
|
||
await gridBucket.delete(buffer._id);
|
||
return true;
|
||
});
|
||
};
|
||
|
||
export const updateRawTextBufferExpiredTime = async ({
|
||
sourceId,
|
||
expiredTime
|
||
}: {
|
||
sourceId: string;
|
||
expiredTime: Date;
|
||
}) => {
|
||
return retryFn(async () => {
|
||
return MongoRawTextBufferSchema.updateOne(
|
||
{ 'metadata.sourceId': sourceId },
|
||
{ $set: { 'metadata.expiredTime': expiredTime } }
|
||
);
|
||
});
|
||
};
|
||
|
||
export const clearExpiredRawTextBufferCron = async () => {
|
||
const gridBucket = getGridBucket();
|
||
|
||
const clearExpiredRawTextBuffer = async () => {
|
||
addLog.debug('Clear expired raw text buffer start');
|
||
|
||
const data = await MongoRawTextBufferSchema.find(
|
||
{
|
||
'metadata.expiredTime': { $lt: new Date() }
|
||
},
|
||
'_id'
|
||
).lean();
|
||
|
||
for (const item of data) {
|
||
try {
|
||
await gridBucket.delete(item._id);
|
||
} catch (error) {
|
||
addLog.error('Delete expired raw text buffer error', error);
|
||
}
|
||
}
|
||
addLog.debug('Clear expired raw text buffer end');
|
||
};
|
||
|
||
setCron('*/10 * * * *', async () => {
|
||
if (
|
||
await checkTimerLock({
|
||
timerId: TimerIdEnum.clearExpiredRawTextBuffer,
|
||
lockMinuted: 9
|
||
})
|
||
) {
|
||
try {
|
||
await clearExpiredRawTextBuffer();
|
||
} catch (error) {
|
||
addLog.error('clearExpiredRawTextBufferCron error', error);
|
||
}
|
||
}
|
||
});
|
||
};
|