Files
leaudit-platform-frontend/app/hooks/dify-dataset-manager/document-detail.ts
T

228 lines
7.9 KiB
TypeScript

import { message } from 'antd';
import { useCallback, useEffect, useRef, useState } from 'react';
import {
fetchUploadFileInfo,
downloadOriginalFile,
updateDocumentByFile,
fetchIndexingStatus,
} from '~/api/dify-dataset/api/documentApi';
import { fetchSegments } from '~/api/dify-dataset/api/segmentApi';
import type { Segment, IndexingStatus } from '~/api/dify-dataset/type';
import type { Document } from '~/api/dify-dataset/type/documentTypes';
import type { DocumentDetailSegmentationSettings } from '~/types/dify-dataset-manager/document-detail';
import { DEFAULT_DOCUMENT_DETAIL_SETTINGS } from '~/types/dify-dataset-manager/document-detail';
/**
* 文档详情状态管理 Hook
*/
export function useDocumentDetail(datasetId: string, document: Document | null) {
// 分段设置状态
const [settings, setSettings] = useState<DocumentDetailSegmentationSettings>(DEFAULT_DOCUMENT_DETAIL_SETTINGS);
// 预览状态
const [previewSegments, setPreviewSegments] = useState<Segment[]>([]);
const [previewLoading, setPreviewLoading] = useState(false);
const [showPreview, setShowPreview] = useState(false);
// 保存状态
const [saving, setSaving] = useState(false);
// 处理状态(嵌入进度)
const [isProcessing, setIsProcessing] = useState(false);
const [indexingStatus, setIndexingStatus] = useState<IndexingStatus | null>(null);
const pollingTimerRef = useRef<ReturnType<typeof setInterval> | null>(null);
/**
* 停止轮询
*/
const stopPolling = useCallback(() => {
if (pollingTimerRef.current) {
clearInterval(pollingTimerRef.current);
pollingTimerRef.current = null;
}
}, []);
/**
* 轮询索引状态
*/
const pollIndexingStatus = useCallback(async (batch: string) => {
try {
const response = await fetchIndexingStatus(datasetId, batch);
const status = response.data?.[0];
if (status) {
setIndexingStatus(status.indexing_status);
if (status.indexing_status === 'completed') {
// 停止轮询
stopPolling();
setIsProcessing(false);
message.success('文档处理完成');
// 刷新分段预览
setPreviewLoading(true);
try {
const segmentResponse = await fetchSegments(datasetId, document?.id || '', 1, 50);
setPreviewSegments(segmentResponse.data || []);
setShowPreview(true);
} catch (err) {
console.error('刷新分段失败:', err);
} finally {
setPreviewLoading(false);
}
} else if (status.indexing_status === 'error') {
// 停止轮询
stopPolling();
setIsProcessing(false);
message.error(status.error || '处理失败');
}
}
} catch (err) {
console.error('获取索引状态失败:', err);
}
}, [datasetId, document?.id, stopPolling]);
/**
* 开始轮询
*/
const startPolling = useCallback((batch: string) => {
stopPolling();
pollingTimerRef.current = setInterval(() => {
pollIndexingStatus(batch);
}, 2000);
// 立即执行一次
pollIndexingStatus(batch);
}, [stopPolling, pollIndexingStatus]);
// 当文档变化时重置设置
useEffect(() => {
if (document) {
// 可以从文档中读取已有的设置,这里使用默认值
setSettings(DEFAULT_DOCUMENT_DETAIL_SETTINGS);
setPreviewSegments([]);
setShowPreview(false);
setIsProcessing(false);
setIndexingStatus(null);
stopPolling();
}
}, [document?.id, stopPolling]);
// 组件卸载时清理定时器
useEffect(() => {
return () => {
stopPolling();
};
}, [stopPolling]);
/**
* 更新设置
*/
const updateSettings = useCallback((key: keyof DocumentDetailSegmentationSettings, value: any) => {
setSettings(prev => ({ ...prev, [key]: value }));
}, []);
/**
* 重置设置
*/
const handleReset = useCallback(() => {
setSettings(DEFAULT_DOCUMENT_DETAIL_SETTINGS);
setPreviewSegments([]);
setShowPreview(false);
}, []);
/**
* 预览分段
*/
const handlePreview = useCallback(async () => {
if (!document) return;
setPreviewLoading(true);
setShowPreview(true);
try {
// 获取当前文档的分段作为预览
const response = await fetchSegments(datasetId, document.id, 1, 50);
setPreviewSegments(response.data || []);
if (response.data?.length === 0) {
message.info('该文档暂无分段数据');
}
} catch (err: any) {
console.error('预览分段失败:', err);
message.error(err.message || '预览失败');
} finally {
setPreviewLoading(false);
}
}, [datasetId, document]);
/**
* 保存并处理
* 流程:获取原始文件 → 下载 → 用新参数重新上传 → 轮询嵌入状态
*/
const handleSaveAndProcess = useCallback(async () => {
if (!document) return;
setIsProcessing(true);
setSaving(true);
setIndexingStatus('waiting');
try {
// 1. 获取原始文件信息
message.loading({ content: '正在获取原始文件信息...', key: 'save-process' });
const uploadFileInfo = await fetchUploadFileInfo(datasetId, document.id);
// 2. 下载原始文件(通过代理路由)
message.loading({ content: '正在下载原始文件...', key: 'save-process' });
const file = await downloadOriginalFile(uploadFileInfo);
// 3. 用新参数重新上传
message.loading({ content: '正在应用新配置并重新处理...', key: 'save-process' });
const result = await updateDocumentByFile(datasetId, document.id, file, {
indexing_technique: settings.indexingTechnique,
process_rule: {
mode: 'custom',
rules: {
pre_processing_rules: [
{ id: 'remove_extra_spaces', enabled: settings.removeExtraSpaces },
{ id: 'remove_urls_emails', enabled: settings.removeUrlsEmails },
],
segmentation: {
separator: settings.separator.replace(/\\n/g, '\n'),
max_tokens: settings.maxTokens,
chunk_overlap: settings.chunkOverlap,
},
},
},
});
message.success({ content: '文档正在处理中...', key: 'save-process' });
// 4. 开始轮询嵌入状态
startPolling(result.batch);
} catch (err: any) {
console.error('保存设置失败:', err);
message.error({ content: err.message || '保存失败', key: 'save-process' });
setIsProcessing(false);
setIndexingStatus(null);
} finally {
setSaving(false);
}
}, [datasetId, document, settings, startPolling]);
return {
// 状态
settings,
previewSegments,
previewLoading,
showPreview,
saving,
isProcessing,
indexingStatus,
// 方法
updateSettings,
handleReset,
handlePreview,
handleSaveAndProcess,
};
}
export type UseDocumentDetailReturn = ReturnType<typeof useDocumentDetail>;