feat：完成上传文档时调整嵌入参数模块初版
This commit is contained in:
PingChuan
@@ -11,7 +11,6 @@ import {
|
||||
message,
|
||||
Empty,
|
||||
Spin,
|
||||
Upload,
|
||||
} from 'antd';
|
||||
import {
|
||||
SearchOutlined,
|
||||
@@ -28,7 +27,8 @@ import {
|
||||
} from '@ant-design/icons';
|
||||
import type { ColumnsType } from 'antd/es/table';
|
||||
import type { Document, IndexingStatus } from '~/api/dify-dataset/type/documentTypes';
|
||||
import { deleteDocument, toggleDocumentStatus, uploadDocument } from '~/api/dify-dataset/api/documentApi';
|
||||
import { deleteDocument, toggleDocumentStatus } from '~/api/dify-dataset/api/documentApi';
|
||||
import DocumentUpload from './document-upload';
|
||||
import '../../styles/components/dify-dataset-manager/index.css';
|
||||
|
||||
interface DocumentListProps {
|
||||
@@ -63,9 +63,11 @@ export default function DocumentList({
|
||||
onViewDocument,
|
||||
}: DocumentListProps) {
|
||||
const [searchValue, setSearchValue] = useState('');
|
||||
const [uploading, setUploading] = useState(false);
|
||||
const [deletingId, setDeletingId] = useState<string | null>(null);
|
||||
|
||||
// 显示上传页面的状态
|
||||
const [showUploadPage, setShowUploadPage] = useState(false);
|
||||
|
||||
/**
|
||||
* 获取状态标签配置
|
||||
*/
|
||||
@@ -141,29 +143,29 @@ export default function DocumentList({
|
||||
};
|
||||
|
||||
/**
|
||||
* 处理文件上传
|
||||
* 点击上传按钮，显示上传页面
|
||||
*/
|
||||
const handleUpload = async (file: File) => {
|
||||
const handleUploadClick = () => {
|
||||
if (!datasetId) {
|
||||
message.error('请先选择知识库');
|
||||
return false;
|
||||
return;
|
||||
}
|
||||
setShowUploadPage(true);
|
||||
};
|
||||
|
||||
setUploading(true);
|
||||
try {
|
||||
await uploadDocument(datasetId, file, (percent) => {
|
||||
console.log('上传进度:', percent);
|
||||
});
|
||||
message.success('上传成功，正在处理...');
|
||||
onRefresh();
|
||||
} catch (err: any) {
|
||||
console.error('上传文件失败:', err);
|
||||
message.error(err.message || '上传失败');
|
||||
} finally {
|
||||
setUploading(false);
|
||||
}
|
||||
/**
|
||||
* 关闭上传页面
|
||||
*/
|
||||
const handleUploadClose = () => {
|
||||
setShowUploadPage(false);
|
||||
};
|
||||
|
||||
return false;
|
||||
/**
|
||||
* 上传成功回调
|
||||
*/
|
||||
const handleUploadSuccess = () => {
|
||||
setShowUploadPage(false);
|
||||
onRefresh();
|
||||
};
|
||||
|
||||
// 过滤文档
|
||||
@@ -271,115 +273,114 @@ export default function DocumentList({
|
||||
];
|
||||
|
||||
return (
|
||||
<div className="document-list-page">
|
||||
{/* 页面头部 */}
|
||||
<div className="page-header">
|
||||
<div className="header-left">
|
||||
<h1>文档</h1>
|
||||
{/* <p className="page-description">
|
||||
知识库的所有文件都在这里显示，整个知识库都可以被接到 Dify 引用或通过 Chat 插件进行索引。
|
||||
</p> */}
|
||||
</div>
|
||||
<div className="header-actions">
|
||||
<Tooltip title="刷新">
|
||||
<Button
|
||||
icon={<ReloadOutlined />}
|
||||
onClick={onRefresh}
|
||||
loading={loading}
|
||||
/>
|
||||
</Tooltip>
|
||||
<Upload
|
||||
beforeUpload={handleUpload}
|
||||
showUploadList={false}
|
||||
accept=".txt,.md,.pdf,.docx,.doc,.csv,.xlsx,.xls"
|
||||
disabled={!datasetId}
|
||||
>
|
||||
<Button
|
||||
type="primary"
|
||||
icon={<CloudUploadOutlined />}
|
||||
loading={uploading}
|
||||
disabled={!datasetId}
|
||||
>
|
||||
添加文件
|
||||
</Button>
|
||||
</Upload>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 搜索栏 */}
|
||||
<div className="document-search-bar">
|
||||
<Input
|
||||
placeholder="搜索文档..."
|
||||
prefix={<SearchOutlined />}
|
||||
value={searchValue}
|
||||
onChange={(e) => setSearchValue(e.target.value)}
|
||||
allowClear
|
||||
style={{ width: 280 }}
|
||||
<>
|
||||
{/* 上传页面 */}
|
||||
{showUploadPage ? (
|
||||
<DocumentUpload
|
||||
datasetId={datasetId}
|
||||
onClose={handleUploadClose}
|
||||
onSuccess={handleUploadSuccess}
|
||||
/>
|
||||
</div>
|
||||
|
||||
{/* 文档表格 */}
|
||||
<div className="document-table-wrapper">
|
||||
{loading && documents.length === 0 ? (
|
||||
<div className="loading-state">
|
||||
<Spin size="large" />
|
||||
<div className="loading-text">加载中...</div>
|
||||
) : (
|
||||
<div className="document-list-page">
|
||||
{/* 页面头部 */}
|
||||
<div className="page-header">
|
||||
<div className="header-left">
|
||||
<h1>文档</h1>
|
||||
</div>
|
||||
<div className="header-actions">
|
||||
<Tooltip title="刷新">
|
||||
<Button
|
||||
icon={<ReloadOutlined />}
|
||||
onClick={onRefresh}
|
||||
loading={loading}
|
||||
/>
|
||||
</Tooltip>
|
||||
<Button
|
||||
type="primary"
|
||||
icon={<CloudUploadOutlined />}
|
||||
onClick={handleUploadClick}
|
||||
disabled={!datasetId}
|
||||
>
|
||||
添加文件
|
||||
</Button>
|
||||
</div>
|
||||
</div>
|
||||
) : filteredDocuments.length === 0 ? (
|
||||
<div className="empty-state">
|
||||
<Empty description={searchValue ? '未找到匹配的文档' : '暂无文档'}>
|
||||
{!searchValue && (
|
||||
<Upload
|
||||
beforeUpload={handleUpload}
|
||||
showUploadList={false}
|
||||
accept=".txt,.md,.pdf,.docx,.doc,.csv,.xlsx,.xls"
|
||||
|
||||
{/* 搜索栏 */}
|
||||
<div className="document-search-bar">
|
||||
<Input
|
||||
placeholder="搜索文档..."
|
||||
prefix={<SearchOutlined />}
|
||||
value={searchValue}
|
||||
onChange={(e) => setSearchValue(e.target.value)}
|
||||
allowClear
|
||||
style={{ width: 280 }}
|
||||
/>
|
||||
</div>
|
||||
|
||||
{/* 文档表格 */}
|
||||
<div className="document-table-wrapper">
|
||||
{loading && documents.length === 0 ? (
|
||||
<div className="loading-state">
|
||||
<Spin size="large" />
|
||||
<div className="loading-text">加载中...</div>
|
||||
</div>
|
||||
) : filteredDocuments.length === 0 ? (
|
||||
<div className="empty-state">
|
||||
<Empty description={searchValue ? '未找到匹配的文档' : '暂无文档'}>
|
||||
{!searchValue && (
|
||||
<Button
|
||||
type="primary"
|
||||
icon={<CloudUploadOutlined />}
|
||||
onClick={handleUploadClick}
|
||||
>
|
||||
上传第一个文档
|
||||
</Button>
|
||||
)}
|
||||
</Empty>
|
||||
</div>
|
||||
) : (
|
||||
<Table
|
||||
className="document-table"
|
||||
columns={columns}
|
||||
dataSource={filteredDocuments}
|
||||
rowKey="id"
|
||||
loading={loading}
|
||||
pagination={false}
|
||||
size="small"
|
||||
scroll={{ x: 'max-content' }}
|
||||
/>
|
||||
)}
|
||||
</div>
|
||||
|
||||
{/* 底部分页器 */}
|
||||
{filteredDocuments.length > 0 && (
|
||||
<div className="document-pagination">
|
||||
<span className="pagination-total">共 {total} 条</span>
|
||||
<div className="pagination-controls">
|
||||
<Button
|
||||
size="small"
|
||||
disabled={page <= 1}
|
||||
onClick={() => onPageChange(page - 1)}
|
||||
>
|
||||
<Button type="primary" icon={<CloudUploadOutlined />}>
|
||||
上传第一个文档
|
||||
</Button>
|
||||
</Upload>
|
||||
)}
|
||||
</Empty>
|
||||
</div>
|
||||
) : (
|
||||
<Table
|
||||
className="document-table"
|
||||
columns={columns}
|
||||
dataSource={filteredDocuments}
|
||||
rowKey="id"
|
||||
loading={loading}
|
||||
pagination={false}
|
||||
size="small"
|
||||
scroll={{ x: 'max-content' }}
|
||||
/>
|
||||
)}
|
||||
</div>
|
||||
|
||||
{/* 底部分页器 */}
|
||||
{filteredDocuments.length > 0 && (
|
||||
<div className="document-pagination">
|
||||
<span className="pagination-total">共 {total} 条</span>
|
||||
<div className="pagination-controls">
|
||||
<Button
|
||||
size="small"
|
||||
disabled={page <= 1}
|
||||
onClick={() => onPageChange(page - 1)}
|
||||
>
|
||||
上一页
|
||||
</Button>
|
||||
<span className="pagination-info">
|
||||
第 {page} 页 / 共 {Math.ceil(total / pageSize)} 页
|
||||
</span>
|
||||
<Button
|
||||
size="small"
|
||||
disabled={page >= Math.ceil(total / pageSize)}
|
||||
onClick={() => onPageChange(page + 1)}
|
||||
>
|
||||
下一页
|
||||
</Button>
|
||||
</div>
|
||||
上一页
|
||||
</Button>
|
||||
<span className="pagination-info">
|
||||
第 {page} 页 / 共 {Math.ceil(total / pageSize)} 页
|
||||
</span>
|
||||
<Button
|
||||
size="small"
|
||||
disabled={page >= Math.ceil(total / pageSize)}
|
||||
onClick={() => onPageChange(page + 1)}
|
||||
>
|
||||
下一页
|
||||
</Button>
|
||||
</div>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
</>
|
||||
);
|
||||
}
|
||||
|
||||
@@ -0,0 +1,914 @@
|
||||
import { useState, useEffect, useRef, useCallback } from 'react';
|
||||
import {
|
||||
Input,
|
||||
Button,
|
||||
InputNumber,
|
||||
Checkbox,
|
||||
Select,
|
||||
Card,
|
||||
Empty,
|
||||
Spin,
|
||||
message,
|
||||
Divider,
|
||||
Tooltip,
|
||||
Progress,
|
||||
Upload,
|
||||
} from 'antd';
|
||||
import type { UploadFile, UploadProps } from 'antd';
|
||||
import {
|
||||
QuestionCircleOutlined,
|
||||
FileTextOutlined,
|
||||
CheckCircleOutlined,
|
||||
LoadingOutlined,
|
||||
ExclamationCircleOutlined,
|
||||
InboxOutlined,
|
||||
DeleteOutlined,
|
||||
ArrowLeftOutlined,
|
||||
} from '@ant-design/icons';
|
||||
import type { IndexingStatus } from '~/api/dify-dataset/type/documentTypes';
|
||||
import type { Segment } from '~/api/dify-dataset/type';
|
||||
import {
|
||||
uploadDocumentWithConfig,
|
||||
updateDocumentByFile,
|
||||
fetchIndexingStatus,
|
||||
} from '~/api/dify-dataset/api/documentApi';
|
||||
import { fetchSegments } from '~/api/dify-dataset/api/segmentApi';
|
||||
|
||||
const { Dragger } = Upload;
|
||||
|
||||
interface DocumentUploadProps {
|
||||
datasetId: string;
|
||||
onClose: () => void;
|
||||
onSuccess: () => void;
|
||||
}
|
||||
|
||||
/**
|
||||
* 分段设置配置
|
||||
*/
|
||||
interface SegmentationSettings {
|
||||
separator: string;
|
||||
maxTokens: number;
|
||||
chunkOverlap: number;
|
||||
removeExtraSpaces: boolean;
|
||||
removeUrlsEmails: boolean;
|
||||
indexingTechnique: 'high_quality' | 'economy';
|
||||
}
|
||||
|
||||
/**
|
||||
* 默认分段设置
|
||||
*/
|
||||
const DEFAULT_SETTINGS: SegmentationSettings = {
|
||||
separator: '\\n\\n',
|
||||
maxTokens: 1024,
|
||||
chunkOverlap: 50,
|
||||
removeExtraSpaces: true,
|
||||
removeUrlsEmails: false,
|
||||
indexingTechnique: 'high_quality',
|
||||
};
|
||||
|
||||
/**
|
||||
* 单个文档的上传状态
|
||||
*/
|
||||
type DocumentStage = 'pending' | 'uploading' | 'indexing' | 'completed' | 'error';
|
||||
|
||||
/**
|
||||
* 上传的文档信息（支持多文件）
|
||||
*/
|
||||
interface UploadedDocument {
|
||||
file: File;
|
||||
documentId: string;
|
||||
batch: string;
|
||||
stage: DocumentStage;
|
||||
indexingStatus: IndexingStatus;
|
||||
uploadProgress: number;
|
||||
error?: string;
|
||||
settings: SegmentationSettings;
|
||||
segments: Segment[];
|
||||
}
|
||||
|
||||
/**
|
||||
* 索引状态配置
|
||||
*/
|
||||
const INDEXING_STATUS_CONFIG: Record<IndexingStatus, { text: string; percent: number }> = {
|
||||
waiting: { text: '等待处理...', percent: 10 },
|
||||
parsing: { text: '解析文档...', percent: 30 },
|
||||
cleaning: { text: '清洗文本...', percent: 50 },
|
||||
splitting: { text: '分段处理...', percent: 70 },
|
||||
indexing: { text: '建立索引...', percent: 85 },
|
||||
completed: { text: '处理完成', percent: 100 },
|
||||
paused: { text: '已暂停', percent: 0 },
|
||||
error: { text: '处理失败', percent: 0 },
|
||||
};
|
||||
|
||||
/**
|
||||
* 支持的文件格式
|
||||
*/
|
||||
const SUPPORTED_FORMATS = 'TXT, MARKDOWN, MDX, PDF, HTML, XLSX, XLS, DOCX, CSV, VTT, PROPERTIES, MD, HTM';
|
||||
|
||||
/**
|
||||
* 文档上传组件
|
||||
* 支持多文件上传，两步流程：选择文件 → 上传并配置分段
|
||||
*/
|
||||
export default function DocumentUpload({
|
||||
datasetId,
|
||||
onClose,
|
||||
onSuccess,
|
||||
}: DocumentUploadProps) {
|
||||
// 步骤控制
|
||||
const [step, setStep] = useState<1 | 2>(1);
|
||||
|
||||
// 文件相关
|
||||
const [selectedFiles, setSelectedFiles] = useState<File[]>([]);
|
||||
const [fileList, setFileList] = useState<UploadFile[]>([]);
|
||||
|
||||
// 多文档状态管理
|
||||
const [uploadedDocuments, setUploadedDocuments] = useState<UploadedDocument[]>([]);
|
||||
// 当前选中查看的文档索引
|
||||
const [currentDocIndex, setCurrentDocIndex] = useState(0);
|
||||
|
||||
// 当前显示的分段设置（来自当前选中的文档）
|
||||
const [currentSettings, setCurrentSettings] = useState<SegmentationSettings>(DEFAULT_SETTINGS);
|
||||
|
||||
// 预览相关
|
||||
const [previewLoading, setPreviewLoading] = useState(false);
|
||||
|
||||
// 轮询定时器（支持多个文档）
|
||||
const pollingTimersRef = useRef<Map<string, NodeJS.Timeout>>(new Map());
|
||||
|
||||
// 清理所有轮询定时器
|
||||
useEffect(() => {
|
||||
return () => {
|
||||
pollingTimersRef.current.forEach(timer => clearInterval(timer));
|
||||
pollingTimersRef.current.clear();
|
||||
};
|
||||
}, []);
|
||||
|
||||
/**
|
||||
* 停止指定文档的轮询
|
||||
*/
|
||||
const stopPolling = useCallback((documentId: string) => {
|
||||
const timer = pollingTimersRef.current.get(documentId);
|
||||
if (timer) {
|
||||
clearInterval(timer);
|
||||
pollingTimersRef.current.delete(documentId);
|
||||
}
|
||||
}, []);
|
||||
|
||||
/**
|
||||
* 停止所有轮询
|
||||
*/
|
||||
const stopAllPolling = useCallback(() => {
|
||||
pollingTimersRef.current.forEach(timer => clearInterval(timer));
|
||||
pollingTimersRef.current.clear();
|
||||
}, []);
|
||||
|
||||
/**
|
||||
* 加载分段预览
|
||||
*/
|
||||
const loadSegmentsPreview = useCallback(async (documentId: string, docIndex: number) => {
|
||||
setPreviewLoading(true);
|
||||
try {
|
||||
const response = await fetchSegments(datasetId, documentId, 1, 50);
|
||||
const segments = response.data || [];
|
||||
// 更新对应文档的分段
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === docIndex ? { ...doc, segments } : doc
|
||||
));
|
||||
} catch (err: any) {
|
||||
console.error('加载分段预览失败:', err);
|
||||
message.error('加载分段预览失败');
|
||||
} finally {
|
||||
setPreviewLoading(false);
|
||||
}
|
||||
}, [datasetId]);
|
||||
|
||||
/**
|
||||
* 轮询索引状态
|
||||
*/
|
||||
const pollIndexingStatus = useCallback(async (batch: string, documentId: string, docIndex: number) => {
|
||||
try {
|
||||
const response = await fetchIndexingStatus(datasetId, batch);
|
||||
const documentStatus = response.data?.[0];
|
||||
|
||||
if (documentStatus) {
|
||||
const status = documentStatus.indexing_status as IndexingStatus;
|
||||
|
||||
// 更新文档状态
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) => {
|
||||
if (idx !== docIndex) return doc;
|
||||
return { ...doc, indexingStatus: status };
|
||||
}));
|
||||
|
||||
if (status === 'completed') {
|
||||
stopPolling(documentId);
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === docIndex ? { ...doc, stage: 'completed' } : doc
|
||||
));
|
||||
// message.success(`文档 "${uploadedDocuments[docIndex]?.file.name}" 处理完成！`);
|
||||
// 自动加载分段预览
|
||||
loadSegmentsPreview(documentId, docIndex);
|
||||
} else if (status === 'error') {
|
||||
stopPolling(documentId);
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === docIndex ? { ...doc, stage: 'error', error: documentStatus.error || '处理失败' } : doc
|
||||
));
|
||||
}
|
||||
}
|
||||
} catch (err) {
|
||||
console.error('获取索引状态失败:', err);
|
||||
}
|
||||
}, [datasetId, stopPolling, loadSegmentsPreview, uploadedDocuments]);
|
||||
|
||||
/**
|
||||
* 开始轮询
|
||||
*/
|
||||
const startPolling = useCallback((batch: string, documentId: string, docIndex: number) => {
|
||||
// 先停止之前的轮询
|
||||
stopPolling(documentId);
|
||||
|
||||
// 开始新的轮询
|
||||
const timer = setInterval(() => {
|
||||
pollIndexingStatus(batch, documentId, docIndex);
|
||||
}, 2000);
|
||||
pollingTimersRef.current.set(documentId, timer);
|
||||
|
||||
// 立即执行一次
|
||||
pollIndexingStatus(batch, documentId, docIndex);
|
||||
}, [stopPolling, pollIndexingStatus]);
|
||||
|
||||
/**
|
||||
* 构建上传配置
|
||||
*/
|
||||
const buildConfig = (s: SegmentationSettings) => ({
|
||||
indexing_technique: s.indexingTechnique,
|
||||
process_rule: {
|
||||
mode: 'custom' as const,
|
||||
rules: {
|
||||
pre_processing_rules: [
|
||||
{ id: 'remove_extra_spaces' as const, enabled: s.removeExtraSpaces },
|
||||
{ id: 'remove_urls_emails' as const, enabled: s.removeUrlsEmails },
|
||||
],
|
||||
segmentation: {
|
||||
separator: s.separator.replace(/\\n/g, '\n'),
|
||||
max_tokens: s.maxTokens,
|
||||
},
|
||||
},
|
||||
},
|
||||
});
|
||||
|
||||
/**
|
||||
* 更新当前文档的设置
|
||||
*/
|
||||
const updateCurrentSettings = (key: keyof SegmentationSettings, value: any) => {
|
||||
const newSettings = { ...currentSettings, [key]: value };
|
||||
setCurrentSettings(newSettings);
|
||||
// 同步更新到文档列表
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === currentDocIndex ? { ...doc, settings: newSettings } : doc
|
||||
));
|
||||
};
|
||||
|
||||
/**
|
||||
* 处理文件选择变化
|
||||
*/
|
||||
const handleFileChange: UploadProps['onChange'] = ({ fileList: newFileList }) => {
|
||||
setFileList(newFileList);
|
||||
// 提取实际文件对象
|
||||
const files = newFileList
|
||||
.filter(f => f.originFileObj)
|
||||
.map(f => f.originFileObj as File);
|
||||
setSelectedFiles(files);
|
||||
};
|
||||
|
||||
/**
|
||||
* 移除文件
|
||||
*/
|
||||
const handleRemoveFile = (file: UploadFile) => {
|
||||
const newFileList = fileList.filter(f => f.uid !== file.uid);
|
||||
setFileList(newFileList);
|
||||
const files = newFileList
|
||||
.filter(f => f.originFileObj)
|
||||
.map(f => f.originFileObj as File);
|
||||
setSelectedFiles(files);
|
||||
};
|
||||
|
||||
/**
|
||||
* 上传单个文件
|
||||
*/
|
||||
const uploadSingleFile = async (file: File, index: number): Promise<void> => {
|
||||
try {
|
||||
// 更新状态为上传中
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === index ? { ...doc, stage: 'uploading' as DocumentStage } : doc
|
||||
));
|
||||
|
||||
const config = buildConfig(DEFAULT_SETTINGS);
|
||||
const result = await uploadDocumentWithConfig(
|
||||
datasetId,
|
||||
file,
|
||||
config,
|
||||
(percent) => {
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === index ? { ...doc, uploadProgress: percent } : doc
|
||||
));
|
||||
}
|
||||
);
|
||||
|
||||
// 更新文档信息
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === index ? {
|
||||
...doc,
|
||||
documentId: result.document.id,
|
||||
batch: result.batch,
|
||||
stage: 'indexing' as DocumentStage,
|
||||
indexingStatus: 'waiting' as IndexingStatus,
|
||||
} : doc
|
||||
));
|
||||
|
||||
// 开始轮询索引状态
|
||||
startPolling(result.batch, result.document.id, index);
|
||||
} catch (err: any) {
|
||||
console.error(`上传文档 ${file.name} 失败:`, err);
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === index ? {
|
||||
...doc,
|
||||
stage: 'error' as DocumentStage,
|
||||
error: err.message || '上传失败',
|
||||
} : doc
|
||||
));
|
||||
}
|
||||
};
|
||||
|
||||
/**
|
||||
* 点击"下一步"：立即上传所有文件
|
||||
*/
|
||||
const handleNextStep = async () => {
|
||||
if (selectedFiles.length === 0) {
|
||||
message.warning('请先选择文件');
|
||||
return;
|
||||
}
|
||||
|
||||
// 初始化所有文档状态
|
||||
const docs: UploadedDocument[] = selectedFiles.map(file => ({
|
||||
file,
|
||||
documentId: '',
|
||||
batch: '',
|
||||
stage: 'pending' as DocumentStage,
|
||||
indexingStatus: 'waiting' as IndexingStatus,
|
||||
uploadProgress: 0,
|
||||
settings: { ...DEFAULT_SETTINGS },
|
||||
segments: [],
|
||||
}));
|
||||
setUploadedDocuments(docs);
|
||||
setCurrentDocIndex(0);
|
||||
setCurrentSettings({ ...DEFAULT_SETTINGS });
|
||||
setStep(2);
|
||||
|
||||
// 依次上传所有文件
|
||||
for (let i = 0; i < selectedFiles.length; i++) {
|
||||
await uploadSingleFile(selectedFiles[i], i);
|
||||
}
|
||||
};
|
||||
|
||||
/**
|
||||
* 切换查看的文档
|
||||
*/
|
||||
const handleDocumentChange = (docId: string) => {
|
||||
const index = uploadedDocuments.findIndex(doc => doc.documentId === docId || doc.file.name === docId);
|
||||
if (index !== -1) {
|
||||
setCurrentDocIndex(index);
|
||||
const doc = uploadedDocuments[index];
|
||||
setCurrentSettings(doc.settings);
|
||||
}
|
||||
};
|
||||
|
||||
/**
|
||||
* 修改参数后重新处理当前文档
|
||||
*/
|
||||
const handleReprocess = async () => {
|
||||
const currentDoc = uploadedDocuments[currentDocIndex];
|
||||
if (!currentDoc || !currentDoc.documentId) return;
|
||||
|
||||
// 更新状态
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === currentDocIndex ? {
|
||||
...doc,
|
||||
stage: 'uploading' as DocumentStage,
|
||||
uploadProgress: 0,
|
||||
segments: [],
|
||||
} : doc
|
||||
));
|
||||
|
||||
try {
|
||||
const config = buildConfig(currentSettings);
|
||||
const result = await updateDocumentByFile(
|
||||
datasetId,
|
||||
currentDoc.documentId,
|
||||
currentDoc.file,
|
||||
config,
|
||||
(percent) => {
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === currentDocIndex ? { ...doc, uploadProgress: percent } : doc
|
||||
));
|
||||
}
|
||||
);
|
||||
|
||||
// 更新 batch
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === currentDocIndex ? {
|
||||
...doc,
|
||||
batch: result.batch,
|
||||
stage: 'indexing' as DocumentStage,
|
||||
indexingStatus: 'waiting' as IndexingStatus,
|
||||
} : doc
|
||||
));
|
||||
|
||||
startPolling(result.batch, currentDoc.documentId, currentDocIndex);
|
||||
} catch (err: any) {
|
||||
console.error('重新处理失败:', err);
|
||||
setUploadedDocuments(prev => prev.map((doc, idx) =>
|
||||
idx === currentDocIndex ? {
|
||||
...doc,
|
||||
stage: 'error' as DocumentStage,
|
||||
error: err.message || '重新处理失败',
|
||||
} : doc
|
||||
));
|
||||
message.error(err.message || '重新处理失败');
|
||||
}
|
||||
};
|
||||
|
||||
/**
|
||||
* 返回上一步
|
||||
*/
|
||||
const handlePrevStep = () => {
|
||||
// 检查是否有文档正在处理
|
||||
const hasProcessing = uploadedDocuments.some(doc =>
|
||||
doc.stage === 'uploading' || doc.stage === 'indexing'
|
||||
);
|
||||
if (hasProcessing) {
|
||||
message.warning('还有文档正在处理中，请等待完成');
|
||||
return;
|
||||
}
|
||||
stopAllPolling();
|
||||
setStep(1);
|
||||
setUploadedDocuments([]);
|
||||
setCurrentDocIndex(0);
|
||||
setCurrentSettings(DEFAULT_SETTINGS);
|
||||
};
|
||||
|
||||
/**
|
||||
* 返回文档列表
|
||||
*/
|
||||
const handleGoToDocuments = () => {
|
||||
stopAllPolling();
|
||||
const hasCompleted = uploadedDocuments.some(doc => doc.stage === 'completed');
|
||||
if (hasCompleted) {
|
||||
onSuccess();
|
||||
}
|
||||
onClose();
|
||||
};
|
||||
|
||||
/**
|
||||
* 获取当前文档
|
||||
*/
|
||||
const getCurrentDocument = (): UploadedDocument | null => {
|
||||
return uploadedDocuments[currentDocIndex] || null;
|
||||
};
|
||||
|
||||
/**
|
||||
* 获取当前文档的进度
|
||||
*/
|
||||
const getCurrentProgress = () => {
|
||||
const doc = getCurrentDocument();
|
||||
if (!doc) return 0;
|
||||
if (doc.stage === 'uploading') {
|
||||
return doc.uploadProgress;
|
||||
}
|
||||
if (doc.stage === 'indexing' || doc.stage === 'completed') {
|
||||
return INDEXING_STATUS_CONFIG[doc.indexingStatus]?.percent || 0;
|
||||
}
|
||||
return 0;
|
||||
};
|
||||
|
||||
/**
|
||||
* 获取当前文档的状态文本
|
||||
*/
|
||||
const getStatusText = () => {
|
||||
const doc = getCurrentDocument();
|
||||
if (!doc) return '';
|
||||
if (doc.stage === 'uploading') {
|
||||
return `正在上传... ${doc.uploadProgress}%`;
|
||||
}
|
||||
if (doc.stage === 'indexing') {
|
||||
return INDEXING_STATUS_CONFIG[doc.indexingStatus]?.text || '处理中...';
|
||||
}
|
||||
if (doc.stage === 'completed') {
|
||||
return '处理完成';
|
||||
}
|
||||
if (doc.stage === 'error') {
|
||||
return doc.error || '处理失败';
|
||||
}
|
||||
return '';
|
||||
};
|
||||
|
||||
/**
|
||||
* 判断当前文档是否正在处理
|
||||
*/
|
||||
const isCurrentDocProcessing = () => {
|
||||
const doc = getCurrentDocument();
|
||||
return doc?.stage === 'uploading' || doc?.stage === 'indexing';
|
||||
};
|
||||
|
||||
/**
|
||||
* 获取所有文档的完成状态统计
|
||||
*/
|
||||
const getCompletionStats = () => {
|
||||
const completed = uploadedDocuments.filter(doc => doc.stage === 'completed').length;
|
||||
const total = uploadedDocuments.length;
|
||||
return { completed, total };
|
||||
};
|
||||
|
||||
/**
|
||||
* 渲染步骤指示器（两步流程）
|
||||
*/
|
||||
const renderSteps = () => (
|
||||
<div className="upload-steps">
|
||||
<div className={`step-item ${step === 1 ? 'active' : ''} ${step > 1 ? 'completed' : ''}`}>
|
||||
<span className="step-number">1</span>
|
||||
<span className="step-title">选择数据源</span>
|
||||
</div>
|
||||
<div className={`step-divider ${step > 1 ? 'completed' : ''}`}></div>
|
||||
<div className={`step-item ${step === 2 ? 'active' : ''}`}>
|
||||
<span className="step-number">2</span>
|
||||
<span className="step-title">文本分段与清洗</span>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
|
||||
/**
|
||||
* 渲染第一步：选择文件（支持多文件）
|
||||
*/
|
||||
const renderStep1 = () => (
|
||||
<div className="upload-step-content step1">
|
||||
<h2 className="step-heading">上传文本文件</h2>
|
||||
<p className="step-description">
|
||||
文档需上传至知识智能理解法治知识库，广东烟草智能理解将按照于知识库，你可以在聊后指数文档所据案中检索它
|
||||
</p>
|
||||
|
||||
<div className="file-drop-zone">
|
||||
<Dragger
|
||||
fileList={fileList}
|
||||
onChange={handleFileChange}
|
||||
beforeUpload={() => false}
|
||||
multiple={true}
|
||||
accept=".txt,.md,.mdx,.pdf,.html,.htm,.xlsx,.xls,.docx,.csv,.vtt,.properties"
|
||||
showUploadList={false}
|
||||
>
|
||||
<p className="ant-upload-drag-icon">
|
||||
<InboxOutlined />
|
||||
</p>
|
||||
<p className="ant-upload-text">拖拽文件或文件夹至此，或者 <span className="upload-link">选择文件</span></p>
|
||||
<p className="ant-upload-hint">
|
||||
已支持 {SUPPORTED_FORMATS}，每个文件不超过 15MB。支持批量上传多个文件。
|
||||
</p>
|
||||
</Dragger>
|
||||
</div>
|
||||
|
||||
{/* 已选文件列表 */}
|
||||
{selectedFiles.length > 0 && (
|
||||
<div className="selected-files-section">
|
||||
<h3 className="section-subtitle">嵌入已就绪 ({selectedFiles.length} 个文件)</h3>
|
||||
<div className="selected-files-list">
|
||||
{fileList.map((file) => (
|
||||
<div key={file.uid} className="selected-file-item">
|
||||
<FileTextOutlined className="file-icon" />
|
||||
<div className="file-info">
|
||||
<span className="file-name">{file.name}</span>
|
||||
<span className="file-size">
|
||||
{file.originFileObj
|
||||
? `${file.originFileObj.type?.split('/')[1]?.toUpperCase() || 'FILE'}，${(file.originFileObj.size / 1024 / 1024).toFixed(2)}MB`
|
||||
: ''}
|
||||
</span>
|
||||
</div>
|
||||
<Button
|
||||
type="text"
|
||||
icon={<DeleteOutlined />}
|
||||
onClick={() => handleRemoveFile(file)}
|
||||
className="remove-file-btn"
|
||||
/>
|
||||
</div>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
)}
|
||||
|
||||
<div className="step-actions">
|
||||
<Button
|
||||
type="primary"
|
||||
onClick={handleNextStep}
|
||||
disabled={selectedFiles.length === 0}
|
||||
className="next-btn"
|
||||
>
|
||||
下一步 →
|
||||
</Button>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
|
||||
/**
|
||||
* 渲染第二步：分段配置与预览
|
||||
* 左侧始终显示配置面板，右侧预览框内显示进度或分段内容
|
||||
*/
|
||||
const renderStep2 = () => {
|
||||
const currentDoc = getCurrentDocument();
|
||||
const isProcessing = isCurrentDocProcessing();
|
||||
const stats = getCompletionStats();
|
||||
|
||||
return (
|
||||
<div className="upload-step-content step2">
|
||||
{/* 分段配置与预览 */}
|
||||
<div className="document-detail-content">
|
||||
{/* 左侧设置区域 */}
|
||||
<div className="settings-panel">
|
||||
<div className="settings-section">
|
||||
<h3 className="section-title">分段设置</h3>
|
||||
|
||||
{/* 分段标识符 */}
|
||||
<div className="setting-item">
|
||||
<label className="setting-label">
|
||||
分段标识符
|
||||
<Tooltip title="系统会在遇到指定分隔符时自动分段，默认值为 \n\n（按段落分段）">
|
||||
<QuestionCircleOutlined className="help-icon" />
|
||||
</Tooltip>
|
||||
</label>
|
||||
<Input
|
||||
value={currentSettings.separator}
|
||||
onChange={(e) => updateCurrentSettings('separator', e.target.value)}
|
||||
placeholder="\n\n"
|
||||
className="setting-input"
|
||||
disabled={isProcessing}
|
||||
/>
|
||||
</div>
|
||||
|
||||
{/* 分段最大长度 */}
|
||||
<div className="setting-item">
|
||||
<label className="setting-label">
|
||||
分段最大长度
|
||||
<Tooltip title="指定每个分段允许的最大字符数（100-4000），超过此限制系统会强制分段">
|
||||
<QuestionCircleOutlined className="help-icon" />
|
||||
</Tooltip>
|
||||
</label>
|
||||
<div className="setting-input-with-suffix">
|
||||
<InputNumber
|
||||
value={currentSettings.maxTokens}
|
||||
onChange={(value) => updateCurrentSettings('maxTokens', value || 1024)}
|
||||
min={100}
|
||||
max={4000}
|
||||
className="setting-input-number"
|
||||
disabled={isProcessing}
|
||||
/>
|
||||
<span className="input-suffix">characters</span>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 分段重叠长度 */}
|
||||
<div className="setting-item">
|
||||
<label className="setting-label">
|
||||
分段重叠长度
|
||||
<Tooltip title="相邻分段之间重叠的字符数，有助于保持上下文连贯性">
|
||||
<QuestionCircleOutlined className="help-icon" />
|
||||
</Tooltip>
|
||||
</label>
|
||||
<div className="setting-input-with-suffix">
|
||||
<InputNumber
|
||||
value={currentSettings.chunkOverlap}
|
||||
onChange={(value) => updateCurrentSettings('chunkOverlap', value || 50)}
|
||||
min={0}
|
||||
max={500}
|
||||
className="setting-input-number"
|
||||
disabled={isProcessing}
|
||||
/>
|
||||
<span className="input-suffix">characters</span>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<Divider />
|
||||
|
||||
{/* 文本预处理规则 */}
|
||||
<div className="settings-section">
|
||||
<h3 className="section-title">文本预处理规则</h3>
|
||||
<div className="checkbox-group">
|
||||
<Checkbox
|
||||
checked={currentSettings.removeExtraSpaces}
|
||||
onChange={(e) => updateCurrentSettings('removeExtraSpaces', e.target.checked)}
|
||||
disabled={isProcessing}
|
||||
>
|
||||
替换掉连续的空格、换行符和制表符
|
||||
</Checkbox>
|
||||
<Checkbox
|
||||
checked={currentSettings.removeUrlsEmails}
|
||||
onChange={(e) => updateCurrentSettings('removeUrlsEmails', e.target.checked)}
|
||||
disabled={isProcessing}
|
||||
>
|
||||
删除所有 URL 和电子邮件地址
|
||||
</Checkbox>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<Divider />
|
||||
|
||||
{/* 索引方式 */}
|
||||
<div className="settings-section">
|
||||
<h3 className="section-title">索引方式</h3>
|
||||
<div className="index-options">
|
||||
<div
|
||||
className={`index-option ${currentSettings.indexingTechnique === 'high_quality' ? 'active' : ''} ${isProcessing ? 'disabled' : ''}`}
|
||||
onClick={() => !isProcessing && updateCurrentSettings('indexingTechnique', 'high_quality')}
|
||||
>
|
||||
<span className="option-radio"></span>
|
||||
<span className="option-label">高质量</span>
|
||||
<span className="option-badge recommended">推荐</span>
|
||||
</div>
|
||||
<div
|
||||
className={`index-option ${currentSettings.indexingTechnique === 'economy' ? 'active' : ''} ${isProcessing ? 'disabled' : ''}`}
|
||||
onClick={() => !isProcessing && updateCurrentSettings('indexingTechnique', 'economy')}
|
||||
>
|
||||
<span className="option-radio"></span>
|
||||
<span className="option-label">经济</span>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 操作按钮 */}
|
||||
<div className="settings-actions">
|
||||
<Button onClick={handlePrevStep} disabled={isProcessing}>
|
||||
<ArrowLeftOutlined /> 上一步
|
||||
</Button>
|
||||
<Button
|
||||
type="primary"
|
||||
onClick={handleReprocess}
|
||||
loading={isProcessing}
|
||||
disabled={isProcessing || !currentDoc?.documentId}
|
||||
>
|
||||
更新嵌入配置
|
||||
</Button>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 右侧预览区域 */}
|
||||
<div className="preview-panel">
|
||||
<Card
|
||||
title={
|
||||
<div className="preview-header">
|
||||
<span>预览</span>
|
||||
{uploadedDocuments.length > 0 && (
|
||||
<>
|
||||
<Select
|
||||
value={currentDoc?.documentId || currentDoc?.file.name}
|
||||
style={{ width: 500 }}
|
||||
onChange={handleDocumentChange}
|
||||
options={uploadedDocuments.map((doc, idx) => ({
|
||||
value: doc.documentId || doc.file.name,
|
||||
label: (
|
||||
<span className="file-select-option">
|
||||
{doc.stage === 'completed' && <CheckCircleOutlined style={{ color: '#52c41a', marginRight: 4 }} />}
|
||||
{(doc.stage === 'uploading' || doc.stage === 'indexing') && <LoadingOutlined style={{ color: '#00684a', marginRight: 4 }} />}
|
||||
{doc.stage === 'error' && <ExclamationCircleOutlined style={{ color: '#ff4d4f', marginRight: 4 }} />}
|
||||
{doc.file.name}
|
||||
</span>
|
||||
),
|
||||
}))}
|
||||
/>
|
||||
{!isProcessing && currentDoc?.segments && (
|
||||
<span className="segment-count">
|
||||
{currentDoc.segments.length} 段块
|
||||
</span>
|
||||
)}
|
||||
</>
|
||||
)}
|
||||
</div>
|
||||
}
|
||||
className="preview-card"
|
||||
>
|
||||
{/* 处理进度（在预览框内显示） */}
|
||||
{isProcessing ? (
|
||||
<div className="preview-processing">
|
||||
<div className="processing-file">
|
||||
<FileTextOutlined className="file-icon" />
|
||||
<span className="file-name">{currentDoc?.file.name}</span>
|
||||
<LoadingOutlined className="status-icon loading" />
|
||||
</div>
|
||||
<Progress
|
||||
percent={getCurrentProgress()}
|
||||
status="active"
|
||||
strokeColor={{
|
||||
'0%': '#00684a',
|
||||
'100%': '#52c41a',
|
||||
}}
|
||||
/>
|
||||
<div className="status-text">{getStatusText()}</div>
|
||||
|
||||
{/* 索引阶段详情 */}
|
||||
{currentDoc?.stage === 'indexing' && (
|
||||
<div className="indexing-stages">
|
||||
<div className={`stage-item ${['waiting', 'parsing', 'cleaning', 'splitting', 'indexing', 'completed'].includes(currentDoc.indexingStatus) ? 'active' : ''}`}>
|
||||
<span className="stage-dot"></span>
|
||||
<span>等待处理</span>
|
||||
</div>
|
||||
<div className={`stage-item ${['parsing', 'cleaning', 'splitting', 'indexing', 'completed'].includes(currentDoc.indexingStatus) ? 'active' : ''}`}>
|
||||
<span className="stage-dot"></span>
|
||||
<span>解析文档</span>
|
||||
</div>
|
||||
<div className={`stage-item ${['cleaning', 'splitting', 'indexing', 'completed'].includes(currentDoc.indexingStatus) ? 'active' : ''}`}>
|
||||
<span className="stage-dot"></span>
|
||||
<span>清洗文本</span>
|
||||
</div>
|
||||
<div className={`stage-item ${['splitting', 'indexing', 'completed'].includes(currentDoc.indexingStatus) ? 'active' : ''}`}>
|
||||
<span className="stage-dot"></span>
|
||||
<span>分段处理</span>
|
||||
</div>
|
||||
<div className={`stage-item ${['indexing', 'completed'].includes(currentDoc.indexingStatus) ? 'active' : ''}`}>
|
||||
<span className="stage-dot"></span>
|
||||
<span>建立索引</span>
|
||||
</div>
|
||||
<div className={`stage-item ${currentDoc.indexingStatus === 'completed' ? 'active' : ''}`}>
|
||||
<span className="stage-dot"></span>
|
||||
<span>完成</span>
|
||||
</div>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
) : currentDoc?.stage === 'error' ? (
|
||||
<div className="preview-error">
|
||||
<ExclamationCircleOutlined className="error-icon" />
|
||||
<div className="error-text">{currentDoc.error || '处理失败'}</div>
|
||||
</div>
|
||||
) : previewLoading ? (
|
||||
<div className="preview-loading">
|
||||
<Spin size="large" />
|
||||
<div className="loading-text">加载中...</div>
|
||||
</div>
|
||||
) : (currentDoc?.segments?.length ?? 0) === 0 ? (
|
||||
<div className="preview-empty">
|
||||
<Empty description="等待处理完成后显示分段预览" />
|
||||
</div>
|
||||
) : (
|
||||
<div className="preview-segments">
|
||||
{currentDoc?.segments.map((segment, index) => (
|
||||
<div key={segment.id} className="segment-item">
|
||||
<div className="segment-header">
|
||||
<span className="segment-index">#{index + 1}</span>
|
||||
<span className="segment-chars">
|
||||
{segment.word_count} 字符
|
||||
</span>
|
||||
</div>
|
||||
<div className="segment-content">
|
||||
{segment.content}
|
||||
</div>
|
||||
</div>
|
||||
))}
|
||||
</div>
|
||||
)}
|
||||
</Card>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 完成状态底部操作 */}
|
||||
{stats.completed > 0 && (
|
||||
<div className="completion-actions">
|
||||
<span className="completion-stats">
|
||||
{stats.completed}/{stats.total} 个文档处理完成
|
||||
</span>
|
||||
<Button type="primary" onClick={handleGoToDocuments}>
|
||||
前往文档
|
||||
</Button>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
);
|
||||
};
|
||||
|
||||
return (
|
||||
<div className="document-upload-page">
|
||||
{/* 页面头部 */}
|
||||
<div className="upload-header">
|
||||
<Button
|
||||
type="text"
|
||||
icon={<ArrowLeftOutlined />}
|
||||
onClick={onClose}
|
||||
className="back-btn"
|
||||
>
|
||||
知识库
|
||||
</Button>
|
||||
{renderSteps()}
|
||||
</div>
|
||||
|
||||
{/* 内容区域 */}
|
||||
<div className="upload-content">
|
||||
{step === 1 && renderStep1()}
|
||||
{step === 2 && renderStep2()}
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
}
|
||||
Reference in New Issue
Block a user