leaudit-platform-frontend/app/api/dify-dataset/client.ts

/**
 * Dify Dataset 客户端 API 模块
 *
 * 提供浏览器端调用 Dify 知识库 API 的函数
 * 通过 Remix API Routes 代理请求
 *
 * @module api/dify-dataset/client
 */

import axios from 'axios';
import type {
    Dataset,
    DatasetsResponse,
    DocumentsResponse,
    SegmentsResponse,
    Document,
    OperationResult,
    IndexingStatusResponse,
    UploadFileInfo,
    UpdateDatasetRequest,
} from './types';

// ============================================================================
// 基础配置
// ============================================================================

/**
 * API 基础 URL
 * 指向 Remix API Routes（/api/dataset/*）
 */
const API_URL = '/api/dataset';

// ============================================================================
// 知识库 API
// ============================================================================

/**
 * 获取知识库列表
 *
 * @param page - 页码，默认 1
 * @param limit - 每页数量，默认 20
 * @returns 知识库列表响应
 */
export async function fetchDatasets(
    page: number = 1,
    limit: number = 20
): Promise<DatasetsResponse> {
    const params = new URLSearchParams({
        page: page.toString(),
        limit: limit.toString(),
    });

    const response = await axios.get<DatasetsResponse>(
        `${API_URL}/datasets?${params}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 获取单个知识库详情
 *
 * @param datasetId - 知识库 ID
 * @returns 知识库详情
 */
export async function fetchDataset(datasetId: string): Promise<Dataset> {
    const response = await axios.get<Dataset>(
        `${API_URL}/datasets/${datasetId}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 更新知识库详情
 *
 * @param datasetId - 知识库 ID
 * @param data - 更新数据
 * @returns 更新后的知识库详情
 */
export async function updateDataset(
    datasetId: string,
    data: UpdateDatasetRequest
): Promise<Dataset> {
    console.log('[Dataset Client] 更新知识库:', { datasetId, data });

    const response = await axios.patch<Dataset>(
        `${API_URL}/datasets/${datasetId}`,
        data,
        {
            headers: { 'Content-Type': 'application/json' },
            withCredentials: true,
        }
    );
    return response.data;
}

// ============================================================================
// 文档 API
// ============================================================================

/**
 * 获取知识库文档列表
 *
 * @param datasetId - 知识库 ID
 * @param page - 页码，默认 1
 * @param limit - 每页数量，默认 20
 * @param keyword - 搜索关键词
 * @returns 文档列表响应
 */
export async function fetchDocuments(
    datasetId: string,
    page: number = 1,
    limit: number = 20,
    keyword?: string
): Promise<DocumentsResponse> {
    const params = new URLSearchParams({
        page: page.toString(),
        limit: limit.toString(),
    });

    if (keyword) {
        params.append('keyword', keyword);
    }

    console.log('[Dataset Client] 获取文档列表:', { datasetId, page, limit, keyword });

    const response = await axios.get<DocumentsResponse>(
        `${API_URL}/datasets/${datasetId}/documents?${params}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 获取单个文档详情
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @returns 文档详情
 */
export async function fetchDocument(
    datasetId: string,
    documentId: string
): Promise<Document> {
    const response = await axios.get<Document>(
        `${API_URL}/datasets/${datasetId}/documents/${documentId}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 删除文档
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @returns 操作结果
 */
export async function deleteDocument(
    datasetId: string,
    documentId: string
): Promise<OperationResult> {
    console.log('[Dataset Client] 删除文档:', { datasetId, documentId });

    const response = await axios.delete<OperationResult>(
        `${API_URL}/datasets/${datasetId}/documents/${documentId}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 启用/禁用文档
 * Dify API: PATCH /datasets/{dataset_id}/documents/status/{action}
 * action: enable / disable / archive / un_archive
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @param enabled - 是否启用
 * @returns 操作结果
 */
export async function toggleDocumentStatus(
    datasetId: string,
    documentId: string,
    enabled: boolean
): Promise<OperationResult> {
    const action = enabled ? 'enable' : 'disable';
    console.log('[Dataset Client] 切换文档状态:', { datasetId, documentId, action });

    const response = await axios.patch<OperationResult>(
        `${API_URL}/datasets/${datasetId}/documents/status/${action}`,
        { document_ids: [documentId] },
        {
            headers: { 'Content-Type': 'application/json' },
            withCredentials: true,
        }
    );
    return response.data;
}

// ============================================================================
// 文档分段 API
// ============================================================================

/**
 * 获取文档分段列表
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @param page - 页码，默认 1
 * @param limit - 每页数量，默认 20
 * @param keyword - 搜索关键词
 * @returns 分段列表响应
 */
export async function fetchSegments(
    datasetId: string,
    documentId: string,
    page: number = 1,
    limit: number = 20,
    keyword?: string
): Promise<SegmentsResponse> {
    const params = new URLSearchParams({
        page: page.toString(),
        limit: limit.toString(),
    });

    if (keyword) {
        params.append('keyword', keyword);
    }

    console.log('[Dataset Client] 获取分段列表:', { datasetId, documentId, page, limit });

    const response = await axios.get<SegmentsResponse>(
        `${API_URL}/datasets/${datasetId}/documents/${documentId}/segments?${params}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 删除分段
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @param segmentId - 分段 ID
 * @returns 操作结果
 */
export async function deleteSegment(
    datasetId: string,
    documentId: string,
    segmentId: string
): Promise<OperationResult> {
    console.log('[Dataset Client] 删除分段:', { datasetId, documentId, segmentId });

    const response = await axios.delete<OperationResult>(
        `${API_URL}/datasets/${datasetId}/documents/${documentId}/segments/${segmentId}`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 启用/禁用分段
 * Dify API: POST /datasets/{dataset_id}/documents/{document_id}/segments/{segment_id}
 * 通过更新分段的方式来切换状态
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @param segmentId - 分段 ID
 * @param enabled - 是否启用
 * @returns 操作结果
 */
export async function toggleSegmentStatus(
    datasetId: string,
    documentId: string,
    segmentId: string,
    enabled: boolean
): Promise<OperationResult> {
    console.log('[Dataset Client] 切换分段状态:', { datasetId, documentId, segmentId, enabled });

    const response = await axios.post<OperationResult>(
        `${API_URL}/datasets/${datasetId}/documents/${documentId}/segments/${segmentId}`,
        { segment: { enabled } },
        {
            headers: { 'Content-Type': 'application/json' },
            withCredentials: true,
        }
    );
    return response.data;
}

// ============================================================================
// 文件上传 API
// ============================================================================

/**
 * 上传文件到知识库
 *
 * @param datasetId - 知识库 ID
 * @param file - 文件对象
 * @param onProgress - 上传进度回调
 * @returns 创建的文档信息
 */
export async function uploadDocument(
    datasetId: string,
    file: File,
    onProgress?: (percent: number) => void
): Promise<any> {
    const formData = new FormData();
    formData.append('file', file);
    formData.append('data', JSON.stringify({
        indexing_technique: 'high_quality',
        process_rule: {
            mode: 'automatic',
        },
    }));

    console.log('[Dataset Client] 上传文档:', { datasetId, fileName: file.name });

    const response = await axios.post(
        `${API_URL}/datasets/${datasetId}/documents`,
        formData,
        {
            withCredentials: true,
            onUploadProgress: (progressEvent) => {
                if (progressEvent.total && onProgress) {
                    const percent = Math.round((progressEvent.loaded * 100) / progressEvent.total);
                    onProgress(percent);
                }
            },
        }
    );
    return response.data;
}

/**
 * 获取文档嵌入状态（索引进度）
 *
 * @param datasetId - 知识库 ID
 * @param batch - 上传文档的批次号
 * @returns 索引状态列表
 */
export async function fetchIndexingStatus(
    datasetId: string,
    batch: string
): Promise<IndexingStatusResponse> {
    console.log('[Dataset Client] 获取索引状态:', { datasetId, batch });

    const response = await axios.get<IndexingStatusResponse>(
        `${API_URL}/datasets/${datasetId}/documents/${batch}/indexing-status`,
        { withCredentials: true }
    );
    return response.data;
}

/**
 * 获取文档上传文件信息
 *
 * @param datasetId - 知识库 ID
 * @param documentId - 文档 ID
 * @returns 上传文件信息
 */
export async function fetchUploadFileInfo(
    datasetId: string,
    documentId: string
): Promise<UploadFileInfo> {
    console.log('[Dataset Client] 获取上传文件信息:', { datasetId, documentId });

    const response = await axios.get<UploadFileInfo>(
        `${API_URL}/datasets/${datasetId}/documents/${documentId}/upload-file`,
        { withCredentials: true }
    );
    return response.data;
}