feat: 知识库设置页面增加 retrieval_model 检索配置功能

1. 召回测试页面增加 Score 阈值参数配置
2. 知识库设置页面新增检索模型配置:
   - 检索方式 (向量/全文/混合/关键字检索)
   - Reranking 模型 (默认开启，不可关闭)
   - Top K 返回数量
   - Score 阈值 (默认开启，可调节数值)
3. 修复 Dify API 字段名问题 (retrieval_model_dict)
4. 优化数据加载流程，使用详情接口获取完整配置

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

app/api/dify-dataset/api/datasetApi.ts

@@ -7,7 +7,7 @@
  */
 
 import axios from 'axios';
-import type { Dataset, DatasetsResponse } from '../type';
+import type { Dataset, DatasetsResponse, UpdateDatasetRequest } from '../type';
 
 /**
  * API 基础 URL
@@ -76,3 +76,27 @@ export async function updateDatasetName(
     );
     return response.data;
 }
+
+/**
+ * 更新知识库设置（包含检索模型配置）
+ *
+ * @param datasetId - 知识库 ID
+ * @param settings - 更新的设置项
+ * @returns 更新后的知识库详情
+ */
+export async function updateDatasetSettings(
+    datasetId: string,
+    settings: UpdateDatasetRequest
+): Promise<Dataset> {
+    console.log('[Dataset Client] 更新知识库设置:', { datasetId, settings });
+
+    const response = await axios.patch<Dataset>(
+        `${API_URL}/datasets/${datasetId}`,
+        settings,
+        {
+            headers: { 'Content-Type': 'application/json' },
+            withCredentials: true,
+        }
+    );
+    return response.data;
+}

app/api/dify-dataset/type/datasetTypes.ts

@@ -12,8 +12,8 @@ export interface Dataset {
     name: string;
     description: string;
     permission: 'only_me' | 'all_team_members';
-    data_source_type: 'upload_file' | 'notion_import' | 'website_crawl';
+    data_source_type: 'upload_file' | 'notion_import' | 'website_crawl' | null;
-    indexing_technique: 'high_quality' | 'economy';
+    indexing_technique: 'high_quality' | 'economy' | null;
     app_count: number;
     document_count: number;
     word_count: number;
@@ -21,6 +21,20 @@ export interface Dataset {
     created_at: number;
     updated_by: string;
     updated_at: number;
+    /** 嵌入模型提供商 */
+    embedding_model_provider?: string | null;
+    /** 嵌入模型名称 */
+    embedding_model?: string | null;
+    /** 嵌入模型是否可用 */
+    embedding_available?: boolean;
+    /** 检索模型配置（Dify API 返回字段名为 retrieval_model_dict） */
+    retrieval_model_dict?: RetrievalModel;
+    /** 标签 */
+    tags?: string[];
+    /** 文档形式 */
+    doc_form?: string | null;
+    /** 供应商 */
+    provider?: string;
 }
 
 /**

app/components/dify-dataset-manager/dataset-settings.tsx

@@ -1,10 +1,18 @@
-import { Form, Input, Button, Card, Spin } from 'antd';
+import { Form, Input, Button, Card, Spin, Divider, Select, Slider, InputNumber, Tooltip, Checkbox } from 'antd';
-import { SaveOutlined } from '@ant-design/icons';
+import { SaveOutlined, QuestionCircleOutlined, CheckCircleFilled } from '@ant-design/icons';
-import { useDatasetSettings } from '~/hooks/dify-dataset-manager/dataset-settings';
+import { useDatasetSettings, type SearchMethod } from '~/hooks/dify-dataset-manager/dataset-settings';
 import type { DatasetSettingsProps } from '~/types/dify-dataset-manager/dataset-settings';
 
 const { TextArea } = Input;
 
+// 检索方式选项
+const SEARCH_METHOD_OPTIONS: { label: string; value: SearchMethod; description: string }[] = [
+    { label: '向量检索', value: 'semantic_search', description: '基于语义理解的智能检索，适合需要理解上下文的场景' },
+    { label: '全文检索', value: 'full_text_search', description: '基于关键词匹配的传统检索方式' },
+    { label: '混合检索', value: 'hybrid_search', description: '结合向量和全文检索，综合效果最佳' },
+    { label: '关键字检索', value: 'keyword_search', description: '精确关键字匹配' },
+];
+
 /**
  * 知识库设置组件
  * 用于修改知识库名称和描述
@@ -18,11 +26,19 @@ export default function DatasetSettings({
     const {
         saving,
         hasChanges,
+        retrievalSettings,
         handleValuesChange,
         handleSave,
         handleReset,
+        updateRetrievalSettings,
     } = useDatasetSettings(dataset, form, onDatasetUpdated);
 
+    // 是否需要显示 Reranking 提示（语义检索和混合检索需要，且强制开启）
+    const showRerankingInfo = retrievalSettings.searchMethod === 'semantic_search' || retrievalSettings.searchMethod === 'hybrid_search';
+    // 权重设置：由于 Reranking 强制开启，混合检索时由 Reranking 模型决定排序，不需要手动设置权重
+    // 所以这里始终不显示权重设置
+    const showWeightsOption = false;
+
     if (!dataset) {
         return (
             <div className="settings-loading">
@@ -81,12 +97,6 @@ export default function DatasetSettings({
                                 {dataset.indexing_technique === 'high_quality' ? '高质量' : '经济'}
                             </span>
                         </div>
-                        {/* <div className="info-item">
-                            <span className="info-label">Embedding 模型：</span>
-                            <span className="info-value">
-                                {dataset.embedding_model || '默认模型'}
-                            </span>
-                        </div> */}
                         <div className="info-item">
                             <span className="info-label">文档数量：</span>
                             <span className="info-value">{dataset.document_count}</span>
@@ -102,24 +112,167 @@ export default function DatasetSettings({
                             </span>
                         </div>
                     </div>
-
-                    {/* 操作按钮 */}
-                    <div className="form-actions">
-                        <Button onClick={handleReset} disabled={!hasChanges}>
-                            重置
-                        </Button>
-                        <Button
-                            type="primary"
-                            icon={<SaveOutlined />}
-                            onClick={handleSave}
-                            loading={saving}
-                            disabled={!hasChanges}
-                        >
-                            保存
-                        </Button>
-                    </div>
                 </Form>
             </Card>
+
+            {/* 检索设置卡片 */}
+            <Card className="settings-card" style={{ marginTop: 16 }}>
+                <h3 style={{ marginBottom: 16, fontSize: 16, fontWeight: 500 }}>
+                    检索设置
+                    <Tooltip title="配置知识库的默认检索参数，影响召回效果">
+                        <QuestionCircleOutlined style={{ marginLeft: 8, color: '#8c8c8c', fontSize: 14 }} />
+                    </Tooltip>
+                </h3>
+
+                {/* 检索方式 */}
+                <div className="setting-item" style={{ marginBottom: 16 }}>
+                    <label style={{ display: 'block', marginBottom: 8, fontWeight: 500 }}>
+                        检索方式
+                    </label>
+                    <Select
+                        value={retrievalSettings.searchMethod}
+                        onChange={(value) => updateRetrievalSettings('searchMethod', value)}
+                        style={{ width: '100%' }}
+                        options={SEARCH_METHOD_OPTIONS.map(opt => ({
+                            value: opt.value,
+                            label: (
+                                <div>
+                                    <span>{opt.label}</span>
+                                    <span style={{ color: '#8c8c8c', fontSize: 12, marginLeft: 8 }}>
+                                        {opt.description}
+                                    </span>
+                                </div>
+                            ),
+                        }))}
+                    />
+                </div>
+
+                {/* Reranking 设置（语义检索和混合检索时显示，默认开启不可关闭） */}
+                {showRerankingInfo && (
+                    <div className="setting-item" style={{ marginBottom: 16 }}>
+                        <Checkbox
+                            checked={true}
+                            disabled={true}
+                        >
+                            <span style={{ color: '#262626' }}>
+                                启用 Reranking 模型
+                                <CheckCircleFilled style={{ marginLeft: 6, color: '#52c41a', fontSize: 12 }} />
+                            </span>
+                            <Tooltip title="Reranking 模型已默认开启，用于对检索结果进行重新排序，提高相关性">
+                                <QuestionCircleOutlined style={{ marginLeft: 4, color: '#8c8c8c' }} />
+                            </Tooltip>
+                        </Checkbox>
+                    </div>
+                )}
+
+                {/* 混合检索权重设置 */}
+                {showWeightsOption && (
+                    <div className="setting-item" style={{ marginBottom: 16 }}>
+                        <label style={{ display: 'block', marginBottom: 8, fontWeight: 500 }}>
+                            语义权重
+                            <Tooltip title="混合检索中语义检索的权重，值越大语义检索占比越高">
+                                <QuestionCircleOutlined style={{ marginLeft: 4, color: '#8c8c8c' }} />
+                            </Tooltip>
+                        </label>
+                        <div style={{ display: 'flex', alignItems: 'center', gap: 12 }}>
+                            <span style={{ fontSize: 12, color: '#8c8c8c' }}>关键词</span>
+                            <Slider
+                                value={retrievalSettings.weights}
+                                onChange={(value) => updateRetrievalSettings('weights', value)}
+                                min={0}
+                                max={1}
+                                step={0.1}
+                                style={{ flex: 1 }}
+                            />
+                            <span style={{ fontSize: 12, color: '#8c8c8c' }}>语义</span>
+                            <InputNumber
+                                value={retrievalSettings.weights}
+                                onChange={(value) => updateRetrievalSettings('weights', value ?? 0.7)}
+                                min={0}
+                                max={1}
+                                step={0.1}
+                                size="small"
+                                style={{ width: 70 }}
+                            />
+                        </div>
+                    </div>
+                )}
+
+                <Divider />
+
+                {/* Top K 设置 */}
+                <div className="setting-item" style={{ marginBottom: 16 }}>
+                    <label style={{ display: 'block', marginBottom: 8, fontWeight: 500 }}>
+                        返回数量 (Top K)
+                        <Tooltip title="每次检索返回的最大结果数量">
+                            <QuestionCircleOutlined style={{ marginLeft: 4, color: '#8c8c8c' }} />
+                        </Tooltip>
+                    </label>
+                    <div style={{ display: 'flex', alignItems: 'center', gap: 12 }}>
+                        <Slider
+                            value={retrievalSettings.topK}
+                            onChange={(value) => updateRetrievalSettings('topK', value)}
+                            min={1}
+                            max={20}
+                            style={{ flex: 1 }}
+                        />
+                        <InputNumber
+                            value={retrievalSettings.topK}
+                            onChange={(value) => updateRetrievalSettings('topK', value ?? 3)}
+                            min={1}
+                            max={20}
+                            size="small"
+                            style={{ width: 70 }}
+                        />
+                    </div>
+                </div>
+
+                {/* Score 阈值设置（默认开启不可关闭，但可调节数值） */}
+                <div className="setting-item" style={{ marginBottom: 16 }}>
+                    <label style={{ display: 'block', marginBottom: 8, fontWeight: 500 }}>
+                        Score 阈值
+                        <CheckCircleFilled style={{ marginLeft: 6, color: '#52c41a', fontSize: 12 }} />
+                        <Tooltip title="Score 阈值已默认开启，只返回相似度分数高于阈值的结果，过滤低质量匹配">
+                            <QuestionCircleOutlined style={{ marginLeft: 4, color: '#8c8c8c' }} />
+                        </Tooltip>
+                    </label>
+                    <div style={{ display: 'flex', alignItems: 'center', gap: 12 }}>
+                        <Slider
+                            value={retrievalSettings.scoreThreshold}
+                            onChange={(value) => updateRetrievalSettings('scoreThreshold', value)}
+                            min={0}
+                            max={1}
+                            step={0.01}
+                            style={{ flex: 1 }}
+                        />
+                        <InputNumber
+                            value={retrievalSettings.scoreThreshold}
+                            onChange={(value) => updateRetrievalSettings('scoreThreshold', value ?? 0.5)}
+                            min={0}
+                            max={1}
+                            step={0.01}
+                            size="small"
+                            style={{ width: 70 }}
+                        />
+                    </div>
+                </div>
+
+                {/* 操作按钮 */}
+                <div className="form-actions" style={{ marginTop: 24, display: 'flex', justifyContent: 'flex-end', gap: 8 }}>
+                    <Button onClick={handleReset} disabled={!hasChanges}>
+                        重置
+                    </Button>
+                    <Button
+                        type="primary"
+                        icon={<SaveOutlined />}
+                        onClick={handleSave}
+                        loading={saving}
+                        disabled={!hasChanges}
+                    >
+                        保存
+                    </Button>
+                </div>
+            </Card>
         </div>
     );
 }

app/components/dify-dataset-manager/retrieve-test.tsx

@@ -1,5 +1,5 @@
 import { SearchOutlined, FileSearchOutlined } from '@ant-design/icons';
-import { Button, Tag, Input, Slider, Spin, Select, Flex } from 'antd';
+import { Button, Tag, Input, Slider, Spin, Select, Flex, Switch, InputNumber, Tooltip } from 'antd';
 import type { RetrieveRecord } from '~/api/dify-dataset/type';
 import { useRetrieveTest } from '~/hooks/dify-dataset-manager/retrieve-test';
 import type { RetrieveTestProps } from '~/types/dify-dataset-manager/retrieve-test';
@@ -97,6 +97,10 @@ export default function RetrieveTest({ datasetId }: RetrieveTestProps) {
         setSearchMethod,
         topK,
         setTopK,
+        scoreThresholdEnabled,
+        setScoreThresholdEnabled,
+        scoreThreshold,
+        setScoreThreshold,
         handleRetrieve,
     } = useRetrieveTest(datasetId);
 
@@ -229,6 +233,46 @@ export default function RetrieveTest({ datasetId }: RetrieveTestProps) {
                             {topK}
                         </span>
                     </Flex>
+
+                    {/* Score 阈值设置 */}
+                    <Flex align="center" gap={12}>
+                        <Tooltip title="开启后，只返回相似度分数高于阈值的结果">
+                            <span style={{
+                                fontSize: 13,
+                                color: colors.textSecondary,
+                                whiteSpace: 'nowrap',
+                                cursor: 'help',
+                            }}>
+                                Score 阈值:
+                            </span>
+                        </Tooltip>
+                        <Switch
+                            size="small"
+                            checked={scoreThresholdEnabled}
+                            onChange={setScoreThresholdEnabled}
+                        />
+                        {scoreThresholdEnabled && (
+                            <>
+                                <Slider
+                                    value={scoreThreshold}
+                                    onChange={setScoreThreshold}
+                                    min={0}
+                                    max={1}
+                                    step={0.01}
+                                    style={{ flex: 1 }}
+                                />
+                                <InputNumber
+                                    value={scoreThreshold}
+                                    onChange={(value) => setScoreThreshold(value ?? 0.5)}
+                                    min={0}
+                                    max={1}
+                                    step={0.01}
+                                    size="small"
+                                    style={{ width: 70 }}
+                                />
+                            </>
+                        )}
+                    </Flex>
                 </Flex>
             </Flex>
 

app/hooks/dify-dataset-manager/dataset-settings.ts

@@ -1,8 +1,77 @@
 import { useState, useEffect, useCallback } from 'react';
 import { message } from 'antd';
 import type { FormInstance } from 'antd';
-import type { Dataset } from '~/api/dify-dataset/type/datasetTypes';
+import type { Dataset, RetrievalModel } from '~/api/dify-dataset/type/datasetTypes';
-import { updateDatasetName } from '~/api/dify-dataset/api/datasetApi';
+import { updateDatasetSettings, fetchDataset } from '~/api/dify-dataset/api/datasetApi';
+import { DIFY_CONFIG } from '~/config/api-config';
+
+/**
+ * 检索方法类型
+ */
+export type SearchMethod = 'keyword_search' | 'semantic_search' | 'full_text_search' | 'hybrid_search';
+
+/**
+ * 检索设置表单值
+ */
+export interface RetrievalSettingsFormValues {
+    searchMethod: SearchMethod;
+    topK: number;
+    scoreThresholdEnabled: boolean;
+    scoreThreshold: number;
+    rerankingEnable: boolean;
+    weights: number; // 混合检索的语义权重 (0-1)
+}
+
+/**
+ * 默认检索设置
+ */
+const DEFAULT_RETRIEVAL_SETTINGS: RetrievalSettingsFormValues = {
+    searchMethod: 'semantic_search',
+    topK: 3,
+    scoreThresholdEnabled: true,  // 默认开启
+    scoreThreshold: 0.5,
+    rerankingEnable: true,        // 默认开启
+    weights: 0.7,
+};
+
+/**
+ * 从 Dataset 的 retrieval_model 转换为表单值
+ */
+function retrievalModelToFormValues(model?: RetrievalModel): RetrievalSettingsFormValues {
+    if (!model) {
+        return { ...DEFAULT_RETRIEVAL_SETTINGS };
+    }
+    return {
+        searchMethod: model.search_method || 'semantic_search',
+        topK: model.top_k ?? 3,
+        scoreThresholdEnabled: model.score_threshold_enabled ?? false,
+        scoreThreshold: model.score_threshold ?? 0.5,
+        rerankingEnable: model.reranking_enable ?? false,
+        weights: model.weights ?? 0.7,
+    };
+}
+
+/**
+ * 从表单值转换为 API 请求的 retrieval_model
+ */
+function formValuesToRetrievalModel(values: RetrievalSettingsFormValues): RetrievalModel {
+    // 语义检索和混合检索需要 Reranking，强制开启
+    const needReranking = values.searchMethod === 'semantic_search' || values.searchMethod === 'hybrid_search';
+
+    return {
+        search_method: values.searchMethod,
+        reranking_enable: needReranking,  // 强制开启，不受用户控制
+        reranking_mode: null,
+        reranking_model: {
+            reranking_provider_name: DIFY_CONFIG.rerankingProviderName,
+            reranking_model_name: DIFY_CONFIG.rerankingModelName,
+        },
+        weights: values.searchMethod === 'hybrid_search' ? values.weights : null,
+        top_k: values.topK,
+        score_threshold_enabled: true,    // 强制开启，不受用户控制
+        score_threshold: values.scoreThreshold,  // 用户可调节数值
+    };
+}
 
 /**
  * 知识库设置状态管理 Hook
@@ -15,6 +84,11 @@ export function useDatasetSettings(
     const [saving, setSaving] = useState(false);
     const [hasChanges, setHasChanges] = useState(false);
 
+    // 检索设置状态（注意：Dify API 返回的字段名是 retrieval_model_dict）
+    const [retrievalSettings, setRetrievalSettings] = useState<RetrievalSettingsFormValues>(
+        () => retrievalModelToFormValues(dataset?.retrieval_model_dict)
+    );
+
     // 初始化表单数据
     useEffect(() => {
         if (dataset) {
@@ -22,20 +96,53 @@ export function useDatasetSettings(
                 name: dataset.name,
                 description: dataset.description || '',
             });
+            console.log('[DatasetSettings] 初始化检索设置, retrieval_model_dict:', dataset.retrieval_model_dict);
+            setRetrievalSettings(retrievalModelToFormValues(dataset.retrieval_model_dict));
             setHasChanges(false);
         }
     }, [dataset, form]);
 
+    /**
+     * 更新检索设置
+     */
+    const updateRetrievalSettings = useCallback(<K extends keyof RetrievalSettingsFormValues>(
+        key: K,
+        value: RetrievalSettingsFormValues[K]
+    ) => {
+        setRetrievalSettings(prev => {
+            const newSettings = { ...prev, [key]: value };
+            // 检查是否有变化
+            checkForChanges(newSettings);
+            return newSettings;
+        });
+    }, [dataset]);
+
+    /**
+     * 检查是否有变化
+     */
+    const checkForChanges = useCallback((newRetrievalSettings?: RetrievalSettingsFormValues) => {
+        const values = form.getFieldsValue();
+        const currentRetrieval = newRetrievalSettings || retrievalSettings;
+        const originalRetrieval = retrievalModelToFormValues(dataset?.retrieval_model_dict);
+
+        const nameChanged = values.name !== dataset?.name;
+        const retrievalChanged =
+            currentRetrieval.searchMethod !== originalRetrieval.searchMethod ||
+            currentRetrieval.topK !== originalRetrieval.topK ||
+            currentRetrieval.scoreThresholdEnabled !== originalRetrieval.scoreThresholdEnabled ||
+            currentRetrieval.scoreThreshold !== originalRetrieval.scoreThreshold ||
+            currentRetrieval.rerankingEnable !== originalRetrieval.rerankingEnable ||
+            currentRetrieval.weights !== originalRetrieval.weights;
+
+        setHasChanges(nameChanged || retrievalChanged);
+    }, [form, dataset, retrievalSettings]);
+
     /**
      * 处理表单值变化
      */
     const handleValuesChange = useCallback(() => {
-        const values = form.getFieldsValue();
+        checkForChanges();
-        const changed =
+    }, [checkForChanges]);
-            values.name !== dataset?.name ||
-            values.description !== (dataset?.description || '');
-        setHasChanges(changed);
-    }, [form, dataset]);
 
     /**
      * 保存设置
@@ -50,11 +157,18 @@ export function useDatasetSettings(
             const values = await form.validateFields();
             setSaving(true);
 
-            // 目前只支持修改名称
+            // 构建完整的更新请求
-            const updatedDataset = await updateDatasetName(dataset.id, values.name);
+            await updateDatasetSettings(dataset.id, {
+                name: values.name,
+                retrieval_model: formValuesToRetrievalModel(retrievalSettings),
+            });
+
+            // PATCH 接口返回的数据可能不完整，重新获取详情
+            const fullDataset = await fetchDataset(dataset.id);
+            console.log('[DatasetSettings] 保存后获取完整数据:', fullDataset);
 
             message.success('保存成功');
-            onDatasetUpdated(updatedDataset);
+            onDatasetUpdated(fullDataset);
             setHasChanges(false);
         } catch (err: any) {
             console.error('保存设置失败:', err);
@@ -62,7 +176,7 @@ export function useDatasetSettings(
         } finally {
             setSaving(false);
         }
-    }, [dataset, form, onDatasetUpdated]);
+    }, [dataset, form, retrievalSettings, onDatasetUpdated]);
 
     /**
      * 重置表单
@@ -73,6 +187,7 @@ export function useDatasetSettings(
                 name: dataset.name,
                 description: dataset.description || '',
             });
+            setRetrievalSettings(retrievalModelToFormValues(dataset.retrieval_model_dict));
             setHasChanges(false);
         }
     }, [dataset, form]);
@@ -81,11 +196,13 @@ export function useDatasetSettings(
         // 状态
         saving,
         hasChanges,
+        retrievalSettings,
 
         // 方法
         handleValuesChange,
         handleSave,
         handleReset,
+        updateRetrievalSettings,
     };
 }
 

app/hooks/dify-dataset-manager/index.ts

@@ -2,7 +2,7 @@ import { useState, useEffect, useCallback } from 'react';
 import { message } from 'antd';
 import type { Dataset } from '~/api/dify-dataset/type/datasetTypes';
 import type { Document } from '~/api/dify-dataset/type/documentTypes';
-import { fetchDatasets } from '~/api/dify-dataset/api/datasetApi';
+import { fetchDatasets, fetchDataset } from '~/api/dify-dataset/api/datasetApi';
 import { fetchDocuments } from '~/api/dify-dataset/api/documentApi';
 import type { MenuTab } from '~/types/dify-dataset-manager/layout';
 import { DEFAULT_DOCUMENT_PAGE_SIZE } from '~/types/dify-dataset-manager/index';
@@ -58,20 +58,26 @@ export function useDatasetManager() {
     }, [documentPageSize]);
 
     /**
-     * 加载知识库（获取第一个知识库）
+     * 加载知识库（获取第一个知识库，再获取详情以包含 retrieval_model）
      */
     const loadDataset = useCallback(async () => {
         setLoadingDataset(true);
         try {
             console.log('[DatasetManager] 加载知识库...');
+            // 先获取列表，找到第一个知识库的 ID
             const response = await fetchDatasets(1, 1);
-            console.log('[DatasetManager] 知识库响应:', response);
+            console.log('[DatasetManager] 知识库列表响应:', response);
 
             if (response && response.data && response.data.length > 0) {
-                const firstDataset = response.data[0];
+                const firstDatasetId = response.data[0].id;
-                setDataset(firstDataset);
+
+                // 再获取详情，包含完整的 retrieval_model 等字段
+                const fullDataset = await fetchDataset(firstDatasetId);
+                console.log('[DatasetManager] 知识库详情响应:', fullDataset);
+
+                setDataset(fullDataset);
                 // 立即加载文档
-                await loadDocuments(firstDataset.id, 1);
+                await loadDocuments(firstDatasetId, 1);
             } else {
                 setError('未找到知识库，请先在Dify中创建知识库');
             }

app/hooks/dify-dataset-manager/retrieve-test.ts

@@ -9,7 +9,12 @@ import type { SearchMethod } from '~/types/dify-dataset-manager/retrieve-test';
  * 构建完整的 retrieval_model 参数（匹配 Dify API 规范）
  * 根据检索方式启用 Reranking（语义搜索和混合搜索需要启用）
  */
-function buildRetrievalModel(searchMethod: SearchMethod, topK: number): RetrievalModel {
+function buildRetrievalModel(
+    searchMethod: SearchMethod,
+    topK: number,
+    scoreThresholdEnabled: boolean,
+    scoreThreshold: number
+): RetrievalModel {
     // 语义搜索和混合搜索需要启用 Reranking
     const needReranking = searchMethod === 'semantic_search' || searchMethod === 'hybrid_search';
 
@@ -23,8 +28,8 @@ function buildRetrievalModel(searchMethod: SearchMethod, topK: number): Retrieva
         },
         weights: null,
         top_k: topK,
-        score_threshold_enabled: false,
+        score_threshold_enabled: scoreThresholdEnabled,
-        score_threshold: null,
+        score_threshold: scoreThresholdEnabled ? scoreThreshold : null,
     };
 }
 
@@ -38,6 +43,9 @@ export function useRetrieveTest(datasetId: string) {
     // 默认使用语义搜索
     const [searchMethod, setSearchMethod] = useState<SearchMethod>('semantic_search');
     const [topK, setTopK] = useState<number>(5);
+    // Score 阈值相关状态
+    const [scoreThresholdEnabled, setScoreThresholdEnabled] = useState(false);
+    const [scoreThreshold, setScoreThreshold] = useState<number>(0.5);
 
     /**
      * 执行检索
@@ -55,7 +63,7 @@ export function useRetrieveTest(datasetId: string) {
 
         setRetrieving(true);
         try {
-            const retrievalModel = buildRetrievalModel(searchMethod, topK);
+            const retrievalModel = buildRetrievalModel(searchMethod, topK, scoreThresholdEnabled, scoreThreshold);
             console.log('[Hook] 检索参数:', { datasetId, query: searchQuery, retrievalModel });
 
             const response = await retrieveDataset(datasetId, searchQuery, retrievalModel);
@@ -69,7 +77,7 @@ export function useRetrieveTest(datasetId: string) {
         } finally {
             setRetrieving(false);
         }
-    }, [datasetId, searchQuery, searchMethod, topK]);
+    }, [datasetId, searchQuery, searchMethod, topK, scoreThresholdEnabled, scoreThreshold]);
 
     return {
         // 状态
@@ -81,6 +89,10 @@ export function useRetrieveTest(datasetId: string) {
         setSearchMethod,
         topK,
         setTopK,
+        scoreThresholdEnabled,
+        setScoreThresholdEnabled,
+        scoreThreshold,
+        setScoreThreshold,
 
         // 方法
         handleRetrieve,

app/routes/api.dataset.datasets.$datasetId.tsx

@@ -54,15 +54,22 @@ export async function loader({ request, params }: LoaderFunctionArgs) {
 }
 
 /**
- * PATCH /api/dataset/datasets/:datasetId - 修改知识库名称
+ * PATCH /api/dataset/datasets/:datasetId - 修改知识库设置
  *
  * Dify API: PATCH /datasets/{dataset_id}
  *
- * 请求体: { "name": "新的知识库名称" }
+ * 请求体支持以下字段:
+ * - name (string): 知识库名称（必填）
+ * - retrieval_model (object): 检索模型配置（选填）
+ *   - search_method: 'keyword_search' | 'semantic_search' | 'full_text_search' | 'hybrid_search'
+ *   - reranking_enable: boolean
+ *   - reranking_model: { reranking_provider_name, reranking_model_name }
+ *   - weights: number | null (混合检索的语义权重)
+ *   - top_k: number
+ *   - score_threshold_enabled: boolean
+ *   - score_threshold: number | null
  *
- * 注意：
+ * 注意：删除知识库功能不对外开放
- * - 仅允许修改知识库名称，其他字段不开放修改
- * - 删除知识库功能不对外开放
  */
 export async function action({ request, params }: ActionFunctionArgs) {
     try {
@@ -89,7 +96,7 @@ export async function action({ request, params }: ActionFunctionArgs) {
         if (method === 'PATCH') {
             const body = await request.json();
 
-            // 只允许修改 name 字段
+            // name 是必填字段
             if (!body.name || typeof body.name !== 'string') {
                 return new Response(
                     JSON.stringify({ error: '请提供有效的知识库名称 (name)' }),
@@ -97,17 +104,48 @@ export async function action({ request, params }: ActionFunctionArgs) {
                 );
             }
 
-            // 只传递 name 字段，忽略其他字段
+            const trimmedName = body.name.trim();
-            const allowedBody = { name: body.name.trim() };
+            if (trimmedName.length === 0) {
-
-            if (allowedBody.name.length === 0) {
                 return new Response(
                     JSON.stringify({ error: '知识库名称不能为空' }),
                     { status: 400, headers: { 'Content-Type': 'application/json' } }
                 );
             }
 
-            console.log('[API] Update Dataset Name:', { datasetId, name: allowedBody.name });
+            // 构建允许的请求体
+            const allowedBody: Record<string, any> = {
+                name: trimmedName,
+            };
+
+            // 可选: retrieval_model 检索模型配置
+            if (body.retrieval_model && typeof body.retrieval_model === 'object') {
+                const rm = body.retrieval_model;
+
+                // 验证 search_method
+                const validSearchMethods = ['keyword_search', 'semantic_search', 'full_text_search', 'hybrid_search'];
+                if (rm.search_method && !validSearchMethods.includes(rm.search_method)) {
+                    return new Response(
+                        JSON.stringify({ error: '无效的检索方法 (search_method)' }),
+                        { status: 400, headers: { 'Content-Type': 'application/json' } }
+                    );
+                }
+
+                allowedBody.retrieval_model = {
+                    search_method: rm.search_method,
+                    reranking_enable: rm.reranking_enable ?? false,
+                    reranking_mode: rm.reranking_mode ?? null,
+                    reranking_model: rm.reranking_model ?? {
+                        reranking_provider_name: '',
+                        reranking_model_name: '',
+                    },
+                    weights: rm.weights ?? null,
+                    top_k: rm.top_k ?? 3,
+                    score_threshold_enabled: rm.score_threshold_enabled ?? false,
+                    score_threshold: rm.score_threshold_enabled ? (rm.score_threshold ?? null) : null,
+                };
+            }
+
+            console.log('[API] Update Dataset Settings:', { datasetId, body: allowedBody });
 
             const apiUrl = `${API_BASE_URL}/dify_dataset/datasets/${datasetId}`;
             const response = await fetch(apiUrl, {

app/types/dify-dataset-manager/retrieve-test.ts

@@ -21,6 +21,8 @@ export type SearchMethod = 'semantic_search' | 'full_text_search' | 'hybrid_sear
 export interface RetrieveOptions {
     searchMethod: SearchMethod;
     topK: number;
+    scoreThresholdEnabled: boolean;
+    scoreThreshold: number;
 }
 
 /**
@@ -32,4 +34,6 @@ export interface RetrieveTestState {
     retrieving: boolean;
     searchMethod: SearchMethod;
     topK: number;
+    scoreThresholdEnabled: boolean;
+    scoreThreshold: number;
 }