feat: 知识库朴素语义搜索（输入→检索→结果列表）

2026-05-27 19:25:41 +08:00
parent 0c7d2d7400
commit 783e212bb1
4 changed files with 193 additions and 8 deletions
@@ -1,8 +1,9 @@
 """
-知识库 API — 上传 / 列表 / 删除 / 来源筛选
+知识库 API — 上传 / 列表 / 删除 / 来源筛选 / 语义搜索
 """

 from typing import Optional
+from pydantic import BaseModel

 from fastapi import APIRouter, Depends, HTTPException, UploadFile, File, Query, status
 from sqlmodel import Session
@@ -104,3 +105,28 @@ def get_grouped_knowledge_items(
    """
    svc = KnowledgeService()
    return svc.get_grouped_items()
+
+
+class SearchRequest(BaseModel):
+    query: str
+    top_k: int = 5
+
+
+@router.post("/search")
+def search_knowledge(
+    body: SearchRequest,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """
+    语义检索：输入一段文字，返回最相关的知识库条目及相似度。
+    查询向量用 type="query"（区分于存入时的 type="db"）。
+    三角色均可读。
+    """
+    svc = KnowledgeService()
+    results = svc.search_similar(query_text=body.query, top_k=body.top_k)
+    return {
+        "results": results,
+        "query": body.query,
+        "count": len(results),
+    }
@@ -232,7 +232,10 @@ class KnowledgeService:
    def search_similar(self, query_text: str, top_k: int = 5) -> list[dict]:
        """
        语义检索：查询句转为向量，用 SQL 余弦距离（<=>）在数据库层检索
-        返回 top_k 条相似笔记，含相似度分数
+        返回 top_k 条相似笔记，含相似度分数 + 原文片段（SQL 端截断前 200 字）。
+
+        注意：当前取前 200 字是已知妥协（整篇向量检索无法定位中段命中点），
+        Phase 4a 做切块检索（chunk）时可优化为取最相关片段。
        """
        query_vector = self.embedder.embed_single(query_text, embed_type="query")
        vec_str = "[" + ",".join(str(v) for v in query_vector) + "]"
@@ -243,6 +246,9 @@ class KnowledgeService:
                    ki.id,
                    ki.title,
                    ki.source_type,
+                    ki.author,
+                    ki.tags,
+                    SUBSTRING(ki.content_md, 1, 200) AS snippet,
                    1 - (ke.embedding <=> '{vec_str}'::vector) AS similarity
                FROM knowledge_embeddings ke
                JOIN knowledge_items ki ON ke.knowledge_id = ki.id
@@ -252,10 +258,20 @@ class KnowledgeService:
            """
            stmt = text(sql)
            rows = session.execute(stmt).all()
-            return [
-                {"id": r.id, "title": r.title, "source_type": r.source_type, "similarity": round(r.similarity, 4)}
-                for r in rows
-            ]
+            results = []
+            for r in rows:
+                tags = r.tags or {}
+                source_detail = tags.get("source_detail") if isinstance(tags, dict) else None
+                results.append({
+                    "id": r.id,
+                    "title": r.title,
+                    "source_type": r.source_type,
+                    "author": r.author,
+                    "source_detail": source_detail,
+                    "snippet": r.snippet,
+                    "similarity": round(r.similarity, 4),
+                })
+            return results

    def get_item_count(self) -> int:
        with Session(engine) as session:
@@ -1,6 +1,6 @@
 import { useState, useEffect } from 'react'
-import { Card, Upload, Table, Button, Space, Select, Popconfirm, message, Tag } from 'antd'
-import { DeleteOutlined, ReloadOutlined, UploadOutlined } from '@ant-design/icons'
+import { Card, Upload, Table, Button, Space, Select, Popconfirm, message, Tag, Input } from 'antd'
+import { DeleteOutlined, ReloadOutlined, UploadOutlined, SearchOutlined, ClearOutlined } from '@ant-design/icons'
 import useAuthStore from '../../stores/authStore'
 import knowledgeService from '../../services/knowledgeService'
 import KnowledgeTree from '../../components/KnowledgeTree/KnowledgeTree'
@@ -34,6 +34,11 @@ export default function KnowledgeBase() {
  const [sourceDetailFilter, setSourceDetailFilter] = useState('')
  const [selectedTreeNode, setSelectedTreeNode] = useState(null) // { type, detail }

+  // 搜索状态
+  const [searchQuery, setSearchQuery] = useState('')
+  const [searchResults, setSearchResults] = useState([])
+  const [searchLoading, setSearchLoading] = useState(false)
+
  const fetchItems = async () => {
    setLoading(true)
    try {
@@ -109,6 +114,30 @@ export default function KnowledgeBase() {
    }
  }

+  // 搜索处理
+  const handleSearch = async () => {
+    if (!searchQuery.trim()) {
+      setSearchResults([])
+      return
+    }
+    setSearchLoading(true)
+    try {
+      const data = await knowledgeService.searchItems(searchQuery.trim())
+      setSearchResults(data.results || [])
+    } catch {
+      message.error('搜索失败')
+      setSearchResults([])
+    } finally {
+      setSearchLoading(false)
+    }
+  }
+
+  // 清空搜索 → 恢复树形浏览（不丢树状态）
+  const handleClearSearch = () => {
+    setSearchQuery('')
+    setSearchResults([])
+  }
+
  // 树节点选中 → 联动过滤
  const handleNodeSelect = (node) => {
    setSelectedTreeNode(node)
@@ -229,6 +258,107 @@ export default function KnowledgeBase() {
        <span style={{ fontSize: 12, color: '#aaa' }}>支持批量上传，系统自动解析 yaml frontmatter 并生成语义向量</span>
      </div>

+      {/* 搜索栏（搜索模式/浏览模式互斥，清空恢复树形浏览不丢状态） */}
+      <div style={{ display: 'flex', alignItems: 'center', gap: 8, marginBottom: 12 }}>
+        <Input
+          placeholder="输入一段文字，语义搜索知识库…"
+          value={searchQuery}
+          onChange={e => setSearchQuery(e.target.value)}
+          onPressEnter={handleSearch}
+          style={{ width: 340, borderRadius: 10 }}
+          prefix={<SearchOutlined style={{ color: '#aaa' }} />}
+          suffix={
+            searchQuery ? (
+              <ClearOutlined
+                style={{ color: '#aaa', cursor: 'pointer' }}
+                onClick={handleClearSearch}
+              />
+            ) : null
+          }
+        />
+        <Button
+          icon={<SearchOutlined />}
+          onClick={handleSearch}
+          loading={searchLoading}
+          style={{ borderRadius: 10 }}
+        >
+          搜索
+        </Button>
+        {searchQuery && (
+          <span style={{ fontSize: 12, color: '#888' }}>
+            {searchResults.length > 0
+              ? `找到 ${searchResults.length} 条相关结果`
+              : '无相关结果'}
+          </span>
+        )}
+      </div>
+
+      {/* 搜索结果列表（搜索模式显示，浏览模式隐藏） */}
+      {searchQuery && searchResults.length > 0 && (
+        <div style={{ marginBottom: 12 }}>
+          <div style={{ fontSize: 13, color: '#666', marginBottom: 8, fontWeight: 500 }}>
+            搜索结果
+          </div>
+          {searchResults.map(item => (
+            <Card
+              key={item.id}
+              size="small"
+              style={{ borderRadius: 10, marginBottom: 8 }}
+              bodyStyle={{ padding: '12px 16px' }}
+            >
+              <div style={{ display: 'flex', justifyContent: 'space-between', alignItems: 'flex-start' }}>
+                <div style={{ flex: 1 }}>
+                  <div style={{ fontWeight: 600, fontSize: 14, color: '#3b4a3b', marginBottom: 4 }}>
+                    {item.title}
+                  </div>
+                  <div style={{ display: 'flex', gap: 8, alignItems: 'center', marginBottom: 6 }}>
+                    <Tag color="default" style={{ borderRadius: 6, fontSize: 12 }}>
+                      {SOURCE_TYPE_LABEL[item.source_type] || item.source_type}
+                    </Tag>
+                    {item.author && (
+                      <span style={{ fontSize: 12, color: '#888' }}>作者：{item.author}</span>
+                    )}
+                    {item.source_detail && (
+                      <span style={{ fontSize: 12, color: '#888' }}>出处：{item.source_detail}</span>
+                    )}
+                  </div>
+                  <div
+                    style={{
+                      fontSize: 12,
+                      color: '#555',
+                      background: '#f5f5f5',
+                      borderRadius: 6,
+                      padding: '6px 10px',
+                      lineHeight: 1.6,
+                      overflow: 'hidden',
+                      textOverflow: 'ellipsis',
+                      display: '-webkit-box',
+                      WebkitLineClamp: 3,
+                      WebkitBoxOrient: 'vertical',
+                    }}
+                  >
+                    {item.snippet}
+                  </div>
+                </div>
+                  <div style={{
+                    marginLeft: 16,
+                    minWidth: 56,
+                    textAlign: 'center',
+                    background: item.similarity >= 0.7 ? '#e6f4ff' : '#f5f5f5',
+                    borderRadius: 8,
+                    padding: '4px 8px',
+                  }}>
+                    <div style={{ fontSize: 18, fontWeight: 700, color: item.similarity >= 0.7 ? '#1677ff' : '#666' }}>
+                      {Math.max(0, Math.round(item.similarity * 100))}%
+                    </div>
+                    <div style={{ fontSize: 11, color: '#888' }}>相关度</div>
+                  </div>
+              </div>
+            </Card>
+          ))}
+        </div>
+      )}
+
      {/* 主体：左侧树 + 右侧列表（flex 左右并列，宽度稳定） */}
      <div style={{ display: 'flex', gap: 12, alignItems: 'flex-start' }}>
        {/* 左侧树 */}
@@ -52,6 +52,19 @@ const knowledgeService = {
    const resp = await http.get('/knowledge/grouped')
    return resp.data
  },
+
+  /**
+   * 语义检索：输入一段文字，返回最相关的知识库条目
+   * @param {string} queryText - 查询文字
+   * @param {number} topK - 返回条数，默认 5
+   */
+  async searchItems(queryText, topK = 5) {
+    const resp = await http.post('/knowledge/search', {
+      query: queryText,
+      top_k: topK,
+    })
+    return resp.data
+  },
 }

 export default knowledgeService