redis缓存替换+pgvector向量替换

ai-chat-service/dev.config.yaml

@@ -53,7 +53,37 @@ dependOn:
     address: "localhost:50054"
     accessToken: "ang1chubdev1ozhome256487d22sapguuv1ozhom"
   tokenizer:
-    address: "http://192.168.239.161:3002"
+    address: "http://127.0.0.1:3002"
+vector:
+  # 向量后端：tencent / pgvector
+  provider: "pgvector"
+  # 历史问答命中阈值
+  threshold: 0.99
+  tencent:
+    url: "http://lb-4u4r1fk4-1ys6gv3rpmdan420.clb.ap-guangzhou.tencentclb.com:60000"
+    username: "root"
+    pwd: "YaUfVueWZJ20e4ghyLlBT8Dou5OapwpFTUq50oft"
+    database: "ai-chat"
+    timeout: 5
+    maxIdleConnPerHost: 2
+    readConsistency: "eventualConsistency"
+    idleConnTimeout: 60
+  pgvector:
+    dsn: "postgres://postgres:postgres@127.0.0.1:15432/ai_chat?sslmode=disable"
+    table: "chat_record_vectors"
+    dimensions: 1024
+    maxLifeTime: 3600
+    maxOpenConn: 10
+    maxIdleConn: 10
+embedding:
+  provider: "openai-compatible"
+  # 智谱 OpenAI 兼容网关；可被项目根目录 .env 覆盖
+  base_url: "https://open.bigmodel.cn/api/paas/v4"
+  # 默认故意设成错误值，真实 key 请放到项目根目录 .env
+  api_key: "__INVALID_SET_AI_CHAT_EMBEDDING_API_KEY__"
+  # embedding-2 固定 1024 维，和当前 pgvector 表结构一致
+  model: "embedding-2"
+  timeout: 10
 vectorDB:
   # 访问地址
   url: "http://lb-4u4r1fk4-1ys6gv3rpmdan420.clb.ap-guangzhou.tencentclb.com:60000"
@@ -69,4 +99,4 @@ vectorDB:
   # 读一致性： strongConsistency(强一致性)，eventualConsistency(最终一致性)
   readConsistency: "eventualConsistency"
   # 空闲连接超时时长s
-  idleConnTimeout: 60
+  idleConnTimeout: 60