增加向量化的库

virtual-patient-common/src/main/java/com/supervision/config/VectorEmbeddingClient.java

@@ -2,10 +2,11 @@ package com.supervision.config;
 
 import cn.hutool.http.HttpUtil;
 import cn.hutool.json.JSONUtil;
-import org.slf4j.Logger;
+import lombok.Data;
-import org.slf4j.LoggerFactory;
+import lombok.extern.slf4j.Slf4j;
 import org.springframework.ai.document.Document;
 import org.springframework.ai.embedding.*;
+import org.springframework.core.SpringProperties;
 import org.springframework.util.Assert;
 
 import java.util.ArrayList;
@@ -13,9 +14,9 @@ import java.util.List;
 import java.util.Map;
 import java.util.concurrent.atomic.AtomicInteger;
 
+@Slf4j
 public class VectorEmbeddingClient extends AbstractEmbeddingClient {
 
-    private final Logger logger = LoggerFactory.getLogger(getClass());
 
     @Override
     public List<Double> embed(Document document) {
@@ -30,17 +31,13 @@ public class VectorEmbeddingClient extends AbstractEmbeddingClient {
     @Override
     public EmbeddingResponse call(EmbeddingRequest request) {
         Assert.notEmpty(request.getInstructions(), "At least one text is required!");
-        if (request.getInstructions().size() != 1) {
-            logger.warn(
-                    "Ollama Embedding does not support batch embedding. Will make multiple API calls to embed(Document)");
-        }
         List<List<Double>> embeddingList = new ArrayList<>();
+        String embeddingUrl = SpringProperties.getProperty("embeddingUrl");
         for (String inputContent : request.getInstructions()) {
-            // TODO 这里需要吧inputContent转化为向量数据
+            // 这里需要吧inputContent转化为向量数据
-            String post = HttpUtil.post("http://192.168.10.42:8000/embeddings/", JSONUtil.toJsonStr(Map.of("text", inputContent)));
+            String post = HttpUtil.post(embeddingUrl, JSONUtil.toJsonStr(Map.of("text", inputContent)));
-            String o = JSONUtil.parseObj(post).getStr("embeddings");
+            EmbeddingData bean = JSONUtil.toBean(post, EmbeddingData.class);
-            List<Double> embedding = JSONUtil.toList(o, Double.class);
+            embeddingList.add(bean.embeddings);
-            embeddingList.add(embedding);
         }
         var indexCounter = new AtomicInteger(0);
         List<Embedding> embeddings = embeddingList.stream()
@@ -48,4 +45,9 @@ public class VectorEmbeddingClient extends AbstractEmbeddingClient {
                 .toList();
         return new EmbeddingResponse(embeddings);
     }
+
+    @Data
+    private static class EmbeddingData {
+        private List<Double> embeddings;
+    }
 }

virtual-patient-common/src/main/java/com/supervision/config/VectorSimilarityConfiguration.java

@@ -3,6 +3,7 @@ package com.supervision.config;
 import org.springframework.ai.vectorstore.RedisVectorStore;
 import org.springframework.context.annotation.Bean;
 import org.springframework.context.annotation.Configuration;
+import org.springframework.core.SpringProperties;
 
 @Configuration
 public class VectorSimilarityConfiguration {
@@ -14,11 +15,14 @@ public class VectorSimilarityConfiguration {
 
     @Bean
     public RedisVectorStore redisVectorStore(VectorEmbeddingClient vectorEmbeddingClient) {
+        String property = SpringProperties.getProperty("spring.ai.vectorstore.redis.uri");
         RedisVectorStore.RedisVectorStoreConfig config = RedisVectorStore.RedisVectorStoreConfig.builder()
-                .withURI("redis://:123456@192.168.10.137:6380")
+                .withURI(property)
                 // 定义搜索过滤器使用的元数据字段
                 .withMetadataFields(
-                        RedisVectorStore.MetadataField.tag("medicalId"),
+                        // 问题的ID
+                        RedisVectorStore.MetadataField.tag("questionId"),
+                        // 类型 1标准问 2相似问 3自定义
                         RedisVectorStore.MetadataField.tag("type"))
                 .build();
         return new RedisVectorStore(config, vectorEmbeddingClient);

virtual-patient-web/src/main/java/com/supervision/VirtualPatientApplication.java

@@ -5,8 +5,10 @@ import org.springframework.boot.SpringApplication;
 import org.springframework.boot.autoconfigure.SpringBootApplication;
 import org.springframework.cloud.client.discovery.EnableDiscoveryClient;
 import org.springframework.cloud.openfeign.EnableFeignClients;
+import org.springframework.context.annotation.EnableAspectJAutoProxy;
 import org.springframework.scheduling.annotation.EnableScheduling;
 
+@EnableAspectJAutoProxy(proxyTargetClass = true)
 @SpringBootApplication
 @MapperScan(basePackages = {"com.supervision.**.mapper"})
 @EnableScheduling

virtual-patient-web/src/main/java/com/supervision/controller/TestController.java

@@ -49,6 +49,7 @@ public class TestController {
 
     private final OllamaChatClient chatClient;
 
+
     @GetMapping("testMatchQuestion")
     public String test(String question) {
         String template = """
@@ -70,8 +71,15 @@ public class TestController {
         return call.getResult().getOutput().getContent();
     }
 
+    @GetMapping("testJedis")
+    public void testJedis() {
+        Map<String, Object> stringObjectMap = redisVectorStore.getJedis().ftConfigGet("spring-ai-index", "");
+        System.out.println(1);
+    }
+
     @GetMapping("testRedisVectorStore")
     public void testRedisVectorStore() {
+
         List<AskTemplateQuestionLibrary> list = askTemplateQuestionLibraryService.list();
         for (AskTemplateQuestionLibrary askTemplateQuestionLibrary : list) {
             String description = askTemplateQuestionLibrary.getDescription();
@@ -84,14 +92,20 @@ public class TestController {
         }
     }
 
+
     @GetMapping("testQuestion")
-    public List<Map<String, String>> testQuestion(String question) {
+    public List<Document> testQuestion(String question) {
         List<Document> documents = redisVectorStore.similaritySearch(SearchRequest
                 .query(question)
-                .withTopK(5));
+                .withTopK(5).withSimilarityThreshold(0.5));
-        return documents.stream().map(document -> Map.of("content", document.getContent(), "id", document.getId())).collect(Collectors.toList());
+//        return documents.stream().map(document -> Map.of("content", document.getContent(), "id", document.getId())).collect(Collectors.toList());
+        documents.forEach(e -> {
+            double v = Double.parseDouble(String.valueOf(e.getMetadata().get("vector_score")));
+            // 降序
+            e.getMetadata().put("originalScore", 1 - (v * 2));
+        });
+        return documents;
     }
-
     @GetMapping("testExpireTime")
     public String testExpireTime() {
         return "OK";