聊聊Spring AI的ChromaVectorStore

序

本文主要研究一下Spring AI的ChromaVectorStore

示例

pom.xml

		<dependency>
			<groupId>org.springframework.ai</groupId>
			<artifactId>spring-ai-starter-vector-store-chroma</artifactId>
		</dependency>

配置

spring:
  ai:
    vectorstore:
      type: chroma
      chroma:
        initialize-schema: true
        collectionName: "test1"
        client:
          host: http://localhost
          port: 8000

代码

    @Test
    public void testAddAndSearch() {
        List<Document> documents = List.of(
                new Document("Spring AI rocks!! Spring AI rocks!! Spring AI rocks!! Spring AI rocks!! Spring AI rocks!!", Map.of("meta1", "meta1")),
                new Document("The World is Big and Salvation Lurks Around the Corner"),
                new Document("You walk forward facing the past and you turn back toward the future.", Map.of("meta2", "meta2")));

        // Add the documents to Milvus Vector Store
        chromaVectorStore.add(documents);

        // Retrieve documents similar to a query
        List<Document> results = this.chromaVectorStore.similaritySearch(SearchRequest.builder().query("Spring").topK(5).build());
        log.info("results:{}", JSON.toJSONString(results));
    }

输出如下：

results:[{"contentFormatter":{"excludedEmbedMetadataKeys":[],"excludedInferenceMetadataKeys":[],"metadataSeparator":"\n","metadataTemplate":"{key}: {value}","textTemplate":"{metadata_string}\n\n{content}"},"formattedContent":"distance: 0.4350912\nmeta1: meta1\n\nSpring AI rocks!! Spring AI rocks!! Spring AI rocks!! Spring AI rocks!! Spring AI rocks!!","id":"53ce7adb-07ba-429c-b443-40edffde2c89","metadata":{"distance":0.4350912,"meta1":"meta1"},"score":0.5649088025093079,"text":"Spring AI rocks!! Spring AI rocks!! Spring AI rocks!! Spring AI rocks!! Spring AI rocks!!"},{"contentFormatter":{"$ref":"$[0].contentFormatter"},"formattedContent":"distance: 0.57093126\n\nThe World is Big and Salvation Lurks Around the Corner","id":"cd79cfe8-8503-4e2e-bb1e-ec0e294bc800","metadata":{"distance":0.57093126},"score":0.42906874418258667,"text":"The World is Big and Salvation Lurks Around the Corner"},{"contentFormatter":{"$ref":"$[0].contentFormatter"},"formattedContent":"distance: 0.59360236\nmeta2: meta2\n\nYou walk forward facing the past and you turn back toward the future.","id":"f266f142-3044-4b09-8178-55abe6ef84c5","metadata":{"distance":0.59360236,"meta2":"meta2"},"score":0.40639764070510864,"text":"You walk forward facing the past and you turn back toward the future."}]

源码

ChromaVectorStoreAutoConfiguration

org/springframework/ai/vectorstore/chroma/autoconfigure/ChromaVectorStoreAutoConfiguration.java

@AutoConfiguration
@ConditionalOnClass({ EmbeddingModel.class, RestClient.class, ChromaVectorStore.class, ObjectMapper.class })
@EnableConfigurationProperties({ ChromaApiProperties.class, ChromaVectorStoreProperties.class })
@ConditionalOnProperty(name = SpringAIVectorStoreTypes.TYPE, havingValue = SpringAIVectorStoreTypes.CHROMA,
		matchIfMissing = true)
public class ChromaVectorStoreAutoConfiguration {

	@Bean
	@ConditionalOnMissingBean(ChromaConnectionDetails.class)
	PropertiesChromaConnectionDetails chromaConnectionDetails(ChromaApiProperties properties) {
		return new PropertiesChromaConnectionDetails(properties);
	}

	@Bean
	@ConditionalOnMissingBean
	public ChromaApi chromaApi(ChromaApiProperties apiProperties,
			ObjectProvider<RestClient.Builder> restClientBuilderProvider, ChromaConnectionDetails connectionDetails,
			ObjectMapper objectMapper) {

		String chromaUrl = String.format("%s:%s", connectionDetails.getHost(), connectionDetails.getPort());

		var chromaApi = new ChromaApi(chromaUrl, restClientBuilderProvider.getIfAvailable(RestClient::builder),
				objectMapper);

		if (StringUtils.hasText(connectionDetails.getKeyToken())) {
			chromaApi.withKeyToken(connectionDetails.getKeyToken());
		}
		else if (StringUtils.hasText(apiProperties.getUsername()) && StringUtils.hasText(apiProperties.getPassword())) {
			chromaApi.withBasicAuthCredentials(apiProperties.getUsername(), apiProperties.getPassword());
		}

		return chromaApi;
	}

	@Bean
	@ConditionalOnMissingBean(BatchingStrategy.class)
	BatchingStrategy chromaBatchingStrategy() {
		return new TokenCountBatchingStrategy();
	}

	@Bean
	@ConditionalOnMissingBean
	public ChromaVectorStore vectorStore(EmbeddingModel embeddingModel, ChromaApi chromaApi,
			ChromaVectorStoreProperties storeProperties, ObjectProvider<ObservationRegistry> observationRegistry,
			ObjectProvider<VectorStoreObservationConvention> customObservationConvention,
			BatchingStrategy chromaBatchingStrategy) {
		return ChromaVectorStore.builder(chromaApi, embeddingModel)
			.collectionName(storeProperties.getCollectionName())
			.initializeSchema(storeProperties.isInitializeSchema())
			.observationRegistry(observationRegistry.getIfUnique(() -> ObservationRegistry.NOOP))
			.customObservationConvention(customObservationConvention.getIfAvailable(() -> null))
			.batchingStrategy(chromaBatchingStrategy)
			.build();
	}

	static class PropertiesChromaConnectionDetails implements ChromaConnectionDetails {

		private final ChromaApiProperties properties;

		PropertiesChromaConnectionDetails(ChromaApiProperties properties) {
			this.properties = properties;
		}

		@Override
		public String getHost() {
			return this.properties.getHost();
		}

		@Override
		public int getPort() {
			return this.properties.getPort();
		}

		@Override
		public String getKeyToken() {
			return this.properties.getKeyToken();
		}

	}

}

ChromaVectorStoreAutoConfiguration在spring.ai.vectorstore.type为chroma时启用，它根据ChromaApiProperties创建ChromaApi，再根据ChromaVectorStoreProperties创建ChromaVectorStore

ChromaApiProperties

org/springframework/ai/vectorstore/chroma/autoconfigure/ChromaApiProperties.java

@ConfigurationProperties(ChromaApiProperties.CONFIG_PREFIX)
public class ChromaApiProperties {

	public static final String CONFIG_PREFIX = "spring.ai.vectorstore.chroma.client";

	private String host = "http://localhost";

	private int port = 8000;

	private String keyToken;

	private String username;

	private String password;

	//......
}	

ChromaApiProperties主要是配置spring.ai.vectorstore.chroma.client，它提供了host、port、keyToken、username、password这些属性

ChromaVectorStoreProperties

org/springframework/ai/vectorstore/chroma/autoconfigure/ChromaVectorStoreProperties.java

@ConfigurationProperties(ChromaVectorStoreProperties.CONFIG_PREFIX)
public class ChromaVectorStoreProperties extends CommonVectorStoreProperties {

	public static final String CONFIG_PREFIX = "spring.ai.vectorstore.chroma";

	private String collectionName = ChromaVectorStore.DEFAULT_COLLECTION_NAME;

	public String getCollectionName() {
		return this.collectionName;
	}

	public void setCollectionName(String collectionName) {
		this.collectionName = collectionName;
	}

}

ChromaVectorStoreProperties提供了spring.ai.vectorstore.chroma，它从CommonVectorStoreProperties继承了initializeSchema属性，自己提供了collectionName属性

小结

Spring AI提供了spring-ai-starter-vector-store-chroma用于自动装配ChromaVectorStore。要注意的是embeddingDimension默认是1536，如果出现cannot unpack non-iterable coroutine object错误，记得把chroma版本降到0.6.2。

doc

• vectordbs/chroma