Using web search

Quarkus LangChain4j currently supports the Tavily search engine. To use it, add the quarkus-langchain4j-tavily extension to your project. You’ll need to specify the API key, this is done by the quarkus.langchain4j.tavily.api-key property.

After this, you can inject the search engine into your application using

@Inject
WebSearchEngine engine;

and then use it by calling its search method.

If you want to let an chat model use web search by itself, there are generally two recommended ways to accomplish this: either by implementing a tool that uses it, or as a content retriever inside a RAG pipeline. The chatbot-web-search example in the quarkus-langchain4j repository demonstrates using web search as a tool.

Using Web search as a tool

To use web search as a tool that the LLM can decide to execute (and the relevant search results will be the return value of the tool execution), you can either use the provided tool from the upstream LangChain4j project, in class dev.langchain4j.web.search.WebSearchTool, or implement your own tool if that one does not fit your requirements. The samples/chatbot-web-search example demonstrates how to use the provided tool.

Using Web search in a RAG pipeline

There is also a provided content retriever, dev.langchain4j.rag.content.retriever.WebSearchContentRetriever that uses a web search engine to retrieve relevant documents. For inspiration, the retrieval augmentor that wraps it may look like this:

@ApplicationScoped
public class WebSearchRetrievalAugmentor implements Supplier<RetrievalAugmentor> {

    @Inject
    WebSearchEngine webSearchEngine;

    @Inject
    ChatLanguageModel chatModel;

    @Override
    public RetrievalAugmentor get() {
        return DefaultRetrievalAugmentor.builder()
                .queryTransformer((question) -> {
                    // before actually querying the engine, we need to transform the
                    // user's question into a suitable search query
                    String query = chatModel.generate("Transform the user's question into a suitable query for the " +
                            "Tavily search engine. The query should yield the results relevant to answering the user's question." +
                            "User's question: " + question.text());
                    return Collections.singleton(Query.from(query));
                }).contentRetriever(new WebSearchContentRetriever(webSearchEngine, 10))
                .build();
    }
}

Tavily configuration reference

Configuration property fixed at build time - All other configuration properties are overridable at runtime

Configuration property	Type	Default
`quarkus.langchain4j.tavily.base-url` Base URL of the Tavily API Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_BASE_URL`	string	`https://api.tavily.com`
`quarkus.langchain4j.tavily.api-key` API key for the Tavily API Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_API_KEY`	string	required
`quarkus.langchain4j.tavily.max-results` Maximum number of results to return Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_MAX_RESULTS`	int	`5`
`quarkus.langchain4j.tavily.timeout` The timeout duration for Tavily requests. Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_TIMEOUT`	Duration	`60S`
`quarkus.langchain4j.tavily.log-requests` Whether requests to Tavily should be logged Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_LOG_REQUESTS`	boolean	`false`
`quarkus.langchain4j.tavily.log-responses` Whether responses from Tavily should be logged Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_LOG_RESPONSES`	boolean	`false`
`quarkus.langchain4j.tavily.search-depth` The search depth to use. This can be "basic" or "advanced". Basic is the default. Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_SEARCH_DEPTH`	`basic`, `advanced`	`basic`
`quarkus.langchain4j.tavily.include-answer` Include a short answer to original query. Default is false. Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_INCLUDE_ANSWER`	boolean	`false`
`quarkus.langchain4j.tavily.include-raw-content` Include the cleaned and parsed HTML content of each search result. Default is false. Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_INCLUDE_RAW_CONTENT`	boolean	`false`
`quarkus.langchain4j.tavily.include-domains` A list of domains to specifically include in the search results. Default is [], which includes all domains. Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_INCLUDE_DOMAINS`	list of string	`[]`
`quarkus.langchain4j.tavily.exclude-domains` A list of domains to specifically exclude from the search results. Default is [], which doesn’t exclude any domains. Environment variable: `QUARKUS_LANGCHAIN4J_TAVILY_EXCLUDE_DOMAINS`	list of string	`[]`

Configuration property

Type

Default

quarkus.langchain4j.tavily.base-url

Base URL of the Tavily API

Environment variable: QUARKUS_LANGCHAIN4J_TAVILY_BASE_URL

string

https://api.tavily.com

quarkus.langchain4j.tavily.api-key

API key for the Tavily API

Environment variable: QUARKUS_LANGCHAIN4J_TAVILY_API_KEY

string