package org.codelibs.fess.llm;

import java.util.List;
import java.util.Map;

/**
 * Interface for LLM (Large Language Model) clients.
 * Implementations provide integration with different LLM providers
 * such as Ollama, OpenAI, and Google Gemini.
 *
 * In addition to low-level chat operations, this interface defines
 * high-level RAG workflow methods that allow each provider to optimize

        assertEquals("ollama", request.getAttribute(Constants.SEARCH_LOG_ACCESS_TYPE));
    }

    @Test
    public void test_mockRequest_attributeOverwrite() {
        final MockHttpServletRequest request = new MockHttpServletRequest();

        request.setAttribute(Constants.SEARCH_LOG_ACCESS_TYPE, "ollama");

    }

    @Test
    public void test_addSearchLog_accessType_ollama() {
        final SearchLog searchLog = callCreateSearchLogAndGetResult("ollama");
        assertEquals("ollama", searchLog.getAccessType());
    }

    @Test
    public void test_addSearchLog_accessType_openai() {
        final SearchLog searchLog = callCreateSearchLogAndGetResult("openai");

    /**
     * Gets the configured LLM type from the system configuration.
     *
     * @return The LLM type string from configuration (e.g., "ollama", "openai", "gemini")
     */
    protected String getLlmType() {
        return ComponentUtil.getFessConfig().getSystemProperty("rag.llm.name", "ollama");
    }

    /**
     * Checks if RAG chat feature is enabled.
     *
     * @return true if RAG chat is enabled, false otherwise

            // Set LLM type name as Access Type for search log
            request.setAttribute(Constants.SEARCH_LOG_ACCESS_TYPE,
                    ComponentUtil.getFessConfig().getSystemProperty("rag.llm.name", "ollama"));

            final Map<String, String[]> fields = parseFieldFilters(request);
            final String[] extraQueries = parseExtraQueries(request);
            final ChatResult result;

    default void setRagLlmName(final String value) {
        setSystemProperty(Constants.RAG_LLM_NAME, value);
    }

    default String getRagLlmName() {
        return getSystemProperty(Constants.RAG_LLM_NAME, "ollama");
    }

    Integer getLdapMaxUsernameLengthAsInteger();

    default String getLdapSecurityPrincipal(final String username) {
        final String value;

# Depuración { #debugging }

Puedes conectar el depurador en tu editor, por ejemplo con Visual Studio Code o PyCharm.

## Llama a `uvicorn` { #call-uvicorn }

En tu aplicación de FastAPI, importa y ejecuta `uvicorn` directamente:

{* ../../docs_src/debugging/tutorial001_py310.py hl[1,15] *}

### Acerca de `__name__ == "__main__"` { #about-name-main }

En realidad, todos (o la mayoría) de los frameworks web funcionan de esta misma manera.

Nunca llamas directamente a esas funciones. Son llamadas por tu framework (en este caso, **FastAPI**).

* que una query de base de datos devuelva los resultados
* etc.

Como el tiempo de ejecución se consume principalmente esperando operaciones de <abbr title="Input and Output - Entrada y salida">I/O</abbr>, las llaman operaciones "I/O bound".

# Koşullu OpenAPI { #conditional-openapi }

Gerekirse, ayarlar ve environment variable'ları kullanarak OpenAPI'yi ortama göre koşullu şekilde yapılandırabilir, hatta tamamen devre dışı bırakabilirsiniz.

## Güvenlik, API'ler ve dokümantasyon hakkında { #about-security-apis-and-docs }

Production ortamında dokümantasyon arayüzlerini gizlemek, API'nizi korumanın yolu olmamalıdır.