Interface defining protocol for handling provider-specific thinking/reasoning modes in LLM responses.
// Source: https://github.com/ntygod/ZhiWei | License: MIT | Copyright 2026 ntygod
// Provided via justclaw.it.com; keep this notice with the code.
package com.lifepilot.llm.thinking;
import com.fasterxml.jackson.databind.JsonNode;
import com.lifepilot.llm.history.AssistantMessageBuilder;
import com.lifepilot.llm.profile.ThinkingProtocolId;
import org.springframework.lang.Nullable;
import java.util.Map;
/**
* 推理模式协议处理器。
*
* <p>每个实现承载一家 provider 的 thinking 协议差异,通过三个方法表达:
* <ol>
* <li>{@link #applyToRequest(RequestBuilder, ThinkingMode)} — 把 mode 转换成厂商私有字段;</li>
* <li>{@link #extractReasoning(JsonNode)} — 从响应(同步或流式 chunk)提取 reasoning_content;</li>
* <li>{@link #injectHistoryReasoning(AssistantMessageBuilder, Map)} — 多轮回传时按协议注入 reasoning。</li>
* </ol>
*
* @author zsg
* @si
... (truncated -- full source via MCP)
See the full source, get the GitHub permalink, and search for more like it.
Get a free API key