LangChain 代码节点#

¥LangChain Code node

使用 LangChain Code 节点导入 LangChain。这意味着如果你需要的功能 n8n 尚未创建相应的节点，你仍然可以使用它。通过配置 LangChain 代码节点连接器，你可以将其用作普通节点、根节点或子节点。

¥Use the LangChain Code node to import LangChain. This means if there is functionality you need that n8n hasn't created a node for, you can still use it. By configuring the LangChain Code node connectors you can use it as a normal node, root node or sub-node.

在此页面，你可以找到节点参数、节点配置指南以及更多资源的链接。

¥On this page, you'll find the node parameters, guidance on configuring the node, and links to more resources.

Not available on Cloud

此节点仅适用于自托管的 n8n。

¥This node is only available on self-hosted n8n.

节点参数#

¥Node parameters

添加代码#

¥Add Code

添加你的自定义代码。选择“执行”或“提供数据”模式。你只能使用一种模式。

¥Add your custom code. Choose either Execute or Supply Data mode. You can only use one mode.

与代码节点节点不同，LangChain 代码节点不支持 Python。

¥Unlike the Code node, the LangChain Code node doesn't support Python.

执行：像使用 n8n 自身的 Code 节点一样使用 LangChain Code 节点。此步骤从工作流中获取输入数据，对其进行处理，并将其作为节点输出返回。此模式需要一个主输入和一个主输出。你必须在“输入”和“输出”中创建这些连接。

¥Execute: use the LangChain Code node like n8n's own Code node. This takes input data from the workflow, processes it, and returns it as the node output. This mode requires a main input and output. You must create these connections in Inputs and Outputs.

提供数据：将 LangChain Code 节点用作子节点，并将数据发送到根节点。此用例使用除 main 之外的输出。

¥Supply Data: use the LangChain Code node as a sub-node, sending data to a root node. This uses an output other than main.

默认情况下，你无法在此节点中加载内置或外部模块。自托管用户可以启用内置和外部模块。

¥By default, you can't load built-in or external modules in this node. Self-hosted users can enable built-in and external modules.

输入#

¥Inputs

选择输入类型。

¥Choose the input types.

主要输入是所有 n8n 工作流中都存在的普通连接器。如果你在节点中设置了主输入和输出，则必须执行代码。

¥The main input is the normal connector found in all n8n workflows. If you have a main input and output set in the node, Execute code is required.

输出#

¥Outputs

选择输出类型。

¥Choose the output types.

主要输出是所有 n8n 工作流中常见的标准连接器。如果你在节点中设置了主输入和输出，则必须执行代码。

¥The main output is the normal connector found in all n8n workflows. If you have a main input and output set in the node, Execute code is required.

节点输入输出配置#

¥Node inputs and outputs configuration

通过配置 LangChain 代码节点连接器（输入和输出），你可以将其用作应用节点、根节点或子节点。

¥By configuring the LangChain Code node connectors (inputs and outputs) you can use it as an app node, root node or sub-node.

节点类型	输入	输出	代码模式
应用节点类似于代码节点。	Main	Main	执行
根节点	Main;至少一种其他类型的	Main	执行
子节点	*	非主节点类型。必须与要连接的输入类型匹配。	提供数据
包含子节点的子节点	非主节点类型	非主节点类型。必须与要连接的输入类型匹配。	提供数据

内置方法#

¥Built-in methods

n8n 提供以下方法，以便更轻松地在 LangChain 代码节点中执行常见任务。

¥n8n provides these methods to make it easier to perform common tasks in the LangChain Code node.

Method	Description
`this.addInputData(inputName, data)`	Populate the data of a specified non-main input. Useful for mocking data. `inputName` is the input connection type, and must be one of: `ai_agent`, `ai_chain`, `ai_document`, `ai_embedding`, `ai_languageModel`, `ai_memory`, `ai_outputParser`, `ai_retriever`, `ai_textSplitter`, `ai_tool`, `ai_vectorRetriever`, `ai_vectorStore` `data` contains the data you want to add. Refer to Data structure for information on the data structure expected by n8n.
`this.addOutputData(outputName, data)`	Populate the data of a specified non-main output. Useful for mocking data. `outputName` is the input connection type, and must be one of: `ai_agent`, `ai_chain`, `ai_document`, `ai_embedding`, `ai_languageModel`, `ai_memory`, `ai_outputParser`, `ai_retriever`, `ai_textSplitter`, `ai_tool`, `ai_vectorRetriever`, `ai_vectorStore` `data` contains the data you want to add. Refer to Data structure for information on the data structure expected by n8n.
`this.getInputConnectionData(inputName, itemIndex, inputIndex?)`	Get data from a specified non-main input. `inputName` is the input connection type, and must be one of: `ai_agent`, `ai_chain`, `ai_document`, `ai_embedding`, `ai_languageModel`, `ai_memory`, `ai_outputParser`, `ai_retriever`, `ai_textSplitter`, `ai_tool`, `ai_vectorRetriever`, `ai_vectorStore` `itemIndex` should always be `0` (this parameter will be used in upcoming functionality) Use `inputIndex` if there is more than one node connected to the specified input.
`this.getInputData(inputIndex?, inputName?)`	Get data from the main input.
`this.getNode()`	Get the current node.
`this.getNodeOutputs()`	Get the outputs of the current node.
`this.getExecutionCancelSignal()`	Use this to stop the execution of a function when the workflow stops. In most cases n8n handles this, but you may need to use it if building your own chains or agents. It replaces the Cancelling a running LLMChain code that you'd use if building a LangChain application normally.

模板和示例#

¥Templates and examples