XML output parser

The XMLOutputParser takes language model output which contains XML and parses it into a JSON object.

The output parser also supports streaming outputs.

Currently, the XML parser does not contain support for self closing tags, or attributes on tags.

Usage

tip

See this section for general instructions on installing integration packages.

npm

npm install @langchain/core

Yarn

yarn add @langchain/core

pnpm

pnpm add @langchain/core

import { XMLOutputParser } from "@langchain/core/output_parsers";

const XML_EXAMPLE = `<?xml version="1.0" encoding="UTF-8"?>
<userProfile>
  <userID>12345</userID>
  <name>John Doe</name>
  <email>john.doe@example.com</email>
  <roles>
    <role>Admin</role>
    <role>User</role>
  </roles>
  <preferences>
    <theme>Dark</theme>
    <notifications>
      <email>true</email>
      <sms>false</sms>
    </notifications>
  </preferences>
</userProfile>`;

const parser = new XMLOutputParser();

const result = await parser.invoke(XML_EXAMPLE);

console.log(JSON.stringify(result, null, 2));
/*
{
  "userProfile": [
    {
      "userID": "12345"
    },
    {
      "name": "John Doe"
    },
    {
      "email": "john.doe@example.com"
    },
    {
      "roles": [
        {
          "role": "Admin"
        },
        {
          "role": "User"
        }
      ]
    },
    {
      "preferences": [
        {
          "theme": "Dark"
        },
        {
          "notifications": [
            {
              "email": "true"
            },
            {
              "sms": "false"
            }
          ]
        }
      ]
    }
  ]
}
*/

API Reference:

• XMLOutputParser from @langchain/core/output_parsers

Streaming

import { XMLOutputParser } from "@langchain/core/output_parsers";
import { FakeStreamingLLM } from "@langchain/core/utils/testing";

const XML_EXAMPLE = `<?xml version="1.0" encoding="UTF-8"?>
<userProfile>
  <userID>12345</userID>
  <roles>
    <role>Admin</role>
    <role>User</role>
  </roles>
</userProfile>`;

const parser = new XMLOutputParser();

// Define your LLM, in this example we'll use demo streaming LLM
const streamingLLM = new FakeStreamingLLM({
  responses: [XML_EXAMPLE],
}).pipe(parser); // Pipe the parser to the LLM

const stream = await streamingLLM.stream(XML_EXAMPLE);
for await (const chunk of stream) {
  console.log(JSON.stringify(chunk, null, 2));
}
/*
{}
{
  "userProfile": ""
}
{
  "userProfile": "\n"
}
{
  "userProfile": [
    {
      "userID": ""
    }
  ]
}
{
  "userProfile": [
    {
      "userID": "123"
    }
  ]
}
{
  "userProfile": [
    {
      "userID": "12345"
    },
    {
      "roles": ""
    }
  ]
}
{
  "userProfile": [
    {
      "userID": "12345"
    },
    {
      "roles": [
        {
          "role": "A"
        }
      ]
    }
  ]
}
{
  "userProfile": [
    {
      "userID": "12345"
    },
    {
      "roles": [
        {
          "role": "Admi"
        }
      ]
    }
  ]
}
{
  "userProfile": [
    {
      "userID": "12345"
    },
    {
      "roles": [
        {
          "role": "Admin"
        },
        {
          "role": "U"
        }
      ]
    }
  ]
}
{
  "userProfile": [
    {
      "userID": "12345"
    },
    {
      "roles": [
        {
          "role": "Admin"
        },
        {
          "role": "User"
        }
      ]
    }
  ]
}
*/

API Reference:

• XMLOutputParser from @langchain/core/output_parsers
• FakeStreamingLLM from @langchain/core/utils/testing