Dynamic json object with DynamicTools or DynamicStrcuturedTools

[KeenanFernandes2000]

I have an agent that generates a json object dynamically, how do I get the agent to push that json object into a dynamicTool or dynamicStructuredTool.

this is my tool for reference:
new DynamicTool({
name: "wordDocGen",
description:
"Tool to compile all document generated content and create a word document using a template provided",
func: async function () {
let data = {};
// /Document Generator/
const loadImage = async (imgPath: string) => {
if (
imgPath.startsWith("http://") ||
imgPath.startsWith("https://")
) {
const response = await fetch(imgPath);
const buffer = await response.arrayBuffer();
return Buffer.from(buffer);
} else {
return fs.promises.readFile(imgPath);
}
};

      // Set up the image module
      const imageModule = new ImageModule({
        getImage: async function (tagValue: any) {
          return await loadImage(tagValue);
        },
        getSize: function () {
          return [500, 500];
        },
      });
      console.log(data);
      // Setting up tempalte for content
      const content = fs.readFileSync(
        "././public/assets/templates/allTemplate.docm"
      );

      const zip = new PizZip(content);
      const doc = new Docxtemplater(zip, {
        modules: [imageModule],
        linebreaks: true,
      });
      await doc.renderAsync(data);
      const buf = doc.getZip().generate({ type: "nodebuffer" });
      const outputFilePath = "././public/audio/documentGen.docm";
      fs.writeFileSync(outputFilePath, buf);

      return "done";
    },
  })

[dosubot[bot]]

🤖

Hey @KeenanFernandes2000! 👋 Back at it again, I see? Hope you're doing well and ready to dive into some more coding adventures! 😄

To modify your agent to push a dynamically generated JSON object into a DynamicTool or DynamicStructuredTool, you need to adjust the runnableAgent sequence. Specifically, you need to generate the JSON object and push it into the action and action_input fields of the JSON blob in the Action section of the agent's output.

Here's an example of how you can do this:

const runnableAgent = RunnableSequence.from([
  {
    input: (i: { input: string; steps: AgentStep[] }) => i.input,
    agent_scratchpad: (i: { input: string; steps: AgentStep[] }) =>
      formatLogToString(i.steps),
  },
  prompt,
  model,
  {
    // This is the new step where you generate the JSON object
    action: (i: { input: string; steps: AgentStep[] }) => {
      // Generate your JSON object here
      const jsonObject = {
        action: "wordDocGen",
        action_input: {
          // Your dynamic data here
        },
      };
      // Convert the JSON object to a string and return it
      return JSON.stringify(jsonObject);
    },
  },
  outputParser,
]);

In this example, I've added a new step in the runnableAgent sequence. This step generates a JSON object with the action field set to "wordDocGen" (the name of your tool) and the action_input field set to the dynamic data you want to pass to the tool. The JSON object is then converted to a string and returned.

Please replace the comment // Your dynamic data here with the actual dynamic data you want to pass to the tool. The data should match the schema of the tool's input.

The structure of the JSON object that the agent needs to generate dynamically is defined by the jsonSchema object. It is an object with properties name, age, and fav_food. The name and age properties are required.

Here is the structure:

{
  "title": "Person",
  "description": "Identifying information about a person.",
  "type": "object",
  "properties": {
    "name": { "title": "Name", "description": "The person's name", "type": "string" },
    "age": { "title": "Age", "description": "The person's age", "type": "integer" },
    "fav_food": {
      "title": "Fav Food",
      "description": "The person's favorite food",
      "type": "string"
    }
  },
  "required": ["name", "age"]
}

The generated JSON object should look like this:

{ "name": "John Doe", "age": 30, "fav_food": "chocolate chip cookies" }

This is based on the context provided in the code snippet where the runnable.invoke method is called with a description, and the response is logged to the console.

You can find more information about this in the examples/src/agents/structured_chat_runnable.ts and examples/src/chains/openai_functions_json_schema.ts files in the langchainjs repository.

Sources

• examples/src/agents/structured_chat_runnable.ts
• examples/src/chains/openai_functions_json_schema.ts

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.

[KeenanFernandes2000]

@dosu-bot what if my data is the json object and be infinitely bigger

this is the main structure of my dynamic data:
{
title: "title of the document",
body: [{heading: "heading of the section ", content: "content for the section},{heading: "heading of the section ", content: "content for the section}]
}

but in my case there can be as many headings and content for each headings in the array. How can i just push the full json object to the tool

[dosubot[bot]]

The error message "Invalid value for 'content': expected a string, got null" indicates that the content field in your JSON object was expected to be a string, but a null value was found instead. This typically happens when the DynamicTool or similar processing logic in your code checks the type of the content field and does not accept null as a valid input.

To resolve this issue, you can modify your code to check for null values and replace them with an empty string (or another default string value) before processing. Here's a quick example of how you might adjust your code to handle null content values:

function processInput(input: any) {
  const content = input.content != null ? input.content : "";
  // Use 'content' in your processing logic, now guaranteed to be a string
}

This adjustment ensures that your processing logic always receives a string for the content field, even if the original input was null, thereby avoiding the error.

Details

Sources

• libs/langchain-community/src/tools/dynamic.ts
• langchain-core/src/tools.ts
• langchain-core/src/utils/@cfworker/json-schema/src/index.ts
• langchain-core/src/utils/@cfworker/json-schema/src/validate.ts
• examples/src/guides/deployment/error_handling.ts
• langchain-core/src/utils/fast-json-patch/src/helpers.ts
• examples/src/guides/deployment/stream_error_handling.ts
• langchain/src/tools/json.ts
• libs/langchain-scripts/src/utils.ts
• langchain-core/src/utils/json.ts
• langchain-core/src/load/serializable.ts
• libs/langchain-google-common/src/utils/stream.ts

About Dosu
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

To continue the conversation, mention @dosu.