How to Generate an Example JSON¶

This guide explains how to use the ExampleJSONGenerator to create a valid JSON example from an SQLModel. Providing a good example to the WorkflowOr-chestrator is one of the most effective ways to improve the accuracy and reliability of data extraction.

When to Use This Feature¶

You should generate an example JSON whenever you don’t have one on hand. While the WorkflowOrchestrator can generate an example automatically, doing so adds an extra LLM call to every extraction task. For production use, it is more cost-effective and consistent to generate the example once and reuse it.

Step 1: Define the Models¶

First, you need SQLModel definitions to serve as the schema. This example uses a Product with related ProductSpecs.

from typing import Optional
from sqlmodel import Field, SQLModel, Relationship

class ProductSpecs(SQLModel, table=True):
    id: Optional[int] = Field(default=None, primary_key=True)
    display_size: float
    ram_gb: int
    storage_gb: int
    product_id: Optional[int] = Field(default=None, foreign_key="product.id")
    product: Optional["Product"] = Relationship(back_populates="specs")

class Product(SQLModel, table=True):
    id: Optional[int] = Field(default=None, primary_key=True)
    name: str
    price: float
    manufacturer: str
    specs: list["ProductSpecs"] = Relationship(back_populates="product")

Step 2: Initialize the Generator¶

Set up the ExampleJSONGenerator with an LLM client and your root model.

from extrai.llm_providers import GeminiClient
from extrai.core import ExampleJSONGenerator

# Replace with your actual API key
llm_client = GeminiClient(api_key="YOUR_GEMINI_API_KEY")

generator = ExampleJSONGenerator(llm_client=llm_client, output_model=Product)

Step 3: Generate the Example¶

Call the generate_example method to get the JSON string. This method communicates with the LLM and validates the output to ensure it conforms to your model’s schema.

import asyncio
import json

async def main():
    example = await generator.generate_example()
    parsed_json = json.loads(example)
    print(json.dumps(parsed_json, indent=2))

if __name__ == "__main__":
    asyncio.run(main())

Step 4: See the Results¶

The generated JSON will represent a valid, nested structure that conforms to your SQLModel definitions.

[
  {
    "_type": "Product",
    "_temp_id": "prod_001",
    "id": 101,
    "name": "Smartphone X1",
    "price": 799.99,
    "manufacturer": "GlobalTech",
    "specs_ref_ids": [
      "specs_001"
    ]
  },
  {
    "_type": "ProductSpecs",
    "_temp_id": "specs_001",
    "id": 201,
    "display_size": 6.7,
    "ram_gb": 8,
    "storage_gb": 256,
    "product_id": 101,
    "product_ref_id": "prod_001"
  }
]

This output can be saved and reused in the WorkflowOrchestrator to improve extraction accuracy.