E-Commerce Scraping with Extracto

Whether you are building a price-tracking engine, aggregating competitor catalogs, or running sentiment analysis on Amazon reviews, extracting structured E-Commerce data is notoriously fragile. Store owners constantly change UI layouts and rely on heavy React/Next.js hydration that breaks traditional scrapers.

Extracto solves this by rendering the cart and product pages visually, and using AI (GPT-4o or Mistral) to pull exactly the schema you requested.

Example: Universal Product Extraction

Instead of writing custom `BeautifulSoup` parsing logic for Amazon, Walmart, BestBuy, and Shopify, you simply write one universal Extracto definition in Python:

import asyncio
from extracto.config import CrawlerConfig
from extracto.crawler_engine import CrawlerEngine
from extracto.schema import PropertyDefinition, DataType

# 1. Define standard Pydantic-style Schema
schema = [
    PropertyDefinition(name="product_title", type=DataType.STRING, description="The main product name"),
    PropertyDefinition(name="price", type=DataType.NUMBER, description="Current price in USD"),
    PropertyDefinition(name="sku", type=DataType.STRING, description="The unique item identifier"),
    PropertyDefinition(name="in_stock", type=DataType.BOOLEAN, description="Is the item currently available to purchase?")
]

# 2. Extract from ANY E-Commerce site without changing the schema
config = CrawlerConfig(
    start_urls=["https://store.example.com/product/1234"],
    properties=schema,
    llm_provider="mistral"
)

# 3. Run
engine = CrawlerEngine(config)
asyncio.run(engine.run())
        

Handling Dynamic React/SPA Stores

Many modern E-commerce storefronts don't load data instantly. They render a skeleton, make GraphQL calls, and hydrate the page. Extracto handles this natively with Playwright, waiting for the network idle state before the AI scans the page visually.