Extraction de données immobilières avec Firecrawl

Extrayez des données structurées de listes immobilières (Zillow, Redfin, Realtor.com) en JSON prêt pour la génération vidéo. Gère le rendu JavaScript et les mesures anti-bot automatiquement.

Spar Skills Guide Bot
Data & IAIntermédiaire1 vues0 installations08/03/2026
Claude CodeCursorWindsurf
web-scrapingreal-estatedata-extractionfirecrawltypescript

name: firecrawl-scraper description: Extract property data from real estate listing URLs using Firecrawl AI. Use when scraping Zillow, Redfin, Realtor.com, or any property listing site. Returns structured data ready for video generation. allowed-tools: Read, Grep, Glob, Bash

Firecrawl Property Extraction

Overview

Firecrawl transforms any real estate listing URL into structured JSON data for video generation. It handles JavaScript rendering, anti-bot measures, and image extraction automatically.

Quick Start

import Firecrawl from '@mendable/firecrawl-js';
import { z } from 'zod';

const firecrawl = new Firecrawl({
  apiKey: process.env.FIRECRAWL_API_KEY
});

const result = await firecrawl.extract({
  urls: [listingUrl],
  prompt: 'Extract property details for video generation',
  schema: PropertySchema
});

Supported Sites

| Site | URL Pattern | Data Quality | |------|-------------|--------------| | Zillow | zillow.com/homedetails/* | Excellent | | Redfin | redfin.com//home/ | Excellent | | Realtor.com | realtor.com/realestateandhomes-detail/* | Excellent | | Trulia | trulia.com/home/* | Good | | Homes.com | homes.com/property/* | Good | | MLS Sites | Varies by region | Good | | Broker Sites | Any | Variable |

Property Schema

See rules/property-extraction.md for complete schema.

const PropertySchema = z.object({
  address: z.string(),
  city: z.string(),
  state: z.string(),
  zipCode: z.string(),
  price: z.number(),
  bedrooms: z.number(),
  bathrooms: z.number(),
  sqft: z.number(),
  lotSize: z.string().optional(),
  yearBuilt: z.number().optional(),
  propertyType: z.string(),
  description: z.string(),
  features: z.array(z.string()),
  images: z.array(z.string()),
  agent: z.object({
    name: z.string(),
    phone: z.string().optional(),
    brokerage: z.string().optional(),
  }).optional(),
});

Advanced Extraction

Competitor Analysis

const CompetitorSchema = z.object({
  listings: z.array(z.object({
    address: z.string(),
    price: z.number(),
    daysOnMarket: z.number(),
    pricePerSqft: z.number(),
  })),
  marketTrends: z.object({
    medianPrice: z.number(),
    averageDaysOnMarket: z.number(),
    inventoryCount: z.number(),
  }),
});

Market Data

See rules/market-data.md

Best Practices

  1. Rate Limiting: Max 10 requests/minute on standard plan
  2. Error Handling: Always wrap in try/catch
  3. Image Quality: Request high-res images when available
  4. Caching: Cache results for 24 hours to save credits
  5. Validation: Always validate extracted data with Zod

API Integration

Next.js Route Handler

// /app/api/scrape/route.ts
export async function POST(request: Request) {
  const { url } = await request.json();

  const firecrawl = new Firecrawl({
    apiKey: process.env.FIRECRAWL_API_KEY!
  });

  const result = await firecrawl.extract({
    urls: [url],
    prompt: 'Extract property listing data',
    schema: PropertySchema,
  });

  return Response.json({
    success: true,
    data: result.data
  });
}

Credit Usage

| Operation | Credits | |-----------|---------| | /scrape (single page) | 1 | | /crawl (per page) | 1 | | /extract (AI) | Tokens-based | | /map (URL discovery) | 1 per 100 URLs |

Error Handling

try {
  const result = await firecrawl.extract({ ... });
} catch (error) {
  if (error.statusCode === 429) {
    // Rate limited - wait and retry
  } else if (error.statusCode === 403) {
    // Site blocked - try alternative approach
  } else {
    // Log and return fallback
  }
}

Skills similaires