This repository shows how to build a Gemini-powered web scraper using Python and LLMs to extract structured data from complex web pages — without writing custom parsing logic.

📖 Read the full tutorial → How to Leverage Gemini AI for Web Scraping

• Fetches HTML from any public webpage
• Converts HTML to Markdown using markdownify
• Sends it to Gemini AI with a natural language prompt
• Extracts structured data in JSON format

• google-generativeai – Gemini API for LLM-powered parsing
• requests – For basic HTTP requests (if not using a proxy)
• beautifulsoup4 – For basic HTML parsing (optional)
• markdownify – Converts HTML into cleaner Markdown
• python-dotenv – For managing API keys and environment variables

1. Clone this repo:

git clone https://github.com/yourusername/gemini-ai-web-scraper.git
cd gemini-ai-web-scraper

2. Install dependencies:

pip install google-generativeai python-dotenv requests beautifulsoup4 markdownify

3. Add your Gemini API Key in the script or as environment variable.

Web scraping with Gemini AI can hit blocks, CAPTCHAs, and anti-bot systems. Crawlbase Smart Proxy solves that.

• Avoid IP blocks with automatic rotation
• Bypass CAPTCHAs seamlessly
• Skip proxy management
• Get clean, parsed HTML for better AI input

import requests
import time
proxy_url = "http://_USER_TOKEN_@smartproxy.crawlbase.com:8012"
proxies = {"http": proxy_url, "https": proxy_url}
url = "https://example.com/protected-page"
time.sleep(2) # Mimic human behavior
response = requests.get(url, proxies=proxies, verify=False)
print(response.text)

Replace _USER_TOKEN_ with your Crawlbase Smart Proxy token. Get one after signup on Crawlbase.