d4vinci/Scrapling — reverse-engineered prompt

Build me a Python web scraping framework that feels easy for regular people but can still handle serious crawling.

I want users to be able to grab one page, scrape text or product cards with simple CSS or XPath style selectors, and then scale the same code into a full crawler. The parser should be smart enough to remember selected elements and find them again if the website layout changes later. Include different ways to fetch pages, from simple requests to JavaScript heavy pages with a stealth browser mode for sites that block obvious bots.

Please add a spider system for bigger jobs with concurrent crawling, sessions, proxy rotation, pause and resume, live stats, and streaming results. Make the examples really clear so someone can scrape a product list in a few lines. Also include a small CLI and an MCP server so AI coding tools can use the scraper. Add tests, docs, and a clean package setup so it can be published and installed normally.