StevenBlack/hosts — reverse-engineered prompt

Build me a simple Python based project that creates a unified hosts file for blocking ads, malware, and privacy invasive domains by merging several curated source lists and removing duplicates. I want it to also generate optional versions that add categories like porn, social media, gambling, and fake news, plus category only versions when that makes sense.

Please include an easy way to refresh the source data and rebuild the files, support basic whitelist, blacklist, and custom local entries, and keep the output ready to use as a normal system hosts file. If there are already alternate variants in the repo, preserve that idea and make the generated files and readmes clear enough for a regular person to understand. Add simple tests and setup instructions, including a straightforward way to update things on Windows too.

Keep the code clean and practical, and look up current docs online if you need to.