I want to share my latest project for the M5Stack Cardputer ADV. It’s called Wikipedia Offline ADV.

I wanted to see if it was possible to fit all of human knowledge into this tiny device and make it searchable without any Wi-Fi connection. The result is a hyper-optimized, standalone Wikipedia search engine wrapped in a retro-futuristic "ELIZA OS" terminal interface.

🚀 Key Features

• 6.4 Million Articles Offline: The entire database is packed into an 18.3 GB archive stored on your SD card.
• Lightning Fast: I built an O(log N) binary pointer search engine. It bypasses standard FAT32 limits by packing file IDs and local offsets into 64-bit pointers. Searches take mere milliseconds.
• Pseudo-AI Assistant: It’s not just a dumb search bar. The onboard NLP engine parses your intent (e.g., "Who is the president of..." or "What is the capital of..."). It uses a custom TF-IDF Lite algorithm to score sentences and extract the exact answer before printing the full 32KB article buffer.
• Fuzzy Levenshtein Radar: Typo on the tiny keyboard? No problem. The engine scans a 250-article sliding window on the SD card, calculates Levenshtein distance, and offers a paginated list of the closest matches.
• Zero-Flicker CRT Rendering: Custom Unified Canvas with hardware clipping for a smooth 60FPS CRT scanline effect.

🛠 How to Install

1. Download Wikipedia Offline ADV from M5Burner.
2. The Database: The firmware requires the database to work! Download the 6.54 GB .zip archive from the official GitHub release (hosted on Google Drive).
3. Extract the files and place them directly into the root directory of a FAT32-formatted MicroSD card (32GB or larger required). Do not put them in folders!
4. Insert the SD card and boot up.

🔗 GitHub Repository & Full Guide: Zeloksa/Wikipedia-Offline-ADV

If you like my work, you can support my late-night coding sessions here: Boosty ☕