Takeout X-Ray is a static web application designed to analyze personal data archives locally in the browser. It processes large datasets without a backend, ensuring user privacy by avoiding data uploads to external servers.

The tool utilizes DuckDB-Wasm, Web Workers, and streaming parsers to handle archives weighing hundreds of megabytes or gigabytes. By processing data in 64-kilobyte chunks via TransformStream, the application avoids memory overflow issues common with standard JSON parsing.

The application supports 13 export formats from 12 services, including Telegram, WhatsApp, Discord, X, Spotify, YouTube, TikTok, Chrome, ChatGPT, and Claude.ai. It uses a zip-inspector to read only the central directory of archives, allowing users to select specific data slices without extracting entire files.

To maintain performance, the tool processes data in batches of 5,000 rows. These batches are registered in a virtual file system and ingested by DuckDB using C++ native parsers, with temporary files explicitly dropped to manage memory consumption effectively.

The software includes specialized adapters to handle unique data structures, such as DAGs in ChatGPT conversations or JavaScript wrappers in X/Twitter archives. It also performs analytical queries directly in SQL, calculating metrics like response speeds and word frequency while accounting for time zones.