Implements a composable filter pipeline that processes HTML documents through a stack of filter classes. Each filter can modify the document, add entries, or extract metadata. The stack supports insertion, removal, and ordering of filters, and runs them in sequence. This pattern is cleanly isolated and could be extracted to build any document transformation pipeline.

Base class for scraping documentation from URLs. Handles HTTP requests with rate limiting, retries, and response caching via a HAR-based cache. Subclasses define version-specific URLs and customize the scraping behavior. The rate-limiting and caching logic are reusable for any web scraping project that needs polite fetching.

A robust IndexedDB abstraction that manages a single database connection with queued operations, schema versioning, graceful fallback to network when IndexedDB is unavailable, and localStorage migration for legacy data. It handles corruption detection, transaction abortion recovery, and multiple stores for pages and indexes. This module is independently reusable as a resilient client-side storage layer.

A filter that converts documentation images (PNG, GIF, JPEG) to WebP format, optionally embeds small images as base64 data URIs, and replaces image URLs. It uses image_optim for lossless compression and caches results in the document. This component solves a common problem in documentation aggregation: reducing bandwidth and storage overhead while maintaining visual fidelity.