Skip to content

overview.md

Latest commit

 

History

History
48 lines (35 loc) · 2.91 KB

overview.md

File metadata and controls

48 lines (35 loc) · 2.91 KB
Raw

Overview

Condense is a high-performance, stateless Node.js optimization engine for images, media, code, WebAssembly, fonts, and binary archives. Designed from first principles to operate entirely in-memory, Condense eliminates temporary disk I/O and easily integrates into high-throughput web services, serverless microservices, containers, and edge runtimes.

Supported Asset Formats

Condense provides specialized optimization pipelines for:

  • Raster & Vector Images: PNG, JPEG, WebP, AVIF, GIF, and SVG with perceptual SSIM/PSNR tuning and responsive ladder generation.
  • SVG Sprites: In-memory batch packing into consolidated <symbol> spritesheets.
  • Audio & Video Streams: MP4, MP3, WAV, OGG with direct stream piping, thumbnail extraction, and faststart support.
  • Code & Modern Text: HTML, CSS, JavaScript, TypeScript, JSX/TSX (via esbuild), JSON, YAML, GraphQL, SCSS, LESS, and cross-document coordinated token mangling.
  • OpenType / TrueType Fonts: SFNT & WOFF metadata table stripping (DSIG, hdmx, LTSH, PCLT).
  • PDF Documents: In-memory comment stripping, metadata removal, and text/stream recompression.
  • ZIP Archives: Pure in-memory recursive decompression, asset optimization, and maximum-ratio repacking with fflate.
  • WebAssembly: Custom section and debug symbol stripping for .wasm binaries.

Architectural Pillars (v1.0.0)

  1. Pillar 1: Next-Gen Intelligence & Perceptual Compression

    • Perceptual image optimization (optimizePerceptualImage) via binary-search SSIM/PSNR targeting.
    • Responsive multi-breakpoint image matrix generation with <picture> markup helpers (generateResponsiveMatrix).
    • Cross-document HTML/CSS/JS token shortening (mangleTokens).

  2. Pillar 2: Enterprise Binary & Asset Processors

    • In-memory SVG spritesheet consolidation (packSvgSprites).
    • In-memory recursive ZIP archive compression (optimizeZip).
    • SFNT & WOFF font table stripping (optimizeFont).
    • In-memory PDF minification (optimizePdf).

  3. Pillar 3: Developer Platform & Enterprise Telemetry

    • Fluent Chainable Pipeline API (createPipeline).
    • Multi-threaded Worker Pool leveraging Node worker_threads (WorkerPool, getWorkerPool).
    • Production preset management engine (presetService).
    • Real-time bandwidth, financial ROI, and carbon savings tracking (telemetryService).

Optimization Modes

  • quality: Best perceptual fidelity with visually lossless, conservative compression.
  • balanced: Practical sweet spot for production environments balancing speed, size, and fidelity.
  • extreme: Maximum compression prioritizing file size reduction with aggressive transformations.

Entry Points

  • src/index.js: Comprehensive public SDK, fluent pipeline, and Express middleware exports.
  • bin/cli.js: Standalone CLI for single-file, batch directory, and server execution.
  • src/app.js: Stateless HTTP Express application.