Analyser

Drop a file. Find out everything. Upload nothing.

A small, local, forensic workbench for photos, sound, video, documents, archives, 3D models and 200+ other formats. Every byte is read and crunched in your browser - no servers, no accounts, no analytics, nothing ever leaves your machine.

Open the app →  ·  Changelog  ·  About

---

Note

This README is a placeholder. It looks the part, but the full docs - architecture deep-dive, format-by-format coverage, contribution guide - are still being written. Poke around the source in the meantime; it is friendlier than it looks.

What it does

Drag anything onto the page and Analyser figures out what it is, then tells you everything it can:

   your file ──▶  [ classify ]  ──▶  photo · audio · video · doc · archive · 3D · unknown
                                          │
                                          ▼
                                 read · decode · measure · visualise
                                          │
                                          ▼
                                 metadata · forensics · previews

• 📷 Photos - full EXIF / IPTC / XMP / ICC, GPS on a map, colour palette, histogram, sharpness, focus point, OCR in 32 languages, QR scan, LSB steganography planes, perceptual hash, and AI-generation markers.
• 🎵 Audio - waveform, real-time FFT spectrogram, loudness (LUFS), pitch, BPM, clipping, stereo vectorscope, embedded tags, lyrics and cover art (auto-analysed as a photo).
• 🎬 Video - container, codec, resolution, frame rate, automatic scene detection, frame stepping, contact sheets and audio extraction.
• 📄 Documents - viewers for PDF, Word, Excel, PowerPoint and EPUB, with text, tables, images and OCR.
• 🗺️ Data & code - CSV/TSV tables, SVG, MIDI scores, subtitles, GPX/KML/GeoJSON plotted on a map, and pretty-printed source.
• 📦 Archives & 3D - browse ZIPs and folders as a tree, inspect STL models in interactive WebGL.
• 🔍 Everything else - 200+ proprietary formats identified by magic bytes, with a hex dump and SHA-256 for anything unrecognised.

Why it is different

🔒 100% local	Files are read with the File API and processed on-device. The network is never touched.
✈️ Offline-first	Installable PWA. Cache the heavy tools once and it works on a plane.
🪶 No framework	Vanilla ES modules, hand-written parsers, lazy-loaded everything.
⚡ Instant	No upload round-trip - analysis starts the moment you drop.

Run it locally

git clone https://github.com/KostaJovanovic/lab.git
cd lab

# any static file server works - it is just HTML/CSS/JS
python -m http.server 8000
#   ...or...
npx serve .

# then open http://localhost:8000

No build step. No node_modules. No config.

Under the hood

HTML + CSS + vanilla JS (ES modules) · service-worker caching · View Transitions SPA router · hand-rolled radix-2 FFT · a pile of lazy-loaded WASM (FFmpeg, ImageMagick, Tesseract, pdf.js) that only download when you actually need them.

assets/js/
  core/       app.js (entry + file classification), formats.js (catalog),
              util.js (helpers), search.js, navigate.js, binutil.js
  renderers/  photo · audio · video · pdf · archive · svg · csv · stl · ...
              the per-type renderers (+ proprietary.js, 200+ formats by magic bytes)
  parsers/    parsers-*.js   lazy per-domain metadata parser chunks
  lib/        plist · cfbf · sqlite · *-loader   shared binary + WASM helpers
assets/css/   analyser.css, fonts.css
assets/img/   banner, favicons, app icons
assets/vendor/  third-party libraries

Status

🚧 Actively built, versioned per commit, shipped to lab.valjdakosta.com. Expect things to move fast and occasionally wobble.

---

Made with too much curiosity by valjdakosta

_{Nothing here phones home. Promise.}