TheFileConvert
Every file. Any format. Free.
TheFileConvert is a free, privacy-first file conversion and compression platform. Most tools run entirely in the browser using JavaScript, Canvas, Web Crypto, and WebAssembly-adjacent libraries — no account, no subscription, and no server upload for local tools.
Live site: thefileconvert.com
Principles
- Local over server — process files on-device whenever technically possible.
- Free, no account — no login, no paywalled tiers, no email requirement.
- Honest labeling — every tool declares a status (
available, experimental, coming-soon) and a processing mode (local, server-assisted, unsupported). See /tools/status.
- No AI, no paid APIs — this is a static/serverless app designed to run near domain-cost.
Tech stack
- Next.js 16 (App Router, TypeScript strict, React 19)
- Tailwind CSS 4
- pdf-lib (PDF creation/editing) + pdfjs-dist (PDF rendering to images)
- JSZip (archive creation/extraction)
- tesseract.js (local, in-browser OCR — lazy-loaded only on
/pdf/ocr; see docs/OCR.md)
- Everything else (image processing, hashing, text/data tools) uses native browser APIs (
Canvas, Web Crypto, DOMParser) — no extra dependency
No database, no auth provider, no payment processor, no AI API.
Getting started
npm install
npm run dev
Open http://localhost:3000.
Project structure
src/
app/ Route pages (App Router) — one folder per tool/category
components/
home/ Homepage sections
layout/ Header, Footer
tools/ Shared tool UI: DropZone, FileWorkflow, TextWorkflow, ToolCard, ...
ui/ Small primitives: Button, Badge
lib/
tools/ Tool registry (metadata: name, category, accepted types, status, FAQ)
processors/ Pure conversion logic (pdf.ts, image.ts, archive.ts, data.ts, markdown.ts, text-documents.ts, ocr.ts)
file-detection/ Extension + magic-byte sniffing
security/ File validation, SVG sanitization, ZIP-bomb guard, archive path sanitization
download/ Object URL lifecycle + ZIP-of-results download
format.ts Byte formatting, filename helpers
e2e/ Playwright end-to-end tests (+ fixtures/)
See docs/ARCHITECTURE.md for the full design, docs/TOOLS.md for how to add a new tool, docs/PRIVACY-MODEL.md for the privacy architecture, and docs/OCR.md for the OCR engine, self-hosted assets, and verified-language policy.
Testing
npm run test # unit tests (Vitest)
npm run test:watch # unit tests, watch mode
npm run test:e2e # Playwright end-to-end tests (builds + serves the app first)
E2E tests cover real file uploads and downloads — merged PDFs are re-parsed with pdf-lib to check page counts, converted images are checked for correct magic bytes, and error states are exercised with intentionally bad input.
Building & running
npm run build
npm run start
Linting & types
npm run lint
npx tsc --noEmit
Known limitations
- PDF compress performs a lossless structural re-save (compressed cross-reference streams). It does not re-sample embedded images, so savings are modest on already-optimized files.
- SVG to PNG is marked experimental — very complex SVGs (filters, external references) may not rasterize perfectly.
- OCR PDF is marked experimental — strong accuracy on clean/typed scans (verified, see
docs/OCR.md), but no rotation auto-detection, no deskew, no handwriting support, and heavy scan noise reduces accuracy significantly.
- DOCX/PPTX/XLSX conversion, PDF password protect/unlock, and audio/video conversion are not implemented in V1. They would require server-side processing (e.g. LibreOffice, ffmpeg) or licensed libraries that don't have a reliable, free, purely client-side equivalent yet. They're listed as "Coming soon" on
/tools/status rather than faked.
- Markdown support covers common syntax (headings, bold/italic, links, lists, blockquotes, fenced code, hr) — it is not a full CommonMark implementation.
Deployment
Designed for Vercel's free tier: fully static/serverless, no database, no long-running processes. See docs/ARCHITECTURE.md for details.
