PaperLab

PaperLab is an independent, open-source PDF workshop: a Go API and a static, vanilla-JS front end with 41 document tools. It is not affiliated with iLovePDF or any other PDF brand.

What is included

• Organise: merge, split, remove pages, extract pages, organise, rotate, crop, N-up, booklet.
• Optimise: compress, repair, OCR, PDF/A, flatten.
• Convert: PDF to Word/PowerPoint/Excel/JPG/text/Markdown, extract images; Word/PowerPoint/Excel/images/HTML to PDF; scan to PDF.
• Edit: edit, sign, watermark, page numbers, metadata.
• Security: protect, unlock, redact, compare.
• Forms: fill, extract form data.
• Intelligence: summarise, chat with PDF, translate.

The UI also has a visual Studio (edit, sign, organise, redact, crop) and a grounded Chat view. Uploads are asynchronous: create a job, poll it, then download outputs. Jobs and files expire automatically.

Requirements

• Go 1.27.2+
• Poppler (pdftoppm, pdftotext, pdfinfo)
• Ghostscript
• qpdf
• Tesseract (optional, for OCR)
• LibreOffice (optional, for high-fidelity Office conversion)
• Chromium/Chrome (optional, for HTML/URL to PDF)

Tools whose required engine is missing are marked unavailable in /api/tools.

Run locally

cp .env.example .env
set -a; . ./.env; set +a
go build -o bin/paperlab ./cmd/server
./bin/paperlab

Open http://localhost:8080. The API health check is at /api/health.

The Go binary does not parse .env itself. Export it as shown above, or pass variables inline.

Docker

The image includes all engines and is the full-fidelity way to run the project:

cp .env.example .env
docker compose up --build

Open http://localhost:8080. Jobs are stored in the paperlab-data volume. The image runs as an unprivileged user and contains Poppler, Ghostscript, qpdf, Tesseract, LibreOffice, Chromium and the fonts used by conversions.

Configuration

.env.example documents the available settings. Copy it to .env; the real .env is ignored by git.

Variable	Default	Purpose
PAPERLAB_ADDR	:8080	Listen address.
PAPERLAB_SERVE_WEB	true	Serve web/ from Gin.
PAPERLAB_WEB_DIR	web	Static front-end directory.
PAPERLAB_DATA_DIR	temp dir	Job staging and output storage.
PAPERLAB_MAX_UPLOAD_MB	250	Upload limit.
PAPERLAB_MAX_FILES	30	Files per job.
PAPERLAB_JOB_TTL_MIN	120	Time before job files are removed.
PAPERLAB_JOB_TIMEOUT_MIN	20	Processing timeout.
PAPERLAB_API_KEY	empty	Optional X-API-Key / Bearer token.
PAPERLAB_CORS_ORIGINS	*	Comma-separated allowed front-end origins.
PAPERLAB_TRUSTED_PROXIES	empty	IPs/CIDRs of reverse proxies allowed to set X-Forwarded-For.
PAPERLAB_RATE_LIMIT_PER_MIN	240	Per-IP request limit.
PAPERLAB_ALLOW_LOCAL_URLS	false	Allow HTML-to-PDF to fetch private/local URLs.
PAPERLAB_CHROME_PATH	auto-detected	Chromium/Chrome binary for HTML/URL to PDF.
PAPERLAB_LIBREOFFICE_PATH	auto-detected	LibreOffice binary for Office conversion.
PAPERLAB_AI_BASE_URL	empty	OpenAI-compatible API base URL.
PAPERLAB_AI_API_KEY	empty	AI provider key.
PAPERLAB_AI_MODEL	gpt-4o-mini	AI model name.
PAPERLAB_LIBRETRANSLATE_URL	empty	LibreTranslate base URL.

Without an AI provider, summaries and chat answers use local extraction and retrieval. Without LibreOffice, .docx, .pptx, .xlsx and .csv use the native fallbacks; legacy .doc, .ppt, .xls and OpenDocument files need LibreOffice. Translation requires an AI provider or LibreTranslate; otherwise the tool is reported as unavailable.

API

• POST /api/jobs — multipart files, optional assets, tool, and JSON options; returns a job.
• GET /api/jobs/:id — status, metadata and output list.
• GET /api/jobs/:id/files/:fileID — download an output.
• DELETE /api/jobs/:id — delete a job.
• GET /api/tools — tool inventory and availability.
• GET /api/config — public limits and engine availability.
• GET /api/health — health and engine versions.
• POST /api/inspect/pdf, POST /api/inspect/forms, POST /api/analyze/redact, POST /api/chat — interactive helpers.

Front end

web/ is plain ES modules with no build step. Gin serves it when PAPERLAB_SERVE_WEB=1. To serve it from a different origin, set the API origin in web/assets/js/config.js, with ?api=https://..., or in localStorage.paperlab_api. If the API requires a key, store it in localStorage.paperlab_api_key.

Google Drive (optional)

Drive import/export stays hidden until public Google identifiers are set in web/assets/js/config.js:

window.PAPERLAB_GOOGLE = {
  clientId: "….apps.googleusercontent.com",
  apiKey: "…",
  appId: "…", // Cloud project number
};

In Google Cloud, enable the Drive API and Google Picker API, configure an External OAuth consent screen with the drive.file scope, create a Web OAuth client with the site and localhost JavaScript origins, and restrict the API key to those origins plus the Drive/Picker APIs. The front end uses the GIS token model, so there is no client secret and no server-side token storage; tokens live in the tab's sessionStorage.

Checks

gofmt -l cmd internal scripts
go mod tidy
git diff --exit-code go.mod go.sum
go vet ./...
go test -race ./...
go build ./...
node --test web/assets/js/google-drive.test.mjs

The Docker image is the full-fidelity target because it includes LibreOffice and Chromium; a bare Go install may report those engines as unavailable.

Contributing

Bug reports, tests, documentation and focused pull requests are welcome. Open an issue for a large change first, keep the change focused, and run the checks above. By contributing, you agree that your contribution is licensed under Apache-2.0.

Security problems should be reported privately through GitHub's Security tab rather than a public issue.

License

PaperLab is Copyright 2026 Harshal Sawant and licensed under the Apache License 2.0. The full text is in LICENSE. Third-party components, including the bundled fonts, PDF.js and the Go modules, are listed in THIRD_PARTY_NOTICES.md. The Docker image also contains Ghostscript (AGPL-3.0), Poppler (GPL-2.0), LibreOffice (MPL-2.0), Tesseract, qpdf and Chromium; see that file before redistributing the image.

PaperLab is an independent project and is not affiliated with iLovePDF or any other PDF brand.