How pdf2md is licensed

Copyright 2026 Siarhei Dudko

Two copyrights meet in this repository, and they are not interchangeable.

What is not ours

The PDF parser at the centre of this project is a third-party Rust crate published under the MIT licence. Its copyright notice and licence text, with those of every other crate linked into the WebAssembly modules, are in packages/core/THIRD-PARTY-NOTICES.md.

We did not write it, we do not claim it, and nothing below takes a single right away from it. It is a Cargo dependency, pulled from crates.io at an exact version, statically linked into every WebAssembly module this repository builds, and its licence travels with each of them. Anyone who wants the parser under MIT — to fork, to sell, to embed in a commercial product — should take it from crates.io, where it is offered on those terms. The MIT licence permits this repository to be distributed under the terms below; it does not withdraw the MIT grant from the code it covers, and this file cannot and does not attempt to.

What is ours

Everything else here — the WebAssembly bindings in crates/pdf-to-md-wasm, the npm packages @dudko.dev/pdf-to-md-core and @dudko.dev/pdf-to-md-cli, the HTTP API, the web application served from pdf2md.dudko.dev, the container images pdf2md and pdf2md-api, and the build and deployment tooling — is offered under:

• PolyForm Noncommercial License 1.0.0 — LICENSES/PolyForm-Noncommercial-1.0.0.md, SPDX PolyForm-Noncommercial-1.0.0. Any noncommercial purpose is permitted: personal study, research, experiment, hobby projects and amateur pursuits, plus use by any charitable organization, educational institution, public research organization, public safety or health organization, environmental protection organization, or government institution — the last group regardless of the source of its funding. Free, with no time limit.
• PolyForm Free Trial License 1.0.0 — LICENSES/PolyForm-Free-Trial-1.0.0.md, SPDX PolyForm-Free-Trial-1.0.0, granted in addition to the above. It permits evaluating whether the software suits a particular application for less than 32 consecutive days on behalf of you or your company. It does not permit redistribution: it is for trying, not for shipping.

The practical shape of this: the parsing is MIT and always will be, and what is licensed here is the packaging around it — the typed API, the loader that works in both a browser and Node, the CLI, the API service and the web application. That is what was built here, and that is what these terms cover.

Commercial use

Neither licence above permits commercial use of the parts that are ours beyond that 32-day evaluation. Commercial use means anything done for commercial advantage or monetary compensation — including embedding the software in a product or service offered to third parties for a fee, whether or not PDF conversion is the part being charged for, and including a company's own internal business use.

There are two ways to obtain it, depending on where the software runs:

• You run it — self-hosting an image, embedding a package, or using the CLI in a business. This needs a separate written licence from the copyright holder. Ask: sergey@dudko.dev
• We run it — the hosted service at pdf2md.dudko.dev and its API at pdf2md-api.dudko.dev. Paid accounts are not open yet. When they are, buying an account will carry commercial permission to use that service, on the terms presented to and accepted by the buyer at the time of purchase. It will not grant the right to run the software yourself; that is the first door.

A third way exists and costs nothing: take the parser crate from crates.io under MIT and build your own bindings. That option is real, it is deliberate, and saying so here is more useful than pretending otherwise.

Nothing on the web grants a licence

Rights under this licence come from this file, and from a written agreement signed by the copyright holder. They do not come from a web page.

No document published at any address — pdf2md.dudko.dev included — grants, widens or varies any right in this file, and no such document is incorporated into it by reference. Where a hosted service has terms, those terms govern that service for the person who accepted them; they say nothing about the software, and they cannot be relied on by anyone else.

This is deliberate. A licence that took its terms from whatever is published at a URL would be a licence whose terms could be changed by whoever holds that domain, and a licensee could not know at the moment of accepting it what they had accepted. Neither should be true of a licence.

The licensor

The copyright holder for the parts described under "What is ours" is Siarhei Dudko, reachable at sergey@dudko.dev.

The address and the website are conveniences for reaching that person; they are not the identity, and a message going undelivered neither grants a licence nor removes the need for one.

Third-party code

The wasm modules statically link third-party Rust crates — the parser itself, lopdf, regex, ttf-parser, unicode-normalization, include_dir and their dependencies — under permissive terms (MIT, Apache-2.0, BSD, Unicode-3.0). None of them restrict the terms above, and each keeps its own notice requirements. Those notices are reproduced in packages/core/THIRD-PARTY-NOTICES.md, which travels with the npm packages and with both container images; regenerate it with pnpm notices whenever the dependencies change.