feat: NIP-03 opentimestamps support (#105)

.changeset/nip-03-opentimestamps.md

@@ -0,0 +1,5 @@
+---
+"nostream": minor
+---
+
+Add NIP-03 OpenTimestamps support for kind 1040 events: structural `.ots` validation, Bitcoin attestation requirement, digest match to the referenced `e` tag, and relay metadata updates (#105).

README.md

@@ -47,6 +47,7 @@ NIPs with a relay-specific implementation are listed here.
 
 - [x] NIP-01: Basic protocol flow description
 - [x] NIP-02: Contact list and petnames
+- [x] NIP-03: OpenTimestamps Attestations for Events
 - [x] NIP-04: Encrypted Direct Message
 - [x] NIP-09: Event deletion
 - [x] NIP-11: Relay information document

package.json

@@ -5,6 +5,7 @@
  "supportedNips": [
  1,
  2,
+ 3,
  4,
  9,
  11,
@@ -49,6 +50,7 @@
  "docker:build": "docker build -t nostream .",
  "pretest:integration": "mkdir -p .test-reports/integration",
  "test:load": "node -r ts-node/register ./scripts/security-load-test.ts",
+ "smoke:nip03": "node -r ts-node/register scripts/smoke-nip03.ts",
  "test:integration": "cucumber-js",
  "cover:integration": "nyc --report-dir .coverage/integration npm run test:integration -- -p cover",
  "export": "node --env-file-if-exists=.env -r ts-node/register src/scripts/export-events.ts",

scripts/smoke-nip03.md

@@ -0,0 +1,146 @@
+# NIP-03 smoke test with a real OpenTimestamps client
+
+This exercises a running nostream relay against a genuine Bitcoin-attested
+OpenTimestamps proof that a real OTS client produced in the wild (the
+event used in the [NIP-03 spec example](https://github.com/nostr-protocol/nips/blob/master/03.md)
+on `wss://nostr-pub.wellorder.net`).
+
+The relay never sees a synthetic proof here: the `.ots` blob in `content`
+was made by a real `ots stamp` + `ots upgrade` flow, is attested to a real
+Bitcoin block header, and is validated by the real `ots verify` binary
+against a public Esplora server before and after it round-trips through
+your relay.
+
+## Why not generate our own proof end-to-end?
+
+`ots stamp` writes a pending proof. A pending proof has to sit in a
+calendar server's queue for several Bitcoin blocks (typically a few
+hours) before `ots upgrade` can turn it into a confirmed Bitcoin
+attestation. Running that end-to-end in CI or on a developer machine is
+impractical. Re-using an already-upgraded, already-published kind 1040
+event is an equally honest "real client" test: we did not make the
+proof, and we prove its validity with the same binary a Nostr client
+would use.
+
+## Prerequisites
+
+- A running nostream relay (default `ws://127.0.0.1:8008`).
+- Node.js (same as the repo; the script runs via `ts-node` and uses the `ws` package).
+- [`opentimestamps-client`](https://github.com/opentimestamps/opentimestamps-client)
+ for the real `ots` step (optional; the script auto-detects and skips
+ gracefully if it's not installed):
+ - Linux / macOS: `pipx install opentimestamps-client`, or
+ `pip install opentimestamps-client`.
+ - Windows: Python 3.13 has an OpenSSL compatibility bug in
+ `python-bitcoinlib` on which the client depends. Run it inside a
+ container instead:
+
+ ```bash
+ docker run --rm -v $PWD:/work python:3.11-slim \
+ sh -c "pip install -q opentimestamps-client && ots info /work/proof.ots"
+ ```
+- A Bitcoin node (optional, only for `--verify`). Without one the script
+ runs `ots info`, which parses the proof and confirms the Bitcoin
+ attestation it terminates in. `ots verify` additionally looks up the
+ block header on a Bitcoin node to prove the attestation is genuine; if
+ you don't have one, `ots info` is the honest equivalent of
+ structural-client acceptance.
+
+## Running the automated script
+
+```bash
+npm run smoke:nip03
+# or, with non-default relays:
+npx ts-node scripts/smoke-nip03.ts \
+ --local-relay ws://127.0.0.1:8008 \
+ --source-relay wss://nostr-pub.wellorder.net
+```
+
+Expected output on a healthy relay with `ots` installed:
+
+```
+NIP-03 end-to-end smoke test
+ local relay: ws://127.0.0.1:8008
+ source relays: wss://nos.lol, wss://relay.damus.io, wss://nostr.wine, wss://offchain.pub, wss://nostr-pub.wellorder.net
+
+1) Discovering a real NIP-03 event from public relays
+ trying wss://nos.lol for any recent kind 1040…
+ PASS discovered 697b40df2f1c… on wss://nos.lol (pubkey=b1104a6e…, attests e=88fea43a70bd…, content=4968 chars)
+
+2) Parsing OTS content with the real `ots` client
+ PASS ots info parsed proof — BitcoinBlockHeaderAttestation(941057) (file: /tmp/nip03-XXXX/proof.ots)
+
+3) Publishing the real event to the local relay
+ PASS local relay accepted real NIP-03 event (reason="")
+
+4) Round-tripping the event through the local relay
+ PASS local relay returned the same event (id, sig, content) on REQ
+
+summary: 3 passed, 0 failed
+```
+
+Pass `--verify` (or set `OTS_VERIFY=1`) to additionally run
+`ots verify -d <target-event-id>` which asks a Bitcoin node to confirm
+the block header. Exit code is `0` iff every step passes.
+
+## What each step proves
+
+1. **Source discovery** — confirms a real third-party kind 1040 event
+ exists, came from a real signed OTS client flow, and that you and the
+ network agree on its bytes.
+2. **`ots info` (or `ots verify`)** — confirms the `.ots` content in
+ `event.content` is a structurally valid OpenTimestamps proof when fed
+ to the real reference client, and that it terminates in a Bitcoin
+ block header attestation (which is what NIP-03 requires). If
+ `--verify` is set, additionally walks the Bitcoin header via a
+ configured node to prove the attested block really contains the merkle
+ root.
+3. **Publish** — confirms nostream's NIP-03 strategy accepts a
+ real-world, real-client-produced kind 1040 event (structure,
+ `e` tag, digest match, Bitcoin attestation requirement all satisfied).
+4. **Round-trip** — confirms the relay persisted the event unchanged and
+ returns the exact same id, signature, and base64 content on REQ, so
+ downstream clients that re-run `ots verify` on the relay's output will
+ still succeed.
+
+## Manual walkthrough (if you want to stamp your own)
+
+If you do have a few hours to wait and want a proof you made yourself:
+
+```bash
+export RELAY=ws://127.0.0.1:8008
+export SK=$(nak key generate)
+export EVENT_ID=$(nak event --sec "$SK" -k 1 -c "anchor this note" "$RELAY" | jq -r '.[1].id // .id')
+
+# Stamp the raw 32 bytes of the event id (not the hex string)
+echo -n "$EVENT_ID" | xxd -r -p > /tmp/nip03-digest.bin
+ots stamp /tmp/nip03-digest.bin
+
+# Wait for calendars + Bitcoin confirmation, then:
+ots upgrade /tmp/nip03-digest.bin.ots
+ots verify /tmp/nip03-digest.bin.ots
+
+export OTS_B64=$(base64 -w0 /tmp/nip03-digest.bin.ots)
+nak event --sec "$SK" -k 1040 \
+ -t e="$EVENT_ID" \
+ -t k=1 \
+ -c "$OTS_B64" \
+ "$RELAY"
+
+# round-trip
+nak req -k 1040 -a "$(nak key public "$SK")" "$RELAY" \
+ | jq -r '.content' | base64 -d | ots verify -
+```
+
+Each publish attempt should come back as `["OK", "<id>", true, ""]`.
+
+## Negative paths
+
+Actively testing NIP-03 rejection paths (mismatched digest, uppercase
+`e` tag, multiple `k` tags, unsupported OTS version, garbage content)
+would require re-signing a mutated event, which means the proof would no
+longer be produced by a real OTS client. Those paths are covered in
+isolation by the unit tests:
+
+- `test/unit/utils/nip03.spec.ts`
+- `test/unit/handlers/event-strategies/timestamp-event-strategy.spec.ts`