From 5d97b9db473d91e74c5669a974847243df019b96 Mon Sep 17 00:00:00 2001
From: William Casarin <jb55@jb55.com>
Date: Wed, 24 Apr 2024 17:04:07 -0700
Subject: [PATCH] translate: don't translate repost data

This is apparently accounting for many translation requests.

I pushed a fix for future damus versions, but in the meantime this stops
us from burning money.

Follow up question: why are reposts passing the apple-ai language check?

Cc: Terry Yiu <terryyiu@gmail.com>
Reported-by: Semisol <hi@semisol.dev>
Signed-off-by: William Casarin <jb55@jb55.com>
---
 package.json           |  1 +
 src/translate.js       | 10 ++++++++++
 test/translate.test.js | 26 ++++++++++++++++++++++++--
 3 files changed, 35 insertions(+), 2 deletions(-)

diff --git a/package.json b/package.json
index c3b68b7..2f0b63f 100644
--- a/package.json
+++ b/package.json
@@ -11,6 +11,7 @@
   },
   "scripts": {
     "test": "ALLOW_HTTP_AUTH=\"true\" DEEPL_KEY=123 tap test/*.test.js",
+    "test-translate": "ALLOW_HTTP_AUTH=\"true\" TRANSLATION_PROVIDER=mock tap test/translate.test.js",
     "debug-test": "ALLOW_HTTP_AUTH=\"true\" DEEPL_KEY=123 tap test/*.test.js --timeout=2400",
     "start": "node src/index.js",
     "mock_deepl": "node test_utils/mock_deepl.js",
diff --git a/src/translate.js b/src/translate.js
index a691274..c46e8ed 100644
--- a/src/translate.js
+++ b/src/translate.js
@@ -100,12 +100,22 @@ async function translate_payload(api, res, payload, trans_id) {
   delete api.translation.queue[trans_id]
 }
 
+function payload_is_data(q) {
+  try {
+    return Object.keys(JSON.parse(q)).length > 0
+  } catch {
+    return false
+  }
+}
+
 async function handle_translate(api, req, res) {
   let id
   try {
     const source = req.query.source.toUpperCase()
     const target = req.query.target.toUpperCase()
     const q = req.query.q
+    if (payload_is_data(q))
+      return util.invalid_request(res, `payload is data`)
     const payload = { source, target, q }
     const validation_res = await validate_payload(payload)
     if (validation_res.ok === false)
diff --git a/test/translate.test.js b/test/translate.test.js
index 6785cf1..081ce15 100644
--- a/test/translate.test.js
+++ b/test/translate.test.js
@@ -29,6 +29,28 @@ tap.test('translate_payload - Existing translation in database (mocked db)', asy
   t.end();
 });
 
+tap.test(`translate_payload - reject json data (mocked server)`, async (t) => {
+  const api = await generate_test_api(t, {
+    simulate_existing_translation_in_db: false,
+    simulate_account_found_in_db: true,
+  });
+
+  const payload = JSON.stringify({"hello": 1})
+  const expected_result = {
+    error: "payload is data",
+  };
+
+  const url = `/translate?source=EN&target=JA&q=${encodeURIComponent(payload)}`
+  const test_data = await generate_test_request_data(api, url)
+
+  const res = await api.test_request
+    .get(url)
+    .set('Authorization', 'Nostr ' + test_data.auth_note_base64)
+
+  t.same(res.statusCode, 400, 'Response should be 400');
+  t.same(res.body, expected_result, 'Response should match expected value');
+})
+
 tap.test('translate_payload - New translation (mocked server)', async (t) => {
   const api = await generate_test_api(t, {
     simulate_existing_translation_in_db: false,
@@ -158,11 +180,11 @@ async function generate_test_api(t, config) {
   return api;
 }
 
-async function generate_test_request_data(api) {
+async function generate_test_request_data(api, q) {
   let test_privkey = '10a9842fadc0aae2a649a1b707bf97e48c787b8517af4728ba3ec304089451be';
   let test_pubkey = nostr.getPublicKey(test_privkey);
 
-  let query_url = '/translate?source=EN&target=JA&q=Hello'
+  let query_url = q || '/translate?source=EN&target=JA&q=Hello'
   let full_query_url = api.router.base_url + query_url;
 
   let auth_note_template = {