From bb1e38cd08c1945b20b497caeff69a65b8329629 Mon Sep 17 00:00:00 2001
From: Vincent <git@vincenttunru.com>
Date: Mon, 15 Jul 2024 14:19:41 +0200
Subject: [PATCH] Port breach data fetching cronjob to TS

---
 package.json                                  |  3 ++-
 src/db/tables/breaches.js                     |  4 ++--
 .../syncBreaches.ts}                          | 22 +++++++++----------
 src/scripts/s3.js                             |  6 ++++-
 4 files changed, 20 insertions(+), 15 deletions(-)
 rename src/scripts/{syncBreaches.js => cronjobs/syncBreaches.ts} (86%)

diff --git a/package.json b/package.json
index 5633cbe48..a2ef69c89 100644
--- a/package.json
+++ b/package.json
@@ -12,6 +12,7 @@
     "dev:cron:first-data-broker-removal-fixed": "tsx --tsconfig tsconfig.cronjobs.json src/scripts/cronjobs/firstDataBrokerRemovalFixed.tsx",
     "dev:cron:monthly-activity": "tsx --tsconfig tsconfig.cronjobs.json src/scripts/cronjobs/monthlyActivity.tsx",
     "dev:cron:db-delete-unverified-subscribers": "tsx --tsconfig tsconfig.cronjobs.json src/scripts/cronjobs/deleteUnverifiedSubscribers.ts",
+    "dev:cron:db-pull-breaches": "tsx --tsconfig tsconfig.cronjobs.json src/scripts/cronjobs/syncBreaches.ts",
     "dev:cron:onerep-limits-alert": "tsx --tsconfig tsconfig.cronjobs.json src/scripts/cronjobs/onerepStatsAlert.ts",
     "dev:nimbus": "node --watch-path config/nimbus.yaml src/scripts/build/nimbusTypes.js",
     "build": "npm run get-location-data && npm run build-glean && npm run build-nimbus && next build && npm run build-cronjobs",
@@ -27,7 +28,7 @@
     "cron:monthly-activity": "node dist/scripts/cronjobs/monthlyActivity.js",
     "cron:breach-alerts": "node src/scripts/emailBreachAlerts.js",
     "cron:db-delete-unverified-subscribers": "node dist/scripts/cronjobs/deleteUnverifiedSubscribers.js",
-    "cron:db-pull-breaches": "node src/scripts/syncBreaches.js",
+    "cron:db-pull-breaches": "node dist/scripts/cronjobs/syncBreaches.js",
     "cron:remote-settings-pull-breaches": "node scripts/updatebreaches.js",
     "cron:onerep-limits-alert": "node dist/scripts/cronjobs/onerepStatsAlert.js",
     "db:migrate": "node -r dotenv-flow/config node_modules/knex/bin/cli.js migrate:latest --knexfile src/db/knexfile.js",
diff --git a/src/db/tables/breaches.js b/src/db/tables/breaches.js
index 9cbf75cc0..993012001 100644
--- a/src/db/tables/breaches.js
+++ b/src/db/tables/breaches.js
@@ -88,8 +88,8 @@ async function upsertBreaches(hibpBreaches) {
 /**
  * Update logo path of a breach by name
  *
- * @param {string} name 
- * @param {string} faviconUrl
+ * @param {string} name
+ * @param {string | null} faviconUrl
  */
 // Not covered by tests; mostly side-effects. See test-coverage.md#mock-heavy
 /* c8 ignore start */
diff --git a/src/scripts/syncBreaches.js b/src/scripts/cronjobs/syncBreaches.ts
similarity index 86%
rename from src/scripts/syncBreaches.js
rename to src/scripts/cronjobs/syncBreaches.ts
index d8eafb610..02d7e4a41 100644
--- a/src/scripts/syncBreaches.js
+++ b/src/scripts/cronjobs/syncBreaches.ts
@@ -12,13 +12,14 @@
 import { readdir } from "node:fs/promises";
 import os from "node:os";
 import Sentry from "@sentry/nextjs";
-import { req, formatDataClassesArray } from "../utils/hibp.js";
+import { req, formatDataClassesArray } from "../../utils/hibp.js";
 import {
   getAllBreaches,
   upsertBreaches,
   updateBreachFaviconUrl,
-} from "../db/tables/breaches.js";
-import { uploadToS3 } from "./s3.js";
+} from "../../db/tables/breaches.js";
+import { uploadToS3 } from "../s3.js";
+import type { Breach } from "../../app/functions/universal/breach.js";
 
 const SENTRY_SLUG = "cron-sync-breaches";
 
@@ -32,7 +33,7 @@ const checkInId = Sentry.captureCheckIn({
   status: "in_progress",
 });
 
-export async function getBreachIcons(breaches) {
+export async function getBreachIcons(breaches: Breach[]) {
   // make logofolder if it doesn't exist
   const logoFolder = os.tmpdir();
   console.log(`Logo folder: ${logoFolder}`);
@@ -82,20 +83,19 @@ export async function getBreachIcons(breaches) {
 }
 
 // Get breaches and upserts to DB
-const breachesResponse = await req("/breaches");
-const breaches = [];
+const breachesResponse: Breach[] = await req("/breaches");
+const breaches: Breach[] = [];
 const seen = new Set();
 for (const breach of breachesResponse) {
   breach.DataClasses = formatDataClassesArray(breach.DataClasses);
-  breach.LogoPath = /[^/]*$/.exec(breach.LogoPath)[0];
+  breach.LogoPath = /[^/]*$/.exec(breach.LogoPath)![0];
   breaches.push(breach);
   seen.add(breach.Name + breach.BreachDate);
 
   // sanity check: corrupt data structure
   if (!isValidBreach(breach))
     throw new Error(
-      "Breach data structure is not valid",
-      JSON.stringify(breach),
+      "Breach data structure is not valid: " + JSON.stringify(breach),
     );
 }
 
@@ -128,10 +128,10 @@ setTimeout(process.exit, 1000);
 /**
  * Null check for some required field
  *
- * @param {object} breach breach object from HIBP
+ * @param breach breach object from HIBP
  * @returns Boolean is it a valid breach
  */
-function isValidBreach(breach) {
+function isValidBreach(breach: Breach) {
   return (
     breach.Name !== undefined &&
     breach.BreachDate !== undefined &&
diff --git a/src/scripts/s3.js b/src/scripts/s3.js
index 3a40eda17..11f5fcc78 100644
--- a/src/scripts/s3.js
+++ b/src/scripts/s3.js
@@ -24,6 +24,10 @@ const s3 = new S3({
   },
 });
 
+/**
+ * @param {string} fileName
+ * @param {Buffer} fileStream
+ */
 export async function uploadToS3(fileName, fileStream) {
   console.log("Attempt to upload to s3: ", fileName);
   const uploadParams = {
@@ -37,7 +41,7 @@ export async function uploadToS3(fileName, fileStream) {
       params: uploadParams,
     }).done();
     console.log("Successfully uploaded data to " + Bucket + "/" + fileName);
-  } catch (err) {
+  } catch (/** @type {any} */ err) {
     console.error(err, err.stack);
   }
 }