From 230fd4acbf91e4c1c75fff73b37d49006975da6c Mon Sep 17 00:00:00 2001
From: Tom Udding <tomudding@users.noreply.github.com>
Date: Sat, 24 Aug 2024 22:00:56 +0200
Subject: [PATCH] feat: check GEWISDB API to ensure it is safe to sync

---
 .env.dist    |  4 ++++
 importdb.php | 63 +++++++++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 66 insertions(+), 1 deletion(-)

diff --git a/.env.dist b/.env.dist
index d25e1dbd1f..5ebd7439a8 100644
--- a/.env.dist
+++ b/.env.dist
@@ -42,6 +42,10 @@ MH_SMTP_BIND_ADDR=0.0.0.0:25
 # Settings for Pwned Passwords API to check if passwords are present in known breaches
 PWNED_PASSWORDS_HOST=https://pwned-passwords.gewis.nl/api
 
+# Settings for GEWISDB API health check
+GEWISDB_API_HOST=https://database.test.gewis.nl/api
+GEWISDB_API_KEY='thiskeyisnotvalid'
+
 # Google Calendar API (Option Calendar) settings
 DOCKER_GOOGLE_API_KEY=unknown
 DOCKER_GOOGLE_CALENDAR_KEY=unknown
diff --git a/importdb.php b/importdb.php
index 3fa2cbe561..d494beeec9 100644
--- a/importdb.php
+++ b/importdb.php
@@ -8,6 +8,68 @@
  * It is a simple PostgreSQL to MySQL copy script.
  */
 
+$apiKey = getenv('GEWISDB_API_KEY');
+$apiHost = getenv('GEWISDB_API_HOST');
+
+if (
+    false === $apiKey
+    || false === $apiHost
+) {
+    echo 'API: no sync, environment variables are not set properly...' . PHP_EOL;
+    exit(1);
+}
+
+$ch = curl_init();
+
+$headers = [
+    sprintf('Authorization: Bearer %s', $apiKey),
+];
+
+curl_setopt($ch, CURLOPT_URL, $apiHost . '/health');
+curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
+curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
+curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
+
+$response = curl_exec($ch);
+
+if (false === $response) {
+    echo 'API: no sync, unexpected cURL error...' . PHP_EOL;
+    curl_close($ch);
+    exit(1);
+}
+
+$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
+
+if (
+    200 === $httpCode
+    || 403 === $httpCode
+) {
+    if (!json_validate($response)) {
+        echo 'API: no sync, invalid JSON returned...' . PHP_EOL;
+        curl_close($ch);
+        exit(1);
+    }
+
+    $health = json_decode($response, true);
+
+    if (
+        $health['healthy']
+        && !$health['sync_paused']
+    ) {
+        echo 'API: sync, healthy and syncs are allowed...' . PHP_EOL;
+    } else {
+        echo 'API: no sync, sync is paused or API is not healthy...' . PHP_EOL;
+        curl_close($ch);
+        exit(1);
+    }
+} else {
+    echo 'API: no sync, unexpected response...' . PHP_EOL;
+    curl_close($ch);
+    exit(1);
+}
+
+curl_close($ch);
+
 echo 'Commencing sync with GEWISDB...' . PHP_EOL;
 
 try {
@@ -182,7 +244,6 @@ function ($a) use ($i) {
         /**
          * Removing old data
          */
-        
         // Tables without primary keys are skipped
         if (0 === count($pks[$table])) continue;