From f3770fbc62ceb181c53d2f8910e62a279e7f0137 Mon Sep 17 00:00:00 2001
From: Shamarvey1 <ajitlookforbest@gmail.com>
Date: Thu, 16 Apr 2026 19:08:59 +0530
Subject: [PATCH] fix: prevent duplicate dataset uploads using file_id check

---
 openml_OS/models/api/v1/Api_data.php | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/openml_OS/models/api/v1/Api_data.php b/openml_OS/models/api/v1/Api_data.php
index 294d1300..c27bedab 100644
--- a/openml_OS/models/api/v1/Api_data.php
+++ b/openml_OS/models/api/v1/Api_data.php
@@ -1276,6 +1276,18 @@ private function data_upload() {
     /* * * *
      * THE ACTUAL INSERTION
      * * * */
+    // 🔍 Check for duplicate dataset using file_id
+    // 🔍 Check for duplicate dataset using file_id
+    $existing = $this->Dataset->getWhereSingle('file_id = ' . $file_id);
+    if ($existing != false) {
+      $this->returnError(
+        135,
+        $this->version,
+        $this->openmlGeneralErrorCode,
+        "Dataset with same file already exists"
+      );
+      return;
+    }
     $id = $this->Dataset->insert($dataset);
     if (!$id) {
       $this->returnError(134, $this->version);