From 0e25aaf1d92c08a4a7b94de214e46213a05b8478 Mon Sep 17 00:00:00 2001
From: Senichenkov <Petua41@yandex.ru>
Date: Thu, 28 Nov 2024 12:17:04 +0300
Subject: [PATCH] Fix UB found by clang UB sanitizer

- Check that column is numeric *before* casting it to INumericType
in DataStats::GetMedianAD
- Don't process empty vector in GetPartition in gfd_validation.cpp
- Disable tests causing float-to-long cast when result is infinity
---
 src/core/algorithms/gfd/gfd_validation.cpp    |  4 ++++
 src/core/algorithms/statistics/data_stats.cpp |  2 +-
 src/tests/test_types.cpp                      | 14 ++++++++++----
 3 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/src/core/algorithms/gfd/gfd_validation.cpp b/src/core/algorithms/gfd/gfd_validation.cpp
index 1f9bf64dec..39a3e4b0c9 100644
--- a/src/core/algorithms/gfd/gfd_validation.cpp
+++ b/src/core/algorithms/gfd/gfd_validation.cpp
@@ -25,6 +25,10 @@ std::vector<std::vector<vertex_t>> GetPartition(std::vector<vertex_t> const& can
                                                 config::ThreadNumType const& threads_num) {
     std::vector<std::vector<vertex_t>> result = {};
 
+    if (candidates.empty()) {
+        return {};
+    }
+
     int musthave = candidates.size() / threads_num;
     int oversized_num = candidates.size() % threads_num;
 
diff --git a/src/core/algorithms/statistics/data_stats.cpp b/src/core/algorithms/statistics/data_stats.cpp
index 809cf42863..121b044c8a 100644
--- a/src/core/algorithms/statistics/data_stats.cpp
+++ b/src/core/algorithms/statistics/data_stats.cpp
@@ -464,8 +464,8 @@ Statistic DataStats::GetMedianAD(size_t index) const {
         return all_stats_[index].median_ad;
     }
     mo::TypedColumnData const& col = col_data_[index];
-    auto const& type = static_cast<mo::INumericType const&>(col.GetType());
     if (!col.IsNumeric()) return {};
+    auto const& type = static_cast<mo::INumericType const&>(col.GetType());
 
     std::vector<std::byte const*> data = DeleteNullAndEmpties(index);
     std::byte* median = MedianOfNumericVector(data, type);
diff --git a/src/tests/test_types.cpp b/src/tests/test_types.cpp
index 7e11b6b55f..38fcb95e18 100644
--- a/src/tests/test_types.cpp
+++ b/src/tests/test_types.cpp
@@ -182,10 +182,16 @@ TYPED_TEST(TestNumeric, Pow) {
 
     test(0, 100);
     test(22, 12);
-    test(123, 321);
-    test(Type(2.72), Type(1.3123141));
-    test(-102, 11);
-    test(-123, 123);
+    // 123^321 won't fit into long (i. e. IntType) -- it's UB
+    if constexpr (!std::is_same_v<Type, long>) {
+        test(123, 321);
+    }
+    test(Type(2.72), 1.3123141);
+    // -102^11 and -123^123 won't fit into long (i. e. IntType) -- it's UB
+    if constexpr (!std::is_same_v<Type, long>) {
+        test(-102, 11);
+        test(-123, 123);
+    }
     test(-21, -7);
 }