#5843 MathLib::toLongNumber()/MathLib::toULongNumber() now use min/max values for bigint/biguint type in case the number to converted cannot be converted properly

lib/mathlib.cpp

@@ -24,6 +24,7 @@
 #include <cmath>
 #include <cctype>
 #include <limits>
+#include <limits>
 
 MathLib::biguint MathLib::toULongNumber(const std::string & str)
 {
@@ -64,7 +65,13 @@ MathLib::biguint MathLib::toULongNumber(const std::string & str)
     }
 
     if (isFloat(str)) {
-        return static_cast<biguint>(std::atof(str.c_str()));
+        // Things are going to be less precise now: the value can't b represented in the biguint type.
+        // Use min/max values as an approximation. See #5843
+        const double doubleval = std::atof(str.c_str());
+        if (doubleval > (double)std::numeric_limits<biguint>::max())
+            return std::numeric_limits<biguint>::max();
+        else
+            return static_cast<biguint>(doubleval);
     }
 
     biguint ret = 0;
@@ -112,7 +119,15 @@ MathLib::bigint MathLib::toLongNumber(const std::string & str)
     }
 
     if (isFloat(str)) {
-        return static_cast<bigint>(std::atof(str.c_str()));
+        // Things are going to be less precise now: the value can't be represented in the bigint type.
+        // Use min/max values as an approximation. See #5843
+        const double doubleval = std::atof(str.c_str());
+        if (doubleval > (double)std::numeric_limits<bigint>::max())
+            return std::numeric_limits<bigint>::max();
+        else if (doubleval < (double)std::numeric_limits<bigint>::min())
+            return std::numeric_limits<bigint>::min();
+        else
+            return static_cast<bigint>(doubleval);
     }
 
     bigint ret = 0;