Fix overflow of Mod_Float for large number (#1522)

Fix issue: https://github.com/microsoft/Power-Fx/issues/1367

Problem this PR fix:
Start from 1e+16 to 1e+20, current mod_float return 0 for most of the
case (mostly wrong value). And overflow from 1e+21 or higher.
This PR use C# % function to do mod for the wrong/overflow issues.

Note on PA (Need to fix as well): mod result is always 0 from 1e18 (mod
123). So, it shared the same problem. mod result is wrong for > 1e17. It
did not return overflow error for 1e306 but return 0 (wrong value).

---------

Co-authored-by: Carlos Figueira <carlosff@microsoft.com>

src/libraries/Microsoft.PowerFx.Interpreter/Functions/LibraryMath.cs

@@ -760,20 +760,14 @@ namespace Microsoft.PowerFx.Functions
                 return CommonErrors.DivByZeroError(irContext);
             }
 
-            // r = a – N × floor(a/b)
-            double q = Math.Floor(arg0 / arg1);
-            if (IsInvalidDouble(q))
-            {
-                return CommonErrors.OverflowError(irContext);
-            }
+            double result = arg0 % arg1;
 
-            double result = arg0 - (arg1 * ((long)q));
-
-            // We validate the reminder is in a valid range.
-            // This is mainly to support very large numbers (like 1E+308) where the calculation could be incorrect
-            if (result < -Math.Abs(arg1) || result > Math.Abs(arg1))
+            // % result has the same sign with dividend.
+            // Using the following fomular to have the sign of mod result is the same as divisor as in Excel and PA Mod.
+            // If result has different sign with divisor, plus result with divisor to flip the sign to be the same with divisor.
+            if (Math.Sign(result) * Math.Sign(arg1) == -1)
             {
-                return CommonErrors.OverflowError(irContext);
+                result = result + arg1;
             }
 
             return new NumberValue(irContext, result);
@@ -783,25 +777,22 @@ namespace Microsoft.PowerFx.Functions
         {
             decimal arg0 = arg0dv.Value;
             decimal arg1 = arg1dv.Value;
-            decimal q;
-
+            
             // Both decimal zero and negative zero will satisfy this test
             if (arg1 == 0m)
             {
                 return CommonErrors.DivByZeroError(irContext);
             }
 
-            // r = a – N × floor(a/b)
-            try
-            {
-                q = decimal.Floor(arg0 / arg1);
-            }
-            catch (OverflowException)
-            {
-                return CommonErrors.OverflowError(irContext);
-            }
+            decimal result = arg0 % arg1;
 
-            decimal result = arg0 - (arg1 * q);
+            // % result has the same sign with dividend.
+            // Using the following fomular to have the sign of mod result is the same as divisor as in Excel and PA Mod.
+            // If result has different sign with divisor, plus result with divisor to flip the sign to be the same with divisor.
+            if (Math.Sign(result) * Math.Sign(arg1) == -1)
+            {
+                result = result + arg1;
+            }
 
             return new DecimalValue(irContext, result);
         }

src/tests/Microsoft.PowerFx.Core.Tests/ExpressionTestCases/ErrorKinds.txt

@@ -410,9 +410,6 @@ Error({Kind:ErrorKind.Div0})
 >> Mod(2,0)
 Error({Kind:ErrorKind.Div0})
 
->> Mod(10^200,-10^-200)
-Error({Kind:ErrorKind.Numeric})
-
 >> Mod(Sqrt(-1),If(Char(0)="a",1))
 Error(Table({Kind:ErrorKind.Numeric},{Kind:ErrorKind.InvalidArgument}))
 

src/tests/Microsoft.PowerFx.Core.Tests/ExpressionTestCases/FloatLarge.txt

@@ -72,7 +72,7 @@ Error({Kind:ErrorKind.Numeric})
 100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
 
 >> Mod(1E+300,1E-20)
-Error({Kind:ErrorKind.Numeric})
+6.0682893566359166E-21
 
 >> Dec2Hex([1,1e-45,1e45,2])
 Table({Value:"1"},{Value:"0"},{Value:Error({Kind:ErrorKind.Numeric})},{Value:"2"})

src/tests/Microsoft.PowerFx.Core.Tests/ExpressionTestCases/Mod_Decimal.txt

@@ -0,0 +1,19 @@
+#SETUP: disable:NumberIsFloat
+
+>> Mod(1E+27, 3)
+1
+
+>> Mod(-1E+27, -3)
+-1
+
+>> Mod(1E+19, 123)
+37
+
+>> Mod(1.2345E+26, 67.89)
+0.81
+
+>> Mod(79228162514264337593543.1234, 3.5678)
+1.3776
+
+>> Mod(79228162514264337593543950335, 123)
+99

src/tests/Microsoft.PowerFx.Core.Tests/ExpressionTestCases/Mod_Float.txt

@@ -0,0 +1,35 @@
+#SETUP: NumberIsFloat
+
+>> Mod(1E+307, 3)
+1
+
+>> Mod(1E+307, -3)
+-2
+
+>> Mod(-1E+307, 3)
+2
+
+>> Mod(-1E+307, -3)
+-1
+
+>> Mod(1.7976931348623157E+300, 3)
+1
+
+>> Mod(1.2345E+260, 67.89)
+41.28195726574958
+
+>> Mod(1E+13, 123)
+16
+
+>> Mod(1E+18, 123)
+16
+
+>> Mod(1E+18, -123)
+-107
+
+// Not exact result due to floating point semantics
+>> Mod(10^200,-10^-200)
+-5.8180083079168396E-201
+
+>> Mod(1E+400, 3)
+Errors: Error 4-10: Numeric value is too large.|Error 0-14: The function 'Mod' has some invalid arguments.|Error 4-10: Invalid argument type (Error). Expecting a Number value instead.