I2527: Fixed NaN issue in variance functions

src/Parsers/Kusto/KustoFunctions/KQLAggregationFunctions.cpp

@@ -599,7 +599,18 @@ bool TakeAnyIf::convertImpl(String & out, IParser::Pos & pos)
 
 bool Variance::convertImpl(String & out, IParser::Pos & pos)
 {
-    return directMapping(out, pos, "varSamp");
+    const String fn_name = getKQLFunctionName(pos);
+    if (fn_name.empty())
+        return false;
+
+    const String expr = getArgument(fn_name, pos);
+    out = std::format(
+        "IF (isNaN(varSamp(if(toTypeName({0}) = 'Nullable(Nothing)', throwIf(toTypeName({0}) = 'Nullable(Nothing)', "
+        "'summarize operator: Failed to resolve scalar expression named null'), {0})) AS variance_{1}), 0, variance_{1})",
+        expr,
+        generateUniqueIdentifier());
+
+    return true;
 }
 
 bool VarianceIf::convertImpl(String & out, IParser::Pos & pos)
@@ -610,14 +621,30 @@ bool VarianceIf::convertImpl(String & out, IParser::Pos & pos)
 
     const String expr = getArgument(fn_name, pos);
     const String predicate = getArgument(fn_name, pos);
-    out = std::format("varSampIf({}, {})", expr, predicate);
+    out = std::format(
+        "IF (isNaN(varSampIf((if(toTypeName({0}) = 'Nullable(Nothing)', throwIf(toTypeName({0}) = 'Nullable(Nothing)', "
+        "'summarize operator: Failed to resolve scalar expression named null'), {0})), {1}) AS variance_{2}), 0, variance_{2})",
+        expr,
+        predicate,
+        generateUniqueIdentifier());
 
     return true;
 }
 
 bool VarianceP::convertImpl(String & out, IParser::Pos & pos)
 {
-    return directMapping(out, pos, "varPop");
+    const String fn_name = getKQLFunctionName(pos);
+    if (fn_name.empty())
+        return false;
+
+    const String expr = getArgument(fn_name, pos);
+    out = std::format(
+        "IF (isNaN(varPop(if(toTypeName({0}) = 'Nullable(Nothing)', throwIf(toTypeName({0}) = 'Nullable(Nothing)', "
+        "'summarize operator: Failed to resolve scalar expression named null'), {0})) AS variance_{1}), 0, variance_{1})",
+        expr,
+        generateUniqueIdentifier());
+
+    return true;
 }
 
 bool CountDistinct::convertImpl(String & out, IParser::Pos & pos)

src/Parsers/tests/KQL/gtest_KQL_AggregateFunctions.cpp

@@ -110,18 +110,6 @@ INSTANTIATE_TEST_SUITE_P(ParserKQLQuery_Aggregate, ParserTest,
             "Customers | summarize by FirstName, LastName, Age",
             "SELECT\n    FirstName,\n    LastName,\n    Age\nFROM Customers\nGROUP BY\n    FirstName,\n    LastName,\n    Age"
         },
-        {
-            "Customers | summarize variance(Age)",
-            "SELECT varSamp(Age) AS variance_Age\nFROM Customers"
-        },
-        {
-            "Customers | summarize variancep(Age)",
-            "SELECT varPop(Age) AS variancep_Age\nFROM Customers"
-        },
-        {
-            "Customers | summarize varianceif(Age, Age < 30)",
-            "SELECT varSampIf(Age, Age < 30) AS varianceif_Age\nFROM Customers"
-        },
         {
             "Customers | summarize z=arg_max(Age, FirstName, LastName) by Occupation",
             "SELECT\n    Occupation,\n    argMax(FirstName, Age) AS FirstName,\n    argMax(LastName, Age) AS LastName,\n    argMax(Age, Age) AS z\nFROM Customers\nGROUP BY Occupation"
@@ -135,3 +123,21 @@ INSTANTIATE_TEST_SUITE_P(ParserKQLQuery_Aggregate, ParserTest,
             "SELECT uniqCombined64Merge(18)(xy) AS Column1\nFROM\n(\n    SELECT uniqCombined64MergeState(18)(arrayJoin([x, y])) AS xy\n    FROM\n    (\n        SELECT\n            uniqCombined64State(18)(Education) AS x,\n            uniqCombined64State(18)(Occupation) AS y\n        FROM Customers\n    )\n)"
         }
 })));
+
+INSTANTIATE_TEST_SUITE_P(ParserKQLQuery_Aggregate, ParserRegexTest,
+    ::testing::Combine(
+        ::testing::Values(std::make_shared<DB::ParserKQLStatement>()),
+        ::testing::ValuesIn(std::initializer_list<ParserTestCase>{
+        {
+            "Customers | summarize variance(Age)",
+            R"(SELECT IF\(isNaN\(varSamp\(if\(toTypeName\(Age\) = \'Nullable\(Nothing\)\', throwIf\(toTypeName\(Age\) = \'Nullable\(Nothing\)\', \'summarize operator: Failed to resolve scalar expression named null\'\), Age\)\) AS variance_\d+\), 0, variance_\d+\) AS variance_Age\nFROM Customers)"
+        },
+        {
+            "Customers | summarize variancep(Age)",
+            R"(SELECT IF\(isNaN\(varPop\(if\(toTypeName\(Age\) = \'Nullable\(Nothing\)\', throwIf\(toTypeName\(Age\) = \'Nullable\(Nothing\)\', \'summarize operator: Failed to resolve scalar expression named null\'\), Age\)\) AS variance_\d+\), 0, variance_\d+\) AS variancep_Age\nFROM Customers)"
+        },
+        {
+            "Customers | summarize varianceif(Age, Age < 30)",
+            R"(SELECT IF\(isNaN\(varSampIf\(if\(toTypeName\(Age\) = \'Nullable\(Nothing\)\', throwIf\(toTypeName\(Age\) = \'Nullable\(Nothing\)\', \'summarize operator: Failed to resolve scalar expression named null\'\), Age\), Age < 30\) AS variance_\d+\), 0, variance_\d+\) AS varianceif_Age\nFROM Customers)"
+        }
+})));

tests/queries/0_stateless/02366_kql_summarize.sql

@@ -132,6 +132,9 @@ print '-- variance/variancep/varianceif --';
 Customers | summarize variance(Age);
 Customers | summarize variancep(Age);
 Customers | summarize varianceif(Age, Age < 30);
+Customers | summarize variance(null); -- { clientError Code: 395 }
+Customers | summarize variancep(null); -- { clientError Code: 395 }
+Customers | summarize varianceif(null, Age < 30); -- { clientError Code: 395 }
 
 print '-- arg_max --';
 Customers | summarize arg_max(Age); -- { clientError NUMBER_OF_ARGUMENTS_DOESNT_MATCH }