table function tests

Author: EpsilonPrime

grammar/FuncTestCaseLexer.g4

@@ -11,6 +11,7 @@ Whitespace    : [ \t\n\r]+ -> channel(HIDDEN) ;
 TripleHash: '###';
 SubstraitScalarTest: 'SUBSTRAIT_SCALAR_TEST';
 SubstraitAggregateTest: 'SUBSTRAIT_AGGREGATE_TEST';
+SubstraitTableTest: 'SUBSTRAIT_TABLE_TEST';
 SubstraitInclude: 'SUBSTRAIT_INCLUDE';
 SubstraitDependency: 'SUBSTRAIT_DEPENDENCY';
 

grammar/FuncTestCaseParser.g4

@@ -15,7 +15,7 @@ header
     ;
 
 version
-    : TripleHash (SubstraitScalarTest | SubstraitAggregateTest) Colon FormatVersion
+    : TripleHash (SubstraitScalarTest | SubstraitAggregateTest | SubstraitTableTest) Colon FormatVersion
     ;
 
 include
@@ -37,6 +37,7 @@ testCase
 testGroup
     : testGroupDescription? (testCase)+                          #scalarFuncTestGroup
     | testGroupDescription? (aggFuncTestCase)+                   #aggregateFuncTestGroup
+    | testGroupDescription? (tableFuncTestCase)+                 #tableFuncTestGroup
     ;
 
 arguments
@@ -76,6 +77,18 @@ aggFuncTestCase
     : aggFuncCall ( OBracket funcOptions CBracket )? Eq result
     ;
 
+tableFuncTestCase
+    : functionName=identifier OParen arguments CParen ( OBracket funcOptions CBracket )? Eq multiRowResult
+    ;
+
+multiRowResult
+    : OBracket (rowTuple (Comma rowTuple)*)? CBracket DoubleColon structType
+    ;
+
+rowTuple
+    : OParen (literal (Comma literal)*)? CParen
+    ;
+
 aggFuncCall
     : tableData funcName=identifier OParen qualifiedAggregateFuncArgs? CParen           #multiArgAggregateFuncCall
     | tableRows functName=identifier OParen aggregateFuncArgs? CParen                   #compactAggregateFuncCall
@@ -344,12 +357,16 @@ parameterizedType
     | precisionTimestampType
     | precisionTimestampTZType
     | funcType
+    | structType
 // TODO implement the rest of the parameterized types
-//  | Struct isnull='?'? Lt expr (Comma expr)* Gt #struct
 //  | NStruct isnull='?'? Lt Identifier expr (Comma Identifier expr)* Gt #nStruct
 //  | Map isnull='?'? Lt key=expr Comma value=expr Gt #map
   ;
 
+structType
+    : Struct isnull=QMark? OAngleBracket dataType (Comma dataType)* CAngleBracket
+  ;
+
 numericParameter
   : IntegerLiteral #integerLiteral
   ;

tests/baseline.json

@@ -6,12 +6,12 @@
     "num_aggregate_functions": 29,
     "num_scalar_functions": 166,
     "num_window_functions": 11,
-    "num_table_functions": 3,
+    "num_table_functions": 4,
     "num_function_overloads": 536
   },
   "coverage": {
-    "total_test_count": 1140,
+    "total_test_count": 1158,
     "num_function_variants": 536,
-    "num_covered_function_variants": 241
+    "num_covered_function_variants": 247
   }
 }

tests/cases/table_generic/explode_array.test

@@ -0,0 +1,20 @@
+### SUBSTRAIT_TABLE_TEST: v1.0
+### SUBSTRAIT_INCLUDE: '/extensions/functions_table_generic.yaml'
+
+# basic_int: Basic integer array explosion
+explode([1, 2, 3]::list<i32>) = [(1), (2), (3)]::struct<i32>
+
+# basic_string: Basic string array explosion
+explode(['a', 'b', 'c']::list<str>) = [('a'), ('b'), ('c')]::struct<str>
+
+# empty: Empty arrays produce zero rows
+explode([]::list<i32>) = []::struct<i32>
+
+# single_element: Single element arrays
+explode([42]::list<i32>) = [(42)]::struct<i32>
+
+# two_elements: Two element arrays
+explode([10, 20]::list<i32>) = [(10), (20)]::struct<i32>
+
+# large_array: Larger array
+explode([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]::list<i32>) = [(1), (2), (3), (4), (5), (6), (7), (8), (9), (10)]::struct<i32>

tests/cases/table_generic/posexplode.test

@@ -0,0 +1,17 @@
+### SUBSTRAIT_TABLE_TEST: v1.0
+### SUBSTRAIT_INCLUDE: '/extensions/functions_table_generic.yaml'
+
+# basic_int: Basic integer array with positions
+posexplode([10, 20, 30]::list<i32>) = [(0, 10), (1, 20), (2, 30)]::struct<i64,i32>
+
+# basic_string: Basic string array with positions
+posexplode(['a', 'b', 'c']::list<str>) = [(0, 'a'), (1, 'b'), (2, 'c')]::struct<i64,str>
+
+# empty: Empty arrays produce zero rows
+posexplode([]::list<i32>) = []::struct<i64,i32>
+
+# single_element: Single element arrays start at position 0
+posexplode([42]::list<i32>) = [(0, 42)]::struct<i64,i32>
+
+# two_elements: Two element arrays
+posexplode([100, 200]::list<i32>) = [(0, 100), (1, 200)]::struct<i64,i32>

tests/cases/table_generic/unnest.test

@@ -0,0 +1,23 @@
+### SUBSTRAIT_TABLE_TEST: v1.0
+### SUBSTRAIT_INCLUDE: '/extensions/functions_table_generic.yaml'
+
+# single_list_int: Single list unnest (same as explode)
+unnest([1, 2, 3]::list<i32>) = [(1), (2), (3)]::struct<i32>
+
+# single_list_string: Single list unnest with strings
+unnest(['a', 'b']::list<str>) = [('a'), ('b')]::struct<str>
+
+# single_list_empty: Empty list produces zero rows
+unnest([]::list<i32>) = []::struct<i32>
+
+# two_lists_same_length: Zip two lists of same length
+unnest([1, 2, 3]::list<i32>, [4, 5, 6]::list<i32>) = [(1, 4), (2, 5), (3, 6)]::struct<i32,i32>
+
+# two_lists_different_types: Zip lists of different types
+unnest([1, 2]::list<i32>, ['a', 'b']::list<str>) = [(1, 'a'), (2, 'b')]::struct<i32,str>
+
+# three_lists: Zip three lists
+unnest([1, 2]::list<i32>, [3, 4]::list<i32>, [5, 6]::list<i32>) = [(1, 3, 5), (2, 4, 6)]::struct<i32,i32,i32>
+
+# four_lists: Zip four lists
+unnest([1, 2]::list<i32>, [3, 4]::list<i32>, [5, 6]::list<i32>, [7, 8]::list<i32>) = [(1, 3, 5, 7), (2, 4, 6, 8)]::struct<i32,i32,i32,i32>