unit-test cases

internal/pkg/bifs/stats.go

@@ -356,6 +356,9 @@ func BIF_mean(collection *mlrval.Mlrval) *mlrval.Mlrval {
 		return value_if_not
 	}
 	n := BIF_count(collection)
+	if n.AcquireIntValue() == 0 {
+		return mlrval.VOID
+	}
 	sum := BIF_sum(collection)
 	return BIF_divide(sum, n)
 }
@@ -595,6 +598,8 @@ func bif_percentiles(
 		p, ok := ps[i].GetNumericToFloatValue()
 		if !ok {
 			outputs[i] = mlrval.ERROR.Copy()
+		} else if len(sorted_array) == 0{
+			outputs[i] = mlrval.VOID
 		} else {
 			if interpolate_linearly {
 				outputs[i] = GetPercentileLinearlyInterpolated(sorted_array, len(sorted_array), p)
@@ -619,5 +624,9 @@ func bif_percentiles(
 // ================================================================
 // TODO:
 // * mode & antimode ... doing strings ... :|
+// * mode ties are non-deterministic ... fix that ...
+// * olh: min/max recurse but other stats funcs do not
+// * behavior of moments w/ too-short data
+// * percentile panics? any panics?
 
 // * also: str contains ... but extend to arrays? or nah?

test/cases/dsl-stats/count/various/cmd

@@ -0,0 +1 @@
+mlr -n --ofmtf 6 --xtab put -f ${CASEDIR}/mlr

test/cases/dsl-stats/count/various/expout

@@ -0,0 +1,20 @@
+count_0                 (error)
+count_0_type            error
+count_null              (error)
+count_null_type         error
+count_empty_array       0
+count_empty_array_type  int
+count_array_1           1
+count_array_1_type      int
+count_array_3           3
+count_array_3_type      int
+count_array_nested      3
+count_array_nested_type int
+count_empty_map         0
+count_empty_map_type    int
+count_map_1             1
+count_map_1_type        int
+count_map_3             3
+count_map_3_type        int
+count_map_nested        3
+count_map_nested_type   int

test/cases/dsl-stats/count/various/mlr

@@ -0,0 +1,26 @@
+end {
+    outputs = {};
+
+    outputs["count_0"] = count(0);
+    outputs["count_null"] = count(null);
+    outputs["count_nonesuch"] = count(nonesuch);
+
+    outputs["count_empty_array"] = count([]);
+    outputs["count_array_1"] = count([7]);
+    outputs["count_array_3"] = count([7,8,9]);
+    outputs["count_array_nested"] = count([7,[80,90],9]);
+
+    outputs["count_empty_map"] = count({});
+    outputs["count_map_1"] = count({ "a" : 7} );
+    outputs["count_map_3"] = count({ "a" : 7, "b" : 8, "c" : 9 } );
+    outputs["count_map_nested"] = count({ "a" : 7, "b" : [80,90], "c" : 9 });
+
+    typed_outputs = {};
+
+    for (k, v in outputs) {
+        typed_outputs[k] = v;
+        typed_outputs[k."_type"] = typeof(v);
+    }
+
+    emit typed_outputs;
+}

test/cases/dsl-stats/distinct_count/various/cmd

@@ -0,0 +1 @@
+mlr -n --ofmtf 6 --xtab put -f ${CASEDIR}/mlr

test/cases/dsl-stats/distinct_count/various/expout

@@ -0,0 +1,32 @@
+distinct_count_0                 (error)
+distinct_count_0_type            error
+distinct_count_null              (error)
+distinct_count_null_type         error
+distinct_count_empty_array       0
+distinct_count_empty_array_type  int
+distinct_count_array_1           1
+distinct_count_array_1_type      int
+distinct_count_array_3a          3
+distinct_count_array_3a_type     int
+distinct_count_array_3b          2
+distinct_count_array_3b_type     int
+distinct_count_array_3c          1
+distinct_count_array_3c_type     int
+distinct_count_array_3d          1
+distinct_count_array_3d_type     int
+distinct_count_array_nested      2
+distinct_count_array_nested_type int
+distinct_count_empty_map         0
+distinct_count_empty_map_type    int
+distinct_count_map_1             1
+distinct_count_map_1_type        int
+distinct_count_map_3a            3
+distinct_count_map_3a_type       int
+distinct_count_map_3b            2
+distinct_count_map_3b_type       int
+distinct_count_map_3c            1
+distinct_count_map_3c_type       int
+distinct_count_map_3d            1
+distinct_count_map_3d_type       int
+distinct_count_map_nested        2
+distinct_count_map_nested_type   int

test/cases/dsl-stats/distinct_count/various/mlr

@@ -0,0 +1,32 @@
+end {
+    outputs = {};
+
+    outputs["distinct_count_0"] = distinct_count(0);
+    outputs["distinct_count_null"] = distinct_count(null);
+    outputs["distinct_count_nonesuch"] = distinct_count(nonesuch);
+
+    outputs["distinct_count_empty_array"] = distinct_count([]);
+    outputs["distinct_count_array_1"] = distinct_count([7]);
+    outputs["distinct_count_array_3a"] = distinct_count([7,8,9]);
+    outputs["distinct_count_array_3b"] = distinct_count([7,7,9]);
+    outputs["distinct_count_array_3c"] = distinct_count([7,7,7]);
+    outputs["distinct_count_array_3d"] = distinct_count([null,null,null]);
+    outputs["distinct_count_array_nested"] = distinct_count([7,[7],7]);
+
+    outputs["distinct_count_empty_map"] = distinct_count({});
+    outputs["distinct_count_map_1"] = distinct_count({ "a" : 7} );
+    outputs["distinct_count_map_3a"] = distinct_count({ "a" : 7, "b" : 8, "c" : 9 } );
+    outputs["distinct_count_map_3b"] = distinct_count({ "a" : 7, "b" : 7, "c" : 9 } );
+    outputs["distinct_count_map_3c"] = distinct_count({ "a" : 7, "b" : 7, "c" : 7 } );
+    outputs["distinct_count_map_3d"] = distinct_count({ "a" : null, "b" : null, "c" : null } );
+    outputs["distinct_count_map_nested"] = distinct_count({ "a" : 7, "b" : [7], "c" : 7 });
+
+    typed_outputs = {};
+
+    for (k, v in outputs) {
+        typed_outputs[k] = v;
+        typed_outputs[k."_type"] = typeof(v);
+    }
+
+    emit typed_outputs;
+}

test/cases/dsl-stats/mode/various/cmd

@@ -0,0 +1 @@
+mlr -n --ofmtf 6 --xtab put -f ${CASEDIR}/mlr

test/cases/dsl-stats/mode/various/expout

@@ -0,0 +1,24 @@
+mode_0                 (error)
+mode_0_type            error
+mode_null              (error)
+mode_null_type         error
+mode_empty_array       
+mode_empty_array_type  empty
+mode_array_1           7
+mode_array_1_type      string
+mode_array_3a          7
+mode_array_3a_type     string
+mode_array_3b          7
+mode_array_3b_type     string
+mode_array_nested      9
+mode_array_nested_type string
+mode_empty_map         
+mode_empty_map_type    empty
+mode_map_1             7
+mode_map_1_type        string
+mode_map_3a            7
+mode_map_3a_type       string
+mode_map_3b            7
+mode_map_3b_type       string
+mode_map_nested        9
+mode_map_nested_type   string

test/cases/dsl-stats/mode/various/mlr

@@ -0,0 +1,28 @@
+end {
+    outputs = {};
+
+    outputs["mode_0"] = mode(0);
+    outputs["mode_null"] = mode(null);
+    outputs["mode_nonesuch"] = mode(nonesuch);
+
+    outputs["mode_empty_array"] = mode([]);
+    outputs["mode_array_1"] = mode([7]);
+    outputs["mode_array_3a"] = mode([7,8,9]);
+    outputs["mode_array_3b"] = mode([7,8,7]);
+    outputs["mode_array_nested"] = mode([7,[8,8,8,8,8,8],9,9,9]);
+
+    outputs["mode_empty_map"] = mode({});
+    outputs["mode_map_1"] = mode({ "a" : 7} );
+    outputs["mode_map_3a"] = mode({ "a" : 7, "b" : 8, "c" : 9 } );
+    outputs["mode_map_3b"] = mode({ "a" : 7, "b" : 8, "c" : 7 } );
+    outputs["mode_map_nested"] = mode({ "a" : 7, "b" : [8,8,8,8,8,8], "c" : 9, "d": 9, "e": 9 });
+
+    typed_outputs = {};
+
+    for (k, v in outputs) {
+        typed_outputs[k] = v;
+        typed_outputs[k."_type"] = typeof(v);
+    }
+
+    emit typed_outputs;
+}

test/cases/dsl-stats/moments/numeric-000/cmd

@@ -0,0 +1 @@
+mlr --ofmtf 6 --ojson --from test/input/abixy head -n 0 then put -q -f test/input/test-moments.mlr

test/cases/dsl-stats/moments/numeric-000/expout

@@ -0,0 +1,26 @@
+[
+{
+  "a_count": 0,
+  "a_sum": 0,
+  "a_sum2": 0,
+  "a_sum3": 0,
+  "a_sum4": 0,
+  "a_mean": "",
+  "a_var": "",
+  "a_stddev": "",
+  "a_meaneb": "",
+  "a_skewness": "",
+  "a_kurtosis": "",
+  "m_count": 0,
+  "m_sum": 0,
+  "m_sum2": 0,
+  "m_sum3": 0,
+  "m_sum4": 0,
+  "m_mean": "",
+  "m_var": "",
+  "m_stddev": "",
+  "m_meaneb": "",
+  "m_skewness": "",
+  "m_kurtosis": ""
+}
+]

test/cases/dsl-stats/moments/numeric-001/cmd

@@ -0,0 +1 @@
+mlr --ofmtf 6 --ojson --from test/input/abixy head -n 1 then put -q -f test/input/test-moments.mlr

test/cases/dsl-stats/moments/numeric-001/expout

@@ -0,0 +1,26 @@
+[
+{
+  "a_count": 1,
+  "a_sum": 1,
+  "a_sum2": 1,
+  "a_sum3": 1,
+  "a_sum4": 1,
+  "a_mean": 1,
+  "a_var": "",
+  "a_stddev": "",
+  "a_meaneb": "",
+  "a_skewness": "",
+  "a_kurtosis": "",
+  "m_count": 1,
+  "m_sum": 1,
+  "m_sum2": 1,
+  "m_sum3": 1,
+  "m_sum4": 1,
+  "m_mean": 1,
+  "m_var": "",
+  "m_stddev": "",
+  "m_meaneb": "",
+  "m_skewness": "",
+  "m_kurtosis": ""
+}
+]

test/cases/dsl-stats/moments/numeric-002/cmd

@@ -0,0 +1 @@
+mlr --ofmtf 6 --ojson --from test/input/abixy head -n 2 then put -q -f test/input/test-moments.mlr

test/cases/dsl-stats/moments/numeric-002/expout

@@ -0,0 +1,26 @@
+[
+{
+  "a_count": 2,
+  "a_sum": 3,
+  "a_sum2": 5,
+  "a_sum3": 9,
+  "a_sum4": 17,
+  "a_mean": 1.500000,
+  "a_var": 0.500000,
+  "a_stddev": 0.707107,
+  "a_meaneb": 0.500000,
+  "a_skewness": 0.000000,
+  "a_kurtosis": -2.000000,
+  "m_count": 2,
+  "m_sum": 3,
+  "m_sum2": 5,
+  "m_sum3": 9,
+  "m_sum4": 17,
+  "m_mean": 1.500000,
+  "m_var": 0.500000,
+  "m_stddev": 0.707107,
+  "m_meaneb": 0.500000,
+  "m_skewness": 0.000000,
+  "m_kurtosis": -2.000000
+}
+]

test/cases/dsl-stats/moments/numeric-003/cmd

@@ -0,0 +1 @@
+mlr --ofmtf 6 --ojson --from test/input/abixy head -n 3 then put -q -f test/input/test-moments.mlr

test/cases/dsl-stats/moments/numeric-003/expout

@@ -0,0 +1,26 @@
+[
+{
+  "a_count": 3,
+  "a_sum": 6,
+  "a_sum2": 14,
+  "a_sum3": 36,
+  "a_sum4": 98,
+  "a_mean": 2,
+  "a_var": 1.000000,
+  "a_stddev": 1.000000,
+  "a_meaneb": 0.577350,
+  "a_skewness": 0.000000,
+  "a_kurtosis": -1.500000,
+  "m_count": 3,
+  "m_sum": 6,
+  "m_sum2": 14,
+  "m_sum3": 36,
+  "m_sum4": 98,
+  "m_mean": 2,
+  "m_var": 1.000000,
+  "m_stddev": 1.000000,
+  "m_meaneb": 0.577350,
+  "m_skewness": 0.000000,
+  "m_kurtosis": -1.500000
+}
+]

test/cases/dsl-stats/moments/numeric-004/cmd

@@ -0,0 +1 @@
+mlr --ofmtf 6 --ojson --from test/input/abixy head -n 4 then put -q -f test/input/test-moments.mlr

test/cases/dsl-stats/moments/numeric-004/expout

@@ -0,0 +1,26 @@
+[
+{
+  "a_count": 4,
+  "a_sum": 10,
+  "a_sum2": 30,
+  "a_sum3": 100,
+  "a_sum4": 354,
+  "a_mean": 2.500000,
+  "a_var": 1.666667,
+  "a_stddev": 1.290994,
+  "a_meaneb": 0.645497,
+  "a_skewness": 0.000000,
+  "a_kurtosis": -1.360000,
+  "m_count": 4,
+  "m_sum": 10,
+  "m_sum2": 30,
+  "m_sum3": 100,
+  "m_sum4": 354,
+  "m_mean": 2.500000,
+  "m_var": 1.666667,
+  "m_stddev": 1.290994,
+  "m_meaneb": 0.645497,
+  "m_skewness": 0.000000,
+  "m_kurtosis": -1.360000
+}
+]

test/cases/dsl-stats/moments/numeric-all/cmd

@@ -0,0 +1 @@
+mlr --ofmtf 6 --ojson --from test/input/abixy put -q -f test/input/test-moments.mlr

test/cases/dsl-stats/moments/numeric-all/expout

@@ -0,0 +1,26 @@
+[
+{
+  "a_count": 10,
+  "a_sum": 55,
+  "a_sum2": 385,
+  "a_sum3": 3025,
+  "a_sum4": 25333,
+  "a_mean": 5.500000,
+  "a_var": 9.166667,
+  "a_stddev": 3.027650,
+  "a_meaneb": 0.957427,
+  "a_skewness": 0.000000,
+  "a_kurtosis": -1.224242,
+  "m_count": 10,
+  "m_sum": 55,
+  "m_sum2": 385,
+  "m_sum3": 3025,
+  "m_sum4": 25333,
+  "m_mean": 5.500000,
+  "m_var": 9.166667,
+  "m_stddev": 3.027650,
+  "m_meaneb": 0.957427,
+  "m_skewness": 0.000000,
+  "m_kurtosis": -1.224242
+}
+]

test/cases/dsl-stats/null_count/various/cmd

@@ -0,0 +1 @@
+mlr -n --ofmtf 6 --xtab put -f ${CASEDIR}/mlr

test/cases/dsl-stats/null_count/various/expout

@@ -0,0 +1,20 @@
+null_count_0                (error)
+null_count_0_type           error
+null_count_null             (error)
+null_count_null_type        error
+null_count_empty_array      0
+null_count_empty_array_type int
+null_count_array_1          0
+null_count_array_1_type     int
+null_count_array_2          0
+null_count_array_2_type     int
+null_count_array_3          2
+null_count_array_3_type     int
+null_count_empty_map        0
+null_count_empty_map_type   int
+null_count_map_1            0
+null_count_map_1_type       int
+null_count_map_2            0
+null_count_map_2_type       int
+null_count_map_3            2
+null_count_map_3_type       int