Merge 9a5ffe2 into c935722

Author: igormunkin

ydb/library/yql/public/udf/arrow/block_item.h

@@ -26,6 +26,11 @@ class TBlockItem {
     template <typename T, typename = std::enable_if_t<TPrimitiveDataType<T>::Result>>
     inline explicit TBlockItem(T value);
 
+    inline explicit TBlockItem(bool value) {
+        Raw.Simple.bool_ = value ? 1 : 0;
+        Raw.Simple.Meta = static_cast<ui8>(EMarkers::Present);
+    }
+
     inline explicit TBlockItem(TStringRef value) {
         Raw.String.Value = value.Data();
         Raw.String.Size = value.Size();
@@ -109,6 +114,10 @@ class TBlockItem {
                 #define FIELD(type) type type##_;
                 PRIMITIVE_VALUE_TYPES(FIELD);
                 #undef FIELD
+                // According to the YQL <-> arrow type mapping convention,
+                // boolean values are processed as 8-bit unsigned integer
+                // with either 0 or 1 as a condition payload.
+                ui8 bool_;
                 ui64 Count;
             };
             union {
@@ -167,6 +176,9 @@ UDF_ASSERT_TYPE_SIZE(TBlockItem, 16);
 PRIMITIVE_VALUE_TYPES(VALUE_AS)
 PRIMITIVE_VALUE_TYPES(VALUE_GET)
 PRIMITIVE_VALUE_TYPES(VALUE_CONSTR)
+// XXX: TBlockItem constructor with <bool> parameter is implemented above.
+VALUE_AS(bool)
+VALUE_GET(bool)
 
 #undef VALUE_AS
 #undef VALUE_GET

ydb/library/yql/udfs/common/string/string_udf.cpp

@@ -174,19 +174,27 @@ namespace {
     XX(AsciiToUpper, to_upper)        \
     XX(AsciiToTitle, to_title)
 
+// NOTE: The functions below are marked as deprecated, so block implementation
+// is not required for them. Hence, STROKA_FIND_UDF provides only the scalar
+// one at the moment.
 #define STROKA_FIND_UDF_MAP(XX) \
-    XX(Contains, Contains)      \
     XX(StartsWith, StartsWith)  \
     XX(EndsWith, EndsWith)      \
     XX(HasPrefix, StartsWith)   \
     XX(HasSuffix, EndsWith)
 
+// NOTE: The functions below are marked as deprecated, so block implementation
+// is not required for them. Hence, STRING_TWO_ARGS_UDF provides only the
+// scalar one at the moment.
 #define STRING_TWO_ARGS_UDF_MAP(XX)                    \
     XX(StartsWithIgnoreCase, AsciiHasPrefixIgnoreCase) \
     XX(EndsWithIgnoreCase, AsciiHasSuffixIgnoreCase)   \
     XX(HasPrefixIgnoreCase, AsciiHasPrefixIgnoreCase)  \
     XX(HasSuffixIgnoreCase, AsciiHasSuffixIgnoreCase)
 
+// NOTE: The functions below are marked as deprecated, so block implementation
+// is not required for them. Hence, STROKA_UDF provides only the scalar one at
+// the moment.
 #define STROKA_UDF_MAP(XX) \
     XX(Reverse, ReverseInPlace)
 
@@ -207,6 +215,30 @@ namespace {
         return valueBuilder->NewString(input);
     }
 
+    BEGIN_SIMPLE_STRICT_ARROW_UDF(TContains, bool(TOptional<char*>, char*)) {
+        Y_UNUSED(valueBuilder);
+        if (!args[0])
+            return TUnboxedValuePod(false);
+
+        const TString haystack(args[0].AsStringRef());
+        const TString needle(args[1].AsStringRef());
+        return TUnboxedValuePod(haystack.Contains(needle));
+    }
+
+    struct TContainsKernelExec : public TBinaryKernelExec<TContainsKernelExec> {
+        template <typename TSink>
+        static void Process(TBlockItem arg1, TBlockItem arg2, const TSink& sink) {
+            if (!arg1)
+                return sink(TBlockItem(false));
+
+            const TString haystack(arg1.AsStringRef());
+            const TString needle(arg2.AsStringRef());
+            sink(TBlockItem(haystack.Contains(needle)));
+        }
+    };
+
+    END_SIMPLE_ARROW_UDF(TContains, TContainsKernelExec::Do);
+
     SIMPLE_STRICT_UDF(TReplaceAll, char*(TAutoMap<char*>, char*, char*)) {
         if (TString result(args[0].AsStringRef()); SubstGlobal(result, args[1].AsStringRef(), args[2].AsStringRef()))
             return valueBuilder->NewString(result);
@@ -277,6 +309,8 @@ namespace {
         return args[0];
     }
 
+    // NOTE: String::Find is marked as deprecated, so block implementation is
+    // not required for them. Hence, only the scalar one is provided.
     SIMPLE_STRICT_UDF_WITH_OPTIONAL_ARGS(TFind, i64(TAutoMap<char*>, char*, TOptional<ui64>), 1) {
         Y_UNUSED(valueBuilder);
         const TString haystack(args[0].AsStringRef());
@@ -285,6 +319,9 @@ namespace {
         return TUnboxedValuePod(haystack.find(needle, pos));
     }
 
+    // NOTE: String::ReverseFind is marked as deprecated, so block
+    // implementation is not required for them. Hence, only the scalar one is
+    // provided.
     SIMPLE_STRICT_UDF_WITH_OPTIONAL_ARGS(TReverseFind, i64(TAutoMap<char*>, char*, TOptional<ui64>), 1) {
         Y_UNUSED(valueBuilder);
         const TString haystack(args[0].AsStringRef());
@@ -293,6 +330,8 @@ namespace {
         return TUnboxedValuePod(haystack.rfind(needle, pos));
     }
 
+    // NOTE: String::Substring is marked as deprecated, so block implementation
+    // is not required for them. Hence, only the scalar one is provided.
     SIMPLE_STRICT_UDF_WITH_OPTIONAL_ARGS(TSubstring, char*(TAutoMap<char*>, TOptional<ui64>, TOptional<ui64>), 1) {
         const TString input(args[0].AsStringRef());
         const ui64 from = args[1].GetOrDefault<ui64>(0);

ydb/library/yql/udfs/common/string/test/canondata/test.test_BlockFind_/results.txt

@@ -14,6 +14,13 @@
                                     "String"
                                 ]
                             ];
+                            [
+                                "contains";
+                                [
+                                    "DataType";
+                                    "Bool"
+                                ]
+                            ];
                             [
                                 "levenstein";
                                 [
@@ -27,22 +34,27 @@
                 "Data" = [
                     [
                         "fdsa";
+                        %false;
                         "3"
                     ];
                     [
                         "aswedfg";
+                        %true;
                         "5"
                     ];
                     [
                         "asdadsaasd";
+                        %true;
                         "8"
                     ];
                     [
                         "gdsfsassas";
+                        %true;
                         "8"
                     ];
                     [
                         "";
+                        %false;
                         "2"
                     ]
                 ]

ydb/library/yql/udfs/common/string/test/cases/BlockFind.sql

@@ -2,5 +2,6 @@
 pragma UseBlocks;
 SELECT
     value,
+    String::Contains(value, "as") AS contains,
     String::LevensteinDistance(value, "as") AS levenstein
 FROM Input;