Support StartsWith/EndsWith/StringContains arrow kernels

4 files changed, 78 insertions, 1 deletions

diff --git a/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp b/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp
index db7aedc5857..d6cf42c96c1 100644
--- a/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp
+++ b/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp
@@ -4654,8 +4654,11 @@ struct TBlockRules {
         {">",  { "Greater" } },
         {">=", { "GreaterOrEqual" } },
 
-        // size kernel
+        // string kernels
         {"Size", { "Size" } },
+        {"StartsWith", { "StartsWith" } },
+        {"EndsWith", { "EndsWith" } },
+        {"StringContains", { "StringContains" } },
     };
 
     TBlockRules()
diff --git a/ydb/library/yql/minikql/invoke_builtins/mkql_builtins.cpp b/ydb/library/yql/minikql/invoke_builtins/mkql_builtins.cpp
index 7a651a660c8..8bca6fdc39b 100644
--- a/ydb/library/yql/minikql/invoke_builtins/mkql_builtins.cpp
+++ b/ydb/library/yql/minikql/invoke_builtins/mkql_builtins.cpp
@@ -109,6 +109,7 @@ void RegisterDefaultOperations(IBuiltinFunctionRegistry& registry, TKernelFamily
     RegisterSubstring(registry);
     RegisterFind(registry);
     RegisterWith(registry);
+    RegisterWith(kernelFamilyMap);
     RegisterInversePresortString(registry);
     RegisterInverseString(registry);
     RegisterNanvl(registry);
diff --git a/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.cpp b/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.cpp
index 1cf606ab3d2..a291f463eb8 100644
--- a/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.cpp
+++ b/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.cpp
@@ -246,6 +246,48 @@ struct TStrGreaterOrEqualOp {
     }
 };
 
+struct TStrStartsWithOp {
+    static inline bool Do(std::string_view left, std::string_view right) {
+        return left.starts_with(right);
+    }
+
+    static inline bool DoWithEmptyLeft(size_t rightLen) {
+        return rightLen == 0;
+    }
+
+    static constexpr bool DoWithEmptyRight() {
+        return true;
+    }
+};
+
+struct TStrEndsWithOp {
+    static inline bool Do(std::string_view left, std::string_view right) {
+        return left.ends_with(right);
+    }
+
+    static inline bool DoWithEmptyLeft(size_t rightLen) {
+        return rightLen == 0;
+    }
+
+    static constexpr bool DoWithEmptyRight() {
+        return true;
+    }
+};
+
+struct TStrContainsOp {
+    static inline bool Do(std::string_view left, std::string_view right) {
+        return left.contains(right);
+    }
+
+    static inline bool DoWithEmptyLeft(size_t rightLen) {
+        return rightLen == 0;
+    }
+
+    static constexpr bool DoWithEmptyRight() {
+        return true;
+    }
+};
+
 template<typename TInput1, typename TInput2, typename TOp>
 void AddCompareStringKernel(TKernelFamilyBase& kernelFamily) {
     // ui8 type is used as bool replacement
@@ -325,6 +367,18 @@ void RegisterStringKernelSize(TKernelFamilyBase& kernelFamily) {
     AddSizeStringKernel<NUdf::TUtf8>(kernelFamily);
 }
 
+void RegisterStringKernelStartsWith(TKernelFamilyBase& kernelFamily) {
+    AddCompareStringKernels<TStrStartsWithOp>(kernelFamily);
+}
+
+void RegisterStringKernelEndsWith(TKernelFamilyBase& kernelFamily) {
+    AddCompareStringKernels<TStrEndsWithOp>(kernelFamily);
+}
+
+void RegisterStringKernelContains(TKernelFamilyBase& kernelFamily) {
+    AddCompareStringKernels<TStrContainsOp>(kernelFamily);
+}
+
 void RegisterSizeBuiltin(TKernelFamilyMap& kernelFamilyMap) {
     auto family = std::make_unique<TKernelFamilyBase>();
 
@@ -333,5 +387,19 @@ void RegisterSizeBuiltin(TKernelFamilyMap& kernelFamilyMap) {
     kernelFamilyMap["Size"] = std::move(family);
 }
 
+void RegisterWith(TKernelFamilyMap& kernelFamilyMap) {
+    auto family = std::make_unique<TKernelFamilyBase>();
+    RegisterStringKernelStartsWith(*family);
+    kernelFamilyMap["StartsWith"] = std::move(family);
+
+    family = std::make_unique<TKernelFamilyBase>();
+    RegisterStringKernelEndsWith(*family);
+    kernelFamilyMap["EndsWith"] = std::move(family);
+
+    family = std::make_unique<TKernelFamilyBase>();
+    RegisterStringKernelContains(*family);
+    kernelFamilyMap["StringContains"] = std::move(family);
+}
+
 }
 }
diff --git a/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.h b/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.h
index 77a5db506ed..f17f51d757e 100644
--- a/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.h
+++ b/ydb/library/yql/minikql/invoke_builtins/mkql_builtins_string_kernels.h
@@ -13,6 +13,11 @@ void RegisterStringKernelGreater(TKernelFamilyBase& kernelFamily);
 void RegisterStringKernelGreaterOrEqual(TKernelFamilyBase& kernelFamily);
 
 void RegisterStringKernelSize(TKernelFamilyBase& kernelFamily);
+void RegisterStringKernelStartsWith(TKernelFamilyBase& kernelFamily);
+void RegisterStringKernelEndsWith(TKernelFamilyBase& kernelFamily);
+void RegisterStringKernelContains(TKernelFamilyBase& kernelFamily);
+
 void RegisterSizeBuiltin(TKernelFamilyMap& kernelFamilyMap);
+void RegisterWith(TKernelFamilyMap& kernelFamilyMap);
 }
 }