YQL-19884: Add Ascii{Starts,Ends}WithIgnoreCase functions to String UDF

commit_hash:4b86982498876ec14632c0a018a940c3393bb5d6

1 files changed, 31 insertions, 5 deletions

diff --git a/yql/essentials/udfs/common/string/string_udf.cpp b/yql/essentials/udfs/common/string/string_udf.cpp
index 161c51c060b..feb95dced93 100644
--- a/yql/essentials/udfs/common/string/string_udf.cpp
+++ b/yql/essentials/udfs/common/string/string_udf.cpp
@@ -153,8 +153,25 @@ SIMPLE_STRICT_UDF_OPTIONS(TReverse, TOptional<char*>(TOptional<char*>),
         }                                                              \
     }
 
-#define STRING_TWO_ARGS_UDF(udfName, function)                                 \
-    BEGIN_SIMPLE_STRICT_ARROW_UDF(T##udfName, bool(TOptional<char*>, char*)) { \
+#define STRING_TWO_ARGS_UDF_DEPRECATED_2025_02(udfName, function)        \
+    SIMPLE_STRICT_UDF_OPTIONS(T##udfName, bool(TOptional<char*>, char*), \
+        builder.SetMaxLangVer(NYql::MakeLangVersion(2025, 1)))           \
+    {                                                                    \
+        Y_UNUSED(valueBuilder);                                          \
+        if (args[0]) {                                                   \
+            const TString haystack(args[0].AsStringRef());               \
+            const TString needle(args[1].AsStringRef());                 \
+            return TUnboxedValuePod(function(haystack, needle));         \
+        } else {                                                         \
+            return TUnboxedValuePod(false);                              \
+        }                                                                \
+    }
+
+#define STRING_ASCII_CMP_IGNORE_CASE_UDF(udfName, function)                    \
+    BEGIN_SIMPLE_STRICT_ARROW_UDF_OPTIONS(T##udfName,                          \
+        bool(TOptional<char*>, char*),                                         \
+        builder.SetMinLangVer(NYql::MakeLangVersion(2025, 2)))                 \
+    {                                                                          \
         Y_UNUSED(valueBuilder);                                                \
         if (args[0]) {                                                         \
             const TString haystack(args[0].AsStringRef());                     \
@@ -380,12 +397,19 @@ SIMPLE_STRICT_UDF_OPTIONS(TReverse, TOptional<char*>(TOptional<char*>),
     XX(HasPrefix, StartsWith)   \
     XX(HasSuffix, EndsWith)
 
-#define STRING_TWO_ARGS_UDF_MAP(XX)                    \
+// NOTE: The functions below are marked as deprecated, so block implementation
+// is not required for them. Hence, STRING_TWO_ARGS_UDF_DEPRECATED_2025_02
+// provides only the scalar one at the moment.
+#define STRING_TWO_ARGS_UDF_MAP_DEPRECATED_2025_02(XX) \
     XX(StartsWithIgnoreCase, AsciiHasPrefixIgnoreCase) \
     XX(EndsWithIgnoreCase, AsciiHasSuffixIgnoreCase)   \
     XX(HasPrefixIgnoreCase, AsciiHasPrefixIgnoreCase)  \
     XX(HasSuffixIgnoreCase, AsciiHasSuffixIgnoreCase)
 
+#define STRING_ASCII_CMP_IGNORE_CASE_UDF_MAP(XX)            \
+    XX(AsciiStartsWithIgnoreCase, AsciiHasPrefixIgnoreCase) \
+    XX(AsciiEndsWithIgnoreCase, 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.
@@ -898,7 +922,8 @@ SIMPLE_STRICT_UDF_OPTIONS(TReverse, TOptional<char*>(TOptional<char*>),
     STROKA_CASE_UDF_MAP(STROKA_CASE_UDF)
     STROKA_ASCII_CASE_UDF_MAP(STROKA_ASCII_CASE_UDF)
     STROKA_FIND_UDF_MAP(STROKA_FIND_UDF)
-    STRING_TWO_ARGS_UDF_MAP(STRING_TWO_ARGS_UDF)
+    STRING_TWO_ARGS_UDF_MAP_DEPRECATED_2025_02(STRING_TWO_ARGS_UDF_DEPRECATED_2025_02)
+    STRING_ASCII_CMP_IGNORE_CASE_UDF_MAP(STRING_ASCII_CMP_IGNORE_CASE_UDF)
     IS_ASCII_UDF_MAP(IS_ASCII_UDF)
 
     static constexpr ui64 padLim = 1000000;
@@ -914,7 +939,8 @@ SIMPLE_STRICT_UDF_OPTIONS(TReverse, TOptional<char*>(TOptional<char*>),
         STROKA_CASE_UDF_MAP(STRING_REGISTER_UDF)
         STROKA_ASCII_CASE_UDF_MAP(STRING_REGISTER_UDF)
         STROKA_FIND_UDF_MAP(STRING_REGISTER_UDF)
-        STRING_TWO_ARGS_UDF_MAP(STRING_REGISTER_UDF)
+        STRING_TWO_ARGS_UDF_MAP_DEPRECATED_2025_02(STRING_REGISTER_UDF)
+        STRING_ASCII_CMP_IGNORE_CASE_UDF_MAP(STRING_REGISTER_UDF)
         IS_ASCII_UDF_MAP(STRING_REGISTER_UDF)
         STRING_STREAM_PAD_FORMATTER_UDF_MAP(STRING_REGISTER_UDF)
         STRING_STREAM_NUM_FORMATTER_UDF_MAP(STRING_REGISTER_UDF)