Mark UDF strictness for json, json2, yson2

18 files changed, 43 insertions, 28 deletions

diff --git a/ydb/library/yql/minikql/dom/CMakeLists.darwin.txt b/ydb/library/yql/minikql/dom/CMakeLists.darwin.txt
index fb33f53eec..4d9af8a9e9 100644
--- a/ydb/library/yql/minikql/dom/CMakeLists.darwin.txt
+++ b/ydb/library/yql/minikql/dom/CMakeLists.darwin.txt
@@ -10,7 +10,7 @@
 add_library(yql-minikql-dom)
 target_compile_options(yql-minikql-dom PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(yql-minikql-dom PUBLIC
diff --git a/ydb/library/yql/minikql/dom/CMakeLists.linux-aarch64.txt b/ydb/library/yql/minikql/dom/CMakeLists.linux-aarch64.txt
index 7fad754cfa..5d15504310 100644
--- a/ydb/library/yql/minikql/dom/CMakeLists.linux-aarch64.txt
+++ b/ydb/library/yql/minikql/dom/CMakeLists.linux-aarch64.txt
@@ -10,7 +10,7 @@
 add_library(yql-minikql-dom)
 target_compile_options(yql-minikql-dom PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(yql-minikql-dom PUBLIC
diff --git a/ydb/library/yql/minikql/dom/CMakeLists.linux.txt b/ydb/library/yql/minikql/dom/CMakeLists.linux.txt
index 7fad754cfa..5d15504310 100644
--- a/ydb/library/yql/minikql/dom/CMakeLists.linux.txt
+++ b/ydb/library/yql/minikql/dom/CMakeLists.linux.txt
@@ -10,7 +10,7 @@
 add_library(yql-minikql-dom)
 target_compile_options(yql-minikql-dom PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(yql-minikql-dom PUBLIC
diff --git a/ydb/library/yql/udfs/common/json/CMakeLists.darwin.txt b/ydb/library/yql/udfs/common/json/CMakeLists.darwin.txt
index ff69837d58..260a11f271 100644
--- a/ydb/library/yql/udfs/common/json/CMakeLists.darwin.txt
+++ b/ydb/library/yql/udfs/common/json/CMakeLists.darwin.txt
@@ -19,7 +19,7 @@ target_link_libraries(json_udf INTERFACE
 add_global_library_for(json_udf.global json_udf)
 target_compile_options(json_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(json_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/json/CMakeLists.linux-aarch64.txt b/ydb/library/yql/udfs/common/json/CMakeLists.linux-aarch64.txt
index 4c0c37d600..84be2050d9 100644
--- a/ydb/library/yql/udfs/common/json/CMakeLists.linux-aarch64.txt
+++ b/ydb/library/yql/udfs/common/json/CMakeLists.linux-aarch64.txt
@@ -20,7 +20,7 @@ target_link_libraries(json_udf INTERFACE
 add_global_library_for(json_udf.global json_udf)
 target_compile_options(json_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(json_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/json/CMakeLists.linux.txt b/ydb/library/yql/udfs/common/json/CMakeLists.linux.txt
index 4c0c37d600..84be2050d9 100644
--- a/ydb/library/yql/udfs/common/json/CMakeLists.linux.txt
+++ b/ydb/library/yql/udfs/common/json/CMakeLists.linux.txt
@@ -20,7 +20,7 @@ target_link_libraries(json_udf INTERFACE
 add_global_library_for(json_udf.global json_udf)
 target_compile_options(json_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(json_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/json/json_udf.cpp b/ydb/library/yql/udfs/common/json/json_udf.cpp
index e994e589bd..49ffec496c 100644
--- a/ydb/library/yql/udfs/common/json/json_udf.cpp
+++ b/ydb/library/yql/udfs/common/json/json_udf.cpp
@@ -104,6 +104,8 @@ namespace {
                 if (!typesOnly) {
                     builder.Implementation(new TGetField);
                 }
+
+                builder.IsStrict();
                 return true;
             } else {
                 return false;
diff --git a/ydb/library/yql/udfs/common/json2/CMakeLists.darwin.txt b/ydb/library/yql/udfs/common/json2/CMakeLists.darwin.txt
index 46e29176e6..a26132211d 100644
--- a/ydb/library/yql/udfs/common/json2/CMakeLists.darwin.txt
+++ b/ydb/library/yql/udfs/common/json2/CMakeLists.darwin.txt
@@ -21,7 +21,7 @@ target_link_libraries(json2_udf INTERFACE
 add_global_library_for(json2_udf.global json2_udf)
 target_compile_options(json2_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(json2_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/json2/CMakeLists.linux-aarch64.txt b/ydb/library/yql/udfs/common/json2/CMakeLists.linux-aarch64.txt
index 6127f5aac1..808af57b77 100644
--- a/ydb/library/yql/udfs/common/json2/CMakeLists.linux-aarch64.txt
+++ b/ydb/library/yql/udfs/common/json2/CMakeLists.linux-aarch64.txt
@@ -22,7 +22,7 @@ target_link_libraries(json2_udf INTERFACE
 add_global_library_for(json2_udf.global json2_udf)
 target_compile_options(json2_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(json2_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/json2/CMakeLists.linux.txt b/ydb/library/yql/udfs/common/json2/CMakeLists.linux.txt
index 6127f5aac1..808af57b77 100644
--- a/ydb/library/yql/udfs/common/json2/CMakeLists.linux.txt
+++ b/ydb/library/yql/udfs/common/json2/CMakeLists.linux.txt
@@ -22,7 +22,7 @@ target_link_libraries(json2_udf INTERFACE
 add_global_library_for(json2_udf.global json2_udf)
 target_compile_options(json2_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(json2_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/json2/as_json_node.h b/ydb/library/yql/udfs/common/json2/as_json_node.h
index 442e4a90ba..4e190f8eb7 100644
--- a/ydb/library/yql/udfs/common/json2/as_json_node.h
+++ b/ydb/library/yql/udfs/common/json2/as_json_node.h
@@ -43,6 +43,8 @@ namespace NJson2Udf {
             if (!typesOnly) {
                 builder.Implementation(new TAsJsonNode<TSource>(builder.GetSourcePosition()));
             }
+
+            builder.IsStrict();
             return true;
         }
 
diff --git a/ydb/library/yql/udfs/common/json2/sql_exists.h b/ydb/library/yql/udfs/common/json2/sql_exists.h
index 32d5b84897..78aac2ce57 100644
--- a/ydb/library/yql/udfs/common/json2/sql_exists.h
+++ b/ydb/library/yql/udfs/common/json2/sql_exists.h
@@ -67,6 +67,9 @@ namespace NJson2Udf {
             if (!typesOnly) {
                 builder.Implementation(new TSqlExists(builder.GetSourcePosition()));
             }
+            if constexpr (!ThrowException) {
+                builder.IsStrict();
+            }
             return true;
         }
 
diff --git a/ydb/library/yql/udfs/common/json2/sql_query.h b/ydb/library/yql/udfs/common/json2/sql_query.h
index 85ed288fec..5b2f6a8a79 100644
--- a/ydb/library/yql/udfs/common/json2/sql_query.h
+++ b/ydb/library/yql/udfs/common/json2/sql_query.h
@@ -60,6 +60,8 @@ namespace NJson2Udf {
                 5. Bool. True - throw on error, false - otherwise
                 6. Resource<JsonNode>?. Default value to return on error. Ignored if 4th argument is true
             */
+            // we can't mark TSqlQuery as strict due to runtime throw policy setting
+            // TODO: optimizer can mark SqlQuery as strict if 3th/5th arguments are literal booleans
             builder.Args()
                 ->Add(inputOptionalType)
                 .Add(jsonPathType)
diff --git a/ydb/library/yql/udfs/common/json2/sql_value.h b/ydb/library/yql/udfs/common/json2/sql_value.h
index 5960c0749e..319fd855e8 100644
--- a/ydb/library/yql/udfs/common/json2/sql_value.h
+++ b/ydb/library/yql/udfs/common/json2/sql_value.h
@@ -156,6 +156,8 @@ namespace NJson2Udf {
                 .Done()
                 .Returns(returnType);
 
+            builder.IsStrict();
+
             if (!typesOnly) {
                 builder.Implementation(new TSqlValue(builder.GetSourcePosition()));
             }
diff --git a/ydb/library/yql/udfs/common/yson2/CMakeLists.darwin.txt b/ydb/library/yql/udfs/common/yson2/CMakeLists.darwin.txt
index e4f115e005..2b0d613f77 100644
--- a/ydb/library/yql/udfs/common/yson2/CMakeLists.darwin.txt
+++ b/ydb/library/yql/udfs/common/yson2/CMakeLists.darwin.txt
@@ -21,7 +21,7 @@ target_link_libraries(yson2_udf INTERFACE
 add_global_library_for(yson2_udf.global yson2_udf)
 target_compile_options(yson2_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(yson2_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/yson2/CMakeLists.linux-aarch64.txt b/ydb/library/yql/udfs/common/yson2/CMakeLists.linux-aarch64.txt
index 74fb6300eb..05c8a4a027 100644
--- a/ydb/library/yql/udfs/common/yson2/CMakeLists.linux-aarch64.txt
+++ b/ydb/library/yql/udfs/common/yson2/CMakeLists.linux-aarch64.txt
@@ -22,7 +22,7 @@ target_link_libraries(yson2_udf INTERFACE
 add_global_library_for(yson2_udf.global yson2_udf)
 target_compile_options(yson2_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(yson2_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/yson2/CMakeLists.linux.txt b/ydb/library/yql/udfs/common/yson2/CMakeLists.linux.txt
index 74fb6300eb..05c8a4a027 100644
--- a/ydb/library/yql/udfs/common/yson2/CMakeLists.linux.txt
+++ b/ydb/library/yql/udfs/common/yson2/CMakeLists.linux.txt
@@ -22,7 +22,7 @@ target_link_libraries(yson2_udf INTERFACE
 add_global_library_for(yson2_udf.global yson2_udf)
 target_compile_options(yson2_udf.global PRIVATE
   -DUDF_ABI_VERSION_MAJOR=2
-  -DUDF_ABI_VERSION_MINOR=27
+  -DUDF_ABI_VERSION_MINOR=28
   -DUDF_ABI_VERSION_PATCH=0
 )
 target_link_libraries(yson2_udf.global PUBLIC
diff --git a/ydb/library/yql/udfs/common/yson2/yson2_udf.cpp b/ydb/library/yql/udfs/common/yson2/yson2_udf.cpp
index 1278ec565c..36a90f5a70 100644
--- a/ydb/library/yql/udfs/common/yson2/yson2_udf.cpp
+++ b/ydb/library/yql/udfs/common/yson2/yson2_udf.cpp
@@ -88,6 +88,7 @@ public:
                 builder.Implementation(new TOptions);
             }
 
+            builder.IsStrict();
             return true;
         } else {
             return false;
@@ -526,7 +527,7 @@ SIMPLE_UDF_OPTIONS(TConvertToStringDict, TStringDictType(TOptional<TNodeResource
         return (options.AutoConvert ? &ConvertToDictImpl<false, true, &ConvertToString<false, true, false>> : &ConvertToDictImpl<false, false, &ConvertToString<false, false, false>>)(args[0], valueBuilder, Pos_);
 }
 
-SIMPLE_UDF(TAttributes, TDictType(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TAttributes, TDictType(TAutoMap<TNodeResource>)) {
     const auto x = args[0];
     if (IsNodeType<ENodeType::Attr>(x)) {
         return x;
@@ -549,7 +550,7 @@ SIMPLE_UDF_OPTIONS(TGetLength, TOptional<ui64>(TAutoMap<TNodeResource>, TOptiona
         return (options.AutoConvert ? &GetLengthImpl<false, true> : &GetLengthImpl<false, false>)(args[0], valueBuilder, Pos_);
 }
 
-SIMPLE_UDF_OPTIONS(TLookup, TOptional<TNodeResource>(TAutoMap<TNodeResource>, char*, TOptional<TOptionsResource>), builder.OptionalArgs(1)) {
+SIMPLE_STRICT_UDF_OPTIONS(TLookup, TOptional<TNodeResource>(TAutoMap<TNodeResource>, char*, TOptional<TOptionsResource>), builder.OptionalArgs(1)) {
     return LookupImpl(args[0], args[1], valueBuilder, Pos_);
 }
 
@@ -655,15 +656,15 @@ SIMPLE_UDF_OPTIONS(TYPathDict, TOptional<TDictType>(TAutoMap<TNodeResource>, cha
         return (options.AutoConvert ? &YPathImpl<&ConvertToDictImpl<false, true>> : &YPathImpl<&ConvertToDictImpl<false, false>>)(args[0], args[1], valueBuilder, Pos_);
 }
 
-SIMPLE_UDF(TSerialize, TYson(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TSerialize, TYson(TAutoMap<TNodeResource>)) {
     return valueBuilder->NewString(SerializeYsonDomToBinary(args[0]));
 }
 
-SIMPLE_UDF(TSerializeText, TYson(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TSerializeText, TYson(TAutoMap<TNodeResource>)) {
     return valueBuilder->NewString(SerializeYsonDomToText(args[0]));
 }
 
-SIMPLE_UDF(TSerializePretty, TYson(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TSerializePretty, TYson(TAutoMap<TNodeResource>)) {
     return valueBuilder->NewString(SerializeYsonDomToPrettyText(args[0]));
 }
 
@@ -679,7 +680,7 @@ SIMPLE_UDF_OPTIONS(TSerializeJson, TOptional<TJson>(TAutoMap<TNodeResource>, TOp
     return {};
 }
 
-SIMPLE_UDF(TWithAttributes, TOptional<TNodeResource>(TAutoMap<TNodeResource>, TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TWithAttributes, TOptional<TNodeResource>(TAutoMap<TNodeResource>, TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     auto x = args[0];
     auto y = args[1];
@@ -724,52 +725,52 @@ TUnboxedValuePod IsTypeImpl(TUnboxedValuePod y) {
     return TUnboxedValuePod(IsNodeType<Type>(y));
 }
 
-SIMPLE_UDF(TIsString, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsString, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::String>(*args);
 }
 
-SIMPLE_UDF(TIsInt64, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsInt64, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::Int64>(*args);
 }
 
-SIMPLE_UDF(TIsUint64, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsUint64, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::Uint64>(*args);
 }
 
-SIMPLE_UDF(TIsBool, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsBool, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::Bool>(*args);
 }
 
-SIMPLE_UDF(TIsDouble, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsDouble, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::Double>(*args);
 }
 
-SIMPLE_UDF(TIsList, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsList, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::List>(*args);
 }
 
-SIMPLE_UDF(TIsDict, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsDict, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::Dict>(*args);
 }
 
-SIMPLE_UDF(TIsEntity, bool(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TIsEntity, bool(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return IsTypeImpl<ENodeType::Entity>(*args);
 }
 
-SIMPLE_UDF(TEquals, bool(TAutoMap<TNodeResource>, TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TEquals, bool(TAutoMap<TNodeResource>, TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return TUnboxedValuePod(EquateDoms(args[0], args[1]));
 }
 
-SIMPLE_UDF(TGetHash, ui64(TAutoMap<TNodeResource>)) {
+SIMPLE_STRICT_UDF(TGetHash, ui64(TAutoMap<TNodeResource>)) {
     Y_UNUSED(valueBuilder);
     return TUnboxedValuePod(HashDom(args[0]));
 }
@@ -914,6 +915,7 @@ public:
             if (!typesOnly) {
                 builder.Implementation(new TFrom(builder.GetSourcePosition(), typeHelper, inputType));
             }
+            builder.IsStrict();
             return true;
         } else {
             return false;
@@ -1062,6 +1064,7 @@ public:
             switch (typeId) {
                 case TDataType<TYJson>::Id:
                     builder.Args()->Add<TAutoMap<TYJson>>().Add(optionsType).Done().Returns(builder.Resource(NodeResourceName));
+                    builder.IsStrict();
                     break;
                 case TDataType<TUtf8>::Id:
                     builder.Args()->Add<TAutoMap<TUtf8>>().Add(optionsType).Done().Returns(builder.Optional()->Item(builder.Resource(NodeResourceName)).Build());
@@ -1131,6 +1134,7 @@ const TStringRef& TParse<TJson, true>::Name() {
 
 }
 
+// TODO: optimizer that marks UDFs as strict if Yson::Options(false as Strict) is given
 SIMPLE_MODULE(TYson2Module,
     TOptions,
     TParse<TYson>,