Merge pull request #68883 from ClickHouse/backport/24.6/68323

Backport #68323 to 24.6: Fix small value DateTime64 constant folding in nested subquery for remote

src/Analyzer/ConstantNode.cpp

@@ -11,6 +11,7 @@
 #include <IO/Operators.h>
 
 #include <DataTypes/FieldToDataType.h>
+#include <DataTypes/DataTypeDateTime64.h>
 
 #include <Parsers/ASTLiteral.h>
 #include <Parsers/ASTFunction.h>
@@ -162,6 +163,7 @@ QueryTreeNodePtr ConstantNode::cloneImpl() const
 ASTPtr ConstantNode::toASTImpl(const ConvertToASTOptions & options) const
 {
     const auto & constant_value_literal = constant_value->getValue();
+    const auto & constant_value_type = constant_value->getType();
     auto constant_value_ast = std::make_shared<ASTLiteral>(constant_value_literal);
 
     if (!options.add_cast_for_constants)
@@ -169,7 +171,26 @@ ASTPtr ConstantNode::toASTImpl(const ConvertToASTOptions & options) const
 
     if (requiresCastCall())
     {
-        auto constant_type_name_ast = std::make_shared<ASTLiteral>(constant_value->getType()->getName());
+        /** Value for DateTime64 is Decimal64, which is serialized as a string literal.
+          * If we serialize it as is, DateTime64 would be parsed from that string literal, which can be incorrect.
+          * For example, DateTime64 cannot be parsed from the short value, like '1', while it's a valid Decimal64 value.
+          * It could also lead to ambiguous parsing because we don't know if the string literal represents a date or a Decimal64 literal.
+          * For this reason, we use a string literal representing a date instead of a Decimal64 literal.
+          */
+        const auto & constant_value_end_type = removeNullable(constant_value_type); /// if Nullable
+        if (WhichDataType(constant_value_end_type->getTypeId()).isDateTime64())
+        {
+            const auto * date_time_type = typeid_cast<const DataTypeDateTime64 *>(constant_value_end_type.get());
+            DecimalField<Decimal64> decimal_value;
+            if (constant_value_literal.tryGet<DecimalField<Decimal64>>(decimal_value))
+            {
+                WriteBufferFromOwnString ostr;
+                writeDateTimeText(decimal_value.getValue(), date_time_type->getScale(), ostr, date_time_type->getTimeZone());
+                constant_value_ast = std::make_shared<ASTLiteral>(ostr.str());
+            }
+        }
+
+        auto constant_type_name_ast = std::make_shared<ASTLiteral>(constant_value_type->getName());
         return makeASTFunction("_CAST", std::move(constant_value_ast), std::move(constant_type_name_ast));
     }
 

tests/queries/0_stateless/03222_datetime64_small_value_const.reference

@@ -0,0 +1,18 @@
+0	1970-01-01 00:00:00.000
+0	1970-01-01 00:00:05.000
+0	1970-01-01 00:45:25.456789
+0	1970-01-01 00:53:25.456789123
+0	\N
+1	1970-01-01 00:00:00.000
+5	1970-01-01 00:00:00.000
+2	1970-01-01 00:00:02.456
+3	1970-01-01 00:00:04.811
+4	1970-01-01 00:10:05.000
+4	1970-01-01 00:10:05.000
+1	1970-01-01 00:00:00.000
+2	1970-01-01 00:00:02.456
+3	1970-01-01 00:00:04.811
+5	1970-01-01 00:00:00.000
+0
+0
+5

tests/queries/0_stateless/03222_datetime64_small_value_const.sql

@@ -0,0 +1,44 @@
+-- Tags: shard
+set session_timezone = 'UTC'; -- don't randomize the session timezone
+SET allow_experimental_analyzer = 1;
+
+select *, (select toDateTime64(0, 3)) from remote('127.0.0.1', system.one) settings prefer_localhost_replica=0;
+select *, (select toDateTime64(5, 3)) from remote('127.0.0.1', system.one) settings prefer_localhost_replica=0;
+select *, (select toDateTime64('1970-01-01 00:45:25.456789', 6)) from remote('127.0.0.1', system.one) settings prefer_localhost_replica=0;
+select *, (select toDateTime64('1970-01-01 00:53:25.456789123', 9)) from remote('127.0.0.1', system.one) settings prefer_localhost_replica=0;
+select *, (select toDateTime64(null,3)) from remote('127.0.0.1', system.one) settings prefer_localhost_replica=0;
+
+create database if not exists shard_0;
+create database if not exists shard_1;
+
+drop table if exists shard_0.dt64_03222;
+drop table if exists shard_1.dt64_03222;
+drop table if exists distr_03222_dt64;
+
+create table shard_0.dt64_03222(id UInt64, dt DateTime64(3)) engine = MergeTree order by id;
+create table shard_1.dt64_03222(id UInt64, dt DateTime64(3)) engine = MergeTree order by id;
+create table distr_03222_dt64 (id UInt64, dt DateTime64(3)) engine = Distributed(test_cluster_two_shards_different_databases, '', dt64_03222);
+
+insert into shard_0.dt64_03222 values(1, toDateTime64('1970-01-01 00:00:00.000',3));
+insert into shard_0.dt64_03222 values(2, toDateTime64('1970-01-01 00:00:02.456',3));
+insert into shard_1.dt64_03222 values(3, toDateTime64('1970-01-01 00:00:04.811',3));
+insert into shard_1.dt64_03222 values(4, toDateTime64('1970-01-01 00:10:05',3));
+insert into shard_1.dt64_03222 values(5, toDateTime64(0,3));
+
+--Output : 1,5 2,3,4 4 1,2,3,5 0 0 5
+select id, dt from distr_03222_dt64 where dt = (select toDateTime64(0,3)) order by id;
+select id, dt from distr_03222_dt64 where dt > (select toDateTime64(0,3)) order by id;
+select id, dt from distr_03222_dt64 where dt > (select toDateTime64('1970-01-01 00:10:00.000',3)) order by id;
+select id, dt from distr_03222_dt64 where dt < (select toDateTime64(5,3)) order by id;
+
+select count(*) from distr_03222_dt64 where dt > (select toDateTime64('2024-07-20 00:00:00',3));
+select count(*) from distr_03222_dt64 where dt > (select now());
+select count(*) from distr_03222_dt64 where dt < (select toDateTime64('2004-07-20 00:00:00',3));
+
+
+drop table if exists shard_0.dt64_03222;
+drop table if exists shard_1.dt64_03222;
+drop table if exists distr_03222_dt64;
+
+drop database shard_0;
+drop database shard_1;