database/mariadb-columnstore-engine
1
0
Fork 0
mirror of https://github.com/mariadb-corporation/mariadb-columnstore-engine.git synced 2025-07-29 08:21:15 +03:00
Code Activity

fix(PrimProc): MCOL-5394 JSON_ARRAYAGG in MCS works differently than in InnoDB

Browse Source
This commit is contained in:
Aleksei Antipovskii
2025-03-27 03:38:42 +01:00
committed by Alexey Antipovsky
parent 21ebd1ac20
commit c618fa284d
4 changed files with 315 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

144
dbcon/joblist/groupconcat.cpp
View File

@ -498,9 +498,16 @@ void GroupConcator::outputRow(std::ostringstream& oss, const rowgroup::Row& row)
case CalpontSystemCatalog::INT:
case CalpontSystemCatalog::BIGINT:
{
int64_t intVal = row.getIntField(*i);
if (fIsJsonArrayAgg && row.isNullValue(*i))
{
oss << "null";
}
else
{
int64_t intVal = row.getIntField(*i);
oss << intVal;
oss << intVal;
}
break;
}
@ -508,7 +515,14 @@ void GroupConcator::outputRow(std::ostringstream& oss, const rowgroup::Row& row)
case CalpontSystemCatalog::DECIMAL:
case CalpontSystemCatalog::UDECIMAL:
{
oss << fixed << row.getDecimalField(*i);
if (fIsJsonArrayAgg && row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << fixed << row.getDecimalField(*i);
}
break;
}
@ -518,18 +532,25 @@ void GroupConcator::outputRow(std::ostringstream& oss, const rowgroup::Row& row)
case CalpontSystemCatalog::UINT:
case CalpontSystemCatalog::UBIGINT:
{
uint64_t uintVal = row.getUintField(*i);
int scale = (int)row.getScale(*i);
if (scale == 0)
if (fIsJsonArrayAgg && row.isNullValue(*i))
{
oss << uintVal;
oss << "null";
}
else
{
oss << fixed
<< datatypes::Decimal(datatypes::TSInt128((int128_t)uintVal), scale,
datatypes::INT128MAXPRECISION);
uint64_t uintVal = row.getUintField(*i);
int scale = (int)row.getScale(*i);
if (scale == 0)
{
oss << uintVal;
}
else
{
oss << fixed
<< datatypes::Decimal(datatypes::TSInt128((int128_t)uintVal), scale,
datatypes::INT128MAXPRECISION);
}
}
break;
@ -541,7 +562,8 @@ void GroupConcator::outputRow(std::ostringstream& oss, const rowgroup::Row& row)
{
if (fIsJsonArrayAgg)
{
auto maybeJson = row.getStringField(*i).safeString(""); // XXX: MULL??? it is not checked anywhere.
auto maybeJson =
row.getStringField(*i).safeString("null"); // XXX: MULL??? it is not checked anywhere.
const auto j = json::parse(maybeJson, nullptr, false);
if (j.is_discarded())
{
@ -562,56 +584,121 @@ void GroupConcator::outputRow(std::ostringstream& oss, const rowgroup::Row& row)
case CalpontSystemCatalog::DOUBLE:
case CalpontSystemCatalog::UDOUBLE:
{
oss << setprecision(15) << row.getDoubleField(*i);
if (fIsJsonArrayAgg && row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << setprecision(15) << row.getDoubleField(*i);
}
break;
}
case CalpontSystemCatalog::LONGDOUBLE:
{
oss << setprecision(15) << row.getLongDoubleField(*i);
if (fIsJsonArrayAgg && row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << setprecision(15) << row.getLongDoubleField(*i);
}
break;
}
case CalpontSystemCatalog::FLOAT:
case CalpontSystemCatalog::UFLOAT:
{
oss << row.getFloatField(*i);
if (fIsJsonArrayAgg && row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << row.getFloatField(*i);
}
break;
}
case CalpontSystemCatalog::DATE:
{
if (fIsJsonArrayAgg)
oss << std::quoted(DataConvert::dateToString(row.getUintField(*i)));
{
if (row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << std::quoted(DataConvert::dateToString(row.getUintField(*i)));
}
}
else
{
oss << DataConvert::dateToString(row.getUintField(*i));
}
break;
}
case CalpontSystemCatalog::DATETIME:
{
if (fIsJsonArrayAgg)
oss << std::quoted(DataConvert::datetimeToString(row.getUintField(*i)));
{
if (row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << std::quoted(DataConvert::datetimeToString(row.getUintField(*i)));
}
}
else
{
oss << DataConvert::datetimeToString(row.getUintField(*i));
}
break;
}
case CalpontSystemCatalog::TIMESTAMP:
{
if (fIsJsonArrayAgg)
oss << std::quoted(DataConvert::timestampToString(row.getUintField(*i), fTimeZone));
{
if (row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << std::quoted(DataConvert::timestampToString(row.getUintField(*i), fTimeZone));
}
}
else
{
oss << DataConvert::timestampToString(row.getUintField(*i), fTimeZone);
}
break;
}
case CalpontSystemCatalog::TIME:
{
if (fIsJsonArrayAgg)
oss << std::quoted(DataConvert::timeToString(row.getUintField(*i)));
{
if (row.isNullValue(*i))
{
oss << "null";
}
else
{
oss << std::quoted(DataConvert::timeToString(row.getUintField(*i)));
}
}
else
{
oss << DataConvert::timeToString(row.getUintField(*i));
}
break;
}
@ -621,7 +708,7 @@ void GroupConcator::outputRow(std::ostringstream& oss, const rowgroup::Row& row)
}
}
i++;
++i;
}
}
@ -832,10 +919,12 @@ class GroupConcatOrderByRow
ordering::CompareRule* fRule;
};
class GroupConcatOrderBy::SortingPQ : public priority_queue<GroupConcatOrderByRow, vector<GroupConcatOrderByRow>, less<GroupConcatOrderByRow>>
class GroupConcatOrderBy::SortingPQ
: public priority_queue<GroupConcatOrderByRow, vector<GroupConcatOrderByRow>, less<GroupConcatOrderByRow>>
{
public:
using BaseType = std::priority_queue<GroupConcatOrderByRow, vector<GroupConcatOrderByRow>, less<GroupConcatOrderByRow>>;
using BaseType =
std::priority_queue<GroupConcatOrderByRow, vector<GroupConcatOrderByRow>, less<GroupConcatOrderByRow>>;
using size_type = BaseType::size_type;
SortingPQ(size_type capacity) : BaseType()
@ -1071,9 +1160,10 @@ void GroupConcatOrderBy::createNewRGData()
rowgroup::RGDataSizeType GroupConcatOrderBy::getDataSize() const
{
return fMemSize
+ fOrderByQueue->capacity() * sizeof(GroupConcatOrderByRow)
+ (fDistinct ? fDistinctMap->size() : 0) * 32 /* TODO: speculative unordered_map memory consumption per item, replace it with counting allocator */;
return fMemSize + fOrderByQueue->capacity() * sizeof(GroupConcatOrderByRow) +
(fDistinct ? fDistinctMap->size() : 0) * 32 /* TODO: speculative unordered_map memory consumption per
item, replace it with counting allocator */
;
}
void GroupConcatOrderBy::processRow(const rowgroup::Row& row)
@ -1083,7 +1173,7 @@ void GroupConcatOrderBy::processRow(const rowgroup::Row& row)
return;
// this row is skipped if any concatenated column is null.
if (concatColIsNull(row))
if (!fIsJsonArrayAgg && concatColIsNull(row))
return;
// if the row count is less than the limit
@ -1356,7 +1446,7 @@ void GroupConcatNoOrder::initialize(const rowgroup::SP_GroupConcat& gcc)
void GroupConcatNoOrder::processRow(const rowgroup::Row& row)
{
// if the row count is less than the limit
if (fCurrentLength < fGroupConcatLen && concatColIsNull(row) == false)
if (fCurrentLength < fGroupConcatLen && (fIsJsonArrayAgg || concatColIsNull(row) == false))
{
copyRow(row, &fRow);

32
mysql-test/columnstore/basic/r/func_jsonarrayagg.result
View File

@ -73,47 +73,47 @@ JSON_VALID(JSON_ARRAYAGG(b))
1
SELECT JSON_ARRAYAGG(a), JSON_ARRAYAGG(b) FROM t1;
JSON_ARRAYAGG(a) JSON_ARRAYAGG(b)
[1,1,2,2,2,2,3,1,1,2,2,2,2,3] ["Hello","World","This","Will","Work","!","Hello","World","This","Will","Work","!"]
[1,1,2,2,2,2,3,1,1,2,2,2,2,3] ["Hello","World","This","Will","Work","!",null,"Hello","World","This","Will","Work","!",null]
SELECT JSON_ARRAYAGG(a), JSON_ARRAYAGG(b) FROM t1 GROUP BY a;
JSON_ARRAYAGG(a) JSON_ARRAYAGG(b)
[1,1,1,1] ["Hello","World","Hello","World"]
[2,2,2,2,2,2,2,2] ["This","Will","Work","!","This","Will","Work","!"]
[3,3]
[3,3] [null,null]
#
# DISTINCT and LIMIT
#
SELECT JSON_ARRAYAGG(b LIMIT 1) FROM t1;
JSON_ARRAYAGG(b LIMIT 1)
["Hello","World","This","Will","Work","!","Hello","World","This","Will","Work","!"]
["Hello","World","This","Will","Work","!",null,"Hello","World","This","Will","Work","!",null]
SELECT JSON_ARRAYAGG(b LIMIT 2) FROM t1;
JSON_ARRAYAGG(b LIMIT 2)
["Hello","World","This","Will","Work","!","Hello","World","This","Will","Work","!"]
["Hello","World","This","Will","Work","!",null,"Hello","World","This","Will","Work","!",null]
SELECT JSON_ARRAYAGG(b LIMIT 1) FROM t1 GROUP BY b;
JSON_ARRAYAGG(b LIMIT 1)
["!","!"]
["Hello","Hello"]
["This","This"]
["Will","Will"]
["Work","Work"]
["World","World"]
[null,null]
SELECT JSON_ARRAYAGG(b LIMIT 2) FROM t1 GROUP BY a;
JSON_ARRAYAGG(b LIMIT 2)
["Hello","World","Hello","World"]
["This","Will","Work","!","This","Will","Work","!"]
[null,null]
SELECT JSON_ARRAYAGG(DISTINCT a) FROM t1;
JSON_ARRAYAGG(DISTINCT a)
[3,2,1]
SELECT JSON_ARRAYAGG(DISTINCT b) FROM t1;
JSON_ARRAYAGG(DISTINCT b)
["Will","World","Work","!","This","Hello"]
["Will","World","Work",null,"!","This","Hello"]
SELECT JSON_ARRAYAGG(DISTINCT a LIMIT 2) FROM t1;
JSON_ARRAYAGG(DISTINCT a LIMIT 2)
[3,2,1]
SELECT JSON_ARRAYAGG(DISTINCT b LIMIT 2) FROM t1;
JSON_ARRAYAGG(DISTINCT b LIMIT 2)
["Will","World","Work","!","This","Hello"]
["Will","World","Work",null,"!","This","Hello"]
#
# JSON aggregation
#
@ -156,7 +156,7 @@ DROP TABLE t1;
CREATE TABLE t1 (a INT)ENGINE=COLUMNSTORE;
SELECT JSON_ARRAYAGG(a) FROM t1;
JSON_ARRAYAGG(a)
NULL
DROP TABLE t1;
#
#
@ -195,10 +195,10 @@ JSON_ARRAYAGG(DISTINCT a)
INSERT INTO t1 VALUES (NULL,NULL), (NULL,NULL);
SELECT JSON_ARRAYAGG(a) FROM t1;
JSON_ARRAYAGG(a)
[1,2,3,1,2,3]
[1,2,3,1,2,3,null,null]
SELECT JSON_ARRAYAGG(DISTINCT a) FROM t1;
JSON_ARRAYAGG(DISTINCT a)
[3,2,1]
[null,2,3,1]
DROP TABLE t1;
CREATE TABLE t1(a VARCHAR(10), b INT)ENGINE=COLUMNSTORE;
INSERT INTO t1 VALUES (1,1), (2,2), (3,3);
@ -212,10 +212,10 @@ JSON_ARRAYAGG(DISTINCT a)
INSERT INTO t1 VALUES (NULL,NULL), (NULL,NULL);
SELECT JSON_ARRAYAGG(a) FROM t1;
JSON_ARRAYAGG(a)
[1,2,3,1,2,3]
[1,2,3,1,2,3,null,null]
SELECT JSON_ARRAYAGG(DISTINCT a) FROM t1;
JSON_ARRAYAGG(DISTINCT a)
[3,2,1]
[null,2,3,1]
DROP TABLE t1;
#
#
@ -233,13 +233,13 @@ JSON_ARRAYAGG(a ORDER BY a ASC)
INSERT INTO t1 VALUES (NULL);
SELECT JSON_ARRAYAGG(a) FROM t1;
JSON_ARRAYAGG(a)
["red","blue"]
["red","blue",null]
SELECT JSON_ARRAYAGG(a ORDER BY a DESC) FROM t1;
JSON_ARRAYAGG(a ORDER BY a DESC)
["red","blue"]
["red","blue",null]
SELECT JSON_ARRAYAGG(a ORDER BY a ASC) FROM t1;
JSON_ARRAYAGG(a ORDER BY a ASC)
["blue","red"]
[null,"blue","red"]
DROP TABLE t1;
set group_concat_max_len=64;
create table t1 (a varchar(254))ENGINE=COLUMNSTORE;

100
mysql-test/columnstore/bugfixes/MCOL-5394-json_arrayagg.result Normal file
View File

@ -0,0 +1,100 @@
DROP DATABASE IF EXISTS mcol_5394 ;
CREATE DATABASE mcol_5394 ;
USE mcol_5394 ;
CREATE TABLE t1 (a int, k int, b VARCHAR(10)) engine=columnstore;
INSERT INTO t1 VALUES
(1, 1, "alfa"),
(1, 2, null),
(2, 3, "doi"),
(1, 4, "unu"),
(3, 5, "trei"),
(4, 6, null),
(4, 7, null),
(1, 8, "one");
SELECT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a;
a JSON_ARRAYAGG(b)
1 ["alfa",null,"unu","one"]
2 ["doi"]
3 ["trei"]
4 [null,null]
SELECT JSON_ARRAYAGG(b) FROM t1;
JSON_ARRAYAGG(b)
["alfa",null,"doi","unu","trei",null,null,"one"]
PREPARE p1 FROM "SELECT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a";
EXECUTE p1;
a JSON_ARRAYAGG(b)
1 ["alfa",null,"unu","one"]
2 ["doi"]
3 ["trei"]
4 [null,null]
EXECUTE p1;
a JSON_ARRAYAGG(b)
1 ["alfa",null,"unu","one"]
2 ["doi"]
3 ["trei"]
4 [null,null]
deallocate prepare p1;
PREPARE p3 FROM
"SELECT SQL_BUFFER_RESULT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a";
EXECUTE p3;
a JSON_ARRAYAGG(b)
1 ["alfa",null,"unu","one"]
2 ["doi"]
3 ["trei"]
4 [null,null]
EXECUTE p3;
a JSON_ARRAYAGG(b)
1 ["alfa",null,"unu","one"]
2 ["doi"]
3 ["trei"]
4 [null,null]
deallocate prepare p3;
PREPARE p4 FROM "SELECT JSON_ARRAYAGG(b) FROM t1";
EXECUTE p4;
JSON_ARRAYAGG(b)
["alfa",null,"doi","unu","trei",null,null,"one"]
EXECUTE p4;
JSON_ARRAYAGG(b)
["alfa",null,"doi","unu","trei",null,null,"one"]
deallocate prepare p4;
SELECT JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]') FROM t1;
JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]')
["alfa", null, "doi", "unu", "trei", null, null, "one", true, false]
PREPARE p1 FROM
"SELECT a, JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]') FROM t1 GROUP BY a";
EXECUTE p1;
a JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]')
1 ["alfa", null, "unu", "one", true, false]
2 ["doi", true, false]
3 ["trei", true, false]
4 [null, null, true, false]
EXECUTE p1;
a JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]')
1 ["alfa", null, "unu", "one", true, false]
2 ["doi", true, false]
3 ["trei", true, false]
4 [null, null, true, false]
deallocate prepare p1;
PREPARE p4 FROM
"SELECT JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]') FROM t1";
EXECUTE p4;
JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]')
["alfa", null, "doi", "unu", "trei", null, null, "one", true, false]
EXECUTE p4;
JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]')
["alfa", null, "doi", "unu", "trei", null, null, "one", true, false]
deallocate prepare p4;
SELECT a, JSON_ARRAYAGG(b) as jarray
FROM t1
GROUP BY a
HAVING jarray= JSON_ARRAY("trei");
a jarray
3 ["trei"]
TRUNCATE TABLE t1;
SELECT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a;
a JSON_ARRAYAGG(b)
SELECT JSON_ARRAYAGG(b) FROM t1;
JSON_ARRAYAGG(b)
NULL
DROP TABLE t1;
DROP DATABASE mcol_5394;

82
mysql-test/columnstore/bugfixes/MCOL-5394-json_arrayagg.test Normal file
View File

@ -0,0 +1,82 @@
--source ../include/have_columnstore.inc
--disable_warnings
DROP DATABASE IF EXISTS mcol_5394 ;
--enable_warnings
CREATE DATABASE mcol_5394 ;
USE mcol_5394 ;
CREATE TABLE t1 (a int, k int, b VARCHAR(10)) engine=columnstore;
INSERT INTO t1 VALUES
(1, 1, "alfa"),
(1, 2, null),
(2, 3, "doi"),
(1, 4, "unu"),
(3, 5, "trei"),
(4, 6, null),
(4, 7, null),
(1, 8, "one");
--sorted_result
SELECT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a;
--sorted_result
SELECT JSON_ARRAYAGG(b) FROM t1;
PREPARE p1 FROM "SELECT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a";
--sorted_result
EXECUTE p1;
--sorted_result
EXECUTE p1;
deallocate prepare p1;
PREPARE p3 FROM
"SELECT SQL_BUFFER_RESULT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a";
--sorted_result
EXECUTE p3;
--sorted_result
EXECUTE p3;
deallocate prepare p3;
PREPARE p4 FROM "SELECT JSON_ARRAYAGG(b) FROM t1";
--sorted_result
EXECUTE p4;
--sorted_result
EXECUTE p4;
deallocate prepare p4;
--sorted_result
SELECT JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]') FROM t1;
PREPARE p1 FROM
"SELECT a, JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]') FROM t1 GROUP BY a";
--sorted_result
EXECUTE p1;
--sorted_result
EXECUTE p1;
deallocate prepare p1;
PREPARE p4 FROM
"SELECT JSON_MERGE_PRESERVE(JSON_ARRAYAGG(b), '[true, false]') FROM t1";
--sorted_result
EXECUTE p4;
--sorted_result
EXECUTE p4;
deallocate prepare p4;
--sorted_result
SELECT a, JSON_ARRAYAGG(b) as jarray
FROM t1
GROUP BY a
HAVING jarray= JSON_ARRAY("trei");
TRUNCATE TABLE t1;
SELECT a, JSON_ARRAYAGG(b) FROM t1 GROUP BY a;
SELECT JSON_ARRAYAGG(b) FROM t1;
DROP TABLE t1;
DROP DATABASE mcol_5394;