From 82ade618286889c8d21a73baeb94143fcce3b86c Mon Sep 17 00:00:00 2001
From: wyb <wybb86@gmail.com>
Date: Tue, 6 Aug 2024 12:28:44 +0800
Subject: [PATCH] [Enhancement] Support json logical type in parquet files()
 (#49385)

Signed-off-by: wyb <wybb86@gmail.com>
(cherry picked from commit 914ca69eeb17bd275e1346c92b652cc37d23c7f0)
---
 be/src/exec/parquet_schema_builder.cpp        |   2 +
 test/sql/test_files/R/json_parquet            |  42 ++++++++++++++++++
 test/sql/test_files/T/json_parquet            |  18 ++++++++
 .../test_files/parquet_format/json.parquet    | Bin 0 -> 911 bytes
 4 files changed, 62 insertions(+)
 create mode 100644 test/sql/test_files/R/json_parquet
 create mode 100644 test/sql/test_files/T/json_parquet
 create mode 100644 test/sql/test_files/parquet_format/json.parquet

diff --git a/be/src/exec/parquet_schema_builder.cpp b/be/src/exec/parquet_schema_builder.cpp
index 0a2186e3d9062..8a3ac6bc1f9ea 100644
--- a/be/src/exec/parquet_schema_builder.cpp
+++ b/be/src/exec/parquet_schema_builder.cpp
@@ -88,6 +88,8 @@ static Status get_parquet_type_from_primitive(const ::parquet::schema::NodePtr&
             auto decimal_logical_type = std::dynamic_pointer_cast<const parquet::DecimalLogicalType>(logical_type);
             *type_desc = TypeDescriptor::create_decimalv3_type(TYPE_DECIMAL128, decimal_logical_type->precision(),
                                                                decimal_logical_type->scale());
+        } else if (logical_type->is_JSON()) {
+            *type_desc = TypeDescriptor::create_json_type();
         } else {
             *type_desc = TypeDescriptor::create_varbinary_type(TypeDescriptor::MAX_VARCHAR_LENGTH);
         }
diff --git a/test/sql/test_files/R/json_parquet b/test/sql/test_files/R/json_parquet
new file mode 100644
index 0000000000000..7b021596645d3
--- /dev/null
+++ b/test/sql/test_files/R/json_parquet
@@ -0,0 +1,42 @@
+-- name: test_json_parquet
+
+create database db_${uuid0};
+use db_${uuid0};
+
+shell: ossutil64 mkdir oss://${oss_bucket}/test_files/parquet_format/${uuid0} >/dev/null || echo "exit 0" >/dev/null
+
+shell: ossutil64 cp --force ./sql/test_files/parquet_format/json.parquet oss://${oss_bucket}/test_files/parquet_format/${uuid0}/ | grep -Pv "(average|elapsed)"
+-- result:
+0
+
+Succeed: Total num: 1, size: 911. OK num: 1(upload 1 files).
+-- !result
+
+select * from files('path' = 'oss://${oss_bucket}/test_files/parquet_format/${uuid0}/*', 'format' = 'parquet');
+-- result:
+1	["a", "b"]
+2	["c", "d"]
+-- !result
+
+create table t1 as select * from files('path' = 'oss://${oss_bucket}/test_files/parquet_format/${uuid0}/*', 'format' = 'parquet');
+
+desc t1;
+-- result:
+id	bigint	YES	true	None	
+json_data	json	YES	false	None	
+-- !result
+
+create table t2 (id bigint, array_col array<varchar(100)>);
+
+insert into t2 select * from files('path' = 'oss://${oss_bucket}/test_files/parquet_format/${uuid0}/*', 'format' = 'parquet');
+-- result:
+-- !result
+
+select * from t2;
+-- result:
+1	["a","b"]
+2	["c","d"]
+-- !result
+
+
+shell: ossutil64 rm -rf oss://${oss_bucket}/test_files/parquet_format/${uuid0}/ > /dev/null
diff --git a/test/sql/test_files/T/json_parquet b/test/sql/test_files/T/json_parquet
new file mode 100644
index 0000000000000..ad1faa49adb25
--- /dev/null
+++ b/test/sql/test_files/T/json_parquet
@@ -0,0 +1,18 @@
+-- name: test_json_parquet
+
+create database db_${uuid0};
+use db_${uuid0};
+
+shell: ossutil64 mkdir oss://${oss_bucket}/test_files/parquet_format/${uuid0} >/dev/null || echo "exit 0" >/dev/null
+shell: ossutil64 cp --force ./sql/test_files/parquet_format/json.parquet oss://${oss_bucket}/test_files/parquet_format/${uuid0}/ | grep -Pv "(average|elapsed)"
+
+select * from files('path' = 'oss://${oss_bucket}/test_files/parquet_format/${uuid0}/*', 'format' = 'parquet');
+
+create table t1 as select * from files('path' = 'oss://${oss_bucket}/test_files/parquet_format/${uuid0}/*', 'format' = 'parquet');
+desc t1;
+
+create table t2 (id bigint, array_col array<varchar(100)>);
+insert into t2 select * from files('path' = 'oss://${oss_bucket}/test_files/parquet_format/${uuid0}/*', 'format' = 'parquet');
+select * from t2;
+
+shell: ossutil64 rm -rf oss://${oss_bucket}/test_files/parquet_format/${uuid0}/ > /dev/null
diff --git a/test/sql/test_files/parquet_format/json.parquet b/test/sql/test_files/parquet_format/json.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a720e76ace1902dd8b561f1574e471fb2001da7f
GIT binary patch
literal 911
zcmaJ=O>5gg5FM$)xFy#W68In%V$$LmyC0zxU3%*=l-z<NR@!y4skBm8D-&Z3{X_XT
zIrhhNR<aYTO{2pypYOeyLFZq;&k2y2#1rD7F(5+#7XUoR!PkD)JN&Tff?WJl{swC<
zDWj8!rgYJHne{kb?AwVO#@_~Yb|VT#JV3@@4siolav(>?Zt(iHE>*&rVaH?ey>Jvx
z#>)}$LIR<!A{g|q{HqZBVH^zi%7Wous}B&4P#_SS8~=7j<fv_aLT)iu{t7Xcad_fn
z4#$@{X}{NhaC{HJf<eU;XLWGzZD>~JO`#I0xVRHF{SsYn=u(QDCyEsUebwYSj$8=C
z>=aLteqJ?2wNA{svOTm+Z$)P8syeF2L?Q14Kb6XeYoY0Offb#C=cuXV7UI9uyL_J-
zvClp}!`#KUljyJc=;MK2=IWL8DPNv;H>DQLlp2?_p}9;o(`&n^C96%v^v_1L+gb@-
z%Tn3QZ2EaRr>K{<w@G?y`hhW9q?x)FR(JmxwPD&|eukOahziYlYnIl0->!Xkm;=W7
w^uzR1w>fR3K^=#ri!B824&<IzYjaboXa^9rdZUxHkvWfMZy|gIaE`y?AM4HO4FCWD

literal 0
HcmV?d00001