From c86fad7cbd91d01feb2fe7180305fb50fe54efb4 Mon Sep 17 00:00:00 2001
From: Qi Chen <kaka11.chen@gmail.com>
Date: Thu, 26 Oct 2023 21:50:18 +0800
Subject: [PATCH] [Fix](orc-reader) Fix orc decimal128 scale issue. (#25977)

---
 be/src/apache-orc                             |   2 +-
 .../scripts/create_preinstalled_table.hql     |  19 ++++++++++++++++++
 .../orc_decimal_table/orc_decimal_table       | Bin 0 -> 1169 bytes
 .../external_table_p0/hive/test_hive_orc.out  |  18 +++++++++++++++++
 .../hive/test_hive_orc.groovy                 |  10 +++++++++
 5 files changed, 48 insertions(+), 1 deletion(-)
 create mode 100644 docker/thirdparties/docker-compose/hive/scripts/preinstalled_data/orc_table/orc_decimal_table/orc_decimal_table

diff --git a/be/src/apache-orc b/be/src/apache-orc
index a7c0af50f8..e52ba9094d 160000
--- a/be/src/apache-orc
+++ b/be/src/apache-orc
@@ -1 +1 @@
-Subproject commit a7c0af50f8ca8ff7cddaf8675473a037f8b13143
+Subproject commit e52ba9094dd8fa3e86e6043e42d6d561fe092d7c
diff --git a/docker/thirdparties/docker-compose/hive/scripts/create_preinstalled_table.hql b/docker/thirdparties/docker-compose/hive/scripts/create_preinstalled_table.hql
index d7e93bb74f..2fbdbbe6db 100644
--- a/docker/thirdparties/docker-compose/hive/scripts/create_preinstalled_table.hql
+++ b/docker/thirdparties/docker-compose/hive/scripts/create_preinstalled_table.hql
@@ -1743,5 +1743,24 @@ LOCATION
 
 msck repair table parquet_timestamp_nanos;
 
+CREATE TABLE `orc_decimal_table`(
+    id INT,
+    decimal_col1 DECIMAL(8, 4),
+    decimal_col2 DECIMAL(18, 6),
+    decimal_col3 DECIMAL(38, 12),
+    decimal_col4 DECIMAL(9, 0),
+    decimal_col5 DECIMAL(27, 9),
+    decimal_col6 DECIMAL(9, 0))
+ROW FORMAT SERDE
+  'org.apache.hadoop.hive.ql.io.orc.OrcSerde'
+STORED AS INPUTFORMAT
+  'org.apache.hadoop.hive.ql.io.orc.OrcInputFormat'
+OUTPUTFORMAT
+  'org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat'
+LOCATION
+  '/user/doris/preinstalled_data/orc_table/orc_decimal_table';
+
+msck repair table orc_decimal_table;
+
 
 show tables;
diff --git a/docker/thirdparties/docker-compose/hive/scripts/preinstalled_data/orc_table/orc_decimal_table/orc_decimal_table b/docker/thirdparties/docker-compose/hive/scripts/preinstalled_data/orc_table/orc_decimal_table/orc_decimal_table
new file mode 100644
index 0000000000000000000000000000000000000000..c63a5e8ea37f22b6c871730f59665ccb4b010f93
GIT binary patch
literal 1169
zcmeYdau#G@;9?VE;b0A5&}LxZ66RuNU|<m9;b0YF<6sitl8_5v@MB<jeCa__Dg&cp
zC104br=A<LqMx&Jvaqdy;?c~Mq`U>`Ie7^p2N)O%7#JRJdI(mT%onB{3sRW~QW<Pq
z%<Zk<%ztR2keQL;W)l-L)9nUE<`<03%ni&ob0jb@G%zqc_5y2Nn86peH~^%z6t0!0
zO)lohp*ucacTXHSbLqm3D|eWtCZ;7UNzch+a7vq&<|f9-!w`haonYHF{gj1`y`7Vd
znHMv#F_bYdJpS|)>{4;QFlB3yb(wJMES&j|O`2o~b18%I86KZAUwY2+I-Izo)W8U_
z=?PdjD_@whCP=rRH+QjduyC@nv!<T2cX4nsGq(cBW1I{O2UwVx<QW)_tY>_({}K0r
zHIKMKY*q#aCJrM8hU_0zb9N`5Tx%S!KlkpjI|jc^S`J=d5Mg5gDd34^V0f{0-@Lp3
z%wK$+f8@`V-3Q;lIW+O_k>498tIoN(?bybyKPNYA+O(Jvs)~W1fx(YOLV`hpL7ag>
zV%3&=r64Q;HjTlZfnin4oA0mOQM9i5({%RhrL)_Zp^{)DgajBEgk%^Pejodhaq;=q
zHwllVm6;)?>|$UjJ-v~u!GNbFakiP~N`+f5+0xky3&mn;{+Dl(i<Lce{QShEo6i`N
zfATPhzA<^Uh?755=1M_udi%HMApx^*g_IpR)wfsJugob?TW#C7yBjvHec#vm_JUSr
zhI;*XrV2)e)IMweCI^9*iyd;&63m_bmuDn6t6F}WcjU;WS;vkX+y5@VP+W;CuJGpG
z+4*S?Oj~Dm%;uP;<o7&jSJT^#opRhKTzjl%tT0&_Fnf~z%mp7;U5QilG)wR-O7lFn
z|Ju)h%K>HGXD;m6b?F-W8-=$_iIz5;yN$#CBpqJ(CcdeB(~l@#w&fd4)85}wydCd$
zG@4)U+??$0^aACRQ&oTeXRF<0({s4z_marIpaLcNn{I`^*(PB^t8PWFWs<$L=DrOR
zL#m&z_hAExcj33TObLv-Ftc-q+#S7XCi8PyKB~-15kJ_yqC}2O`2VTAe+Ko9|Bt97
zUp^!AX`lb6nu8mDuxPX=ew6Lm|L5~`%N6W1rgzG>2zCgIteK_}zmIqI_v>Fjf7qBO
z>muVKdqn1ltja5fwo)GV;tpkNZ<*Vbjm6f%Z*E#1e94!*K#4<ao|I$2F$0ZFA8#J2
z_PHX$R5a=54xebJ$Yi}KLTZK1s%o=}*6dgoBjuG^w{t@5`sr!h0j@J%M4w7^-=?Hx
zr8|3-x^A_>rkn9Icboj4?#E^Fdux$cW2dRvy}b&@{=HhT<Ez2ZUcVcUrtZ2rCEI!C
zpZpE}Zp+&fecoK!(RpaqHM<jTGiS{SytHV|Th-+|F1r`^<(YjizsJw9kV$}1qM<=R
ViHU(nV?Uz-d(#(YX8#~(NdWCv-RA%R

literal 0
HcmV?d00001

diff --git a/regression-test/data/external_table_p0/hive/test_hive_orc.out b/regression-test/data/external_table_p0/hive/test_hive_orc.out
index 4f1841c7d3..aff1b185cf 100644
--- a/regression-test/data/external_table_p0/hive/test_hive_orc.out
+++ b/regression-test/data/external_table_p0/hive/test_hive_orc.out
@@ -83,6 +83,24 @@ tablets	tinyint_col	179	182	182	187	183	181	177	183	177	187	183	202	202	186	528
 -- !only_partition_col --
 3600	3600
 
+-- !decimals1 --
+1	1.1234	12.123456	123.123456789876	12	1234.123456789	123
+2	1234.1234	123456789123.123456	12345678912345678912345678.123456789876	123456789	123456789123456789.123456780	987654321
+3	1234.0000	123456789123.000000	12345678912345678912345678.000000000000	123456789	123456789123456789.000000000	987654321
+4	0.0000	0.000000	123.123456789876	12	0E-9	123
+5	1.1234	12.123456	0E-12	0	1234.123456789	0
+
+-- !decimals2 --
+3	1234.0000	123456789123.000000	12345678912345678912345678.000000000000	123456789	123456789123456789.000000000	987654321
+
+-- !decimals3 --
+1	1.1234	12.123456	123.123456789876	12	1234.123456789	123
+2	1234.1234	123456789123.123456	12345678912345678912345678.123456789876	123456789	123456789123456789.123456780	987654321
+
+-- !decimals4 --
+4	0.0000	0.000000	123.123456789876	12	0E-9	123
+5	1.1234	12.123456	0E-12	0	1234.123456789	0
+
 -- !select_top50 --
 4	55	999742610	400899305488827731	false	6.5976813E8	7.8723304616937395E17	\N	base tennis pit vertical friday	2022-08-19T07:29:58	\N	tablets	smallint_col	2019-02-07	[7.53124931825377e+17]	["NbSSBtwzpxNSkkwga"]	tablets	smallint_col
 2	49	999613702	105493714032727452	\N	6.3322381E8	9.8642324410240179E17	Unveil bright recruit participate. Suspect impression camera mathematical revelation. Fault live2 elbow debt west hydrogen current.	how literary	2022-09-03T17:20:21	481707.1065	tablets	boolean_col	2020-01-12	[]	["HoMrAnn", "wteEFvIwoZsVpVQdscMb", NULL, "zcGFmv", "kGEBBckbMtX", "hrEtCGFdPWZK"]	tablets	boolean_col
diff --git a/regression-test/suites/external_table_p0/hive/test_hive_orc.groovy b/regression-test/suites/external_table_p0/hive/test_hive_orc.groovy
index e971d82340..b6bccfb286 100644
--- a/regression-test/suites/external_table_p0/hive/test_hive_orc.groovy
+++ b/regression-test/suites/external_table_p0/hive/test_hive_orc.groovy
@@ -66,6 +66,14 @@ suite("test_hive_orc", "all_types,p0,external,hive,external_docker,external_dock
         qt_only_partition_col """select count(p1_col), count(p2_col) from orc_all_types;"""
     }
 
+    // decimals
+    def decimals = {
+        qt_decimals1 """select * from orc_decimal_table order by id;"""
+        qt_decimals2 """select * from orc_decimal_table where id = 3 order by id;"""
+        qt_decimals3 """select * from orc_decimal_table where id < 3 order by id;"""
+        qt_decimals4 """select * from orc_decimal_table where id > 3 order by id;"""
+    }
+
     String enabled = context.config.otherConfigs.get("enableHiveTest")
     if (enabled != null && enabled.equalsIgnoreCase("true")) {
         try {
@@ -86,6 +94,7 @@ suite("test_hive_orc", "all_types,p0,external,hive,external_docker,external_dock
             search_in_int()
             search_mix()
             only_partition_col()
+            decimals()
 
             sql """drop catalog if exists ${catalog_name}"""
 
@@ -103,3 +112,4 @@ suite("test_hive_orc", "all_types,p0,external,hive,external_docker,external_dock
         }
     }
 }
+