Skip to content

Commit 33187c8

Browse files
prasadns14agirish
authored andcommitted
Revert "MD-2954: Fixed TPCDS100 hive datagen script (#337)" (#402)
This reverts commit 7c5ea70.
1 parent 44f5b00 commit 33187c8

File tree

6 files changed

+60281
-60317
lines changed

6 files changed

+60281
-60317
lines changed

framework/resources/Datasources/hive_storage/change_metadata2.ddl

Lines changed: 24 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -34,45 +34,45 @@ ALTER TABLE remove_columns2 REPLACE COLUMNS (
3434
ALTER TABLE modify_columntypes1 REPLACE COLUMNS (
3535
int_col string,
3636
bigint_col varchar(50),
37-
date_col date,
38-
time_col string,
37+
date_col int,
38+
time_col bigint,
3939
timestamp_col varchar(20),
40-
interval_col string,
41-
varchar_col string,
42-
float_col float,
43-
double_col double,
44-
bool_col boolean
40+
interval_col date,
41+
varchar_col int,
42+
float_col int,
43+
double_col float,
44+
bool_col int
4545
);
4646

4747
ALTER TABLE modify_columntypes2 REPLACE COLUMNS (
4848
int_col float,
49-
bigint_col bigint,
49+
bigint_col int,
5050
date_col varchar(15),
51-
time_col string,
52-
timestamp_col timestamp,
53-
interval_col string,
51+
time_col timestamp,
52+
timestamp_col date,
53+
interval_col boolean,
5454
varchar_col varchar(5),
5555
float_col string,
56-
double_col double,
57-
bool_col boolean
56+
double_col bigint,
57+
bool_col float
5858
);
5959

6060
alter table seq_modify_columntypes1 replace columns (
6161
l_orderkey float,
6262
l_partkey double,
6363
l_suppkey varchar(1),
6464
l_linenumber string,
65-
l_quantity double,
66-
l_extendedprice double,
67-
l_discount double,
65+
l_quantity bigint,
66+
l_extendedprice int,
67+
l_discount float,
6868
l_tax varchar(20),
69-
l_returnflag string,
70-
l_linestatus string,
71-
l_shipdate date,
69+
l_returnflag boolean,
70+
l_linestatus int,
71+
l_shipdate timestamp,
7272
l_commitdate string,
73-
l_receiptdate date,
73+
l_receiptdate int,
7474
l_shipinstruct varchar(2),
75-
l_shipmode string,
75+
l_shipmode boolean,
7676
l_comment double
7777
);
7878
ALTER TABLE dpp_compressed_mktevents ADD COLUMNS (newcol STRING);
@@ -87,10 +87,7 @@ ALTER TABLE changemetadata_orders_partitioned ADD PARTITION (year=1997) location
8787

8888
ALTER TABLE changemetadata_orders_partitioned1 replace columns (
8989
o_orderkey INT,
90-
o_custkey INT
91-
);
92-
93-
ALTER TABLE changemetadata_orders_partitioned1 add columns (
90+
o_custkey INT,
9491
o_orderdate STRING,
9592
o_orderpriority STRING,
9693
o_clerk STRING,
@@ -106,9 +103,9 @@ ALTER TABLE changemetadata_orders_partitioned2 replace columns (
106103
o_custkey STRING,
107104
o_orderstatus STRING,
108105
o_totalprice DOUBLE,
109-
o_orderdate STRING,
106+
o_orderdate BOOLEAN,
110107
o_orderpriority VARCHAR(1),
111108
o_clerk DOUBLE,
112109
o_shippriority INT
113110
);
114-
ALTER TABLE changemetadata_orders_partitioned2 DROP IF EXISTS PARTITION(year=1990);
111+
ALTER TABLE changemetadata_orders_partitioned2 DROP IF EXISTS PARTITION(year=1990)

framework/resources/Datasources/hive_storage/tpcds100/tpcds100_hive.ddl

Lines changed: 40 additions & 73 deletions
Original file line numberDiff line numberDiff line change
@@ -624,9 +624,7 @@ create external table tpcds100_parquet.customer
624624
c_email_address string,
625625
c_last_review_date string
626626
)
627-
STORED AS PARQUET
628-
LOCATION '/drill/testdata/tpcds_sf100/parquet/customer'
629-
TBLPROPERTIES ("parquet.page.size"="4096","parquet.block.size"="12288");
627+
STORED AS PARQUET;
630628

631629
drop table if exists tpcds100_parquet.customer_address;
632630
create external table tpcds100_parquet.customer_address
@@ -645,9 +643,7 @@ create external table tpcds100_parquet.customer_address
645643
ca_gmt_offset int,
646644
ca_location_type string
647645
)
648-
STORED AS PARQUET
649-
LOCATION '/drill/testdata/tpcds_sf100/parquet/customer_address'
650-
TBLPROPERTIES ("parquet.dictionary.page.size"="65536");
646+
STORED AS PARQUET;
651647

652648
drop table if exists tpcds100_parquet.customer_demographics;
653649
create external table tpcds100_parquet.customer_demographics
@@ -662,9 +658,7 @@ create external table tpcds100_parquet.customer_demographics
662658
cd_dep_employed_count int,
663659
cd_dep_college_count int
664660
)
665-
STORED AS PARQUET
666-
LOCATION '/drill/testdata/tpcds_sf100/parquet/customer_demographics'
667-
TBLPROPERTIES ("parquet.enable.dictionary"="false", "parquet.compression"="SNAPPY", "parquet.page.size"="4096","parquet.block.size"="12288");
661+
STORED AS PARQUET;
668662

669663
drop table if exists tpcds100_parquet.household_demographics;
670664
create external table tpcds100_parquet.household_demographics
@@ -675,17 +669,15 @@ create external table tpcds100_parquet.household_demographics
675669
hd_dep_count int,
676670
hd_vehicle_count int
677671
)
678-
STORED AS PARQUET
679-
LOCATION '/drill/testdata/tpcds_sf100/parquet/household_demographics'
680-
TBLPROPERTIES ("parquet.compression"="GZIP");
672+
STORED AS PARQUET;
681673

682674
drop table if exists tpcds100_parquet.item;
683-
create external table tpcds100_parquet.item
675+
create external table tpcds100_parquet.item
684676
(
685677
i_item_sk int,
686678
i_item_id string,
687-
i_rec_start_date date,
688-
i_rec_end_date date,
679+
i_rec_start_date string,
680+
i_rec_end_date string,
689681
i_item_desc string,
690682
i_current_price double,
691683
i_wholesale_cost double,
@@ -705,8 +697,7 @@ create external table tpcds100_parquet.item
705697
i_manager_id int,
706698
i_product_name string
707699
)
708-
STORED AS PARQUET
709-
LOCATION '/drill/testdata/tpcds_sf100/parquet/item';
700+
STORED AS PARQUET;
710701

711702
drop table if exists tpcds100_parquet.promotion;
712703
create external table tpcds100_parquet.promotion
@@ -731,8 +722,7 @@ create external table tpcds100_parquet.promotion
731722
p_purpose string,
732723
p_discount_active string
733724
)
734-
STORED AS PARQUET
735-
LOCATION '/drill/testdata/tpcds_sf100/parquet/promotion';
725+
STORED AS PARQUET;
736726

737727
drop table if exists tpcds100_parquet.time_dim;
738728
create external table tpcds100_parquet.time_dim (
@@ -747,9 +737,7 @@ create external table tpcds100_parquet.time_dim (
747737
t_sub_shift string,
748738
t_meal_time string
749739
)
750-
STORED AS PARQUET
751-
LOCATION '/drill/testdata/tpcds_sf100/parquet/time_dim'
752-
TBLPROPERTIES ("parquet.enable.dictionary"="false", "parquet.compression"="GZIP", "parquet.block.size"="32768", "parquet.page.size"="1024");
740+
STORED AS PARQUET;
753741

754742
drop table if exists tpcds100_parquet.date_dim;
755743
create external table tpcds100_parquet.date_dim (
@@ -782,12 +770,10 @@ create external table tpcds100_parquet.date_dim (
782770
d_current_quarter string,
783771
d_current_year string
784772
)
785-
STORED AS PARQUET
786-
LOCATION '/drill/testdata/tpcds_sf100/parquet/date_dim'
787-
TBLPROPERTIES ("parquet.enable.dictionary"="false", "parquet.page.size"="134217728");
773+
STORED AS PARQUET;
788774

789775
drop table if exists tpcds100_parquet.store;
790-
create external table tpcds100_parquet.store (
776+
create external table tpcds100_parquet.store (
791777
s_store_sk int,
792778
s_store_id string,
793779
s_rec_start_date string,
@@ -818,12 +804,10 @@ create external table tpcds100_parquet.store (
818804
s_gmt_offset double,
819805
s_tax_precentage double
820806
)
821-
STORED AS PARQUET
822-
LOCATION '/drill/testdata/tpcds_sf100/parquet/store'
823-
TBLPROPERTIES ("parquet.dictionary.page.size"="32768", "parquet.block.size"="32768", "parquet.page.size"="32768");
807+
STORED AS PARQUET;
824808

825809
drop table if exists tpcds100_parquet.store_sales;
826-
create external table tpcds100_parquet.store_sales (
810+
create external table tpcds100_parquet.store_sales (
827811
ss_sold_date_sk int,
828812
ss_sold_time_sk int,
829813
ss_item_sk int,
@@ -841,18 +825,17 @@ create external table tpcds100_parquet.store_sales (
841825
ss_ext_discount_amt double,
842826
ss_ext_sales_price double,
843827
ss_ext_wholesale_cost double,
844-
ss_ext_list_price double,
828+
ss_ext_list_price double,
845829
ss_ext_tax double,
846830
ss_coupon_amt double,
847831
ss_net_paid double,
848832
ss_net_paid_inc_tax double,
849833
ss_net_profit double
850834
)
851-
STORED AS PARQUET
852-
LOCATION '/drill/testdata/tpcds_sf100/parquet/store_sales';
835+
STORED AS PARQUET;
853836

854837
drop table if exists tpcds100_parquet.warehouse;
855-
create external table tpcds100_parquet.warehouse (
838+
create external table tpcds100_parquet.warehouse (
856839
w_warehouse_sk int,
857840
w_warehouse_id string,
858841
w_warehouse_name string,
@@ -868,8 +851,7 @@ create external table tpcds100_parquet.warehouse (
868851
w_country string,
869852
w_gmt_offset double
870853
)
871-
STORED AS PARQUET
872-
LOCATION '/drill/testdata/tpcds_sf100/parquet/warehouse';
854+
STORED AS PARQUET;
873855

874856
drop table if exists tpcds100_parquet.ship_mode;
875857
create external table tpcds100_parquet.ship_mode (
@@ -880,29 +862,26 @@ create external table tpcds100_parquet.ship_mode (
880862
sm_carrier string,
881863
sm_contract string
882864
)
883-
STORED AS PARQUET
884-
LOCATION '/drill/testdata/tpcds_sf100/parquet/ship_mode';
865+
STORED AS PARQUET;
885866

886867
drop table if exists tpcds100_parquet.reason;
887868
create external table tpcds100_parquet.reason (
888869
r_reason_sk int,
889870
r_reason_id string,
890871
r_reason_desc string
891872
)
892-
STORED AS PARQUET
893-
LOCATION '/drill/testdata/tpcds_sf100/parquet/reason';
873+
STORED AS PARQUET;
894874

895875
drop table if exists tpcds100_parquet.income_band;
896-
create external table tpcds100_parquet.income_band (
876+
create external table tpcds100_parquet.income_band (
897877
ib_income_band_sk int,
898878
ib_lower_bound int,
899879
ib_upper_bound int
900880
)
901-
STORED AS PARQUET
902-
LOCATION '/drill/testdata/tpcds_sf100/parquet/income_band';
881+
STORED AS PARQUET;
903882

904883
drop table if exists tpcds100_parquet.call_center;
905-
create external table tpcds100_parquet.call_center (
884+
create external table tpcds100_parquet.call_center (
906885
cc_call_center_sk int,
907886
cc_call_center_id string,
908887
cc_rec_start_date string,
@@ -935,11 +914,10 @@ create external table tpcds100_parquet.call_center (
935914
cc_gmt_offset double,
936915
cc_tax_percentage double
937916
)
938-
STORED AS PARQUET
939-
LOCATION '/drill/testdata/tpcds_sf100/parquet/call_center';
917+
STORED AS PARQUET;
940918

941919
drop table if exists tpcds100_parquet.web_site;
942-
create external table tpcds100_parquet.web_site (
920+
create external table tpcds100_parquet.web_site (
943921
web_site_sk int,
944922
web_site_id string,
945923
web_rec_start_date string,
@@ -967,11 +945,10 @@ create external table tpcds100_parquet.web_site (
967945
web_gmt_offset double,
968946
web_tax_percentage double
969947
)
970-
STORED AS PARQUET
971-
LOCATION '/drill/testdata/tpcds_sf100/parquet/web_site';
948+
STORED AS PARQUET;
972949

973950
drop table if exists tpcds100_parquet.store_returns;
974-
create external table tpcds100_parquet.store_returns (
951+
create external table tpcds100_parquet.store_returns (
975952
sr_returned_date_sk int,
976953
sr_return_time_sk int,
977954
sr_item_sk int,
@@ -993,11 +970,10 @@ create external table tpcds100_parquet.store_returns (
993970
sr_store_credit double,
994971
sr_net_loss double
995972
)
996-
STORED AS PARQUET
997-
LOCATION '/drill/testdata/tpcds_sf100/parquet/store_returns';
973+
STORED AS PARQUET;
998974

999975
drop table if exists tpcds100_parquet.web_page;
1000-
create external table tpcds100_parquet.web_page (
976+
create external table tpcds100_parquet.web_page (
1001977
wp_web_page_sk int,
1002978
wp_web_page_id string,
1003979
wp_rec_start_date string,
@@ -1013,8 +989,7 @@ create external table tpcds100_parquet.web_page (
1013989
wp_image_count int,
1014990
wp_max_ad_count int
1015991
)
1016-
STORED AS PARQUET
1017-
LOCATION '/drill/testdata/tpcds_sf100/parquet/web_page';
992+
STORED AS PARQUET;
1018993

1019994
drop table if exists tpcds100_parquet.catalog_page;
1020995
create external table tpcds100_parquet.catalog_page (
@@ -1028,23 +1003,19 @@ create external table tpcds100_parquet.catalog_page (
10281003
cp_description string,
10291004
cp_type string
10301005
)
1031-
STORED AS PARQUET
1032-
LOCATION '/drill/testdata/tpcds_sf100/parquet/catalog_page'
1033-
TBLPROPERTIES ("parquet.dictionary.page.size"="5242880", "parquet.page.size"="262144", "parquet.block.size"="262144");
1006+
STORED AS PARQUET;
10341007

10351008
drop table if exists tpcds100_parquet.inventory;
1036-
create external table tpcds100_parquet.inventory (
1009+
create external table tpcds100_parquet.inventory (
10371010
inv_date_sk int,
10381011
inv_item_sk int,
10391012
inv_warehouse_sk int,
10401013
inv_quantity_on_hand int
10411014
)
1042-
STORED AS PARQUET
1043-
LOCATION '/drill/testdata/tpcds_sf100/parquet/inventory'
1044-
TBLPROPERTIES ("parquet.block.size"="10485760");
1015+
STORED AS PARQUET;
10451016

10461017
drop table if exists tpcds100_parquet.catalog_returns;
1047-
create external table tpcds100_parquet.catalog_returns (
1018+
create external table tpcds100_parquet.catalog_returns (
10481019
cr_returned_date_sk int,
10491020
cr_returned_time_sk int,
10501021
cr_item_sk int,
@@ -1073,11 +1044,10 @@ create external table tpcds100_parquet.catalog_returns (
10731044
cr_store_credit double,
10741045
cr_net_loss double
10751046
)
1076-
STORED AS PARQUET
1077-
LOCATION '/drill/testdata/tpcds_sf100/parquet/catalog_returns';
1047+
STORED AS PARQUET;
10781048

10791049
drop table if exists tpcds100_parquet.web_returns;
1080-
create table tpcds100_parquet.web_returns (
1050+
create external table tpcds100_parquet.web_returns (
10811051
wr_returned_date_sk int,
10821052
wr_returned_time_sk int,
10831053
wr_item_sk int,
@@ -1103,8 +1073,7 @@ create table tpcds100_parquet.web_returns (
11031073
wr_account_credit double,
11041074
wr_net_loss double
11051075
)
1106-
STORED AS PARQUET
1107-
LOCATION '/drill/testdata/tpcds_sf100/parquet/web_returns';
1076+
STORED AS PARQUET;
11081077

11091078
drop table if exists tpcds100_parquet.web_sales;
11101079
create external table tpcds100_parquet.web_sales (
@@ -1143,11 +1112,10 @@ create external table tpcds100_parquet.web_sales (
11431112
ws_net_paid_inc_ship_tax double,
11441113
ws_net_profit double
11451114
)
1146-
STORED AS PARQUET
1147-
LOCATION '/drill/testdata/tpcds_sf100/parquet/web_sales';
1115+
STORED AS PARQUET;
11481116

11491117
drop table if exists tpcds100_parquet.catalog_sales;
1150-
create external table tpcds100_parquet.catalog_sales (
1118+
create external table tpcds100_parquet.catalog_sales (
11511119
cs_sold_date_sk int,
11521120
cs_sold_time_sk int,
11531121
cs_ship_date_sk int,
@@ -1183,5 +1151,4 @@ create external table tpcds100_parquet.catalog_sales (
11831151
cs_net_paid_inc_ship_tax double,
11841152
cs_net_profit double
11851153
)
1186-
STORED AS PARQUET
1187-
LOCATION '/drill/testdata/tpcds_sf100/parquet/catalog_sales';
1154+
STORED AS PARQUET;

0 commit comments

Comments
 (0)