tpcds 10t plan shape for external table

regression-test/data/nereids_tpcds_shape_sf10t_orc/shape/query1.out

@@ -0,0 +1,37 @@
+-- This file is automatically generated. You should know what you did if you want to edit this
+-- !ds_shape_1 --
+PhysicalCteAnchor ( cteId=CTEId#0 )
+--PhysicalCteProducer ( cteId=CTEId#0 )
+----PhysicalProject
+------hashAgg[GLOBAL]
+--------PhysicalDistribute[DistributionSpecHash]
+----------hashAgg[LOCAL]
+------------PhysicalProject
+--------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_returns.sr_returned_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF0 d_date_sk->[sr_returned_date_sk]
+----------------PhysicalProject
+------------------PhysicalOlapScan[store_returns] apply RFs: RF0
+----------------PhysicalProject
+------------------filter((date_dim.d_year = 2000))
+--------------------PhysicalOlapScan[date_dim]
+--PhysicalResultSink
+----PhysicalTopN[MERGE_SORT]
+------PhysicalDistribute[DistributionSpecGather]
+--------PhysicalTopN[LOCAL_SORT]
+----------PhysicalProject
+------------hashJoin[INNER_JOIN shuffleBucket] hashCondition=((ctr1.ctr_store_sk = ctr2.ctr_store_sk)) otherCondition=((cast(ctr_total_return as DOUBLE) > cast((avg(cast(ctr_total_return as DECIMALV3(38, 4))) * 1.2) as DOUBLE))) build RFs:RF3 ctr_store_sk->[ctr_store_sk,s_store_sk]
+--------------PhysicalProject
+----------------hashJoin[INNER_JOIN broadcast] hashCondition=((store.s_store_sk = ctr1.ctr_store_sk)) otherCondition=() build RFs:RF2 s_store_sk->[ctr_store_sk]
+------------------PhysicalProject
+--------------------hashJoin[INNER_JOIN shuffle] hashCondition=((ctr1.ctr_customer_sk = customer.c_customer_sk)) otherCondition=() build RFs:RF1 c_customer_sk->[ctr_customer_sk]
+----------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF1 RF2 RF3
+----------------------PhysicalProject
+------------------------PhysicalOlapScan[customer]
+------------------PhysicalProject
+--------------------filter((store.s_state = 'TN'))
+----------------------PhysicalOlapScan[store] apply RFs: RF3
+--------------hashAgg[GLOBAL]
+----------------PhysicalDistribute[DistributionSpecHash]
+------------------hashAgg[LOCAL]
+--------------------PhysicalDistribute[DistributionSpecExecutionAny]
+----------------------PhysicalCteConsumer ( cteId=CTEId#0 )
+

regression-test/data/nereids_tpcds_shape_sf10t_orc/shape/query10.out

@@ -0,0 +1,47 @@
+-- This file is automatically generated. You should know what you did if you want to edit this
+-- !ds_shape_10 --
+PhysicalResultSink
+--PhysicalTopN[MERGE_SORT]
+----PhysicalDistribute[DistributionSpecGather]
+------PhysicalTopN[LOCAL_SORT]
+--------PhysicalProject
+----------hashAgg[GLOBAL]
+------------PhysicalDistribute[DistributionSpecHash]
+--------------hashAgg[LOCAL]
+----------------PhysicalProject
+------------------filter((ifnull($c$1, FALSE) OR ifnull($c$2, FALSE)))
+--------------------hashJoin[LEFT_SEMI_JOIN bucketShuffle] hashCondition=((c.c_customer_sk = catalog_sales.cs_ship_customer_sk)) otherCondition=()
+----------------------hashJoin[LEFT_SEMI_JOIN shuffle] hashCondition=((c.c_customer_sk = web_sales.ws_bill_customer_sk)) otherCondition=()
+------------------------PhysicalProject
+--------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((customer_demographics.cd_demo_sk = c.c_current_cdemo_sk)) otherCondition=() build RFs:RF5 cd_demo_sk->[c_current_cdemo_sk]
+----------------------------PhysicalProject
+------------------------------hashJoin[INNER_JOIN shuffle] hashCondition=((c.c_current_addr_sk = ca.ca_address_sk)) otherCondition=() build RFs:RF4 ca_address_sk->[c_current_addr_sk]
+--------------------------------hashJoin[RIGHT_SEMI_JOIN shuffle] hashCondition=((c.c_customer_sk = store_sales.ss_customer_sk)) otherCondition=() build RFs:RF3 c_customer_sk->[ss_customer_sk]
+----------------------------------PhysicalProject
+------------------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF2 d_date_sk->[ss_sold_date_sk]
+--------------------------------------PhysicalProject
+----------------------------------------PhysicalOlapScan[store_sales] apply RFs: RF2 RF3
+--------------------------------------PhysicalProject
+----------------------------------------filter((date_dim.d_moy <= 6) and (date_dim.d_moy >= 3) and (date_dim.d_year = 2001))
+------------------------------------------PhysicalOlapScan[date_dim]
+----------------------------------PhysicalProject
+------------------------------------PhysicalOlapScan[customer] apply RFs: RF4 RF5
+--------------------------------PhysicalProject
+----------------------------------filter(ca_county IN ('Campbell County', 'Cleburne County', 'Escambia County', 'Fairfield County', 'Washtenaw County'))
+------------------------------------PhysicalOlapScan[customer_address]
+----------------------------PhysicalOlapScan[customer_demographics]
+------------------------PhysicalProject
+--------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((web_sales.ws_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF1 d_date_sk->[ws_sold_date_sk]
+----------------------------PhysicalProject
+------------------------------PhysicalOlapScan[web_sales] apply RFs: RF1
+----------------------------PhysicalProject
+------------------------------filter((date_dim.d_moy <= 6) and (date_dim.d_moy >= 3) and (date_dim.d_year = 2001))
+--------------------------------PhysicalOlapScan[date_dim]
+----------------------PhysicalProject
+------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((catalog_sales.cs_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF0 d_date_sk->[cs_sold_date_sk]
+--------------------------PhysicalProject
+----------------------------PhysicalOlapScan[catalog_sales] apply RFs: RF0
+--------------------------PhysicalProject
+----------------------------filter((date_dim.d_moy <= 6) and (date_dim.d_moy >= 3) and (date_dim.d_year = 2001))
+------------------------------PhysicalOlapScan[date_dim]
+

regression-test/data/nereids_tpcds_shape_sf10t_orc/shape/query11.out

@@ -0,0 +1,57 @@
+-- This file is automatically generated. You should know what you did if you want to edit this
+-- !ds_shape_11 --
+PhysicalCteAnchor ( cteId=CTEId#0 )
+--PhysicalCteProducer ( cteId=CTEId#0 )
+----PhysicalUnion
+------PhysicalProject
+--------hashAgg[GLOBAL]
+----------PhysicalDistribute[DistributionSpecHash]
+------------hashAgg[LOCAL]
+--------------PhysicalProject
+----------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF1 d_date_sk->[ss_sold_date_sk]
+------------------PhysicalProject
+--------------------hashJoin[INNER_JOIN shuffle] hashCondition=((customer.c_customer_sk = store_sales.ss_customer_sk)) otherCondition=() build RFs:RF0 c_customer_sk->[ss_customer_sk]
+----------------------PhysicalProject
+------------------------PhysicalOlapScan[store_sales] apply RFs: RF0 RF1
+----------------------PhysicalProject
+------------------------PhysicalOlapScan[customer]
+------------------PhysicalProject
+--------------------filter(d_year IN (1998, 1999))
+----------------------PhysicalOlapScan[date_dim]
+------PhysicalProject
+--------hashAgg[GLOBAL]
+----------PhysicalDistribute[DistributionSpecHash]
+------------hashAgg[LOCAL]
+--------------PhysicalProject
+----------------hashJoin[INNER_JOIN broadcast] hashCondition=((web_sales.ws_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF3 d_date_sk->[ws_sold_date_sk]
+------------------PhysicalProject
+--------------------hashJoin[INNER_JOIN shuffle] hashCondition=((customer.c_customer_sk = web_sales.ws_bill_customer_sk)) otherCondition=() build RFs:RF2 c_customer_sk->[ws_bill_customer_sk]
+----------------------PhysicalProject
+------------------------PhysicalOlapScan[web_sales] apply RFs: RF2 RF3
+----------------------PhysicalProject
+------------------------PhysicalOlapScan[customer]
+------------------PhysicalProject
+--------------------filter(d_year IN (1998, 1999))
+----------------------PhysicalOlapScan[date_dim]
+--PhysicalResultSink
+----PhysicalTopN[MERGE_SORT]
+------PhysicalDistribute[DistributionSpecGather]
+--------PhysicalTopN[LOCAL_SORT]
+----------PhysicalProject
+------------hashJoin[INNER_JOIN shuffleBucket] hashCondition=((t_s_firstyear.customer_id = t_w_secyear.customer_id)) otherCondition=((if((year_total > 0.00), (cast(year_total as DECIMALV3(38, 8)) / year_total), 0.000000) > if((year_total > 0.00), (cast(year_total as DECIMALV3(38, 8)) / year_total), 0.000000))) build RFs:RF6 customer_id->[customer_id]
+--------------PhysicalProject
+----------------filter((t_w_secyear.dyear = 1999) and (t_w_secyear.sale_type = 'w'))
+------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF6
+--------------PhysicalProject
+----------------hashJoin[INNER_JOIN bucketShuffle] hashCondition=((t_s_firstyear.customer_id = t_w_firstyear.customer_id)) otherCondition=() build RFs:RF5 customer_id->[customer_id,customer_id]
+------------------hashJoin[INNER_JOIN shuffle] hashCondition=((t_s_secyear.customer_id = t_s_firstyear.customer_id)) otherCondition=() build RFs:RF4 customer_id->[customer_id]
+--------------------PhysicalProject
+----------------------filter((t_s_secyear.dyear = 1999) and (t_s_secyear.sale_type = 's'))
+------------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF4 RF5
+--------------------PhysicalProject
+----------------------filter((t_s_firstyear.dyear = 1998) and (t_s_firstyear.sale_type = 's') and (t_s_firstyear.year_total > 0.00))
+------------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF5
+------------------PhysicalProject
+--------------------filter((t_w_firstyear.dyear = 1998) and (t_w_firstyear.sale_type = 'w') and (t_w_firstyear.year_total > 0.00))
+----------------------PhysicalCteConsumer ( cteId=CTEId#0 )
+

regression-test/data/nereids_tpcds_shape_sf10t_orc/shape/query12.out

@@ -0,0 +1,26 @@
+-- This file is automatically generated. You should know what you did if you want to edit this
+-- !ds_shape_12 --
+PhysicalResultSink
+--PhysicalTopN[MERGE_SORT]
+----PhysicalDistribute[DistributionSpecGather]
+------PhysicalTopN[LOCAL_SORT]
+--------PhysicalProject
+----------PhysicalWindow
+------------PhysicalQuickSort[LOCAL_SORT]
+--------------PhysicalDistribute[DistributionSpecHash]
+----------------hashAgg[GLOBAL]
+------------------PhysicalDistribute[DistributionSpecHash]
+--------------------hashAgg[LOCAL]
+----------------------PhysicalProject
+------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((web_sales.ws_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF1 d_date_sk->[ws_sold_date_sk]
+--------------------------PhysicalProject
+----------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((web_sales.ws_item_sk = item.i_item_sk)) otherCondition=() build RFs:RF0 i_item_sk->[ws_item_sk]
+------------------------------PhysicalProject
+--------------------------------PhysicalOlapScan[web_sales] apply RFs: RF0 RF1
+------------------------------PhysicalProject
+--------------------------------filter(i_category IN ('Books', 'Electronics', 'Men'))
+----------------------------------PhysicalOlapScan[item]
+--------------------------PhysicalProject
+----------------------------filter((date_dim.d_date <= '2001-07-15') and (date_dim.d_date >= '2001-06-15'))
+------------------------------PhysicalOlapScan[date_dim]
+

regression-test/data/nereids_tpcds_shape_sf10t_orc/shape/query13.out

@@ -0,0 +1,34 @@
+-- This file is automatically generated. You should know what you did if you want to edit this
+-- !ds_shape_13 --
+PhysicalResultSink
+--hashAgg[GLOBAL]
+----PhysicalDistribute[DistributionSpecGather]
+------hashAgg[LOCAL]
+--------PhysicalProject
+----------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF4 d_date_sk->[ss_sold_date_sk]
+------------PhysicalProject
+--------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_addr_sk = customer_address.ca_address_sk)) otherCondition=((((ca_state IN ('IL', 'TN', 'TX') AND ((store_sales.ss_net_profit >= 100.00) AND (store_sales.ss_net_profit <= 200.00))) OR (ca_state IN ('ID', 'OH', 'WY') AND ((store_sales.ss_net_profit >= 150.00) AND (store_sales.ss_net_profit <= 300.00)))) OR (ca_state IN ('IA', 'MS', 'SC') AND ((store_sales.ss_net_profit >= 50.00) AND (store_sales.ss_net_profit <= 250.00))))) build RFs:RF3 ca_address_sk->[ss_addr_sk]
+----------------PhysicalProject
+------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_hdemo_sk = household_demographics.hd_demo_sk)) otherCondition=((((household_demographics.hd_dep_count = 1) AND ((((customer_demographics.cd_marital_status = 'D') AND (customer_demographics.cd_education_status = 'Primary')) AND ((store_sales.ss_sales_price >= 50.00) AND (store_sales.ss_sales_price <= 100.00))) OR (((customer_demographics.cd_marital_status = 'W') AND (customer_demographics.cd_education_status = '2 yr Degree')) AND ((store_sales.ss_sales_price >= 150.00) AND (store_sales.ss_sales_price <= 200.00))))) OR ((((customer_demographics.cd_marital_status = 'M') AND (customer_demographics.cd_education_status = 'College')) AND ((store_sales.ss_sales_price >= 100.00) AND (store_sales.ss_sales_price <= 150.00))) AND (household_demographics.hd_dep_count = 3)))) build RFs:RF2 hd_demo_sk->[ss_hdemo_sk]
+--------------------PhysicalProject
+----------------------hashJoin[INNER_JOIN broadcast] hashCondition=((customer_demographics.cd_demo_sk = store_sales.ss_cdemo_sk)) otherCondition=() build RFs:RF1 cd_demo_sk->[ss_cdemo_sk]
+------------------------PhysicalProject
+--------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store.s_store_sk = store_sales.ss_store_sk)) otherCondition=() build RFs:RF0 s_store_sk->[ss_store_sk]
+----------------------------PhysicalProject
+------------------------------filter((store_sales.ss_net_profit <= 300.00) and (store_sales.ss_net_profit >= 50.00) and (store_sales.ss_sales_price <= 200.00) and (store_sales.ss_sales_price >= 50.00))
+--------------------------------PhysicalOlapScan[store_sales] apply RFs: RF0 RF1 RF2 RF3 RF4
+----------------------------PhysicalProject
+------------------------------PhysicalOlapScan[store]
+------------------------PhysicalProject
+--------------------------filter(((((customer_demographics.cd_marital_status = 'M') AND (customer_demographics.cd_education_status = 'College')) OR ((customer_demographics.cd_marital_status = 'D') AND (customer_demographics.cd_education_status = 'Primary'))) OR ((customer_demographics.cd_marital_status = 'W') AND (customer_demographics.cd_education_status = '2 yr Degree'))))
+----------------------------PhysicalOlapScan[customer_demographics]
+--------------------PhysicalProject
+----------------------filter(hd_dep_count IN (1, 3))
+------------------------PhysicalOlapScan[household_demographics]
+----------------PhysicalProject
+------------------filter((customer_address.ca_country = 'United States') and ca_state IN ('IA', 'ID', 'IL', 'MS', 'OH', 'SC', 'TN', 'TX', 'WY'))
+--------------------PhysicalOlapScan[customer_address]
+------------PhysicalProject
+--------------filter((date_dim.d_year = 2001))
+----------------PhysicalOlapScan[date_dim]
+