should be it

external/duckdb/benchmark/h2oai/group/h2oai.benchmark.in

@@ -0,0 +1,24 @@
+# name: ${FILE_PATH}
+# description: ${DESCRIPTION}
+# group: [h2oai]
+
+require httpfs load_only
+
+name Q${QUERY_NUMBER_PADDED}
+group H2OAI
+
+cache h2oai.duckdb
+
+retry load 5
+
+load benchmark/h2oai/group/queries/load.sql
+
+run benchmark/h2oai/group/queries/q${QUERY_NUMBER_PADDED}.sql
+
+result_query ${RESULT_COLUMNS}
+${RESULT_QUERY}
+----
+${RESULT_ANSWER}
+
+cleanup
+DROP TABLE ans

external/duckdb/benchmark/h2oai/group/q01.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q01.benchmark
+# description: Run query 01 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=1
+QUERY_NUMBER_PADDED=01
+RESULT_COLUMNS=III
+RESULT_QUERY=SELECT COUNT(DISTINCT id1), SUM(v1), COUNT(*) FROM ans
+RESULT_ANSWER=95	28498857	96

external/duckdb/benchmark/h2oai/group/q02.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q02.benchmark
+# description: Run query 02 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=2
+QUERY_NUMBER_PADDED=02
+RESULT_COLUMNS=IIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id1), COUNT(DISTINCT id2), SUM(v1), COUNT(*) FROM ans
+RESULT_ANSWER=95	95	28498857	9216

external/duckdb/benchmark/h2oai/group/q03.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q03.benchmark
+# description: Run query 03 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=3
+QUERY_NUMBER_PADDED=03
+RESULT_COLUMNS=IIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id3), SUM(v1), SUM(v3), COUNT(*) FROM ans
+RESULT_ANSWER=95000	28498857	4749467.631946747	95001

external/duckdb/benchmark/h2oai/group/q04.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q04.benchmark
+# description: Run query 04 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=4
+QUERY_NUMBER_PADDED=04
+RESULT_COLUMNS=IIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id4), SUM(v1), SUM(v2), SUM(v3), COUNT(*) FROM ans
+RESULT_ANSWER=95	287.9894309270616821	767.8529216923457105	4799.873270453372	96

external/duckdb/benchmark/h2oai/group/q05.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q05.benchmark
+# description: Run query 05 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=5
+QUERY_NUMBER_PADDED=05
+RESULT_COLUMNS=IIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id6), SUM(v1), SUM(v2), SUM(v3), COUNT(*) FROM ans
+RESULT_ANSWER=95000	28498857	75988394	474969574.04777884	95001

external/duckdb/benchmark/h2oai/group/q06.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q06.benchmark
+# description: Run query 06 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=6
+QUERY_NUMBER_PADDED=06
+RESULT_COLUMNS=IIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id4), COUNT(DISTINCT id5), SUM(median_v3), SUM(sd_v3), COUNT(*) FROM ans
+RESULT_ANSWER=95	95	460771.2164439997	266006.9046221105	9216

external/duckdb/benchmark/h2oai/group/q07.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q07.benchmark
+# description: Run query 07 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=7
+QUERY_NUMBER_PADDED=07
+RESULT_COLUMNS=III
+RESULT_QUERY=SELECT COUNT(DISTINCT id3), SUM(range_v1_v2), COUNT(*) FROM ans
+RESULT_ANSWER=95000	379850	95001

external/duckdb/benchmark/h2oai/group/q08.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q08.benchmark
+# description: Run query 08 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=8
+QUERY_NUMBER_PADDED=08
+RESULT_COLUMNS=III
+RESULT_QUERY=SELECT COUNT(DISTINCT id6), SUM(largest2_v3), COUNT(*) FROM ans;
+RESULT_ANSWER=95000	18700554.77963206	190002

external/duckdb/benchmark/h2oai/group/q09.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q09.benchmark
+# description: Run query 09 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=9
+QUERY_NUMBER_PADDED=09
+RESULT_COLUMNS=IIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id2), COUNT(DISTINCT id4), SUM(r2), COUNT(*) FROM ans;
+RESULT_ANSWER=95	95	9.940515516534347	9216

external/duckdb/benchmark/h2oai/group/q10.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/group/q10.benchmark
+# description: Run query 10 from the H2OAI benchmark
+# group: [group]
+
+template benchmark/h2oai/group/h2oai.benchmark.in
+QUERY_NUMBER=10
+QUERY_NUMBER_PADDED=10
+RESULT_COLUMNS=IIIIIIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT id1), COUNT(DISTINCT id2), COUNT(DISTINCT id3), COUNT(DISTINCT id4), COUNT(DISTINCT id5), COUNT(DISTINCT id6), SUM(v3), SUM(count), COUNT(*) FROM ans;
+RESULT_ANSWER=95	95	95000	95	95	95000	474969574.04781127	10000000	9999993

external/duckdb/benchmark/h2oai/group/queries/load.sql

@@ -0,0 +1 @@
+CREATE TABLE x_group AS SELECT * FROM read_csv_auto('https://github.com/duckdb/duckdb-data/releases/download/v1.0/G1_1e7_1e2_5_0.csv.gz');

external/duckdb/benchmark/h2oai/group/queries/q01.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id1, sum(v1) AS v1 FROM x_group GROUP BY id1;

external/duckdb/benchmark/h2oai/group/queries/q02.sql

@@ -0,0 +1,2 @@
+CREATE TEMP TABLE ans AS SELECT id1, id2, sum(v1) AS v1 FROM x_group GROUP BY id1, id2;
+

external/duckdb/benchmark/h2oai/group/queries/q03.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id3, sum(v1) AS v1, avg(v3) AS v3 FROM x_group GROUP BY id3;

external/duckdb/benchmark/h2oai/group/queries/q04.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id4, avg(v1) AS v1, avg(v2) AS v2, avg(v3) AS v3 FROM x_group GROUP BY id4;

external/duckdb/benchmark/h2oai/group/queries/q05.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id6, sum(v1) AS v1, sum(v2) AS v2, sum(v3) AS v3 FROM x_group GROUP BY id6;

external/duckdb/benchmark/h2oai/group/queries/q06.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id4, id5, quantile_cont(v3, 0.5) AS median_v3, stddev(v3) AS sd_v3 FROM x_group GROUP BY id4, id5;

external/duckdb/benchmark/h2oai/group/queries/q07.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id3, max(v1)-min(v2) AS range_v1_v2 FROM x_group GROUP BY id3;

external/duckdb/benchmark/h2oai/group/queries/q08.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id6, v3 AS largest2_v3 FROM (SELECT id6, v3, row_number() OVER (PARTITION BY id6 ORDER BY v3 DESC) AS order_v3 FROM x_group WHERE v3 IS NOT NULL) sub_query WHERE order_v3 <= 2

external/duckdb/benchmark/h2oai/group/queries/q09.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id2, id4, pow(corr(v1, v2), 2) AS r2 FROM x_group GROUP BY id2, id4;

external/duckdb/benchmark/h2oai/group/queries/q10.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT id1, id2, id3, id4, id5, id6, sum(v3) AS v3, count(*) AS count FROM x_group GROUP BY id1, id2, id3, id4, id5, id6;

external/duckdb/benchmark/h2oai/join/h2oai.benchmark.in

@@ -0,0 +1,26 @@
+# name: ${FILE_PATH}
+# description: ${DESCRIPTION}
+# group: [h2oaijoin]
+
+require httpfs load_only
+
+name Q${QUERY_NUMBER_PADDED}
+group H2OAI Join
+
+storage persistent
+
+cache h2oaijoin.duckdb
+
+retry load 5
+
+load benchmark/h2oai/join/queries/load.sql
+
+run benchmark/h2oai/join/queries/q${QUERY_NUMBER_PADDED}.sql
+
+result_query ${RESULT_COLUMNS}
+${RESULT_QUERY}
+----
+${RESULT_ANSWER}
+
+cleanup
+DROP TABLE ans

external/duckdb/benchmark/h2oai/join/q01.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/join/q01.benchmark
+# description: Run query 01 from the H2OAI join benchmark
+# group: [join]
+
+template benchmark/h2oai/join/h2oai.benchmark.in
+QUERY_NUMBER=1
+QUERY_NUMBER_PADDED=01
+RESULT_COLUMNS=III
+RESULT_QUERY=SELECT COUNT(DISTINCT id4), SUM(v2), COUNT(*) FROM ans
+RESULT_ANSWER=9	347720187.3928393	8998860

external/duckdb/benchmark/h2oai/join/q02.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/join/q02.benchmark
+# description: Run query 02 from the H2OAI join benchmark
+# group: [join]
+
+template benchmark/h2oai/join/h2oai.benchmark.in
+QUERY_NUMBER=2
+QUERY_NUMBER_PADDED=02
+RESULT_COLUMNS=IIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT medium_id1), COUNT(DISTINCT medium_id4), COUNT(DISTINCT medium_id5), SUM(v2), COUNT(*) FROM ans
+RESULT_ANSWER=10	10	9000	449999844.93783	8998412

external/duckdb/benchmark/h2oai/join/q03.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/join/q03.benchmark
+# description: Run query 03 from the H2OAI join benchmark
+# group: [join]
+
+template benchmark/h2oai/join/h2oai.benchmark.in
+QUERY_NUMBER=3
+QUERY_NUMBER_PADDED=03
+RESULT_COLUMNS=IIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT medium_id1), COUNT(DISTINCT medium_id4), COUNT(DISTINCT medium_id5), SUM(v2), COUNT(*) FROM ans
+RESULT_ANSWER=10	10	9000	449999844.9375197	10000000

external/duckdb/benchmark/h2oai/join/q04.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/join/q04.benchmark
+# description: Run query 04 from the H2OAI join benchmark
+# group: [join]
+
+template benchmark/h2oai/join/h2oai.benchmark.in
+QUERY_NUMBER=4
+QUERY_NUMBER_PADDED=04
+RESULT_COLUMNS=IIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT medium_id1), COUNT(DISTINCT medium_id2), COUNT(DISTINCT medium_id4), SUM(v2), COUNT(*) FROM ans
+RESULT_ANSWER=10	9000	10	449999844.9375197	8998412

external/duckdb/benchmark/h2oai/join/q05.benchmark

@@ -0,0 +1,10 @@
+# name: benchmark/h2oai/join/q05.benchmark
+# description: Run query 05 from the H2OAI join benchmark
+# group: [join]
+
+template benchmark/h2oai/join/h2oai.benchmark.in
+QUERY_NUMBER=5
+QUERY_NUMBER_PADDED=05
+RESULT_COLUMNS=IIIIIII
+RESULT_QUERY=SELECT COUNT(DISTINCT big_id1), COUNT(DISTINCT big_id2), COUNT(DISTINCT big_id4), COUNT(DISTINCT big_id5), COUNT(DISTINCT big_id6), SUM(v2), COUNT(*) FROM ans
+RESULT_ANSWER=10	10000	10	10000	9000000	449860428.61554617	9000000

external/duckdb/benchmark/h2oai/join/queries/load.sql

@@ -0,0 +1,4 @@
+CREATE TABLE IF NOT EXISTS x AS SELECT * FROM read_csv_auto('https://github.com/duckdb/duckdb-data/releases/download/v1.0/J1_1e7_NA_0_0.csv.gz');
+CREATE TABLE IF NOT EXISTS small AS SELECT * FROM read_csv_auto('https://github.com/duckdb/duckdb-data/releases/download/v1.0/J1_1e7_1e1_0_0.csv.gz');
+CREATE TABLE IF NOT EXISTS medium AS SELECT * FROM read_csv_auto('https://github.com/duckdb/duckdb-data/releases/download/v1.0/J1_1e7_1e4_0_0.csv.gz');
+CREATE TABLE IF NOT EXISTS big AS SELECT * FROM read_csv_auto('https://github.com/duckdb/duckdb-data/releases/download/v1.0/J1_1e7_1e7_0_0.csv.gz');

external/duckdb/benchmark/h2oai/join/queries/q01.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT x.*, small.id4 AS small_id4, v2 FROM x JOIN small USING (id1);

external/duckdb/benchmark/h2oai/join/queries/q02.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT x.*, medium.id1 AS medium_id1, medium.id4 AS medium_id4, medium.id5 AS medium_id5, v2 FROM x JOIN medium USING (id2);

external/duckdb/benchmark/h2oai/join/queries/q03.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT x.*, medium.id1 AS medium_id1, medium.id4 AS medium_id4, medium.id5 AS medium_id5, v2 FROM x LEFT JOIN medium USING (id2);

external/duckdb/benchmark/h2oai/join/queries/q04.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT x.*, medium.id1 AS medium_id1, medium.id2 AS medium_id2, medium.id4 AS medium_id4, v2 FROM x JOIN medium USING (id5);

external/duckdb/benchmark/h2oai/join/queries/q05.sql

@@ -0,0 +1 @@
+CREATE TEMP TABLE ans AS SELECT x.*, big.id1 AS big_id1, big.id2 AS big_id2, big.id4 AS big_id4, big.id5 AS big_id5, big.id6 AS big_id6, v2 FROM x JOIN big USING (id3);