Fix bugs with TPCH implementation (cmu-db#204)

src/main/java/com/oltpbenchmark/benchmarks/tpch/TPCHUtil.java

@@ -99,4 +99,19 @@ public static String getRegionFromRegionKey(int regionKey) {
         }
     }
 
+    /**
+     * Generates a random brand string of the form 'Brand#MN' where M and N are
+     * two single character strings representing two numbers randomly and
+     * independently selected within [1 .. 5]
+     *
+     * @param rand Random generator to use
+     * @return A random brand conforming to the TPCH specification
+     */
+    public static String randomBrand(RandomGenerator rand) {
+        int M = rand.number(1, 5);
+        int N = rand.number(1, 5);
+
+        return String.format("Brand#%d%d", M, N);
+    }
+
 }

src/main/java/com/oltpbenchmark/benchmarks/tpch/TPCHWorker.java

@@ -41,7 +41,7 @@ public TPCHWorker(TPCHBenchmark benchmarkModule, int id) {
     protected TransactionStatus executeWork(Connection conn, TransactionType nextTransaction) throws UserAbortException, SQLException {
         try {
             GenericQuery proc = (GenericQuery) this.getProcedure(nextTransaction.getProcedureClass());
-            proc.run(conn, rand);
+            proc.run(conn, rand, this.configuration.getScaleFactor());
         } catch (ClassCastException e) {
             throw new RuntimeException(e);
         }

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/GenericQuery.java

@@ -31,11 +31,11 @@ public abstract class GenericQuery extends Procedure {
 
     protected static final Logger LOG = LoggerFactory.getLogger(GenericQuery.class);
 
-    protected abstract PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException;
+    protected abstract PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException;
 
-    public void run(Connection conn, RandomGenerator rand) throws SQLException {
+    public void run(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
 
-        try (PreparedStatement stmt = getStatement(conn, rand); ResultSet rs = stmt.executeQuery()) {
+        try (PreparedStatement stmt = getStatement(conn, rand, scaleFactor); ResultSet rs = stmt.executeQuery()) {
             while (rs.next()) {
                 //do nothing
             }

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q1.java

@@ -52,7 +52,7 @@ public class Q1 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         String delta = String.valueOf(rand.number(60, 120));
 
         PreparedStatement stmt = this.getPreparedStatement(conn, query_stmt);

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q10.java

@@ -27,7 +27,7 @@
 
 public class Q10 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""      
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                c_custkey,
                c_name,
@@ -63,7 +63,7 @@ public class Q10 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // DATE is the first day of a randomly selected month from the second month of 1993 to the first month of 1995
         int year = rand.number(1993, 1995);
         int month = rand.number(year == 1993 ? 2 : 1, year == 1995 ? 1 : 12);

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q11.java

@@ -28,7 +28,7 @@
 
 public class Q11 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""            
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                ps_partkey,
                SUM(ps_supplycost * ps_availqty) AS VALUE
@@ -39,7 +39,7 @@ public class Q11 extends GenericQuery {
             WHERE
                ps_suppkey = s_suppkey
                AND s_nationkey = n_nationkey
-               AND n_name = 'ETHIOPIA'
+               AND n_name = ?
             GROUP BY
                ps_partkey
             HAVING
@@ -58,17 +58,17 @@ public class Q11 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // NATION is randomly selected within the list of values defined for N_NAME in Clause 4.2.3
         String nation = TPCHUtil.choice(TPCHConstants.N_NAME, rand);
 
         // FRACTION is chosen as 0.0001 / SF
-        // TODO: we should technically pass dbgen's SF down here somehow
-        double fraction = 0.0001;
+        double fraction = 0.0001 / scaleFactor;
 
         PreparedStatement stmt = this.getPreparedStatement(conn, query_stmt);
-        stmt.setDouble(1, fraction);
-        stmt.setString(2, nation);
+        stmt.setString(1, nation);
+        stmt.setDouble(2, fraction);
+        stmt.setString(3, nation);
         return stmt;
     }
 }

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q12.java

@@ -29,37 +29,48 @@
 
 public class Q12 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""            
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
-               ps_partkey,
-               SUM(ps_supplycost * ps_availqty) AS VALUE
+                l_shipmode,
+                SUM(
+                    CASE
+                        WHEN
+                            o_orderpriority = '1-URGENT' OR o_orderpriority = '2-HIGH'
+                        THEN
+                            1
+                        ELSE
+                            0
+                    END
+                ) AS high_line_count,
+                SUM(
+                    CASE
+                        WHEN
+                            o_orderpriority <> '1-URGENT' AND o_orderpriority <> '2-HIGH'
+                        THEN
+                            1
+                        ELSE
+                            0
+                    END
+                ) AS low_line_count
             FROM
-               partsupp,
-               supplier,
-               nation
+                orders,
+                lineitem
             WHERE
-               ps_suppkey = s_suppkey
-               AND s_nationkey = n_nationkey
-               AND n_name = 'ETHIOPIA'
+                o_orderkey = l_orderkey
+                AND l_shipmode IN (?, ?)
+                AND l_commitdate < l_receiptdate
+                AND l_shipdate < l_commitdate
+                AND l_receiptdate >= DATE ?
+                AND l_receiptdate < DATE ? + INTERVAL '1' YEAR
             GROUP BY
-               ps_partkey
-            HAVING
-               SUM(ps_supplycost * ps_availqty) > (
-               SELECT
-                  SUM(ps_supplycost * ps_availqty) * ?
-               FROM
-                  partsupp, supplier, nation
-               WHERE
-                  ps_suppkey = s_suppkey
-                  AND s_nationkey = n_nationkey
-                  AND n_name = ? )
-               ORDER BY
-                  VALUE DESC
+                l_shipmode
+            ORDER BY
+                l_shipmode
             """
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // SHIPMODE1 is randomly selected within the list of values defined for Modes in Clause 4.2.2.13
         String shipMode1 = TPCHUtil.choice(TPCHConstants.MODES, rand);
 

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q13.java

@@ -27,7 +27,7 @@
 
 public class Q13 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""                        
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                c_count,
                COUNT(*) AS custdist
@@ -55,7 +55,7 @@ public class Q13 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // WORD1 is randomly selected from 4 possible values: special, pending, unusual, express
         String word1 = TPCHUtil.choice(new String[]{"special", "pending", "unusual", "express"}, rand);
 

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q14.java

@@ -27,7 +27,7 @@
 
 public class Q14 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""        
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                100.00 * SUM(
                CASE
@@ -49,7 +49,7 @@ public class Q14 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // DATE is the first day of a month randomly selected from a random year within [1993 .. 1997]
         int year = rand.number(1993, 1997);
         int month = rand.number(1, 12);

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q15.java

@@ -42,7 +42,7 @@ CREATE view revenue0 (supplier_no, total_revenue) AS
             """
     );
 
-    public final SQLStmt query_stmt = new SQLStmt("""                        
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                s_suppkey,
                s_name,
@@ -71,7 +71,7 @@ CREATE view revenue0 (supplier_no, total_revenue) AS
     );
 
     @Override
-    public void run(Connection conn, RandomGenerator rand) throws SQLException {
+    public void run(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // With this query, we have to set up a view before we execute the
         // query, then drop it once we're done.
         try (Statement stmt = conn.createStatement()) {
@@ -85,7 +85,7 @@ public void run(Connection conn, RandomGenerator rand) throws SQLException {
                 String sql = createview_stmt.getSQL();
                 sql = sql.replace("?", String.format("'%s'", date));
                 stmt.execute(sql);
-                super.run(conn, rand);
+                super.run(conn, rand, scaleFactor);
             } finally {
                 String sql = dropview_stmt.getSQL();
                 stmt.execute(sql);
@@ -95,7 +95,7 @@ public void run(Connection conn, RandomGenerator rand) throws SQLException {
     }
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         return this.getPreparedStatement(conn, query_stmt);
     }
 }

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q16.java

@@ -30,7 +30,7 @@
 
 public class Q16 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""      
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                p_brand,
                p_type,
@@ -66,12 +66,8 @@ AND p_size IN (?, ?, ?, ?, ?, ?, ?, ?)
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
-        // BRAND = Brand#MN where M and N are two single character strings representing two numbers randomly and
-        // independently selected within [1 .. 5];
-        int M = rand.number(1, 5);
-        int N = rand.number(1, 5);
-        String brand = String.format("BRAND#%d%d", M, N);
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
+        String brand = TPCHUtil.randomBrand(rand);
 
         // TYPE is made of the first 2 syllables of a string randomly selected within the
         // list of 3-syllable strings defined for Types in Clause 4.2.2.13

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q17.java

@@ -28,7 +28,7 @@
 
 public class Q17 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""      
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                SUM(l_extendedprice) / 7.0 AS avg_yearly
             FROM
@@ -49,12 +49,8 @@ public class Q17 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
-        // BRAND = 'Brand#MN' where MN is a two character string representing two numbers randomly and independently
-        // selected within [1 .. 5]
-        int M = rand.number(1, 5);
-        int N = rand.number(1, 5);
-        String brand = String.format("BRAND#%d%d", M, N);
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
+        String brand = TPCHUtil.randomBrand(rand);
 
         // CONTAINER is randomly selected within the list of 2-syllable strings defined for Containers in Clause
         // 4.2.2.13

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q18.java

@@ -65,7 +65,7 @@ public class Q18 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // QUANTITY is randomly selected within [312..315]
         int quantity = rand.number(312, 315);
 

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q19.java

@@ -18,6 +18,7 @@
 package com.oltpbenchmark.benchmarks.tpch.procedures;
 
 import com.oltpbenchmark.api.SQLStmt;
+import com.oltpbenchmark.benchmarks.tpch.TPCHUtil;
 import com.oltpbenchmark.util.RandomGenerator;
 
 import java.sql.Connection;
@@ -69,7 +70,7 @@ AND l_shipmode IN ('AIR', 'AIR REG')
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // QUANTITY1 is randomly selected within [1..10]
         int quantity1 = rand.number(1, 10);
 
@@ -81,17 +82,9 @@ protected PreparedStatement getStatement(Connection conn, RandomGenerator rand)
 
         // BRAND1, BRAND2, BRAND3 = 'Brand#MN' where each MN is a two character string representing two numbers
         // randomly and independently selected within [1 .. 5]
-        int M;
-        int N;
-        M = rand.number(1, 5);
-        N = rand.number(1, 5);
-        String brand1 = String.format("BRAND#%d%d", M, N);
-        M = rand.number(1, 5);
-        N = rand.number(1, 5);
-        String brand2 = String.format("BRAND#%d%d", M, N);
-        M = rand.number(1, 5);
-        N = rand.number(1, 5);
-        String brand3 = String.format("BRAND#%d%d", M, N);
+        String brand1 = TPCHUtil.randomBrand(rand);
+        String brand2 = TPCHUtil.randomBrand(rand);
+        String brand3 = TPCHUtil.randomBrand(rand);
 
         PreparedStatement stmt = this.getPreparedStatement(conn, query_stmt);
         stmt.setString(1, brand1);

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q2.java

@@ -77,7 +77,7 @@ public class Q2 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         int size = rand.number(1, 50);
         String type = TPCHUtil.choice(TPCHConstants.TYPE_S3, rand);
         String region = TPCHUtil.choice(TPCHConstants.R_NAME, rand);

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q20.java

@@ -29,7 +29,7 @@
 
 public class Q20 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt("""     
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                s_name,
                s_address
@@ -72,7 +72,7 @@ public class Q20 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // COLOR is randomly selected within the list of values defined for the generation of P_NAME
         String color = TPCHUtil.choice(TPCHConstants.P_NAME_GENERATOR, rand) + "%";
 

src/main/java/com/oltpbenchmark/benchmarks/tpch/procedures/Q21.java

@@ -28,7 +28,7 @@
 
 public class Q21 extends GenericQuery {
 
-    public final SQLStmt query_stmt = new SQLStmt(""" 
+    public final SQLStmt query_stmt = new SQLStmt("""
             SELECT
                s_name,
                COUNT(*) AS numwait
@@ -74,7 +74,7 @@ public class Q21 extends GenericQuery {
     );
 
     @Override
-    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand) throws SQLException {
+    protected PreparedStatement getStatement(Connection conn, RandomGenerator rand, double scaleFactor) throws SQLException {
         // NATION is randomly selected within the list of values defined for N_NAME in Clause 4.2.3
         String nation = TPCHUtil.choice(TPCHConstants.N_NAME, rand);