TER-59 Command to check duplicates in table

Author: jirkasemmler

src/Table/Teradata/TeradataTableQueryBuilder.php

@@ -23,7 +23,7 @@ class TeradataTableQueryBuilder implements TableQueryBuilderInterface
     ];
 
     private const INVALID_PKS_FOR_TABLE = 'invalidPKs';
-    const PK_CONSTRAINT_NAME = 'kbc_pk';
+    private const PK_CONSTRAINT_NAME = 'kbc_pk';
 
     public function getCreateTempTableCommand(string $schemaName, string $tableName, ColumnCollection $columns): string
     {
@@ -145,7 +145,7 @@ public function getCreateTableCommand(
 
         if ($primaryKeys !== []) {
             $columnsSql .= sprintf(
-                ",\nCONSTRAINT %s PRIMARY KEY (%s)",
+                ',\nCONSTRAINT %s PRIMARY KEY (%s)',
                 self::PK_CONSTRAINT_NAME,
                 implode(
                     ', ',
@@ -187,7 +187,7 @@ public function getCreateTableCommandFromDefinition(
     public function getAddPrimaryKeyCommand(string $schemaName, string $tableName, array $columns): string
     {
         return sprintf(
-            "ALTER TABLE %s.%s ADD CONSTRAINT %s PRIMARY KEY (%s);",
+            'ALTER TABLE %s.%s ADD CONSTRAINT %s PRIMARY KEY (%s);',
             TeradataQuote::quoteSingleIdentifier($schemaName),
             TeradataQuote::quoteSingleIdentifier($tableName),
             self::PK_CONSTRAINT_NAME,
@@ -198,10 +198,34 @@ public function getAddPrimaryKeyCommand(string $schemaName, string $tableName, a
     public function getDropPrimaryKeyCommand(string $schemaName, string $tableName): string
     {
         return sprintf(
-            "ALTER TABLE %s.%s DROP CONSTRAINT %s;",
+            'ALTER TABLE %s.%s DROP CONSTRAINT %s;',
             TeradataQuote::quoteSingleIdentifier($schemaName),
             TeradataQuote::quoteSingleIdentifier($tableName),
             self::PK_CONSTRAINT_NAME
         );
     }
+
+    /**
+     * @param string[] $columns
+     */
+    public function getCommandForDuplicates(string $schemaName, string $tableName, array $columns): string
+    {
+        $formattedColumns = implode(
+            ',',
+            array_map(fn($item) => TeradataQuote::quoteSingleIdentifier($item), $columns)
+        );
+        return sprintf(
+            <<<SQL
+SELECT MAX("_row_number_") AS "count" FROM
+(
+    SELECT ROW_NUMBER() OVER (PARTITION BY %s ORDER BY %s) AS "_row_number_" FROM %s.%s
+) "data"
+SQL
+            ,
+            $formattedColumns,
+            $formattedColumns,
+            TeradataQuote::quoteSingleIdentifier($schemaName),
+            TeradataQuote::quoteSingleIdentifier($tableName),
+        );
+    }
 }

tests/Functional/Teradata/Table/TeradataTableQueryBuilderTest.php

@@ -108,13 +108,12 @@ public function testGetDropTableCommand(): void
      * @dataProvider createTableTestSqlProvider
      */
     public function testGetCreateCommand(
-        array  $columns,
-        array  $primaryKeys,
-        array  $expectedColumnNames,
-        array  $expectedPKs,
+        array $columns,
+        array $primaryKeys,
+        array $expectedColumnNames,
+        array $expectedPKs,
         string $expectedSql
-    ): void
-    {
+    ): void {
         $sql = $this->qb->getCreateTableCommand(
             $this->getDatabaseName(),
             self::TABLE_GENERIC,
@@ -346,10 +345,9 @@ public function createTableTestFromDefinitionSqlProvider(): Generator
      */
     public function testGetCreateTableCommandFromDefinition(
         TeradataTableDefinition $definition,
-        string                  $expectedSql,
-        bool                    $createPrimaryKeys
-    ): void
-    {
+        string $expectedSql,
+        bool $createPrimaryKeys
+    ): void {
         $this->cleanDatabase($this->getDatabaseName());
         $this->createDatabase($this->getDatabaseName());
         $sql = $this->qb->getCreateTableCommandFromDefinition($definition, $createPrimaryKeys);
@@ -370,7 +368,7 @@ public function testGetCreateTableCommandFromDefinition(
         }
     }
 
-    public function testAddAndDropPK()
+    public function testAddAndDropPK(): void
     {
         $testDb = $this->getDatabaseName();
         $tableName = self::TABLE_GENERIC;
@@ -411,4 +409,51 @@ public function testAddAndDropPK()
         $ref1 = new TeradataTableReflection($this->connection, $testDb, $tableName);
         $this->assertEmpty($ref1->getPrimaryKeysNames());
     }
+
+
+    public function testDeduplication(): void
+    {
+        $testDb = $this->getDatabaseName();
+        $tableName = self::TABLE_GENERIC;
+
+        // definition for table
+        $definition = new TeradataTableDefinition(
+            $testDb,
+            $tableName,
+            false,
+            new ColumnCollection(
+                [
+                    TeradataColumn::createGenericColumn('col1'),
+                    TeradataColumn::createGenericColumn('col2'),
+                    TeradataColumn::createGenericColumn('col3'),
+                ]
+            ),
+            []
+        );
+
+        $sql = $this->qb->getCreateTableCommandFromDefinition($definition, true);
+        $this->connection->executeQuery($sql);
+
+        foreach ([['1', '1', '1'], ['2', '2', '2'], ['3', '3', '3']] as $i) {
+            $this->connection->executeStatement(sprintf(
+                'INSERT INTO %s.%s VALUES (%s)',
+                TeradataQuote::quoteSingleIdentifier($testDb),
+                TeradataQuote::quoteSingleIdentifier($tableName),
+                implode(',', $i)
+            ));
+        }
+        $duplicatedSql = $this->qb->getCommandForDuplicates($testDb, $tableName, ['col2', 'col3']);
+        $data = $this->connection->fetchOne($duplicatedSql);
+        $this->assertEquals('1', $data);
+
+        $this->connection->executeStatement(sprintf(
+            'INSERT INTO %s.%s VALUES (5,3,3)',
+            TeradataQuote::quoteSingleIdentifier($testDb),
+            TeradataQuote::quoteSingleIdentifier($tableName)
+        ));
+
+        $duplicatedSql = $this->qb->getCommandForDuplicates($testDb, $tableName, ['col2', 'col3']);
+        $data = $this->connection->fetchOne($duplicatedSql);
+        $this->assertEquals('2', $data);
+    }
 }

tests/Unit/Table/Teradata/TeradataTableQueryBuilderTest.php

@@ -75,8 +75,6 @@ public function testGetPKCommand(): void
             'ALTER TABLE "myDB"."myTable" ADD CONSTRAINT kbc_pk PRIMARY KEY ("my","rules");',
             $this->qb->getAddPrimaryKeyCommand('myDB', 'myTable', ['my', 'rules'])
         );
-
-
     }
 
     public function testGetDropPKCommand(): void
@@ -85,6 +83,16 @@ public function testGetDropPKCommand(): void
             'ALTER TABLE "myDB"."myTable" DROP CONSTRAINT kbc_pk;',
             $this->qb->getDropPrimaryKeyCommand('myDB', 'myTable')
         );
+    }
 
+    public function testGetDuplicationCommand(): void
+    {
+        $this->assertEquals(
+            'SELECT MAX("_row_number_") AS "count" FROM
+(
+    SELECT ROW_NUMBER() OVER (PARTITION BY "my","rules" ORDER BY "my","rules") AS "_row_number_" FROM "myDB"."myTable"
+) "data"',
+            $this->qb->getCommandForDuplicates('myDB', 'myTable', ['my', 'rules'])
+        );
     }
 }