Pass configured database collation to table creation statements

matomo-org · Sep 10, 2024 · 1907ff7 · 1907ff7
1 parent 1d7429a
commit 1907ff7
Show file tree

Hide file tree

Showing 7 changed files with 148 additions and 4 deletions.
diff --git a/config/global.ini.php b/config/global.ini.php
@@ -45,11 +45,12 @@
 ; Matomo should work correctly without this setting but we recommend to have a charset set.
 charset = utf8
 
-; In some database setups the collation for the connection changes after a "SET NAMES" statement
-; is issued, unless a "COLLATE" argument is added.
+; In some database setups the collation used for queries and creating tables can have unexpected
+; values, or change after a database version upgrade.
 ; If you encounter "Illegal mix of collation" errors, setting this config to the value matching
 ; your existing database tables can help.
 ; This setting will only be used if "charset" is also set.
+; Matomo should work correctly without this setting but we recommend to have a collation set.
 collation =
 
 ; Database error codes to ignore during updates

diff --git a/core/Db/Schema/Mysql.php b/core/Db/Schema/Mysql.php
@@ -678,10 +678,15 @@ public function getTableCreateOptions(): string
     {
         $engine = $this->getTableEngine();
         $charset = $this->getUsedCharset();
+        $collation = $this->getUsedCollation();
         $rowFormat = $this->getTableRowFormat();
 
         $options = "ENGINE=$engine DEFAULT CHARSET=$charset";
 
+        if ('' !== $collation) {
+            $options .= " COLLATE=$collation";
+        }
+
         if ('' !== $rowFormat) {
             $options .= " $rowFormat";
         }
@@ -784,6 +789,11 @@ protected function getUsedCharset(): string
         return $this->getDbSettings()->getUsedCharset();
     }
 
+    protected function getUsedCollation(): string
+    {
+        return $this->getDbSettings()->getUsedCollation();
+    }
+
     private function getTablePrefix()
     {
         return $this->getDbSettings()->getTablePrefix();

diff --git a/core/Db/Schema/Tidb.php b/core/Db/Schema/Tidb.php
@@ -36,12 +36,17 @@ public function getTableCreateOptions(): string
     {
         $engine = $this->getTableEngine();
         $charset = $this->getUsedCharset();
+        $collation = $this->getUsedCollation();
         $rowFormat = $this->getTableRowFormat();
 
+        if ('utf8mb4' === $charset && '' === $collation) {
+            $collation = 'utf8mb4_0900_ai_ci';
+        }
+
         $options = "ENGINE=$engine DEFAULT CHARSET=$charset";
 
-        if ('utf8mb4' === $charset) {
-            $options .= ' COLLATE=utf8mb4_0900_ai_ci';
+        if ('' !== $collation) {
+            $options .= " COLLATE=$collation";
         }
 
         if ('' !== $rowFormat) {

diff --git a/core/Db/Settings.php b/core/Db/Settings.php
@@ -33,6 +33,11 @@ public function getUsedCharset()
         return strtolower($this->getDbSetting('charset'));
     }
 
+    public function getUsedCollation()
+    {
+        return strtolower($this->getDbSetting('collation') ?? '');
+    }
+
     public function getRowFormat()
     {
         return $this->getUsedCharset() === 'utf8mb4' ? 'ROW_FORMAT=DYNAMIC' : '';

diff --git a/tests/PHPUnit/Integration/Db/Schema/MariadbTest.php b/tests/PHPUnit/Integration/Db/Schema/MariadbTest.php
@@ -68,4 +68,45 @@ public function testOptimize()
         // should optimize when forced
         $this->assertTrue($schema->optimizeTables(['table3', 'table4'], true), true);
     }
+
+    /**
+     * @dataProvider getTableCreateOptionsTestData
+     */
+    public function testTableCreateOptions(array $optionOverrides, string $expected): void
+    {
+        if (DatabaseConfig::getConfigValue('schema') !== 'Mariadb') {
+            self::markTestSkipped('Mariadb is not available');
+        }
+
+        foreach ($optionOverrides as $name => $value) {
+            DatabaseConfig::setConfigValue($name, $value);
+        }
+
+        $schema = Db\Schema::getInstance();
+
+        self::assertSame($expected, $schema->getTableCreateOptions());
+    }
+
+    public function getTableCreateOptionsTestData(): iterable
+    {
+        yield 'defaults' => [
+            [],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 ROW_FORMAT=DYNAMIC'
+        ];
+
+        yield 'override charset' => [
+            ['charset' => 'utf8mb3'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb3'
+        ];
+
+        yield 'override collation' => [
+            ['collation' => 'utf8mb4_general_ci'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci ROW_FORMAT=DYNAMIC'
+        ];
+
+        yield 'override charset and collation' => [
+            ['charset' => 'utf8mb3', 'collation' => 'utf8mb3_general_ci'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb3_general_ci'
+        ];
+    }
 }
diff --git a/tests/PHPUnit/Integration/Db/Schema/MysqlTest.php b/tests/PHPUnit/Integration/Db/Schema/MysqlTest.php
@@ -68,4 +68,45 @@ public function testOptimize()
         // should optimize when forced
         $this->assertTrue($schema->optimizeTables(['table3', 'table4'], true));
     }
+
+    /**
+     * @dataProvider getTableCreateOptionsTestData
+     */
+    public function testTableCreateOptions(array $optionOverrides, string $expected): void
+    {
+        if (DatabaseConfig::getConfigValue('schema') !== 'Mysql') {
+            self::markTestSkipped('Mysql is not available');
+        }
+
+        foreach ($optionOverrides as $name => $value) {
+            DatabaseConfig::setConfigValue($name, $value);
+        }
+
+        $schema = Db\Schema::getInstance();
+
+        self::assertSame($expected, $schema->getTableCreateOptions());
+    }
+
+    public function getTableCreateOptionsTestData(): iterable
+    {
+        yield 'defaults' => [
+            [],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 ROW_FORMAT=DYNAMIC'
+        ];
+
+        yield 'override charset' => [
+            ['charset' => 'utf8mb3'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb3'
+        ];
+
+        yield 'override collation' => [
+            ['collation' => 'utf8mb4_general_ci'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci ROW_FORMAT=DYNAMIC'
+        ];
+
+        yield 'override charset and collation' => [
+            ['charset' => 'utf8mb3', 'collation' => 'utf8mb3_general_ci'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb3_general_ci'
+        ];
+    }
 }
diff --git a/tests/PHPUnit/Integration/Db/Schema/TidbTest.php b/tests/PHPUnit/Integration/Db/Schema/TidbTest.php
@@ -43,4 +43,45 @@ public function testOptimize()
         $this->assertFalse($schema->optimizeTables(['table3', 'table4']));
         $this->assertFalse($schema->optimizeTables(['table3', 'table4'], true));
     }
+
+    /**
+     * @dataProvider getTableCreateOptionsTestData
+     */
+    public function testTableCreateOptions(array $optionOverrides, string $expected): void
+    {
+        if (DatabaseConfig::getConfigValue('schema') !== 'Tidb') {
+            self::markTestSkipped('Tidb is not available');
+        }
+
+        foreach ($optionOverrides as $name => $value) {
+            DatabaseConfig::setConfigValue($name, $value);
+        }
+
+        $schema = Db\Schema::getInstance();
+
+        self::assertSame($expected, $schema->getTableCreateOptions());
+    }
+
+    public function getTableCreateOptionsTestData(): iterable
+    {
+        yield 'defaults' => [
+            [],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci ROW_FORMAT=DYNAMIC'
+        ];
+
+        yield 'override charset' => [
+            ['charset' => 'utf8mb3'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb4_0900_ai_ci'
+        ];
+
+        yield 'override collation' => [
+            ['collation' => 'utf8mb4_general_ci'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci ROW_FORMAT=DYNAMIC'
+        ];
+
+        yield 'override charset and collation' => [
+            ['charset' => 'utf8mb3', 'collation' => 'utf8mb3_general_ci'],
+            'ENGINE=InnoDB DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb3_general_ci'
+        ];
+    }
 }