Detect default collation to be used during database creation

core/Db/Schema.php

@@ -89,6 +89,17 @@ private function getSchema(): SchemaInterface
         return $this->schema;
     }
 
+    /**
+     * Returns the default collation for a charset.
+     *
+     * @param string $charset
+     * @return string
+     */
+    public function getDefaultCollationForCharset(string $charset): string
+    {
+        return $this->getSchema()->getDefaultCollationForCharset($charset);
+    }
+
     /**
      * Get the table options to use for a CREATE TABLE statement.
      *

core/Db/Schema/Mysql.php

@@ -669,6 +669,31 @@ public function supportsComplexColumnUpdates(): bool
         return true;
     }
 
+    /**
+     * Returns the default collation for a charset.
+     *
+     * @param string $charset
+     *
+     * @return string
+     * @throws Exception
+     */
+    public function getDefaultCollationForCharset(string $charset): string
+    {
+        $result = $this->getDb()->fetchRow(
+            'SHOW COLLATION WHERE `Default` = "Yes" AND `Charset` = ?',
+            [$charset]
+        );
+
+        if (!isset($result['Collation'])) {
+            throw new Exception(sprintf(
+                'Failed to detect default collation for character set "%s"',
+                $charset
+            ));
+        }
+
+        return $result['Collation'];
+    }
+
     public function getDefaultPort(): int
     {
         return 3306;
@@ -770,8 +795,9 @@ public function getSupportedReadIsolationTransactionLevel(): string
     protected function getDatabaseCreateOptions(): string
     {
         $charset = DbHelper::getDefaultCharset();
+        $collation = $this->getDefaultCollationForCharset($charset);
 
-        return "DEFAULT CHARACTER SET $charset";
+        return "DEFAULT CHARACTER SET $charset COLLATE $collation";
     }
 
     protected function getTableEngine()

core/Db/Schema/Tidb.php

@@ -9,8 +9,6 @@
 
 namespace Piwik\Db\Schema;
 
-use Piwik\DbHelper;
-
 /**
  * Mariadb schema
  */
@@ -27,6 +25,18 @@ public function supportsComplexColumnUpdates(): bool
         return false;
     }
 
+    public function getDefaultCollationForCharset(string $charset): string
+    {
+        $collation = parent::getDefaultCollationForCharset($charset);
+
+        if ('utf8mb4' === $charset && 'utf8mb4_bin' === $collation) {
+            // replace the TiDB default "utf8mb4_bin" with a better default
+            return 'utf8mb4_0900_ai_ci';
+        }
+
+        return $collation;
+    }
+
     public function getDefaultPort(): int
     {
         return 4000;
@@ -78,16 +88,4 @@ public function getSupportedReadIsolationTransactionLevel(): string
         // TiDB doesn't support READ UNCOMMITTED
         return 'READ COMMITTED';
     }
-
-    protected function getDatabaseCreateOptions(): string
-    {
-        $charset = DbHelper::getDefaultCharset();
-        $options = "DEFAULT CHARACTER SET $charset";
-
-        if ('utf8mb4' === $charset) {
-            $options .= ' COLLATE=utf8mb4_0900_ai_ci';
-        }
-
-        return $options;
-    }
 }

core/Db/SchemaInterface.php

@@ -135,6 +135,15 @@ public function addMaxExecutionTimeHintToQuery(string $sql, float $limit): strin
      */
     public function supportsComplexColumnUpdates(): bool;
 
+    /**
+     * Returns the default collation for a charset used by this database engine.
+     *
+     * @param string $charset
+     *
+     * @return string
+     */
+    public function getDefaultCollationForCharset(string $charset): string;
+
     /**
      * Return the default port used by this database engine
      *

core/DbHelper.php

@@ -210,7 +210,7 @@ public static function tableHasIndex($table, $indexName)
      * @return string
      * @throws Tracker\Db\DbException
      */
-    public static function getDefaultCharset()
+    public static function getDefaultCharset(): string
     {
         $result = Db::get()->fetchRow("SHOW CHARACTER SET LIKE 'utf8mb4'");
 
@@ -233,6 +233,19 @@ public static function getDefaultCharset()
         return 'utf8mb4';
     }
 
+    /**
+     * Returns the default collation for a charset.
+     *
+     * @param string $charset
+     *
+     * @return string
+     * @throws Exception
+     */
+    public static function getDefaultCollationForCharset(string $charset): string
+    {
+        return Schema::getInstance()->getDefaultCollationForCharset($charset);
+    }
+
     /**
      * Returns sql queries to convert all installed tables to utf8mb4
      *

tests/PHPUnit/Integration/Db/Schema/TidbTest.php

@@ -44,6 +44,16 @@ public function testOptimize()
         $this->assertFalse($schema->optimizeTables(['table3', 'table4'], true));
     }
 
+    public function testGetDefaultCollationForCharsetReplacesUtf8mb4Binary(): void
+    {
+        $schema = new Db\Schema\Tidb();
+
+        self::assertSame(
+            'utf8mb4_0900_ai_ci',
+            $schema->getDefaultCollationForCharset('utf8mb4')
+        );
+    }
+
     /**
      * @dataProvider getTableCreateOptionsTestData
      */

tests/PHPUnit/Integration/DbHelperTest.php

@@ -119,6 +119,25 @@ public function testAddOriginHintToQuery()
         self::assertEquals($expected, $result);
     }
 
+    public function testGetDefaultCollationForCharset(): void
+    {
+        $charset = 'utf8mb4';
+        $collation = DbHelper::getDefaultCollationForCharset($charset);
+        $expectedPrefix = $charset . '_';
+
+        // exact collation depends on the database used
+        // but should always start with the charset
+        self::assertStringStartsWith($expectedPrefix, $collation);
+    }
+
+    public function testGetDefaultCollationForCharsetThrowsForInvalidCharset(): void
+    {
+        self::expectException(\Exception::class);
+        self::expectExceptionMessage('Failed to detect default collation for character set "invalid"');
+
+        DbHelper::getDefaultCollationForCharset('invalid');
+    }
+
     private function assertDbExists($dbName)
     {
         $dbs = Db::fetchAll("SHOW DATABASES");