Fix RAND() function behavior

packages/mysql-on-sqlite/src/sqlite/class-wp-pdo-mysql-on-sqlite.php

@@ -436,6 +436,13 @@ class WP_PDO_MySQL_On_SQLite extends PDO {
 	 */
 	private $connection;
 
+	/**
+	 * User-defined functions registered on the SQLite connection.
+	 *
+	 * @var WP_SQLite_PDO_User_Defined_Functions
+	 */
+	private $user_defined_functions;
+
 	/**
 	 * A service for managing MySQL INFORMATION_SCHEMA tables in SQLite.
 	 *
@@ -724,7 +731,7 @@ public function __construct(
 		$this->connection->query( 'PRAGMA foreign_keys = ON' );
 
 		// Register SQLite functions.
-		WP_SQLite_PDO_User_Defined_Functions::register_for( $this->connection->get_pdo() );
+		$this->user_defined_functions = WP_SQLite_PDO_User_Defined_Functions::register_for( $this->connection->get_pdo() );
 
 		// Load MySQL grammar.
 		if ( null === self::$mysql_grammar ) {
@@ -4395,6 +4402,25 @@ private function translate_function_call( WP_Parser_Node $node ): string {
 		}
 
 		switch ( $name ) {
+			case 'RAND':
+				/*
+				 * Unseeded RAND() compiles to a fast native SQLite expression.
+				 * In MySQL, unseeded RAND() uses the thread-level random state,
+				 * independent of any RAND(N) seeding, so we don't need PHP UDF.
+				 *
+				 * We map SQLite's RANDOM() to a float in [0, 1) by masking to
+				 * 53 bits (IEEE 754 double mantissa width) and dividing by 2^53.
+				 * This avoids the edge case where masking to 63 bits and dividing
+				 * by 2^63 could round to 1.0 due to loss of precision in double.
+				 *
+				 * Seeded RAND(N) falls through to the UDF, which implements
+				 * MySQL's deterministic LCG (Linear Congruential Generator).
+				 * The UDF also handles RAND(NULL) as RAND(0), matching MySQL.
+				 */
+				if ( 0 === count( $args ) ) {
+					return '((RANDOM() & ((1 << 53) - 1)) / ((1 << 53) * 1.0))';
+				}
+				return $this->translate_sequence( $node->get_children() );
 			case 'DATE_FORMAT':
 				list ( $date, $mysql_format ) = $args;
 
@@ -6609,6 +6635,7 @@ private function flush(): void {
 		$this->last_column_meta         = array();
 		$this->is_readonly              = false;
 		$this->wrapper_transaction_type = null;
+		$this->user_defined_functions->flush();
 	}
 
 	/**

packages/mysql-on-sqlite/src/sqlite/class-wp-sqlite-pdo-user-defined-functions.php

@@ -95,6 +95,39 @@ public static function register_for( $pdo ): self {
 		'_helper_like_to_glob_pattern' => '_helper_like_to_glob_pattern',
 	);
 
+	/**
+	 * First element of the RAND(N) LCG state (the value the output is derived from).
+	 *
+	 * @var int|null
+	 */
+	private $rand_seed1 = null;
+
+	/**
+	 * Second element of the RAND(N) LCG state (the paired value used in the recurrence).
+	 *
+	 * @var int|null
+	 */
+	private $rand_seed2 = null;
+
+	/**
+	 * Last seed value passed to RAND(N) in the current statement.
+	 *
+	 * Used to detect whether the rand sequence is advancing with the same seed
+	 * (e.g. "SELECT RAND(3) FROM t"), or reseeding (starting a new sequence).
+	 *
+	 * @var int|null
+	 */
+	private $rand_last_seed = null;
+
+	/**
+	 * Clear any per-statement state held by the UDFs.
+	 */
+	public function flush(): void {
+		$this->rand_seed1     = null;
+		$this->rand_seed2     = null;
+		$this->rand_last_seed = null;
+	}
+
 	/**
 	 * A helper function to throw an error from SQLite expressions.
 	 *
@@ -167,19 +200,74 @@ public function md5( $field ) {
 	}
 
 	/**
-	 * Method to emulate MySQL RAND() function.
+	 * Method to emulate MySQL's seeded RAND(N) function.
 	 *
-	 * SQLite does have a random generator, but it is called RANDOM() and returns random
-	 * number between -9223372036854775808 and +9223372036854775807. So we substitute it
-	 * with PHP random generator.
+	 * Implements MySQL's deterministic LCG (Linear Congruential Generator),
+	 * producing bit-exact output for a given seed.
 	 *
-	 * This function uses mt_rand() which is four times faster than rand() and returns
-	 * the random number between 0 and 1.
+	 * Known divergences from MySQL:
 	 *
-	 * @return int
+	 *  1. In MySQL, RAND(N) behaves differently depending on whether the seed
+	 *     is constant expression or varies per invocation:
+	 *      - Constant seed (e.g. "SELECT RAND(3) FROM t"):
+	 *        LCG is initialized once per statement and advanced for each row.
+	 *      - Non-constant seed (e.g. "SELECT RAND(col) FROM t"):
+	 *        LCG is initialized for every row with its seed value.
+	 *
+	 *     A SQLite UDF cannot tell whether the seed expression is constant, so
+	 *     we just compare the seed against its last value. This diverges from
+	 *     MySQL in rare cases, and we can consider improving it in the future.
+	 *
+	 *  2. The LCG state is shared across call sites in the same query, so
+	 *     "SELECT RAND(1), RAND(1)" yields different results here than in MySQL.
+	 *     This is a rare edge case that we can consider improving in the future.
+	 *
+	 * Unseeded RAND() never reaches this function. The AST driver translates it
+	 * directly to a more efficient SQLite-native expression.
+	 *
+	 * @param int|float|string|null $seed Seed value.
+	 *
+	 * @return float A value in [0, 1).
 	 */
-	public function rand() {
-		return mt_rand( 0, 1 );
+	public function rand( $seed ) {
+		// Requires 64-bit PHP. Seed * 0x10000001 can exceed PHP_INT_MAX on 32-bit.
+		$max_value = 0x3FFFFFFF;
+
+		if ( null === $seed ) {
+			// MySQL treats NULL seed as 0.
+			$seed = 0;
+		} elseif ( ! is_int( $seed ) ) {
+			/*
+			 * MySQL rounds float values and numeric strings take the same path.
+			 * Reduce the value to a 32-bit range using "fmod" to avoid firing
+			 * the "out-of-range float to int" cast deprecation on PHP 8.1+.
+			 */
+			$seed = (int) fmod( round( (float) $seed, 0, PHP_ROUND_HALF_EVEN ), 0x100000000 );
+		}
+
+		// Initialize MySQL's internal 30-bit seeds.
+		if ( $seed !== $this->rand_last_seed ) {
+			/*
+			 * MySQL casts to uint32, and the intermediate results wrap at 32-bit
+			 * unsigned boundaries. We emulate this with & 0xFFFFFFFF masks.
+			 */
+			$seed_u32             = $seed & 0xFFFFFFFF;
+			$this->rand_seed1     = ( ( $seed_u32 * 0x10001 + 55555555 ) & 0xFFFFFFFF ) % $max_value;
+			$this->rand_seed2     = ( ( $seed_u32 * 0x10000001 ) & 0xFFFFFFFF ) % $max_value;
+			$this->rand_last_seed = $seed;
+		}
+
+		/*
+		 * MySQL's LCG recurrence:
+		 *   seed1 = (seed1 * 3 + seed2) % max_value
+		 *   seed2 = (seed1 + seed2 + 33) % max_value
+		 *
+		 * Note that seed1 is updated first and the new value is used for seed2.
+		 */
+		$this->rand_seed1 = ( $this->rand_seed1 * 3 + $this->rand_seed2 ) % $max_value;
+		$this->rand_seed2 = ( $this->rand_seed1 + $this->rand_seed2 + 33 ) % $max_value;
+
+		return (float) $this->rand_seed1 / (float) $max_value;
 	}
 
 	/**