Add MySQL-judged REGEXP test suites

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

@@ -61,6 +61,7 @@ public static function register_for( $pdo ): self {
 		'localtimestamp'               => 'now',
 		'isnull'                       => 'isnull',
 		'regexp'                       => 'regexp',
+		'regexp_like'                  => 'regexp_like',
 		'field'                        => 'field',
 		'log'                          => 'log',
 		'least'                        => 'least',
@@ -565,12 +566,15 @@ public function isnull( $field ) {
 	}
 
 	/**
-	 * Method to emulate MySQL REGEXP() function.
+	 * Method to emulate MySQL REGEXP operator.
 	 *
-	 * @param string $pattern Regular expression to match.
-	 * @param string $field   Haystack.
+	 * SQLite runs "X REGEXP Y" as "regexp(Y, X)". MySQL runs the operator
+	 * as REGEXP_LIKE(), so this method uses regexp_like().
 	 *
-	 * @return integer 1 if matched, 0 if not matched.
+	 * @param string|null $pattern Regular expression to match.
+	 * @param string|null $field   Haystack.
+	 *
+	 * @return int|null 1 if matched, 0 if not matched, NULL if an argument is NULL.
 	 */
 	public function regexp( $pattern, $field ) {
 		/*
@@ -587,17 +591,62 @@ public function regexp( $pattern, $field ) {
 		 * be reasonably safe since PHP does not allow null bytes in
 		 * regular expressions anyway.
 		 */
-		if ( "\x00" === $pattern[0] ) {
-			$pattern = substr( $pattern, 1 );
-			$flags   = '';
-		} else {
-			// Otherwise, the search is case-insensitive.
-			$flags = 'i';
+		if ( is_string( $pattern ) && 0 === strpos( $pattern, "\x00" ) ) {
+			return $this->regexp_like( $field, substr( $pattern, 1 ), 'c' );
+		}
+		return $this->regexp_like( $field, $pattern );
+	}
+
+	/**
+	 * Method to emulate MySQL REGEXP_LIKE() function.
+	 *
+	 * @param string|null $subject    Haystack.
+	 * @param string|null $pattern    Regular expression to match.
+	 * @param string|null $match_type Optional. Letters that change how the pattern matches.
+	 *                                Without "c" or "i", the match is case-insensitive.
+	 *
+	 * @throws Exception When the match type has an unknown letter.
+	 * @return int|null 1 if matched, 0 if not matched, NULL if an argument is NULL.
+	 */
+	public function regexp_like( $subject, $pattern, $match_type = '' ) {
+		if ( null === $match_type ) {
+			return null;
 		}
+
+		/*
+		 * Apply the letters in order, so that the last of "c" and "i" wins, like in MySQL.
+		 * The "u" letter needs no flag, as PCRE treats only "\n" as a line break by default.
+		 */
+		$case_insensitive = true;
+		$flags            = '';
+		$length           = strlen( $match_type );
+		for ( $i = 0; $i < $length; $i++ ) {
+			$letter = $match_type[ $i ];
+			if ( 'c' === $letter ) {
+				$case_insensitive = false;
+			} elseif ( 'i' === $letter ) {
+				$case_insensitive = true;
+			} elseif ( 'm' === $letter ) {
+				$flags .= 'm';
+			} elseif ( 'n' === $letter ) {
+				$flags .= 's';
+			} elseif ( 'u' !== $letter ) {
+				throw new Exception( 'Incorrect arguments to regexp_like' );
+			}
+		}
+		if ( $case_insensitive ) {
+			$flags .= 'i';
+		}
+
+		// Like MySQL, reject an invalid match type even for a NULL subject or pattern.
+		if ( null === $subject || null === $pattern ) {
+			return null;
+		}
+
 		$pattern = str_replace( '/', '\/', $pattern );
 		$pattern = '/' . $pattern . '/' . $flags;
 
-		return preg_match( $pattern, $field );
+		return preg_match( $pattern, $subject );
 	}
 
 	/**

packages/mysql-on-sqlite/tests/WP_MySQL_On_SQLite_Tests.php

@@ -266,6 +266,52 @@ public static function regexpOperators() {
 		);
 	}
 
+	/**
+	 * @dataProvider regexpLikeQueries
+	 */
+	public function testRegexpLike( $query, $expected_result ) {
+		$this->assertQuery( $query );
+		$this->assertSame( $expected_result, $this->last_result[0]->result );
+	}
+
+	public static function regexpLikeQueries() {
+		return array(
+			'no match type'          => array( "SELECT REGEXP_LIKE('ABC', 'b') AS result", '1' ),
+			'empty match type'       => array( "SELECT REGEXP_LIKE('ABC', 'b', '') AS result", '1' ),
+			'case-sensitive'         => array( "SELECT REGEXP_LIKE('ABC', 'b', 'c') AS result", '0' ),
+			'case-insensitive'       => array( "SELECT REGEXP_LIKE('ABC', 'b', 'i') AS result", '1' ),
+			'last "i" wins'          => array( "SELECT REGEXP_LIKE('ABC', 'b', 'ci') AS result", '1' ),
+			'last "c" wins'          => array( "SELECT REGEXP_LIKE('ABC', 'b', 'ic') AS result", '0' ),
+			'single line'            => array( "SELECT REGEXP_LIKE('a\\nb', '^b') AS result", '0' ),
+			'multiline'              => array( "SELECT REGEXP_LIKE('a\\nb', '^b', 'm') AS result", '1' ),
+			'dot without line break' => array( "SELECT REGEXP_LIKE('a\\nb', 'a.b') AS result", '0' ),
+			'dot with line break'    => array( "SELECT REGEXP_LIKE('a\\nb', 'a.b', 'n') AS result", '1' ),
+			'Unix line breaks'       => array( "SELECT REGEXP_LIKE('a\\nb', '^b', 'mu') AS result", '1' ),
+			'repeated letters'       => array( "SELECT REGEXP_LIKE('a\\nb', '^b', 'mm') AS result", '1' ),
+			'NULL subject'           => array( "SELECT REGEXP_LIKE(NULL, 'b') AS result", null ),
+			'NULL pattern'           => array( "SELECT REGEXP_LIKE('ABC', NULL) AS result", null ),
+			'NULL match type'        => array( "SELECT REGEXP_LIKE('ABC', 'b', NULL) AS result", null ),
+			'NULL REGEXP subject'    => array( "SELECT NULL REGEXP 'b' AS result", null ),
+			'NULL REGEXP pattern'    => array( "SELECT 'ABC' REGEXP NULL AS result", null ),
+			'NULL NOT REGEXP'        => array( "SELECT NULL NOT REGEXP 'b' AS result", null ),
+		);
+	}
+
+	/**
+	 * @dataProvider regexpLikeInvalidMatchTypeQueries
+	 */
+	public function testRegexpLikeWithInvalidMatchType( $query ) {
+		$this->assertQueryError( $query, 'Incorrect arguments to regexp_like' );
+	}
+
+	public static function regexpLikeInvalidMatchTypeQueries() {
+		return array(
+			'subject and pattern' => array( "SELECT REGEXP_LIKE('ABC', 'b', 'x')" ),
+			'NULL subject'        => array( "SELECT REGEXP_LIKE(NULL, 'b', 'x')" ),
+			'NULL pattern'        => array( "SELECT REGEXP_LIKE('ABC', NULL, 'x')" ),
+		);
+	}
+
 	public function testInsertDateNow() {
 		$this->assertQuery(
 			"INSERT INTO _dates (option_name, option_value) VALUES ('first', now());"

packages/mysql-on-sqlite/tests/regexp/WP_MySQL_On_SQLite_Regexp_Instr_Tests.php

@@ -0,0 +1,14 @@
+<?php
+
+require_once __DIR__ . '/WP_MySQL_On_SQLite_Regexp_Test_Case.php';
+
+/**
+ * Check REGEXP_INSTR() against MySQL's results for the REGEXP test cases.
+ */
+class WP_MySQL_On_SQLite_Regexp_Instr_Tests extends WP_MySQL_On_SQLite_Regexp_Test_Case {
+	const CHECKS = array( 'regexp_instr', 'regexp_instr_end' );
+
+	protected function setUp(): void {
+		$this->markTestSkipped( 'REGEXP_INSTR() is not supported yet.' );
+	}
+}