Jekis/jscodematcher

## jscodematcher
<?php

/**
 * This is a regex builder.
 * Final regular expression finds the js code in the <script> tag.
 * It is complex because any js code will be found.
 */


$jsCodeGroups = [];

// Match js comment.
$jsCodeGroups['comment-1'] = [
    'name' => 'c1',
    'exp' => '//[^\n]*\n',
    'match_chars' => '// Foo comment here',
];

// Match js comment.
$jsCodeGroups['comment-2'] = [
    'name' => 'c2',
    'exp' => '/\*.*?\*/',
    'match_chars' => '/* Foo comment here */',
];

// Match any character except single quote and double quote.
$jsCodeGroups['not-a-quote'] = [
    'name' => 'notq',
    'exp' => '[^\'"]', // In regexp it must look like text: [^'"]
    'match_chars' => 'var a = 1; alert(1);',
];

// Match any js string wrapped in single quotes.
$jsCodeGroups['single-quotes-wrap'] = [
    'name' => 'sqw',
    'exp' => "'(?:\\\\'|[^'])*'", // In regexp it must look like text: '(?:\\'|[^'])*'
    'match_chars' => "'I\'m a foo string'",
];

// Match any js string wrapped in double quotes.
$jsCodeGroups['double-quotes-wrap'] = [
    'name' => 'dqw',
    'exp' => '"(?:\\\\"|[^"])*"', // In regexp it must look like text: "(?:\\"|[^"])*"
    'match_chars' => '"Another \"foo string\""',
];

// Match any new line.
$jsCodeGroups['new-line'] = [
    'name' => 'nl',
    'exp' => '\n', // In regexp it must look like text: \n
    'match_chars' => "\n",
];

// Generate regular expression from all groups.
$jsCodeGroupsRegexps = [];
foreach ($jsCodeGroups as $gData) {
    if (!empty($gData['name'])) {
        $jsCodeGroupsRegexps[] = sprintf('(?<%s>%s)', $gData['name'], $gData['exp']);
    } else {
        $jsCodeGroupsRegexps[] = sprintf('(%s)', $gData['exp']);
    }
}
$jsCodeGroupsRegexp = join('|', $jsCodeGroupsRegexps);

// And the final regular expression.
// TODO: <script> tag could have attributes.
$finalRegExp = sprintf('~<script>(?<js>(%s)*?)?</script>~si', $jsCodeGroupsRegexp);
// <script>(?<js>((?<c1>//[^\n]*\n)|(?<c2>/\*.*?\*/)|(?<notq>[^'"])|(?<sqw>'(?:\\'|[^'])*')|(?<dqw>"(?:\\"|[^"])*")|(?<nl>\n))*?)?</script>

$html = <<<HTML
<script>
/*start<!--*/
    var num = 1;
    var singleQuote = 'foo <script> \'foo\' </script> foo';
    var doubleQuote = "foo <script> \"foo\" </script> foo";
    // foo <script> foo </script> foo
    /* foo <script> foo </script> foo */
    /*
        foo <script> foo </script> foo
    */
    alert(num);
/*-->end*/
</script>
HTML;

printf("RegExp: %s\n", $finalRegExp);

if (preg_match($finalRegExp, $html, $matches)) {
    if (isset($matches['js']) && strpos($matches['js'], 'start') && strpos($matches['js'], 'end')) {
        echo 'Correct RegExp: js code found!';
    } else {
        echo 'Wrong RegExp: found js code is incorrect.';
    }
} else {
    echo 'Wrong RegExp: Nothing matched';
}

exit;
	<?php

	/**
	* This is a regex builder.
	* Final regular expression finds the js code in the <script> tag.
	* It is complex because any js code will be found.
	*/


	$jsCodeGroups = [];

	// Match js comment.
	$jsCodeGroups['comment-1'] = [
	'name' => 'c1',
	'exp' => '//[^\n]*\n',
	'match_chars' => '// Foo comment here',
	];

	// Match js comment.
	$jsCodeGroups['comment-2'] = [
	'name' => 'c2',
	'exp' => '/\.?\*/',
	'match_chars' => '/* Foo comment here */',
	];

	// Match any character except single quote and double quote.
	$jsCodeGroups['not-a-quote'] = [
	'name' => 'notq',
	'exp' => '[^\'"]', // In regexp it must look like text: [^'"]
	'match_chars' => 'var a = 1; alert(1);',
	];

	// Match any js string wrapped in single quotes.
	$jsCodeGroups['single-quotes-wrap'] = [
	'name' => 'sqw',
	'exp' => "'(?:\\\\'\|[^'])'", // In regexp it must look like text: '(?:\\'\|[^'])'
	'match_chars' => "'I\'m a foo string'",
	];

	// Match any js string wrapped in double quotes.
	$jsCodeGroups['double-quotes-wrap'] = [
	'name' => 'dqw',
	'exp' => '"(?:\\\\"\|[^"])"', // In regexp it must look like text: "(?:\\"\|[^"])"
	'match_chars' => '"Another \"foo string\""',
	];

	// Match any new line.
	$jsCodeGroups['new-line'] = [
	'name' => 'nl',
	'exp' => '\n', // In regexp it must look like text: \n
	'match_chars' => "\n",
	];

	// Generate regular expression from all groups.
	$jsCodeGroupsRegexps = [];
	foreach ($jsCodeGroups as $gData) {
	if (!empty($gData['name'])) {
	$jsCodeGroupsRegexps[] = sprintf('(?<%s>%s)', $gData['name'], $gData['exp']);
	} else {
	$jsCodeGroupsRegexps[] = sprintf('(%s)', $gData['exp']);
	}
	}
	$jsCodeGroupsRegexp = join('\|', $jsCodeGroupsRegexps);

	// And the final regular expression.
	// TODO: <script> tag could have attributes.
	$finalRegExp = sprintf('~<script>(?<js>(%s)*?)?</script>~si', $jsCodeGroupsRegexp);
	// <script>(?<js>((?<c1>//[^\n]\n)\|(?<c2>/\.?\/)\|(?<notq>[^'"])\|(?<sqw>'(?:\\'\|[^'])')\|(?<dqw>"(?:\\"\|[^"])")\|(?<nl>\n))*?)?</script>

	$html = <<<HTML
	<script>
	/start<!--/
	var num = 1;
	var singleQuote = 'foo <script> \'foo\' </script> foo';
	var doubleQuote = "foo <script> \"foo\" </script> foo";
	// foo <script> foo </script> foo
	/* foo <script> foo </script> foo */
	/*
	foo <script> foo </script> foo
	*/
	alert(num);
	/-->end/
	</script>
	HTML;

	printf("RegExp: %s\n", $finalRegExp);

	if (preg_match($finalRegExp, $html, $matches)) {
	if (isset($matches['js']) && strpos($matches['js'], 'start') && strpos($matches['js'], 'end')) {
	echo 'Correct RegExp: js code found!';
	} else {
	echo 'Wrong RegExp: found js code is incorrect.';
	}
	} else {
	echo 'Wrong RegExp: Nothing matched';
	}

	exit;