voku\helper\HtmlMin::minify()

Complexity

Conditions	9
Paths	66

Size

Total Lines

124

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	46
CRAP Score	9.0007

Importance

Changes

0

<?php
    public function minify($html, $decodeUtf8Specials = false): string
    {
        $html = (string) $html;
        if (!isset($html[0])) {
            return '';
        }

        $html = \trim($html);
        if (!$html) {
            return '';
        }

        // init
        static $CACHE_SELF_CLOSING_TAGS = null;
        if ($CACHE_SELF_CLOSING_TAGS === null) {
            $CACHE_SELF_CLOSING_TAGS = \implode('|', self::$selfClosingTags);
        }

        // reset
        $this->protectedChildNodes = [];

        // save old content
        $origHtml = $html;
        $origHtmlLength = \strlen($html);

        // -------------------------------------------------------------------------
        // Minify the HTML via "HtmlDomParser"
        // -------------------------------------------------------------------------

        if ($this->doOptimizeViaHtmlDomParser) {
            $html = $this->minifyHtmlDom($html, $decodeUtf8Specials);
        }

        // -------------------------------------------------------------------------
        // Trim whitespace from html-string. [protected html is still protected]
        // -------------------------------------------------------------------------

        // Remove extra white-space(s) between HTML attribute(s)
        $html = (string) \preg_replace_callback(
            '#<([^/\s<>!]+)(?:\s+([^<>]*?)\s*|\s*)(/?)>#',
            function ($matches) {
                return '<' . $matches[1] . (string) \preg_replace('#([^\s=]+)(\=([\'"]?)(.*?)\3)?(\s+|$)#s', ' $1$2', $matches[2]) . $matches[3] . '>';
            },
            $html
        );

        if ($this->doRemoveSpacesBetweenTags) {
            // Remove spaces that are between > and <
            $html = (string) \preg_replace('/(>) (<)/', '>$2', $html);
        }

        // -------------------------------------------------------------------------
        // Restore protected HTML-code.
        // -------------------------------------------------------------------------

        $html = (string) \preg_replace_callback(
            '/<(?<element>' . $this->protectedChildNodesHelper . ')(?<attributes> [^>]*)?>(?<value>.*?)<\/' . $this->protectedChildNodesHelper . '>/',
            [$this, 'restoreProtectedHtml'],
            $html
        );

        // -------------------------------------------------------------------------
        // Restore protected HTML-entities.
        // -------------------------------------------------------------------------

        if ($this->doOptimizeViaHtmlDomParser) {
            $html = HtmlDomParser::putReplacedBackToPreserveHtmlEntities($html);
        }

        // ------------------------------------
        // Final clean-up
        // ------------------------------------

        $html = \str_replace(
            [
                'html>' . "\n",
                "\n" . '<html',
                'html/>' . "\n",
                "\n" . '</html',
                'head>' . "\n",
                "\n" . '<head',
                'head/>' . "\n",
                "\n" . '</head',
            ],
            [
                'html>',
                '<html',
                'html/>',
                '</html',
                'head>',
                '<head',
                'head/>',
                '</head',
            ],
            $html
        );

        // self closing tags, don't need a trailing slash ...
        $replace = [];
        $replacement = [];
        foreach (self::$selfClosingTags as $selfClosingTag) {
            $replace[] = '<' . $selfClosingTag . '/>';
            $replacement[] = '<' . $selfClosingTag . '>';
            $replace[] = '<' . $selfClosingTag . ' />';
            $replacement[] = '<' . $selfClosingTag . '>';
        }
        $html = \str_replace(
            $replace,
            $replacement,
            $html
        );

        $html = (string) \preg_replace('#<\b(' . $CACHE_SELF_CLOSING_TAGS . ')([^>]*+)><\/\b\1>#', '<\\1\\2>', $html);

        // ------------------------------------
        // check if compression worked
        // ------------------------------------

        if ($origHtmlLength < \strlen($html)) {
            $html = $origHtml;
        }

        return $html;
    }