TreeHouse\IoBundle\Item\Modifier\Data\Transformer\HtmlToMarkdownTransformer::transform()

Complexity

Conditions	7
Paths	11

Size

Total Lines

71

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	23
CRAP Score	7.1591

Importance

Changes

0

<?php
    public function transform($value)
    {
        if (is_null($value)) {
            return $value;
        }

        if (is_scalar($value)) {
            $value = (string) $value;
        }

        if (!is_string($value)) {
            throw new TransformationFailedException(
                sprintf('Expected a string to transform, got %s instead', json_encode($value))
            );
        }

        // replace non-breaking spaces, somehow this results in a question mark when markdownifying
        $value = str_replace(['&nbsp;', "\xC2\xA0"], ' ', $value);

        // remove leading spaces/tabs
        $value = preg_replace('/^[ \t]+/m', '', $value);

        // purify the html first
        $value = $this->purifier->purify($value);

        // perform some replacements...
        $replacements = [
            [['/>\s+</', '/\s+<\//'],            ['><', '</']],   # remove whitespace/newlines between tags: this can cause
                                                                  # trailing whitespace after markdownifying
            [['/\s+<br\/?>/', '/<br\/?>\s+/'],   '<br>'],         # also remove whitespace/newlines around <br> tags
            ['/([^>])\n([^<])/',                 '\\1<br>\\2'],   # replace newlines with <br> if the newline is not between 2 tags
            ['/(<(p|li)>)<br\s?\/?>/i',          '\\1'],          # remove <br>'s at the beginning of a paragraph
            ['/<br\s?\/?>(<\/(p|li)>)/i',        '\\1'],          # remove <br>'s at the end of a paragraph
        ];

        foreach ($replacements as list($search, $replace)) {
            $value = preg_replace($search, $replace, $value);
        }

        // strip tags in headings
        foreach(range(1, 6) as $headingSize) {
            $value = preg_replace_callback('/(<h'.$headingSize.'>)(.*)(<\/h'.$headingSize.'>)/iU', function ($matches) {
                if (count($matches) !== 4) {
                    return $matches[0];
                }
                return $matches[1] . trim(strip_tags(str_replace('<br>', ' ', $matches[2]))) . $matches[3];
            }, $value);
        }
        
        // remove any double bullets
        $value = preg_replace('/(<li>\s*)[\*|\-]{1}/im', '\\1', $value);

        // convert to markdown
        $value = @$this->converter->parseString($value);

        // Fix different types of bullets. What this does is check each line if it starts with any of "-ו○",
        // not followed by another bullet, and normalizes it to "* text".
        $value = preg_replace('/^[\-ו○]\s*([^\-ו○])/mu',  '* $1', $value);

        // Now make sure there's a newline before 2 consecutive lines that start with a bullet.
        // This could lead to superfluous newlines, but they will be corrected later on.
        $value = preg_replace('/(\n\* [^\n]+){2,}/', "\n$0", "\n" . $value);

        // remove trailing spaces/tabs
        $value = preg_replace('/[ \t]+$/m', '', $value);

        // remove excessive newlines
        $value = preg_replace('/\n{3,}/m', "\n\n", $value);

        return trim($value);
    }
}