Markdownify가 변환하지 못하고 남긴 <pre><code> 블럭을 Makrdown으로 변환

code-to-md.php

<?php

function markdownFiles(string $directory) : \Generator
{
    $recurDirIter = new RecursiveDirectoryIterator("./posts/");
    $recurIterIter = new RecursiveIteratorIterator($recurDirIter);
    $regexIter = new RegexIterator($recurIterIter, '/^.+\.md$/', \RecursiveRegexIterator::GET_MATCH);

    foreach ($regexIter as $filePath) {
        yield $filePath[0];
    }
}

function convertMarkdownFile(string $markdownFile) : void
{
    $content = file($markdownFile, \FILE_IGNORE_NEW_LINES);
    $markdownContent = convertContent($content);
    backupOriginalFile($markdownFile);
    saveToMakrdownFile($markdownContent, $markdownFile);
}

function convertContent(array $content) : array
{
    $convertedContent = [];
    $inCode = false;
    foreach ($content as $line) {
        if ($inCode == false) {
            if (preg_match('/<pre>\s*<code\s+class\s*=\s*"(language-)?(?<lang>[^"]+)"\s*>(?<code>.*)<\/code>\s*<\/pre>$/i', $line, $m) ||
                preg_match('/<pre>\s+class\s*=\s*"lang:(?<lang>[^"]+)"\s*>(?<code>.*)\s*<\/pre>$/i', $line, $m)
            ) {
                if (isset($m["lang"]) && strlen($m["lang"]) > 0) {
                    $beginMark = ['```'.$m["lang"]];
                    $endMark = ['```'];
                } else {
                    $beginMark = ["`"];
                    $endMark = ["`"];
                }
                $convertedContent = array_merge(
                    $convertedContent,
                    $beginMark,
                    decodeHtmlEntity([$m["code"]]),
                    $endMark
                );
                continue;
            }
            if (preg_match('/<pre>\s*<code\s+class\s*=\s*"(language-)?(?<lang>[^"]+)"\s*>(?<code>.*)$/', $line, $m) ||
                preg_match('/<pre\s+class\s*=\s*"lang:(?<lang>[^"]+)"\s*>(?<code>.*)$/', $line, $m)
            ) {
                $inCode = true;
                $codeBlock = ["```{$m['lang']}", "{$m['code']}"];
            } else {
                $convertedContent[] = $line;
            }
        } else {
            if (preg_match('/^(?<code>.*)<\/code>\s*<\/pre>$/', $line, $m) ||
                preg_match('/^(?<code>.*)\s*<\/pre>$/', $line, $m)
            ) {
                $inCode = false;
                $codeBlock[] = "{$m['code']}";
                $codeBlock[] = "```";
                $convertedContent = array_merge($convertedContent, decodeHtmlEntity($codeBlock));
                unset($codeBlock);
            } else {
                $codeBlock[] = $line;
            }
        }
    }

    return $convertedContent;
}

function decodeHtmlEntity(array $codeBlock) : array
{
    return array_map(
        function ($line) {
            return html_entity_decode($line, ENT_QUOTES);
        },
        $codeBlock
    );
}

function backupOriginalFile(string $markdownFile) : void
{
    rename($markdownFile, $markdownFile.".saved");
}

function saveToMakrdownFile(array $content, string $filePath) : void
{
    if (file_put_contents($filePath, implode("\n", $content)) === false) {
        die("Save error.");
    }
}

foreach (markdownFiles("./posts/") as $markdownFile) {
    convertMarkdownFile($markdownFile);
}