123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297
<?php
class Markdown {
private static $currentFilePath = '';
private static $webPath = '';
public static function parse($text, $baseUrl = '', $webPath = '') {
self::$currentFilePath = $baseUrl;
self::$webPath = $webPath;
$text = str_replace(["\r\n", "\r"], "\n", $text);
$codeBlocks = [];
$inlineCodes = [];
$htmlTags = [];
$text = preg_replace_callback('/<(img|a|p|div|span|picture|source|br|hr)[^>]*>/i', function($matches) use (&$htmlTags) {
$id = count($htmlTags);
$placeholder = "XHTMLTAGREPLACEX" . $id . "XHTMLTAGREPLACEX";
$htmlTags[$placeholder] = $matches[0];
return $placeholder;
}, $text);
$text = preg_replace_callback('/<\/(img|a|p|div|span|picture|source|br|hr)>/i', function($matches) use (&$htmlTags) {
$id = count($htmlTags);
$placeholder = "XHTMLTAGREPLACEX" . $id . "XHTMLTAGREPLACEX";
$htmlTags[$placeholder] = $matches[0];
return $placeholder;
}, $text);
$text = preg_replace_callback('/<!--.*?-->/s', function($matches) use (&$htmlTags) {
$id = count($htmlTags);
$placeholder = "XHTMLTAGREPLACEX" . $id . "XHTMLTAGREPLACEX";
$htmlTags[$placeholder] = $matches[0];
return $placeholder;
}, $text);
$text = preg_replace_callback('/&[a-zA-Z]+;|&#[0-9]+;|&#x[0-9a-fA-F]+;/', function($matches) use (&$htmlTags) {
$id = count($htmlTags);
$placeholder = "XHTMLTAGREPLACEX" . $id . "XHTMLTAGREPLACEX";
$htmlTags[$placeholder] = $matches[0];
return $placeholder;
}, $text);
$text = preg_replace_callback('/```([a-zA-Z0-9\-_]*)\n?(.*?)\n?```/s', function($matches) use (&$codeBlocks) {
$id = count($codeBlocks);
$placeholder = "XCODEBLOCKREPLACEX" . $id . "XCODEBLOCKREPLACEX";
$codeBlocks[$placeholder] = trim($matches[2]);
return "\n" . $placeholder . "\n";
}, $text);
$text = preg_replace_callback('/`([^`\n]+?)`/', function($matches) use (&$inlineCodes) {
$id = count($inlineCodes);
$placeholder = "XINLINECODEREPLACEX" . $id . "XINLINECODEREPLACEX";
$inlineCodes[$placeholder] = $matches[1];
return $placeholder;
}, $text);
$text = htmlspecialchars($text, ENT_QUOTES, 'UTF-8');
$text = preg_replace_callback('/^###### (.+?)$/m', function($m) {
$id = self::generateHeaderId($m[1]);
return '<h6 id="' . $id . '">' . $m[1] . '</h6>';
}, $text);
$text = preg_replace_callback('/^##### (.+?)$/m', function($m) {
$id = self::generateHeaderId($m[1]);
return '<h5 id="' . $id . '">' . $m[1] . '</h5>';
}, $text);
$text = preg_replace_callback('/^#### (.+?)$/m', function($m) {
$id = self::generateHeaderId($m[1]);
return '<h4 id="' . $id . '">' . $m[1] . '</h4>';
}, $text);
$text = preg_replace_callback('/^### (.+?)$/m', function($m) {
$id = self::generateHeaderId($m[1]);
return '<h3 id="' . $id . '">' . $m[1] . '</h3>';
}, $text);
$text = preg_replace_callback('/^## (.+?)$/m', function($m) {
$id = self::generateHeaderId($m[1]);
return '<h2 id="' . $id . '">' . $m[1] . '</h2>';
}, $text);
$text = preg_replace_callback('/^# (.+?)$/m', function($m) {
$id = self::generateHeaderId($m[1]);
return '<h1 id="' . $id . '">' . $m[1] . '</h1>';
}, $text);
$text = preg_replace_callback('/!\[([^\]]*?)\]\(([^)]+?)\)/', function($matches) {
$alt = $matches[1];
$src = $matches[2];
if (!preg_match('/^(https?:\/\/|\/\/|data:)/', $src)) {
$baseUrl = self::$currentFilePath;
$src = rtrim($baseUrl, '/') . '/' . ltrim($src, '/');
}
return '<img src="' . htmlspecialchars($src, ENT_QUOTES, 'UTF-8') . '" alt="' . htmlspecialchars($alt, ENT_QUOTES, 'UTF-8') . '" class="markdown-img">';
}, $text);
$text = preg_replace_callback('/\[([^\]]+?)\]\(([^)]+?)\)/', function($matches) {
$linkText = $matches[1];
$linkUrl = $matches[2];
$processedUrl = self::processMarkdownLink($linkUrl);
return '<a href="' . $processedUrl . '" class="markdown-link">' . $linkText . '</a>';
}, $text);
$text = preg_replace('/(?<!XINLINECODEREPLACEX)\*\*\*([^*\n]+?)\*\*\*(?!XINLINECODEREPLACEX)/', '<strong><em>$1</em></strong>', $text);
$text = preg_replace('/(?<!XINLINECODEREPLACEX)\*\*([^*\n]+?)\*\*(?!XINLINECODEREPLACEX)/', '<strong>$1</strong>', $text);
$text = preg_replace('/(?<!XINLINECODEREPLACEX)(?<!\*)\*([^*\n]+?)\*(?!\*)(?!XINLINECODEREPLACEX)/', '<em>$1</em>', $text);
$text = preg_replace('/(?<!XINLINECODEREPLACEX)___([^_\n]+?)___(?!XINLINECODEREPLACEX)/', '<strong><em>$1</em></strong>', $text);
$text = preg_replace('/(?<!XINLINECODEREPLACEX)(?<!_)__([^_\n]+?)__(?!_)(?!XINLINECODEREPLACEX)/', '<strong>$1</strong>', $text);
$text = preg_replace('/(?<!XINLINECODEREPLACEX)(?<!_)_([^_\n]+?)_(?!_)(?!XINLINECODEREPLACEX)/', '<em>$1</em>', $text);
$text = preg_replace('/~~([^~\n]+?)~~/', '<del>$1</del>', $text);
$text = preg_replace('/^\s*---\s*$/m', '<hr class="markdown-hr">', $text);
$text = preg_replace('/^\s*\*\*\*\s*$/m', '<hr class="markdown-hr">', $text);
$text = preg_replace('/^> (.+?)$/m', '<blockquote class="markdown-blockquote">$1</blockquote>', $text);
$text = preg_replace('/^(\s*)[\*\-\+] (.+?)$/m', '$1<li class="markdown-li">$2</li>', $text);
$text = preg_replace('/^(\s*)\d+\. (.+?)$/m', '$1<li class="markdown-li markdown-li-ordered">$2</li>', $text);
$text = self::wrapLists($text);
$text = preg_replace_callback('/(?:^\|.+\|\s*$\n?)+/m', function($matches) {
return self::parseTable($matches[0]);
}, $text);
$text = self::wrapParagraphs($text);
foreach ($codeBlocks as $placeholder => $content) {
$escapedContent = htmlspecialchars($content, ENT_QUOTES, 'UTF-8');
$codeHtml = '<pre class="code-block"><code>' . $escapedContent . '</code></pre>';
$text = str_replace($placeholder, $codeHtml, $text);
}
foreach ($inlineCodes as $placeholder => $content) {
$escapedContent = htmlspecialchars($content, ENT_QUOTES, 'UTF-8');
$codeHtml = '<code class="inline-code">' . $escapedContent . '</code>';
$text = str_replace($placeholder, $codeHtml, $text);
}
foreach ($htmlTags as $placeholder => $content) {
if (preg_match('/<img\s/i', $content)) {
$content = preg_replace_callback('/src\s*=\s*["\']([^"\']+)["\']/i', function($matches) {
$src = $matches[1];
if (!preg_match('/^(https?:\/\/|\/\/|data:)/', $src)) {
$baseUrl = self::$currentFilePath;
$src = rtrim($baseUrl, '/') . '/' . ltrim($src, '/');
}
return 'src="' . htmlspecialchars($src, ENT_QUOTES, 'UTF-8') . '"';
}, $content);
$content = preg_replace_callback('/<img([^>]*?)>/i', function($m) {
$attrs = $m[1];
$style = '';
if (preg_match('/width\s*=\s*["\']?(\d+)["\']?/i', $attrs, $widthMatch)) {
$style .= 'width: ' . $widthMatch[1] . 'px; ';
}
if (preg_match('/height\s*=\s*["\']?(\d+)["\']?/i', $attrs, $heightMatch)) {
$style .= 'height: ' . $heightMatch[1] . 'px; ';
}
if ($style) {
$style = 'max-width: 100%; ' . $style;
if (preg_match('/style\s*=\s*["\']([^"\']*)["\']/', $attrs)) {
$attrs = preg_replace('/style\s*=\s*["\']([^"\']*)["\']/', 'style="$1 ' . $style . '"', $attrs);
} else {
$attrs .= ' style="' . trim($style) . '"';
}
}
return '<img' . $attrs . '>';
}, $content);
if (preg_match('/class\s*=\s*["\']([^"\']*)["\']/', $content, $classMatch)) {
if (strpos($classMatch[1], 'markdown-img') === false) {
$newClass = trim($classMatch[1] . ' markdown-img');
$content = preg_replace('/class\s*=\s*["\']([^"\']*)["\']/', 'class="' . $newClass . '"', $content);
}
} else {
$content = preg_replace('/<img\s/', '<img class="markdown-img" ', $content);
}
}
$text = str_replace($placeholder, $content, $text);
}
return $text;
}
private static function processMarkdownLink($linkUrl) {
if (preg_match('/^(https?:\/\/|\/\/|mailto:|#)/', $linkUrl)) {
return htmlspecialchars($linkUrl, ENT_QUOTES, 'UTF-8');
}
$urlParts = parse_url($linkUrl);
$path = $urlParts['path'] ?? $linkUrl;
$existingQuery = $urlParts['query'] ?? '';
$fragment = $urlParts['fragment'] ?? '';
$extension = strtolower(pathinfo($path, PATHINFO_EXTENSION));
$isMarkdown = in_array($extension, ['md', 'markdown']);
$baseUrl = self::$currentFilePath;
$newUrl = rtrim($baseUrl, '/') . '/' . ltrim($path, '/');
if (substr($newUrl, -1) !== '/') {
$newUrl .= '/';
}
if ($existingQuery) {
$newUrl .= '?' . $existingQuery;
} elseif ($isMarkdown) {
$newUrl .= '?view=default';
}
if ($fragment) {
$newUrl .= '#' . $fragment;
}
return htmlspecialchars($newUrl, ENT_QUOTES, 'UTF-8');
}
private static function resolveRelativePath($currentDir, $relativePath) {
$parts = $currentDir ? explode('/', $currentDir) : [];
$relParts = explode('/', $relativePath);
foreach ($relParts as $part) {
if ($part === '' || $part === '.') {
continue;
} elseif ($part === '..') {
if (count($parts) > 0) {
array_pop($parts);
}
} else {
$parts[] = $part;
}
}
return implode('/', $parts);
}
private static function generateHeaderId($headerText) {
$headerText = strip_tags($headerText);
$id = strtolower($headerText);
$id = preg_replace('/[^a-z0-9]+/', '-', $id);
$id = trim($id, '-');
return $id;
}
private static function wrapLists($text) {
$lines = explode("\n", $text);
$result = [];
$inList = false;
$listType = '';
$lastWasListItem = false;
foreach ($lines as $line) {
if (preg_match('/^(\s*)<li class="markdown-li( markdown-li-ordered)?"/', $line, $matches)) {
$isOrdered = !empty($matches[2]);
$newListType = $isOrdered ? 'ol' : 'ul';
if (!$inList) {
$result[] = "<$newListType class=\"markdown-list\">";
$listType = $newListType;
$inList = true;
} elseif ($listType !== $newListType) {
if (!($listType === 'ol' && $newListType === 'ol')) {
$result[] = "</$listType>";
$result[] = "<$newListType class=\"markdown-list\">";
$listType = $newListType;
}
}
$result[] = $line;
$lastWasListItem = true;
} else {
if ($inList && trim($line) === '' && $lastWasListItem) {
$lastWasListItem = false;
continue;
}
if ($inList && trim($line) !== '') {
$result[] = "</$listType>";
$inList = false;
}
if (trim($line) !== '') {
$result[] = $line;
$lastWasListItem = false;
}
}
}
if ($inList) {
$result[] = "</$listType>";
}
return implode("\n", $result);
}
private static function parseTable($table) {
$table = trim($table);
$rows = explode("\n", $table);
$html = '<table class="markdown-table">';
$isHeader = true;
foreach ($rows as $row) {
if (empty(trim($row))) continue;
if (preg_match('/^\|[\s\-\|:]+\|$/', $row)) {
$isHeader = false;
continue;
}
$cells = explode('|', trim($row, '|'));
$cells = array_map('trim', $cells);
$tag = $isHeader ? 'th' : 'td';
$class = $isHeader ? 'markdown-th' : 'markdown-td';
$html .= '<tr class="markdown-tr">';
foreach ($cells as $cell) {
$html .= "<$tag class=\"$class\">$cell</$tag>";
}
$html .= '</tr>';
if ($isHeader) $isHeader = false;
}
$html .= '</table>';
return $html;
}
private static function wrapParagraphs($text) {
$paragraphs = preg_split('/\n\s*\n/', $text);
$result = [];
foreach ($paragraphs as $paragraph) {
$paragraph = trim($paragraph);
if (empty($paragraph)) continue;
if (preg_match('/^XCODEBLOCKREPLACEX\d+XCODEBLOCKREPLACEX$/', $paragraph)) {
$result[] = $paragraph;
}
elseif (preg_match('/^<(h[1-6]|ul|ol|blockquote|pre|hr|table|div)/i', $paragraph)) {
$result[] = $paragraph;
}
else {
if (preg_match('/<(p|div|a|picture|img)[^>]*>/', $paragraph)) {
$result[] = $paragraph;
} else {
$paragraph = preg_replace('/\n(?!<)/', '<br>', $paragraph);
$result[] = '<p class="markdown-p">' . $paragraph . '</p>';
}
}
}
return implode("\n\n", $result);
}
}
?>