function markup(&$body, $track_cites = false, $op = false)
{
global $board, $config, $markup_urls;
$modifiers = extract_modifiers($body);
$body = preg_replace('@<tinyboard (?!escape )([\\w\\s]+)>(.+?)</tinyboard>@us', '', $body);
$body = preg_replace('@<(tinyboard) escape ([\\w\\s]+)>@i', '<$1 $2>', $body);
if (isset($modifiers['raw html']) && $modifiers['raw html'] == '1') {
return array();
}
$body = str_replace("\r", '', $body);
$body = utf8tohtml($body);
if (mysql_version() < 50503) {
$body = mb_encode_numericentity($body, array(0x10000, 0xffffff, 0, 0xffffff), 'UTF-8');
}
foreach ($config['markup'] as $markup) {
if (is_string($markup[1])) {
$body = preg_replace($markup[0], $markup[1], $body);
} elseif (is_callable($markup[1])) {
$body = preg_replace_callback($markup[0], $markup[1], $body);
}
}
if ($config['markup_urls']) {
$markup_urls = array();
$body = preg_replace_callback('/((?:https?:\\/\\/|ftp:\\/\\/|irc:\\/\\/)[^\\s<>()"]+?(?:\\([^\\s<>()"]*?\\)[^\\s<>()"]*?)*)((?:\\s|<|>|"|\\.||\\]|!|\\?|,|,|")*(?:[\\s<>()"]|$))/', 'markup_url', $body, -1, $num_links);
if ($num_links > $config['max_links']) {
error($config['error']['toomanylinks']);
}
if ($num_links < $config['min_links'] && $op) {
error(sprintf($config['error']['notenoughlinks'], $config['min_links']));
}
}
if ($config['markup_repair_tidy']) {
$body = str_replace(' ', ' ', $body);
}
if ($config['auto_unicode']) {
$body = unicodify($body);
if ($config['markup_urls']) {
foreach ($markup_urls as &$url) {
$body = str_replace(unicodify($url), $url, $body);
}
}
}
$tracked_cites = array();
// Cites
if (isset($board) && preg_match_all('/(^|\\s)>>(\\d+?)([\\s,.)?]|$)/m', $body, $cites, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
if (count($cites[0]) > $config['max_cites']) {
error($config['error']['toomanycites']);
}
$skip_chars = 0;
$body_tmp = $body;
$search_cites = array();
foreach ($cites as $matches) {
$search_cites[] = '`id` = ' . $matches[2][0];
}
$search_cites = array_unique($search_cites);
$query = query(sprintf('SELECT `thread`, `id` FROM ``posts_%s`` WHERE ' . implode(' OR ', $search_cites), $board['uri'])) or error(db_error());
$cited_posts = array();
while ($cited = $query->fetch(PDO::FETCH_ASSOC)) {
$cited_posts[$cited['id']] = $cited['thread'] ? $cited['thread'] : false;
}
foreach ($cites as $matches) {
$cite = $matches[2][0];
// preg_match_all is not multibyte-safe
foreach ($matches as &$match) {
$match[1] = mb_strlen(substr($body_tmp, 0, $match[1]));
}
if (isset($cited_posts[$cite])) {
$replacement = '<a onclick="highlightReply(\'' . $cite . '\', event);" href="' . $config['root'] . $board['dir'] . $config['dir']['res'] . ($cited_posts[$cite] ? $cited_posts[$cite] : $cite) . '.html#' . $cite . '">' . '>>' . $cite . '</a>';
$body = mb_substr_replace($body, $matches[1][0] . $replacement . $matches[3][0], $matches[0][1] + $skip_chars, mb_strlen($matches[0][0]));
$skip_chars += mb_strlen($matches[1][0] . $replacement . $matches[3][0]) - mb_strlen($matches[0][0]);
if ($track_cites && $config['track_cites']) {
$tracked_cites[] = array($board['uri'], $cite);
}
}
}
}
// Cross-board linking
if (preg_match_all('/(^|\\s)>>>\\/(' . $config['board_regex'] . 'f?)\\/(\\d+)?([\\s,.)?]|$)/um', $body, $cites, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
if (count($cites[0]) > $config['max_cites']) {
error($config['error']['toomanycross']);
}
$skip_chars = 0;
$body_tmp = $body;
if (isset($cited_posts)) {
// Carry found posts from local board >>X links
foreach ($cited_posts as $cite => $thread) {
$cited_posts[$cite] = $config['root'] . $board['dir'] . $config['dir']['res'] . ($thread ? $thread : $cite) . '.html#' . $cite;
}
$cited_posts = array($board['uri'] => $cited_posts);
} else {
$cited_posts = array();
}
$crossboard_indexes = array();
$search_cites_boards = array();
foreach ($cites as $matches) {
$_board = $matches[2][0];
$cite = @$matches[3][0];
if (!isset($search_cites_boards[$_board])) {
$search_cites_boards[$_board] = array();
}
$search_cites_boards[$_board][] = $cite;
}
$tmp_board = $board['uri'];
foreach ($search_cites_boards as $_board => $search_cites) {
$clauses = array();
foreach ($search_cites as $cite) {
if (!$cite || isset($cited_posts[$_board][$cite])) {
continue;
}
$clauses[] = '`id` = ' . $cite;
}
$clauses = array_unique($clauses);
if ($board['uri'] != $_board) {
if (!openBoard($_board)) {
continue;
}
// Unknown board
}
if (!empty($clauses)) {
$cited_posts[$_board] = array();
$query = query(sprintf('SELECT `thread`, `id` FROM ``posts_%s`` WHERE ' . implode(' OR ', $clauses), $board['uri'])) or error(db_error());
while ($cite = $query->fetch(PDO::FETCH_ASSOC)) {
$cited_posts[$_board][$cite['id']] = $config['root'] . $board['dir'] . $config['dir']['res'] . ($cite['thread'] ? $cite['thread'] : $cite['id']) . '.html#' . $cite['id'];
}
}
$crossboard_indexes[$_board] = $config['root'] . $board['dir'] . $config['file_index'];
}
// Restore old board
if (!$tmp_board) {
unset($GLOBALS['board']);
} elseif ($board['uri'] != $tmp_board) {
openBoard($tmp_board);
}
foreach ($cites as $matches) {
$_board = $matches[2][0];
$cite = @$matches[3][0];
// preg_match_all is not multibyte-safe
foreach ($matches as &$match) {
$match[1] = mb_strlen(substr($body_tmp, 0, $match[1]));
}
if ($cite) {
if (isset($cited_posts[$_board][$cite])) {
$link = $cited_posts[$_board][$cite];
$replacement = '<a ' . ($_board == $board['uri'] ? 'onclick="highlightReply(\'' . $cite . '\', event);" ' : '') . 'href="' . $link . '">' . '>>>/' . $_board . '/' . $cite . '</a>';
$body = mb_substr_replace($body, $matches[1][0] . $replacement . $matches[4][0], $matches[0][1] + $skip_chars, mb_strlen($matches[0][0]));
$skip_chars += mb_strlen($matches[1][0] . $replacement . $matches[4][0]) - mb_strlen($matches[0][0]);
if ($track_cites && $config['track_cites']) {
$tracked_cites[] = array($_board, $cite);
}
}
} elseif (isset($crossboard_indexes[$_board])) {
$replacement = '<a href="' . $crossboard_indexes[$_board] . '">' . '>>>/' . $_board . '/' . '</a>';
$body = mb_substr_replace($body, $matches[1][0] . $replacement . $matches[4][0], $matches[0][1] + $skip_chars, mb_strlen($matches[0][0]));
$skip_chars += mb_strlen($matches[1][0] . $replacement . $matches[4][0]) - mb_strlen($matches[0][0]);
}
}
}
$tracked_cites = array_unique($tracked_cites, SORT_REGULAR);
if ($config['strip_superfluous_returns']) {
$body = preg_replace('/\\s+$/', '', $body);
}
if ($config['markup_paragraphs']) {
$paragraphs = explode("\n", $body);
$bodyNew = "";
$tagsOpen = false;
// Matches <a>, <a href="" title="">, but not <img/> and returns a
$matchOpen = "#<([A-Z][A-Z0-9]*)+(?:(?:\\s+\\w+(?:\\s*=\\s*(?:\".*?\"|'.*?'|[^'\">\\s]+))?)+\\s*|\\s*)>#i";
// Matches </a> returns a
$matchClose = "#</([A-Z][A-Z0-9]*/?)>#i";
$tagsOpened = array();
$tagsClosed = array();
foreach ($paragraphs as $paragraph) {
// Determine if RTL based on content of line.
if (strlen(trim($paragraph)) > 0) {
$paragraphDirection = is_rtl($paragraph) ? "rtl" : "ltr";
} else {
$paragraphDirection = "empty";
}
// Add in a quote class for >quotes.
if (strpos($paragraph, ">") === 0) {
$quoteClass = "quote";
} else {
$quoteClass = "";
}
// If tags are closed, start a new line.
if ($tagsOpen === false) {
$bodyNew .= "<p class=\"body-line {$paragraphDirection} {$quoteClass}\">";
}
// If tags are open, add the paragraph to our temporary holder instead.
if ($tagsOpen !== false) {
$tagsOpen .= $paragraph;
// Recheck tags to see if we've formed a complete tag with this latest line.
if (preg_match_all($matchOpen, $tagsOpen, $tagsOpened) === preg_match_all($matchClose, $tagsOpen, $tagsClosed)) {
sort($tagsOpened[1]);
sort($tagsClosed[1]);
// Double-check to make sure these are the same tags.
if (count(array_diff_assoc($tagsOpened[1], $tagsClosed[1])) === 0) {
// Tags are closed! \o/
$bodyNew .= $tagsOpen;
$tagsOpen = false;
}
}
} else {
if (preg_match_all($matchOpen, $paragraph, $tagsOpened) === preg_match_all($matchClose, $paragraph, $tagsClosed)) {
sort($tagsOpened[1]);
sort($tagsClosed[1]);
// Double-check to make sure these are the same tags.
if (count(array_diff_assoc($tagsOpened[1], $tagsClosed[1])) === 0) {
$bodyNew .= $paragraph;
}
} else {
// Tags are open!
$tagsOpen = $paragraph;
}
}
// If tags are open, do not close it.
if (!$tagsOpen) {
$bodyNew .= "</p>";
} else {
if ($tagsOpen !== false) {
$tagsOpen .= "<br />";
}
}
}
if ($tagsOpen !== false) {
$bodyNew .= $tagsOpen;
}
$body = $bodyNew;
} else {
$body = preg_replace("/^\\s*>.*\$/m", '<span class="quote">$0</span>', $body);
$body = preg_replace("/\n/", '<br/>', $body);
}
if ($config['markup_repair_tidy']) {
$tidy = new tidy();
$body = str_replace("\t", '	', $body);
$body = $tidy->repairString($body, array('doctype' => 'omit', 'bare' => true, 'literal-attributes' => true, 'indent' => false, 'show-body-only' => true, 'wrap' => 0, 'output-bom' => false, 'output-html' => true, 'newline' => 'LF', 'quiet' => true), 'utf8');
$body = str_replace("\n", '', $body);
}
// replace tabs with 8 spaces
$body = str_replace("\t", '	', $body);
return $tracked_cites;
}
function markup(&$body, $track_cites = false)
{
global $board, $config, $markup_urls;
$modifiers = extract_modifiers($body);
$body = preg_replace('@<tinyboard (?!escape )([\\w\\s]+)>(.+?)</tinyboard>@us', '', $body);
$body = preg_replace('@<(tinyboard) escape ([\\w\\s]+)>@i', '<$1 $2>', $body);
if (isset($modifiers['raw html']) && $modifiers['raw html'] == '1') {
return array();
}
$body = str_replace("\r", '', $body);
$body = utf8tohtml($body);
if (mysql_version() < 50503) {
$body = mb_encode_numericentity($body, array(0x10000, 0xffffff, 0, 0xffffff), 'UTF-8');
}
if ($config['markup_code']) {
$code_markup = array();
$body = preg_replace_callback($config['markup_code'], function ($matches) use(&$code_markup) {
$d = count($code_markup);
$code_markup[] = $matches;
return "<code {$d}>";
}, $body);
}
foreach ($config['markup'] as $markup) {
if (is_string($markup[1])) {
$body = preg_replace($markup[0], $markup[1], $body);
} elseif (is_callable($markup[1])) {
$body = preg_replace_callback($markup[0], $markup[1], $body);
}
}
if ($config['markup_urls']) {
$markup_urls = array();
$body = preg_replace_callback('/((?:https?:\\/\\/|ftp:\\/\\/|irc:\\/\\/)[^\\s<>()"]+?(?:\\([^\\s<>()"]*?\\)[^\\s<>()"]*?)*)((?:\\s|<|>|"|\\.||\\]|!|\\?|,|,|")*(?:[\\s<>()"]|$))/', 'markup_url', $body, -1, $num_links);
if ($num_links > $config['max_links']) {
error($config['error']['toomanylinks']);
}
}
if ($config['markup_repair_tidy']) {
$body = str_replace(' ', ' ', $body);
}
if ($config['auto_unicode']) {
$body = unicodify($body);
if ($config['markup_urls']) {
foreach ($markup_urls as &$url) {
$body = str_replace(unicodify($url), $url, $body);
}
}
}
$tracked_cites = array();
// Cites
if (isset($board) && preg_match_all('/(^|\\s)>>(\\d+?)([\\s,.)?]|$)/m', $body, $cites, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
if (count($cites[0]) > $config['max_cites']) {
error($config['error']['toomanycites']);
}
$skip_chars = 0;
$body_tmp = $body;
$search_cites = array();
foreach ($cites as $matches) {
$search_cites[] = '`id` = ' . $matches[2][0];
}
$search_cites = array_unique($search_cites);
$query = query(sprintf('SELECT `thread`, `id` FROM ``posts_%s`` WHERE ' . implode(' OR ', $search_cites), $board['uri'])) or error(db_error());
$cited_posts = array();
while ($cited = $query->fetch(PDO::FETCH_ASSOC)) {
$cited_posts[$cited['id']] = $cited['thread'] ? $cited['thread'] : false;
}
foreach ($cites as $matches) {
$cite = $matches[2][0];
// preg_match_all is not multibyte-safe
foreach ($matches as &$match) {
$match[1] = mb_strlen(substr($body_tmp, 0, $match[1]));
}
if (isset($cited_posts[$cite])) {
$replacement = '<a onclick="highlightReply(\'' . $cite . '\');" href="' . $config['root'] . $board['dir'] . $config['dir']['res'] . link_for(array('id' => $cite, 'thread' => $cited_posts[$cite])) . '#' . $cite . '">' . '>>' . $cite . '</a>';
$body = mb_substr_replace($body, $matches[1][0] . $replacement . $matches[3][0], $matches[0][1] + $skip_chars, mb_strlen($matches[0][0]));
$skip_chars += mb_strlen($matches[1][0] . $replacement . $matches[3][0]) - mb_strlen($matches[0][0]);
if ($track_cites && $config['track_cites']) {
$tracked_cites[] = array($board['uri'], $cite);
}
}
}
}
// Cross-board linking
if (preg_match_all('/(^|\\s)>>>\\/(' . $config['board_regex'] . 'f?)\\/(\\d+)?([\\s,.)?]|$)/um', $body, $cites, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
if (count($cites[0]) > $config['max_cites']) {
error($config['error']['toomanycross']);
}
$skip_chars = 0;
$body_tmp = $body;
if (isset($cited_posts)) {
// Carry found posts from local board >>X links
foreach ($cited_posts as $cite => $thread) {
$cited_posts[$cite] = $config['root'] . $board['dir'] . $config['dir']['res'] . ($thread ? $thread : $cite) . '.html#' . $cite;
}
$cited_posts = array($board['uri'] => $cited_posts);
} else {
$cited_posts = array();
}
$crossboard_indexes = array();
$search_cites_boards = array();
foreach ($cites as $matches) {
$_board = $matches[2][0];
$cite = @$matches[3][0];
if (!isset($search_cites_boards[$_board])) {
$search_cites_boards[$_board] = array();
}
$search_cites_boards[$_board][] = $cite;
}
$tmp_board = $board['uri'];
foreach ($search_cites_boards as $_board => $search_cites) {
$clauses = array();
foreach ($search_cites as $cite) {
if (!$cite || isset($cited_posts[$_board][$cite])) {
continue;
}
$clauses[] = '`id` = ' . $cite;
}
$clauses = array_unique($clauses);
if ($board['uri'] != $_board) {
if (!openBoard($_board)) {
continue;
}
// Unknown board
}
if (!empty($clauses)) {
$cited_posts[$_board] = array();
$query = query(sprintf('SELECT `thread`, `id`, `slug` FROM ``posts_%s`` WHERE ' . implode(' OR ', $clauses), $board['uri'])) or error(db_error());
while ($cite = $query->fetch(PDO::FETCH_ASSOC)) {
$cited_posts[$_board][$cite['id']] = $config['root'] . $board['dir'] . $config['dir']['res'] . link_for($cite) . '#' . $cite['id'];
}
}
$crossboard_indexes[$_board] = $config['root'] . $board['dir'] . $config['file_index'];
}
// Restore old board
if ($board['uri'] != $tmp_board) {
openBoard($tmp_board);
}
foreach ($cites as $matches) {
$_board = $matches[2][0];
$cite = @$matches[3][0];
// preg_match_all is not multibyte-safe
foreach ($matches as &$match) {
$match[1] = mb_strlen(substr($body_tmp, 0, $match[1]));
}
if ($cite) {
if (isset($cited_posts[$_board][$cite])) {
$link = $cited_posts[$_board][$cite];
$replacement = '<a ' . ($_board == $board['uri'] ? 'onclick="highlightReply(\'' . $cite . '\');" ' : '') . 'href="' . $link . '">' . '>>>/' . $_board . '/' . $cite . '</a>';
$body = mb_substr_replace($body, $matches[1][0] . $replacement . $matches[4][0], $matches[0][1] + $skip_chars, mb_strlen($matches[0][0]));
$skip_chars += mb_strlen($matches[1][0] . $replacement . $matches[4][0]) - mb_strlen($matches[0][0]);
if ($track_cites && $config['track_cites']) {
$tracked_cites[] = array($_board, $cite);
}
}
} elseif (isset($crossboard_indexes[$_board])) {
$replacement = '<a href="' . $crossboard_indexes[$_board] . '">' . '>>>/' . $_board . '/' . '</a>';
$body = mb_substr_replace($body, $matches[1][0] . $replacement . $matches[4][0], $matches[0][1] + $skip_chars, mb_strlen($matches[0][0]));
$skip_chars += mb_strlen($matches[1][0] . $replacement . $matches[4][0]) - mb_strlen($matches[0][0]);
}
}
}
$tracked_cites = array_unique($tracked_cites, SORT_REGULAR);
$body = preg_replace("/^\\s*>.*\$/m", '<span class="quote">$0</span>', $body);
if ($config['strip_superfluous_returns']) {
$body = preg_replace('/\\s+$/', '', $body);
}
$body = preg_replace("/\n/", '<br/>', $body);
// Fix code markup
if ($config['markup_code']) {
foreach ($code_markup as $id => $val) {
$code = isset($val[2]) ? $val[2] : $val[1];
$code_lang = isset($val[2]) ? $val[1] : "";
$code = "<pre class='code lang-{$code_lang}'>" . str_replace(array("\n", "\t"), array(" ", "	"), htmlspecialchars($code)) . "</pre>";
$body = str_replace("<code {$id}>", $code, $body);
}
}
if ($config['markup_repair_tidy']) {
$tidy = new tidy();
$body = str_replace("\t", '	', $body);
$body = $tidy->repairString($body, array('doctype' => 'omit', 'bare' => true, 'literal-attributes' => true, 'indent' => false, 'show-body-only' => true, 'wrap' => 0, 'output-bom' => false, 'output-html' => true, 'newline' => 'LF', 'quiet' => true), 'utf8');
$body = str_replace("\n", '', $body);
}
// replace tabs with 8 spaces
$body = str_replace("\t", ' ', $body);
return $tracked_cites;
}
/**
* Recursively turn relative links into absolute links
* @static
*/
function FixLinks(&$content, $server, $offset)
{
$pos = mb_strpos($content, 'href="', $offset);
if ($pos <= 0) {
return;
}
$pos = $pos + 6;
$pos2 = mb_strpos($content, '"', $pos);
if ($pos2 <= 0) {
return;
}
//well formed link
$check = mb_strpos($content, '>', $pos);
if ($check !== false && $check < $pos2) {
SimpleBlogCommon::FixLinks($content, $server, $pos2);
return;
}
$title = mb_substr($content, $pos, $pos2 - $pos);
//internal link
if (mb_strpos($title, 'mailto:') !== false) {
SimpleBlogCommon::FixLinks($content, $server, $pos2);
return;
}
if (mb_strpos($title, '://') !== false) {
SimpleBlogCommon::FixLinks($content, $server, $pos2);
return;
}
if (mb_strpos($title, '/') === 0) {
$replacement = $server . $title;
} else {
$replacement = $server . common::GetUrl($title);
}
$content = mb_substr_replace($content, $replacement, $pos, $pos2 - $pos);
SimpleBlogCommon::FixLinks($content, $server, $pos2);
}
public static function fixcase_callback($word)
{
$word = $word[1];
$word = mb_strtolower($word, 'UTF-8');
if ($word == "de") {
return $word;
}
$word = StringUtils::mb_ucfirst($word, 'UTF-8');
if (mb_substr($word, 1, 1, 'UTF-8') == "'") {
if (mb_substr($word, 0, 1, 'UTF-8') == "D") {
$word = mb_strtolower($word, 'UTF-8');
}
$next = mb_substr($word, 2, 1, 'UTF-8');
$next = mb_strtoupper($next, 'UTF-8');
$word = mb_substr_replace($word, $next, 2, 1, 'UTF-8');
}
return $word;
}
function testMbSubstrReplaceWithUtf8()
{
$str = 'âønæë';
$this->assertSame(mb_substr_replace($str, 'ñ', 1), 'âñønæë');
// no len
$this->assertSame(mb_substr_replace($str, 'ñ', 0, 2), 'ñnæë');
$this->assertSame(mb_substr_replace($str, 'ñx', 2, 3), 'âøñx');
$this->assertSame(mb_substr_replace($str, 'z', 2, 10), 'âøz');
// end+
$this->assertSame(mb_substr_replace($str, 'ñ', 1, 2), 'âñæë');
}
