<?php
namespace XF\Str;
use function count, strlen;
class MentionFormatter
{
const STRUCTURED_MENTION_REGEX = '#(?<=^|\s|[\](,/\'"]|--|@)@\[(\d+):(\'|"|"|)(.*)\\2\]#iU';
protected $placeholders = [];
protected $mentionedUsers = [];
public function getMentionsBbCode($message)
{
$disabledTags = array_map(
function($v) { return preg_quote($v, '#'); },
$this->getMentionDisabledBbCodeTags()
);
$message = $this->setupPlaceholders($message,
'#\[(' . implode('|', $disabledTags) . ')([= ][^\]]*)?](.*)\[/\\1]#siU'
);
$matches = $this->getPossibleMentionMatches($message);
$usersByMatch = $this->getMentionMatchUsers($matches);
$prefix = \XF::options()->userMentionKeepAt ? '@' : '';
$message = $this->applyMentionUserMatches(
$message, $matches, $usersByMatch,
function($user) use ($prefix)
{
return '[USER=' . $user['user_id'] . ']' . $prefix . $user['username'] . '[/USER]';
}
);
$message = $this->restorePlaceholders($message);
return $message;
}
protected function getMentionDisabledBbCodeTags(): array
{
$bbCodeRules = \XF::app()->bbCode()->rules('mentions');
$disabledTags = [];
foreach ($bbCodeRules->getTags() AS $tagName => $tag)
{
if (!empty($tag['stopAutoLink']) || !empty($tag['plain']))
{
$disabledTags[] = $tagName;
}
}
// technically mentions can be parsed in quotes, but it's likely they already have
// and this isn't necessarily text that we want to attribute to the poster
$disabledTags[] = 'quote';
return $disabledTags;
}
public function getMentionsStructuredText($message)
{
$message = $this->setupPlaceholders($message, self::STRUCTURED_MENTION_REGEX);
$matches = $this->getPossibleMentionMatches($message);
$usersByMatch = $this->getMentionMatchUsers($matches);
$prefix = \XF::options()->userMentionKeepAt ? '@' : '';
$message = $this->applyMentionUserMatches(
$message, $matches, $usersByMatch,
function($user) use ($prefix)
{
if (strpos($user['username'], ']') !== false)
{
if (strpos($user['username'], "'") !== false)
{
$username = '"' . $prefix . $user['username'] . '"';
}
else
{
$username = "'" . $prefix . $user['username'] . "'";
}
}
else
{
$username = $prefix . $user['username'];
}
return '@[' . $user['user_id'] . ':' . $username . ']';
}
);
$message = $this->restorePlaceholders($message);
return $message;
}
public function getMentionedUsers()
{
return $this->mentionedUsers;
}
protected function setupPlaceholders($message, $regex)
{
$this->placeholders = [];
return preg_replace_callback($regex, function($match)
{
$replace = "\x1A" . count($this->placeholders) . "\x1A";
$this->placeholders[$replace] = $match[0];
return $replace;
}, $message);
}
protected function restorePlaceholders($message)
{
if ($this->placeholders)
{
$message = strtr($message, $this->placeholders);
$this->placeholders = [];
}
return $message;
}
protected function getPossibleMentionMatches($message)
{
$min = 2;
if (!preg_match_all(
'#(?<=^|\s|[\](,/\'"]|--)@(?!\[|\s)(([^\s@]|(?<@| ){' . $min . '}((?>[:,.!?](?=[^\s:,.!?[\]()])|' . $this->getTagEndPartialRegex(true) . '+?))*)#iu',
$message, $matches, PREG_OFFSET_CAPTURE | PREG_SET_ORDER
))
{
return [];
}
return $matches;
}
protected function getTagEndPartialRegex($negated)
{
return '[' . ($negated ? '^' : '') . ':;,.!?\s@\'"*/)\]\[-]';
}
protected function getMentionMatchUsers(array $matches)
{
$db = \XF::db();
$matchKeys = array_keys($matches);
$whereParts = [];
$matchParts = [];
$usersByMatch = [];
foreach ($matches AS $key => $match)
{
if (utf8_strlen($match[1][0]) > 50)
{
// longer than max username length
continue;
}
$sql = 'user.username LIKE ' . $db->quote($db->escapeLike($match[1][0], '?%'));
$whereParts[] = $sql;
$matchParts[] = 'IF(' . $sql . ', 1, 0) AS match_' . $key;
}
if (!$whereParts)
{
return [];
}
$userResults = $db->query("
SELECT user.user_id, user.username,
" . implode(', ', $matchParts) . "
FROM xf_user AS user
WHERE (" . implode(' OR ', $whereParts) . ")
ORDER BY LENGTH(user.username) DESC
");
while ($user = $userResults->fetch())
{
$userInfo = [
'user_id' => $user['user_id'],
'username' => $user['username'],
'lower' => utf8_strtolower($user['username'])
];
foreach ($matchKeys AS $key)
{
if (!empty($user["match_$key"]))
{
$usersByMatch[$key][$user['user_id']] = $userInfo;
}
}
}
return $usersByMatch;
}
protected function applyMentionUserMatches($message, array $matches, array $usersByMatch, \Closure $tagReplacement)
{
$this->mentionedUsers = [];
if (!$usersByMatch)
{
return $message;
}
$newMessage = '';
$lastOffset = 0;
$mentionedUsers = [];
$endMatch = $this->getTagEndPartialRegex(false);
foreach ($matches AS $key => $match)
{
if ($match[0][1] > $lastOffset)
{
$newMessage .= substr($message, $lastOffset, $match[0][1] - $lastOffset);
}
else if ($lastOffset > $match[0][1])
{
continue;
}
$lastOffset = $match[0][1] + strlen($match[0][0]);
$haveMatch = false;
if (!empty($usersByMatch[$key]))
{
$testName = utf8_strtolower($match[1][0]);
$testOffset = $match[1][1];
foreach ($usersByMatch[$key] AS $userId => $user)
{
// It's possible for the byte length to change between the lower and standard versions
// due to conversions like İ -> i (2 byte to 1). Therefore, we try to check whether either
// length matches the name.
$lowerLen = strlen($user['lower']);
$originalLen = strlen($user['username']);
if ($testName === $user['lower'])
{
$nameLen = $lowerLen;
}
else if (utf8_strtolower(substr($message, $testOffset, $lowerLen)) === $user['lower'])
{
$nameLen = $lowerLen;
}
else if (
$lowerLen !== $originalLen
&& utf8_strtolower(substr($message, $testOffset, $originalLen)) === $user['lower']
)
{
$nameLen = $originalLen;
}
else
{
$nameLen = null;
}
$nextTestOffsetStart = $testOffset + ($nameLen ?: 0);
if (
$nameLen
&& (
!isset($message[$nextTestOffsetStart])
|| preg_match('#' . $endMatch . '#i', $message[$nextTestOffsetStart])
)
)
{
$mentionedUsers[$userId] = $user;
$newMessage .= $tagReplacement($user);
$haveMatch = true;
$lastOffset = $testOffset + strlen($user['username']);
break;
}
}
}
if (!$haveMatch)
{
$newMessage .= $match[0][0];
}
}
$newMessage .= substr($message, $lastOffset);
$this->mentionedUsers = $mentionedUsers;
return $newMessage;
}
}