Legacy Cyrillic in Hash Tagging

[motd2]

Cyrillic in Hash Tagging - resolved
dbtech/usertag/includes/class_core.php
line 915
find

 if (preg_match('/[^\w-]/i', $letter) OR in_array($letter, array(';', '', ' ', "\r", "\n", "\t", "\s", ".", ","))) // in_array($letter, array(';', '', ' ', "\r", "\n", "\t", "\s", ".", ","))

replace

 if (preg_match('/[^\w-]/iu', $letter) OR in_array($letter, array(';', '', ' ', "\r", "\n", "\t", "\s", ".", ","))) // in_array($letter, array(';', '', ' ', "\r", "\n", "\t", "\s", ".", ","))

test on vB3, vB4 (utf-8, windows-1251) English and Cyrillic.
Probably it will work with any encoding.

 

[Fillip H.]

I'll test it with iso-8859-1 and get back to you

 

[Fillip H.]

motd2

Update: If possible, I'd like you to beta test a change for me that may improve hash tagging altogether.

In /dbtech/usertag/includes/class_core.php find:

			case 'hash':
				$max = 100;
				$min = 1;

				$post = preg_replace('/&#\d{3,5};/', '', $post);
				while (($at = strpos($post, '#', $at)) !== false)
				{
					$at++;
					$username = '';

					if ((in_array(substr($post, $at - 2, 1), array(' ', "\r", "\n", "\t", "\s", ".", ",")) !== false) OR (($at - 2) < 0))
					{
						for ($x = 0; $x <= $max; $x += strlen($letter))
						{
							$letter = $post[$offset = $at + $x];
							if (preg_match('/[^\w-]/iu', $letter) OR in_array($letter, array(';', '', ' ', "\r", "\n", "\t", "\s", ".", ","))) // in_array($letter, array(';', '', ' ', "\r", "\n", "\t", "\s", ".", ","))
							{
								$matches[] = $username;
								break;
							}

							if ($letter == '&' AND preg_match('/&#\d{3,5};/', $post, $multi, NULL, $offset))
							{
								$letter = $multi[0];
							}

							$username .= $letter;
						}
					}
				}
				break;

Replace with

			case 'hash':
				$post = preg_split('/[\s,]+/u', $post);
				foreach ($post as $word)
				{
					if ($word{0} == '#')
					{
						// This was a match
						$matches[] = substr($word, 1);
					}
				}
				break;

I have successfully tested this with the word посещения on my local test installation (ISO-8859-1 in the Language settings, latin1_swedish_ci in the database).

Could you please test this out with your database schema, and just generally try to break it?

I'll put this change live in an update on Monday

 

[motd2]

Fillip H.

if you use one word посещения - it works

if you use the word and text
#посещения
here text message
more text

all the words are combined into one hashtag and notification does not work

 

[Fillip H.]

I'm unable to replicate this issue.