It always skips needed text parts for some reason. Simple example:
Able to reproduce with this code.
<?php $content .= ' <tr> <td>русский текст</td> <td>'.123'.</td> </tr> '; $content .= ' <tr> <td>русский текст 2</td> <td>'.123'.</td> </tr> '; "<td></td>"
Gonna file a bug for it, thanks!
This is because our multi-caret functionality tries to be intelligent, so that it only selects whole words or instances and not partial matches (unless relevant). We have a bug open that will address this:
Although with this code sample it does seem to be overzealous. Your bug report can handle that though, thanks @Defman!