Make building of search results work for multi-byte encoded characters

[Kristian-Krastev]

When a search is made, the information that is shown in every result snippet is taken from database tables 'bookshelves', 'books', 'chapters', 'pages', but the string data in them may be encoded in format different than one-to-one byte format. For example for every cyrillic character are used up to four bytes for representation.

Operations on strings in this context are not multi-byte safe and the returned snippets contain 'broken' information.
I think the multibyte string methods are a good alternative for the solution of that problem.

[ssddanbrown]

Thanks @Kristian-Krastev for offering this PR. Could you provide a minimal example of content and search term that causes breakage? Would help so I can add a test case to prevent regression.

[Kristian-Krastev]

Sure,
i have created a page with content:

На мен ми трябва нещо добро
Вкарай ги готовите в мойто число
Младо маняче за милион
Накрая да забравиме кво е било

and in the input search field i enter (the third row from the content):

Младо маняче за милион

[ssddanbrown]

Thanks for confirming the content! PR now merged for next feature release.