Item9289: Backlink search is A-Z centric and cannot handle I18N or special characters

Priority:	CurrentState:	AppliesTo:	Component:	WaitingFor:
Normal	Confirmed	Engine	I18N

Our backlinks feature is flawed.

There are several problems

• If you have a topic called "Hello++Hello" the search code barfs because of the + signs in a regex.
• If the topic name you search for is part of an I18N word like SuperDuperØsters then it finds SuperDuper as a backlink because it does not see the Ø as a letter. Same with accented characters.

There are two problems.

First the regex search cannot handle regex characters. We need to escape the regex characters of %TOPIC. Maybe we need a BACKLINKS macro which can be defined only in template land where we can escape the regex characters in TOPIC?

2nd the regex in the backlinks templates has a [A-Za-z0-9] in them.

We cannot use A-Z in ANY regex in our user interface. Most countries in this world has other characters then A-Z.

Even if we cannot get it perfect, we can at least make it near perfect. We probably have to use a character class. Maybe a negative one that lists all the characters we do not see as part of a topic name. We are depending on the name filter and I am not sure we can get to this. But at least we can use the default. Not many admins will ever change the namefilter anyway.

-- KennethLavrsen - 09 Jul 2010

Defer to 1.1.5

-- GeorgeClark - 13 Dec 2011

Defer to 2.0. UTF8 core rework is planned for a major release.

-- GeorgeClark - 18 Feb 2012