* dashes also split words into multiple tokens, so they are now
stripped from search entirely (see: 'weather-beaten')
* only search for the full search input if the tokenizer didn't yield
any fulltext tokens. This should fix search performance after 00f048d3ae
* fixes#497
* port extended client file handling from other branch
* class DBC -> DBCReader now initializes a DBCFile
which itself is a BinaryFile
* update DBCReader to use the new DB wrappers multi-insert feature
* fix melee attack power from spells counting towards generic attack power
* combine melee and ranged attack power from the same spell into
generic attack power, if amounts are equal
* filters should test string input during initialization, so errors
can be passed on to the PageTemplate.
* have strings be compared for identity by default and (NOT) LIKE as
optional operator
* note: if a string criterium gets processed via callback function
it will not get prechecked
* readd regular indizes for name cols. There are cases where entities
are named in a way that does not work with FT indizes.
ex. "XT:9" is two tokens "XT", "9" which are too short to be indexed.
* additionally to FT search also exact match col to search string.
* set path and domain on consent cookie
* if page uses https, tag cookies as secure
* use samesite=lax so cookies get sent if user visits via external link
* implement tooltip scaling calculation for spells with attribute
SPELL_ATTR0_LEVEL_DAMAGE_CALCULATION
* I don't know why i loaded item scales as spell scales a decade ago.
Lets hope removing that didn't break stuff.
* rework map pos -> zone pos calculation. It was kinda stupid to run a
query for each individual spawn point. So cache all wma entries per
map and do the calculation in php.
* sqlgen 'spawns' is now about twice as fast
* fix converting min/max reputation requirements to TC condition
on detail page (it still lacks the exact amount)
* also Game::getReputationLevelForPoints was off by 1
* do not use stopwords in fulltext search
apparentyl there is a bug in innodb where including stopwords in a search causes the lookup to fail entirely.
e.g. innodb_ft_enable_stopword must be disabled when the index is edited (rows are added/removed)
* don't create a MATCH AGAINST search from empty search strings after sanitization
* drop fulltext indizes for locale zhCN
logographic languages need special treatment, which handling may differ by db provider
* use LIKE search by default for locale zhCN. Added config option to use fulltext if supported by db.
