parse-unicode-class ( name -- class )

Vocabulary

regexp.parser

Inputs

name

an object

Outputs

class

an object

Definition

USING: assocs combinators combinators.short-circuit kernel
regexp.classes sequences splitting unicode unicode.data ;

IN: regexp.parser

: parse-unicode-class ( name -- class )
    {
        {
            [
                dup
                { [ length 1 = ] [ first "clmnpsz" member? ] }
                1&&
            ]
            [ >upper first <category-range-class> ]
        }
        {
            [ dup >title categories member? ]
            [ simple-category-table at <category-class> ]
        }
        {
            [ "script=" ?head ]
            [
                [ simple-script-table at ] [ <script-class> ]
                [ "script=" prepend bad-class ] ?if
            ]
        }
        [ bad-class ]
    } cond ;