pcre2

Vocabulary

pcre2

Definition

USING: destructors ;

IN: pcre2

TUPLE: pcre2 < disposable handle ;

Methods

USING: accessors destructors pcre2 pcre2.ffi ;

M: pcre2 dispose* handle>> pcre2_code_free ;


USING: accessors alien.data alien.strings arrays assocs
combinators destructors endian grouping io.encodings.string
io.encodings.utf8 kernel make math pcre2 pcre2.ffi pcre2.private
sequences sorting ;

M:: pcre2 findall ( subject obj -- matches )
    [
        subject utf8 encode
        :> subject_bytes subject_bytes length :> subject_length
        obj handle>> :> re re pcre2-utf? :> utf? re pcre2-crlf?
        :> crlf? re f pcre2_match_data_create_from_pattern
        &pcre2_match_data_free :> match_data 0 :> start_offset!
        0 :> options! re subject_bytes subject_length
        start_offset options match_data f pcre2_match
        :> rc re pcre2-name-count :> name_count name_count
        [ f ] [
            re pcre2-name-table re pcre2-name-entry-size
            [ rot * memory>byte-array ] [ <groups> ] bi
            [ 2 cut [ be> ] [ utf8 alien>string ] bi* ]
            map>alist sort-keys
        ] if-zero :> name_table rc 0 < [
            rc
            { { PCRE2_ERROR_NOMATCH [ ~array~ ] } [ throw ] }
            case
        ] [
            match_data pcre2_get_ovector_pointer rc
            assert-positive 2 * PCRE2_SIZE <c-direct-array>
            :> ovector [
                [
                    f ovector first2 subject subseq 2array ,
                    name_table [
                        ovector rot 2 * tail-slice
                        first2 subject subseq 2array ,
                    ] assoc-each
                ] { } make , [
                    f :> break?! 0 options! ovector second
                    start_offset! ovector first2 = [
                        ovector first subject_length =
                        2147483656 options!
                    ] [
                        match_data pcre2_get_startchar
                        :> startchar start_offset startchar <=
                        ~quotation~ ~quotation~ if
                    ] if
                    [ f ] [
                        re subject_bytes subject_length
                        start_offset options match_data f
                        pcre2_match
                        :> rc rc PCRE2_ERROR_NOMATCH =
                        ~quotation~ ~quotation~ if
                    ] if
                ] loop
            ] { } make
        ] if
    ] with-destructors ;