use v6.d;
unit class MaildirIndexer::Email;
use MaildirIndexer::LogTimelineSchema;

has IO $.path;
has %!headers;
has $!body;

method BUILD(:%!headers,:$!body,:$!path = IO) { }

method message-id(--> Str:D) { split-refs(%!headers<message-id>)[0] // '' }
method refs(--> Iterable) {
    return (
        |split-refs(%!headers<in-reply-to>),
        |split-refs(%!headers<references>).reverse,
    );
}

multi split-refs(Any --> Iterable) { return () }
multi split-refs(Str:D $str --> Iterable) {
    my @result;
    MaildirIndexer::LogTimelineSchema::Parse::Header.log: {
        @result = $/<refs>».Str if $str ~~ m{'<' $<refs> = (<-[<>]>+)+ % [ '>' .*? '<' ] '>' };
    }
    return @result;
}

method addresses (--> Iterable) {
    return (
        |extract-addresses(%!headers<sender>),
        |extract-addresses(%!headers<x-original-sender>),
        |extract-addresses(%!headers<from>),
        |extract-addresses(%!headers<to>),
        |extract-addresses(%!headers<reply-to>),
        |extract-addresses(%!headers<mailing-list>),
        |extract-addresses(%!headers<x-original-from>),
    ).unique;
    # we should add a hack for info@meetup.com, where we keep the
    # whole "from", since it's the only way to distinguish between
    # different groups
}

# subset of the grammar of p5's Email::Address, ignoring comments and
# quoting
my grammar Address {
    token CTL { <[ \x00 .. \x1F \x7F ]> }
    token special { <[ \( \) \< \> \[ \] \: \; \@ \\ \, \. \" ]> }

    token atext { <-CTL> & <-special> & \S }
    token dot-atom { <ws> [ [<atext>+]+ % '.' ] <ws> }

    token local-part { <.dot-atom> }
    token domain { <.dot-atom> }

    token addr { <local-part> '@' <domain> }

    rule TOP { ^ .*? [ <addr>+ % .+? ] .*? $ }
}

multi sub extract-addresses(Any --> Iterable) { return () }
multi sub extract-addresses(Str:D $str --> Iterable) {
    my @result;
    MaildirIndexer::LogTimelineSchema::Parse::Header.log: {
        with Address.parse($str) {
            @result = $_<addr>».Str;
        }
    }
    return @result;
}