kraken/stdlib/regex.krak

import io
import vector
import string
import mem
import conversions

fun regex(in: char*):regex {
    return regex(string::string(in))
}
fun regex(in: string::string):regex {
    var out.construct(in):regex
    return out
}

obj regexState(Object) {
    var character: char
    var next_states: vector::vector<regexState*>
    fun construct(charIn:char): regexState* {
        character = charIn
        next_states.construct()
        return this
    }
    fun construct(): regexState* {
        return construct(conversions::to_char(0))
    }
    fun copy_construct(old:regexState*): void {
        character = old->character
        next_states.copy_construct(&old->next_states)
    }
    fun destruct():void {
        next_states.destruct()
    }
    fun match(input: char): vector::vector<regexState*> {
        return next_states.filter(fun(it:regexState*, input:char):bool { return it->character == input; }, input)

        io::print("in match for: "); io::println(character)
        io::println("pre")
        for (var i = 0; i < next_states.size; i++;)
            io::println(next_states[i]->character)
        var nx = next_states.filter(fun(it:regexState*, input:char):bool { return it->character == input; }, input)
        io::println("next")
        for (var i = 0; i < nx.size; i++;)
            io::println(nx[i]->character)
        //return next_states.filter(fun(it:regexState*, input:char):bool { return it->character == input; }, input)
        return nx
    }
    fun is_end():bool {
        return next_states.any_true(fun(state: regexState*):bool { return state->character == 1; })
    }
}

obj regex(Object) {
    var regexString: string::string
    var begin: regexState
    fun construct(regexStringIn: string::string): regex* {
        begin.construct()
        regexString.copy_construct(&regexStringIn)

        var traverse = &begin
        for (var i = 0; i < regexString.length(); i++;) {
            var next = mem::new<regexState>()->construct(regexString[i])
            traverse->next_states.add(next)
            traverse = next
        }
        traverse->next_states.add(mem::new<regexState>()->construct(conversions::to_char(1)))
        return this
    }
    fun copy_construct(old:regex*):void {
        begin.copy_construct(&old->begin)
        regexString.copy_construct(&old->regexString)
    }
    fun destruct():void {
        begin.destruct()
        regexString.destruct()
    }
    fun long_match(to_match: char*): int { return long_match(string::string(to_match)); }
    fun long_match(to_match: string::string): int {
        var next = vector::vector(&begin)
        var longest = 0
        for (var i = 0; i < to_match.length(); i++;) {
            if (next.size == 0)
                return longest
            if (next.any_true(fun(state: regexState*):bool { return state->is_end(); }))
                longest = i
            //next = next.flatten_map<regexState*>(fun(state: regexState*): vector::vector<regexState*> { return state->match(to_match[i]); })
            //next = next.flatten_map<regexState*>(fun(state: regexState*): vector::vector<regexState*> { return state->match('a'); })
            //next = next.flatten_map(fun(state: regexState*): vector::vector<regexState*> { return state->match('a'); })
            next = next.flatten_map(fun(state: regexState*, c:char): vector::vector<regexState*> { return state->match(c); }, to_match[i])
        }
        /*io::println("ok, ending with")*/
        /*for (var i = 0; i < next.size; i++;) {*/
            /*io::println(next[i]->character)*/
            /*io::println(conversions::to_int(next[i]->next_states[0]->character))*/
        /*}*/
        if (next.any_true(fun(state: regexState*):bool { return state->is_end(); }))
            return to_match.length()
        return longest
    }
}