tokenizer.kt

/**
 * Created by aziis98 on 29/11/2016.
 * Copyright 2016 Antonio De Lucreziis
 */

fun <T> split(source: List<T>, predicate: (T, T) -> Boolean): List<List<T>> {
    fun _split(predicate: (T, T) -> Boolean, acc: List<List<T>>, source: List<T>): Pair<List<List<T>>, List<T>> {
        if (acc.isEmpty()) {
            val a = source[0]
            val b = source[1]
            val xs = source.subList(2, source.size)

            return if (predicate(a, b)) {
                listOf(listOf(a, b))
            }
            else {
                listOf(listOf(a), listOf(b))
            } to xs
        }
        else {
            val (a, xs) = source.pairHead()

            return if (predicate(acc.last().last(), a)) {
                (acc.init() + listOf(acc.last() + listOf(a)))
            }
            else {
                acc + listOf(listOf(a))
            } to xs
        }
    }

    fun __split(predicate: (T, T) -> Boolean, acc: List<List<T>>, source: List<T>): List<List<T>> =
            if (source.isEmpty())
                acc
            else
                _split(predicate, acc, source).let { (acc, source) -> __split(predicate, acc, source) }

    return __split(predicate, listOf(), source)
}

fun stdGluer(a: Char, b: Char) = arrayOf(
        a.isLetter() && b.isLetter(),
        a.isDigit() && b.isDigit(),
        a.isDigit() && b == '.',
        a == '.' && b.isDigit()
).any { it }

fun tokenizeText(source: String, predicate: (Char, Char) -> Boolean = ::stdGluer): List<String> {
    return split(source.toList(), predicate)
            .map { String(it.toCharArray()) }
}

fun main(args: Array<String>) {
    val test = "15.0 + 74 / 2"

    println(tokenizeText(test).filter(String::isNotBlank))
}

fun <T> List<T>.init() = subList(0, size - 1)
fun <T> List<T>.tail() = subList(1, size)

fun <T> List<T>.pairHead() = first() to subList(1, size)
fun boolOr(vararg conditions: Boolean) = conditions.any()