<?php

namespace App\Models;

class ChatLib {

	public function addMessage($msg) {
		$tokens = $this->tokenize($msg->text_content);
		if (empty($tokens)) return;
		$tokens[] = '';
		foreach ($tokens as $num => $token) {
			if ($num === 0) {
				$this->addTransition([], $token);
			} else {
				$start = max(0, $num - $this->size);
				$end = $num;
				for ($i = $start; $i < $end; ++$i) {
					$this->addTransition(array_slice($tokens, $i, $end - $i), $token);
					if ($end - $i < 3) break;
				}
			}
		}
	}

	public function compile() {
		foreach ($this->transitions as $key => $value) {
			$this->transitions[$key] = $this->index($this->transitions[$key]);
			if (empty($this->transitions[$key])) {
				unset($this->transitions[$key]);
			}
		}
		echo 'size: ', number_format(strlen(json_encode($this->transitions)), 0), PHP_EOL;
	}

	public function generate($limit = 100) {
		$tokens = [];
		$generated = '';
		while (strlen($generated) < $limit) {
			$next = $this->randomNext($tokens);
			if (empty($next)) break;
			$tokens[] = $next;
			$generated .= $next;
		}
		return $generated;
	}

	private function index($arr) {
		$result = [];
		$sum = 0;
		foreach ($arr as $key => $entry) {
			$weight = $entry['count'];
			if ($weight == 1) continue;
			$lower = $sum;
			$sum += $weight;
			$examples = [];
			if (is_array(end($entry['examples']))) {
				// already processed
				$examples = $entry['examples'];
			} else {
				$subsum = 0;
				foreach ($entry['examples'] as $example => $subweight) {
					$sublower = $subsum;
					$subsum += $subweight * $subweight;
					$examples[] = [$example, $sublower, $subsum];
				}
			}
			$result[] = [$key, $lower, $sum, $examples];
		}
		return $result;
	}

	private function randomNext($tokens) {
		$cnt = count($tokens);
		for ($size = min($this->size, $cnt); $size >= 0; --$size) {
			$cmb = $this->generalize(array_slice($tokens, $cnt - $size, $size));
			if (isset($this->transitions[$cmb])) {
				$pick = $this->pick($this->transitions[$cmb]);
				if (!is_null($pick)) {
					return $this->exampleOf($pick);
				}
			}
		}
		return '';
	}

	private function pick($options) {
		if (empty($options)) return null;
		$max = end($options)[2];
		$num = random_int(0, $max);
		$min_index = 0;
		$max_index = count($options) - 1;
		while ($min_index < $max_index) {
			$cur_index = intval(($min_index + $max_index) / 2);
			$cur_low = $options[$cur_index][1];
			$cur_high = $options[$cur_index][2];
			if ($cur_low > $num) {
				$max_index = $cur_index;
			} else if ($cur_high < $num) {
				$min_index = $cur_index + 1;
			} else {
				$min_index = $cur_index;
				break;
			}
		}
		return $options[$min_index];
	}

	private function addTransition($state, $next) {
		$cmb = $this->generalize($state);
		if (!isset($this->transitions[$cmb])) {
			$this->transitions[$cmb] = [];
		}
		$this->increment($this->transitions[$cmb], $next);
	}

	private function increment(&$which, $token) {
		$generalized = $this->generalize([$token]);
		if (!isset($which[$generalized])) {
			$which[$generalized] = [
				'count' => 1,
				'examples' => [],
			];
			$which[$generalized]['examples'][$token] = 1;
		} else {
			++$which[$generalized]['count'];
			if (!isset($which[$generalized]['examples'][$token])) {
				$which[$generalized]['examples'][$token] = 1;
			} else {
				++$which[$generalized]['examples'][$token];
			}
		}
	}

	private function tokenize($str) {
		return array_values(array_filter(preg_split('/\b/u', $str)));
	}

	private function generalize($tokens) {
		$str = '';
		foreach ($tokens as $token) {
			$replaced = preg_replace('/\d+/', '0', $token);
			$replaced = strtolower($replaced);
			$str .= empty($replaced) ? $token : $replaced;
		}
		return $str;
	}

	private function exampleOf($pick) {
		$example = $this->pick($pick[3]);
		return $example[0];
	}

	private $size = 5;
	private $transitions = [];

}