<?php

declare(strict_types=1);

class Robot
{
    public string $type;
    public int $count;
    public float $intensity;
    public string $lastAccessed;

    public function __construct(string $type, int $count, float $intensity, string $lastAccessed)
    {
        $this->type = $type;
        $this->count = $count;
        $this->intensity = $intensity;
        $this->lastAccessed = $lastAccessed;
    }
}

class RobotsMonth
{
    public int $year;
    public int $month;
    public RobotsTypeMonth $robots;

    public function __construct(string $markup, int $year, int $month)
    {
        $this->year = $year;
        $this->month = $month;
        $this->robots = new RobotsTypeMonth($markup);
    }
}


/**
 * A class that represents the variation of robots in a month
 */
class RobotsTypeMonth extends ArrayObject
{
    public function __construct(string $markup)
    {
        $this->aggragateRobots($this->processMarkup($markup));
    }

    public function processMarkup(string $markup): array
    {
        $xml = new SimpleXMLElement($markup, LIBXML_NOERROR | LIBXML_ERR_NONE | LIBXML_ERR_FATAL);
        $result = $xml->xpath('//table/tbody/tr');
        return $result;
    }

    public function unescapeString(string $str): string
    {
        $repl = [
            ['[\x20]', ' '],
            ['\-',     '-'],
            ['\;',     ';'],
            ['\/',     '/'],
            ['\.',     '.'],
        ];

        $str = trim($str, "/");
        foreach ($repl as list($s, $r)) {
            $str = str_replace_recursive($s, $r, $str);
        }
        return $str;
    }

    public function aggragateRobots(array $parent): void
    {
        foreach ($parent as $key => $node) {
            // skip header
            if (isset($node->th)) continue;
            $type = $this->unescapeString((string)$node->td[0]);
            $robot = new Robot($type, (int)$node->td[1], (float)$node->td[2], (string)$node->td[3],);
            $this[] = $robot;
        }
    }
}



$years = [
    2018,
    2019,
    2020,
    2021,
    2022,
    2023,
    2024,
];

$monthRange = [
    [1, 12],
    [1, 12],
    [1, 12],
    [1, 12],
    [1, 12],
    [1, 12],
    [1, 04],
];


function str_replace_recursive(string $search, string $replace, string $subject): string
{
    if (strpos($subject, $search) !== false) {
        $subject = str_replace($search, $replace, $subject);
        return str_replace_recursive($search, $replace, $subject);
    }
    return $subject;
}

function getRangeFormatted(array $monthRange, int $i): array
{
    $array = [];
    foreach (range(...$monthRange[$i]) as $m) {
        $array[] = sprintf("%02d", $m);
    }
    return $array;
}

$aggragated = [];
foreach ($years as $i => $y) {
    foreach (getRangeFormatted($monthRange, $i) as $m) {
        $raw = file_get_contents(__DIR__ . "/$y/$m.html");
        $content = mb_convert_encoding($raw, 'utf-8', 'euc-jp');

        $robots = new RobotsMonth($content, (int)$y, (int)$m);
        $aggragated[] = $robots;
    }
}

$counter = [];

foreach ($aggragated as $m) {
    assert($m instanceof RobotsMonth);
    foreach ($m->robots as $robot) {
        if (!array_key_exists("$robot->type", $counter)) {
            $counter["$robot->type"] = 0;
        }
        $counter["$robot->type"] += $robot->count;
    }
}

unset($counter['no_user_agent']);
unset($counter['wordpress']);

printf("%-9s\0%s\0%s\n", "COUNT", "PERCENTAGE", "USER-AGENT",);

$sum = array_sum(array_values($counter));
foreach ($counter as $k => $v) {
    printf("%-9d\0%-6.1f\0 %s\n", $v, $v / $sum * 100 , $k,);
}