Tリーグの試合結果からデータを一覧で取得

scrape-tleague-game-stat.php

<?php
$urls=<<<EOF
https://tleague.jp/match/?season=2018&month=201810
https://tleague.jp/match/?season=2018&month=201811
https://tleague.jp/match/?season=2018&month=201812
https://tleague.jp/match/?season=2018&month=201901
https://tleague.jp/match/?season=2018&month=201902
https://tleague.jp/match/?season=2018&month=201903
https://tleague.jp/match/?season=2019&month=201910
https://tleague.jp/match/?season=2019&month=201911
https://tleague.jp/match/?season=2019&month=201912
EOF;


foreach (explode("\n", $urls) as $url){
    $html = getPage(trim($url));
    $dom = new DOMDocument("1.0");
    if (@$dom->loadHTML($html)){
        $xpath = new DOMXPath($dom);
        foreach ($xpath->query('//li/div[@class="field"]') as $k=>$line) {
            if ($k < 1) continue;
            $data = [];
            $props = ['date', 'time', 'sex', 'home', 'result', 'away', 'arena', 'media', 'ticket'];
            $tmp = [];
            foreach ($props as $prop){
                $_d = $xpath->query('div[@class="cell-'.$prop.'"]', $line);
                $d = trim($_d->item(0)->textContent);
                switch ($prop){
                  case 'date':
                    $tmp[$prop] = substr($d, 0, 10);
                    break;
                  case 'time':
                    $d = trim(preg_replace('@（.*?）@', '', $d));
                    $tmp[$prop] = $d;
                    break;
                  case 'sex':
                  case 'home':
                  case 'away':
                  case 'arena':
                    $tmp[$prop] = $d;
                    break;
                  case 'result':
                    $result_link = $xpath->query('div/a', $_d->item(0));
                    $href = $result_link->item(0)->getAttribute('href');
                    $match_url = 'https://tleague.jp' . $href;
                    $page = new DOMDocument("1.0");
                    if (@$page->loadHTML(getPage($match_url))){
                        $xpage = new DOMXPath($page);
                        $text = $xpage->query('//ul[@class="item-spec reset"]//li')->item(2)->textContent;
                        $num = 0;
                        if (preg_match('/入場者数：([\d,]+)人/', $text, $m)){
                            $num = intval(str_replace(',', '', $m[1]));
                        }
                        $tmp[$prop] = $num;
                    }
                    break;
                }
            }
            echo implode(',', $tmp). "\n";
        }
    }
}


function getPage($url)
{
    $cache_file = __DIR__.'/cache/'.md5($url).'.html';
    if (file_exists($cache_file)){
        $html = file_get_contents($cache_file);
    }
    else {
        sleep(1);
        ob_start();
        system('wget -O - "'. $url .'"');
        $html = ob_get_clean();
        file_put_contents($cache_file, $html);
    }
    return $html;
}