wangyating/wandoujia.php

<?php
require 'vendor/autoload.php';

use PhpOffice\PhpSpreadsheet\Spreadsheet;
use PhpOffice\PhpSpreadsheet\Writer\Xlsx;
use QL\QueryList;

$urls=getList();
$apps=[];
foreach ($urls as $url) {
    preg_match("/\d+/",$url,$catid);
    for ($i=1;$i<20;$i++){
        $data=getPages($catid[0],$i);
        if ($data){
            $apps=array_merge($apps,$data);
        }
    }
}

$spreadsheet = new Spreadsheet();
try {
    $title="豌豆荚";
    $sheet = $spreadsheet->getActiveSheet();
    $sheet->setTitle($title);
    $sheet->fromArray($apps);
    $writer = new Xlsx($spreadsheet);
    $writer->save('/www/ai/'.$title.".xlsx");
    $spreadsheet->disconnectWorksheets();
} catch (\PhpOffice\PhpSpreadsheet\Exception $e) {
}



function getPages($catid,$page){
    echo "正在处理：$catid\t$page".PHP_EOL;
    $url="https://www.wandoujia.com/wdjweb/api/category/more?catId={$catid}&subCatId=0&page=$page&ctoken=35CLqdFYoYWt80ByPrajl22L";
    $html=json_decode(file_get_contents($url),true)['data']['content'];
    return getListByHtml($html);

}

function getListByHtml($html){
    $rules = array (
        'title' =>
            array (
                0 => 'h2>a',
                1 => 'text',
            ),
        'link' =>
            array (
                0 => 'h2>a',
                1 => 'href',
            ),
    );
    $data=(new QL\QueryList)->html($html)->rules($rules)->range('.card')->queryData();
    foreach ($data as &$datum) {
        preg_match("/\d+/",$datum['link'],$appid);
        $appid=$appid[0];
        $datum['down_url']="https://www.wandoujia.com/apps/{$appid}/download/dot?ch=detail_normal_dl";
    }
    return $data;
}


function getFirstAppList($url){
    $rules = array (
        'title' =>
            array (
                0 => 'h2>a',
                1 => 'text',
            ),
        'link' =>
            array (
                0 => 'h2>a',
                1 => 'href',
            ),
    );
    $data = QueryList::get($url)->rules($rules)->range('.app-box>.card')->queryData();
    return $data;
}


function getList(){
    $data = (new QL\QueryList)->get('https://www.wandoujia.com/category/app?pos=w/crumb/appcategory')->find('.cate-link')->attrs('href');
    return $data->all();
}