<?php


namespace App\Services;


use Exception;
use ErrorException;
use Log;
use QL\QueryList;

class XiurenjiService
{
    public $domainUrl = "https://www.xiurenji.com";
    public $xiurenRootUrl = "https://www.xiurenji.com/XiuRen/";
//    public $rootDir = "/Users/shixuesen/Documents/tmp/xiuren/";
    public $rootDir = "/Volumes/Backup/images/xiuren/";
    public $queryInstance;
    public function __construct()
    {
        $this->queryInstance = QueryList::getInstance();
    }

    public function scrapeAlbum()
    {
        $pageSize = 20;
        $pageCount = $this->getEncodeHtmlContent("https://www.xiurenji.com/XiuRen/index.html")->find(".page span")->htmls()->get(0);
        print_r($pageCount);
        if ((int)$pageCount > 0) {
            for ($i = 0; $i <= ceil($pageCount / $pageSize); $i++) {
                $urlSuffix = "";
                if ($i == 0) {
                    $urlSuffix = "index.html";
                } else {
                    $urlSuffix = "index" . $i . ".html";
                }
                $this->scrapePageAlbum($this->xiurenRootUrl . $urlSuffix);
//                exit;
            }
        }

    }

    public function scrapePageAlbum($url) {

        $pageContent = $this->getEncodeHtmlContent($url);
//        dump($pageContent);
        $items = $pageContent->find(".dan a")->getElements();
        foreach ($items as $item) {
            dump($item->getAttribute("href"));
            $this->scrapeSingleAlbum($this->domainUrl . $item->getAttribute("href"));
        }

    }

    public function scrapeSingleAlbum($url) {
        $pageContent = $this->getEncodeHtmlContent($url);
        $pageSize = 3;
        $items = $pageContent->find(".ina p:nth-child(2)")->texts();
        $pageItems = $pageContent->find(".page a:eq(-2)")->htmls();
        if (count($pageItems) <= 0) {
            dump("this album is error: ". $url);
            Log::error("this album is error: " . $url);
            return;
        }
//        dump($pageItems);exit;
        foreach ($items as $item) {
//            $imageNum = substr($item, strrpos($item, "[") + 1,  strrpos($item, "P]") - strrpos($item, "[") - 1);
//            $pageCount = ceil($imageNum / $pageSize);
            $pageCount = (int)$pageItems[0];
//            dump("pageCount: ". $pageCount . "");exit;
            $slashPos = strpos($url, "XiuRen/") + 7;
            $dotPos = strrpos($url, ".");
            $albumCode = substr($url, $slashPos, $dotPos - $slashPos);
//            $albumName0 = substr($item, strpos($item, "["), strrpos($item, "]") - strpos($item, "[") + 1);
            $albumName = ltrim(substr($item, 6, strrpos($item, "]") - 5));
            if (mb_strlen($albumName) <= 12 || mb_strlen($albumName) >= 50 || !str_contains($albumName, "No")) {
                dump("old Album:  " . $albumName);
                $albumName = $pageContent->find(".ina p b:nth-child(2)")->texts();
//                $albumName = ltrim(substr($item, 6, strrpos(substr($item, 0, strrpos($item, "@")), " ") - 5));
                dump("new Album:  " . urldecode($albumName[0]));
                $albumName = urldecode($albumName[0]);
            }
            $imageNo = 1;
            $this->parseContent($this->rootDir. $albumName, $pageContent, $imageNo);
            dump("albumName: ". $albumName);
//            dump("item: ". $item);
//            exit;
            for ($i = 1; $i < $pageCount; $i++) {
                $pageContent = $this->getEncodeHtmlContent($this->xiurenRootUrl . $albumCode . "_" . $i . ".html");
                $this->parseContent($this->rootDir. $albumName, $pageContent, $imageNo);
            }
//            dump("slashPos: " . $slashPos . " dotPos: " . $dotPos . " albumCode: ". $albumCode);
//            dump($item);
//            exit;
        }
    }

    public function parseContent($dir, $pageContent, &$imageNo)
    {
        if (!is_dir($dir)) {
            try {
                mkdir($dir);
            } catch (Exception $e) {
                Log::error($e->getTraceAsString());
                return;
            }
        }
        $images = $pageContent->find(".img p img")->getElements();
//        dump($images);exit;
        foreach ($images as $image) {
            usleep(random_int(10, 100) * 100);
            $imageUrl = $image->getAttribute("src");
            $trueImageUrl = "https://x1.plmn5.com/U". substr($imageUrl, 2);
            $trueImageUrl = "https://www.xiurenji.com" .  $imageUrl;
            $fileInfo = pathinfo($trueImageUrl);
            if (file_exists($dir . "/" .$fileInfo["basename"])) {
                rename($dir . "/" . $fileInfo["basename"], $dir . "/" . $imageNo . "-" . $fileInfo["basename"]);
                $imageNo++;
                continue;
            }
            if (file_exists($dir . "/" . $imageNo . "-" . $fileInfo["basename"])) {
                $imageNo++;
                continue;
            }
            dump($fileInfo);
            $attempts = 0;
            $content = "";
            do {
                try {
                    $curl_handle=curl_init();
                    curl_setopt($curl_handle, CURLOPT_URL, $trueImageUrl);
                    curl_setopt($curl_handle, CURLOPT_CONNECTTIMEOUT, 2000);
                    curl_setopt($curl_handle, CURLOPT_RETURNTRANSFER, 1);
                    curl_setopt($curl_handle, CURLOPT_USERAGENT, 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/72.0.3626.109 Safari/537.36');
                    curl_setopt($curl_handle, CURLOPT_REFERER, $this->xiurenRootUrl);
                    curl_setopt($curl_handle, CURLOPT_FOLLOWLOCATION, true);
                    $content = curl_exec($curl_handle);
                    if ($content === false) {
                        $le = new Exception("get image has error: " . curl_error($curl_handle));
                        curl_close($curl_handle);
                        throw $le;
                    }
                    curl_close($curl_handle);
//                    $content = file_get_contents($trueImageUrl);
                } catch (ErrorException | Exception $e) {
                    echo $e->getTraceAsString() . "\n";
                    $sleepTime = 10000 * random_int(100, 1000);
                    echo "wait for $trueImageUrl sleep {$sleepTime} nano second \n";
                    usleep($sleepTime);
                    $attempts ++;
                    continue;
                }
                break;
            } while($attempts < 100);
            if ($content != "") {
                file_put_contents($dir . "/" . $imageNo . "-" . $fileInfo["basename"], $content);
            } else {
                Log::error("image content is empty ". $trueImageUrl);
            }
            $imageNo++;
//            dump("current imageNo: " . $imageNo);
        }

    }

    public function getEncodeHtmlContent($url) {
        $attempts = 0;
        $html = "";
        do {
            try {
                $html = iconv('gb2312','UTF-8//IGNORE',file_get_contents($url));
            } catch (Exception $e) {
                echo $e->getTraceAsString() . "\n";
                $sleepTime = 10000 * random_int(100, 1000);
                echo "sleep {$sleepTime} nano second \n";
                usleep($sleepTime);
                $attempts ++;
                continue;
            }
            break;
        } while($attempts < 100);
        return $this->queryInstance->setHtml($html);
    }

}