queryList/app/Service/WxCrawler.php

<?php


namespace App\Service;

use App\Log;
use QL\QueryList;

class WxCrawler
{
    //微信内容div正则
    private $wxContentDiv = '/<div class="rich_media_content.*?">(.*?)<\/div>/s';

    //微信图片样式
    private $imageStyle = 'style="height: auto !important;visibility: visible !important;"';

    /**
     * 爬取内容
     * @param  $url
     * @return false|string
     */
    private function _get($url)
    {
        return file_get_contents($url);
    }

    public function crawByUrl($url)
    {
        $content = $this->_get($url);
        $basicInfo = $this->articleBasicInfo($content);

        $content_html = $this->contentHandle($content);
        return array_merge($basicInfo,['content' => $content_html]);
    }
    /**
     * 处理微信文章源码
     * @param  $content
     * @return [带图html文本，无图html文本]
     */
    private function contentHandle($content)
    {
        $content_html_pattern = $this->wxContentDiv;
        preg_match_all($content_html_pattern, $content, $html_matches);
        if(empty(array_filter($html_matches))) {
            return [];
        }
        $content_html = $html_matches[1][0];

        //去除掉hidden隐藏
        $content_html = str_replace('style="visibility: hidden;"','',$content_html);

        //过滤掉iframe
//        $content_html = preg_replace('/<iframe(.*?)<\/iframe>/','',$content_html);

        // 过滤掉voice
        $content_html = preg_replace('/<mpvoice(.*?)<\/mpvoice>/','',$content_html);


        // 处理图片
//        Log::logInfo('图片处理开始'.date('Y-m-d H:i:s'), [], 'getDetailByLink');
        $content_html = preg_replace_callback('/(<img.*?src=[\'|\"])(.*?(?:[\'|\"]))(.*?[\/]?>)/', function($matches) {
//            return $matches[1] . $this->getImg($matches[2]).'" '.$this->imageStyle . $matches[3];
            return $matches[1] . urldecode(htmlspecialchars_decode($matches[2])) . " " . $this->imageStyle . ' src="' . urldecode(htmlspecialchars_decode($matches[2])) . ' ' .$matches[3];
        }, $content_html);
//        Log::logInfo('图片处理结束'.date('Y-m-d H:i:s'), [], 'getDetailByLink');
        $content_html = preg_replace_callback('/background-image: url\((.*?)\)/', function($matches) {
            return 'background-image: url(' . $this->getImg($matches[1]).') ';
//            return 'background-image: url(' . urldecode(htmlspecialchars_decode(($matches[1]))).') ';
        }, $content_html);
        $content_html = preg_replace_callback('/background: url\((.*?)\)/', function($matches) {
            return 'background: url(' . $this->getImg($matches[1]).') ';
//            return 'background-image: url(' . urldecode(htmlspecialchars_decode(($matches[1]))).') ';
        }, $content_html);
        // 处理视频封面
        $content_html = preg_replace_callback('/data-cover="(.*?)"/', function($matches) {
//            return 'data-cover="' . $this->getImg($matches[1]).'" '.$this->imageStyle;
            return 'data-cover="' . urldecode(htmlspecialchars_decode(($matches[1]))). '" ' .$this->imageStyle;
        }, $content_html);
        // 处理音频
//        $content_html = preg_replace_callback('/voice_encode_fileid="(.*?)"/', function($matches) {
//            return 'src="' . config('link.voice_media').$matches[1] .'" ';
//        }, $content_html);

//        $content_html = preg_replace_callback('/<mpvoice .*?(\t|\r\n|\n|\s)*<\/mpvoice>/is', function($matches) {
//            $mpVoice = urldecode($matches[0]);
//            preg_match('/voice_encode_fileid="(.*?)"/', $mpVoice, $voiceSrc);
//            preg_match('/name="(.*?)"/', $mpVoice, $voiceTitle);
//            $title = isset($voiceTitle[1]) ? $voiceTitle[1] : '-';
//            preg_match('/src="(.*?)"/', $mpVoice, $voiceLength);
//            $voiceLength = htmlspecialchars_decode($voiceLength[1]);
//            $info = parse_url($voiceLength);
//            parse_str($info['query'], $outPut);
//            $playLength = isset($outPut['play_length']) ? $outPut['play_length'] : '00:00';
//            $voiceHtmlNew = '<div class="audio" style="background: #f7f7f7;position: relative;border-radius: 8px;padding: 20px;line-height: 1.4;">
//    <img class="icon" style="width: 30px;height: 30px;display: inline-block;vertical-align: middle;" src="./voice.svg" alt="">
//    <div class="info" style="width: calc(100% - 40px);display: inline-block;vertical-align: middle;padding-left: 10px;box-sizing: border-box;" voice_encode_fileid="'. $voiceSrc[1] .'">
//      <h3 style="margin: 0;padding: 0;display: block;font-weight: 700;font-size: 17px;color: rgba(0,0,0,0.9);overflow: hidden;text-overflow: ellipsis;display: -webkit-box;-webkit-box-orient: vertical;-webkit-line-clamp: 2;word-wrap: break-word;-webkit-hyphens: auto;">'. $playLength .'</h3>
//      <span class="desc" style="display: block;font-size: 12px;color: rgba(0,0,0,0.5);width: auto;overflow: hidden;text-overflow: ellipsis;white-space: nowrap;word-wrap: normal;padding: 4px 72px 4px 0;">' .$title. '</span>
//    </div>
//  </div>';
//            return $voiceHtmlNew;
//        }, $content_html);

        //添加微信样式
//        var_dump(htmlspecialchars_decode($content_html));die;
        $content_html = '<div class="rich_media_content" style="max-width: 677px;margin-left: auto;margin-right: auto;">'.$content_html. '</div>';

        return $content_html;
    }
    /**
     * 获取文章的基本信息
     * @param  $content
     * @return $basicInfo
     */
    private function articleBasicInfo($content)
    {
        $item = [
            'title' => 'title',         //标题
            'description' => 'digest',         //描述
            'msg_source_url' => 'content_url',    //文章链接
            'image' => 'cover',       //封面图片链接
        ];
        $basicInfo = [];
        foreach ($item as $k => $v) {
            if($k == 'msg_source_url')
                $pattern = '/var '.$k.' = \'(.*?)\';/s';
            else
                $pattern = '<meta property="og:'.$k.'" content="(.*?)" />';
            preg_match_all($pattern,$content,$matches);
            if(array_key_exists(1, $matches) && !empty($matches[1][0])){
                $basicInfo[$v] = $this->htmlTransform($matches[1][0]);
            }else{
                $basicInfo[$v] = '';
            }
        }
        return $basicInfo;
    }
    /**
     * 特殊字符转换
     * @param  $string
     * @return $string
     */
    private function htmlTransform($string)
    {
        $string = str_replace('&quot;','"',$string);
        $string = str_replace('&amp;','&',$string);
        $string = str_replace('amp;','',$string);
        $string = str_replace('&lt;','<',$string);
        $string = str_replace('&gt;','>',$string);
        $string = str_replace('&nbsp;',' ',$string);
        $string = str_replace("\\", '',$string);
        $string = str_replace("x26", '&',$string);
        return $string;
    }

    /**
     * @param $url
     * @return string
     */
    private function getImg($url){
        $url = htmlspecialchars_decode($url);
        $url = trim($url, "'");
        $url = trim($url, '"');
        $refer = "http://www.qq.com/";
        $opt = [
            'http'=>[
                'header'=>"Referer: " . $refer
            ]
        ];

        if(strstr($url, 'mpvideo')) return $url;
        $context = stream_context_create($opt);
        //接受数据流
        $ossUrl = '';
        $file_contents = file_get_contents(urldecode($url),false, $context);

        $path = '/mnt/queryList/public/upload/';
        $fileType = self::getImgType($url);
        $fileName = time().rand(0,99999) . '.'.$fileType;
        $filePath = $path . $fileName;
        if(strstr($url, 'wx_fmt=svg') !== false) {
            $im = new \Imagick($url);
            $svg = '<?xml version="1.0" encoding="UTF-8" standalone="no"?>'.$file_contents;
            $im->readImageBlob($svg);
            $im->setImageFormat($fileType);
            $srcImage = $im->getImageGeometry(); //获取源图片宽和高
            $im->resizeImage($srcImage['width'], $srcImage['height'], \imagick::FILTER_LANCZOS, 1, false);
            $im->writeImage($filePath);
            $im->clear();
            $im->destroy();
        } else {
            if(strstr($url, 'wx_fmt=gif') === false) {
                $imageSteam =  Imagecreatefromstring($file_contents);

                if(!file_exists($path))
                    mkdir($path,0777,true);
                //生成新图片
                imagejpeg($imageSteam, $filePath);
            } else {
                $im = new \Imagick($url);
                $im->writeImages($filePath, true);
                $im->clear();
                $im->destroy();
            }

        }
        if(file_exists($filePath)) {
            $ossClient = new OssServices('weixiaomeng');
            // 上传阿里云
            $ossFile = $ossClient->upload($fileType, $filePath, 'upload/weixiaomeng/' . date("Y-m-d", time()));
            unlink($filePath);
            $ossUrl = $ossFile['oss-request-url'];
        }
        return empty($ossUrl) ? $url : $ossUrl;
    }

    /*
     *  获取图片类型
     * */
    public static function getImgType($url)
    {
        $info = getimagesize($url);
        $suffix = false;
        if($mime = $info['mime']){
            $suffix = explode('/',$mime)[1];
        }
        return $suffix;
    }
}