information-system/application/third_party/htmlcompressor/controllers/index_gm.php

<?php

if (!defined('BASEPATH')) exit('No direct script access allowed');

class Index_gm extends CI_Controller
{

    public function __construct(){
        parent::__construct();
        $this->load->library('simple_html_dom_lib');
    }

    public function index(){
        $this->load->view('welcome');
    }
	
	//选择处理方式
	public function choose_way(){
		$this->htmlsource = $this->input->post('htmlsource');
        $this->websitehost = $this->input->post('websitehost');
        $this->template_name = $this->input->post('template_name');
        $this->create_amp = $this->input->post('create_amp');
        $this->debug = $this->input->post('debug');
		
		/*$this->htmlsource = file_get_contents('aa.htm');
		$this->websitehost = 'https://data.chinarundreisen.com';
		$this->create_amp = 'true';
		$this->template_name = 'gm';*/
		
		if (empty($this->htmlsource) || empty($this->websitehost)) {
            $this->output->set_status_header(500);
            echo 'error:htmlsource or websitehost is empty!';
            log_message('error', "htmlsource or websitehost is empty! " . $this->websitehost);
            return false;
        }
		
		//域名后面不能有/
        if (substr($this->websitehost, -1, 1) == '/') {
            $this->websitehost = substr($this->websitehost, 0, -1);
        }
		
		//进行页面解析
		$this->html_object = str_get_html($this->htmlsource);
		
		//PC或AMP的构造方式
		if (!empty($this->create_amp)) {
            echo $this->create_amp();
            return;
        }else{
			echo $this->optimize();
			return;
		}
		
	}
	
	//html优化处理
    function optimize(){
		//提取css样式地址
		$link_css_array = array();
		$css_content = '';
		foreach ($this->html_object->find('link') as $link_css) {
			//将所有的样式链接存储为一个数组
			if ($link_css->rel == 'stylesheet' && !empty($link_css->href)) {
				$link_css_array[] = $link_css->href;
				$link_css->outertext = ''; //删除链接
			}
		}
		
		//循环下载所有的css样式，拼接成字符串
		foreach ($link_css_array as $item) {
			$get_http_temp = GET_HTTP($this->format_url($item, $this->websitehost));
			if ($get_http_temp == false) {
				$this->output->set_status_header(404);
				echo 'CSS文件下载错误';
				log_message('error', "optimize CSS文件下载错误! " . $item);
				return FALSE;
			}
			$css_content .= $get_http_temp;
		}

		//找出所有style写的内置样式
		foreach ($this->html_object->find('style') as $style_css) {
			if ($style_css->type == "text/css") {
				$css_content .= $style_css->innertext;
			}
		}
		
		//提取和下载所有JS脚本，包括链接文件和页面脚本
		$link_js_array = array();
		$js_inline_content = '';
		$js_jquery_content = '';
		foreach ($this->html_object->find('script') as $link_script) {
			if (!empty($link_script->src)) {
				if (empty($link_script->defer) && empty($link_script->async)) { //没有defer标记的才处理，否则让js延后加载
					$link_js_array[] = $link_script->src;
					$link_script->outertext = ''; //删除链接，移动到页底
				}
			} else {
				//网页内的js不需要提取
				//$js_content.= $link_script->innertext;//js的内容
				// $js_content.= $link_script;//js的内容，包含<script>标签
				//含有$表示调用了jquery的函数，添加延迟加载defer
				//包含gtm的js引入就跳出本次循环
				if(strpos($link_script, 'https://www.googletagmanager.com/gtm.js')){
					continue;
				}
				if (strpos($link_script, '$') !== FALSE) {
					$js_jquery_content .= $link_script->innertext;
				} else {
					$js_inline_content .= $link_script; //js的内容，包含<script>标签
				}
				$link_script->outertext = ''; //删除js，移动到页底
			}
		}
		
		$js_scr_content = '';
		foreach ($link_js_array as $item) {
			$get_http_temp = GET_HTTP($this->format_url($item, $this->websitehost));
			if ($get_http_temp == false) {
				$this->output->set_status_header(404);
				echo 'JS文件下载错误';
				log_message('error', "optimize js文件下载错误! " . $item);
				return FALSE;
			}
			$js_scr_content .= $get_http_temp;
		}
		
		//把网页内容和css提交到purifycss处理
		$purifycss_server = 'http://184.172.113.216:33033/';
		if (strpos($this->websitehost, 'chinarundreisen.com') !== false) {//德语站点使用自己的css处理服务器
			$purifycss_server = 'http://158.177.67.52:33033/';
		}
		$optimize_css = GET_HTTP($purifycss_server, 'html_source=' . urlencode($this->htmlsource) . '&html_css=' . urlencode($css_content), 'POST');
		if (empty($optimize_css)) {
			$this->output->set_status_header(500);
			echo 'css精简错误';
			return FALSE;
		}
		
		//把精简的css添加到head前面
		$this->html_object = str_replace('</head>', '<style type="text/css">' . $optimize_css . "</style></head>", $this->html_object);
		//删除多余空格和换行符
		$this->html_object = str_replace(array("    ", "\t", "\n", "\r"), " ", $this->html_object);
		//循环n次，把双空格替换为一个空格
		for ($i = 0; $i <= 4; $i++) {
			$this->html_object = str_replace("  ", " ", $this->html_object);
		}

		//在最后加载原始css文件和js文件
		//把css移动到页面底部，延迟加载
		$lastload_js = '<noscript id="deferred-styles">';
		foreach ($link_css_array as $item) {
			$lastload_js .= '<link rel="stylesheet" type="text/css" href="' . $item . '"/>';
		}
		$lastload_js .= '</noscript><script>var loadDeferredStyles=function(){var addStylesNode=document.getElementById("deferred-styles");var replacement=document.createElement("div");replacement.innerHTML=addStylesNode.textContent;document.body.appendChild(replacement);addStylesNode.parentElement.removeChild(addStylesNode)};var raf=requestAnimationFrame||mozRequestAnimationFrame||webkitRequestAnimationFrame||msRequestAnimationFrame;if(raf){raf(function(){window.setTimeout(loadDeferredStyles,0)})}else{window.addEventListener("load",loadDeferredStyles)};</script>';

		$js_content = $js_scr_content . $js_jquery_content;
		//延迟加载js，需要把返回的js代码保存到一个文件中，然后替换占位符，以便加载js文件
		$lastload_js .= '<script async src="##DEFERSCRIPTDEFER##"></script>';
		$lastload_js .= $js_inline_content;

		$this->html_object = str_replace('</body>', $lastload_js . '</body>', $this->html_object);
		echo json_encode(array('result' => 'ok', 'data' => array('html_object' => $this->html_object, 'js_content' => $js_content)));
    }
	
	//生成AMP版本 用移动端版本生成
    /*
        1.加载站点AMP模板，模板只有导航、头部、底部
        2.删除全部js引用和内链代码
        3.删除全部css引用，内联css放入<style amp-custom>
        4.图片全部替换成amp-img，设置宽高，设置layout="responsive"
        5.去掉所有元素中的style
        6.去掉样式表中的!important
        7.替换读取源文件title description keywords canonical
        8.todo:增加application/ld+json
		要求: 图片文件必须带有域名，否则的获取尺寸会出错，如 https://data.asiahighlights.com/pic/abc.jpg
     */
	function create_amp(){
		//读取站点的AMP模板
        $amp_template = $this->load->view('amp-template/' . $this->template_name, null, true);
		
		//生成移动版本，删除多余元素
        foreach ($this->html_object->find('.hidden-xs') as $hidden_item) {
            $hidden_item->outertext = '';
        }
		
		foreach ($this->html_object->find('title') as $title) {
            $amp_template = str_replace('<!--@TITLE@-->', $title->innertext, $amp_template);
        }
		
		foreach ($this->html_object->find('meta') as $meta) {
            if ($meta->name == 'description') {
                $amp_template = str_replace('<!--@DESCRIPTION@-->', $meta->content, $amp_template);
            }
            if ($meta->name == 'keywords') {
                $amp_template = str_replace('<!--@KEYWORDS@-->', $meta->content, $amp_template);
            }
        }
		
		foreach ($this->html_object->find('script') as $script) {
            $script->outertext = '';
        }
		
		foreach ($this->html_object->find('noscript') as $noscript) {
            $noscript->outertext = '';
        }
		
		foreach ($this->html_object->find('link') as $link) {
            if ($link->rel == 'stylesheet' && !empty($link->href)) {
                $link->outertext = '';
            }
            if ($link->rel == 'canonical' && !empty($link->href)) {
                $amp_template = str_replace('<!--@CANONICAL@-->', $link->href, $amp_template);
            }
        }
		
		//获取图片尺寸，AMP要求必须填写width和height
        //格式化图片URL
        foreach ($this->html_object->find('img') as $image) {
            if (!empty($image->src)) {
                $img_src_host = parse_url($image->src, PHP_URL_HOST);
                if (empty($img_src_host)) {
                    $img_src_host = $this->websitehost;
                }
                $image->src = $this->format_url($image->src, $img_src_host);
            }
            if (!empty($image->originalsrc)) {
                $img_src_host = parse_url($image->src, PHP_URL_HOST);
                if (empty($img_src_host)) {
                    $img_src_host = $websitehost;
                }
                $image->originalsrc = $this->format_url($image->originalsrc, $img_src_host);
            }
        }

        //拼接请求的图片url，如 https://data.asiahighlights.com/imagesize.php?photo=/pic/logo-ah.png,/pic/ah-slide-logo.png
        $request_size = array();
        foreach ($this->html_object->find('img') as $image) {
            $img_src = $image->src;
            if (!empty($image->originalsrc)) {
                $img_src = $image->originalsrc;
            }
            if (strpos($img_src, 'facebook.com') !== false  || strpos($img_src, 'dmca.com') !== false) {//facebook dmca会添加一个跟踪代码，需要删除掉，否则获取图片尺寸会报错
                $image->outertext = '';
                continue;
            }
            $img_src_urls = parse_url($img_src);
            $request_size[$img_src_urls['host']][] = $img_src_urls['path'];
        }

        //请求图片尺寸
        $image_sizes = array();
        foreach ($request_size as $host => $path) {
            $parse_url = "https://{$host}/imagesize.php?photo=" . urlencode(implode(',', $path));
            $size_data = GET_HTTP($parse_url);
            if (!empty($size_data)) {
                $size_data = json_decode($size_data);
                foreach ($size_data as $size_item) {
                    $size_item->photo = "https://{$host}" . $size_item->photo;
                    $image_sizes[$size_item->photo] = $size_item;
                }
            }
        }

        //获取图片尺寸 end
        foreach ($this->html_object->find('img') as $image) {
            $img_src = $image->src;
            if (!empty($image->originalsrc)) {
                $img_src = $image->originalsrc;
            }
            if(empty($image_sizes[$img_src])){
                //如果尺寸不存在则跳过
                continue;
            }
            $img_size = $image_sizes[$img_src];
            $img_alt = $image->alt;
            $img_class = $image->class;
            if (!empty($img_size)) {
				//添加webp图片支持
				$img_src_webp='';
				$layout = 'responsive';
				if(substr($img_src, -4) == '.jpg'){
					if($img_size->width < 400){
						$layout = 'fixed';
					}
					$img_src_jpg=" <amp-img fallback layout=\"{$layout}\" class=\"{$img_class}\" alt=\"{$img_alt}\" src=\"{$img_src}\" width=\"{$img_size->width}\" height=\"{$img_size->height}\"></amp-img>";
					
					$img_src_webp=substr($img_src, 0, strlen($img_src) - 4).'.webp';
					
					$img_src_webp = " <amp-img layout=\"{$layout}\" class=\"{$img_class}\" alt=\"{$img_alt}\" src=\"{$img_src_webp}\" width=\"{$img_size->width}\" height=\"{$img_size->height}\">{$img_src_jpg}</amp-img>";
				}
				if(!empty($img_src_webp)){
					$image->outertext =$img_src_webp;
				}else{
					$image->outertext = " <amp-img layout=\"{$layout}\" class=\"{$img_class}\" alt=\"{$img_alt}\" src=\"{$img_src}\" width=\"{$img_size->width}\" height=\"{$img_size->height}\"></amp-img>";
				}
            }
        }
		
		//只将中间主要部分用于css精简，提高速度
		foreach ($this->html_object->find('#content_main') as $content_main) {
            $amp_template = str_replace('<!--@CUSTOM-CONENT@-->', $content_main, $amp_template);
        }
		
		//产品页面多做一些特殊的处理
		foreach ($this->html_object->find('.tab-nav') as $tabnav) {
            $amp_template = str_replace($tabnav, '', $amp_template);
        }
		
		//产品页面删除表单
		foreach ($this->html_object->find('.sticky') as $sticky) {
            $amp_template = str_replace($sticky, '@AMPFOOTER-BUTTON@', $amp_template);
        }
		
		foreach ($this->html_object->find('form') as $form) {
            $amp_template = str_replace($form, '', $amp_template);
        }
		
		if(!empty($form)){
			$replace_text = '<script async="" custom-element="amp-form" src="https://cdn.ampproject.org/v0/amp-form-0.1.js"></script><style amp-custom>';
			$amp_template = str_replace('<style amp-custom>',$replace_text, $amp_template);
		}
		
		//产品页面处理结束
		
		$link_css_array = array();
		$css_content = '';
		foreach ($this->html_object->find('link') as $link_css) {
			//将所有的样式链接存储为一个数组
			if ($link_css->rel == 'stylesheet' && !empty($link_css->href)) {
				$link_css_array[] = $link_css->href;
				$link_css->outertext = ''; //删除链接
			}
		}
		
		//循环下载所有的css样式，拼接成字符串
		foreach ($link_css_array as $item) {
			//排除一定不会用到的css，提高精简的速度
			if(strpos($item,'base.css')){
				$item = 'https://data.chinarundreisen.com/min/?f=/css/base.css';
			}
			
			if(strpos($item,'reiseroute.css')){
				$item = 'https://data.chinarundreisen.com/min/?f=/css/reiseroute-amp.css';
			}
			
			$get_http_temp = GET_HTTP($this->format_url($item, $this->websitehost));
			if ($get_http_temp == false) {
				$this->output->set_status_header(404);
				echo 'CSS文件下载错误';
				log_message('error', "optimize CSS文件下载错误! " . $item);
				return FALSE;
			}
			$css_content .= $get_http_temp;
		}
		
		foreach ($this->html_object->find('style') as $style) {
            $css_content .= $style->innertext;
            $style->outertext = '';
        }
		
		//把网页内容和css提交到purifycss处理
		$purifycss_server = 'http://158.177.67.52:33033/';
		$optimize_css = GET_HTTP($purifycss_server, 'html_source=' . urlencode($content_main) . '&html_css=' . urlencode($css_content), 'POST');
		if (empty($optimize_css)) {
			$this->output->set_status_header(500);
			echo 'css精简错误';
			return FALSE;
		}
		
        $optimize_css = str_replace('!important', '', $optimize_css);
        $optimize_css = str_replace('@charset "utf-8";', '', $optimize_css);
        $amp_template = str_replace('/*@CUSTOM-CSS@*/', $optimize_css, $amp_template);

        //替换掉写在组件上的样式
        $amp_template = preg_replace('/style=.+?[\'|"]/i', '', $amp_template);
		echo json_encode(array('result' => 'ok', 'data' => array('amp' => $amp_template)));
	}
	
	//格式化url，保证请求的URL有域名，//更换为对应的域名路径
    function format_url($url, $host = ''){
        if (substr($url, 0, 8) == 'https://' || substr($url, 0, 7) == 'http://') {
            $url = str_replace('http://', 'https://', $url);//http要改为https
            return urldecode($url);
        }

        if (substr($url, 0, 2) == '//') { //https或http
            return urldecode('https://'.substr($url, 2, strlen($url)-2)); //只替换开头的//，部分url不规范，有//在url中间
        }

        return urldecode($host . $url);
    }

}