php提取网页正文内容的例子

作者:袖梨 2022-06-24

Html2Article-php实现的提取网页正文部分,最近研究百度结果页的资讯采集,其中关键环节就是从采集回的页面中提取出文章。

因为难点在于如何去识别并保留网页中的文章部分,而且删除其它无用的信息,并且要做到通用化,不能像火车头那样根据目标站来制定采集规则,因为搜索引擎结果中有各种的网页。

抓回一个页面的数据,如何匹配出正文部分,郑晓在下班路上想了个思路是:

1. 提取出body标签部分–>剔除所有链接–>剔除所有script、注释–>剔除所有空白标签(包括标签内不含中文的)–>获取结果。
2. 直接匹配出非链接的、 符合在div、p、h标签中的中文部分???
还是会有不少其它多余信息啊,比如底部信息等。。 如何搞?不知道大家有木有什么思路或建议?

这个类是从网上找到的一个php实现的提取网页正文部分的算法,郑晓在本地也测试了下,准确率非常高。

 代码如下 复制代码

class Readability {
    // 保存判定结果的标记位名称
    const ATTR_CONTENT_SCORE = "contentScore";

    // DOM 解析类目前只支持 UTF-8 编码
    const DOM_DEFAULT_CHARSET = "utf-8";

    // 当判定失败时显示的内容
    const MESSAGE_CAN_NOT_GET = "Readability was unable to parse this page for content.";

    // DOM 解析类(PHP5 已内置)
    protected $DOM = null;

    // 需要解析的源代码
    protected $source = "";

    // 章节的父元素列表
    private $parentNodes = array();

    // 需要删除的标签
    // Note: added extra tags from http://www.111com.net

    private $junkTags = Array("style", "form", "iframe", "script", "button", "input", "textarea",
                                "noscript", "select", "option", "object", "applet", "basefont",
                                "bgsound", "blink", "canvas", "command", "menu", "nav", "datalist",
                                "embed", "frame", "frameset", "keygen", "label", "marquee", "link");

    // 需要删除的属性
    private $junkAttrs = Array("style", "class", "onclick", "onmouseover", "align", "border", "margin");


    /**
     * 构造函数
     *      @param $input_char 字符串的编码。默认 utf-8,可以省略
     */
    function __construct($source, $input_char = "utf-8") {
        $this->source = $source;

        // DOM 解析类只能处理 UTF-8 格式的字符
        $source = mb_convert_encoding($source, 'HTML-ENTITIES', $input_char);

        // 预处理 HTML 标签,剔除冗余的标签等
        $source = $this->preparSource($source);

        // 生成 DOM 解析类
        $this->DOM = new DOMDocument('1.0', $input_char);
        try {
            //libxml_use_internal_errors(true);
            // 会有些错误信息,不过不要紧 :^)
            if (!@$this->DOM->loadHTML(''.$source)) {
                throw new Exception("Parse HTML Error!");
            }

            foreach ($this->DOM->childNodes as $item) {
                if ($item->nodeType == XML_PI_NODE) {
                    $this->DOM->removeChild($item); // remove hack
                }
            }

            // insert proper
            $this->DOM->encoding = Readability::DOM_DEFAULT_CHARSET;
        } catch (Exception $e) {
            // ...
        }
    }


    /**
     * 预处理 HTML 标签,使其能够准确被 DOM 解析类处理
     *
     * @return String
     */
    private function preparSource($string) {
        // 剔除多余的 HTML 编码标记,避免解析出错
        preg_match("/charset=([\w|\-]+);?/", $string, $match);
        if (isset($match[1])) {
            $string = preg_replace("/charset=([\w|\-]+);?/", "", $string, 1);
        }

        // Replace all doubled-up
tags with

tags, and remove fonts.
        $string = preg_replace("/[ \r\n\s]*/i", "

", $string);
        $string = preg_replace("/<\/?font[^>]*>/i", "", $string);

        // @see https://github.com/feelinglucky/php-readability/issues/7
        //   - from http://stackoverflow.com/questions/7130867/remove-script-tag-from-html-content
        $string = preg_replace("#(.*?)#is", "", $string);

        return trim($string);
    }


    /**
     * 删除 DOM 元素中所有的 $TagName 标签
     *
     * @return DOMDocument
     */
    private function removeJunkTag($RootNode, $TagName) {
       
        $Tags = $RootNode->getElementsByTagName($TagName);
       
        //Note: always index 0, because removing a tag removes it from the results as well.
        while($Tag = $Tags->item(0)){
            $parentNode = $Tag->parentNode;
            $parentNode->removeChild($Tag);
        }
       
        return $RootNode;
       
    }

    /**
     * 删除元素中所有不需要的属性
     */
    private function removeJunkAttr($RootNode, $Attr) {
        $Tags = $RootNode->getElementsByTagName("*");

        $i = 0;
        while($Tag = $Tags->item($i++)) {
            $Tag->removeAttribute($Attr);
        }

        return $RootNode;
    }

    /**
     * 根据评分获取页面主要内容的盒模型
     *      判定算法来自:http://code.google.com/p/arc90labs-readability/  
     *      这里由郑晓博客转发
     * @return DOMNode
     */
    private function getTopBox() {
        // 获得页面所有的章节
        $allParagraphs = $this->DOM->getElementsByTagName("p");

        // Study all the paragraphs and find the chunk that has the best score.
        // A score is determined by things like: Number of

's, commas, special classes, etc.
        $i = 0;
        while($paragraph = $allParagraphs->item($i++)) {
            $parentNode   = $paragraph->parentNode;
            $contentScore = intval($parentNode->getAttribute(Readability::ATTR_CONTENT_SCORE));
            $className    = $parentNode->getAttribute("class");
            $id           = $parentNode->getAttribute("id");

            // Look for a special classname
            if (preg_match("/(comment|meta|footer|footnote)/i", $className)) {
                $contentScore -= 50;
            } else if(preg_match(
                "/((^|\\s)(post|hentry|entry[-]?(content|text|body)?|article[-]?(content|text|body)?)(\\s|$))/i",
                $className)) {
                $contentScore += 25;
            }

            // Look for a special ID
            if (preg_match("/(comment|meta|footer|footnote)/i", $id)) {
                $contentScore -= 50;
            } else if (preg_match(
                "/^(post|hentry|entry[-]?(content|text|body)?|article[-]?(content|text|body)?)$/i",
                $id)) {
                $contentScore += 25;
            }

            // Add a point for the paragraph found
            // Add points for any commas within this paragraph
            if (strlen($paragraph->nodeValue) > 10) {
                $contentScore += strlen($paragraph->nodeValue);
            }

            // 保存父元素的判定得分
            $parentNode->setAttribute(Readability::ATTR_CONTENT_SCORE, $contentScore);

            // 保存章节的父元素,以便下次快速获取
            array_push($this->parentNodes, $parentNode);
        }

        $topBox = null;
       
        // Assignment from index for performance.
        //     See http://www.peachpit.com/articles/article.aspx?p=31567&seqNum=5
        for ($i = 0, $len = sizeof($this->parentNodes); $i < $len; $i++) {
            $parentNode      = $this->parentNodes[$i];
            $contentScore    = intval($parentNode->getAttribute(Readability::ATTR_CONTENT_SCORE));
            $orgContentScore = intval($topBox ? $topBox->getAttribute(Readability::ATTR_CONTENT_SCORE) : 0);

            if ($contentScore && $contentScore > $orgContentScore) {
                $topBox = $parentNode;
            }
        }
       
        // 此时,$topBox 应为已经判定后的页面内容主元素
        return $topBox;
    }


    /**
     * 获取 HTML 页面标题
     *
     * @return String
     */
    public function getTitle() {
        $split_point = ' - ';
        $titleNodes = $this->DOM->getElementsByTagName("title");

        if ($titleNodes->length
            && $titleNode = $titleNodes->item(0)) {
            // @see http://stackoverflow.com/questions/717328/how-to-explode-string-right-to-left
            $title  = trim($titleNode->nodeValue);
            $result = array_map('strrev', explode($split_point, strrev($title)));
            return sizeof($result) > 1 ? array_pop($result) : $title;
        }

        return null;
    }


    /**
     * Get Leading Image Url
     *
     * @return String
     */
    public function getLeadImageUrl($node) {
        $images = $node->getElementsByTagName("img");

        if ($images->length && $leadImage = $images->item(0)) {
            return $leadImage->getAttribute("src");
        }

        return null;
    }


    /**
     * 获取页面的主要内容(Readability 以后的内容)
     *
     * @return Array
     */
    public function getContent() {
        if (!$this->DOM) return false;

        // 获取页面标题
        $ContentTitle = $this->getTitle();

        // 获取页面主内容
        $ContentBox = $this->getTopBox();
       
        //Check if we found a suitable top-box.
        if($ContentBox === null)
            throw new RuntimeException(Readability::MESSAGE_CAN_NOT_GET);
       
        // 复制内容到新的 DOMDocument
        $Target = new DOMDocument;
        $Target->appendChild($Target->importNode($ContentBox, true));

        // 删除不需要的标签
        foreach ($this->junkTags as $tag) {
            $Target = $this->removeJunkTag($Target, $tag);
        }

        // 删除不需要的属性
        foreach ($this->junkAttrs as $attr) {
            $Target = $this->removeJunkAttr($Target, $attr);
        }

        $content = mb_convert_encoding($Target->saveHTML(), Readability::DOM_DEFAULT_CHARSET, "HTML-ENTITIES");

        // 多个数据,以数组的形式返回
        return Array(
            'lead_image_url' => $this->getLeadImageUrl($Target),
            'word_count' => mb_strlen(strip_tags($content), Readability::DOM_DEFAULT_CHARSET),
            'title' => $ContentTitle ? $ContentTitle : null,
            'content' => $content
        );
    }

    function __destruct() { }
}


使用起来也非常简单,实例化时传入网页的html源码和相应的编码,然后直接调用其getContent方法即可返回提取到的正文部分,提取出的文章中可能还会含有少部分链接,可以自己后期再修改

相关文章

精彩推荐