grab article content

grab_article.php

<?php
error_reporting(-1);

$parent_nodes = array();

$html  = file_get_contents("http://www.csdn.net/article/2013-01-09/2813524-CSDN-morning-paper");

$doc = new DOMDocument();

$doc->encoding = "utf-8";

try {

    @$doc->loadHTML($html);

	$ptext = $doc->getElementsByTagName("p");
	
} catch (Exception $e) {
    // not do anything
}

foreach($ptext as $item)
{
	$parent_node = $item->parentNode;
	$content_score = intval($parent_node->getAttribute("contentScore"));
	$class_name = $parent_node->getAttribute("class");
    $id = $parent_node->getAttribute("id");

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

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

    // // Add points for any commas within this paragraph
    if (strlen($item->nodeValue) > 10)
    {
        $content_score += strlen($item->nodeValue);
    }

    $parent_node->setAttribute("contentScore", $content_score);

    array_push($parent_nodes, $parent_node);
}

$box = $doc->createElement('div', '');

for ($i = 0, $len = count($parent_nodes); $i < $len; $i++)
{
    $parent_node = $parent_nodes[$i];
    $content_score = intval($parent_node->getAttribute("contentScore"));
    $org_content_score = intval($box->getAttribute("contentScore"));

    if($content_score && $content_score > $org_content_score)
    {
        $box = $parent_node;
    }
}

$content = $doc->saveHTML($box);

echo $content;