Scraping HN Front Page - Handeling Simple HTML Dom Error

Question

I'm using 'Simple HTML Dom' to scrape the HN Front Page (news.ycombinator.com), which works great most of the time.

However, every now and then they promote a job/company that lacks the elements that the scraper is looking for, i.e. score, username and number of comments.

This of course, breaks the array and thus the output of my script:

 <?php

// 2012-02-12 Maximilian (Extract news.ycombinator.com's Front Page)

// Set the header during development
//header ("content-type: text/xml");

// Call the external PHP Simple HTML DOM Parser (http://simplehtmldom.sourceforge.net/manual.htm)
include('lib/simple_html_dom.php');

date_default_timezone_set('Europe/Berlin');

// Download 'news.ycombinator.com' content
//$tmp = file_get_contents('http://news.ycombinator.com');
//file_put_contents('get.tmp', $tmp);

// Retrieve the content
$html = file_get_html('tc.tmp');

// Set the extraction pattern for each item
$title = $html->find("tr td table tr td.title a");
$score = $html->find("tr td.subtext span");
$user = $html->find("tr td.subtext a[href^=user]");
$link = $html->find("tr td table tr td.title a");
$time = $html->find("tr td.subtext");
$additionals = $html->find("tr td.subtext a[href^=item?id]");

// Construct the feed by looping through the items
for($i=0;$i<29;$i++) {

$cr=1;

// Check if the item points to an external website
if (!strstr($link[$i]->href,'http')) {

$url = 'http://news.ycombinator.com/'.$link[$i]->href;
$description = "Join the discussion on Hacker News.";


} else {

$url = $link[$i]->href;

// Getting content here

if (empty($abstract)) {

$description ="Failed to load any relevant content. Please try again later.";

} else {

$description = $abstract;

}

}
// Put all the items together
  $result .= '<item><id>f'.$i.'</id><title>'.htmlspecialchars(trim($title[$i]->plaintext)).'</title><description><![CDATA['.$description.']]></description><pubDate>'.str_replace('  | '.$additionals[$i]->plaintext,'',str_replace($score[$i]->plaintext.' by '.$user[$i]->plaintext.' ','',$time[$i]->plaintext)).'</pubDate><score>'.$score[$i]->plaintext.'</score><user>'.$user[$i]->plaintext.'</user><comments>'.$additionals[$i]->plaintext.'</comments><id>'.substr($additionals[$i]->href,8).'</id><discussion>http://news.ycombinator.com/'.$additionals[$i]->href.'</discussion><link>'.htmlspecialchars($url).'</link></item>'; 
}

$output = '<rss><channel><id>news.ycombinator.com Frontpage</id><buildDate>'.date('Y-m-d H:i:s').'</buildDate>'.$result.'</channel></rss>';

file_put_contents('tc.xml', $output);


?>

Here's an example of the correct output

<item>
<id>f0</id>
<title>Show HN: Bootswatch, free swatches for your Bootstrap site</title>
<description><![CDATA[Easy to Install Simply download the CSS file from the swatch of your choice and replace the one in Bootstrap. No messing around with hex values. Whole New Feel We've all been there with the black bar and blue buttons. See how a splash of color and typography can transform the feel of your site. Modular Changes are contained in just two LESS files, enabling modification and ensuring forward compatibility.]]></description>
<pubDate>3 hours ago</pubDate>
<score>196 points</score>
<user>parkov</user>
<comments>30 comments</comments>
<id>3594540</id>
<discussion>http://news.ycombinator.com/item?id=3594540</discussion>
<link>http://bootswatch.com</link>
</item>
<item>
<id>f1</id>
<title>Louis CK inspires Jim Gaffigan to sell comedy special for $5 online</title>
<description><![CDATA[Dear Internet Friends,Inspired by the brilliant Louis CK, I have decided to debut my all-new hour stand-up special on my website, Jimgaffigan.com.Beginning sometime in April, “Jim Gaffigan: Mr. Universe” will be available exclusively for download for only $5. A dollar from each download will go directly to The Bob Woodruff Foundation; a charity dedicated to serving injured Veterans and their families.I am confident that the low price of my new comedy special and the fact that 20% of each $5 download will be donated to this very noble cause will prevent people from stealing it. Maybe I’m being naïve, but I trust you guys.]]></description>
<pubDate>57 minutes ago</pubDate>
<score>25 points</score>
<user>rkudeshi</user>
<comments>4 comments</comments>
<id>3595285</id>
<discussion>http://news.ycombinator.com/item?id=3595285</discussion>
<link>http://www.whosay.com/jimgaffigan/content/218011</link>
</item>

And here's an example of incorrect output. Note that the elements are not empty, thus I cannot seem to catch the error and simply jump to the next item. Everything past the promotion post will break:

<item>
<id>f14</id>
<title>Build the next Legos: We're hiring an iOS Developer &amp; Web Developer (YC S11)</title>
<description><![CDATA[Interested in building the next generation of toys on digital devices such as the iPad? That’s what we’re doing here at Launchpad Toys with apps like Toontastic (Named one of the “Top 10 iPad Apps of 2011” by the New York Times and was recently added to the iTunes Hall of Fame) and an awesom]]><![CDATA[e suite of others we have under development. We’re looking for creative and playful coders that have made games or highly visual apps/sites in the past for our two open development positions. As a kid, you probably played with Legos endlessly and grew up to be a hacker because you still love building things. Sounds like you? Email us at howdy@launchpadtoys.com with a couple links to some projects and code that we can look at along with your resume.]]></description>
<pubDate>2 hours ago</pubDate>
<score>14 points</score>
<user>bproper</user>
<comments>7 comments</comments>
<id>3594944</id>
<discussion>http://news.ycombinator.com/item?id=3594944</discussion>
<link>http://launchpadtoys.com/blog/2012/02/iosdeveloper-webdeveloper/</link>
</item>
<item>
<id>f15</id>
<title>SOPA foe Fred Wilson supports a blacklist on pirate sites</title>
<description><![CDATA[VC Fred Wilson says Google, Bing, Facebook, and Twitter should warn people when they try to log in at known pirate sites: "We don't need legislation." Fred Wilson says: If they try to pass antipiracy legislation, it will once again be 'war.' (Credit: Greg Sandoval/CNET) Fred Wilson, a well-known ven]]><![CDATA[ture capitalist from New York, says he's in favor of creating a blacklist for Web sites found to traffic in pirated films, music, and other intellectual property. The co-founder of Union Square Ventures told a gathering of media executives at the Paley Center for Media yesterday that he believes a good antipiracy measure would be for Google, Twitter, Facebook, and other major sites to issue warnings to people when they try to connect with a known pirate site. Fred Wilson, a co-founder of Union Square Ventures, says 'Our children have been taught to steal.' (Credit: Union Square Ventures) Wilson favors establishing an independent group to create a "black and white list." "The blacklist are those sites we all know are bad news," he told the audience in New York.]]></description>
<pubDate>14 points by bproper 2 hours ago  | 7 comments</pubDate>
<score>24 points</score>
<user>andrewcross</user>
<comments>12 comments</comments>
<id>3594558</id>
<discussion>http://news.ycombinator.com/item?id=3594558</discussion>
<link>http://news.cnet.com/8301-31001_3-57377862-261/post-sopa-influential-tech-investor-favors-blacklisting-pirate-sites/</link>
</item>

So here's my question: How can I handle a situation where a particular element is missing and find() doesn't throw an error? Do I have to start from scratch, or is there a better approach in scraping the HN front page?

For anyone curious, here's the whole XML file: http://thequeue.org/api/tc.xml

Answer 1

You have to work by chunks in order to handle that, there seems to be a dummy spacer element that can help you with that:

$news = preg_split('/<tr style="height:5px"><\/tr>/',$html->find('tbody',2)->innertext);

And then use subselectors:

foreach($news as $article){
    $article = str_get_html($article)
    // No upvote arrow found so its not a valid article
    if(count($article->find('img')) === 0){
        continue;
    }
}

And for the other elements you use the same selectors

Answer 2

We'll thanks to Ivan's trail of thought, I am now splitting the initially scraped HTML into an array, each node representing a post. Then, going through every single post in a loop, I'll check if the up voting arrow image exists. If not, I'll not add it to the result. In the end everything will be stitched back together and the sponsored post is left out. Here's the code:

$array = explode('<tr style="height:5px"></tr>',$html);
foreach ($array as $post) {

    if (!strstr($post,'grayarrow.gif')){}else{

    $clean .=  $post;

    }

}
unset($array);
$html = str_get_html($clean.'</body></html>');