Support Has Moved!

Something like this should do the trick (only displays feeds from the last 24 hours, give or take an hour based on cache life):

<?php
include('simplepie.inc');
//$feed = new SimplePie('http://digg.com/rss/index.xml');
$feed = new SimplePie('feed://www.tuaw.com/rss.xml');

// Display the name of the feed
echo '<h1>' . $feed->get_feed_title() . '</h1>';

// Create a new array to hold data in
$new = array();

// Loop through all of the items in the feed
for ($x=0; $x < $feed->get_item_quantity(); $x++) {

    // Make sure that we have a handle for the current item.
    $item = $feed->get_item($x);

    // Calculate 24 hours ago
    $yesterday = time() - (24*60*60);

    // Compare the timestamp of the feed item with 24 hours ago.
    if ($item->get_date('U') > $yesterday) {

        // If the item was posted within the last 24 hours, store the item in our array we set up.
        $new[] = $item;
    }
}

// Loop through all of the items in the new array and display whatever we want.
foreach($new as $item) {
    echo '<h3>' . $item->get_title() . '</h3>';
    echo '<p>' . $item->get_date('j M Y, H:i:s O') . '</p>';
    echo $item->get_description();
    echo '<hr />';
}
?>

focusfriends -- Agreed.  You can't *rely* on the item's date, but this will work in a pinch.  For best results, you should aggregate the data in a database like you said.

Can you please post the full multifeed code unsing the last 24 hours hack; I can´t figure out at which location implementing it... the full php-script would be the easiest way understanding the implementation...

Thanks in advance,
Daniel

focusfriends wrote:

The only way to do this properly is to make your own (database) archive of your periodic feed crawls and apply a "signature" to each item so you know if it's new or not. And then you can use your own "insert stamp" in your "only today's items" judgement. Be aware though that you can't rely on the pubdate or item link (guid) in your "new item" judgement.

I am looking into applying some functionality to new items and I also considered a database.  However, the problem of what data is new exists at a fundamental level.  Each time the Feed is updated it might only include one new item, but at this stage I don't know what is new and what is old, this is pre-database data, raw from SimplePie.  So how do I know what data to put into the database?  Do I put it all in and risk duplicates?  Or do I look at the pubDate and not put anything in that already has that pubDate?  It seems we already dismissed the pubDate as being inaccurate. 

So the question stands, how can we know what item is fresh and what item is old?  I suppose we could test titles to see if they are unique before putting the item into the database, but that feels a little like a hack. 

Some WordPress plugins that take feeds and submit them into WordPress as stories seem to know what is a new story and what is old, I wonder how they do that?

Last edited by esteban (23 April 2007 02:54:53)

We're addressing a number of these issues in 1.0:

1) Multifeeds support is integrated directly into the core so there's no more hacking the data together.

2) SimplePie parses out the data and provides an API for accessing that data.  If you want to build an aggregator, it is up to you to determine what is new and what is old based on your own database solution.

Many people use the <guid> value ($item->get_id()) to keep track, but currently if there is no <guid>, get_id() returns false.  One thing that still needs to be implemented, but is still on-track for 1.0 is that when a guid is not available, SimplePie will generate a unique guid hash from a combination of the title, permalink URL, and pubDate of the item.  That should allow a fairly reliable, unchanging unique identifier for the news item in question.

In the meantime, I would suggest generating a hash of your own using the aforementioned pieces of data and either the md5() or sha1() PHP functions.

Thanks guys, sounds like v1.0 with a proper API will be awesome.