Zend_Feed_Reader is a component used to consume RSS and Atom feeds of any version, including RDF/RSS 1.0, RSS 2.0, Atom 0.3 and Atom 1.0. The API for retrieving feed data is deliberately simple since Zend_Feed_Reader is capable of searching any feed of any type for the information requested through the API. If the typical elements containing this information are not present, it will adapt and fall back on a variety of alternative elements instead. This ability to choose from alternatives removes the need for users to create their own abstraction layer on top of the component to make it useful or have any in-depth knowledge of the underlying standards, current alternatives, and namespaced extensions.

Internally, Zend_Feed_Reader works almost entirely on the basis of making XPath queries against the feed XML's Document Object Model. The DOM is not exposed though a chained property API like Zend_Feed though the underlying DOMDocument, DOMElement and DOMXPath objects are exposed for external manipulation. This singular approach to parsing is consistent and the component offers a plugin system to add to the Feed and Entry level API by writing Extensions on a similar basis.

Performance is assisted in three ways. First of all, Zend_Feed_Reader supports caching using Zend_Cache to maintain a copy of the original feed XML. This allows you to skip network requests for a feed URI if the cache is valid. Second, the Feed and Entry level API is backed by an internal cache (non-persistant) so repeat API calls for the same feed will avoid additional DOM or XPath use. Thirdly, importing feeds from a URI can take advantage of HTTP Conditional GET requests which allow servers to issue an empty 304 response when the requested feed has not changed since the last time you requested it. In the final case, an instance of Zend_Cache will hold the last received feed along with the ETag and Last-Modified header values sent in the HTTP response.

In relation to Zend_Feed, Zend_Feed_Reader was formulated as a free standing replacement for Zend_Feed but it is not backwards compatible with Zend_Feed. Rather it is an alternative following a different ideology focused on being simple to use, flexible, consistent and extendable through the plugin system. Zend_Feed_Reader is also not capable of constructing feeds and delegates this responsibility to Zend_Feed_Writer, its sibling in arms.

Importing a feed with Zend_Feed_Reader is not that much different to Zend_Feed. Feeds can be imported from a string, file, URI or an instance of type Zend_Feed_Abstract. Importing from a URI can additionally utilise a HTTP Conditional GET request. If importing fails, an exception will be raised. The end result will be an object of type Zend_Feed_Reader_FeedInterface, the core implementations of which are Zend_Feed_Reader_Feed_Rss and Zend_Feed_Reader_Feed_Atom (Zend_Feed took all the short names!). Both objects support multiple (all existing) versions of these broad feed types.

In the following example, we import an RDF/RSS 1.0 feed and extract some basic information that can be saved to a database or elsewhere.

The example above demonstrates Zend_Feed_Reader's API, and it also demonstrates some of its internal operation. In reality, the RDF feed selected does not have any native date or author elements, however it does utilise the Dublin Core 1.1 module which offers namespaced creator and date elements. Zend_Feed_Reader falls back on these and similar options if no relevant native elements exist. If it absolutely cannot find an alternative it will return NULL, indicating the information could not be found in the feed. You should note that classes implementing Zend_Feed_Reader_FeedInterface also implement the SPL Iterator and Countable interfaces.

Feeds can also be imported from strings, files, and even objects of type Zend_Feed_Abstract.

Zend_Feed_Reader does its best not to stick you in a narrow confine. If you need to work on a feed outside of Zend_Feed_Reader, you can extract the base DOMDocument or DOMElement objects from any class, or even an XML string containing these. Also provided are methods to extract the current DOMXPath object (with all core and Extension namespaces registered) and the correct prefix used in all XPath queries for the current Feed or Entry. The basic methods to use (on any object) are saveXml(), getDomDocument(), getElement(), getXpath() and getXpathPrefix(). These will let you break free of Zend_Feed_Reader and do whatever else you want.

• saveXml() returns an XML string containing only the element representing the current object.

• getDomDocument() returns the DOMDocument object representing the entire feed (even if called from an Entry object).

• getElement() returns the DOMElement of the current object (i.e. the Feed or current Entry).

• getXpath() returns the DOMXPath object for the current feed (even if called from an Entry object) with the namespaces of the current feed type and all loaded Extensions pre-registered.

• getXpathPrefix() returns the query prefix for the current object (i.e. the Feed or current Entry) which includes the correct XPath query path for that specific Feed or Entry.

Here's an example where a feed might include an RSS Extension not supported by Zend_Feed_Reader out of the box. Notably, you could write and register an Extension (covered later) to do this, but that's not always warranted for a quick check. You must register any new namespaces on the DOMXPath object before use unless they are registered by Zend_Feed_Reader or an Extension beforehand.

These days, many websites are aware that the location of their XML feeds is not always obvious. A small RDF, RSS or Atom graphic helps when the user is reading the page, but what about when a machine visits trying to identify where your feeds are located? To assist in this, websites may point to their feeds using <link> tags in the <head> section of their HTML. To take advantage of this, you can use Zend_Feed_Reader to locate these feeds using the static findFeedLinks() method.

This method calls any URI and searches for the location of RSS, RDF and Atom feeds assuming the website's HTML contains the relevant links. It then returns a value object where you can check for the existence of a RSS, RDF or Atom feed URI.

The returned object is an ArrayObject subclass called Zend_Feed_Reader_Collection_FeedLink so you can cast it to an array, or iterate over it, to access all the detected links. However, as a simple shortcut, you can just grab the first RSS, RDF or Atom link using its public properties as in the example below. Otherwise, each element of the ArrayObject is a simple array with the keys "type" and "uri" where the type is one of "rdf", "rss" or "atom".

Based on these links, you can then import from whichever source you wish in the usual manner.

This quick method only gives you one link for each feed type, but websites may indicate many links of any type. Perhaps it's a news site with a RSS feed for each news category. You can iterate over all links using the ArrayObject's iterator.

In an attempt to simplify return types, with Zend Framework 1.10 return types from the various feed and entry level methods may include an object of type Zend_Feed_Reader_Collection_CollectionAbstract. Despite the special class name which I'll explain below, this is just a simple subclass of SPL's ArrayObject.

The main purpose here is to allow the presentation of as much data as possible from the requested elements, while still allowing access to the most relevant data as a simple array. This also enforces a standard approach to returning such data which previously may have wandered between arrays and objects.

The new class type acts identically to ArrayObject with the sole addition being a new method getValues() which returns a simple flat array containing the most relevant information.

A simple example of this is Zend_Feed_Reader_FeedInterface::getCategories(). When used with any RSS or Atom feed, this method will return category data as a container object called Zend_Feed_Reader_Collection_Category. The container object will contain, per category, three fields of data: term, scheme and label. The "term" is the basic category name, often machine readable (i.e. plays nice with URIs). The scheme represents a categorisation scheme (usually a URI identifier) also known as a "domain" in RSS 2.0. The "label" is a human readable category name which supports HTML entities. In RSS 2.0, there is no label attribute so it is always set to the same value as the term for convenience.

To access category labels by themselves in a simple value array, you might commit to something like:

It's a contrived example, but the point is that the labels are tied up with other information.

However, the container class allows you to access the "most relevant" data as a simple array using the getValues() method. The concept of "most relevant" is obviously a judgement call. For categories it means the category labels (not the terms or schemes) while for authors it would be the authors' names (not their email addresses or URIs). The simple array is flat (just values) and passed through array_unique() to remove duplication.

The above example shows how to extract only labels and nothing else thus giving simple access to the category labels without any additional work to extract that data by itself.

Retrieving information from a feed (we'll cover entries and items in the next section though they follow identical principals) uses a clearly defined API which is exactly the same regardless of whether the feed in question is RSS, RDF or Atom. The same goes for sub-versions of these standards and we've tested every single RSS and Atom version. While the underlying feed XML can differ substantially in terms of the tags and elements they present, they nonetheless are all trying to convey similar information and to reflect this all the differences and wrangling over alternative tags are handled internally by Zend_Feed_Reader presenting you with an identical interface for each. Ideally, you should not have to care whether a feed is RSS or Atom so long as you can extract the information you want.

Note: While determining common ground between feed types is itself complex, it should be noted that RSS in particular is a constantly disputed "specification". This has its roots in the original RSS 2.0 document which contains ambiguities and does not detail the correct treatment of all elements. As a result, this component rigorously applies the RSS 2.0.11 Specification published by the RSS Advisory Board and its accompanying RSS Best Practices Profile. No other interpretation of RSS 2.0 will be supported though exceptions may be allowed where it does not directly prevent the application of the two documents mentioned above.

Of course, we don't live in an ideal world so there may be times the API just does not cover what you're looking for. To assist you, Zend_Feed_Reader offers a plugin system which allows you to write Extensions to expand the core API and cover any additional data you are trying to extract from feeds. If writing another Extension is too much trouble, you can simply grab the underlying DOM or XPath objects and do it by hand in your application. Of course, we really do encourage writing an Extension simply to make it more portable and reusable, and useful Extensions may be proposed to the Framework for formal addition.

Here's a summary of the Core API for Feeds. You should note it comprises not only the basic RSS and Atom standards, but also accounts for a number of included Extensions bundled with Zend_Feed_Reader. The naming of these Extension sourced methods remain fairly generic - all Extension methods operate at the same level as the Core API though we do allow you to retrieve any specific Extension object separately if required.

Given the variety of feeds in the wild, some of these methods will undoubtedly return NULL indicating the relevant information couldn't be located. Where possible, Zend_Feed_Reader will fall back on alternative elements during its search. For example, searching an RSS feed for a modification date is more complicated than it looks. RSS 2.0 feeds should include a <lastBuildDate> tag and (or) a <pubDate> element. But what if it doesn't, maybe this is an RSS 1.0 feed? Perhaps it instead has an <atom:updated> element with identical information (Atom may be used to supplement RSS's syntax)? Failing that, we could simply look at the entries, pick the most recent, and use its <pubDate> element. Assuming it exists... Many feeds also use Dublin Core 1.0 or 1.1 <dc:date> elements for feeds and entries. Or we could find Atom lurking again.

The point is, Zend_Feed_Reader was designed to know this. When you ask for the modification date (or anything else), it will run off and search for all these alternatives until it either gives up and returns NULL, or finds an alternative that should have the right answer.

In addition to the above methods, all Feed objects implement methods for retrieving the DOM and XPath objects for the current feeds as described earlier. Feed objects also implement the SPL Iterator and Countable interfaces. The extended API is summarised below.

Retrieving information for specific entries or items (depending on whether you speak Atom or RSS) is identical to feed level data. Accessing entries is simply a matter of iterating over a Feed object or using the SPL Iterator interface Feed objects implement and calling the appropriate method on each.

The extended API for entries is identical to that for feeds with the exception of the Iterator methods which are not needed here.

Extending Zend_Feed_Reader allows you to add methods at both the feed and entry level which cover the retrieval of information not already supported by Zend_Feed_Reader. Given the number of RSS and Atom extensions that exist, this is a good thing since Zend_Feed_Reader couldn't possibly add everything.

There are two types of Extensions possible, those which retrieve information from elements which are immediate children of the root element (e.g. <channel> for RSS or <feed> for Atom) and those who retrieve information from child elements of an entry (e.g. <item> for RSS or <entry> for Atom). On the filesystem these are grouped as classes within a namespace based on the extension standard's name. For example, internally we have Zend_Feed_Reader_Extension_DublinCore_Feed and Zend_Feed_Reader_Extension_DublinCore_Entry classes which are two Extensions implementing Dublin Core 1.0 and 1.1 support.

Extensions are loaded into Zend_Feed_Reader using Zend_Loader_PluginLoader, so their operation will be familiar from other Zend Framework components. Zend_Feed_Reader already bundles a number of these Extensions, however those which are not used internally and registered by default (so called Core Extensions) must be registered to Zend_Feed_Reader before they are used. The bundled Extensions include:

The Core Extensions are somewhat special since they are extremely common and multi-faceted. For example, we have a Core Extension for Atom. Atom is implemented as an Extension (not just a base class) because it doubles as a valid RSS module - you can insert Atom elements into RSS feeds. I've even seen RDF feeds which use a lot of Atom in place of more common Extensions like Dublin Core.

The additional non-Core Extensions are offered but not registered to Zend_Feed_Reader by default. If you want to use them, you'll need to tell Zend_Feed_Reader to load them in advance of importing a feed. Additional non-Core Extensions will be included in future iterations of the component.

Registering an Extension with Zend_Feed_Reader, so it is loaded and its API is available to Feed and Entry objects, is a simple affair using the Zend_Loader_PluginLoader. Here we register the optional Slash Extension, and discover that it can be directly called from the Entry level API without any effort. Note that Extension names are case sensitive and use camel casing for multiple terms.

In the simple example above, we checked how frequently a feed is being updated using the getUpdatePeriod() method. Since it's not part of Zend_Feed_Reader's core API, it could only be a method supported by the newly registered Syndication Extension.

As you can also notice, the new methods from Extensions are accessible from the main API using PHP's magic methods. As an alternative, you can also directly access any Extension object for a similar result as seen below.