SQL Server Reference Guide

Programming Applications: RSS Feeds from SQL Server

The average person today has access to more information than at perhaps any time in history: print media, billboards, television, radio, and, of course, several methods that the modern computer provides. With this plethora of avenues for information, users struggle to find a way to limit the stream of data.

In the Business Intelligence software world, the "clicks" paradigm is sometimes used to determine the best methods of delivering information to busy professionals.

In this paradigm, five levels (or "clicks") indicate the way the user gets data:

The 4^th "click" is reserved for the technical professionals who design the other four levels.

At the first level (0-click), E-mail has been the favorite delivery technology. Unfortunately, unsolicited E-mail and corporate chatter have rendered this method largely ineffective.

To deliver data that a user wants to see, the best method is to have them sign up for it. One technology that directly addresses this need is Really Simple Syndication, or RSS.

The long and short of RSS is that it's an XML file with specific elements. One set of elements in the file contains the information you want to publish to the user. The user enters the location of the RSS file into an "aggregator" program, which looks a bit like an E-mail client. When the RSS file is polled by the aggregator, the program displays the items that changed. You can create a headline, a short description, and a URL for the aggregator. This information can be as varied as corporate news, inventory changes, server statuses, blogs, and more.

There are a few ways to create these RSS files, but a database is ideal for storing the information for the RSS elements. A program can update the database with the information, and SQL Server can create an RSS file from the data. Or, SQL Server can store the data, and another program can extract it to create the RSS file. In either case, the aggregator (or "RSS Reader") checks the file location periodically and compares it to a cache of the previously checked RSS file. The new items are displayed for the user to read.

Let's take a step-by-step look at the process to create a database table for an RSS feed, and then explore the ways that SQL Server or another program can create the RSS file from that table.

To begin, let's examine how an RSS file is structured. It's an XML file, so it uses a header, footer, and in between there are specific tags to enclose "elements." These elements are the data for the XML content. To mark the elements, tags names are enclosed with less-than and greater-than symbols <>. To show the completion of the data element, the end-tags are also enclosed the same way, but begin with a forward slash /. An "outer" element must enclose an "inner" element. A simple tag arrangement looks a bit like this:

<Element 1>
	<Element 2> Data Goes in Here </Element 2>
</Element 1>

A "well-formed" RSS file looks like this:

<?xml version="1.0" encoding="utf-8" ?>
<channel>
<title>SiteName</title>
<description> Site Description </description>
<link>http://www.website.com</link>

<item>
 <title>Sample Headline 1</title>
 <pubdate>07/16/2004</pubdate>
 <description>Item Description 1</description>
 <link>http://www.someurl.com</link>
</item>

<item>
 <title>Sample Headline 2</title>
 <pubdate>07/17/2004</pubdate>
 <description>Item Description 2</description>
 <link>http://www.someurl.com</link>
</item>

</channel>
The first section of the file:
<?xml version="1.0" encoding="utf-8" ?>

is the header portion that tags this file as XML. The main tag of:

<channel>

sets the XML file to be accepted as an RSS file. The elements that follow:

<title>SiteName</title>
<description> Site Description </description>
<link>http://www.website.com</link>

are slightly static. If the SQL Server database services only one feed, then this information can be hard-coded into the RSS file before or after it's accessed by the database extract. If not, the elements between <title> and </title> can be stored in a table, along with the description and link elements. A final line, </channel>, also needs to be added after the database extract to complete the opening tag.

The "meat" of the RSS file is stored in the <item> tags. Each of these tags is the equivalent of an individual E-mail. Four elements are stored here: title, pubdate, description, and link.

The <title> element is the text to be displayed in bold in the aggregator. The <pubdate> displays the published date in the "date" column of the aggregator. The <description> tag is the text that appears in the main window, and the <link> tag redirects the aggregator when the user clicks the description.

To store these elements, use the code below to create a table to hold the data:

CREATE TABLE RSSFeed 
	(
	ItemID int identity (1, 1) NOT NULL 
	, ItemName varchar (100) NULL
	, PublishDate smalldatetime NULL
	, ItemDesc varchar (500) NULL
	, ItemURL varchar(100) NULL
	)
GO

There are only four elements that are required, so this code creates just those fields, along with a key. In more practical use, the data might be spread over several tables that already exist. If that's the case, a view accomplishes the same thing as this table.

The code below fills this new table with some sample values:

INSERT INTO RSSFeed ( 
ItemName
, PublishDate 
, ItemDesc 
, ItemURL 
)
VALUES 
(
 'Sample Headline 1'
, getdate()-3
, 'Item Description 1'
, 'http://www.someurl.com'
)
INSERT INTO RSSFeed ( 
ItemName
, PublishDate 
, ItemDesc 
, ItemURL 
)
VALUES 
(
 'Sample Headline 2'
, getdate()-2
, 'Item Description 2',
'http://www.someurl.com'
)
GO 

The beauty of storing the RSS data in a database is that any program can keep the published items up-to-date. That program can be completely unaware of the RSS creation process.

It's important to remember that an RSS database operation is always either an insert or delete situation. The elements are similar to E-mail messages; once an E-mail has arrived, the sender can't reach into your in-box and change it.

This makes sense, since the information in an RSS feed is usually time-sensitive. There's seldom a need to re-write a simple message; normally, you just send an update.

For that reason, the SELECT statement used to extract the RSS message commonly uses a date range as a filter. Another alternative is to code the publishing application to remove older entries from the file.

The important part of that decision is to consider the data's timeliness. If the data is "stale" within a week, that would be the extract range. If users might not see the data for a month or more, the RSS file should include that range.

There are three choices for extracting element data for the RSS file. The first choice is to have an external program access the database and create the file from the elements provided.

Another option is to use T-SQL to create the entire file. By setting NOCOUNT ON, and removing the headers from the output, the SQL Statement below will create a complete RSS file as an output:

SET NOCOUNT ON
SELECT 
'<?xml version="1.0" encoding="utf-8" ?>' + CHAR(13)
, '<rss version="2.0">' + CHAR(13)
,'<channel>' + CHAR(13)
,'<title>SiteName</title>' + CHAR(13)
,'<description> Site Description </description>' + CHAR(13)
,'<link>http://www.website.com</link>' + CHAR(13)
SELECT
'<item>' + CHAR(13)
, '<title>', LTRIM(RTRIM(ItemName)) AS title, '</title>' + CHAR(13)
, '<date>', CONVERT(VARCHAR(10),PublishDate,101) AS pubdate, '</date>' + CHAR(13)
, '<description>' + LTRIM(RTRIM(ItemDesc)) AS description, '</description>' + CHAR(13)
, '<link>', LTRIM(RTRIM(ItemURL)) AS link, '</link>' + CHAR(13)
,'</item>', CHAR(13)
FROM 
RSSFeed 
SELECT
'</channel>' + CHAR(13)
,'</rss>'

The osql program or some other client can create the file on a regular basis using this select statement. Note that the code shown above does not specify a date range.

Another method is to combine the Web Assistant Wizard with SQL Server's XML features and the following query:

SELECT 
/*	
These items need to be in a format or header file:
<?xml version="1.0" encoding="utf-8" ?>
<rss version="2.0">
<channel>
<title>SiteName</title>
<description> Site Description </description>
<link>http://www.website.com</link>
*/
	 LTRIM(RTRIM(ItemName)) AS title
	, CONVERT(VARCHAR(10),PublishDate,101) AS pubdate
	, LTRIM(RTRIM(ItemDesc)) AS description 
	, LTRIM(RTRIM(ItemURL)) AS link
/* 

And this needs to be at the end of the format or footer file:

</channel> 
</rss>
*/
FROM 
	RSSFeed item
FOR XML AUTO, ELEMENTS

A format file is required with this method, containing the header and footer elements in the comments. You can explore basic information about SQL's XML features in my article, "XML Overview".

There are several other articles at InformIT dealing with RSS aggregators and other applications. Make sure you check them out.

Online Resources

Everything you ever wanted to know about RSS origins and flavors is on the definitive resource for XML dialects – XML.com.

BlogSpace has a rundown on various RSS aggregators (readers) for various operating systems here.

InformIT Tutorials and Sample Chapters

Kevin Meltzer and Brent Michalski show you more about RSS, and also give examples of other tags.

Tristan Louis dives into RSS a little further in this article.