XML for storing content...

andrewtayloruk

Hi,

I've been playing around with a CMS/DMS based around PHP+XML+XSLT+SQL.

Initial data storage in a RDBMS, all successful queries are executed then the results stored in XML chunks. PHP then uses XSLT to display XHTML output. This is also cached.

All content is entered via a CMS, once a site user requests content it checks for a generated XHTML page and if it exists returns this rather than hitting the database.

If the page doesn't already exist; it checks to see if the results exist in XML form (i.e. the results of the proposed query have been used on another page), if this exists, they're turned in to a XHTML page for future use and displayed to the client. If neither the XHTML nor the XML exists, the query hasn't been ran recently so the XML is generated from the DB, turned in to an XML chunk for later, the XML chunk is then used to create an XHTML page for later use and shown to the client.

The content stored back in the RDBMS should be stored as content, not HTML, there should be no mark up tags at all, no hrefs, no <li>, no <img src>. So how do you do it? What's the correct way, if I'm storing loads of text that's fine, but, what about news articles, should I store all the images separately and then push them in to the content later, that's fine, what about links, should all my links be at the bottom of the page and not mixed in with the content. What about paragraph tags?

What are everyone’s opinions on this?

Andrew

Shrike

This sounds on first impression like a nice proposal. I like the idea of storing 'chunks' of data in XML - perhaps reusable page components? It does however beg the question "Why bother with the XML cache?" since there is already an XHTML cache. If the XML chunks are reusable page elements whereas the XHTML is a complete page then this is understandable 🙂

andrewtayloruk

Yes, you're exactly right. Try and think of how many times you query the same set of tables; you may do

$query = 'SELECT field1, field2 FROM table';

$query = 'SELECT field1,field2,table2.field3 FROM table LEFT JOIN table2 ON (table1.primary = table2.foreign)';

$query = 'SELECT field1,field2,table2.field3 FROM table LEFT JOIN table2 ON (table1.primary = table2.foreign) WHERE field2 LIKE '%AS%';

And so on and so on.

Depending on your application, it may be better to have some kind of super query -

$query = 'SELECT table1.field1,table1.field2,table1.field3,table1.field4,table2.field1,table2.field2,table2.field3,table2.field4,table2.field5,table3.field1,table3.field2,table3.field3,table3.field4 FROM table1 LEFT JOIN table2 ON (table1.primary = table2.foreign) LEFT JOIN table3 ON (table2.primary = table3.foreign) ORDER BY table1.field1';

Chunk this down to an XML result then use different XSLT sheets depending on the page requirement, remember you can use XSLT selectively, so you could just display 1 or 2 records, or just certain fields.

It also has benefits if you send your content out to other applications, so now you're using one XML chunk to build an XHTML page and the same chunk to generate two RSS feeds and a feed to suppliers database.

The other thing to remember is, it's only the first user that takes the hit on performance, they generate the chunk and the complete XHTML for that page and any other pages which use the same data.

If you're going to go to the effort to safe hits on the DB, you may as well go to town.

But I'm still confused on how to store my content in the DB, should it have HTML tags or should I add some kind of pseudo tags, if I'm storing images seperately

 <page_content>
  <title>My Title</title>
  <author>Me</author>
  <author_email>me@.com<author_email>
  <summary>This is a summary of this article</summary>
  <body>This is my body text, I really want to add an image in here, should I use an img src tag or should I have a seperate element?</body>
  <images>
    <image>
     <name>myimage.jpg</name>
     <size>46kb</size>
     <height>10</height>
     <width>20</height>
     <alt>This is an image</alt>
    </image>
  </images>
 </page_content>

If I'm having seperate image element (which makes sense in theory) how do I then get this in to the body content when I output it in HTML?

Anyone ?

Weedpacket

Originally posted by andrewtayloruk
But I'm still confused on how to store my content in the DB, should it have HTML tags or should I add some kind of pseudo tags, if I'm storing images seperately

If you're going down the XSLT road already, you might as well use some other dialect of XML instead of XHTML. One reason is that if the site design changes, you only need to update the XSLT program.

Of course, there's little advantage to storing XML in a database unless either the entire database is implemented in XML or if the documents (and from your example I'm guessing that this is the case) have the potential to be arbitrarily complex. If it's something that could be straightforwardly implemented as a consistently-defined pile of field-value pairs then you'd be better off just storing them in the RDBMS directly (modulo caching!).

The Apache Organisation have been playing around with developing an XML-based database (in which queries are written in XPath instead of SQL), and a lot of the code for quite a few of their projects (Cocoon, Ant, etc.) is written in XML. I mention this because you might want to consider seeing if there is already an XML dialect that matches your needs out there, to save you designing your own. And while I've been typing this I've been trying to remember that other site, and I just have: www.xml.org - kinda obvious, really....