The XMP Trick

This nice bit of XQuery comes from my colleague Kelly Stirman:

You may have heard of XMP. It’s a standard developed by Adobe to embed XML meta-data in rich media binary objects.

The idea is that when you start a photo shoot you put into the camera the location, subject and even the rights associated with the photos.

Then each image is embedded with this information and, as it is passed around, systems can make use of the information without anyone having to enter the meta-data again.

It’s really cool and a huge time and headache saver for people who handle photos.

And it’s even cooler that, using XQuery and the standard MarkLogic extensions, you can load the image into MarkLogic and Query the XMP directly:

Step 1: Load an image with XMP.

How about this picture a boat I took in Maine recently:

Using the MarkLogic update extenstions to XQuery, you can just go ahead and pull it from my blog and upload it into the MarkLogic Server you have set up in the Tutorial:

     xdmp:document-load(
        "http://xquery.typepad.com/photos/uncategorized/maineboat.jpg",
        <options xmlns="xdmp:document-load">
          <uri>/xmp/maine-boat.jpg</uri>
       </options>
     )

 Step 2: Query the image with XQuery and some other helpful MarkLogic builtins:

    declare namespace dc="http://purl.org/dc/elements/1.1/"
    declare namespace tiff="http://ns.adobe.com/tiff/1.0/"
    declare namespace rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
    declare namespace x="adobe:ns:meta/" 

    (: get the image :) 

    let $doc := doc("/xmp/maine-boat.jpg")

    (: make a string out of the image using the MarkLogic Server extension xdmp:quote :)

    let $node := xdmp:quote($doc/node())

    (: isolate the <x:xmpmeta> node in the string using substring-before and substring-after :)

    let $xmp := substring-after($node,"<x:xmpmeta")
    let $xmp := substring-before($xmp,"</x:xmpmeta>")     

    (: put back the stuff we took out with out after and before :)

 let $xmp := concat("<x:xmpmeta",$xmp,"</x:xmpmeta>")

    (: make THAT into a node using the very handy MarkLogic extension xdmp:unquote :)

    let $xmp := xdmp:unquote($xmp)

    return

    (: return what we want - in this case the basic meta-data. try just $xmp to see the entire node :)

       <image>
                   {$xmp/x:xmpmeta/rdf:RDF/rdf:Description/dc:title}
                   {$xmp/x:xmpmeta/rdf:RDF/rdf:Description/dc:creator}
                   {$xmp/x:xmpmeta/rdf:RDF/rdf:Description/dc:subject}
       </image> 

Very nice!  We are storing the original binary (you can display it as an image any time) and are easily able to unlock the content contained within it. 

Welcome to the new world of self-describing content!

 MT