How Prince PDF Export Works

The following describes how the GET:pages/{pageid}/pdf feature works.

• User-Agent does a GET on /deki/pages/{pageid}/pdf
• READ permissions are checked
• DekiXmlParser is in invoked to get the XHTML
• XDoc is created with html, head, title, body tags
• XDoc head and body section is populated with parser.Head, parser.MainBody
• Locate all the img tags using XPATH, append an authtoken to all relative links (so images on restricted pages can be downloaded)
• Add the XHTML Doctype to the begining of the XDoc to make prince happy
• Convert the XDoc to a MemoryStream so we can pipe it into stdin when invoking prince
• Invoke prince to generate the PDF and capture stdout
• Set the ContentDisposition header using filename={page_title}.pdf
• write the stdout stream to the browser  

The prince binary is invoked as follows:

/usr/bin/prince --input=xml -s http://yoursite.com/skins/common/prince.css --baseurl=http://yoursite.com - -o -

The arguments are this:

--input=xml:  use prince's xml/xhtml parser

-s:  the css stylesheet to use

--baseurl: base url to use for relative links

-: read input from stdin

-o -: write output to stdout