Styling XML documents

Presentation of XML documents

Ideally, XML documents do not contain presentation markup. Tags and attributes should describe what the document is about, not how the document looks.

<PLAY>
<TITLE>The Tragedy of Hamlet, Prince of Denmark</TITLE>
  <ACT>
    <SCENE title="Elsinore. A platform before the castle.">
      <STAGEDIR>
        FRANCISCO at his post. Enter to him BERNARDO
      </STAGEDIR>
.
      <SPEECH speaker="BERNARDO">
        <LINE>Who's there?</LINE>
      </SPEECH>
.
      <SPEECH speaker="FRANCISCO">
        <LINE>
          Nay, answer me: stand, and unfold yourself.
        </LINE>
      </SPEECH>
.
      <SPEECH speaker="BERNARDO">
        <LINE>Long live the king!</LINE>
      </SPEECH>

...

can be presented as this.

CSS and XSL

• CSS2 already does that. CSS started as a method for styling HTML documents, was generalized to XML docs
• XSL goes further. It adds:
  • Additional properties for complex formattingMore later
  • Complex pagination model
  • Content generation

Three Specifications

• Source tree: any XML document.
• Result tree: (originally) Formatting Objects, in practice any XML document type that a device knows how to display.
• The transformation language: XSLT. It is even more general in that it is able to transform a document in any XML vocabulary into a document in any other XML vocabulary.
• XSLT needs a mechanism to access nodes in source tree: XPath

In order to generate contents, XSL introduced a two-step process: (1) transformation of the original document into an intermediary format that contains the original and generated text, with markup describing formatting semantics, and (2) rendering of this format. Since it seemed convenient that that intermediary document should be expressed as XML, the transformation part was generalized to a AnyML to OtherML and became a separate document: XST Transformations (XSLT)

XSLT

• General-purpose transformation language, that can be used to generate TOCs, indexes, that's what it was originally intended for, and more..

  Very simple example: changes all foo elements to comments.

  A stylesheet is a set of templates that describe what to output when a chosen type of markup is met in the source file

  <xsl:template match="node()|@*">
    <xsl:copy>
      <xsl:apply-templates select="node()|@*"/>
    </xsl:copy>
  </xsl:template>
  .
  <xsl:template match="foo">
    <xsl:comment><xsl:apply-templates/></xsl:comment>
  </xsl:template>
  .

• As a consequence,Started to be used to generate HTML for display of any type of XML data
• Since that became very popular (mainly because HTML is widespread) XSLT was finished first and became a Recommendation in 1999.

XPath

Part of XSLT that allows access to the input XML documentNecessary for specifying which part of the source to retrieve and modify into the output

Examples:

<xsl:value-of select="ACT/SCENE/SPEECH/@speaker"/>

will return all speaker attributes nodes as a string: "BERNARDO FRANCISCO BERNARDO..."

or:

 /ACT[3]/SCENE[position() < 3]/
 SPEECH[@speaker="Hamlet"]/LINE[substring(.,"beer")]

Points to Act V, Scene 1:

<SPEECH speaker="Hamlet">
<LINE>No, faith, not a jot; but to follow him thither with</LINE>
<LINE>modesty enough, and likelihood to lead it: as</LINE>
<LINE>thus: Alexander died, Alexander was buried,</LINE>
<LINE>Alexander returneth into dust; the dust is earth; of</LINE>
<LINE>earth we make loam; and why of that loam, whereto he</LINE>
<LINE>was converted, might they not stop a beer-barrel?</LINE>
<LINE>Imperious Caesar, dead and turn'd to clay,</LINE>
<LINE>Might stop a hole to keep the wind away:</LINE>
<LINE>O, that that earth, which kept the worldin awe,</LINE>
<LINE>Should patch a wall to expel the winter flaw!</LINE>

XPath could also be generalized and was therefore jointly designed by XSLT and XLink WGs

Formatting Objects

An XML vocabulary for presentation semantics.

Example

<xsl:template match="TITLE">
  <fo:block font-family="Helvetica" 
            font-size="20pt"
            space-after.optimum="2em" 
            keep-with-next.within-page="always">
    <xsl:apply-templates/>
  </fo:block>
</xsl:templates

Elements and Attributes

Elements (formatting objects)

• <fo:layout-master-set>, <fo:page-sequence>
• <fo:block>, <fo:inline>
• <fo:footnote>, <fo:external-graphics>

Attributes (properties)

• CSS2 properties with additional values (e.g. backslant for text-style)
• Other properties (e.g. hyphenation, writing-mode, keeps & breaks)

XSL Processing

• off-line formatting (typically: paper documentation)
• server sends XML file and XSL style sheet, client transforms and displaysBrowsers do (or plan to do) transformations to XHTML or FOs
• server receives CC/PP, transforms its XML to FOs/XHTML/SVG/DisplayML, client renders.

Examples/Applications

• XML to XHTML: presentation of data for the web (example: XML file, HTML output).
• XML to FOs: formatting of paged documents (example: MathML spec, this slideshow)
• FOs can include other XML vocabularies, like SVG and MathML

Current Situation: XSLT 1.1

• New features, that were required and already implemented
  • Multiple output documents
  • RTF to node-sets
  • Extension Functions
• Because of technical problems with the model, 1.1 spec is now `on hold'.
• The WG is working on XSLT2.0 which will stay compatible with 1.1 (#1 requirement)

Current Situation: XSLT 2.0, XPath 2.0

• Requirements documents published 15 February 2001.
• XPath2.0 currently developed by XSL WG and XML Query WGSo that common functionality is shared (XPath is a kind of query language, so why not extend it to meet the needs database people.)
• XSLT2.0, like XML Query 1.0, will be based on XPath2.0
• First Working Draft of XPath expected May

Current Situation: XSL 1.0

• Currently in CR: model complete, testing implementability of features.
• Already 6 or 7 implementations: off-line and browsers.
• Will move to Recommendation in the next few weeks.