Guide to making XML documents

1.1. TEI and OUCS

The Text Encoding Initiative (TEI) Guidelines are an international and interdisciplinary standard that enables libraries, museums, publishers, and individual scholars to represent a variety of literary and linguistic texts for online research, teaching, and preservation.

The TEI standard is maintained by a consortium of leading institutions and projects worldwide; Oxford is one of these institutions. Two of the major players in the TEI are members of OUCS: Lou Burnard and Sebastian Rahtz. Lou joined the Text Encoding Initiative project as its European Editor back in 1989 (a post he still holds), while Sebastian is one of the consortium's directors and actively develops the TEI itself.

1.2. OUCS Web Site Accessibility

Since 2002 it has been law to provide documents (including web pages) in accessible formats to users of alternative technologies such as screen readers. The relevant legislation is the Special Educational Needs and Disability Act (SENDA) 2001 which is part 4 of the Disability discrimination Act (DDA). This act brought Education establishments into line with commercial providers in the way that they provide information and services to the disabled community.

The W3C organisation have created various standards for web accessibility. These are:

• Priority 1 (A)
• Priority 2 (AA)
• Priority 3 (AAA)

Priority level details

All OUCS pages should reach level 1 standard. The University has decided that its web sites should ideally conform to the level 2 standard and meet as many level 3 points as possible.

The following document includes details on how to make your XML documents accessible to as wide an audience as possible. Please make sure that you follow these accessibility guidelines - it's the LAW!

2.1. XML Schema

The rules of the TEI XML format are stored in a schema (we use the RELAXNG schema language) file. This file defines the structure of how XML is to be written and is the key to transforming the text from one format to another. In order to write a valid TEI XML document the schema has to be followed. Luckily there are many XML editors that look after the schema for you and show any errors when the document is tested against the schema.

2.2. XSLT Stylesheets

An XSLT Stylesheet or Extensible Stylesheet Language Transformation Stylesheet is basically a set of rules to process a XML document. It turns an XML rendition of a file into the final version of a file. OUCS uses two versions of XSLT files, one turns an XML file into a web page (HTML format), the other turns XML into PDF format for printing.

2.3. CSS Stylesheets

CSS or Cascading Style Sheets are files containing information on how a document is to be presented e.g. bold, red headings or grey backgrounds. There are two versions used by OUCS: one displays the XML file directly and is fairly simple; the other displays the final web page and is fairly complex.

2.4. Version Control System

This is the method by which documents are stored. OUCS uses an open source system called Subversion to store XML documents. It is not restricted in the type of documents it can accept, allowing us to store just about anything we might need.

3.1. A Few Definitions

Elements and Tags

XML documents have lots of elements, one example is the title element. This begins with a start-tag <title> and is closed by the end-tag </title>. Any text between the start and end tags is therefore defined as the title of the document. Most XML tags work in this way: a start tag, some text, followed by an end tag. There are some elements that are self closing (i.e. they have no end tag); where appropriate these will be highlighted later in this document.

Content and Data

Any text between tags is the content of the element. This can be of two forms: the actual information or data; and other elements. Where the two occur together this is termed mixed content.

Attributes

All elements can have additional properties beside the element name and content. These properties are the attributes of an element and they consist of name-value pairs. For example a <div> element can have the attribute id="xxx", where xxx represents a name or number. In the example below, the id is 'email':

<div id="email">  <head>Configuring your email client</head>  <p> text....</p></div>

XML structure and nesting tags

XML is very strict on its element structure, especially compared to HTML. In XML, tags usually have to be started and ended. They must be nested properly and used in the correct place within the document hierarchy. This generally means that you cannot open a new tag e.g. <p> without closing the previous <p> tag. (N.B. there are exceptions to this rule e.g. self-closing tags).

3.2. Dissecting the OUCS XML Template

Viewing the OUCS template code in your editor shows the document structure. The complete page structure is shown below:

First comes the declaration that the file is a TEI document <TEI.2>. This is effectively the start tag for the document, all other elements must be correctly arranged or nested inside the <TEI.2> tags for the document to be valid TEI XML.

The first element inside <TEI.2> is the <teiHeader> element. Everything within this element is part of the document's Metadata (Metadata is data about the document, e.g. its title, author, creation date etc.). OUCS documents have a number of fields in the <teiHeader>; some have to be manually completed, such as the title of the document, while others are automatically added on document submission e.g. Last changed by information. Usually, when writing your own documents, you should complete the following metadata elements:

• title
• author
• date

It will also be necessary to complete an extra sections in the header recording who is responsible

• authority

The end of the metadata is marked by the closing </teiHeader> tag.

After the metadata comes the body of the document. This can be split into three sections:

<front>

contains any prefatory matter (headers, title page, prefaces, dedications, etc.) found before the start of a text proper.

<body>

contains the whole body of a single unitary text, excluding any front or back matter.

<back>

contains any appendices, etc., following the main part of a text.

The majority of OUCS documents only use the body section for the text. This is shown in the next example:

If you do want to include front and/or back additions to your document, the sections are coded in the following manner:

4.1. Sectioning your text

Your text may be just a series of paragraphs, or these paragraphs may be grouped together into chapters, sections, subsections, etc. In the former case, each paragraph is embedded inside a the <p> element. In the latter case, the <body> may be divided into a series of <div> elements, which may be further subdivided. An example of div structure is shown below:

Sectioning your document has important effects on the OUCS web site. Each div used is processed when the document is converted into html. Major divisions are treated as separate web pages and help to form the basis of the internal page navigation system. Each division is also sequentially numbered: 1, 2, 3 ... Where a div section is within another div, it is treated as a subsection and numbered accordingly e.g. 2.1, 2.2, 2.3....

Sectioning documents also influences the HTML output to browsers. The title of a document is always given the <h1> tag, major divisions are thus given the <h2> tag and minor section divisions are given <h3>, <h4>, <h5> etc. depending on how deep they are nested within the document.

Correct structural markup for documentation is important for accessibility. When documents are marked up in a structured way, they allow users of alternative technologies to discover the main sections and subsections more quickly and more easily. The structure allows users to jump from one section to another, without the need to read all of the information on the page. Documents that do not use structured markup pose a problem (to screen reader users in particular), as it is very difficult to find out what is on a page without reading all of the text. Where structural markup has not been used, the author has often employed styles (bold, italic, etc.) to indicate different sections and headings. While obvious to sighted readers, the structure is lost to screen reader users who must read the page to find out if it is of interest to them.

It is a requirement for authors to structure their documents in an accessible manner: relying on style alone is to be avoided as this results in inaccessible documents.

The following elements can be used to divide up your text:

<p>

marks paragraphs in prose.

<div>

contains a subdivision of the front, body, or back of a text.

When structural divisions smaller than a <div> are necessary, inner <div> elements may be used, without limit to the depth of nesting (see example above).

The value of every id attribute must be unique within a document. They may be used to derive the names of HTML pages, so giving sensible mnemonic names is a good idea.

4.2. Headings and Closings

Here is an example of their use:

N.B. At present it is not possible to use the <head> tag without using the <div> tag first.

4.3. Marking Highlighted Phrases

4.4. Cross References and Links

Explicit cross references or links from one point to another in a text in the same XML document may be encoded using the elements described in section 4.4.1. Simple Cross References. References or links to elements of some other XML document, or to parts of non-XML documents, may be encoded using the TEI extended pointers described in section 4.4.2. Extended Pointers.

Accessibility of your links is important. The text you use can either enhance a user's understanding of where the link will lead, or leave them clueless. The worst phrase you can use for a link is Click Here or simply Here: in both instances the user is left with no clear idea of where the link could lead. This problem is compounded for a screen reader user: they can get lists of all links from any given page, but if the author of the page has just said Click Here or Here, they will get a list consisting of just that. The user will be left stranded on the page with no clear way to move forwards in their search for information.

An accessible link is one that conveys both where the link will go and the information the user is likely to find. By default our system will add a title attribute to any link you make on your page when it is transformed into HTML. However, while this is good practice and a nice failsafe measure, it will only add the same text as the link text. This might be adequate in some circumstances, but to make your links more accessible you should add your own additional text using the n attribute. People browsing with modern visual browsers will see your additional link information when they mouse over your link, and screen reader users will have more information about where the link will take them as the title attribute is read out to them.

4.5. Addresses

4.6. Lists

4.7. Tables

The <table> element can also take the align, summary, width, border, frame, rules, cellspacing and cellpadding attributes defined in HTML, and the conversion to HTML will pass them straight through.

4.8. Figures and Graphics

Not all the components of a document are necessarily textual. The most straight forward text will often contain diagrams or illustrations, to say nothing of documents in which image and text are inextricably intertwined, or electronic resources in which the two are complementary. This poses accessibility issues for users who cannot see the images. What are they? Are they important to the text, or just page decoration? Is the image a graph or simple picture? Has the author provided extra information about the graphic for those that cannot see it? If you do not provide alternative text for graphics or other accessibiity features in the page coding, the page will be inaccessible to some visitors.

The following tags and attributes are used to add images to web pages:

A picture is inserted into a document using the url attribute of the <figure> element:

Usually, a graphic will have at the least an identifying title, which should be encoded using the <head> element. Images which are given a head tag have this text automatically converted to a figure caption and are numbered sequentially throughout the document. It is also essential to include a brief description of the image using <figDesc>. If the image is difficult to describe in just a few words, you should provide an alternative page where a full account of the image can be given to the user: this extra information should be provided via a [d] link. These are normal url links to normal web pages. By convention the [d] link should be provided next to the image in question; users needing greater detail about a given image will click on the [d] link for more information.

If the image is for decoration only (very rare on OUCS pages), it is still necessary to include the <figDesc> element in your document, but in this case it should be left blank. By convention the image is then considered just page decoration and unimportant to the reader.

If you want to control the way text flows around an image, use a rend value, as described in the Rends section.

4.9. Getting a TEI file to display a newsfeed

A newsfeed can be displayed by putting a <xptr> element with a rend of rss and a type of transclude inside a <p> element. The url attribute has the URL of the newsfeed. Our XSL can cope with newsfeeds written in RSS 2.0, RSS 1.0 and Atom 1.0.

This will produce output like the following:

By default, 10 items of the feed will get output together with an RSS icon that allows people to subscribe to the newsfeed.

Gotcha: the web page will not change when new items get added to the feed unless you arrange for your page not to be cached by AxKit. Please contact webmaster@oucs.ox.ac.uk to get this done.

4.9.1. rsssummary gives a different style of output

Other components can be added to the rend to control what gets output and how it gets output.

A different style of output is delivered if you add rsssummary to the rend:

<p> <xptr rend="rss rsssummary" type="transclude" url="http://newsrss.bbc.co.uk/rss/newsonline_uk_edition/front_page/rss.xml" /> </p>

This will produce output like the following:

NHS 'cover-up' unacceptable - HuntThere should be "no hiding place" for anyone involved in a cover-up by England's NHS regulator after baby deaths at a Cumbria hospital, the health secretary says.

Lloyds to return to private handsGeorge Osborne uses his annual Mansion House speech to announce the government is preparing to sell its stake in Lloyds Bank back to private investors.

Royal baby's gender to be 'surprise'The Duke and Duchess of Cambridge do not know the gender of their baby, it is revealed - as it is confirmed the birth will take place at St Mary's hospital in London.

Families win Iraq deaths damages bidFamilies of soldiers killed in Iraq can sue the government for negligence and bring damages claims under the Human Rights Act, the Supreme Court rules.

MP Evans 'refutes all allegations'Commons Deputy Speaker Nigel Evans says he "refutes all allegations" after he is arrested on suspicion of three counts of indecent assault.

Obama calls for nuclear arsenal cutsSpeaking at Berlin's Brandenburg Gate, President Barack Obama calls for reductions to be made in US and Russian nuclear stockpiles.

Kerry calls Karzai to defuse tensionThe US Secretary of State John Kerry calls the Afghan leader to defuse tension over the opening of a Taliban office, an Afghan official tells the BBC.

Violent protests mar Brazil matchProtesters clash with police in Brazil's city of Fortaleza ahead of a football match against Mexico, as public discontent over poor public services widens.

Work Programme 'not doing enough'Providers of the government's flagship Work Programme tell ministers the costs of helping sick and disabled jobseekers into employment cannot be met under the scheme.

Ian Brady 'chronically psychotic'Moors Murderer Ian Brady remains "chronically psychotic", an expert at high-security Ashworth Hospital tells his mental health tribunal.

4.10. Using HTML elements in a TEI file

Although TEI is a rich language so that most of what can be coded in HTML can also be coded in TEI, there are occasions when you may want to use some HTML in a TEI document.

If you wish to do this, you need to introduce a namespace that you can use to say that a particular element belongs to HTML rather than to TEI. Usually, the name html is used for this namespace.

If you want some HTML elements to appear in the <head> element of the HTML that gets generated, you should put these elements between the <fileDesc> and the <revisionDesc> elements (that appear in the <teiHeader>).

4.11. Forms to collect data and send a message

It is possible to provide a form (in a TEI file) that collects some data from a user and sends that data to someone in an e-mail message. There are details about this in a document on FormMail.

4.12. Accessibility of documentation

Accessibility of our documentation is paramount to ensure documents are accessible to all readers and for OUCS to stay on the correct side of the law. It is necessary for all OUCS authors to familiarise themselves with the ways and means to make their documents as accessible as possible.

Authors need to make sure that they follow the following guidelines:

• do not make links with the text here or click here, make links that mean something out of context of the sentence they are in. Similarly do not use the same titles for lots of different links on a page when they actually point to different places.
• When using graphics always provide the <figDesc> element. If necessary go the extra step and make a [d] link for longer explanations of figures
• When using tables, make sure they are comprehensible when they are linearised. Always include the summary attribute regardless of whether the table is for layout or data. The latter requires you to give some details of the table's content.
• When you have finished making a web page, you can check its accessibility using online services such as those found in the Complete List of Web Accessibility Evaluation Tools (compiled by the Web Accessibility Initiative (WAI)).

Please use these checkers and make any changes required.

5.1. Generated Divisions

When an index or table of contents is to be encoded (rather than one being generated) for some reason, the <list> element discussed in section 4.6. Lists should be used.