XML 101

Micah Dubinko

Brain Attic, L.L.C.

...Presented By

Micah Dubinko (micah@dubinko.info)

IBM Certified Solution Developer, XML and Related Technologies

Columnist on xml.com

Infoworld Innovator for 2004

http://wiki.brainattic.info

http://dubinko.info/blog/

Available for consulting!

What is XML?

• Foundation: characters (not bytes, nor bits, nor anything digital)
• Specifically Unicode characters
• Unicode defines ~100k unique code points
• US-ASCII, such as it is, is valid Unicode (UTF-8)
• UTF-16 is also supported by all XML processors
• Further reading: Mike J. Brown's XML tutorial

What is XML like?

• Similar to HTML (Derivative of SGML)
• But no predefined tag names, and tighter rules
• In the upcoming slides, each title shows a part of XML
• Well-formed vs. Valid
• But first, a look at the rules...

Reading (selected) XML syntax rules

• Not as hard as it looks
• Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
• document ::= prolog element Misc*
• element ::= EmptyElemTag | STag content ETag
• STag ::= '<' Name (S Attribute)* S? '>'
• Attribute ::= Name Eq AttValue
• ETag ::= '</' Name S? '>'

Character content

• Without content, there'd be nothing to "mark up"
• Beware ISO-8859-1 and "Windows" encoding
• Nearly any Unicode character allowed, except control sequences
• Built in escapes for: < > ' " and &
• Numeric escapes for other characters: &#x1234;
• Custom-defined entity references: '&copyright; -- possible through DTDs

<![CDATA[ sections ]]>

• Special ranges where any character, especially <, is OK
• In fact normal character escapes within don't work
• The only thing not allowed is ]]>
• Cannot nest
• Just another way an XML document can contain character content
• Many APIs and data models (like XSLT) don't differentiate CDATA sections

<Elements/>

• Know these, and you know one third of XML
• No overlaps, like <em><strong>huh?</em></strong>
• An XML document has one (and only one) "root element"
• <e>with content</e> or <empty/>
• XML Namespaces complicates things
• <prefix:e xmlns:prefix="http://some_long_url/that's/hard/to/remember"> content </prefix:e>

thisSlide="Attributes"

• Additional information attached to elements
• Unique and non-ordered
• Quotes around value not optional, but can be single or double
• XML Namespaces really complicates things
• If it starts with xmlns, it's sort of not really an attribute
• Default namespaces don't apply
• To process, you might need to look ahead

<!-- Comments -->

• Generally for humans (don't stick machine-readable stuff here)
• Don't nest
• Don't include -- anywhere (including ---; --------------; etc.)
• "conditional comments" are terrible hack <!--[if IE 5]> .. <![endif]-->

<?Processing instructions?>

• Advantage: PIs are not constrained by DTD or schema validation
• Disadvantage: PIs are not constrained by DTD or schema validation
• Contents often look like attributes, but are just characters.
• The <?xml declaration at the top is not a PI, though a few APIs incorrectly treat it as such

<?xml version='1.0'?>

• The XML prolog, usually a good idea to include
• version: required
• encoding: recommended (and required when not UTF-8 or UTF-16)
• standalone: rarely used

<!DOCTYPE declaration>

• Often required by specific XML vocabularies (like XHTML)
• Something of a holdover from SGML
• Declared name must match root element name
• Can refer to an external DTD
• Can refer to an internal DTD subset
• Unpleasant interactions with namespaces

Document Type Definitions (DTDs)

• Use different syntax than what we've seen so far
• Can declare elements and attributes
• Can define attributes of type ID (unique)
• Can define character entities (in fact, the only way to do so)

...Questions?

Ask micah@dubinko.info