XML in a Nutshell, 3rd Edition

Attribute Types

In merely well-formed XML, attribute values can be any string of text. The only restrictions are that any occurrences of < or & must be escaped as < and &, and whichever kind of quotation mark, single or double, is used to delimit the value must also be escaped. However, a DTD allows you to make somewhat stronger statements about the content of an attribute value. Indeed, these are stronger statements than can be made about the contents of an element. For instance, you can say that an attribute value must be unique within the document, that it must be a legal XML name token, or that it must be chosen from a fixed list of values.

There are 10 attribute types in XML. They are:

• CDATA

• NMTOKEN

• NMTOKENS

• Enumeration

• ENTITY

• ENTITIES

• ID

• IDREF

• IDREFS

• NOTATION

These are the only attribute types allowed. A DTD cannot say that an attribute value must be an integer or a date between 1966 and 2004, for example.

CDATA

A CDATA attribute value can contain any string of text acceptable in a well-formed XML attribute value. This is the most general attribute type. For example, you would use this type for an alt attribute of an image element because there’s no particular form the text in such an attribute has to follow.

<!ATTLIST image alt CDATA #IMPLIED>

You would also use this for other kinds of data such as prices, URLs, email and snail mail addresses, citations, and other types that—while they have more structure than a simple string of text—don’t match any of the other attribute types. ...