An Introduction to DTD
K. Yue, copyright 2000,
November 22, 2000

Introduction

• A DTD is a grammar that is used to determine the validity of an XML document.
• An XML document satisfying the rules of a DTD is said to be validated.
• DTD is used to specify additional constraints and rules for a given vocabulary, such as:
  • element nesting rules.
  • attribute name and value constraints.
• DTD allows XML parsers to capture errors as soon as possible.

Example

Consider the XML document:

<person>
<name>Adam</name>
<spouse>M3774</spouse>
<spouse>Eva</spouse>
</person>

Potential questions:

• What to do with two spouses?
• Is "Eva" a person? Maybe it should be a <pet>.
• Any information about "M3774" or "Eva"?

Example:

A better approach is to specify the constraints using DTD.

• A person may only have up to one spouse.
• A spouse must refer to a person in the same XML doc.
  

A possible DTD declaration for this:

<!ELEMENT person (name, pet+)>
<!ATTLIST person
          id ID #REQUIRED
          spouse IDREF #IMPLIED>

An XML document satisfying the DTD:

<person id="p12324" spouse="p10001">
<name>Adam</name>
<pet>Eva</pet>
</person>
<person id="p10001">
<name>M3774</name>
</person>

DTD Declarations

• The <DOCTYPE> tag is used in XML to associate to DTD declarations.
• DTD declares the following constructs:
  • ELEMENT: XML element types
  • ATTLIST: attributes of an element
  • ENTITY: reusable content referenced by the &…; syntax
  • NOTATION: external contents not to be parsed.

Element Declarations

• Format: <!ELEMENT element_name element_declaration>
• Element_declaration can be:
  • EMPTY: empty element
  • ANY: may contain anything; no parsing checking (thus defeating the purpose of DTD).
  • PCDATA (parsed character data): text.
  • Content model: a declaration of contents enclosed by ( & ) for specifying children elements.
• Content model:
  • Declare children elements
  • May use the order operators:
    • , (comma): sequencing.
    • | (vertical bar): option, choice, "or".
  • May use cardinality operators:
    • +: one or many.
    • *: zero or many.
    • ?: zero or one.

Example:

<!ELEMENT name (first, middleinitial?, last)>

OK:

<name><first>Bun</first><last>Yue</last></name>

Not OK:

<name><last>Yue</last><first>Bun</first ></name>
<name>The one and only:
<middleinitial>K</middleinitial>
<first>Bun</first><last>Yue</last></name>

Example:

Can you spot the potential implications?

<!ELEMENT bookcollection (book+) >
<!ELEMENT book (author, publisher, ISBN, chapter*)>
<!ELEMENT author (#PCDATA)>
<!ELEMENT publisher (#PCDATA)>
<!ELEMENT ISBN (#PCDATA)>
<!ELEMENT chapter (#PCDATA)>
<!ELEMENT bookcollection (book+) >
<!ELEMENT book (author, publisher, ISBN, chapter*)>
<!ELEMENT author (#PCDATA)>
<!ELEMENT publisher (#PCDATA)>
<!ELEMENT ISBN (#PCDATA)>
<!ELEMENT chapter (#PCDATA)>

ATTList Declarations

• To declare attribute properties of an element.
• Format: <!ATTLIST element_name attribute_declarations>
• Attribute declarations declare one or more attributes.
• Each attribute declaration includes the attribute name, its type and a setting.

Example:

<!ATTLIST person
    comment CDATA #IMPLIED>
<!ATTLIST person
    ssn ID #REQUIRED
    gender (male|female) #IMPLIED
    age NMTOKEN #IMPLIED
    iq NMTOKEN "100">

OK:

<person ssn="123456789">…
</person>
<person ssn="111-11-1111"
  gender="female" iq="200">…</person>
<person ssn="abc">…</person>

Not OK:

<person>…</person>
<person ssn="123" gender="unknown">…</person>
<person ssn="123" age="???">…</person>

• Attribute data types may be:
  • CDATA: string, least restrictive.
  • ID: an unique name string within the document: like a primary key of the element.
  • IDREF: refer to an ID value of some other elements.
  • IDREFS: refer to many ID values separated by a space.
  • ENTITY: name of a pre-defined external entity.
  • ENTITIES: many entity names separated by a space.
  • NMTOKEN: a name formed by alphanumeric characters only (including ".", " "-" and ":").
  • NMTOKENS: many NMTOKENS separated by a white space.
  • NOTATION: for referencing data other than XML.
  • Enumeration: providing explicit choices separated by | within a pair of parenthesis.
    
• Attribute values and settings may be:
  • #REQUIRD: mandatory.
  • #IMPLIED: optional.
  • Default value: if the attribute is missing, assume default value.
  • #FIXED and default value: only one possible value of the attribute: the default value.

Example:

simpleAddress.DTD

<!ELEMENT addressBook (person)+>
<!ELEMENT person (name,email*,link?)>
<!ATTLIST person id ID #REQUIRED>
<!ATTLIST person
    gender (male|female) #IMPLIED
    luckynumber (NMTOKENS) #IMPLIED>
<!ELEMENT name (#PCDATA|(first,last))|(last,first))>
<!ELEMENT first (NMTOKEN)>
<!ELEMENT last (#PCDATA)>
<!ELEMENT email (#PCDATA)>
<!ELEMENT link EMPTY>
<!ATTLIST link
    spouse IDREF #IMPLIED
    children IDREFS #IMPLIED>

An XML document satisfying the DTD:

<?xml version="1.0"?>
<!DOCTYPE addressBook SYSTEM "http://yourserver/dtd/addressBook.dtd">
<addressBook>
<person id="123456789" gender="male" luckynumber="7 12 3">
  <name>
    <last>Hope</last> <first>Bob</first>
  </name>
  <email>BobHope@hollywood.com</email>
</person>
<person id="222222222" gender="female">
  <name>Deborah King</name>
  <link spouse="222222223" children="222222226 222222227" />
</person>
<person id="222222223" gender="male">
  <name>Jim King</name>
</person>
<person id="222222226" gender="male">
  <name>John King</name>
</person>
<person id="222222227" gender="female">
  <name>Jane King</name>
</person>
</addressBook>