Search the Catalog
DocBook: The Definitive Guide

DocBook: The Definitive Guide

By Norman Walsh & Leonard Muellner
1st Edition October 1999
1-56592-580-7, Order Number: 5807
652 pages, $36.95 , Includes CD-ROM

Chapter 2. Creating DocBook Documents

This chapter explains in concrete, practical terms how to make DocBook documents. It's an overview of all the kinds of markup that are possible in DocBook documents. It explains how to create several kinds of DocBook documents: books, sets of books, chapters, articles, and reference manual entries. The idea is to give you enough basic information to actually start writing. The information here is intentionally skeletal; you can find "the details" in the reference section of this book.

Before we can examine DocBook markup, we have to take a look at what an SGML or XML system requires.

2.1. Making an SGML Document

SGML requires that your document have a specific prologue. The following sections describe the features of the prologue.

2.1.1. An SGML Declaration

SGML documents begin with an optional SGML Declaration. The declaration can precede the document instance, but generally it is stored in a separate file that is associated with the DTD. The SGML Declaration is a grab bag of SGML defaults. DocBook includes an SGML Declaration that is appropriate for most DocBook documents, so we won't go into a lot of detail here about the SGML Declaration.

In brief, the SGML Declaration describes, among other things, what characters are markup delimiters (the default is angle brackets), what characters can compose tag and attribute names (usually the alphabetical and numeric characters plus the dash and the period), what characters can legally occur within your document, how long SGML "names" and "numbers" can be, what sort of minimizations (abbreviation of markup) are allowed, and so on. Changing the SGML Declaration is rarely necessary, and because many tools only partially support changes to the declaration, changing it is best avoided, if possible.

Wayne Wholer has written an excellent tutorial on the SGML Declaration; if you're interested in more details, see

2.1.2. A Document Type Declaration

All SGML documents must begin with a document type declaration. This identifies the DTD that will be used by the document and what the root element of the document will be. A typical doctype declaration for a DocBook document looks like this:

<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V3.1//EN">

This declaration indicates that the root element, which is the first element in the hierarchical structure of the document, will be <book> and that the DTD used will be the one identified by the public identifier -//OASIS//DTD DocBook V3.1//EN. See Section 2.3.1" later in this chapter.

2.1.3. An Internal Subset

It's also possible to provide additional declarations in a document by placing them in the document type declaration:

<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V3.1//EN" [
<!ENTITY nwalsh "Norman Walsh">
<!ENTITY chap1 SYSTEM "chap1.sgm">
<!ENTITY chap2 SYSTEM "chap2.sgm">

These declarations form what is known as the internal subset. The declarations stored in the file referenced by the public or system identifier in the DOCTYPE declaration is called the external subset and it is technically optional. It is legal to put the DTD in the internal subset and to have no external subset, but for a DTD as large as DocBook that wouldn't make much sense.


The internal subset is parsed first and, if multiple declarations for an entity occur, the first declaration is used. Declarations in the internal subset override declarations in the external subset.

2.1.4. The Document (or Root) Element

Although comments and processing instructions may occur between the document type declaration and the root element, the root element usually immediately follows the document type declaration:

<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V3.1//EN" [
<!ENTITY nwalsh "Norman Walsh">
<!ENTITY chap1 SYSTEM "chap1.sgm">
<!ENTITY chap2 SYSTEM "chap2.sgm">

You cannot place the root element of the document in an external entity.

2.1.5. Typing an SGML Document

If you are entering SGML using a text editor such as Emacs or vi, there are a few things to keep in mind.[1] Using a structured text editor designed for SGML hides most of these issues.

  • DocBook element and attribute names are not case-sensitive. There's no difference between <Para> and <pArA>. Entity names are case-sensitive, however.

    If you are interested in future XML compatibility, input all element and attribute names strictly in lowercase.

  • If attribute values contain spaces or punctuation characters, you must quote them. You are not required to quote attribute values if they consist of a single word or number, although it is not wrong to do so.

    When quoting attribute values, you can use either a straight single quote ('), or a straight double quote ("). Don't use the "curly" quotes (" and ") in your editing tool.

    If you are interested in future XML compatibility, always quote all attribute values.

  • Several forms of markup minimization are allowed, including empty tags. Instead of typing the entire end tag for an element, you can type simply </>. For example:

    This is <emphasis>important</>: never stick the tines of a fork
    in an electrical outlet.

    You can use this technique for any and every tag, but it will make your documents very hard to understand and difficult to debug if you introduce errors. It is best to use this technique only for inline elements containing a short string of text.

    Empty start tags are also possible, but may be even more confusing. For the record, if you encounter an empty start tag, the SGML parser uses the element that ended last:

    This is <emphasis>important</>.  So is <>this</>.

    Both "important" and "this" are emphasized.

    If you are interested in future XML compatibility, don't use any of these tricks.

  • The null end tag (net) minimization feature allows constructions like this:

    This is <emphasis/important/: never stick the tines of a fork
    in an electrical outlet.

    If, instead of ending a start tag with >, you end it with a slash, then the next occurrence of a slash ends the element.

    If you are interested in future XML compatibility, don't use net tag minimization either.

If you are willing to modify both the declaration and the DTD, even more dramatic minimizations are possible, including completely omitted tags and "shortcut" markup.

NoteRemoving Minimizations

Although we've made a point of reminding you about which of these minimization features are not valid in XML, that's not really a sufficient reason to avoid using them. (The fact that many of the minimization features can lead to confusing, difficult-to-author documents might be.)

If you want to convert one of these documents to XML at some point in the future, you can run it through a program like sgmlnorm, which will remove all the minimizations and insert the correct, verbose markup. The sgmlnorm program is part of the SP and Jade distributions, which are on the CD-ROM.

2.2. Making an XML Document

In order to create DocBook documents in XML, you'll need an XML version of DocBook. We've included one on the CD, but it hasn't been officially adopted by the OASIS DocBook Technical Committee yet. If you're interested in the technical details, Appendix B, describes the specific differences between SGML and XML versions of DocBook.

XML, like SGML, requires a specific prologue in your document. The following sections describe the features of the XML prologue.

2.2.1. An XML Declaration

XML documents should begin with an XML declaration. Unlike the SGML declaration, which is a grab bag of features, the XML declaration identifies a few simple aspects of the document:

<?xml version="1.0" standalone="no"?>

Identifying the version of XML ensures that future changes to the XML specification will not alter the semantics of this document. The standalone declaration simply makes explicit the fact that this document cannot "stand alone," and that it relies on an external DTD. The complete details of the XML declaration are described in the XML specification.

2.2.2. A Document Type Declaration

Strictly speaking, XML documents don't require a DTD. Realistically, DocBook XML documents will have one.

The document type declaration identifies the DTD that will be used by the document and what the root element of the document will be. A typical doctype declaration for a DocBook document looks like this:

<?xml version='1.0'?>
<!DOCTYPE book PUBLIC "-//Norman Walsh//DTD DocBk XML V3.1.4//EN"

This declaration indicates that the root element will be <book> and that the DTD used will be the one indentified by the public identifier -//Norman Walsh//DTD DocBk XML V3.1.4//EN. External declarations in XML must include a system identifier (the public identifier is optional). In this example, the DTD is stored on a web server.

System identifiers in XML must be URIs. Many systems may accept filenames and interpret them locally as file: URLs, but it's always correct to fully qualify them.

2.2.3. An Internal Subset

It's also possible to provide additional declarations in a document by placing them in the document type declaration:

<?xml version='1.0'?>
<!DOCTYPE book PUBLIC "-//Norman Walsh//DTD DocBk XML V3.1.4/EN"
                         "" [
<!ENTITY nwalsh "Norman Walsh">
<!ENTITY chap1 SYSTEM "chap1.sgm">
<!ENTITY chap2 SYSTEM "chap2.sgm">

These declarations form what is known as the internal subset. The declarations stored in the file referenced by the public or system identifier in the DOCTYPE declaration is called the external subset, which is technically optional. It is legal to put the DTD in the internal subset and to have no external subset, but for a DTD as large as DocBook, that would make very little sense.


The internal subset is parsed first in XML and, if multiple declarations for an entity occur, the first declaration is used. Declarations in the internal subset override declarations in the external subset.

2.2.4. The Document (or Root) Element

Although comments and processing instructions may occur between the document type declaration and the root element, the root element usually immediately follows the document type declaration:

<?xml version='1.0'?>
<!DOCTYPE book PUBLIC "-//Norman Walsh//DTD DocBk XML V3.1.4//EN"
                         "" [
<!ENTITY nwalsh "Norman Walsh">
<!ENTITY chap1 SYSTEM "chap1.sgm">
<!ENTITY chap2 SYSTEM "chap2.sgm">

The important point is that the root element must be physically present immediately after the document type declaration. You cannot place the root element of the document in an external entity.

2.2.5. Typing an XML Document

If you are entering SGML using a text editor such as Emacs or vi, there are a few things to keep in mind. Using a structured text editor designed for XML hides most of these issues.

  • In XML, all markup is case-sensitive. In the XML version of DocBook, you must always type all element, attribute, and entity names in lowercase.

  • You are required to quote all attribute values in XML.

    When quoting attribute values, you can use either a straight single quote ('), or a straight double quote ("). Don't use the "curly" quotes (" and ") in your editing tool.

  • Empty elements in XML are marked with a distinctive syntax: <xref/>.

  • Processing instructions in XML begin and end with a question mark: <?pitarget data?>.

  • XML was designed to be served, received, and processed over the Web. Two of its most important design principles are ease of implementation and interoperability with both SGML and HTML.

    The markup minimization features in SGML documents make it more difficult to process, and harder to write a parser to interpret it; these minimization features also run counter to the XML design principles named above. As a result, XML does not support them.

    Luckily, a good authoring environment can offer all of the features of markup minimization without interfering with the interoperability of documents. And because XML tools are easier to write, it's likely that good, inexpensive XML authoring environments will be available eventually.

2.2.6. XML and SGML Markup Considerations in This Book

Conceptually, almost everything in this book applies equally to SGML and XML. But because DocBook V3.1 is an SGML DTD, we naturally tend to use SGML conventions in our writing. If you're primarily interested in XML, there are just a few small details to keep in mind.

  • XML is case-sensitive, while the SGML version of DocBook is not. In this book, we've chosen to present the element names using mixed case (Book, indexterm, XRef, and so on), but in the DocBook XML DTD, all element, attribute, and entity names are strictly lowercase.

  • Empty element start tags in XML are marked with a distinctive syntax: <xref/>. In SGML, the trailing slash is not present, so some of our examples need slight revisions to be valid XML elements.

  • Processing instructions in XML begin and end with a question mark: <?pitarget data?>. In SGML, the trailing question mark is not present, so some of our examples need slight revisions to be valid XML elements.

  • Generally we use public identifiers in examples, but whenever system identifiers are used, don't forget that XML system identifiers must be Uniform Resource Indicators (URIs), in which SGML system identifiers are usually simple filenames.

For a more detailed discussion of DocBook and XML, see Appendix B.

2.3. Public Identifiers, System Identifiers, and Catalog Files

When a DTD or other external file is referenced from a document, the reference can be specified in three ways: using a public identifier, a system identifier, or both. In XML, the system identifier is generally required and the public identifier is optional. In SGML, neither is required, but at least one must be present.[2]

A public identifier is a globally unique, abstract name, such as the following, which is the official public identifier for DocBook V3.1:

-//OASIS//DTD DocBook V3.1//EN

The introduction of XML has added some small complications to system identifiers. In SGML, a system identifier generally points to a single, local version of a file using local system conventions. In XML, it must point with a Uniform Resource Indicator (URI). The most common URI today is the Uniform Resource Locator (URL), which is familiar to anyone who browses the Web. URLs are a lot like SGML system identifiers, because they generally point to a single version of a file on a particular machine. In the future, Uniform Resource Names (URN), another form of URI, will allow XML system identifiers to have the abstract characteristics of public identifiers.

The following filename is an example of an SGML system identifier:

An equivalent XML system identifier might be:

The advantage of using the public identifier is that it makes your documents more portable. For any system on which DocBook is installed, the public identifier will resolve to the appropriate local version of the DTD (if public identifiers can be resolved at all).

Public identifiers have two disadvantages:

  • Because XML does not require them, and because system identifiers are required, developing XML tools may not provide adequate support for public identifiers. To work with these systems you must use system identifiers.

  • Public identifiers aren't magical. They're simply a method of indirection. For them to work, there must be a resolution mechanism for public identifiers. Luckily, several years ago, SGML Open (now OASIS) described a standard mechanism for mapping public identifiers to system identifers using catalog files.

    See OASIS Technical Resolution 9401:1997 (Amendment 2 to TR 9401).

2.3.1. Public Identifiers

An important characteristic of public identifiers is that they are globally unique. Referring to a document with a public identifier should mean that the identifier will resolve to the same actual document on any system even though the location of that document on each system may vary. As a rule, you should never reuse public identifiers, and a published revision should have a new public identifier. Not following these rules defeats one purpose of the public identifier.

A public identifier can be any string of upper- and lowercase letters, digits, any of the following symbols: "'", "(", ")", "+", ",", "-", ".", "/", ":", "=", "?", and white space, including line breaks. Formal public identifiers

Most public identifiers conform to the ISO 8879 standard that defines formal public identifiers. Formal public identifiers, frequently referred to as FPI, have a prescribed format that can ensure uniqueness:[3]

prefix//owner-identifier//text-class text-description//language//display-version

Here are descriptions of the identifiers in this string:


The prefix is either a "+" or a "-" Registered public identifiers begin with "+"; unregistered identifiers begin with "-".

(ISO standards sometimes use a third form beginning with ISO and the standard number, but this form is only available to ISO.)

The purpose of registration is to guarantee a unique owner-identifier. There are few authorities with the power to issue registered public identifiers, so in practice unregistered identifiers are more common.

The Graphics Communication Association (GCA) can assign registered public identifiers. They do this by issuing the applicant a unique string and declaring the format of the owner identifier. For example, the Davenport Group was issued the string "A00002" and could have published DocBook using an FPI of the following form:

+//ISO/IEC 9070/RA::A00002//...

Another way to use a registered public identifier is to use the format reserved for internet domain names. For example, O'Reilly can issue documents using an FPI of the following form:


As of DocBook V3.1, the OASIS Technical Committee responsible for DocBook has elected to use the unregistered owner identifier, OASIS, thus its prefix is -.



Identifies the person or organization that owns the identifier. Registration guarantees a unique owner identifier. Short of registration, some effort should be made to ensure that the owner identifier is globally unique. A company name, for example, is a reasonable choice as are Internet domain names. It's also not uncommon to see the names of individuals used as the owner-identifier, although clearly this may introduce collisions over time.

The owner-identifier for DocBook V3.1 is OASIS. Earlier versions used the owner-identifier Davenport.


The text class identifies the kind of document that is associated with this public identifier. Common text classes are


An SGML or XML document.


A DTD or part of a DTD.


A collection of element declarations.


A collection of entity declarations.


Data that is not in SGML or XML.

DocBook is a DTD, thus its text class is DTD.


This field provides a description of the document. The text description is free-form, but cannot include the string //.

The text description of DocBook is DocBook V3.1.

In the uncommon case of unavailable public texts (FPIs for proprietary DTDs, for example), there are a few other options available (technically in front of or in place of the text description), but they're rarely used. [4]


Indicates the language in which the document is written. It is recommended that the ISO standard two-letter language codes be used if possible.

DocBook is an English-language DTD, thus its language is EN.


This field, which is not frequently used, distinguishes between public texts that are the same except for the display device or system to which they apply.

For example, the FPI for the ISO Latin 1 character set is:

-//ISO 8879-1986//ENTITIES Added Latin 1//EN

A reasonable FPI for an XML version of this character set is:

-//ISO 8879-1986//ENTITIES Added Latin 1//EN//XML

2.3.2. System Identifiers

System identifiers are usually filenames on the local system. In SGML, there's no constraint on what they can be. Anything that your SGML processing system recognizes is allowed. In XML, system identifiers must be URIs (Uniform Resource Identifiers).

The use of URIs as system identifiers introduces the possibility that a system identifier can be a URN. This allows the system identifier to benefit from the same global uniqueness benefit as the public identifier. It seems likely that XML system identifiers will eventually move in this direction.

2.3.3. Catalog Files

Catalog files are the standard mechanism for resolving public identifiers into system identifiers. Some resolution mechanism is necessary because DocBook refers to its component modules with public identifiers, and those must be mapped to actual files on the system before any piece of software can actually load them.

The catalog file format was defined in 1994 by SGML Open (now OASIS). The formal specification is contained in OASIS Technical Resolution 9401:1997.

Informally, a catalog is a text file that contains a number of keyword/value pairs. The most frequently used keywords are PUBLIC, SYSTEM, SGMLDECL, DTDDECL, CATALOG, OVERRIDE, DELEGATE, and DOCTYPE.


The PUBLIC keyword maps public identifiers to system identifiers:

PUBLIC "-//OASIS//DTD DocBook V3.1//EN" "docbook/3.1/docbook.dtd"

The SYSTEM keyword maps system identifiers to system identifiers:


The SGMLDECL keyword identifies the system identifier of the SGML Declaration that should be used:

SGMLDECL "docbook/3.1/docbook.dcl"

Like SGMLDECL, DTDDECL identifies the SGML Declaration that should be used. DTDDECL associates a declaration with a particular public identifier for a DTD:

DTDDECL "-//OASIS//DTD DocBook V3.1//EN" "docbook/3.1/docbook.dcl"

Unfortunately, it is not supported by the free tools that are available. The practical benefit of DTDDECL can usually be achieved, albeit in a slightly cumbersome way, with multiple catalog files.


The CATALOG keyword allows one catalog to include the content of another. This can make maintenance somewhat easier and allows a system to directly use the catalog files included in DTD distributions. For example, the DocBook distribution includes a catalog file. Rather than copying each of the declarations in that catalog into your system catalog, you can simply include the contents of the DocBook catalog:

CATALOG "docbook/3.1/catalog"

The OVERRIDE keyword indicates whether or not public identifiers override system identifiers. If a given declaration includes both a system identifer and a public identifier, most systems attempt to process the document referenced by the system identifier, and consequently ignore the public identifier. Specifying

in the catalog informs the processing system that resolution should be attempted first with the public identifier.


The DELEGATE keyword allows you to specify that some set of public identifiers should be resolved by another catalog. Unlike the CATALOG keyword, which loads the referenced catalog, DELEGATE does nothing until an attempt is made to resolve a public identifier.

The DELEGATE entry specifies a partial public identifier and an alternate catalog:

DELEGATE "-//OASIS" "/usr/sgml/oasis/catalog"

Partial public identifers are simply initial substring matches. Given the preceding entry, if an attempt is made to match any public identifier that begins with the string -//OASIS, the alternate catalog /usr/sgml/oasis/catalog will be used instead of the current catalog.


The DOCTYPE keyword allows you to specify a default system identifier. If an SGML document begins with a DOCTYPE declaration that specifies neither a public identifier nor a system identifier (or is missing a DOCTYPE declaration altogether), the DOCTYPE declaration may provide a default:

DOCTYPE BOOK n:/share/sgml/docbook/3.1/docbook.dtd

A small fragment of an actual catalog file is shown in Example 2-1.

Example 2-1. A Sample Catalog

-- Comments are delimited by pairs of double-hyphens,      (1)
   as in SGML and XML comments. --

OVERRIDE YES                                               (2)

SGMLDECL "n:/share/sgml/docbook/3.1/docbook.dcl"           (3)

DOCTYPE  BOOK  n:/share/sgml/docbook/3.1/docbook.dtd       (4)

PUBLIC "-//OASIS//DTD DocBook V3.1//EN"                    (5)

SYSTEM ""      (6)
Catalog files may also include comments.
This catalog specifies that public identifiers should be used in favor of system identifiers, if both are present.
The default declaration specified by this catalog is the DocBook declaration.
Given an explicit (or implied) SGML DOCTYPE of
use n:/share/sgml/docbook/3.1/docbook.dtd as the default system identifier. Note that this can only apply to SGML documents because the DOCTYPE declaration above is not a valid XML element.
Map the OASIS public identifer to the local copy of the DocBook V3.1 DTD.
Map a system identifer for the XML version of DocBook to a local version.

A few notes:

  • It's not uncommon to have several catalog files. See below, Section".

  • Like attributes on elements you can quote, the public identifier and system identifier are surrounded by either single or double quotes.

  • White space in the catalog file is generally irrelevant. You can use spaces, tabs, or new lines between keywords and their arguments.

  • When a relative system identifier is used, it is considered to be relative to the location of the catalog file, not the document being processed. Locating catalog files

Catalog files go a long way towards making documents more portable by introducing a level of indirection. A problem still remains, however: how does a processor locate the appropriate catalog file(s)? OASIS outlines a complete interchange packaging scheme, but for most applications the answer is simply that the processor looks for a file called catalog or CATALOG.

Some applications allow you to specify a list of directories that should be examined for catalog files. Other tools allow you to specify the actual files.

Note that even if a list of directories or catalog files is provided, applications may still load catalog files that occur in directories in which other documents are found. For example, SP and Jade always load the catalog file that occurs in the directory in which a DTD or document resides, even if that directory is not on the catalog file list.

2.4. Physical Divisions: Breaking a Document into Physical Chunks

The rest of this chapter describes how you can break documents into logical chunks, such as books, chapters, sections, and so on. Before we begin, and while the subject of the internal subset is fresh in your mind, let's take a quick look at how to break documents into separate physical chunks.

Actually, we've already told you how to do it. If you recall, in the preceding sections we had declarations of the form:

<!ENTITY name SYSTEM "filename">
If you refer to the entity name in your document after this declaration, the system will insert the contents of the file filename into your document at that point. So, if you've got a book that consists of three chapters and two appendixes, you might create a file called book.sgm, which looks like this:

<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V3.1//EN" [
<!ENTITY chap1 SYSTEM "chap1.sgm">
<!ENTITY chap2 SYSTEM "chap2.sgm">
<!ENTITY chap3 SYSTEM "chap3.sgm">
<!ENTITY appa SYSTEM "appa.sgm">
<!ENTITY appb SYSTEM "appb.sgm">
<book><title>My First Book</title>

You can then write the chapters and appendixes conveniently in separate files. Note that these files do not and must not have document type declarations.

For example, Chapter 1 might begin like this:

<chapter id="ch1"><title>My First Chapter</title>
<para>My first paragraph.</para>

But it should not begin with its own document type declaration:

<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
<chapter id="ch1"><title>My First Chapter</title>
<para>My first paragraph.</para>

2.5. Logical Divisions: The Categories of Elements in DocBook

DocBook elements can be divided broadly into these categories:

Divisions, which divide books into parts
Components, which divide books or divisions into chapters
Sections, which subdivide components
Meta-information elements
Block elements
Inline elements

In the rest of this section, we'll describe briefly the elements that make up these categories. This section is designed to give you an overview. It is not an exhaustive list of every element in DocBook.

For more information about any specific element and the elements that it may contain, consult the reference page for the element in question.

2.5.1. Sets

A Set contains two or more Books. It's the hierarchical top of DocBook. You use the Set tag, for example, for a series of books on a single subject that you want to access and maintain as a single unit, such as the manuals for an airplane engine or the documentation for a programming language.

2.5.2. Books

A Book is probably the most common top-level element in a document. The DocBook definition of a book is very loose and general. Given the variety of books authored with DocBook and the number of different conventions for book organization used in countries around the world, attempting to impose a strict ordering of elements can make the content model extremely complex. But DocBook gives you free reign. It's very reasonable to use a local customization layer to impose a more strict ordering for your applications.

Books consist of a mixture of the following elements:


Dedication pages almost always occur at the front of a book.

Navigational Components

There are a few component-level elements designed for navigation: ToC, for Tables of Contents; LoT, for Lists of Titles (for lists of figures, tables, examples, and so on); and Index, for indexes.


Divisions are the first hierarchical level below Book. They contain Parts and References. Parts, in turn, contain components. References contain RefEntrys. These are discussed more thoroughly in Section 2.9".

Books can contain components directly and are not required to contain divisions.


These are the chapter-like elements of a Book.

2.5.3. Components

Components are the chapter-like elements of a Book or Part: Preface, Chapter, Appendix, Glossary, and Bibliography. Articles can also occur at the component level. We describe Articles in more detail in the section titled Section 2.8". Components generally contain block elements and/or sections, and some can contain navigational components and RefEntrys.

2.5.4. Sections

There are several flavors of sectioning elements in DocBook:

Sect1Sect5 elements

The Sect1Sect5 elements are the most common sectioning elements. They can occur in most component-level elements. These numbered section elements must be properly nested (Sect2s can only occur inside Sect1s, Sect3s can only occur inside Sect2s, and so on). There are five levels of numbered sections.

Section element

The Section element, introduced in DocBook V3.1, is an alternative to numbered sections. Sections are recursive, meaning that you can nest them to any depth desired.

SimpleSect element

In addition to numbered sections, there's the SimpleSect element. It is a terminal section that can occur at any level, but it cannot have any other sectioning element nested within it.


A BridgeHead provides a section title without any containing section.

RefSect1RefSect3 elements

These elements, which occur only in RefEntrys, are analogous to the numbered section elements in components. There are only three levels of numbered section elements in a RefEntry.

GlossDiv, BiblioDiv, and IndexDiv

Glossarys, Bibliographys, and Indexes can be broken into top-level divisions, but not sections. Unlike sections, these elements do not nest.

2.5.5. Meta-Information

All of the elements at the section level and above include a wrapper for meta-information about the content. See, for example, BookInfo.

The meta-information wrapper is designed to contain bibliographic information about the content (Author, Title, Publisher, and so on) as well as other meta-information such as revision histories, keyword sets, and index terms.

2.5.6. Block Elements

The block elements occur immediately below the component and sectioning elements. These are the (roughly) paragraph-level elements in DocBook. They can be divided into a number of categories: lists, admonitions, line-specific environments, synopses of several sorts, tables, figures, examples, and a dozen or more miscellaneous elements. Lists

There are seven list elements in DocBook:


A list of CallOuts and their descriptions. CallOuts are marks, frequently numbered and typically on a graphic or verbatim environment, that are described in a CalloutList, outside the element in which they occur.


A list of glossary terms and their definitions.


An unordered (bulleted) list. There are attributes to control the marks used.


A numbered list. There are attributes to control the type of enumeration.


A repeating set of named items. For example, a list of states and their capitals might be represented as a SegmentedList.


An unadorned list of items. SimpleLists can be inline or arranged in columns.


A list of terms and definitions or descriptions. (This list of list types is a VariableList.) Admonitions

There are five types of admonitions in DocBook: Caution, Important, Note, Tip, and Warning.

All of the admonitions have the same structure: an optional Title followed by paragraph-level elements. The DocBook DTD does not impose any specific semantics on the individual admonitions. For example, DocBook does not mandate that Warnings be reserved for cases where bodily harm can result. Line-specific environments

These environments preserve whitespace and line breaks in the source text. DocBook does not provide the equivalent of HTML's BR tag, so there's no way to interject a line break into normal running text.


The Address element is intended for postal addresses. In addition to being line-specific, Address contains additional elements suitable for marking up names and addresses.


A LiteralLayout does not have any semantic association beyond the preservation of whitespace and line breaks. In particular, while ProgramListing and Screen are frequently presented in a fixed-width font, a change of fonts is not necessarily implied by LiteralLayout.


A ProgramListing is a verbatim environment, usually presented in Courier or some other fixed-width font, for program sources, code fragments, and similar listings.


A Screen is a verbatim or literal environment for text screen-captures, other fragments of an ASCII display, and similar things. Screen is also a frequent catch-all for any verbatim text.


ScreenShot is actually a wrapper for a Graphic intended for screen shots of a GUI for example.


A Synopsis is a verbatim environment for command and function synopsis. Examples, figures, and tables

Examples, Figures, and Tables are common block-level elements: Example, InformalExample, Figure, InformalFigure, Table, and InformalTable.

The distinction between formal and informal elements is that formal elements have titles while informal ones do not. The InformalFigure element was introduced in DocBook V3.1. In prior versions of DocBook, you could only achieve the effect of an informal figure by placing its content, unwrapped, at the location where the informal figure was desired. Paragraphs

There are three paragraph elements: Para, SimPara (simple paragraphs may not contain other block-level elements), and FormalPara (formal paragraphs have titles). Equations

There are two block-equation elements, Equation and InformalEquation (for inline equations, use InlineEquation).

Informal equations don't have titles. For reasons of backward-compatibility, Equations are not required to have titles. However, it may be more difficult for some stylesheet languages to properly enumerate Equations if they lack titles. Graphics

Graphics occur most frequently in Figures and ScreenShots, but they can also occur without a wrapper. DocBook considers a Graphic a block element, even if it appears to occur inline. For graphics that you want to be represented inline, use InlineGraphic.

DocBook V3.1 introduced a new element to contain graphics and other media types: MediaObject and its inline cousin, InlineMediaObject. These elements may contain video, audio, image, and text data. A single media object can contain several alternative forms from which the presentation system can select the most appropriate object. Questions and answers

DocBook V3.1 introduced the QandASet element, which is suitable for FAQs (Frequently Asked Questions) and other similar collections of Questions and Answers. Miscellaneous block elements

The following block elements are also available:


A block quotation. Block quotations may have Attributions.


An environment for marking up all the parameters and options of a command.


A short introduction, typically a quotation, at the beginning of a document. Epigraphs may have Attributions.


An environment for marking up the return value and arguments of a function.


A summary of the main points discussed in a book component (chapter, section, and so on).


A set of related error messages.


A procedure. Procedures contain Steps, which may contain SubSteps.


A sidebar.

2.5.7. Inline Elements

Users of DocBook are provided with a surfeit of inline elements. Inline elements are used to mark up running text. In published documents, inline elements often cause a font change or other small change, but they do not cause line or paragraph breaks.

In practice, writers generally settle on the tagging of inline elements that suits their time and subject matter. This may be a large number of elements or only a handful. What is important is that you choose to mark up not every possible item, but only those for which distinctive tagging will be useful in the production of the finished document for the readers who will search through it.

The following comprehensive list may be a useful tool for the process of narrowing down the elements that you will choose to mark up; it is not intended to overwhelm you by its sheer length. For convenience, we've divided the inlines into several subcategories.

The classification used here is not meant to be authoritative, only helpful in providing a feel for the nature of the inlines. Several elements appear in more than one category, and arguments could be made to support the placement of additional elements in other categories or entirely new categories. Traditional publishing inlines

These inlines identify things that commonly occur in general writing:


An abbreviation, especially one followed by a period.


An often pronounceable word made from the initial (or selected) letters of a name or phrase.


Emphasized text.


A footnote. The location of the Footnote element identifies the location of the first reference to the footnote. Additional references to the same footnote can be inserted with FootnoteRef.


A span of text.


An inline quotation.


A trademark. Cross references

The cross reference inlines identify both explicit cross references, such as Link, and implicit cross references like GlossTerm. You can make the most of the implicit references explicit with a LinkEnd attribute.


A spot in the document.


An inline bibliographic reference to another published work.


A citation to a reference page.


The title of a cited work.


The first occurrence of a term.


A glossary term.


A hypertext link.


A link that addresses its target indirectly, through an entity.


A link that addresses its target by means of a URL (Uniform Resource Locator).


A cross reference to another part of the document. Markup

These inlines are used to mark up text for special presentation:


A word or phrase in a language other than the primary language of the document.


A word meant specifically as a word and not representing anything else.


Data, generally text, displayed or presented by a computer.


Inline text that is some literal value.


A string of formatting markup in text that is to be represented literally.


A character or string indicating the start of an input field in a computer display.


Content that may or must be replaced by the user.


A component of SGML markup.


Data entered by the user. Mathematics

DocBook does not define a complete set of elements for representing equations. No one has ever pressed the DocBook maintainers to add this functionality, and the prevailing opinion is that incorporating MathML using a mechanism like namespaces is probably the best long-term solution.


A mathematical equation or expression occurring inline.


A subscript (as in H2O, the molecular formula for water).


A superscript (as in x2, the mathematical notation for x multiplied by itself). User interfaces

These elements describe aspects of a user interface:


A graphical user interface (GUI) keyboard shortcut.


The text on a button in a GUI.


Graphic and/or text appearing as a icon in a GUI.


The text of a label in a GUI.


The name of a menu in a GUI.


The name of a terminal menu item in a GUI.


The name of a submenu in a GUI.


The text printed on a key on a keyboard.


The internal, frequently numeric, identifier for a key on a keyboard.


A combination of input actions.


The symbolic name of a key on a keyboard.


A selection or series of selections from a menu.


The conventional name of a mouse button.


A key combination for an action that is also accessible through a menu. Programming languages and constructs

Many of the technical inlines in DocBook are related to programming.


A response to a user event.


The name of a class, in the object-oriented programming sense.


A programming or system constant.


An error code.


An error message.


The classification of an error message.


The name of a function or subroutine, as in a programming language.


An element of a GUI.


The name of a formal specification of a GUI.


Inline text that is some literal value.


The actual text of a message component in a message set.


A value or a symbolic reference to a value.


A unit of data associated with some part of a computer system.


Content that may or must be replaced by the user.


The value returned by a function.


A field in a structure (in the programming language sense).


The name of a structure (in the programming language sense).


A name that is replaced by a value before processing.


A unit of information.


The classification of a value.


The name of a variable. Operating systems

These inlines identify parts of an operating system, or an operating environment:


The name of a software program.


The name of an executable program or other software command.


A software environment variable.


The name of a file.


A name that identifies the physical medium on which some information resides.


The actual text of a message component in a message set.


An option for a software command.


A value or a symbolic reference to a value.


A character or string indicating the start of an input field in a computer display.


A system-related item or term. General purpose

There are also a number of general-purpose technical inlines.


The name of a software program.


The name of a database, or part of a database.


An email address.


The name of a file.


A physical part of a computer system.


An object containing or pointing to graphical data that will be rendered inline.


Inline text that is some literal value.


A name that identifies the physical medium on which some information resides.


An option for a software command.


Optional information.


Content that may or must be replaced by the user.


A name that is replaced by a value before processing.


A unit of information.


The classification of a value.

2.6. Making a DocBook Book

A typical Book, in English at least, consists of some meta-information in a BookInfo (Title, Author, Copyright, and so on), one or more Prefaces, several Chapters, and perhaps a few Appendixes. A Book may also contain Bibliographys, Glossarys, Indexes and a Colophon.

Example 2-2 shows the structure of a typical book. Additional content is required where the ellipses occur.

Example 2-2. A Typical Book

<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
  <title>My First Book</title>
  <copyright><year>1998</year><holder>Jane Doe</holder></copyright>
<preface><title>Foreword</title> ... </preface>
<chapter> ... </chapter>
<chapter> ... </chapter>
<chapter> ... </chapter>
<appendix> ... </appendix>
<appendix> ... </appendix>
<index> ... </index>

2.7. Making a Chapter

Chapters, Prefaces, and Appendixes all have a similar structure. They consist of a Title, possibly some additional meta-information, and any number of block-level elements followed by any number of top-level sections. Each section may in turn contain any number of block-level elements followed by any number from the next section level, as shown in Example 2-3.

Example 2-3. A Typical Chapter

<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
<chapter><title>My Chapter</title>
<para> ... </para>
<sect1><title>First Section</title>
<para> ... </para>
<example> ... </example>

2.8. Making an Article

For documents smaller than a book, such as: journal articles, white papers, or technical notes, Article is frequently the most logical starting point. The body of an Article is essentially the same as the body of a Chapter or any other component-level element, as shown in Example 2-4

Articles may include Appendixes, Bibliographys, Indexes and Glossarys.

Example 2-4. A Typical Article

<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
  <title>My Article</title>
<para> ... </para>
<sect1><title>On the Possibility of Going Home</title>
<para> ... </para>
<bibliography> ... </bibliography>

2.9. Making a Reference Page

The reference page or manual page in DocBook was inspired by, and in fact designed to reproduce, the common UNIX "manpage" concept. (We use the word "page" loosely here to mean a document of variable length containing reference material on a specific topic.) DocBook is rich in markup tailored for such documents, which often vary greatly in content, however well-structured they may be. To reflect both the structure and the variability of such texts, DocBook specifies that reference pages have a strict sequence of parts, even though several of them are actually optional.

Of the following sequence of elements that may appear in a RefEntry, only two are obligatory: RefNameDiv and RefSect1.


The DocInfo element contains meta-information about the reference page (which should not be confused with RefMeta, which it precedes). It marks up information about the author of the document, or the product to which it pertains, or the document's revision history, or other such information.


RefMeta contains a title for the reference page (which may be inferred if the RefMeta element is not present) and an indication of the volume number in which this reference page occurs. The ManVolNum is a very UNIX-centric concept. In traditional UNIX documentation, the subject of a reference page is typically identified by name and volume number; this allows you to distinguish between the uname command, "uname(1)" in volume 1 of the documentation and the uname function, "uname(3)" in volume 3.

Additional information of this sort such as conformance or vendor information specific to the particular environment you are working in, may be stored in RefMiscInfo.


The first obligatory element is RefNameDiv, which is a wrapper for information about whatever you're documenting, rather than the document itself. It can begin with a RefDescriptor if several items are being documented as a group and the group has a name. The RefNameDiv must contain at least one RefName, that is, the name of whatever you're documenting, and a single short statement that sums up the use or function of the item(s) at a glance: their RefPurpose. Also available is the RefClass, intended to detail the operating system configurations that the software element in question supports.

If no RefEntryTitle is given in the RefMeta, the title of the reference page is the RefDescriptor, if present, or the first RefName.


A RefSynopsisDiv is intended to provide a quick synopsis of the topic covered by the reference page. For commands, this is generally a syntax summary of the command, and for functions, the function prototype, but other options are possible. A Title is allowed, but not required, presumably because the application that processes reference pages will generate the appropriate title if it is not given. In traditional UNIX documentation, its title is always "Synopsis".


Within RefEntrys, there are only three levels of sectioning elements: RefSect1, RefSect2, and RefSect3.

Example 2-5 shows the beginning of a RefEntry that illustrates one possible reference page:

Example 2-5. A Sample Reference Page

<refentry id="printf">


<refpurpose>print formatted output</refpurpose>


#include &lt;stdio.h&gt;
  <funcdef>int <function>printf</function></funcdef>
  <paramdef>const char *<parameter>format</parameter></paramdef>

  <funcdef>int <function>fprintf</function></funcdef>
  <paramdef>FILE *<parameter>strm</parameter></paramdef>
  <paramdef>const char *<parameter>format</parameter></paramdef>

  <funcdef>int <function>sprintf</function></funcdef>
  <paramdef>char *<parameter>s</parameter></paramdef>
  <paramdef>const char *<parameter>format</parameter></paramdef>


<indexterm><primary>printing function</primary></indexterm>

<function>printf</function> places output on the standard
output stream stdout.

2.10. Making Front- and Backmatter

DocBook contains markup for the usual variety of front- and backmatter necessary for books and articles: indexes, glossaries, bibliographies, and tables of contents. In many cases, these components are generated automatically, at least in part, from your document by an external processor, but you can create them by hand, and in either case, store them in DocBook.

Some forms of backmatter, like indexes and glossaries, usually require additional markup in the document to make generation by an application possible. Bibliographies are usually composed by hand like the rest of your text, unless you are automatically selecting bibliographic entries out of some larger database. Our principal concern here is to acquaint you with the kind of markup you need to include in your documents if you want to construct these components.

Frontmatter, like the table of contents, is almost always generated automatically from the text of a document by the processing application. If you need information about how to mark up a table of contents in DocBook, please consult the reference page for ToC.

2.10.1. Making an Index

In some highly-structured documents such as reference manuals, you can automate the whole process of generating an index successfully without altering or adding to the original source. You can design a processing application to select the information and compile it into an adequate index. But this is rare.

In most cases--and even in the case of some reference manuals--a useful index still requires human intervention to mark occurrences of words or concepts that will appear in the text of the index. Marking index terms

Docbook distinguishes two kinds of index markers: those that are singular and result in a single page entry in the index itself, and those that are multiple and refer to a range of pages.

You put a singular index marker where the subject it refers to actually occurs in your text:

The tiger<indexterm>
<primary>Big Cats</primary>
is a very large cat indeed.
This index term has two levels, primary and secondary. They correspond to an increasing amount of indented text in the resultant index. DocBook allows for three levels of index terms, with the third labeled tertiary.

There are two ways that you can index a range of text. The first is to put index marks at both the beginning and end of the discussion. The mark at the beginning asserts that it is the start of a range, and the mark at the end refers back to the beginning. In this way, the processing application can determine what range of text is indexed. Here's the previous tiger example recast as starting and ending index terms:

The tiger<indexterm id="tiger-desc" class="startofrange">
<primary>Big Cats</primary>
is a very large cat indeed…
So much for tigers<indexterm startref="tiger-desc" class="endofrange">. Let's talk about

Note that the mark at the start of the range identifies itself as the start of a range with the Class attribute, and provides an ID. The mark at the end of the range points back to the start.

Another way to mark up a range of text is to specify that the entire content of an element, such as a chapter or section, is the complete range. In this case, all you need is for the index term to point to the ID of the element that contains the content in question. The Zone attribute of indexterm provides this functionality.

One of the interesting features of this method is that the actual index marks do not have to occur anywhere near the text being indexed. It is possible to collect all of them together, for example, in one file, but it is not invalid to have the index marker occur near the element it indexes.

Suppose the discussion of tigers in your document comprises a whole text object (like a Sect1 or a Chapter) with an ID value of tiger-desc. You can put the following tag anywhere in your document to index that range of text:

<indexterm zone="tiger-desc">
<primary>Big Cats</primary>

DocBook also contains markup for index hits that point to other index hits (of the same type such as "See Cats, big" or "See also Lions"). See the reference pages for See and SeeAlso. Printing an index

After you have added the appropriate markup to your document, an external application can use this information to build an index. The resulting index must have information about the page numbers on which the concepts appear. It's usually the document formatter that builds the index. In this case, it may never be instantiated in DocBook.

However, there are applications that can produce an index marked up in DocBook. The following example includes some one- and two-level IndexEntry elements (which correspond to the primary and secondary levels in the indexterms themselves) that begin with the letter D:

<!DOCTYPE index PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
  <primaryie>database (bibliographic), 253, 255</primaryie>
     <secondaryie>structure, 255</secondaryie>
     <secondaryie>tools, 259</secondaryie>
  <primaryie>dates (language specific), 179</primaryie>
  <primaryie>DC fonts, <emphasis>172</emphasis>, 177</primaryie>
     <secondaryie>Math fonts, 177</secondaryie>

2.10.2. Making a Glossary

Glossarys, like Bibliographys, are often constructed by hand. However, some applications are capable of building a skeletal index from glossary term markup in the document. If all of your terms are defined in some glossary database, it may even be possible to construct the complete glossary automatically.

To enable automatic glossary generation, or simply automatic linking from glossary terms in the text to glossary entries, you must add markup to your documents. In the text, you markup a term for compilation later with the inline GlossTerm tag. This tag can have a LinkEnd attribute whose value is the ID of the actual entry in the glossary.[5]

For instance, if you have this markup in your document:

<glossterm linkend="xml">Extensible Markup Language</glossterm> is a new standard… 

your glossary might look like this:

<!DOCTYPE glossary PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
<glossary><title>Example Glossary</title>

<glossentry id="xml"><glossterm>Extensible Markup Language</glossterm>
  <para>Some reasonable definition here.</para>
  <glossseealso otherterm="sgml">


Note that the GlossTerm tag reappears in the glossary to mark up the term and distinguish it from its definition within the GlossEntry. The ID that the GlossEntry referenced in the text is the ID of the GlossEntry in the Glossary itself. You can use the link between source and glossary to create a link in the online form of your document, as we have done with the online form of the glossary in this book.

2.10.3. Making a Bibliography

There are two ways to set up a bibliography in DocBook: you can have the data raw or cooked. Here's an example of a raw bibliographical item, wrapped in the Biblioentry element:

<biblioentry xreflabel="Kites75">
     <holder>Product Development International Holding N. V.</holder>
    <publishername>Plenary Publications International, Inc.</publishername>
  <subtitle>Ancient Craft to Modern Sport</subtitle>
    <title>The Family Creative Workshop</title>
      <othername role=middle>Davenport</othername>
      <contrib>Editor in Chief</contrib>

The "raw" data in a Biblioentry is comprehensive to a fault--there are enough fields to suit a host of different bibliographical styles, and that is the point. An abundance of data requires processing applications to select, punctuate, order, and format the bibliographical data, and it is unlikely that all the information provided will actually be output.

All the "cooked" data in a Bibliomixed entry in a bibliography, on the other hand, is intended to be presented to the reader in the form and sequence in which it is provided. It even includes punctuation between the fields of data:

  <bibliomset relation=article>
    <surname>Walsh</surname>, <firstname>Norman</firstname>.
    <title role=article>Introduction to Cascading Style Sheets</title>.
  <bibliomset relation=journal>
    <title>The World Wide Web Journal</title> 
    <publishername>O'Reilly & Associates, Inc.</publishername> and
    <corpname>The World Wide Web Consortium</corpname>.
    <pubdate>Winter, 1996</pubdate></bibliomset>.

Clearly, these two ways of marking up bibliographical entries are suited to different circumstances. You should use one or the other for your bibliography, not both. Strictly speaking, mingling the raw and the cooked may be "kosher" as far as the DTD is concerned, but it will almost certainly cause problems for most processing applications. >



Many of these things are influenced by the SGML declaration in use. For the purpose of this discussion, we assume you are using the standard DocBook declaration.


This is not absolutely true. SGML allows for the possibility that the reference could be implied by the application, but this is very rarely the case.


Essentially, it can ensure that two different owners won't accidentally tread on each other. Nothing can prevent a given owner from reusing public identifiers, except maybe common sense.


See Appendix A of [MALER96], for more details.


Some sophisticated formatters might even be able to establish the link simply by examining the content of the terms and the glossary. In that case, the author is not required to make explicit links.

Back to: DocBook: The Definitive Guide Home | O'Reilly Bookstores | How to Order | O'Reilly Contacts
International | About O'Reilly | Affiliated Companies | Privacy Policy

© 2001, O'Reilly & Associates, Inc.