doctools_lang_intro (n) - Linux Manuals

doctools_lang_intro: doctools language introduction

NAME

doctools_lang_intro - doctools language introduction

DESCRIPTION

This document is an informal introduction to version 1 of the doctools markup language based on a multitude of examples. After reading this a writer should be ready to understand the two parts of the formal specification, i.e. the doctools language syntax specification and the doctools language command reference.

FUNDAMENTALS

In the broadest terms possible the doctools markup language is LaTeX-like, instead of like SGML and similar languages. A document written in this language consists primarily of text, with markup commands embedded into it.

Each markup command is a Tcl command surrounded by a matching pair of [ and ]. Inside of these delimiters the usual rules for a Tcl command apply with regard to word quotation, nested commands, continuation lines, etc. I.e.

  ... [list_begin enumerated] ...
  ... [call [cmd foo] \\
          [arg bar]] ...
  ... [term {complex concept}] ...
  ... [opt "[arg key] [arg value]"] ...

BASIC STRUCTURE

The most simple document which can be written in doctools is
    [manpage_begin NAME SECTION VERSION]
    [description]
    [manpage_end]
This also shows us that all doctools documents are split into two parts, the header and the body. Everything coming before [description] belongs to the header, and everything coming after belongs to the body, with the whole document bracketed by the two manpage_* commands. Before and after these opening and closing commands we have only whitespace.

In the remainder of this section we will discuss only the contents of the header, the structure of the body will be discussed in the section Text structure.

The header section can be empty, and otherwise may contain only an arbitrary sequence of the four so-called header commands, plus whitespace. These commands are

titledesc
moddesc
require
copyright

They provide, through their arguments, additional information about the document, like its title, the title of the larger group the document belongs to (if applicable), the requirements of the documented packages (if applicable), and copyright assignments. All of them can occur multiple times, including none, and they can be used in any order. However for titledesc and moddesc only the last occurrence is taken. For the other two the specified information is accumulated, in the given order. Regular text is not allowed within the header.

Given the above a less minimal example of a document is

[manpage_begin NAME SECTION VERSION]
[copyright {YEAR AUTHOR}]
[titledesc TITLE]
[moddesc   MODULE_TITLE]
[require   PACKAGE VERSION]
[require   PACKAGE]
[description]
[manpage_end]
Remember that the whitespace is optional. The document
    [manpage_begin NAME SECTION VERSION]
    [copyright {YEAR AUTHOR}][titledesc TITLE][moddesc MODULE_TITLE]
    [require PACKAGE VERSION][require PACKAGE][description]
    [manpage_end]
has the same meaning as the example before.

On the other hand, if whitespace is present it consists not only of any sequence of characters containing the space character, horizontal and vertical tabs, carriage return, and newline, but it may contain comment markup as well, in the form of the comment command.

[comment { ... }]
[manpage_begin NAME SECTION VERSION]
[copyright {YEAR AUTHOR}]
[titledesc TITLE]
[moddesc   MODULE_TITLE][comment { ... }]
[require   PACKAGE VERSION]
[require   PACKAGE]
[description]
[manpage_end]
[comment { ... }]

ADVANCED STRUCTURE

In the simple examples of the last section we fudged a bit regarding the markup actually allowed to be used before the manpage_begin command opening the document.

Instead of only whitespace the two templating commands include and vset are also allowed, to enable the writer to either set and/or import configuration settings relevant to the document. I.e. it is possible to write

[include FILE]
[vset VAR VALUE]
[manpage_begin NAME SECTION VERSION]
[description]
[manpage_end]
Even more important, these two commands are allowed anywhere where a markup command is allowed, without regard for any other structure. I.e. for example in the header as well.
[manpage_begin NAME SECTION VERSION]
[include FILE]
[vset VAR VALUE]
[description]
[manpage_end]
The only restriction include has to obey is that the contents of the included file must be valid at the place of the inclusion. I.e. a file included before manpage_begin may contain only the templating commands vset and include, a file included in the header may contain only header commands, etc.

TEXT STRUCTURE

The body of the document consists mainly of text, possibly split into sections, subsections, and paragraphs, with parts marked up to highlight various semantic categories of text, and additional structure through the use of examples and (nested) lists.

This section explains the high-level structural commands, with everything else deferred to the following sections.

The simplest way of structuring the body is through the introduction of paragraphs. The command for doing so is para. Each occurrence of this command closes the previous paragraph and automatically opens the next. The first paragraph is automatically opened at the beginning of the body, by description. In the same manner the last paragraph automatically ends at manpage_end.

[manpage_begin NAME SECTION VERSION]
[description]
 ...
[para]
 ...
[para]
 ...
[manpage_end]
Empty paragraphs are ignored.

A structure coarser than paragraphs are sections, which allow the writer to split a document into larger, and labeled, pieces. The command for doing so is section. Each occurrence of this command closes the previous section and automatically opens the next, including its first paragraph. The first section is automatically opened at the beginning of the body, by description (This section is labeled "DESCRIPTION"). In the same manner the last section automatically ends at manpage_end.

Empty sections are not ignored. We are free to (not) use paragraphs within sections.

[manpage_begin NAME SECTION VERSION]
[description]
 ...
[section {Section A}]
 ...
[para]
 ...
[section {Section B}]
 ...
[manpage_end]
Between sections and paragraphs we have subsections, to split sections. The command for doing so is subsection. Each occurrence of this command closes the previous subsection and automatically opens the next, including its first paragraph. A subsection is automatically opened at the beginning of the body, by description, and at the beginning of each section. In the same manner the last subsection automatically ends at manpage_end.

Empty subsections are not ignored. We are free to (not) use paragraphs within subsections.

[manpage_begin NAME SECTION VERSION]
[description]
 ...
[section {Section A}]
 ...
[subsection {Sub 1}]
 ...
[para]
 ...
[subsection {Sub 2}]
 ...
[section {Section B}]
 ...
[manpage_end]

TEXT MARKUP

Having handled the overall structure a writer can impose on the document we now take a closer at the text in a paragraph.

While most often this is just the unadorned content of the document we do have situations where we wish to highlight parts of it as some type of thing or other, like command arguments, command names, concepts, uris, etc.

For this we have a series of markup commands which take the text to highlight as their single argument. It should be noted that while their predominant use is the highlighting of parts of a paragraph they can also be used to mark up the arguments of list item commands, and of other markup commands.

The commands available to us are

arg
Its argument is a the name of a command argument.
class
Its argument is a class name.
cmd
Its argument is a command name (Tcl command).
const
Its argument is a constant.
emph
General, non-semantic emphasis.
file
Its argument is a filename / path.
fun
Its argument is a function name.
method
Its argument is a method name
namespace
Its argument is namespace name.
opt
Its argument is some optional syntax element.
option
Its argument is a command line switch / widget option.
package
Its argument is a package name.
sectref
Its argument is the title of a section or subsection, i.e. a section reference.
syscmd
Its argument is a command name (external, system command).
term
Its argument is a concept, or general terminology.
type
Its argument is a type name.
uri
Its argument is a uniform resource identifier, i.e an external reference. A second argument can be used to specify an explicit label for the reference in question.
usage
The arguments describe the syntax of a Tcl command.
var
Its argument is a variable.
widget
Its argument is a widget name.

The example demonstrating the use of text markup is an excerpt from the doctools language command reference, with some highlighting added. It shows their use within a block of text, as the arguments of a list item command (call), and our ability to nest them.

  ...
  [call [cmd arg_def] [arg type] [arg name]] [opt [arg mode]]]

  Text structure. List element. Argument list. Automatically closes the
  previous list element. Specifies the data-[arg type] of the described
  argument of a command, its [arg name] and its i/o-[arg mode]. The
  latter is optional.
  ...

ESCAPES

Beyond the 20 commands for simple markup shown in the previous section we have two more available which are technically simple markup. However their function is not the marking up of phrases as specific types of things, but the insertion of characters, namely [ and ]. These commands, lb and rb respectively, are required because our use of [ and ] to bracket markup commands makes it impossible to directly use [ and ] within the text.

Our example of their use are the sources of the last sentence in the previous paragraph, with some highlighting added.

  ...
  These commands, [cmd lb] and [cmd lb] respectively, are required
  because our use of [lb] and [rb] to bracket markup commands makes it
  impossible to directly use [lb] and [rb] within the text.
  ...

CROSS-REFERENCES

The last two commands we have to discuss are for the declaration of cross-references between documents, explicit and implicit. They are keywords and see_also. Both take an arbitrary number of arguments, all of which have to be plain unmarked text. I.e. it is not allowed to use markup on them. Both commands can be used multiple times in a document. If that is done all arguments of all occurrences of one of them are put together into a single set.
keywords
The arguments of this command are interpreted as keywords describing the document. A processor can use this information to create an index indirectly linking the containing document to all documents with the same keywords.
see_also
The arguments of this command are interpreted as references to other documents. A processor can format them as direct links to these documents.

All the cross-reference commands can occur anywhere in the document between manpage_begin and manpage_end. As such the writer can choose whether she wants to have them at the beginning of the body, or at its end, maybe near the place a keyword is actually defined by the main content, or considers them as meta data which should be in the header, etc.

Our example shows the sources for the cross-references of this document, with some highlighting added. Incidentally they are found at the end of the body.

  ...
  [see_also doctools_intro]
  [see_also doctools_lang_syntax]
  [see_also doctools_lang_cmdref]
  [keywords markup {semantic markup}]
  [keywords {doctools markup} {doctools language}]
  [keywords {doctools syntax} {doctools commands}]
  [manpage_end]

EXAMPLES

Where ever we can write plain text we can write examples too. For simple examples we have the command example which takes a single argument, the text of the argument. The example text must not contain markup. If we wish to have markup within an example we have to use the 2-command combination example_begin / example_end instead.

The first opens an example block, the other closes it, and in between we can write plain text and use all the regular text markup commands. Note that text structure commands are not allowed. This also means that it is not possible to embed examples and lists within an example. On the other hand, we can use templating commands within example blocks to read their contents from a file (Remember section Advanced structure).

The source for the very first example in this document (see section Fundamentals), with some highlighting added, is

  [example {
    ... [list_begin enumerated] ...
  }]
Using example_begin / example_end this would look like
  [example_begin]
    ... [list_begin enumerated] ...
  [example_end]

LISTS

Where ever we can write plain text we can write lists too. The main commands are list_begin to start a list, and list_end to close one. The opening command takes an argument specifying the type of list started it, and this in turn determines which of the eight existing list item commands are allowed within the list to start list items.

After the opening command only whitespace is allowed, until the first list item command opens the first item of the list. Each item is a regular series of paragraphs and is closed by either the next list item command, or the end of the list. If closed by a list item command this command automatically opens the next list item. A consequence of a list item being a series of paragraphs is that all regular text markup can be used within a list item, including examples and other lists.

The list types recognized by list_begin and their associated list item commands are:

arguments
(arg_def) This opens an argument (declaration) list. It is a specialized form of a term definition list where the term is an argument name, with its type and i/o-mode.
commands
(cmd_def) This opens a command (declaration) list. It is a specialized form of a term definition list where the term is a command name.
definitions
(def and call) This opens a general term definition list. The terms defined by the list items are specified through the argument(s) of the list item commands, either general terms, possibly with markup (def), or Tcl commands with their syntax (call).
enumerated
(enum) This opens a general enumerated list.
itemized
(item) This opens a general itemized list.
options
(opt_def) This opens an option (declaration) list. It is a specialized form of a term definition list where the term is an option name, possibly with the option's arguments.
tkoptions
(tkoption_def) This opens a widget option (declaration) list. It is a specialized form of a term definition list where the term is the name of a configuration option for a widget, with its name and class in the option database.

Our example is the source of the definition list in the previous paragraph, with most of the content in the middle removed.

  ...
  [list_begin definitions]
  [def [const arg]]

  ([cmd arg_def]) This opens an argument (declaration) list. It is a
  specialized form of a definition list where the term is an argument
  name, with its type and i/o-mode.

  [def [const itemized]]

  ([cmd item])
  This opens a general itemized list.

  ...
  [def [const tkoption]]

  ([cmd tkoption_def]) This opens a widget option (declaration) list. It
  is a specialized form of a definition list where the term is the name
  of a configuration option for a widget, with its name and class in the
  option database.

  [list_end]
  ...
Note that a list cannot begin in one (sub)section and end in another. Differently said, (sub)section breaks are not allowed within lists and list items. An example of this illegal construct is
  ...
  [list_begin itemized]
  [item]
  ...
  [section {ILLEGAL WITHIN THE LIST}]
  ...
  [list_end]
  ...

FURTHER READING

Now that this document has been digested the reader, assumed to be a writer of documentation should be fortified enough to be able to understand the formal doctools language syntax specification as well. From here on out the doctools language command reference will also serve as the detailed specification and cheat sheet for all available commands and their syntax.

To be able to validate a document while writing it, it is also recommended to familiarize oneself with one of the applications for the processing and conversion of doctools documents, i.e. either Tcllib's easy and simple dtplite, or Tclapps' ultra-configurable dtp.

BUGS, IDEAS, FEEDBACK

This document, will undoubtedly contain bugs and other problems. Please report such in the category doctools of the Tcllib SF Trackers [http://sourceforge.net/tracker/?group_id=12883]. Please also report any ideas for enhancements you may have.

KEYWORDS

doctools commands, doctools language, doctools markup, doctools syntax, markup, semantic markup

CATEGORY

Documentation tools

COPYRIGHT

Copyright (c) 2007 Andreas Kupries <andreas_kupries [at] users.sourceforge.net>

SEE ALSO

doctools_intro, doctools_lang_cmdref, doctools_lang_faq, doctools_lang_syntax