Texi2HTML – Texinfo to HTML v4.13

Last Update: 30 June 2010

for Version 4.13 of the texi2html script.

Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies.

Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one.

Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that this permission notice may be stated in a translation approved by the Free Software Foundation.

Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies.

Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one.

Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that this permission notice may be stated in a translation approved by the Free Software Foundation.

Texi2HTML

This manual, last updated 30 June 2010, describes version 4.13 of the texi2html Perl script which converts Texinfo into HTML.

Please send bug reports concerning this manual to the Texi2HTML developement list bug-texinfo@gnu.org. Please state the exact version of the manual which contains the bug, as given above.

This manual is currently under construction and of course incomplete.  ;-)

1 Overview

Texinfo is the official documentation format of the GNU project. It uses a single source file to produce both online information and printed output.

It is often desirable to have a way to produce HTML from Texinfo sources, as GNU-Info files are produced. It is much simpler to run a converter than it is to rewrite all the documentation in HTML, especially considering that there is so much Texinfo documentation in the world.

Some time ago makeinfo wasn’t able to produce HTML output format, but people still wanted documentation in HTML. This was the birthing hour for texi2html. The basic purpose of texi2html is to convert Texinfo documents into HTML.

Since then, HTML support in makeinfo has improved, but texi2html is still stronger in many areas, including the degree to which it allows customization. With texi2html, some important aspects of the resulting HTML files may be specified via command line options, and configuration files provide an even finer degree of control over the final output, allowing most every aspect of the final output not specified in the Texinfo input file to be specified. Configuration files are written in perl, like the main program, and anything which may be specified on the command line may also be specified within a configuration file.

For an example of the kind of pages texi2html is capable of producing, have a look at the following sites: the Singular Manual, the Cederqvist (CVS Manual).

1.1 Why texi2html and not makeinfo?

You would like to produce HTML files from your existing Texinfo files? There are two programs you can use to do this. The first is makeinfo (see Generating HTML in GNU Texinfo). The second is texi2html.

The design goal of makeinfo’s HTML output was to produce readable HTML output. It is now possible to use CSS for HTML customization. Another possibility is to use intermediate formats, like docbook or makeinfo XML and XSL stylesheets to customize the resulting document. Still the output produced by makeinfo isn’t customizable.

The current development of texi2html tries to provide for producing the more interesting and sophisticated HTML pages that today’s Internet users have come to expect. The goal behind texi2html is to generate attractive HTML by default but also to allow users considerable freedom to affect the final style and design of the output HTML pages. This is achieved via command line options and flexible configuration files.

In contrast to the HTML produced by makeinfo --html (the makeinfo program is part of the Texinfo distribution), the texi2html program, among other differences, allows for the customization of the entire page layout, including headers, footers, style sheets, etc., allows for customization of the low level HTML formatting, provides for splitting documents at various levels, and provides for using the latex2html program to convert @tex sections of the Texinfo source.

The focus on HTML is still present but with the help of the customization files it is now possible to use texi2html to produce other formats as well. texi2html may for example be turned into a texinfo to roff translator with the help of a customization file provided with the distribution.

texi2html should reasonably convert all Texinfo 4.8 constructs. If you find it does not, please send a bug report to the texi2html-bug@nongnu.org email list.

2 Obtaining texi2html

The latest version of the source code for texi2html should be available from www.nongnu.org/texi2html/. texi2html is also available with teTeX and TeX Live.

3 Installation of texi2html

3.1 Requirements

To install texi2html, you must first obtain a copy of the source distribution. See section Obtaining texi2html.

texi2html requires perl version 5.00405 or above to be run. An older perl 5 version with File::Spec is also likely to work. The current version has been lightly tested on a wide range of perl, but has not been tested extensively on versions of perl below 5.6. To play nice with encodings you also need the Encode and Unicode::Normalize modules.

To rebuild the script perl isn’t required in most cases. For more information about advanced build features, see Advanced build features.

3.2 Configuring the source and rebuilding

texi2html is a standard Automake-based distribution. If you have a source version, you should run ./configure to configure the sources and make to build the script.

./configure accepts options to select the installation directory for the ‘texi2html’ file, the default directories texi2html will use to look for configuration files, and other details. Run ./configure --help for more information. Running ./configure creates ‘texi2html_configured.pl’ from ‘texi2html.pl’, and also builds the make configuration files (‘Makefile’s).

Running make combines five files into the final ‘texi2html’ program file:

• ‘texi2html_configured.pl’ contains the base program,
• ‘MySimple.pm’ handles the command line options,
• ‘texi2html.init’ is the default configuration file, and
• ‘T2h_i18n.pm’ is used for internationalization.
• ‘translations.pl’ contains the translations of the strings used in documents.

Running make also rebuilds the documentation if needed.

3.3 Installing

make install performs the installation to the locations specified to the ./configure script. This usually involves placing the actual ‘texi2html’ file someplace in your path, such as ‘/usr/local/bin’ or ‘/usr/bin’.

Installing texi2html in your path should be sufficient to run it. To use default initialization files, or a configuration file for LaTeX2HTML when using latex2html to convert @tex sections (see section Expanding @tex and @math regions using LaTeX2HTML), install them in the package data directory specified to configure. This is ‘/usr/local/share/texi2html/’ by default, but depends on the value of the ‘--pkgdatadir=dir’ option passed to the ./configure script. See section Use initialization files for fine tuning for more.

3.4 Advanced build features

This section documents features that are unlikely to be used but deserve a bit of documentation.

A ./configure switch, ‘--with-unicode’ allows to choose whether the unicode code should be used or not. The default is to detect it with a test. This code requires Encode and Unicode::Normalize modules.

A similar ./configure switch, ‘--with-unidecode’ allows to choose whether the perl module Text::Unidecode should be used or not. The default is to detect it with a test. This code requires the Text::Unidecode module.

perl isn’t needed to build the script. the script is build by ‘./configure’ and a shell script launched by make which is a simple wrapper around a sed one-liner. The perl command can be specified with the environment variable $PERL, otherwise it is detected. perl is required to rebuild the documentation as the HTML documentation is rebuild with texi2html itself.

Old style translations are managed by a script manage_i18n.pl, created by ./configure. manage_i18n.pl requires Data::Dumper to function normally. If this module isn’t there ./configure detects it and manage_i18n.pl doesn’t really rebuild the translations, but only copy files. It is possible to use the ./configure switch ‘--enable-translations’

to override the ./configure detection. For more about translations, see Internationalization.

It is possible to build from outside of the source directory, for example the following should work:

tar xzvf texi2html-4.13.tar.gz
mkdir texi2html_build
cd texi2html_build
../texi2html-4.13/configure && make

All these features enables to build texi2html on a platform in order to run it on another platform, a kind of cross-building. The ./configure switches and $PERL allows to specify everything needed for the build of the texi2html script.

4 Invoking texi2html

To produce an HTML manual, run texi2html with a Texinfo file as an argument. For example, this manual is created with:

$ texi2html texi2html.texi

texi2html can accept more than one manual on the command line, and will proceed with each of the manuals in turn.

The behaviour of texi2html may be changed with command line options. These command line options are always associated with corresponding perl variables which may appear in init files, and these variables are presented in this chapter each time a switch is described.

Boolean command line switches always have a corresponding negated switch, obtained by prepending ‘no’ or ‘no-’ to the switch name. For example ‘--nomenu’ does the reverse of ‘--menu’ .

When more than one manual is processed, the command line apply to all the manuals, except for command-line switches setting the output file names.

4.1 General options

Miscellaneous general options:

‘--error-limit=num’

Quit after num errors (default 1000), (variable $ERROR_LIMIT ).

‘--help’

Display a short help and exit.

‘--verbose’

Be verbose.

‘--version’

Display version information and exit.

4.2 Specifying where to split the generated document

The HTML manual resulting from the processing of the Texinfo source may be split into files at different levels. This is specified with the option ‘--split’ which takes an argument, namely the level of splitting (variable: $SPLIT ). This level may be:

‘chapter’

The document is split at @chapter, @appendix, or @unnumbered.

‘section’

The document is split at the same places as it is using the ‘chapter’ argument, and also at @section, @appendixsec or @unnumberedsec.

‘node’

The document is split at every sectioning command. It is not necessarily split at each node, if the @node structure doesn’t correspond with the sectioning command structure (see below).

‘none’

The document isn’t split. This is the default.

There are two kinds of commands which may be used to define sectioning elements in Texinfo: @node and the structuring commands (@top, @section, @appendixsubsec, and so on). A node just preceding a structuring command is considered to be part of the same sectioning element as that command. If the @node Top isn’t associated with a structuring command it also defines a sectioning element.

By default, nodes which aren’t associated with a structuring command are not considered to be sectioning commands. They are always considered to be part of a sectioning element defined by a structuring command. It is possible to change this behaviour via the ‘--use-nodes’ option (variable $USE_NODES ). In this case, nodes not associated with structuring commands are also considered to be sectioning commands defining a sectioning element.

This default behaviour mimics texi2dvi behaviour, which ignores @node commands for the purprose of sectioning, while the second looks like makeinfo behaviour (see Two Paths in GNU Texinfo).

As an illustration, the following table shows how a sample Texinfo document is divided into sectioning elements when ‘--use-nodes’ is used and not:

@node node1
@chapter node 1
node1 text

@node node2
node2 text

@node node3
node3 text
@chapter node 3
chapter text

@node node1
@chapter node 1
node1 text

@node node2
node2 text

@node node3
node3 text
@chapter node 3
chapter text

@node node1
@chapter node 1
node1 text

@node node2
node2 text

@node node3
node3 text
@chapter node 3
chapter text

4.3 Setting output file and directory names

The manual name is constructed by stripping the ‘.texi’, ‘.txi’, ‘.texinfo’, or ‘.txinfo’ extension from the Texinfo file name.

By default, texi2html generates the manual file in the current directory if the manual isn’t split. A ‘.html’ file extension is appended to the manual name.

If the manual is split the files are put in a directory named after the manual name. The file name is constructed using the manual name as basename. An underscore followed by a number is appended to the basename for each files corresponding with sectioning elements, with the exception of the top element. For the top element there is nothing appended. The files containing special elements pages have an underscore and a 3 letter code corresponding to their content (‘toc’ for table of contents, ‘abt’ for about, ‘ovr’ for overview, ‘fot’ for footnotes if they are separated) appended. Lastly, an ‘.html’ file extension is appended.

Thus, if the texinfo file ‘afile.texi’ is processed and split at chapters into 3 files, the generated files (in directory ‘afile’) will be:

afile.html         --> @node Top or @top section
afile_1.html       --> Chapter 1
afile_2.html       --> Chapter 2
afile_toc.html     --> Table of Contents
afile_abt.html     --> About Page

This default behavior may be modified by several command line options. If the output isn’t split, the prefix file name may be overrided by the ‘--output’ command line option (variable $OUT ). If the output is split, and ‘--output’ is set, the files are placed in the directory specified by the argument to the option.

The basename may be overridden with ‘--prefix’ (variable $PREFIX ). If ‘--short-ext’ is given, ‘.htm’ is appended instead of ‘.html’ in the final step (variable $SHORTEXTN ). The ‘--top-file’ option overrides the top element file name (variable $TOP_FILE ). This can be used to name the top element file ‘index.html’. Similarly, ‘--toc-file’ changes the name of the table of contents file (variable $TOC_FILE ).

Reusing the example above, but this time calling texi2html like so:

$ texi2html -split chapter -prefix manual -short-ext -top-file index.htm -toc-file contents.htm afile.texi

we get, in ‘manual’:

index.htm          --> @node Top or @top section
manual_1.htm       --> Chapter 1
manual_2.htm       --> Chapter 2
contents.htm       --> Table of Contents
manual_abt.htm     --> About Page

The file names generated by texi2html differ from those generated by makeinfo. makeinfo uses the @setfilename to determine the manual name(1). Also makeinfo uses the node name to construct the file names while splitting at nodes.

It is possible to get the same behaviour out of texi2html by specifying the ‘--node-files’ option (variable $NODE_FILES ). The default is false for this option. If the output isn’t split at nodes, texi2html will still output files named after the nodes, without real content but redirecting to the right file. This trick enables the generated HTML manual to be a target for the cross-references of other manuals generated by makeinfo or texi2html.

In case the files are named after the node names, another command-line option, ‘--transliterate-file-names’

can be set to trigger ASCII transliteration of node file names (variable $TRANSLITERATE_FILE_NAMES ). Transliteration is set in the default case.

4.4 Specifying which regions get expanded

The default for texi2html is to expand the @ifhtml, @html, and @menu regions, all the @ifnot regions except @ifnothtml, and no other @if regions.

It is possible to expand other regions by setting ‘--if<region>’ , where ‘<region>’ is replaced by the literal name of the region (for example, ‘--iftex’). Symetrically, if ‘--no-if<region>’ is specified, the ‘<region>’ region is ignored. The configuration file array, @EXPAND , holds the names of regions which should be expanded. The only region name present in @EXPAND in the default case is ‘html’.

If ‘--nomenu’ is set, the @menu sections are not expanded (variable $SHOW_MENU ). The default is to expand @menu sections.

4.5 Command line options related to Texinfo language features

Miscalleneous Texinfo related things may be specified via command line options.

‘--document-language=lang’

Sets the document language similar to the Texinfo directive, @documentlanguage lang (variable $DOCUMENTLANGUAGE ). The default is ‘en’, that is, use the english language strings.

‘-Dvar’

Sets var. Equivalent to, @set var 1, in Texinfo.

‘-Uvar’

Clears var. Equivalent to, @clear var, in Texinfo.

‘-Pdir’

Prepend dir to the list of directories to search for @include files (the associated array is @PREPEND_DIRS , empty in the default case).

‘-Idir’

Append dir to the list of directories to search for @include files (the associated array is @INCLUDE_DIRS , empty in the default case).

The include files are always searched for in the current directory.

4.6 Page layout related command line options

If the ‘--frames’ option is specified, HTML frames are used. A file describing the frame layout is generated, and the document page is associated with a frame where the short table of content appears (variable $FRAMES ). The default is not to use frames.

It is also possible to suppress the section navigation panel with ‘--no-headers’ (variable $HEADERS , the default is to output all the navigation panels), and to specify whether footnotes should appear at the foot of the same page which contains the reference to the note with ‘--footnote-style’ set to ‘end’ or on a separate page with ‘--footnote-style’ set to ‘separate’ (variable $FOOTNOTESTYLE ). The default is to have separated footnotes.

4.7 Customizing the HTML and text style

Miscalleneous style changes may be achieved with command line options.

‘--doctype=DTD’

‘--frameset-doctype=DTD’

You can specify the document DTD by setting these options. ‘--frameset-doctype’ applies to the file describing the frames when frames are used (corresponding variables are $DOCTYPE and $FRAMESET_DOCTYPE ).

The default for the document doctype for HTML is:

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">

And for the frameset doctype:

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Frameset//EN" "http://www.w3.org/TR/html4/frameset.dtd">

‘--iso’

If this option is set, ISO8859 entities are used for some special symbols, like Copyright © (variable $USE_ISO ). It is the default.

‘--css-include=file’

This command line switch provides for the inclusion of an external Cascading Style Sheet (CSS) file. More than one file may be specified, and ‘-’ stands for the standard input (array @CSS_FILES ).

The option use is the same than for makeinfo and is described extensively in HTML CSS in GNU Texinfo. Briefly, the CSS @import lines from the external file CSS file are pasted before the texi2html CSS rules, and the external file CSS rules are pasted after the texi2html CSS rules.

‘--css-ref=URL’

This command line switch provides for the inclusion of an reference to a Cascading Style Sheet (CSS) URL. More than one URL may be specified (array @CSS_REFS ).

‘--html-xref-prefix=path’

This option sets the base directory for external HTML texinfo manuals (variable $EXTERNAL_DIR ). Defaults to ‘../’.

‘--def-table’

If this option is set, HTML tables are used to format definition commands, rather than HTML definition tables (variable $DEF_TABLE ). Default is false.

‘--short-ref’

If this option is set, cross-references are given without section numbers (variable $SHORT_REF ). Default is false.

‘--number-sections’

If this option is set, sections are numbered (variable $NUMBER_SECTIONS ). This is the default.

‘--toc-links’

If this option is set, links from headings to TOC entries are created (variable $TOC_LINKS ). Default is false.

4.8 Expanding @tex and @math regions using LaTeX2HTML

It is possible to use LaTeX2HTML to process @tex regions and @math{} commands. This is an attractive way to display mathematical constructs in the HTML manual. The ‘--l2h’ option activates this feature (variable $L2H). It is usually desirable to expand @tex sections when this option is specified (see section Specifying which regions get expanded). The default is not to use this feature.

The ‘--l2h-l2h=program’ option enables changing the name/location of the LaTeX2HTML program processing TeX regions (variable $L2H_L2H ). The default is latex2html.

‘--l2h-tmp’ sets the directory used for temporary files, this name shouldn’t contain a dot ‘.’ (variable is $L2H_TMP ). Defaults to the current dir.

The file specified by ‘--l2h-file’ is used as LaTeX2HTML init file. It is searched at the same places than init files (see section Use initialization files for fine tuning), and the default is ‘l2h.init’.

4.9 Use initialization files for fine tuning

Initialization variables are read first from ‘/usr/local/share/texi2html/Config’ (the exact location being changeable with the ‘--pkgdatadir=dir’ option of the configure script, see Installation of texi2html), ‘/usr/local/etc/texi2html/Config’ (the exact location being changeable with the ‘--sysconfdir=dir’ option of the configure script, see Installation of texi2html), from ‘./Config’ then from ‘$HOME/.texi2html/Config’. Any command-line option can override the corresponding option set in init file, and the option ‘--init-file’ specifies an init file to be loaded, with later settings overriding earlier ones.

The init files specified with ‘--init-file’ are searched first in the current directory, then in the ‘$HOME/.texi2html/’ directory, in the ‘/usr/local/etc/texi2html/’ directory and lastly in the ‘/usr/local/share/texi2html/’ directory.

A file is also included based on the language selected, by $DOCUMENTLANGUAGE , ‘--document-language’ or @documentlanguage. All the files with name the language name in ‘/usr/local/share/texi2html/i18n/’, ‘/usr/local/etc/texi2html/i18n/’, ‘$HOME/.texi2html/i18n/’ and then ‘./i18n/’ are included.

The default initialization options are defined in the ‘texi2html.init’ file contained in the texi2html distribution (which gets included near the beginning of the texi2html script that gets installed).

To customize texi2html it is best if you copy the appropriate sections from the ‘texi2html.init’ contents into an appropriate local initialization file, make the necessary changes there, and then have texi2html read this initialization file by one of the means described above.

Some init files are provided with texi2html, for example ‘book.init’ which produces an output more in line with what could be in a book, or ‘chm.init’ outputs files that can be used to produce a CHM file.

5 Overview of initialization files content and loading

The initialization files are perl files, read as explained in Use initialization files for fine tuning. You don’t need to know much of perl to do some simple changes in variable values, however, to be able to really take advantage of all the features of the initialization file, a good knowledge of perl is required.

In initialization file two kind of variables appear. These are normal variables (including arrays and hashes) and references on functions. The later permits the dynamic redefinition of functions used to produce the HTML manual. You should be able to change the value of some normal variables without a deep knowledge of perl, by looking at the existing examples. The possible mistakes in that case could be omitted ‘;’, and bad quoting.

Initialization file are loaded from the main program by the mean of a require, while in the Texi2HTML::Config namespace. This means that the namespace of the main program and the namespace of initialization files are distinct, which ensures that no name clash should happen. The variables are declared with use vars, such that it should be possible to use the use strict pragma in the initialization file code.

To avoid messing with the variables in the main namespace all the global variables which could be of use in the init files are in the Texi2HTML namespace. Notice that the functions of the main program are still in the main namespace.

Since texi2html can proceed more than one file on the command line, you should make sure that you initialize the variables that are used during a manual formatting. The handlers explained later can be used for that (see section Bypassing normal formatting).

5.1 Setting the encodings

There are four encodings relevant for texi2html, they are associated with corresponding configuration variables. If these variables are defined they determine a corresponding value in %Texi2HTML::THISDOC which is otherwise autodetected:

1. The variable $DOCUMENT_ENCODING corresponds with the document encoding. If defined, this variable sets $Texi2HTML::THISDOC{'DOCUMENT_ENCODING'} . If not defined, the encoding appearing in @documentencoding will be used to set $Texi2HTML::THISDOC{'DOCUMENT_ENCODING'} . The @documentencoding value appears in $Texi2HTML::THISDOC{'documentencoding'} .
2. The texinfo files encoding. If $IN_ENCODING is set, this sets $Texi2HTML::THISDOC{'IN_ENCODING'} . Otherwise, when $Texi2HTML::THISDOC{'DOCUMENT_ENCODING'} is set, $Texi2HTML::THISDOC{'IN_ENCODING'} is also set if the encoding is supported by perl.
3. The out files encoding. It is associated with the variable $OUT_ENCODING . If defined, $Texi2HTML::THISDOC{'OUT_ENCODING'} is set accordingly. If not defined, the value of $Texi2HTML::THISDOC{'ENCODING_NAME'} or $Texi2HTML::THISDOC{'IN_ENCODING'} is used if one of these variables is set.
4. The encoding advertized in out files, associated with the variable $ENCODING_NAME . It sets $Texi2HTML::THISDOC{'ENCODING_NAME'} if defined.

   If unset the value of this variable is based on the other ENCODING values, and if they are all undefined, the variable $DEFAULT_ENCODING is used.

The values for the encoding related variables are set in the default init_out function reference (see section Preparing the output).

In general the $DOCUMENT_ENCODING and $IN_ENCODING are set to the right values. $OUT_ENCODING is also rightly set according to $ENCODING_NAME. To force a given encoding for the output, the $ENCODING_NAME value may be set. The current default output encoding is UTF-8.

5.2 Redefining functions in initialization files

To redefine a function you must replace the corresponding funtion reference with a reference on your function. Thus you should write your function, give it a name you are certain it is unique in the Texi2HTML::Config namespace, and override the value of the function reference with your own function reference. When another function from the main program (or from another functions of an initialization file) calls the reference, your function will be used.

For example the function reference corresponding with the function called when doing an anchor is called $anchor. Thus if you want to override the corresponding function you could write:

# override the function reference
$anchor = \&my_own_function;

# the function reference now refers to
sub my_own_function {
# process arguments and return an html anchor
}

5.3 Conventions used for function prototypes

As the functions are defined by a reference name, we will always use the reference name in function prototypes. For the function arguments we will use \@array for a reference on an array and similarly \%hash for a reference on a hash.

Thus, the prototype for the function associated with the function reference ‘$formatting_function’ will be:

Function Reference: $text formatting_function $arg1 \@arg2

formatting_function takes as first argument $arg2, as second argument a reference on an array \@arg2 and returns the formatted text $text.

To redefined the corresponding function, you should write:

$formatting_function = \&my_formatting_function

sub my_formatting_function($ $)
{
    my $arg1 = shift;
    my $arg2 = shift;
    # prepare $formatted_text
    .....
    return $formatted_text
}

6 Fine tuning of the page layout

Some features of the page layout might be specified with command line options, the corresponding variables are described in Page layout related command line options. Fine tuning of the page layout may be achieved with redefinition of other variables and function references in the initialization files.

6.1 The different categories of pages and sectioning elements

The following sectioning elements can be associated with pages:

Normal elements

These are normal sections or nodes. Their association with pages is determined by the splitting of the document. See section Specifying where to split the generated document.

Top element

The top element is the higher element in the document structure. If there is a @top section it is the element associated with that section. Otherwise it is the element associated with the @node Top. If there is no @node Top the first element is the top element.

The top element is formatted differently than a normal element if there is a @top section or the @node Top isn’t associated with a sectioning command.

Misc elements

These elements are associated with pages if the document is split. There are four misc elements:

1. Table of contents
2. Short table of contents, also called Overview
3. Footnotes page
4. About page

The About page shouldn’t be present for documents consisting in only one sectioning element, or for documents unsplit and without navigation information. The Footnote page should only be present if the footnotes appear on a separated page (see section Page layout related command line options), however a footnote element is present if the document isn’t split. The Table of contents should only be formatted if @contents is present in the document. Similarly the Overview should only appear if @shortcontents or @summarycontents is present. The Table of contents and the Overview may also be directly included within the document, not as separate pages (see section Table of contents and Short table of contents).

6.2 Page layout and navigation panel overview

A page is broken up in three parts. A page header, the sections and a page footer. A common element in the page layout is a navigation panel with icons or text linking to other sections or pages. Another common element is a rule, separating sections or footer. The navigation panel and the rules may be part of the sections or part of headers or footers. You may use the variables $SMALL_RULE , $DEFAULT_RULE , $MIDDLE_RULE and $BIG_RULE

for rules of different sizes. The defaults are

$SMALL_RULE = '<hr size="1">';
$DEFAULT_RULE = '<hr>';
$MIDDLE_RULE = '<hr size="2">';
$BIG_RULE = '<hr size="6">';

In the header some important meta data may be defined, like the title or style information, and textual informations may be present in comments. All this doesn’t appear directly in the displayed HTML, though.

The page layout is mainly controlled by functions, the precise functions called depending on the document splitting. The navigation panel, however, can be customized with variables.

Element labels

There are 19 items associated with elements. Each of these is associated with a name and a reference to the element they represent, when such an element exists. The element is either a global element or an element relative to the current element. The relative elements are found with respect with the document structure defined by the section structuring commands (@chapter, @unnumbered…) or by the nodes (in that case the node directions are specified on node line or in menu organization). These items are called element labels. They may be associated with a button (see section Specifying the buttons formatting), and used in the formatting functions (see section Main program variables and usefull functions).

Here is the list:

‘ ’

An empty button

Top

Top element. The associated name is $TOP_HEADING if that variable is defined. This variable is not set by default.

Contents

Table of contents

About

About (help) page

Overview

Overview, short table of contents

First

First element in reading order

Last

Last element in reading order

Index

The first chapter with @printindex. The associated name is $INDEX_CHAPTER , if the variable is set. This variable is not set by default.

This

The current element

Back

Preceding element in reading order

FastBack

Beginning of this chapter or previous chapter if the element is a chapter

Prev

Previous section on the same level

NodePrev

Previous node

Forward

Next element in reading order

FastForward

Next chapter

Next

Next section on the same level

NodeNext

Next node

Following

Next node in node reading order

Up

Up section

NodeUp

Up node

FileNext

Forward element first in the next page (or file)

FilePrev

Backward element first in the previous page (or file)

6.3 Customization of the navigation panels buttons

A lot of customization of the navigation panel may be achieved without redefining functions, with variables redefinition. In case it isn’t enough, it is also possible to redefine the function doing the navigation panel formatting.

6.3.1 Controlling the navigation panel panel at a high level

The global formatting of the navigation panels may be changed with the following variables:

$VERTICAL_HEAD_NAVIGATION

A vertical navigation panel will be used for the header navigation panel if this variable is true.

$ICONS

Icons are used instead of textual buttons if this variable is true.

$HEADERS

If this variable is false there is no section navigation, no navigation panels for the elements within the pages, only at the beginning and the end of the page (see section Page layout related command line options).

6.3.2 Specifying the buttons formatting

Several arrays and hashes enable a precise control on the buttons and their display. The following arrays determine the buttons present in navigation panels:

@SECTION_BUTTONS

This array is used for the navigation panel buttons present at the begining of sectioning elements. If split at node or section they are also used at the page footer, and in the case of section navigation at the page header.

@SECTION_FOOTER_BUTTONS

@NODE_FOOTER_BUTTONS

This array is used for the navigation panel buttons present at the footer of pages when split at node or at section.

If $WORDS_IN_PAGE is set and the output is split at nodes, these buttons are only present if there are more than $WORDS_IN_PAGE

words in the sectioning element text. This counting is very rough and include punctuation marks, html elements, numbers. The default is to include the buttons after 300 words.

@CHAPTER_BUTTONS

This array is used for the buttons appearing at the page footer if split at chapter, and at the page header if split at chapter and there is no section navigation.

@MISC_BUTTONS

These buttons appear at the beginning of special and sections and at the end of these section pages if the output is split.

@LINKS_BUTTONS

These are used for <link> elements if they are output in the headers.

The array specify the buttons displayed in navigation panels, and how the button is displayed. Each element is associated with a button of the navigation panel from left to right. The signification of the array element value is the following:

reference on a function

The function is called with argument a boolean true if the navigation panel should be vertical. Should return the formatted button text.

reference on a scalar

The scalar value is printed. For some possibly usefull scalars, Accessing elements informations.

reference on an array

In this case the first array element should be a reference on text and the second element an element label. In that case a link to the element associated with the element label with the scalar value text is generated.

For example if the buttons array element is

[ 'Next', \$Texi2HTML::NODE{Next} ] 

The button will be a link to the next section with text $Texi2HTML::NODE{Next} .

element label

If icons are not used, the button is a link to the corresponding element which text is defined by the value associated with the element label in the %NAVIGATION_TEXT hash, surrounded by ‘[’ and ‘]’. If the element label is ‘ ’, there is no ‘[’ and ‘]’. The element of the %NAVIGATION_TEXT hash are defined dynamically, in the init_out function reference (see section Preparing the output).

If icons are used, the button is an image with file determined by the value associated with the element label in the %ACTIVE_ICONS

hash if the the link really leads to an element, or in the %PASSIVE_ICONS

hash if there is no element to link to. Of course if there is a link to the element the icon links to that element. The button name and the button description are used in HTML attributes to have a textual description of the icon. The corresponding strings are in %BUTTONS_NAME for the button name and %NAVIGATION_TEXT for the description.

If $USE_ACCESSKEY is set, the accesskey attribute is used in navigation. In that case the %BUTTONS_ACCESSKEY

hash is used for the access key.

Similarly, if If $USE_REL_REV is set, the rel attribute is used in navigation. In that case the %BUTTONS_REL hash is used for the rel attribute.

6.3.3 Changing the navigation panel formatting

If you are not satisfied with this scheme, it is possible to control exactly the formatting of navigation panels by redefining a function reference. The function controlling the display of navigation panel is associated with the following function reference:

Function Reference: $navigation_text print_navigation \@buttons $vertical

\@buttons is an array reference which should hold the specification of the buttons for that navigation panel. $vertical is true if the navigation panel should be vertical. Returns the formatted navigation panel in $navigation_text.

6.4 Main program variables and usefull functions

In the functions controlling the page layout some global variables set by the main program are available, with value corresponding with the current layout element.

6.4.1 Accessing elements informations

Four hashes are available, with key the elements labels (as described in Element labels) and values:

%Texi2HTML::NAME

The formatted element name

%Texi2HTML::HREF

The element hypertext reference

%Texi2HTML::NODE

The element node name

%Texi2HTML::NO_TEXI

The element name after removal of texi commands

If $USE_NODE_TARGET is set, the node anchors are used as target for the section HREF, if there is a node associated to that section.

6.4.2 Accessing global informations

Three kinds of global informations are available, miscalleneous global strings, flags set by @set and special flags and section lines.

Strings that can be set by some @-commands

Some information may be set up by @-commands, by a variable in the init file, and sometime on the command-line. This is the case, for example, for @footnotestyle, ‘--footnote-style’ and $FOOTNOTESTYLE. The value associated with these should be read by calling the get_conf function, in the Texi2HTML::Config namespace, for example:

if (get_conf('footnotestyle') eq 'separate') { .... }

When it exists, the associated variable that sets the default value have the same name, in upper case. Currently the following information corresponding with @-commands is available through get_conf: kbdinputstyle, paragraphindent, setchapternewpage, headings, footnotestyle, exampleindent, firstparagraphindent, everyheading, everyfooting, evenheading, evenfooting, oddheading, oddfooting, setcontentsaftertitlepage, setshortcontentsaftertitlepage, frenchspacing, fillcolumn, documentlanguage, novalidate.

Information nthat can be set on the command line and in init files, or that could be modified on a per-output file basis are also available through get_conf:

SPLIT

How the manual is split.

SPLIT_SIZE

The split size, only relevant for info output.

doctype

The current doctype.

headers

True if headers are to be output.

Global strings

The %Texi2HTML::THISDOC hash holds some global informations:

fulltitle

title set by @settitle. If there is no @settitle other possibilities are tried (@title, @shorttitlepage…).

fulltitle_no_texi

fulltitle without texi formatting

fulltitle_texi

fulltitle with texi commands

title

title set by @title.

title_no_texi

title without texi formatting

title_texi

title with texi commands

author

Authors list set by @author.

authors

A reference on an array containing each author set by @author.

copying_comment

Text appearing in @copying with all the texinfo commands removed, put in comments.

program

The name and version of texi2html.

program_homepage

Homepage for texi2html.

program_authors

Authors of texi2html.

file_base_name

base name of the texinfo manual file.

filename

This is a reference on a hash that holds the filenames for special elements. These files may not be used in certain cases, for example the toc element file name may not be relevant if table of contents is not output separately. The keys are

doc

the document file if not split, if split should be the top element file.

top

Top element file name.

toc

Table of contents element file name.

stoc

Overview (also called short table of contents) element file name.

about

About element file name.

foot

Footnotes element file name.

frame

Main frame file.

toc_frame

Table of contents frame file name.

input_file_name

Name of the texinfo manual file given on the command line.

destination_directory

Destination directory for the resulting files.

extension

Extension for the output files.

toc_file

The file name of the table of contents, should always be valid, even when table of contents are output directly in the document.

inline_contents

A reference on a hash containing two key, one for each type of table of contents:

contents

The associated value is a reference on an array containg the line resulting from formatting the table of contents, including a heading and a reference.

shortcontents

The associated value is a reference on an array containg the line resulting from formatting the short table of contents, including a heading and a reference.

today

The date. May be overriden by $DATE.

css_import_lines

reference on an array containing the @import lines of CSS files.

css_lines

reference on an array containing the normal lines of CSS files.

Flags

Flags defined by @set may be accessed through the %main::value hash. The key is the flag name, the value is the flag value at the end of the document.

Section lines

The following array references or arrays holds formatted lines:

$Texi2HTML::THIS_SECTION

Lines of the current element.

$Texi2HTML::OVERVIEW

Lines of short table of contents. See section Special pages formatting.

$Texi2HTML::TOC_LINES

Lines of table of contents. See section Special pages formatting.

$Texi2HTML::TITLEPAGE

The title page formatted. See section Formatting of title page.

6.4.3 Function usefull in page formatting

The usefull function is a function used to print an array of lines, which also counts the number of words in the array, if needed.

Function: $words_number main::print_lines $filehandle \@lines_array

$filehandle is the opened filehandle the function should write to. \@lines_array is the array line the function should write to the file. If this argument is omitted, the function uses $Texi2HTML::THIS_SECTION . $words_number is the number of words in the array, only defined if split at nodes and $WORDS_IN_PAGE is defined.

6.5 Preparing the output

After the texinfo file has been parsed, some information is available which can be used to modify some variables and prepare the outputting. For example the document language, the document encoding, values set with @set or @setfilename and other similar @-commands are not known before the texinfo parsing.

The following function reference may be redefined to be called after texinfo processing and before document generation:

Function Reference: init_out

This function perform the initialization of variables and any other task before document outputting.

In the default case the $BODYTEXT (see section Customizing the page header) and the hashes %NAVIGATION_TEXT , %BUTTONS_NAME (see section Specifying the buttons formatting), %BUTTONS_GOTO (see section Formatting of about text) are initialized. Indeed the initialization of these variables is dependent upon the document language selection. Similarly the encoding variables are set based on the information now available (see section Setting the encodings).

To perform the default initializations and also add more code, you could do as in the following example (save the default function reference and call it in your own function) :

my $default_init_out = $init_out;
$init_out = \&makeinfo_like_init_out;
sub makeinfo_like_init_out() 
{
   &$default_init_out();
   $NAVIGATION_TEXT{'Following'} = ' > ';
}

6.6 Finalizing the output

If you want to do some cleaning after the document was generated (close files, write at the end of files and so on), the following function reference may be redefined:

Function Reference: finish_out

This function is called after the document generation.

The default is to do nothing.

6.7 Customizing the texi2html css lines

If the variable $CSS_LINES is set it is used for the css entries. For example if you don’t want any css entries, set

$CSS_LINES = '';

If this variable is undef (as in th edefault case), it is possible to modify the texi2html css lines by modifying the entries or adding to the %css_map hash. Each key is a css selector, the corresponding value is a style string.

Another possiblility is to modify the array corresponding with the array reference $Texi2HTML::THISDOC{'css_import_lines'} that contains the @import lines of CSS files, and similarly it is possible to modify the array corresponding with the array reference $Texi2HTML::THISDOC{'css_lines'} that contains the normal CSS files lines (for details on what corresponds with those different lines, see HTML CSS in GNU Texinfo). The right place to modify these arrays is in a function appearing in the @command_handler_process array (see section Bypassing normal formatting). Later, the CSS lines are allready expanded, by the function reference below.

In th edefault case, the resulting css lines are in $Texi2html::THISDOC{'CSS_LINES'} . It is also possible to change completely the way $Texi2html::THISDOC{'CSS_LINES'} are generated by redefining the following function reference:

Function Reference: css_lines \@import_lines \@rule_lines

This function should be used to construct the variable $Texi2html::THISDOC{'CSS_LINES'}. \@import_lines are the @import lines of the files specified with ‘--include-css’ , and \@rule_lines are the css commands lines of these files. See section Customizing the HTML and text style.

6.8 Customizing the page header

It is possible to add lines to the text within the <head> HTML elements, by defining the variable $EXTRA_HEAD . Similarly it is possible to add text just after the <body> element with the variable $AFTER_BODY_OPEN . These variables are empty by default.

The HTML encoding of the resulting document is defined by $ENCODING_NAME . If the variable isn’t defined, the @documentencoding value is used, or the $OUT_ENCODING value, if set. $ENCODING_NAME may influence the value of $OUT_ENCODING, which corresponds with the encoding used when writing to the resulting files. See section Setting the encodings.

The description of the document may be specified in $DOCUMENT_DESCRIPTION . If this variable is undef, the text associated with @documentdescription is used, and if there isn’t such test a default description is constructed using the document title and the name of the first section of the file. The value used during document formatting is $Texi2HTML::THISDOC{'DOCUMENT_DESCRIPTION'} .

The <body> element attributes may be set by defining the variable $BODYTEXT . The resulting attributes are in $Texi2HTML::THISDOC{'BODYTEXT'} . If you want to define that variable dynamically, you should use the init_out function reference (see section Preparing the output).

<link> element are used in the header if $USE_LINKS

is set. @LINKS_BUTTONS determines which links are used. %BUTTONS_REL determines the link type associated with the rel attribute.

The default functions call the function associated with $print_head_navigation to format the navigation panel for the page header. Thus you can control parts of the formatting by redefining the function reference.

Function Reference: print_head_navigation $filehandle \@buttons

$filehandle is the opened filehandle the function should write to. \@buttons is an array reference which should hold the specification of the buttons for the navigation panel.

If you want even more control, you can have full control over the page header formatting by redefining three function references. The function associated with $print_page_head is called for all the pages, and after that, the function associated with $print_chapter_header is called if the document is split at chapters, or the function associated with $print_section_header is called if the document is split at sections.

Function Reference: print_page_head $filehandle

$filehandle is the opened filehandle the function should write to. This function should print the page head, including the <body> element.

Function Reference: print_chapter_header $filehandle

$filehandle is the opened filehandle the function should write to. This function is called if the document is split at chapters, after print_page_head.

Function Reference: print_section_header $filehandle

$filehandle is the opened filehandle the function should write to. This function is called if the document is split at sections, after print_page_head.

6.9 Customizing the sections

The functions associated with the following function references are used for the formatting of sections:

Function Reference: print_section $filehandle $first_in_page $previous_is_top

$filehandle is the opened filehandle the function should write to. $first_in_page is true if this section is the first section in the page. $previous_is_top is true if this section is the section following the Top section. This function should print the current section contents.

Function Reference: end_section $filehandle $last_element_or_before_top

$filehandle is the opened filehandle the function should write to. $last_element_or_before_top is true if this section precedes the top element or is the last one in page, or before the special elements.

6.10 Customizing the page footer

It is possible to add text just before the </body> element with the variable $PRE_BODY_CLOSE . Nothing is added by default.

A date is collected to be output in the footer. You can change it by defining $DATE in the initialization file.

The default functions call the function associated with $print_foot_navigation to format the navigation panel for the page footer. Thus you can control parts of the formatting by redefining the function reference.

Function Reference: print_foot_navigation $filehandle \@buttons

$filehandle is the opened filehandle the function should write to. \@buttons is an array reference which should hold the specification of the buttons for the navigation panel.

If you want even more control, you can have more control over the page footer formatting by redefining three function references. The function associated with $print_chapter_footer is called if the document is split at chapters, or the function associated with $print_section_footer is called if the document is split at sections. After that the function associated with $print_page_foot is called.

Function Reference: print_page_foot $filehandle

$filehandle is the opened filehandle the function should write to. This function should print the page foot, including the </body> element.

Function Reference: print_chapter_footer $filehandle

$filehandle is the opened filehandle the function should write to. This function is called if the document is split at chapters, before print_page_foot.

Function Reference: print_section_footer $filehandle

$filehandle is the opened filehandle the function should write to. This function is called if the document is split at sections, before print_page_foot.

6.11 Special pages formatting

For the special elements, two things must be formatted: the content and the page layout

6.11.1 Customizing the content of the special pages

The label for the special elements, except for the Top element is formatted according to the function reference $misc_element_label :

Function Reference: $misc_element_label misc_element_label $identifier $page_name

$identifier is the identifier associated with the special element. $page_name is the special element name. It should return a label that can be used for references to the special element.

6.11.1.1 Top element text formatting

The top element formatting is controlled by three function which also controls the layout of the top element page or section. The associated function references are:

Function Reference: print_Top_header $filehandle $begin_page

$filehandle is the opened filehandle the function should write to. $begin_page is true if the element is the first in a page. This function should begin the Top element. At the time this function is called the top element text hasn’t been parsed.

Function Reference: print_Top $filehandle $has_top_heading

$filehandle is the opened filehandle the function should write to. $has_top_heading is true if there is a @heading command or @titlefont command appearing in the Top element text. This function should be used to format the Top element text and navigation panel.

Function Reference: print_Top_footer $filehandle $end_page

$filehandle is the opened filehandle the function should write to. $end_page is true if the element is the last in a page. This function should end the Top element.

6.11.1.2 Table of contents and Short table of contents

Two possibilities exist for the formatting of table of contents (and short table of contents). In the default case, the table of contents are in separate elements, at the end of the document if the document is unsplit or in separate files. This is consistent with makeinfo where menus are used for navigation. Another mode may be selected by setting $INLINE_CONTENTS . In that case the table of contents are not output as separate elements but are instead output where the corresponding @-command, for example @contents, is set. This behaviour is more consistent with texi2dvi. If @setcontentsaftertitlepage appears in the document, and even if $INLINE_CONTENTS is set, the table of contents are merged in the title (which isn’t output in the default case, see Formatting of title page).

Several variables may be used to control the formatting of table of contents and short table of contents:

$CONTENTS

If the variable is true a table of contents is done even if there is no @contents command. If it is defined and false, no table of contents is done even if there is a @contents command.

$SHORTCONTENTS

If the variable is true a short table of contents is done even if there is no @summarycontents command. If it is defined and false, no short table of contents is done even if there is a @summarycontents command.

$BEFORE_OVERVIEW

The variable value is inserted before the short table of contents text.

$AFTER_OVERVIEW

The variable value is inserted after the short table of contents text.

$BEFORE_TOC_LINES

The variable value is inserted before the table of contents text.

$AFTER_TOC_LINES

The variable value is inserted after the table of contents text.

$NO_BULLET_LIST_STYLE

This should contain a css style used for the list style when there is no bullet.

$NO_BULLET_LIST_CLASS

This should contain the class assocciated with the $NO_BULLET_LIST_STYLE css style.

$NO_BULLET_LIST_ATTRIBUTE

This should contain an attribute text used for the list element when there is no bullet. For example it is used in the tables of if they are formatted with a list.

More control on the table of contents and short table of contents formatting may be achieved by redefining a function with the following associated function reference:

Function Reference: toc_body \@elements

\@elements is an array reference contining informations about all the elements of the document. Each of the entry of this array is an hash reference which entries correspond with different informations about the element. Interesting keys have the following meaning:

top

true if the element is the top element,

index_page

true if the element is an index page added because of index splitting,

toc_level

level of the element in the table of content. Highest level is 1 for the top element and for chapters, appendix and so on, 2 for section, unnumberedsec and so on...

tocid

label used for reference linking to the element in table of contents,

file

the file containing the element, usefull to do href to that file in case the document is split,

text

text of the element, with section number,

name

text of the element, without section number.

This function doesn’t return anything but should fill the array corresponding with the $Texi2HTML::TOC_LINES and $Texi2HTML::OVERVIEW references with the table of contents and short table of contents.

Another function reference is used to add a heading and a reference, to be used with $INLINE_CONTENTS or merged in the title. Its output is not used when the table of contents are separate elements.

Function Reference: \ @inline_contents_lines inline_contents $filehandle $command $element

This function reference returns a reference on an array holding the lines containing the contents, heading and reference. $filehandle is a reference on the currently opened file if the function is called because a @contents or @shortcontents command was encountered, it is undef otherwise. $command is either ‘contents’ or ‘shortcontents’. $element is a hash reference containing informations about the table of contents context. Relevant keys are:

target

The identifier associated with the table of contents, used for example to do references to the table of contents using href in HTML.

id

The identifier associated with the element, used to do labels. In general the same than the target, but not necessarily.

file

The file name containing the table of contents.

6.11.1.3 Formatting of footnotes text

The footnotes text is allready formatting when @footnote commands are expanded. See section Customizing the footnotes formatting.

6.11.1.4 Formatting of about text

The default about element contains an explaination of the buttons used in the document (@SECTION_BUTTONS, Specifying the buttons formatting) and an example locating the buttons targets in an example. The formatting of this text may be influenced by the following hashes and variables:

$PRE_ABOUT

$AFTER_ABOUT

This variable may be a scalar or a function reference. If it is a scalar, the value is used. If this is a function reference it is expanded and the returned text is used. The text is added before or after the main about text.

%BUTTONS_GOTO

The keys of this hash are element labels (see Element labels). The value is the text associated with the element label in the about text. The element of the hash are defined dynamically, you should in the init_out function reference (see section Preparing the output).

%BUTTONS_EXAMPLE

The keys of this hash are element labels (see Element labels). The value is the text associated with the element label in the about example, typically a section number.

If this is not enough and you want to control exactly the formatting of the about text, you can redefine the function associated with the following function reference:

Function Reference: $about_text print_about

This function should return the about text.

6.11.1.5 Formatting of title page

The title page is first formatted using the text appearing in the @titlepage section, and put in $Texi2HTML::TITLEPAGE . More formatting can be done using the following function reference:

Function Reference: titlepage

This function should complete $Texi2HTML::TITLEPAGE.

In the default case, in this function the title is output if there is no titlepage, the table of contents and short table of contents are also added if they are to be output and @setcontentsaftertitlepage or @setshortcontentsaftertitlepage appear in the document (see section Table of contents and Short table of contents).

In the default case the resulting title page output is not used in the document, except if the top node is not associated with any content.

6.11.2 Customizing the layout of the special pages

The formatting of each of the special pages, or section in case the document is not split, is controlled by a function. The associated function reference is called accordingly:

print_Top

print_Top_header

print_Top_footer

Formatting of top element page or section. It is also used for the formatting of the top element text (see section Top element text formatting).

print_Toc

Formatting of table of contents page or section

print_Overview

Formatting of short table of contents page or section

print_About

Formatting of about (help) page or section

print_Footnotes

Formatting of footnotes section or page in case footnotes are on a separated page or the document isn’t split.

In the default case, $print_Top calls $print_Top_header for the header and $print_Top_footer for the footer of top element. All the other function call $print_misc which in turn calls $print_misc_header for the headers and $print_misc_footer

for the footers.

6.12 Customizing the file and target names

File names

It is possible to specify the file names with more control than with the command line options (see section Setting output file and directory names). First the extension may be overrided by the variable $EXTENSION

value. The variable should be undef if no extension is to be added. Two function references enable further customization. One is usefull in case $NODE_FILES is true and it is used to customize the node file name.

Function Reference: $node_file node_file_name \%node

\%node is a hash reference with the following interesting keys (there are much more keys):

texi

The texinfo node name.

with_section

True if associated with a section.

The result is the node file name $node_file.

The other is used to customize the file names associated with each element, and the name of the file associated with the special elements.

Function Reference: $file element_file_name \%element $type $docu_name

\%element is undefined for the special elements (about, overview, table of contents, footnotes). Otherwise it is a hash reference with the following interesting keys (there are much more keys):

texi

The texinfo element name.

number

The number associated with a section.

doc_nr

A number incremented whenever a new file should begin, based on how the document is split (see section Specifying where to split the generated document).

text

The element text.

name

The element text without section number.

$type is empty for normal elements. For the top element it is ‘top’, for the table of contents it is ‘toc’, for the overview it is ‘stoc’, for the footnotes it is ‘foot’ and for about is ‘about’. If frames are used (see section Page layout related command line options), the function reference is also called for ‘frame’, the frame file name, and ‘toc_frame’ the table of content frame file name. $docu_name is the basename of the texinfo manual. The result is the element or special element file name.

target names

Similarly target and id may be set. The id is placed where the item is located, the target is used to construct references to that item. In general they should be equal, but not always, for example in the default case, the target for a section is the node id. The following function reference, is for target items (nodes, anchors, floats):

Function Reference: ($target,$id) node_target_name \%node, $default_target, $default_id

\%node is the same as in the node_file_name function reference above. $default_target is the target already set (it is also in $node->{'target'}), and $default_id is similarly the id already set.

For element associated with files (which may be nodes), the function reference is:

Function Reference: ($target,$id) element_target_name \%element, $default_target, $default_id

the \%element is the same than in element_file_name, and $default_target and $default_id are the target and id already set.

Placed items (floats, footnotes, index entries, anchors, contents, shortcontents and headings) file and target may also be set. In the default case, they should be rightly set, so be careful when changing them. The following function reference can be used:

Function Reference: ($target, $id, $file) placed_target_file_name \%placed_item, \%element, $default_target, $default_id, $default_file, $context

\%placed_item is a hash reference describing the placed item, in the same vein than above. the \%element is the same than in element_file_name, corresponding with the element containing the placed item. $default_file, default_id and $default_target are the file, id and target already set. $context describes the context, it is empty in the normal cases, and can also be set to ‘footnotes’ if in footnotes, or to ‘no_associated_element’ if the placed item is out of any element (typically in @titlepage, @copying).

For special elements, the %misc_pages_targets hash is used to set the target and id. The possibilities for the keys are ‘Overview’, ‘Contents’, ‘Footnotes’ and ‘About’.

6.13 Generation of external files for index entries

Within the document, @printindex commands are expanded as explained in Customizing the formatting of index lists. In case you want to do something special with index entries, outside of the document, you should first set the variable $IDX_SUMMARY true. After that some function reference will be called for each non empty index. For each index there are 3 function references, one called for initialization, one called for each index entry and the last one called for finalization.

Function Reference: index_summary_file_begin $index_name $is_printed $manual_name

$index_name is the two letters name for the index. This function is called for each index appearing in the document, before index_summary_file_entry. $is_printed is true if there is a @printindex for that index. $manual_name is the manual basename.

Function Reference: index_summary_file_entry $index_name $entry_text $entry_reference $formatted_entry $texi_entry $entry_element_reference $entry_element_header $is_printed $manual_name

This function is called for each entry of an index. index_name is the name of the index. $entry_text is the entry in plain text, $formatted_entry is the index entry formatted, $texi_entry is the entry with texinfo commands. $entry_reference is the reference placed at the index entry place, in the form ‘file#id’. $entry_element_header is the formatted header of the element containing the index entry. entry_element_header is the reference to the beginning of the element containing the index entry, in the form ‘file#id’. $is_printed is true if there is a @printindex for that index. $manual_name is the manual basename.

Function Reference: index_summary_file_end $index_name $is_printed $manual_name

$index_name is the two letters name for the index. This function is called for each index appearing in the document, after index_summary_file_entry. $is_printed is true if there is a @printindex for that index. $manual_name is the manual basename.

7 Customizing HTML and text style in init files

Some simple customization may be achieved with the redefinition of the variables associated with the command line options. For the description and an explanation of the meaning of these variables, Customizing the HTML and text style.

Other variables and hash entries can be modified in initialization file to achieve more customization. Lastly, functions references corresponding with functions called from the main program and initialization files may be redefined.

7.1 Three contexts for expansions: preformatted, normal and string

There are three contexts of interest, one is the normal context, the other is a special context, called the preformatted context and the last is the string context. The preformatted context occurs when the spacing between words is kept. This is the case, for example, in @display or @example regions, and in menu comments (see section Menu formatting). The preformatted regions are usually rendered in <pre> elements in HTML. The string context occurs when rendering strings without formatting elements, in comments or titles for example.

7.2 Three passes: macro expansion, document structure and output

There are three passes in texi2html. During pass 0, the @macro are expanded, in pass 1 the document structure is gathered and in pass 2 the result is output. In most cases you shouldn’t care about it, as almost all of the output customization is done in pass 2. Only if you want to do something before the pass 2 should you care.

7.3 Customizing the formatting of commands without argument

This includes the commands whose name is a nonletter character like @@, the commands with lettered characters and braces but whose braces should be empty, like @TeX{}, or some commands associated with accentted letters like @AA{}. If there happens to be something within the braces, it is put after the command, thus

@TeX{something}

leads to the same than

@TeX{} something

Each of these categories of commands have three associated hashes, one for normal context, the other for preformatted context and the last in strings. The keys of the hashes are the command names, the associated value is the text replacing the command.

The hashes are:

To change the HTML resulting from these constructs, just change the value. For example, if you want ­ to be outputted for @- in normal and preformatted context, write in your init file:

$simple_map{'-'} = '­';
$simple_map_pre{'-'} = '­';

7.4 Punctuation commands

The formatting of a punctuation character followed by is determined by the hash %colon_command_punctuation_characters . If a command is preceded by a character in th is hash, it is replaced by the associated value. In the default case, the associated value is also the character, so this leave the punctuation character unmodified.

The following function reference may be redefined to handle characters that are in %colon_command_punctuation_characters:

Function Reference: $punctuation $colon_command $character

The $character is a character appearing in %colon_command_punctuation_characters and preceding a command. In the default case the associated value in %colon_command_punctuation_characters is returned.

7.5 Customizing accent, style and other simple commands

The formatting of the HTML produced by style and indicatric commands (@tt, @code, @email, @titlefont), the accentuation related commands taking argument (@', @udotaccent, @dotless) and miscalleneous commands (@email, @verb, @w, @uref, @math, @asis) is controlled by two hash in the default case, %style_map for normal context, %style_map_pre for preformatted context and %style_map_texi in string context.

The key of the hashes are the command names. There are two possibilities for the values corresponding with two interfaces. The values may be strings or hash references, and you can chose the interface depending on the one you prefer. The interface with hash reference is a bit more flexible but might also be regarded as more complex. If you don’t like either of these interfaces you can define your own.

Some remarks are in order:

• The nonlettered accent commands which following character is considered to be the argument (like in @`a) should be keys of the hash %accent_map hash, even if no value is associated.
• @math is handled differently if LaTeX2HTML is used.

7.5.1 An interface for commands formatting with a hash reference

The key of the hashes are the command names. The value determine how the command argument is formatted. This value is a reference on a hash. In this hash each key corresponds with a type of information for the formatting, and the value is the corresponding information. For example, in

$style_map{'command'} = { 'args' => ['code'], 'inline_attribute' => 'code'};

the arguments for @command are interpreted as specified by the values associated with the ‘args’ key while the inline_attribute associated with that command is ‘code’.

The following keys in the hashes associated with each command have the following meaning:

‘args’

The value associated is a reference on an array. Each element of the array defines how the arguments (separated by ‘,’ in the texinfo code) for the @-command should be formatted. The possibilities are

normal

for normal text,

code

for text with ‘---’, ‘--’, ‘''’ and ‘``’ kept as is,

keep

if the texinfo should be kept as is, without interpretation of the @-commands.

For example, we have

$style_map{'email'}->{'args'} = ['code', 'normal'];

because ‘---’, ‘--’, ‘''’ and ‘``’ should be kept as is in the first argument of @email.

The default is ‘['normal']’.

‘attribute’

If the associated value is a word, it is considered to be an XML element name, and the argument is enclosed between the element opening and the element closing. For example, if the value is elem, the resulting HTML is <elem>arg</elem>.

If the text is a word followed by some text, the word and is interpreted as above, and the text is considered to be the attributes text of the element. Thus elem class="elem" leads to <elem class="elem">arg</elem>. This works only if there is only one argument.

‘inline_attribute’

Like an attribute, except that it is closed at each paragraph end and reopened at each beginning of paragraph. This is in fact more used than ‘attribute’ since it allows to have well formed HTML/XML.

‘inline_begin’

The associated value is added in front of the text and each time a paragraph is restarted while within the command.

‘inline_end’

The associated value is added after the text and each time a paragraph is ended while within the command.

‘begin’

The associated value is added in front of the text.

‘begin’

The associated value is added after the text.

‘quotes’

If the corresponding value is true, the result is enclosed in quotes $OPEN_QUOTE_SYMBOL and $CLOSE_QUOTE_SYMBOL , with defaults ‘`’ and ‘'’.

‘function’

The corresponding value should be a function reference. The corresponding function is called with the following arguments:

$command

The @-command name

$args

A reference on an array containing the arguments of the @-command.

$command_stack

A reference on an array containing the name of the @-commands containing the @-command being formatted, latest on top.

$state

A reference on a hash containing a lot of informations about the context of the @-command.

$line_nr

An opaque structure containing the information about the line number of the @-command. It can be used to call main::line_error or main::line_warn with first argument a message, and second argument $line_nr.

7.5.2 Defining the style and indicatric commands interface

If you don’t like this scheme, it is possible to change how those commands are processed by redefining the following function reference:

Function Reference: $resulting_text style $style $command $text $args $no_close $no_open $line_nr $state $command_stack

$command is the @-command, $style is the value associated with the $command in the %style_map, %style_map_pre or %style_map_texi hashes. The $text is the text appearing within the @-command braces. args is a reference on an array contening the command arguments formatted according to the same conventions than with the reference hash style (provided the value associated with the @-command is a hash reference with a $arg key as described in Reference hash args). If $text is split in paragraphs each paragraph is passed through the function, and $no_close is true if it is not the last paragraph, while $no_open is true if it is not the first paragraph. $line_nr is an opaque structure containing the information about the line number of the @-command. It can be used to call main::echo_error or main::echo_warning with first argument a message, and second argument $line_nr. $state is a reference on a hash containing a lot of informations about the context of the @-command. $command_stack is a reference on an array containing the name of the @-commands containing the @-command being formatted.

7.6 Formatting of special simple commands

The formatting of special simple commands is controlled by functions. To customize the output, the corresponding function references should be redefined. All these functions return a formatted text.

The formatting of anchors is controlled by $anchor_label .

Function Reference: $anchor_label anchor_label $identifier $anchor

$identifier is the anchor identifier, $anchoris the @anchor argument.

In the default case, it uses a function reference, $anchor

that can do a reference target or link. It is especially relevant for HTML but can be used in other formats, it is a rather common element of different formats.

Function Reference: $anchor anchor $identifier $href $text $attributes

If $identifier is not empty, this value should be used to create a target for links (typically associated with a name or id attribute in HTML). The $href argument specifies a hpertextual reference which should be used to link to a target. In case both $identifier and $href are given the text produced should be both a target for $identifier and a link to $href. $text is the text to be displayed. $attributes are additional attributes. It should be reasonable to assume that the attributes are for a <a> HTML element.

To customize the images produced by @image, the first possibility is to modify the @IMAGE_EXTENSIONS , which holds a list of filename extensions for image files. It is also possible to redefine the function used to determine the filename of the image:

Warning: This description is wrong. The API is still moving, so don’t count on it.

Function Reference: $filename image_files $basename $extension

$basename is the first @image argument, $extension is the corresponding @image argument. This function reference should return an array of image filenames without path that the main program should look for.

Last, it is possible to control the formatting of @image by redefining:

Function Reference: $image image $file_path $basename $preformatted $file_name $alt_text $width $height $raw_alt $extension $working_dir $file_relative_path

$file_path is the image file name with the path from the output directory to the source manual directory prepended, $basename the file name without extension (the first @image argument). $preformatted is true if the image appears in preformatted text. $file_name is the file name without path but with extension. $alt_text is the alternate text, it may be undefined. $width and $height are the corresponding arguments of @image, $raw_alt is the unmodified alt argument of @image and $extension holds the corresponding @image argument. $working_dir is the path to working dir relative to the output directory. $file_relative_path is the file name relative to the $working_dir.

The formatting of @sp is controlled by:

Function Reference: $sp sp $number $preformatted

$number is the numeric argument of @sp. $preformatted is true if the @sp appears in preformatted text.

The formatting of @acronym and @abbr is controlled by:

Function Reference: $acronym acronym_like $acronym_texi $acronym_text $with_explanation \@explanation_lines $explanation_text $explanation_simply_formatted

$acronym_texi is the acronym argument with texinfo @-commands, $acronym_text is formatted.

The other arguments are related with the explanation, the second arg of the acronym. $with_explanation is true if the second argument of the acronym command is present. If an explanation exists, coming from previous @acronym or as an arg of this command, the other args are defined: \@explanation_lines is a reference on an array containing the simply fomatted explanation lines, $explanation_text is the explanation text formatted, $explanation_simply_formatted is the explanation with a light formatting, unabling in HTML (or XML) the explanation to be in an attribute.

7.7 Processing special characters in text

Some characters are processed especially in text: ‘---’, ‘--’, ‘``’ and ‘''’. This is done only if in normal text and not in some commands (@code, @env…). A function reference is called to process the text and should take care of those constructs. It may also be used to transform the text, for example set it in upper case if it is in @sc. This function should also take care of protecting special characters

Function Reference: $processed_text normal_text $text $in_raw_text $in_preformatted $in_code $in_math $in_simple $command_stack

The function processes $text and returns $processed_text. The other arguments give some information about the context of the text. $in_raw_text is true if the text appears in special place where there is no formatting, typically in comments. $in_preformatted is true if in a preformatted environemnt, and $in_code is true if in a special command like @code, @env where ‘---’, ‘--’, ‘``’ and ‘''’ should not be touched. $in_math is true if in @math. $in_simple is true if in string context. $command_stack is an array containing the name of the formatting @-command that enclose the text.

In the default case the ‘---’, ‘--’, ‘``’ and ‘''’ constructs are expanded if needed and the text is upper-cased if in @sc. Special characters (‘&’, ‘"’, ‘<’ and ‘>’ in HTML) are protected if needed.

Some characters are special, for example we have ‘&’, ‘"’, ‘<’ and ‘>’ in HTML. In some cases some pieces of text don’t go through the above function, but still needs to be protected to appear in text. This is done by the function associated with the function reference

Function Reference: $protected_text protect_text $text

The function processes the unprotected text $text and returns the resulting protected text $protected_text.

Empty lines are processed by the following function reference, which could be usefull if empty lines are to be removed for example

Function Reference: $resulting_text empty_line $empty_line $state

This function processes an $empty_line and returns the resulting text $resulting_text. $state is a structure that holds informations about the state of the parsing. Empty lines are left as is by default except right after a definition @-command.

7.8 Customizing of line command output

@-commands that appear on a line and take the line as argument may be formatted especially, if they are in the %line_command_map . In that case the following function reference is used to format the @-command:

Function Reference: $resulting_text line_command $command $arg_text $arg_texi $state

$command is the @-command. $arg_text is the @-command formatted argument, $arg_texi is the @-command argument without any formatting. $state is a structure that holds informations about the state of the parsing. The resulting text is $resulting_text.

In the default case, @-commands appearing in @titlepage, @title, @subtitle, @author are formatted by this function.

7.9 Customizing strings written by texi2html

texi2html writes some strings in the generated document at various places, at the page footers, on the help page, for special section headings, buttons alt text and so on. These strings are customizable. The string chosen depends on the language of the document (set by ‘--document-language’ , $DOCUMENTLANGUAGE or @documentlanguage). This is the basis for internationalization as it allows for strings translations.

The Gettext framework is used for those strings. libintl-perl is used as a gettext implementation, and more precisely the pure perl implementation is used, to be sure to have a consistent gettext-like implementation which is not the case if the system one is used. libintl-perl is shipped in texi2html and installed to be sure that it is available. It is also possible to use the system libintl (currently decided at build-time).

The texi2html_document domain is used for those strings. These strings are texinfo strings, which may have @-commands, and the variable parts of the strings are denoted by ‘{arg_name}’ as is common practice for perl gettext. For more on internationalization, see Internationalization.

7.9.1 Old style translations

A gettext-like framework appeared in version 5.0. For reference the previous handling of strings (up to 1.82) is still documented here. With that system, the strings are found in a hash reference, $LANGUAGES . Each key is a language code. The associated value is also a hash reference. The key is an english string and the associated value is the string replacing the english string, if present. For example, we have

$LANGUAGES->{'fr'} = {
              ' Up ' => 'Plus haut',
};

It means that whenever the string ‘ Up ’ is to be written and the language is ‘fr’, ‘Plus haut’ is written. It is possible to customize the english strings by redefining the ‘en’ language hash.

When a string contains a substring like ‘{’ name ‘}’ it means that the string will be expanded by texi2html. For example, if we have

$LANGUAGES->{'fr'} = {
              'See {node_file_href}' => 'Voir {node_file_href}',
};

‘{node_file_href}’ will be expanded to an href for a node in a file by texi2html in the string.

When a @documentlanguage appears in the document and the language wasn’t set on the command line, it may be convenient for the user to redefine some variables based on the new language. There is a function reference that may be used for that, it is called each time a @documentlanguage is encountered:

Function Reference: $translate_names

This function is called each time @documentlanguage is encountered and the language wasn’t set on the command line. It should be used to retranslate some strings based on the new language.

7.10 References

7.10.1 Reference to external manual

The references are produced with two function references, one for the hypertextual reference construction, the other for the full reference to external manual.

Function Reference: $href external_href $node $node_identifier $xml_node_identifier $manual_file_name

$node is the node name, with @-commands. $node_identifer is the node name mapped to an identifier acceptable as a file name. $xml_node_identifier is the node name mapped to an identifier acceptable as an XML identifier. Those identifiers are built as explained in HTML Xref in GNU Texinfo, thus allowing for cross references to external manuals. $file is the manual or file name of the external reference. This function should return an href leading to the external manual.

The default for this function is to make a reference compatible with makeinfo (see HTML Xref in GNU Texinfo).

Function Reference: $text external_ref $command $section $book $file $href $cross_ref_name \@args_texi \@formatted_args $node

This function formats a reference to an external texinfo manual. The $command is the ref command (ref, xref or pxref, in text, at sentence beginning or in parenthesis). The optionnal $section argument is the section in the book and book is the book title. $node_and_file is manual file name. $href it an hypertextual reference to the distant manual constructed using the above function. $cross_ref_name is an optionnal cross reference name appearing in the reference command. \@args_texi is a reference on an array containing the @-command arguments, not formatted, with \@formatted_args contains the formatted @-command arguments. $node is the node name, formatted. This function returns the text corresponding with the external html manual reference. This function returns the full formatted text of the external reference.

7.10.2 Reference to an internal node

A function reference is available for internal references.

Function Reference: $text internal_ref $command $href $short_name $name $is_section \@args_texi \@formatted_args

This function formats a reference to a node in the current manual. The $command is the ref command (ref, xref or pxref, in text, at sentence beginning or in parenthesis). $href it an hypertextual reference linking to the corresponding node or section. $short_name and $name hold the text for the reference but $short_name can be the node name which is assumed to be shorter than the section name. $is_section is a boolean true if the reference is a reference to a section. \@args_texi is a reference on an array containing the @-command arguments, not formatted, with \@formatted_args contains the formatted @-command arguments. This function returns the full formatted text of the internal reference.

7.11 Commands used for centering and flushing of text

When a command controlling the alignement of text is used (@center, @flushleft and @flushright), the main program takes care of opening and closing paragraphs. The alignement commands are the key of the %paragraph_style hash. The value is used in the function doing the formatting of the paragraphs. See section Formatting (or not) a paragraph and a preformatted region.

A function references allows for a customization of the formatting of the text appearing in the command block.

Function Reference: $result paragraph_style_command $command $text

$command is the command name, $text is the text appearing within the command. This function returns a formatted text. The default is to return the text unmodified.

7.12 Formatting (or not) a paragraph and a preformatted region

7.12.1 Paragraph and preformatted region formatting

The formatting of a paragraph region or a preformatted region, is controlled by function references:

Function Reference: $paragraph_text paragraph $text $alignement $index $formatting_command $formatting_command_formatted \$paragraph_number $format $item_number $enumerate_style $number $command_stack_at_end $command_stack_at_begin

This function formats a paragraph. $text is the text of the paragraph, $alignement is the empty string when no alignement command has been seen, otherwise it is the current alignement command name. See section Commands used for centering and flushing of text. $indent holds ‘noindent’ or ‘indent’ if the corresponding @-command appeared in the paragraph. $command_stack_at_end and $command_stack_at_begin are arrays containing the opened @-commands at end and at beginning of the paragraph, latest on top.

The remaining arguments are usefull when the paragraph appears within a list or table. It is usefull whenever the paragraph has to be formatted differently when appearing in such environments. Moreover in that case the format command (@itemize…) may have an associated formatting command. $formatting_command is this formatting command (like @minus). $formatting_command_formatted is the command formatted in html in case the formatting command is a leading command (like @minus) which should be leading the first paragraph. \$paragraph_number is a reference on the number of paragraphs in that format command. The corresponding variable should be increased when a paragraph is added. $format is the format command. See section Formatting individual table and list items.

If the $format is an enumerate, $item_number is the number of the item in the list, $enumerate_style is the argument of the enumerate, $number is the number or letter corresponding with this item.

Function Reference: $preformatted_text preformatted $text $style $region_name $formatting_command $formatting_command_formatted \$preformatted_number $format $item_number $enumerate_style $number $command_stack_at_end $command_stack_at_begin

This function formats a preformatted region. $text is the text of the preformatted region, $style is the css style associated with that preformatted region (see section Customizing the texi2html css lines). $region_name is the name of the command opening the preformatted region (example…, see Formatting of complex formats (@example, @display…)) or a identifier for the preformatted context (for example menu-comment, see Menu formatting). The alignment commands are not taken into account, as the spaces are preserved in preformatted regions, you should flush and center by hand. $command_stack_at_end and $command_stack_at_begin are arrays containing the opened @-commands at end and at beginning of the preformatted region, latest on top.

The remaining arguments are usefull when the preformatted region appears within a list or table. It is usefull whenever the preformatted region has to be formatted differently when appearing in such environments. Moreover in that case the format command (@itemize…) may have an associated formatting command. $formatting_command is this formatting command (like @minus). $formatting_command_formatted is the command formatted in html in case the formatting command is a leading command (like @minus) which should be leading the first preformatted region. \$preformatted_number is a reference on the number of preformatted regions in that format command. The corresponding variable should be increased when a preformatted region is added. $format is the format command. See section Formatting individual table and list items.

If the $format is an enumerate, $item_number is the number of the item in the list, $enumerate_style is the argument of the enumerate, $number is the number or letter corresponding with this item.

7.12.2 Avoiding paragraphs in formats

It is possible to avoid that a format closes the previous paragraph or preformatted region and reopens one, by putting the format command in a hash, %format_in_paragraph with a true value. This only makes sense for few commands since otherwise the nesting of formats and paragraphs could become wrong.

If the value of %no_paragraph_commands associated with a command is true, no paragraph is started by the command if outside of a paragraph (after an empty line, for example). If the value is set to 0, it will start a paragraph. If the value is not set, reasonable defaults are set.

It is also possible to stop a paragraph when an @-command happens by putting the @-command in the %stop_paragraph_command hash associated with a true value.

7.13 Formatting of complex formats (@example, @display…)

Here we see how a whole complex format is formatted. For the formatting of the text, see Formatting (or not) a paragraph and a preformatted region.

The formatting of the complex formats is ultimately controlled by a function, however the default for this function uses a hash and changing the hash values should be enough in most cases. This hash is called %complex_format_map . It has a key for each of the complex format commands (example, smallexample, lisp, smalllisp, display, smalldisplay, format, smallformat).

The associated value is a reference on a hash. The keys are:

begin

The begin should lead to the beginning of the formatted HTML.

end

The end should lead to the end of the formatted HTML.

class

The HTML class. If not defined, the command name.

pre_style

The preformatted style. If not defined the corresponding CSS style is used.

style

If the associated value is code, the format is assumed to be in code style, where with ‘---’, ‘--’, ‘''’ and ‘``’ kept as is. If the key is absent the format inherits the code style and the font from the enclosing context.

The enclosed text will be formatted as described in Formatting (or not) a paragraph and a preformatted region, and the name of the complex format will be available to the function formatting the text.

If you aren’t satisfied with this scheme, you can redefine the following function reference for a better control over the complex format formatting:

Function Reference: $complex_format_text complex_format $format_name $preformatted_text

$format_name is the complex format name, $preformatted_text is the text allready formatted as described in Formatting (or not) a paragraph and a preformatted region. This function returns the whole complex format.

7.14 Customizing the formatting of lists and tables

The formatting of lists and tables is done at two levels:

• At the level of the whole region (table or list),
• At the level of the individual items, rows or cells of the list or table.

7.14.1 Formatting individual table and list items

In texinfo it is possible to give @itemize or table command (hereafter called a format command) a formatting command. For example @minus is the formatting command here:

@table @minus

The default is to apply the command to the text item, however it is possible to avoid it. The hash %special_list_commands has an entry for each of the format command. Each of these entries is a hash reference. If a formatting command is a key of the hash reference, then the formatting command is not applied to the text item for that format command. For example, if we have:

$special_list_commands{'itemize'} = { 'bullet' => '' };

and we have the following @itemize:

@itemize @bullet
@item an item
@end itemize

then @bullet will not be applied to an item.

More control of the text before formatting of the line or the item is achieved with the following function reference:

Function Reference: ( $result_line, $open_command) format_list_item_texi $format $line $prepended $command $number

The $format is the list or table @-command, $line is the item line, $command is the format command, $prepended is set to the text folllowing the format command on the format argument line. The $number is the number of the item, as set in @enumerate. The $result_line replaces the item argument. $open_command is an obsolete return code that can be set to anything.

lists

The items of lists are formatted using the following function reference:

Function Reference: $list_item list_item $text $format $command $formatted_command $item_number $enumerate_style $number $prepended_texi $prepended_formatted

This function formats the text between @item commands. $text is the text corresponding with the item. $format is the type of format, ‘itemize’ or ‘enumerate’. $command is the formatting command given in argument to @itemize, $formatted_command is this command formatted if it is a leading command, like @minus.

If the $format is an enumerate, $item_number is the number of the item in the list, $enumerate_style is the argument of the enumerate, $number is the number or letter corresponding with this item.

If the $format is an itemize, $prepended_texi is the text that appeared on the itemize line, maybe after the formatting command (if any), and $prepended_formatted is the corresponding text, formatted.

two column tables

The two columns tables (@table, @ftable and @vtable), items are formatted using two function references, one for the first line located on the @item line corresponding with the first column, the other for the text appearing on the following lines, corresponding with the second column text.

Function Reference: $table_item table_item $item_text $index_label_text $format $command $command_stack $item_command $formatted_index_entry

This function is used to format the text on the @item line. $item_text is the text line. In case there is an index entry associated with the @item (as with @ftable and @vtable), $index_label_text is the text inserted at the place where an index entry appears. See section Formatting of index entries. $format is the type of format, ‘table’, ‘ftable’ or ‘vtable’. $command is the formatting command given in argument to the table format command. $command_stack is an array with all the @-commands opened, latest on top. $item_command is the item command, ‘@item’ or ‘@itemx’. $formatted_index_entry is the index entry formatted.

Function Reference: $table_line table_line $text

This function is used to format the text on the lines following the @item line. $text is the corresponding text.

multitable

The multitable elements formatting is controlled by the functions associated with two function references. One for a cell, and the other for a row.

Function Reference: $multitable_cell cell $text $item_command \@columnfractions \@prototype_row \@prototype_lengths $column_number

This function is used to format the text of a multitable cell, the text following a @item or a @tab. $text is the corresponding text. $item_command is the command used to introduce the row, such that it is possible to distinguish between @item and @headitem. \@columnfractions is a reference on an array containing the @columnfraction arguments, if any, and \@prototype_row is a reference on an array containing the row prototypes given on the @multitable line, if any. \@prototype_lengths array contains the lengths of the row prototypes formatted. $column_number is the maximal number of columns.

Function Reference: $multitable_row row $text $item_command \@columnfractions \@prototype_row \@prototype_lengths $column_number

This function is used to format a multitable row. $text is the row text, with cells allready formatted with the $cell

function reference. $item_command, \@columnfractions, \@prototype_row, \@prototype_lengths and $column_number are the same than in the function reference above.

In the default case, this function is interlinked with $begin_format_texi (see section Customizing format opening) and @multitable formatting since a stack of possible nested multitables is kept to know the cell number.

7.14.2 Formatting of a whole table or list

If the Texinfo command is a key of the %format_map , the associated value is used to specify the formatting of the construct, otherwise a function is called. The value in %format_map associated with a command is interpreted similarly with values associated with more simpler commands:

• If the text is a word, it is considered to be an HTML element name, and the whole table or list is enclosed between the element opening and the element closing.
• If the text is a word followed by some text, the word and is interpreted as above, and the text is considered to be the attributes text of the element.
• If the text is empty nothing is added to the text.

In case the %format_map isn’t used, a function reference called $table_list should be redefined, the associated function will be called each time a command isn’t found in %format_map.

Function Reference: $whole_table_list table_list $format_command $text $command $formatted_command $item_nr $enumerate_style $prepended_texi $prepended_formatted \@columnfractions \@prototype_row \@prototype_lengths $column_number

$format_command is the Texinfo command name, $text is the formatted items. $command is the format command given in argument to the format command, $formatted_command is the same, but formatted. $prepended_texi is the remaining text on the format command line, $prepended_formatted is the same, but formatted. Only relevant in @enumerate, $item_nr is the item number, and $enumerate_style is the @enumerate style. Only relevant in @multitable \@columnfractions is a reference on an array containing the @columnfraction arguments, if any, \@prototype_row is a reference on an array containing the row prototypes given on the @multitable line, if any, \@prototype_lengths array contains the lengths of the row prototypes formatted and $column_number is the maximal number of columns.

If you still want to use %format_map but differently from the default, it is possible to redefine the following function reference:

Function Reference: $whole_table_list format $command $format $text

$command is the @-command, $format is the entry associated with $command in %format_map. $text is the formatted items.

7.15 Definition commands formatting

The formatting of definition commands is controlled by a main hash, 3 strings and another hash, and and five functions. The mainhash describes how the text on the definition line is interpreted, the functions control the formatting of the definition line and the definition function text.

7.15.1 Customizing the interpretation of a definition line

The keys of the hash %def_map are definition command names. There are two types of entries:

• If the command is a shortcut for another definition command the value is a text and the definition command is replaced by the text.

  For example if we have:

  $def_map{'deftruc'} = '@defvr {A truc}';
  

  and a line like

  @deftruc var
  

  the line will be transformed in

  @defvr {A truc} var
  

• If the command isn’t a shortcut, it is associated with an array reference. The first element is ‘f’, ‘v’ or ‘t’ corresponding with the index type (‘f’ for function, ‘v’ for variable, ‘t’ for type).

  The remaining of the array describes how to interpret the text following the definition command on the definition command line. The entry item specify what corresponds with the next bracketed item or word. Currently the possibilities are ‘category’, ‘name’, ‘type’, ‘class’, ‘arg’ and ‘argtype’. ‘arg’ means that the arguments are not mixed with type definitions, with ‘argtype’ types are mixed with definitions. When there is no ‘arg’ nor ‘argtype’ it is the same than ‘argtype’ (like makeinfo).

  For example if we have

  def_map{'defvr'} = [ 'v', 'category', 'name' ];
  

  The first bracketed item following @defvr is considered to be the category and the next one is the name. The index associated with the definition line is the variables index.

Some characters are special with regard with definition parsing, they are delimiters, the can have a role in definition argument determination, and also hae a special meaning in arguments parsing. This is not very well documented in the texinfo manual, so it is subject to change. Strings allow to determine the delimiters:

$def_argument_separator_delimiters

Characters that separate arguments, currently ()[],.

$def_always_delimiters

Character that are always delimiters, if they appear in a type or a parameter, ()[].

$def_in_type_delimiters

Character that are considered as delimiters only if in a type. In a parameter they are part of the parameter.

7.15.2 Customization of the definition formatting

Five functions are used when formatting a definition command:

category name

Function Reference: $category definition_category $category $class $style $command

This function precise a category name associating a class $class (if given) with $category. The $style of the definition may be ‘f’, for function, ‘v’, for variable or ‘t’, for type. The $command is the definition @-command.

Function Reference: $entry definition_index_entry $name $class $style $command

This function precise a name associating a class $class (if given) with $name. This is used to do an index enntry associated with th edefinition command. The $style of the definition may be ‘f’, for function, ‘v’, for variable or ‘t’, for type. The $command is the definition @-command.

formatting of the definition line

Function Reference: $line def_line $class_category_class $name $type $arguments $index_label \@arguments_array \@arguments_type_array \@unformatted_arguments_array $command $class_name $category $class $style $original_command

This function formats the definition line. $class_category is the category formatted with $definition_category , $name, $type and arguments are the element of the definition line. $index_label is the text inserted at the place where an index entry appears. See section Formatting of index entries. \@arguments_array is an array holding the definition arguments, formatted. \@arguments_type_array holds the type of the definition arguments, like ‘name’, ‘type’ and similar arguments, ‘paramtype’. ‘delimiter’ and ‘param’. \@unformatted_arguments_array holds the arguments without @-command substitution. $command is the definition command, after substitution. $class_name is the class applied on name, formatted as specified in definition_index_entry. $category and $class are the corresponding arguments. $style corresponds with the index style, as explained above. $original_command is the unmodified definition @-command.

definition text

Function Reference: $definition_text def_item $text

This function formats the definition text, $text.

the whole definition

Function Reference: $definition def $text

This function formats the whole definition. The definition line and text formatted by the above functions are in $text.

7.16 Customizing headings formatting

A function controls the formatting of sectioning element headings, with the corresponding function reference:

Function Reference: $heading_text heading \%element_reference

The \%element_reference is a reference on a hash corresponding with the sectioning element. The following keys are of interest:

text

The heading text

text_nonumber

The heading text without section number

node

true if the sectioning element is a node without associated structuring command

level

The level of the element in the document tree. ‘0’ is for @top, ‘1’ for @chapter and so on

tag_level

the sectioning element name, with @raisesections and @lowersections taken into account

top

true if it is the top element

It is also possible to customize the heading text with section number with the following function reference (called for headings and nodes):

Function Reference: $result_texi heading_texi $heading_command $heading $number

$heading_command is the sectioning @-command of that heading. $heading is the texinfo for that heading. $number is the heading number classicaly computed with dots between numbers, and letters for top level appendix numbering. This function should return the texinfo text corresponding with the numbered heading.

The label associated with the heading that can appear before the heading itself and even before the navigation panel is customized with the following function reference:

Function Reference: $element_label element_label $identifier \%element_reference $command $unformatted_line

$identifier is the identifier associated with the heading. \%element_reference is the same as above. $command is the @-command appearing on the line, and $unformatted_line is the line, unformatted.

Additionally, for @node and sectionning @-commands the formatting of the label, navigation panel and heading is controlled by:

Function Reference: $element_heagin_text element_heading \%element_reference $command $command_texi_arg $formatted_arg $in_preformatted $one_section $element_heading $first_in_page $is_top $previous_is_top $unformatted_line $element_id $new_element

\%element_reference is the same as above. $command is the heading @-command. $command_texi_arg is the argument of the @-command, unformatted. $formatted_arg is is the argument of the @-command, formatted. $in_preformatted is true if in preformatted environment. $one_section is true if there is only one section. $first_in_page is true if this is the first heading in a page. $is_top is true if the heading is considered as a top element heading. $previous_is_top is true if the previous helement was a top element. $unformatted_line holds the whole line, unformatted. $element_id is the id of the heading. $new_element is true if the heading is the first of an element block.

7.17 Formatting of special regions (@verbatim, @cartouche, @quotation)

Regions corresponding with raw text, like @verbatim, @html, @tex or the content of the file given in @verbatiminclude argument are formatted according to the following function reference:

Function Reference: $raw_region raw $command $text

$command is the command name, $text is the raw text. In the default case, if $command is verbatiminclude the text is the content of the @verbatiminclude file argument.

If LaTeX2HTML is used, @tex regions are handled differently, (see section Bypassing normal formatting).

The @cartouche command formatting is controlled by the function reference:

Function Reference: $cartouche cartouche $text

$text is the text appearing within the cartouche.

The formatting of @quotation and @smallquotation is controlled by two function references. The first one is usefull in case the @quotation has an argument, as it allows to prepend a string to the quotation text:

Function Reference: $prepended_string quotation_prepend_text $command $text

$command is the @-command. $text is the argument of the quotation with @-commands not interpreted. This function can return a string which will be prepended to the quotation text.

The whole quotation is formatted by:

Function Reference: $quotation quotation $command $quotation_text $argument_text $argument_text_texi

$command is the @-command. $quotation_text is the quotation text, formatted, with the text prepended by the function above. $argument_text is the argument of the @quotation, formatted. $argument_text_texi is the argument of the @quotation, simply formatted.

7.18 Menu formatting

There are two possibilities for menu formatting:

• format the whole menu in a preformatted environment, like in Formatting of complex formats (@example, @display…);
• format the menu in table with more specialized formatting for each part of the menu;

The simple formatting in a preformatted is used if $SIMPLE_MENU is true, otherwise the format with tables is used (this is the default).

If $USE_ACCESSKEY is set, the accesskey attribute is used in anchors. In that case the %BUTTONS_ACCESSKEY

hash is used for the access key.

To understand how the formatting of menus is controlled, the different parts of a menu are first described, then how to control the formatting of each of these parts, for each possible formatting.

7.18.1 The structure of a menu

In texi2html, a menu is considered to be composed of 2 parts, the menu entries and the menu comments. Menu entries are further divided in an entry link and optionnaly an entry description. The entry link consists in a node name and an optionnal menu entry name.

A menu entry begins with ‘*’ at the beginning of the line. It begins with the entry link, followed by the description. The description spans until the next menu entry, or an empty line not contained within a command block which begun in the description. An empty line or starts a menu comment, which spans until the next menu entry.

Here is an illustration of these rules:

@menu
* entry name: node name.        description begins
   description continues
* another menu entry::
   description begins
                    description continues

   A menu comment, after an empty line

* node::                        description begins
still in description.

* last entry::         description begins @emph{text

of the description, even if there is an empty line,
because we are in @emph}.
@end menu

7.18.2 The formatting of the different menu components

If in a preformatted context (and $SIMPLE_MENU isn’t set), the menu link and description are put in the same preformatted environment. This can be avoided with $SEPARATE_DESCRIPTION .

Two function references are associated with the formatting of the different parts of a menu:

Function Reference: $link menu_link $section \%state $href $node $name $ending $has_name $command_stack $preformatted

$section is the section name corresponding with the link, $href is the link hypertextual reference. $href may be absent. \%state holds informations about the current context. $node is the node name, $name is the name of the node. $ending is the text ending the link entry, in general ‘::’ followed by some spaces. $has_name is true if the entry has an explicit name, otherwise $name has been constructed using the formatted node name. $command_stack is an array containing the commands enclosing the menu link. It is used in the default case to detect if the menu link is right in the @menu or not, since if it is not right below the menu the formatting is simpler. $preformatted is true if in preformatted context. See section Three contexts for expansions: preformatted, normal and string.

This command is not called if $SIMPLE_MENU is set.

Function Reference: $description menu_description $description_text \%state $element_text

$description_text is the text of the menu description. The formatted link is also here if in preformatted context and $SEPARATE_DESCRIPTION is not set. \%state should be used similarly than for the menu link. $element_text is the heading of the element associated with the node. $command_stack and $preformatted are the same than for the menu link.

The menu comment part is formatted like a normal command, called menu_comment. It is only used if not in preformatted environment and if just below a @menu since otherwise one cannot tell if it is a menu commment or normal text. The default is to have it be formatted like a Formatting of complex formats (@example, @display…), with

$complex_format_map{'menu_comment'} =
{
   'begin' => "<tr><th colspan=\"3\" align=\"left\" valign=\"top\">",
   'end' => "</th></tr>", 'class' => 'menu-comment',
}

7.18.3 Simple menu formatting in a preformatted environment

If the menu is to be formatted in a single preformatted environment, an entry for ‘menu’ and ‘detailmenu’ should be added to the %complex_format_map hash (see section Formatting of complex formats (@example, @display…)). In the default case, if the user didn’t add an entry himself, a very simple entry is used, with:

$complex_format_map->{'menu'} = { 'begin' => '' , 'end' => '',
    'class' => 'menu-preformatted' };

7.18.4 The formatting of the menu in a table

In the default case, the name of the section corresponding with the node is used instead of the node name. If $NODE_NAME_IN_MENU is true, however, node names are used. If $AVOID_MENU_REDUNDANCY

is true and menu entry equal menu description the description isn’t printed. This is the default. Likewise, if node or section name equal entry name, do not print entry name.

A symbol, $MENU_SYMBOL is put at the beginning of menu entries when the node name is used. The default is ‘&bull;’. If $UNNUMBERED_SYMBOL_IN_MENU is true it is also put at the beginning of unnumbered section names. This is not done by default.

The menu comments are considered to be preformatted text. The style associated with this preformatted text is determined by $MENU_PRE_STYLE . Default is ‘font-family: serif’. The entry similar with an entry in %complex_format_map (see section Formatting of complex formats (@example, @display…)) used when the menu appears in a preformatted enviroment is in $MENU_PRE_COMPLEX_FORMAT , and, in the default case is:

$MENU_PRE_COMPLEX_FORMAT = {
              'pre_style' => $MENU_PRE_STYLE, 
              'class' => 'menu-preformatted'
   };

The css class associated with menu comments is menu-comments.

The following function reference controls the formatting of a wole menu or a detailmenu in that case:

Function Reference: $menu menu_command $command $menu_components_text

$command is the menu command, currently ‘menu’, ‘detailmenu’ or ‘direntry’. $menu_components_text is the formatted menu components text, obtained as explained above.

7.19 Indices formatting

Two different things needs to be handled for indices formatting, the place where the index term appears, the index entry, and the index list itself. The indexing commands like @cindex determines where index entries appear, and the index list is printed with a @printindex command.

7.19.1 Formatting of index entries

Index entry places in the main text may be the target for hypertext references. Their formatting is controlled by the function associated with the following function reference:

Function Reference: $target index_entry_label $identifier $preformatted $entry $index_name $index_command $texi_entry $formatted_entry

$identifier should be used to create a target for links (typically associated with a name or id attribute in HTML). $preformatted is true if the index entry appeared in preformatted text. $entry is the index entry with all the @-commands removed. $index_name is the index name, $command is the index command which may be a index command like @cindex, but also a definition or a table. $texi_entry is th eindex entry with @-commands, and $formatted_entry the entry formatted.

Regular index entries are (like @cindex) are formatted using the following function reference:

Function Reference: $index_entry index_entry_command $command $index_name $label $entry_texi $entry_formatted

$command, $index_name, $entry_texi and $entry_formatted are the same as above, and $label is what could be used as a label, formatted using the function above.

7.19.2 Customizing the formatting of index lists

There is an elaborate default index formatting in texi2html, with index summary by letter linking to index entries grouped by letters too, with the possibility of index pages split accross files. This system may be completly bypassed by redefining the function reference that is called when @printindex is encountered:

Function Reference: $index_text printindex $index_name

$index_name is the index name appearing on the @printindex line. The index formatted should be returned by this function reference.

If the default index formatting is used, there are still possibilities to customize part of the formatting. The index entries are sorted alphabetically. A whole index list is considered to be composed of letter entries. A letter entry is composed by all the index entries beginning with that letter. A letter may be a non alphabetical character, but we call it letter here.

An index summary appears at the beginning and at the end of an index list, and should be used to jump directly to a letter entry. Indices lists may be split across pages, thus the different letters may appear on different files. The number of index entries appearing on each page is determined by a variable $SPLIT_INDEX if set. The default is to split indices after 100 entries.

The formatting of all these elements is controlled by the following function references:

formatting of a letter in a summary

Function Reference: $letter summary_letter $letter $file $identifier

This function is used to format a letter appearing in a summary, refering to a letter entry in the index list. $letter is the letter. $file is the file name where the letter entry appears. More precisely, it is empty when the letter entry is on the same page than the summary, it contains the file name when the index page is split accross page. $identifier is an identifier for the target letter entry.

formatting of a summary

Function Reference: $summary index_summary \@alphabetical_letters \@nonalphabetical_letters

\@alphabetical_letters and \@nonalphabetical_letters contain the formatted summary letters, formatted with the above function.

formatting of an index entry

Function Reference: $entry index_entry $entry_href $entry_text $section_href $section_heading

$entry_href is a reference to the place where the index entry appeared, $entry_text is the corresponding text. $section_href is a reference to the beginning of the sectioning element containing the index entry, $section_heading is the heading of the element.

formatting of letter entry

Function Reference: $letter_entry index_letter $letter $identifier $index_entries_text

This function formats a letter entry, consisting in all the index entries beginning with this letter. $letter is the letter, $identifier should be used to create a target for links (typically links from summaries), and $index_entries_text is the text of the index entries formatted as described above.

formatting of whole index

Function Reference: $index print_index $index_text $index_name

$index_text is the text of all the index entries grouped by letter appearing in that page formatted as above. It is undef if there are no entries or theindex name isn’t known. index_name is the name of the index, the argument of @printindex.

7.20 Floats and lists of floats

Floats appear in the @float environment, optionaly with a style and a label, and with optionnal @caption and @shortcaption. Their list appear after a @listoffloats.

A hash reference is associated with each float, it is available in some formatting functions. The keys are:

caption_texi

shortcaption_texi

A reference on an array containing the caption or shortcaption lines, with texi @-commands.

style_texi

The style with texi @-commands.

style_id

The unique identifier associated with the style.

style

The style formatted.

nr

The number with the same conventions than makeinfo (use the chapter number a dot and then the number of the float of that style in the chapter, or an absolute number if in unnumbered).

chapter_nr

The number of the chapter containing the float.

nr_in_chapter

The number of the float in the chapter.

absolut_nr

The number of the float in the document.

texi

The label with @-commands.

name

The label formatted.

id

The unique identifier associated with the label. Usefull to make an anchor or a reference.

target

The target that can be used to refer to that float.

element

A reference on a structure representing the element the float appear in.

7.20.1 Formatting a float

First there is an occasion to construct a texinfo text for the caption, using the caption texinfo lines and the informations in the float structure. The returned lines will be formatted in the main program. A function reference is used here:

Function Reference: (\@caption_lines_returned, \@shortcaption_lines_returned) caption_shortcaption \%float \@caption_lines \@shortcaption_lines

\%float is the structure defined above. \@caption_lines and \@shortcaption_lines are references on arrays containing the texinfo lines for caption and short caption. \@caption_lines_returned and \@shortcaption_lines_returned are references on an array containing the texinfo lines for the caption and shortcaption.

Then the float is formatted with the following function reference:

Function Reference: $text float $float_text \%float $caption_text $shortcaption_text

$float_text is the text appearing within the @float, formatted. \%float is still the structure defined above. $caption_text and $shortcaption_text are the caption and short caption build with the above function and formatted.

It is also possible to do something when a caption or a shortcaption appear with t hefollowing function reference:

Function Reference: $text caption_shortcaption_command $command $formatted_caption \@texi_lines \%float

$command is the @-command, ‘caption’ or ‘shortcaption’. $formatted_caption is the caption text, formatted, while \@texi_lines is a reference on an array containing the caption lines, this time without any formatting. \%float is still the structure defined above.

In the default case this function reference returns an empty string.

7.20.2 Formatting lists of floats

A list of floats is introduced by @listoffloats. The argument of @listoffloats is the style. First the style texinfo can be modified with the following function reference:

Function Reference: $style_texi_returned listoffloats_style $style_texi

$style_texi is the @listoffloats argument with texinfo @-commands kept. It is possible to make changes to the $style_texi and return a modified string, still with @-commands. The modified string is formatted in the main program.

After that, for each of the floats with that style, first there is a possibility to modify the float style and the float caption before they are formatted in the main program, with the following function references:

Function Reference: $float_style_texi_returned listoffloats_float_style $style_texi \%float

$style_texi is the style, and \%float is the structure described above. This function reference returns a style to be formatted in the main program.

Function Reference: ( \@caption_texi_returned, $caption_or_shortcaption) listoffloats_caption \%float

\%float is the structure described above. This function reference returns a caption to be formatted in the main program, \@caption_texi_returned, and a string, $caption_or_shortcaption that is either ‘caption’ or ‘shortcaption’ that can be used by the main program if this information is needed.

Each entry is formatted by:

Function Reference: $listoffloats_entry listoffloats_entry $style_texi \%float $float_style $caption $href

$style_texi is the style with @-commands, $float_style is the style returned by the above function and formatted. $caption is the caption returned by the above function formatted. \%float is the structure corresponding with the float, and $href is an href pointing to the float location.

Lastly, the whole @listoffloats is formatted by:

Function Reference: $listoffloats listoffloats $style_texi $style \@listoffloats_entries

$style_texi is the style with @-commands, $style is the style returned by the above function and formatted. The array reference \@listoffloats_entries holds the entries formatted by the above function.

7.21 Customizing the footnotes formatting

Each footnote is associated with a footnote entry. Several footnote entries are grouped in a footnote section. When a footnote appears, two things must be formatted: in the main text the place where the footnote appear and the footnote text.

Two functions, with corresponding function references control the formatting of the footnotes:

Function Reference: (\@lines $text_for_document) foot_line_and_ref $number_in_doc $number_in_page $footnote_id $place_id $document_file $footnote_file \@lines \%state

$number_in_doc is the footnote number in the whole document, $number_in_page is the footnote number in the current page. $footnote_id is an identifier for the footnote in the footnote text which should be used to make target for references to that footnote, while $place_id is an identifier for the location of the footnote in the main document. Similarly, $document_file is the file name of the file containing the text where the footnote appears in the main document, while $footnote_file is the file name of the file where the footnote text appears.

\@lines is a reference on an array containing the footnote text lines, allready formatted. And \%state holds informations about the context at the footnote place in the main document. As usual the most usefull entry is preformatted which is true if the footnote appears in a preformatted context.

This function returns a reference on an array, \@lines containing the updated footnote text for the footnote entry, and $text_for_document, the text appearing at the footnote place in the main document, linking to the footnote entry.

The following function is only used when footnotes are at the bottom of a page and the document is split. For customization of the footnotes page in case they are on a separated page or section, Customizing the layout of the special pages. For the determination of the footnote locations, Page layout related command line options.

Function Reference: foot_section \@footnotes_lines

This function formats a group of footnotes. \@footnotes_lines is a reference on an array holding the lines of all the footnote entries formatted as explained above. This function modifies the reference.

7.22 Customizing format opening

The following function reference is called when a format is opened. A format is any @-command that ends with a @end except @-commands that only select if the input is processed (like @ignore or @ifhtml) or raw @-commands (like @verbatim and @html).

Function Reference: $line begin_format_texi $command $line \%state

The $command is the format command, the $line is the line following the @-command, \%state is a reference on a hash containing many formatting information. It can modify the line and return something else.

In the default case, it is used to keep track of the multitable nesting. As a consequence, it is linked with the multitable formating. See Multitable formatting.

7.23 Bypassing normal formatting

It is possible to bypass completely the normal formatting of @-commands with braces and raw regions (@html, @tex, @xml… regions). In that case the @-commands and the text within are passed to a user defined function early, in a pass when no expansion of output takes place, called the collecting pass. Another user defined function is called during the output expansion phase.

Moreover, arbitrary user defined functions may be called between the different texinfo parsing and outputting passes. This could be used, for example to initialize some things before collecting the @-commands and their text, expanding them between the collecting and expansion phase and doing some cleaning after the expansion pass. These possibilities are used for the interface to LaTeX2HTML (see section Expanding @tex and @math regions using LaTeX2HTML), and the examples are taken from that use.

The @-commands that are keys of the %command_handler hash are collected in the collecting pass and expanded in the expansion pass using user defined functions. The associated value is a reference on a hash used to specify the user defined function references. The key of the hash reference are 'init' for the function reference called during the collecting pass, and 'expand' during the expansion pass. Here is an example for an @-command with braces:

$command_handler{'math'} =
     { 'init' => \&Texi2HTML::LaTeX2HTML::to_latex,
       'expand' => \&Texi2HTML::LaTeX2HTML::do_tex
     };

And an example for a raw region @-command:

$command_handler{'tex'} =
     { 'init' => \&Texi2HTML::LaTeX2HTML::to_latex,
       'expand' => \&Texi2HTML::LaTeX2HTML::do_tex
     };

The function references are called like:

Function Reference: $status $command_handler{'$command'}->{'init'} $command $text $count

$command is the @-command name, $text is the text appearing within the @-command. $count is a counter counting how many times this @-command appeared. $status is a boolean which should be true if the collecting was succesfull. If false the @-command and the text is discarded.

Function Reference: $result $command_handler{'$command'}->{'expand'} $command $count $state $text

$command is the @-command name, $count is a counter counting how many times this @-command appeared. $state is a reference on a hash containing many informations about the context. $text should be empty. $result is the expanded resulting text.

There are five places for user defined functions, associated with arrays:

@command_handler_setup

The function references in that array are called before anything is done, including collecting the output file names. The input file names directory are available.

@command_handler_init

The function references in that array are as soon as the file names are known. It may be at different moments, before processing anything, right after @sefilename, or at the end of the first pass (after @macro and @include expansions). At that time the information available is essentially the file names.

@command_handler_names

The function references in that array are called right after the collecting pass. At that time all the special @-commands have been collected as explained above but no output has been produced, the element (node and section) names hasn’t been processed and expanded.

@command_handler_process

The function references in that array are called after the element names have been processed, but before the main output initialization.

@command_handler_output

The function references in that array are called rigth before the main output processing, so that more informations are available, like the title.

@command_handler_finish

he function references in that array are called after the end of the output generation.

Here is an example of these arrays use:

push @command_handler_init, \&Texi2HTML::LaTeX2HTML::init;
push @command_handler_process, \&Texi2HTML::LaTeX2HTML::latex2html;
push @command_handler_finish, \&Texi2HTML::LaTeX2HTML::finish;

7.24 Handling special regions

Special regions @titlepage, @documentdescription and @copying are removed from the document before the last pass in the default case. They can be kept if the value associated with the @-command in the %region_formats_kept hash is true.

The @insertcopying @-command is formatted by

Function Reference: $insertcopying insertcopying $text $comment $simple_text

$text is the text appearing in @copying, formatted. $comment is the text with texi removed, should be very simple text. $simple_text is the text formatted in string context.

The title page handling is described in Formatting of title page.

7.25 Customizing other commands, and unknown commands

Many commands without braces are available in texinfo, sometimes with a specific syntax. For example we have @sp, @noindent, @documentlanguage, @oddheading, @headings, @shortcontents, @shorttitlepage or @comment. texi2html interprets some of these commands and some functions or variables are used for their formatting or to access their information. In the default case, however, most of these constructs are ignored.

It is possible to change how the things following these commands on the line are handled, what is considered to be an arg for those commands and it is also possible to keep them instead of discarding them such that it is possible to handle them specially, with the same function than the one used for unknown commands.

Those special commands without braces are the key of a hash: %misc_command . The associated value is a reference on a hash enabling to set the properties of these commands. The keys of this hash reference is the name of a property, the value is the value of the property. For example here we have line for the arg property for the command @-command.

$misc_command{'command'} = {'arg' => 'line', 'skip' => 'space'};

The properties and possible values are:

skip

This property enables to set what is skipped after the command arguments. Here are the possible values:

line

The remaining of the line is skipped.

space

Spaces are skipped but not newline.

whitespace

Spaces are skipped

linewhitespace

Spaces are skipped if there are only spaces remaining on the line.

linespace

Spaces are skipped, but not newline if there are only spaces remaining on the line

arg

If the associated value is line the line is considered to be the argument. If it is a number it is the number of args (separated by spaces).

keep

If true the args and the macro are kept, otherwise they are discarded. The defaut is to have keep undef for all the commands. If keep is true for @verbatiminclude the default action for this macro isn’t done.

Commands which don’t appear in the hashes %simple_map , %simple_map_pre , %simple_map_texi and %misc_command, or that appear in %misc_command but with keep true are processed by the following function reference:

Function Reference: ($result_line, $result, $result_text, $message) unknown $command $line $pass

$command is the @-command, $line is the line following the $command. $pass is the pass of texi2html (see section Three passes: macro expansion, document structure and output). $result is a boolean. If it is true then the other return values are taken into account otherwise the default actions are used. In case $result is true, $result_line is the new line to be processed further, $result_text is the resulting formatted text and $message, if defined is a message outputted to the output with line number added by texi2html.

Commands with braces not specified above nor in %style_map , %style_map_pre and %style_map_texi are processed by the following function reference

Function Reference: ($result, $result_text, $message) unknown_style $command $text

$command is the @-command, $text is the text appearing within the braces (allready formatted). $result is a boolean. If it is true then the other return values are taken into account otherwise the default actions are used. In case $result is true, $result_text is the resulting formatted text and $message, if defined is a message outputted to the output with line number added by texi2html.

Appendix A Internationalization

The strings written in the document are selected based on the document language. This can be used to customize the strings, as described in Customizing strings written by texi2html. This also enables translation of the strings.

The gdt subroutine is used for translated strings:

Function: gdt ($string, \%variables_hash, \%state)

with $string the string to be translated, \%variables_hash a reference on a hash holding the variable parts of the translated string, and \%state a hash reference determining the context of expansion (use the document state, expansion in string, no expansion...).

Translated strings are texinfo strings, which may have @-commands. In translated strings, the variables parts of the string are not denoted by %s and the like, but by ‘{arg_name}’ (which is common for perl gettext). For example, in the following, ‘{section}’ will be replaced by the section name:

see {section}

This is for 2 reasons, first changing the order of printf arguments is only available since perl 5.8.0, second the order of the argument may not be predictable when commands expansion may lead to different orders depending on the output format.

The expansion of those strings happens that way:

1. First the string is translated. The locale is @documentlanguage.@documentencoding.

   If the @documentlanguage is like ll_CC, ll_CC is tried first, and then ll. If the encoding is not us-ascii, us-ascii is also tried. The idea is that if there is a us-ascii encoding, it means that all the characters in the charset may be expressed as @-commands. For example there is a fr.us-ascii locale that can accomodate any encoding, since all the latin1 characters have associated @-commands. For the ja translations, there is only ja.utf-8 since there are no @-commands for ja letters.

2. Next the args in string are protected, for example ‘{arg_name}’ becomes ‘@internal_translation_open_brace{}arg_name@internal_translation_close_brace{}’ (this part is skipped if there is no expansion).
3. Next the string is expanded as a texinfo string. ‘@internal_translation_open_brace{}’ expands as { and ‘@internal_translation_close_brace{}’ expands as }, such that in the end one still gets ‘{arg_name}’ within an expanded string (this part is skipped if there is no expansion).
4. Then the in string arguments are substituted, for example ‘{arg_name}’ is substituted by the corresponding argument.

For example, in the following ‘{date}’, ‘{program_homepage}’ and ‘{program}’ are the argument of the string. Since they are used in @uref, their order in not predictable. The ‘{'duplicate'=>1}’ means the the document state should be used when expanding the string. ‘{date}’, ‘{program_homepage}’ and ‘{program}’ are substituted after the expansion, which means that they should already be acceptable output:

gdt('This document was generated on @i{{date}} using @uref{{program_homepage}, @i{{program}}}.', {
           'date' => $date, 'program_homepage' => $Texi2HTML::THISDOC{'program_homepage'}, 'program' => $Texi2HTML::THISDOC{'program'} },{'duplicate'=>1});

This approach is a bit complicated, however what is interesting is that it allows to have translation available in different encodings for charset that are covered by @-commands, and also to specify how the formatting for some commands is done independently of the output format but still allow it to be language dependent. For example, the ‘@pxref’ string may be:

see {node_file_href} section `{section}\' in @cite{{book}}

which allows to specify a string independently of the output format but with a rich formatting that may be differently translated in other languages.

It is also possible to use more regular %s escapes, and also avoid any expansion (with ’keep_texi’ in the state).

B Incompatibilities with previous versions

texi2html has accumulated a lot of incompatibilities with previous versions. They are documented in the ‘NEWS’ file, we discuss them here nevertheless. Most of the incompatibilities were introduced in version 1.68. API changed also a lot between 1.78 and 1.80. Between 1.82 and 5.0 API changed also quite a bit, and more importantly, the whole internationalization handling was changed.

• API changes between 1.82 and 5.0:
  • complex format handling API changed. $complex_format_map is replaced by a real hash %complex_format_map and the begin and end are no more evaluated.
  • In the default case a gettext-like framework is now used for in document string translations. The old framework is still available by setting $I18N_PERL_HASH.
  • Information that can be set by @-commands should now be accessed through get_conf.
  • $format_list_item_texi should now modify directly the @item argument and apply the @-command given as @table argument. Correspondingly, some arguments of $table_item have been removed.
  • Formatting of titlepage changed, and @title, @subtitle and @author are now handled by $line_command
  • $SEPARATED_FOOTNOTES has been replaced by $FOOTNOTESTYLE, that takes the same argument than @footnotestyle.
  • replace $LANG by $DOCUMENTLANGUAGE.
  • replace $DO_CONTENTS by $CONTENTS and $DO_SCONTENTS by $SHORTCONTENTS.
  • replace $TRANSLITERATE_NODE by $TRANSLITERATE_FILE_NAMES.
  • replace $SECTION_NAVIGATION by $HEADERS.
  • @verbatiminclude is now a $raw command argument distinct from @verbatim.
  • @direntry is formatted like a menu and passed down to $menu_command.
  • add $NODE_FILENAMES to use node names for files associated with nodes, such that $NODE_FILES is only means creating redirection files.
  • Remove the user from footer, and the corresponding USER variable and $Texi2HTML::THISDOC{user}.
  • Don’t treat _setfilename as a value set. It should be available through get_conf. Same for all the other @-commands previously avaiable as values.
  • change in class attribute for html elements: for ul to class "no-bullet" ($NO_BULLET_LIST_ATTRIBUTE). @top and @centerchap have now their command name as class. @node heading class is now "node-heading", instead of "node". The whole header is in an element with class "header" (was "node" when !$HEADER_IN_TABLE).
  • print_element_header is obsoleted, things should be directly done in element_heading.
  • Add ’inline_attribute’ for an XML attribute that should be closed at paragraph end and reopened at paragraph begin. Consequently the closing and opening of ’attribute’ isn’t done anymore in paragraph and preformatted.
  • USE_ISO now only determines if '', `` --- and -- are transformed into entities. t2h_default_set_iso_symbols sets or unsets USE_ISO, %things_map/%pre_map/%simple_format_simple_map_texi, $OPEN_QUOTE_SYMBOL and $CLOSE_QUOTE_SYMBOL. %iso_symbols is unused now.
  • change in the normal_text function reference API, now there is an input variable true if in @math.
  • change the external_ref API, fourth argument is now the file name and add the node as an extra argument.
  • rename $Texi2HTML::THISDOC{css_lines} as $Texi2HTML::THISDOC{css_rule_lines}
• API changes between 1.78 and 1.80:
  • what was done in the beginning of the $print_section function reference is now done in $print_element_header.
  • there is a new argument for $normal_text.
  • there is a new argument for $menu_link and $simple_menu_link, and the name argument is always set, even if there is no explicit name. The new argument is true if there really was an explicit name.
  • $internal_ref and $external_ref don’t change inforef to xref anymore.
  • in $table_item the text is not formatted with the format command, the text_formatted argument is.
  • definition index entries are now formatted with $definition_index_entry, not with $definition_category anymore.
  • $printindex is called if an index appears, even if the index is not defined or there are no index entries.
  • new argument (@-command name) for $quotation and $quotation_prepend_text.
  • change in the heading API. THIS_HEADER is not used anymore. New function reference, element_heading to format a node or a section heading, including navigation and label. Accordingly, print_Top and one_section don’t print the element header anymore. Also it is reported if the element is a new element, is the main element and more arguments, and top element heading is always done in heading.
  • print_element_header and print_navigation now return their result.
  • the copying key of %Texi2HTML::THISDOC is now called copying_comment.
  • TOC_LIST_ATTRIBUTE is now called NO_BULLET_LIST_ATTRIBUTE.
  • TOC_LIST_STYLE is now called NO_BULLET_LIST_STYLE.
  • the $unknown function reference has a new argument, the pass number.
  • ‘--sec-nav’ is replaced by ‘--headers’.
  • ‘--Verbose’ is replaced by ‘--verbose’.
  • ‘--lang’ is replaced by ‘--document-language’.
  • ‘--separated-footnotes’ is replaced by ‘--footnote-style’.
  • ‘--lang’ is replaced by ‘--document-language’.
  • &$menu_comment is removed, menu_comment is now handled like an @-command.
  • @detailmenu is now formatted more like @menu, and the &$menu function reference is replaced by &$menu_command. &$menu is kept for backward compatibility. If &$menu is defined, @detailmenu is ignored.
  • the API for the formatting of menus completly changed. $simple_menu_link is removed, everything should be done in $menu_link.
  • image API changed, and is unstable, so not documented.
  • image file paths are not completed anymore in the default case. The previous behaviour can be restored with $COMPLETE_IMAGE_PATHS set to true.
  • in %misc_command ‘texi’ is not used anymore. The value and macros are expanded as they should be unconditionnally.
  • there is a new ’style’ key in $complex_format_map, to be able to differentiate complex formats inheriting fonts and code_style (like @format, @display).
  • $EXTENSION should be undef if one doesn’t want an extension to be added.
  • THISDOC{’title’} and similar are now for @title since only one @title should appear in the document. @settitle is tried first to set fulltitle.
  • Configuration variables are modified anymore, instead the variable value is put in $Texi2HTML::THISDOC{’VAR’}. This is the case for DO_CONTENTS, DO_SCONTENTS, CSS_LINES, BODYTEXT, DOCUMENT_DESCRIPTION, DOCUMENT_ENCODING, IN_ENCODING, ENCODING_NAME, OUT_ENCODING.

    For example, if $CSS_LINES is defined, the value is put in $Texi2HTML::THISDOC{’CSS_LINES’} which is used for formatting, and if $CSS_LINES is not defined, $Texi2HTML::THISDOC{’CSS_LINES’} is autodetected.

  • When there is no section and $USE_NODES is not set don’t split by node. This behaviour and the previous aren’t documented, so it could change in the future.
• API changes between 1.76 and 1.78:
  • paragraph has new arguments with indentation information, added as the third argument, and other context information. The formatting linked with commands opened before the paragraph and closed after the paragraph are done in the formatting function. Similar things are done for preformatted.
  • normal_text changed completely. There are much more arguments to give informations about the context, and normal_text now does more text manipulation.
  • New arguments for image the alt text, the height and width, the path to working dir and the path to image file relative to working dir. More image formatting is done in the formatting function.
  • New argument for empty_line.
  • End of line removal is done in formatting function of definition line formatting.
  • node_file_name now should only returns the node file, since the redirection file isn’t used anymore. element_file_name only is used for file names, whatever NODES_FILES is.
• changes between 1.66 and 1.68:
  • When the manual is split the default is to put resulting files in a directory with name the manual file basename (previously they were left in the current directory). To avoid that, call texi2html with ‘-output .’. This has been changed to be compatible with makeinfo and also because it fits better with the cross manuals reference scheme.
  • The option ‘--output’ signification changed. It now has the same meaning than for makeinfo. It seems that in 1.66 it was the same than ‘-out-file’. ‘--output’ new meaning allows to replace ‘-out-file’ and ‘-subdir’ with a unique option.

    More precisely ‘-out-file’ forces the output to be unsplit while ‘--output’ behaves differently when split (it specifies the directory where the manual files should be outputted) and unsplit (it specifies the output file). ‘-subdir’ is retained for backward compatibility.

    If you want a backward compatibility you can use ‘-subdir’ for the output directory if the document is split, and ‘-out-file’ if the document isn’t split. This hasn’t been tested extensively though.

  • Many options has been obsoleted but they are retained for backward compatibility.
  • The init files are searched in new directories, however they are still searched for in the old directories for backward compatibility.
  • the option ‘--glossary’ doesn’t do anything. Likely nothing specific is done regarding bibliographies. This has been decided because this added some semantics to the texinfo formatting language that weren’t part of texinfo.

    It should be possible to do something similar with macros. See for example ‘glossary.texi’ for glossary and ‘my-bib-macros.texi’ for bibliography in the directory ‘examples’. In the web2c package there is an example of use of BibTeX, see http://tug.org/texlive/devsrc/Build/source/TeX/texk/web2c/doc/ (the examples for bibliography are taken from the texinfo home page http://www.gnu.org/software/texinfo/texinfo.html).

  • don’t use T2H_CENTER_IMAGE. @center should be used insead, it will give the right output for all the formats.
  • If a directory creation fails the program aborts. This is much safer.
  • The interface for internationalisation changed, although the previous wasn’t documented a lot.
  • the API described in this manual changed a lot. A important change was to use the Texi2HTML::Config names space instead of variables prefixed with ‘T2H_’ or ‘t2h_’. To cope with the change the prefix should be removed from variables in init files. Some variables are now in %Texi2HTML::THISDOC.
  • @ifinfo regions are not expanded by default. This may lead to warnings or errors especially if the Top node is enclosed in @ifinfo, as some node won’t appear in menus. The quick fix is to call texi2html with the option ‘--ifinfo’ and the right way should be to make more use of @ifnottex.
  • The code appearing before the first node is now outputted, it was ignored before. ‘--ignore-preamble-text’ revert to the previous behaviour. Enclosing in @ifnothtml would be much cleaner.

C How little texi2html texinfo differs from GNU texinfo

For features documented in the texinfo manual, the texinfo interpretation by texi2html shouldn’t differ from the interpretation of makeinfo or texi2dvi. However for constructs with unspecified behaviour texi2html often doesn’t lead to the same result than makeinfo or texi2dvi. makeinfo and texi2dvi are also inconsistent in most of these cases (or broken). You are urged not to use these features unless absolutely necessary. This information is only here to help understand why texi2html differ from other texinfo interpreters, it may be inacurate and the texi2html behaviour may change in the future and was different in the past.

@-commands with text on the line

In the texinfo manual it is specified that block @-commands should appear on a line without text and the closing @end should also be on a line by himself. With texi2html it is possible to add text before and after the command, so the following is right:

something @example the example @end example after the example

makeinfo and texi2dvi may also accept text before the command and text after the @end command, sometimes ignoring it after the @end.

This is a feature you should especially not rely on.

special @-commands handling

The special @-commands are commands like @pagesizes, @sp, @evenheading, @raisesections, @defindex and a lot more. In many cases makeinfo and texi2dvi don’t parse those commands the same way too. texi2html may also show some differences in parsing of the arguments of these commands, in case there are wrong arguments, and also ignore differently things following those commands. How user defined macros, set and values are expanded in those commands may also be different. Part of the specification of how these commands are handled is configureable (see section Customizing other commands, and unknown commands), but not what happens during the beginning of the parsing for some of those commands.

features different between makeinfo and texi2dvi

When makeinfo or texi2dvi use a feature which is reserved for one or the other translator, texi2html uses that feature. So for example @definfoenclose which is ignored by texi2dvi is taken into account and @kbdinputstyle which is ignored by makeinfo is taken into account.

user defined macros and values

In this area makeinfo and texi2dvi also differ a lot. The reference implementation is the makeinfo implementation as texi2dvi is easily broken when macros are not used simply.

• @rmacro and @macro behave exactly the same. In fact this goes against a documented behaviour, however if a user don’t want a recursive macro he can simply avoid reusing the macro in the definition. If somebody report that the feature is usefull we could try to implement it.
• It is possible to escape the end of a macro definition with

  \@end macro
  

  with the ‘\’ being removed after the first expansion. Otherwise it is not possible to produce a \@end macro in a macro.

• @unmacro is interpreted during the macro argument expansion. Don’t know what makeinfo exactly do.
• Some @value may be expanded later than the others, those that are in special commands, like @node.

@, in @node

Like texi2dvi but unlike makeinfo @, don’t break @node arguments like a regular ‘,’.

Things before first node or preamble

Things before the first node or before the preamble may not be exactly interpreted or discarded as makeinfo or texi2dvi do.

encodings

texi2html knows more encodings, in fact all encodings perl knows about.

commands in @ifset and @ifclear

texi2html doesn’t need a proper nesting of internal @ifset or @ifclear if they are in ignored or raw regions (like @html or @verbatim). For example the following is accepted by texi2html and not by makeinfo:

@ifset notset
@ignore
@ifset
@end ignore
@end ifset

In @ifset and @ifclear texi2html also accepts a lot more of invalid constructs. For example the following is accepted by texi2html but not by makeinfo:

@set flag
@ifset flag
@itemize
@item my item
@end ifset
text
@ifset flag
@end itemize
@end ifset

Appendix D Command Line Option Index