\input texinfo   @c -*-texinfo-*-

@setfilename aspell.info
@settitle GNU Aspell 0.60.3
@syncodeindex pg cp
@documentencoding ISO-8859-1

@copying
This is the user's manual for Aspell

Copyright @copyright{} 2000--2005 Kevin Atkinson.

@quotation
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.1 or
any later version published by the Free Software Foundation; with no
Invariant Sections, no Front-Cover Texts and no Back-Cover Texts.  A
copy of the license is included in the section entitled "GNU Free
Documentation License".
@end quotation
@end copying

@dircategory GNU Packages
@direntry
* Aspell: (aspell).        GNU Aspell spelling checker
@end direntry

@titlepage
@title GNU Aspell 0.60.3
@author Kevin Atkinson (@email{kevina@@gnu.org})
@page
@vskip 0pt plus 1filll
@insertcopying
@end titlepage

@contents

@ifnottex
@node Top
@top GNU Aspell
GNU Aspell is a spell checker designed to eventually replace Ispell.
It can either be used as a library or as an independent spell checker.

@menu
* Introduction::                
* Support::                     
* Basic Usage::                 
* Customizing Aspell::          
* Working With Dictionaries::   
* Writing programs to use Aspell::  
* Adding Support For Other Languages::  
* Implementation Notes::        
* Languages Which Aspell can Support::  
* Language Related Issues::     
* To Do::                       
* Installing::                  
* ChangeLog::                   
* Authors::                     
* Copying::                     

@detailmenu
 --- The Detailed Node Listing ---

Basic Usage

* Spellchecking Individual Files::  
* Using Aspell as a Replacement for Ispell::  
* Using Aspell with other Applications::  

Customizing Aspell

* Specifying Options::          
* The Options::                 
* Dumping Configuration Values::  
* Notes on Various Options::    

Notes on Various Options

* Notes on Various Filters and Filter Modes::  
* Notes on the Prefix Option::  
* Notes on Typo-Analysis::      
* Notes on the Different Suggestion Modes::  

Working With Dictionaries

* Using aspell-import::         
* How Aspell Selects an Appropriate Dictionary::  
* Listing Available Dictionaries::  
* Dumping the Contents of the Word List::  
* Creating an Individual Word List::  
* Working With Affix Info in Word Lists::  
* Using Multi Dictionaries::    
* Dictionary Naming::           
* AWLI files::                  

Writing programs to use Aspell

* Through the C API::           
* Through A Pipe::              
* Notes on Storing Replacement Pairs::  

Adding Support For Other Languages

* The Language Data File::      
* Compiling the Word List::     
* Phonetic Code::               
* The Simple Soundslike::       
* Replacement Tables::          
* Affix Compression::           
* Controlling the Behavior of Run-together Words::  
* Creating A New Character Set::  
* Format of the Character Set Data File::  
* Creating An Official Dictionary Package::  

Implementation Notes

* Aspell Suggestion Strategy::  
* Notes on 8-bit Characters::   

Languages Which Aspell can Support

* Supported::                   
* Unsupported::                 
* Multiple Scripts::            
* Planned Dictionaries::        
* References::                  

Language Related Issues

* Compound Words::              
* Words With Symbols in Them::  
* Unicode Normalization::       
* German Sharp S::              
* Context Sensitive Spelling::  

To Do

* Important Items::             
* Other Items::                 
* Notes on Various Items::      

Notes on Various Items

* Word skipping by context::    
* Hidden Markov Model::         
* Email the Personal Dictionary::  

Copying

* GNU Free Documentation License::
* GNU Lesser General Public License::

@end detailmenu
@end menu

@end ifnottex

@node Introduction
@chapter Introduction

GNU Aspell is a spell checker designed to eventually replace Ispell.  It
can either be used as a library or as an independent spell checker.  Its
main feature is that it does a much better job of suggesting possible
replacements for a misspelled word than just about any other spell
checker out there for the English language.  Unlike Ispell, Aspell can
also easily check documents in UTF-8 without having to use a special
dictionary.  Aspell will also do its best to respect the current locale
setting.  Other advantages over Ispell include support for using
multiple dictionaries at once and intelligently handling personal
dictionaries when more than one Aspell process is open at once.

The latest version of Aspell can always be found at @uref{http://aspell.net}

@section Comparison to other spell checker engines

@multitable {Alternate Dictionaries} {Aspell} {Ispell} {Netscape} {Microsoft}
@item             @tab Aspell @tab Ispell @tab Netscape @tab Microsoft
@item             @tab        @tab        @tab  4.0     @tab  Word 97
@item Open Source @tab x      @tab  x
@item Suggestion  @tab 88-98  @tab 54     @tab 55-70?   @tab  71
@item Intelligence
@item Personal part @tab x    @tab x      @tab x
@item of Suggestions
@item Alternate Dictionaries @tab x @tab x @tab ?        @tab ?
@item International Support @tab x @tab x @tab ?         @tab ?
@end multitable

The Suggestion Intelligence is based on a small test kernel of
misspelled/correct word pairs.  Go to @uref{http://aspell.net/test}
for more info and how you can help contribute to the test kernel.  The
current scores for Aspell are 88 in @emph{fast} mode, 93 in
@emph{normal} mode, and 98 in @emph{bad spellers} mode: for more
information about the various suggestion modes @ref{Notes on the
Different Suggestion Modes}.


If you have any other information you would like to add to this chart
please contact me at @email{kevina@@gnu.org}.

@subsection Comparison to Ispell

@subsubsection Features that only Aspell has

@itemize @bullet
@item
Is an actual library that other programs can link to instead of
having to use it through a pipe.

@item
Does a much better job of suggesting possible replacements for a
misspelled word than Ispell does or for that matter any other spell
checker I have seen.  If you know a spell checker that does a better job
please let me know.

@item
Can learn from user's misspellings.

@item
Can easily check documents in UTF-8 without having to use a special
dictionary.

@item
Has support for using multiple dictionaries at once.

@item
Is multiprocess intelligent.  When a personal dictionary (or
replacement list) is saved it will now first update the list against
the dictionary on disk in case another process modified it.

@item
Can share the memory used in the main word list between processes.

@item
A better, more complete word list for the English language.  Word
lists are provided for American, British, and Canadian spelling.
Special care has been taken to only include one spelling for each word
in any particular word list.  The word list included in Ispell by
contrast only included support for American and British and also tends
to included multiple spellings for a word which can mask some spelling
errors.
@end itemize

@subsubsection Things that, currently, only Ispell has

@itemize @bullet
@item
Lower memory footprint

@item
Ability to deal with arbitrary multi-character letters such as old
ASCII encodings of accented letters.

@item
Perhaps better support for spell checking (La)TeX files.

@end itemize

For a detailed description of how Aspell differs from Ispell, @xref{Differences
From Ispell}.

@node Support
@chapter Support

Support for Aspell can be found on the Aspell mailing lists.
Instructions for joining the various mailing lists (and an archive of
them) can be found off the Aspell home page at
@uref{http://aspell.net}.

Bug reports should be submitted via the Sourceforge Tracker at
@uref{http://sourceforge.net/@/tracker/?group_id=245} rather than
being posted to the mailing lists.

@node Basic Usage
@chapter Basic Usage

For a quick reference on the Aspell utility use the command
@command{aspell --help}.

@menu
* Spellchecking Individual Files::  
* Using Aspell as a Replacement for Ispell::  
* Using Aspell with other Applications::  
@end menu

@node Spellchecking Individual Files
@section Spellchecking Individual Files

To use Aspell to spellcheck a file use:

@example
aspell check [@var{options}] @var{filename}
@end example

@noindent
at the command line where @code{@var{filename}} is the file you want to
check and @code{@var{options}} is any number of optional options.  Some of
the more useful ones include:

@table @b
@item --mode=@var{mode}
the mode to use when checking files.  The available modes are
@code{none}, @code{url}, @code{email}, @code{sgml}, @code{tex},
@code{texinfo}, @code{nroff}, among others.  For more information on the
various modes see @ref{Notes on Various Filters and Filter Modes}.

@item --dont-backup
don't create a backup file.

@item --sug-mode=@var{mode}
the suggestion mode to use where mode is one of @code{ultra},
@code{fast}, @code{normal}, or @code{bad-spellers}.  For more
information on these modes see @ref{Notes on the Different Suggestion
Modes}.

@item --lang=@var{name}/-l @var{name}
the language the document is written in.  The default depends on the
current locale.

@item --encoding=@var{name}
encoding the document is expected to be in.  The default depends on the
current locale.

@item --master=@var{name}/-d @var{name}
the main dictionary to use.

@item --keymapping=@var{name}
the keymapping to use.  Either @option{aspell} for the default mapping
or @option{ispell} to use the same mapping that the Ispell utility
uses.
@end table

For more information on the available options, please see
@ref{Customizing Aspell}.

For example to check the file @file{foo.txt}:

@example
aspell check foo.txt
@end example

@noindent
and to check the file @file{foo.txt} using the @option{bad-spellers}
suggestion mode and the American English dictionary:

@example
aspell check --sug-mode=bad-spellers -d en_US foo.txt
@end example

If the @option{mode} option is not given then Aspell will use the
extension of the file to determine the current mode.  If the extension
is @file{.tex}, then @option{tex} mode will be used, if the extension
is @file{.html}, @file{.htm}, @file{.php}, or @file{.sgml} it will
check the file in @option{sgml} mode, otherwise it will use
@option{url} mode. 

For more information on the various modes that can be used, see
@ref{Notes on Various Filters and Filter Modes}.

If Aspell was compiled with curses support and the @env{TERM}
environment variable is set to a capable terminal type than Aspell
will use a nice full screen interface.  Otherwise it will use a
simpler ``dumb'' terminal interface where the misspelled word is
surrounded by two '*'.  In either case the interface should be self
explanatory.

If Aspell is compiled with a version of the curses library that support
wide characters than Aspell can also check UTF-8 text.  Furthermore, the
document will be displayed in the encoding defined by the current
locale.  This encoding does not necessary have to be the same encoding
that the document is in.  This means that is is possible to check an
8-bit encoding such as ISO-8859-1 on an UTF-8 terminal.  To do so
simply set the @option{encoding} option to @samp{iso-8859-1}.
Furthermore it is also possible to check an UTF-8 document on an 8-bit
terminal provided that the document can be successfully converted into
that encoding.

@node Using Aspell as a Replacement for Ispell
@section Using Aspell as a Replacement for Ispell

As of GNU Aspell 0.60.1 Aspell should be able to completely replace
Ispell for most applications.  The Ispell compatibility script should
work for most applications which expect Ispell.  However their are
some differences which you should be aware of.

@subsection As a Drop In Replacement

Aspell can be used as a drop in replacement for Ispell for programs
that use Ispell through a pipe such as Emacs and LyX.  It can also be
used with programs that simply call the @command{ispell} command and
expect the original file to be overwritten with the corrected version.

If you do not have Ispell installed on your system and have installed
the Ispell compatibly script than you should not need to do anything
as most applications will that expect Ispell will work as expected
with Aspell via the Ispell compatibility script.

Otherwise, the recommended way to use Aspell as a replacement for
Ispell is to change the @command{ispell} command from within the
program being used.  If the program uses @command{ispell} in pipe mode
simple change @command{ispell} to @command{aspell}.  If the program
calls the @command{ispell} command to check the file change
@command{ispell} with @command{aspell check}.

If that is impossible than the @command{run-with-aspell} script can be
used.  This script modifies the patch so that programs see the Ispell
compatibility script instead of the actual true @command{ispell}
command.  The format of the script is:

@example
run-with-aspell @var{command}
@end example

@noindent
where @var{command} is the name of the program with any optional
arguments.

The old method of mapping Ispell to Aspell is discouraged because it
can create compatibility problems with programs that actually require
Ispell such as Ispell's own scripts.

@anchor{Differences From Ispell}
@subsection Differences From Ispell

Nevertheless, Aspell is not Ispell, nor is it meant to completely
emulate the behavior of Ispell.  The @command{aspell} command is not
identical to the @command{ispell} command when not used in ``pipe''
mode.  If an application expects the @command{ispell} command than
the Ispell compatibility script should be used.  

@subsubsection Functionally of the Ispell Compatibility Script

The Ispell compatibility script provides the following Ispell
functionally.

@itemize @bullet
@item
The ability to check a file when called with out any mode parameters.

@item
The pipe or -a mode.

@item 
The list of -l mode.

@item
The version or -v mode.  A single line is returned which, while not
being identical to the line Ispell returns, is sufficient to fool most
programs.

@item
The munch or -c mode.

@item
The expand or -e mode.

@item
The ability to dump the affix file when called with '-D'.  However the
format of the affix file is different.  Furthermore, not all languages
have an affix file.

@end itemize

However the Ispell script currently unable to emulate the 
the '-A' pipe mode.  This is different from the normal pipe mode in
that the special @code{&Include_File&} command is recognized.

@subsubsection Recognized Options

Aspell, and thus the ispell compatibility script, recognizes most of
the options that Ispell uses except for the '-S', '-w' and '-T'
options.  The Aspell command will simply ignore these options if it
sees them.

@subsubsection Check Mode Compatibility

The interface used by Aspell when checking individual files is
slightly different than Ispell's.  In particular the default
keymappings are not the same as the ones Ispell uses.  However Aspell
supports using the Ispell keymappings via the @option{keymapping}
option.  To use the Ispell keymappings set the @option{kepmapping}
option to @code{ispell}.  This can be done on the command line by
adding using the command:

@example
  aspell check --keymapping=ispell @dots{}
@exdent or with the Ispell compatibility script
  ispell --keymapping=ispell @dots{}
@end example

The Ispell keymapping can always be used when the Ispell compatibly
script is called by uncommenting the indicated line in the
@command{ispell} script.

@subsubsection Pipe Mode Compatibility

The Aspell pipe mode should be identical to the Ispell pipe mode except
if the line starts with a '$$' as that will trigger special Aspell
only commands or if the line starts with a '~' which is ignored by Aspell.

@subsubsection Other Differences

The compiled dictionary format is completely different than Ispell's.
Furthermore the format of the language data files' are different than
Ispell's affix file.  However, all known Ispell dictionaries been
converted to Aspell format, except for Albanian (sq) as I was unable
to find the source word list.

The naming and format of the personal dictionary is also different.
However, ispell personal dictionaries can be imported using the
@command{aspell-import} script.  @xref{Using aspell-import}.  The
Ispell personal dictionary is simply a list of words while the Aspell
one is a list of words with a header line.  Thus it is also fairly
easy to convert between the two.  @xref{Format of the Personal Dictionary}.

@subsubsection Missing Functionally

The only major area where Ispell is superior to Aspell is in the
handling of multi character letters such as old ASCII encoding of
accented characters.

However, Aspell can handle UTF-8 documents far better than Ispell
can.

@node Using Aspell with other Applications
@section Using Aspell with other Applications

@subsection With Emacs and XEmacs

The easiest way to use Aspell with Emacs or XEmacs is to add this
line:

@verbatim
(setq-default ispell-program-name "aspell")
@end verbatim

to the end of your .emacs file.

For some reason version 3.0 of ispell.el (the lisp program that
(x)emacs uses) want to reverse the suggestion list.  To fix this add
this line:

@verbatim
(setq-default ispell-extra-args '("--reverse"))
@end verbatim

after the previous line in your .emacs file and it should solve the problem.

Ispell.el, version 3.1 (December 1, 1998) and better, has the list
reversing problem fixed.  You can find it at
@uref{http://www.kdstevens.com/~stevens/ispell-page.html}.

@subsection With LyX

Version 1.0 of LyX provides support for Aspell's learning from user's
mistakes feature.

To use Aspell with LyX 1.0 either change the @option{spell_command}
option in the @file{.lyxrc} file or use the @command{run-with-aspell}
utility.

@subsection With VIM

@c @emph{(The following section was written by ``R.  Marc'', rmarc at
@c copacetic net.)}

To use Aspell in VIM you simply need to add the following line to your
@file{.vimrc} file:

@verbatim
map ^T :w!<CR>:!aspell check %<CR>:e! %<CR>
@end verbatim

I use @kbd{Ctrl-T} since that's the way you spell check in
@command{pico}.  In order to add a control character to your
@code{.vimrc} you must type @kbd{Ctrl-v} first.  In this case
@kbd{Ctrl-v Ctrl-t}.

A more useful way to use Aspell, IMHO, is in combination with Newsbody
(@uref{http://www.image.dk/~byrial/newsbody/}) which is how I use it
since VIM is my editor for my mailer and my news reader.
@verbatim
map ^T\\1\\2<CR>:e! %<CR>
map \\1 :w!<CR>
map \\2 :!newsbody -qs -n % -p aspell check \\%f<CR>
@end verbatim

@subsection With Pine

To use Aspell in Pine simply change the option @option{speller} to

@example
aspell --mode=email check
@end example

To change the @option{speller} option go to the main menu.  Type
@kbd{S} for @emph{setup}, @kbd{C} for @emph{config}, then @kbd{W} for
@emph{where is}.  Type in @kbd{speller} as the word to find.  The
speller option should be highlighted now.  Hit enter, type in the
above line, and hit enter again.  Then type @kbd{E} for @emph{exit
setup} and @kbd{Y} to save the change.

If you have a strong desire to check other people's comments change
@option{speller} to

@example
aspell check
@end example

@noindent
instead which will avoid switching Aspell into email mode.

@node Customizing Aspell
@chapter Customizing Aspell

The behavior of Aspell can be changed by any number of options which
can be specified at either the command line, the environment variable
@env{ASPELL_CONF}, a personal configuration file, or a global
configuration file.  Options specified on the command line override
options specified by the environment variable.  Options specified by
the environment variable override options specified by either of the
configuration files.  Finally options specified by the personal
configuration file override options specified in the global
configuration file.  Options specified in the environment variable
@env{ASPELL_CONF}, a personal configuration file, or a global
configuration file will take effect no matter how Aspell is used which
includes being used by other applications.

Aspell has three basic type of options: @dfn{boolean}, @dfn{value},
and @dfn{list}.

@dfn{Boolean} options are either enabled or disabled, @dfn{value}
options take a specific value, and @dfn{list} options can either have
entries added or removed from the list.

@menu
* Specifying Options::          
* The Options::                 
* Dumping Configuration Values::  
* Notes on Various Options::    
@end menu

@node Specifying Options
@section Specifying Options

@subsection At the Command Line

All options specified at the command line have the following basic format:

@example
--@var{option}[=@var{value}]
@end example

@noindent
where the @samp{=} can be replaced by whitespace.

However some options also have single letter abbreviations of the form:

@example
@var{letter} [@var{optional_whitespace} @var{value}]
@end example

Any non-ASCII characters are expected to be in the encoding
specified by the current locale.

To reset an option to the default value prefix the option with a
@samp{reset-} and don't specify a value.

@subsubsection Value options

To specify a value option simply specify the option with its
corresponding value.  For example to set the filter mode to TeX use
@samp{--mode=tex}.

If a value option has a single letter shortcut simply specify the single
letter shortcut with its corresponding value.  For example to use a the
accented version of the American English dictionary use @samp{-d
en_US-w_accents}.

@subsubsection Boolean options

To enable a boolean option simply specify the option without any
corresponding value, or prefix it with an @samp{enable-}.  For example
to create a backup file use @samp{--backup}.  To disable a boolean
option prefix the option name with a @samp{dont-} or @samp{disable-}.
For avoid creating a backup file use @samp{--dont-backup}.  Boolean options
can also be set directly like a value option where the value is either
"true" or "false", for example @samp{--backup=true}.

If a boolean option has a single letter abbreviation simply give the
letter corresponding to either enabling or disabling the option
without any corresponding value.  For example to consider run-together
words legal use @samp{-C} or to consider them illegal use
@samp{-B}

@subsubsection List options

To add a value to the list, prefix the option name with a
@samp{add-} and then specify the value to add.  For example to add
the URL filter use @samp{--add-filter url}.  To remove a value from
a list option prefix the option name with a @samp{rem-} and then
specify the value to remove.  For example to remove the URL filter use
@samp{--rem-filter url}.  To remove all items from a list prefix the
option name with a @samp{clear-} without specify any value.  For
example to remove all filters use @samp{--clear-filter}.

A list option can also be set directly, in which case it will be set to
a single value.  To directly set a list option to multiple values prefix
the option name with a @samp{lset-} and separate each value with a
@samp{:}.  For example to use the URL and TeX filter use
@samp{--lset-filter url:tex}.


@subsection Via a Configuration File

Aspell can also accept options via a personal or global configuration
file.  The exact files to used are specified by the options
@option{per-conf} and @option{conf} respectively but the personal
configuration file is normally @file{.aspell.conf} located in the
@env{HOME} directory and the global one is normally @file{aspell.conf}
which is located in the @file{etc} directory which is normally
@file{/usr/etc} or @file{/usr/local/etc}.  To find out the particular
values for your particular system use @command{aspell dump config}.

Each line of the configuration file has the format:

@example
@var{option} [@var{value}]
@end example

There may be any number of spaces between the option and the value
however it can only be spaces, i.e. there is no @samp{=} between the
option name and the value.

Comments may also be included by preceding them with a @samp{#} as
anything from a @samp{#} to a newline is ignored.  Blank lines are also
allowed.

To include a literal @samp{#} use @samp{\#}.  To include a literal
@samp{\} use @samp{\\}.  Any other non-alpha character can also be
protected by a @samp{\} if necessary.

Any non-ASCII characters are expected to be in UTF-8.

To reset an option to the default value prefix the option with a
@samp{reset-} and don't specify a value.

Values set in the personal configuration file override those in the
global file.  Options specified at either the command line or via an
environment variable override those specified by either configuration
file.

@quotation Note
Filters and corresponding options also may be assembled
inside a special meta filter file named @file{@var{metafilter}.flt}.  A
filter has to be loaded via adding a @code{add-filter @var{filtername}}
line to the meta filter file before its options may be specified.
@end quotation

@subsubsection Value options

To specify a value option simply include the option followed by the
corresponding option.  For example to set the default language to
German use @option{lang german}.

@subsubsection Boolean options

To specify a boolean option simply include the option followed by a
@samp{true} to enable it or a @samp{false} to disable it.  For example
to allow run-together words use @samp{run-together true}.

@subsubsection List options

To add a value to the list, prefix the option name with a
@samp{add-} and then specify the value to add.  For example to add
the URL filter use @samp{add-filter url}.  To remove a value from
a list option prefix the option name with a @samp{rem-} and then
specify the value to remove.  For example to remove the URL filter use
@samp{rem-filter url}.  To remove all items from a list prefix the
option name with a @samp{clear-} without specifying any value.  For
example to remove all filters use @samp{clear-filter}.

A list option can also be set directly, in which case it will be set to
a single value.  To directly set a list option to multiple values prefix
the option name with a @samp{lset-} and separate each value with a
@samp{:}.  For example to use the URL and TeX filter use
@samp{lset-filter url:tex}.  To include a literal @samp{:} use
@samp{\:}.

@subsection Setting Options via an Environment Variable

The environment variable @env{ASPELL_CONF} may also be used and it
overrides any options set in the configuration file.  The format of
the string is exactly the same as the configuration file except that
semicolons (@samp{;}) are used instead of newlines.

@node The Options
@section The Options

The following is a list of available options broken down by category.
Each entry has the following format:

@quotation
@table @b
@item @var{option}[,@var{single-letter-abbreviation}]
@i{(@var{type})} @var{description}
@end table
@end quotation

Where single letter options are specified as they would appear at the
command line, ie with the preceding dash.  Boolean single letter
options are specified in the following format:

@quotation
-<abbreviation to enable>|-<abbreviation to disable>
@end quotation

@var{option} is one of the following: @emph{boolean}, @emph{string},
@emph{file}, @emph{dir}, @emph{integer}, or @emph{list}.

@emph{String}, @emph{file}, @emph{dir}, and @emph{integer} types are all
value options which can only take a specific type of value.

@subsection Dictionary Options

The following options may be used to control which dictionaries to use
and how they behave (for more information see @ref{How Aspell Selects
an Appropriate Dictionary}):

@table @b

@item master,-d
@i{(string)}
Base name of the dictionary to use.  If this option is specified than
Aspell will either use this dictionary or die.

@item dict-dir
@i{(dir)}
Location of the main word list.

@item lang
@i{(string)}
Language to use, it follows the same format of the @env{LANG}
environment variable on most systems.  It consists of the two letter
@acronym{ISO 639} language code and an optional two letter
@acronym{ISO 3166} country code after a dash or underscore.  The
default value is based on the value of the @env{LC_MESSAGES} locale.

@item size
@i{(string)}
The preferred size of the word list.

@item variety
@i{(list)}
Any extra information to distinguish two different words lists that
have the same lang and size.

@item word-list-path
@i{(list)}
Search path for word list information files.

@c @item module-search-order (@i{list})
@c list of available modules, modules that come first on this list have a
@c higher priority. Currently there is only one speller module.

@item personal,-p
@i{(file)}
Personal word list file name.

@item repl
@i{(file)}
Replacements list file name.

@item extra-dicts
@i{(list)}
Extra dictionaries to use.

@item dict-alias
@i{(list)}
create dictionary aliases.  Each entry has the form @samp{@var{from}
@var{to}}.  Will override any system dictionaries that are present.

@end table

@subsection Encoding Options

These options control the encoding the document is expected to be in and
how it is displayed.

@table @b

@item encoding
@i{(string)}
The encoding the input text is in.  Valid values include, but not
limited to, @samp{iso-8859-*}, @samp{utf-8}, @samp{ucs-2},
@samp{ucs-4}.  When using the Aspell utility the default encoding is
based on the current locale.  Thus if your locale currently uses the
@samp{utf-8} encoding than everything will be in @acronym{UTF-8}.  The
@samp{ucs-2} and @samp{ucs-4} encodings are intended to be used by other
programs using the Aspell library and is not supported by the Aspell
utility.

@item normalize
@i{(boolean)}
Perform Unicode normalization.  Enabled by default.

@item norm-strict
@i{(boolean)}
Avoid lossy conversions when normalizing.  Lossy conversions includes
compatibility mappings such as splitting the letter @samp{OE} (U+152)
into @samp{O} and @samp{E} (when the combined letter is not available),
and mappings which will remove accents.  Disabled by default except when
creating dictionaries.

@item norm-form
@i{(string)}
The normalization form the output should be in.  This option primarily
effects the normalization form of the suggestions as when spell
checkering as the actual text is unchanged unless there is an error.
Valid values are @samp{none}, @samp{nfd} for fully decomposition
(Normalization Form D), @samp{nfc} for Normalization Form C, or
@samp{comp} for fully composed.  @samp{comp} is like @samp{nfc} except
that @emph{full} composition is used rather than @emph{canonical}
composition.  The @option{normalize} option must be enabled for this
option to be used.

@item norm-required
@i{(boolean)}
Set to true when the current language requires Unicode normalization.
This is generally the case when private use characters are used
internally by Aspell or when Normalization Form C is not the same as
full composition.

@end table

@subsection Checker Options

These options control the behavior of Aspell when checking
documents.

@table @b
@item ignore,-W
@i{(integer)}
Ignore words with N characters or less

@item ignore-repl
@i{(boolean)}
Ignore commands to store replacement pairs.

@item save-repl
@i{(boolean)}
Save the replacement word list on save all.

@item keyboard
@i{(file)}
The base name of the keyboard definition file to use (@pxref{Notes on
Typo-Analysis})

@item sug-mode
@i{(mode)}
Suggestion mode = @samp{ultra} | @samp{fast} | @samp{normal} | @samp{slow} |
@samp{bad-spellers} (@pxref{Notes on the Different Suggestion Modes})

@item ignore-case
@i{(boolean)}
Ignore case when checking words.

@item ignore-accents
@i{(boolean)}
Ignore accents when checking words -- @emph{currently ignored}.

@end table

@subsection Filter Options

These options modify the behavior of the Aspell filter interface in
general (for more information see @pxref{Notes on Various Filters and
Filter Modes}).

@table @b

@item filter
@i{(list)}
filters to use

@item filter-path
@i{(list)}
Where to look when loading filter and filter modes.

@item mode
@i{(string)}
Sets the filter mode.  Possible values include, but not limited to,
@samp{none}, @samp{url}, @samp{email}, @samp{sgml}, or @samp{tex}.  (The
shortcut options @option{-e} may be used for email, @option{-H} for
HTML, or @option{-t} for @TeX{}).

@end table

These options belong to filters packaged along with Aspell standard
distribution.  These options may be prefixed by the keyword
@code{f-} in order to explicitly indicate that they are options
recognized by a filter and not by Aspell itself.

@subsubsection email
This filter hides quoting characters and email preamble and other
parts of an email which need not to be spell checked.

@table @b
@item email-quote
@i{(list)}
Email quote characters.

@item email-margin
@i{(integer)}
The number of characters that can appear before the quote character
@end table

@subsubsection html
This filter converts an HTML source file into a format which eases
spell checking of HTML texts by Aspell.

@table @b
@item html-check
@i{(list)}
HTML attributes to always check, such as alt= (alternate text).

@item html-skip
@i{(list)}
HTML tags to always skip the contents of, such as <script>.
@end table

@subsubsection sgml
This filter is identical to the HTML filter except that it options has
different default values which are currently the empty list.

@subsubsection tex/latex
This filter hides all LaTeX commands and corresponding parameters not
being readable text in LaTeX output from Aspell.

@table @b
@item tex-command
@i{(list)}
@TeX{} commands

@item tex-check-comments
@i{(boolean)}
check @TeX{} comments

@c @item tex-multi-byte
@c (@i{list}) TeX multi byte letter en|decoding

@end table

@subsubsection texinfo
This filter hides all Texinfo command from Aspell.  It can also hide
Texinfo parameters and environments not corresponding to readable text.

@table @b
@item texinfo-ignore
@i{(list)} Texinfo command to ignore the parameters of.

@item texinfo-ignore-env
@i{(list)} Texinfo environments to ignore.

@end table

@subsubsection context
@c FIXME: Shorten
This filter can be used to spell check source
codes, HTML sources and other texts which consist of different
contexts.  These contexts must be separated by pairs of unique
delimiters.  The different contexts may not be dependent upon each
other except for initial context which is assumed if not any other
context applies.

@table @b
@item context-visible-first
@i{(boolean)} Switches the context which should be visible to Aspell.
Per default the initial context is assumed to be invisible as one
would expect when spell checking source files of programs where
relevant parts are contained in string constants and comments but not
in the remaining code.  If set to true the initial context is visible
while the delimited ones are hidden.

@item add|rem-context-delimiters
@i{(list)} Add or remove pairs of delimiters.  This allows you to
specify the character, or sequences of characters, which should be
used to switch contexts and therefore have to be escaped by @samp{\} if
they should appear literally.  The two delimiting chars belonging to
one pair have to be separated by a space character.  If multiple pairs
are specified by one @option{add|rem-@/context-delimiters} call the
different pairs have to be separated by a literal comma.  Per default
the delimiters are set to C/C++ comment and string constant
delimiters.  If the end of line delimits a context than this has to be
indicated by the literal @samp{\0} string.
@end table

@subsection Run-together Word Options

These may be used to control the behavior of run-together words (for
more information @pxref{Controlling the Behavior of Run-together
Words}):
@table @b

@item run-together,-C|-B
@i{(boolean)}
consider run-together words legal

@item run-together-limit
@i{(integer)}
maximum numbers that can be strung together

@item run-together-min
@i{(integer)}
minimal length of interior words
@end table

@subsection Miscellaneous Options

Miscellaneous other options that don't fall under any other category
@table @b

@item conf
@i{(file)}
main configuration file

@item conf-dir
@i{(dir)}
location of main configuration file

@item data-dir
@i{(dir)}
location of language data files

@item local-data-dir
@i{(dir)}
alternative location of language data files.  This directory is
searched before @option{data-dir}.  It defaults to the same directory
the actual main word list is in (which is not necessarily
@option{dict-dir}).

@item home-dir
@i{(dir)}
location for personal files

@item per-conf
@i{(file)}
personal configuration file

@item prefix
@i{(dir)}
prefix directory

@item set-prefix
@i{(boolean)}
set the prefix based on executable location (only works on WIN32 and
when compiled with @option{--enable-win32-relocatable})
@end table

@subsection Aspell Utility Options

@table @b

@item backup,-b|-x
@i{(boolean)}
Create a backup file by appending @file{.bak} to the file name.  (Only
applies when the command is @command{check}).

@item time
@i{(boolean)}
Time load time and suggest time in @command{pipe} mode.

@item byte-offsets
@i{(boolean)}
Use byte offsets instead of character offsets in @command{pipe} mode.

@item reverse
@i{(boolean)}
Reverse the order of the suggestions list in @command{pipe} mode.

@item keymapping
@i{(string)}
the keymapping to use.  Either @option{aspell} for the default mapping
or @option{ispell} to use the same mapping that the Ispell utility
uses.

@item guess
@i{(boolean)}
make possible root/affix combinations not in the dictionary in
@command{pipe} mode.

@item suggest
@i{(boolean)}
Suggest possible replacements in @command{pipe} mode.  If false Aspell
will simply report the misspelling and make no attempt at suggestions
or possible corrections.
@end table


@node Dumping Configuration Values
@section Dumping Configuration Values

To find out the current value of all the options use the command
@command{aspell dump config}.  This will dump the current
configuration to standard output.  The format of the contents dumped
is such that it can be used as either the global or personal
configuration file.

To find out the current value of a particular option use
@command{aspell config @var{option}}.  This will print out the value of
@var{option} to @code{stdout} and nothing else.

@node Notes on Various Options
@section Notes on Various Options

@menu
* Notes on Various Filters and Filter Modes::  
* Notes on the Prefix Option::  
* Notes on Typo-Analysis::      
* Notes on the Different Suggestion Modes::  
@end menu

@node Notes on Various Filters and Filter Modes
@subsection Notes on Various Filters and Filter Modes

Aspell now has filter support.  You can either select from individual
filters or choose a filter mode.  To select a filter mode use the
@option{mode} option.  You may choose from @samp{none}, @samp{url},
@samp{email}, @samp{sgml}, @samp{ccpp}, @samp{tex} and any other
available on your system.  The default mode is @samp{url}.  Individual
filters can be added with the option @option{add-filter} and removed
with the @option{rem-filter} option.  The currently available filters
are @samp{url}, @samp{email}, @samp{sgml} and @samp{tex}, @samp{latex}
(alias for @samp{tex}), @samp{nroff}, @samp{context}, as well as a
bunch of filters which translate the text from one format to another.

To check which filters are available use @command{aspell dump filters}.
To check which filter modes are available use @command{aspell dump
modes}.  The @command{aspell help} command will also list all available
filter and filter modes.

@subsubsection None Filter Mode

The @option{none} mode is exactly what it says.  It turns off all filters.

@subsubsection URL Filter

The @option{url} filter/mode skips over URLs, host names, and email
addresses.  Because this filter is almost always useful and rarely
does any harm it is enabled in all modes except @option{none}.  To
turn it off either select the @option{none} mode or use
@option{rem-filter} option @emph{after} the desired mode is selected.

@subsubsection Email Filter

The @option{email} filter mode skips over quoted text.  It currently
does not support skipping over headers however a future version
should.  In the meantime I suggest you use Aspell with Newsbody which
can be found at @uref{http://home.worldonline.dk/~byrial/newsbody/}.
The option @option{email-skip} controls the number of characters that
can appear before the email quote character, the default is 10.  The
option @option{add|rem-email-quote} controls the characters that are
considered quote characters, the defaults are @samp{>} and @samp{|}.

@subsubsection SGML Filter

The SGML filter allows you to spell check SGML, HTML, XHTML, and XML
files. In most cases everything within a tag @samp{<tag attrib=value
attrib2="a whole sentence">} will be skipped by the spell checker. The
SGML/HTML/XML that Aspell supports is a slight superset of most DTDs
(Document Type Definitions) and can spell check the often non-conforming
HTML found on the web.

Two configuration options, @option{sgml-skip} and @option{sgml-check},
allow you to control what is spell checked. The tag and attribute names
specified are case insensitive.

@table @b
@item sgml-skip
This is a list of tags whose contents will also be skipped by the spell
checker.  For example, if you wish to leave a misspelling in a document
and not have them flagged as misspellings, you could surround them with
a <nospellcheck> tag:
@example
  <TD><FONT size=2><NOSPELLCHECK>leviosa</NOSPELLCHECK>
  is what Mr. Potter said</FONT></TD>
@end example
@noindent
And put that word in the skip config directive:
@example
add-sgml-skip nospellcheck
@end example

@item sgml-check

This is a list of attributes whose values you do want spell checked. By
default, 'alt' (<img> alternate text) is a member of the check list
since it is text that is seen by a web page viewer. You may also want
'value' to be on the check list since that is the text put on buttons:
@example
add-sgml-check value
@end example
@noindent
In this case @samp{<input type=button value="Donr">} will be flagged as a
misspelling.
@end table

This filter will also translate SGML characters of the form
@samp{&#num;}.  Other SGML characters such as @samp{&amp;} will simply
be skipped over so that the word @samp{amp}, for example, will not be
spell checked.  Eventually full support for properly translating SGML
characters will be added.

@subsubsection HTML Filter

The @option{html} filter is like the SGML Filter Mode but specialized for
HTML.  By default, 'script' and 'style' are members of the skip list in
HTML mode.

@subsubsection @TeX{}/LaTeX Filter

The @option{tex} (all lowercase) filter mode skips over @TeX{}
commands and parameters and/or options to certain commands.  It also
skips over @TeX{} comments by default.  The option
@option{[dont-]tex-check-comments} controls whether or not Aspell will
skip over @TeX{} comments.  The option @option{add|rem-tex-command}
controls which @TeX{} commands should have certain parameters and/or
options also skipped over.  Commands that are not specified will have
all their parameters and/or options checked.  The format for each item
is

@example
<command> <a list of p,P,o and Os>
@end example

The first item is simply the command name.  The second item controls
which parameters to skip over.  A 'p' skips over a parameter while a
'P' doesn't.  Similar an 'o' will skip over an optional parameter
while a 'O' doesn't.  The first letter on the list will apply to the
first parameter, the second letter will apply to the second parameter
etc.  If there are more parameters than letters Aspell will simply
check them as normal.  For example the option

@example
add-tex-command rule pp
@end example

@noindent
will skip over the first two parameters of the @code{rule} command
while the option

@example
add-tex-command foo Pop
@end example

@noindent
will @emph{check} the first parameter of the @code{foo} command, skip
over the next optional parameter, if it is present, and will skip over
the second parameter --- even if the optional parameter is not present
--- and will check any additional parameters.

A @samp{*} at the end of the command is simply ignored.  For example
the option

@example
enlargethispage p
@end example

@noindent
will ignore the first parameter in both @option{enlargethispage} and
@option{enlargethispage*}.

To remove a command simple use the @option{rem-tex-command} option.
For example

@example
rem-tex-command foo
@end example

@noindent
will remove the command foo, if present, from the list of @TeX{}
commands.

The TeX filter mode is also available via @option{latex} alias name.

@c The TeXfilter mode also contains a decoding and a encoding filter for
@c @emph{babel} character codes like the German Umlauts:

@c @itemize @bullet
@c @item
@c @code{@"a} -> @code{\"a} -> @code{"a}

@c @item
@c @code{@"o} -> @code{\"o} -> @code{"o}

@c @item
@c @code{@"u} -> @code{\"u} -> @code{"u}

@c @item
@c @code{@"s} -> @code{\"s} -> @code{"s}
@c @end itemize
@c @end quotation


@c @quotation
@c @option{add|rem-tex-multi-byte} conversion

@c Changes list of multi character coded TeX(babel) characters recognized
@c by Aspell.  In case of German umlauts mentioned above this would mean
@c that Aspell would decode from their multi character representation to
@c their proper single char representation.  Given the German word
@c @code{St@"arke} (strength) which within TeX/LaTeXdocument has to be
@c written as @code{St"arke} or as @code{St\"arke} would split it into
@c the two words @code{St} and @code{arke} if it does not know anything
@c about the multi character encoding @code{"a} or @code{\"a} of
@c @code{@"a}.  On the other hand if it knows about it than Aspell will
@c recognize the word properly and will not try to make any strange
@c suggestion.

@c Each multi character coding conversion has to be specified the
@c following way:
@c @example
@c @i{char}:@i{rep}[:@i{rep}[@dots{}]]
@c @end example
@c where @code{@i{char}} is the character encoded by multiple characters
@c and rep stands for a specific representation of that character.  For
@c each character may be specified as many representations as available
@c for it.
@c @end quotation

@subsubsection Texinfo Filter

The @option{texinfo} filter allows you to spell check Texinfo files.
It will skip over any Texinfo commands and their parameters when
appropriate.  It will also skip over some Texinfo environments such as
@command{example}.  The list option @option{texinfo-ignore}
controls which commands to ignore the parameters of and the list option
@option{texinfo-ignore-env} controls which Texinfo
environments to ignore.

The Texinfo filter has special code to deal with the @command{@@table}
and related commands.  It will apply the formatting command to each of
the @command{@@item} or @command{@@itemx} commands just like Texinfo
will.  This means that if the formatting command is @command{@@code}
and and the @command{@@code} command is a member of the
@option{texinfo-ignore} option than the Texinfo filter will ignore the
parameter of the @command{@@item} command as if the parameter was also
the parameter of the @command{@@code} command.

The Texinfo filter will also skip over the @samp{\input texinfo} line.

@subsubsection Nroff Filter

The @option{nroff} filter mode allows you to check the spelling of
Nroff documents. The mode is enabled by giving
@option{--add-filter=nroff} or @option{-n} command line option to
@command{aspell}. It is also automatically enabled if the first three
characters of the file being checked are @code{.\"} (a @command{nroff} comment
marker) or the file name ends in a one of the following suffixes:

@itemize
@item single decimal digit from @samp{0} to @samp{9}
@item letter @samp{n}
@item @samp{tmac}
@end itemize

@noindent
This filter mode skips following @command{nroff} language
elements:

@itemize
@item Comments
@item Requests
@item Names of @command{nroff} registers (both traditional two-letter
names and GNU nroff long names)
@item Arguments to the following requests: @code{ds}, @code{de}, @code{nr},
@code{do}, @code{so}.
@item Arguments to font switch (@code{\f}) and size switch (@code{\s})
escapes
@item Arguments to extended charset escape in both traditional
(@code{\(}) and extended (@code{\[comp1 comp2 @dots{}]}) form.
@end itemize

@subsubsection Context Filter

The @emph{context} filter allows Aspell to distinguish between visible
and invisible contexts.  The visible ones will be spell checked and
the invisible ones will be ignored.  The contexts are distinguished by
the fact that the visible/invisible ones are delimited by specific and
unique delimiter characters or character sequences.  Whether the
delimited contexts should be visible or invisible only stated by the
value of the @option{[dont-]context-visible-first} option and not by
the delimiters.

The context delimiters are specified as pairs of delimiters via the
@option{add|rem-@/context-delimiters} option.  The delimiters enclosing
a specific context are specified as a space separated pair.  If more
than one delimiter pair is specified by one call of
@option{add|rem-@/context-delimiters} they have to be combined to a
comma separated list.  To indicate that a context is always closed by
end of line use @code{\0} sequence as closing delimiter.

@subsubsection Ccpp Filter Mode

The @option{ccpp} filter mode will limit spell checking to C/C++ comments
and string literals. Any code in between will be left alone.

@node Notes on the Prefix Option
@subsection Notes on the Prefix Option

The @option{prefix} option is there to allow Aspell to easily be
relocated.  Changing @option{prefix} will change all directory names
relative to the new prefix that are not explicitly set.  For example
if @option{prefix} was @file{/usr/local/aspell} and @option{dict-dir}
has a default value of @file{/usr/local/aspell/dict} than changing
@option{prefix} to @file{/opt/aspell} will also change the default
value of @option{dict-dir} to @file{/opt/aspell/dict}.  Note that
modifying @option{prefix} will only effect the default compiled in
values of directories.  If a directory option is explicitly given a
value than changing the value of @option{prefix} has no effect on that
directory option.

@node Notes on Typo-Analysis
@subsection Notes on Typo-Analysis and the Keyboard Definition File

Aspell .33 and better will, in general, give a higher priority to
certain misspellings which are likely to be due to typos such as
@code{teh} instead of @code{the} or @code{hapoy} instead of
@code{happy}.  However in order to do this well Aspell needs to know the
layout of the keyboard via the keyboard definition file.  The keyboard
definition file simply identifies the keys on the keyboard and which of
them are right next to each other.  It has an extension of @file{.kbd}
and all non-ASCII characters are expected to be in UTF-8.

To identify a key use:

@example
key @var{base} @var{other} @dots{}
@end example
@noindent

where @var{base} is the base character that the key types, and
@var{other} are other keys that the key can produce.  For example

@example
key a A @'a @'A
@end example

It generally is only necessary to list keys which type more than one
distinct letter as Aspell can derive the rest from the language data
file.  For example, it is not necessary to include the previously
mentioned key.

To identify two keys as being right next to each other simply list the
type keys right after each other.  For example the line:

@example
as
@end example

@noindent
will indicate that @samp{a} and @samp{s} are right next to each other.
If @samp{as} is listed as a entry it is not necessary to list @samp{sa}
as an entry as that will be done automatically.  Also by @dfn{right
next to each other} I mean two keys that are close enough together that
it is easy to type one instead of the other.  On most keyboards this
means keys that are to the left or to the right of each other and
@emph{not} keys that are below or above it.

The default for this option is normally @option{standard}.  However
the default can be changed via the language data file.  The normal
default, @option{standard}, should work well for most QWERTY like
keyboard layouts.  It may need minor adjusting for foreign keyboards.
The @option{dvorak} option can be used for a Dvorak layout.

@node Notes on the Different Suggestion Modes
@subsection Notes on the Different Suggestion Modes

In order to understand what these suggestion modes do, a basic
understanding of how Aspell works is required.  For that see
@ref{Aspell Suggestion Strategy}.

The suggestion modes are as follows.

@table @b

@item ultra
This method will use the fastest method available to come up with
decent suggestions.  This currently means that it will look for
soundslikes within one edit distance.  This method will also use the
replacement table if one is available.  In this mode Aspell gets about
87% of the words from my small test kernel of misspelled words.  (Go
to @uref{http://aspell.net/test} for more info on the test kernel as
well as comparisons of this version of Aspell with previous versions
and other spell checkers.)

@item fast
This method is currently identical to @option{ultra}.

@item normal
This mode will use what ever method is necessary to return good
suggestions in most cases in a reasonable amount of time.  This
currently means it will looks for soundslikes within two edit distance
apart.  This mode gets 93% of the words.

@item slow
Like @option{normal} except that ``reasonable amount of time'' is not
a consideration.  In most cases it will return the same results as
@option{normal}.  The biggest difference is that it will try an ngram
scan if the normal methods of finding a suggestion fail.  
@c FIXME: Explain what this means.

@item bad-spellers
This method is like @option{slow} but is tailored more for the bad
speller, where as the other modes are tailored more
to strike a good balance between typos and true misspellings.  This
mode never performs typo-analysis and returns a @emph{huge} number of
words for the really bad spellers who can't seem to get the spelling
anything close to what it should be.  If the misspelled word looks
anything like the correct spelling it is bound to be found
@emph{somewhere} on the list of 100 or more suggestions.  This mode
gets 98% of the words.
@end table

If jump tables were not used then the @option{normal} option is
identical to @option{fast}.  And the @option{slow} option is identical
to the @option{normal} if jump tables were used.


@node Working With Dictionaries
@chapter Working With Dictionaries

@menu
* Using aspell-import::         
* How Aspell Selects an Appropriate Dictionary::  
* Listing Available Dictionaries::  
* Dumping the Contents of the Word List::  
* Creating an Individual Word List::  
* Working With Affix Info in Word Lists::  
* Format of the Personal and Replacement Dictionaries::  
* Using Multi Dictionaries::    
* Dictionary Naming::           
* AWLI files::                  
@end menu

@node Using aspell-import
@section Using @command{aspell-import}

The @command{aspell-import} Perl script will look for old personal
dictionaries and will import them into GNU Aspell.  It will look for
both Ispell and Aspell ones.  To use it, just run it from the command
prompt.  If you get an error about @file{/usr/bin/perl} not being
found, then instead try @command{perl @var{bindir}/aspell-import}.  When
running the script if you get a message like:

@verbatim
Error: No word lists can be found for the language "de".
@end verbatim

This means that you have not installed support for the given language,
in this case @code{de} for German.  To rectify the situation download
and install a dictionary designed to work with GNU Aspell 0.50 or
better.


@node How Aspell Selects an Appropriate Dictionary
@section How Aspell Selects an Appropriate Dictionary

If the @option{master} option is set in any fashion (via the command
line, the @env{ASPELL_CONF} environment variable, or a configuration
file) Aspell will look for a dictionary of that name.  If one could
not be found, it will complain.

Otherwise it will use the value of the @option{lang} option to search
for an appropriate dictionary.  If more than one dictionary is found
for the given language string then it will look for a dictionary with
a matching variety if the @option{variety} option is set.  If it is not
set it will look for a dictionary without a variety.  If after matching
the @option{lang} and @option{variety} there is still more than one
dictionary available it will find one with the size closest to the
value of the @option{size} option.  The default size is 60.  If Aspell
cannot find a dictionary based on the @option{lang} option then it
will give up and complain.

If the @option{lang} option is not explicitly set its value will be
based on the @env{LC_MESSAGES} locale.  This locale is generally taken
from the @env{LC_MESSAGES} environment variable or the @env{LANG}
environment variable if @env{LC_MESSAGES} is not set.  However, if
Aspell is being used as a library from within another program which
already explicitly set the locale then it will use the locale of the
library rather than the environment variables.  If Aspell cannot
determine the language from the @env{LC_MESSAGES} locale than it will
default to @code{en_US}.

The list option @option{dict-alias} can be used to influence which dictionary
is selected by creating an alias from one dictionary name to another.
This option is most useful when there is more than one dictionary for a
given language.  For example @samp{add-dict-alias en_US en_US-w_accents}
will cause Aspell to choose the accented version of the American English
dictionary instead of the non-accented version.  To add an alias use:

@example
add-dict-alias @var{NAME} @var{VAL}
@end example

@node Listing Available Dictionaries
@section Listing Available Dictionaries

For a list of available dictionaries use the command @command{aspell
dump dicts}.  This will form a list of dictionaries that Aspell will
search when a dictionary is not specifically given.

@node Dumping the Contents of the Word List
@section Dumping the Contents of the Word List

The dump command in @command{aspell} will simply dump the contents of
a word list to @file{stdout} in a format that can be read back in with
@command{aspell create}.

If no word list is specified the command will act on the default one.
For example the command

@example
aspell dump personal
@end example

@noindent
will simply dump the contents of the current personal word list to
@file{stdout}.

@node Creating an Individual Word List
@section Creating an Individual Word List

To create an individual main word list from a list of words use the
command

@example
aspell --lang=@var{lang} create master ./@var{base} < @var{wordlist}
@end example

@noindent
where @var{base} is the name of the word list and
@var{wordlist} is the list of words separated by white space.
The name of the word list will automatically be converted to all
lowercase.  The @code{./} is important because without it Aspell will
create the word list in the normal word list directory.  If you are
trying to create a word list in a language other than English check
the Aspell @option{data-dir} (usually @file{/usr/share/aspell}, use
@code{aspell dump config} to find out what it is on your system) to
see if a language data file exists for your language.  If not you will
need to create one.  For more information on using Aspell with other
languages @ref{Adding Support For Other Languages}.

This will create the file @file{@var{base}} in the current directory.
To use the new word list copy the file to the normal word list
directory (use @code{aspell config} to find out what it is) and use
the option @option{--master=@var{base}}.

During the creating of the dictionary you may get a number of warnings
or errors about invalid words or affixes.  By default Aspell will skip
any invalid words and remove invalid affixes.  If you rather that Aspell
simply accepts all words given then the option
@option{--dont-validate-words} can be specified.  To avoid checking if
affixes are valid use the option @option{--dont-validate-affixes}.
However, rather than disable checking, it is preferable to clean the
input word list.  This can be done by using the command

@example
aspell --local-data-dir=./ --lang=@var{lang} clean < @var{wordlist} > @var{result}
@end example

@noindent
which will clean the word list and output the results to @var{result}.
By default it will remove invalid characters from the beginning and end
of a word before resorting to skipping the word.  If you rather it just
skip the words than add the keyword strict:

@example
aspell --local-data-dir=./ --lang=@var{lang} clean strict < @var{wordlist} > @var{result}
@end example

The option @option{--clean-words} can be be added when creating a
dictionary if you want Aspell to remove invalid characters from the
beginning and end of a word like the "clean" command does. In addition
the options @option{--dont-skip-invalid-words} and
@option{--dont-clean-affixes} can be specified to turn the warnings into
errors.

The compiled dictionary file are endian order dependent.  When a
dictionary is loaded the endian order is checked.  Please do not
distribute the compiled dictionaries unless you are only distributing
them for a particular platform such as you would a binary.

Aspell is now also able to use special @code{multi} dictionaries.  for
more information @ref{How Aspell Selects an Appropriate Dictionary}.

A personal and replacement word list can be created in a similar fashion.

@c FIXME: add notes about how affix compression works when creating
@c dictionaries.

@subsection Format of the Replacement Word List

The replacement word list has each replacement pair on its own line in
the following format

@example
@i{misspelled_word} @i{correction}
@end example

@node Working With Affix Info in Word Lists
@section Working With Affix Info in Word Lists

@subsection The Munch Command

The @command{munch} command takes a list of words from standard input
and outputs a list of possible root words and affixes.  The root may,
however, be invalid as it does not check them against the exiting
dictionary.  For example the command:
@example
echo brother | aspell -l en munch
@exdent produces
brother broth/R brothe/R
@end example

@subsection The Expand Command

The @command{expand} command is the reverse of @command{munch}, it
expands affix flags to produce a list of words.  For example:
@example
echo both/R | aspell -l en expand
@exdent produces
both bother
@end example

It formal usage is:
@example
aspell expand [@var{level}] [@var{limit}]
@end example
@noindent
Where @var{level} is the expansion level.  Valid values are between 1
and 3.  Level 1 is the default if not otherwise specified.  Level 2
causes the original root/affix to be included, for example:
@example
both/R both bother
@end example
@noindent
Level 3 causes multiple lines to be printed, one for each generated
word, with the original root/affix combination followed by the word it
creates:
@example
both/R both
both/R bother
@end example
@noindent
Levels larger than 3 may also be supported, but should not used as
they may eventually be removed.

If a @var{limit} parameter is given than only expansions which effect
the first @var{limit} letters will be expanded.  If a base word is not
completely expanded for a given affix flag that flag will be left on
the word.  Note that prefixes are always expanded.

@subsection The Munch-list Command

The @command{munch-list} command will reduce the size of word list via
affix compression.  It will reduce a list of words to a minimal (or
close to it) set of roots and affixes that will match the same list of
words.  The list of words is read from standard input and the result,
the ``munched'' list, is written to standard out.  It's usage is:

@example
aspell munch-list [keep] [single|multi] [simple] < @var{infile} > @var{outfile}
@end example

@noindent
where @option{simple}, @option{single}, @option{multi}, and
@option{keep} are literal values.

The default algorithm used should give near optimum results.  In some
cases the set of words returned is, provably, the minimum number
possible.  In the typical case the number of words returned is within
1% of the optimum number.

By default Aspell will remove redundant affix flags.  The @option{keep}
flag will avoid removing them, which can be useful if you want to
include all possible expansions for each base word.

When cross products are involved it may be beneficial to list a base
word more than once.  Unfortunately, the current version of Aspell can
not correctly handle multiple base words in a dictionary.  Therefore,
the current default behavior is to only include the one with the most
expansions.  All of them can be included via the @option{multi} flag.
Once aspell is able to handle multiple base words the default will be
to include them all.  The @option{single} flag can be used to only
include one of them.

The @option{simple} flag will select an alternate faster algorithm.
This algorithm is very similar to the @command{munch} command
distributed with MySpell (the Open Office spell checker), however, it
doesn't give nearly as good results.  It does okay for the English
word list but not for some other languages such as German; the normal
algorithm reduced a list of 312,002 German words to 79,420 base words
while the simple algorithm only reduced it to 115,927 words.  This
algorithm may disappear in a future version of Aspell.

@node Format of the Personal and Replacement Dictionaries
@section Format of the Personal and Replacement Dictionaries

@anchor{Format of the Personal Dictionary}
@subsection Format of the Personal Dictionary
                                   
The personal dictionary generally has a filename of the form:

@example
.aspell.@var{lang}.pws
@end example

@noindent
And the file itself contains two parts.  The first part is a
header line of the form:

@example
personal_ws-1.1 @var{lang} @var{num} @i{[}@var{encoding}@i{]}
@end example  

@noindent
where @var{num} is the number of words in the list.  This number is
only used as a hint, and thus does not have to be accurate.  When
creating a new dictionary it is perfectly acceptable for @var{num} to
be 0.  The @var{encoding} is optional and specifies the encoding of
the word list.  If it is left out the encoding is expected to be in
the default encoding for the language as specified by the
@option{data-encoding} option.  @xref{data-encoding}.

The second part of simply a word list with one word per line.

@subsection Format of the Personal Replacement Dictionary

The personal replacement dictionary generally has a filename of the form:

@example
.aspell.@var{lang}.prepl
@end example

@noindent
And the file itself contains two parts.  The first part is a
header line of the form:

@example
personal_repl-1.1 @var{lang} @var{num} @i{[}@var{encoding}@i{]}
@end example  

@noindent
where @var{num} is currently unused and thus always 0.  As with the
personal dictionary the @var{encoding} is optional.

The second part simply a list of replacements with one replacement
per line with each replacement pair has the following format:

@example
@var{misspelled_word} @var{correction}
@end example

@node Using Multi Dictionaries
@section Using Multi Dictionaries

As with previous versions of Aspell you can specify the main
dictionary to use via the @option{-d} or @option{--master} option.
However as of @acronym{Aspell .32} you can now also:

@enumerate
@item
Specify more than word list to use with the
@option{extra-dicts} option.

@item
Specify special @emph{multi} dictionaries.
@end enumerate

The @option{extra-dicts} is a list option.  To add a dictionary use
@option{add-extra-dicts} or to remove a dictionary from the list
use @option{rem-extra-dicts}.

A @emph{multi} dictionary is a special file which is basically a list
of dictionary files to use.  A @emph{multi} dictionary must end in
@file{.multi} and has roughly the same format as a configuration file
with the only accepted key being @option{add}.

@node Dictionary Naming
@section Dictionary Naming

In order for Aspell to be able to correctly recognize a dictionary
based on the setting of the @env{LANG} environment variable the
dictionaries need to be located somewhere Aspell can find them and
they need to be @emph{multi} dictionaries.  Where aspell looks for
dictionaries depends on the value of the @option{dict-dir} and
@option{word-list-path} option.

@option{dict-dir} is generally @file{@var{prefix}/lib/aspell}, and
@option{word-list-path} is generally empty.

Each dictionary that you expect Aspell to be able to find needs to
have a name in the following format:

@example
@var{language}[_@var{region}][-@var{variety}][-@var{size}].multi
@end example

@noindent
where @var{language} is the two letter language code, @var{region} is
the two letter region code, @var{variety} is any extra information to
distinguish the word list from other ones with the same language and
spelling.  Multiple varieties can be used by separating them with a '-'.
Finally, @var{size} is the size of the dictionary.  If no size is
specified that the default size of 60 will be assumed.

For example:

@example
en.multi
en_US.multi
en-medical.multi
en-medical-85.multi
en-85.multi
de.multi
@end example

@node AWLI files
@section AWLI files

In order for Aspell to find dictionaries that are located in odd
places or not named according to @ref{Dictionary Naming}, an AWLI file
needs to be created for the dictionary and located in some place where
Aspell can find it.

Each AWLI file has a name in the following format:

@example
@var{language}[@var{region}][-@var{variety}][-@var{size}]-@var{module}.awli
@end example

@noindent
where the names have the same meaning as in @ref{Dictionary Naming},
and @var{module} is the speller module to use, which should be set
to @var{default} for now since there is only one speller module.

Each @file{awli} file for an Aspell word list should then contain
exactly one line which contains the full path of the main word list.

@node Writing programs to use Aspell
@chapter Writing programs to use Aspell

There are two main ways to use Aspell from within your application.
Through the external C API or through a pipe.  The internal
Aspell API can be used directly but that is not recommended as the
actual Aspell API is constantly changing.

@menu
* Through the C API::           
* Through A Pipe::              
* Notes on Storing Replacement Pairs::  
@end menu

@node Through the C API
@section Through the C API

The Aspell library contains two main classes and several helper
classes.  The two main classes are @code{AspellConfig} and
@code{AspellSpeller}. The @code{AspellConfig} class is used to set
initial defaults and to change spell checker specific options.  The
@code{AspellSpeller} class does most of the real work.  The
@code{C API} is responsible for managing the dictionaries, checking
if a word is in the dictionary, and coming up with suggestions among
other things. There are many helper classes the important ones are
@code{AspellWordList}, @code{AspellMutableWordList},
@code{Aspell*Enumeration}.  The @code{AspellWordList} classes is used
for accessing the suggestion list, as well as the personal and
suggestion word list currently in use.  The
@code{AspellMutableWordList} is used to manage the personal, and
perhaps other, word lists.  The @code{Aspell*Enumeration} classes are
used for iterating through a list.

@subsection Usage

To use Aspell your application should include @file{aspell.h}.  In
order to ensure that all the necessary libraries are linked in libtool
should be used to perform the linking.  When using libtool simply
linking with @code{-laspell} should be all that is necessary.  When
using shared libraries you might be able to simply link
@code{-laspell}, but this is not recommended.  This version of Aspell
uses the CVS version of libtool however released versions of libtool
should also work.

When your application first starts you should get a new configuration
class with the command:

@smallexample
AspellConfig * spell_config = new_aspell_config();
@end smallexample

@noindent
which will create a new @code{AspellConfig} class.  It is allocated
with @command{new} and it is your responsibility to delete it with
@code{delete_aspell_config}.  Once you have the config class you
should set some variables.  The most important one is the language
variable.  To do so use the command:

@smallexample
aspell_config_replace(spell_config, "lang", "en_US");
@end smallexample

@noindent
which will set the default language to use to American English.  The
language is expected to be the standard two letter ISO 639 language
code, with an optional two letter ISO 3166 country code after an
underscore.  You can set the preferred size via the @option{size}
option, any extra info via the @option{variety} option, and the
encoding via the @option{encoding} option.  Other things you might
want to set is the preferred spell checker to use, the search path for
dictionaries, and the like --- see @ref{The Options}, for a list of
all available options.

Whenever a new document is created a new @code{AspellSpeller} class
should also be created.  There should be one speller class per
document.  To create a new speller class use the
@code{new_aspell_speller} and then cast it up using
@code{to_aspell_speller} like so:

@smallexample
AspellCanHaveError * possible_err = new_aspell_speller(spell_config);
AspellSpeller * spell_checker = 0;
if (aspell_error_number(possible_err) != 0)
  puts(aspell_error_message(possible_err));
else
  spell_checker = to_aspell_speller(possible_err);
@end smallexample

@noindent
which will create a new @code{AspellSpeller} class using the defaults
found in @code{spell_config}.  To find out which dictionary is
selected the @option{lang}, @option{size}, and @option{variety} options
may be examined.  To find out the exact name of the dictionary the
@option{master} option may be examined as well as the
@option{master-flags} options to see if there were any special flags
that were passed on to the module.  The @option{module} option way
also be examined to figure out which speller module was selected, but
since there is only one this option will always be the same.

If for some reason you want to use different defaults simply clone
@code{spell_config} and change the setting like so:

@smallexample
AspellConfig * spell_config2 = aspell_config_clone(spell_config);
aspell_config_replace(spell_config2, "lang","nl");
possible_err = new_aspell_speller(spell_config2);
delete_aspell_config(spell_config2);
@end smallexample

Once the speller class is created you can use the @code{check} method
to see if a word in the document is correct like so:

@smallexample
int correct = aspell_speller_check(spell_checker, @var{word}, @var{size});
@end smallexample

@noindent
@var{word} is expected to be a @code{const char *} character
string.  If the encoding is set to be @code{ucs-2} or
@code{ucs-4} @var{word} is expected to be a cast
from either @code{const u16int *} or @code{const u32int *}
respectively.  @code{u16int} and @code{u32int} are generally
@code{unsigned short} and @code{unsigned int} respectively.
@var{size} is the length of the string or @code{-1} if the string
is null terminated.  If the string is a cast from @code{const u16int
*} or @code{const u32int *} then @code{@i{size}} is the amount of
space in bytes the string takes up after being cast to @code{const
char *} and not the true size of the string.  @code{sspell_speller_check}
will return @code{0} if it is not found and non-zero otherwise.

If the word is not correct, then the @code{suggest} method can be used
to come up with likely replacements.

@smallexample
AspellWordList * suggestions = aspell_speller_suggest(spell_checker,
                                                      @var{word}, @var{size});
AspellStringEnumeration * elements = aspell_word_list_elements(suggestions);
const char * word;
while ( (word = aspell_string_enumeration_next(aspell_elements)) != NULL )
@{
  // add to suggestion list
@}
delete_aspell_string_enumeration(elements);
@end smallexample

Notice how @code{elements} is deleted but @code{suggestions} is not.
The value returned by @code{suggestions} is only valid to the next
call to @code{suggest}.  Once a replacement is made the
@code{store_repl} method should be used to communicate the replacement
pair back to the spell checker (for the reason, @pxref{Notes on
Storing Replacement Pairs}).  Its usage is as follows:

@smallexample
aspell_speller_store_repl(spell_checker, @var{misspelled_word}, @var{size},
                          @var{correctly_spelled_word}, @var{size});
@end smallexample

If the user decided to add the word to the session or personal
dictionary the the word can be be added using the
@code{add_to_session} or @code{add_to_personal} methods respectively
like so:

@smallexample
aspell_speller_add_to_session|personal(spell_checker, @i{word}, @i{size});
@end smallexample

It is better to let the spell checker manage these words rather than
doing it yourself so that the words have a chance of appearing in the
suggestion list.

Finally, when the document is closed the @code{AspellSpeller} class
should be deleted like so:

@smallexample
delete_aspell_speller(spell_checker);
@end smallexample

@subsection API Reference

Methods that return a boolean result generally return @code{false} on
error and @code{true} otherwise.  To find out what went wrong use the
@code{error_number} and @code{error_message} methods.  Unless
otherwise stated methods that return a @code{const char *} will return
@code{NULL} on error.  In general, the character string returned is
only valid until the next method which returns a @code{const char *}
is called.

For the details of the various classes please see the header files.
In the future I will generate class references using some automated
tool.

@subsection Examples

Two simple examples are included in the examples directory.  The
@code{example-c} program demonstrates most of the Aspell library
functionality and the @code{list-dicts} lists the available
dictionaries.

@subsection Notes About Thread Safety

Aspell should be thread safe, when used properly, as long as the
underlying compiler, C and C++ library is thread safe.  Aspell
objects, including the AspellSpeller class, should not be used by
multiple threads unless they are protected by locks or it is only
accessed by read-only methods.  A method is read-only only if a
@code{const} object is passed in.  Many methods that seam to be
read-only are not because they may store state information in the
object.

@node Through A Pipe
@section Through A Pipe

When given the @command{pipe} or @command{-a} command, Aspell goes
into a pipe mode that is compatible with @command{ispell -a}.  Aspell
also defines its own set of extensions to Ispell pipe mode.

@subsection Format of the Data Stream

In this mode, Aspell prints a one-line version identification message,
and then begins reading lines of input.  For each input line, a single
line is written to the standard output for each word checked for
spelling on the line.  If the word was found in the main dictionary,
or your personal dictionary, then the line contains only a @samp{*}.

If the word is not in the dictionary, but there are suggestions, then
the line contains an @samp{&}, a space, the misspelled word, a space,
the number of near misses, the number of characters between the
beginning of the line and the beginning of the misspelled word, a
colon, another space, and a list of the suggestions separated by
commas and spaces.

Finally, if the word does not appear in the dictionary, and there are
no suggestions, then the line contains a @samp{#}, a space, the
misspelled word, a space, and the character offset from the beginning
of the line.  Each sentence of text input is terminated with an
additional blank line, indicating that Aspell has completed processing
the input line.

These output lines can be summarized as follows:

@example
@strong{OK}: *
@strong{Suggestions}: & @i{original} @i{count} @i{offset}: @i{miss}, @i{miss}, @dots{}
@strong{None}: # @i{original} @i{offset}
@end example

When in the @option{-a} mode, Aspell will also accept lines of single
words prefixed with any of @samp{*}, @samp{&}, @samp{@@},
@samp{+}, @samp{-}, @samp{~}, @samp{#}, @samp{!}, @samp{%}, or
@samp{^}.  A line starting with @samp{*} tells Aspell to insert the
word into the user's dictionary.  A line starting with @samp{&} tells
Aspell to insert an all-lowercase version of the word into the user's
dictionary.  A line starting with @samp{@@} causes Aspell to accept
this word in the future.  A line starting with @samp{+}, followed
immediately by a valid mode will cause Aspell to parse future input
according the syntax of that formatter.  A line consisting solely of a
@samp{+} will place Aspell in @TeX{}/LaTeX mode (similar to the
@option{-t} option) and @samp{-} returns Aspell to its default mode
(which is Nroff unless otherwise specified).
(but these commands are obsolete).  A line @samp{~}, is ignored for
Ispell compatibility.  A line prefixed with @samp{#} will cause the
personal dictionaries to be saved.  A line prefixed with @samp{!}
will turn on terse mode (see below), and a line prefixed with
@samp{%} will return Aspell to normal (non-terse) mode.  Any input
following the prefix characters @samp{+}, @samp{-}, @samp{#},
@samp{!}, @samp{~}, or @samp{%} is ignored, as is any input
following.  To allow spell-checking of lines beginning with these
characters, a line starting with @samp{^} has that character removed
before it is passed to the spell-checking code.  It is recommended
that programmatic interfaces prefix every data line with an uparrow to
protect themselves against future changes in Aspell.

To summarize these:

@multitable @columnfractions .1 .9
@item @kbd{*@var{word}}
 @tab Add a word to the personal dictionary
@item @kbd{&@var{word}}
 @tab Insert the all-lowercase version of the word in the personal dictionary
@item @kbd{@@@var{word}}
 @tab Accept the word, but leave it out of the dictionary
@item @kbd{#}
 @tab Save the current personal dictionary
@item @kbd{~} @tab Ignored for Ispell compatibility.
@item @kbd{+} @tab Enter @TeX{} mode.
@item @kbd{+@var{mode}} @tab Enter the mode specified by @var{mode}.
@item @kbd{-} @tab Enter the default mode.
@item @kbd{!} @tab Enter terse mode
@item @kbd{%} @tab Exit terse mode
@item @kbd{^} @tab Spell-check the rest of the line
@end multitable

In terse mode, Aspell will not print lines beginning with @samp{*},
which indicate correct words.  This significantly improves running
speed when the driving program is going to ignore correct words
anyway.

In addition to the above commands which are designed for Ispell
compatibility Aspell also supports its own extensions.  All Aspell
extensions follow the following format.

@example
$$@var{command} [@var{data}]
@end example

Where @var{data} may or may not be required depending on the
particular command.  Aspell currently supports the following commands:

@multitable @columnfractions .33 .67
@item @code{cs @var{option},@var{value}} @tab Change a configuration option.
@item @code{cr @var{option}} @tab Prints the value of a configuration option.
@item @code{pp}
 @tab Returns a list of all words in the current personal wordlist.
@item @code{ps}
 @tab Returns a list of all words in the current session dictionary.
@item @code{l} @tab Returns the current language name.
@item @code{ra @var{mis},@var{cor}} @tab Add the word pair to the
 replacement dictionary for later use. Returns nothing.
@end multitable

Anything returned is returned on its own line.  All lists returned
have the following format

@example
@i{num of items}: @i{item1}, @i{item2}, @i{etc}
@end example

@c FIXME: Add note about byte-offset option.

@emph{(Part of the preceding section was directly copied out of the
Ispell manual)}

@node Notes on Storing Replacement Pairs
@section Notes on Storing Replacement Pairs

The @code{store_repl} method and the @code{$$ra} should be used
because Aspell is able to learn from users misspellings.  For example
on the first pass a user misspells @emph{beginning} as @emph{beging}
so Aspell suggests:

@example
begging, begin, being, Beijing, bagging, @dots{}.
@end example

@noindent
However the user then tries @emph{begning} and Aspell suggests

@example
beginning, beaning, begging, @dots{}
@end example

@noindent
so the user selects @emph{beginning}.  However, later on in the
document the user misspells it as @emph{begng} (@strong{not}
@emph{beging}).  Normally Aspell will suggest.

@example
began, begging, begin, begun, @dots{}
@end example

@noindent
However because it knows the user misspelled @emph{beginning} as
@emph{beging} it will instead suggest:

@example
beginning, began, begging, begin, begun @dots{}
@end example

I myself often misspelled beginning (and still do) as something close
to begging and two many times wind up writing sentences such as
"begging with @dots{}".

Please also note that replacements commands has a memory.  Which means
if you first store the replacement pair:

@example
sicolagest -> psycolagest
@end example

@noindent
then store the replacement pair

@example
psycolagest -> psychologist
@end example

@noindent
The replacement pair

@example
sicolagest -> psychologist
@end example

@noindent
will also get stored so that you don't have to worry about it.

@node Adding Support For Other Languages
@chapter Adding Support For Other Languages

Before you consider adding support for Aspell, first make sure that
someone else has not already done it.  A good number of dictionaries
off the Aspell home page at @uref{http://aspell.net}.  If your
language is not listed above please send me a note and I will work
with you on adding support.

Adding a language to Aspell is fairly straightforward.  You basically
need to create the language data file, and compile a new word list.

@menu
* The Language Data File::      
* Compiling the Word List::     
* Phonetic Code::               
* The Simple Soundslike::       
* Replacement Tables::          
* Affix Compression::           
* Controlling the Behavior of Run-together Words::  
* Creating A New Character Set::  
* Format of the Character Set Data File::  
* Creating An Official Dictionary Package::  
@end menu

@node The Language Data File
@section The Language Data File

The basic format of the language data file is the same as it is for the
Aspell configuration file.  It is named @file{@var{lang}.dat} and is
located in the architecture independent data dir for Aspell (option
@option{data-dir}) which is usually @file{@var{prefix}/share/aspell}.
Use @command{aspell config} to find out where it is in your
installation.  By convention the language name should be the two letter
ISO 639 language code if it exists, if not use the three letter code.

The language data file has several mandatory fields, and several
optional ones.  All fields are case sensitive and should be in all
lower case.

The two mandatory fields are @option{name} and @option{charset}.

@option{name} is the name of the language and should be the same as the
file name (without the @file{.dat}).

@option{charset} is the 8-bit character set Aspell will expect the
word lists to be formatted in.  If possible choose from one of the
standard ones provided with Aspell.  These are @samp{iso-8859-*},
@samp{koi8-*}, or @samp{viscii}.  If your language does not require
any non-ascii characters choose @samp{iso-8859-1}.  If one of these
standard character sets is not suitable for your language than you can
create a new one.  @xref{Creating A New Character Set}.

The optional fields are as follows:

@table @option

@anchor{data-encoding}
@item data-encoding

The encoding the language data files are expected to be in as well as
the default encoding to use when saving the personal dictionaries.  It
can be either @samp{utf-8} or any of the 8-bit encoding that Aspell
supports.  If not set, then it defaults to @option{charset}.

@item special

Non-letter characters that can appear in your language such as the
@samp{'} and @samp{-}. The format for the value is a list separated by
spaces.  Each item of the list has the following format.

@example
<char> <begin><middle><end>
@end example

@var{char} is the non letter character in question.
@var{begin}, @var{middle}, @var{end} are either a
@samp{-} or a @samp{*}.  A star for @var{begin} means that the
character can begin a word, a @samp{-} means it can't.  The same is
true for @var{middle} and @var{end}. For example, the entry
for the @samp{'} in English is:

@example
' -*-
@end example

To include more than one middle character just list them one after
another on the same line.  For example, to make both the @samp{'}
and the @samp{-} a middle character, use the following line in the
language data file:

@example
special ' -*- - -*-
@end example

@item soundslike

The name of the soundslike data for the language.  The data is
expected to be in the file @file{@var{name}_phonetic.dat}.

If @var{name} is @samp{simpile} then a very simple soundslike is
used.  This is nearly as powerful as full phonetic soundslike but it
can be computed a lot faster.  (@pxref{The Simple Soundslike})

If the soundslike name is @samp{none}, or this option is not specified,
than no soundslike will be used.  The effective soundslike is the word
converted to all lowercase and possibly with accents stripped
depending on the @option{store-as} option.  For languages with
phonetic spelling the difference will not be very noticeable.
However, for languages with non-phonetic spelling there will be a
noticeable difference.  The difference you notice will depend on the
quality of the soundslike data file.  If you do not notice much of a
difference for a language with non-phonetic spelling that is a good
indication that the soundslike data is not rough enough---or the words
you are trying are not that badly misspelled.

@item invisible-soundslike

Avoid storing the soundslike information with the word.  Instead it is
computed as needed.  This option defaults to true if the soundslike is
@samp{none} or @samp{simpile}, and false when a phonetic soundslike is
used.

@item repl-table

@xref{Replacement Tables}.

@item keyboard

The base name of the keyboard definition file to use.  For more
information see @ref{Notes on Typo-Analysis}.

@item sug-split-char

A list of characters which specifies which characters to insert between
two words when a word is split.  This is a list option.

@item affix
@itemx affix-compress
@itemx partially-expand

@xref{Affix Compression}.

@item store-as

How the words are indexed in the dictionary.  If "stripped" then the
word is indexed in a lower case and de-accented form.  If "lower", then
the word is indexed in a lower case form but with accent info still
intact.  This just controls how the word is indexed, not how it is
stored.  The default is "stripped" unless affix compression is used.

@c @item ignore-accents

@c @item affix-char

@c Unimplemented

@c @item flag-char

@c Unimplemented

@item norm-required

Should be set to true if your language makes use of private use
characters or when Normalization Form C is not the same as
full composition.

@item normalize

@item norm-form

@end table

Additional options includes options to control how run-together words
are handled the same way as they are in the normal configuration
files.  for more information, please @ref{Controlling the Behavior of
Run-together Words}.

@node Compiling the Word List
@section Compiling the Word List

Once you have a working language data file installed in the right
place you are ready to compile the main word list.  To find out what
to do, see @ref{Working With Dictionaries}.  This section also
includes instructions for creating the AWLI file.

@node Phonetic Code
@section Phonetic Code

@c @emph{(The following section was originally written by Bj@"orn Jacke,
@c bjoern.jacke at gmx de)}

Aspell is in fact the spell checker that comes up with the best
suggestions if it finds an unknown word.  One reason is that it does
not just compare the word with other words in the dictionary (like
Ispell does) but also uses phonetic comparisons with other words.

The new table driven phonetic code is very flexible and setting up
phonetic transformation rules for other languages is not difficult but
there can be a number of stumbling blocks --- that's why I wrote this
section.

The main phonetic code is free of any language specific code and
should be powerful enough to allow setting up rules for any language.
Anything which is language specific is kept in a plain text file and
can easily be edited.  So it's even possible to write phonetic
transformation rules if you don't have any programming skills.  All
you need to know is how words of the language are written and how they
are pronounced.

@subsection Syntax of the transformation array

In the translation array there are two strings on each line; the first
one is the search string (or switch name) and the second one is the
replacement string (or switch parameter).  The line

@example
version   @var{version}
@end example

@noindent
is also required to appear somewhere in the translation array.  The
version string can be anything but it should be changed whenever a new
version of the translation array is released.  This is important
because it will keep Aspell from using a compiled dictionary with the
wrong set of rules.  For example, if when coming up with suggestion
for @code{hallo}, Aspell will use the new rules to come up with the
soundslike say @code{H*L*}, but if @samp{hello} is stored in the
dictionary using the old rules as @code{HL} instead of @code{H*L*}
Aspell will never be able to come up with @samp{hello}.  So to solve
this problem Aspell checks if the version strings match and aborts
with an error if they don't.  Thus it is important to update it
whenever a new version of the translation array is released.  This is
only a problem with the main word list as the personal word lists are
now stored as simple word lists with a single header line (i.e. no
soundslike data).

Each non switch line represents one replacement (transformation) rule.
Words beginning with the same letter must be grouped together; the
order inside this group does not depend on alphabetical issues but it
gives priorities; the higher the rule the higher the priority.  That's
why the first rule that matches is applied.  In the following example:

@example
GH   _
G    K
@end example

@noindent
@samp{GH -> _} has higher priority than @samp{G -> K}

@samp{_} represents the empty string ``''.  If @samp{GH -> _} came
after @samp{G -> K}, the second rule would never match because the
algorithm would stop searching for more rules after the first match.
The above rules transform any @samp{GH} to an empty string (delete
them) and transforms any other @samp{G} to @samp{K}.

At the end of the first string of a line (the search string) there may
optionally stand a number of characters in brackets.  One (only one!)
of these characters must fit.  It's comparable with the @samp{[ ]}
brackets in regular expressions.  The rule @samp{DG(EIY) -> J} for
example would match any @samp{DGE}, @samp{DGI} and
@samp{DGY} and replace them with @samp{J}.  This way you can
reduce several rules to one.

Before the search string, one or more dashes @samp{-} may be placed.
Those search strings will be matched totally but only the beginning of
the string will be replaced.  Furthermore, for these rules no follow-up
rule will be searched (what this is will be explained later).  The
rule @samp{TCH-- }-> _ will match any word containing
@samp{TCH} (like @samp{match}) but will only replace the first
character @samp{T} with an empty string.  The number of dashes
determines how many characters from the end will not be replaced.
After the replacement, the search for transformation rules continues
with the not replaced @samp{CH}!

If a @samp{<} is appended to the search string, the search for
replacement rules will continue with the replacement string and not with
the next character of the word.  The rule @samp{PH< -> F} for example
would replace @samp{PH} with @samp{F} and then again start to search for
a replacement rule for @samp{F@dots{}}.  If there would also be rules
like @samp{FO }-> @samp{O} and @samp{F -> _} then words like
@samp{PHOXYZ} would be transformed to @samp{OXYZ} and any occurrences of
@samp{PH} that are not followed by an @samp{O} will be deleted like
@samp{PHIXYZ -> IXYZ}.  The second replacement however is not applied if
the priority of this rule is lower than the priority of the first rule.

Priorities are added to a rule by putting a number between 0 and 9 at
the end of the search string, for example @samp{ING6 -> N}.
The higher the number the higher is the priority.

Priorities are especially important for the previously mentioned
follow-up rules.  Follow-up rules are searched beginning from the last
string of the first search string.  This is a bit complicated but I
hope this example will make it clearer:

@example
CHS      X
CH       G

HAU--1   H

SCH      SH
@end example

In this example @samp{CHS} in the word @samp{FUCHS} would be
transformed to @samp{X}.  If we take the word @samp{DURCHSCHNITT} then
things look a bit different.  Here @samp{CH} belongs together and
@samp{SCH} belongs together and both are spoken separately.  The
algorithm however first finds the string @samp{CHS} which may not be
transformed like in the previous word @samp{FUCHS}.  At this point the
algorithm can find a follow up rule.  It takes the last character of
the first matching rule (@samp{CHS}) which is @samp{S} and looks for
the next match, beginning from this character.  What it finds is
clear: It finds @samp{SCH -> SH}, which has the same priority
(no priority means standard priority, which is 5).  If the priority is
the same or higher the follow-up rule will be applied.  Let's take a
look at the word @samp{SCHAUKEL}.  In this word @samp{SCH} belongs
together and may not be taken apart.  After the algorithm has found
@samp{SCH }-> @samp{SH} it searches for a follow-up rule for
@samp{H+}@samp{AUKEL}.  It finds @samp{HAU--1 -> H}, but does not
apply it because its priority is lower than the one of the first rule.
You see that this is a very powerful feature but it also can easily
lead to mistakes.  If you really don't need this feature you can turn
it off by putting the line:

@example
followup      0
@end example

@noindent
at the beginning of the phonetic table file.  As mentioned, for rules
containing a @samp{-} no follow-up rules are searched but giving such
rules a priority is not totally senseless because they can be
follow-up rules and in that case the priority makes sense again.
Follow-up rules of follow-up rules are not searched because this is in
fact not needed very often.

The control character @samp{^} says that the search string only
matches at the beginning of words so that the rule @samp{RH -> R} will
only apply to words like @samp{RHESUS} but not @samp{PERHAPS}.  You
can append another @samp{^} to the search string.  In that case the
algorithm treats the rest of the word totally separately from the
first matched string at the beginning.  This is useful for prefixes
whose pronunciation does not depend on the rest of the word and vice
versa like @samp{OVER^^} in English for example.

The same way as @samp{^} works does @samp{$} only apply to words
that end with the search string.  @samp{GN$ -> N} only
matches on words like @samp{SIGN} but not @samp{SIGNUM}.  If
you use @samp{^} and @samp{$} together, both of them must fit
@samp{ENOUGH^$ -> NF} will only match the word
@samp{ENOUGH} and nothing else.

Of course you can combine all of the mentioned control characters but
they must occur in this order: @samp{< - priority ^ $}.  All
characters must be written in CAPITAL letters.

If absolutely no rule can be found --- might happen if you use strange
characters for which you don't have any replacement rule --- the next
character will simply be skipped and the search for replacement rules
will continue with the rest of the word.

If you want double letters to be reduced to one you must set up a rule
like @samp{LL- -> L}.  If double letters in the resulting phonetic
word should be allowed, you must place the line:

@example
collapse_result     0
@end example

@noindent
at the beginning of your transformation table file; otherwise set the
value to `1'.  The English rules for example strip all vowels from
words and so the word "GOGO" would be transformed to "K" and not to
"KK" (as desired) if @code{collapse_result} is set to 1.  That's why
the English rules have @code{collapse_result} set to @code{0}.

By default, all accents are removed from a word before it is matched to
the soundslike rules.  If you do not want this then add the line

@example
remove_accents      0
@end example

at the beginning of your file.  The exact definition of an accent is
language dependent and is controlled via the character set file.  If you
set remove_accents to '0' then you should also set "store-as" to "lower"
in the language data file (not the phonetic transformation file)
otherwise Aspell will have problems when both the accented and the
de-accented version of a word appearing in the dictionary; it will
consider one of them as incorrectly spelled.

@subsection How do I start finally?

Before you start to write an array of transformation rules, you should
be aware that you have to do some work to make sure that things you do
will result in correct transformation rules.

@subsubsection Things that come in handy

First of all, you need to have a large word list of the language you
want to make phonetics for.  It should contain about as many words as
the dictionary of the spell checker.  If you don't have such a list,
you will probably find an Ispell dictionary at
@uref{http://fmg-www.cs.ucla.edu/geoff/ispell-dictionaries.html} which
will help you.  You can then make affix expansion via @command{ispell
-e} and then pipe it through @command{tr " " "\n"} to put one word on
each line.  After that you eventually have to convert special
characters like @samp{@'e} from Ispell's internal representation to
latin1 encoding.  @command{sed s/e'/@'e/g} for example would replace
all @samp{e'} with @samp{@'e}.

The second is that you know how to use regular expressions and know
how to use @command{grep}.  You should for example know that:

@example
grep ^[^aeiou]qu[io] wordlist | less
@end example

@noindent
will show you all words that begin with any character but @samp{a},
@samp{e}, @samp{i}, @samp{o} or @samp{u} and then continue with
@samp{qui} or @samp{quo}.  This stuff is important for example to
find out if a phonetic replacement rule you want to set up is valid
for all words which match the expression you want to replace.  Taking
a look at the regex(7) man page is a good idea.

@subsubsection What the phonetic code should do

Normal text comparison works well as long as the typer misspells a
word because he pressed one key he didn't really want to press.  In
these cases, mostly one character differs from the original word.

In cases where the writer didn't know about the correct spelling of
the word, the word may have several characters that differ from the
original word but usually the word would still sound like the
original.  Someone might think that `tough' is spelled `taff'.  No
spell checker without phonetic code will come to the idea that this
might be `tough', but a spell checker who knows that `taff' would be
pronounced like `tough' will make good suggestions to the user.  Another
example could be `funetik' and `phonetic'.

From these examples you can see that the phonetic transformation should
not be too fussy and too precise.  If you implement a whole phonetic
dictionary as you can find it in books this will not be very useful
because then there could still be many characters differing from the
misspelled and the desired word.  What you should do if you implement
the phonetic transformation table is to reduce the number of used
letters to the only really necessary ones.

Characters that sound similar should be reduced to one.  In English
language for example `Z' sounds like `S' and that's why the
transformation rule @samp{Z -> S} is present in the
replacement table.  ``PH is spoken like ``F and so we have a
@samp{PH -> F} rule.

If you take a closer look you will even see that vowels sound very
similar in the English language: `contradiction', `cuntradiction',
`cantradiction' or `centradiction' in fact sound nearly the same,
don't they? Therefore the English phonetic replacement rules not only
reduce all vowels to one but even remove them all (removing is done by
just setting up no rule for those letters).  The phonetic code of
``contradiction'' is ``KNTRTKXN'' and if you try to read this
letter-monster loud you will hear that it still sound a bit like
`contradiction'.  You also see that ``D'' is transformed to ``T''
because they nearly sound the same.

If you think you have found a regularity you should @emph{always} take
your word list and @command{grep} for the corresponding regular
expression you want to make a transformation rule for.  An example: If
you come to the idea that all English words ending on `ough' sound
like `AF' at the end because you think of `enough' and `tough'.  If
you then @code{grep} for the corresponding regular expression by
@command{grep -i ough$ wordlist} you will see that the rule you wanted
to set up is not correct because the rule doesn't fit to words like
`although' or `bough'.  So you have to define your rule more precisely
or you have to set up exceptions if the number of words that differ
from the desired rule is not too big.

Don't forget about follow-up rules which can help in many cases but
which also can lead to confusion and unwanted side effects.  It's also
important to write exceptions in front of the more general rules
(@samp{GH} before @samp{G} etc.).

If you think you have set up a number of rules that may produce some
good results try them out! If you run Aspell as @command{aspell
--lang=@var{your_language} pipe} you get a prompt at which you can type
in words.  If you just type words Aspell checks them and eventually
makes suggestions if they are misspelled.  If you type in @code{$$Sw
@var{word}} you will see the phonetic transformation and you can test
out if your work does what you want.

Another good way to check that changes you make to your rules don't
have any bad side effects is to create another list from your word
list which contains not only the word of the word list but also the
corresponding phonetic version of this word on the same line.  If you
do this once before the change and once after the change you can make
a diff (see @command{man diff}) to see what @emph{really} changed.  To
do this use the command @command{aspell --lang=@var{your_language}
soundslike}.  In this mode aspell will output the the original word
and then its soundslike separated by a tab character for each word you
give it.  If you are interested in seeing how the algorithm works you
can download a set of useful programs from
@uref{http://members.xoom.com/maccy/spell/phonet-utils.tar.gz}.  This
includes a program that produces a list as mentioned above and another
program which illustrates how the algorithm works.  It uses the same
transformation table as Aspell and so it helps a lot during the
process of creating a phonetic transformation table for Aspell.

During your work you should write down your basic ideas so that other
people are able to understand what you did (and you still know about
it after a few weeks).  The English table has a huge documentation
appended as an example.

Now you can start experimenting with all the things you just read and
perhaps set up a nice phonetic transformation table for your language
to help Aspell to come up with the best correction suggestions ever
seen also for your language.  Take a look at the Aspell homepage to
see if there is already a transformation table for your language.  If
there is one you might also take a look at it to see if it could be
improved.

If you think that this section helped you or if you think that this is
just a waste of time you can send any feedback to
@email{bjoern.jacke@@gmx.de}.

@node The Simple Soundslike
@section The Simple Soundslike

The simple soundslike goes something like this:

@example
sl0[0] = lookup0(word[0])
for (i = 1; i < size; i++)
  sl0[i] = lookup(word[i]);
s = 0;
for (i = 0; i < size; i++)
  sl.append(al0[i]) unless sl0[i] == 0 || sl0[i] == sl0[i-1];
@end example

Basically each character can be converted to another character or deleted.
A separate lookup table is used for the first character.  If the same
soundslike letter is repeated, the duplicate is removed.

By default all accents are removed, and all vowels are deleted unless
they appear at the start of the word in which case they are converted
to a @samp{*}.  The exact behavior can be customized via the character
data file.

The simplified soundslike has the advantage that it is very fast to
compute and thus does not need to be stored with a word.  Also, when
affix compression is used and the @option{partially-expand} is given
the results will be identical to the results when affix compression is
not used.

Of course it is not nearly as powerful as the phonetic soundslike.

@node Replacement Tables
@section Replacement Tables

When phonetic code is not used a replacement table can be used instead.
To enable the use of a replacement table add the line @code{repl-table
@var{lang}}, in which case the replacement table is excepted to be in
the file @file{@var{lang}_repl.dat}.  A complete file name can also be
specified in place of @var{lang}.  For compatibility with MySpell the
replacement table can also be part of the affix file, in which case
@option{repl-table} will be @file{@var{lang}_affix.dat"}.

Replacement table syntax:

@example
REP [number_of_replacement_definitions]
REP [what] [replacement]
REP [what] [replacement]
@end example

For example a possible English replacement table definition
to handle misspelled consonants:

@example
REP 8
REP f ph
REP ph f
REP f gh
REP gh f
REP j dg
REP dg j
REP k ch
REP ch k
@end example

@node Affix Compression
@section Affix Compression

Aspell, as of version 0.60, now has support for affix compression.
The codebase comes from MySpell found in OpenOffice.

To add support for affix compression add the following lines to the
language data file.

@example
affix          @var{lang}
affix-compress true
@end example

The line @samp{affix @var{lang}} adds support for recognizing affix
information, and the line @samp{affix-compress true} enables affix
compression.

The affix file is expected to be named @file{@var{lang}_affix.dat}.  It is
the exact same format as those used by MySpell.  More information can
be found in the myspell/ directory of the distribution or at
@uref{http://lingucomponent.openoffice.org/dictionary.html}.

Affix compression can also be used with soundslike lookup.  Aspell
does this by only storing the soundslike for the root word.  When a
word is misspelled it will search for a soundslike close to all
possible roots of the misspelled word.

When no soundslike information, or the simple soundslike, is used it
may be beneficial to specify the option @option{partially-expand}
which will partially expand a word with affix information so that the
affix flags do not effect the first 3 letters of the word.  This will
allow Aspell to get more accurate results when scanning the list for near
misses since the full word can be used and not just the root.
Specifying this option, however, will also effectively expand any
prefixes.  Thus this option should not be used for prefix heavy
languages such as Hebrew.

An existing word list, without affix info, can be affix compressed
using using @command{aspell munch-list}.

@subsection Format of the Affix File

@c (as written in affix.readme)

An affix is either a  prefix or a suffix attached to root words to make
other words.  For example supply -> supplied by dropping the "y" and
adding an "ied" (the suffix).

Here is an example of how to define one specific suffix borrowed
from the English affix file.

@example
SFX D Y 4
SFX D   0     d          e
SFX D   y     ied        [^aeiou]y
SFX D   0     ed         [^ey]
SFX D   0     ed         [aeiou]y
@end example

This file is space delimited and case sensitive.  So this information
can be interpreted as follows:

The first line has 4 fields:

@multitable @columnfractions .05 .15 .80
@item
1 @tab @t{SFX} @tab indicates this is a suffix
@item
2 @tab @t{D} @tab is the name of the character which represents this suffix
@item
3 @tab @t{Y} @tab indicates it can be combined with prefixes (cross product)
@item
4 @tab @t{4} @tab indicates that sequence of 4 affix entries are needed to
properly store the affix information
@end multitable

The remaining lines describe the unique information for the 4 affix
entries that make up this affix.  Each line can be interpreted
as follows: (note fields 1 and 2 are used as a check against line 1 info)

@multitable @columnfractions .05 .15 .80
@item
1 @tab @t{SFX} @tab indicates this is a suffix
@item
2 @tab @t{D} @tab is the name of the character which represents this affix
@item
3 @tab @t{y} @tab the string of chars to strip off before adding affix (a 0
here indicates the NULL string)
@item
4 @tab @t{ied} @tab the string of affix characters to add (a 0 here
indicates the NULL string)
@item
5 @tab @t{[^aeiou]y} @tab the conditions which must be met before the affix
can be applied
@end multitable

Field 5 is interesting.  Since this is a suffix, field 5 tells us that
there are 2 conditions that must be met.  The first condition is that
the next to the last character in the word must @emph{not} be any of the
following "a", "e", "i", "o" or "u".  The second condition is that
the last character of the word must end in "y".

@subsection When Compared With Ispell

Now for comparison purposes, here is the same information from the
Ispell @file{english.aff} compression file which was used as the basis
for the OOo one.

@example
flag *D:
    E           >       D               # As in create > created
    [^AEIOU]Y   >       -Y,IED          # As in imply > implied
    [^EY]       >       ED              # As in cross > crossed
    [AEIOU]Y    >       ED              # As in convey > conveyed
@end example

The Ispell information has exactly the same information but in a
slightly different (case-insensitive) format:

Here are the ways to see the mapping from Ispell .aff format to our
OOo format.

@enumerate
@item
The ispell english.aff has flag D under the "suffix" section so
you know it is a suffix.

@item
The D is the character assigned to this suffix

@item
@samp{*} indicates that it can be combined with prefixes

@item
Each line following the : describes the affix entries needed
to define this suffix

@itemize @bullet
@item
The first field is the conditions that must be met.
@item
The second field is after the > if a "-" occurs is the string to strip
off (can be blank).
@item
The third field is the string to add (the affix)
@end itemize
@end enumerate

In addition all chars in ispell aff files are in uppercase.

@subsection Specifying Affix Flags

Affix flags are specified in the word list by specifying them after
the @samp{/} character:

@example
@var{word}/@var{flags}
@end example

For example:

@example
create/DG
@end example

@noindent
will associate the @samp{D} and @samp{G} flag with the word create.

@node Controlling the Behavior of Run-together Words
@section Controlling the Behavior of Run-together Words

Aspell currently has support for unconditionally accepting run-together
words.

Support for unconditionally accepting run-together words can either be
turned on in the language data file or as a normal option via the
@option{run-together} option.  The @option{run-together-limit} options
controls the maximum number of words that can be strung together, the
default is normally 2.  The @option{run-together-min} options
controls the minimum length of the individual components of the run
together word, the default is normally 3.  Both the
@option{run-together-limit} and @option{run-together-min} option may
be specified in both the language data file or as a normal option.

@c FIXME: Add note about compound word support when suggesting.

@node Creating A New Character Set
@section Creating A New Character Set

If there is not a standard character set for your language than you
can invent one.  The new charset will only be used by Aspell
internally.  If the option @option{data-encoding} is set to
@samp{utf-8}, and your current locale character type is always set to
@samp{utf-8}, than you can use UTF-8 for everything and not
worry yourself that an 8-bit character set is being used internally.
If your language has no more than 210 distinct symbols, including
different capitalizations and accents, than Aspell can support it.

In order to create a new character set you will need to download
the file @file{aspell-lang-@var{version}.tar.bz2} from
@uref{ftp://ftp.gnu.org/@/gnu/aspell/} and follow
the instructions in the @file{README} file.  Version 0.1, and 0.2 of
mkchardata @emph{will not} work as the format of the character data file
has changed.

@node Format of the Character Set Data File
@section Format of the Character Set Data File

The character set data files that Aspell uses can be fine tuned.  If
this is done, that character set file should be renamed to
@file{@var{lang}.cset}.

The line(s) starting with @samp{=} do not specify the exact name of the
file, but rather the name of the Unicode mapping that is used.  Unless
you change the Unicode mapping, this line should not be changed.

@node Creating An Official Dictionary Package
@section Creating An Official Dictionary Package

Once you have a basic dictionary working, you should consider creating
an official package so that it can be distributed with Aspell.  To do so
download
@url{ftp://alpha.gnu.org/@/gnu/aspell/@/aspell-lang-@var{version}.tar.bz2}.

@node Implementation Notes
@appendix Implementation Notes

@menu
* Aspell Suggestion Strategy::  
* Notes on 8-bit Characters::   
@end menu

@node Aspell Suggestion Strategy
@appendixsec Aspell Suggestion Strategy

The magic behind my spell checker comes from merging Lawrence Philips
excellent metaphone algorithm and Ispell's near miss strategy which is
inserting a space or hyphen, interchanging two adjacent letters,
changing one letter, deleting a letter, or adding a letter.

The process goes something like this.

@enumerate
@item
Convert the misspelled word to its soundslike equivalent (its
metaphone for English words).

@item
Find all words that have a soundslike within one or two edit distances
from the original word's soundslike.  The edit distance is the total
number of deletions, insertions, exchanges, or adjacent swaps needed
to make one string equivalent to the other.  When set to only look for
soundslikes within one edit distance it tries all possible soundslike
combinations and checks if each one is in the dictionary.  When set to
find all soundslike within two edit distances it scans through the
entire dictionary and quickly scores each soundslike.  The scoring is
quick because it will give up if the two soundslikes are more than two
edit distances apart.

@item
Find misspelled words that have a correctly spelled replacement by the
same criteria of step number 2 and 3.  That is the misspelled word in
the word pair (such as ``teh -> the'') would appear in the suggestions
list as if it was a correct spelling.

@item
Score the result list and return the words with the lowest score.  The
score is roughly the weighed average of the weighed edit distance of
the word to the misspelled word and the soundslike equivalent of the
two words.  The weighted edit distance is like the edit distance
except that the various edits have weights attached to them.

@item
Replace the misspelled words that have correctly spelled replacements
with their replacements and remove any duplicates that might arise
because of this.
@end enumerate

Please note that the soundslike equivalent is a rough approximation of
how the words sounds.  It is not the phoneme of the word by any means.
For more details about exactly how each step is performed please see
the file @file{suggest.cc}.  For more information on the metaphone
algorithm please see the data file @file{english_phonet.dat}.

@node Notes on 8-bit Characters
@appendixsec Notes on 8-bit Characters

There is a very good reason I use 8-bit characters in Aspell. Speed
and simplicity. While many parts of my code can fairly easily be
converted to some sort of wide character as my code is clean. Other
parts can not be.

One of the reasons because is many, many places I use a direct lookup
to find out various information about characters. With 8-bit
characters this is very feasible because there is only 256 of
them. With 16-bit wide characters this will waste a LOT of space. With
32-bit characters this is just plain impossible. Converting the lookup
tables to another form is certainly possible but degrades performance
significantly.

Furthermore, some of my algorithms relay on words consisting only on a
small number of distinct characters (often around 30 when case and
accents are not considered). When the possible character can consist
of any Unicode character this number becomes several thousand, if
that. In order for these algorithms to still be used, some sort of
limit will need to be placed on the possible characters the word can
contain. If I impose that limit, I might as well use some sort of
8-bit characters set which will automatically place the limit on what
the characters can be.

There is also the issue of how I should store the word lists in
memory? As a string of 32 bit wide characters. Now that is using up 4
times more memory than characters would and for languages that can fit
within an 8-bit character that is, in my view, a gross waste of
memory. So maybe I should store them is some variable width format
such as UTF-8. Unfortunately, way, way too many of the algorithms will
simply not work with variable width characters without significant
modification which will very likely degrade performance. So the
solution is to work with the characters as 32-bit wide characters and
than convert it to a shorter representation when storing them in the
lookup tables. Now that can lead to an inefficiency. I could also use
16 bit wide characters however that may not be good enough to hold all
future versions of Unicode and therefore has the same problems.

As a response to the space waste used by storing word lists in some
sort of wide format some one asked:

@quotation
Since hard drives are cheaper and cheaper, you could store a dictionary
in a usable (uncompressed) form and use it directly with memory
mapping. Then the efficiency would directly depend on the disk caching
method, and only the used part of the dictionaries would really be
loaded into memory. You would no more have to load plain dictionaries
into main memory, you'll just want to compute some indexes (or
something like that) after mapping.
@end quotation

However, the fact of the matter is that most of the dictionary will be
read into memory anyway if it is available. If it is not available
than there would be a good deal of disk swaps. Making characters
32-bit wide will increase the chance that there are more disk swaps.
So the bottom line is that it is more efficient to convert characters
from something like UTF-8 into some sort of 8-bit character. I could
also use some sort of disk space lookup table such as the Berkeley
Database. However this will @strong{definitely} degrade performance.

The bottom line is that keeping Aspell 8-bit internally is a very well
though out decision that is not likely to change any time soon. Feel
free to challenge me on it, but, don't expect me to change my mind
unless you can bring up some point that I have not thought of before
and quite possibly a patch to solve cleanly convert Aspell to Unicode
internally without a serious performance lost OR serious memory usage
increase.

@node Languages Which Aspell can Support
@appendix Languages Which Aspell can Support

Even though Aspell will remain 8-bit internally it should still be
able to support any written languages not based on a logographic
script.  The only logographic writing system in current use are those
based on h@`anzi which includes Chinese, Japanese, and sometimes Korean.

@menu
* Supported::                   
* Unsupported::                 
* Multiple Scripts::            
* Planned Dictionaries::        
* References::                  
@end menu

@node Supported
@appendixsec Supported

Aspell 0.60 should be able to support the following languages:

@include lang-supported.texi

Dictionaries marked as @dfn{0.50} are available for Aspell 0.50.  Ones
marked as @dfn{0.60} are available for Aspell 0.60 only.  Ones marked as
@dfn{Planned} should eventually be available.  Ones marked as
@dfn{Maybe} might be available in the future.
@xref{Planned Dictionaries}, for more info.

@appendixsubsec Notes on Latin Languages

Any word that can be written using one of the Latin ISO-8859 character
sets (ISO-8859-1,2,3,4,9,10,13,14,15,16) can be written, in decomposed
form, using the ASCII characters, the 23 additional letters:

@example
U+00C6 LATIN CAPITAL LETTER AE
U+00D0 LATIN CAPITAL LETTER ETH
U+00D8 LATIN CAPITAL LETTER O WITH STROKE
U+00DE LATIN CAPITAL LETTER THORN
U+00DE LATIN SMALL LETTER THORN
U+00DF LATIN SMALL LETTER SHARP S
U+00E6 LATIN SMALL LETTER AE
U+00F0 LATIN SMALL LETTER ETH
U+00F8 LATIN SMALL LETTER O WITH STROKE
U+0110 LATIN CAPITAL LETTER D WITH STROKE
U+0111 LATIN SMALL LETTER D WITH STROKE
U+0126 LATIN CAPITAL LETTER H WITH STROKE
U+0127 LATIN SMALL LETTER H WITH STROKE
U+0131 LATIN SMALL LETTER DOTLESS I
U+0138 LATIN SMALL LETTER KRA
U+0141 LATIN CAPITAL LETTER L WITH STROKE
U+0142 LATIN SMALL LETTER L WITH STROKE
U+014A LATIN CAPITAL LETTER ENG
U+014B LATIN SMALL LETTER ENG
U+0152 LATIN CAPITAL LIGATURE OE
U+0153 LATIN SMALL LIGATURE OE
U+0166 LATIN CAPITAL LETTER T WITH STROKE
U+0167 LATIN SMALL LETTER T WITH STROKE
@end example

and the 14 modifiers:

@example
U+0300 COMBINING GRAVE ACCENT
U+0301 COMBINING ACUTE ACCENT
U+0302 COMBINING CIRCUMFLEX ACCENT
U+0303 COMBINING TILDE
U+0304 COMBINING MACRON
U+0306 COMBINING BREVE
U+0307 COMBINING DOT ABOVE
U+0308 COMBINING DIAERESIS
U+030A COMBINING RING ABOVE
U+030B COMBINING DOUBLE ACUTE ACCENT
U+030C COMBINING CARON
U+0326 COMBINING COMMA BELOW
U+0327 COMBINING CEDILLA
U+0328 COMBINING OGONEK
@end example

Which is a total of 37 additional Unicode code points.

All ISO-8859 character leaves the characters 0x00 - 0x1F, and 0x80 -
0x9F unmapped as they are generally used as control characters.  Of
those, 0x01 - 0x0F, 0x11 - 0x1F and 0x80 - 0x9F may be mapped to
anything in Aspell.  This is a total of 62 characters which can be
remapped in any ISO-8859 character set.  Thus, by remapping 37 of the 62
characters to the previously specified Unicode code-points, any modified
ISO-8859 character set can be used for any Latin languages covered by
ISO-8859.  Of course decomposing every single accented character wastes
a lot of space, so only characters that cannot be represented in the
precomposed form should be broken up.  By using this trick it is
possible to store foreign words in the correctly accented form in the
dictionary even if the precomposed character is not in the current
character set.

Any letter in the Unicode range U+0000 - U+0249, U+1E00 - U+1EFF (Basic
Latin, Latin-1 Supplement, Latin Extended-A, Latin Extended-B, and Latin
Extended Additional) can be represented using around 175 basic letters,
and 25 modifiers which is less than 210 and can thus fit in an Aspell
8-bit character set.  Since this Unicode range covers any possible Latin
language this special character set can be used to represent any word
written using the Latin script if so desired.

@appendixsubsec Syllabic

Syllabic languages use a separate symbol for each syllable of the
language.  Even thought most of them have more than 210 distinct
symbols Aspell can still support them by breaking them up.

@appendixsubsubsec The Ethiopic Syllabary

Even though the Ethiopic script has more than 210 distinct characters
Aspell can still handle it.  The idea is to split each character into
two parts based on the Consonant and Vowel parts.  This encoding of the
syllabary is far more useful to Aspell than if they were stored in UTF-8
or UTF-16.  In fact, the exiting suggestion strategy of Aspell will work
well with this encoding without any additional modifications.  However,
additional improvements may be possible by taking advantage of the
consonant-vowel structure of this encoding.

In fact, the split consonant-vowel representation may prove to be so
useful that it may be beneficial to encode other syllabary in this
fashion, even if they are less than 210 of them.

The code to break up a syllabary into the consonant-vowel part is part
of the Unicode normalization process.

@appendixsubsubsec The Yi Syllabary

A very large syllabary with 819 distinct symbols.  However, like
Ethiopic, it should be possible to support this script by breaking it
up.

@appendixsubsubsec The Ojibwe Syllabary

With only 120 distinct symbols, Aspell can actually support this one as
is.  However, as previously mentioned, it may be beneficial to break it
up into the consonant-vowel representation anyway.

@node Unsupported
@appendixsec Unsupported

These languages, when written in the given script, are currently
unsupported by Aspell for one reason or another.

@include lang-unsupported.texi

@appendixsubsec The Thai, Khmer, and Lao Scripts

The Thai, Khmer, and Lao scripts presents a different problem for
Aspell.  The problem is not that there are more than 210 unique symbols,
but that there are no spaces between words.  This means that there is no
easy way to split a sentence into individual words.  However, it is
still possible to spell check these scripts, it is just a lot more
difficult.  I will be happy to work with someone who is interested in
adding Thai, Khmer, or Lao support to Aspell, but it is not likely
something I will do on my own in the foreseeable future.

@appendixsubsec Languages which use H@`anzi Characters

H@`anzi Characters are used to write Chinese, Japanese, Korean, and were
once used to write Vietnamese.  Each h@`anzi character represents a
syllable of a spoken word and also has a meaning.  Since there are
around 3,000 of them in common usage it is unlikely that Aspell will
ever be able to support spell checking languages written using h@`anzi
until full Unicode support is implemented.  However, I am not even sure
if these languages need spell checking since h@`anzi characters are
generally not entered in directly.  Furthermore even if Aspell could
spell check h@`anzi the existing suggestion strategy will not work well
at all, and thus a completely new strategy will need to be developed.
However, if it is the case that h@`anzi needs to be spell checked and
you know something about the issues involved please fell free to contact
me.

@appendixsubsec Japanese

Modern Japanese is written in a mixture of @dfn{hiragana},
@dfn{katakana}, @dfn{kanji}, and sometimes @dfn{romaji}.  @dfn{Hiragana}
and @dfn{katakana} are both syllabaries unique to Japan, @dfn{kanji} is
a modified form of h@`anzi, and @dfn{romaji} uses the Latin alphabet.
With some work, Aspell should be able to check the non-kanji part of
Japanese text.  However, based on my limited understanding of Japanese
hiragana is often used at the end of kanji.  Thus if Aspell was to
simply separate out the hiragana from kanji it would end up with a lot
of word endings which are not proper words and will thus be flagged as
misspellings.  However, this can be fairly easily rectified as text is
tokenized into words before it is converted into Aspell's internal
encoding.  In fact, some Japanese text is written in entirely in one
script.  For example books for children and foreigners are sometimes
written entirely in hiragana.  Thus, Aspell, in its current state, could
prove at least somewhat useful for spell checking Japanese.

@appendixsubsec Hangul

Korean is generally written in hangul or a mixture of han and hangul.  In
Hangul letters individual letters, known as jamo, are grouped together
in syllable blocks.  Unicode allows Hangul to be stored in one of three
ways, (A) Individual jamo letters (Hangul Compatibility Jamo, U+3130 -
U+318F), (D) decomposed jamo (Hangul Jamo, U+1100 - U+11FF), and (C)
precoposed sylable blocks (Hangul Syllables, U+AC00 - U+D7AF).  In order
for Aspell to work with Hangul it needs to be form A.  Unfortunately the
existing Normalization code in Aspell will not be able to adequately
deal with converting Hangul from form D and C to form A and back again.
However, once this code is written, Aspell should be able to spell check
Hangul with out any problem.

@node Multiple Scripts
@appendixsec Languages Written in Multiple Scripts

Aspell should be able to check text written in the same language but
in multiple scripts with some work.  If the number of unique symbols
in both scripts is less than 210, then a special character set can be
used to allow both scripts to be encoded in the same dictionary.
However this may not be the most efficient solution.  An alternate
solution is to store each script in its own dictionary and allow
Aspell to choose the correct dictionary based on which script the
given word is written in.  Aspell currently does not support this mode
of spell checking but it is something that I hope to eventually
support.

@node Planned Dictionaries
@appendixsec Notes on Planned Dictionaries

Kevin Patrick Scannell has word lists available for the following
languages based on his web crawling software
(@uref{http://borel.slu.edu/crubadan/}) but needs someone to proofread
them:

@include crubadan.texi

@noindent
If you are interested, please contact him at scannell at slu edu.

A free spell checker for the Arabic (ar) languages called Duali can be
found at @uref{http://@/www.arabeyes.org/@/project.php?@/proj=duali}.
It uses its own form of affix compression.  This needs to be converted
to Aspell format.  The author, Mohammed Elzubeir, expressed an interest
in converting it but I have not heard from him in a while.  If you are
interested in helping with this conversion please let me know.

An Ispell hash table has been discovered for Albanian (sq) at
@uref{http://@/www.7kosova.com/kde-shqip/@/ispell/ispell.html}.
However, the raw word list is not provided and the author has not
been responding to emails.  If you know how to disassemble a Ispell hash
table or the where-abouts of the word list used to create the hash
table I would appreciate hearing from you.

A dictionary marked as "Planned" or "Maybe" but not listed in
the section means that someone has expressed an interest in creating
one.  If you are interested in helping please contact me at
@email{kevina@@gnu.org} so that I can put you in touch with them.

@node References
@appendixsec References

The information in this chapter was gathered from numerous sources,
including:

@itemize

@item
ISO 639-2 Registration Authority,
@uref{http://www.loc.gov/@/standards/iso639-2/}

@item
Languages and Scripts (Official Unicode Site),
@uref{http://www.unicode.org/@/onlinedat/languages-scripts.html}

@item
Omniglot - a guide to written language, @uref{http://www.omniglot.com/}

@item
Wikipedia - The Free Encyclopedia, @uref{http://wikipedia.org/}

@item
Ethnologue - Languages of the World,
@uref{http://www.ethnologue.com/}

@item
World Languages - The Ultimate Language Store,
@uref{http://www.worldlanguage.com/}

@item
South African Languages Web,
@uref{http://www.languages.web.za/}

@item
The Languages and Writing Systems of Africa (Global Advisor Newsletter),
@uref{http://www.intersolinc.com/@/newsletters/africa.htm}

@end itemize

Special thanks goes to Era Eriksson for helping me with the information in
this chapter.

@node Language Related Issues
@appendix Language Related Issues

Here are some language related issues that a good spell checker needs to
handle.  If you have any more information about any of these issues, or
of a new issue not discussed here, please email me at
@email{kevina@@gnu.org}.

@menu
* Compound Words::              
* Words With Symbols in Them::  
* Unicode Normalization::       
* German Sharp S::              
* Context Sensitive Spelling::  
@end menu

@node Compound Words
@appendixsec Compound Words

In some languages, such as German, it is acceptable to string two words
together, thus forming a compound word.  However, there are rules to
when this can be done.  Furthermore, it is not always sufficient to
simply concatenate the two words.  For example, sometimes a letter is
inserted between the two words.  Aspell currently has support for
unconditionally stringing words together.  I tried implementing more
sophisticated support for compound words in Aspell but it was too
limiting and no one used it.

After receiving feedback from several people it seems that acceptable
support for compound words involved two basically independent parts.
If this is not sufficient for your language please let me know.

@heading Part One

Describes how the word needs to be changed when forming a compound

@example
CMP <flag> <strip> <add> <cond> <cond2>

<flag>  is the compound flag
<strip> is the string to strip or 0 for the null string
<add>   is the string to add or 0 for the null string
<cond>  is the condition to match at the end of the current word
<cond2> is the condition to match at the beginning of the next word
@end example

@noindent
All but the last field are the same as a suffix entry in the exiting
affix code.

<cond> is a simplified regular expression.  Some examples:
@example
. (for anything)
e
[^aeiou]y
[^ey]
[aeiou]y
@end example

It does not seem necessary to change the beginning of a word when
forming compounds

@heading Part Two

Describes the position a word can appear in (beginning, middle, or
end) and with which words.

To do this each word can be assigned a category.  Then each category
can be given a set of rules to describe how it can be used in a
compound word for example

@example
A + B: indicates that category A may appear at the beginning of a
  word when followed by a category B word.  When combined it is then
  considered a category B word.
A + C + B: here a C word may only appear between an A or B word
A + A + B
A + A
A + A + A
etc..
@end example

I have not decided if a word should be allowed to belong to more than
one category as a new category can be created in necessary to mean
words in both category A and B for example.

@appendixsubsec To Implement

To implement support for compound words based on the above description
the following will need to be done:

@enumerate

@item
expand the affix code to support special compound flags as
described in part one

@item
write code to store the conditions as described in part two

@item
expand the compound checking code to check against the conditions

@item
expand the dictionary format to store the necessary compound info
with the word

@end enumerate

I don't know when I will be able to actually implement this.  If you
would like to try please let me know.

@node Words With Symbols in Them
@appendixsec Words With Spaces or Other Symbols in Them

Many languages, including English, have words with non-letter symbols in
them.  For example the apostrophe.  These symbols generally appear in
the middle of a word, but they can also appear at the end, such as in an
abbreviation.  If a symbol can @emph{only} appear as part of a word then
Aspell can treat it as if it were a letter.

However, the problem is most of these symbols have other uses.  For
example, the apostrophe is often used as a single quote and the
abbreviations marker is also used as a period.  Thus, Aspell cannot
blindly treat them as if they were letters.

Aspell currently handles the case where the symbol can only appear in
the middle of the word fairly well.  It simply assumes that if there is
a letter both before and after the symbol than it is part of the word.
This works most of the time but it is not fool proof.  For example,
suppose the user forgot to leave a space after the period:

@display
  @dots{} and the dog went up the tree.Then the cat @dots{}
@end display

@noindent
Aspell would think ``tree.Then'' is one word.  A better solution
might be to then try to check ``tree'' and ``Then'' separately.
But what if one of them is not in the dictionary?  Should Aspell assume
``tree.Then'' is one word?

The case where the symbol can appear at the beginning or end of the word
is more difficult to deal with.  The symbol may or may not actually be
part of the word.  Aspell currently handles this case by first trying to
spell check the word with the symbol and if that fails, try it without.
The problem is, if the word is misspelled, should Aspell assume the
symbol belongs with the word or not?  Currently Aspell assumes it does,
which is not always the correct thing to do.

Numbers in words present a different challenge to Aspell.  If Aspell
treats numbers as letters than every possible number a user might write
in a document must be specified in the dictionary.  This could easily
be solved by having special code to assume all numbers are correctly
spelled.  Yet, what about something like ``4th''.  Since the ``th''
suffix can appear after any number we are left with the same
problem.  The solution would be to have a special symbol for ``any
number''.

Words with spaces in them, such as foreign phrases, are even more
trouble to deal with.  The basic problem is that when tokenizing a
string there is no good way to keep phrases together. One solution is to
use trial and error.  If a word is not in the dictionary try grouping it
with the previous or next word and see if the combined word is in the
dictionary.  But what if the combined word is not, should the misspelled
word be grouped when looking for suggestions?  One solution is to also
store each part of the phrase in the dictionary, but tag it as part of a
phrase and not an independent word.

To further complicate things, most applications that use spell checkers
are accustom to parsing the document themselves and sending it to the
spell checker a word at a time.  In order to support words with spaces in
them a more complicated interface will be required.

@node Unicode Normalization
@appendixsec Unicode Normalization

Because Unicode contains a large number of precomposed characters there
are multiple ways a character can be represented.  For example letter
@"o can either be represented as

@example
U+00F6 LATIN SMALL LETTER O WITH DIAERESIS
@exdent or
U+0061 LATIN SMALL LETTER O + U+0308 COMBINING DIAERESIS
@end example

By performing normalization first, Aspell will only see one of these
representations.  The exact form of normalization depends on the
language.  Give the choice of:

@enumerate
@item Precomposed character
@item Base letter + combining character(s)
@item Base letter only
@end enumerate

@noindent
if the precomposed character is in the target character set, then (1),
if both base and combining character is present, then (2), otherwise (3).

Unicode Normalization is now implemented in Aspell 0.60.

@node German Sharp S
@appendixsec German Sharp S

The German Sharp S or Eszett does not have an uppercase equivalent.
Instead when @samp{@ss{}} is converted to @samp{SS}.  The conversion of
@samp{@ss{}} to @samp{SS} requires a special rule, and increases the
length of a word, thus disallowing inplace case conversion.
Furthermore, my general rule of converting all words to lowercase before
looking them up in the dictionary won't work because the conversion of
@samp{SS} to lowercase is ambiguous; it can be @samp{ss} or
@samp{@ss{}}.  I do plan on dealing with this eventually.

@node Context Sensitive Spelling
@appendixsec Context Sensitive Spelling

In some language, such as Luxembourgish, the spelling of a word depends
on which words surround it.  For example the the letter @samp{n} at the
end of a word will disappear if it is followed by another word starting
with a certain letter such as an @samp{s}.  However, it can probably get
more complicated than that.  I would like to know how complicated before
I attempt to implement support for context sensitive spelling.


@node To Do
@appendix To Do

@menu
* Important Items::             
* Other Items::                 
* Notes on Various Items::      
@end menu

@node Important Items
@appendixsec Important Items

Words in bold indicate how you should refer to the item when
discussing it with me or others.

@appendixsubsec Things that need to be done

These items need to be done before I consider Aspell finished. If you
are interested in helping me with one of these tasks please email
me. Good C++ skills are needed for most of these tasks involving
coding.

@itemize @bullet
@item
Create a generic filter to handle multi-character letters such as
@samp{"a} or @samp{\"a} for @"a.  This filter should make use of the
already exiting normalization code if possible.
@item
Make Aspell @strong{Thread safe}. Even though Aspell itself is not
multi-threaded I would like it to be thread safe so that it can be
used by multi-threaded programs. There are several areas of Aspell
that that are potently thread unsafe (such as accessing a global pool)
and several classes which have the potential of being used by more
than one thread (such as the personal dictionary). @emph{[In
Progress]}.
@item
Enhance @strong{ispell.el} so that it will work better with GNU
Aspell.  @emph{[In Progress]}.
@item
Clean up copyright notices and bring the Aspell package up to
@strong{GNU Standards}. @emph{[In Progress]}.
@end itemize

@appendixsubsec Things I would like to get done

I would like to get these done. However, I may still consider Aspell
finished without. They will probably eventually get
implemented. However, I could still use help with them.

@itemize @bullet
@item
Better support for @strong{compound words}. The support for
@emph{conditional} compound words found in Aspell versions 0.50
and earlier is no longer available since no one seems to be using
it. Support for @emph{unconditional} compound words is still 
available. @xref{Compound Words}.
@item
Be able to accept @strong{words with spaces in them} as many languages
have words, such as a word in a foreign phrase, which only makes sense
when followed by other words. @xref{Words With Symbols in Them}.
@item
Reorganize manual to make it easier to understand and to make it
possible to break out useful man pages.
@item
Support @strong{soundslike lookup with affix compression}.  I think it
is possible, although I don't know how effective it will be.  The basic
idea is to affix compress the soundslike codes and then match the
codes up with affix compressed words.  If you are interested,
email @email{aspell-devel@@gnu.org}, and I will explain it in more
detail.
@item
Use Lawrence Philips' new @strong{Double Metaphone algorithm}. See
@uref{http://aspell.net/@/metaphone/}. The main task involved
here is converting the algorithm into table form. This will take some
time but there is no real programming experience required. If you want
to help with Aspell but don't have any real programming experience,
this would be a great place to start.
@item
Rank suggestions based on @strong{frequency information}.  Both global
frequency and document specific frequency can be used.  The latter will
require that the whole document be made available to the spell checker.
Also use frequency information to flag words which are found in the
dictionary but not in common usage, and thus might not be what was
intended.
@item
Support a @strong{"dual-script" mode} where Aspell can use a separate
dictionary depending on which script it detects the current word in, the
two dictionaries can have nothing in common, ie an English one and a
Russian one for example.  This will @emph{not} support two languages that
use the same script as that is a lot more complicated.  For example if
the word is misspelled which dictionary should it use for the
suggestions?
@item
Write a @strong{GUI} for the Aspell utility. Ideally it should be able to do
everything the Aspell utility can do and not just be able spell check
a document.
@item
Develop a @strong{more powerful C API} for Aspell.  Ideally this API should
allow one to perform all the tasks the Aspell utility can do.  This
included the ability to check whole documents, and create
dictionaries, among other things.  
@item
Create a @strong{C++ interface} for Aspell, possibly on top of the C one.

@end itemize

@node Other Items
@appendixsec Other Items

These items all sound like good ideas however I am not sure when I
will get to implementing them if ever.  Words in bold indicate how you
should refer to the item when discussing it with me or others.

@itemize @bullet
@item
Come up with a plug-in for @command{gEdit} the gnome text editor.

@item
Change languages (and thus dictionaries) based on the information in
the actual document.

@item
Come up with a mode that will skip words based on the symbols that
(almost) always surround the word.  @xref{Word skipping by context}.

@item
Create two @strong{server modes} for Aspell.  One that uses the DICT
protocol and one that uses @command{ispell -a} method of communication
via some arbitrary port.

@item
Come up with @strong{thread safe personal dictionaries}.

@item
Use the @strong{Hidden Markov Model} to base the suggestions on not
only the word itself but on the context around the word. @xref{Hidden
Markov Model}.

@item
Having a way to @strong{email the personal dictionary} and/or
replacement list to a particular address either periodically or when
it grows to a certain size. @xref{Email the Personal Dictionary}.
@end itemize

The following good ideas were found in the ispell @file{WISHES} file
so I thought I would pass them on.

@itemize @bullet
@item
Ispell should be smart enough to ignore hyphenation signs, such as the
@TeX{} @samp{\-} hyphenation indicator.

@item
(Jeff Edmonds) The personal dictionary should be able to remove
certain words from the master dictionary, so that obscure words like
"wether" wouldn't mask favorite typos.

@item
(Jeff Edmonds) It would be wonderful if Ispell could correct inserted
spaces such as "th e" for "the" or even "can not" for "cannot".

@item
Since Ispell has dictionaries available to it, it is conceivable that
it could automatically determine the language of a particular file by
choosing the dictionary that produced the fewest spelling errors on
the first few lines.
@end itemize

@node Notes on Various Items
@appendixsec Notes on Various Items

@menu
* Word skipping by context::    
* Hidden Markov Model::         
* Email the Personal Dictionary::  
@end menu

@node  Word skipping by context
@appendixsubsec Word skipping by context

This was posted on the Aspell mailing list on January 1, 1999:

I had an idea on a great general way to determine if a word should be
skipped.  Determine the words to skip based on the symbols that
(almost) always surround the word.

For example when asked to check the following C++ code:
@example
cout << "My age is: " << num << endl;
cout << "Next year I will be " << num + 1 << endl;
@end example

@code{cout}, @code{num}, and @code{endl} will all be skipped.
@code{cout} will be skipped because it is always preceded by a
@samp{<<}.  @code{num} will be skipped because it is always preceded
by a @samp{<<}.  And @code{endl} will be skipped because it is always
between a @samp{<<} and a @samp{;}.

Given the following HTML code.
@example
<table width=50% cellspacing=0 cellpadding=1>
<tr><td>One<td>Two<td>Three
<tr><td>1<td>2<td>3
</table>

<table cellspacing=0 cellpadding=1>
</table>
@end example

@code{table}, @code{width} @code{cellspacing}, @code{cellpadding},
@code{tr}, @code{td} will all be skipped because they are always
enclosed in @samp{<>}.  Now of course @code{table} and
@code{width} would be marked as correct anyway however there is no
harm in skipping them.

So I was wondering if anyone on this list has any experience in
writing this sort of context recognition code or could give me some
pointers in the right direction.

This sort of word skipping will be very powerful if done right.  I
imagine that it could replace specific spell checker modes for @TeX{},
Nroff, SGML etc because it will automatically be able to figure out
where it should skip words.  It could also probably do a very good job
on programming languages code.

If you are interested in helping me out with this or just have general
comments about the idea please let me know.

@node Hidden Markov Model
@appendixsubsec Hidden Markov Model

Knud Haugaard S@o{}rensen suggested this one.  From his email on the
Aspell mailing list:

consider these examples:

@example
a fone number.
 -> a phone number.

a fone dress.
 -> a fine dress.
@end example

the example illustrates that the right correction might depend on the
context of the word.

So I suggested that you take a look on HMM to solve this problem.

This might also provide a good base to include grammar correction in
Aspell.

see this link
@uref{http://www.cse.ogi.edu/CSLU/HLTsurvey/ch1node7.html}.

I think it is a great idea.
 However unfortunately it will probably be very complicated to implement.
 Perhaps in the far future.

@node Email the Personal Dictionary
@appendixsubsec Email the Personal Dictionary

Someone suggested in a personal email:

@quotation
Have you thought of adding a function to Aspell, that - when the
personal dictionary has grown significantly - sends the user's
personal dictionary to the maintainer of the corresponding Aspell
dictionary? (if the user allows it)

It would be a very useful service to the dictionary maintainers, and I
think most users can see their benefit in it too.
@end quotation

And I replied:

@quotation
Yes I have considered something like that but not for the personal
dictionaries but rather the replacement word list in order to get
better test data for @uref{http://aspell.sourceforge.net/test/}.
@end quotation

The problem is I don't know of a good way to do this since Aspell can
also be used as a library.  It also is not a real high priority,
especially since I would first need to learn how to send email within
a C++ program.

@c @node Installing
@c @appendix Installing

@include readme.texi


@node ChangeLog
@appendix ChangeLog

@heading Changes from 0.60.2 to 0.60.3 (June 28, 2005)

@itemize @bullet
@item
Fixed bugs involving several of the C API functions.
@item
Fixed bug where @samp{ultra} or @samp{fast} mode will not return any
suggestions when soundslike lookup is not used.
@item
Made a minor, yet significant, optimization to the suggestion code.
This speed things up by an order of magnitude in some cases.
@item
Avoid using the slow ngram scan except when the @option{sug-mode} is 
@samp{slow} or @samp{bad-speller}.
@item
Fixed a bug in curses mode which causes word-wrap to not work
correctly in some cases.
@item
Fixed a bug in pipe mode with a missing newline.
@item
Fixed the @command{spell} compatibility script.
@item
Several other minor bugs fixed.
@item
Made note about the change in behavior of the @option{-l} command line
switch.
@item
Other manual update/fixes.
@item
Updated to Libtool 1.5.18, Automake 1.9.6, and Makeinfo 4.8.
@end itemize

@heading Changes from 0.60.1 to 0.60.2 (December 18, 2004)

@itemize @bullet
@item
Added the @command{munch-list} command to the aspell utility.  The
@command{munch} program in the @file{myspell/} directory will
disappear in Aspell 0.61.  The @command{munchlist} script will also
likely disappear or be replaced when Aspell 0.61 is released since it
doesn't work correctly anyway.
@item
Several important bug fixes some of which rendered some non-English
languages unusable.
@item
Other minor changes.
@end itemize

@heading Changes from 0.60.1 to 0.60.1.1 (November 20, 2004)

@itemize @bullet
@item
Fix bug involving checking of capitalized word when affix compression
is used.
@item
Compile fixes.
@item
Added an option to disable using the ``wide'' curses version in case it
causes compile problems.
@item
Minor manual updates
@item
Avoided including some unnecessary files in the distribution.
@end itemize

@heading Changes from 0.60 to 0.60.1 (November 7, 2004)

@itemize @bullet
@item
Lots of compile fixes for various platforms.
@item
Miscellaneous bug fixes.
@item
Added Nroff filter thanks to Sergey Poznyakoff.
@item
The default filter mode when in pipe mode is now nroff for compatibility with
Ispell.
@item
Added Texinfo filter.
@item
Added a section detailing the differences between Ispell and Aspell.
@item
Updated the section on thread safety.
@item
Other miscellaneous manual changes such as updating the To Do and Authors
section.
@end itemize

@heading Changes from 0.50.5 to 0.60 (August 27, 2004)

@itemize @bullet
@item
Added support for Affix Compression.  Affix compression stores the root
word and then a list of prefixes and suffixes that the word can take,
and thus saves a lot of space.  The codebase comes from MySpell found in
OpenOffice.  It uses the same affix file that OpenOffice (and Mozilla)
use.  Affix compression will even work with soundslike lookup to a
limited extent.
@item
Added support for accepting all input and printing all output in UTF-8
or some other encoding different from the one Aspell uses.  This
includes support for Unicode normalization.  Aspell can now support any
language with no more than 210 distinct characters, including different
capitalizations and accents, @emph{even if} there is not an existing
8-bit encoding that supports the language.
@item
Added support for loadable filters and customizable filter modes thanks
to Christoph Hinterm@"uller.
@item
Enhanced SGML filter to also support skipping sgml tags such as "script"
blocks thanks to Tom Snyder.
@item
Added gettext support thanks to Sergey Poznyakoff
@item
Reworked the compiled dictionary format.  Compiled dictionaries now take
up less space (less than 80% for the English language) and
creating them is significantly faster (over 4 times for the
English language).
@item
Reworked suggestion code.  It is significantly faster when dealing with
short words (up to 10 times).  Also added support for MySpell
Replacement Tables and n-gram lookup.  In addition, added basic support
for compound words.
@item
Manual has has been converted to texinfo format thanks to the work of
Chris Martin.
@item
Reworked the build system so that a single Makefile is used for most
of the code.
@item
All data, by default, is now included in
@file{@var{libdir}/aspell-0.60}.  Also added a built time option to
increment the major version number of the shared library.  This should
allow both Aspell version 0.50 and 0.60 to coexist.  The major version
number is @emph{not} incremented by default as Aspell 0.60 is binary
compatible with Aspell 0.50.  @xref{Binary Compatibility}.
@item
The code to handle dictionaries has been rewritten.  Because of this
support for the dictionary option @option{strip-accents} has been
removed.  In addition the @option{ignore-accents} option is currently
unimplemented.
@item
Lots of other minor changes due to massive overhaul of the source code.

@end itemize

@heading Changes from 0.50.4.1 to 0.50.5 (Feb 10, 2004)

@itemize @bullet
@item
Reworked url filter which fixed several bugs and now accepts
"bla.bla/kdkdl" as a url.
@item
Fixed bug in which the url filter was coming before all other filters
when it was suppose to come after. This solved a number of problems
where the url filter was interfering with other filters.
@item
Small bug fix in SGML filter.
@item
Added code page charsets, ie cp125?.dat.
@item
Added natural (split) keyboard data file as "split.kbd"
@item
Compile fixes for the upcoming Gcc 3.4
@item
Removed Solaris link hack as it was causing more problems than it
fixes.
@item
Compile fixes for Sun WorkShop 6 compiler, but there may still be some
problems, especially with linking.
@item
Included patch to help compile with Microsoft Visual C++ 6.
@item
Minor manual fixes.
@item
Updated the TODO section to reflect the current progress with the next
major version of Aspell (0.51).
@item
Updated to Autoconf 2.59, Automake 1.82, and Libtool 1.5.2.
@end itemize

@heading Changes from 0.50.4 to 0.50.4.1 (Oct 11, 2003)

@itemize @bullet
@item
Fixed major bug in pipe mode which caused the last character to be
chopped off words before they were stored.
@item
Minor formating fixes in the manual.
@end itemize

@heading Changes from 0.50.3 to 0.50.4 (Sep 26, 2003)

@itemize @bullet
@item
Minor changes in URL filter to avoid treating the double quote
character as part of the URL, and to avoid treating words ending in
more than one period as a URL.

@item
Document fixes in Aspell API

@item
Small compile fixes, including one for GCC 3.3

@item
Updated Win32 section since a port now exists thanks to Thorsten
Maerz.

@item
Complain instead of doing nothing or aborting for unimplemented
functions in Aspell utility.

@item
Portability bug fixes.

@item
Upgraded to Autoconf 2.57, Automake 1.7.7, Libtool 1.5 (no longer use
CVS version  of libtool).
@end itemize

@heading Changes from 0.50.2 to 0.50.3 (Nov 23, 2002)

@itemize @bullet
@item
Hopefully fixed the ispell alignment error problem when Aspell is used
with ispell.el.

@item
Fixed a problem with personal dictionaries on NFS mounted home
directories.

@item
Compiled libaspell-common directory into libaspell for now to avoid
forcing applications to relink whenever a new Aspell version is out
which was due to the use of the libtool '-release' flag.

@item
Fixed Makefiles so that Aspell can be built outside the source tree
(i.e.  with VPATH).

@item
Updated the section on compiling with Win32.

@item
Updated to Autoconf 2.56.
@end itemize

@heading Changes from 0.50.1 to 0.50.2 (Sep 28, 2002)

@itemize @bullet
@item
Fixed a number of bugs in ispell compatibility mode

@item
Fixed a number of bugs with the handling of replacement pairs

@item
Other miscellaneous bug fixes

@item
Additional Win32 portability fixes

@item
Added the Ukrainian KOI8-U charset.
@end itemize

@heading Changes from 0.50 to 0.50.1 (Aug 28, 2002)

@itemize @bullet
@item
A rather large number of portability fixes for non GNU/Linux platforms.

@item
Fixed pkglibdir and pkgdatadir in configure.

@item
Reintroduced some configure options from Aspell .33.7 included
dict-dir, data-dir, curses, curses-include, win32-relocatable.

@item
Fixed Aspell so it will now compile with -O3 when using gcc.

@item
Updated note on Win32 support.

@item
Other minor manual improvements.

@item
Portability fixes in dictionary files

@item
Official dictionary package for the Slovak language.
@end itemize

@heading Changes from .33.7.1 to 0.50 (Aug 23, 2002)

@itemize @bullet
@item
A complete overhaul of the source code which included merging Pspell
into Aspell.

@item
Changed the way dictionaries and languages are handled.

@item
Added Dvorak keymap.

@item
Added the ability to list the available dictionaries

@item
Improved the spell checking interface a bit.

@item
Added support for using the Ispell keymapping when checking files.

@item
Complete rewrite of the filter interface.
 It should now be fairly easy to add new filters to Aspell.

@item
Added some preliminary developer documentation.

@item
Lots of other changes due to the massive overhaul of the source code.
@end itemize

@heading Changes from .33.7 to .33.7.1 (Aug 20, 2001)

@itemize @bullet
@item
Minor manual fixes.

@item
Compile fix for Gcc 3.0 and Solaris.
@end itemize

@heading Changes from .33.6.3 to .33.7 (Aug 2, 2001)

@itemize @bullet
@item
Updates to Autoconf 2.50 and switched to the HEAD branch of libtools.

@item
Fixed a bug which caused Aspell to crash when typo-analysis is not
used such as when sug-mode is @strong{fast} or @strong{bad spellers}.

@item
Added support for typo-analysis even when a soundslike is not used.

@item
Fixed a bug which causes extended charters to display incorrectly on
some platforms

@item
Compile fixes so that it will compile with Gcc 3.0.

@item
Compile fixed which should allow Aspell to compile with Egcs 1.1.  I
have not been able to actually test it though.  Please let me know at
kevina@@users.sourceforge.net if you have have tried with Egcs 1.1.

@item
Compile and configuration script fixes so that USE_FILE_INO will
properly be defined and Aspell will compile correctly when it is
defined.

@item
More ANSI C++ compliance fixes.
@end itemize

@heading Changes from .33.6.2 to .33.6.3 (June 3, 2001)

@itemize @bullet
@item
Fixed a build problem in the manual/ directory by including
manual-text and manual-html in the distribution.
@end itemize

@heading Changes from .33.6.1 to .33.6.2 (June 3, 2001)

@itemize @bullet
@item
Compile fix so that Aspell will work correctly when not installed in
/usr/local.

@item
Avoided regenerating the manual unless configured with enable-maintainer-mode.

@item
Added the missing documentation files in the scowl directory.
@end itemize

@heading Changes from .33.6 to .33.6.1 (May 29, 2001)

@itemize @bullet
@item
Fixed a formating problem with the manual involving <.

@item
Added a note about creating pwli files.

@item
Removed the space after between the -L and the directory name in the
pspell-module/Makefile  which caused problems on some platforms.

@item
Added the configure option AM_MAINTAINER_MODE to avoid enabling rules
which often causes generated build files to be rebuild with the wrong
version of Libtool by default.  I don't know why I didn't think to do
this a long time ago.
@end itemize

@heading Changes from .33.5 to .33.6 (May 18, 2001)

@itemize @bullet
@item
Fixed a minor bug where some words would have random compound tags
attached to them.

@item
Fixed a compile problem on many platforms where fileno is defined as a
macro.

@item
Updated the description for a few of Aspell's options.

@item
Removed the note of Aspell not being able to run when compiled with
the upcoming Gcc 3.0 compiler as things seam to work now.

@item
Added a note about Aspell not being able to compile with Egcs 1.1.

@item
Added hack to deal with Libtool's interdependencies problem.  See bug
#416981 for Pspell for more info.
@end itemize

@heading Changes from .33 to .33.5 (April 5, 2001)

@itemize @bullet
@item
@strong{dump master} correctly detects which dictionary and language
to use based on the @env{LANG} environment variable.

@item
Fixed a problem on Win32 which involves path names that begin with
<Drive Letter>:.

@item
Bug fixes and enhancements so that Aspell can once again run under
MinGW.  You can even use the new full screen interface if Aspell is
compiled with PDCurses.

@item
Some major modifications to make Aspell more C++ compliant in order to
get Aspell to compile under the upcoming Gcc 3.0 compiler.  This
included only using STL features found in the standard version of C++.
(Which means Aspell will no longer require using the SGI version of
the STL) This should also make compiling C++ under non-gcc compilers a
lot simpler.  Please note that Aspell still has some problems with the
upcoming Gcc 3.0 compiler.

@item
Minor changes to remove some -Wall warnings.

@item
Added a hack to that Aspell will properly compile as a shared library
under Solaris.

@item
Added a few import missing words to the English word list.
@end itemize

@heading Changes from .32.6 to .33 (January 28, 2001)

@itemize @bullet
@item
Added a new new curses based interface to replace the dumb terminal
interface everyone has been bitching about.

@item
Added the ability to give higher priority to words such as "the"
instead of "teh" which are likely to be due to typos.

@item
Reorganized the manual so that it is hopefully easier to follow.

@item
Ability to automatically select the best dictionary to used based on
the setting of the @env{LANG} environment variable.

@item
Expanded the medium dictionary size to include more words which
included the original words found in ispell and eliminated the large
size for now.

@item
Added three special variant add-on dictionaries.

@item
Switched to the multi-language branch of the CVS version of libtool.

@end itemize

@heading Changes from .32.5 to .32.6 (Nov 8, 2000)

@itemize @bullet
@item
Fixed a bug that where Aspell will crash when reading in accented
characters on some platforms.  This fixes bug # 112435.

@item
Fixed some other bugs so that it will run under Win32 under CygWin.
Unfortunately it still won't run properly under Mingw.

@item
Fixed the mmap test in configure so that it won't fail on some
platforms that use munmap(char *, int) instead of munmap(void *, int).

@item
Upgraded to the latest CVS version of libtool which fixed the problem
with using GNU Make under Solaris.

@item
Added an option to copy files instead of using symbolic links for the
special @strong{multi} dictionary files.
@end itemize

@heading Changes from .32.1 to .32.5 (August 18, 2000)

@itemize @bullet
@item
Changed my email from kevinatk at home com to kevina at users
sourceforge net please make a note of the new email address.

@item
Added an option to control if the personal replacement dictionary is
saved when the save_all_wls method is called.

@item
Brought back the ability to dump the master word list even in the case
of the special @strong{multi} lists.

@item
Added a large number of hacker related words and some other slang
terms to the medium size word list.

@item
Added an @strong{ispell} and @strong{spell} compatibility script for
systems which don't have ispell installed.  They are located in the
scripts/ directory and are not installed by default.

@item
Manual fixes.

@item
Added a note on not using GNU Make on Solaris.
@end itemize

@heading Changes from .32 to .32.1 (August 5, 2000)

@itemize @bullet
@item
Minor compile fixes for recent gcc snapshot.

@item
Fixed naming of pwli files.

@item
Fixed a bug when aspell will crash when used with certain single
letter flags.  This bug was most noticeable when used with Emacs.

@item
Word list changes, see SCOWL Readme.

@item
Other miscellaneous changes.
@end itemize

@heading Changes from .31.1 to .32 (July 23, 2000)

@itemize @bullet
@item
Added support for optionally doing without the soundslike data.

@item
Greatly reduced the amount of memory used when creating word lists.

@item
Added support for ignoring accents when coming up with suggestions.

@item
Added support for local-data-dir which is searched before data-dir.

@item
Added support for specifying which words may be used in compounds and where
 they may be used.

@item
Added support for having more than one main word list as well as a
special @strong{multi} word list files which will allow multiple word
lists to be treated as one.

@item
Aspell now uses a completely new word list.

@item
The apostrophe (') is no longer considered part of the word when it as
at the end of the word such as in @samp{dogs'}.
@end itemize

@heading Changes from .31 to .31.1 (June 18, 2000)

@itemize @bullet
@item
Fixed a bug where Aspell would not create a complete dictionary file
on some platforms when the data is 8-bit.

@item
Added a workaround so Aspell will work with ispell.el 3.3.

@item
Minor compile fixes so it would compile better with the very latest
gcc (CVS Version).

@item
Removed note about compiling in Win32 as I was now able to get it to work.
@end itemize

@heading Changes from .30.1 to .31 (June 11, 2000)

@itemize @bullet
@item
Added support for spell checking run together words.

@item
Added an option to produce a list of misspelled words from standard input.

@item
More robust error reporting when reading in language data files.

@item
Fixed a bug when that will cause Aspell to crash if the
@strong{special} line was not defined in the language data file.

@item
Update Pspell Module.

@item
Minor bug fixes.

@item
Added cross references in ``The Aspell Utility Chapter'' for easier
use.
@end itemize

@heading Changes from .30 to .30.1 (April 29, 2000)

@itemize @bullet
@item
Ported Aspell to Win32 platforms.

@item
Portability fixes which may help aspell compile on other platforms.

@item
Aspell will no longer fail if for some reason the mmap fails, instead
it will just read the file in as normal and free the memory when done.

@item
Minor changes in the format of the main word list as a result of the
changes, the old format should still work in most cases.

@item
Fixed a bug when aspell was ignoring the extension of file names such
as .html or .tex when checking files.

@item
Fixed a bug when aspell will go into an infinite loop when creating
the main word list from a word list which has duplicates in it.

@item
Minor changes to the manual for better clarity.
@end itemize

@heading Changes from .29.1 to .30 (April 2, 2000)

@itemize @bullet
@item
Fixed many of the capitalization bugs found in previous versions of
Aspell.

@item
Changed the format of the main word list yet again.

@item
Fixed a bug so that @code{aspell check} will work on the PowerPC.

@item
Added ability to change configuration options in the middle of a session.

@item
Added words from /usr/dict/words found on most Linux systems as well
as a bunch of commonly used abbreviation to the word list.

@item
Fixed a bug when aspell will dump core after reporting certain errors
when compiled with gcc 2.95 or higher.  This involved reworked the
Exception heritage to get around a bug in gcc 2.95.

@item
Added a few more commands to the list of default commands the @TeX{}
filter knows about.

@item
Aspell will now check if a word only contains valid characters before
adding it to any dictionaries.  This might mean that you have to
manually delete a few words from your personal word list.

@item
Added option to ignore case when checking a document.

@item
Adjusted the parameters of the @strong{normal} suggest mode to so that
significantly less far fetched results are returned in cases such as
tomatoe, which went from 100 suggestions down to 32, at the expense of
getting slightly lower results (less than 1%),

@item
Improved the edit distance algorithm for slightly faster results.

@item
Removed the @samp{$$m} command in pipe mode, you should now use
@samp{$$cs mode,@var{mode}} to set the mode and @strong{$$cr mode} to
find out the current mode.

@item
Reworked parts of Aspell to use Pspell services to avoid duplicating code.

@item
Added a module for the newly released Pspell.  It will get installed
with the rest of aspell.

@item
Miscellaneous other bug fixes.
@end itemize

@heading Changes from .29 to .29.1 (Feb 18, 2000)

@itemize @bullet
@item
Improved the @TeX{} filter so that it will accept '@@' at the begging
of a command name and ignore trailing '*'s.  It also now has better
defaults for which parameters to skip.

@item
Reworked the main dictionary so that it can be memory mapped in.  This
decreases startup time and allows multiple aspell processes to use the
same memory for the main word list.  This also also made Aspell 64 bit
clean so that it should work on an alpha now.

@item
Fix so that aspell can compile on platforms that gnu as is not
available for.

@item
Fixed issue with flock so it would compile on FreeBSD.

@item
Minor changes in the code to make it more C++ compliant although I am
sure there will still be problems when using some other compiler other
than gcc or egcs.

@item
Added some comments to the header files to better document a few of
the classes.
@end itemize

@heading Changes from .28.3 to .29 (Feb 6, 2000)

@itemize @bullet
@item
Fixed a bug in the pipe mode with lines that start with
@samp{^$$}.

@item
Added support for ignoring all words less than or equal to a specified length

@item
New soundslike code based thanks to the contribution of Bj@"orn Jacke.
It now gets all of its data from a table making it easier for other
people to add soundslike code for their native language.  He also
converted the metaphone algorithm to table form, eliminating the need
for the old metaphone code.

@item
Major redesign of the suggestion code for better results.

@item
Changed the format of the personal word lists.  In most cases it
should be converted automatically.

@item
Changed the format of the main word list.

@item
Name space cleanup for more consistent naming.  I now use name spaces
which means that gcc 2.8.* and egcs 1.0.* will no longer cut it.

@item
Used file locks when reading and saving the personal dictionaries so
that it truly multiprocess safe.

@item
Added rudimentary filter support.

@item
Reworked the configuration system once again.  However, the changes to
the end user who does not directly use my library should be minimal.

@item
Rewrote my code that handles parsing command line parameters so that
it no longer uses popt as it was causing to many problems and didn't
integrate well with my new configuration system.

@item
Fixed pipe mode so that it will properly ignore lines starting with
'~' for better ispell compatibility.

@item
Aspell now has a new home page at
@uref{http://aspell.sourceforge.net/}.  Please make note of the new
URL.

@item
Miscellaneous manual fixes and clarifications.
@end itemize

@heading Changes from .28.2.1 to .28.3 (Nov 20, 1999)

@itemize @bullet
@item
Fixed a bug that caused aspell to crash when spell checking words over
60 characters long.

@item
Reworked @strong{aspell check} so that
@enumerate
@item
You no longer have to hit enter when making a choice.

@item
It will now overwrite the original file instead of creating a new
file.  An optional backup can be made by using the -b option.
@end enumerate

@item
Fixed a few bugs in data.cc.
@end itemize

@heading Changes from .28.2 to .28.2.1 (Aug 25, 1999)

@itemize @bullet
@item
Fixed the version number for the shared library.

@item
Fixed a problem with undefined references when linking to the shared library.
@end itemize

@heading Changes from .28.1 to .28.2 (Aug 25, 1999)

@itemize @bullet
@item
Fixed a bunch of bugs in the language and configuration classes.

@item
Minor changed in the code so that it could compile with the new gcc
2.95(.1).

@item
Changed the output of @code{dump config} so that default values are
given the value @code{<default>}.  This means that the output can
be used to created a configuration file.

@item
Added notes on using aspell with VIM.
@end itemize

@heading Changes from .28 to .28.1 (July 27, 1999)

@itemize @bullet
@item
Removed some debug output

@item
Changed notes on compiling with gcc 2.8.* as I managed to get it to
compile on my school account

@item
Avoided included @strong{stdexcept} in @file{const_string.hh} so that
I could get to compile on my schools account with gcc 2.8.1.
@end itemize

@heading Changes from .27.2 to .28 (July 25, 1999)

@itemize @bullet
Provided an iterator for the replacement classes.

@item
Added support for dumping and creating and merging the personal and
replacement word lists.

@item
Changed the aspell utility command line a bit, it now used popt.

@item
Totally reworked aspell configuration system.  Now aspell can get
configuration from any of 5 sources: the command line, the environment
variable @env{ASPELL_CONF}, the personal configuration file, the
global configuration file, and finally the compiled in defaults.

@item
Totally reworked the language class in preparation for my new language
code.  See @url{http://aspell.sourceforge.net/international/} for more
information of what I have in store.

@item
Added some options to the configure script: --enable-dict-dir=DIR,
--enable-doc-dir=DIR, --enable-debug, and --enable-opt

@item
Removed some old header files.

@item
Reorganized the directory structure a bit

@item
Made the text version of the manual pages slightly easier to read

@item
Used the @samp{\url} command for urls for better formating of the
printed version.
@end itemize

@heading Changes from .27.1 to .27.2 (Mar 1, 1999)

@itemize @bullet
@item
Fixed a major bug that caused aspell to dump core when used without
any arguments

@item
Fixed another major bug that caused aspell to do nothing when used in
interactive mode.

@item
Added an option to exit in Aspell's interactive mode.

@item
Removed some old documentation files from the distribution.

@item
Minor changes on to the section on using Aspell with egcs.

@item
Minor changes to remove -Wall warnings.
@end itemize

@heading Changes from .27 to .27.1 (Feb 24, 1999)

@itemize @bullet
@item
Fixed a minor compile problem.

@item
Updated the section on using Aspell with egcs to it it more clear why
the patch is necessary.
@end itemize

@heading Changes from .26.2 to .27 (Feb 22, 1999)

@itemize @bullet
@item
Totally reworked the C++ library which means you may need to change
some things in your code.

@item
Added support for detachable and multiple personal dictionaries in the
C++ class library.

@item
The C++ class library now throws exceptions.

@item
Reworked aspell ability to learn from users misspellings a bit so that
it now has a memory.  For more information see @ref{Notes on Storing
Replacement Pairs}.

@item
Upgraded autoconf to version 2.13 and automake to version 1.4 for
better portability.

@item
Fixed the configuration so the @code{make dist} will work.  For now on
aspell will be distributed with @code{make dist}.

@item
Added support to skip over URL's, email addresses and host names.

@item
Added support for dumping the master and personal word list.  You can
now also merge a personal word list.  Type aspell -help for help on
using this feature.


@item
Reorganized the source code.

@item
Started using proper version numbers for the shared library.

@item
Fixed a bug that caused a spell to crash when adding certain
replacement pairs.

@item
Fixed the problem with duplicate lines when exiting pipe mode for
good.
@end itemize

@heading Changed from .26.1 to .26.2 (Jan 3, 1998)

@itemize @bullet
@item
Fixed another compile problem.  Hopefully this time it will really
compile OK on other peoples machines.
@end itemize

@heading Changed from .26 to .26.1 (Jan 3, 1998)

@itemize @bullet
@item
Fixed a small compile problem in @file{as_data.cc}.
@end itemize

@heading Changed from .25.1 to .26 (Jan 3, 1999)

@itemize @bullet
@item
Fixed a bug that causes duplicates items to be displayed in the
suggestion list for good.  (If it still does it please send be email.)

@item
Added the ability for aspell to learn form the users misspellings.

@item
Library Interface changes.  Still more to come @dots{}.

@item
Is now multiprocess safe.  When a personal dictionary (or replacement
list) is saved it will now first update he list against the dictionary
on disk in case another process modified it.

@item
Fixed the bug that caused duplicate output when used non interactively
in pipe mode.

@item
Dropped support for gcc 2.7.2 as the C++ compiler.

@item
Updated the How Aspell Works (@ref{Aspell Suggestion Strategy}.)

@item
Added support for the @env{ASPELL_DATA_DIR} environment variable.
@end itemize

@heading Changes from .25 to .25.1 (Dec 10, 1998)

@itemize @bullet
@item
Fixed the version number so that Aspell reports the correct version
number.

@item
Changed the note on gcc 2.7.2 compilers to make it clear that only the
C++ compiler cannot be gcc 2.7.2, it is OK if the C compiler is gcc
2.7.2.

@item
Updated the TODO list and reorganized it a bit.

@item
Fixed the directory so that all the documentation will get installed
in @verb{#${prefix}/doc/aspell#} instead of half of it in
@verb{#${prefix}/doc/aspell#} and half of it in
@verb{#${prefix}/doc/kspell#}.
@end itemize

@heading Changes from .24 to .25 (Nov 23, 1998)

@itemize @bullet
@item
Total rework of how the main word list is stored.  Start up time
decreased to about 1/3 of what it was in .24 and memory usage
decreased to about 2/3.  (When used with the provided word list on a
Linux system).

Also the format and default locations of the main word list data files
changed in the process and the data is now machine dependent.  The
personal word list format, however, stayed the same.

@item
Changed the scoring method to produce slightly better results with
words like the vs.  teh.  And other simpler misspellings where two
letters are swapped.

@item
Fixed the very unpredictable behavior of the @samp{*}, @samp{&},
@samp{@@} commands in the pipe mode.

@item
Added documentations for Aspell pipe mode (also known as
@command{ispell -a} compatibility mode)

@item
Added a bunch of Aspell specific extensions to the pipe mode and
documented them.

@item
Documented the @code{to_soundslike} and @code{soundslike} methods for
the @code{aspell} class.

@item
Changed the scoring method to produce better results for words like
@emph{fone} vs @emph{phone} and other words that have a spelling that
doesn't directly relate to how the word sounds by using the phoneme
equivalent of the word in the scoring of it.

@item
Added the @code{to_phoneme} and @code{have_phoneme} methods to the
@code{SC_Language} class.

@item
Added the @code{to_phoneme} method to the @code{aspell} class.

@item
Added the framework for being able to learn from the users
misspelling.  Right now it just keep a log of replacements.

@item
Redid @file{stl_rope-30.diff}.  For some reason the version of patch
on my system refused it.

@item
Rewrite of the ``@emph{Using as a replacement for Ispell}'' section
and added the @code{run-with-aspell} utility as a replacement of the
old method of mapping Ispell to Aspell.

@item
Fixed a bug that caused duplicate words to appear in the suggestion
list.
@end itemize

@heading Changes from .23 to .24 (Nov 8, 1998)

@itemize @bullet
@item
Fixed my code so that it can once again compile with g++ 2.7.2.

@item
Rewrote the How It Works chapter.

@item
Rewrote the Requirement section and added noted on compiling with g++
2.7.2.

@item
Added a To Do chapter.

@item
Added a Glossary and References chapter.

@item
Other minor documentation improvements.

@item
Internal code documentation improvements.
@end itemize

@heading Changes from .22.1 to .23 (Oct 31, 1998)

@itemize @bullet
@item
Minor documentation fixes.

@item
Changed the scoring strategy for words with 3 or less letters.  This
cut the number of words returned for these roughly in half.

@item
Expanded the word list to also include @strong{american.0} and
@strong{american.1} from the Ispell distribution.  It now includes
@strong{english.0}, @strong{english.1}, @strong{american.0} and
@strong{american.1} from the directory @file{languages/english}
provided with Ispell 3.1.20.

@item
Added a link to the location of the latest Ispell.el in the
documentation.

@item
Started a C interface and added some rough documentation for it.
@end itemize

@heading Changes from .22 to .22.1 (Oct 27, 1998)

@itemize @bullet
@item
Minor bug fixes.  I was deleting arrays with delete rather than
delete[].  I was suppressed that this had not created a problem.

@item
Added a simple test program to test for a memory leak present on some
systems.  (Only debian slink at the moment.) See the file
memleak-test.cc for more info.
@end itemize

@heading Changes from .21 to .22 (Oct 26, 1998)

@itemize @bullet
@item
Major redesign or the scoring method.  It now uses absolute distances
rather than relative scores for more consistent results.  See
@file{suggest.cc} for more info.

@item
Suggest code rewritten is several places however the core process
stayed the same.

@item
The @code{suggest_ultra} method temporarily does nothing.  It should
be working again by the next release.
@end itemize

@heading Changes from .20 to .21 (Oct 13, 1998)

@itemize @bullet
@item
Added documentation for aspell::Error

@item

Changed the library name from @code{libspell} to @code{libaspell}.  It
should never have been @code{libspell} in the first place.  Sorry for
the incompatibility.

@item
Added @file{as_error.hh} to the list of files copied to the include
directory so that you can actually use the library outside of the
source dir.

@item
Fixed bug that caused a segmentation fault with words where the only
suggestions was inserting a space or hyphen such as in
@strong{ledgerline}.

@item
Added the @strong{score} method to @code{aspell}.

@item
Changed the scoring method to deal with word when the user uses "f" in
place of "ph" a lot better.
@end itemize

@heading Changes from .11 to .20 (Oct 10, 1998)

@itemize @bullet
@item
@emph{Name change}.  Everything that was Kspell is now Aspell.  Sorry,
the name Kspell was already used by KDE and I didn't want to cause any
confusion.

@item
Fixed a bug that causes a segmentation fault when the @env{HOME}
environment variable doesn't exist.
@end itemize

@heading Changes from .10 to .11 (Sep 12, 1998)

@itemize @bullet
@item
Overhaul of the SC_Language class

@item
Added documentation for international support

@item
Added documentation for the C++ library

@item
Other minor bug fixes.
@end itemize

@node Authors
@appendix Authors

The folowing people or companies have contributed a non-trival amount of
code to Aspell and thus own the Copyright to part of Aspell.

@table @asis

@item Jose Da Silva
Bug fixes and enhancements to @command{word-list-compress}.

@item Sergey Poznyakoff
Wrote the Nroff filter.

@item Tom Snyder
Enhanced the SGML filter to also support skipping sgml tags such as
"script" blocks.

@item Kevin B. Hendricks (and Contributers)
Wrote MySpell which is a simpile spell checker library that supports
affix compression.  Aspell affix compression code is based on his code.

@item Christoph Hinterm@"uller
Added support for for loadable filters.

@item Melvin Hadasht
Wrote a locale independent version of strtol and strtod.  
Wrote the original loadable filter support however his code has been 
completely rewritten by Christoph Hinterm@"uller and Kevin Atkinson.

@item Bj@"orn Jacke
Wrote the generic soundslike algorithm which gets all of its data from a
file, thus eliminating almost all need for language specific code from
Aspell.

@item Silicon Graphics Computer Systems, Inc.
@itemx Hewlett-Packard Company
Parts of the SGI STL code were used in various places throughout the
Aspell source.

@end table

In addition the authors of some of translated messages did not release
their work into the Public Domain, and thus own the copyright to the
translated text.  See the files @file{*.po} in the @file{po} directory
for more details.

The folowing people also contributed to the development of Aspell but do
not own the Copyright to part of Aspell.

@table @asis

@item Sergey Poznyakoff
Added gettext support.

@item Chris Martin
Converted the manual to texinfo.

@item Lawrence Philips
Wrote the original metaphone algorithm; however, he released his work
into the Public Domain.

@item Michael Kuhn
Converted the metaphone algorithm into C code and made some enhansments
to the orignal algorithm.  He also released his work into the Public
Domain.

@item Geoff Kuenning (and contributers)
The authors of Ispell.  Many of the ideas used in Aspell, espacally with
the affix code, were taken from Ispell.  However none of the orignal
Ispell code is used in Aspell.

@end table

@node Copying
@appendix Copying

Copyright @copyright{} 2000--2004 Kevin Atkinson.

Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.1 or
any later version published by the Free Software Foundation; with no
Invariant Sections, no Front-Cover Texts and no Back-Cover Texts.  A
copy of the license is included in the section entitled "GNU Free
Documentation License".

The library and utility program is copyright @copyright{} 2000--2004 by
Kevin Atkinson.  You can redistribute it and/or modify it under the
terms of the GNU Lesser General Public License as (LGPL) published by
the Free Software Foundation; either version 2.1 of the License, or
(at your option) any later version.

Certain parts of the library, as indicated at the top of the source
file, are under a weaker license.  However, all parts of the library
are LGPL Compatible.

@menu
* GNU Free Documentation License::
* GNU Lesser General Public License::
@end menu

@include fdl.texi
@include lgpl.texi

@c @node Index
@c @unnumbered Index

@c @printindex cp

@bye