unifont-6.3.20131215.tar.gz

[unifont.git] / doc / unifont.texi

\input texinfo   @c -*-texinfo-*-
@c %**start of header
@setfilename unifont.info
@settitle Unifont
@c %**end of header

@paragraphindent none

@copying
This tutorial describes Unifont, a bitmap-based font covering the
Unicode Basic Multilingual Plane, and its utility programs.

Copyright @copyright{} 2008--2013 Paul Hardy

@quotation
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with no
Invariant Sections, with no Front-Cover Texts and no Back-Cover Texts.
@end quotation

@end copying

@dircategory Fonts

@direntry
* Unifont (unifont).  A bitmap-based font covering the Unicode BMP.
@end direntry

@titlepage

@title Unifont
@author Paul Hardy
@page
@vskip 0pt plus 1filll

@insertcopying

@end titlepage

@contents


@node Top, Introduction, (dir), (dir)

@menu
* Introduction:: General overview.
* Tutorial:: Tutorial on Unifont utilities and Unifont modification.
* Reference:: Detailed description of each Unifont utility.
@end menu


@node Introduction, Tutorial, Top, Top
@chapter Introduction


This document describes the process of using the GNU Unifont utilities
to create a font.  The steps described in the "Using Graphical Tools"
section in the "Tutorial" chapter are more or less the steps that
I (Paul Hardy) followed to add thousands of glyphs to GNU Unifont,
except that I didn't have the luxury of just typing @code{make}
to make a new font while adding those glyphs.

I streamlined the font build process after I was done drawing the
Unicode 5.1 glyphs.

I know that plain ASCII text is *so* last millennium, especially for
a package related to Unicode.  Yet ASCII can be read with anything;
hence this format.

If you have questions, please email
@code{unifoundry@@unifoundry.com}.
You can check for the latest Unifont news at
@code{http://savannah.gnu.org/projects/unifont} and
@code{http://unifoundry.com}.
You can also submit a bug report through the
@code{http://savannah.gnu.org/projects/unifont} page.

DISCLAIMER: Donald Knuth warned in his Metafont book that if someone
started designing type, they would never again be able to look at
a page of text normally and just read its content.  There is a
point of no return beyond which a serious font designer begins
looking at how individual letters in a font on a page are drawn,
and how they might be improved.  Be warned!


--- Paul Hardy (@code{unifoundry@@unifoundry.com})  2008, 2013


@node Tutorial, Reference, Introduction, Top
@chapter Tutorial

This chapter provides a step-by-step tutorial on using the Unifont
utility programs to modify a font in the GNU Unifont format.

@menu
* Unicode:: Brief Overview of The Unicode Standard.
* Unifont Structure:: The format of Unifont files.
* Hex File Format:: The @code{unifont.hex} file format.
* Using Graphical Tools:: The Unifont graphical utilities.
* Using Hexdraw:: The Unifont ASCII utility for text editors.
* Checking Coverage:: Checking Unicode Basic Multilingual Plane coverage.
* Custom Builds:: Customizing the composition of a Unifont derivative.
* Seeing the Big Picture (Literally!):: Creating a Unifont poster.
* Combining Circles:: Glyphs with zero width.
* Installing Fonts on GNU/Linux:: font installation on Unix/Linux.
* Creating a Brand New Font:: advice on adding a new Unicode script.
* Updates to Unicode:: modifying Unifont for Unicode updates.
@end menu

@node Unicode, Unifont Structure, Tutorial, Tutorial
@section Unicode


Unicode is an international standard to encode all the world's
scripts with one universal scheme.  Unicode is the default encoding
for web pages and is gaining popularity in many other applications.
To learn more about Unicode, look at code charts, and see the
latest developments, check out

@example
http://unicode.org
@end example

Unifont follows the Unicode encoding scheme.  Unicode
defines the numeric value of a character, but does not define
one particular font.  There can be (and are) many fonts that
support a subset of Unicode characters.

In 1998, Roman Czyborra observed that there was still no font,
free or commercial, with complete Unicode coverage.  He envisioned
a low-quality bitmapped font as an easy way to produce a font
that covered much of the Unicode standard.


@node Unifont Structure, Hex File Format, Unicode, Tutorial
@section Unifont Structure


GNU Unifont is a dual-width pixel font.  Roman Czyborra
began this font in 1998 with a goal of having one glyph
rendered for each visible character in the Unicode Basic
Multilingual Plane (Plane 0, the first 65,536 characters).
His original writing on this is at @code{http://czyborra.com/unifont/}.

(Note that the term "character" is used very loosely here for
simplicity; the Unicode Standard has a stricter definition
of what constitutes a character.)

The font is dual-width.  Each character is 16 pixels tall, and
either 8 or 16 pixels wide.  The characters are stored in a
unique .hex file format invented by Roman Czyborra as a convenient
way of giving each character exactly a one line specification.
Conversion between this .hex format and BDF font format is trivial.


@node Hex File Format, Using Graphical Tools, Unifont Structure, Tutorial
@section Hex File Format


By convention, files containing the Unifont native font format
have the extension ".hex".  Their format is extremely simple, consisting
of two fields separated with a colon (":") and ending with a newline.

The first field is the Unicode code point, in hexadecimal.  For all
Plane 0 code points, this is a four digit hexadecimal number.  Hexadecimal
digits are (in order) 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A, B, C, D, E,
and F.  The Unicode Standard uses a hexadecimal number to assign
each character a location.  These locations are called "code points"
and their range is 0 through 10FFFF, inclusive.

The range 0 through FFFF, inclusive, is called the Basic Multilingual
Plane (BMP), or Plane 0.  This plane contains glyphs for most of
the world's modern writing scripts.

Because the original goal of GNU Unifont was to cover the entire
Unicode Basic Multilingual Plane, a four digit hexadecimal number
suffices.  In the future, I plan to allow the utilities to go beyond
the BMP now that coverage of the Unicode BMP is finally complete.

The programs will probably handle glyphs beyond the BMP properly, but
that capability is considered experimental, as the focus was to cover
the BMP.  The C programs allow up to an eight digit hexadecimal code
point, but not all the utilities have been thoroughly tested beyond
the BMP.

The second field is a string of hexadecimal digits.  There are 32
digits for a character that is 8 pixels wide, and 64 digits for a
character that is 16 pixels wide.

The good news is you don't have to worry about these long digit
strings.  Roman Czyborra wrote a utility, @code{hexdraw}, to convert
.hex fonts to a form that can be edited with a plain text editor,
then converted back into .hex format.

Paul Hardy wrote two utilities to do the same thing except with
bitmapped graphics images for editing with a graphics editor:
@code{unihex2bmp} converts a block of 256 characters into a graphics
file, and @code{unibmp2hex} converts such a graphics file back into
.hex format.  These bitmaps display the 256 characters in a block
arranged in a 16 by 16 character grid.  The graphics editor must
maintain the image as a monochrome (black and white) file, with
one bit per pixel.  After conversion from a .bmp file back to
a .hex file, the next step is conversion to a BDF font file.  A BDF
file can only encode a pixel being on or off (i.e., black or white
only with no intermediate shades of gray).

Andrew Miller later converted @code{unihex2bmp} and @code{unibmp2hex}
to Perl, then transformed them into
@code{unihex2png} and @code{unipng2hex}, respectively.  These programs
convert Unifont .hex files to and from Portable Network Graphics files.


@node Using Graphical Tools, Using Hexdraw, Hex File Format, Tutorial
@section Hex File Format


Let's look at an example.  Suppose you want to modify the Coptic letters
in the range U+2C80..U+2CFF ("U+" is Unicode shorthand).  These
letters are in the upper half of the block U+2C00..U+2CFF.  The
Unicode utilities in this package refer to this as "page" 2C.
("Page" is not a Unicode term --- it is just a term unique to this
package to refer to a block of 256 code points/characters).

The steps to follow will be:

@enumerate
@item
Convert .hex version of the page 2C range as a 16 by 16
bitmapped grid.

@item
Modify the bitmap in any graphics editor, being careful
to re-save it as a Windows Bitmap (.bmp) or Wireless
Bitmap file when finished.

@item
Convert the modified bitmap back into a .hex font file.

@item
Merge the results with the original @code{unifont.hex} file
(or whatever its name might be).

@item
Run @code{unidup} on the resulting file to guard against
duplicate character definitions.

@item
Create the new bitmapped version of the font.

@item
Check the compiled font for duplicates.

@item
If there are duplicates, remove them and go back to Step 5.

@item
Create the new TrueType version or other versions of the font.

@end enumerate

@strong{Step 1:} Convert the .hex range into a bitmap grid.
Assuming our font file is named @code{unifont.hex}, type

@example
unihex2bmp -p2C < unifont.hex > uni2C.bmp
@end example

@strong{Step 2:} Modify @code{uni2C.bmp} with your favorite graphics editor.
Note that whatever graphics editor you use must preserve the
file as a black and white bitmap (monochrome), with one bit
per pixel.  During editing, you can draw guidelines outside
the actual 16x16 font pixel area; they will be ignored when
converting back into .hex format.  You can also erase the
grid borders between code points on purpose or by accident,
and it will have no effect on the generated .hex file.  Just
don't erase the code point numbers on the outer edges of
the grid.  The conversion from .bmp back to .hex only looks
at the "U+0000" in the upper left-hand corner of the bitmap graphic
and other code point numbers, and at each code point's
16x16 pixel area inside its 32x32 pixel grid area.  Every other
artifact in the final graphics file outside these areas is ignored.

If a new version of Unicode adds glyphs to a page that were
previously unassigned, be sure to remove the newly-assigned
code points from the @code{unassigned.hex} file because the code
point is no longer unassigned.

@strong{Step 3:} Convert the edited .bmp file back into .hex format:

@example
unibmp2hex < uni2C.bmp > uni2C.hex
@end example

@noindent
Note that the conversion from a bitmap image to a .hex file
can't distinguish between a legitimate single- or double-width
space character and a code point that does not have an assigned
value.  Therefore, space glyphs are separately contained in the
@code{spaces.hex} file.

@strong{Step 4:} Merge the results with the original @code{unifont.hex} file.
This requires several sub-steps:

@itemize
@item
Edit the original @code{unifont.hex} file and delete the
lines that begin with "2C".

@item
Insert the @code{uni2C.hex} file into @code{unifont.hex}, either with
a text editor such as @code{emacs} or @code{vi}, or with a GNU/Linux
command such as:
       
@example
sort uni2C.hex unifont.hex > new-unifont.hex
@end example

This second option (using @code{sort}) is preferred, because
@code{unidup} (in Step 5) might miss duplicate code points
if your final result isn't in proper order.

@end itemize


@strong{Step 5:} Make sure there are no duplicates with @code{unidup}:

@example
unidup < unifont.hex
@end example

@noindent
or

@example
unidup < new-unifont.hex
@end example

@noindent
depending on the name of your final font file.  If there
is no output, your modified font contains no duplicates.

This editing is best done on an input .hex file, such as
@code{unifont-base.hex}.

@strong{Step 6:} Create the new bitmapped version of the font.  In the
@code{font/} directory, type

@example
make hex
@end example

@strong{Step 7:} Check the compiled font for duplicates.  Change to the
@code{font/compiled/} directory and run

@example
unidup < mynewfontfile.hex
@end example

@noindent
for whatever font file you created.

@strong{Step 8:} If there are duplicates, remove them in the @code{font/}
directory and go back to Step 5.

@strong{Step 9:} Create the new TrueType version of the font and all other
bitmapped versions.  From the @code{font/} directory, type

@example
make distclean && make
@end example

@noindent
Then be prepared to wait a long time unless you are using
a computer with plenty of RAM and CPU horsepower.  Your
computer should have at least 256 Megabytes of virtual
memory (RAM), and at least 250 Megabytes of free disk space.

To only create a BDF font, in the @code{font/} directory just type

@example
make bdf
@end example

To only create a BDF and PCF font, in the @code{font/} directory type

@example
make pcf
@end example

Creating a BDF font is the first step in creating a PCF font
(not counting generating the compiled master ".hex" input file).
BDF fonts can be created just with the tools in this package.
PCF fonts are created by running @code{bdftopcf} on the BDF font.
TrueType fonts require FontForge.

The Unifont package also includes two new programs
for working with Portable Network Graphics (PNG) files instead
of BMP files.  These utilities are @code{unihex2png} and
@code{unipng2hex}.  They work in a similar manner to the corresponding
programs @code{unihex2bmp} and @code{unibmp2hex}, respectively.

To use @code{unihex2png} instead of @code{unihex2bmp}, continuing
the example of the Coptic script in the U+2Cxx range, type:

@example
unihex2png -p2C -i unifont.hex -o uni2C.png
@end example

Note that with @code{unihex2bmp} specifying input and output files
is optional, while with @code{unihex2png} they must be specified
explicitly.  More specifically, @code{unihex2png} will read a .hex
file format input from STDIN if no "-i" argument is specified,
but the name of the binary PNG file must always be specified with
the "-o" option.

Then edit the resulting PNG file to your heart's content.  When done,
convert the file back into a @code{unifont.hex} format file.  In
this example, type:

@example
unipng2hex -i uni2C.png -o uni2C.hex
@end example

Similar to @code{unihex2png}, the binary PNG file must be specified
with "-i" but the .hex format file will be written to STDOUT if the
"-o" option is omitted.

Finally, merge  your changes in with your main .hex font file as
described previously in this section.


@node Using Hexdraw, Checking Coverage, Using Graphical Tools, Tutorial
@section Using Hexdraw


Roman Czyborra's original utility to edit glyphs is the @code{hexdraw}
Perl script.  Using the same script as in the previous chapter, Coptic,
here are the steps for modifying @code{unifont.hex} using @code{hexdraw}.

First, realize that Unifont has tens of thousands of glyphs
(characters, using the term character loosely).  In this example,
out of the tens of thousands of glyphs, we want to modify the range
U+2C80..U+2CFF (only 128 glyphs).

The range U+2C80..U+2CFF could be extracted from @code{unifont.hex} by
using the @code{egrep} utility to look for lines beginning with "2C8"
through "2CF", or that range could be isolated by copying @code{unifont.hex}
into another file, and deleting all lines except the desired range.

The following steps will probably minimize typographical errors,
but they aren't the only way.

@enumerate
@item
"Grep" the desired block of 256 glyphs (using the @code{grep} utility)
and convert this into a text representation for editing.

@item
Edit the block with a text editor, such as @code{emacs} or @code{vi}.

@item
Convert the edited text file back into .hex format.

@item
Delete the edited block range from the original font file.

@item
Merge the two .hex files into one file.

@item
Check for duplicates with @code{unidup}.

@item
Generate new fonts as described in the "Using Graphical Tools" section above.

@end enumerate

@strong{Step 1:} Extract the desired block with @code{grep}:

@example
grep "^2C" unifont.hex | hexdraw > uni2C.txt
@end example

@strong{Step 2:} Edit @code{uni2C.txt} with a text editor.

@strong{Step 3:} Convert the text file back into .hex format:

@example
hexdraw < uni2C.txt > uni2C.hex
@end example

@strong{Step 4:} Delete the lines in the original @code{unifont.hex}
file that begin with "2C".

@strong{Step 5:} Merge the two files:

@example
sort unifont.hex uni2C.hex > new-unifont.hex
@end example

@noindent
or use Roman's @code{hexmerge} utility:

@example
hexmerge unifont.hex uni2C.hex > new-unifont.hex
@end example

@strong{Step 6:} Check for duplicates:

@example
unidup < new-unifont.hex
@end example

@noindent
Of course, remove any reported duplicates.

@strong{Step 7:} Build the font as in the "Using Graphical Tools" section
above.  This can be as simple as typing

@example
make
@end example

@noindent
in the @code{font/} directory.

I (Paul Hardy) had only used @code{hexdraw} in the very beginning of my
work on Unifont.  Once I got my graphics programs working,
I ignored it for months.  Then I wanted to go through all of the
Yi Syllables and Yi Radicals --- over 1000 glyphs --- to fine-tune
their horizontal alignment after I drew them.  @code{hexdraw} turned out
to be the perfect tool for this.  By printing hyphens ("-") as
place holders where a pixel is off, it allowed me to verify space
to the left and right of each character.  I later used @code{hexdraw}
for similar fine-tuning of spacing on Hangul and other glyphs.
It is ideal for the task.


@node Checking Coverage, Custom Builds, Using Hexdraw, Tutorial
@section Checking Coverage


There should never be duplicates in a .hex file.  If there are, remove
them before the .hex font is turned into a BDF or other font file.  The
recommendations above include making liberal use of @code{unidup} to avoid
such a situation.

The @code{unipagecount} program will print a hexadecimal number of code
points that have coverage within each 256 code point block.  The
hexadecimal number will therefore range from 0 (no coverage) to
100 (= 256 decimal; full coverage).  If a number is ever more than
100 hexadecimal, there's an extra character (glyph) for that page.

To further look at the coverage within just one 256 code point
page (using page 2C, containing Coptic, as our example) use

@example
unipagecount -p2C < unifont.hex
@end example

Note that the "page number" can use upper- or lower-case letters:
@code{-p2C} or @code{-p2c} will both work.  That will print a 16 x 16 grid
of U+2C00..U+2CFF.  Of course, without placeholder glyphs for the
unassigned code points from @code{unassigned.hex} in the U+2C00..U+2CFF
range, unipagecount can give a lower number than the true coverage.

Using the @code{-l} flag with @code{unipagecount} will produce an HTML
table with links to corresponding graphics images.  You can get
an idea of how this works in the @code{font/compiled/} directory after
running @code{make}; the @code{index.html} file in that directory will have
a table with links to the 256 glyph maps in the @code{font/compiled/bmp/}
subdirectory.

With @code{unipagecount}, the background color of the cells will range from
red (for 0% complete in that 256 code point block) to orange
(a little coverage) to yellow (more coverage) to green
(complete coverage).  If a cell looks light red or pink,
the corresponding code page probably has duplicate characters.
Verify that with

@example
sort unifont.hex | unidup
@end example

@noindent
(substituting the name of your .hex file for @code{unifont.hex}).

To see the coverage of each Unicode script, copy the file
@code{font/coverage.dat} into the same directory as the
@code{unifont.hex} file you're examining.  Then run

@example
unicoverage < unifont.hex > coverage.out
@end example

This will give you all the scripts within the Unicode Basic
Multilingual Plane, in order, with a percentage (0.0% through
100.0%) of each script's coverage.  Note that to get the true
coverage of assigned code points, you'll have to merge @code{unassigned.hex}
with the rest of @code{unifont.hex} if not done by default in your setup.

Using the .hex files in @code{font/hexsrc/}, you can create a font with
all available glyphs with

@example
sort font/hexsrc/*.hex >unifont-whole.hex
@end example

@noindent
and run @code{unicoverage} using the resulting @code{unifont-whole.hex} file.


@node Custom Builds, Seeing the Big Picture (Literally!), Checking Coverage, Tutorial
@section Custom Builds


The font can be built from within the @code{font/} directory by simply typing

@example
make
@end example

@noindent
From the top-level directory (one level above the @code{font/} directory),
typing

@example
make BUILDFONT=1
@end example

@noindent
will also build the font.  The font is not built by default by typing
@code{make} from the top-level directory, because a pre-built version
already exists in the @code{font/precompiled/} directory.  Font files
are architecture-independent, so the only reason to build the font is
if you modify its composition.

By default, source glyphs are read from the @code{font/hexsrc/} directory.
Glyphs for unassigned code points are built into the font by default,
but glyphs for Private Use Area code points are not built into the font.

All of the .hex file names can be replaced selectively on the
@code{make} command line to override their default values.
Their locations are relative to the @code{font/} directory.
The list of component hex file variables is:

@table @code

@item UNIFONTBASE
The bulk of Unifont scripts

@item CJK
Most of the CJK Ideographs

@item HANGUL
Hangul Syllables block

@item SPACES
Space glyphs, single- and double-width

@item UNASSIGNED
Glyphs for unassigned code points

@item PUA
Glyphs for the Private Use Area

@end table

So, for example, to build a font that includes four-digit hexadecimal
code point glyphs (as white digits on a black background) for the
Private Use Area, type

@example
make PUA="hexsrc/pua.hex"
@end example

@noindent
because those glyphs reside in the @code{font/hexsrc/pua.hex} file.

To build a font that includes your own special PUA glyphs, type

@example
make PUA="mycoolPUA.hex"
@end example

@noindent
or whatever the name of your PUA glyph .hex file is named.

To create a build that includes the supplied PUA glyphs but not the
unassigned code point glyphs, type

@example
make PUA="hexsrc/pua.hex" UNASSIGNED=""
@end example

If you create a custom font build of your own in one gigantic file,
you can build with just this file by declaring all the ordinary files
to be null:

@example
make UNIFONTBASE="mycustomfont.hex" \ @*
CJK="" HANGUL="" UNASSIGNED="" PUA=""
@end example

Note that this command did not include an override for the glyphs for spaces
contained in the @code{font/hexsrc/spaces.hex} file; that is, the variable
SPACES was not redefined on the command line.  You could also pass the
argument SPACES="", but just be aware that those spaces glyphs are in
a separate file for a reason.  When graphical (".bmp") glyph files are
converted back into hex string (".hex") format, the @code{unibmp2hex} utility
doesn't know if a blank glyph area is a space glyph or not, so it doesn't
encode anything.  The @code{font/hexsrc/spaces.hex} file contains glyphs that
are strings of 0s, with length depending on whether the space is nominally
a single- or double-width space.  (Unifont does not distinguish between
finer spacing used in traditional typesetting, such as a thin space, en space,
em space, or quad space; all spaces are either 8 pixels wide or 16 pixels
wide.)


@node Seeing the Big Picture (Literally!), Combining Circles, Custom Builds, Tutorial
@section Seeing the Big Picture (Literally!)


The GNU Unifont 6.3 release introduced a new program, @code{unifontpic}.
This produces a chart of all the Basic Multilingual Plane glyphs in
Unifont.  By default the chart is arranged as a 256-by-256 glyph
table.  Specifying the @code{-l} option produces a chart that is
16 glyphs wide by 4,096 glyphs long.  See unifontpic(1) for more
information.

The "long" version, created with @code{unifontpic -l}, only produces
16 glyphs per row.  This is more useful for scrolling through on
a computer screen.

GIMP, the GNU Image Manipulation Program, will properly display
the resulting long .bmp file (at least under GNOME), but not all
graphics utilities can.  The output file is over 65,536 pixel rows tall,
which causes problems with some graphics programs.

To generate a chart with all your glyphs in one giant @code{unifont.hex}
file, type the command

@example
unifontpic < unifont.hex > unifont.bmp
@end example

@noindent
The output is a monochrome Bitmap Graphics Format (.bmp) file.
If you prefer PNG files, use your favorite graphics program or
conversion program to convert the BMP file to a PNG file.

This utility is especially useful if you're customizing the font,
for example if adding your own Private Use Area glyphs.

The default 256-by-256 code point chart will render satisfactorily
on a sheet of paper approximately 3 feet by 3 feet (1 meter by 1 meter)
at 120 dots per inch.  Thus the square format is suitable for printing.


@node Combining Circles, Installing Fonts on GNU/Linux, Seeing the Big Picture (Literally!), Tutorial
@section Combining Circles


The earliest versions of Unifont (before the 5.1 release) used glyphs
that showed dashed circles for combining characters.  This is how the
glyphs appear in The Unicode Standard code charts.  In version 5.1,
Paul Hardy was able to get combining characters to appear superimposed
correctly in the TrueType version of the font.  There are no plans
to try to get combining characters to work in a BDF or PCF version
owing to the limitations of those bitmapped font formats.

With combining characters working, Paul Hardy created variations of
Unifont with and without combining circles, the idea being that the
version without combining circles would be used to create the TrueType
font.  The variation with combining circles was kept for reference.

Unifont Version 6.2 simplified the build to produce only one font variation,
without combining circles.

Unifont Version 6.3 introduced a new utility, @code{unigencircles},
to superimpose combining circles over glyphs with code points in
a @code{combining.txt} file.

The latest Unifont release contains a parallel set of font files
named @code{unifont_sample.*}.  These "Unifont Sample" font files
contain glyphs with combining circles where appropriate.  The
"Unifont Sample" font is therefore not intended for general-purpose
writing, but only for illustrating each individual glyph as represented
in The Unicode Standard.

To run @code{unigencircles}, start with the file
@code{font/ttfsrc/combining.txt} and type a command of this form:

@example
unigencircles combining.txt < unifont.hex > unifont-circles.hex
@end example

@noindent
where @code{unifont.hex} is a single file containing all the glyphs you
wish to render.  You could create such a file from the @code{font/}
directory with the command

@example
sort hexsrc/*.hex >unifont.hex
@end example

Because the output is another .hex file, you can use all Unifont
utilities with this resulting file just as you would with the .hex files
in @code{font/hexsrc/}.

If you want to build a font from this generated @code{unifont.hex} file,
you could type

@example
make UNIFONTBASE="unifont-circles.hex" CJK="" HANGUL="" \ @*
UNASSIGNED="" PUA=""
@end example

@noindent
as discussed above.  In this case, if you included
@code{font/hexsrc/spaces.hex} in the @code{unifont.hex} input file,
you should also set SPACES="" on the command line so that you only
read in your final custom @code{unifont-circles.hex} file.


@node Installing Fonts on GNU/Linux, Creating a Brand New Font, Combining Circles, Tutorial
@section Installing Fonts on GNU/Linux


The original standard font format of Unifont was a BDF font.  The newer
PCF font format loads much faster when a program begins, and so is preferable
for installations using the X Window System and similar environments.

Compress PCF fonts using

@example
gzip -9 fontname.pcf
@end example

Copy the resulting @code{fontname.pcf.gz} file to
@code{/usr/share/fonts/X11/misc/} or place in a
local font directory if your windowing software supports that (for
example, @code{~/.fonts/}).

Copy TrueType fonts to @code{/usr/share/fonts/truetype/} uncompressed or place
in your local font directory.  Note: on some versions of Unix, such as
Solaris, the name of the TrueType directory might be TrueType and
might be under @code{/usr/share/fonts/X11/} --- examine the system fonts
directories, then modify the font @code{make install} rule accordingly.

On most flavors of GNU/Linux with the latest @code{xset} utility (including
the latest Debian and Red Hat releases), the following command should
allow you to start using the font immediately:

@example
xset fp rehash
@end example

The safest way to make sure the system knows about the new fonts will
be to restart the X Window System or even reboot.


@node Creating a Brand New Font, Updates to Unicode, Installing Fonts on GNU/Linux, Tutorial
@section Creating a Brand New Font


The original tools will only produce glyphs that are 16 pixels tall, and
either 8 or 16 pixels wide.  The utilities @code{unihex2png},
@code{unipng2hex}, @code{hexdraw}, and @code{hex2bdf} now also support
glyph heights of 24 and 32 pixels, and glyph widths of up to 32 pixels,
but this is not fully tested.  These new dimensions are currently
available for experimental use.  See the respective sections for each
of these programs in the Reference chapter of this document, or their
respective man pages.

This current release is furthermore only fully
tested over the Unicode Basic Multilingual Plane (code points from
U+0000 through U+FFFF, with none above that range --- this will change
in the future).  The tools can be used to create and manipulate
any bitmapped font within those constraints.

To create a brand new font (or even to add a new script to Unifont
in the future), plan out the basic dimensions of the characters:

@itemize

@item
How far above the lowest pixel will the baseline appear
(in other words, how many rows are necessary for descenders
such as in the glyphs for `g', `q', and `y')?

@item
How many pixels must be empty on top and bottom for accents
(in other words, what will the height of capital letters be)?

@item
Must glyphs be centered, left-aligned, or right-aligned?

@item
For a Latin font, what will the "x-height" be (the height of
a lower-case "x" and related letters, such as "n" and "r")?

@end itemize

Consistent capital heights, x-heights, descender depths, and centering
will produce a better looking font.  Look over the entire script and
plan out a template grid that is consistent for the entire script.
Then use that template for each glyph you draw for the script.

Unifont characters for the most part leave the right-most column
of pixels blank if possible for left-to-right scripts.  Centering is
done around the fourth pixel (if a glyph is eight pixels wide) or
around the eighth pixel (if a glyph is 16 pixels wide).

Experimenting and (above all) having fun with these utilities is
the best way to learn.


@node Updates to Unicode, , Creating a Brand New Font, Tutorial
@section Updates to Unicode


If a currently unassigned code point is assigned to a character
in the future, the font can be updated as follows:

@enumerate

@item
Delete the code point's entry from @code{font/hexsrc/unassigned.hex},
as that code point will no longer be unassigned.

@item
Determine which existing .hex file should contain the
newly defined character (examine the files to see
which one contains other glyphs in the script.

@itemize

@item
@code{unifont-base.hex} contains most scripts

@item
@code{wqy.hex} contains most CJK ideographs; its name pays homage
to the Wen Quan Yi font, the source of almost all of its glyphs

@item
@code{hangul-syllables.hex} contains the Hangul Syllables block
(U+AC00..U+D7A3); this should never have new code points added as
it covers the fixed range of the Unicode Hangul Syllables block

@item
@code{spaces.hex} contains the list of single- and double-width spaces

@end itemize

If in doubt (for example, if a new script is added to
Unicode and you're not sure which .hex file to augment),
add the new glyphs to @code{unifont-base.hex}.

@item
Update the appropriate .hex file.

@item
Consider if @code{font/coverage.dat} has to be updated.
Follow its existing format to insert a new script,
being sure to change any previously unassigned ranges
before or after the newly added script.

@item
Make a new .hex version of the font, and verify that
you didn't introduce any duplicates.

@item
Run @code{unipagecount} and/or @code{unicoverage} as described
previously to verify that you have not mistakenly deleted
any existing characters.

@end enumerate

Enjoy!


@node Reference, , Tutorial, Top
@chapter Reference

@menu
* bdfimplode::
* hex2bdf::
* hex2sfd::
* hexbraille::
* hexdraw::
* hexmerge::
* johab2ucs2::
* unibdf2hex::
* unibmp2hex::
* unicoverage::
* unidup::
* unifontchojung::
* unifontksx::
* unifontpic::
* unigencircles::
* unigenwidth::
* unihex2bmp::
* unihex2png::
* unihexgen::
* unipagecount::
* unipng2hex::
@end menu

@include bdfimplode.texi
@include hex2bdf.texi
@include hex2sfd.texi
@include hexbraille.texi
@include hexdraw.texi
@include hexmerge.texi
@include johab2ucs2.texi
@include unibdf2hex.texi
@include unibmp2hex.texi
@include unicoverage.texi
@include unidup.texi
@include unifontchojung.texi
@include unifontksx.texi
@include unifontpic.texi
@include unigencircles.texi
@include unigenwidth.texi
@include unihex2bmp.texi
@include unihex2png.texi
@include unihexgen.texi
@include unipagecount.texi
@include unipng2hex.texi

@bye