NAME¶

otftotfm - create TeX font metrics from OpenType fonts

SYNOPSIS¶

otftotfm [-a] [options] fontfile [texname]

DESCRIPTION¶

Otftotfm creates the font metric and encoding files required to use an OpenType font with TeX. You supply an OpenType ".otf" or ".ttf" font file, a base ".enc" encoding, and a TeX name "texname" for the resulting font, and say which OpenType features should be turned on. Then otftotfm generates and installs the corresponding TeX-related metric files (".tfm" TeX font metrics, ".vf" virtual fonts, and ".enc" encoding files). It works on both PostScript-flavored and TrueType-flavored OpenType fonts, although TrueType-flavor support will only work easily with pdftex.

The easiest way to use otftotfm is with the -a option; see Automatic Mode below. Without -a, otftotfm writes all its output files to the current directory.

After running "otftotfm fontfile texname" and installing the results (manually or with -a), you can use the OpenType font in plain TeX with a command like this:

    \font\myfont=texname at 10pt


    {\myfont This text uses the OpenType font.}

LaTeX users will generally make a ".fd" input file so that commands like "\renewcommand{\rmdefault}{TeXName}" work correctly. See the EXAMPLE section for more; check the DIAGNOSTICS and FREQUENTLY ASKED QUESTIONS sections if you have trouble.

EXAMPLE¶

This section uses MinionPro to show one way to install OpenType fonts for LaTeX. We begin with six fonts: "MinionPro-Regular.otf", "MinionPro-It.otf", "MinionPro-Semibold.otf", "MinionPro-SemiboldIt.otf", "MinionPro-Bold.otf", and "MinionPro-BoldIt.otf".

Our first task is to decide how to encode the fonts. The "encoding scheme" is used by TeX to decide how to typeset accents and symbols like "$". The "LY1" encoding scheme has reasonable accent support and is a good choice for many OpenType fonts. LY1 corresponds to the "texnansx.enc" encoding file, so we will supply otftotfm with the "-e texnansx" option.

Next, we decide on a naming scheme for the font metric files. Let's use the OpenType font names as a base. (There's generally no need to follow the six-character "Karl Berry" naming scheme.) Just in case we come back later and add a different encoding scheme, we'll prepend "LY1--" to each name.

We're now ready to run otftotfm for the first set of fonts. Note the "-fkern -fliga" options, which access pair kerns and the default "f" ligatures.

    otftotfm -a -e texnansx MinionPro-Regular.otf \


            -fkern -fliga LY1--MinionPro-Regular


    otftotfm -a -e texnansx MinionPro-It.otf \


            -fkern -fliga LY1--MinionPro-It


    otftotfm -a -e texnansx MinionPro-Semibold.otf \


            -fkern -fliga LY1--MinionPro-Semibold


    otftotfm -a -e texnansx MinionPro-SemiboldIt.otf \


            -fkern -fliga LY1--MinionPro-SemiboldIt


    otftotfm -a -e texnansx MinionPro-Bold.otf \


            -fkern -fliga LY1--MinionPro-Bold


    otftotfm -a -e texnansx MinionPro-BoldIt.otf \


            -fkern -fliga LY1--MinionPro-BoldIt

The small-caps fonts are generated with an additional "-fsmcp" option. We append "--fsmcp" to the font metric names as well, differentiating them from the regular fonts. Although MinionPro's italic fonts support small-caps, the LaTeX font selection scheme can't access them easily, so we've left them off.

    otftotfm -a -e texnansx MinionPro-Regular.otf \


            -fkern -fliga -fsmcp LY1--MinionPro-Regular--fsmcp


    otftotfm -a -e texnansx MinionPro-Semibold.otf \


            -fkern -fliga -fsmcp LY1--MinionPro-Semibold--fsmcp


    otftotfm -a -e texnansx MinionPro-Bold.otf \


            -fkern -fliga -fsmcp LY1--MinionPro-Bold--fsmcp

To get old-style numerals, just add the "-fonum" option to each invocation -- and, to reduce confusion, append "--fonum" to the font metric names.

At this point, all our font metric files are installed, and it's finally time to create the ".fd" file. (The ".fd" format is documented in The LaTeX Companion.) Let's call the LaTeX font family "MinionPro". Then the ".fd" file is "LY1MinionPro.fd", and it contains:

    \DeclareFontFamily{LY1}{MinionPro}{}


    \DeclareFontShape{LY1}{MinionPro}{m}{n}%


            { <-> LY1--MinionPro-Regular }{}


    \DeclareFontShape{LY1}{MinionPro}{m}{it}{ <-> LY1--MinionPro-It }{}


    \DeclareFontShape{LY1}{MinionPro}{m}{sc}%


            { <-> LY1--MinionPro-Regular--fsmcp }{}


    \DeclareFontShape{LY1}{MinionPro}{sb}{n}%


            { <-> LY1--MinionPro-Semibold }{}


    \DeclareFontShape{LY1}{MinionPro}{sb}{it}%


            { <-> LY1--MinionPro-SemiboldIt }{}


    \DeclareFontShape{LY1}{MinionPro}{sb}{sc}%


            { <-> LY1--MinionPro-Semibold--fsmcp }{}


    \DeclareFontShape{LY1}{MinionPro}{b}{n}{ <-> LY1--MinionPro-Bold }{}


    \DeclareFontShape{LY1}{MinionPro}{b}{it}%


            { <-> LY1--MinionPro-BoldIt }{}


    \DeclareFontShape{LY1}{MinionPro}{b}{sc}%


            { <-> LY1--MinionPro-Bold--fsmcp }{}


    \DeclareFontShape{LY1}{MinionPro}{bx}{n}%


            { <-> ssub * MinionPro/b/n }{}


    \DeclareFontShape{LY1}{MinionPro}{bx}{it}%


            { <-> ssub * MinionPro/b/it }{}


    \DeclareFontShape{LY1}{MinionPro}{bx}{sc}%


            { <-> ssub * MinionPro/b/sc }{}

We're now ready to use MinionPro in LaTeX, with lines like this in the document preamble:

    \usepackage[LY1]{fontenc}


    \renewcommand{\rmdefault}{MinionPro}


    \renewcommand{\bfdefault}{b}

Of course, we're free at any time to add more MinionPro variants with otftotfm; they'll become accessible to LaTeX as soon as we edit the "MinionPro.fd" file.

OPTIONS¶

With long options, you need type only as many characters as will make the option unique.

ENCODINGS¶

Otftotfm interprets encoding files as Unicode. For example, say an input encoding has "/dotlessi" at position 10. Otftotfm detects that position 10 should contain Unicode character U+0131 LATIN SMALL LETTER DOTLESS I, and uses the font's glyph for that character (possibly modified by any active features). The selected glyph might not be named "dotlessi"; only the Unicode value matters.

Otftotfm assigns Unicode values to glyph names using a table published by Adobe (SEE ALSO has a reference), with extensions for TeX. For more fine-grained control, add UNICODING commands to the input encoding file. These commands have the following format:

    % UNICODING glyph =: choice1 [choice2 ...] ;

    % UNICODING pi1 =: uni03D6 ;

    % UNICODING Delta =: uni0394 uni2206 ;

    % UNICODING Delta =: Deltagreek Delta ;

    % UNICODING ff =: ; fi =: ; fl =: ; ffi =: ; ffl =: ;

    % UNICODING ff =: ff emptyslot ;

LIGKERN comments in the encoding can add ligatures and inhibit kerns, as in afm2tfm(1). To add a ligature, say:

    % LIGKERN glyph1 glyph2 =: result ;

    % LIGKERN glyph1 {} glyph2 ;

    % LIGKERN a {} * ;

    % LIGKERN * {} * ;

Otftotfm also supports extended syntax for setting kern values and inhibiting ligatures. To add an n-unit kern between two glyphs, say:

    % LIGKERNX glyph1 {n} glyph2 ;

    % LIGKERNX glyph1 {L} glyph2 ;

You can set the --boundary-char and --altselector-char from an encoding file with commands like this:

    % LIGKERN || = boundarychar ;


    % LIGKERNX ^^ = altselectorchar ;

Otftotfm has a default set of eight ligatures, namely:

    space l =: lslash ; space L =: Lslash ;


    question quoteleft =: questiondown ; exclam quoteleft =: exclamdown ;


    hyphen hyphen =: endash ; endash hyphen =: emdash ;


    quoteleft quoteleft =: quotedblleft ;


    quoteright quoteright =: quotedblright

The POSITION command shifts a glyph within its bounding box. The syntax is

    % POSITION glyph pdx pdy adx ;

    % POSITION T 10 0 20

    % POSITION degree 0 20 0

The CODINGSCHEME command specifies the coding scheme for fonts using this encoding. This is a string, less than 40 characters long and containing no parentheses, that classifies the encoding for TeX's purposes. Sample coding schemes include "TEX TEXT", "TEX MATH ITALIC", and "EXTENDED TEX FONT ENCODING - LATIN". For example:

    % CODINGSCHEME EXTENDED TEX FONT ENCODING - LATIN

Finally, the WARNMISSING command makes any glyphs not supported by the input font appear as black boxes. The dvips(1) processor will also print a warning when encountering these glyphs. For example:

    % WARNMISSING yes

The --unicoding, --ligkern, --position, --coding-scheme, and --warn-missing options add UNICODING, LIGKERN/LIGKERNX, POSITION, CODINGSCHEME, and WARNMISSING commands to an encoding, and can override commands in the encoding itself. Some common encoding files have commands that are inappropriate for OpenType fonts; for example, "t1.enc" hard-codes f-ligatures, which can cause problems with small-cap fonts. Supply the --no-encoding-commands option to ignore all commands from the encoding file. Commands from options like --ligkern are processed in any case.

GLYPH PATTERNS¶

The --include-subs and --include-alternates options, and their --exclude and --*-filter variants, accept the following types of pattern.

• Glyph names. Example: "Aacute". For PostScript-flavored fonts, use otfinfo(1)'s -g option to see a font's glyph names, and "cfftot1 font.otf | t1testpage" to generate a PostScript file showing each glyph.
• Glyph name patterns using the shell-style glob-matching rules: "*" matches any number of characters, "?" matches any single character, and "[...]" matches any character in a set. Example: "*.end".
• Unicode category properties in angle brackets. Examples: "<Letter>", "<UppercaseLetter>", "<Lu>". The complete list of both short and long names: Letter/L, UppercaseLetter/Lu, LowercaseLetter/Ll, TitlecaseLetter/Lt, ModifierLetter/Lm, OtherLetter/Lo; Number/N, DecimalNumber/Nd, LetterNumber/Nl, OtherNumber/No; Punctuation/P, ConnectorPunctuation/Pc, DashPunctuation/Pd, OpenPunctuation/Ps, ClosePunctuation/Pe, InitialPunctuation/Pi, FinalPunctuation/Pf, OtherPunctuation/Po; Symbol/S, MathSymbol/Sm, CurrencySymbol/Sc, ModifierSymbol/Sk, OtherSymbol/So; Mark/M, SpacingMark/Mc, EnclosingMark/Me, NonspacingMark/Mn; Separator/Z, SpaceSeparator/Zs, LineSeparator/Zl, ParagraphSeparator/Zp; Other/C, Surrogate/Cs, Format/Cf, Control/Cc, PrivateUse/Co, Unassigned/Cn. Category values current as of Unicode 4.0.
• Unicode ranges. Example: "U+007f-U+008C".

The "!" prefix negates a pattern, and you can separate multiple patterns by spaces.

FEATURE DIRECTORY¶

This section lists features common to Western OpenType fonts and describes how otftotfm handles them for common fonts. Please send the author mail if otftotfm does not handle a feature you need, or you believe it handles some feature incorrectly.

aalt, Access All Alternates

Lets the user choose between all available alternate forms for a character. This includes things like superscript and subscript variants, different styles (swash, for example), and even ornaments. The --altselector-feature=aalt option can help an --altselector-char provide useful access to alternates, but the aalt feature isn't usually useful on its own. Try the salt and calt features instead.

c2sc, Small Capitals From Capitals

Replaces capital letters with small capitals: a sort of converse of the more conventional smcp feature, which replaces lower-case letters with small capitals. Supported.

calt, Contextual Alternates

Lets the user choose between context-appropriate swash forms for each character. For example, given the word "DREW" in a cursive typeface, the "R E W" might be translated to calmer forms than the initial "D". There may be more than one choice for a given letter, in which case the user should be able to select among them. TeX can't support complex contextual alternates, or alternate selection, but otftotfm supports some fonts quite well. The input encoding should have lots of empty space for variants, and it should specify a boundary character. See also cswh.

case, Case-Sensitive Forms

Shifts punctuation marks up to a position that works well with all-capital-letter sequences. For example, the hyphen character, which generally centers vertically on the x-height, is raised up to center vertically on a capital letter. Also replaces text figures with lining figures, and accent marks with forms more appropriate for capitals. Supported.

cpsp, Capital Spacing

Adds a bit of space on either side of each capital letter. Supported. (However, the OpenType tag registry suggests that cpsp be on by default, but applying to all-caps text only; TeX cannot easily implement that contextual intelligence.)

cswh, Contextual Swash

Lets the user choose between context-appropriate swash forms for each character. For example, in the words "Ab AC", the first "A" might be translated to a swash form, while the second might not. There may be more than one choice for a given letter, in which case the user should be able to select among them. Otftotfm supports some fonts quite well. The input encoding should have lots of empty space for swash variants, and it should specify a boundary character. See also calt and swsh.

dlig, Discretionary Ligatures

Activates uncommon ligatures, such as "c_t", "s_p", and "s_t". Supported.

dnom, Denominators

Replaces digits and some punctuation marks with smaller forms sitting on the baseline, intended for fraction denominators. Supported.

fina, Terminal Forms

Substitutes appropriate forms for letters occurring at the ends of words. This feature doesn't select swash variants; it's intended for normal use, and the specification recommends that it be on by default. Partially supported: TeX will only treat spaces as the ends of words, where a correct implementation would probably include punctuation too. See cswh for selecting swash variants active at the ends of words.

frac, Fractions

Replaces simple sequences like "1/2" with nice-looking fractions. Supported, but beware: many fonts will translate "11/32" into "1" + "1/3" + "2".

hist, Historical Forms

Replaces characters with historical variants. Usually, this means at least translating regular "s" to long "s". Supported.

kern, Kerning

Adjusts the space between characters (pair kerning). Generally supported, and you should probably turn it on. As a special case, "-fkern" can also read kerning information from the "kern" table in conventional TrueType fonts.

liga, Standard Ligatures

Activates common ligatures, such as "f_f", "f_i", "f_f_j", and (in some Adobe fonts) "T_h". Generally supported, and you should probably turn it on.

lnum, Lining Figures

Uses lining figures, the set of digits that are all about as high as capital letters. Supported. Compare onum; see also pnum and tnum.

numr, Numerators

Replaces digits and some punctuation marks with smaller, raised forms intended for fraction numerators. Supported, but not usually useful.

onum, Oldstyle Figures

Uses old-style figures, also known as text figures. This is the set of digits that have ascenders and descenders like lower-case letters. Supported. Compare lnum; see also pnum and tnum.

ordn, Ordinals

Designed for Spanish and French. Replaces ordinal numbers, such as "2.o", with forms where the "o" is raised, and replaces the sequence "No" with an integrated glyph. Supported.

ornm, Ornaments

Replaces some alphabetic characters in the font with ornaments, and links the bullet character to a set of all bullet-like ornaments, from which the user can choose. Partially supported: TeX can handle alphabetic substitutions, but not bullet choice.

pnum, Proportional Figures

Digits will have different widths. Supported. Compare tnum; see also lnum and onum.

salt, Stylistic Alternates

Lets the user choose between stylistic alternate forms for a character. The --altselector-char mechanism provides useful access to this feature. If you turn on salt globally, otftotfm takes the first alternate form whenever there's more than one choice. See also aalt and ss01; salt is generally more useful than aalt for TeX, since it refers exclusively to stylistic alternates.

sinf, Scientific Inferiors

Replaces digits and some punctuation marks with smaller, lowered forms intended for subscripts. Supported. Compare subs.

size, Optical Size

This feature stores information about the range of optical sizes for which the font was intended. There is no point in selecting it with otftotfm, since it should not change the font's appearance in any way.

smcp, Small Capitals

Replaces lower-case letters with small capitals. Supported. Compare c2sc.

ss01-ss20, Stylistic Sets 1-20

Replaces characters with a uniform set of stylistic alternates. Differs from features like salt in that a Stylistic Set is uniform: an ssXX feature should never involve selection from a set of possible alternate characters. Supported.

subs, Subscript

Replaces characters with smaller, lowered forms intended for subscripts. Supported. Compare sinf; some fonts support sinf but not subs.

sups, Superscript

Replaces digits, some punctuation marks, and some lower-case letters with smaller, raised forms intended for superscripts. Supported.

swsh, Swash

Activates all swash forms for each character. There may be more than one swash form, in which case otftotfm will pick the first one listed. Supported, except that swash variants other than the first are inaccessible. Note that some fonts with swash variants support the cswh feature exclusively.

tnum, Tabular Figures

All digits will have the same width, so that tables and the like will align visually. Supported. Compare pnum; see also lnum and onum.

zero, Slashed Zero

Replaces the zero character with a slashed zero. Supported.

DIAGNOSTICS AND TROUBLESHOOTING¶

no writable directory found in $TEXMF

Otftotfm could not find a writable directory in your $TEXMFVAR or $TEXMF path. Did you create a $HOME/.texmf-var or $HOME/texmf directory? If so, run the command "kpsewhich --expand-path='$TEXMF'" to verify that directory is not being found. You may need to set your TEXMF environment variable, to '{!!'"$HOME"'/texmf,!!$TEXMFMAIN}', for instance (note the different kinds of quotes; on my machine, this expands to '{!!/home/kohler/texmf,!!$TEXMFMAIN}').

'char' has no encoding, ignoring kern removal

(or ligature removal, lig/kern removal, or ligature)

These messages indicate a slight problem with your encoding file: one of the LIGKERN commands referred to a character not present in the encoding. This might be due to a misspelling in the LIGKERN command or the encoding file, or it might be an oversight. Either fix the encoding file or ignore the warning.

can't map 'char' to Unicode

Another encoding file problem: One of the glyph names in an UNICODING block could not be converted to Unicode. This is problematic since UNICODING exists wholly to translate glyph names into Unicode. Fix the encoding file or ignore the warning.

not enough room in encoding, ignoring N glyph(s) ...

There wasn't space in the encoding for all the glyphs referred to by the features you selected. For example, maybe the font had more ligatures than there were empty slots in the encoding. Fix this warning by selecting fewer features, or by using an encoding with more empty slots, such as the 7t.enc encoding distributed with otftotfm.

The '-a' option did not install my font correctly.

Try again with the '--verbose' option, which causes otftotfm to explain its behavior. Note that by default, otftotfm will not re-install files already present in your system's TeX search paths (in the current directory, for instance). Use '--force' to override this behavior.

FREQUENTLY ASKED QUESTIONS¶

How can I get a small-caps "SS" in place of the German sharp-S?

Supply the option '--unicoding "germandbls =: SSsmall"'.

How can I prevent f-ligatures from forming in a small-caps font?

This should happen automatically, but some overzealous encoding files add f-ligatures even when the font doesn't request them. Try the "--no-encoding-commands" option if this is a problem for you.

Otftotfm seems to take a long time.

Use the -V option to see what it's doing. Often the culprit is the updmap(1) program; if you're planning to run otftotfm multiple times, give it the --no-updmap option and run updmap manually when you're done.

How can I refer to the different forms of phi?

Otftotfm follows TeX practice and widely-distributed TeX encoding vectors, so "/phi" in an input encoding vector should map to a "straight" phi and "/phi1" should map to a "loopy" phi. Note that TeX practice differs from the PostScript standard naming conventions, in which "/phi" is "loopy" and "/phi1" is "straight"; this means that otftotfm may map "/phi" in an input encoding vector to a font's "/phi1" glyph, and vice versa. Perhaps most unambiguously, you can use "/uni03D5" for the "straight" form and "/uni03C6" for the "loopy" form.

How can I get lining figures (that is, normal line-height digits) with small caps ('-fsmcp')?

Many fonts use old-style figures by default with small caps. Since the default is not specified, it's wise to explicitly supply '-flnum' or '-fonum'.

BUGS¶

See the documentation for --pl above if you have problems running otftotfm's output through fontinst.

SEE ALSO¶

pltotf(1), tftopl(1), vptovf(1), afm2tfm(1), dvips(1), cfftot1(1), otfinfo(1), t1dotlessj(1), t1testpage(1), ttftotype42(1), kpsewhich(1), updmap(1)

Adobe Type 1 Font Format

Adobe Technical Notes #5176, The Compact Font Format Specification, and #5177, The Type 2 Charstring Format

OpenType Specification, Version 1.4

A Directory Structure for TeX Files, http://www.tug.org/tds/

Kpathsea: A library for path searching, http://www.tug.org/kpathsea/

Sivan Toledo, Exploiting Rich Fonts, TUGboat 21(2), 2000, http://www.tug.org/TUGboat/Articles/tb21-2/tb67tole.pdf

Michel Goossens, Frank Mittelbach, and Alexander Samarin, The LaTeX Companion (for information on the .fd file format)

Adobe Systems, "Unicode and Glyph Names". Refers to the glyphlist.txt file used to translate glyph names to Unicode code points. http://partners.adobe.com/public/developer/opentype/index_glyph.html

AUTHOR¶

Eddie Kohler (ekohler@gmail.com)

Thanks to Karl Berry, Marco Kuhlmann, Adam Lindsay, Bruce D'Arcus, Thomas Esser, Claire Connelly, Nelson H.F. Beebe, and Ryuji Suzuki for suggestions, bug reports, and help. Particular thanks to Achim Blumensath and Michael Zedler for suggestions and patches, some of them extensive.