Reference

ttfbuilder.plx(1p)

assemble a font from another font

NAME

ttfbuilder - assemble a font from another font

DESCRIPTION

ttfbuilder is a font subsetting program gone wild. It's aim is to allow a user to describe a new font in terms of the glyph pallette of a source font. Thus the new font may include ligatures of glyphs in the source font, or positional movements or whatever.

The main features of ttfbuilder are

•

Ability to create glyphs that are not in any cmap and to reference such glyphs via postscript name, glyph id or Unicode cmap entry.

•

Ability to work with an attachment points database. Thus ligatures are assembled by describing which attachment points should coincide, rather than having to give absolute locations in terms of shifting.

•

Ability to change the name of the font and change strings in the name table.

ttfbuilder is controlled via a description file which describes the glyphs in the new font, in terms of glyphs in the source font. This description file is an XML file with the following key elements:

glyph

This describes a glyph and the attributes allow setting of the postscript name and Unicode id for the glyph. The glyph element has children which describe what goes into the glyph.

The "overstrike" attribute controls whether an attachment to a base glyph can extend beyond origin or the advance width of the base glyph. The default is to allow the diacritics to extend and even occur to the left of the origin or to the right of the advance. if "overstrike" is set to "left" for a glyph, then any diacritics may not extend to the left of the origin and the glyph will be shifted right to ensure this, if necessary. Likewise if "overstrike" is set to "right" then the advance will be increased to ensure that a diacritic does not extend beyond the advance width. values of "true" or 1 or "both" or "all" guard both sides of the glyph.

The default action for all glyphs is not to guard. This can be changed using the -d option with 16 setting the default "left" value and 32 setting the default "right" value.

base

A base character is a reference to a glyph in the source font (via Unicode id, postscript name, glyph id) which is used in building the parent glyph. If there is more than one base glyph in a glyph, then the base glyphs are concatenated in sequence according to their advance widths, creating a single glyph. If a glyph only contains a single base glyph with no attachments, and the base glyph has not been shifted in any way, then the resulting font will include the glyph directly rather than by reference.

attach

A base glyph may have attachments, which may have their own attachments in their turn. An attachment is a reference to a glyph in the source font and also the name of an attachment point on the attachment and one on its parent which are used to position the attachment so that the attachment points coincide.

The "attach" element takes two parameters describing the attachment point on the base ("at") and the attachment point on the diacritic ("with"). If these are missing, then the glyphs are aligned centrally in the x direction and with no adjustment vertically.

advance

It is possible to override the default value of the advance width for any glyph. Thus the advance element may occur is a child of either glyph, base or attach and it sets the value of the advance width for its parent to the value given in the width attribute. The default value of the advance width for a glyph is the widest advance width taken from each of the child glyphs (including attachments) in their position within the glyph. Thus if an attachment is positioned far enough to the right, it may well cause the advance width of the glyph to increase beyond that of the base glyph the attachment is on.

rsb

This allows the advance of a glyph to be specified in terms of the right side bearing rather than an absolute advance value.

lsb

In the case where overstrike is disallowed, sets the guard space to the left of the glyph.

shift

It is also possible to shift glyphs, at least base and attach glyphs. Shifting occurs after attachment (for obvious reasons).

string

In the names section of the description file it is possible to specify strings which cause changes to the name section of the font. The string element takes a num attribute which specifies which string to change. It is also possible to specify which platform, encoding and or language id the change should be made to.

There is one special value for the num attribute, which is "name". This causes the name of the font (string id 1) and also the full font name (string id 4) to be assembled from the font name and style (string id 2).

font attributes

Various font attributes can be set from a TTFBuilder configuration file:

    ascent      amount of space needed above the baseline for a glyph
    descent     space below the baseline needed for a glyph
    linegap     how much space to put between lines
    cp          OS/2 codepages field (in binary)
    coverage    OS/2 Unicode block coverage (in binary)

Binary values are written in hex (as in: 0003)

The DTD for the configuration file is:

    <!ELEMENT font (names?, glyphs)>
    <!ATTLIST font
        ascent      CDATA #IMPLIED
        descent     CDATA #IMPLIED
        linegap     CDATA #IMPLIED
        cp          CDATA #IMPLIED
        coverage    CDATA #IMPLIED>


    <!ELEMENT names (string)+>


    <!ELEMENT string (#PCDATA)>
    <!ATTLIST string
        num     CDATA #REQUIRED
        pid     CDATA #IMPLIED
        eid     CDATA #IMPLIED
        lid     CDATA #IMPLIED>


    <!ELEMENT glyphs (glyph)+>


    <!ELEMENT glyph (property* ⎪ note? ⎪ (advance ⎪ rsb ⎪ lsb)+ ⎪ base)+>
    <!ATTLIST glyph
        PSNAme      CDATA #IMPLIED
        UID         CDATA #IMPLIED
        GID         CDATA #IMPLIED
        overstrike  (0 ⎪ false ⎪ none ⎪ 1 ⎪ true ⎪ all ⎪ both ⎪ left ⎪ right)  #IMPLIED>


    <!ELEMENT base (advance ⎪ rsb ⎪ lsb ⎪ attach ⎪ shift)*>
    <!ATTLIST base
        PSName  CDATA #IMPLIED
        UID     CDATA #IMPLIED
        GID     CDATA #IMPLIED>


    <!ELEMENT attach (advance ⎪ rsb ⎪ lsb ⎪ shift)*>
    <!ATTLIST attach
        PSName  CDATA #IMPLIED
        UID     CDATA #IMPLIED
        GID     CDATA #IMPLIED
        with    CDATA #IMPLIED
        at      CDATA #IMPLIED>


    <!ELEMENT advance EMPTY>
    <!ATTLIST advance
        width   CDATA #REQUIRED>


    <!ELEMENT rsb EMPTY>
    <!ATTLIST rsb
        width   CDATA #REQUIRED>


    <!ELEMENT lsb EMPTY>
    <!ATTLIST lsb
        width   CDATA #REQUIRED>


    <!ELEMENT shift EMPTY>
    <!ATTLIST shift
        x       CDATA #IMPLIED
        y       CDATA #IMPLIED>


    <!ELEMENT property EMPTY>
    <!ATTLIST property
        name    CDATA #REQUIRED
        value   CDATA #REQUIRED>


    <!ELEMENT note (#PCDATA)>

From this small language, quite a lot can be done.

Attachment Points

One of the most powerful mechanisms for relating glyphs is that of attachment points. This concept is concerned with attaching diacritics to a base character and the attachment is achieved by specifying an attachment point on the base character and one on the diacritic. The attachment points are usually designed as single point paths in the glyph and their location or path number are held in a separate database. When the attaching of the diacritic to the base character occurs, then the diacritic is positioned so that the two attachment points coincide.

ttfbuilder works with attachment point databases represented in XML.

The DTD for an attachment point database is:

    <!ELEMENT font (glyph)*>
    <!ATTLIST font
        name    CDATA #IMPLIED
        upem    CDATA #IMPLIED


    <!ELEMENT glyph (property* ⎪ (point ⎪ compound)* ⎪ note)>
    <!ATTLIST glyph
        PSName  CDATA #IMPLIED
        UID     CDATA #IMPLIED
        GID     CDATA #IMPLIED>


    <!ELEMENT point (location ⎪ contour)+>
    <!ATTLIST point
        type    CDATA #REQUIRED>


    <!ELEMENT location EMPTY>
    <!ATTLIST location
        x       CDATA #REQUIRED
        y       CDATA #REQUIRED>


    <!ELEMENT contour EMPTY>
    <!ATTLIST contour
        num     CDATA #REQUIRED>


    <!ELEMENT compound EMPTY>
    <!ATTLIST compound
        bbox    CDATA #REQUIRED
        PSName  CDATA #IMPLIED
        UID     CDATA #IMPLIED
        GID     CDATA #IMPLIED>


    <!ELEMENT property EMPTY>
    <!ATTLIST property
        name    CDATA #REQUIRED
        value   CDATA #REQUIRED>


    <!ELEMENT note (#PCDATA)>

A "font" contains "glyphs" which have attachment "points". Each point has a name and either a contour (path number from 0) or a location of an attachment point (real or virtual) in terms of "x" and "y" co-ordinates in em units.

A "glyph" may also be a compound glyph in which case the boxes representing the location of the components of the "compound" are listed. Each component lists a bounding box describing the location of the component in relation to the main glyph. This is a 4 element string, separated by comma and optional whitespace. Each element is a co-ordinate in em units. The sequence of values is: "xMin", "yMin", "xMax", "yMax". The compound also indicates which glyph this component refers to.

Each glyph may also contain properties. A property is a name value pair. There may only be one property with any particular name attribute.

A glyph may have textual notes associated with it.

Usage

ttfbuilder needs a configuration file, an attachment point database and a source font. It also needs to know where to store the resulting font and can take a further file to write a new attachment point database to, which represents the attachment point database for the generated font.