<HTML>
<HEAD>
<!-- Created with AOLpress/2.0 -->
<!-- AP: Created on: 16-Aug-2003 -->
<!-- AP: Last modified: 11-Nov-2009 -->
<TITLE>The dialog for editing Contextual or Chaining Contextual features</TITLE>
<LINK REL="icon" href="ffanvil16.png">
<LINK REL="stylesheet" TYPE="text/css" HREF="FontForge.css">
</HEAD>
<BODY>
<DIV id="in">
<H1>
Editing Contextual or Chaining Contextual features
</H1>
<H2>
What is a Contextual Positioning feature?
</H2>
<P>
There are two types of contextual positioning modes: Contextual Positioning
and Chained Contextual Positioning. In the simplest form of the first you
may specify a list glyphs and specify repositioning to occur if that sequence
is matched. For example you might specify (in English)<BR>
<CODE> 5 t h</CODE><BR>
and if that sequence were found raise the "t" and "h" to 5<SUP>th</SUP>.
You can also specify classes of glyphs so you could say something like:<BR>
<CODE> [0-9] t h</CODE><BR>
to raise "t" and "h" after any digit.
<P>
After applying this lookup, a word processor will skip over the three glyphs
matched by the pattern, at least it will in a simple "Contextual Positioning"
lookup. The "Chained Contextual Positioning" lookup is a bit more general
and allows you to divide the pattern into three parts: A part before the
current glyph (called backtracking), a part including the current glyph,
and a part after the current glyph (called lookahead). Positioning changes
may only be made to the part including the current glyph, and the word processor
will advance by the number of glyphs in that subset of the pattern.
<H2>
What is a Contextual Substitution feature?
</H2>
<P>
Substitutions come in three types. The first two are similar to the two types
for positioning, the third is designed to handle a very specific case of
arabic typography and is applied backwards.
<P>
Suppose you had a script font where most letters join at the base line, but
after some letters (b,o,v,w) the join is near the x-height, so a special
version of each lower case letter needs to be created designed for a left
side join near the x-height. You would want to be able to say:<BR>
<CODE> [bovw] [a-z]</CODE><BR>
Note that this just specifies the cases in which the substitution may be
applied. It does not specify the substitution itself, that is done in a separate
lookup.
<P>
You may pull down an example script font with this substitution from
<A HREF="http://fontforge.sf.net/sfds/FormalScript.tgz">FormalScript.tgz</A>,
this example is worked out in detail in the
<A HREF="editexample6-5.html#Conditional">tutorial</A>.
<H2>
More complete descriptions
</H2>
<P>
For more information on contextual features see Adobe's Docs:
<UL>
<LI>
<A HREF="http://partners.adobe.com/public/developer/opentype/index_table_formats2.html">The
GPOS table</A>, for positioning glyphs
<LI>
<A HREF="http://partners.adobe.com/public/developer/opentype/index_table_formats1.html">The
GSUB table</A>, for substituting glyphs
<LI>
<A HREF="http://partners.adobe.com/public/developer/opentype/index_tag3.html">The
feature tag registry</A>.
</UL>
<H2>
How do these relate to Apple Advanced Typography features?
</H2>
<P>
In some cases a contextual or chaining contextual substitution can be converted
into one of Apple's contextual glyph substitution subtables. See the page
on <A HREF="gposgsub.html#sometimes">Apple Advanced Typography </A>for more
information as to when and caveats about how.
<H2>
Creating a contextual feature
</H2>
<P>
Before you can edit one of these features you must create it from the
<A HREF="fontinfo.html#Contextual">Element->Font Info->Contextual
dialog</A>.
<H2>
Editing a feature
</H2>
<H3>
<IMG SRC="contextchain-format.png" WIDTH="476" HEIGHT="503" ALIGN="Right">Format
</H3>
<P>
Except for the reverse chaining features, any of these may come in one of
three formats.
<P>
The simplest format is a list of <A HREF="#Glyphs">glyphs</A>. You may specify
several glyph sequences in a single feature. So the script example above
could be specified (quite inefficiently) by 4*26 lines like:<BR>
<CODE> b a</CODE><BR>
<CODE> b b</CODE><BR>
<CODE> ...<BR>
o a</CODE><BR>
<CODE> ...</CODE>
<P>
The next format allows you to specify a list of
<A HREF="#Classes">classes</A>. In the script example we would define 2
classes:<BR>
<CODE> [bovw]</CODE><BR>
<CODE> [ac-np-ux-z]</CODE><BR>
and define two patterns as:<BR>
<CODE> class1 [class2]<BR>
class1 [class1]</CODE>
<P>
The third format is the most general and allows you to specify your pattern
by a <A HREF="#Coverage">separate list of glyphs</A> for each glyph
position:<BR>
<CODE> [bovw] [a-z]</CODE><BR>
<BR CLEAR=ALL>
<H3>
<IMG SRC="contextchain-coverage.png" WIDTH="476" HEIGHT="503" ALIGN="Right"><A
NAME="Coverage">Coverage</A>
</H3>
<P>
This display shows a list whose entries are coverage tables -- that is lists
of glyphs. A match occurs if the current glyph matches one of the glyphs
named in the current coverage table.
<P>
In the example at right there is only a single coverage table here (but there
could be more).
<P>
In a Chaining Contextual feature it is also possible to specify a list of
coverage tables to match glyphs before the current glyph, and another list
to match glyphs after.
<P>
This feature does not directly describe how transformations are to happen
to the glyphs, instead it invokes a list of nested lookups that are to be
applied to specific positions in the match string (positions in the backtrack
and lookahead lists may not be transformed).
<P>
In the example at right, the lookup 'high' will be applied to any glyph that
matches the coverage table at the top of the list. Again the buttons under
the sequence lookup list allow you to add, change, remove or reorder these
transformations. (And yes, the order the transformations are applied can
matter in complex situations).
<H4>
Creating or editing a lookup position pair
</H4>
<P>
You can change a lookup by clicking on it, this will produce a pull down
menu of all lookups that can be applied. You can change the position by editing
it.
<P>
You can add a new entry by pressing the <New> button, and then selecting
a lookup.
<H3>
Setting a coverage table
</H3>
<P>
You can change a coverage table by editing it. You may enter either glyph
names or unicode characters (which will be converted to a glyph name -- a
few characters which have special meaning: space, right parenthesis, asterix;
must have their names typed in full). If you click on the little box on the
right you will get a dialog containing a font view. In this view you may
select glyphs to your heart's content, when you press OK these glyphs become
your table.<BR CLEAR=ALL>
<H3>
<IMG SRC="contextchain-glyphlists.png" WIDTH="476" HEIGHT="498" ALIGN="Right"><A
NAME="Glyphs">Glyphs</A>
</H3>
<P>
This format allows you to specify several glyph lists to match. In the example
at right the string "A,B,C" will be matched in the glyphs before the current
glyph, then "D E" starting at the current location, and finally "F G H" after
them. If everything matched then the lookup "Smallcaps" will be applied to
location 0 (in this case "D") , and "superscript" to location 1 (here, "E").
<P>
The order is significant, word processors will stop at the first match they
find so in the following pattern strings:<BR>
<CODE> b a</CODE><BR>
<CODE> b a f</CODE><BR>
the second entry would never be matched because "b a" would be applied first.
While:<BR>
<CODE> b a f</CODE><BR>
<CODE> b a</CODE><BR>
would apply match "b a f" when an "f" was present, and "b a"
otherwise.<BR CLEAR=ALL>
<H3>
<IMG SRC="contextchain-pickglyph.png" WIDTH="476" HEIGHT="503" ALIGN="Right">Selecting
glyphs
</H3>
<P>
The text field may be manipulated as above. However here we are specifying
a string of glyphs each of which must be matched, rather than a class of
glyphs any of which could match. So in the example at right, "DE" would match,
but "E" would not.
<P>
As above a set of lookups can be applied after the match occurs.<BR CLEAR=ALL>
<H3>
<IMG SRC="contextchain-class.png" WIDTH="476" HEIGHT="503" ALIGN="Right"><A
NAME="Classes">Classes</A>
</H3>
<P>
When specifying a class match, you must specify at least one set of glyph
classes. (you may also specify separate sets of classes for the backtrack
and lookahead lists, or you use the same set for everything). You must also
specify a list of patterns where each pattern is a list of numbers and each
number is the index of the class into the class list.
<P>
Every set of classes has a magic class at index 0 which matches whatever
glyphs are not explicitly mentioned in another class. FontForge displays
this by calling it "{Everything Else}". You may not delete, edit nor reorder
this class. Other classes are just unordered lists of glyph names (rather
like coverage tables above), however any specific glyph may belong to at
most one class at a time (whereas a glyph may occur in as many coverage tables
as desired).<BR CLEAR=ALL>
<H3>
<IMG SRC="contextchain-classlist.png" WIDTH="476" HEIGHT="503" ALIGN="Right">A
list of Class numbers
</H3>
<P>
Once you have set up your classes, you may then edit the patterns you want
to match. To remind you of what your classes are, FontForge displays the
class list underneath the pattern. Clicking on a class will insert that class's
index into the pattern.
<P>
Finally you must set the nested lookups. This is exactly the same as previous
sequence / lookup settings.<BR CLEAR=ALL>
<P ALIGN=Center>
-- <A HREF="fontinfo.html">Up</A> -- <A HREF="overview.html">TOC</A> --
</DIV>
</BODY></HTML>
