
unistrok file format

#comment
unicode_index some_unused_number "|" stroke_desc [ "|" filter ] 

the unicode index must be in hex and <=0x10FFFF
nobody cares about the second number. Its a hal-code, so just
put a zero there, its not used by kanjipad.

then there is a pipe, "|", which is used a separator.

next is the stroke data:

each stroke is separated by a space. each stroke is composed of a set
of 1 or more directions. (Think of the numeric keypad for direction mnemonics)

DIRECTIONS:
1	diagonal down left
2	down
3	diagonal down right
4	left
6	right
7	diagonal up left
8	up
9	diagonal up right
b	shorthand for "62" 
c	shorthand for "26"
x	shorthand for "21"
y	shorthand for "23"

last is the filter. A filter is optional, if it exists its after another pipe. A filter is a set
of space separated rules. Each rule is in the format:

{Code}{Stroke Id} "-" {Code}{Stroke ID} ["!"]

what it means is to favor a strokeset when the value before the "-" is greater than the value after it.
the "!" at the end is optional: it means that the rule is mandatory, and much more important than
non-mandatory rules. (read "!" as "must!"). Stroke Id's are 1-based offsets into the set of
strokes. (x vals increase to the right, y vals increase downwards )

Stroke codes are:

FILTER STROKE CODES:
x	first x in the stroke
y	first y in the stroke
i	last x in the stroke
j	last y in the stroke
a	average x of the stroke
b	average y of the stroke
l	stroke length
!	MUST




