keming_machine.txt

--- Pixel 0
- Lowheight bit
- encoding: has pixel - it's low height
- used by the diacritics system to quickly look up if the character is low height without parsing the Pixel 1

### Legends
#
#  A·B < unset for lowheight miniscules, as in e
#  |·| < space we don't care
#  C·D < middle hole for majuscules, as in C
#  E·F < middle hole for miniscules, as in c
#  G·H
#  ――― < baseline
#  |·|
#  J·K

--- Pixel 1
- A..K Occupied (1024)
- Is ABGH are all Ys instead of Bars? (2)
    - Say, A is Bar but E is wye (e.g. Ꮨ), this condition is false; this character must be encoded as ABDFGH(B).
- encoding:
    - <MSB> Y0000000 JK000000 ABCDEFGH <LSB>
    - Y: Bar/Wye Mode
    - A..K: arguments
- B-type will contract the space by 2 pixels, while Y-type will do it by 1

# Capital/lower itself is given using the pixel 0 due to the diacritics processing

--- Examples
- AB(B): T
- ABCEGH(B): C
- ABCEFGH(Y): K
- ABCDEG: Ꮅ
- ABCDEFGH: B,D,O
- ABCDFH: Ч
- ABCEG: Г
- ABGH: Ꮖ
- ACDEG: Ꮀ
- ACDEFGH: h,Ƅ
- ACDFH: ߆
- ACEGH: L
- AH(Y): \
- BDEFGH: J
- BDFGH: ɺ,ป
- BG(Y): /
- CD: Ⴕ
- CDEF(Y): Φ
- CDEFGH: a,c,e,i,o,φ,ϕ
- CDEFGHJK: g
- CDEFGHK: ƞ

- AB(Y): Y
- ABCD(Y): V
- CDEF(Y): v
- EFGH(Y): ʌ
- CDGH(Y): A

--- Rules
# Legend: _ dont care
#         @ must have a bit set
#         ` must have a bit unset
- ͟A͟B͟C͟D͟E͟F͟G͟H͟J͟K͟ ͟ ͟ ͟A͟B͟C͟D͟E͟F͟G͟H͟J͟K͟
- _@_`___`__ — `_________ # Γe,TJ ; Ye,YJ,Ve,VJ,TA,ΓA,VA,Vʌ,YA,Yʌ,yA,yʌ,/a,/d
- _@_@___`__ — `___`_@___ # Pɺ but NOT Po,PJ
- _@_@___`__ — `___@_____ # Fo,PJ (always 1 px)
- ___`_`____ — `___@_`___ # Cꟶ,Kꟶ,Lꟶ,Γꟶ
- ___`_`____ — `_@___`___ # CꟵ,KꟵ,LꟵ,ΓꟵ
-----------------------------------------------------
- _`________ — @_`___`___ # eꞀ,LT ; eY,LY,eV,LV,AT,AꞀ,AY,Ay,λY,λy,a\,b\
- _`___`_@__ — @_@___`___ # Lꟼ but NOT oꟼ,bꟼ
- _`___@____ — @_@___`___ # oꟼ,bꟼ (always 1 px)
- _`___@_`__ — __`_`_____ # ⱶƆ,ⱶJ
- _`_@___`__ — __`_`_____ # ⱵƆ,ⱵJ


--- Implementation
code: |
    val posTable = intArrayOf(7,6,5,4,3,2,1,0,9,8)

    class RuleMask(s: String) {

        private var careBits = 0
        private var ruleBits = 0

        init {
            s.forEachIndexed { index, char ->
                when (char) {
                    '@' -> {
                        careBits = careBits or (1 shl posTable[index])
                        ruleBits = ruleBits or (1 shl posTable[index])
                    }
                    '`' -> {
                        careBits = careBits or (1 shl posTable[index])
                    }
                }
            }
        }

        fun matches(shapeBits: Int) = ((shapeBits and careBits) and ruleBits) == 0

    }

--- Pixel 2
dot removal for diacritics:
- All 24 bits are used to put replacement character
- encoding:
    - <MSB> RRRRRRRR GGGGGGGG BBBBBBBB <LSB>