The Keyboard Module



In [22]:

    
import pandas
from keymonkey import *

The keyboard Module knows about the keyboard model and layout

The Data Structure for Keyboards and Layouts

The KeySwitch structure holds data about each physical key on the keyboard. A collection of KeySwitches is a Keyboard. The PC_keybord holds a typical PC or laptop keyboard. Easy we can define other spacial keyboards for further study. A random sample of the netrics of a typical keyboard is showed bellow:



In [2]:

    
list(map(lambda _: random.choice([v for k,v in PC_keyboard.items()]), range(5)))









    Out[2]:





[KeySwitch(keycode=42, finger=3, hand='L', line=2, effort=0.0, special=0),
 KeySwitch(keycode=29, finger=6, hand='R', line=1, effort=3.0, special=0),
 KeySwitch(keycode=27, finger=3, hand='L', line=1, effort=2.0, special=0),
 KeySwitch(keycode=59, finger=7, hand='R', line=3, effort=2.0, special=0),
 KeySwitch(keycode=31, finger=7, hand='R', line=1, effort=2.0, special=0)]

Some well known layouts is defined first. The KeyAction data structure holds the lower, upper, altenativer lower and alternative upper keys for a layout assignment. Thus is possible to optise 2 layouts at the same time, like Greek and English.

A random choise of assignments of KeyActions to KeyCaps is given bellow:



In [3]:

    
list(map(lambda _: random.choice([(k,v) for k,v in Querty.items()]), range(5)))









    Out[3]:





[(KeySwitch(keycode=58, finger=6, hand='R', line=3, effort=2.0, special=0),
  KeyAction(a='m', sa='M', b='Greek_mu', sb='Greek_MU')),
 (KeySwitch(keycode=59, finger=7, hand='R', line=3, effort=2.0, special=0),
  KeyAction(a='comma', sa='less', b='comma', sb='less')),
 (KeySwitch(keycode=24, finger=0, hand='L', line=1, effort=2.0, special=0),
  KeyAction(a='q', sa='Q', b='semicolon', sb='colon')),
 (KeySwitch(keycode=42, finger=3, hand='L', line=2, effort=0.0, special=0),
  KeyAction(a='g', sa='G', b='Greek_gamma', sb='Greek_GAMMA')),
 (KeySwitch(keycode=28, finger=3, hand='L', line=1, effort=2.5, special=0),
  KeyAction(a='t', sa='T', b='Greek_tau', sb='Greek_TAU'))]

This correspond to this cost assignments



In [4]:

    
display(Image("../docs/keyboard-cost.png"))

Graphical display of a keyboard

Using the X11 system we can get with the help of a bash script a graphical iamge of a given layout. We can also try it with a simple command.



In [5]:

    
showkbd(Querty,label="The Querty Layout")

We can mutate keyboards. Here we exchange keycode 32, with keycode 58

Mutations and Crossovers



In [6]:

    
mutA=Querty.copy()
kbd_swap_keys(mutA,32,58)
showkbd(mutA,label="Change key 'O' with key 'M'")

Or ask to do randomly a number of mutations :



In [7]:

    
mutB=Querty.copy()
kbd_mutate(mutB,mutations=6)



In [8]:

    
showkbd(mutB,label="A Querty with 6 mutations")

We can get the distance and the key diffs, from a specific keyboard easy



In [9]:

    
kbd_diffs(mutB,Querty)









    Out[9]:





[40, 41, 42, 43, 51, 54, 56, 25, 59, 30, 31]



In [10]:

    
kbd_distance(mutB,Querty)









    Out[10]:





11



In [11]:

    
from keyboard import *
genB=kbd_crossover(mutB,Querty,percent=0.5,debug=True)









    



A_54:w -> B_25:c
A_56:c -> B_54:b
A_59:h -> B_43:comma



In [12]:

    
showkbd(genB,label="A 50% crossover with Querty")



In [13]:

    
kbd_distance(genB,mutB)









    Out[13]:





5



In [14]:

    
showkbd_side(genB,mutB)

Initial Generation creation



In [15]:

    
populationSize   = 50
initialMutations = 31
population=generate_population_from(Querty,populationSize,initialMutations)



In [16]:

    
showkbd_side(population[0],population[1])
showkbd_side(population[2],population[3])
showkbd_side(population[4],population[5])

Calculate Population Distances



In [19]:

    
pm= population_matrix(population)
#pandas.DataFrame(pm)



In [18]:

    
mean=numpy.mean(pm)
print("Distance average: {}".format(mean))









    



Distance average: 10.1888

Read a corpus Text



In [23]:

    
strokes = read_corpus("../data/longos_small.txt")



In [ ]: