from FSA import *

# stems
M_WA = compileRE('(tu)|(toto)|(geni)')
M_MI = compileRE('(ti)|(ji)|(lima)')
KI_VI = compileRE('(tu)|(su)|(tanda)')

# these need no prefix
N_WORD = compileRE('(simu)|(meza)|(nchi)')

# prefixes
M = compileRE('m')
WA = compileRE('wa')
MI = compileRE('mi')
KI = compileRE('ki')
VI = compileRE('vi')

# suffix: singular only (m-, ki-, N class)
NI = compileRE('(ni)?')

# singular words in each class
M2 = minimize(union(M_WA, M_MI))
M_WORD = concatenation(M, M2)
KI_WORD = concatenation(KI, KI_VI)

# plural words in each class
WA_WORD = concatenation(WA, M_WA)
MI_WORD = concatenation(MI, M_MI)
VI_WORD = concatenation(VI, KI_VI)

# singular and plural words
SINGULAR1 = minimize(union(M_WORD, KI_WORD, N_WORD))
SINGULAR = determinize(concatenation(SINGULAR1, NI))
PLURAL = minimize(union(WA_WORD, MI_WORD, VI_WORD))

# union of singular and plural words
swahili_noun = minimize(union(SINGULAR, PLURAL))