qutuf / SourceCode /Models /Normalization /NormalizationRulesDict.py
Boulbaba's picture
Upload 210 files
21baa2f verified
'''
Created on ูกูกโ€/ู ูฃโ€/ูขู ูกู 
@Created by: Muhammad Altabba
'''
#from Lexicon.NormalizationRulesDict import SecondNormDict
'''
This dictionary stores regular expressions... for the normalization phase.
To see this file clear just copy and paste on notepad!
'''
FirstNormDict = dict()
#FirstNormDict[char()] = char()+char();
FirstNormDict[chr(1600)] = '' #hyphen
FirstNormDict[chr(1570)] = 'ุกุง' #alef-mada
'''
This dictionary stores regular expressions... for the normalization phase.
To see this file clear just copy and paste on notepad!
'''
SecondNormDict = dict()
SecondNormDict[chr(1611)] = '' # fathatan ู‹ ู‹ ู‹ ู‹ ู‹ (SHIFT+W)
SecondNormDict[chr(1612)] = '' #damatan ูŒ ูŒ ูŒ ูŒ ูŒ (SHIFT+R)
SecondNormDict[chr(1613)] = '' #kasratan ู ู ู ู ู (SHIFT+S)
SecondNormDict[chr(1614)] = '' #fathaูŽูŽูŽูŽูŽ (SHIFT+Q)
SecondNormDict[chr(1615)] = '' #dama ู ู ู ู ู (SHIFT+E)
SecondNormDict[chr(1616)] = '' #kasra ู ู ู ู ู (SHIFT+A)
SecondNormDict[chr(1617)] = '' #shada ู‘ ู‘ ู‘ ู‘ ู‘ (SHIFT+`)
SecondNormDict[chr(1618)] = '' #sokon ู’ ู’ ู’ ู’ ู’ (SHIFT+X)
#SecondNormDict['ุฆ'] = 'ุก'
#SecondNormDict['ุค'] = 'ุก'
#SecondNormDict['ุฃ'] = 'ุก'
#SecondNormDict['ุฅ'] = 'ุก'