Class: UniversalDetector::SJISDistributionAnalysis

Inherits:
CharDistributionAnalysis show all
Defined in:
lib/CharDistributionAnalysis.rb

Constant Summary

Constants inherited from CharDistributionAnalysis

CharDistributionAnalysis::ENOUGH_DATA_THRESHOLD, CharDistributionAnalysis::SURE_NO, CharDistributionAnalysis::SURE_YES

Instance Method Summary collapse

Methods inherited from CharDistributionAnalysis

#feed, #get_confidence, #got_enough_data, #reset

Constructor Details

#initialize ⇒ SJISDistributionAnalysis

Returns a new instance of SJISDistributionAnalysis.



197
198
199
200
201
202
 
# File 'lib/CharDistributionAnalysis.rb', line 197

def initialize
    super
    @_mCharToFreqOrder = JISCharToFreqOrder
    @_mTableSize = JIS_TABLE_SIZE
    @_mTypicalDistributionRatio = JIS_TYPICAL_DISTRIBUTION_RATIO
end

Instance Method Details

#get_order(aStr) ⇒ Object 



204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
 
# File 'lib/CharDistributionAnalysis.rb', line 204

def get_order(aStr)
    # for sjis encoding, we are interested 
    #   first  byte range: 0x81 -- 0x9f , 0xe0 -- 0xfe
    #   second byte range: 0x40 -- 0x7e,  0x81 -- oxfe
    # no validation needed here. State machine has done that
    if (aStr[0] >= 0x81) && (aStr[0] <= 0x9F)
        order = 188 * (aStr[0] - 0x81)
    elsif (aStr[0] >= 0xE0) and (aStr[0] <= 0xEF)
        order = 188 * (aStr[0] - 0xE0 + 31)
    else
        return -1;
    end
    order = order + aStr[1] - 0x40
    if aStr[1] > 0x7F
        order =- 1
    end
    return order
end