Class: CharDet::EUCJPDistributionAnalysis

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

Instance Method Summary collapse

Methods inherited from CharDistributionAnalysis

#feed, #get_confidence, #got_enough_data, #reset

Constructor Details

#initialize ⇒ EUCJPDistributionAnalysis

Returns a new instance of EUCJPDistributionAnalysis.



230
231
232
233
234
235
 
# File 'lib/rchardet/chardistribution.rb', line 230

def initialize
  super()
  @charToFreqOrder = JISCharToFreqOrder
  @tableSize = JIS_TABLE_SIZE
  @typicalDistributionRatio = JIS_TYPICAL_DISTRIBUTION_RATIO
end

Instance Method Details

#get_order(aStr) ⇒ Object 



237
238
239
240
241
242
243
244
245
246
247
248
 
# File 'lib/rchardet/chardistribution.rb', line 237

def get_order(aStr)
  # for euc-JP encoding, we are interested 
  #   first  byte range: 0xa0 -- 0xfe
  #   second byte range: 0xa1 -- 0xfe
  # no validation needed here. State machine has done that
  if aStr[0, 1] >= "\xA0"
    bytes = aStr.bytes.to_a
    return 94 * (bytes[0] - 0xA1) + bytes[1] - 0xa1
  else
    return -1
  end
end