Class: CharDet::EUCTWDistributionAnalysis

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

Instance Method Summary collapse

Methods inherited from CharDistributionAnalysis

#feed, #got_enough_data, #reset

Constructor Details

#initialize ⇒ EUCTWDistributionAnalysis 



101
102
103
104
105
106
 
# File 'lib/rchardet/chardistribution.rb', line 101

def initialize
  super()
  @charToFreqOrder = EUCTWCharToFreqOrder
  @tableSize = EUCTW_TABLE_SIZE
  @typicalDistributionRatio = EUCTW_TYPICAL_DISTRIBUTION_RATIO
end

Instance Method Details

#get_confidence ⇒ Object 



121
122
123
124
125
126
127
 
# File 'lib/rchardet/chardistribution.rb', line 121

def get_confidence
  if @freqChars <= MINIMUM_DATA_THRESHOLD
    return SURE_NO
  end

  super
end

#get_order(aStr) ⇒ Object 



108
109
110
111
112
113
114
115
116
117
118
119
 
# File 'lib/rchardet/chardistribution.rb', line 108

def get_order(aStr)
  # for euc-TW encoding, we are interested 
  #   first  byte range: 0xc4 -- 0xfe
  #   second byte range: 0xa1 -- 0xfe
  # no validation needed here. State machine has done that
  if aStr[0, 1] >= "\xC4"
    bytes = aStr.bytes.to_a
    return 94 * (bytes[0] - 0xC4) + bytes[1] - 0xA1
  else
    return -1
  end
end