Class: CharDet::JapaneseContextAnalysis

Inherits:
Object
  • Object
show all
Defined in:
lib/rchardet/jpcntx.rb

Direct Known Subclasses

EUCJPContextAnalysis, SJISContextAnalysis

Instance Method Summary collapse

Constructor Details

#initialize ⇒ JapaneseContextAnalysis

Returns a new instance of JapaneseContextAnalysis.



123
124
125
 
# File 'lib/rchardet/jpcntx.rb', line 123

def initialize
  reset()
end

Instance Method Details

#feed(aBuf, aLen) ⇒ Object 



135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
 
# File 'lib/rchardet/jpcntx.rb', line 135

def feed(aBuf, aLen)
  return if @done

  # The buffer we got is byte oriented, and a character may span in more than one
  # buffers. In case the last one or two byte in last buffer is not complete, we
  # record how many byte needed to complete that character and skip these bytes here.
  # We can choose to record those bytes as well and analyse the character once it
  # is complete, but since a character will not make much difference, by simply skipping
  # this character will simply our logic and improve performance.
  i = @needToSkipCharNum
  while i < aLen
    order, charLen = get_order(aBuf[i, 2])
    i += charLen
    if i > aLen
      @needToSkipCharNum = i - aLen
      @lastCharOrder = -1
    else
      if (order != -1) and (@lastCharOrder != -1)
        @totalRel += 1
        if @totalRel > MAX_REL_THRESHOLD
          @done = true
          break
        end
        @relSample[JP2_CHAR_CONTEXT[@lastCharOrder][order]] += 1
      end
      @lastCharOrder = order
    end
  end
end

#get_confidence ⇒ Object 



169
170
171
172
173
174
175
176
 
# File 'lib/rchardet/jpcntx.rb', line 169

def get_confidence
  # This is just one way to calculate confidence. It works well for me.
  if @totalRel > MINIMUM_DATA_THRESHOLD
    return (@totalRel - @relSample[0]) / @totalRel
  else
    return DONT_KNOW
  end
end

#get_order(aStr) ⇒ Object 



178
179
180
 
# File 'lib/rchardet/jpcntx.rb', line 178

def get_order(aStr)
  return -1, 1
end

#got_enough_data ⇒ Object 



165
166
167
 
# File 'lib/rchardet/jpcntx.rb', line 165

def got_enough_data
  return @totalRel > ENOUGH_REL_THRESHOLD
end

#reset ⇒ Object 



127
128
129
130
131
132
133
 
# File 'lib/rchardet/jpcntx.rb', line 127

def reset
  @totalRel = 0 # total sequence received
  @relSample = [0] * NUM_OF_CATEGORY # category counters, each interger counts sequence in its category
  @needToSkipCharNum = 0 # if last byte in current buffer is not the last byte of a character, we need to know how many bytes to skip in next buffer
  @lastCharOrder = -1 # The order of previous char
  @done = false # If this flag is set to constants.True, detection is done and conclusion has been made
end