#

[0] Fix | Delete

# kconv.rb - Kanji Converter.

[1] Fix | Delete

#

[2] Fix | Delete

# $Id: kconv.rb 30112 2010-12-07 11:47:39Z naruse $

[3] Fix | Delete

#

[4] Fix | Delete

# ----

[5] Fix | Delete

#

[6] Fix | Delete

# kconv.rb implements the Kconv class for Kanji Converter. Additionally,

[7] Fix | Delete

# some methods in String classes are added to allow easy conversion.

[8] Fix | Delete

#

[9] Fix | Delete

[10] Fix | Delete

require 'nkf'

[11] Fix | Delete

[12] Fix | Delete

#

[13] Fix | Delete

# Kanji Converter for Ruby.

[14] Fix | Delete

#

[15] Fix | Delete

module Kconv

[16] Fix | Delete

#

[17] Fix | Delete

# Public Constants

[18] Fix | Delete

#

[19] Fix | Delete

[20] Fix | Delete

#Constant of Encoding

[21] Fix | Delete

[22] Fix | Delete

# Auto-Detect

[23] Fix | Delete

AUTO = NKF::AUTO

[24] Fix | Delete

# ISO-2022-JP

[25] Fix | Delete

JIS = NKF::JIS

[26] Fix | Delete

# EUC-JP

[27] Fix | Delete

EUC = NKF::EUC

[28] Fix | Delete

# Shift_JIS

[29] Fix | Delete

SJIS = NKF::SJIS

[30] Fix | Delete

# BINARY

[31] Fix | Delete

BINARY = NKF::BINARY

[32] Fix | Delete

# NOCONV

[33] Fix | Delete

NOCONV = NKF::NOCONV

[34] Fix | Delete

# ASCII

[35] Fix | Delete

ASCII = NKF::ASCII

[36] Fix | Delete

# UTF-8

[37] Fix | Delete

UTF8 = NKF::UTF8

[38] Fix | Delete

# UTF-16

[39] Fix | Delete

UTF16 = NKF::UTF16

[40] Fix | Delete

# UTF-32

[41] Fix | Delete

UTF32 = NKF::UTF32

[42] Fix | Delete

# UNKNOWN

[43] Fix | Delete

UNKNOWN = NKF::UNKNOWN

[44] Fix | Delete

[45] Fix | Delete

#

[46] Fix | Delete

# Public Methods

[47] Fix | Delete

#

[48] Fix | Delete

[49] Fix | Delete

# call-seq:

[50] Fix | Delete

# Kconv.kconv(str, to_enc, from_enc=nil)

[51] Fix | Delete

#

[52] Fix | Delete

# Convert <code>str</code> to <code>to_enc</code>.

[53] Fix | Delete

# <code>to_enc</code> and <code>from_enc</code> are given as constants of Kconv or Encoding objects.

[54] Fix | Delete

def kconv(str, to_enc, from_enc=nil)

[55] Fix | Delete

opt = ''

[56] Fix | Delete

opt += ' --ic=' + from_enc.to_s if from_enc

[57] Fix | Delete

opt += ' --oc=' + to_enc.to_s if to_enc

[58] Fix | Delete

[59] Fix | Delete

::NKF::nkf(opt, str)

[60] Fix | Delete

end

[61] Fix | Delete

module_function :kconv

[62] Fix | Delete

[63] Fix | Delete

#

[64] Fix | Delete

# Encode to

[65] Fix | Delete

#

[66] Fix | Delete

[67] Fix | Delete

# call-seq:

[68] Fix | Delete

# Kconv.tojis(str) => string

[69] Fix | Delete

#

[70] Fix | Delete

# Convert <code>str</code> to ISO-2022-JP

[71] Fix | Delete

def tojis(str)

[72] Fix | Delete

kconv(str, JIS)

[73] Fix | Delete

end

[74] Fix | Delete

module_function :tojis

[75] Fix | Delete

[76] Fix | Delete

# call-seq:

[77] Fix | Delete

# Kconv.toeuc(str) => string

[78] Fix | Delete

#

[79] Fix | Delete

# Convert <code>str</code> to EUC-JP

[80] Fix | Delete

def toeuc(str)

[81] Fix | Delete

kconv(str, EUC)

[82] Fix | Delete

end

[83] Fix | Delete

module_function :toeuc

[84] Fix | Delete

[85] Fix | Delete

# call-seq:

[86] Fix | Delete

# Kconv.tosjis(str) => string

[87] Fix | Delete

#

[88] Fix | Delete

# Convert <code>str</code> to Shift_JIS

[89] Fix | Delete

def tosjis(str)

[90] Fix | Delete

kconv(str, SJIS)

[91] Fix | Delete

end

[92] Fix | Delete

module_function :tosjis

[93] Fix | Delete

[94] Fix | Delete

# call-seq:

[95] Fix | Delete

# Kconv.toutf8(str) => string

[96] Fix | Delete

#

[97] Fix | Delete

# Convert <code>str</code> to UTF-8

[98] Fix | Delete

def toutf8(str)

[99] Fix | Delete

kconv(str, UTF8)

[100] Fix | Delete

end

[101] Fix | Delete

module_function :toutf8

[102] Fix | Delete

[103] Fix | Delete

# call-seq:

[104] Fix | Delete

# Kconv.toutf16(str) => string

[105] Fix | Delete

#

[106] Fix | Delete

# Convert <code>str</code> to UTF-16

[107] Fix | Delete

def toutf16(str)

[108] Fix | Delete

kconv(str, UTF16)

[109] Fix | Delete

end

[110] Fix | Delete

module_function :toutf16

[111] Fix | Delete

[112] Fix | Delete

# call-seq:

[113] Fix | Delete

# Kconv.toutf32(str) => string

[114] Fix | Delete

#

[115] Fix | Delete

# Convert <code>str</code> to UTF-32

[116] Fix | Delete

def toutf32(str)

[117] Fix | Delete

kconv(str, UTF32)

[118] Fix | Delete

end

[119] Fix | Delete

module_function :toutf32

[120] Fix | Delete

[121] Fix | Delete

# call-seq:

[122] Fix | Delete

# Kconv.tolocale => string

[123] Fix | Delete

#

[124] Fix | Delete

# Convert <code>self</code> to locale encoding

[125] Fix | Delete

def tolocale(str)

[126] Fix | Delete

kconv(str, Encoding.locale_charmap)

[127] Fix | Delete

end

[128] Fix | Delete

module_function :tolocale

[129] Fix | Delete

[130] Fix | Delete

#

[131] Fix | Delete

# guess

[132] Fix | Delete

#

[133] Fix | Delete

[134] Fix | Delete

# call-seq:

[135] Fix | Delete

# Kconv.guess(str) => encoding

[136] Fix | Delete

#

[137] Fix | Delete

# Guess input encoding by NKF.guess

[138] Fix | Delete

def guess(str)

[139] Fix | Delete

::NKF::guess(str)

[140] Fix | Delete

end

[141] Fix | Delete

module_function :guess

[142] Fix | Delete

[143] Fix | Delete

#

[144] Fix | Delete

# isEncoding

[145] Fix | Delete

#

[146] Fix | Delete

[147] Fix | Delete

# call-seq:

[148] Fix | Delete

# Kconv.iseuc(str) => true or false

[149] Fix | Delete

#

[150] Fix | Delete

# Returns whether input encoding is EUC-JP or not.

[151] Fix | Delete

#

[152] Fix | Delete

# *Note* don't expect this return value is MatchData.

[153] Fix | Delete

def iseuc(str)

[154] Fix | Delete

str.dup.force_encoding(EUC).valid_encoding?

[155] Fix | Delete

end

[156] Fix | Delete

module_function :iseuc

[157] Fix | Delete

[158] Fix | Delete

# call-seq:

[159] Fix | Delete

# Kconv.issjis(str) => true or false

[160] Fix | Delete

#

[161] Fix | Delete

# Returns whether input encoding is Shift_JIS or not.

[162] Fix | Delete

def issjis(str)

[163] Fix | Delete

str.dup.force_encoding(SJIS).valid_encoding?

[164] Fix | Delete

end

[165] Fix | Delete

module_function :issjis

[166] Fix | Delete

[167] Fix | Delete

# call-seq:

[168] Fix | Delete

# Kconv.isjis(str) => true or false

[169] Fix | Delete

#

[170] Fix | Delete

# Returns whether input encoding is ISO-2022-JP or not.

[171] Fix | Delete

def isjis(str)

[172] Fix | Delete

/\A [\t\n\r\x20-\x7E]*

[173] Fix | Delete

(?:

[174] Fix | Delete

(?:\x1b \x28 I [\x21-\x7E]*

[175] Fix | Delete

|\x1b \x28 J [\x21-\x7E]*

[176] Fix | Delete

|\x1b \x24 @ (?:[\x21-\x7E]{2})*

[177] Fix | Delete

|\x1b \x24 B (?:[\x21-\x7E]{2})*

[178] Fix | Delete

|\x1b \x24 \x28 D (?:[\x21-\x7E]{2})*

[179] Fix | Delete

)*

[180] Fix | Delete

\x1b \x28 B [\t\n\r\x20-\x7E]*

[181] Fix | Delete

)*

[182] Fix | Delete

\z/nox =~ str.dup.force_encoding('BINARY') ? true : false

[183] Fix | Delete

end

[184] Fix | Delete

module_function :isjis

[185] Fix | Delete

[186] Fix | Delete

# call-seq:

[187] Fix | Delete

# Kconv.isutf8(str) => true or false

[188] Fix | Delete

#

[189] Fix | Delete

# Returns whether input encoding is UTF-8 or not.

[190] Fix | Delete

def isutf8(str)

[191] Fix | Delete

str.dup.force_encoding(UTF8).valid_encoding?

[192] Fix | Delete

end

[193] Fix | Delete

module_function :isutf8

[194] Fix | Delete

end

[195] Fix | Delete

[196] Fix | Delete

class String

[197] Fix | Delete

# call-seq:

[198] Fix | Delete

# String#kconv(to_enc, from_enc)

[199] Fix | Delete

#

[200] Fix | Delete

# Convert <code>self</code> to <code>to_enc</code>.

[201] Fix | Delete

# <code>to_enc</code> and <code>from_enc</code> are given as constants of Kconv or Encoding objects.

[202] Fix | Delete

def kconv(to_enc, from_enc=nil)

[203] Fix | Delete

from_enc = self.encoding if !from_enc && self.encoding != Encoding.list[0]

[204] Fix | Delete

Kconv::kconv(self, to_enc, from_enc)

[205] Fix | Delete

end

[206] Fix | Delete

[207] Fix | Delete

#

[208] Fix | Delete

# to Encoding

[209] Fix | Delete

#

[210] Fix | Delete

[211] Fix | Delete

# call-seq:

[212] Fix | Delete

# String#tojis => string

[213] Fix | Delete

#

[214] Fix | Delete

# Convert <code>self</code> to ISO-2022-JP

[215] Fix | Delete

def tojis; Kconv.tojis(self) end

[216] Fix | Delete

[217] Fix | Delete

# call-seq:

[218] Fix | Delete

# String#toeuc => string

[219] Fix | Delete

#

[220] Fix | Delete

# Convert <code>self</code> to EUC-JP

[221] Fix | Delete

def toeuc; Kconv.toeuc(self) end

[222] Fix | Delete

[223] Fix | Delete

# call-seq:

[224] Fix | Delete

# String#tosjis => string

[225] Fix | Delete

#

[226] Fix | Delete

# Convert <code>self</code> to Shift_JIS

[227] Fix | Delete

def tosjis; Kconv.tosjis(self) end

[228] Fix | Delete

[229] Fix | Delete

# call-seq:

[230] Fix | Delete

# String#toutf8 => string

[231] Fix | Delete

#

[232] Fix | Delete

# Convert <code>self</code> to UTF-8

[233] Fix | Delete

def toutf8; Kconv.toutf8(self) end

[234] Fix | Delete

[235] Fix | Delete

# call-seq:

[236] Fix | Delete

# String#toutf16 => string

[237] Fix | Delete

#

[238] Fix | Delete

# Convert <code>self</code> to UTF-16

[239] Fix | Delete

def toutf16; Kconv.toutf16(self) end

[240] Fix | Delete

[241] Fix | Delete

# call-seq:

[242] Fix | Delete

# String#toutf32 => string

[243] Fix | Delete

#

[244] Fix | Delete

# Convert <code>self</code> to UTF-32

[245] Fix | Delete

def toutf32; Kconv.toutf32(self) end

[246] Fix | Delete

[247] Fix | Delete

# call-seq:

[248] Fix | Delete

# String#tolocale => string

[249] Fix | Delete

#

[250] Fix | Delete

# Convert <code>self</code> to locale encoding

[251] Fix | Delete

def tolocale; Kconv.tolocale(self) end

[252] Fix | Delete

[253] Fix | Delete

#

[254] Fix | Delete

# is Encoding

[255] Fix | Delete

#

[256] Fix | Delete

[257] Fix | Delete

# call-seq:

[258] Fix | Delete

# String#iseuc => true or false

[259] Fix | Delete

#

[260] Fix | Delete

# Returns whether <code>self</code>'s encoding is EUC-JP or not.

[261] Fix | Delete

def iseuc; Kconv.iseuc(self) end

[262] Fix | Delete

[263] Fix | Delete

# call-seq:

[264] Fix | Delete

# String#issjis => true or false

[265] Fix | Delete

#

[266] Fix | Delete

# Returns whether <code>self</code>'s encoding is Shift_JIS or not.

[267] Fix | Delete

def issjis; Kconv.issjis(self) end

[268] Fix | Delete

[269] Fix | Delete

# call-seq:

[270] Fix | Delete

# String#isjis => true or false

[271] Fix | Delete

#

[272] Fix | Delete

# Returns whether <code>self</code>'s encoding is ISO-2022-JP or not.

[273] Fix | Delete

def isjis; Kconv.isjis(self) end

[274] Fix | Delete

[275] Fix | Delete

# call-seq:

[276] Fix | Delete

# String#isutf8 => true or false

[277] Fix | Delete

#

[278] Fix | Delete

# Returns whether <code>self</code>'s encoding is UTF-8 or not.

[279] Fix | Delete

def isutf8; Kconv.isutf8(self) end

[280] Fix | Delete

end

[281] Fix | Delete

[282] Fix | Delete