Navigation

← Previous Changeset
Next Changeset →

Changeset 1594

Timestamp:

07/24/05 22:46:33 (3 years ago)

Author:

nahi

Message:

varname change: @encoding => @internal_encoding

Files:

trunk/lib/xsd/charset.rb (modified) (9 diffs)

Legend:

: Unmodified
: Added
: Removed
: Modified
: Copied
: Moved

trunk/lib/xsd/charset.rb

r1552	r1594
1	1	# XSD4R - Charset handling library.
2		# Copyright (C) 2001, 2003 NAKAMURA, Hiroshi <nahi@ruby-lang.org>.
	2	# Copyright (C) 2001, 2003, 2005 NAKAMURA, Hiroshi <nahi@ruby-lang.org>.
3	3
4	4	# This program is copyrighted free software by NAKAMURA, Hiroshi. You can
…	…
11	11
12	12	module Charset
13		@encoding = $KCODE
	13	@internal_encoding = $KCODE
14	14
15	15	class XSDError < StandardError; end
…	…
27	27	begin
28	28	require 'xsd/iconvcharset'
29		@encoding = 'UTF8'
30		sjtag = (/(mswin\|bccwin\|mingw\|cygwin\|emx)/ =~ RUBY_PLATFORM) ? 'cp932' : 'shift_jis'
31		EncodingConvertMap[['UTF8', 'EUC' ]] = Proc.new { \|str\| IconvCharset.safe_iconv("euc-jp", "utf-8", str) }
32		EncodingConvertMap[['EUC' , 'UTF8']] = Proc.new { \|str\| IconvCharset.safe_iconv("utf-8", "euc-jp", str) }
33		EncodingConvertMap[['EUC' , 'SJIS']] = Proc.new { \|str\| IconvCharset.safe_iconv(sjtag, "euc-jp", str) }
34		EncodingConvertMap[['UTF8', 'SJIS']] = Proc.new { \|str\| IconvCharset.safe_iconv(sjtag, "utf-8", str) }
35		EncodingConvertMap[['SJIS', 'UTF8']] = Proc.new { \|str\| IconvCharset.safe_iconv("utf-8", sjtag, str) }
36		EncodingConvertMap[['SJIS', 'EUC' ]] = Proc.new { \|str\| IconvCharset.safe_iconv("euc-jp", sjtag, str) }
	29	@internal_encoding = 'UTF8'
	30	sjtag = (/(mswin\|bccwin\|mingw\|cygwin\|emx)/ =~ RUBY_PLATFORM) ? 'cp932' :
	31	'shift_jis'
	32	EncodingConvertMap[['UTF8', 'EUC' ]] =
	33	Proc.new { \|str\| IconvCharset.safe_iconv("euc-jp", "utf-8", str) }
	34	EncodingConvertMap[['EUC' , 'UTF8']] =
	35	Proc.new { \|str\| IconvCharset.safe_iconv("utf-8", "euc-jp", str) }
	36	EncodingConvertMap[['EUC' , 'SJIS']] =
	37	Proc.new { \|str\| IconvCharset.safe_iconv(sjtag, "euc-jp", str) }
	38	EncodingConvertMap[['UTF8', 'SJIS']] =
	39	Proc.new { \|str\| IconvCharset.safe_iconv(sjtag, "utf-8", str) }
	40	EncodingConvertMap[['SJIS', 'UTF8']] =
	41	Proc.new { \|str\| IconvCharset.safe_iconv("utf-8", sjtag, str) }
	42	EncodingConvertMap[['SJIS', 'EUC' ]] =
	43	Proc.new { \|str\| IconvCharset.safe_iconv("euc-jp", sjtag, str) }
37	44	rescue LoadError
38	45	begin
39	46	require 'nkf'
40		EncodingConvertMap[['EUC' , 'SJIS']] = Proc.new { \|str\| NKF.nkf('-sXm0', str) }
41		EncodingConvertMap[['SJIS', 'EUC' ]] = Proc.new { \|str\| NKF.nkf('-eXm0', str) }
	47	EncodingConvertMap[['EUC' , 'SJIS']] =
	48	Proc.new { \|str\| NKF.nkf('-sXm0', str) }
	49	EncodingConvertMap[['SJIS', 'EUC' ]] =
	50	Proc.new { \|str\| NKF.nkf('-eXm0', str) }
42	51	rescue LoadError
43	52	end
…	…
45	54	begin
46	55	require 'uconv'
47		@encoding = 'UTF8'
	56	@internal_encoding = 'UTF8'
48	57	EncodingConvertMap[['UTF8', 'EUC' ]] = Uconv.method(:u8toeuc)
49	58	EncodingConvertMap[['UTF8', 'SJIS']] = Uconv.method(:u8tosjis)
…	…
68	77	#
69	78	def Charset.encoding
70		@encoding
	79	@internal_encoding
71	80	end
72	81
73	82	def Charset.encoding=(encoding)
74	83	warn("xsd charset is set to #{encoding}") if $DEBUG
75		@encoding = encoding
	84	@internal_encoding = encoding
76	85	end
77	86
78	87	def Charset.encoding_label
79		charset_label(@encoding)
	88	charset_label(@internal_encoding)
80	89	end
81	90
82	91	def Charset.encoding_to_xml(str, charset)
83		encoding_conv(str, @encoding, charset_str(charset))
	92	encoding_conv(str, @internal_encoding, charset_str(charset))
84	93	end
85	94
86	95	def Charset.encoding_from_xml(str, charset)
87		encoding_conv(str, charset_str(charset), @encoding)
	96	encoding_conv(str, charset_str(charset), @internal_encoding)
88	97	end
89	98
…	…
95	104	else
96	105	raise CharsetConversionError.new(
97		"Converter not found: #{ ~~enc_from } -> #{ enc_to~~ }")
	106	"Converter not found: #{enc_from} -> #{enc_to}")
98	107	end
99	108	end
…	…
113	122	# us_ascii = '[\x00-\x7F]'
114	123	us_ascii = '[\x9\xa\xd\x20-\x7F]' # XML 1.0 restricted.
115		USASCIIRegexp = Regexp.new("\\A#{ ~~us_ascii~~ }*\\z", nil, "NONE")
	124	USASCIIRegexp = Regexp.new("\\A#{us_ascii}*\\z", nil, "NONE")
116	125
117	126	twobytes_euc = '(?:[\x8E\xA1-\xFE][\xA1-\xFE])'
118	127	threebytes_euc = '(?:\x8F[\xA1-\xFE][\xA1-\xFE])'
119		character_euc = "(?:#{ ~~us_ascii }\|#{ twobytes_euc }\|#{ threebytes_euc~~ })"
120		EUCRegexp = Regexp.new("\\A#{ ~~character_euc~~ }*\\z", nil, "NONE")
	128	character_euc = "(?:#{us_ascii}\|#{twobytes_euc}\|#{threebytes_euc})"
	129	EUCRegexp = Regexp.new("\\A#{character_euc}*\\z", nil, "NONE")
121	130
122	131	# onebyte_sjis = '[\x00-\x7F\xA1-\xDF]'
123	132	onebyte_sjis = '[\x9\xa\xd\x20-\x7F\xA1-\xDF]' # XML 1.0 restricted.
124	133	twobytes_sjis = '(?:[\x81-\x9F\xE0-\xFC][\x40-\x7E\x80-\xFC])'
125		character_sjis = "(?:#{ ~~onebyte_sjis }\|#{ twobytes_sjis~~ })"
126		SJISRegexp = Regexp.new("\\A#{ ~~character_sjis~~ }*\\z", nil, "NONE")
	134	character_sjis = "(?:#{onebyte_sjis}\|#{twobytes_sjis})"
	135	SJISRegexp = Regexp.new("\\A#{character_sjis}*\\z", nil, "NONE")
127	136
128	137	# 0xxxxxxx
…	…
133	142	# 11110uuu 10uuuzzz 10yyyyyy 10xxxxxx
134	143	fourbytes_utf8 = '(?:[\xF0-\xF7][\x80-\xBF][\x80-\xBF][\x80-\xBF])'
135		character_utf8 = "(?:#{ us_ascii }\|#{ twobytes_utf8 }\|#{ threebytes_utf8 }\|#{ fourbytes_utf8 })"
136		UTF8Regexp = Regexp.new("\\A#{ character_utf8 }*\\z", nil, "NONE")
	144	character_utf8 =
	145	"(?:#{us_ascii}\|#{twobytes_utf8}\|#{threebytes_utf8}\|#{fourbytes_utf8})"
	146	UTF8Regexp = Regexp.new("\\A#{character_utf8}*\\z", nil, "NONE")
137	147
138	148	def Charset.is_us_ascii(str)
…	…
163	173	is_sjis(str)
164	174	else
165		raise UnknownCharsetError.new("Unknown charset: #{ ~~code~~ }")
	175	raise UnknownCharsetError.new("Unknown charset: #{code}")
166	176	end
167	177	end

Navigation

Changeset 1594

Legend:

trunk/lib/xsd/charset.rb

Download in other formats: