ruby-changes:15919
From: naruse <ko1@a...>
Date: Mon, 17 May 2010 15:28:30 +0900 (JST)
Subject: [ruby-changes:15919] Ruby:r27860 (trunk): * enc/iso_2022_jp.h: add CP50220.
naruse 2010-05-17 15:28:16 +0900 (Mon, 17 May 2010) New Revision: 27860 http://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=rev&revision=27860 Log: * enc/iso_2022_jp.h: add CP50220. * enc/trans/iso2022.trans: add converter for CP50220. Modified files: trunk/ChangeLog trunk/enc/iso_2022_jp.h trunk/enc/trans/iso2022.trans trunk/test/ruby/test_transcode.rb Index: ChangeLog =================================================================== --- ChangeLog (revision 27859) +++ ChangeLog (revision 27860) @@ -1,3 +1,9 @@ +Mon Apr 5 09:20:08 2010 NARUSE, Yui <naruse@r...> + + * enc/iso_2022_jp.h: add CP50220. + + * enc/trans/iso2022.trans: add converter for CP50220. + Mon May 17 09:37:25 2010 NAKAMURA Usaku <usa@r...> * lib/fileutils.rb (FileUtils::Entry_#entries): returns pathname in Index: enc/trans/iso2022.trans =================================================================== --- enc/trans/iso2022.trans (revision 27859) +++ enc/trans/iso2022.trans (revision 27860) @@ -358,6 +358,18 @@ } static const rb_transcoder +rb_cp50220_decoder = { + "CP50220", "cp51932", cp50221_decoder, + TRANSCODE_TABLE_INFO, + 1, /* input_unit_length */ + 3, /* max_input */ + 3, /* max_output */ + asciicompat_decoder, /* asciicompat_type */ + 1, iso2022jp_init, iso2022jp_init, /* state_size, state_init, state_fini */ + NULL, fun_si_cp50221_decoder, NULL, fun_so_cp50221_decoder +}; + +static const rb_transcoder rb_cp50221_decoder = { "CP50221", "cp51932", cp50221_decoder, TRANSCODE_TABLE_INFO, @@ -370,7 +382,8 @@ }; static ssize_t -fun_so_cp50221_encoder(void *statep, const unsigned char *s, size_t l, unsigned char *o, size_t osize) +fun_so_cp5022x_encoder(void *statep, const unsigned char *s, size_t l, + unsigned char *o, size_t osize) { unsigned char *sp = statep; unsigned char *output0 = o; @@ -425,11 +438,121 @@ 5, /* max_output */ asciicompat_encoder, /* asciicompat_type */ 1, iso2022jp_init, iso2022jp_init, /* state_size, state_init, state_fini */ - NULL, NULL, NULL, fun_so_cp50221_encoder, + NULL, NULL, NULL, fun_so_cp5022x_encoder, finish_iso2022jp_encoder, iso2022jp_encoder_reset_sequence_size, finish_iso2022jp_encoder }; +static const char *tbl0208 = + "\x21\x23\x21\x56\x21\x57\x21\x22\x21\x26\x25\x72\x25\x21\x25\x23" \ + "\x25\x25\x25\x27\x25\x29\x25\x63\x25\x65\x25\x67\x25\x43\x21\x3C" \ + "\x25\x22\x25\x24\x25\x26\x25\x28\x25\x2A\x25\x2B\x25\x2D\x25\x2F" \ + "\x25\x31\x25\x33\x25\x35\x25\x37\x25\x39\x25\x3B\x25\x3D\x25\x3F" \ + "\x25\x41\x25\x44\x25\x46\x25\x48\x25\x4A\x25\x4B\x25\x4C\x25\x4D" \ + "\x25\x4E\x25\x4F\x25\x52\x25\x55\x25\x58\x25\x5B\x25\x5E\x25\x5F" \ + "\x25\x60\x25\x61\x25\x62\x25\x64\x25\x66\x25\x68\x25\x69\x25\x6A" \ + "\x25\x6B\x25\x6C\x25\x6D\x25\x6F\x25\x73\x21\x2B\x21\x2C"; + +static ssize_t +fun_so_cp50220_encoder(void *statep, const unsigned char *s, size_t l, + unsigned char *o, size_t osize) +{ + unsigned char *output0 = o; + unsigned char *sp = statep; + + if (sp[0] == G0_JISX0201_KATAKANA) { + int c = sp[2] & 0x7F; + const char *p = tbl0208 + (c - 0x21) * 2; + if (sp[1] != G0_JISX0208_1983) { + *o++ = 0x1b; + *o++ = '$'; + *o++ = 'B'; + } + sp[0] = G0_JISX0208_1983; + *o++ = *p++; + if (l == 2 && s[0] == 0x8E) { + if (s[1] == 0xDE) { + *o++ = *p + 1; + return o - output0; + } + else if (s[1] == 0xDF && (0x4A <= c && c <= 0x4E)) { + *o++ = *p + 2; + return o - output0; + } + } + *o++ = *p; + } + + if (l == 2 && s[0] == 0x8E) { + const char *p = tbl0208 + (s[1] - 0xA1) * 2; + if ((0xA1 <= s[1] && s[1] <= 0xB5) || + (0xC5 <= s[1] && s[1] <= 0xC9) || + (0xCF <= s[1] && s[1] <= 0xDF)) { + if (*sp != G0_JISX0208_1983) { + *o++ = 0x1b; + *o++ = '$'; + *o++ = 'B'; + *sp = G0_JISX0208_1983; + } + *o++ = *p++; + *o++ = *p; + return o - output0; + } + + sp[2] = s[1]; + sp[1] = sp[0]; + sp[0] = G0_JISX0201_KATAKANA; + return o - output0; + } + + o += fun_so_cp5022x_encoder(statep, s, l, o, osize); + return o - output0; +} + +static ssize_t +finish_cp50220_encoder(void *statep, unsigned char *o, size_t osize) +{ + unsigned char *sp = statep; + unsigned char *output0 = o; + + if (*sp == G0_ASCII) + return 0; + + if (sp[0] == G0_JISX0201_KATAKANA) { + int c = sp[2] & 0x7F; + const char *p = tbl0208 + (c - 0x21) * 2; + if (sp[1] != G0_JISX0208_1983) { + *o++ = 0x1b; + *o++ = '$'; + *o++ = 'B'; + } + sp[0] = G0_JISX0208_1983; + *o++ = *p++; + *o++ = *p; + } + + *o++ = 0x1b; + *o++ = '('; + *o++ = 'B'; + *sp = G0_ASCII; + + return o - output0; +} + +static const rb_transcoder +rb_cp50220_encoder = { + "CP51932", "CP50220", cp50221_encoder, + TRANSCODE_TABLE_INFO, + 1, /* input_unit_length */ + 3, /* max_input */ + 5, /* max_output */ + asciicompat_encoder, /* asciicompat_type */ + 3, iso2022jp_init, iso2022jp_init, /* state_size, state_init, state_fini */ + NULL, NULL, NULL, fun_so_cp50220_encoder, + finish_cp50220_encoder, + iso2022jp_encoder_reset_sequence_size, finish_cp50220_encoder +}; + void Init_iso2022(void) { @@ -437,7 +560,9 @@ rb_register_transcoder(&rb_iso2022jp_encoder); rb_register_transcoder(&rb_stateless_iso2022jp_to_eucjp); rb_register_transcoder(&rb_eucjp_to_stateless_iso2022jp); + rb_register_transcoder(&rb_cp50220_decoder); rb_register_transcoder(&rb_cp50221_decoder); + rb_register_transcoder(&rb_cp50220_encoder); rb_register_transcoder(&rb_cp50221_encoder); } Index: enc/iso_2022_jp.h =================================================================== --- enc/iso_2022_jp.h (revision 27859) +++ enc/iso_2022_jp.h (revision 27860) @@ -5,16 +5,34 @@ ENC_REPLICATE("ISO-2022-JP-2", "ISO-2022-JP"); ENC_ALIAS("ISO2022-JP2", "ISO-2022-JP-2"); +/* Windows Codepage 50220 + * a ISO-2022-JP variant. + * This includes + * * US-ASCII + * * JIS X 0201 Latin + * * JIS X 0201 Katakana + * * JIS X 0208 + * * NEC special characters + * * NEC selected IBM extended characters + * and this implementation doesn't include + * * User Defined Characters + * + * So this CP50220 has the same characters of CP51932. + * + * See http://legacy-encoding.sourceforge.jp/wiki/index.php?cp50220 + */ +ENC_REPLICATE("CP50220", "ISO-2022-JP"); + /* Windows Codepage 50221 * a ISO-2022-JP variant. * This includes * * US-ASCII + * * JIS X 0201 Latin * * JIS X 0201 Katakana * * JIS X 0208 * * NEC special characters * * NEC selected IBM extended characters * and this implementation doesn't include - * * JIS X 0201 Latin * * User Defined Characters * * So this CP50221 has the same characters of CP51932. Index: test/ruby/test_transcode.rb =================================================================== --- test/ruby/test_transcode.rb (revision 27859) +++ test/ruby/test_transcode.rb (revision 27860) @@ -1373,8 +1373,17 @@ assert_equal("\u5fde", "\e$B\x7A\x21".encode("utf-8", "cp50221")) assert_equal("\u72be", "\e$B\x7B\x21".encode("utf-8", "cp50221")) assert_equal("\u91d7", "\e$B\x7C\x21".encode("utf-8", "cp50221")) + assert_equal("\e(I!_\e(B", "\xA1\xDF".encode("cp50220","sjis")) end + def test_cp50221 + assert_equal("\e$B!#!,\e(B".force_encoding("cp50220"), + "\xA1\xDF".encode("cp50220","sjis")) + assert_equal("\e$B%*!+%,%I%J!+%N!+%P%\\%^!+%Q%]%\"\e(B".force_encoding("cp50220"), + "\xB5\xDE\xB6\xDE\xC4\xDE\xC5\xDE\xC9\xDE\xCA\xDE\xCE\xDE\xCF\xDE\xCA\xDF\xCE\xDF\xB1". + encode("cp50220", "sjis")) + end + def test_iso_2022_jp_1 # check_both_ways("\u9299", "\x1b$(Dd!\x1b(B", "iso-2022-jp-1") # JIS X 0212 8 1 end -- ML: ruby-changes@q... Info: http://www.atdot.net/~ko1/quickml/