|
| 1 | +#! /usr/bin/perl |
| 2 | +# |
| 3 | +# Copyright (c) 2007, PostgreSQL Global Development Group |
| 4 | +# |
| 5 | +# $PostgreSQL: pgsql/src/backend/utils/mb/Unicode/UCS_to_EUC_JIS_2004.pl,v 1.1 2007/03/25 11:56:02 ishii Exp $ |
| 6 | +# |
| 7 | +# Generate UTF-8 <--> EUC_JIS_2004 code conversion tables from |
| 8 | +# "euc-jis-2004-std.txt" (http://x0213.org) |
| 9 | + |
| 10 | +require"ucs2utf.pl"; |
| 11 | + |
| 12 | +$TEST = 1; |
| 13 | + |
| 14 | +# first generate UTF-8 --> EUC_JIS_2004 table |
| 15 | + |
| 16 | +$in_file ="euc-jis-2004-std.txt"; |
| 17 | + |
| 18 | +open( FILE,$in_file ) ||die("cannot open$in_file" ); |
| 19 | + |
| 20 | +reset'array'; |
| 21 | +reset'array1'; |
| 22 | +reset'comment'; |
| 23 | +reset'comment1'; |
| 24 | + |
| 25 | +while($line = <FILE> ){ |
| 26 | +if ($line =~/^0x(.*)[\t]*U\+(.*)\+(.*)[\t]*#(.*)$/) { |
| 27 | +$c =$1; |
| 28 | +$u1 =$2; |
| 29 | +$u2 =$3; |
| 30 | +$rest ="U+" .$u1 ."+" .$u2 .$4; |
| 31 | +$code =hex($c); |
| 32 | +$ucs =hex($u1); |
| 33 | +$utf1 = &ucs2utf($ucs); |
| 34 | +$ucs =hex($u2); |
| 35 | +$utf2 = &ucs2utf($ucs); |
| 36 | +$str =sprintf"%08x%08x",$utf1,$utf2; |
| 37 | +$array1{$str } =$code; |
| 38 | +$comment1{$str } =$rest; |
| 39 | +$count1++; |
| 40 | +next; |
| 41 | +}elsif ($line =~/^0x(.*)[\t]*U\+(.*)[\t]*#(.*)$/) { |
| 42 | +$c =$1; |
| 43 | +$u =$2; |
| 44 | +$rest ="U+" .$u .$3; |
| 45 | +}else { |
| 46 | +next; |
| 47 | +} |
| 48 | + |
| 49 | +$ucs =hex($u); |
| 50 | +$code =hex($c); |
| 51 | +$utf = &ucs2utf($ucs); |
| 52 | +if($array{$utf }ne"" ){ |
| 53 | +printfSTDERR"Warning: duplicate UTF8: %04x\n",$ucs; |
| 54 | +next; |
| 55 | +} |
| 56 | +$count++; |
| 57 | + |
| 58 | +$array{$utf } =$code; |
| 59 | +$comment{$code } =$rest; |
| 60 | +} |
| 61 | +close( FILE ); |
| 62 | + |
| 63 | +$file ="utf8_to_euc_jis_2004.map"; |
| 64 | +open( FILE,">$file" ) ||die("cannot open$file" ); |
| 65 | +print FILE"/*\n"; |
| 66 | +print FILE" * This file was generated by UCS_to_EUC_JIS_2004.pl\n"; |
| 67 | +print FILE" */\n"; |
| 68 | +print FILE"static pg_utf_to_local ULmapEUC_JIS_2004[] = {\n"; |
| 69 | + |
| 70 | +for$index (sort {$a<=>$b}keys(%array ) ){ |
| 71 | +$code =$array{$index }; |
| 72 | +$count--; |
| 73 | +if($count == 0 ){ |
| 74 | +printf FILE" {0x%08x, 0x%06x}/*%s */\n",$index,$code,$comment{$code }; |
| 75 | +}else { |
| 76 | +printf FILE" {0x%08x, 0x%06x},/*%s */\n",$index,$code,$comment{$code }; |
| 77 | +} |
| 78 | +} |
| 79 | + |
| 80 | +print FILE"};\n"; |
| 81 | +close(FILE); |
| 82 | + |
| 83 | +if ($TEST == 1) { |
| 84 | +$file1 ="utf8.data"; |
| 85 | +$file2 ="euc_jis_2004.data"; |
| 86 | +open( FILE1,">$file1" ) ||die("cannot open$file1" ); |
| 87 | +open( FILE2,">$file2" ) ||die("cannot open$file2" ); |
| 88 | + |
| 89 | +for$index (sort {$a<=>$b}keys(%array ) ){ |
| 90 | +$code =$array{$index }; |
| 91 | +if ($code > 0x00 &&$code != 0x09 &&$code != 0x0a &&$code != 0x0d && |
| 92 | +$code != 0x5c && |
| 93 | +($code < 0x80 || |
| 94 | + ($code >= 0x8ea1 &&$code <= 0x8efe) || |
| 95 | + ($code >= 0x8fa1a1 &&$code <= 0x8ffefe) || |
| 96 | + ($code >= 0xa1a1 &&$code <= 0x8fefe))) { |
| 97 | +for ($i = 3;$i >= 0;$i--) { |
| 98 | +$s =$i * 8; |
| 99 | +$mask = 0xff <<$s; |
| 100 | +print FILE1pack("C", ($index &$mask) >>$s)if$index &$mask; |
| 101 | +print FILE2pack("C", ($code &$mask) >>$s)if$code &$mask; |
| 102 | +} |
| 103 | +print FILE1"\n"; |
| 104 | +print FILE2"\n"; |
| 105 | +} |
| 106 | +} |
| 107 | +} |
| 108 | + |
| 109 | +$file ="utf8_to_euc_jis_2004_combined.map"; |
| 110 | +open( FILE,">$file" ) ||die("cannot open$file" ); |
| 111 | +print FILE"/*\n"; |
| 112 | +print FILE" * This file was generated by UCS_to_EUC_JIS_2004.pl\n"; |
| 113 | +print FILE" */\n"; |
| 114 | +print FILE"static pg_utf_to_local_combined ULmapEUC_JIS_2004_combined[] = {\n"; |
| 115 | + |
| 116 | +for$index (sort {$acmp$b}keys(%array1 ) ){ |
| 117 | +$code =$array1{$index }; |
| 118 | +$count1--; |
| 119 | +if($count1 == 0 ){ |
| 120 | +printf FILE" {0x%s, 0x%s, 0x%06x}/*%s */\n",substr($index, 0, 8),substr($index, 8, 8),$code,$comment1{$index }; |
| 121 | +}else { |
| 122 | +printf FILE" {0x%s, 0x%s, 0x%06x},/*%s */\n",substr($index, 0, 8),substr($index, 8, 8),$code,$comment1{$index }; |
| 123 | +} |
| 124 | +} |
| 125 | + |
| 126 | +print FILE"};\n"; |
| 127 | +close(FILE); |
| 128 | + |
| 129 | +if ($TEST == 1) { |
| 130 | +for$index (sort {$acmp$b}keys(%array1 ) ){ |
| 131 | +$code =$array1{$index }; |
| 132 | +if ($code > 0x00 &&$code != 0x09 &&$code != 0x0a &&$code != 0x0d && |
| 133 | +$code != 0x5c && |
| 134 | +($code < 0x80 || |
| 135 | + ($code >= 0x8ea1 &&$code <= 0x8efe) || |
| 136 | + ($code >= 0x8fa1a1 &&$code <= 0x8ffefe) || |
| 137 | + ($code >= 0xa1a1 &&$code <= 0x8fefe))) { |
| 138 | + |
| 139 | +$v1 =hex(substr($index, 0, 8)); |
| 140 | +$v2 =hex(substr($index, 8, 8)); |
| 141 | + |
| 142 | +for ($i = 3;$i >= 0;$i--) { |
| 143 | +$s =$i * 8; |
| 144 | +$mask = 0xff <<$s; |
| 145 | +print FILE1pack("C", ($v1 &$mask) >>$s)if$v1 &$mask; |
| 146 | +print FILE2pack("C", ($code &$mask) >>$s)if$code &$mask; |
| 147 | +} |
| 148 | +for ($i = 3;$i >= 0;$i--) { |
| 149 | +$s =$i * 8; |
| 150 | +$mask = 0xff <<$s; |
| 151 | +print FILE1pack("C", ($v2 &$mask) >>$s)if$v2 &$mask; |
| 152 | +} |
| 153 | +print FILE1"\n"; |
| 154 | +print FILE2"\n"; |
| 155 | +} |
| 156 | +} |
| 157 | +close(FILE1); |
| 158 | +close(FILE2); |
| 159 | +} |
| 160 | + |
| 161 | +# then generate EUC_JIS_2004 --> UTF-8 table |
| 162 | + |
| 163 | +$in_file ="euc-jis-2004-std.txt"; |
| 164 | + |
| 165 | +open( FILE,$in_file ) ||die("cannot open$in_file" ); |
| 166 | + |
| 167 | +reset'array'; |
| 168 | +reset'array1'; |
| 169 | +reset'comment'; |
| 170 | +reset'comment1'; |
| 171 | + |
| 172 | +while($line = <FILE> ){ |
| 173 | +if ($line =~/^0x(.*)[\t]*U\+(.*)\+(.*)[\t]*#(.*)$/) { |
| 174 | +$c =$1; |
| 175 | +$u1 =$2; |
| 176 | +$u2 =$3; |
| 177 | +$rest ="U+" .$u1 ."+" .$u2 .$4; |
| 178 | +$code =hex($c); |
| 179 | +$ucs =hex($u1); |
| 180 | +$utf1 = &ucs2utf($ucs); |
| 181 | +$ucs =hex($u2); |
| 182 | +$utf2 = &ucs2utf($ucs); |
| 183 | +$str =sprintf"%08x%08x",$utf1,$utf2; |
| 184 | +$array1{$code } =$str; |
| 185 | +$comment1{$code } =$rest; |
| 186 | +$count1++; |
| 187 | +next; |
| 188 | +}elsif ($line =~/^0x(.*)[\t]*U\+(.*)[\t]*#(.*)$/) { |
| 189 | +$c =$1; |
| 190 | +$u =$2; |
| 191 | +$rest ="U+" .$u .$3; |
| 192 | +}else { |
| 193 | +next; |
| 194 | +} |
| 195 | + |
| 196 | +$ucs =hex($u); |
| 197 | +$code =hex($c); |
| 198 | +$utf = &ucs2utf($ucs); |
| 199 | +if($array{$code }ne"" ){ |
| 200 | +printfSTDERR"Warning: duplicate UTF8: %04x\n",$ucs; |
| 201 | +next; |
| 202 | +} |
| 203 | +$count++; |
| 204 | + |
| 205 | +$array{$code } =$utf; |
| 206 | +$comment{$utf } =$rest; |
| 207 | +} |
| 208 | +close( FILE ); |
| 209 | + |
| 210 | +$file ="euc_jis_2004_to_utf8.map"; |
| 211 | +open( FILE,">$file" ) ||die("cannot open$file" ); |
| 212 | +print FILE"/*\n"; |
| 213 | +print FILE" * This file was generated by UCS_to_EUC_JIS_2004.pl\n"; |
| 214 | +print FILE" */\n"; |
| 215 | +print FILE"static pg_local_to_utf LUmapEUC_JIS_2004[] = {\n"; |
| 216 | + |
| 217 | +for$index (sort {$a<=>$b}keys(%array ) ){ |
| 218 | +$code =$array{$index }; |
| 219 | +$count--; |
| 220 | +if($count == 0 ){ |
| 221 | +printf FILE" {0x%06x, 0x%08x}/*%s */\n",$index,$code,$comment{$code }; |
| 222 | +}else { |
| 223 | +printf FILE" {0x%06x, 0x%08x},/*%s */\n",$index,$code,$comment{$code }; |
| 224 | +} |
| 225 | +} |
| 226 | + |
| 227 | +print FILE"};\n"; |
| 228 | +close(FILE); |
| 229 | + |
| 230 | +$file ="euc_jis_2004_to_utf8_combined.map"; |
| 231 | +open( FILE,">$file" ) ||die("cannot open$file" ); |
| 232 | +print FILE"/*\n"; |
| 233 | +print FILE" * This file was generated by UCS_to_EUC_JIS_2004.pl\n"; |
| 234 | +print FILE" */\n"; |
| 235 | +print FILE"static pg_local_to_utf_combined LUmapEUC_JIS_2004_combined[] = {\n"; |
| 236 | + |
| 237 | +for$index (sort {$a<=>$b}keys(%array1 ) ){ |
| 238 | +$code =$array1{$index }; |
| 239 | +$count1--; |
| 240 | +if($count1 == 0 ){ |
| 241 | +printf FILE" {0x%06x, 0x%s, 0x%s}/*%s */\n",$index,substr($code, 0, 8),substr($code, 8, 8),$comment1{$index }; |
| 242 | +}else { |
| 243 | +printf FILE" {0x%06x, 0x%s, 0x%s},/*%s */\n",$index,substr($code, 0, 8),substr($code, 8, 8),$comment1{$index }; |
| 244 | +} |
| 245 | +} |
| 246 | + |
| 247 | +print FILE"};\n"; |
| 248 | +close(FILE); |