KidzSearch Encyclopedia:AutoEd/fullwidth.js

//<source lang=javascript>

function autoEdFullwidth(str) { //MAIN FUNCTION describes list of fixes

//Spacing, periods and interpucts
 str = str.replace(/　/g, ' '); //fullwidth space
 //str = str.replace(/・/g, '·'); //fullwidth interpunct
 str = str.replace(/。/g, '.'); //replace ideographic period with a regular period (from TheFarix's AWB cleanup module)
 //str = str.replace(/・・・/g, '...'); //three interpuncts normally mean an ellipsis
 //str = str.replace(/···/g, '...'); //catch already converted triple interpuncts - it's after the interpunct rule so we only need one go-through to catch partial replacements
 str = str.replace(/…/g, '...'); //it's a little ballsy, I guess, but we're already doing lots of replacements, so... (see also WP:ELLIPSES)

//Common punctuation
 str = str.replace(/！/g, '!');
 str = str.replace(/（/g, '(');
 str = str.replace(/）/g, ')');
 str = str.replace(/？/g, '?');

//Numbers
 str = str.replace(/０/g, '0');
 str = str.replace(/１/g, '1');
 str = str.replace(/２/g, '2');
 str = str.replace(/３/g, '3');
 str = str.replace(/４/g, '4');
 str = str.replace(/５/g, '5');
 str = str.replace(/６/g, '6');
 str = str.replace(/７/g, '7');
 str = str.replace(/８/g, '8');
 str = str.replace(/９/g, '9');

//Latin letters, uppercase
 str = str.replace(/Ａ/g, 'Α');
 str = str.replace(/Ｂ/g, 'B');
 str = str.replace(/Ｃ/g, 'C');
 str = str.replace(/Ｄ/g, 'D');
 str = str.replace(/Ｅ/g, 'E');
 str = str.replace(/Ｆ/g, 'F');
 str = str.replace(/Ｇ/g, 'G');
 str = str.replace(/Ｈ/g, 'H');
 str = str.replace(/Ｉ/g, 'I');
 str = str.replace(/Ｊ/g, 'J');
 str = str.replace(/Ｋ/g, 'K');
 str = str.replace(/Ｌ/g, 'L');
 str = str.replace(/Ｍ/g, 'M');
 str = str.replace(/Ｎ/g, 'N');
 str = str.replace(/Ｏ/g, 'O');
 str = str.replace(/Ｐ/g, 'P');
 str = str.replace(/Ｑ/g, 'Q');
 str = str.replace(/Ｒ/g, 'R');
 str = str.replace(/Ｓ/g, 'S');
 str = str.replace(/Ｔ/g, 'T');
 str = str.replace(/Ｕ/g, 'U');
 str = str.replace(/Ｖ/g, 'V');
 str = str.replace(/Ｗ/g, 'W');
 str = str.replace(/Ｘ/g, 'X');
 str = str.replace(/Ｙ/g, 'Y');
 str = str.replace(/Ｚ/g, 'Z');

//Latin letters, lowercase
 str = str.replace(/ａ/g, 'a');
 str = str.replace(/ｂ/g, 'b');
 str = str.replace(/ｃ/g, 'c');
 str = str.replace(/ｄ/g, 'd');
 str = str.replace(/ｅ/g, 'e');
 str = str.replace(/ｆ/g, 'f');
 str = str.replace(/ｇ/g, 'g');
 str = str.replace(/ｈ/g, 'h');
 str = str.replace(/ｉ/g, 'i');
 str = str.replace(/ｊ/g, 'j');
 str = str.replace(/ｋ/g, 'k');
 str = str.replace(/ｌ/g, 'l');
 str = str.replace(/ｍ/g, 'm');
 str = str.replace(/ｎ/g, 'n');
 str = str.replace(/ｏ/g, 'o');
 str = str.replace(/ｐ/g, 'p');
 str = str.replace(/ｑ/g, 'q');
 str = str.replace(/ｒ/g, 'r');
 str = str.replace(/ｓ/g, 's');
 str = str.replace(/ｔ/g, 't');
 str = str.replace(/ｕ/g, 'u');
 str = str.replace(/ｖ/g, 'v');
 str = str.replace(/ｗ/g, 'w');
 str = str.replace(/ｘ/g, 'x');
 str = str.replace(/ｙ/g, 'y');
 str = str.replace(/ｚ/g, 'z');

//Other punctuation
 str = str.replace(/＂/g, '"');
 str = str.replace(/＃/g, '#');
 str = str.replace(/＄/g, '$');
 str = str.replace(/％/g, '%');
 str = str.replace(/＆/g, '&');
 str = str.replace(/＇/g, '\);
 str = str.replace(/＊/g, '*');
 str = str.replace(/＋/g, '+');
 str = str.replace(/，/g, ', ');
 //str = str.replace(/、/g, ', '); //replace ideographic comma with a regular comma (from TheFarix's AWB cleanup module)
 str = str.replace(/－/g, '-');
 str = str.replace(/．/g, '. ');
 str = str.replace(/／/g, '\/');
 str = str.replace(/：/g, ': ');
 str = str.replace(/；/g, '; ');
 str = str.replace(/＜/g, '<');
 str = str.replace(/＝/g, '='); //this replacement could break template usage, use = as an alternative inside templates
 str = str.replace(/＞/g, '>');
 str = str.replace(/＠/g, '@');
 str = str.replace(/［/g, '['); //this replacement could break wikimarkup usage
 str = str.replace(/＼/g, '\\');
 str = str.replace(/］/g, ']'); //this replacement could break wikimarkup usage
 str = str.replace(/＾/g, '^');
 str = str.replace(/＿/g, '_');
 str = str.replace(/｀/g, '`');
 str = str.replace(/｛/g, '{'); //this replacement could break wikimarkup usage, use Template:( as an alternative
 str = str.replace(/｜/g, '|'); //this replacement could break template usage, use | as an alternative inside templates
 str = str.replace(/｝/g, '}'); //this replacement could break wikimarkup usage, use } as an alternative
 str = str.replace(/～/g, '〜'); //ASCII tilde is, apparently, treated as a diacritic in some fonts, so this should be a better replacement
 str = str.replace(/￠/g, '¢');
 str = str.replace(/￡/g, '£');
 str = str.replace(/￢/g, '¬');
 str = str.replace(/￤/g, '¦');
 str = str.replace(/￥/g, '¥');
 str = str.replace(/￦/g, '₩');
 str = str.replace(/￨/g, '│');
 str = str.replace(/￩/g, '←');
 str = str.replace(/￪/g, '↑');
 str = str.replace(/￫/g, '→');
 str = str.replace(/￬/g, '↓');
 str = str.replace(/￭/g, '■');
 str = str.replace(/￮/g, '○');
 str = str.replace(/『/g, '「'); //some Japanese quote replacement (from TheFarix's AWB cleanup module)
 str = str.replace(/』/g, '」'); //some Japanese quote replacement (from TheFarix's AWB cleanup module)

 return str;

}

//</source>