// ==UserScript==
// @name          ThaiWrapTest
// @namespace     http://mm.co.th/poonlap/thaiwraptest
// @description	  Wraping Thai long line with JavaScript.
// @include       *
// ==/UserScript==
/*  The code was taken from http://siit.net/members/art/thaiwrap.html
 *  by K.Arthit Suriyawongkul (http://bact.blogspot.com)
 *  I just put it in userscript format and add some content-type detection.
 *  It will do Thai line wrapping only for the page which has content-type tis-620, windows-874 or UTF-8.
 *  --- Poonlap Veerathanabutr <poonlap@mm.co.th> 2006-04-29
 */
<!--
function thaiWrap(separator) {
 /* compare function for sorting */
 function cnum(a,b){return a-b;}
 
 /* unambiguous words that are common, like prepositions */
 /* for example, not include "@7H-" since it may be part of "@7H-", "'H2"/"'H2", "7-"/"@7-" */
 /* @G|-"9H|0|C
I|DI|C+I|C|6|+#7-|A%0|
1|@7H-|I'"|I2|A%I'|1I|@#20|6H|I3|D!H|C
H|I-|
1|2
|6|1I|9I|'2!|*H'|"1|1H'|-7H|B"|*2!2#|@H2|CI|C*H|C|D'I|C+!H|C+
H|@%G
|C
%I|D
%|@2|
H'"|	1|I|@#G'|@I2|@
I2 */
 /* revised + new words by thep at linux.thai.net */ 
 cw="(\u0e40\u0e1b\u0e47\u0e19|\u0e2d\u0e22\u0e39\u0e48|\u0e08\u0e30|\u0e43\u0e0a\u0e49|\u0e44\u0e14\u0e49|\u0e43\u0e2b\u0e49|\u0e43\u0e19|\u0e08\u0e36\u0e07|\u0e2b\u0e23\u0e37\u0e2d|\u0e41\u0e25\u0e30|\u0e01\u0e31\u0e1a|\u0e40\u0e19\u0e37\u0e48\u0e2d\u0e07|\u0e14\u0e49\u0e27\u0e22|\u0e16\u0e49\u0e32|\u0e41\u0e25\u0e49\u0e27|\u0e17\u0e31\u0e49\u0e07|\u0e40\u0e1e\u0e23\u0e32\u0e30|\u0e0b\u0e36\u0e48\u0e07|\u0e0b\u0e49\u0e33|\u0e44\u0e21\u0e48|\u0e43\u0e0a\u0e48|\u0e15\u0e49\u0e2d\u0e07|\u0e01\u0e31\u0e19|\u0e08\u0e32\u0e01|\u0e16\u0e36\u0e07|\u0e19\u0e31\u0e49\u0e19|\u0e1c\u0e39\u0e49|\u0e04\u0e27\u0e32\u0e21|\u0e2a\u0e48\u0e27\u0e19|\u0e22\u0e31\u0e07|\u0e17\u0e31\u0e48\u0e27|\u0e2d\u0e37\u0e48\u0e19|\u0e42\u0e14\u0e22|\u0e2a\u0e32\u0e21\u0e32\u0e23\u0e16|\u0e40\u0e17\u0e48\u0e32|\u0e43\u0e15\u0e49|\u0e43\u0e2a\u0e48|\u0e43\u0e14|\u0e44\u0e27\u0e49|\u0e43\u0e2b\u0e21\u0e48|\u0e43\u0e2b\u0e0d\u0e48|\u0e40\u0e25\u0e47\u0e01|\u0e43\u0e01\u0e25\u0e49|\u0e44\u0e01\u0e25|\u0e40\u0e02\u0e32|\u0e0a\u0e48\u0e27\u0e22|\u0e09\u0e1a\u0e31\u0e1a|\u0e04\u0e49\u0e19|\u0e40\u0e23\u0e47\u0e27|\u0e40\u0e02\u0e49\u0e32|\u0e40\u0e0a\u0e49\u0e32)";
 /* leading chars */
 lc="[\u0e40-\u0e44]|\\(|\\[|\\{|\"";
 /* "following" chars */
 fc="\u0e2f|[\u0e30-\u0e3A]|[\u0e45-\u0e4e]|\\)|\\]|\\}|\"";
 /* thai chars */
 tc="\u0e01-\u0e3a\u0e40-\u0e4f\u0e5a\u0e5b";
 
 r=new Array();
 l=new Array();
 
 /* following chars followed by leading chars */
 r[0]=new RegExp("("+fc+")(?=("+lc+"))");
 l[0]=1;
 /* l >= 0 means using this value as length */
 
 /* thai followed by non-thai */
 r[1]=new RegExp("(["+tc+"])(?![\\)\\]\\}\"]|["+tc+"])");
 l[1]=1;
 
 /* non-thai followed by thai */
 r[2]=new RegExp("([^"+tc+"\\(\\(\\[\\{\"])(?=["+tc+"])");
 l[2]=1;
 
 /* non-leading char followed by known word */
 r[3]=new RegExp("([^"+lc+"])(?=("+lc+")*"+cw+"("+fc+")?)");
 l[3]=1;
 
 /* known word followed by non-following chars */
 r[4]=new RegExp("(("+lc+")*"+cw+"("+fc+")*)(?!"+fc+")");
 l[4]=-1;
 /* l < 0 means using length from match() function */
 
 /* end-of-word symbols */
 r[5]=new RegExp("([\u0e45\u0e46\u0e33])");
 l[5]=1;

 /* not break 
24, 
24 to 
2-4, 
2-4 */
 /* not work with Opera */ 
 /*r[6]=new RegExp("([\u0e30\u0e32-\u0e3a])(?=([^\u0e15]([\u0e30-\u0e3a\0e47\u0e4d])))")*/;
 /*l[6]=1;*/
 
 /* do breaks */
 function F(n){
  var p,a,c,x,t,e;
  if(n.nodeType==3){
   /* find all possible break points */
   p=new Array();
   for(i=0;i<r.length;i++){
    t=n.data.search(r[i]);
    if(t>=0){
     if(l[i]>=0){p.push(t+l[i]);}
     else{e=n.data.match(r[i]);p.push(t+e[1].length);}
    }
   }
   /* use the left-most break point */
   if(p.length>0){
    p.sort(cnum);
    if(p[0]>=0){
     a=n.splitText(p[0]);
     /*n.parentNode.insertBefore(document.createElement("WBR"),a);*/
     /* Use zero-width space instead of <WBR> */
     /* as Opera doesn't support <WBR>, http://www.quirksmode.org/oddsandends/wbr.html */
     n.parentNode.insertBefore(document.createTextNode(separator),a);
    }
   }
  }else{
   if(n.tagName!="STYLE"&&n.tagName!="SCRIPT"){
    for(c=0;x=n.childNodes[c];++c){F(x);}}
  }
 }
 
 /* do breaks for each body in every frames */
 function G(w){
  var fm=w.frames;
  if(fm.length<=0){
   F(document.body);
  }else{
   for(var i=0;i<fm.length;i++){
    D=fm[i].document;
    F(D.body);
   }
  }
 }
 
 G(window.self);
}
metas = document.getElementsByTagName("meta");
mayBeThai = 0;
thai_charset = /(tis-620|utf-8|windows-874)/i;
for ( i = 0; i < metas.length ; i++ ) {
	if( thai_charset.test(metas[i].content) ) {
		//alert(metas[i].content);
		mayBeThai = 1;
	}
}
if( mayBeThai == 1 ) {
  thaiWrap("|");
}
//-->