iapt-platform
/
mint
mirror of https://github.com/iapt-platform/mint.git


			
							123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310
							<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
	<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
	<script>
		function nextPar(arr,index)
		{
			if(index<arr.length-1)
			{
				for(var j=index+1;j<arr.length;j++)//循环arr的长度次
				{
					if(arr[j].id>0)
					{
						return(j)//把J的值返回成为函数值
					}
				}
				return(-1)
			}
			else{
				return(-1);
			}
		}
	
		function do_final(){
			var biaodian = document.getElementById("biaodian").value.split("$")
			var biaodian_no = document.getElementById("biaodian_no").value.split("$")
			var strPali = document.getElementById("input_pali").value;
			var strPaliPar=strPali.split("\n");/*按回车拆分原文为数组*/
			var strTran = document.getElementById("output_translate").value;
			var strTranPar=strTran.split("\n");//按回车拆分译文为数组
			var strTranList = new Array()//创建原本译文数组
			var newTran = new Array()//创建结果译文数组
			
		for(var i=0;i<strTranPar.length;i++)//依照译文行数进行循环
		{
			
			parPart=strTranPar[i].split("\t")//依照tab将译文一维数组拆分成二维数组
			var newPar=new Object()//创建对象
			newPar.id=parPart[0]*10
			newPar.text=parPart[1]//赋值：译文内容
			newPar.id=newPar.id/10//赋值：巴利原文段落号

			if(newPar.text){
				if(newPar.id>0){
					strTranList.push(newPar)//将一维矩阵塞入译文二维数组末端
				}
				else{
					for(ibd in biaodian){
						eval("newPar.text=newPar.text.replace(/"+biaodian[ibd]+"/g,'"+biaodian[ibd]+"\t')")
						eval("newPar.text=newPar.text.replace(/\t"+biaodian[ibd]+"/g,'"+biaodian[ibd]+"')")
					}
					splitText=newPar.text.split("\t")
					for(x in splitText){
						var newSplitObj=new Object()//创建对象
						newSplitObj.id=0
						newSplitObj.text=splitText[x]
						var strNoSpace=newSplitObj.text.replace(/ /g,"")
						if(strNoSpace.length>0){
							
							var isBdNo=false
							for(ibdno in biaodian_no){
								if(strNoSpace==biaodian_no[ibdno]){
									isBdNo=true;
									break;
								}
							}
							if(isBdNo)
							{
								strTranList[strTranList.length-1].text+=newSplitObj.text
							}
							else
							{
								strTranList.push(newSplitObj)
							}
						}
					}
				}
				}
				
			}//依照tab将译文一维数组转换成二维数组
			
			//for(x in strTranList){
			//	document.getElementById("info").innerHTML+=(strTranList[x].id+"-"+strTranList[x].text+"<br />")
			//}
			//return;
			
			var currPar=0;//定义当前译文段落号，起始值0
			var currId=strTranList[currPar].id//定义当前译文ID，赋值：当前译文对应的巴利原文段落号
			while(currPar<strTranList.length-1)//执行次数=strTranList的长度-1
			{
				var iNextPar=nextPar(strTranList,currPar)//定义下一标题译文段落号
				var nextId=strTranList[iNextPar].id//定义下一标题ID，赋值：对应的巴利原文段落号
				if(iNextPar==currPar+1){//紧挨着
					if(nextId==currId+1){//正好
						newTran.push(strTranList[currPar].text)//推送当前段落译文到结果译文数组
					}
					else{  //不够 拆分
						//拆分后的段落数
						var iSplitNum = nextId-currId;
						var strCombinPali="";
						//生成合并pali 并计算每段比例
						for(var k=currId-1;k<nextId-1;k++){
							strCombinPali=strCombinPali+strPaliPar[k]
						}
						splitText=new Array()
						iCurrSplitBegin=0
						for(var k=0;k<iSplitNum;k++){
							newRate=strPaliPar[currId-1+k].length/strCombinPali.length
							iCurrSplitEnd=iCurrSplitBegin+(strTranList[currPar].text.length*newRate)
							newText=strTranList[currPar].text.slice(iCurrSplitBegin,iCurrSplitEnd)
							iCurrSplitBegin=iCurrSplitEnd
							newTran.push(newText)	
							//document.getElementById("info").innerHTML=document.getElementById("info").innerHTML+newRate+" "+currId+" "+nextId+"<br>"
						}
					}
				}
				else//不挨着
				{
					if(nextId-currId==iNextPar-currPar){//正好，下一标题原文号-当前标题原文号=下一标题译文号-当前标题译文号
						for(var iNew=currPar;iNew<iNextPar;iNew++)
						{
							newTran.push(strTranList[iNew].text)
						}
					}
					else if(nextId-currId>iNextPar-currPar){//拆，原文段数>译文段数
						newTran.push(strTranList[currPar].text)
						//拆分后的段落数
						var iSplitNum = nextId-currId-1;
						
						//生成合并译文
						var strCombinTran="";
						for(var k=currPar+1;k<iNextPar;k++){
							strCombinTran=strCombinTran+strTranList[k].text
						}						
						//生成合并pali 并计算每段比例
						var strCombinPali="";//定义合并巴利原文
						for(var k=currId;k<nextId-1;k++)
						{
							strCombinPali=strCombinPali+strPaliPar[k]
						}//将巴利原文合并
						splitText=new Array()//定义新数组
						iCurrSplitBegin=0
						for(var k=0;k<iSplitNum;k++)//循环巴利原文段落数次
						{
							newRate=strPaliPar[currId+k].length/strCombinPali.length//从巴利原文数组中提取当前段落字符数和总字符数之比
							iCurrSplitEnd=iCurrSplitBegin+(strCombinTran.length*newRate)//结束编码=起始编码+中间编码*比例
							newText=strCombinTran.slice(iCurrSplitBegin,iCurrSplitEnd)//从起始到结束，取出
							iCurrSplitBegin=iCurrSplitEnd
							newTran.push(newText)	
							//document.getElementById("info").innerHTML=document.getElementById("info").innerHTML+newRate+" "+currId+" "+nextId+"<br>"
						}
					}
					else{//合并
					newTran.push(strTranList[currPar].text)
						//拆分后的段落数
						var iSplitNum = nextId-currId-1;
						var tranNum=new Array()
						//生成合并译文
						var strCombinTran="";
						for(var k=currPar+1;k<iNextPar;k++){
							strCombinTran=strCombinTran+strTranList[k].text
							tranNum.push(strCombinTran.length)
						}
						var paliNum=new Array()
						//生成合并pali 并计算每段比例
						var strCombinPali="";
						for(var k=currId;k<nextId-1;k++)
						{
							strCombinPali=strCombinPali+strPaliPar[k]
							paliNum.push(strCombinPali.length)
						}
						var theoNum=new Array()
						var PCRate=strCombinTran.length/strCombinPali.length//译巴字数比
						for(var k=0;k<nextId-currId-1;k++)//得到理论字数
						{
							theoNum.push(paliNum[k]*PCRate)
							//document.getElementById("info").innerHTML=document.getElementById("info").innerHTML+paliNum[k]+" "+theoNum[k]+"<br>"
						}
function tranPos(J)//输入译文段号返回合并段号
	{
	if(tranNum[J]-theoNum[0]<0)
		{
		return(0)
		}
	else if(J==0)
		{
		return(0)
		}
	else if(tranNum[J]>=theoNum[nextId-currId-2])
		{
		return(nextId-currId-2)
		}
	else
	{
	
	for(var i=1;i<nextId-currId-1;i++)
		{
		if(tranNum[J]>theoNum[i])//译文J长度大于原文i理论值
			{}
		else//原文i理论值>译文J长度>原文i-1理论值
			{
			if(Math.abs(tranNum[J]-theoNum[i])<Math.abs(tranNum[J]-theoNum[i-1]))//更靠近原文i理论值
				{
				if(Math.abs(tranNum[J]-theoNum[i])<Math.abs(tranNum[J+1]-theoNum[i]))//译文J比译文J+1更靠近理论值
					{
					return(i)
					}
				else//译文J+1比译文J更靠近理论值
					{
					return(i)
					}
				}
			else//更靠近原文i-1理论值
				{
				if(Math.abs(tranNum[J-1]-theoNum[i-1])<Math.abs(tranNum[J]-theoNum[i-1]))//译文J-1比译文J更靠近理论值i-1
					{
					return(i)
					}
				else//译文J比译文J-1更靠近理论值i-1
					{
					return(i-1)
					}
				}
			}
		}
	}
	
}
						var temtrancom=""
						for(var k=0;k<iNextPar-currPar-1;k++)
						{
						var nextPos=tranPos(k+1)
						var currPos=tranPos(k)
						if(k==iNextPar-currPar-2)	
							{
							temtrancom=temtrancom+strTranList[k+currPar+1].text
							newTran.push(temtrancom)
							temtrancom=""
							}
						else
						{
							if(nextPos>currPos)
								{
								if(nextPos-currPos>1)
									{
									for(var x=1;x<nextPos-currPos;x++)
										{
										newTran.push(" ")
										}
									}
								else
									{}
								temtrancom=temtrancom+strTranList[k+currPar+1].text
								newTran.push(temtrancom)
								temtrancom=""
									
									
								}
							else
								{
								temtrancom=temtrancom+strTranList[k+currPar+1].text
								}
						
						}
						//document.getElementById("info").innerHTML=document.getElementById("info").innerHTML+tranNum[k]+" "+tranPos(k)+" "+theoNum[tranPos(k)]+" "+(k+currPar+1)+"<br>"
						}

										
					}
					
				}
				currPar=iNextPar
				currId=strTranList[currPar].id
			}
			newTran.push(strTranList[strTranList.length-1].text)
			
			var output=""
			for(var i=0;i<newTran.length;i++){
				output+=(i+1)+"\t"+newTran[i]+"\n"
			}
			document.getElementById("output_final").value=output
			//output=output.replace(/#/g,"");


		}
	
		</script>
</head>
<body>
<h3>标点替换表</h3>
<input type="input" id="biaodian" value="。$？$；$：$:$’$”" />
<h3>标点不替换表</h3>
<input type="input" id="biaodian_no" value="’$”" />
<h3>巴利原文</h3>
从“XXX_Pali.txt”文件中复制纯文本黏贴于此<br>
<textarea id="input_pali" rows="15" cols="100">
</textarea>
<h3>添加译文</h3>
从xls文件中复制“行号”“内容”两列内容黏贴于此<br>
<textarea id="output_translate"  rows="15" cols="100"></textarea><br>
<button type="button" onclick="do_final()">批量处理</button><br>
<h3>对读结果</h3>
可在文本编辑器中进行中巴对读查看结果<br>
<textarea id="output_final"  rows="15" cols="100"></textarea><br>
<div id="test">
</div>
<div id="info">
</div>
</body>
</html>