我试图用node.js从pile语料库中正确读写100万行，问题是我得到了一个有100万行的输出文件，但第100万行是原始文件的900 kish行。
这是我使用的：

fs =require("fs");

var buffer = '';
var rs1 = fs.createReadStream('00.jsonl',"ascii");
var ws = fs.createWriteStream('pile_1M.txt', {flags: 'w',encoding: 'ascii'});
var lines_num=0;

var max_read=1000000;

rs1.on('data', function(chunk) {
    rs1.pause()
  
    var lines = (buffer+chunk).split(/\r?\n/g);
  
    for (var i = 0; i < lines.length && lines_num<max_read; ++i) {
        lines_num++;     
        ws.write(lines[i]+"\n");}
    }
         
    if(lines_num>=max_read){
        rs1.close()
        console.log("close")    
    } 
    
    rs1.resume(); 
    
});

rs1.on('close', function () {
    ws.close();       
});

编辑：
感谢@jfriend00，我提出了以下工作解决方案：

fs =require("fs");

var buffer = '';
var rs1 = fs.createReadStream('/media/user/hdd2/pile/00.jsonl',"utf-8");
var ws = fs.createWriteStream('pile_1M_2.txt', {flags: 'w',encoding: 'utf-8'});
var lines_num=0;

var max_read=100000;

function json_line_to_text(line){
var index=line.indexOf(', "meta": {"');
var res=line.substring(10,index-1);
if(index==-1){
    res=line;
}
return res;
}

rs1.on('data', function(chunk) {
  rs1.pause();
  
  var lines = (buffer+chunk).split(/\r?\n/g);
  
    for (var i = 0; i < lines.length && lines_num < max_read; ++i) {

        if (i == lines.length - 1) {
            buffer = lines[i];
        } else {
            
            if (lines_num < max_read) {
                ws.write(json_line_to_text(lines[i]) + "\n");
                lines_num++; 
            }
        }

    }  
      
    if(lines_num>=max_read){
      rs1.close()
      console.log("close")    
    } 
    
    rs1.resume(); 
    
});

rs1.on('close', function () {
    ws.close();       
});

**我包含了jsonline_to_text函数，我最初认为这是问题的真正原因，但事实并非如此。