详解nodejs爬虫程序解决gbk等中文编码问题

<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>

<meta charset=utf-8"/>

res.on('data', function(data) {

  htmlData.push(data);
  htmlDataLength += data.length;
 });
var bufferHtmlData = Buffer.concat(htmlData,htmlDataLength);

decodeHtmlData = iconv.decode(bufferHtmlData,'gbk');

var $ = cheerio.load(decodeHtmlData, {decodeEntities: false});

 $('title','head').each(function(i, e) {

  htmlHeadTitle = $(e).text();
  console.log(htmlHeadTitle);
 });

fromEncoding: function(buf) {
          buf = ensureBuffer(buf);
          var idx = 0, len = 0,
            newBuf = new Buffer(len*2),unicode,gbkcode;
          for (var i = 0, _len = buf.length; i < _len; i++, len++) {
            if (!!(buf[i] & 0x80)) {//the high bit is 1, so this byte is gbkcode's high byte.skip next byte
              i++;
            }
          }
          var newBuf = new Buffer(len*2);
          for (var i = 0, j = 0, _len = buf.length; i < _len; i++, j++) {
            var temp = buf[i], gbkcode, unicode;
            if (temp & 0x80) {
              gbkcode = (temp << 8) + buf[++i];
              unicode = table[gbkcode] || iconv.defaultCharUnicode.charCodeAt(0);//not found in table, replace with defaultCharUnicode
            }else {
              unicode = temp;
            }
            newBuf[j*2] = unicode & 0xFF;//low byte
            newBuf[j*2+1] = unicode >> 8;//high byte
          }
          return newBuf.toString('ucs2');
        }

var cheerio = require('cheerio');
var http = require('http');
var iconv = require('iconv-lite');
var htmlData = [];
var htmlDataLength = 0;
var count = 0;

http.globalAgent = 'Mozilla/5.0 (Windows NT 6.1; rv:2.0.1) Gecko/20100101 Firefox/4.0.1';
http.get('http://www.cr173.com', function(res) {

 res.on('data', function(data) {

  htmlData.push(data);
  htmlDataLength += data.length;
  count ++;
 });

 res.on('end',function(){

  callback(htmlData);
 });

});

function callback(htmlData){

 console.log(count);
 var bufferHtmlData = Buffer.concat(htmlData,htmlDataLength);
 var charset = '';
 var decodeHtmlData;
 var htmlHeadTitle = '';
 var htmlHeadCharset = '';
 var htmlHeadContent = '';
 var index = 0;

 var $ = cheerio.load(bufferHtmlData, {decodeEntities: false});

 $('meta','head').each(function(i, e) {

  htmlHeadCharset = $(e).attr('charset');
  htmlHeadContent = $(e).attr('content');

  if(typeof(htmlHeadCharset) != 'undefined'){

   charset = htmlHeadCharset;
  }

  if(typeof(htmlHeadContent) != 'undefined'){

   if(htmlHeadContent.match(/charset=/ig)){

    index = htmlHeadContent.indexOf('=');
    charset = htmlHeadContent.substring(index+1);
   }
  }
 });

 //此处为什么需要对整个网页进行转吗，是因为cheerio这个组件不能够返回buffer,iconv则无法转换之
 if(charset.match(/gb/ig)){

  decodeHtmlData = iconv.decode(bufferHtmlData,'gbk');
 }
 else{//因为有可能返回的网页中不存在charset字段，因此默认都是按照utf8进行处理

  decodeHtmlData = iconv.decode(bufferHtmlData,'utf8');
 }

 var $ = cheerio.load(decodeHtmlData, {decodeEntities: false});

 $('title','head').each(function(i, e) {

  htmlHeadTitle = $(e).text();
  console.log(htmlHeadTitle);
 });

 console.log(charset);

}