[文本处理] [已解决]批处理如何获取“言情后花园www.yqhhy.cc”的网页内容

Dim objIE, Url, Text
Url = "http://www.yqhhy.cc/54/54335/2868608.html"
Set objIE = CreateObject("InternetExplorer.Application")
objIE.Visible = False
objIE.Navigate Url
Do Until objIE.ReadyState = 4
  WScript.Sleep 200
Loop
Text = objIE.Document.GetElementById("content").InnerText
Set objIE = Nothing
Dim objREG
Set objREG = New RegExp
objREG.Global = True
objREG.IgnoreCase = True
objREG.Pattern = "[\s]+(.*)言情后花园（www.yqhhy.cc），提供精品言情小说阅读！[0-9a-z]+"
Text = objREG.Replace(Text, "$1")
Set objREG = Nothing
CreateObject("Scripting.FileSystemObject").OpenTextFile("a.txt",2,True).Write Text
WScript.Echo "ok"
复制代码

//&cls&cscript -nologo -e:jscript "%~f0"&pause&exit

//编码转换
function BintoStr(strBin,strCharset){
    try{
        var stream = new ActiveXObject('ADODB.Stream')
        stream.Type = 1
        stream.Mode = 3
        stream.Open()
        stream.Write(strBin)
        stream.Position = 0
        stream.Type = 2
        stream.Charset = strCharset
        return stream.ReadText
    }
    catch(e){}
}

//获取网页内容
function getHtmlTxt(url, strCharset){
    try{
        var http = new ActiveXObject('MSXML2.XMLHTTP')
    }
    catch(e){
        var http = new ActiveXObject('WinHttp.WinHttpRequest.5.1')
    }
    http.open('GET', url, false)
    http.send()
    return BintoStr(http.ResponseBody, strCharset);
}

//纯实现，可直接下载http://www.yqhhy.cc/modules/article/packdown12.php?id=54335&type=txt
//章节
var chapter=[2868608,2868609,2868610,2868611,2868612,2868613,2868614,2868615,2868616,2868617,2868618,2868619,2868620,2868621,2868622,2868623,2868624,2868625,2868626,2868627,2868628,2868629,2868630,2868631,2868632,2868633,2868634,2868635,2868636,2868637,2868638,2868639,2868640,2868641,2868642,2868643,2868644,2868645,2868646,2868647,2868648,2868649,2868650,2868651,2868652,2868653,2868654,2868655,2868656,2868657,2868658,2868659,2868660,2868661,2868662,2868663,2869257,2870837,2872851,2872852,2875074,2875340,2879312,2881646,2881647,2884297,2886886,2888072,2888073,2888797,2894304,2894305,2896700,2896701,2898723,2910028];
var fso = new ActiveXObject('Scripting.FileSystemObject');
for(var i=0;i<chapter.length;i++){
    var text = getHtmlTxt('http://www.yqhhy.cc/54/54335/'+chapter[i]+'.html', 'gbk').replace(/&nbsp;/g,' ').replace(/<br \/>|<br\/>|<br>/g,'\r\n');
    var m = text.match(/<h1\s?[^<]*?>[\s\S]+?<\/h1>|<div id=\"content\">[\s\S]+?<\/div>/ig);
    if(m){
        var title = m[0].replace(/<[^<]+>/g,'').replace(/^\s*|\s*$/g,'');
        var content = m[1].replace(/<[^<]+>/g,'').replace(/.*言情后花园(.*www.yqhhy.cc)?.*/ig,'').replace(/(\r?\n)+/g,'\r\n');
    }
    var f=fso.OpenTextFile('结果.txt', 8, true)
    f.Write(title+'\r\n'+content+'\r\n');
    f.Close();
    WSH.echo((i+1)+'、'+title);
}
复制代码