var xhr = (function () {
    var aXMLHttpVers = ['MSXML2.XMLHTTP.6.0', 'MSXML2.XMLHTTP.3.0', 'MSXML2.XMLHTTP', 'Microsoft.XMLHTTP'];
    for (var i = 0; i < aXMLHttpVers.length; i++) {
        try {
            return new ActiveXObject(aXMLHttpVers[i]);
        } catch (error) { }
    }
    showError('Can\'t build XMLHTTP automation object.');
    WScript.Quit(1);
})();
var fso = new ActiveXObject('Scripting.FileSystemObject'),
    shell = new ActiveXObject('WScript.Shell'),
    curDir,
    url = 'http://www.chinapoesy.com/',
    aTSIndex = ['TangShiAllIndex2.html', 'TangShiAllIndex1.html', 'TangShiAllIndex.html'],
    aTSUrl = [],
    arr,
    outFile = 'tangshi.txt',//ÌÆÊ«Îı¾
    reTSIndex = /<a[^>]+href=(['"])(.+?)\1[^>]*>\s*µÚ\d+¾í\s*<\/a>/gi,
    rePoem = /<li[^>]+class="LiTitle"[^>]*>\s*<a[^>]+class="Green"[^>]*>\s*(.+?)\s*<\/a>[\S\s]+?<\/li>\s*<li[^>]+class="LiContent[^"]*"[^>]*>\s*<b>(.+?)<\/b>([\S\s]+?)<\/li>/gi,
    rePB = /<[^>]*>|\r|\n/g,
    reNextPage = /<a[^>]+href="\/([^"]+)"[^>]*>\s*<img\s+src="\/Images\/Pager\/nextn.gif"\s+border="0"[^>]*>\s*<\/a>/i,
    aFields = ['×÷Õß', '±êÌâ', 'ÕýÎÄ'],
    sSplit = '\t',
    ts,
    sHtml,
    i,
    l,
    timer;

timer = new Date();
curDir = shell.CurrentDirectory = fso.GetParentFolderName(WScript.ScriptFullName);
try {
    ts = fso.OpenTextFile(outFile, 2, true);
} catch (err) {
    showError(err, 'Writing to ' + outFile);
    WScript.Quit(2);
}

for (i = aTSIndex.length; i >= 0; i--) {
    sHtml = getHtml(url + aTSIndex[i]);
    while (arr = reTSIndex.exec(sHtml)) {
        aTSUrl.push(arr[2]);
    }
}

ts.WriteLine(aFields.join(sSplit));
for (i = 0, l = aTSUrl.length; i < l; i++) {
    writePoem(url + aTSUrl[i]);
    //next page
    while (arr = reNextPage.exec(sHtml)) {
        writePoem(url + arr[1]);
    }
}
ts.Close();
WScript.Echo('Mission complete.\nTime elapsed: ' + (new Date() - timer) / 1000 + 's');
WScript.Quit();

function getHtml(URL) {
    xhr.open('GET', URL, false);
    xhr.send();
    if (200 === xhr.status) {
        return xhr.responseText;
    }
    showError('fetch URI "' + URL + '" failed.\nstatus: ' + xhr.status);
    return '';
}
function writePoem(URL) {
    sHtml = getHtml(URL);
    while (arr = rePoem.exec(sHtml)) {
        ts.WriteLine([arr[1], arr[2], arr[3].replace(rePB, '')].join(sSplit));
    }
}
function showError(err, source) {
    WScript.Echo('[object Error]' === Object.prototype.toString.call(err) ?
        [
            err.name,
            'source: ' + (undefined === source ? '' : source),
            'number: ' + (err.number >>> 0).toString(16),
            'equipment: ' + (err.number >> 16 & 0x1FFF),
            'code: ' + (err.number & 0xFFFF),
            'Information: ' + err.message
        ].join('\n')
        :
        err);
}
¸´ÖÆ´úÂë

var fso = new ActiveXObject('Scripting.FileSystemObject');
var http = new ActiveXObject('Microsoft.XMLHTTP');

//var map = {'ÔªÇú':'YuanQu','Ê«¾­':'ShiJing','³þ´Ç':'ChuCi','ÀÖ¸®Ê«¼¯':'YueFu','ÆäËü¹Å´ú':'Other','½üÏÖ´úÊ«':'XianDai','Íâ¹úÊ«´Ê':'ForeignPoesy'};
var map = {'Ê«¾­':'ShiJing','³þ´Ç':'ChuCi', '½üÏÖ´úÊ«':'XianDai'};
var home = 'http://www.chinapoesy.com/';

for (var key in map) {
    var n = 0;
    var url = home + map[key] + 'Index' + (key == 'Íâ¹úÊ«´Ê' ? '.aspx' : '.html');
    var reg = /href='(.+?\.html)'\r?\n\s*target="_blank"/g;
    while(url != '') {
        var txt = getText(url);
        if(!n++) {
            var m1 = txt.match(/\(°´ÕÕÈËÆøÅÅÃû\).*?href='(.*?\.html)'/);  //°´ÕÕÈËÆøÅÅÃûÓÐ "¸ü¶à..." £¿
            txt = m1 ? getText(home + m1[1]) : txt.match(/\(°´ÕÕÈËÆøÅÅÃû\)[\s\S]+?<\/table>/)[0];
        }
        var m2 = txt.match(/href="\/([^"]*\.html)"[^<>]*><img src="\/Images\/Pager\/nextn\.gif"/); //ÓÐ "ÏÂÒ»Ò³" £¿
        url = m2 ? home + m2[1] : '';
        txt = m2 ? txt.match(/>²éѯÖÐ ÇëÉÔºó¡­[\s\S]+?<\/table>/)[0] : txt;
        while(arr=reg.exec(txt)) getPoesy(home + arr[1]);
    }
}

//»ñÈ¡ÍøÒ³ÄÚÈÝ
function getText(url) {
    http.open('GET', url, false);
    http.send();
    with(new ActiveXObject('ADODB.Stream')){
        Mode = 3; Type = 1;
        Open();
        Write(http.responseBody);
        Position = 0;
        Type = 2;
        Charset = 'UTF-8';
        var s = ReadText(-1);
    }
    return s;
}

//»ñÈ¡Ê«´Ê
function getPoesy(url) {
    var reg = /href='(.+?\.html)'\s*target\s*=\s*"_blank"([\s\S]+?\(\d+\))/g;
    while(url != '') {
        var txt = getText(url);
        var m = txt.match(/href="\/([^"]*\.html)"[^<>]*><img src="\/Images\/Pager\/nextn\.gif"/);  //ÓÐ "ÏÂÒ»Ò³" £¿
        url = m ? home + m[1] : '';
        txt = txt.match(/>±à¼­Ëü<\/a>[\s\S]+?>¹ØÓÚÎÒÃÇ<\/a>/);
        while(arr=reg.exec(txt)) {
            var s = getText(home + arr[1]);                                  //¸è´ÊÄÚÈÝ
            s = s.split('("#loading").css("display","none");')[1];
            if(!s) return 0;        //ijЩÌض¨Ê«´ÊÍøÖ·´ò²»¿ª£¬Óöµ½ÕâÖÖÇé¿ö¼ÌÐø
            s = s.split('<script type="text/javascript">')[0];
            //ÖÐÎļ°±êµã×Ö·ûºóÃæ¼Ó <br />
            s = s.replace(/([\u4E00-\u9FFF¡££¿£¡£¬¡¢£º¡±£©¡·¡¿¡­ ¡ª])(?:\r?\n|<\/p>)/g, '$1<br />');
            //ɾ³ýËùÓбêÇ©£¬±£Áô»»ÐбêÇ©£¬É¾³ý¿Õ¸ñºÍhtmlתÒå¿Õ¸ñ
            s = s.replace(/<(?!\/?br)[^>]*>|&nbsp;|[ 	¡¡£ ]+/g, '');
            s = s.replace(/<\/?br>|<br\/>/g, '<br />');   //<br>¡¢</br>¡¢<br/> Ì滻Ϊ <br />
            s = s.replace(/\r?\n/g, '');
            s = s.replace(/^(?:<br \/>)+/, '');           //ɾ³ý¿ªÍ·¶àÓàµÄ±êÇ©
            s = s.replace(/(<br \/>)+$/, '$1');           //ɾ³ý½áβ¶àÓàµÄ±êÇ©
            s = s.replace(/&middot;/g, '¡¤').replace(/&ldquo;/g, '¡°').replace(/&rdquo;/g, '¡±');
            s = s.replace(/&mdash;/g, '¡ª').replace(/&quot;/g, '"');
            s = s.replace(/(?:<br \/>)+/, '\t');          //µÚÒ»´ÎÁ¬Ðø¶à¸ö»»ÐбêÇ©Ì滻Ϊ Tab
            s = s.replace(/(<br \/>)+/, '$1');            //µÚ¶þ´ÎÁ¬Ðø¶à¸ö»»ÐбêÇ©Ì滻Ϊһ¸ö
            var s1 = arr[2].replace(/[ 	¡¡£ \r\n]+/g, '').replace(/.*>/, '');  //±êÌâ+ÈËÆøÖ¸Êý
            writeToFile(s1 + '\t' + s);
        }
    }
}

//дÈëÎı¾
function writeToFile(str) {
    var objFile = fso.OpenTextFile(key + '.Log', 8, true);
    objFile.WriteLine(str);
    objFile.Close();
}

WSH.Echo('Done')
¸´ÖÆ´úÂë

s = s.replace(/<(?!\/?br)[^>]*>|&nbsp;|[ 	¡¡£ ]+/g, '').replace(/<\/?br>|<br\/>/g, '<br />');
¸´ÖÆ´úÂë

var fso  = new ActiveXObject('Scripting.FileSystemObject');
var http = new ActiveXObject('Microsoft.XMLHTTP');
var home = 'http://www.chinapoesy.com/';

for (var i=0; i<=2; i++) {
    var map = [], map_ = [];
    var url = home + 'TangshiAllindex' + i + '.html';
    var txt = getText(url);
    txt = txt.split('class="Normal">Ͼí</a>')[1];
    var reg = /href='(.+?\.html)'\r?\n\s*target="_blank"/g;

    while(arr = reg.exec(txt)) {
        url = home + arr[1];
        while(url != '') {
            var s = getText(url);           //¸è´ÊÒ³Ãæ
            var m = s.match(/href="\/([^"]*\.html)"[^<>]*><img src="\/Images\/Pager\/nextn\.gif"/);//ÓÐ "ÏÂÒ»Ò³" £¿
            url = m ? home + m[1] : '';
            s = s.split('<!-- AspNetPager V6.0.0 for VS2005 End -->')[1];
            if(!s) continue;
            var a = s.split(/<\/li>\r?\n\s*<\/ul>\r?\n\s*\<\/div>/);
            for(var j=0; j<a.length-1; j++) {
                var ti = a[j].match(/<b>(.*?)<\/b>/)[1];                         //Ê«´Ê±êÌâ
                var au = a[j].match(/class="Green">\r?\n\s*([^\r\n]*)/)[1];      //×÷ÕßÃû
                var li = a[j].match(/class="LiTitle"><a href='(.*?\.html)'/)[1]; //±êÌâÁ¬½Ó
                if(!map_[au]) {
                    getPopularity(home + li);
                    map_[au] = 1;
                }
                writeToFile(getPoesy(a[j]));
            }
        }
    }
}

//»ñÈ¡ÍøÒ³ÄÚÈÝ
function getText(url) {
    http.open('GET', url, false);
    http.send();
    with(new ActiveXObject('ADODB.Stream')){
        Mode = 3; Type = 1;
        Open();
        Write(http.responseBody);
        Position = 0;
        Type = 2;
        Charset = 'UTF-8';
        var s = ReadText(-1);
    }
    return s;
}

//»ñÈ¡ÈËÆøÖ¸Êý
function getPopularity(url) {
    while(url != '') {
        var s= getText(url);
        var m = s.match(/href="\/([^"]*\.html)"[^<>]*><img src="\/Images\/Pager\/nextn\.gif"/); //ÓÐ "ÏÂÒ»Ò³" £¿
        url = m ? home + m[1] : '';
        s = s.split('<!-- AspNetPager V6.0.0 for VS2005 End -->')[1];
        if(!s) return 0;
        var reg = /title="([^"]*)\)">[\s\S]+?(\(\d*\))/g;
        while(arr = reg.exec(s)) map[au + arr[1]] = arr[2];
    }
}

//»ñÈ¡Ê«´Ê
function getPoesy(str) {
    var s = str.replace(/>ÆÀÂÛ<\/a>/, '>');
    s = s.replace(/([\u4E00-\u9FFF¡££¿£¡£¬¡¢£º¡±£©¡·¡¿¡­ ¡ª])(?:\r?\n|<\/p>)/g, '$1<br />');
    s = s.replace(/<(?!\/?br)[^>]*>|&nbsp;|[ 	¡¡£ ]+/g, '');
    s = s.replace(/<\/?br>|<br\/>/g, '<br />');
    s = s.replace(/\r?\n/g, '');
    s = s.replace(/^(?:<br \/>)+/, '');
    s = s.replace(/(<br \/>)+$/, '$1');
    s = s.replace(/(?:<br \/>)+/, '\t');
    s = s.replace(/(<br \/>)+/, '$1');
    return ti + map[au+ti] + '\t' + s;
}

//дÈëÎı¾
function writeToFile(str) {
    var objFile = fso.OpenTextFile('È«ÌÆÊ«.Log', 8, true);
    objFile.WriteLine(str);
    objFile.Close();
}

WSH.Echo('Done')
¸´ÖÆ´úÂë

var fso = new ActiveXObject('Scripting.FileSystemObject');
var http = new ActiveXObject('Microsoft.XMLHTTP');

var home = 'http://www.chinapoesy.com/';
var url = home + 'SongCiAllIndex_1.html';
var reg = /href='(.+?\.html)'\r?\n\s*target="_blank"/g;

while(url != '') {
    var txt = getText(url);
    var m = txt.match(/href="\/([^"]*\.html)"[^<>]*><img src="\/Images\/Pager\/nextn\.gif"/); //ÓÐ "ÏÂÒ»Ò³" £¿
    url = m ? home + m[1] : '';
    txt = txt.match(/>²éѯÖÐ ÇëÉÔºó¡­[\s\S]+?<\/table>/)[0];
    while(arr = reg.exec(txt)) getPoesy(home + arr[1]);
}

//»ñÈ¡ÍøÒ³ÄÚÈÝ
function getText(url) {
    http.open('GET', url, false);
    http.send();
    with(new ActiveXObject('ADODB.Stream')){
        Mode = 3; Type = 1;
        Open();
        Write(http.responseBody);
        Position = 0;
        Type = 2;
        Charset = 'UTF-8';
        var s = ReadText(-1);
    }
    return s;
}

//»ñÈ¡Ê«´Ê
function getPoesy(url) {
    var reg = /href='(.+?\.html)'\s*target\s*=\s*"_blank"([\s\S]+?)(\(\d+\))/g;
    while(url != '') {
        var txt = getText(url);
        var m = txt.match(/href="\/([^"]*\.html)"[^<>]*><img src="\/Images\/Pager\/nextn\.gif"/);  //ÓÐ "ÏÂÒ»Ò³" £¿
        url = m ? home + m[1] : '';
        txt = txt.match(/>±à¼­Ëü<\/a>[\s\S]+?>¹ØÓÚÎÒÃÇ<\/a>/);
        while(arr = reg.exec(txt)) {
            var s = getText(home + arr[1]);                         //¸è´ÊÄÚÈÝ
            s = s.split('("#loading").css("display","none");')[1];
            if(!s) return 0;
            s = s.split('<script type="text/javascript">')[0];
            s = s.replace(/([\u4E00-\u9FFF¡££¿£¡£¬¡¢£º¡±£©¡·¡¿¡­ ¡ª])(?:\r?\n|<\/p>)/g, '$1<br />');
            s = s.replace(/<(?!\/?br)[^>]*>|&nbsp;|[ 	¡¡£ ]+/g, '');
            s = s.replace(/<\/?br>|<br\/>/g, '<br />');
            s = s.replace(/\r?\n/g, '');
            s = s.replace(/^(?:<br \/>)+/, '');
            s = s.replace(/(<br \/>)+$/, '$1');
            s = s.replace(/(?:<br \/>)+/, '\t');
            s = s.replace(/(<br \/>)+/, '$1'); 
            var s1 = arr[2].match(/title\s*=\s*"([^"]*?)\)"/)[1];  //±êÌâ
            var s2 = arr[3];                                       //ÈËÆøÖ¸Êý
            writeToFile(s1 + s2 + '\t' + s);
        }
    }
}

//дÈëÎı¾
function writeToFile(str) {
    var objFile = fso.OpenTextFile('È«ËδÊ.Log', 8, true);
    objFile.WriteLine(str);
    objFile.Close();
}

WSH.Echo('Done')
¸´ÖÆ´úÂë

(function(e,t){function i(e,t,i,r,n){this.level=e,this.reChild=t,this.a2=r,this.next=n,this.tableText="",this.tableID=i}function r(e,t,i){var n,a,h,u=s(e);if(null===t.next)return n=t.getContent(u),void o.WriteLine([n[0],i[2],i[1],n[1]].join(f));for("author"===t.level&&(d[i[0]]&&(n=RegExp("<a[^>]+href=[\"']([^\"']+)[\"'][^>]*>\\s*"+l(d[i[0]])+"\\s*</a>","i").exec(u))&&(u=s(x+n[1])),a=T[i[0]]);;){for(;n=t.getChildren(u,a);){if(i.push("author"===t.level?"×÷Õߣº"+n[1]:n[1]),"title"===t.level){h=n[1]+".txt";try{o=p.OpenTextFile(h,2,!0)}catch(e){c(e,"Opening file "+h);continue}}r(x+n[0],t.next,i),i.pop(),"title"===t.level&&o.Close()}if(!(n=b.exec(u)))break;u=s(x+n[1])}}function n(e){var t,i,n,l,a=["TangShiAllIndex2.html","TangShiAllIndex1.html","TangShiAllIndex.html"],h=/<a[^>]+href=(['"])(.+?)\1[^>]*>\s*µÚ\d+¾í\s*<\/a>/gi,u=/<li[^>]+class="LiTitle"[^>]*>\s*<a[^>]+href=["']([^"']+)["'][^>]+?class="Green"[^>]*>\s*(.+?)\s*<\/a>[\S\s]+?<\/li>/gi,g={};try{o=p.OpenTextFile(e,2,!0)}catch(t){return c(t,"Writing to "+e),!1}for(l=a.length-1;l>=0;l--)for(t=s(x+a[l]);n=h.exec(t);)for(i=s(x+n[2]);;){for(;n=u.exec(i);)g[n[1]]||(r(x+n[1],oArc.next.next,["È«ÌÆÊ«","×÷Õߣº"+n[2]]),g[n[1]]=!0);if(!(n=b.exec(i)))break;i=s(x+n[1])}o.Close()}function l(e){return e.replace(/[()]/g,"\\$&")}function a(e){return e.replace(/^\s+|\s+$|/g,"")}function s(e){return u.open("GET",e,!1),u.send(),200===u.status?u.responseText:(c('fetch URI "'+e+'" failed.\nstatus: '+u.status),"")}function c(e,t){WScript.Echo("[object Error]"===Object.prototype.toString.call(e)?[e.name,"source: "+(void 0===t?"":t),"number: "+(e.number>>>0).toString(16),"equipment: "+(e.number>>16&8191),"code: "+(65535&e.number),"Information: "+e.message].join("\n"):e)}var o,h,u=function(){for(var e=["MSXML2.XMLHTTP.6.0","MSXML2.XMLHTTP.3.0","MSXML2.XMLHTTP","Microsoft.XMLHTTP"],t=0;t<e.length;t++)try{return new ActiveXObject(e[t])}catch(e){}c("Can't build XMLHTTP automation object."),WScript.Quit(1)}(),p=new ActiveXObject("Scripting.FileSystemObject"),g=new ActiveXObject("WScript.Shell"),x="http://www.chinapoesy.com/",b=/<a[^>]+href="\/([^"]+)"[^>]*>\s*<img\s+src="\/Images\/Pager\/nextn.gif"\s+border="0"[^>]*>\s*<\/a>/i,f="\t",d={"½üÏÖ´úÊ«":"½üÏÖ´úÊ«ÈËÈ«¼¯","ÌÆÊ«":"¸ü¶àÌÆÊ«Ê«ÈË","ËδÊ":"¸ü¶àËδʴÊÈË","ÔªÇú":"¸ü¶àÔªÇúÃû¼Ò","Íâ¹úÊ«¸è":"¸ü¶àÍâ¹úÊ«ÈË"},T={"ËδÊ":'id="DDlSongCi"',"ÔªÇú":'id="DDlSongCi"',"ÌÆÊ«":'id="DDlTang"'};i.prototype.getChildren=function(e){var t;return""===this.tableText&&(this.tableText=RegExp("<table[^>]+"+l(this.tableID)+"[^>]*>[\\S\\s]+?</table>","i").exec(e)),(t=this.reChild.exec(this.tableText))?[t[this.a2[0]],t[this.a2[1]]]:(this.tableText="",null)},oArc=new i("content",null,"",null,null),oArc.getContent=function(e,t,i){var r;r=e.split(/<script\s+type="text\/javascript"[^>]*>\s*\$\("#loading"\)\.css\("display","none"\);\s*<\/script>|<(\w+)[^>]+id="Author\d*"[^>]*>[\S\s]*?<\/\1>|<script\s+type="text\/javascript">\s*<!--\s*google_ad_client[^<>]+?google_ad_slot[^<>]+?-->\s*<\/script>/gi);try{return[a(r[1].replace(/(?:<[^<>]*>|&nbsp;)+/g,"")).replace(/[\r\n]/g," "),r[2].replace(/(?:<[^<>]*>|&nbsp;)+/g,function(e){return/(?:<br\s*\/?>|<\/br>)+/i.test(e)?"<br />":""}).replace(/(?:<br \/>|\s)+/gi,"<br />")]}catch(e){return[]}},oArc=new i("article",/<a[^>]+href=["']([^"']+)["'][^>]*>([^<>]+)<\/a>/gi,'id="DLFeelingChina"',[1,2],oArc),oArc.getChildren=function(e,t){var i;return""===this.tableText&&(this.tableText=RegExp("<table[^>]+"+l(t||this.tableID)+"[^>]*>[\\S\\s]+?</table>","i").exec(e)),(i=this.reChild.exec(this.tableText))?[i[this.a2[0]],/\((\d+)\)/.exec(i[this.a2[1]])[1]]:(this.tableText="",null)},oArc=new i("author",/<a[^>]+href=["']([^"']+)["'][^>]*>([^<>]+)<\/a>/gi,'id="DLAtuhor"',[1,2],oArc),oArc.getChildren=function(e,t){var i;return""===this.tableText&&(this.tableText=RegExp("<table[^>]+"+l(t||this.tableID)+"[^>]*>[\\S\\s]+?</table>","i").exec(e)),(i=this.reChild.exec(this.tableText))?[i[this.a2[0]],i[this.a2[1]].replace(/\s+|\(\d*\)/g,"")]:(this.tableText="",null)},oArc=new i("title",RegExp("<a[^>]+href=[\"']/([^\"']+)[\"'][^>]*>("+t.join("|")+")</a>","gi"),'class="top"',[1,2],oArc),h=new Date,g.CurrentDirectory=p.GetParentFolderName(WScript.ScriptFullName),e&&n("È«ÌÆÊ«.txt"),r(x,oArc,[]),WScript.Echo("Mission complete.\nTime elapsed: "+(new Date-h)/1e3+"s"),WScript.Quit()})(!0,["ÌÆÊ«","ËδÊ","È«ËδÊ","ÔªÇú","Ê«¾­","³þ´Ç","ÀÖ¸®Ê«¼¯","½üÏÖ´úÊ«","Íâ¹úÊ«¸è"]);
//µÚÒ»¸ö²ÎÊýÊÇ·ñÏÂÔØÈ«ÌÆÊ«£¬1=ÊÇ£¬0=·ñ;
//µÚ¶þ¸ö²ÎÊýÊdzýÈ«ÌÆÊ«ÒÔÍâµÄÆäËûÀ¸Ä¿ÁÐ±í¡£
¸´ÖÆ´úÂë

blQT = true;
¸´ÖÆ´úÂë