xpath 参考

using System;

using System.Collections.Generic;

using System.Linq;

using System.Web;

using System.Text.RegularExpressions;

using System.Configuration;

/// <summary>

////// </summary>

public static class SearchConst

{

    public static readonly string ARG_CLIENT = "client";

    public static readonly string ARG_WORD = "word";

    public static readonly int DataColumnCount = ;

    public static readonly int ColumnOfUrl = ;

    public static readonly int ColumnOfTitle = ;

    public static readonly int ColumnOfInfo = ;

    public static readonly int ColumnOfAdUrl = ;

    public static readonly string FMT_Date = "yyyy/MM/dd";

    public static readonly string FMT_TIME = "HH:mm:ss";

    public static readonly string UserAgentPC = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:11.0) Gecko/20100101 Firefox/11.0";

    public static readonly string UserAgentMobile = "Mozilla/5.0 (iPhone; CPU iPhone OS 6_0 like Mac OS X) AppleWebKit/536.26 (KHTML, like Gecko) Version/6.0 Mobile/10A403 Safari/8536.25";

    public static readonly string SearchKeyWordPlace = "#{q}";

    public static readonly string DefaultEncode = "UTF-8";

    public static readonly string AttributeHref = "href";

    public static readonly string FILEEXT_ZIP = ".zip";

    public static readonly string FILE_TXT = "source.txt";

    public static readonly string FILE_KEY = "SavePath";

    public static readonly string BATCH_PARALLES_KEY = "BatchParalles";

    public static readonly string FLG_ENABLED = "";

    public static readonly string CLIENT_MONITOR = "BJMOR";

    public static readonly string MSG_E_PAGE_STYLE_CHANGE = "fff";

    public static class Google

    {

        public static readonly string UserAgent = UserAgentPC;

        public static readonly string[] XPATH_ROOT = { "mbEnd", "mbEnd" };

        public static readonly string[] XPATH_CITE = { "//div[@id='mbEnd']//ol/li//cite", "//div[@id='mbEnd']//ol/li//cite" };  //获取url

        public static readonly string[] XPATH_H3 = { "//div[@id='mbEnd']//ol/li//h3", "//div[@id='mbEnd']//ol/li/h3" }; //  //获取标题

        public static readonly string[] XPATH_ADURL = { "//div[@id='mbEnd']//ol/li//h3//a[1]", "//div[@id='mbEnd']//ol/li/h3//a[1]" };

        public static readonly string[] XPATH_INFO = { "//div[@id='mbEnd']//ol/li//div[@class='ac ads-creative']", "//div[@id='mbEnd']//ol/li//div[@class='ads-creative']" };

        // top info

        public static readonly string[] XPATH_ROOT_TOP = { "taw", "taw" };

        public static readonly string[] XPATH_CITE_TOP = { "//div[@id='tads']//ol/li//cite", "//div[@id='tads']//ol/li//cite" };

        public static readonly string[] XPATH_H3_TOP = { "//div[@id='tads']//ol/li//h3", "//div[@id='tads']//ol/li/h3" };

        public static readonly string[] XPATH_ADURL_TOP = { "//div[@id='tads']//ol/li//h3//a[1]", "//div[@id='tads']//ol/li/h3//a[1]" };

        public static readonly string[] XPATH_INFO_TOP = { "//div[@id='tads']//ol/li//div[@class='ac ads-creative']", "//div[@id='tads']//ol/li//div[@class='ads-creative']" };

        //

        public static readonly Regex RegexAdUrl = new Regex(@"adurl=(http[\S]*$)");

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class GoogleM

    {

        public static readonly string UserAgent = UserAgentMobile;

        //info

        public static readonly string[] XPATH_ROOT = { "bottomads", "bottomads" };

        public static readonly string[] XPATH_CITE = { "//div[@id='tadsb']/ol/li//cite", "//div[@id='tadsb']/ol/li//cite" };

        public static readonly string[] XPATH_H3 = { "//div[@id='tadsb']/ol/li//h3", "//div[@id='tadsb']/ol/li//h3" };

        public static readonly string[] XPATH_ADURL = { "//div[@id='tadsb']/ol/li//h3//a", "//div[@id='tadsb']/ol/li//h3//a" };

        public static readonly string[] XPATH_INFO = { "//div[@id='tadsb']/ol/li//div[@class='ac ads-creative']", "//div[@id='tadsb']/ol/li//div[@class='ads-creative']" };

        // top info

        public static readonly string[] XPATH_ROOT_TOP = { "tads", "tads" };

        public static readonly string[] XPATH_CITE_TOP = { "//div[@id='tads']/ol/li//cite", "//div[@id='tads']/ol/li//cite" };

        public static readonly string[] XPATH_H3_TOP = { "//div[@id='tads']/ol/li//h3", "//div[@id='tads']/ol/li//h3" };

        public static readonly string[] XPATH_ADURL_TOP = { "//div[@id='tads']/ol/li//h3//a", "//div[@id='tads']/ol/li//h3//a" };

        public static readonly string[] XPATH_INFO_TOP = { "//div[@id='tads']/ol/li//div[@class='ac ads-creative']", "//div[@id='tads']/ol/li//div[@class='ads-creative']" };

        //

        public static readonly Regex RegexAdUrl = new Regex(@"adurl=(http[\S]*$)");

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class MSN

    {

        public static readonly string UserAgent = UserAgentPC;

        //b_context/b_ad

        public static readonly string[] XPATH_ROOT = { "sidebar", "b_context" };

        public static readonly string[] XPATH_CITE = { "//div[@class='sb_adsNv2']//li//cite", "//ol[@id='b_context']//li[@class='b_ad']//li//cite" };

        public static readonly string[] XPATH_H3 = { "//div[@class='sb_adsNv2']//li//h3", "//ol[@id='b_context']//li[@class='b_ad']//li//h2" };

        public static readonly string[] XPATH_ADURL = { "//div[@class='sb_adsNv2']//li//a", "//ol[@id='b_context']//li[@class='b_ad']//li//a" };

        public static readonly string[] XPATH_INFO = { "//div[@class='sb_adsNv2']//li//p", "//ol[@id='b_context']//li[@class='b_ad']//li//p" };

        //b_results/b_ad

        public static readonly string[] XPATH_ROOT_TOP = { "results_container", "b_results" };

        public static readonly string[] XPATH_CITE_TOP = { "//div[@class='sb_adsWv2']//li//cite", "//ol[@id='b_results']//li[@class='b_ad']//li//cite" };

        public static readonly string[] XPATH_H3_TOP = { "//div[@class='sb_adsWv2']//li//h3", "//ol[@id='b_results']//li[@class='b_ad']//li//h2" };

        public static readonly string[] XPATH_ADURL_TOP = { "//div[@class='sb_adsWv2']//li//a", "//ol[@id='b_results']//li[@class='b_ad']//li//a" };

        public static readonly string[] XPATH_INFO_TOP = { "//div[@class='sb_adsWv2']//li//p", "//ol[@id='b_results']//li[@class='b_ad']//li//p" };

        //

        public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)");

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class Yahoo

    {

        public static readonly string UserAgent = UserAgentPC;

        public static readonly string XPATH_ROOT = "sIn";

        public static readonly string XPATH_CITE1 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/div[@class='a cf']";

        public static readonly string XPATH_H31 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/h3";

        public static readonly string XPATH_ADURL1 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/h3/a";

        public static readonly string XPATH_INFO1 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/p";

        //

        public static readonly string XPATH_ROOT_TOP = "So1";

        public static readonly string XPATH_CITE_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/div[@class='a cf']";

        public static readonly string XPATH_H3_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/h3";

        public static readonly string XPATH_ADURL_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/h3/a";

        public static readonly string XPATH_INFO_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/p";

        //

        public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)");

        public static readonly string NullUrl = "&gt;";

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class Yahoo2

    {

        public static readonly string UserAgent = UserAgentPC;

        public static readonly string XPATH_ROOT_TOP = "contents";

        public static readonly string XPATH_CITE_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/cite";

        public static readonly string XPATH_H3_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/h2/a";

        public static readonly string XPATH_ADURL_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/h2/a";

        public static readonly string XPATH_INFO_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/p[@class='smr']";

        //

        public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)");

        public static readonly string NullUrl = "&gt;";

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class YahooM

    {

        public static readonly string UserAgent = UserAgentMobile;

        public static readonly string XPATH_ROOT = "contentsInner";

        public static readonly string XPATH_CITE = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/cite";

        public static readonly string XPATH_H3 = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/h3";

        public static readonly string XPATH_ADURL = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/h3/a";

        public static readonly string XPATH_INFO = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/p[@class='dtl']";

        public static readonly string XPATH_ROOT_TOP = "contentsInner";

        public static readonly string XPATH_CITE_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/cite";

        public static readonly string XPATH_H3_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/h3";

        public static readonly string XPATH_ADURL_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/h3/a";

        public static readonly string XPATH_INFO_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/p[@class='dtl']";

        //

        public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)");

        public static readonly string NullUrl = "&gt;";

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class BaiDu {

        public static readonly string UserAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko";

        public static readonly string[] XPATH_ROOT = { "ec_im_container", "ec_im_container" }; //第一种情况   。

        public static readonly string[] XPATH_CITE = { "//a/font[@size='-1' and @class]","//a/font[@size='-1' and @class]" }; //第一种情况

        public static readonly string[] XPATH_H3 = { "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]", "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]" };//第一种情况

        public static readonly string[] XPATH_ADURL = { "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]", "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]" };

        public static readonly string[] XPATH_INFO = { "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-click]/font[1]", "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-click]/font[1]" };

        // top info

        public static readonly string[] XPATH_ROOT_TOP = { "content_left", "content_left" };

        public static readonly string[] XPATH_CITE_TOP = { "//table[@data-click]/tbody/tr/td//a[not(@data-is-main-url) and not(contains(@href,'tool'))]/span", "//div[@class and @style]/div/div/a/span[1]|//div/table/tbody/tr/td[2]/div//a/span[1]" };   //前下后上

        public static readonly string[] XPATH_H3_TOP = { "//table/tbody/tr/td/a[ @data-is-main-url]", "//div[@class and @style]/div/div/h3" };    //前下后上

        public static readonly string[] XPATH_ADURL_TOP = { "//table/tbody/tr/td/a[ @data-is-main-url]", "//table/tbody/tr/td/a[ @data-is-main-url]" };    //前下后上

        public static readonly string[] XPATH_INFO_TOP = { "//table[@data-click and @class]/tbody/tr[3]/td/a[not(./span)]|//table[@data-click and @class]/tbody/tr/td/table/tbody/tr/td/div/font/a", "//div[@class and @style]/div/div[not(./span)]/a|//div/table/tbody/tr/td/div/font/a[not(./span)]" };         //前

        //

        //public static readonly Regex RegexAdUrl = new Regex(@"http[\S]*$");

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class CnBing {

        public static readonly string UserAgent = UserAgentPC;

        public static readonly string[] XPATH_ROOT = { "b_context", "b_context" };

        public static readonly string[] XPATH_CITE = { "//div[@class='sb_add sb_adTA']//cite", "//div[@class='sb_add sb_adTA']//cite" };

        public static readonly string[] XPATH_H3 = { "//div[@class='sb_add sb_adTA']//h2/a", "//div[@class='sb_add sb_adTA']//h2/a" };//第一种情况

        public static readonly string[] XPATH_ADURL = { "//div[@class='sb_add sb_adTA']//h2/a", "//div[@class='sb_add sb_adTA']//h2/a" };

        public static readonly string[] XPATH_INFO = { "//div[@class='sb_add sb_adTA']//div[@class='b_caption']/p", "//div[@class='sb_add sb_adTA']//div[@class='b_caption']/p" };

        // top info

        public static readonly string[] XPATH_ROOT_TOP = { "gg", "gg" };

        public static readonly string[] XPATH_CITE_TOP = { "", "" };   //前下后上

        public static readonly string[] XPATH_H3_TOP = { "", "" };    //前下后上

        public static readonly string[] XPATH_ADURL_TOP = { "", "" };    //前下后上

        public static readonly string[] XPATH_INFO_TOP = { "", "" };         //前下部分广告后上

        //

        public static readonly Regex RegexAdUrl = new Regex(@"rturl=(http[\S]*$)");

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class HaoSou {

        public static readonly string UserAgent = UserAgentPC;

            // 右边的广告

        public static readonly string[] XPATH_ROOT = { "side", "side" }; //获取范围

        public static readonly string[] XPATH_CITE = { "//ul[@id='rightbox']/li/p/cite[not(contains(text(),' http://e.360.cn'))]|//div[@id='m-spread-left']//cite", "//ul[@id='rightbox']/li/p/cite[not(contains(text(),' http://e.360.cn'))]|//div[@id='m-spread-left']//cite" }; //第一种情况

        public static readonly string[] XPATH_H3 = { "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'ss'))]|//div[@id='m-spread-left']//h3/a", "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'ss'))]|//div[@id='m-spread-left']//h3/a" };//第一种情况

        public static readonly string[] XPATH_ADURL = { "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'ss'))]|//div[@id='m-spread-left']//h3/a", "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'ss'))]|//div[@id='m-spread-left']//h3/a" };

        public static readonly string[] XPATH_INFO = { "//ul[@id='e_idea_pp']/li//p|//ul[@id='rightbox']/li/p[not(contains(text(),'4000-360-360'))]", "//ul[@id='e_idea_pp']/li//p|//ul[@id='rightbox']/li/p[not(contains(text(),'4000-360-360'))]" };

        // top info

        public static readonly string[] XPATH_ROOT_TOP = {"ss", "sss" };

        public static readonly string[] XPATH_CITE_TOP = { "", "" };   //前下后上

        public static readonly string[] XPATH_H3_TOP = { "", "" };    //前下后上

        public static readonly string[] XPATH_ADURL_TOP = { "", "" };    //前下后上

        public static readonly string[] XPATH_INFO_TOP = { "", "" };         //前下部分广告后上

        //

        //public static readonly Regex RegexAdUrl = new Regex(@"http[\S]*$");

        //

        public static readonly string BAITAI_ID = "";

    }

    public static class Sogou {

        public static readonly string UserAgent = UserAgentPC;

        //right 部分

        public static readonly string[] XPATH_ROOT = { "right" };

        public static readonly string[] XPATH_CITE = { "//div[@class='bizr_fb']" };//绿色的url

        public static readonly string[] XPATH_H3 = { "//h3[@class='bizr_title']" };//#ad_leftresult_0 > h3:nth-child(1)

        public static readonly string[] XPATH_ADURL = { "//h3[@class='bizr_title']/a" };//.h3的url

        public static readonly string[] XPATH_INFO = { "//div[@class='bizr_ft']" };

        //top 部分

        public static readonly string[] XPATH_ROOT_TOP = { "promotion_adv_container" };//*[@id="promotion_adv_container"]/div/div

        public static readonly string[] XPATH_CITE_TOP = { "//div[contains(@class,'biz_rb')and @id]/div//cite" };

        public static readonly string[] XPATH_H3_TOP = { "//h3[@class='biz_title']" };

        public static readonly string[] XPATH_ADURL_TOP = { "//h3[@class='biz_title']/a" };

        public static readonly string[] XPATH_INFO_TOP = { "//div[@class='crown_info_box' or @class='biz_ft']|//div[contains(@id,'box_id')]/table" };//   ""

        //

        //public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)");

        //

        public static readonly string BAITAI_ID = "";

        public static readonly string NullUrl = "&gt;";

    }

}

using System;using System.Collections.Generic;using System.Linq;using System.Web;using System.Text.RegularExpressions;using System.Configuration;
/// <summary>/// SearchHelper の概要の説明です/// </summary>public static class SearchConst{
public static readonly string ARG_CLIENT = "client";
public static readonly string ARG_WORD = "word";
public static readonly int DataColumnCount = 4;
public static readonly int ColumnOfUrl = 0;
public static readonly int ColumnOfTitle = 1;
public static readonly int ColumnOfInfo = 2;
public static readonly int ColumnOfAdUrl = 3;
public static readonly string FMT_Date = "yyyy/MM/dd";
public static readonly string FMT_TIME = "HH:mm:ss";
public static readonly string UserAgentPC = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:11.0) Gecko/20100101 Firefox/11.0";
public static readonly string UserAgentMobile = "Mozilla/5.0 (iPhone; CPU iPhone OS 6_0 like Mac OS X) AppleWebKit/536.26 (KHTML, like Gecko) Version/6.0 Mobile/10A403 Safari/8536.25";
public static readonly string SearchKeyWordPlace = "#{q}";
public static readonly string DefaultEncode = "UTF-8";
public static readonly string AttributeHref = "href";
public static readonly string FILEEXT_ZIP = ".zip";
public static readonly string FILE_TXT = "source.txt";
public static readonly string FILE_KEY = "SavePath";
public static readonly string BATCH_PARALLES_KEY = "BatchParalles";
public static readonly string FLG_ENABLED = "1";
public static readonly string CLIENT_MONITOR = "BJMOR";
public static readonly string MSG_E_PAGE_STYLE_CHANGE = "スポンサーチェックの検索媒体レイアウト変更";

public static class Google {
public static readonly string UserAgent = UserAgentPC;
public static readonly string[] XPATH_ROOT = { "mbEnd", "mbEnd" }; public static readonly string[] XPATH_CITE = { "//div[@id='mbEnd']//ol/li//cite", "//div[@id='mbEnd']//ol/li//cite" }; //获取url public static readonly string[] XPATH_H3 = { "//div[@id='mbEnd']//ol/li//h3", "//div[@id='mbEnd']//ol/li/h3" }; // //获取标题 public static readonly string[] XPATH_ADURL = { "//div[@id='mbEnd']//ol/li//h3//a[1]", "//div[@id='mbEnd']//ol/li/h3//a[1]" }; public static readonly string[] XPATH_INFO = { "//div[@id='mbEnd']//ol/li//div[@class='ac ads-creative']", "//div[@id='mbEnd']//ol/li//div[@class='ads-creative']" }; // top info public static readonly string[] XPATH_ROOT_TOP = { "taw", "taw" }; public static readonly string[] XPATH_CITE_TOP = { "//div[@id='tads']//ol/li//cite", "//div[@id='tads']//ol/li//cite" }; public static readonly string[] XPATH_H3_TOP = { "//div[@id='tads']//ol/li//h3", "//div[@id='tads']//ol/li/h3" }; public static readonly string[] XPATH_ADURL_TOP = { "//div[@id='tads']//ol/li//h3//a[1]", "//div[@id='tads']//ol/li/h3//a[1]" }; public static readonly string[] XPATH_INFO_TOP = { "//div[@id='tads']//ol/li//div[@class='ac ads-creative']", "//div[@id='tads']//ol/li//div[@class='ads-creative']" }; // public static readonly Regex RegexAdUrl = new Regex(@"adurl=(http[\S]*$)"); // public static readonly string BAITAI_ID = "001"; }
public static class GoogleM { public static readonly string UserAgent = UserAgentMobile;
//info public static readonly string[] XPATH_ROOT = { "bottomads", "bottomads" }; public static readonly string[] XPATH_CITE = { "//div[@id='tadsb']/ol/li//cite", "//div[@id='tadsb']/ol/li//cite" }; public static readonly string[] XPATH_H3 = { "//div[@id='tadsb']/ol/li//h3", "//div[@id='tadsb']/ol/li//h3" }; public static readonly string[] XPATH_ADURL = { "//div[@id='tadsb']/ol/li//h3//a", "//div[@id='tadsb']/ol/li//h3//a" }; public static readonly string[] XPATH_INFO = { "//div[@id='tadsb']/ol/li//div[@class='ac ads-creative']", "//div[@id='tadsb']/ol/li//div[@class='ads-creative']" };
// top info public static readonly string[] XPATH_ROOT_TOP = { "tads", "tads" }; public static readonly string[] XPATH_CITE_TOP = { "//div[@id='tads']/ol/li//cite", "//div[@id='tads']/ol/li//cite" }; public static readonly string[] XPATH_H3_TOP = { "//div[@id='tads']/ol/li//h3", "//div[@id='tads']/ol/li//h3" }; public static readonly string[] XPATH_ADURL_TOP = { "//div[@id='tads']/ol/li//h3//a", "//div[@id='tads']/ol/li//h3//a" }; public static readonly string[] XPATH_INFO_TOP = { "//div[@id='tads']/ol/li//div[@class='ac ads-creative']", "//div[@id='tads']/ol/li//div[@class='ads-creative']" }; // public static readonly Regex RegexAdUrl = new Regex(@"adurl=(http[\S]*$)"); // public static readonly string BAITAI_ID = "005"; }
public static class MSN { public static readonly string UserAgent = UserAgentPC; //b_context/b_ad public static readonly string[] XPATH_ROOT = { "sidebar", "b_context" }; public static readonly string[] XPATH_CITE = { "//div[@class='sb_adsNv2']//li//cite", "//ol[@id='b_context']//li[@class='b_ad']//li//cite" }; public static readonly string[] XPATH_H3 = { "//div[@class='sb_adsNv2']//li//h3", "//ol[@id='b_context']//li[@class='b_ad']//li//h2" }; public static readonly string[] XPATH_ADURL = { "//div[@class='sb_adsNv2']//li//a", "//ol[@id='b_context']//li[@class='b_ad']//li//a" }; public static readonly string[] XPATH_INFO = { "//div[@class='sb_adsNv2']//li//p", "//ol[@id='b_context']//li[@class='b_ad']//li//p" }; //b_results/b_ad public static readonly string[] XPATH_ROOT_TOP = { "results_container", "b_results" }; public static readonly string[] XPATH_CITE_TOP = { "//div[@class='sb_adsWv2']//li//cite", "//ol[@id='b_results']//li[@class='b_ad']//li//cite" }; public static readonly string[] XPATH_H3_TOP = { "//div[@class='sb_adsWv2']//li//h3", "//ol[@id='b_results']//li[@class='b_ad']//li//h2" }; public static readonly string[] XPATH_ADURL_TOP = { "//div[@class='sb_adsWv2']//li//a", "//ol[@id='b_results']//li[@class='b_ad']//li//a" }; public static readonly string[] XPATH_INFO_TOP = { "//div[@class='sb_adsWv2']//li//p", "//ol[@id='b_results']//li[@class='b_ad']//li//p" }; // public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)"); // public static readonly string BAITAI_ID = "003"; }
public static class Yahoo { public static readonly string UserAgent = UserAgentPC;
public static readonly string XPATH_ROOT = "sIn"; public static readonly string XPATH_CITE1 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/div[@class='a cf']"; public static readonly string XPATH_H31 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/h3"; public static readonly string XPATH_ADURL1 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/h3/a"; public static readonly string XPATH_INFO1 = "//div[@id='So3']/div[@class='bd']/div[@class='w']/p"; // public static readonly string XPATH_ROOT_TOP = "So1"; public static readonly string XPATH_CITE_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/div[@class='a cf']"; public static readonly string XPATH_H3_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/h3"; public static readonly string XPATH_ADURL_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/h3/a"; public static readonly string XPATH_INFO_TOP = "//div[@id='So1']/div[@class='bd']/div[@class='w']/p"; // public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)"); public static readonly string NullUrl = ">"; // public static readonly string BAITAI_ID = "002"; }
public static class Yahoo2 { public static readonly string UserAgent = UserAgentPC;
public static readonly string XPATH_ROOT_TOP = "contents"; public static readonly string XPATH_CITE_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/cite"; public static readonly string XPATH_H3_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/h2/a"; public static readonly string XPATH_ADURL_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/h2/a"; public static readonly string XPATH_INFO_TOP = "//div[@id='contents']/div[@class='cWrap']/div[@class='listWrap cf']/ul/li/p[@class='smr']"; // public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)"); public static readonly string NullUrl = ">"; // public static readonly string BAITAI_ID = "004"; }
public static class YahooM { public static readonly string UserAgent = UserAgentMobile;
public static readonly string XPATH_ROOT = "contentsInner"; public static readonly string XPATH_CITE = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/cite"; public static readonly string XPATH_H3 = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/h3"; public static readonly string XPATH_ADURL = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/h3/a"; public static readonly string XPATH_INFO = "//div[@id='contentsInner']//aside[@class='So']/div[@class='bd']/ul/li/p[@class='dtl']";

public static readonly string XPATH_ROOT_TOP = "contentsInner"; public static readonly string XPATH_CITE_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/cite"; public static readonly string XPATH_H3_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/h3"; public static readonly string XPATH_ADURL_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/h3/a"; public static readonly string XPATH_INFO_TOP = "//div[@id='contentsInner']/aside[@class='So next-cmm']/div[@class='bd']/ul/li/p[@class='dtl']"; // public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)"); public static readonly string NullUrl = ">"; // public static readonly string BAITAI_ID = "006"; }
public static class BaiDu {
public static readonly string UserAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko";
public static readonly string[] XPATH_ROOT = { "ec_im_container", "ec_im_container" }; //第一种情况好像就一种情况。 public static readonly string[] XPATH_CITE = { "//a/font[@size='-1' and @class]","//a/font[@size='-1' and @class]" }; //第一种情况 public static readonly string[] XPATH_H3 = { "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]", "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]" };//第一种情况 public static readonly string[] XPATH_ADURL = { "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]", "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-is-main-url]" }; public static readonly string[] XPATH_INFO = { "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-click]/font[1]", "//a[contains(@class,'EC_BL')and contains(@id,'dfs')and @data-click]/font[1]" }; // top info public static readonly string[] XPATH_ROOT_TOP = { "content_left", "content_left" }; public static readonly string[] XPATH_CITE_TOP = { "//table[@data-click]/tbody/tr/td//a[not(@data-is-main-url) and not(contains(@href,'tool'))]/span", "//div[@class and @style]/div/div/a/span[1]|//div/table/tbody/tr/td[2]/div//a/span[1]" }; //前下后上 public static readonly string[] XPATH_H3_TOP = { "//table/tbody/tr/td/a[ @data-is-main-url]", "//div[@class and @style]/div/div/h3" }; //前下后上 public static readonly string[] XPATH_ADURL_TOP = { "//table/tbody/tr/td/a[ @data-is-main-url]", "//table/tbody/tr/td/a[ @data-is-main-url]" }; //前下后上 public static readonly string[] XPATH_INFO_TOP = { "//table[@data-click and @class]/tbody/tr[3]/td/a[not(./span)]|//table[@data-click and @class]/tbody/tr/td/table/tbody/tr/td/div/font/a", "//div[@class and @style]/div/div[not(./span)]/a|//div/table/tbody/tr/td/div/font/a[not(./span)]" }; //前下部分广告后上 // //public static readonly Regex RegexAdUrl = new Regex(@"http[\S]*$"); // public static readonly string BAITAI_ID = "007"; }
public static class CnBing {
public static readonly string UserAgent = UserAgentPC;
public static readonly string[] XPATH_ROOT = { "b_context", "b_context" }; public static readonly string[] XPATH_CITE = { "//div[@class='sb_add sb_adTA']//cite", "//div[@class='sb_add sb_adTA']//cite" }; public static readonly string[] XPATH_H3 = { "//div[@class='sb_add sb_adTA']//h2/a", "//div[@class='sb_add sb_adTA']//h2/a" };//第一种情况 public static readonly string[] XPATH_ADURL = { "//div[@class='sb_add sb_adTA']//h2/a", "//div[@class='sb_add sb_adTA']//h2/a" }; public static readonly string[] XPATH_INFO = { "//div[@class='sb_add sb_adTA']//div[@class='b_caption']/p", "//div[@class='sb_add sb_adTA']//div[@class='b_caption']/p" }; // top info public static readonly string[] XPATH_ROOT_TOP = { "なし", "なし" }; public static readonly string[] XPATH_CITE_TOP = { "", "" }; //前下后上 public static readonly string[] XPATH_H3_TOP = { "", "" }; //前下后上 public static readonly string[] XPATH_ADURL_TOP = { "", "" }; //前下后上 public static readonly string[] XPATH_INFO_TOP = { "", "" }; //前下部分广告后上 // public static readonly Regex RegexAdUrl = new Regex(@"rturl=(http[\S]*$)"); // public static readonly string BAITAI_ID = "008"; }
public static class HaoSou {
public static readonly string UserAgent = UserAgentPC; // 右边的广告 public static readonly string[] XPATH_ROOT = { "side", "side" }; //获取范围 public static readonly string[] XPATH_CITE = { "//ul[@id='rightbox']/li/p/cite[not(contains(text(),' http://e.360.cn'))]|//div[@id='m-spread-left']//cite", "//ul[@id='rightbox']/li/p/cite[not(contains(text(),' http://e.360.cn'))]|//div[@id='m-spread-left']//cite" }; //第一种情况 public static readonly string[] XPATH_H3 = { "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'好搜推广'))]|//div[@id='m-spread-left']//h3/a", "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'好搜推广'))]|//div[@id='m-spread-left']//h3/a" };//第一种情况 public static readonly string[] XPATH_ADURL = { "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'好搜推广'))]|//div[@id='m-spread-left']//h3/a", "//ul[@id='rightbox']/li/h3/a[not(contains(text(),'好搜推广'))]|//div[@id='m-spread-left']//h3/a" }; public static readonly string[] XPATH_INFO = { "//ul[@id='e_idea_pp']/li//p|//ul[@id='rightbox']/li/p[not(contains(text(),'4000-360-360'))]", "//ul[@id='e_idea_pp']/li//p|//ul[@id='rightbox']/li/p[not(contains(text(),'4000-360-360'))]" }; // top info public static readonly string[] XPATH_ROOT_TOP = {"なし", "なし" }; public static readonly string[] XPATH_CITE_TOP = { "", "" }; //前下后上 public static readonly string[] XPATH_H3_TOP = { "", "" }; //前下后上 public static readonly string[] XPATH_ADURL_TOP = { "", "" }; //前下后上 public static readonly string[] XPATH_INFO_TOP = { "", "" }; //前下部分广告后上 // //public static readonly Regex RegexAdUrl = new Regex(@"http[\S]*$"); // public static readonly string BAITAI_ID = "009"; }
public static class Sogou { public static readonly string UserAgent = UserAgentPC; //right 部分 public static readonly string[] XPATH_ROOT = { "right" }; public static readonly string[] XPATH_CITE = { "//div[@class='bizr_fb']" };//绿色的url public static readonly string[] XPATH_H3 = { "//h3[@class='bizr_title']" };//#ad_leftresult_0 > h3:nth-child(1) public static readonly string[] XPATH_ADURL = { "//h3[@class='bizr_title']/a" };//.h3的url public static readonly string[] XPATH_INFO = { "//div[@class='bizr_ft']" }; //top 部分 public static readonly string[] XPATH_ROOT_TOP = { "promotion_adv_container" };//*[@id="promotion_adv_container"]/div/div public static readonly string[] XPATH_CITE_TOP = { "//div[contains(@class,'biz_rb')and @id]/div//cite" }; public static readonly string[] XPATH_H3_TOP = { "//h3[@class='biz_title']" }; public static readonly string[] XPATH_ADURL_TOP = { "//h3[@class='biz_title']/a" }; public static readonly string[] XPATH_INFO_TOP = { "//div[@class='crown_info_box' or @class='biz_ft']|//div[contains(@id,'box_id')]/table" };// ""
// //public static readonly Regex RegexAdUrl = new Regex(@"\*\*(http[\S]*$)"); //0 public static readonly string BAITAI_ID = "010"; public static readonly string NullUrl = ">"; }
}

xpath 参考的更多相关文章

【转】XPath 示例
XPath 示例其他版本本主题回顾整个 XPath 参考中出现的语法示例. 所有示例均基于 XPath 语法的示例 XML 文件 (inventory.xml). 有关在测试文件中使用 X ...
XPATH 带命名空间数据的读取
在XML中,很多情况下有命名空间,如果直接使用XPATH 读取是会读到空节点. 解决办法如下: InputStream is=loader.getResourceAsStream("com/ ...
HtmlCleanner结合xpath用法（转载）
HtmlCleaner cleaner = new HtmlCleaner(); TagNode node = cleaner.clean(new URL("http://finance.s ...
scrapy2_初窥Scrapy
递归知识:oop,xpath,jsp,items,pipline等专业网络知识,初级水平并不是很scrapy,可以从简单模块自己写. 初窥Scrapy Scrapy是一个为了爬取网站数据,提取结构性数 ...
较全的IT方面帮助文档
http://www.shouce.ren/post/d/id/108632 XSLT参考手册-新.CHMhttp://www.shouce.ren/post/d/id/108633 XSL-FO参考 ...
selenium java 浏览器操作
环境搭建 selenium 2.53 selenium-java-2.53.0.jar selenium-java-2.53.0-srcs.jar 原代码包拷贝的工程lib下,做build path ...
JDOM 操作XML
http://www.cnblogs.com/hoojo/archive/2011/08/11/2134638.html 可扩展标记语言——eXtensible Markup Language 用户可 ...
python爬虫 scrapy2_初窥Scrapy
sklearn实战-乳腺癌细胞数据挖掘 https://study.163.com/course/introduction.htm?courseId=1005269003&utm_campai ...
[开发笔记]-Linq to xml学习笔记
最近需要用到操作xml文档的方法,学习了一下linq to xml,特此记录. 测试代码: class Program { //参考: LINQ to XML 编程基础 - luckdv - 博客园 ...

随机推荐

java文件复制函数
在写代码中,经常会需要复制文件.这时候就可以把这些函数放到工具类里面. 下面是一个复制文件的函数: public class CopyFileUtil { private static String ...
百度地图TILE算法
Creating primary keyvar LLBAND2 = [75, 60, 45, 30, 15, 0]; var LL2MC2 = [[-.0015702102444, 111320.70 ...
如何正确并完全安装Visual Studio 2015企业版本?
http://blog.csdn.net/code_godfather/article/details/47381631 [简介] 常用网名: 猪头三出生日期: 1981.XX.XX 个人网站: ...
关于rails里集成测试assert_template的写法
assert_template后面只能跟随模板文件名,不能跟随命名路径.比如routes.rb: get 'login' => 'sessions#new' 在集成测试用例里,只能写成asser ...
16、总经理要阅读的书籍 - IT软件人员书籍系列文章
总经理是公司的一个领导角色.他主要负责公司级别的比如规划,战略等等内容.有些公司的总经理,比如软件公司的总经理,往往是一个大的业务员,将更多的大型的软件项目投取过来,让公司能够有钱赚,让公司员工能够跟 ...
使用MiniProfiler给Asp.net MVC和Entity Framework号脉(附源码)
在学习python开发框架pylons/pyramid的过程中,里面有个非常棒的页面性能监控功能,这样在开发过程中,你能清楚的知道当前页面的性能以及其它参数. 这里介绍一下如何给Asp.net MVC ...
Linux下防火墙开启相关端口及查看已开启端口
最近利用Apache Mina实现了一个http服务,发布到linux下发现无法访问,通过HttpClient来发送http请求时,报如下错误: Exception in thread "m ...
JavaScript(一)——简介（简单介绍）
1.JavaScript是个什么东西? 它是个脚本语言,需要有宿主文件,它的宿主文件是HTML文件. 2.它与Java什么关系? 没有什么直接的联系,Java是Sun公司(已被Oracle收购了),J ...
挖一挖C#中那些我们不常用的东西之系列（1）——ToDictionary,ToLookup
这个系列我们看看C#中有哪些我们知道,但是又不知道怎么用,又或者懒得去了解的东西,比如这篇我们要介绍的toDictionary 和ToLookup. 从图中我们看到有四个ToXXX的方法,其中ToAr ...
hadoop2.4.1集群搭建
准备Linux环境修改主机名: $ vim /etc/sysconfig/network NETWORKING=yes HOSTNAME=hadoop001 修改IP: # vim /etc/sys ...

xpath 参考

xpath 参考的更多相关文章

随机推荐

热门专题