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

Android 短信的还原
上篇文章讲到<Android 短信的备份>,本文主要实现Android 短信的还原,即是将一条布局文件: <RelativeLayout xmlns:android="h ...
[读书笔记] CSS权威指南1: 选择器
通配选择器可以与任何元素匹配,就像是一个通配符 /*每一个元素的字体都设置为红色*/ * { color: red; } 元素选择器指示文档元素的选择器. /*为body的字体设置为红色*/ bo ...
[在线] html 转 pdf
http://www.htm2pdf.co.uk/
Grunt安装配置教程:前端自动化工作流
Grunt这货是啥? Grunt 是一个基于任务的 JavaScript 项目命令行构建工具. 最近很火的前端自动化小工具,基于任务的命令行构建工具 http://gruntjs.com Grunt能 ...
java环境变量 windows centos 安装jdk
windows: 1.安装jdk,注意不是jre 2. 计算机→属性→高级系统设置→高级→环境变量,选择下面的那个系统环境变量 3. 系统变量→新建 JAVA_HOME 变量 . 变量值填写jdk的安 ...
双十一来了，别让你的mongodb宕机了
好久没过来吹牛了,前段时间一直赶项目,没有时间来更新博客,项目也终于赶完了,接下来就要面临双十一这场惊心动魄的处女秀考验, 我们项目中会有一个wcf集群,而集群地址则放在mongodb中,所以mong ...
Java并发之ThreadPoolExecutor 线程执行服务
package com.thread.test.thread; import java.util.concurrent.ExecutorService; import java.util.concur ...
烂泥：使用snmpwalk采集设备的OID信息
本文由秀依林枫提供友情赞助,首发于烂泥行天下. 打算开始学习有关监控方面的知识,但是现在很多监控系统都是根据SNMP进行的.而SNMP监控的性能指标很多都是通过snmpwalk采集设备的OID信息得到 ...
springmvc+log4j操作日志记录,详细配置
没有接触过的,先了解一下:log4j教程部分内容来:log4j教程感谢! 需要导入包: log包:log4j-12.17.jar 第一步:web.xml配置 <!-- log4j配置,文件路 ...
高性能MySQL笔记第4章 Schema与数据类型优化
4.1 选择优化的数据类型通用原则更小的通常更好前提是要确保没有低估需要存储的值范围:因为它占用更少的磁盘.内存.CPU缓存,并且处理时需要的CPU周期也更少. 简单就好简 ...

xpath 参考

xpath 参考的更多相关文章

随机推荐

热门专题