/// <summary>
/// 分析用户主页信息
/// </summary>
/// <param name="home"></param>
public static CommonEntityLib.Entities.user.Entity AnalysisUserHome(string home)
{
try
{
CommonEntityLib.Entities.user.Entity weiboUser = new CommonEntityLib.Entities.user.Entity
{
ID = Regex.Match(home, "\"id\":\"(\\d*?)\"").Groups[1].Value
};
//分析uid
if (string.IsNullOrEmpty(weiboUser.ID))
{
return(null);
}
//获取分页爬博文所要信息
weiboUser.IDStr = Regex.Match(home, "'stageId':'(\\d*?)'").Groups[1].Value;
//获取用户发表的微博数量
var mblogNumStr = Regex.Match(home, "\"mblogNum\":\"(\\d*?)\"").Groups[1].Value;
weiboUser.StatusesCount = int.Parse(mblogNumStr);
//获取关注数量
var attNumStr = Regex.Match(home, "\"attNum\":\"(\\d*?)\"").Groups[1].Value;
weiboUser.FriendsCount = int.Parse(attNumStr);
//获取粉丝数量
var fansNumStr = Regex.Match(home, "\"fansNum\":\"(\\d*?)\"").Groups[1].Value;
weiboUser.FollowersCount = int.Parse(fansNumStr);
//获取用户性别
var genderStr = Regex.Match(home, "\"ta\":\"(.*?)\"").Groups[1].Value;
if (genderStr == @"\u4ed6")
{
weiboUser.Gender = "m";
}
//用户昵称
var nickname = Regex.Match(home, "\"name\":\"(.*?)\"").Groups[1].Value;
weiboUser.ScreenName = nickname.NormalU2C();
//用户所在省份
var location = Regex.Match(home, "\"nativePlace\":\"(.*?)\"").Groups[1].Value;
weiboUser.Location = location.NormalU2C();
//用户简介
var description = Regex.Match(home, "\"description\":\"(.*?)\"").Groups[1].Value;
weiboUser.Description = description.NormalU2C();
//头像地址
var profileImageUrl = Regex.Match(home, @"""profile_image_url"":""(?<url>.*?)""").Groups["url"].Value;
weiboUser.ProfileImageUrl = profileImageUrl.Replace(@"\", "");
return(weiboUser);
}
catch (Exception exception)
{
AnalyseCnPageLogger.Error(string.Format("分析用户主页信息失败\r\n{0}", home), exception);
return(null);
}
}
/// <summary>
/// 根据uid获取用户信息
/// </summary>
/// <param name="webLogin">登陆对象</param>
/// <param name="uid"></param>
/// <returns></returns>
public CommonEntityLib.Entities.user.Entity GetUserEntity(IWeiboLogin webLogin, string uid)
{
IWeiboLogin cnWeiboLogin = PlatformType.CN.GetWeiboLogin();
cnWeiboLogin.Web.Cookie = webLogin.Web.Cookie;
if (!CNHttpWork.UserExist(cnWeiboLogin.Web, uid))
{
return(null);
}
CommonEntityLib.Entities.user.Entity res = CNHttpWork.GetUserEntity(cnWeiboLogin.Web, uid);
if (res == null)
{
return(null);
}
string url = "http://weibo.com/" + uid + "/info";
string html = webLogin.Web.GetHTML(url);
if (string.IsNullOrEmpty(html))
{
return(null);
}
var temHtmlList = new Dictionary <string, string>();
if (html.Contains("<script>FM"))
{
var regex = new Regex(@"FM.view\((?<json>.*?)\)</script>");
if (regex.IsMatch(html))
{
foreach (Match match in regex.Matches(html))
{
string jsonStr = match.Groups["json"].Value;
if (!jsonStr.Contains("等级信息") && !jsonStr.Contains("基本信息") && !jsonStr.Contains("他的主页") && !jsonStr.Contains("Pl_Official_Headerv6__1"))
{
continue;
}
var json = DynamicJson.Parse(jsonStr);
string domid = json.domid;
try
{
string ht = json.html;
for (var i = (char)1; i < (char)32; i++)
{
ht = ht.Replace(i.ToString(CultureInfo.InvariantCulture), string.Empty);
}
ht = ht.Replace("\x7F", string.Empty);
temHtmlList.Add(domid, ht);
}
catch (Exception)
{
}
}
}
}
foreach (var thl in temHtmlList)
{
if (thl.Key.Contains("Pl_Official_Headerv6__1"))
{
res.Verified = thl.Value.Contains("verified.weibo.com");
if (res.Verified)
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(thl.Value);
var root = document.DocumentNode;
res.VerifiedReason = root.SelectSingleNode("div/div/div[@class='shadow S_shadow']/div[@class='pf_photo']/a/em")
.Attributes["title"].Value;
}
continue;
}
if (thl.Value.Contains("等级信息"))
{
// 等级
// <span>Lv.2</span>
string lv = Regex.Match(thl.Value, "<span>Lv\\.(?<lv>\\d*)</span>").Groups["lv"].Value;
int uRank;
if (!string.IsNullOrEmpty(lv) && int.TryParse(lv, out uRank))
{
res.Urank = uRank;
}
continue;
}
if (thl.Value.Contains("他的主页"))
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(thl.Value);
var root = document.DocumentNode;
var aTag = root.SelectSingleNode("div/div/table/tr/td/a");
if (aTag != null)
{
res.ProfileUrl = string.Format("http://weibo.com{0}", aTag.Attributes["href"].Value);
}
}
if (thl.Value.Contains("基本信息"))
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(thl.Value);
var root = document.DocumentNode;
HtmlNodeCollection categoryNodeList = root.SelectNodes("//ul[@class='clearfix']/li");
if (categoryNodeList == null)
{
continue;
}
foreach (HtmlNode htmlNode in categoryNodeList)
{
HtmlNodeCollection spans = htmlNode.SelectNodes("span");
if (spans.Count != 2)
{
continue;
}
string txt = spans[0].InnerText;
string val = spans[1].InnerText;
if (txt.Contains("博客"))
{
res.Url = val;
continue;
}
if (txt.Contains("个性域名"))
{
res.Domain = val;
continue;
}
if (txt.Contains("注册时间"))
{
res.CreatedAt = val;
continue;
}
if (txt.Contains("所在地"))
{
res.Location = val;
if (!string.IsNullOrEmpty(res.Location) && res.Location.Split(' ').Count() >= 2)
{
string provice = res.Location.Split(' ')[0].Trim();
string city = res.Location.Split(' ')[1].Trim();
res.Province = ProvinceToCode(provice);
res.City = CityToCode(city);
}
continue;
}
if (txt.Contains("标签"))
{
HtmlNodeCollection tags = spans[1].SelectNodes("a");
if (tags != null)
{
res.Remark = string.Join(",", tags.Select(c => c.InnerText));
}
}
}
}
}
return(res);
}
/// <summary>
/// 分析用户主页信息
/// </summary>
/// <param name="home"></param>
public static CommonEntityLib.Entities.user.Entity AnalysisUserHome(string home)
{
try
{
CommonEntityLib.Entities.user.Entity weiboUser = new CommonEntityLib.Entities.user.Entity
{
ID = Regex.Match(home, "\"id\":\"(\\d*?)\"").Groups[1].Value
};
//分析uid
if (string.IsNullOrEmpty(weiboUser.ID))
return null;
//获取分页爬博文所要信息
weiboUser.IDStr = Regex.Match(home, "'stageId':'(\\d*?)'").Groups[1].Value;
//获取用户发表的微博数量
var mblogNumStr = Regex.Match(home, "\"mblogNum\":\"(\\d*?)\"").Groups[1].Value;
weiboUser.StatusesCount = int.Parse(mblogNumStr);
//获取关注数量
var attNumStr = Regex.Match(home, "\"attNum\":\"(\\d*?)\"").Groups[1].Value;
weiboUser.FriendsCount = int.Parse(attNumStr);
//获取粉丝数量
var fansNumStr = Regex.Match(home, "\"fansNum\":\"(\\d*?)\"").Groups[1].Value;
weiboUser.FollowersCount = int.Parse(fansNumStr);
//获取用户性别
var genderStr = Regex.Match(home, "\"ta\":\"(.*?)\"").Groups[1].Value;
if (genderStr == @"\u4ed6")
weiboUser.Gender = "m";
//用户昵称
var nickname = Regex.Match(home, "\"name\":\"(.*?)\"").Groups[1].Value;
weiboUser.ScreenName = nickname.NormalU2C();
//用户所在省份
var location = Regex.Match(home, "\"nativePlace\":\"(.*?)\"").Groups[1].Value;
weiboUser.Location = location.NormalU2C();
//用户简介
var description = Regex.Match(home, "\"description\":\"(.*?)\"").Groups[1].Value;
weiboUser.Description = description.NormalU2C();
//头像地址
var profileImageUrl = Regex.Match(home, @"""profile_image_url"":""(?<url>.*?)""").Groups["url"].Value;
weiboUser.ProfileImageUrl = profileImageUrl.Replace(@"\", "");
return weiboUser;
}
catch (Exception exception)
{
AnalyseCnPageLogger.Error(string.Format("分析用户主页信息失败\r\n{0}", home), exception);
return null;
}
}
private static CommonEntityLib.Entities.user.Entity AnalyseUserPage(string html)
{
var location = LocationRegex.Match(html).Groups["location"].Value;
string province;
string city;
if (!string.IsNullOrEmpty(location))
{
province = location.Split(' ')[0];
province = ProvinceToCode(province);
city = "0";
if (location.Contains(" "))
{
city = location.Split(' ')[1];
city = CityToCode(city);
}
}
else
{
province = "11";
city = "0";
}
CommonEntityLib.Entities.user.Entity userInformationEntity = new CommonEntityLib.Entities.user.Entity
{
ScreenName = NicknameRegex.Match(html).Groups["nickname"].Value,
Gender = GenderRegex.Match(html).Groups["gender"].Value == "男" ? "m" : "f",
Description = DescriptionRegex.Match(html).Groups["description"].Value,
Province = province,
City = city
};
return userInformationEntity;
}
